Sampling Insects from Trees - LAC Biosafety

INSECT SAMPLINGIN FORESTECOSYSTEMS

METHODS IN ECOLOGYSERIES LISTGeographic Information Systems in Ecology1997, Carol A. Johnston, University of MinnesotaResearchers will find this an invaluable guide to applying and getting the most out ofGeographical Information Systems, one of the most revolutionary and important tools thathave become available to ecological researchers in recent years.An Introduction to Ecological Modelling: Putting Practice into Theory1997, M. Gillman & Rosie Hails, Open University & CEH Oxford“Teachers of courses on ecological modelling will find [this book] a useful source-book at a competitiveprice.”This book aims to open up the exciting area of ecological modeling to a much wider audience.Stable Isotopes in Ecology and Environmental Science1994, edited by Kate Lajtha & Robert Michener, Oregon State University & Boston UniversityThis book, written by two of the leading researchers in the field, explains the background tostable isotope methodology and discuss the use of the methods in varying ecologicalsituations.Geographical Population Analysis: Tools for the Analysis of Biodiversity1994, Brian A. Maurer, Michigan State UniversityThis book discusses methods and statistical techniques that can be used to analyze spatialpatterns in geographic populations. These techniques incorporate ideas from fractal geometryto develop measures of geographic range fragmentation, and can be used to ask questionsregarding the conservation of biodiversity.Molecular Methods in Ecology2000, edited by Allan J. Baker, Royal Ontario MuseumThis book provides both postgraduates and researchers with a guide to choosing andemploying appropriate methodologies for successful research in the field of molecular ecology.Biogenic Trace Gases: Measuring Emissions from Soils and Water1995, edited by P.A. Matson & R.C. Harriss, University of California & University of NewHampshire“The present volume ...will serve as an important tool box for researchers and graduate students in thisdiscipline, and will provide both a range of techniques for field measurements and a conceptualframework for extrapolation strategies.”This how-to guide details the concepts and techniques involved in the detection andmeasurement of trace gases, and the impact they have on ecological studies.Ecological Data: Design, Management and Processing2000, edited by William Michener & James Brunt, University of New MexicoThis book provides a much-needed resource for those involved in designing and implementingecological research, as well as students who are entering the environmental sciences.Population Parameters: Estimation for Ecological Models1999, Hamish McCallum, University of QueenslandThis book brings together a diverse and scattered literature, to provide clear guidance on howto estimate parameters for models of animal populations.

METHODS IN ECOLOGYInsect Sampling inForest EcosystemsEDITED BYSIMON R. LEATHERDepartment of Biological SciencesImperial College of Science, Technology and MedicineSilwood ParkAscotUKSERIES EDITORSJ.H. LAWTON CBE, FRSNatural Environment Research CouncilSwindon, UKG.E. LIKENSInstitute of Ecosystem StudiesMillbrook, USA

© 2005 by Blackwell Science Ltda Blackwell Publishing companyBLACKWELL PUBLISHING350 Main Street, Malden, MA 02148-5020, USA108 Cowley Road, Oxford OX4 1JF, UK550 Swanston Street, Carlton, Victoria 3053,AustraliaThe right of Simon Leather to be identified as theAuthor of the Editorial Material in this Work hasbeen asserted in accordance with the UKCopyright, Designs, and Patents Act 1988.All rights reserved. No part of this publication maybe reproduced, stored in a retrieval system,or transmitted, in any form or by any means,electronic, mechanical, photocopying, recordingor otherwise, except as permitted by the UKCopyright, Designs, and Patents Act 1988, withoutthe prior permission of the publisher.A catalogue record for this title is available from theBritish Library.Set in 9 1 /2 on 12pt Meridienby SNP Best-set Typesetter Ltd., Hong KongPrinted and bound in the United Kingdomby MPG Books, Bodmin, CornwallThe publisher’s policy is to use permanent paperfrom mills that operate a sustainable forestrypolicy, and which has been manufactured frompulp processed using acid-free and elementarychlorine-free practices. Furthermore, the publisherensures that the text paper and cover board usedhave met acceptable environmental accreditationstandards.For further information onBlackwell Publishing, visit our website:www.blackwellpublishing.comFirst published 2005 by Blackwell Science LtdLibrary of Congress Cataloging-in-Publication DataInsect sampling in forest ecosystems / edited bySimon R. Leather.p. cm. – (Methods in ecology)Includes bibliographical references (p. ).ISBN 0-632-05388-7 (pbk. : alk. paper)1. Forest insects–Research–Methodology.2. Forest surveys. 3. Ecological surveys.4. Sampling (Statistics) I. Leather, S. R.(Simon R.) II. Series.SB761.I56 2005634.9¢67¢072–dc222004009772

ContentsContributors, viiMethods in Ecology series, ixPreface, xi1 Sampling theory and practice, 1Simon R. Leather and Allan D. Watt2 Sampling insects from roots, 16Alan C. Gange3 Pitfall trapping in ecological studies, 37B.A. Woodcock4 Sampling methods for forest understory vegetation, 58Claire M.P. Ozanne5 Sampling insects from trees: shoots, stems, and trunks, 77Martin R. Speight6 Insects in flight, 116Mark Young7 Techniques and methods for sampling canopy insects, 146Claire M.P. Ozanne8 Sampling methods for water-filled tree holes and their artificialanalogues, 168S.P. Yanoviak and O.M. Fincke9 Sampling devices and sampling design for aquatic insects, 186Leon Blaustein and Matthew Spencerv

viCONTENTS10 Methods for sampling termites, 221David T. Jones, Robert H.J. Verkerk, and Paul Eggleton11 Parasitoids and predators, 254Nick MillsIndex, 279

ContributorsLeon BlausteinCommunity Ecology Laboratory, Institute of Evolution, University of Haifa,Haifa 31905, IsraelPaul EggletonTermite Research Group, Department of Entomology, The Natural HistoryMuseum, Cromwell Road, London, SW7 5BD, UKO.M. FinckeDepartment of Zoology, University of Oklahoma, Norman, Oklahoma 73019,USAAlan C. GangeSchool of Biological Sciences, Royal Holloway, University of London, Egham,Surrey, TW20 0EX, UKDavid T. JonesTermite Research Group, Department of Entomology, The Natural HistoryMuseum, Cromwell Road, London, SW7 5BD, UKSimon R. LeatherDepartment of Biological Sciences, Imperial College of Science, Technologyand Medicine, Silwood Park, Ascot, Berkshire, SL5 7PY, UKNick MillsInsect Biology, Wellman Hall, University of California at Berkeley, Berkeley,California 94720-3112, USAClaire M.P. OzanneCentre for Research in Ecology and the Environment, School of Life Sciences,Roehampton University of Surrey, West Hill, London SW15 3SN, UKvii

viiiCONTRIBUTORSMartin R. SpeightDepartment of Zoology, University of Oxford, South Parks Road, Oxford, UKMatthew SpencerCommunity Ecology Laboratory, Institute of Evolution, University of Haifa,Haifa 31905, IsraelCurrent address: Department of Mathematics and Statistics, DalhousieUniversity, Halifax, Nova Scotia, B3H 3J5, CanadaRobert H.J. VerkerkDepartment of Biology, Imperial College of Science, Technology and Medicine,Silwood Park, Ascot, Berkshire, SL5 7PY, UKAllan D. WattCentre for Ecology and Hydrology, Hill of Brathens, Glassel, Banchory,Aberdeenshire AB31 4BW, UKB.A. WoodcockCentre for Agri-Environment Research, Department of Agriculture,University of Reading, RG6 6AR, UKS.P. YanoviakDepartment of Zoology, University of Oklahoma, Norman, Oklahoma 73019,USACurrent address: Florida Medical Entomology Laboratory, 200 9 th Street SE,Vero Beach, FL 32962, USAMark YoungCulterty Field Station, Department of Zoology, University of Aberdeen,Newburgh Ellon, Aberdeenshire, AB41 OAA, UK

Methods in Ecology seriesSeries editorsProfessor John H. Lawton is Chief Executive of the UK Natural EnvironmentResearch Council, and holds honorary professorships at Imperial CollegeLondon and the University of York. He is a Fellow of the Royal Society, and hasreceived numerous national and international prizes. Professor Lawton is author,co-author, or editor of six books, a former editor of Ecological Entomology,and has published over 300 scientific articles.Dr Gene E. Likens is President and Director of the Institute of EcosystemStudies in Millbrook, New York, and also holds professorships at CornellUniversity, Yale University, and Rutgers University. He received the 2001National Medal of Science and has received eight honorary degrees. Dr Likensis also author, co-author, or editor of 15 books, and of over 450 publishedscientific articles.About the seriesThe Methods in Ecology series is a useful and ever-growing collection of booksaimed at helping ecologists to choose and apply an appropriate methodology fortheir research. The series is edited by two internationally renowned ecologists,Professor John H. Lawton and Dr Gene E. Likens, and aims to address the needfor a set of concise and authoritative books to guide researchers through thewide range of methods and approaches that are available to ecologists.Each volume is not simply a recipe book, but takes a critical look at differentapproaches to the solution of a problem, whether in the laboratory or in thefield, and whether involving the collection or the analysis of data.Rather than reiterate established methods, authors are encouraged to featurenew technologies, often borrowed from other disciplines, that ecologistscan apply to their work. Innovative techniques, properly used, can offer particularlyexciting opportunities for the advancement of ecology.ix

xMETHODS IN ECOLOGY SERIESThe series strives to be at the cutting edge of the subject, introducing ecologiststo a wide range of techniques that are currently rarely used, but deserve to bebetter known, or it seeks to provide up-to-date methods in more familiar areas.Its main purpose is not only to provide instruction in basic methods (the “howto”), but also to explain the benefits and limitations of each method (the “whythis way?”), as well as showing how to interpret the results, what they mean,and generally to put them in the context of the discipline.Much is now expected of the science of ecology, as humankind struggles with agrowing environmental crisis. Good methodology alone never solved any problem,but bad or inappropriate methodology can only make matters worse.Ecologists now have a powerful and rapidly growing set of methods and toolswith which to confront fundamental problems of a theoretical and applied nature.We hope that this series will be a major contribution towards making thesetechniques known to a much wider audience.

PrefaceInsect sampling, although firmly based on standard ecological census techniques,presents special problems that are not faced by other ecologists. Withthe small size, varied life cycles, rapid rates of increase, and ingenious adaptationsto habitats of insects, ecological entomologists face problems that aresomewhat different to those faced by vertebrate or plant ecologists. That said,these same features make working with insects more amenable than working,for example, with large mammals.Within the entomological world there are many different groups of specialists— those that work in agricultural systems, in desert systems, or with particulargroups of insects. Many of these overlap in their approach and methodology,but some are unique and require specialist knowledge. One such specializationis forest entomology, another is aquatic entomology.Forest ecosystems, whether natural or manmade, present special problems tothe ecologists working beneath their canopies. In contrast to grassland, arable,and moorland ecosystems, where the scientist can stand above the study areaand view the system in large patches, forest ecologists are towered over by theirstudy substrate. Trees are large, dominate the canopy, and are not as amenableto sampling as herbaceous plants. The forest floor, often criss-crossed by surfaceor near-surface roots, also presents its own particular hazards to the researcher.In plantation forests, ridges, furrows, and drains mean that soil sampling, althoughsuperficially a similar exercise to that conducted in an arable ecosystem,is again not quite as simple. Root grafting makes sub-soil sampling onerous inthe extreme.Tropical forests are perhaps even more difficult to work in; the profusion ofendophytic vegetation and the multi-layered structure of the canopy in manytypes of forest can make sampling a nightmare.Study in forest ecosystems is an important part of ecology. In tropical naturalforest ecosystems much work is performed in attempts to quantify the diversityof these unique systems. In temperate and boreal forests equally importantwork is conducted. Furthermore, with the massive increase in plantationforestry (tropical plantation forestry has increased more than threefold inthe last decade), the need to sample for survey and protection purposes hasdramatically increased. This book, although covering all aspects of insect samplingwithin all ecosystems, has a definite bias towards forest ecosystems. Therexi

xiiPREFACEare, however, many common features of insect sampling that can be applied toother ecosystems and every chapter brings these together in an integratedwhole. Special cases do of course exist and each of these gets a chapter to itself.This book brings together the collective expertise obtained over many yearsof intensive fieldwork in tropical and temperate ecosystems by a number ofwell-known entomologists. Each chapter, as well as dealing with samplinga particular stratum of the canopy or specialized group of insects, presents acomprehensive guide to running experiments within and beneath the forestcanopy. Many potentially useful pieces of work conducted in forest ecosystemshave fallen at the final hurdle – the translation of field data to the printed page.Unless surveys and field experiments are realistically designed within a soundbut manageable framework they are doomed to failure. In addition, the failureof many ecologists working in agricultural ecosystems or on parkland trees torecognize the constraints imposed on ecologists working within large scaleforest ecosystems must be redressed.This book attempts to highlight the problems faced by entomologists workingin different ecosystems and to suggest ways in which their methodology can bemodified so as to be understood by ecologists and become accepted within thegeneral fields of ecology and entomology.Simon Leather graduated from the University of Leeds in 1977 with a firstclasshonors degree in Agricultural Zoology. He followed that with a PhD inaphid ecology at the University of East Anglia. He is currently Reader in AppliedEcology in the Department of Biological Sciences at Imperial College’s SilwoodPark campus. He has been researching the population biology of agriculturaland forest pests, particularly insects, for over 25 years. Ten of these years werespent with the British Forestry Commission, where he learnt how to canopysamplethe hard way! He has written and edited several books and has, since1996, been editor (latterly co-editor) of Ecological Entomology.

CHAPTER 1Sampling theory and practiceSIMON R. LEATHER AND ALLAN D. WATTIntroductionThis chapter deals with the need to sample insects, the theory underlying sampling,the need to calibrate samples, and the design of sampling programs, and itevaluates the use of different sampling techniques.Why sample?Sampling is a scientist’s way of collecting information, and the majority of samplingis undertaken to answer specific questions. This was not always the case.Sampling as we know it was first done in a haphazard manner and bore little relationto what we would call sampling today. The first samples taken were basicallya by-product of the desire of natural historians to collect information aboutthe world around them.A brief history of information collectionThe history of information collection can be classified into three mainstages. There is a little overlap, but in the main we can recognize three separatephases.1 The collectorsThis can be classified as the pin, stuff, and draw era. As travel became relativelysafer and people became more interested in what lay beyond their horizonsthere was a rapid expansion both in the number of naturalists traveling to othercontinents and in the number of people employed by naturalists to collect andreturn specimens to Europe. Drawing was also a popular activity and to a certainextent filled the niche now occupied by photography. Many ships’ officerswere accomplished amateur artists and many had an interest in the flora andfauna of the countries they visited. This phase resulted in the acquisition ofmany thousands of specimens of plants and animals, either stuffed, pickled,1

2 CHAPTER 1pressed, or pinned, accompanied by many sketches of the organisms in theirnative settings, although as the majority of the artists had no scientific trainingthese drawings and paintings sometimes bear only a passing resemblance toreality.Although this resulted in the garnering of many examples of plants and animalsthere was little knowledge of the biology or ecology of the organisms. Thisled to a great deal of confusion, particularly in the field of entomology where thesometimes complicated life cycles such as those occurring in dimorphic andpolymorphic species such as aphids led to the misclassification of many species.For example, several aphid species were classified as being more than onespecies, depending on which host plant they were removed from or dependingon which stage of the life cycle they were in at the time of their collection. Othersimilar mistakes occurred in the Lepidoptera where confusion over the identityof members of several mimic species lasted for some time until the larval stageswere recognized. Ladybird beetles such as Adalia bipunctata and Adalia decempunctata,now well known as being extremely variable in their different colorforms were also once misidentified as separate species until their life historieswere fully elucidated.2 The observersThere were of course some collectors who also collected observations. Manynatural historians, as well as having a keen eye for the chase and for sketching,also felt the need to observe the behavior of the animals that they were collecting.These are exemplified by Darwin and Fabre who, as well as making detailedcollections of specimens, also spent many hours observing and recording patternsof behavior. These observations provided plenty of information on the biologyof the species, but as much of it was centered on individuals and theirinteractions with other individuals of the same species did not provide a greatdeal of information on their place in ecosystems, did not always provide accurateinformation about mortality factors and was confounded by a great deal ofunrecognized environmental “noise.”3 Experimental/controlled samplingThe next great step forward in the field of information collecting was the useof experimental studies in controlled conditions. For example, by studyingthe biology of an insect in the laboratory, it is possible to obtain detailed knowledgeof life history parameters such as fecundity, longevity, etc., and it is alsopossible to assign specific values to mortality factors, albeit in a far from naturalenvironment. The main drawback of this type of study is that environmentalvariability is lost and the natural impact of mortality and natality factors iscompromised.The best option is to combine laboratory methods and natural conditions, and

SAMPLING THEORY AND PRACTICE 3to do experimental and manipulative work in the field. The need to obtain accurateestimates of animal numbers in the field led to the development of thetheory of sampling and, incidentally, to the use of statistics in the biologicalsciences.Estimating abundance and predicting population dynamicsThe major use of sampling in entomology is to determine the number of insectsin a given area or location, usually for pest control or conservation purposes.The other main reason for wanting information on insect numbers is to increaseour understanding of the population dynamics of the insect(s) in question andto make predictions of their future abundance.Before one can make a prediction, one needs to know how many insectsthere are in the first place. This is equally true, whether one is going to control apest or to conserve an endangered species. It is not a sound practice (althoughsome modelers do it) to conjure a number out of thin air. There is also a need toknow what factors affect those numbers. There are basically six facts about thepopulation of an insect that are required before sensible predictions of the populationdynamics can be made:1 density — an expression of the species’ abundance in an area;2 dispersion (distribution) — the spatial distribution of individuals of aspecies;3 natality — birth rate;4 mortality — death rate;5 age structure — the relative proportions of individuals in different ageclasses;6 population trend — the trend in the abundance of the study species.It is only from this sort of information that one can start to make some sort of inferencesabout the population dynamics of the insect. The only reliable way toobtain this type of information is to sample.Sampling methodsTo sample an insect requires both a sampling technique and a sampling program.These are different things, although it is noticeable that even in the scientificliterature the two terms are quite often used interchangeably.Sampling techniquesA sampling technique is the method used to collect information from a singlesampling unit. Therefore the focus of a sampling technique is on the equipmentand/or the way the count is accomplished.

4 CHAPTER 1Sampling programsA sampling program, on the other hand, is the procedure for employingthe sampling technique to obtain a sample and make an estimate. Samplingprograms direct how a sample is to be taken, including sampling unit size, numberof sample units, spatial pattern of obtaining sampling units, and timing ofsamples.Before, however, one starts to think about either the sampling unit or thesampling program it is necessary to know something about the insect that isgoing to be sampled.An important starting point is to find out about the life cycle and biology ofthe insect and especially about where it is likely to be found. There is no point insampling terrestrial habitats for something that lives in water. Some insectshave marked changes in distribution during the course of a year, so it is importantthat this is taken into account before any sampling program is undertaken.For example, the bird cherry aphid Rhopalosiphum padi lives on grasses in summerand on the bird cherry tree Prunus padus in autumn, winter, and spring(Dixon & Glen 1971). For those insects that show seasonal changes in habitatuse, it is essential to know when the changeover from habitat to habitat takesplace, at least approximately. Sampling will of course have to be conducted inboth habitats for some period of time to pinpoint this changeover. Thus, a goodknowledge of the biology and ecology of the insect is very important. Anotherimportant consideration is the likely cost of the sampling in terms of both timeand money.Deciding on the approachSampling tools/techniquesThere are a number of tools that can be used to sample insect populations. Onecan sample aerially, for example using suction traps. These are used throughoutBritain by Rothamsted Insect Survey (Knight et al. 1992) and in many otherparts of the world. They are primarily used to trap aphids, and sample at twostandard heights, 1.2 m and 12.2 m. Sticky traps, either with or without attractants,can be used for almost anything that flies and is too weak to get off thesticky board. Light traps are also commonly used to sample aerial populations,although the insects mainly caught are night-flying Lepidoptera. There are variousintercept traps that are used to catch beetles, flies, aphids, and other insects,such as yellow water traps, Malaise traps and window traps. These are discussedfurther in other chapters. It is useful to note that the range of technology is quitevast. A great deal of effort can go into the design and evaluation of traps, and thisis often an essential part of the design of a sampling program. For example, someinsects are more readily caught by certain types of trap (Heathcote 1957,

SAMPLING THEORY AND PRACTICE 5Niemelä et al. 1986). The behavior of the insect will largely determine the typeof trap or sampling tool used (see later chapters).Passive traps versus active trapsSampling and trapping techniques can broadly be classified into two types —passive or active.A passive trap is one that should be neutral and depends entirely on chance.An active trap depends on the behavior of the insect but takes advantage of thebehavior and attracts the insect to the trap by chemical lures, baits, or evencolor — all of which can be varied to give different trapping efficiencies and targets(Finch 1990).What is the advantage of a passive trap over an active trap?Passive traps allow unbiased estimates of insect populations because the insectsare neither attracted nor repelled by the traps. For example, although aerial suctiontraps are powered by a motor and draw air into the collecting tube they areessentially passive in action as they depend on the insect flying into the ambit ofthe trap and do not depend on it being attracted to the area. A big drawback tothe use of passive traps is that they are not very useful at low densities. This is aparticular problem when programs have been designed to monitor the abundanceof occurrence of pests — for example insects on quarantine lists. In thosecases an attractant trap is a much better alternative as they are better detectiontools. They do, however, give a biased estimate of the density per unit area andconversion factors then have to be applied. Thus, when using attractant traps,particularly if they are being used to obtain population estimates, it is vitally importantto know over what range the trap is effective and whether there are directionalas well as distance effects.Direct habitat samplingSometimes, particularly if one is working with a pest species, the most usefulmethod of sampling is one that estimates the population size in the habitat —e.g. a crop or nature reserve. Indirect methods of sampling — e.g. aerial samplingwith a suction trap or pan trapping in a field — only indicate what is present inthe area, and do not tell you what is actually on the plant or in the soil. It will tellyou what is there and gives some idea of whether there are many or few, but unlessit has been backed up by calibration studies it does not tell you how manyinsects there are per plant or per unit area of habitat, or whether they are actuallypresent on the area that you are concerned with; they may just have beenen route somewhere when they were caught. This is particularly true of migratoryinsects.

6 CHAPTER 1What methods are available and what determines their use?Particular methods are dealt with in the following chapters. Here, we considerthe rationale behind the selection of available techniques. When sampling onthe ground there are a number of methods available. Quadrats may be used forsome insects — e.g. predatory surface-active beetles, aphids on plants, etc. However,searching a surface quadrat is no good for cryptic, soil-dwelling nocturnalinsects such as the large pine weevil Hylobius abietis. Whole-plant searches orpart-plant searches are also useful, and if the insects are relatively sedentary cangive good population estimates. Pitfall trapping is useful for surface-active insects,particularly those active during the night, (see Chapter 3). Soil extractionmethods can be used for soil or root-dwelling species (see Chapter 2).Destructive versus non-destructive samplingIf whole plants or small areas of habitat are to be sampled, two approaches canbe used — destructive and non-destructive. Destructive sampling involves theremoval of the sample unit for later assessment (see below), whereas with nondestructivesampling the sample unit is searched or sampled in situ.Both these approaches have their merits and disadvantages. For example,suppose you are counting aphids on a plant. You could sample destructively —i.e. remove the plant or part of the plant from the ground or from the main stem,and bring it back into the laboratory. Alternatively, you could sample nondestructively— for example, examine 100 leaves and record what is found.Destructive sampling is more accurate as the insects are less likely to escapeduring the counting process. One cannot, however, go back and sample thesame plant or area again. This is a particular problem if there are only a limitednumber of plants to begin with, or if the habitat type is rare and easily disturbed.If one is sampling from a large number of uniform plants such as a field of leeksor a forest plantation, destructive sampling may be a useful technique. A disadvantageof destructive sampling is that it is more time consuming, and is thusnot useful in situations where a quick estimate of insect numbers is required,say for a control operation. It is possible with destructive sampling, however, topostpone sampling by storage, be it in the freezer or in some sort of preservative.This is particularly useful in those situations where a large number of sampleshave to be taken in a limited time period and where there is no need for a swiftresult. It means that the actual counting of the insects can be saved for a less busytime of year — e.g. the winter.Non-destructive sampling, on the other hand, does allow re-sampling of theplants and habitats on a frequent or regular basis. This is very useful in sensitiveareas and when local population dynamics are being studied. Non-destructivesampling tends to be quicker than destructive sampling and causes less disturbanceto the habitat. It does however depend on the insects being relativelysedentary or slow to respond to disturbance. Thus the counts will tend to be

SAMPLING THEORY AND PRACTICE 7underestimates. To counter this, as non-destructive sampling is fairly quick,more samples can be taken, although this does not entirely solve the problemsof underestimation.How many samples?There are a number of factors that determine the number of samples that aretaken. The first requirement is to be sure that the sample taken is representativeof the population that is being sampled. To ascertain this it may be necessary toperform stratified sampling. It is not always safe to assume that insects are systematicallydistributed. A number of different distributions are possible. Thepopulation could be randomly distributed, uniformly distributed, or even in anaggregated (clumped or contagious) distribution (Fig. 1.1). These factors allneed considering. It is possible to determine what distribution the populationhas by using the following approach.Variance — mean ratiosThe dispersion of a population determines the relationships between the variances 2 and the arithmetic mean m thus:1 random distribution — the variance is equal to the mean — s 2 =m;2 regular (uniform) distribution — the variance is less than the mean — s 2 m.The distribution of the organism can have a marked influence on the way inwhich you might sample. Take, for example, a site in which the organism youare going to sample has a soil-dwelling pupae. The easiest approach is to do asimple line transect from one corner of the field to another, or if you are(a) (b) (c)Fig. 1.1 Sampling different distributions with a common sample plan: (a) random, (b)uniform or regular, and (c) aggregated distribution. Note that the values returned, even inthis simple example, are very different for the aggregated distribution.

8 CHAPTER 1concerned about the slope or topography you might do another transect acrossfrom the other two corners in the form of an X. Depending on the distribution ofthe organism you may get totally different answers (Fig. 1.1).Stratified samplingSuppose we know that the population we are going to sample varies systematicallyacross the area we are going to sample.We may know, for example, that the insect does not occur in high densities inparticular areas — e.g. where there are lots of stones — but does occur in highnumbers in areas where there is a lot of sand. It may even be something morespecific: for example, if we were sampling trees for insects, we might know thatthe distribution of the insect within the crown of the tree is not uniform. Thepine beauty moth, for example, lays most of its eggs in the upper third of thetree, so one can get a good estimate of the population by just counting eggsfound on the first five whorls and then either multiplying up or just takingthe figure obtained to be representative of the population (Watt & Leather1988a). It really depends on what one is sampling for. If one is sampling forpredictive purposes than the first five whorls is good enough; on the otherhand, if the sampling is part of a detailed population study, then the samplingneeds to be more thorough and to take more account of the distribution of theorganism. Thus, for the pine beauty moth, a branch is taken from every otherwhorl, the number of branches per whorl counted, and the counts are thenmultiplied up accordingly. If one had a very large scale study, one might just takea third-whorl branch at random and multiply up from there (Leather 1993).Of course one would have to have done some whole-tree sampling first todetermine what all the various multiplication factors were going to be. For example,with winter moth eggs on Sitka spruce there is a marked difference inegg distribution, not just in relation to tree height, but also within the branches(Watt et al. 1992) (Fig. 1.2). One could therefore work out various samplingschemes to use.In essence, though, before a sampling scheme can be devised, one needs to dosome preliminary sampling to get a feel for what number or size of sample onewill require.In general, the more samples that are taken the more precise the populationestimates will be. However, time and expense are always constraining factors.Thus the usual approach is to decide on the lowest number of samples that canbe taken to achieve a reasonable population estimate within the error limits set(Box 1.1).One should make such calculations throughout the season. So for example ifyou are sampling cereal aphids at the beginning of a season when numbers arelow you would start with a thousand tillers per field, and make adjustments asthe population rises — but never below 100 tillers per field. There is usually aminimum value that the sampler never falls below and a maximum that

SAMPLING THEORY AND PRACTICE 930Frequency of eggs252015105010 20 30 40 50 60 70 80 90 100Distance from main trunk (cm)Fig. 1.2 The distribution of winter moth eggs along Sitka spruce branches. Data from Wattet al. (1992).Box 1.1 To calculate the number of samples requiredn = (s/m) ¥ cv) 2where n is the number of samples required, s (sigma) is the variance, m (mu) is the unknownmean of the population, and cv is the coefficient of variation of the mean which inturn is defined ascv( X)=s n/mFor practical purposes one needs to collect a series of samples and make preliminary estimatesof the mean (Xe) and the variance (Se) and then use this formulan = (Se/Xe ¥ cv) 2is never exceeded: these are determined by time and the requirement foraccuracy. (For a number of case studies see Chapter 9.)Sampling conceptsChoosing a sample unitWhat is a sampling unit?A sampling unit is a proportion of the habitable space from which insect countsare taken. The units must be distinct and not overlap. A sampling unit can bevery variable in form. For example, it could be direct counts of all the caterpillars

10 CHAPTER 1Box 1.2 Possible types of information required from a sampling scheme1 Estimates of population density per unit area2 Assessments of percentage infestation or parasitism3 Estimation of damage per unit area4 Absolute population countsin 1 m 2 of cereal field, or it could be 20 sweeps of a sweep net down a row. Moreusually, a sample unit is more easily measurable, e.g. a quadrat or template of aknown area.What determines the choice of a sampling unit?The choice of sampling unit is dependent on what information is required. In anagricultural system the grower most frequently wants to know whether a particularinsect pest has reached a threshold level which requires action (e.g.spraying), or what proportion of the crop is infested. In other circumstances(e.g. a population study) the observer may be more interested in the number ofinsects per given unit area (Box 1.2).Criteria for sampling unitsSample units must meet a number of criteria if they are to be useful.1 Each sample unit should have an equal chance of selectionThis is where it is important to know what type of distribution individualswithin the population display. Unfortunately using a totally random samplingscheme in some situations, even agricultural, can be too expensive in termsof time. Certainly in some situations — e.g. in a dense forest — it may not be logisticallypossible to apply a totally random sampling pattern. Therefore mostfields and plots are sampled on a prearranged pattern — e.g. two X’s, a V, a W, orwhatever, with the samples collected along the transects. A degree of randomizationcan then be introduced, for example by varying the distance betweensampling stations or by taking samples from either side of the transect on a randombasis. It is important to avoid bias when sampling. This is particularly easyto introduce when sampling in crops. It is difficult to avoid selecting the leavesthat look infested, e.g. discolored or curling. In cases like that, an element ofchance should be built into the process when arriving at the sample station —e.g. take the first plant on the left, or throw a quadrat to standardize thesampling unit.

SAMPLING THEORY AND PRACTICE 112 The proportion of an insect population using the sample unit as habitat shouldremain constant throughout the sampling eventIf for example the insect moves around depending on time of day, then yourpopulation estimates will vary accordingly — e.g. the beet armyworm Spodopteraexigua is phototactic so sampling at midday will produce lower counts than duringtwilight or dawn as the larvae move into the ground or center of plants athigh light intensities. The large pine weevil Hylobius abietis is another example —it is night active so sampling should be done at the same time each day to keepthings constant. Sampling should therefore be planned to take these factors intoaccount.3 There should be a reasonable balance between the variance produced whendata are collected from a given sample unit and the cost (time, labor, orequipment) in assessing that unitGenerally a preferred sample unit would be the minimum size which wouldallow an adequate number of replications on a given date to produce averageswith meaningful variance. Sampling all the leaves on a plant would providevery accurate information on that plant but as one would only be able to samplea few plants then the population estimate for the site would be extremely poor.Incidence counts are also useful (Ward et al. 1985). These rely on intensive samplingover a number of seasons so that one has a robust relationship betweenthe numbers of insects present and the infestation rate of those plants. This is avery useful technique for non-experts such as farmers. It is however, not a feasibleoption unless the preliminary studies have been completed. Cautionshould also be exercised with this method as the relationship between incidenceand population can change.4 Whenever possible or practical the sample unit should be as near as possible tothe natural habitat unitIn other words the area within which the insect is likely to spend most of its timein a given developmental stage — e.g. a cereal plant for an aphid, a leaf on a treefor an aphid or leaf miner, a branch for a defoliating caterpillar, and so on. Insectswithout discrete habitats — e.g. soil dwellers, predatory beetles, etc. — aresomewhat more problematic and in such cases it is probably wise to rely on randomquadrats etc.5A sample unit should have stabilityOr, if not, then its changes should be easily and continuously measured — e.g.the number of shoots in a cereal crop.

12 CHAPTER 16 The sampling unit must be easily delineated or describedFor example, buds on a branch, leaves, or plants, or quadrats of standard size.7 Ideally a sample unit should be able to be converted to some measure of unit areaThus it is important to count the number of trees in a compartment or plants ina field, etc., and then to be able to convert the counts obtained to numbers perm 2 , for example (Box 1.3). What conversion is used, however, is less importantthat the fact that a conversion of some type is required in order to compare thedensity of different stages of the same insect species. This is essential if the mortalityoccurring between different stages is to be estimated.8 The number and location of sample units should be selected according to thepurpose of the samplingThus one could just sample the ears of cereal plants if one was interested in Sitobionavenae for prediction purposes (George & Gair 1979), whereas whole-plantcounts would be needed for population estimates (Leather et al. 1984).Box 1.3 Sampling the pine beauty mothThe pine beauty moth Panolis flammea has a typical univoltine lifecycle. The adult lays eggson pine needles which hatch into larvae that pass through five instars whilst feeding inthe canopy. The fifth-instar larva stops feeding and passes into a pre-pupal stage that spinsto the ground, burrows into the litter layer, and pupates (Watt & Leather 1988b).Sampling is carried out at all stages of the life cycle. Although each sampling techniquegives a different output, they are all easily converted to a common measure, in this caseindividuals per square meter.Stage Method Output ConversionAdult Pheromone trap Males per trap Calibrated to areacovered by trapEggs Needle counts Eggs per whorl Converted toprojected areacovered by treeLarvae Funnel traps Head capsules per Collecting area offunnelfunnel knownPre-pupae Basin traps Pre-pupae per basin Collecting area ofbasin knownPupae Soil sample Pupae per 15 cm 2 Converted to per m 2measure

SAMPLING THEORY AND PRACTICE 13Informed sampling and collectingAs one works more and more with insects, one gains a knowledge or feeling ofwhere to find particular groups or species. Although this is not strictly sampling,it does help inform the sampling process, and when one requires insects to startcultures or laboratory and field experiments it is certainly useful to be able tolocate relatively large numbers of specimens quickly and easily.In general, insects are small and relatively fragile, their reproductive and developmentrates are highly influenced by environmental factors, in particulartemperature, and many of them, especially in their larval stages, are likely tofeature in the diets of birds and other vertebrates as well as arthropod predators.This tends to mean that insects, except for the brightly colored highly mobilespecies such as butterflies, are more likely to spend most of their day in shelteredor concealed habitats, and in fact many insect species have taken this to the extremeand spend much of their life cycle living and feeding within plant parts —e.g. gall insects, leaf miners, bark beetles. Therefore, if looking for a ready supplyof various insect species, dense clumps of grass, piles of leaves, under rocks andstones, in tree hollows and crevices, under loose bark, under logs, or even infungi, will prove rewarding sites to search. Very dry habitats are unlikely to yieldlarge numbers of individuals or species, but a moist, sheltered hollow under abroad-leaved tree is a sure source of a myriad of different species, albeit not allinsects.Insects, particularly herbivorous ones, are of course closely associated withtheir host plants, and certain times of year and sites on the plant are more likelyto yield results than others. Certain plant species naturally potentially harbormore insect species than others. Oaks, willows, and birches are natural hot spotsfor insects of all descriptions from bark-dwelling Pscoptera to gallers, miners,general defoliators, and sap suckers. Many herbivorous insects depend on aready supply of nitrogen to enable them to develop quickly at the beginning ofthe year. Check meristems, developing buds, young shoots, and flower buds forcaterpillars and sap suckers. Birch aphids Euceraphis punctipennis closely followgrowing shoots. Curled or distorted leaves are often signs that sap suckers or leaftiers are in the vicinity, although be warned that these deformations will persistlong after the insect has completed its life cycle and departed. Similarly, sootymould, sticky leaves, and silken threads are often signs that aphids, other sapsuckers, and web-spinning Lepidoptera are or have been present. Swellings onstems and sap and resin flows may also indicate the presence of stem borers,gallers, and bark beetles.In temperate parts of the world insects spend a large proportion of their lifecycle overwintering (Leather et al. 1993). Many have behavioral adaptationsthat cause them to seek out specific overwintering sites — e.g. negative phototaxisthat causes them to search for dark crevices or thigmotactic responses thatmake them aggregate. If looking for ladybirds during the winter, it is often usefulto look under loose bark, under window sills, or even on fence posts. Aggre-

14 CHAPTER 1gations often form in such situations. If your insect overwinters in the soil, avoidwet places and look for well-drained sites, preferably under trees rather than inthe open. Overwintering is a costly business and insects attempt to minimizecosts by overwintering in sites where the soil is unlikely to freeze, below about10 cm depth. During winter, searching under hedges, in the middle of rottinglogs, and in dense clumps of grass is also likely to repay one’s efforts.In general, think shelter, food, and protection and you are likely to find someinsects in a relatively short space of time.ConclusionsIn this chapter we have tried to give an overview of the philosophy of sampling,the rationale behind the choice of sample unit and technique, and somepointers towards what is the best approach to use in particular situations. Wehave not provided detailed mathematical and statistical formulae or numerousworked examples. Those wishing to acquire more of the mathematicalbackground should consult two excellent textbooks that provide a wealth ofsuch information, Southwood and Henderson (2000) and Sutherland (1996).Chapters within this book provide more specific mathematical and theoreticalapproaches for specific cases, but in the main deal with the practicalities ofsampling either in specific habitats or with problematic guilds or groups.ReferencesDixon, A.F.G. & Glen, D.M. (1971) Morph determination in the bird cherry-oat aphid,Rhopalosiphum padi (L). Annals of Applied Biology, 68, 11–21.Finch, S. (1990) The effectiveness of traps used currently for monitoring populations of thecabbage root fly (Delia radicum). Annals of Applied Biology, 116, 447–454.George, K.S. & Gair, R. (1979) Crop loss assessment on winter wheat attacked by the grainaphid Sitobion avenae (F.). Plant Pathology, 28, 143–149.Heathcote, G.D. (1957) The comparison of yellow cylindrical, flat and water traps, and of Johnsonsuction traps for sampling aphids. Annals of Applied Biology, 45, 133–139.Knight, J.D., Tatchell, G.M., Norton, G.A., & Harrington, R. (1992) FLYPAST: an informationmanagement system for the Rothamsted aphid database to aid pest control research andadvice. Crop Protection, 11, 419–426.Leather, S.R. (1993) Influence of site factor modification on the population development ofthe pine beauty moth (Panolis flammea) in a Scottish lodgepole pine (Pinus contorta) plantation.Forest Ecology & Management, 59, 207–223.Leather, S.R., Bale, J.S., & Walters, K.F.A. (1993) The Ecology of Insect Overwintering. CambridgeUniversity Press, Cambridge.Leather, S.R., Carter, N., Walters , K.F.A., et al. (1984) Epidemiology of cereal aphids on winterwheat in Norfolk, 1979–1981. Journal of Applied Ecology, 21, 103–114.Niemelä, J., Halme, E., Pajunen, T., & Haila, Y. (1986) Sampling spiders and carabid beetleswith pitfall traps: the effect of increased sampling effort. Annales Entomologici Fennici, 52,109–111.

SAMPLING THEORY AND PRACTICE 15Southwood, T.R.E. & Henderson, P.A. (2000) Ecological Methods. 3rd edn. Blackwell Science,Oxford.Sutherland, W.J. (1996) Ecological Census Techniques. Cambridge University Press, Cambridge.Ward, S.A., Rabbinge, R., & Mantel, W.P. (1985) The use of incidence counts for estimation ofaphid populations. 1. Minimum sample size for required accuracy. Netherlands Journal ofPlant Pathology, 91, 93–99.Watt, A.D. & Leather, S.R. (1988a) The distribution of eggs laid by the pine beauty mothPanolis flammea (Denis & Schiff.) (Lep., Noctuidae) on lodgepole pine. Journal of AppliedEntomology, 106, 108–110.Watt, A.D. & Leather, S.R. (1988b). The pine beauty in Scottish lodgepole pine plantations.In Dynamics of Forest Insect Populations: Patterns, Causes, Implications (ed. A.A. Berryman),pp. 243–266. Plenum Press, New York.Watt, A.D., Evans, R., & Varley, T. (1992) The egg-laying behaviour of a native insect, thewinter moth Operophtera brumata (L.) (Lep., Geometridae), on an introduced tree species,Sitka spruce, Picea sitchensis. Journal of Applied Entomology, 114, 1–4.

CHAPTER 2Sampling insects from rootsALAN C. GANGEIntroductionThere are relatively few ecologists who dare to venture below ground, to studythe effects of subterranean insects on plants. If one examines the insect–plantinteraction literature for the last 20 years, fewer than 2 percent of studies dealwith root-feeding insects. From this paucity of information, one is tempted toconclude that subterranean insects are of little consequence in natural systems.However, a quick glance at the agricultural and horticultural literature showsthat there is a rich array of studies involving these insects, since many of themare pests of considerable economic importance. Indeed, root-feeding insectscan be so destructive that several species have been introduced in biologicalcontrol programs against weeds (e.g. Blossey 1993, Cordo et al. 1995, Sheppardet al. 1995).Why is there this apparent lack of interest in ecological studies involving subterraneaninsects? The answer undoubtedly lies in the difficulty of samplingthese animals. Unlike their foliar counterparts, rhizophagous insects are ofteninvisible for part or all of their life cycles. Furthermore, excavation of soil maynot always be sufficient to detect them, since some species feed internally in theroot system. Experiments involving these insects often end in failure, as nondestructivemonitoring of the system is difficult and problems may go undetected.To add to these physical problems, various aspects of the biology of thespecies may also hinder sampling methods. In some cases, the stage in the soil islong-lived and the time span involved may be greater than that allotted to standardresearch projects, which are generally of three years’ duration. The end resultof these problems is that sampling for rhizophagous insects is generally alaborious, time-consuming, and often tedious operation. However, it need notalways be so and a number of ingenious methods have been developed.The most recent comprehensive review of rhizophagous insects and their effectson plants is that of Brown and Gange (1990). This documents that only sixof the 26 orders of insects are well represented as below-ground herbivores, andof these the most important order is the Coleoptera. Diptera and Lepidopteraalso contain species with rhizophagous larvae, while within the Hemiptera theAphididae (aphids), Cercopidae (spittle bugs), Cicadidae (cicadas), and Pseudococcidae(mealy bugs) contain economically important root-feeding species.16

SAMPLING INSECTS FROM ROOTS 17The Collembola also have representatives which feed on roots, though the majorityprobably feed on microorganisms or decaying leaf litter (Hopkin 1997).Collembola apart, the majority of insects associated with roots have a stage oftheir life cycle above ground and these mobile adults can easily be used to identifythe presence of subterranean stages in a particular area. A good way to startwith rhizophagous insect sampling is to understand the visible signs of theirpresence, manifest in the terrestrial environment.External cluesTo determine if a species is present in a location, a variety of trapping methodsfor adults can be used. Suction sampling (e.g. Arnold 1994) can be particularlyeffective, but a number of species have nocturnal adults. Many of these seem tobe attracted to light, and mercury vapor (MV) light traps have been used tomonitor adult numbers of chafer grubs (Coleoptera: Scarabaeidae) nearpastures (Roberts et al. 1982b). Interestingly, adults of the wingless black vineweevil Otiorhynchus sulcatus are also attracted to light, but generally to tungstenbulbs, rather than MV (Labuschagne 1999). Water traps have been used tocapture adults of the cabbage root fly Delia radicum (Bracken 1988), whilepheromone traps have been developed for some species (e.g. the pea and beanweevil Sitona lineatus [Smart et al. 1994]). If the biology of the species is wellknown, then emergence traps (described in Southwood & Henderson 2000)can be very effective (e.g. for S. discoideus [Goldson et al. 1988]). Some species oneclosion leave characteristic evidence, and the empty emergence skins of variouscicada species have been used to estimate nymphal densities below ground(White & Sedcole 1993). Sticky traps, with the sticky side facing downwards,have been used to estimate numbers of grape phylloxera Daktulosphairavitifoliae emerging from grape rootstocks (Hawthorne & Dennehy 1991).Adults of many species feed on foliage in a characteristic manner. A good exampleof this is the leaf-notching produced by O. sulcatus and this can be used asan excellent method of detecting the pest (Labuschagne 1999). However, theeffects of subterranean larval feeding are also often apparent, most commonlymanifest in wilting of foliar tissues, because the main effect of root removal bylarvae is the imposition of drought stress in a plant (Masters 1995). In naturalplant populations, individuals which show unusual drought stress or which diefor reasons not attributable to foliar insects or pathogens (e.g. Strong et al. 1995)should be suspected of having insects attacking the roots. In some cases, internalroot borers produce quantities of frass at the exterior end of their tunnelsand this can be visible at or just below the soil surface. Maron (1998) gives an examplewith ghost moth Hepialus californicus, where frass can be easily seen at thebase of infested bush lupine plants.Subterranean aphids often live in close proximity to ant colonies and a numberof species live entirely within the nest of the ants. In grassland systems, one

18 CHAPTER 2must first find the ant mounds and then sample within these to find the aphids(Pontin 1978). Although the aphids are “cultured” by the ants, a significantnumber are eaten too, and a further method of deciding whether subterraneaninsects are present in any given location is to look for the signs of predation. Forexample, in pasture grassland and amenity turf, birds such as rooks, crows, andmagpies can do significant damage, when searching for large subterraneanlarvae of chafer grubs (Coleoptera: Scarabaeidae) or “leatherjackets” (Diptera:Tipulidae). Indeed, for turf managers, birds represent the best early warningsystem that subterranean larvae are present and may need to be controlled(Fermanian et al. 1997).Field extraction methodsChemical methodsExtraction of insects from soil without disturbance of the soil profile must involvesome form of chemical expulsion or the use of an attractant. Variouschemicals have been used over the years to expel insects from soil, with varyingdegrees of success. These include St Ives fluid (a mixture of disinfectant andother chemicals), potassium permanganate, mustard, formalin, petrol (gasoline),diesel fuel, ammonia, nitric acid, acetic acid, soapy water, and brine. In theearly years, the chemical was poured onto the surface of soil and the appearanceof larvae awaited. There are of course many problems with this approach, notleast toxicity of the chemicals to the operator and to any plant life present.Furthermore, the method is not quantifiable, as the area from which larvaehave appeared is unknown.Of these chemicals, only brine has any merit and is worth consideration.Stewart and Kozicki (1987) developed a successful sampling method for tipulidlarvae in grassland, termed the “brine pipe method.” This involved hammering10 cm diameter plastic pipes into soil to a depth of about 5 cm, and filling thepipes with strong brine solution. The brine slowly percolates into the soil, andon contact with the larvae causes these to rise to the surface, where they float inthe pipe. The method can produce comparable results with more conventionallaboratory-based techniques (below) and can be quantified, by treating the pipeas a soil “core.” Figure 2.1 shows the efficacy of the method. Here, 16 differentfields, all under permanent ryegrass Lolium perenne / clover Trifolium repens pasturewere sampled in the spring of 1999 (Gange, unpublished). Twenty 10 cmdiameter brine pipes were placed randomly in each field. Within 30 cm of eachpipe, a 10 cm diameter ¥ 10 cm deep soil core was taken and tipulid larvae wereextracted from each in the laboratory by wet-sieving (see below). It can be seenthat the brine pipe method provides a good estimator of total abundance whenlarval numbers are high, but tends to underestimate abundance when totalnumbers are low. The most likely reason for this is that the pipe method relies on

SAMPLING INSECTS FROM ROOTS 19100Total larvae per m 2755025y = x00 20 40 60 80Larvae per m 2 from brine pipe methodFig. 2.1 Relation between tipulid larvae extracted by the brine pipe method and byexhaustive hand-sampling (total numbers). Dashed line is the fitted regression (y = 0.817x +21.89), solid line is the line of equality (y = x). At low densities, brine pipes underestimatetotal numbers, but the accuracy of the method improves with increasing density. Theregression predicts that brine pipes will record the total population when density is about 120larvae per m 2 . Data from Gange (unpublished).the percolation of brine into the soil and if larvae are at low density, it is likelythat not all will be affected by the solution. However, if larval numbers are high,then a higher proportion of larvae are likely to come into contact with the solution.Furthermore, the number of pipes required and time to check them meansthat this is a less efficient method from the labor point of view, if larvae are rareor patchy. Interestingly, no other subterranean insects seem to appear in thepipes, but earthworms can also be sampled by this method.Behavioral methodsPerhaps because of their economic importance, tipulids (Diptera: Tipulidae)seem to have been the subject of more published sampling methods than anyother root-feeding insect. The brine pipe method outlined above is particularlyuseful because it enables farmers or turf managers to sample for the insects insitu, and, as it is quantifiable, indices of infestation have been produced againstwhich field counts can be compared. Farmers can then decide whether it iseconomically viable to spray a field to control their numbers (Clements 1984).However, if a source of salt, or water, or pipes is not available, it is still possible todetermine if tipulid larvae are present in a field, by taking advantage of theirnocturnal behavior. An area of grassland is thoroughly soaked with water and atarpaulin or similar item (polyethylene bin liners are an acceptable substitute) islaid over the soil surface (Gratwick 1992). Inspection beneath the tarpaulin inthe early morning should reveal larvae, which have emerged at night to feed onthe surface, but which do not return to the soil because it remains dark under

20 CHAPTER 2the cover. These must be collected quickly, because exposure to light will causethem to burrow rapidly into the soil. If the researcher merely wishes to obtainlarvae for experiments or to start a culture, this is a very easy method for theircollection.BaitsInstead of trying to persuade insects to leave the soil, an alternative method is toprovide them with an attractant in the form of a bait. Perhaps surprisingly, thisis not a widely adopted method, most likely because it produces only semiquantitativeinformation, as the area from which larvae have been attracted isdifficult to measure. However, baits have been developed for wireworms(Coleoptera: Elateridae) (Ward & Keaster 1977) and a bait consisting of a 1 : 1mixture of wheat and corn was used by Belcher (1989) to estimate the proportionof corn fields infested with wireworms in Missouri. An example of the kindof data one can obtain by this method is given in Fig. 2.2. While the method maybe of little use for quantifying insect density on a local scale (e.g. per m 2 ), itis useful for recording density on a regional scale (e.g. proportion of fieldsinfested, etc.) Belcher (1989) mentions that white grubs (Coleoptera: Scarabaeidae)(otherwise known as chafer grubs) were also attracted to the bait.However, this fact does not appear to have been used in any subsequent sam-Aeolus mellillusMelanotuscommunisM. depressusM. similisM. verberansAll species0 5 10 15 20 25 30Percentage of fieldsFig. 2.2 An example of the data that can be obtained by baiting for subterranean larvae.Belcher (1989) sampled cornfields in Missouri and was able to record the percentage of fieldsinfested by different species of wireworm (Coleoptera: Elateridae). Drawn from data inBelcher (1989).

SAMPLING INSECTS FROM ROOTS 21pling program for these often injurious insects. Baits have been used to controlone insect pest, the black field cricket Teleogryllus commodus, which can cause seriousdamage to pastures in Australia. Williams et al. (1982) describe the successof cereal baits impregnated with insecticide in the fight against this insect.Recipes for baits for attracting adults of O. sulcatus are given by Labuschagne(1999) and can be most successful when scouting for the presence of this pest.Most baits for larvae and adults (Labuschagne 1999) appear to be based on acereal/bran mixture, but probably the main criterion for a successful bait is theevolution of CO 2. This is because it is thought that CO 2is the primary stimulusused by insects to orientate themselves to roots in the soil (Brown & Gange1990, Bernklau & Bjostad 1998). This probably explains why another excellentbait for wireworm larvae in ex-pasture is a buried potato. Apart from being anacceptable food source, the potato gives off CO 2and the larvae aggregate towardsit. While not being of much use from a quantitative point of view, thismethod can be used to determine if the insects are present.Hand-sortingThe most laborious, but also probably the most accurate method of extraction inthe field is hand-sorting of extracted soil cores. An excellent example of this isprovided by Penev (1992) who hand-sorted soil cores measuring 25 ¥ 25 cm and30–40 cm deep in the field when sampling for wireworms. For large insectswhich are abundant, this method is often quite rewarding. Gange et al. (1991)hand-sorted turf when sampling for larvae of Phyllopertha horticola (gardenchafer) infesting a golf tee. They used 25 ¥ 25 cm ¥ 10 cm deep quadrats andfound that the number of larvae varied between 1 and 49 per quadrat, equivalentto a range of 16–784 per m 2 . The distribution of larvae was highly aggregated,conforming to a negative binomial distribution (Fig. 2.3).Highly aggregated distributions are observed commonly with subterraneaninsects and result from clumped ovipositional patterns, feeding preferences,and the heterogeneous nature of the soil environment (Brown & Gange 1990).This means that in any situation a large number of quadrats may contain zero orvery few insects, and the overall process of accurately measuring the populationand its spatial distribution may be an extremely time-consuming business. Thetime taken largely depends on the ease of visibility of larvae and their size.For example, Harcourt and Binns (1989) hand-sorted soil cores measuring3600 cm 3 when searching for larvae of the alfalfa snout beetle Otiorhynchus ligusticiand it took them nine minutes for each core. The distribution of larvae wasalso highly aggregated, again conforming to a negative binomial distribution(Fig. 2.4). Meanwhile, Seastedt (1984) sorted soil cores from prairie grasslandmeasuring 2000 cm 3 and it took 40 minutes per core. The best option is to organizea team of people to perform the sampling together. Thus, in the study ofGange et al. (1991), seven people managed to sort 100 cores, each measuring6250 cm 3 , in five hours (equivalent to 21 person minutes per core).

22 CHAPTER 2(a)8Number of quadrats6420(b)8Number of quadrats64201 3 5 7 9 11 13 15 1719 212325272931333537394143454749Number of larvae per 25 ¥ 25 cm quadratFig. 2.3 The spatial distribution of chafer grub larvae (Coleoptera: Scarabaeidae) in a golf tee,as revealed by hand-sorting of quadrats. (a) recorded distribution; (b) fitted negativebinomial. Redrawn from Gange et al. (1991).The number and size of cores are generally determined by the identity of thespecies being sampled. The depth of cores needs to be such that virtually all ofthe root system is sampled, but must also take into account the biology of the insect.Unless published information on a species is available, it is best to performa preliminary experiment to develop a sampling strategy (e.g. De Barro 1991)which minimizes variance, but with a replicate number which is feasible in thetime available. Good examples of the use of binomial sequential samplingmethods are provided by Allsopp (1991) for a sucking insect and Badenhausserand Lerin (1999) for a chewer. In any sampling program, it must be rememberedthat insect vertical distribution in soil can vary in time and space within aseason (e.g. Hanula 1993), and over the course of several seasons (Brown &Gange 1990).To speed up the extraction process, sieving of soil may be used, but this ofcourse depends upon the soil texture. Sieving has been used successfully torecord insects as disparate in size as white grubs Phyllophaga spp (Coleoptera:Scarabaeidae) in pine plantations (Fowler & Wilson 1971) and sugar beet root

SAMPLING INSECTS FROM ROOTS 23108Frequency64200 1 2 3 4 5 6 7 8 9 10 11 12Number of larvae per 16 ¥ 16 cm quadratFig. 2.4 The spatial distribution of larvae of the alfalfa snout beetle Otiorhynchus ligustici infield soil. Solid bars represent the recorded larval abundance, open bars represent a fittednegative binomial distribution. Drawn from data in Harcourt and Binns (1989).maggot Tetanops myopaeformis (Diptera: Otitidae) in cultivated fields (Whitfield& Grace 1985).For large-scale sampling of insects across whole fields, plough transects havebeen used. This is simply where a tractor plough cuts a furrow and the insect larvaeexposed are counted and expressed as numbers per unit length of furrow.The technique has generally been used in grassland where the destructive natureof the method is not too much of a problem (Roberts et al. 1982a, East &Willoughby 1983).Laboratory extraction methodsDissection of rootsIn the majority of studies with rhizophagous insects, sampling involvesremoving soil cores from the field and extracting these in the laboratory.In this situation it is possible to make detailed examinations and dissectionsof roots to determine larval numbers. In cases where the insect lives internallyin the root, this may be the only way in which accurate records of numberscan be obtained. Dissection has been used to record insect attack in a rangeof plant species, including grape (Dutcher & All 1979), sunflower (Rogers1985), purple viper’s bugloss (Forrester 1993), and bush lupine (Strong et al.1995, Maron 1998). In one case, careful dissection has enabled the entire

24 CHAPTER 2entomofauna associated with the roots of Centaurea species to be determined(Müller et al. 1989).Flotation methodsAs with field studies, hand-sorting has been commonly used. However, it is noticeablein a number of long-term studies that this process has then given way toother, more automated forms of extraction. For example, Goldson and Proffitt(1988) and Goldson et al. (1988) used hand-sorting in the early stages of theirwork, but subsequently changed to using flotation methods for the extractionof S. discoideus larvae from lucerne field samples.A variety of flotation methods have been described (Southwood & Henderson2000); these generally involve a thorough mixing of the soil sample withwater, sugar, or salt solution and collecting the insects from the surface. Salt ismost commonly used — e.g. for tipulids (Lauenstein 1986) and Sitona spp(Coleoptera: Curculionidae) (Goldson et al. 1988). The advantages of this socalledpassive extraction technique are that it is inexpensive and relativelyquick. De Barro (1991) compared hand-sorting of sugarcane roots with flotationin water for the wax-covered mealybug Saccharicoccus sacchari. Flotationtook half the time of direct counting and produced identical counts of insects.Furthermore, flotation is particularly useful for the extraction of inactive stagessuch as pupae and eggs. Indeed, this is the standard method for obtaining eggcounts of a range of subterranean Coleoptera (Elvin & Yeargan 1985, Blanket al. 1986) and Diptera (Dosdall et al. 1994).However, if one is trying to obtain an accurate estimate of the spatial distributionof eggs in soil, then flotation is not ideal, particularly for friable soils, inwhich cores easily break up. An ingenious egg sampling method for onion flyDelia antiqua eggs was therefore developed by Havukkala et al. (1992). In this,Petri dishes 15 cm diameter ¥ 2 cm deep with wire gauze bottoms were filledwith soil and then exposed to ovipositing flies in various situations. After oviposition,the dishes were filled with molten agar, from below. After cooling, the resultingsolid was cut into sections and mixed with hot water, and the position ofeggs accurately determined following extraction with flotation. In this way, itwas possible to show how eggs of this insect were distributed within the soil profile(Fig. 2.5). Most eggs were deposited within the top 8 mm of soil, a fact whichcan be used to improve the targeting of insecticides against this pest (Havukkalaet al. 1992).The other disadvantages of flotation are that it is often difficult to get “clean”samples of insects and that dead animals in the soil will also be extracted. It maytherefore be misleading in terms of producing estimates of active populationsizes for some species (McSorley & Walter 1991).To overcome the problem of obtaining clean samples, chemicals such as magnesiumsulfate may be added to the water to ease separation of insects from thesoil material. However, one extraction method that is unique for arthropods is

SAMPLING INSECTS FROM ROOTS 253025Percentage of eggs201510502 4 6 8 10 12 14 16 18 20Depth in soil (mm)Fig. 2.5 The vertical distribution of onion fly Delia antiqua eggs in soil, as revealed by themolten agar technique. Redrawn from Havukkala et al. (1992).hydrocarbon adhesion. As the cuticle of most species is lipophilic, it adheres topetroleum derivatives and makes for a very efficient extraction process. The soilis mixed with a solution of water and a hydrocarbon (usually heptane) and allowedto settle. The insects will be found in the heptane layer. The procedurewas first described by Walter et al. (1987) and has since been improved by Geurset al. (1991) and Kethley (1991).A variation of the “wet” method involves the sieving of insects from thesoil/water solution. With the aid of a continuous stream of water, this methodcan be an improvement on the simple act of flotation and has been usedsuccessfully when the root-feeding community is sought (e.g. Clements et al.1987). Sieving can also be combined with subsequent flotation in magnesiumsulfate (Murray & Clements 1995) to separate small larvae from the debris remainingon the sieve. Another refinement to the flotation method is elutriation,in which air is bubbled through the soil/water mixture in an effort toimprove separation of the insects from vegetative and soil material (e.g. House& Alzugaray 1989).Behavioral methodsIn contrast to passive extraction methods, a variety of active techniques are alsoavailable, which rely on behavioral mechanisms of the insects. As all subterraneaninsects shun light and avoid high temperatures, these methods relyon the production of a temperature gradient to drive them out of soil samples.

26 CHAPTER 2Possibly the most common method is use of the Berlese–Tullgren funnel, thehistory and development of which is described by Southwood andHenderson (2000). Briefly, this technique involves the use of heat and light todrive insects out of a soil sample into a collecting container. The collectingreceptacle is usually filled with a 70% alcohol solution, to preserve specimens.This method can be used to extract the active stages of any subterranean insect,but it is especially useful for microarthropods, such as Collembola, whichcannot easily be obtained by any of the preceding methods.Various modifications to the basic design have been made, usually forparticular root-feeding insects. For example, the soil may be contained withina canister, which allows for regulation of temperature gradients throughthe core (Lussenhop 1971). This method can easily be adapted to collect liveinsects and in this way it is particularly useful for obtaining microarthropod“communities.” Indeed, Klironomos and Kendrick (1995) used the canistermethod to extract Collembola and mites from leaf litter for use in subsequentexperiments.One of the most widely used and efficient variations of the funnel is the Blasdaleversion, for tipulid larval extraction (Blasdale 1974). In this, turf cores areheld in metal cylinders and positioned turf surface downwards in a dish of coldwater. Heat is applied to the soil end of the core and this drives the larvaethrough the core and into the water. It is likely that this method would be oflittle use for most rhizophagous insects, as many species will only movedownwards through a soil profile. Tipulids are an exception, as they often leavethe soil at night to feed on the surface (Gratwick 1992).The use of active extraction methods for root-feeding insect density estimatesis widespread and in general they are inexpensive and produce clean samples.However, their efficiency is often questioned (e.g. McSorley & Walter 1991),as the number of insects extracted can be affected by soil moisture content,whether the soil core is inverted or in its original position, and whether it is intactor broken up. Hammer (1944) found that to extract maximum numbers ofCollembola it was necessary to maintain the core intact and to invert it. Inversionappears to allow animals to leave the soil by natural passages, such asearthworm burrows, which open to the surface. Another problem is that condensationcan form on the inside of the soil container and small animals can becometrapped in this and so not be counted (Haarløv 1947). Furthermore, aparticular problem, especially with Tullgren-type extractors with high temperaturegradients (e.g. Crossley & Blair 1991) is that the temperature generatedinside the core may be detrimental to the insect being sampled. It is a fact that bigfunnels extract relatively more large invertebrates, which Ausden (1996) attributedto the desiccation of microarthropods in large funnels. It is best to runthe extractor with a low temperature gradient and to prevent the soil from dryingout. A simple alteration to the standard Tullgren funnel is to use a very lowwattage light bulb (e.g. 10 W), and to place polyethylene film over the samplecontainer. Indeed, a very simple demonstration of the importance of these

SAMPLING INSECTS FROM ROOTS 27160Collembola per 10 cm core (± s.e.)14012010080604020040 W bulb,uncovered40 W bulb,covered10 W bulb,uncovered10 W bulb,coveredFig. 2.6 Collembola abundance in a rye grass Lolium perenne pasture soil, as measured byTullgren extraction. Each sample was left for one week. Use of too powerful a bulb (40 W)significantly reduces the numbers of animals recorded, while the use of a polyethyleneprotective covering over the sample increases the numbers obtained. Data from Gange(unpublished).modifications can be seen in Fig. 2.6. Here, the use of too powerful a bulb and noprotective covering dramatically reduced the number of Collembola obtainedfrom soil samples.With any behavioral extraction method, there must be a trade-off betweenextraction time and the accuracy of the method. Thus, the use of a high temperaturegradient will speed up the extraction process, but may underestimatenumbers if many microarthropods die without leaving the soil. Use of a lowertemperature gradient means a longer extraction period, but higher estimates ofabundance. However, a further problem with these systems is that in the time ittakes for the insects to be persuaded to leave the soil, considerable reproductioncan have taken place. The slower the process, the worse this situation is likely tobecome. This problem was noted by Pontin (1978), who suggested that subterraneanaphids could produce a large number of offspring while still in the samplingcontainers, leading to overestimates of population size.An excellent comparison of a behavioral and a passive extraction method forroot aphids is provided by Salt et al. (1996). Here, Tullgren funnels were comparedwith flotation to extract the subterranean aphids Pachypappa spp andPachypappella spp from Sitka spruce Picea sitchensis plantation soil (Fig. 2.7). InTullgren funnels, the majority of the aphids extracted were first-instar nymphs,while in water flotation the majority of the aphids were adults and late-instarnymphs. Tullgren estimates of total abundance were significantly higher than

28 CHAPTER 21400Aphids per m 2 (± s.e.)120010008006004002000First-instarnymphs2nd–4th-instarnymphsAdult aphidsTotal aphidsTullgren extractionFlotation extractionFig. 2.7 A comparison of Tullgren extraction with flotation, for measuring the abundanceof spruce root aphids Pachypappa spp and Pachypappella spp. Tullgren extraction produceshigher overall estimates, but these are almost entirely first-instar nymphs. Flotation revealsvery few small, first-instar aphids, but records more adults. Drawn from data in Salt et al.(1996).flotation estimates (Fig. 2.7), caused by the large numbers of first instars obtainedby this method. Salt et al. (1996) suggest that reproduction had occurredin the funnels, leading to the high proportion of first instars, but also acknowledgethat flotation underestimated first-instar numbers, because it is virtuallyimpossible to separate such small animals from the organic debris floating onthe surface. This paper emphasizes why neither active nor passive extractionmethods are ideal for small root-feeding insects. In general, the method usedshould be commensurate with the biology and size of the insect being sampled.Thus, for larger larvae, which are not close to pupation, Tullgren funnels or theirequivalent are very efficient. However, for smaller larvae or insects, inactivestages, or actively reproducing adult insects, flotation is a better choice, with theproviso that great care must be taken to ensure that all individuals, no matterhow small, are found. For the latter scenario, wet-sieving is likely to representthe best way of ensuring that (for example) first-instar aphid nymphs are sampledefficiently.Laboratory visualization methodsWhile not strictly sampling methods, a number of techniques have been usedfor the examination of insect distribution and behavior in soils. These methodshave not been widely used, but offer a lot of promise for the understanding ofinsect responses to soil parameters such as moisture content and temperature.Improved knowledge of rhizophagous insect response to biotic and abiotic fac-

SAMPLING INSECTS FROM ROOTS 29tors in the soil will enable improved targeting and efficacy of pesticides againstinjurious species and a clearer understanding of the interactions between theseinsects and their host plants in natural situations.To understand which insects were feeding on clover roots, Baylis et al. (1986)labeled roots with 32 P and then used autoradiography to see which members ofthe soil fauna were radioactive. This method would be useful if one were simplytrying to determine the structure of the rhizophagous community associatedwith one plant species, but it is of little use for the determination of host plantpreferences in a given community. More recently, Briones et al. (1999) haveused carbon stable isotope analysis to determine the feeding of two collembolanspecies associated with leaf litter. This method assumes that the isotopic compositionof the body tissue of microarthropods gives an accurate estimate of thed 13 C value of their diet. With this approach, Briones et al. (1999) were able toshow feeding preferences for organic matter derived from maize, a C 4plant (ormicroorganisms growing on it), compared with matter derived from C 3plants.This technique represents an important advance in soil biological research andshould be applicable to rhizophagous insects, as has already been achieved withearthworms (e.g. Schmidt et al. 1997).A technique for insect behavioral observation was presented by Lussenhop etal. (1991), who advocated the use of video technology for the observation in situof subterranean insects. The method does allow for a considerable amount of visualobservation time to be achieved. However, the manner in which the experimentalunits (biotrons) are set up may be open to criticism in that theygenerally involve some form of glass observation plate, against which insectsand roots may show unnatural behavior. Nevertheless, for the observation ofsmall organisms and the detection of their feeding behavior, this method doesoffer a number of opportunities. Direct observation of southern corn rootwormlarvae Diabrotica undecimpunctata was successfully used by Brust (1991) to monitorpredation of larvae in the soil and enabled a species of Lasius (Hymenoptera:Formicidae) to be identified as the main predator.To overcome the problem of soil disturbance or insertion of observationchambers into soil, Villani and Gould (1986) and Villani and Wright (1988) developedthe use of radiography for direct observation. Intact blocks of soil weresubjected to X-ray analysis and, as the pictures in Villani and Wright (1988)demonstrate, individual scarab larvae could be clearly seen. The larvae used inthese experiments were all large; smaller individuals or species may be impossibleto detect by this method, so a recourse to hand-sorting must be made (Villani& Nyrop 1991). Nevertheless, Harrison et al. (1993) applied X-ray computed tomographyto the study of the smaller pecan weevil Curculio caryae and were ableto record the burrowing activity of this insect. The X-ray technique is very usefulfor documenting the responses of larvae to changes in abiotic parameters,such as soil moisture, and is considerably less time-consuming than handsortingthe soil to determine larval positions. Such observation methods areparticularly important for documenting the behavior of larvae within a soil

30 CHAPTER 2profile. Results such as those of Villani and Nyrop (1991) clearly show differencesin the behavioral patterns of two species of chafer grub (Coleoptera:Scarabaeidae) and could be used to target insecticides more efficiently in timeand space in the field.ConclusionsThe difficulty of sampling subterranean insects has undoubtedly led to a lack ofstudy by ecologists. However, a number of methods are available for their study,and a summary of the decisions needed to be taken is given in Fig. 2.8. Beforestarting any sampling program, it is wise to understand as much about the biologyof the species involved as possible. Many species have adult stages which arefree-flying above ground.Developing a sampling program for these is a good start, and considerablyeasier than searching for the larval forms in the soil. In situations such as grassland,predators such as birds and mammals provide an excellent indicator of larvalpresence in the soil. Other visible signs in the host plant are drought stress(though not necessarily in hot dry conditions) and poor plant growth not attributableto foliar feeders or pathogenic fungi.There are few in situ extraction methods in the field. Brine pipes work well fortipulid larvae and baits are an under-used method of determining larval presence.All other sampling methods are destructive. Hand-sorting of soil, whetherin field or laboratory, is the most accurate method, but is time-consuming andtedious. For internal root-feeders, there is no other way than excavating theroot system and dissecting it. Passive extraction methods, generally involvingsome form of flotation, are useful for inactive stages, very small insects, andactively reproducing adults. Great care must be taken to separate things such asCollembola or first-instar aphids from soil debris; the use of wet-sieving mayhelp in the capture of these individuals. Hydrocarbon adhesion is excellent,though surprisingly under-used.Active extraction methods rely on heat and light to drive insects out of thesoil. They are good for large, active insects but do not sample inactive stages.They have been widely used for the extraction of small insects, but there areseveral problems with this approach. Adult insects such as aphids can produceconsiderable numbers of offspring within the apparatus, leading to erroneousestimates of population size and structure. Too high a temperature gradient inthe soil core can kill small insects such as Collembola, leading to underestimatesof abundance.Several methods of subterranean insect observation have been developed,the most promising of which is radiographic imaging. This is very good forlarger insects, but needs to be refined to detect small individuals. The use ofcarbon stable isotopes offers great promise for the future.

SAMPLING INSECTS FROM ROOTS 31FieldTraps, scouting, baitsAdultvisibleLab.QuickSlowTullgrenFlotation, wet-sievingCollembolaFieldQuickSlowNoneNoneLab.QuickSlowTullgrenFlotation, wet-sieving,hand-sortingTipulidIdentifythe insectChewerFieldQuickSlowChemical expulsionHand-sorting, sievingLab.QuickSlowTullgrenFlotation, wet-sieving,hand-sorting,root dissectionOtherQuickNoneFieldSlowBaits, hand-sorting,sievingQuickTullgrenSuckerLab.SlowFlotation, wet-sieving,hand-sortingQuickNoneFieldSlowHand-sortingFig. 2.8 Schematic diagram showing the decisions that need to be taken to decide on aparticular sampling method appropriate for any subterranean insect. First, identify the insect;having done so, the nature of its life history (sucker, chewer, etc.) needs to determined. Then,one must ask if the extraction procedure will take place in the laboratory (lab.) or field. Thefinal step is to decide whether answers are required with relative ease (quick), perhaps at theexpense of complete accuracy, or whether some time can be devoted to the procedure, toensure that is as accurate as possible (slow). Each extraction procedure, with respectiveadvantages and disadvantages, is described in the text. For some groups in some situations(e.g. Collembola in the field) there is no realistic method available.

32 CHAPTER 2ReferencesAllsopp, P.G. (1991) Binomial sequential sampling of adult Saccharicoccus sacchari on sugarcane.Entomologia Experimentalis et Applicata, 60, 213–218.Arnold, A.J. (1994) Insect sampling without nets, bags or filters. Crop Protection, 13, 73–76.Ausden, M. (1996) Invertebrates. In Ecological Census Techniques: a Handbook (ed. W.J.Sutherland), pp. 139–177. Cambridge University Press, Cambridge.Badenhausser, I. & Lerin, J. (1999) Binomial and numerical sampling for estimating density ofBaris coerulescens (Coleoptera: Curculionidae) on oilseed rape. Journal of Economic Entomology,92, 875–885.Baylis, J.P., Cherrett, J.M., & Ford, J.B. (1986) A survey of the invertebrates feeding on livingclover roots (Trifolium repens L) using 32 P as a radiotracer. Pedobiologia, 29, 201–208.Belcher, D.W. (1989) Influence of cropping systems on the number of wireworms (Coleoptera:Elateridae) collected in baits in Missouri cornfields. Journal of the Kansas Entomological Society,62, 590–592.Bernklau, E.J. & Bjostad, L.B. (1998) Reinvestigation of host location by western corn rootwormlarvae (Coleoptera: Chrysomelidae): CO 2is the only volatile attractant. Journal of EconomicEntomology, 91, 1331–1340.Blank, R.H., Bell, D.S., & Olson, M.H. (1986) Differentiating between black field cricket andblack beetle damage in Northland pastures under drought conditions. New Zealand Journal ofExperimental Agriculture, 14, 361–367.Blasdale, P. (1974) A method of turf sampling and extraction of leatherjackets. Plant Pathology,23, 14–16.Blossey, B. (1993) Herbivory below ground and biological weed control: life history of a rootboringweevil on purple loosestrife. Oecologia, 94, 380–387.Bracken, G.K. (1988) Seasonal occurrence and infestation potential of cabbage maggot, Deliaradicum (L.) (Diptera: Anthomyiidae), attacking Rutabaga in Manitoba as determined bycaptures of females in water traps. Canadian Entomologist, 120, 609–614.Briones, M.J.I., Ineson, P., & Sleep, D. (1999) Use of d 13 C to determine food selection in collembolanspecies. Soil Biology and Biochemistry, 31, 937–940.Brown, V.K. & Gange, A.C. (1990) Insect herbivory below ground. Advances in EcologicalResearch, 20, 1–58.Brust, G.E. (1991) A method for observing belowground pest–predator interactions in cornagroecosystems. Journal of Entomological Science, 26, 1–8.Clements, R.O. (1984) Control of insect pests in grassland. Span, 27, 77–80.Clements, R.O., Bentley, B.R., & Nuttall, R.M. (1987) The invertebrate population and responseto pesticide treatment of two permanent and two temporary pastures. Annals of AppliedBiology, 111, 399–407.Cordo, H.A., DeLoach, C.J., & Ferrer, R. (1995) Host range of the Argentine root borer Carmentahaematica (Ureta) (Lepidoptera: Sesiidae), a potential biocontrol agent for snakeweeds(Gutierrezia) in the United States. Biological Control, 5, 1–10.Crossley, D.A. & Blair, J.M. (1991) A high-efficiency, “low-technology” Tullgren-typeextractor for soil microarthropods. Agriculture, Ecosystems and Environment, 34, 187–192.De Barro, P.J. (1991) Sampling strategies for above and below ground populations of Saccharicoccussacchari (Cockerell) (Hemiptera: Pseudococcidae) on sugarcane. Journal of theAustralian Entomological Society, 30, 19–20.Dosdall, L.M., Herbut, M.J., & Cowle, N.T. (1994) Susceptibilities of species and cultivars ofcanola and mustard to infestation by root maggots (Delia spp.) (Diptera: Anthomyiidae).Canadian Entomologist, 126, 251–260.

SAMPLING INSECTS FROM ROOTS 33Dutcher, J.D. & All, J.N. (1979) Damage impact of larval feeding by the grape root borer in acommercial Concord grape vineyard. Journal of Economic Entomology, 72, 159–161.East, R. & Willoughby, B.E. (1983) Grass grub (Costelytra zealandica) population collapse in thenorthern North Island. New Zealand Journal of Agricultural Research, 26, 381–390.Elvin, M.K. & Yeargan, K.V. (1985) Spatial distribution of clover root curculio, Sitona hispidulus(Fabricius) (Coleoptera: Curculionidae) eggs in relation to alfalfa crowns. Journal of theKansas Entomological Society, 58, 346–348.Fermanian, T.W., Shurtleff, M.C., Randell, R., Wilkinson, H.T., & Nixon, P.L. (1997) ControllingTurfgrass Pests. Prentice Hall, Upper Saddle River, NJ.Forrester, G.J. (1993) Resource partitioning between two species of Ceutorhynchus (Coleoptera:Curculionidae) on Echium plantagineum in a Mediterranean habitat. Bulletin of EntomologicalResearch, 83, 345–351.Fowler, R.F. & Wilson, L.F. (1971) White grub populations, Phyllophaga spp. in relation to damagedred pine seedlings in Michigan and Wisconsin plantations (Coleoptera: Scarabaeidae).Michigan Entomologist, 4, 23–28.Gange, A.C., Brown, V.K., Barlow, G.S., Whitehouse, D.M., & Moreton, R.J. (1991) Spatial distributionof garden chafer larvae in a golf tee. Journal of the Sports Turf Research Institute, 67,8–13.Geurs, M., Bongers, J., & Brussard, L. (1991) Variations of the heptane flotation method for improvedefficiency of collecting microarthropods from a sandy loam soil. Agriculture, Ecosystemsand Environment, 34, 213–221.Goldson, S.L. & Proffitt, J.R. (1988) The effect of lucerne age on its sensitivity to damage bySitona discoideus Gyllenhall (Coleoptera: Curculionidae) in Canterbury. In Proceedings of the5th Australasian Conference on Grassland Invertebrate Ecology (ed. P.P. Stahle), pp. 323–331. Universityof Melbourne, Victoria.Goldson, S.L., Frampton, E.R. & Proffitt, J.R. (1988) Population dynamics and larval establishmentof Sitona discoideus (Coleoptera: Curculionidae) in New Zealand lucerne. Journal ofApplied Ecology, 25, 177–195.Gratwick, M. (1992) Crop Pests in the UK. Chapman & Hall, London.Haarløv, N. (1947) A new modification of the Tullgren apparatus. Journal of Animal Ecology, 16,115–121.Hammer, M. (1944) Studies on the Oribatids and Collemboles of Greenland. Meddelelser omGrønland, 141, 1–210.Hanula, J.L. (1993) Vertical distribution of black vine weevil (Coleoptera, Curculionidae) immaturesand infection by entomogeneous nematodes in soil columns and field soil. Journalof Economic Entomology, 86, 340–347.Harcourt, D.G. & Binns, M.R. (1989) Sampling technique for larvae of the alfalfa snoutbeetle, Otiorhynchus ligustici (Coleoptera: Curculionidae). Great Lakes Entomologist, 22,121–126.Harrison, R.D., Gardner, W.A., Tollner, W.E., & Kinard, D.J. (1993) X-ray computed tomographystudies of the burrowing behavior of fourth instar pecan weevil (Coleoptera, Curculionidae).Journal of Economic Entomology, 86, 1714–1719.Havukkala, I., Harris, M.O., & Miller, J.R. (1992) Onion fly egg distribution in soil: a new samplingmethod and the effect of two granular insecticides. Entomologia Experimentalis et Applicata,63, 283–289.Hawthorne, D.J. & Dennehy, T.J. (1991) Reciprocal movement of grape phylloxera(Homoptera: Phylloxeridae) alates and crawlers between two differentially phylloxeraresistantgrape cultivars. Journal of Economic Entomology, 84, 230–236.Hopkin, S.P. (1997) Biology of the Springtails (Insecta: Collembola). Oxford University Press,Oxford.

34 CHAPTER 2House, G.J. & Alzugaray, M.D.R. (1989) Influence of cover cropping and no-tillage practices oncommunity composition of soil arthropods in a North Carolina agroecosystem. EnvironmentalEntomology, 18, 302–307.Kethley, J. (1991) A procedure for extraction of microarthropods from bulk samples of sandysoils. Agriculture, Ecosystems and Environment, 34, 193–200.Klironomos, J.N. & Kendrick, W.B. (1995) Stimulative effects of arthropods on endomycorrhizasof sugar maple in the presence of decaying litter. Functional Ecology, 9, 528–536.Labuschagne, L. (1999) Black vine weevil — the Millennium bug? Antenna, 23, 213–220.Lauenstein, G. (1986) Leatherjackets as pests of grasslands: their biology and control. Zeitschriftfür Angewandte Entomologie, 73, 385–432.Lussenhop, J. (1971) A simplified canister-type soil arthropod extractor. Pedobiologia, 11,40–45.Lussenhop, J., Fogel, R., & Pregitzer, K. (1991) A new dawn for soil biology: video analysisof root–soil–microbial–faunal interactions. Agriculture, Ecosystems and Environment, 34, 235–249.Maron, J.L. (1998) Insect herbivory above- and belowground: individual and joint effects onplant fitness. Ecology, 79, 1281–1293.Masters, G.J. (1995) The impact of root herbivory on plant and aphid performance: field andlaboratory evidence. Acta Oecologia, 16, 135–142.McSorley, R. & Walter, D.E. (1991) Comparison of soil extraction methods for nematodes andmicroarthropods. Agriculture, Ecosystems and Environment, 34, 201–207.Müller, H., Stinson, C.S.A., Marquardt, K., & Schroeder, D. (1989) The entomofaunas of rootsof Centaurea maculosa Lam., C. diffusa Lam., and C. vallesiaca Jordan in Europe. Journal of AppliedEntomology, 107, 83–95.Murray, P.J. & Clements, R.O. (1995) Distribution and abundance of three species of Sitona(Coleoptera: Curculionidae) in grassland in England. Annals of Applied Biology, 127, 229–237.Penev, L.D. (1992) Qualitative and quantitative spatial variation in soil wire-worm assemblagesin relation to climatic and habitat factors. Oikos, 63, 180–192.Pontin, A.J. (1978) The numbers and distribution of subterranean aphids and their exploitationby the ant Lasius flavus (Fabr.). Ecological Entomology, 3, 203–207.Roberts, R.J., Ridsdill Smith, T.J., Porter, M.R., & Sawtell, N.L. (1982a) Fluctuations in theabundance of pasture scarabs over an 18-year period of light trapping. In Proceedings of the 3rdAustralasian Conference on Grassland Invertebrate Ecology (ed. K.E. Lee), pp. 75–79. SA GovernmentPrinter, Adelaide.Roberts, R.J., Campbell, A.J., Porter, M.R., & Sawtell, N.L. (1982b) The distribution and abundanceof pasture scarabs in relation to Eucalyptus trees. In Proceedings of the 3rd AustralasianConference on Grassland Invertebrate Ecology (ed. K.E. Lee), pp. 207–214. SA GovernmentPrinter, Adelaide.Rogers, C.E. (1985) Bionomics of Eucosma womonana Kearfott (Lepidoptera: Tortricidae), aroot borer in sunflowers. Environmental Entomology, 14, 42–44.Salt, D.T., Major, E., & Whittaker, J.B. (1996) Population dynamics of root aphids feeding onSitka spruce in two commercial plantations. Pedobiologia, 40, 1–11.Schmidt, O., Scrimgeour, C.H., & Handley, L.L. (1997) Natural abundance of 15 N and 13 C inearthworms from a wheat and wheat–clover field. Soil Biology and Biochemistry, 29,1301–1308.Seastedt, T.R. (1984) Belowground macroarthropods of annually burned and unburned tallgrassprairie. American Midland Naturalist, 111, 405–408.Sheppard, A.W., Aeschlimann, J.P., Sagliocco, J.L., & Vitou, J. (1995) Below-ground herbivoryin Carduus nutans (Asteraceae) and the potential for biological control. Biological Control Scienceand Technology, 5, 261–270.

SAMPLING INSECTS FROM ROOTS 35Smart, L.E., Blight, M.M., Pickett, J.A., & Pye, B.J. (1994) Development of field strategies incorporatingsemiochemicals for the control of the pea and bean weevil, Sitona lineatus L. CropProtection, 13, 127–135.Southwood, T.R.E. & Henderson, P.A. (2000) Ecological Methods. 3rd edn. Blackwell Science,Oxford.Stewart, R.M. & Kozicki, K.R. (1987) DIY assessment of leatherjacket numbers in grassland.Proceedings of Crop Protection in Northern Britain Conference, pp. 349–353. Scottish Crop ResearchInstitute, Dundee.Strong, D.R., Maron, J.L., Connors, P.G., Whipple, A., Harrison, S., & Jefferies, R.L. (1995)High mortality, fluctuation in numbers, and heavy subterranean insect herbivory in bushlupine, Lupinus arboreus. Oecologia, 104, 85–92.Villani, M.G. & Gould, F. (1986) Use of radiographs for movement analysis of the corn wireworm,Melanotus communis (Coleoptera: Elateridae). Environmental Entomology, 15, 462–464.Villani, M.G. & Nyrop, J.P. (1991) Age-dependent movement patterns of Japanese beetle andEuropean chafer (Coleoptera: Scarabeidae) grubs in soil-turfgrass microcosms. EnvironmentalEntomology, 20, 241–251.Villani, M.G. & Wright, R.J. (1988) Use of radiography in behavioral studies of turfgrassinfestingscarab grub species (Coleoptera: Scarabaeidae). Bulletin of the Entomological Society ofAmerica, 34, 132–144.Walter, D.E., Kethley, J., & Moore, J.C. (1987) A heptane flotation method for recoveringmicroarthropods from semiarid soils, with comparison to the Merchant–Crossley highgradientextraction method and estimates of microarthropod biomass. Pedobiologia, 30,221–232.Ward, R.H. & Keaster, A.J. (1977) Wireworm baiting: use of solar energy to enhance early detectionof Melanotus depressus, M. verberans and M. mellillus in midwest cornfields. Journal ofEconomic Entomology, 70, 403–406.White, E.G. & Sedcole, J.R. (1993) A study of the abundance and patchiness of cicada nymphs(Homoptera: Tibicinidae) in a New Zealand sub-alpine shrub-grassland. New Zealand Journalof Ecology, 20, 38–51.Whitfield, G.H. & Grace, B. (1985) Cold hardiness and overwintering survival of the sugarbeetroot maggot (Diptera: Otitidae) in southern Alberta. Annals of the Entomological Society ofAmerica, 78, 501–505.Williams, P. Stahle, P.P., Gagen, S.J., & Murphy, G.D. (1982) Strategies for control of theblack field cricket, Teleogryllus commodus (Walker). In Proceedings of the 3rd Australasian Conferenceon Grassland Invertebrate Ecology (ed. K.E. Lee), pp. 365–369. SA Government Printer,Adelaide.

36 CHAPTER 2Index of methods and approachesMethodology Topics addressed CommentsField extraction methodsChemical Use of irritant chemicals to May be toxic to user; hard to produceexpel insects from soil.density estimates; may kill specimens.Behavioral Use of dark covers on a damp Only works for tipulid larvae; notsoil surface.quantifiable; produces live specimensfor culture.Baits Use of food attractants to Not quantifiable; good for obtainingobtain active larvae or adults. live specimens for culture.Hand sorting Systematic sifting through a Quantifiable, but laborious; not reallydefined volume of soil.suitable for small insects.Laboratory extraction methodsRoot dissection Removal of insects from Quantifiable, but laborious.inside a root system.Flotation Immersion of a defined Quick; good for inactive stages, butvolume of soil in a liquid quantification hampered because(usually water or brine). dead specimens are obtained too.Behavioral Use of temperature and light Quantifiable, but does not sampleto expel insects from soil. inactive stages; Soil factors andoperating conditions affect results.Laboratory visualization methodsUse of stable isotopes, videoor X-ray observationtechniques.Not quantifiable; good for behavioralstudies.

CHAPTER 3Pitfall trapping in ecological studiesB.A. WOODCOCKIntroductionPitfall trapping is one of the oldest, most frequently used, and simple of allinvertebrate sampling techniques, and yet it is also one of the most frequentlymisused. This is because pitfall trapping for surface-active invertebrates is proneto producing non-quantitative data, particularly if used without consideringthe problems associated with this sampling technique. This chapter considersthe practical aspects of pitfall trap design and installation, and then discusses thetheoretical basis of pitfall trapping that must be incorporated into experimentaldesign and analysis if this method is to be used successfully in ecological studies.This chapter provides a comprehensive review of the methods and theory behindpitfall trapping and provides information on suitable experimental protocolsto be applied in pitfall trapping sampling programs.The choice of sampling technique in any invertebrate sampling program is integralto the success of the project. The decision will not only determine the typeof invertebrates that are sampled, but also in what numbers, over what spatialscale, and how quantitative the data produced are. What sampling method willalso be influenced by more pragmatic decisions based on the money and timeavailable to each project. Sampling epigeal invertebrates, those species activeon the soil surface, is a good example of where these problems must be consideredcarefully to produce an effective sampling program within the means ofthe project.The most commonly sampled epigeal invertebrates are the ground beetles(Coleoptera: Carabidae), rove beetles (Coleoptera: Staphylinidae), wanderingspiders (e.g. Aranae: Lycosidae and Clubionidae), and ants (Hymenoptera:Formicidae). These groups are characterized as highly active, mostlypolyphagous, invertebrate predators (Greenslade 1973, Uetz & Unzicker 1976,Thiele 1977, Frank 1991). These characteristics can make these groups hard tosample using many techniques. Their active nature means that they while theymay show specific habitat associations, a spatially and temporally restrictedsampling technique may fail to catch many species. Also, polyphagous predatorsare not associated with either a particular host plant, or specific preyspecies. Any sampling strategy that could be used to target such an associationwould also be ineffective.37

38 CHAPTER 3The sampling technique used most frequently to collect epigeal invertebratesis pitfall trapping. The technique was first developed by Hertz (1927), andshortly after by Barber (1931), who used open-top containers buried with therim level to the substrate surface, so that anything falling into the containerbecomes trapped. While originally conceived as a qualitative technique, thepotential of the method for quantitatively sampling epigeal invertebrate populationswas soon realized (Fichter 1941). From this inauspicious start pitfalltraps have come to dominate epigeal invertebrate sampling (Uetz & Unzicker1976, Thiele 1977). They have been used in practically every terrestrial habitat,from deserts (Thomas & Sleeper 1977, Faragalla & Adam 1985), to forests(Niemelä et al. 1986, Spence & Niemelä 1994), to caves (Barber 1931). The techniquehas also been used to obtain information on the structure of invertebratecommunities (Hammond 1990, Jarosík 1992), habitat associations (Honêk1988, Hanski & Niemelä 1990), activity patterns (Ericson 1978, Den Boer 1981),spatial distribution (Niemelä 1990), relative abundances (Desender & Maelfait1986a, Mommertz et al. 1996), total population estimates (Gist & Crossley 1973,Mommertz et al. 1996), and distribution ranges (Barber 1931, Giblin-Davis et al.1994). Pitfall trapping also plays a role in some pest monitoring programs(Kharboutli & Mack 1993, Obeng-Ofori 1993, Rieske & Raffa 1993, Simmons et al.1998). For the last three-quarters of a century pitfall traps have proved to be oneof the most versatile, useful, and widely used invertebrate sampling techniques.The wide-scale adoption of this technique is due to a number of factors. Basictraps are cheap, and normally require no specialized manufacturing process.Traps are also easy to transport (Lemieux & Lindgren 1999), and quick to install.Perhaps one of the greatest advantages of pitfall traps is that they will samplecontinuously, requiring only periodic emptying. This not only removes biasesassociated with other techniques that sample at one point in time (Topping &Sunderland 1992), but also allows large numbers of invertebrates to be caughtover an entire season with minimal effort. This makes the technique particularlyuseful for sampling invertebrate occurring at low densities (Melbourne1999). The low levels of disturbance, both physically and aesthetically, whichpitfall trap installation and collection causes has made them useful for samplingenvironmentally sensitive areas (Melbourne 1999).Unfortunately, while Fichter (1941) was the first to recognize the values ofpitfall trapping as a quantitative tool, he was also the first to acknowledge itsfailings. As each species has the potential to respond uniquely to pitfall traps, therates at which they are caught can vary. The proportion of each species in thetraps no longer necessarily represents their relative abundance in the samplinghabitat. If pitfall traps are to be used in ecological studies it is necessary that fieldbiologists have a comprehensive understanding of both the advantages and disadvantagesof this method. This must include an understanding not only ofdifferent trap designs, and sampling strategies, but also of what can be done toimprove the quantitative nature of the data.This chapter will first consider the various designs and modifications thathave been developed for pitfall traps, and the implication of how design impacts

PITFALL TRAPPING IN ECOLOGICAL STUDIES 39on the capture rates of different species. Secondly the integral role that samplingstrategy plays in reducing biases associated with pitfall traps is reviewed.Finally the concept of activity–abundance as a tool for the quantitative interpretationof pitfall traps is described. Pitfall traps are a valuable tool in ecology,and like any tool they must be used carefully with an understanding of theirflaws if their use is not to be open to criticism.Pitfall traps, their designs and applicationIt would seem that every study uses a novel design of pitfall trap (Table 3.1); differentsizes, shapes, and construction material are normal. This is often due tothe immediate availability of materials for each study, and has led to a high levelof inconsistency between different research projects. However, pitfall trap designcan influence the capture and retention of different species. Althoughthere is no right or wrong design, knowledge of the effectiveness of differenttrap types will allow traps to be tailored to the experimental requirements andTable 3.1 Examples of the application of various modifications that have been developedfor pitfall traps.Trap type Use in ecological studies ExamplesConventional Habitat associations; spatial Ericson (1978, 1979),patterns; community Niemelä et al. (1990),structure; mark–recapture Niemelä et al. (1992),studies Dennis et al. (1997)Baited traps For aggregated or rare Walsh (1933), Rieske &species; pest monitoring Raffa (1993)Time sorting traps Determination of diel Luff (1978), Kegel (1990),activity patterns Chapman & Armstrong (1997)Barrier trapping Prevents immigration; Baars (1979), Desendersamples from a defined area; & Maelfait (1986a), Momertzclosed mark–recapture et al. (1996)experimentsDrift fences Increasing overall catch; Smith (1976), Morrill et al.identifying directional (1990), Melbourne (1999)movementRamps Biases catch to larger Bostanian et al. (1983)individuals; reduces floodingGutter traps Increases overall catch; Luff (1975, 1978),biasing catch towards larger Lawrence (1982), Spence &species Niemelä (1994)Subterranean Useful for soil active species Kuschel (1991), Owen (1995)or larval stages

40 CHAPTER 3RoofSoil surfaceSupports for roofFunnelCollecting containerKilling/preservative fluidFig. 3.1 A cut away diagram showing the design of a basic pitfall trap suitable for most epigealinvertebrate surveys. This design is only a guideline and should be modified according to thespecific requirements of each sampling program and the materials available for its construction.practical limitations of each project. This section considers what designs of pitfalltrap, and what modifications, are available. Figure 3.1 shows the design of abasic pitfall trap.Trap materialThe material traps have been constructed from has almost always been determinedby what is easily available at the time. Plastic is presently the most frequentlyused material (e.g. Honêk 1988, Niemelä 1990, Dennis et al. 1997),although prior to this both metal (e.g. Hertz 1927, Luff 1975, Smith 1976) andglass (e.g. Briggs 1960, Mitchell 1963a, Greenslade 1964) were frequently used.Species-specific responses to trap material are common (Luff 1975, Obeng-Ofori 1993). Glass is normally found to be the most effective material in terms ofnumbers of individuals captured, and is almost always superior to metals (Luff1975, Obeng-Ofori 1993). The superiority of glass over other materials is particularlyimportant if live catches are required. Glass provides few abrasionswhich insects can use to escape, although if a killing agent or preservative isused other materials like plastic are likely to be as effective (Luff 1975). However,glass is heavy, fragile, and hard to use in the construction of more special-

PITFALL TRAPPING IN ECOLOGICAL STUDIES 41ized trap designs. Rieske and Raffa (1993) suggested the use of Teflon to minimizesurface grip within traps, and so maximize invertebrate retention whereother trap materials are used. Trap color was found by Greenslade (1964) tohave no effect on the catch of Carabidae.Trap shape and sizeWhile default rather than design has dictated that traps are normally circularthere is no reason why they should always be this shape. However, there doesnot appear to be any advantage gained, in terms of overall trap efficiency, by deviatingfrom this standard circular design (Spence & Niemelä 1994). The onlyexception to this is with some of the more extreme pitfall trap designs like guttertraps, which are described below.However, different species do respond in different ways to trap shape, evenwhen there is no difference in overall trap perimeter (Baars 1979, Spence &Niemelä 1994). As a rule where communities are being compared in a givenstudy trap shape should therefore be kept constant. If the shape must be altered,then the perimeter should be kept as constant as possible (Luff 1975, Baars1979). Independent of shape there is considerable variation in the size of traps.For circular traps, diameter may vary from as little as 1.8 cm (Greenslade& Greenslade 1971, Greenslade 1973, Abensberg-Traut & Steven 1995) to over25 cm (Morrill et al. 1990). A modal diameter determined from the literatureis found to be around 6–8 cm. Trap depth is variable but tends to be at least 8–10 cm; anything below this is likely to be particularly prone to escape. Largerpitfall traps catch more individuals than smaller traps (Luff 1975, Baars 1979,Abensberg-Traut & Steven 1995), but this increase in catch is not necessarilyproportional to trap diameter (Morrill et al. 1990).Baars (1979) used simulations to show that when comparing sites the numberand shape of traps was not important providing that total perimeter area oftraps was constant. However, this will vary with the target taxa. Abensberg-Traut and Steven (1995) suggested that for a comparable area many small trapsmay be more efficient than a few large ones, particularly for species that arehighly aggregated like ants. In the case of very small diameter pitfall traps,larger species may be too big to be trapped, and will be excluded from samples(Luff 1975, Abensberg-Traut & Steven 1995). A theoretical basis exists for correctingcatch sizes of traps of different diameters and is discussed by Luff (1975);this may even be applied to traps that differ in shape. Although this correctionmethod has been used to compare traps of dissimilar shapes, and sizes (Luff1975, Spence & Niemelä 1994), Scheller (1984) found that it was not alwayseffective.RoofsRoofs covering the mouth of traps have been used in many pitfall trap studies,

42 CHAPTER 3providing protection from the elements (e.g. Fichter 1941, Honêk 1988,Hammond 1990). The roofs are supported 3–4 cm from the soil surface to allowfree access to the traps. Roofs are useful for traps both with and without preservatives,since rain will cause preservative dilution (Hammond 1990) or maydrown live insects (Briggs 1960). Roofs will also prevent debris, which providesescape routes for insects, falling into traps (Uetz & Unzicker 1976, Morrill et al.1990). Access by birds and small mammals stealing the contents of the traps isprevented (Briggs 1960, Mitchell 1963a), as is their consumption of toxicpreservatives (Marshall & Doty 1990, Hall 1991). Wire barriers have also beenused to prevent the accidental capture of small vertebrates in traps.Unfortunately roofs cause bias in the catches of pitfall traps (Joose 1965,Morrill 1975, Baars 1979). The use of transparent materials for roof coverings,however, minimizes the influence of roofs on the catches of invertebrates(Joose 1965, Baars 1979).FunnelsFunnels have been used in many studies (e.g. Gist & Crossley 1973, Faragalla &Adam 1985, Morrill et al. 1990, Clarke & Bloom 1992), normally to reduceescape where no preservative is used (Vlijm et al. 1961, Uetz & Unzicker 1976).Funnels also reduce desiccation of the trap contents and prevent vertebrate interference(Briggs 1960, Mitchell 1963a). Funnels are placed at the opening ofthe traps, and work on the same principle as lobster pots, making escape difficult.Both capture rate and trap efficiency will probably be influenced by thepresence of a funnel, although as of yet there is no evidence for this.The trap rimThe protrusion of the trap rim can repel invertebrates, although this is dependenton invertebrate size (Morrill et al. 1990, Good & Giller 1991). It has beensuggested that the trap rim should be placed 1–7 mm below the level of the substratesurface (Good & Giller 1991, Obeng-Ofori 1993). However, this is normallyawkward, and for large numbers of traps may be impractical. As a generalrule it is necessary to at least get the rim of the trap level with the substrate surface.After heavy rain erosion of soil around trap rims can occur, requiring thatthe soil be replaced (Hammond 1990).Killing agents, preservatives, and detergentsMark and recapture experiments require that the sampling procedure does notkill the catch (e.g. Ericson 1977, Parmenter & MacMahon 1989, Thomas et al.1998). However, once confined within the trap predatory species will feed onanything small enough to eat, and this may include target species of thesampling program (Mitchell 1963a, Greenslade & Greenslade 1971). Even with

PITFALL TRAPPING IN ECOLOGICAL STUDIES 43daily collection of the traps this can still be a problem. One solution is to placesoil, or another suitable substrate, in the trap to provide a refuge for smallerspecies (Ericson 1979, Honêk 1988). Wire meshes have also been used to separatelarge and small species (Lawrence 1982, Niemelä et al. 1992).When it is not necessary to keep the catch alive a killing solution is normallyused, to stop predation and reduce levels of escape (Uetz & Unzicker 1976,Curtis 1980, Waage 1985, Holopainen & Varis 1986, Lemieux & Lindgren1999). The solution will also normally act as a preservative, reducing the needfor regular collections. The choice of solution (Lemieux & Lindgren 1999) is dependenton: its effectiveness in preventing decay and fouling of specimens; thespeed with which it kills insects before they can escape; whether it will remainnon-volatile when diluted by rain, or concentrated by the sun. Other considerationsinclude legal or health and safety requirements that may prevent the useof potentially harmful chemicals (Hall 1991) (Table 3.2).Table 3.2 Killing fluids/preservatives that have been used in ecological studies, givingsuggested concentrations and listing the advantages and disadvantages of their use in pitfalltraps. The concentrations are only suggestions for a collection interval of between two andfour weeks under temperate conditions. It is suggested that to all of these an unscenteddetergent should be added to reduce surface tension.Preservative Concentrations Advantages and disadvantagesEthylene glycol 25–50% solution Freely available as car antifreeze.Good preservative. Toxic to birdsand mammals. Attractant to someinvertebrates.Propylene glycol 25–50% solution More expensive than ethylene glycol,but considered less toxic. Possibleattractant?Water N/A Freely available. Poor preservative.Formalin 5–10% solution Relatively freely available. Goodpreservative. Possible health andsafety problems. Toxic. Attractant.Saline solution 1% to saturated solution Freely available. Reasonablepreservative, but damages somespecimens. Possible attractant?Alcohol 70% solution Freely available. Good preservative.Attractant. Volatile and will evaporate.Acetic acid 5% solution Freely available. Good preservative.Attractant.Chloral hydrate An additive to above Relatively freely available and can besolutionsused to inhibit bacterial/fungalgrowth. Toxic. Possible attractant?

44 CHAPTER 3A wide variety of solutions have been used in pitfall traps, including propyleneor ethylene glycol (Digweed et al. 1995, Dennis et al. 1997), water (Briggs1960, Holopainen 1992), alcohol (Fichter 1941, Greenslade & Greenslade1971), formalin (Baars 1979, Desender & Maelfait 1986a), kerosene (Faragalla& Adam 1985), brine (van den Berghe 1992, Lemieux & Lindgren 1999), chloralhydrate (Hammond 1990), and benzoic/acetic acid (Scheller 1984). Thequality of these preservatives is variable and has meant that few are in commonuse. Water for example may be freely available, and will kill invertebrates, butsample decomposition is a problem. At present one of the most commonly usedpreservatives in ecological experiments is ethylene glycol (antifreeze). Itspopularity is based on its low cost, free availability, and good preservative andkilling qualities. However, ethylene glycol is sweet-tasting and toxic to bothbirds and mammals, which actively consume it (Beasley 1985, Marshall & Doty1990, Hall 1991). The less toxic propylene glycol has been proposed as an alternative,as it shares essentially the same beneficial qualities as ethylene glycol,although it is more expensive (Hall 1991).Preservative concentration depends on the interval between collection dates.A 50-percent ethylene glycol solution is suitable for most purposes (Epstein &Kulman 1984), although if the trap is to be checked very infrequently, e.g. lessthan once a month, concentrations as high as 100 percent may be required(Clarke & Bloom 1992). This dilution principle can be sensibly applied to mostpreservatives. It is also normal to add a small quantity of unscented detergent tothe killing solution to reduce surface tension. This will increase the efficiencyof traps, as insects slip more easily under the surface of the killing agent/preservative.It should be noted that almost all preservatives will act as attractants for atleast some species of invertebrates. For example, in the Carabidae positivespecies-specific responses have been found for the preservatives formalin(e.g. Luff 1968, Scuhravy 1970, Adis & Kramer 1975, Ericson 1979, Feoktistov1980, Scheller 1984, Holopainen & Varis 1986), ethylene/propylene glycol(Hammond 1990, Holopainen 1990, Holopainen 1992), and benzoic / acetic acid(Scheller 1984). While this will influence the relative proportions of differentspecies caught in pitfall traps, the use of preservatives is often a necessary evil.BaitsThe use of baits is one of the only techniques in pitfall trapping that intentionallybiases the catch size of different species (Walsh 1933, Greenslade 1964,Greenslade & Greenslade 1971). Their use should be strictly for qualitativeanalyses, such as determination of habitat association (Hanski & Niemelä1990), or in producing total species inventories (Romero & Jaffe 1989,Hammond 1990). There is an argument for the quantitative use of baits in theanalysis of population size of a single species occurring at low densities, or onethat is too aggregated to ensure capture with more passive approaches to pitfalltrapping. This is particularly so when monitoring pest populations where warn-

PITFALL TRAPPING IN ECOLOGICAL STUDIES 45ing of population increases in a single species is valuable (Rieske & Raffa 1993,Giblin-Davis et al. 1994, Yasuda 1996). In combination with other samplingmethods, baits have proved useful for sampling ants, whose aggregated distributionoften makes a determination of total species richness difficult(Greenslade & Greenslade 1971, Romero & Jaffe 1989).Solid baits, including carrion, fruit, or dung (Romero & Jaffe 1989, Hammond1990, Hanski & Niemelä 1990, Giblin-Davis et al. 1994), are normally positionedon a platform in the middle of the trap, or suspended immediately aboveit. Liquid baits, such as beer or honey solutions (Greenslade & Greenslade1971), can be placed directly into the collecting vessel, although they may bepoor preservatives and so the catch may require regular collection. Other typesof baits may be highly specific and are used more frequently in pest monitoringprograms. Such baits have included pheromones (Yasuda 1996) and alpha- andbeta-pinenes (Rieske & Raffa 1993).Specialized designsThe evolution of the pitfall trap from its initial simple concept (Hertz 1927,Barber 1931) has produced designs that have increased catch sizes, or allowedseveral normally hard-to-investigate aspects of invertebrate ecology to be considered.These traps include: time-sorting traps for determining diel activity patterns(Houston 1971, Barndt 1976, Luff 1978, Chapman & Armstrong 1997);gutter traps, which are essentially highly elongated conventional pitfall traps,that will increase the overall catch size (Luff 1975, Luff 1978, Lawrence 1982,Spence & Niemelä 1994); drift fences, which are strips of metal or plastic placedon the surface to direct insects towards the pitfall trap, so increasing the overallcatch or identifying the directional movement of insects (Smith 1976, Desender& Maelfait 1986a, Morrill et al. 1990, Melbourne 1999); barrier trapping, whichuses normal pitfall traps in conjunction with an outer barrier preventing the immigrationor emigration of invertebrates from a spatially delimited area (Gist &Crossley 1973, Baars 1979, Holopainen & Varis 1986, Desender & Maelfait1986a, Mommertz et al. 1996); ramp traps, which use a ramp to lead up to thecollection chamber, so that the trap does not need to be buried, and can be usedon rocky ground (Bostanian et al. 1983, Spence & Niemelä 1994). Recently,some entomologists have experimented with subterranean pitfall traps. Thesetraps are placed so that they are in a hole some distance below the soil surface,e.g. 10–20 cm. A column of coarse wire mesh encircling the rim of the trap andextending to the surface prevents soil falling in, while allowing insects crawlingthough the soil to be collected (Kuschel 1991, Owen 1995).Sampling strategyIndependent of the actual design of the trap, the sampling strategy employedcan be used to maximize the quantitative potential of a pitfall trapping program.

46 CHAPTER 3Sampling strategy refers to the number of traps, their spatial arrangement, andthe duration of sampling. Differences in sampling strategy affect not only theproportion of the community sampled, but also the relative abundances ofspecies caught in traps. For these reasons it is important to carefully considersampling strategy before initiating any experiment.Trap numberThe number of traps used to obtain information from a particular sampling areais highly variable in the literature, ranging from as few as two (Melbourne 1999)to as many as 300 (Niemelä et al. 1990). This number usually depends both onthe size of the area to be sampled and on the specific design of the pitfall traps.For example, Melbourne (1999) used drift fences in conjunction with pitfalltraps to increase their effective perimeter, and so used few traps. In the case ofNiemelä et al. (1990) 300 traps were used to sample a 19-hectare woodland.Obrtel (1971) showed that the highest incremental increases in overallspecies richness occurred for the first five traps in beetle communities. However,Stein (1965) considered that fewer than 20 traps would be insufficient todetermine the number of Carabidae species in a site, while Bombosch (1962)found that increasing the number of traps above 70 still caught additionalspecies. It is likely that reliable data on the species at a site can be obtained from12 pitfall traps when considering common temperate Carabidae fauna (Obrtel1971). Use of preliminary studies, or previous published work, may be the mostreliable method for estimating trap number, particularly in long-term studies(Uetz & Unzicker 1976). As a rule, the more pitfall traps the better an individualcommunity will be sampled, although this must be traded off against the extrawork required in processing the data.Spatial arrangementTraps are rarely positioned randomly within a single plot or site, due to practicalproblems of finding them again. Frequently used trapping patterns are lineartransects (e.g. Mitchell 1963b, Honêk 1988, Good & Giller 1991, Kharboutli &Mack 1993), and grids (e.g. Ericson 1979, Epstein & Kulman 1984, Niemeläet al. 1992) (Table 3.3).The arrangement of traps and their number can reduce overall trapping efficiency.Luff (1975) demonstrated theoretically that a correction factor shouldbe applied to traps placed in a grid, as outer traps will shield inner traps andreduce their effective diameter. This will have the effect of reducing the sizes ofcatches. Scheller (1984) experimentally demonstrated this effect, but for onlyone carabid species. However, this does have implications when comparingsites using the same number and type of trap but with different spatialarrangements.The separation between each trap will be dependent on the area of the sam-

PITFALL TRAPPING IN ECOLOGICAL STUDIES 47Table 3.3 Frequently used spatial arrangements of pitfall traps, giving the application ofthese arrangements in ecological studies and the advantages and disadvantages of eachapproach. In all cases it is often advantageous to use an obvious marker for each trap,such as a flag, to aid in relocation.Spatial layout Appearance Advantages and disadvantagesRandom From a practical perspective traps can be extremelyhard to find, although each trap can be consideredas statistically independent. Trap separation isunpredictable.GridTransect Commonly used approach as it provides good evencoverage of the sampling area, while the individualplots are relatively simple to relocate. By adjustingtrap separation, individual traps’ statisticalindependence can be maintained or avoided.Suitable for the identification of the effects ofenvironmental gradients on invertebratecommunities. For example edge effects infragmented woodlands or altitudinal gradients.pling site, the number of traps, and their diameter. It may be desirable to increasetrap separations, ensuring that there is an even coverage of traps overthe whole of the sample plot. Divisions ranging from 0.3 m (Luff 1975) to 30 m(Honêk 1988) have been used, although separations of between 5 and 10 m aremore common (Baars 1979, Holopainen 1990, Niemelä 1990, Kharboutli &Mack 1993). As trap size increases so should separation (Uetz & Unzicker 1976).It is common practice to amalgamate trap contents within a given samplingpoint to reduce small-scale spatial differences in catch sizes between adjacenttraps. If it is required that the catches of different traps are to be independentfrom each other then large separations are required. Digweed et al. (1995) suggestthat a minimum separation of 25 m is required in Carabidae communities iftraps are to be statistically independent.Sampling duration and temporal patternThe duration over which pitfall traps have been used to sample epigeal invertebratesranges from as little as little as two days (Greenslade 1973) to over threeyears (Clarke & Bloom 1992) for an individual experiment. Long-term samplingprograms may sample essentially indefinitely. However, following thework of Baars (1979) and Den Boer (1979), it is now acknowledged to be necessaryfor quantitative work to sample over the entire activity period of the communityin question. Baars (1979) considers this period to be a year, although

48 CHAPTER 3this is somewhat conservative. Most temperate studies ignore at least the winterseason, as catches during this period are low. At a minimum, a reasonablesampling period should be greater than four months (e.g. Obrtel 1971, Epstein& Kulman 1984, Niemelä 1990). Pitfall trap catches based on these long samplingperiods have been shown to have good correlations with the abundance ofseveral species of Carabidae (Baars 1979, Den Boer 1979, Luff 1982). Samplingover such extended periods is necessary as the activity of a species will vary in itsseasonal distribution from site to site; however, the total length and intensityof activity is hypothesized to be approximately the same between sites (Baars1979, Den Boer 1979).Trap catches should therefore only be used to infer differences in populationsize for one species between sites and should not be used to provide informationon the relative population sizes of each species (Baars 1979, Den Boer 1979).However, Baars (1979) states that comparisons of the population sizes ofseveral species from different sites may be possible with samples taken overthe whole year, providing the relationship between the true density and thesize of the pitfall trap catch for each species is known. Such relationships areunknown for most species. Hanski and Niemelä(1990) suggest that, althoughabsolute population sizes may not be known, a large difference in the relativeabundances of two species is enough to infer that one is more abundant than theother.Where it has not been possible to sample for long periods of time, data maystill have some quantitative value. Niemelä et al. (1990) showed that shortersampling periods contain important biological information. Temporal subsamplesof between 10 and 28 days retain the approximate rank and relativeproportions of the dominant species when compared to data from a muchlonger sampling period. However, sample similarity increased and variance decreasedwhen the sampling sub-period was increased. Sampling within theseshort time periods provides extremely limited and potentially unreliable information,and should be avoided where possible.DepletionWhen a killing agent/preservative is used in the trap, depletion of the localpopulation can occur, which may give the impression of a reduction in populationsize (e.g. Luff 1975, Ericson 1979, Digweed et al. 1995). In sampling programsthat occur over a long period, depletion of larval stages early in the seasonmay result in smaller adult populations. This will remain unnoticed unless larvalstages have been specifically identified. In the cases of the Carabidae the subterraneanlifestyle of most larvae (Kegel 1990) means that their capture inpitfalls is low.Although the effects of depletion can be reduced by using traps at moderatedensities (Greenslade 1973, Digweed et al. 1995), Digweed et al. (1995) suggestthat high trap densities could be useful. By using high trap densities the populationspresent within the sphere of influence of the traps are likely to be sampled

PITFALL TRAPPING IN ECOLOGICAL STUDIES 49in their entirety. The success of this will depend on the levels of immigrationinto this sphere of influence (Den Boer 1970).Surrounding vegetation structureAn increase in structural complexity of vegetation around pitfall traps can resultin a reduction in the catch (Greenslade 1964, Melbourne 1999). Melbourne(1999) hypothesized this to be due to an increase in the total surface area availablefor invertebrates to move on in structurally complex habitats. This causes adecrease in the effective number of pitfall traps per unit area, reducing the overallpitfall trap catch. For larger species, or species primarily active on the soil surface,direct impedance by the vegetation will be more likely to reduce catch size(Greenslade 1964). Since vegetation structure will not be static throughout agrowth season the effects may also change apparent population sizes as thedilution/impedance effect of vegetation structure becomes more prominent asthe season progresses (Greenslade 1964, Melbourne 1999).For these reasons, pitfall traps should not be used to compare habitatsthat have field-layer vegetation that is structurally dissimilar (Greenslade 1964,Maelfait & Baert 1975, Melbourne 1999). If such a comparison is integral to thestudy, vegetation surrounding each trap should be removed to standardize theimmediate area surrounding each trap (Greenslade 1964, Penny 1966,Melbourne 1999).Digging-in effectsDigging in effects are a temporary increase in the capture rate of pitfall traps inresponse to the physical disturbance caused by trap installation. These do notrepresent real increases in the density of surface-active invertebrates (Joose1965, Joose & Kapteijn 1968, Greenslade 1973, Digweed et al. 1995). Diggingineffects have been recorded for species of Collembola (Joose 1965, Joose &Kapteijn 1968), ants (Greenslade 1973), and carabids (Digweed et al. 1995),although wandering spiders have not been found to exhibit this behavior(Greenslade 1973). The extent to which digging-in effects occur is normallyspecies-specific (Joose 1965, Greenslade 1973, Digweed et al. 1995). The durationof digging-in effects is also variable. In Collembola it can be as short as a singleday (Joose 1965). Digging-in effects are considered to be minimal for mostgroups after a week (Majer 1978). For this reason it is advisable to ignore thefirst week’s catch during a sampling program.Activity–abundanceAs the rate of capture of most invertebrates is proportional to their activity(Maelfait & Baert 1975, Curtis 1980), the numbers of each species caught will

50 CHAPTER 3not reflect their true abundance. Instead their rate of capture will be proportionalto the interaction between their abundance and activity, this is theconcept of activity–abundance (Tretzel 1954, Heydemann 1957, Thiele 1977).Species that are largely sessile, but occur at high abundances, may be underrepresentedin pitfall traps compared to less abundant but more active species.This redefinition of what pitfall catches represent provides a much sounderconceptual framework for the interpretation of pitfall trap data. However,without information on the activity of each species it is almost impossible to relatepitfall trap catches to the true relative abundances of different species. Informationon the activity of different species is relatively infrequent in theliterature (e.g. Halsall & Wratten 1988). In addition to this there are additionalproblems with activity–abundance, as while activity may be correlated withcapture rate it is likely to be confounded by behavioral peculiarities of eachspecies. These behavioral differences will influence rates of capture independentof the activity and density of different species (e.g. Den Boer 1981, Halsall &Wratten 1988, Morrill et al. 1990, Obeng-Ofori 1993, Topping 1993, Mommertzet al. 1996). Nonetheless, this concept is valuable in the interpretation of pitfalltrap catches.ConclusionsWhile the quantitative nature of pitfall traps is likely to remain questionable,they are still one of the most frequently used collecting techniques for surfacedwelling invertebrates. While this choice may seem irrational in the face of theirmany problems, their use is probably no more questionable than most othersampling techniques used for invertebrates. Every sampling technique willhave inherent biases resulting from individual species behavior. These individualbehaviors will influence not only how often a species comes into contactwith a trap, but also how it responds when it encounters it. It would seem unlikelythat any trapping method has ever provided a perfect representation ofthe relative abundances of each species in a habitat. While there will always besome species that are highly misrepresented in pitfall traps, the vast majority arelikely to be represented at frequencies that at least reflect their true relativeabundances. Determining a priori which species will be highly misrepresentedin pitfall trap samples cannot be achieved on the basis of general morphological,or even taxonomic, trends. Essentially these highly misrepresented species canbe considered as being uncontrolled for random variation, which every collectingtechnique is prone to.It is also important to appreciate the limitations of pitfall trapping in terms ofwhat it does actually catch. The method is not an all-purpose technique suitablefor catching every species from a predetermined taxonomic group, e.g. theCarabidae. Instead the method is more likely to be guild-specific, targeting onlythose species that are highly active on the soil surface. For example, while most

PITFALL TRAPPING IN ECOLOGICAL STUDIES 51species of Carabidae are surface-active insects, member of the genus Dromius arearboreal (Terell-Nield 1990). Such species are going to be largely absent frompitfall trap catches: while they may be part of a taxonomic group targeted by pitfalltraps and present in an area being sampled, they are not part of the guild ofsurface-active invertebrates actually caught by pitfall traps. While species not inthis surface-active guild may still occur in low numbers, it is possible to try toremove them from the dataset by ignoring those species representing the bottom1–5 percent of the total abundance of individuals (Dennis et al. 1997). Thisremoval of some lower percentage of the catch also has the advantage of removingspecies that may not be truly associated with a habitat but are instead intransit through it (Den Boer 1977, Desender & Maelfait 1986b, Dennis et al.1997).With a good understanding of the flaws associated with pitfall traps, and withproper precautions taken to deal with these problems, it should be possible touse this method at least semi-quantitatively. This should always be done tentatively,and highly questionable results should be treated with caution.ReferencesAbensberg-Traut, M. & Steven, D. (1995) The effects of pitfall trap diameter on ant speciesrichness (Hymenoptera: Formicidae) and species composition of the catch in a semi-arideucalypt woodland. Australian Journal of Ecology, 20, 282–287.Adis, J. & Kramer, E. (1975) Formaldehyd-lösung attrahiert Carabus problematicus (Coleoptera:Carabidae). Entomologica Germanica, 2, 121–125.Baars, M.A. (1979) Catches in pitfall traps in relation to mean densities of carabid beetles.Oecologia, 41, 25–46.Barber, H.S. (1931) Traps for cave inhabiting insects. Journal of the Elisha Michell Scientific Society,46, 259–266.Barndt, D. (1976) Untersuchung der diurnalen und saisonalen Aktivität von Käfern mit einerneu entwickelten Electro-bodenfalle. Verhandlungen des Botanischen Vereins der ProvinzBrandenberg, 112, 103–122.Beasley, V.R. (1985) Diagnosis of ethylene glycol (antifreeze) poisoning. Feline Practice, 15,41–46.Bombosch, S. (1962) Untersunchungen über die Auswertbarkeit von Fallenfängen. Zeitschriftfür Angewandte, Zoology, 49, 149–160.Bostanian, N.J., Boivin, G., & Goulet, H. (1983) Ramp pitfall trap. Journal of Economic Entomology,76, 1473–1475.Briggs, J.B. (1960) A comparison of pitfall trapping and soil sampling in assessing populationsof two species of ground beetles (Col.: Carabidae). East Malling Research Station Annual Report,48, 108–12.Chapman, P.A. & Armstrong, G. (1997) Design and use of a time-sorting pitfall trap for predatoryarthropods. Agriculture, Ecosystem and Environment, 65, 15–21.Clarke, W.H. & Bloom, P.E. (1992) An efficient and inexpensive pitfall trap system. EntomologicalNews, 103, 55–59.Curtis, D.J. (1980) Pitfalls in spider community studies (Archnida, Aranae). Journal of Arachnology,8, 271–280.

52 CHAPTER 3Den Boer, P.J. (1970) On the significance of dispersal power for populations of carabid beetles(Coleoptera, Carabidae). Oecologia, 4, 1–28.Den Boer, P.J. (1977) Dispersal power and survival: carabids in a cultivated countryside. LandbouwhogeschoolWageningen The Netherlands Miscellaneous Papers, 14. H. Veenman & Sons,Wageningen.Den Boer, P.J. (1979) The individual behaviour and population dynamics of some carabid beetlesin forests. Miscellaneous Papers LH Wageningen, 18, 157–166.Den Boer, P.J. (1981) On the survival of populations in a heterogeneous and variable environment.Oecologia, 50, 39–53.Dennis, P., Young, M.R., Howard, C.L., & Gordon, I.J. (1997) The response of epigeal beetles(Col.: Carabidae, Staphylinidae) to varied grazing regimes on upland Nardus stricta grasslands.Journal of Applied Ecology, 34, 433–443.Desender, K. & Maelfait, J.P. (1986a) Pitfall trapping with enclosures: a method for estimatingthe relationship between the abundances of coexisting carabid species (Coleoptera:Carabidae). Holarctic Ecology, 9, 245–250.Desender, K. & Maelfait, J.P. (1986b) The relation between dispersal power, commonness andbiological features of carabid beetles (Coleoptera, Carabidae). Annales de la Societe Royale Zoologiquede Belgique, 116, 84–94.Digweed, S.C., Currie, C.R., Cárcamo, H.A., & Spence, J.R. (1995) Digging out the “digging-ineffect” of pitfall traps: influences of depletion and disturbance on catches of ground beetles(Coleoptera: Carabidae). Pedobiologia, 39, 561–567.Epstein, M.E. & Kulman, H.M. (1984) Effects of aprons on pitfall catches of carabid beetles inforests and fields. The Great Lakes Entomologist, 17, 215–221.Ericson, D. (1977) Estimating population parameters of Pterostichus cupreus and P. melanarius(Carabidae) in arable fields by means of capture–recapture. Oikos, 29, 407–417.Ericson, D. (1978) Distribution, activity and density of some Carabidae (Coleoptera) in winterwheat fields. Pedobiologia, 18, 202–217.Ericson, D. (1979) The interpretation of pitfall catches of Pterostichus cupreus and Pt. melanarius(Coleoptera, Carabidae) in cereal fields. Pedobiologia, 19, 320–328.Faragalla, A.A. & Adam, E.E. (1985) Pitfall trapping of tenebrionid and carabid beetles(Coleoptera) in different habitats in the central region of Saudi Arabia. Zeitschrift für AngewandteEntomologie, 99, 466–471.Feoktistov, B.F. (1980) Effectivost lovushek Barberaraznoga tipa. Zooliknesky Zhurnal, 59,1554–1558.Fichter, E. (1941) Apparatus for the comparison of soil surface arthropod populations. Ecology,22, 338–339.Frank, J.H. (1991) Staphylinidae. In An introduction to Immature Insects of North America (ed. F.W.Stehr), pp. 341–352. Kendall-Hunt, Dubuque, Iowa.Giblin-Davis, R.M., Peña, J.E., & Duncan, R.E. (1994) Lethal pitfall trap for the evaluationof semiochemical-mediated attraction of Metamasius hemipterus sericeus (Coleoptera:Curculionidae). Florida Entomologist, 77, 247–255.Gist, C.S. & Crossley, J.D.A. (1973) A method for quantifying pitfall traps. Environmental Entomology,2, 951–952.Good, J.A. & Giller, P.S. (1991) The effect of cereal and grass management on the Staphylinidae(Coleoptera) assemblages in south-west Ireland. Journal of Applied Ecology, 28,810–826.Greenslade, P. & Greenslade, P.J.M. (1971) The use of baits and preservatives in pitfall traps.Journal of the Australian Entomological Society, 10, 253–260.Greenslade, P.J.M. (1964) Pitfall trapping as a method for studying populations of Carabidae(Coleoptera). Journal of Animal Ecology, 33, 301–310.

PITFALL TRAPPING IN ECOLOGICAL STUDIES 53Greenslade, P.J.M. (1973) Sampling ants with pitfall traps: digging in effects. Insectes Sociaux,20, 343–353.Hall, D.W. (1991) The environmental hazard of ethylene glycol in insect pit-fall traps. TheColeopterists Bulletin, 45, 193–194.Halsall, N.B. & Wratten, S.D. (1988) The efficiency of pitfall trapping for polyphagous predatoryCarabidae. Ecological Entomology, 13, 293–299.Hammond, P.M. (1990) Insect abundance and diversity in the Dumoga-Bone national park,N.Sulawesi, with special reference to the beetle fauna of lowland rain forest in the Toraut region.In Insects and Rainforests of South East Asia (Wallacea) (ed. W.J. Knight & J.D. Holloway),pp. 197–254. The Royal Entomological Society of London, London.Hanski, I. & Niemelä, J. (1990) Elevation distributions of dung and carrion beetles in NorthernSulawesi. In Insects and the rain forests of South East Asia (Wallacea) (ed. W.J. Knight & J.D.Holloway), pp. 145–153. The Royal Entomological Society of London, London.Hertz, M. (1927) Huomioita petokuoriaisten olinpaikoista. Luonnon Ystävä, 31, 218–222.Heydemann, B. (1957) Die Biotopstruktur als Raumwiderstand und Raumfulle für dieTierwelt. Verhandlungen der Deutschen Zoologischen Gesellschaft Saarbrücken, 56,332–347.Holopainen, J.K. (1990) Influence of ethylene glycol on the numbers of carabids and other soilarthropods caught in pitfall traps. In The role of Ground Beetles in Environmental and EcologicalStudies (ed. N.E. Stork), pp. 339–341. Intercept, Hampshire, UK.Holopainen, J.K. (1992) Catch and sex ratio of Carabidae (Coleoptera) in pitfall traps filledwith ethylene glycol or water. Pedobiologia, 36, 257–261Holopainen, J.K. & Varis, A.L. (1986) Effects of mechanical barriers and formalin preservativeon pitfall catches of carabid beetles (Coleoptera, Carabidae) in arable fields. Journal of AppliedEntomology, 102, 440–445.Honêk, A. (1988) The effects of crop density and microclimate on pitfall trap catches of Carabidae,Staphylinidae (Coleoptera), and Lycosidae (Aranea) in cereal fields. Pedobiologia, 32,233–242.Houston, W.W.K. (1971) A mechanical time sorting pitfall trap. Entomologist’s Monthly Magazine,107, 214–216.Jarosík, V. (1992) Pitfall trapping and species abundance relationships: a value for carabid beetles(Coleoptera: Carabidae). Acta Entomologica, Bohemoslovaca, 89, 1–12.Joose, E.N.G. (1965) Pitfall-trapping as a method for studying surface dwelling Collembola.Zeitschrift für Morphologie und Ökologio der Tiere, 55, 587–596.Joose, E.N.G. & Kapteijn, J.M. (1968) Activity-stimulating phenomena caused by fielddisturbancein the use of pitfall traps. Oecologia, 1, 385–392.Kegel, B. (1990) Diurnal activity of carabid beetles living on arable land. In The role ofGround Beetles in Environmental and Ecological Studies (ed. N.E. Stork), pp. 65–76. Intercept,Hampshire, UK.Kharboutli, M.S. & Mack, T.P. (1993) Comparison of three methods for sampling arthropodpests and their natural enemies in peanut fields. Journal of Economic Entomology, 86,1802–1810.Kuschel, G. (1991) A pitfall trap for hypogean fauna. Curculio, 31, 5.Lawrence, K.O. (1982) A linear pitfall trap for mole crickets and other soil arthropods. FloridaEntomologist, 65, 376–377.Lemieux, J.P. & Lindgren, B.S. (1999) A pitfall trap for large-scale trapping of Carabidae: comparisonagainst conventional design using two different preservatives. Pedobiologia, 43,245–253.Luff, M.L. (1968) Some effects of formalin on the numbers of Coleoptera caught in pitfall traps.Entomologist’s Monthly Magazine, 104, 115–116.

54 CHAPTER 3Luff, M.L. (1975) Some features influencing the efficiency of pitfall traps. Oecologia, 19,345–357.Luff, M.L. (1978) Diel activity patterns of some field Carabidae. Ecological Entomology, 3,53–62.Luff, M.L. (1982) Population dynamics of Carabidae. Annales of Applied Biology, 101,164–170.Maelfait, J.P. & Baert, L. (1975) Contributions to the knowledge of the arachno- and entomofaunaof different wood habitats. Part I. Sampled habitats, theoretical study of the pitfallmethod, survey of the captured taxa. Biol Jb Dodonaea, 43, 179–196.Majer, J.D. (1978) An improved pitfall trap for sampling ants and other epigaeic invertebrates.Journal of the Australian Entomological Society, 17, 261–262.Marshall, D.A. & Doty, R.L. (1990) Taste responses of dogs to ethylene glycol, propylene glycol,and ethylene glycol-based antifreeze. Journal of the American Veterinary Medical Association, 12,1599–1602.Melbourne, B.A. (1999) Bias in the effects of habitat structure on pitfall traps: an experimentalevaluation. Australian Journal of Ecology, 24, 228–239.Mitchell, B. (1963a) Ecology of two carabid beetles, Bembidion lampros (Herbst) and Trechusquadristriatus (Schrank). I. Life cycles and feeding behaviour. Journal of Animal Ecology, 32,289–299.Mitchell, B. (1963b) Ecology of two carabid beetles, Bembidion lampros (Herbst) and Trechusquadristriatus (Schrank). II. Studies on populations of adults in the field, with special referenceto the technique of pitfall trapping. Journal of Animal Ecology, 32, 377–392.Mommertz, S., Schauer, C., Kösters, N., Lang, A., & Filser, J. (1996) A comparison of D-vac suction,fenced and unfenced pitfall trap sampling of epigeal arthropods in agroecosystems.Annales Zoolgica Fennici, 33, 117–124.Morrill, W.L. (1975) Plastic pitfall trap. Environmental Entomology, 4, 596.Morrill, W.L., Lester, D.G., & Wrona, A.E. (1990) Factors affecting efficacy of pitfall traps forbeetles (Coleoptera: Carabidae and Tenebrionidae). Journal of Entomological Science, 25,284–293.Niemelä, J. (1990) Spatial distribution of carabid beetles in the southern Finnish taiga: thequestion of scale. In The Role of Ground Beetles in Ecological and Environmental Studies (ed. N.E.Stork), pp. 143–155. Intercept, Hampshire, UK.Niemelä, J., Halme, E., Pajunen, T., & Haila, Y. (1986) Sampling spiders and carabid beetleswith pitfall traps: the effects of increased sampling effort. Annales Entomologici Fennici, 52,109–111.Niemelä, J., Halme, E., & Haila, Y. (1990) Balancing sampling effort in pitfall trapping of carabidbeetles. Entomologica Fennica, 1, 233–238.Niemelä, J., Spence, J.R., & Spence, D.H. (1992) Habitat associations and seasonal activity ofground-beetles (Coleoptera: Carabidae) in central Alberta. The Canadian Entomologist, 124,521–540.Obeng-Ofori, D. (1993) The behaviour of 9 stored product beetles at pitfall trap arenas andtheir capture in millet. Entomologica Experimentalis et Applicata, 6, 161–169.Obrtel, R. (1971) Number of pitfall traps in relation to the structure of the catch of soil surfaceColeoptera. Acta Entomologica Bohemoslavaca, 68, 300–309.Owen, J.A. (1995) A pitfall trap for repetitive sampling of hypogean arthropod faunas. Entomologist’sRecord, 107, 225–228.Parmenter, R.R. & MacMahon, J.A. (1989) Animal density estimation using a trapping web design:Field validation experiments. Ecology, 70, 169–179.Penny, M.M. (1966) Studies on certain aspects of the ecology of Nebria brevicolis (F.)(Coleoptera, Carabidae). Journal of Animal Ecology, 35, 505–512.

PITFALL TRAPPING IN ECOLOGICAL STUDIES 55Rieske, L.K. & Raffa, K.F. (1993) Potential use of baited pitfall traps in monitoring pine rootweevil, Hylobius picivorus, and Hylobius radicis (Coleoptera: Curculionidae) populations andinfestation levels. Forest Entomology, 86, 475–485.Romero, H. & Jaffe, K. (1989) A comparison of methods for sampling ants (Hymenoptera,Formicidae) in savannahs. Biotropica, 21, 348–352.Scheller, H.V. (1984) Pitfall trapping as the basis for studying ground beetle (Carabidae) predationin spring barley. Tidsskrift for Planteval, 88, 317–324.Scuhravy, V. (1970) Zur Anlockungsfähigkeit von Formalin für Carabiden in Bodenfallen.Beitrage Entomologie, 20, 371–374.Simmons, C.L., Pedigo, L.P., & Rice, M.E. (1998) Evaluation of seven sampling techniques forwireworms (Coleoptera: Elateridae). Environmental Entomology, 27, 1062–1068.Smith, B.J. (1976) A new application in the pitfall trapping of insects. Transactions of the KentuckyAcademy of Science, 37, 94–97.Spence, J.R. & Niemelä, J.K. (1994) Sampling carabid assemblages with pitfall traps: The madnessand the method. The Canadian Entomologist, 126, 881–894.Stein, W. (1965) Die Zusammensetzung der Caribidenfauna einer weisen mit starkwechselnden Feuchtigkeitsverhältnissen. Zeitschrift für Morphologie und Ökologie der Tiere, 55,83–99.Terell-Nield, C. (1990) Is it possible to age woodlands on the basis of their carabid beetle diversity?The Entomologist, 109, 136–145.Thiele, H.-U. (1977) Carabid Beetles in Their Environment: a Study on Habitat Selection by Adaptationin Physiology and Behavior. Springer, New York.Thomas, C.F.G., Parkinson, L., & Marshall, E.J.P. (1998) Isolating the components ofactivity–density for the carabid beetle Pterostichus melanarius in farmland. Oecologia, 116,103–112.Thomas, J.D.B. & Sleeper, E.L. (1977) The use of pitfall traps for estimating the abundance ofarthropods, with special reference to the Tenebrionidae (Coleoptera). Annals of the EntomologicalSociety of America, 70, 242–248.Topping, C.J. (1993) Behavioural responses of three linyphiid spiders to pitfall traps. EntomologicaExperimentalis et Applicata, 68, 287–293.Topping, C.J. & Sunderland, K.D. (1992) Limitations to the use of pitfall traps in ecologicalstudies exemplified by a study of spiders in a field of winter wheat. Journal of Applied Ecology,29, 485–491.Tretzel, E. (1954) Riefe- und Fortpflanzungszeit bei Spinnen. Zeitschrift für Morphologie undÖkologie der Tiere, 42, 643–691.Uetz, G.W. & Unzicker, J.D. (1976) Pitfall trapping in ecological studies of wandering spiders.Journal of Arachnology, 3, 101–111.van den Berghe, E. (1992) On pitfall trapping invertebrates. Entomological News, 103, 149–156.Vlijm, L., Hartsuyker, L., & Richter, C.J.J. (1961) Ecological studies on carabid beetles. I. Calathusmelanocephalus (Linn.). Archives Nıerlandaises de Zoologie, 14, 410–422.Waage, B.E. (1985) Trapping efficiencies of carabid beetles in glass and plastic pitfall traps containingdifferent solutions. Fauna Norvegica, Series B, 32, 33–36.Walsh, G.B. (1933) Studies in British necrophagous Coleoptera. II. The attractive powers ofvarious natural baits. Entomologist’s Monthly Magazine, 69, 28–32.Yasuda, K. (1996) Attractiveness of pitfall traps to West-Indian sweet potato weevil, Euscepespostfasciatus (Fairmaire) (Coleoptera: Curculionidae). Japanese Journal of Applied Entomologyand Zoology, 40, 97–102.

56 CHAPTER 3Index of methods and approachesMethodology Topics addressed CommentsTrap design and installationTrap material Materials used in pitfall trap Glass is seen to be one of the mostconstruction and their relative effective at preventing escapes, butefficiency in catchingplastic-based pitfall traps are moreinvertebrates.practical.Trap shape and size The effect of trap shape and size Consistency in size and shapeon the capture rates of different within an experiment isspecies.recommended.Roofs The use of roofs to protect Roofs and covers can reduceagainst weather conditions and damage to the catch, although maypreventing damage/capture by effect relative capture rates of targetbirds and mammals.species.Funnels The use of funnels to increase Where the catch is kept alivecapture rates and reduce funnels are useful in reducingdamage to the catch.escape rates.Trap rim Protruding trap rims influence Trap rims must be flush with thecapture rate.substrate surface.Killing and The use of killing agents to Killing agents and preservatives canpreservative agents increase capture rate and act as attractants for some speciesprevent decomposition. and may be toxic to vertebrates.Baits Attractants for infrequently Useful for highly aggregatedoccurring or target species. species or those targeted by pestmonitoring programs.Specialized designs Unconventional designs of Traps considered include driftpitfall traps and their value in fences, gutter traps, time sort traps,asking specific ecological barrier traps, ramp traps, andquestions.subterranean pitfall traps.Sampling strategyTrap numbers Optimal trap numbers and Twelve pitfall traps are suggested asspecies accumulation rates. a suitable number in mostsituations.Spatial arrangement Trap arrangement into grids, Trap arrangement is chosentransects, and randomprimarily for practical reasons,positioning; their relative although it may influence capturebenefits and uses.rates.Sampling duration How duration of trapping will Whole-season sampling periods areinfluence the quantitative recommended.value of the catches.Depletion effects Long or intensive trapping can Depletion may affect larval stages,reduce natural population sizes. influencing future demographicpatterns.Continued

PITFALL TRAPPING IN ECOLOGICAL STUDIES 57Methodology Topics addressed CommentsVegetation Vegetation structure will Comparisons between structurallyimpediment influence capture rates due to different habitats should beimpedance of movement and avoided.dilution effects.Digging-in effects Immediately after pitfall trap After one week most digging ininstallation, capture rates are effects have dissipated.unusually high.Activity–abundance The interaction between This concept is of key importance inindividual species abundance the interpretation of pitfall trapand their relative activity rates. catches.

CHAPTER 4Sampling methods forforest understory vegetationCLAIRE M.P. OZANNEIntroductionThe understory is a varied and complex habitat, forming a key layer in the forestecosystem. Lawrence (1995) defines the understory as the “vegetation layer betweenthe tree canopy and the ground cover in a forest.” Although drawn froma wide taxonomic range, many understory plants share a number of characteristicsassociated with shade tolerance, including longer foliation and moreefficient photosynthesis per unit leaf area (Spurr 1980). Of course, lightdemandingunderstory plants may be also present, but confined to open gaps orglades where light can penetrate to the forest floor.Since the plant composition varies from forest to forest and biome to biomewe would expect many different groups of insect to inhabit the understory. Forthe purposes of effective sampling, these can be divided into four categories: (a)insects associated with understory plants; (b) insects associated with the understoryenvironment (e.g. shade, constancy of microclimate, low wind speed); (c)gap specialists; and (d) resource specialists (e.g. dead wood, coppice, parasites).The understory makes a significant contribution to forest resources because itsupports a distinctive fauna; however, it cannot be totally isolated from otherforest habitats – neither the litter and soil layer below nor the high canopyabove. There are several examples of insects and other invertebrates whichmove between forest layers, interacting with the communities at several levels,e.g. Hymenoptera, Collembola, and Araneae (Oliveira & Campos 1996,Bowden et al. 1976, Simon 1995). This means that techniques and methodologiesdescribed in other chapters in this book may be applicable for investigationsof the understory.In this chapter I shall consider the process of collecting insects that are associatedwith forest understory vegetation (category a) and insects which areactively moving through the understory layer (categories a, b, c, and d).How should techniques be chosen?Collecting insects from vegetation usually has one of two major aims. The first isto generate a species list for the habitat, and the second to obtain subsets or sam-58

METHODS FOR FOREST UNDERSTORY VEGETATION 59ples of the community that are representative of the whole. The understory istaxonomically and structurally diverse and therefore it can be difficult to makedecisions about the number of sampling events, their location, and the techniquesto use. No one sampling technique will enable the entomologist to meetfully either of the two major aims noted above, and so it is likely that severaltechniques will be used in any one investigation. The key to choosing techniquesis a set of clearly defined research questions or study aims.The research questions should define the target insect groups or communities,the type of vegetation from which sampling will be needed, and maydictate the location from which samples should be collected. Each of these willsignal the most appropriate collection technique. The boundaries of the communityunder investigation need to be clearly delineated before a sampling protocolcan be set up and appropriate techniques chosen. For example, if theresearch question focuses on the insect community of a specific understoryplant the design demanded will be different from a study in which the researchquestion relates to the impact of edge effects or fragmentation.In this chapter techniques have been arranged according to the structure ofthe vegetation in which sampling is carried out. Within that, the insect groupsand communities that are most successfully collected with each technique arenoted, and location sensitivity is mentioned where appropriate. Understoryvegetation can range from low grass swards, through herbs of varying structuralcomplexity, to shrubs and small trees and the associated vascular and nonvascularepiphytes, so putting together a well-designed investigation can beboth challenging and exciting.Sampling from low understory vegetation including grassesand herbsSuction or vacuum samplingSuction or vacuum sampling can be used in understory vegetation types fromshort grass through to shrubs and small trees, but is particularly effective for collectinginsects in grasses and herbs. The technique is suited to collecting data oninsects associated with specific plant communities, resources, and locations,e.g. gaps and edges, and is not directly dependent on insect activity; thus it canbe described as a passive sampling method.Suction samplers may be divided into two major types according to the nozzleor hose diameter and common modes of use: wide-hosed (>20 cm diameter)and narrow-hosed (

60 CHAPTER 4Fig. 4.1 Narrow-hosed vacuum sampler (N-type), constructed from a converted garden leafvacuum. (Adapted from Sabre® Manual.)essentially function in the same manner, using a fan to generate a suction current.Air is drawn up a suction tube, one end of which is bisected by a collectingbag constructed of netting, muslin, or cotton (Fig. 4.1). The bag can be detachedfrom the tube and stored or emptied on site.The effectiveness of vacuum sampling is dependent on two main factors: thespeed of airflow and the proficiency with which the operator empties the bag(active insects can easily be lost at this stage!). Narrow-hosed models have beenshown to collect samples that are more representative of field communitiesthan wide-hosed models. For example, they are more effective in collecting insectgroups such as aphids and associated predatory beetles such as Tachyporussp. (Staphylinidae) (Macleod et al. 1994, Stewart & Wright 1995). This effectivenesshas been attributed to the greater airflow rate per unit nozzle area,

METHODS FOR FOREST UNDERSTORY VEGETATION 61and to better extraction of insects from the lower parts of the vegetation (Stewart& Wright 1995). Indeed, W-type models have been found to collect aphidspoorly from soil level and to favor insects in the top sections of plants (Hand1986). N-type models are therefore recommended. Additionally they are easierto manage in the field, being lighter to manipulate and much quieter.Sampling efficiency is also influenced marginally by the method of samplecollection and environmental conditions. The vacuum nozzle can be selectivelyplaced over individual plants or directly onto the vegetation sward; the formeris clearly easier with wide-hosed models. For N-type models a common mode ofuse is to lay down quadrats and run the nozzle slowly over and through the vegetation– e.g. 1 minute for a 25 cm 2 quadrat. A more sophisticated method usesa cylinder or tent with known cross-sectional area which can be placed on theground, the hose inserted through the top, and the contents vacuumed up.Fewer insects will be lost with this method (Stewart & Wright 1995).Species accumulation curves for vacuum sampling strongly link efficiency tosampling effort (Buffington & Redak 1998), so the number of sampling eventsneeds to be optimized. Rain or dew can adversely affect efficiency. Hendersonand Whittaker (1977) demonstrated that when vegetation is wet insects remainattached to the foliage; collection events should therefore be confined to dryconditions. Sampling will only be truly comparative if environmental conditionsand vegetation complexity are held constant.Whatever method is used it will be necessary to clean the samples (i.e. separateinsects from debris). This process can be very laborious if done by hand, butsome insects will cling to plant material and so large fragments need to bechecked carefully. Extraction of samples using Tullgren funnels (Sutherland1996) or CO 2and extraction using black fluorescent bulbs (Buffington & Redak1998) is less time-intensive and very effective. Should samples need to be storedin alcohol prior to cleaning to prevent predation loss, then insects can be extractedsubsequently using flotation techniques (Dondale et al. 1971).Although N-type vacuum samplers are able to collect insects from a widerange of taxonomic groups and produce samples typical of the community, theyare particularly suitable for insects with sufficient surface area for suction to acton, for example, medium to large Collembola and Homoptera (although alateaphids are favored over apterous morphs due to larger wing surface area for suction;Hand 1986). Large heavy species that the current cannot carry into the collectionbag will be under-sampled; for example, large aculeate Hymenoptera(Buffington & Redak 1998).How does vacuum sampling compare with other techniques? The other mostcommonly used technique in low vegetation is the sweep net (see below).Vacuum-sampling has been found more sensitive for detecting communityvariation than sweep-netting in complex vegetation such as Californian coastalsage scrub (Buffington & Redak 1998) and in cotton (Byerly et al. 1978, Ellingtonet al. 1984). Differences were particularly notable for animals with a smallbody size that may not be collected when sweeping because the air draught

62 CHAPTER 4pushes them away from the net. Buffington and Redak (1998) also foundvacuum-sampling to be more effective than sweeping for collecting Dipteraand smaller Hymenoptera.The action of a vacuum sampler is less likely to cause damage to sensitive vegetation,and where several operators are involved in a study it is more consistentthan methods such as sweeping, relying less on experience and good technique(Buffington & Redak 1998).Medium-height vegetation including shrubsSweep-net captureThe sweep net can be used in medium-height vegetation for collecting insectsassociated with specific plants or resources. Nets have been used extensively incrop pest surveys and in some studies for collecting insects in forest understoryand even the canopy (Canaday 1987, Noyes 1989, Lowman et al. 1993). Thesweep net is a passive sampling method that is suited to collecting insects associatedwith specific plant communities. (A sweep net could be used for insects associatedwith specific plant species if the volume of foliage is sufficient to sweepthrough.)Sweep nets are constructed from sturdy cotton material that can withstandvigorous movement though vegetation. The mouth of the net is usually circular,although D-shaped nets are more effective in short vegetation (Southwood& Henderson 2000). Net diameter usually approximates 0.5 m. Sweeping is essentiallya qualitative method but can be made semi-quantitative by standardizingthe number of sweeps in a given area or along a defined transect (25sweeps being effective; Gray & Treloar 1933), or by standardizing the length ofsweeping time (e.g. 5 or 10 minutes). The net should be plied in an energeticfigure-of-eight motion, with sufficient forward movement to prevent overlapof sweeps.The effectiveness of the sweep net is dependent upon the way in which it ismanipulated by the operator, the vegetation structure, and the prevailing environmentalconditions. For example, speed of net motion has an impact on catchsize, with a greater speed collecting more insects (Balogh & Loska 1956), andexperienced samplers may capture more insects because they sweep more vigorouslyand more deeply into the crop (Wise & Lamb 1998). Vegetation densityand crop phenology are also significant (Byerley et al. 1978, Wilson & Gutierrez1980, Ellington et al. 1984). Sweep nets are particularly effective if the insectsare known to be present in the top part of the vegetation (e.g. around seedpods), and in less dense vegetation, where insects can easily be knocked into thenet. If sweep nets are to be used for population estimation in complex vegetationthen calibration is required by means of a pilot study in which netting iscompared with some absolute counting method.

METHODS FOR FOREST UNDERSTORY VEGETATION 63As with vacuum sampling, nets do not collect insects from vegetation if thesurface is wet or damp, and so the timing of collection (e.g. after the evaporationof dew) is an important factor. However, sweep nets have the distinct advantagethat they can be used to collect larger numbers of samples than other methodssuch as interception traps, and they also allow samples to be collected in a randomizedmanner meeting some of the essential requirements of parametricstatistics. The exact number of samples and their location will depend uponthe variability of the community and the study aims.Sweep nets can collect samples that are completely representative of the populationor community of insects (Gadagkar et al. 1990). Wise and Lamb (1998)found that there was stability and repeatability in the relationship betweenvariance and mean, allowing sweep nets to be used to detect whether pestpopulations were above or below level of control. However, there can be biases,for example towards adults and large nymphs of Lygus species (probably due totheir location in the upper parts of the plant) (Wise & Lamb 1998). Linders(1995) found the sweep net to be effective for sampling the weevil Trichosirocalustroglodytes on Plantago lanceolata spikes, but the catches were influenced bydiurnal and seasonal activity. In contrast, sweep nets are not good for capturingLepidoptera (Gadagkar et al. 1990), but sample Hymenoptera, Diptera, smallColeoptera, and arachnids quite well (Canaday 1987).The major advantage of this technique lies in simplicity and portability, sincesweep nets can be easily carried and a large number of samples collected frommany locations. When compared to Malaise traps, sweep nets were found to bemore effective in collecting representative samples of forest Hymenoptera becausethey could be used in many parts of the habitat, although the traps werebetter in wet conditions (Noyes 1989).Malaise trapsMalaise traps are used to capture insects that are moving about above low vegetation.They have been used in studies of succession (Belshaw 1992) and can beused to detect the direction of movement of insects through a habitat. This typeof trap will be particularly effective at collecting insects associated with theunderstory environment (category b) and if located appropriately could alsocapture insects associated with specific resources, e.g. gaps and dead wood (categoriesc and d). Malaise traps will also collect insects associated with specificunderstory plants, provided these are actively flying through the habitat.Malaise traps are a form of flight interception trap. They function on thebasis that flying insects will not detect the vertical portion of the net andwhen they collide with it will cease flying and close their wings. Many insectsare phototactic (Wigglesworth 1972) and once they have alighted on a surfacewill move upwards towards the light. The traps make use of this behaviorby guiding upwardly moving insects by means of a sloping net roof into acollecting jar.

64 CHAPTER 4(a)(b)Fig. 4.2 Two Malaise trap designs with taught net walls and collection bottles set at thehighest point. (a) Cornell design (4-directional); (b) Townes model (bi-directional). AfterMatthews & Matthews (1983).The most common Malaise trap design has two opposing vertical net walls setin a T shape and supported by a pole and guy ropes. The mesh for these walls isusually black to reduce visibility. The walls are topped by an angled net roofwhich slopes upwards at the junction of the T (Fig. 4.2), the roof may be constructedof white or black mesh. At the highest point a removable collecting potis set which is filled with preservative fluid. In some designs the trap is baited, forexample with dry ice to attract biting flies (Strickler & Walker 1993).This type of flight interception trap was first designed by Malaise in 1937.There have been a number of subsequent variations in the design with twomain types, the four-directional Cornell design (Matthews & Matthews 1983)and the bi-directional Townes model (Townes 1962, 1972) (Fig. 4.2). In a comparativestudy, Matthews & Matthews (1983) found the Townes model to bemore effective partly because it captured more insects from higher up in the aircolumn. Alternative designs have been made to hang in low or high vegetation,constructed again from mesh but with four intersecting sheets and a non-meshroof and base tray (a composite interception trap; Basset 1985, Winchester& Scudder 1993, Springate & Basset 1996). These can be suspended frombranches or ropes within the canopy or across flight paths between trees.Location and climate can affect the efficiency of Malaise traps. Location is importantbecause insects often follow specific flight paths through vegetation(Hutcheson 1990, Matthews & Matthews 1983). For this reason traps are usuallyset along vegetation edges and at intersections. Some investigators suggesta north–south orientation with the head of the trap facing the sun’s zenith(Noyes 1989). Traps set in exposed and sheltered situations in the same habitattype may have different efficiencies (Noyes 1989), due partly to differences inflight behavior and partly to microclimate variation. However, where severaltraps are used they should not be located too closely together, or samples will nolonger be independent.Mesh size and color can affect the types of insects captured. For

METHODS FOR FOREST UNDERSTORY VEGETATION 65Hymenoptera, 300 holes per cm 2 are recommended, otherwise very small individualswill be lost (Noyes 1989), but it is more common to use a mesh of 100holes per cm 2 (Darling & Packer 1988). It has been found that green traps catchfewer tabanids than natural-colored mesh (Roberts 1970), but that the use of ablack base with white roof can increase the effectiveness of the trap (Matthews& Matthews 1983). Using a bicolored trap can alter the family balance of thecatch (Darling & Packer 1988); for example, coarse-mesh bicolor traps capturegreater numbers of aculeates and ichneumonids at the expense of small-bodiedHymenoptera. However, on balance, bicolored traps are recommended formost studies.Since Malaise traps are activity dependent, captures will be affected by climaticconditions, more specifically by the number of degree days available forflight and on seasonal variations in height of flight (Matthews & Matthews1983). Scheduling of sample collection therefore needs careful consideration.Malaise traps have the advantage that they are generally inexpensive, and canbe easily set up and left in secure locations. The greatest drawback, however, isthat the traps are easily subject to windthrow and are therefore difficult to use inexposed sites.Clearly Malaise traps will capture insects actively flying through the forestunderstory, including groups such as Hymenoptera, Diptera, Coleoptera, andLepidoptera. However, they will also capture to a lesser extent insects that arecarried passively by air currents and alight on the net, such as Psocoptera andCollembola. They have been used to survey Diptera including Tabanidae(Strickler & Walker 1993), and Disney et al. (1982) found that Calypterateswere very effectively collected. Hymenoptera are also successfully surveyedwith Malaise traps (Darling & Packer 1988, Noyes 1989). Disney et al. (1982)note that Malaise traps are exceptionally effective at capturing swarms of particularspecies, and indeed these can overwhelm a sample bottle, making it difficultto sort out rarely occurring species.Malaise traps only collect a portion of the whole understory community, butthey compare favorably with window traps (see below) for collecting insectssuch as Hymenoptera, probably because they retain more of the insects whichalight on them (Noyes 1989). Although climate-dependent, Malaise traps dofunction when the conditions are damp or wet and therefore have advantagesover vacuum-sampling and sweep-netting, where effectiveness is reduced inunderstory vegetation that is constantly wet (Noyes 1989).Window trapsWindow traps are similar to Malaise traps in principle, i.e. they are flight interceptiontraps. They are usually constructed from a vertical panel of Perspex ormesh with a drop tray below (often plastic piping with drain holes near the topto prevent flooding) (Peck & Davis 1980). They are particularly useful for collectingbeetles, which typically close their wings on encountering the wall and

66 CHAPTER 4therefore fall into the drop tray (Masner & Goulet 1981). Other weakly flyinginsects are caught, but some will be able to take off again and therefore will notbe trapped consistently. Impregnation of the trap with insecticide can increasethe capture rate of small insects such as microhymenoptera (Masner & Goulet1981). In the forest environment these traps are often used to collect insectsassociated with specific understory resources such as dead wood. They are particularlyuseful for surveying dead-wood beetles, a tremendously importantpart of the forest fauna.Color trapsThe response of insects to color, noted for Malaise traps, can be exploited toenhance the efficiency of collection. Color traps, e.g. sticky or water filled, havebeen used successfully to capture insects in temperate and tropical forest understoryas well as in heather and in structurally complex crops such as oilseed rape(Disney et al. 1982, Canaday 1987, Noyes 1989, Usher 1990, Burke & Goulet1998, Bowie 1999). These traps usually collect insects that are flying above orbetween plants within the understory, insects that may belong to category a(those associated with particular plants) or category d (resource specialists suchas predators and parasitoids). Since color traps are essentially active traps theywill be more selective than previously described techniques such as vacuumsampling.The principle is that insects are attracted to colors that mimic the spectral reflectanceof a habitat resource, e.g. flower, leaf, and stem colors. Predatory andparasitic insects are attracted either to the color of plants on which their preyitems feed or to the color of alternative food sources such as pollen-bearingflowers (Bowie 1999). Trap color therefore plays a significant role in the effectivenesswith which different insect groups are caught (Kirk 1984).Although color traps may be sticky or water-filled, here we concentrate onthe latter. Water traps are usually constructed from a colored tray, bowl, orbucket that is filled with water (some insect groups may be attracted by thewater itself; Noyes 1989). Holes drilled near the top will prevent overflow afterrain. The effectiveness of the trap is increased if detergent is added to reduce thesurface tension, and it may be necessary to add a preservative such as sodiumbenzoate or antifreeze, if traps are emptied infrequently. In some habitats andseasons it may be essential to service traps daily to keep up with the capturerates.Color traps will collect insects that are aerial either because they are activelyflying, or because they are part of the air plankton. There will be a higher probabilityof capture of insects that have control and therefore selectivity over landinglocation. Typical groups caught are the Diptera, Homoptera, Hymenoptera,and Coleoptera.Color traps may be set up in lines or grids and are commonly spaced at intervalsof a few meters. The choice of trap distribution is dependent on the research

METHODS FOR FOREST UNDERSTORY VEGETATION 67question. For example if the aim of the study is to determine species richness foran area then a large grid system may be appropriate, whereas the investigationof a habitat gradient may suggest a line of traps. However, it is clear that traplocation within the habitat can significantly affect the catch (e.g. edge effects;Bowie 1999), so this must be “factored out” if it is not of interest.Height of trap above or within vegetation has been found to influence thecatch for insects in rainforest understory, heather and oil seed rape and thereforeshould be standardized (Usher 1990, Bowie 1999). This could be done inthree ways, depending on the target insect groups: (i) standardizing to justabove the height of the understory so that visibility is consistent; (ii) standardizingto the mean height of the food resource (e.g. flower or fruit); (iii) standardizingto a defined height from the ground, perhaps determined as optimum in apilot study.Selecting trap color will again depend upon the research question becausethere is some variability in the responses of insect groups (Table 4.1).Colored water traps have been compared with other collection techniques. Infield crops they have been found to be as effective for adult syrphids as stickytraps set in several orientations and more effective for larvae when placed onthe ground (Bowie 1999). In woodland, white traps were found to be betterthan Malaise traps for capturing Syrphidae (Disney et al. 1982). Thus coloredwater traps are particularly recommended for this group of insects. In comparisonwith other collection techniques, however, colored traps require muchmore sampling effort and are therefore recommended for specialist monitoring(e.g. fauna associated with Gramineae or flowering plants) rather than for generalsurveys. They can be modified to increase their effectiveness for monitoringspecific insects either by sheltering (Finch 1992, Coon & Rinicks 1962) or byshape modification. Shape modification allows the traps to mimic plant resourcesin pattern as well as in color (e.g. yellow spheres on a contrasting backgroundto capture insects in citrus plantations; Cornelius et al. 1999).Color traps are one example of a bait trap. There is of course a plethora ofother baits that will attract insects, including pheromones, CO 2, plant or flowermimics, and there are examples of human bait being used to trap biting flies andmosquitoes (Costantini et al. 1998).Table 4.1 Preferences of insect groups for specific colored water traps.Trap colorWhiteYellowOrangeMore effective for Phoridae and non-cereal aphidsMore effective for Agromyzidae; cereal aphids; Hymenoptera, e.g.Chalcodoidea, CoccinellidaeHigher densities of ChironomidaeDisney et al. (1982), Noyes (1989), De Barro (1991), Parajulee and Slosser (2003).

68 CHAPTER 4Foliage baggingThe techniques described so far for medium-height vegetation have all been relativemethods of estimating population parameters, i.e. the captures are recordedrelative to trapping effort. These can be converted to absolute estimates bycalibration using actual counts of insects on plants or by using techniques thatproduce absolute estimates such as foliage bagging. Foliage or branch bagging isvery straightforward, but usually involves removing part of the vegetation, anactivity that may not be desirable at sensitive forest sites.Bagging involves drawing a net, or a cotton or plastic bag, over the foliagein such a way as to disturb as few insects as possible. In a survey of insects oncotton, Byerly et al. (1978) devised a cotton bag that could be introducedover a branch and drawn back to the branch base. The insects were then allowedto settle for 24 hours and the bag rapidly drawn up over the foliage, trapping ahigh proportion of the insect community. The branch is then clipped off. In somesurveys anaesthetizing chemicals are introduced into the bag to prevent loss ofspecimens when the bag is opened to extract the catch (Basset et al. 1997).The technique has the advantage of being absolute, and when carried out efficientlycan produce representative samples (Byerly et al. 1978). However,there are also a number of disadvantages. Where branches are long and bags donot stretch the whole length, samples are biased towards herbivores and towardsspecies that feed in actively growing tissue. Insect samples are dominatedby less active groups such as Collembola, Psocoptera, and sedentary larvae, e.g.Diptera (Blanton 1990).The branch bagging method has been modified successfully for use in tallvegetation and even tree canopies where it is used as an alternative to chemicalknockdown (Majer & Recher 1988, Schowalter 1995) (see Chapter 7).Tall vegetation including small treesUnderstory shrubs and trees are a structurally significant part of managed andunmanaged forest ecosystems. They are particularly prevalent in tropical biomes,but have been, and still remain, economically important in temperate regions(e.g. coppice with standards).It is possible and indeed effective to use chemical knockdown (Chapter 7) tosample from tall understory vegetation, provided the plants are screened offfrom the canopy above (Floren & Linsenmair 1998). Screening can be arrangedby constructing a tent above the target species to reduce chemical spread and toprevent insects from the high canopy falling into collection sheets. This techniqueis likely to produce the most comprehensive and representative samples,although of course erecting the tent will disturb mobile insects and so communitiesneed to be allowed time to settle before treatment begins.

METHODS FOR FOREST UNDERSTORY VEGETATION 69Other sampling techniques such as beating, branch clipping, and bagging willbe limited to the space that the investigator can reach from the ground unlessaccess systems such as ladders, ropes, and towers are used. Once the ground isleft behind, sampling the understory becomes similar to sampling the canopy(see Chapter 7).Beating traysShrubs and small trees are very difficult to sample quantitatively, and in the pastit has been common to use a beating tray. This is almost entirely a qualitativemethod to be used when constructing species lists or when collecting life-cycledata for particular species. However, some insects are so easily dislodged thatusing a beating tray can give a relative estimate of population densities (e.g.Geometrid larvae; White 1975).A tray is held underneath the shrub or small tree and the branches tappedsharply. This causes insects to fall from the branches onto the tray. Insects arethen collected up by pooter (aspirator) or fine brush, or funnelled into a collectingjar. The tray is usually constructed from cotton stretched onto a foldingwooden frame. White cotton is commonly used since the contrasting color allowseven small insects to be seen and removed. If the tray has a standardizedarea (e.g. 1 m 2 ) and the number of taps or hits is also standardized, it is possibleto make at least some comparisons between vegetation types.The main biases in this technique lie in the selectivity with which insects willfall from the vegetation, the strength of the beating action, and biases involvedin aspiration of samples. It is inevitable that larger insects will be preferentiallycollected from the sheet and more active ones may escape. If used in surveys offorest understory trees or shrubs then limiting the number of people involvedwill produced more standardized samples.Beating trays are often used to collect insects such as Chrysomelidae, Lepidoptera,and Heteroptera. In a comparison between beating trays (drop clothmethod), sweep nets, and absolute sampling on cotton for Lygus lineolaris, it wasfound that neither beating nor sweeping captured as many as the absolutemethod and both were adversely affected by plant height. The beating methodwas found to be more effective than the sweep net for estimating populationdensities but was more time-consuming (Snodgrass 1993).Herms et al. (1990) compared the use of a beating tray in honeylocust treesGleditsia triacanthos with vacuum-sampling. For the most abundant insects(Homoptera and Heteroptera) the beating tray was found to sample earlyinstars more effectively but vacuum sampling was better for adults. Differenceswere ascribed to the capacity of the beating action to dislodge the small insectsfrom unfolding leaflets. However they were able to use the vacuum samplerover a much larger area of the lower canopy of the trees and thereforesamples collected with this technique are likely to be more representative of thewhole community.

70 CHAPTER 4Vacuum sampling in shrubs and treesVacuum sampling can be used quite effectively to collect insects from shrubs andsmall trees. Three main methods can be employed: holding the nozzle over vegetation,e.g. slipping the hose over a branch; searching an area of the canopy for aset length of time (Herms et al. 1990); or vacuuming the foliage along a transector across a 180° arc (Buffington & Redak 1998). The most appropriate methodwill depend on the density and height of the vegetation, but in all cases thenarrow-hosed models will be much easier to use than the heavy wide-hosedtypes.Use of secondary characteristics in sampling from the understoryThe collection techniques discussed so far in this chapter have been absoluteand relative methods of sampling that rely on being able to detect, capture, andremove insects from the habitat. They are generally destructive methods. However,it is possible to make estimates of population densities of insects fromsecondary characteristics such as spider webs, leaf mines, and even insect frass(Sterling & Hambler 1988, Ozanne & Bell 2003). The use of a feature such as leafmines will give an accurate measure of population density without calibration,whereas the use of an indicator must be calibrated (Southwood & Henderson2000). For example, Thorpe and Ridgway (1994) were able to estimate the densitiesof gypsy moth Lymantria dispar in oak woodland by measuring the frassdrop per unit area and the yield of frass pellets per larva (Table 4.2). A further advantageof using this method to investigate insect populations is that estimatesof energy flow can also be made.ConclusionsIt is clear that when they are applied appropriately each of the techniques describedin this chapter can be used successfully to survey insect populations andto investigate community structure and dynamics. To determine which techniquesare the most appropriate (and in many studies several complementarymethods will be needed) some preliminary information should be gathered.Firstly, data on the structure and composition of the understory habitatshould be collected, since the vegetation and the insects that it supports willvary according to forest biome. For example, in tropical rainforests the understoryhas a number of structural layers below the high canopy, together with awide range of vascular epiphytes, whereas in temperate forests understory treesare scarcer, but structurally diverse shrubs and non-vascular epiphytes are common.In addition, the management of primary, secondary, or plantation foresthas a significant impact on the structure and composition of the understorycommunity. Information about the vegetation will help the investigator to con-

METHODS FOR FOREST UNDERSTORY VEGETATION 71Table 4.2 Calibration of frass output to larval density for Lymantria dispar (gypsy moth) inoak (Quercus spp.), following Thorpe and Ridgway (1994).FrasscollectionFrassproductionLarval densityFrass collection from canopy using 50 cm ¥ 50 cm polyethylene funnelsfor 12–16 hours from afternoon to morning, recorded as frass densityper m 2Frass yield per larva recorded for 40 larvae in cups provisioned with oakleavesMean density of larvae per tree, estimated using the equation:C = x d/x yC = 1/(area sampled by each frass sampling device)x d= mean drop (frass per trap)x y= mean yield (frass per larva)(Leibold & Elkinton 1988)Extrapolationto whole treedensitiesMeasure the perimeter of the dripline of each tree, calculate the areaand then multiply by larval densityCalculation carried out for fourth instar and again just before pupation.Frass drop: may be used as a relative population estimate.Larval density: an absolute population estimate but subject to greater error than frass drop.sider which techniques are likely to yield samples representative of the habitat.The second step is to gather some baseline information about the insect groupspresent in the understory and their distribution or the category (see Introduction)into which they best fit. These data can be obtained in a pilot study in whichseveral collecting methods are tested. Armed with this background informationit will be possible to develop clear experimental hypotheses that will direct thechoice of techniques.The understory is a rich environment and a very productive one for researchaiming to answer fundamental ecological questions or to investigate the impactof habitat manipulation and management. With the right tools we can learnmuch more about the role of this forest layer in ecosystem dynamics.ReferencesBalogh, J. & Loska, I. (1956) Untersuchungen über die Zoozönose des Luzernenfeldes. ActaZoologica Academiae Scientiarium Hungaricae, 2, 17–114.Basset, Y. (1985) Comparaison de quelques méthods de piégeage de la faune dendrobie.Bulletin Romand d’Entomologie 3, 1–14.Basset, Y., Springate, N.D., Aberlanc, H.P., & Delvare. G. (1997) A review of methods for samplingarthropods in tree canopies. In Canopy Arthropods (ed. N. Stork, J. Adis, & R. Didham),pp. 27–52. Chapman & Hall, London.

72 CHAPTER 4Belshaw, R. (1992) Tachinid (Diptera) assemblages in habitats of a secondary succession insouthern Britain. The Entomologist, 111, 151–161.Blanton, C.M. (1990) Canopy arthropod sampling: a comparison of collapsible bag and foggingmethods. Journal of Agricultural Entomology, 7, 41–50.Bowden, J., Haines, I.H., & Mercer, D. (1976) Climbing Collembola. Pedobiologia, 16, 298–312.Bowie, M.H. (1999) Effects of distance from field edge on aphidophagous insects in a wheatcrop and observations on trap design and placement. International Journal of Pest Management,45, 69–73.Buffington M.L. & Redak, R.A. (1998) A comparison of vacuum sampling versus sweepnettingfor arthropod biodiversity measurements in California coastal sage scrub. Journalof Insect Conservation, 2, 99–106.Burke, D. & Goulet, H. (1998) Landscape and area effects on beetle assemblages in Ontario.Ecography, 21, 472–479.Byerly, K.F., Gutierrez, A.P., Jones, R., & Luck, R.F. (1978) Comparison of sampling methodsfor some arthropod populations in cotton. Hilgardia, 46, 257–282.Canaday, C.L. (1987) Comparison of insect fauna captured in six different trap types in a Douglas-firforest. Canadian Entomologist, 119, 1101–1108.Coon, B.F. & Rinicks, H.B. (1962) Cereal aphid capture in yellow baffle trays. Journal ofEconomic Entomology, 55, 407–408.Cornelius M.L., Duan, J.J., & Messing, R.H. (1999) Visual stimuli and the response of femaleoriental fruit flies (Diptera: Tephritidae) to fruit-mimicking traps. Journal of Economic Entomology,92, 121–129.Costantini, C., Sagon, N.F., Sanogo, E., & Merzagora, L. (1998) Relationship to human bitingcollections and influence of light and bednet in CDC light-trap catches of west African malariavectors. Bulletin of Entomological Research, 88, 503–511.Darling, D.C. & Packer, L. (1988) Effectiveness of malaise traps in collecting Hymenoptera: theinfluence of trap design, mesh size and location. Canadian Entomologist, 120, 787–796.De Barro, P.J. (1991) Attractiveness of four colours of traps to cereal aphids (Hemiptera: Aphididae)in Southern Australia. Journal of the Australian Entomological Society, 30, 263–264.Dietrick, E.J. (1961) An improved backpack motor fan for suction sampling of insect populations.Journal of Economic Entomology, 54, 394–395.Disney, R.H.L., Erzinclioglu, Y.Z., Henshaw, D.J. de C., et al. (1982) Collecting methods andthe adequacy of attempted fauna surveys, with reference to the Diptera. Field Studies, 5,607–621.Dondale, C.D., Nicholls, C.F., Redner, J.H., Semple, R.B., & Turnbull, A.L. (1971) An improvedBerlese–Tullgren funnel and flotation separator for extracting grassland arthropods. CanadianEntomologist, 103, 1549–1552.Ellington, J., Kiser, K., Ferguson, G., & Cardenas, M. (1984) A comparison of sweepnet,absolute and Insectavac sampling methods in cotton ecosystems. Journal of Economic Entomology,77, 599–605.Finch, S. (1992) Improving the selectivity of water traps for monitoring populations of cabbageroot fly. Annals of Applied Biology, 120, 1–7.Floren, A. & Linsenmair, K.E. (1998) Diversity and recolonisation of arboreal Formicidae andColeoptera in a lowland rain forest in Sabah, Malaysia. Selbyana, 19, 155–161.Gadagkar, R., Chandrashekara, K., & Nair, P (1990) Insect species diversity in the tropics: samplingmethods and a case study. Journal of the Bombay Natural History Society, 87, 337–353.Gray, H. & Treloar, A. (1933) On the enumeration of insect populations by the method of netcollection. Ecology, 14, 356–367.Hand, S.C. (1986) The capture efficiency of the Dietrick vacuum insect net for aphids ongrasses and cereals. Annals of Applied Biology, 108, 233–241.

METHODS FOR FOREST UNDERSTORY VEGETATION 73Henderson, I.F. & Whittaker, T.M. (1977) The efficiency of an insect suction sampler in grassland.Ecological Entomology, 2, 57–60.Herms, D.A., Neilsen, D.G., & Davis Snydor, T. (1990) Comparison of two methods for samplingarboreal insect populations. Journal of Economic Entomology, 83, 869–874.Hutcheson, J. (1990) Characterization of terrestrial insect communities using quantified,Malaise-trapped Coleoptera. Ecological Entomology, 15, 143–151.Kirk, W.D.J. (1984) Ecologically selective coloured traps. Ecological Entomology, 9, 35–41.Lawrence, E. (ed.) (1995) Henderson’s Dictionary of Biological Terms. 11th edn. Longman,London.Liebhold, A.M. & Elkinton, J.S. (1988) Techniques for estimating the density of late-instargypsy moth, Lymantria dispar (Lepidoptera: Lymantriidae) populations using frass drop andfrass production measurements. Environmental Entomology, 17, 381–384.Linders, E.G.A. (1995) Biology of the weevil Trichosirocalus troglodytes and impact on its hostPlantago lanceolata. Acta Oecologia, 16, 703–718.Lowman, M., Moffet, M., & Rinker, H.B. (1993) A new technique for taxonomic and ecologicalsampling in rain forest canopies. Selbyana, 14, 75–79.Macleod, A., Wratten, S.D., & Harwood, R.W.J. (1994) The efficiency of a new lightweight suctionsampler for sampling aphids and their predators in arable land. Annals of Applied Biology,124, 11–17.Majer, J.D. & Recher, H.F. (1988) Invertebrate communities on Western Australian eucalypts:a comparison of branch clipping and chemical knockdown procedures. Australian Journal ofEcology, 13, 269–278.Malaise, R. (1937) A new insect trap. Entomologisk Tidskrift, 58, 148–160.Masner, L. & Goulet, H. (1981) A new model of flight interception trap for some hymenopterousinsects. Entomological News, 92, 199–202.Matthews, R.W. & Matthews, J.R. (1983) Malaise traps: the Townes model catches more insects.Contributions to the American Entomological Institute, 20, 428–432.Noyes, J.S. (1989) A study of five methods of sampling Hymenoptera (Insecta) in atropical rainforest, with special reference to the Parasitica. Journal of Natural History, 23,285–298.Oliveira, M.L. & Campos, L.A.O. (1996) Preferencia por estratos florestais e por substanciasodoriferas em abelhas Euglossinae (Hymenoptera, Apidae). Revista Brasileira de Zoologia, 13,1075–1085.Ozanne, C.M.P. & Bell J.R. (2003) Collecting arthropods and arthropod remains for primatestudies. In Field and Laboratory Methods in Primatology: a Practical Guide (ed. J. Setchell & D.Curtis), pp 214–227. Cambridge University Press, Cambridge.Parajulee, M.N. & Slosser, J.E. (2003) Potential of yellow sticky traps for lady beetle survey incotton. Journal of Economic Entomology, 96, 239–245.Peck, S.B. & Davis, A.E. (1980) Collecting small beetles with large-area “window traps”.Coleopterists’ Bulletin, 34, 237–239Roberts, R.H. (1970) Color of malaise trap and the collection of Tabanidae. Mosquito News, 30,567–571.Schowalter, T.D. (1995) Canopy invertebrate community response to disturbance and theconsequences of herbivory in temperate and tropical forests. Selbyana, 16, 41–48.Simon, U. (1995) Untersuching der Stratozönosen von Spinnen und Weberknechten (Arach.:Araneae, Opilionida) und der Waldkiefer (Pinus sylvestris L.). Diss.FB. UmweltundPlanungswissenchaften, TU Berlin. Wissenschaft und Technick Verlag, Berlin.Snodgrass, G. (1993) Estimating absolute density of nymphs of Lygus lineolaris (Heteroptera:Miridae) in cotton using drop cloth and sweep-net sampling methods. Journal of EconomicEntomology, 86, 1116–1123.

74 CHAPTER 4Southwood, T.R.E. & Henderson, P.A. (2000) Ecological Methods. 3rd edn. Blackwell Science,Oxford.Springate, N.D. & Basset, Y. (1996) Diel activity of arboreal arthropods associated with PapuaNew Guinea trees. Journal of Natural History, 30, 101–112.Spurr, S.H. (1980) Forest Ecology. 3rd edn. Wiley, New York.Sterling, P.H. & Hambler, C. (1988) Coppicing for conservation: do hazel communities benefit?In Woodland Conservation and Research in the Clay Veil of Oxfordshire and Buckinghamshire (ed. K.Kirby & F.J. Wright), pp. 69–80. Research and Survey in Nature Conservation 15. NCC, Peterborough.Stewart, A.J.A. & Wright, A.F. (1995) A new inexpensive suction apparatus for samplingarthropods in grassland. Ecological Entomology, 20, 98–102.Strickler, J.D. & Walker, E.D. (1993) Seasonal abundance and species diversity of adult Tabanidae(Diptera) at Lake Lansing Park-North, Michigan. Great Lakes Entomologist, 26, 107–112.Sutherland, W.J. (1996) Ecological Census Techniques: a Handbook. Cambridge University Press,Cambridge.Thornhill, E.W. (1978) A motorised insect sampler. Pest Articles and News Summaries, 24,205–207.Thorpe, K.W. & Ridgway, R.L. (1994) Effects of trunk barriers on larval gypsy moth (Lepidoptera:Lymantriidae) density in isolated- and contiguous-canopy oak trees. EnvironmentalEntomology, 23, 832–836.Townes, H. (1962) Design for a Malaise trap. Proceedings of the Entomological Society of Washington,64, 253–262.Townes, H. (1972) A light-weight Malaise trap. Entomological News, 83, 239–247.Usher, M.B. (1990) Assessment for conservation values: the use of water traps to assess thearthropod communities of heather moorland. Biological Conservation, 53, 191–198.White, T.C.R. (1975) A quantitative method of beating for sampling larvae of Selidosema suavis(Lepidoptera; Geometridae) in plantations in New Zealand. Canadian Entomologist, 107,403–412.Wigglesworth, V.B. (1972) The Principles of Insect Physiology. 7th edn. Cambridge UniversityPress, Cambridge.Wilson, L.T. & Gutierrez, A.P. (1980) Within-plant distribution of predators on cotton: commentson sampling and predator efficiencies. Hilgardia, 48, 1–11.Winchester, N.N. & Scudder, G.G.E. (1993) Methodology for Sampling Terrestrial Arthropods in B.C.Resources Inventory Committee, B.C. Ministry of Lands and Parks, pp 1–32.Wise, I.L. & Lamb, R.J. (1998) Sampling plant bugs, Lygus spp. (Heteroptera: Miridae), inCanola to make control decisions. The Canadian Entomologist, 130, 837–851.

METHODS FOR FOREST UNDERSTORY VEGETATION 75Index of methods and approachesMethodology Topics addressed CommentsSuction or vacuum Questions of association with Samples a wide range ofsampling specific plant communities, taxonomic groups. Particularlyresources and locations,effective for aphids andparticularly in habitats such as associated predatory beetlesshort grass through to shrubs such as Tachyporus sppand small trees.(Staphylinidae), medium toAbsolute estimates oflarge Collembola andpopulation density.Community structure data.Homoptera. More effective thansweeping for collecting Dipteraand smaller Hymenoptera.Large heavy species are undersampled,e.g. large aculeateHymenoptera. W-type modelshave been found to collectaphids poorly from soil leveland to favor insects in the topsections of plants.Sweep-net capture Questions of association with Samples insects that are knownwhole plant communities or to be present in the top part ofspecific species if there isthe vegetation, e.g. around seedsufficient foliage volume.pod.Relative estimates of population Hymenoptera, Diptera, smalldensity.Coleoptera, and arachnids arequite well sampled.Lepidoptera are under-sampled.Malaise traps Studies of successional change. Small HymenopteraDetection of the direction of Hymenoptera, Dipteramovement of insects through a (especially Tabanidae andhabitat.Calypterates), Coleoptera,Questions of association with Lepidoptera.the understory environment.Association with specificresources if appropriatelylocated, e.g. gaps and deadwood.Association with specificunderstory plants if insects areflighted.Relative estimates of populationdensity and species richness.Continued

76 CHAPTER 4Methodology Topics addressed CommentsWindow traps Association with specific Coleoptera, e.g. dead-woodunderstory resources such as beetles.dead wood.Relative estimates of density Increased capture rate of smalland species richness.insects such asMicrohymenopteraColor traps Association with particular Diptera, Homoptera,plants.Hymenoptera, Coleoptera.Resource specialization, e.g. Brown traps: capturepredators and parasitoids. significantly higher densities ofRelative estimates of population Chironomidae.density and species richness. Yellow traps: more effective forAgromyzidae and cereal aphids;Hymenoptera such as theChalcidoidea; Coccinellidae.White traps: for Phoridae, somespecies of Syrphidae, noncerealaphids.Foliage bagging Absolute estimates of Collembola, Psocoptera, andpopulation density and species sedentary larvae, e.g. Diptera.richness per unit area and perunit plant material.Association with specific plantspecies and with specific plantparts.Beating tray Qualitative sampling, relative Chrysomelidae, Lepidoptera,population estimates, presence Homoptera, Heteroptera,absence data, populationparticularly early instars.structure data.Secondary Relative or absolute estimates Lepidoptera, Coleoptera.characteristics of population density.

CHAPTER 5Sampling insects from trees:shoots, stems, and trunksMARTIN R. SPEIGHTIntroductionMany insect species can be found living on the outside of twigs, shoots, and bark(Speight & Wylie 2001). So aphids, scale insects, booklice, and thrips are relativelyeasy to find and identify. However, unlike foliage, where the majority ofinsects live on the outside of the plant, a large proportion of insects which feedon the shoots, stems, and trunks of trees are to be found inside the plant tissue.Thus borers, tunnelers, and gallers tend to remain hidden for much of their lifecycles, and in many cases this concealed habit renders them very difficult toeven detect, never mind count. Very often, their presence inside shoots, underbark, or in the timber is only advertised by the effects they have on the hostplant. Dead or deformed shoots, holes in bark, or the exudation of frass andresin, are important indicators of the insect within. Another problem often confrontsthe entomologist even when the insects are found: many active borersare in the larval stage, and their taxonomy, without access to the adult specimen,can be difficult or even impossible. Finally, with such a huge size range ofbreeding sites and food items, from the smallest twigs to the thickest trunks,both from living trees and also from dead or moribund ones, the sampling proceduresemployed for these types of insects are many and varied indeed.Table 5.1 summarizes these main habitats, and introduces some of the importantinsect groups to be found in each one.DetectionInsects such as aphids, scales, and mealybugs provide relatively obvious indicationsof their presence, either as the individual insects themselves, or in the waxor so-called “wool” that they produce. Some, such as horse chestnut scalePulvinaria regalis, beech scale Cryptococcus fagi, and pine woolly aphid Pineus piniare detectable from some distance, since in large densities they coat the stems ortrunks of their host trees with white exudations under which they live and/orlay their eggs. However, because so many habitats mentioned in Table 5.1 areinternal to the host plant, it may not be at all obvious in many cases that insectsare present, and one of the first problems facing more rigorous sampling77

78 CHAPTER 5Table 5.1 Major stem habitats on trees and their associated insect groups.Habitat Insect activity Insect examplesShoots/stems Sap feeding Scale insects (Hemiptera: Coccidae)Woolly aphids (Hemiptera: Adelgidae)Aphids (Hemiptera: Aphididae)BoringBark beetles (Coleoptera: Scolytidae)Moth larvae (Lepidoptera: Pyralidae)Longhorn beetles (Coleoptera: Cerambycidae)GirdlingChewingLonghorn beetles (Coleoptera: Cerambycidae)Longhorn beetles (Coleoptera: Cerambycidae)Grasshoppers & crickets (Orthoptera: Acrididae &Gryllidae)Moth larvae (Lepidoptera: various)Bark surface Sap feeding Scale insects (Hemiptera: Coccidae)Woolly aphids (Hemiptera: Adelgidae)Aphids (Hemiptera: Aphididae)DetritusfeedersBook lice (Psocoptera)Bark interior Borers Moth larvae (Lepidoptera: Pyralidae)Bark/sapwood Borers Bark beetles (Coleoptera: Scolytidae)interfaceLonghorn beetles (Coleoptera: Cerambycidae)FungivoresFungus flies (Diptera: Mycetophilidae)Ambrosia beetles (Coleoptera: Platypodidae)Timber Borers “Woodworm” beetles (Coleoptera: Anobiidae)Moth larvae (Lepidoptera: Cossidae & Hepialidae)Longhorn beetles (Coleoptera: Cerambycidae)Jewel beetles (Coleoptera: Buprestidae)Powder post beetles (Coleoptera: Lyctidae)procedures is deciding whether or not a tree contains anything to sample in thefirst place. Numerous indications can be observed in forest stands which betraythe existence of insect activity within the plant tissues, and Table 5.2 provides asummary of some of the most obvious.Sampling methodsThe sampling methods described in this chapter are arranged basically by thetype of habitat or part of the tree where they are normally found (see Table 5.1).Many types of sampling tactic are described, predominantly via the use of examplessourced from the published literature.

SAMPLING INSECTS FROM TREES 79Table 5.2 Examples of evidence of insect activity concealed within plant tissues, eithercurrently or in the past.Part of plant Evidence Causal agentTwigs & Dead or discolored foliage, general Boring by moth larvae or beetle adultshoots tree dieback e.g. Dioryctria cristata (pine shoot moth),Tomicus piniperda (pine shoot beetle)Bent, twisted, or deformed shoot, Boring by moth larvaefoliage still greene.g. Rhyacionia buoliana (Europeanpine shoot moth)Broken or decayed shoot, often Boring by moth larvaewith orange-brown sap and resin e.g. Hypsipyla grandella (mahoganyexudationsshoot borer)Periodic swellings along thin stems Boring by beetle larvaeor twigse.g. Saperda populnea (poplar longhorn)Swellings or deformed buds (galls) Gall wasps, gall woolly aphidse.g. Andricus sppExternal bark Tan to brown fine granules or dust Entrance holes of bark beetles inin bark crevicesmoribund treese.g. Ips spp, Scolytus sppHeavy sap or resin exudation Entrance holes of bark beetles inrelatively vigorous treese.g. Dendroctonus micans (Spruce barkbeetle)White to cream powder or dust on Entrance holes of ambrosia beetlesbark or at base of treee.g. Trypodendron sppSmall circular holes in bark Exit holes of bark or ambrosia beetlese.g. Scolytus spp, Trypodendron sppLarge circular holes in bark Exit holes of woodwaspse.g. Sirex sppMedium to large oval holes in bark Exit holes of longhorn beetlese.g. Phoracantha sppMedium to large roughly circular Bore holes of larvae of goat or woodholes in bark, accompanied by mothsoozing resin and wood debris e.g. Cossus cossusSwellings running spirally around Tunnels of larvae of varicose borerstrunke.g. Agrilus sexdentatus (Buprestidae)Dry earthern tunnels running Termitesmainly up and down trunks and e.g. Coptotermes sppstemsInternal bark/ Engravings with many branches Tunnels of bark beetlessapwood from a central gallery e.g. Tomicus, Scolytus, Ips sppsurface Shallow but wide engravings, Tunnels of longhorn or roundheadoften containing compacted dust beetlesor frasse.g. Tetropium sppContd p. 80

80 CHAPTER 5Table 5.2 (contd)Part of plant Evidence Causal agentDeep oval holes in sapwood Pupation tunnels of longhorn beetlese.g. Phymatodes sppShallow oval holes in sapwood, Pupation chambers of weevilsoften surrounded by wood fibers e.g. Pissodes sppLarge circular holes in sapwood Exit holes of woodwaspse.g. Sirex sppSmall circular holes in sapwood, Entrance/exit holes of powder-postno blue/black stain, exuding white beetlesduste.g. Lyctus sppSmall circular holes in sapwood, Entrance/exit holes of wood wormno blue/black stain, no white dust e.g. Anobium sppSmall circular holes in sapwood, Entrance holes of ambrosia beetlessurrounded by blue or black stain e.g. Trypodendron spp, Platypus sppHeartwood or Large random tunnels, often with Wood-boring termite galleriesinside main smooth, sculptured surface texture e.g. Coptotermes spptrunk One or two wide tunnels, usually Larval tunnels of goat or wood mothscircular in cross sectione.g. Cossus spp, Xyleutes sppSmall tunnels or chambers, Larval/pupal chambers of ambrosiasurrounded by blue/black staining beetlesin wood tissuee.g. Trypodendron spp, Platypus sppShoots and twigs: externalGallsAs with any other part of a tree, sampling of twigs and shoots involves a threedimensionalarena, wherein in order to obtain data representative of the wholetree, the distribution of insect populations within the whole tree has to be considered.This is especially problematic when counts from mature trees are required,since many species of insect are not uniformly distributed throughoutthe entire height of the host plant. For example, the spruce woolly aphid Adelgesabietis (Hemiptera: Adelgidae), shows a clumped distribution within crowns ofeven small trees (Fig. 5.1) (Fidgen et al. 1994). Most adelgid galls occur on lateralshoots of mid-crown branches, so sampling insects at the very top or on thelowest whorls of branches will underestimate overall populations. Very often,pilot surveys, sampling all the of the tree crown, will reveal such dilemmas, sothat subsequent, more detailed, assessments can be directed at the regions of thetree supporting highest population densities, and reliable sampling units can bederived. The main snag with this is of course that whole-canopy sampling maywell be impossible due to the sheer size of the habitat involved.

SAMPLING INSECTS FROM TREES 813025Trees 1–2 m tallTrees 3–4 m tallMean % shoots galled201510501 3 5 7Whorl (1 = lowest)Fig. 5.1 Mean percentage (± s.e.) of shoots of white spruce galled by the adelgid Adelgesabietis, according to tree height and branch whorl. From Fidgen et al. (1994).Galls on shoots, twigs, and indeed foliage are relatively easy to sample, sincethey are large, recognizable, and easy to identify to species from some distance.The insects inside the galls may not be so tractable in terms of abundance or taxonomy,but the effects which they produce on the host plant, such as abnormalshoot or leaf development, leave unmistakable traces. In the case of Adelgesabietis, once trees in a stand have been selected according to the purpose of theinvestigation, and the region of the tree where most galls can be found is determined,the proportion of current year shoots per whorl with one or more gallson them is simple to assess by visual counts (Fidgen et al. 1994). However, it isusually impossible to examine all the shoots or twigs on a tree, even a relativelysmall one, so that sample units have to be established which will provide reliablerepresentations of the whole tree, or, indeed, forest stand. In the case of thegall wasp Andricus sp (Hymenoptera: Cynipidae), which stimulates gall formationon oaks in Arizona, USA, it was found that shoot diameter was related tothe probability of attack (Pires & Price 2000). Therefore, it was possible in thisstudy to relate the probability of attack by gall wasps to the total number ofshoots in diameter categories. In general, the thicker shoots supported moregalls than thinner ones, and in fact pilot studies showed that sampling shootswith a diameter of less than 1.5 mm provided underestimates of the overallinfestations. So Pires and Price (2000) measured at least 100 shoots growingin the current year, and the number of galls per year counted. In order toprovide whole-tree estimates of gall abundance, the total number of shoots fortrees taller than 4 m was estimated, based on the number of branches multipliedby the number of estimated shoots per branch. The probability of attack was

82 CHAPTER 5calculated in relation to the number of sampled shoots in each size (diameter)class.The incidence of galls themselves does not necessarily provide estimates ofinsect abundances since in many species of gall-inducing insects, several individualsmay inhabit one gall. However, in some cases at least, it is possible to estimatethe overall insect population by measuring the size (volume) of thegall — bigger galls contain more insects. Figure 5.2 shows an example fromSouth Australia, which involves the gall wasp Mesostoa kerri (Hymenoptera:Braconidae). This insect causes stem galls on Banksia marginata, and in order toinvestigate the population density of wasps in each gall, Austin and Dangerfield(1998) cut terminal branches supporting fresh (i.e. no exit holes) galls from anumber of trees. Because galls vary in shape and size, the volume of each onewas measured in the laboratory by water displacement in a measuring cylinder,having first removed any attached leaves or twigs. Each gall was then carefullydissected under a stereo-microscope, and adult insects, larvae, or pupae foundwithin the gall chambers counted and preserved. As Figure 5.2 shows, therewas a highly significant relationship between gall volume and number of gallwasps contained within. Thus in subsequent sampling routines, a simple measureof gall numbers combined with gall volume would be able to provide accurateestimates of gall wasp population density. Furthermore, it is possible toleave each gall intact so that the numbers of natural enemies versus gall waspscould be assessed by allowing all insects within a gall to emerge naturally, withoutdisturbance.1412Volume of gall (cm 3 )10864200 50 100 150 200 250 300Number of gall wasps per gallFig. 5.2 Relationship between the size of galls, measured by water displacement, and thenumber of gall wasps Mesostoa kerri inside. From Austin & Dangerfield (1998).

SAMPLING INSECTS FROM TREES 83External feedersMany insects which live externally on stems and shoots can be collected using“traditional” methods such as beating trays (Wearing & Attfield 2002) orcanopy knockdown using insecticidal fogs or sprays (Stork et al. 2001, Speightet al. 2003). However, insects which feed directly on twigs and shoots, such asscales, aphids, or psyllids (all Hemiptera) are usually much more difficult tosample quantitatively, because of their very small size, cryptic appearance,and/or extremely patchy distribution. Very often, microscopic examination isrequired to count adult (and especially juvenile) populations, so that destructivesampling is frequently required. One basic problem is removing tiny insectsfrom the twigs or shoots on which they reside. As long as they do not stick tootightly to the stems, as in the cases of aphids or psyllids, they may be washed offshoots or stems using soapy water. The resulting liquid can then be strainedthrough wire screens onto muslin filters, from where the numbers of insects ofvarious life stages can be counted under a low-powered microscope. This systemof population assessment can work very well. Geiger and Gutierrez (2000)used water-washing as a “rough and ready” tactic for assessing the populationdensity of the leucaena psyllid Heteropsylla cubana (Hemiptera: Psyllidae), andcompared the procedure with a pilot study where labor-intensive absolutecounts where carried out. Water-washing provided very accurate assessmentsof real densities, with the relationships between the two measurements showingr-squared values from 0.88 to 0.98, and regression slopes near to unity.Some species of twig and shoot feeders, though small themselves, produceeasily recognizable signs of their presence. Woolly aphids (Adelgidae) andcertain true aphids (Aphididae) produce waxy secretions under which theylive and reproduce, and these secretions provide visual clues about infestationlevels and distributions within tree canopies. In addition, some species causedeformation to shoots via feeding damage, which can also be observed withoutdestructive sampling. For example, the balsam twig aphid Mindarus abietinusfeeds on the buds and elongating shoots of balsam fir Abies balsamea in the USA,and these aphids form noticeable aggregations covered with powdery wax andhoneydew (Kleintjes 1997). As described above for another species, mid-crownbranch tips are known to provide reliable estimates of whole-tree infestationlevels. In surveys, 20 host trees were selected randomly from the middle of theplot (to avoid edge effects), and two 25 cm branch tips per tree from the midcrownregion were visually assessed for wax and honeydew, and also the numberof distorted shoots per total number of shoots was counted. In this way, grosslevels of insect density could be assessed, but in order to relate these general observationsto actual numbers of insects, microscopic examination on clippedtwigs was required. At the same time, it then proved possible to assess the numbersof predators such as hoverflies and ladybirds associated with the aphids.The above sampling system may sound simple enough, but obtaining acceptablyaccurate population densities of insects such as woolly aphids by counting

84 CHAPTER 5individuals can be difficult because of their very small size and high density. Thehemlock woolly aphid Adelges tsugae (Hemiptera: Adelgidae), for example, isless than 1 mm long, and can occur in densities over 9 per cm of twig length(Gray et al. 1998). An additional problem involves the frequently highlyclumped distributions on host tree twigs. It is important therefore to employsampling systems which cut down the amount of painstaking and tricky directcounting. Indirect estimates of insect density are desirable, especially for broadscalepopulation studies, and some researchers have used a so-called binomialsampling plan which uses data collected from two categories. According to Grayet al. (1998), the precision and accuracy of estimated insect density from binomialsampling is dependent on the precision and accuracy of equations that describethe relationship between the proportion of samples (in this case twigs)with at least a predetermined insect threshold, and the mean insect density. Ifsuch an equation can be found to be reliable, then sampling effort can be muchreduced. Likely problems with this type of technique center on the possible inconsistenciesover wide geographical areas, and changing insect distributionswith life cycle and season.Gray et al. (1998) sampled 16 to 20 hemlocks in each of several sites over severalgenerations of the hemlock woolly aphid, selecting trees which wouldmaximize the range of woolly aphid densities. Four equally spaced branch tipsaround 30 cm long were cut from each of lower and upper tree crowns, and keptcool and humid before they could be examined under a microscope and theadelgids on each twig counted. A twig in this case was defined as the portion ofbranch extending from a terminal bud to the first node below the terminal.Finally, the relationship between the mean number of adelgids per twig and theproportion of twigs infested with a certain minimum number of adelgids wasderived using empirical or theoretical distributions, one example of which isshown in Fig. 5.3.The empirical model shown in the figure is as follows:[ ]( P T )lnx= lna + bln -ln1 -(5.1)where x = mean adelgid density, a and b are regression estimates, and P is theproportion of twigs infested with at least T individuals.In this case, the empirical model only works well for population estimateswithin a generation, but it means that a rapid assessment of twigs in a sample forthe proportion infested with, say, a minimum of three insects provides a reliableestimate of mean numbers overall at that point in time. Hence, as the figureshows, for the combined summer and winter generations from two sites, if 30percent of twigs in the samples had at least three adelgids on them (P T= 0.3), themodel predicts that there will be a mean of 0.66 insects per twig in the wholesample.Scale insects (Hemiptera: Coccidae) can present even more problems thanadelgids and aphids. They are also very small, and many are extremely cryptic

SAMPLING INSECTS FROM TREES 8532ln (mean nos.)10–1–2–3–2.5 –2 –1.5 –1 –0.5 0 0.5 1 1.5Proportion (P ) of twigs infested with 3 or more adelgids (ln[–ln(1 – P T=3 )])Fig. 5.3 Relationship between mean hemlock woolly adelgid density and proportion of twigsinfested with three or more insects, from the empirical distribution model, for threesuccessive generations combined. From Gray et al. (1998).and hard to recognize, especially in the young stages. An additional problem involvesthe high mobility of first-instar nymphs (the crawlers), which usuallyprovide the only really dispersive stage of the insect. On a large tree, eggs,crawlers, and later life stages may be patchily distributed on twigs and shoots,with access problems to the high canopy. Clearly, the larger the number of samplesthat can be taken from all parts of the tree canopy, the more representativethe overall mean abundance on twig or shoot may be, but a compromise has tobe reached wherein sampling effort is matched with the reliability of the results.Sedentary scales are relatively easily sampled, nevertheless, using clipping toolssuch as long-armed pruners, and their population densities assessed using amicroscope as described above. Crawlers, however, require a static trapping system;citriola scale Coccus pseudomagnoliarum, which feeds on the shrub Chinesehackberry, provides an example. Dreistadt (1996) used double-sided transparentsticky tape to trap scale crawlers on the twigs of the host tree. Each twig wasencircled with the tape, so that an accurate length of twig was covered, based onthe width of the tape employed (in this case, approximately 13 mm). Thesemini-sticky traps were replaced weekly at the same spot on the branches eachtime during the active season for the crawlers, though as the crawler stage declined,the tape could be left in position for longer. Dispersing crawlers stuck tothe tape, which could be removed and counted under a fairly high-poweredmicroscope (up to 30¥ magnification). Though in this published example thenumbers caught per trap were not corrected for trap (i.e. twig) diameter, itwould make between-branch and between-tree comparisons more reliable ifthe surface area of the sticky trap was measured and the number of crawlers

86 CHAPTER 5300Mean number of crawlers per trap250200150100500April May June JulyFig. 5.4 Mean (±s.e.) density of citriola scale crawlers caught on sticky-tape traps onhackberry trees in California. From Dreistadt (1996).caught on it corrected to a unit area measure, such as density per cm 2 of trap. Inthis way, seasonal peaks and troughs of scale crawler (and other similarly smallbut mobile insect) populations can be recorded, as exemplified by Fig. 5.4.Shoots and twigs: internalInsects which inhabit shoots and twigs can be grouped under the general title of“borer.” They may be adults, as in the case of some bark beetles (Coleoptera:Scolytidae), larvae, as with some longhorn beetles (Coleoptera: Cerambycidae)and many moth larvae (Lepidoptera: Pyralidae and Tortricidae), or both adultsand larvae may be found in the same shoot, as with some weevils (Coleoptera:Curculionidae). Whatever the insect involved, most will cause the shoot or twigin which they are living to die, deform or otherwise show fairly clear symptomsthat all is not well and attacks are taking place. These symptoms vary with insectspecies and activity, but many can be easily distinguished one from another.A very common example from Europe and elsewhere involves the pine shootbeetle Tomicus piniperda (Coleoptera: Scolytidae). The larvae of this species liveand grow, as with most Scolytidae, under the bark of moribund standing treesand logs (see below), but the young adults have to undergo a process of maturationfeeding before they can mature their eggs or sperm (Speight & Wainhouse1989). This they accomplish by tunneling into the terminal and leader shoots ofhealthy pine trees. No breeding takes place, merely adult feeding, but the effecton the host tree’s shoots can be serious indeed. Damaged shoots eventually fall

SAMPLING INSECTS FROM TREES 87Mean number of fallen shoots (m 2 )140120100806040200Stand age 0–20 yearsStand age 81–100 years200 400 600 800 1000 1200 1400 1600 1800Distance from yard (m)Fig. 5.5 Number of pine shoots (±s.d.) pruned by pine shoot beetles Tomicus piniperda,according to distance from a timber yard. From Långström & Hellqvist (1990).off the trees, especially in high winds, and the easiest way of assessing shoot beetlepopulations in tree canopies is to count the number of fallen shoots on theground below. On the assumption that all damaged shoots in a given area of foresthave an equal chance of falling to the ground, sample plots can be establishedwherein all dead shoots are recognized and counted. During studies onTomicus in Sweden, Långström and Hellqvist (1990) set up marker poles alongtransects leading away from a timber yard (ideal breeding sites for bark beetles),and all fallen pine shoots were collected within 5 m 2 plots centered on the poles.Any dead shoots not killed by bark beetles are easily recognized, lacking as theydo the telltale hollow tunnel up the center of beetle-attacked shoots. In thisexample, the authors were able to show clear declines in attacks to trees as thedistance from the timber yard increased (Fig. 5.5). Any trees within a rough500 m radius of the yard were clearly at most risk from attack.However, it may be necessary to carry out more detailed sampling of T.piniperda populations and the damage done to standing trees by shoot-boring.In this case the luxury of forest floor sampling is unavailable, and more subtleways of estimating damage categories may be required. Kauffman et al. (1998)used damage categories based on visual assessments of the continuing activityof adult bark beetles in pine shoots, again from the ground, but this time usingwhole-tree studies for young pines, or canopy examination using binoculars orladders for bigger trees. Again, the actual insect was not directly observed at thisstage of the sampling. Instead, the structure and especially color of the foliageon shoots was used to categorize damage levels. Changes from dark green,through yellow, to brown indicated increasing intensities of attacks by Tomicus.In order to assess the numbers of beetles emerging from damaged shoots, on

88 CHAPTER 5their way to lay eggs in logs or stressed trees, these authors had to resort to baggingwithin pine canopies, followed by destructive sampling. Clumps of three tofive shoots in various damaged categories were left attached to the trees, butplaced in nylon mesh bags which were sealed around the stems at the bottom sothat any beetles emerging from the shoots were unable to escape. The baggedshoots were later cut from the trees and dissected, providing information on (i)the number of shoots with T. piniperda tunnels, (ii) the number of dead and aliveT. piniperda adults in each tunnel, and (iii) the number of dead and alive T.piniperda in the bag outside the shoot. Using these procedures, detailed populationestimates of young bark beetles could be obtained, and combined with thedamage done to the host pine trees.Unlike most bark beetles, some species of weevil spend their larval and pupalstages within leading shoots of trees. The damage is the same — dead shoots andresultant dieback and deformation, especially of young trees. The white pineweevil Pissodes strobi (Coleoptera: Curculionidae) is an extremely common pestof many native and introduced pine and spruce species in North America. Matureadult female weevils puncture feeding pits and lay eggs in the revealedcambium of the top half of the previous year’s leader of the host tree. The resultinglarvae feed downward beneath the bark, eventually girdling and killingthe shoot. Fully grown larvae then pupate in the shoot, and new adultsre-emerge from the now dead leaders in summer (Nealis 1998). In order tostudy the populations dynamics of P. strobi, it is necessary to assess populationdensities within leading shoots not just of the weevil itself, but also of naturalenemies such as predators and parasitoids that may also reside within theshoots. According to Nealis (1998), the leader itself is thus a meaningful andconvenient sample unit, containing in a single neat “package” most of theagents impacting on weevil survival. In this example, sampling consisted of cuttingthe entire infested leader at the base of the previous year’s growth at regularintervals during the growing season, taking care to spread the destructiveload around even-aged tree stands as much as possible, whilst avoiding rows oftrees at plantation edges. A census of all leaders in the study area gave the frequencyof infested trees in each plantation for the given year. On each samplingoccasion, half the collected leaders were dissected immediately and all insectswithin recorded and counted. The other half were kept intact under controlledconditions (20 °C and 16 : 8 hours light to dark) in 1 m long PVC tubes with plasticcollecting funnels at the open bottom end. Any emerging weevils or other insectsfell into these funnels and could be preserved and subsequently identifiedand counted. It is important to note that some insect species may take a considerabletime to emerge from wood samples such as these, especially as in this casewhere parasitic Hymenoptera required a winter diapause before becomingadult. The maintenance of cut samples of trees may be required for manymonths to ensure all borer individuals have emerged. Shoot-boring Lepidopteraare similar to beetles with the same habits, in that they too usually leave

SAMPLING INSECTS FROM TREES 89fairly obvious evidence of their presence. The family Tortricidae contains manyspecies which attack leading and lateral shoots of trees, again causing shootdeath, tree distortion and considerable dieback. Examples include the notorioustemperate and tropical pine shoot moths Rhyacionia spp and the mainlyNorth American pine shoot borers Eucosma spp. As with the beetles, assessmentsof attack rates of these insects are simple enough to carry out; systematicwalking through young pine stands looking at each tree will show the obvioussigns of bent, stunted, brown, or dead leading shoots (Speight & Wylie 2001).Assessing shoot borer damage visually in this manner is easy, quick, nondestructive,and conservative (because some damage may be overlooked)(Prueitt & Ross 1998). In large plantations, where every tree cannot be examinedindividually, sample grids can be established to sub-sample each standin a routine manner. Thus, in an example from Oregon, USA, Prueitt andRoss (1998) sampled a total of 120 trees per plot for damage caused by Eucosmasonomana, by examining five trees in a row heading north, followed by five treesin the next row heading south, and so on until the full 120 had been studied.More detailed sampling however may be required on occasion, for examplewhen the intensity of shoot attack per tree is of interest. Here, individual whorlsfrom the tops of trees downwards can be sampled, and the levels of shoot borerinfestation scored. Clearly, this system will only work well when the tops of thetrees are within easy reach of the sampling team.Some lepidopteran borers are easiest to assess not by the actual damage theycause to shoots (though this may on occasion provide long-range visual cluesabout the insect’s whereabouts), but by the frass and resin exudations whichtheir activities inside the infested shoots cause (Jactel et al. 2002). Perhapsthe best example of this is the infamous mahogany shoot borer Hypsipyla spp(Lepidoptera: Pyralidae). This pest occurs wherever mahoganies of many generaand species are grown, from Central and South America, through Africa andSouth Asia, to Southeast Asia and Australia (Speight & Wylie 2001). Adult femaleslay a single egg on or near terminal shoots and the young larvae tunnelinto healthy shoots, eventually killing them. The larva’s entrance hole exudessap, resin, boring debris, and frass (feces), and it is possible to assess the progressof insect development by the nature of these exudations, as Table 5.3 shows(Howard 1995). Using such criteria, the stage of development of the insect in aparticular tree can be readily assessed without having to see the larva at all. Becausethis insect is a typical “low-density” pest, where only one larva per youngtree is required to render it worthless for a final timber product, sampling usuallytakes the form of assessments on a stand scale. For example, in field studiescarried out by Newton et al. (1998) in Costa Rica, routine (once a month) inspectionsof each tree in young stands of two different mahogany species provided,simply and efficiently, cumulative data on the percentage of treesattacked (Fig. 5.6). Sampling the actually larvae within the infested shoots wasnot, on this occasion, required.

90 CHAPTER 5Table 5.3 Stages in the development of resin and frass exudations on leading shoots ofmahogany Swietenia mahagoni caused by the mahogany shoot borer Hypsipyla grandella.From Howard (1995).Characteristics of exudationInsect development stage3 mm diameter ball of cream-colored frass Early-instar larva in first stages of tunneling inand resinleading shoot15 to 20 mm frass aggregation, ranging Mature larva, having excavated a tunnel up to 5from pale-wood color to orange or light or even 10 cm long in leaderbrownDarkened, dissipated, sticky to dry frassaggregation, dark red or brownBoring ceased, larva either died, pupated, oradult moth emerged9080% trees in stand attacked706050403020CedrelaSwietenia1000 10 20 30 40 50 60 70 80 90Time (weeks after first assessment)Fig. 5.6 Cumulative number of two mahogany species attacked by Hypsipyla grandella inCosta Rica. From Newton et al. (1998).External bark surfaceCertain species of forest insect spend some or all of their life cycle on the outsideof the main trunks of trees. Longhorn beetles (Coleoptera: Cerambycidae) layeggs on the bark surface and the resulting larvae tunnel beneath the bark. Someadult scale insects such as the horse chestnut scale Pulvinaria regalis (Hemiptera:Coccidae) also oviposit on main trunks and major branches, whilst other scales,for example the beech scale Cryptococcus fagisuga (Hemiptera: Coccoidea) spendall but the active crawler stage on main stem bark.

SAMPLING INSECTS FROM TREES 91CountingSampling of these populations can involve similar techniques to those describedabove for insects on the outside of twigs and shoots, but extra problems arise becauseof the extensive nature of main stems, especially on mature trees. Manyexternal bark insects are not evenly distributed over the whole trunk of trees,and very often tend to be concentrated in certain regions. Figure 5.7 gives an exampleof the vertical distribution of the maritime pine scale Matsucoccus feytaudi(Hemiptera: Margarodidae). The relative tree height in the figure is an indicationof the location on the main trunk as a percentage of the total trunk height.Clearly, scales are most abundant in the middle section of the trunk (Jactel et al.1996). In addition, pine scale, as with so many externally located insects, arevery sensitive to bark texture. As Fig. 5.8 shows, scales seem to avoid very thickand very thin bark, but prefer instead thickish bark with many flakes or smallcracks. This type of surface provides them with shelter, anchorage, and access tothe sap of the bark cambium on which they feed. Any sampling regime for thistype of insect which concentrates on very thick or very thin bark, or at the topsor bottoms of trunks, will seriously underestimate scale population density,whilst concentrating on thickish bark in the middle section of tree trunks willnot be representative of the whole trunk. Once again, pilot studies should becarried out to decide on the optimal regions of the trunk for detailed sampling.Some scale insects are readily visible on the outside of the bark, as in the caseof the horse chestnut scale Pulvinaria regalis. The large adult females lay thousandsof eggs in white waxy secretions on the main stems of lime, sycamore, andhorse chestnut trees in urban areas of western Europe. The adult herself diesMean scale frequency (%)201816141210864205 to 10 10 to 15 15 to 20 20 to 25 25 to 30 30 to 35 35 to 40 40 to 45 45 to 50 50 to 55 55 to 60 60 to 65Relative tree height (%)Fig. 5.7 Vertical distribution of the within-tree populations of second-instar pine scaleMatsucoccus feytaudi. From Jactel et al. (1996).

92 CHAPTER 5Thin bark, few small flakesThin bark, many small flakesThick bark, few small cracks, many flakesThick bark, few vertical cracks, many small cracksVery thick bark, deep vertical cracksVery thick, rough bark, very deep vertical cracks0 1 2 3 4 5Mean number of scales per logFig. 5.8 Distribution of second-instar pine scale Matsucoccus feytaudi in relation to bark type ofhost tree. From Jactel et al. (1996).leaving a very obvious brown body over the wax, providing a readily visible,countable unit. Systematic quadrat sampling using 0.01 m 2 wire quadratsplaced vertically on tree trunks, using ladders for access, was used by Speight(1994) and each female plus egg mass counted within the quadrat. In an initialcalibration exercise, densities of eggs in each egg mass was assessed by microscopecounting after removing the wax with a solvent. Subsequently, the size(mean of shortest and longest body dimension) of a full range of small to largedead adults was related to egg numbers per mass, so that in field sampling themean density of eggs laid per unit bark area could be assessed by estimating themean size of dead females per quadrat.Because scales such as Pulvinaria are so obvious as adults once the eggmasseshave been deposited, widespread surveys of infestation levels can becarried out using visual assessments of whole-tree populations. Speight et al.(1998) developed an “eye-ball” scoring system for Pulvinaria egg-masses ontown trees, where the estimated percentage cover of trunk and main brancheswas split into intervals of 10 percent, each being given a “score” from zero to 10.Thus, trees with no egg-masses in evidence at all received a score of zero, whilsttrees with the bark completely covered by egg-masses were awarded a score of10. In order to standardize the sampling assessments, teams of surveyors weretrained in a pilot study to ensure that one person’s estimate of 30 percent barkcoverage (score of 3) was the same as any other’s. In the final analysis, these visualassessments of bark insect abundance are only truly of use if they can be relatedto actual insect density, especially those of the feeding life stages. In thecase of Pulvinaria, Speight et al. (1998) were able to relate the egg-mass scoresper tree to the mean density of scale insect nymphs feeding on the leaves of the

SAMPLING INSECTS FROM TREES 93Mean square root number of nymphs onleaves per cm 21.41.21.00.80.60.40.20r 2 = 0.7896p < 0.001(± s.e.)0 2 4 6 8 10Egg-mass scoreFig. 5.9 Relationship between egg-mass score on trunks and nymphs per cm 2 of leaf surfacefor horse chestnut scale Pulvinaria regalis. Data from Speight et al. (1998).same tree (Fig. 5.9). In this way, the large-scale visual surveys could be used toinfer important information for the population densities of feeding herbivores.Other scale species are much smaller than Pulvinaria, and though they leaveevidence of their presence via waxy “wool” as before, it is much more difficult tosample insect lifestages accurately. A case in point is the beech scale Cryptococcusfagisuga (= fagi). This non-armored scale feeds and grows on the bark of mainstems of its host trees, and the whole life cycle, bar the dispersing first instars,takes place under the “wool” produced by the insects. As in the previous example,visual estimates of total bark coverage are useful for assessments on a foreststandlevel, and they can be taken down to fairly small-scale measurements. ForCryptococcus, both Gora et al. (1994) and Lunderstädt (1998) recorded scale densityusing a five-level scale, from 1 = very sporadically dispersed wax wool“points” indicating the location of scale colonies, up to 5 = wool “points” coveringlarge areas of bark. Thus it was possible to determine the infestation level ofeach beech tree sampled, measured as the annual mean value of the monthlyobserved scale density, divided into four classes. These were (a) no or very slightinfestation (mean value

94 CHAPTER 5microscope, the waxy coverings of each scale colony were dissected, and classifiedas follows: (a) established first instars (with wax secretion); (b) secondinstars; (c) non-fecund adults; (d) fecund adults with eggs either in their bodiesor laid; and (e) total number of eggs or hatched first instars. It was possible to calibratethe visual scoring technique with actual insect density, such that therewas a positive linear relationship between mean cover score and mean numberof adult scales on the bark, providing a validation of the much easier and lesstime-consuming technique when compared with microscopic examination.Insects such as longhorn (cerambycid) beetles spend little time on the barksurface, being bark cambium or wood borers as larvae, but they lay their eggs onexternal bark, and very often this is the only easy part of the life stage for sampling,since the rest is concealed below the bark, making population density assessmentmuch more difficult (see below). One of the most widespread andpotentially devastating pests of eucalyptus trees all over the tropical and subtropicalworld is the eucalyptus longhorn beetle Phoracantha semipunctata(Speight & Wylie 2001). Adult female beetles lay their eggs in bark cracks andcrevices on trees that are in some way stressed (drought struck, dying, or felled,for example). The simplest way to sample cerambycid eggs laid on bark in thismanner is to visually search for them on the bark of cut and sectioned eucalyptuslogs, making sure to examine bark cracks and flakes (Way et al. 1992). As inother examples described above, the position on the tree trunks of peak ovipositionis not random, but varies with height up the trunk, and also bark texture(Fig. 5.10). From these results it is clear that future intensive sampling could berestricted to the lowest 3 m of trunk in order to study the highest egg densities.3500Number of eggs per 3 m length of trunk300025002000150010005000Loose barkTight bark0–3 m 4–6 m 7–9 m 10–12 m 13–15 mPosition on trunk measured from baseFig. 5.10 Numbers of eggs laid per tree by Phoracantha semipunctata in different ovipositionsites on cut trunks of eucalyptus trees. From Way et al. (1992).

SAMPLING INSECTS FROM TREES 95CapturingInsects moving around on the surface of tree bark, either emerging from withinthe bark or timber (see below) or migrating up or down the trunks, can be capturedusing a variety of methods and their population densities assessed. Onoccasion, it has been possible to employ modern technology in the form of12-volt battery-operated vacuum-cleaners to suck wandering insects off thebark (Jantti et al. 2001), but the time-honored trapping methods are still morewidely used. Effectively, there are two basic trap types for this purpose, stickytraps and collecting traps. In the former, a band of sticky material is placedaround a tree trunk or limb, and insects that wander onto the surface areretained. In the latter, a physical barrier of some sort attached to the bark guidesthe insects into a collecting funnel or pot from which they cannot escape.Sticky bands have been used for many years by horticulturalists to preventpestiferous insects such as moths and weevils from climbing up trees havingemerged from the soil or litter to infest leaves, shoots, or fruits. The technologyis basically very simple, and though the actual sticky substance may be a commerciallyavailable chemical compound such as Tanglefoot® or Hyvis®, simplegrease may suffice for short periods.One example of the use of sticky bands to monitor insect populations movingup tree trunks concerns the Bruce spanworm Operophtera brucerata(Lepidoptera: Geometridae), a close relative of the European winter moth O.brumata, in Quebec, Canada (Hébert & St-Antoine 1999). Like the winter moth,Bruce spanworm adult females are flightless, and after emerging from pupae inthe soil they crawl up trunks of various deciduous tree species to lay eggs in barkcrevices for the winter. In this example, the authors used 15 cm wide strips ofdifferent materials coated with Tanglefoot wrapped tightly around tree trunksat about 1.3 m above the ground. Adult moths stuck to the bands were counted,and considerable numbers of insects were caught (Fig. 5.11). Notice howeverthat unless the bands are changed regularly, in this case once a week, mediumtermmonitoring may underestimate population densities, since there is a limitto the number of individuals that can stick to a band trap, after which it becomessaturated with dead and dying moths and the remainder escape capture. Incidentally,if egg numbers are required, it is possible to use artificial ovipositionsubstrates, again attached to bark, on which insects lay their eggs. In the studyby Hébert and St-Antoine, the success of this technique depends on the type ofsubstrate. As can be seen from the figure, polyurethane foam is most successful.Sticky bands have other practical problems in addition to saturation. The applicationis messy and time-consuming (especially in cold weather when thesticky glue is hard), and the insects caught are effectively unusable for anythingelse. Any cracks or crevices in the bark under the bands allow insects to bypassthe trap and continue up the tree (Webb et al. 1995).Collecting pots or traps of various designs are more efficient than sticky bandsin the main, though they do involve more expense and setting up. They include

96 CHAPTER 5Sticky bands unchangedaSticky bands changed weeklybPolyurethane foamAPolyethylene foamDacronBBFemalesEggs0 100 200 300 400 500 600 700 800Mean eggs or females per 1000 cm 2 (± s.e.)Fig. 5.11 Densities of winter moth eggs and female moths on different artificial substratesand sticky bands, respectively, placed at breast height around the trunks of sugar maple trees.From Hébert & St-Antoine (1999).the so-called lobster-pot traps, circle traps, or trunk window traps. One of theoriginal “lobster-pot” type traps was designed and used by George Gradwell andGeorge Varley in their classic long-term studies on winter moth in WythamWood, Oxfordshire (Agassiz & Gradwell 1977). The trap was made of thin nylonmaterial (actually derived from ladies’ stockings!) with a gaping open endpointing downwards held on a wire framework attached securely to the bark.Insects crawling up the bark entered the trap, ascended the nylon funnel, andentered the capture chamber at the top through a small raised opening throughwhich they could not return. Since the width of bark covered by the gape wasknown, the numbers of winter moth moving up trees per unit surface of treetrunk could be counted over given time periods.These days, nylon stockings have given way to metal gauze and plastic, butthe principle remains the same, and one commercially available trap, the circletrap, is used very extensively, especially in the USA, to monitor numbers of pestssuch as the plum curculio Curculio caryae and the pecan weevil Conotrachelusnenuphar (Coleoptera: Curculionidae) (Great Lakes IPM 2000). So-called crawltraps have also been constructed out of modified inverted metal funnels sealedto the bark to catch pine weevils such as Hylobius pales and scolytids such as Ipsgrandicollis (Hanula et al. 2002).More sophisticated “lobster-pot” type traps are now available for collectingtree-trunk insects. One such trap was first designed and used by Moeed andMeads in 1983, and two versions were employed, one to catch insects movingup tree trunks (the “up-trap”), and the other to intercept those moving downtrunks (the “down-trap”). Construction details can be obtained from the

SAMPLING INSECTS FROM TREES 976000Total number of individuals50004000300020001000Up-trapDown-trap0Collembola Orthoptera Hemiptera Coleoptera Diptera HymenopteraFig. 5.12 Numbers of individuals within common hexapod/insect groups caught over an 18-month period in trunk traps attached to the bark of four tree species in New Zealand. FromMoeed & Meads (1983).publication; essentially arthropods crawling up or down tree trunks come intocontact with the mesh “girdle” around the tree and are channeled into the collectingjar. As can be seen from Fig. 5.12, large numbers of individuals can becaught with this system. It is significant that the majority are moving up thetree, with smaller numbers moving down.A final type of trunk trap is the trunk window trap, which has been used tocollect insects emerging from the bark or wood of standing trees, and also fromfungal polypores growing on tree trunks. A vertical pane of clear Perspex ismounted at 90 degrees to the tree trunk, with a collecting funnel and bottlebelow. Insects attempting to walk over the vertical pane, or indeed those that flyinto it having just taken off from the bark surface, fall into the mesh funnel andare then collected in the jar below where they are preserved (Kaila et al. 1997).Trunk window traps have been used by these authors to compare the populationsof saproxylic Coleoptera in boreal forests in Finland. Figure 5.13 illustratessome of their results. Using detrended correspondence analysis (DCA), it wasclear that beetle populations from dead trees left standing in clear-cut forestswere distinctly different from those from the same habitats in mature, uncut,forests. Forest management has important consequences for dead-wood insects,and the conservation status of such material cannot be overemphasized.Indeed, some authors recommend that trunk window traps, whilst not collectingall insect species equally, are simple and effective tools for samplingsaproxylic beetles in the tropics as well as in temperate regions (Grove 2000).Without considerable modification, trunk traps are not suitable for samplingarthropods walking or crawling on horizontal branches. Instead, it is possible to

DCA 298 CHAPTER 53210–1–2–3–3 –2 –1 0 12Fig. 5.13 Detrended correspondence analysis (DCA) of beetle samples from trunk windowtraps in clear-cut (open symbols) and mature (closed symbols) forests in Finland. Trianglesand squares denote two sites of each type. From Kaila et al. (1997).DCA 1use a form of pitfall trap, a capture system used the world over for sampling mobileanimals on the forest floor. Koponen et al. (1997) designed an arboreal pitfalltrap which they employed to sample arthropods moving on large horizontalbranches of old oak trees in Finland. The trap consists of a plastic collar constructedfrom a water pipe, coated with the non-stick substance, Fluon®, whichis fitted tightly around the branch. Animals encountering the collar fall into a

SAMPLING INSECTS FROM TREES 99plastic funnel and thus into a collecting bottle below. Since the funnel is substantiallybelow the branch, rainwater tends not to fill the bottle; drain holesprevent the trap becoming waterlogged in very bad weather. In their studyusing this trap, Koponen et al. caught a total of 32,938 arthropods from sevenstudy sites with five traps per site, over a 4–5 week interval.Bark/sapwood interfaceSampling insects living inside trees, especially main branches and trunks, can bevery problematic, and almost always involves some destruction. If the trees arealready dead, and the bark can be removed easily, then insects within can befairly readily collected and counted. The difficulty is of course that this type ofsampling will destroy the habitats for these insects, and if, as so many speciesare, they are rare or endangered, then the very sampling procedure may makematters worse.Sampling techniques for insects that spend at least part of their life cycles(usually larvae and pupae) beneath tree bark can be arranged into severalcategories, including hand/eye searching, trap-logging, log dissection, bark removal,emergence trapping or caging, and externally trapping for flying adults.Each will be considered in turn.Hand-searchingDirect searching is probably the most effective sampling method when the aimis to find as many scarce insect (and other animal) species under bark or in timberas possible within a short time (Siitonen & Martikainen 1994). Adult individualsmay be found walking about on the surface of logs, dead trees, andstumps, and such material is usually an irresistible magnet for entomologists!Loose bark can be pulled off, and adults, larvae, and pupae may be collectedwith ease. Experienced people may on occasion be able to rear young stages incontrolled environments, though once removed from their natural habitatsbark-dwelling insect pupae, and larvae especially, are notoriously difficult tocare for until adulthood. As with many other examples of this type of exploration,the intensity of sampling, normally in these cases equated to the areaof bark searched, can be related to the success of discovering new species.Figure 5.14 shows how species accumulation curves can be produced for barkbeetles (Coleoptera: Scolytidae) in old pine stumps in British Columbia, Canada(Safranyik et al. 1999). The curve only begins to level out as bark surfacesampled exceeds 10,000 cm 2 (1 m 2 ). Clearly, perfunctory, rapid hand-searchingwill not reveal the presence of all species in a timber sample.Old, dead trees soon loose their tightly attached bark, revealing evidence ofinsect infestations in the past in the form of tunnel engravings and patterns onthe sapwood surface. Though the actual insects responsible have long gone,

100 CHAPTER 57Mean number of species collected65432100 2000 4000 6000 8000 10,000 12,000Cumulative bark area sampled (cm 2 )Fig. 5.14 Mean species accumulation curves of bark beetles attacking pine stumps accordingto area of bark sampled. From Safranyik et al. (1999).numbers, sizes, and distributions can still be sampled. Furthermore, since barkbeetles for example produce species-specific gallery patterns, taxonomic workcan also be accomplished by examining such material. Macías-Sámano andBorden (2000) investigated the interactions of two scolytid species in grand firin Canada, by examining whole trunks of fallen fir trees from which the barkhad sloughed off naturally. A string marked off in meter intervals was pinnedalong the entire length of the tree trunks, and at each 1 m point along the string,the numbers of galleries of each beetle species were counted in order to studyinterspecific competitive interactions.Trap-loggingMost species of bark- or wood-inhabiting insects are attracted to stressed, moribund,or dead timber; healthy trees are normally defended against borer attackby chemical and/or physical means (Speight et al. 1999). Hence, if logs are leftout in a forest in a susceptible state, ripe for colonization, then adult borers fromthe surrounding habitats can be expected to oviposit in this material, eventuallyproducing new generations of themselves which can then be collected andcounted. In this way, for example, it may be possible to census populations ofinsects in a forest which are otherwise very difficult to find or identify. This canbe especially problematic in species-rich areas such as tropical rainforests.Tavakilian et al. (1997) set out to investigate the host–plant relationships of

SAMPLING INSECTS FROM TREES 101tropical longhorn beetles (Coleoptera: Cerambycidae) in French Guiana. Theonly way to provide incontrovertible evidence of host–plant associations is torear an adult insect from an accurately identified host plant. In this example, theauthors felled 690 trees and lianas in the dry season, when it was thought mostlonghorn beetles would be active. Rare tree species were favored in the hopethat poorly known beetle species would be encountered. The felled trees wereleft on the forest floor for about four months, during which time naturally occurringlonghorn beetles had the opportunity to lay their eggs in the moribundmaterial. After this time, each log was examined for evidence of insect attack,such as oviposition scars or frass exudations. Logs with such evidence were cutinto 80 cm lengths, and placed in cages (see also below). The cages were monitoredtwice per week, and emerging insects collected, preserved, and identifiedwhere possible. As a result of all this effort, around 350 cerambycid species werecollected, 90 of which were undescribed.Bark removal and log dissectionIn situations where the larvae or pupae of bark borers need to be sampled directlyand quantitatively, there is usually no recourse but to take the tree or logapart, or at least to remove sections of its bark. A whole mature tree is impossiblein practical terms to sample in this way, so it is important before intensivesampling begins to establish where in the trunk is the highest likelihood of findingthe target insects. Bark-boring insect species are not uniformly distributedover the whole trunk of trees, but instead tend to congregate in certain areas, relatedto stem girth, height above ground, and especially bark thickness. Put simply,small species of bark beetles for instance tend to be found under thinnerbark, as compared with large species that need the extra space for their tunnelsand galleries that thick bark provides. Figure 5.15 provides an example of threespecies of bark beetle (Coleoptera: Scolytidae) all of which have larval stages ingalleries under the bark of damaged spruce trees. As can be seen, the smallXylechinus is normally restricted to the tops of trees where the bark is thinner,and the other, larger species cannot grow to maturity (Jakuš 1998). Obviouslythere would be no point in sampling this species in areas of trunk from near thebottom of mature trees. These data were collected in a fairly standard fashionthat involves a lot of unavoidable destruction. Spruce tree trunks first had theirbranches removed, and were then divided into 2 m long sections (generallycalled billets). From the middle of each billet, a 50 cm long strip of bark waspeeled, with the width of the bark strip being equal to the girth of the trunk atthe peeling position. This strip was then divided into 4 equal quarters, each ofwhich was the sampling unit. Beneath each sampling unit, bark beetle gallerieswere identified (scolytids have very distinct species-specific gallery patterns)and broods (assemblages of larvae and pupae) counted. The density of bark beetleattack was then calculated by dividing the number of brood systems by thearea of the sample unit, in this case measured in dm 2 .

Pupae per gallery102 CHAPTER 56Density of attack(brood systems per dm 2 )54321Hylurgops palliatusXylechinus pilosusXyloterus lineatus00 5 10 15 20 25 30Distance from top of tree (m)Fig. 5.15 Attack density distribution of three species of bark beetle on bark from snapped oruprooted spruce trees according to position on trunk. From Jakuš (1998).Gallery length (cm)1412108642Gallery length;r 2 = 0.91Pupae per gallery;r 2 = 0.995045403530252015105000 5 10 15 20 25 30 35 40Attacks per 0.1 m 2Fig. 5.16 Relationships between attacks of Ips cembrae and (a) gallery length and (b) pupaeper gallery (fitted curves). From Zhang et al. (1992).Table 5.4 suggests general rules for sampling of beetles under bark, which involvesectioning the main trunk, removing the bark from selected parts of eachsection, and sampling bark beetle populations by counting individuals andmeasuring the dimensions of their galleries or brood chambers. In this way, forexample, Zhang et al. (1992) were able to demonstrate competitive interactionswithin one beetle population as shown in Fig. 5.16. As can be seen, as attack

Table 5.4 Examples from the literature of methods of sampling scolytids under bark.Beetle species Common name Tree species Material sampled Bark removal technique Population measurements Reference& countryIps grandicollis Five-spined Pine, Billets 35 cm long All bark stripped from each Numbers of beetles, alive or dead, Lawsonengraver beetle Australia by 12–20 cm billet counted in each life stage, plus 1993diameter enemy species. Attack densityassessed as egg galleries per100 cm 2 , egg gallery length (cm)per 100 cm 2 bark area in 10replicates of 4 billets eachIps typographus Spruce bark Spruce, Standing trees Circular samples hammered Numbers of individuals of each Gonzalezbeetle Belgium from bark with sharpened development stage. Longitudinal et al. 1996metal tubing, 1 dm 2 samples taken at 4 cardinalcross-section positions around barkcircumference, at 1 m intervals.Circular samples taken at 2 to4m intervals, all aroundcircumference (numbers variedaccording to trunk diameter)Ips cembrae Larch bark Larch, Felled, fire- One rectangular bark Attack density (attacks per Zhang et al.beetle China damaged trees sample (20 ¥ 50 cm) 0.1 m 2 ), number of egg galleries 1992stripped per tree at breast per attack, egg gallery length,height number of egg niches, and pupalchambers per galleryIps grandicollis Five-spined Pine, Billets 70 cm long From 4 equal quadrants of Ranked categories (1 = maternal Stone &engraver Australia by c. 15 cm billet, 7 beetle entrance holes gallery only, no eggs Æ 6 = most Simpsonbeetle diameter — 2 per selected at random, and a 4 ¥ new adults emerged) 1990felled tree 4 cm piece of bark chiseled outIps spp Spruce bark Spruce, Snapped or 50 cm strip of bark peeled Brood systems for each beetle JakušHylurgops spp beetles Slovakia uprooted trees from middle of each section. species counted, position on 1998Pityogenes spp cut into 2 m long Bark strips divided into 4 trunk recorded. Density of beetlePolygraphus spp sections quarters (upper and lower, left attack = number of beetle broodXyloterus spp and right quarters of trunk) – systems divided by area ofeach quarter a sample unit sample, in dm 2

104 CHAPTER 5frequency or density increased, so the gallery length decreased, as did the numberof larvae successfully reaching the pupal stage, indicative of increasing competitionfor limited resources (space and food).Entrance or emergence hole samplingIf data on bark beetle survival and emergence success are required, it may not benecessary to laboriously remove bark; the numbers of holes in the bark madeeither by ovipositing females entering the bark or, more frequently, by newadults leaving their pupal chambers may be all that is needed. Cut logs orbranches are the usual sample section as before. Entrance holes can be difficultto spot, but assessing their density has been used to investigate the relationshipsbetween tree stress and bark beetle attack (Kelsey & Gladwin 2001). For emergencestudies, the cut ends of the logs should be covered with paraffin wax orplastic to avoid the logs drying out, and this material can then be stored undercontrolled environment conditions (or on occasion simply left in the forest),until all adult emergence is thought to be over. In the situations where only onespecies of scolytid is known to be inhabiting the billets, their uniform exit holesare easily recognized and counted. Any different size or shape holes in the barkare likely to be caused by other insects, such as bark beetle parasitoids. For example,Lozano et al. (1997) investigated the interactions between the olive barkbeetle Phloeotribus scarabaeoides and its wasp parasitoid Cheiropachus quadram(Hymenoptera: Pteromalidae) in southern Spain. The emergence holes in olivebark made by the two insects were readily distinguishable, by virtue of their differentsizes (the parasitoid holes being significantly smaller in diameter). Cutlogs measuring 40–60 cm long and 4–8 cm in diameter were placed in an olivegrove in March, and left there all summer to allow bark beetle infestation andparasitoid attack to proceed. Once adults of both species had emerged inautumn, the logs were retrieved and the densities of attack (in numbers perdm 2 ) calculated from their respective emergence holes in each log. A lot of usefulinformation can be obtained over relatively large areas with little effort.Emergence trappingIn addition to merely counting emergence holes, it may be useful on occasionactually to collect the insects which emerge from tree bark, or, indeed, fromdeeper inside timber. The actual species residing under bark may not in facthave been identified, especially when species complexes, natural enemies, orrare species are under investigation. Many workers have used some from ofemergence traps or cages to collect emerging insects for later preservation andstudy. Emergence traps usually consist of some form of bag made from cotton orfine mesh nylon, in which a billet of timber is placed. The bags containing billetsmay either be left in the field or, more conveniently, placed in a rearing roommaintained at around 25 °C and 50 percent relative humidity (Reid & Robb

SAMPLING INSECTS FROM TREES 1051999). To facilitate the collection of insect specimens, bags can be suspendedwith plastic funnels at the bottom end into which insects fall. Below the funnelssmall vials of 70 percent alcohol catch and preserve the collection. If the vials areemptied in a regular basis, say once or twice per week, then information on thetiming of emergences can also acquired. Once emergence is thought to be complete,the logs can be dissected and beetle galleries measured as described above.As might be expected, there are fairly close relationships to be found betweenthe number of emergence holes in bark and the actual number of beetles emerging(Fig. 5.17) (Turchin & Odendaal 1996).It might be considered that there is little extra information to be gained in asampling program that merely collects one species of insect. However, this typeof sampling has been used successfully to study the colonization of new, exposedlogs by beetles such as scolytids, and the interactions with their naturallyoccurring enemies in the forest. For example, Weslien (1992) cut billets ofNorway spruce from a Swedish forest, and exposed them for a varying numberof weeks in the forest to attack from the spruce bark beetle Ips typographus, aswell as any other insect species which might also colonize the bark or timber.After the allotted weeks of exposure, each log was caged and emerging insectscollected as described above. Figure 5.18 presents some of the results. Ips wasfound to colonize logs in the first week of exposure, whereas predators andparasitoids did not take up residence in any numbers until logs had been exposedfor four or even eight weeks. These enemies would appear therefore to beresponding to the growing larvae of Ips under the bark before being stimulatedto enter the bark themselves. Competitors (other scolytid species) also did notappear until exposure had lasted for several weeks; it is likely that they are lessaggressive bark colonizers, and require more moribund bark before attacking it.12001000Number of emergedbeetles800600400200R 2 = 0.75030200 400 600 800 1000 1200Number of emergence holesFig. 5.17 Relationship between number of Dendroctonus frontalis adults emerging from logand number of new emergence holes. From Turchin & Odendaal (1996).

106 CHAPTER 5Ips typographusLarge predatorsSmall predators1 week exposed4 weeks exposed8 weeks exposedParasitoids of IpsCompetitors of Ips1 10 100 1000 10,000Log 10 numbers collectedFig. 5.18 Numbers of insects emerging from caged spruce logs exposed for varying lengths oftime prior to caging. From Weslien (1992).Flight trappingOnce adult bark- and wood-boring insects have emerged from bark or timber,their role is to find a mate, and especially of course to find new, suitable tree hoststo colonize. This is usually accomplished by flight, and since suitable breedingmaterial may be patchily and irregularly distributed throughout the forest, itsdiscovery may involve repeated flight over several days (Madoffe & Bakke1995). Because of this behavior, it is possible to sample bark- and wood-boringinsect populations using flight traps, with or without pheromone baits. Flightand pheromone traps are discussed elsewhere in this book (see Chapter 6), andso only their use for wood and bark borers will be briefly described here. Windowflight traps consist in general of perpendicular plates of transparent plasticmounted over a funnel leading into a collecting jar (see also External bark surface,above). Lower-stem flight traps are variations on the same theme, and can beconstructed out of very basic materials, including cut-down plastic milk bottles(Erbilgin & Raffa 2002). Unless these traps are baited with specific pheromones(Erbilgin et al. 2001), beetles and other bark or wood borers flying about hit theplastic panes by chance and fall into the jar. The very randomness of the system,coupled with the fact that overall abundance is not measured directly, rather activity,means that this type of trap is not used regularly for bark- or wood-borersampling. Window flight traps have been used to study assemblages of barkbeetle species in old-growth forests in Finland, for example (Martikainen et al.1999), and as Fig. 5.19 illustrates, many sample plots may be required before

SAMPLING INSECTS FROM TREES 1072520Number of species15105Old growthOvermatureMature00 2 4 6 8 10Number of sample plotsFig. 5.19 Cumulative number of insect species associated with bark beetles caught inwindow flight traps in Finnish spruce forests. From Martikainen et al. (1999).there is some degree of certainty that all species in the habitat under investigationhave been found. When more precise information is available on the flightpatterns and relative abundance of a single species of beetle, pheromone trapsare rather more useful.Synthetic pheromones are now available commercially for a wide range ofscolytid species, and traps baited with them are used routinely to monitor pestlevels, especially in high-risk areas such as dockyards where imported timbermay be infested by unwanted exotic pests. Most commonly, bark beetle baits areemployed in funnel traps, tree-trunk traps or drainpipe traps, all variations onthe same theme. As sampling tools in forest stands, one major question involvesthe range, or effective sampling area, of a pheromone trap. Both the trap designand the concentration of the chosen pheromone bait come into play here.Figure 5.20 shows one example of pheromone trapping for the spruce bark beetleIps typographus, a species with which so much pheromone work has beendone over the years. Clearly, pheromone-baited window and pipe traps show asimilar attraction radius, whilst sticky traps with the same bait are much less effective.Note that the dose of the chemical attractant has a vital influence on thetrap’s efficiency, but only up to a certain concentration. Perhaps most surprising,for this example at least, is the very small attraction radius exhibited: an individualIps flying beyond about 2 m from a particular trap would appear not tobe influenced at all. This problem thus has important implications for the spacingand distribution of pheromone traps for bark beetles in a forest stand.

108 CHAPTER 52.01.8Effective attraction radius (m)1.61.41.21.00.80.60.40.200 500 1000 1500 2000Dose of methylbutenol (mg/d)Window trapPipe trapSticky trapWindow fittedPipe fittedSticky fittedFig. 5.20 Effective attraction ratio for three types of pheromone trap for Ips typographusaccording to pheromone dose. From Schlyter (1992).8% Recapture rate76543210y = ab 2 /(x + b) 2R 2 = 0.9951% recapture distance0 20 40 60 80 100 120Distance (m)Fig. 5.21 Relationship between percentage recapture of marked Ips typographus withdistance of pheromone trap from release site (fitted model). From Zolubas & Byers (1995).A further problem with pheromone traps for scolytids and other flyingbeetles concerns the fact that beetles tend to disperse widely from their point ofemergence from tree or bark material, and the capture rate of pheromone trapsdeclines dramatically as the distance from the source increases (Fig. 5.21). Inthis examples, it would appear that traps have to be located really very close to

SAMPLING INSECTS FROM TREES 109Total number of beetles caught per day20181614121086420Hylastes longicollisDendroctonus valensMay June July August SeptemberFig. 5.22 Seasonal flight patterns for two species of scolytid assessed by pheromone trapcaptures in Oregon. From Peck et al. (1997).the origin of bark beetle flight patterns — a few tens of meters at most is themaximum range for reliable estimations of flying population density. Possiblyone of the most useful ways of employing pheromone traps in the sampling ofbark beetle populations is to describe peak flight patterns for various species. Assumingthat each species of insect reacts similarly to pheromone trap locationand distribution, comparisons of seasonal flight activity can be made to identifywhen pest species are most active, and hence when forest stands or individualtrees are most at risk. In Oregon, USA, Peck et al. (1997) set up multiple funnel(Lindgren) traps, baited with a cocktail of synthetic pheromones for a variety ofbark beetle species, mixed with ethanol which is a general stimulus for barkbeetles trying to locate suitable host trees. Figure 5.22 depicts the seasonal flightpatterns of two scolytid species sampled from May to September. Species suchas Hylastes longicollis have an obvious and fairly tight peak flight period inJune, whilst others such as Dendroctonus valens are active through much of thesummer season and hence much less predictable in terms of risk evaluation(Peck et al. 1997).Wood borersProbably the most difficult group of insects to sample in trees are those specieswhich bore deep into the timber itself, frequently all the way into the heartwood.It is often well-nigh impossible even to detect their presence, never mindto assess their abundance, and apart from the wholesale felling of forests

110 CHAPTER 5followed by laborious dissection with a chainsaw, quantitative sampling of populationsin the timber is just about impossible. Emergence traps and cages containinglogs or billets will eventually reveal some of these species, mainly mothsand beetles, though some take literally years to complete larval developmentand eventually appear as adults. One or two species do give their presence awayby producing active holes in bark from which frass and resin exude. One exampleis the teak beehole borer Xyleutes ceramica (Lepidoptera: Cossidae). Thelarvae of this so-called wood moth tunnel deep into the heartwood of standing,seemingly healthy trees, excavating large and extensive tunnels. Rarely dotrees exhibit obvious symptoms of infestation; only the final timber product becomesseverely degraded. The hole maintained by the larva on the outside of thebark is the only real external sign of the insect within, but this does allow for thepossibility of inspections in tropical forest stands, scoring trees for usually nomore than presence or absence of the pest. Smaller wood-destroying insectssuch as the well-known group of beetles called woodworm (Coleoptera: Anobiidae)can be sampled quantitatively, but again only via assessing emerging adultpopulations. Seybold and Tupy (1993) used laboratory cages containing billetsof Norway spruce in California to investigate infestations of the anobiid Ernobiusmollis. This species excavates extensive tunnels in the phloem and xylem ofmoribund, especially fire-damaged, timber. Emergences from the cages werefound in this study to continue for over 13 months, suggesting both delayedemergence and, more problematically, possible re-infestation of wood materialwith bark remaining actually within the cages. In the latter case, populationdensity sampling runs the risk of rather severe over-estimation.ConclusionsSampling insects on the twigs, shoots, and trunks of trees presents numerousproblems. These include uneven population distributions over external and internalsurfaces, the frequently concealed and inaccessible nature of the life stageunder investigation, and the general magnitude of the sampling proceduresrequired to provide reliable and representative results. Pilot studies are frequentlyrequired to narrow down sampling areas and to select the best protocols.Destruction of the host trees, and the insects that associate with them, isoften unavoidable. Until the advent of technologies that enable us to see insidebark and timber, such as X-rays or ultrasound, sampling these groups of insectswill always be difficult.ReferencesAgassiz, D. & Gradwell, G. (1977) A trap for wingless female moths. Proceedings and Transactionsof the British Entomological and Natural History Society, 10, 3–4, 69–70

SAMPLING INSECTS FROM TREES 111Austin, A.D. & Dangerfield, P.C. (1998) Biology of Mesostoa kerri (Insecta: Hymenoptera:Braconidae: Mesostoinae), an endemic Australian wasp that causes stem galls on Banksiamarginata. Australian Journal of Botany, 46, 559–569.Dreistadt, S.H. (1996) Citricola scale (Homoptera: Coccidae) abundance on Chinesehackberry and scale control with spray oil or acephate trunk implants. Journal of EconomicEntomology, 89, 481–487.Erbilgin, N. & Raffa, K.F. (2002) Association of declining red pine stands with reduced populationsof bark beetle predators, seasonal increases in root colonizing insects, and incidence ofroot pathogens. Forest Ecology and Management 164, 221–236.Erbilgin, N., Szele, A., Klepzig, K.D., & Raffa, K.F. (2001) Trap type, chirality of alpha-pinene,and geographic region affect sampling efficiency of root and lower stem insects in pine. Journalof Economic Entomology 94, 1113–1121.Fidgen, J.G., Teerling, C.R., & McKinnon, M.L. (1994) Intra- and inter-crown distribution ofthe eastern spruce gall adelgid, Adelges abietis (L.), on young white spruce. Canadian Entomologist,126, 1105–1110.Geiger, C.A. & Gutierrez, A.P. (2000) Ecology of Heteropsylla cubana (Homoptera: Psyllidae):Psyllid damage, tree phenology, thermal relations, and parasitism in the field. EnvironmentalEntomology 29, 76–86.Gonzalez, R., Gregoire, J.C., Drumont, A., & de Windt, N. (1996) A sampling technique to estimatewithin-tree populations of pre-emergent Ips typographus (Col., Scolytidae). Journal ofApplied Entomology, 120, 569–576.Gora, V., Koenig, J., & Lunderstaedt, J. (1994) Physiological defence reactions of youngbeech trees (Fagus sylvatica) to attack by Phyllaphis fagi. Forest Ecology and Management, 70,245–254.Gray, D.R., Evans, R.A., & Salom, S.M. (1998) Hemlock woolly adelgid (Homoptera:Adelgidae) dispersion and the failure of binomial sampling to estimate population density.Environmental Entomology, 27, 564–571.Great Lakes IPM (2000) Insect and animal traps. http://www.greatlakesipm.com [accessedMay 12, 2004].Grove, S. (2000) Trunk window trapping: an effective technique for sampling tropicalsaproxylic beetles. Memoirs of the Queensland Museum, 46(1), 149–160.Hanula, J.L., Meeker, J.R., Miller, D.R., & Barnard, E.L. (2002) Association of wildfire with treehealth and numbers of pine bark beetles, reproduction weevils and their associates inFlorida. Forest Ecology and Management 170, 233–247Hébert, C. & St-Antoine, L. (1999) Oviposition trap to sample eggs of Operophtera bruceata(Lepidoptera: Geometridae) and other wingless geometrid species. Canadian Entomologist,131, 557–565.Howard, F.W. (1995) Reduction of damage to mahogany by mahogany shoot borer andmahogany leaf miner by use of azadirachtin. Journal of Tropical Forest Science, 7, 454–461.Jactel, H., Perthuisot, N., Menassieu, P., Raise, G., & Burban, C. (1996) A sampling design forwithin-tree larval populations of the maritime pine scale, Matsucoccus feytaudi Duc.(Homoptera: Margarodidae), and the relationship between larval population estimates andmale catch in pheromone traps. Canadian Entomologist, 128, 1143–1156.Jactel, H., Goulard, M., Menassieu, P., & Goujon, G. (2002) Habitat diversity in forest plantationsreduces infestations of the pine stem borer Dioryctria sylvestrella. Journal of Applied Ecology,39, 618–628.Jakuš, R. (1998) Patch level variation of bark beetle attack (Col., Scolytidae) on snapped anduprooted trees in Norway spruce primeval natural forest in endemic condition: proportionsof colonized surface and variability of ecological conditions. Journal of Applied Entomology,122, 543–546.

112 CHAPTER 5Jantti, A., Aho, T., Hakkarainen, H., Kuitunen, M., & Jukka, S. (2001) Prey depletion by theforaging of the Eurasian treecreeper, Certhia familiaris, on tree-trunk arthropods. Oecologia(Berlin), 128, 488–491.Kaila, L., Martikainen, P., & Punttila, P. (1997) Dead trees left in clear-cuts benefit saproxylicColeoptera adapted to natural disturbances in boreal forest. Biodiversity and Conservation, 6,1–18.Kauffman, W.C., Waltz, R.D., & Cummings, R.B. (1998) Shoot feeding and overwintering behaviorof Tomicus piniperda (Coleoptera: Scolytidae): implications for management andregulation. Journal of Economic Entomology, 91, 182–190.Kelsey, R. & Gladwin, J. (2001) Attraction of Scolytus unispinosus bark beetles to ethanol inwater-stressed Douglas-fir branches. Forest Ecology and Management, 144, 229–238Kleintjes, P.K. (1997) Midseason insecticide treatment of balsam twig aphids (Homoptera:Aphididae) and their aphidophagous predators in a Wisconsin Christmas tree plantation.Environmental Entomology, 26, 1393–1397.Koponen, S., Rinne, V., & Clayhills, T. (1997) Arthropods on oak branches in SW Finland, collectedby a new trap type. Entomologica Fennica, 8, 177–183.Långström, B. & Hellqvist, C. (1990) Spatial distribution of crown damage and growth lossescaused by recurrent attacks of pine shoot beetles in pine stands surrounding a pulp mill insouthern Sweden. Journal of Applied Entomology, 110, 261–269.Lawson, S.A. (1993) Overwintering mortality of Ips grandicollis Eichh. (Col., Scolytidae) and itsparasitoid, Roptrocerus xylophagorum Ratz. (Hym. Pteromalidae), in South Australia. Journalof Applied Entomology, 115, 240–245.Lozano, C., Kidd, N.A.C., Jervis, M.A., & Campos, M. (1997) Effects of parasitoid spatialheterogeneity, sex ratio and mutual interference on the interaction between the olive barkbeetle Phloeotribus scarabaeoides (Col., Scolytidae) and the pteromalid parasitoid Cheiropachusquadrum (Hum., Pteromalidae). Journal of Applied Entomology, 121, 521–528.Lunderstädt, J. (1998) Impact of external factors on the population dynamics of beech scale(Cryptococcus fagisuga) (Hom., Pseudococcidae) in beech (Fagus sylvatica) stands during thelatency stage. Journal of Applied Entomology, 122, 319–322.Macías-Sámano, J.E. & Borden, J.H. (2000) Interactions between Scolytus ventralis andPityokteines elegans (Coleoptera: Scolytidae) in Abies grandis. Environmental Entomology, 29,28–34.Madoffe, S.S. & Bakke, A. (1995) Seasonal fluctuations and diversity of bark and wood-boringbeetles in lowland forest: implications for management practices. South African Forestry Journal,173, 9–15.Martikainen, P., Siitonen, J., Kaila, L., Punttila, P., & Rauh, J. (1999) Bark beetles (Coleoptera,Scolytidae) and associated beetle species in mature managed and old-growth boreal forestsin southern Finland. Forest Ecology and Management, 116, 233–245.Moeed, A. & Meads, M.J. (1983) Invertebrate fauna of four tree species in Orongorongo Valley,New Zealand, as revealed by trunk traps. New Zealand Journal of Ecology, 6, 39–53.Nealis, V.G. (1998) Population dynamics of the white pine weevil, Pissodes strobi, infesting jackpine, Pinus banksiana, in Ontario, Canada. Ecological Entomology, 23, 305–313.Newton, A.C., Cornelius, J.P., Mesen, J.F., Corea, E.A., & Watt, A.D. (1998) Variation in attackby the mahogany shoot borer, Hypsipyla grandella (Lepidoptera: Pyralidae), in relation tohost growth and phenology. Bulletin of Entomological Research, 88, 319–326.Peck, R.W., Equihua, M.A., & Ross, D.W. (1997) Seasonal flight patterns of bark and ambrosiabeetles (Coleoptera: Scolytidae) in northeastern Oregon. Pan Pacific Entomologist, 73,204–212.Pires, C.S.S. & Price, P.W. (2000) Patterns of host plant growth and attack and establishment ofgall-inducing wasp (Hymenoptera: Cynipidae). Environmental Entomology, 29, 49–54.

SAMPLING INSECTS FROM TREES 113Prueitt, S.C. & Ross, D.W. (1998) Effects of environment and host genetics on Eucosma sonomana(Lepidoptera: Tortricidae) infestation levels. Environmental Entomology, 27, 1469–1472.Reid, M.L. & Robb, T. (1999) Death of vigorous trees benefits bark beetles. Oecologia (Berlin),120, 555–562.Safranyik, L., Shore, T.L., & Linton, D.A. (1999) Attack by bark beetles (Coleoptera:Scolytidae) following spacing of mature lodgepole pine (Pinaceae) stands. Canadian Entomologist,131, 671–685.Schlyter, F. (1992) Sampling range, attraction range, and effective attraction radius: estimatesof trap efficiency and communication distance in coleopteran pheromone and hostattractant systems. Journal of Applied Entomology, 114, 439–454.Seybold, S.J. & Tupy, J.L. (1993) Ernobius mollis (L.) (Coleoptera: Anobiidae) established inCalifornia. Pan Pacific Entomologist, 69, 36–40.Siitonen, J. & Martikainen, P. (1994) Occurrence of rare and threatened insects living on decayingPopulus tremula: a comparison between Finnish and Russian Karelia. ScandinavianJournal of Forest Research, 9, 185–191.Speight, M.R. (1994) Reproductive capacity of the horse chestnut scale insect, Pulvinaria regalisCanard (Hom., Coccidae). Journal of Applied Entomology, 118, 59–67.Speight, M.R. & Wainhouse, D. (1989) Ecology and Management of Forest Insects. Oxford UniversityPress, Oxford.Speight, M.R. & Wylie, F.R. (2001) Insect Pests in Tropical Forestry. CABI, Wallingford.Speight, M.R., Hails, R.S., Gilbert, M., & Foggo, A. (1998) Horse chestnut scale (Pulvinariaregalis) (Homoptera: Coccidae) and urban host tree environment. Ecology (Washington DC),79, 1503–1513.Speight, M.R., Hunter, M.D, & Watt, A.D. (1999) Ecology of Insects: Concepts and Applications.Blackwell, Oxford.Speight, M.R., Intachat, J., Chey, V.K., & Chung, A.Y.C. (2003) Insects in managed canopies. InArthropods of Tropical Forests: Spatio-Temporal Dynamics and Resource Use in the Canopy (ed. Y.Basset, R. Kitching, S. Miller, & V. Novotny). Cambridge University Press, Cambridge.Stone, C. & Simpson, J.A. (1990) Species associations in Ips grandicollis galleries in Pinus taeda.New Zealand Journal of Forestry Science, 20, 75–96.Stork, N.E., Hammond, P.M., Russell, B.L., & Hadwen, W.L. (2001) The spatial distribution ofbeetles within the canopies of oak trees in Richmond Park, UK. Ecological Entomology, 26,302–311.Tavakilian, G., Berkov, A., Meurer, G.B., & Mori, S. (1997) Neotropical tree species and theirfaunas of xylophagous longicorns (Coleoptera: Cerambycidae) in French Guiana. BotanicalReview, 63, 303–355.Turchin, P. & Odendaal, F.J. (1996) Measuring the effective sampling area of a pheromone trapfor monitoring population density of southern pine beetle (Coleoptera: Scolytidae). EnvironmentalEntomology, 25, 582–588.Wainhouse, D. & Howell, R.S. (1983) Intraspecific variation in beech scale populations and insusceptibility of their host Fagus sylvatica. Ecological Entomology, 8, 351–359Way, M.J., Cammell, M.E., & Paiva, M.R. (1992) Studies on egg predation by ants(Hymenoptera: Formicidae) especially on the eucalyptus borer Phoracantha semipunctata(Coleoptera: Cerambycidae) in Portugal. Bulletin of Entomological Research, 82, 425–432.Wearing, C.H. & Attfield, B. (2002) Phenology of the predatory bugs Orius vicinus (Heteroptera:Anthocoridae) and Sejanus albisignata (Heteroptera: Miridae) in Otago, New Zealand, appleorchards. Biocontrol Science and Technology, 12, 481–492.Webb, R.E., White, G.B., & Thorpe, K.W. (1995) Response of gypsy moth (Lepidoptera:Lymantriidae) larvae to sticky barrier bands on simulated trees. Proceedings of the EntomologicalSociety of Washington, 97, 695–700.

114 CHAPTER 5Weslien, J. (1992) The arthropod complex associated with Ips typographus (L.) (Coleoptera,Scolytidae): species composition, phenology, and impact on bark beetle productivity. EntomologicaFennica, 3, 205–213.Zhang, Q.H., Byers, J.A., & Schlyter, F. (1992) Optimal attack density in the larch bark beetle,Ips cembrae (Coleoptera: Scolytidae). Journal of Applied Ecology, 29, 672–678.Zolubas, P. & Byers, J.A. (1995) Recapture of dispersing bark beetle Ips typographus L. (Col.,Scolytidae) in pheromone-baited traps: regression models. Journal of Applied Entomology,119, 285–289.Index of methods and approachesMethodology Topics addressed CommentsExternal Assessments of which trees in Only efficient for external feeders,examination for a stand may be attacked. especially somewhat sessile formsinsects Assessment of attack severity. (e.g. sap-feeders, gall formers etc).Presence of insects on thesurface.Identity of insects.Absolute estimates ofpopulation density andspecies richness.Assessment of communitystructure, e.g. guild.Collection of live specimensfor subsequent experimentalwork on population dynamicsand feeding strategies.External Assessments of which trees in Shapes and sizes of holes, as well asexamination for a stand may be attacked. extruded frass or bore dust, provideevidence of damage Assessment of attack severity. good indication of pest type (e.g.woodwasps, longhorn beetles, barkPresence of insects withinbeetles, ambrosia beetles, powderpostshoots and stems.beetles).Density of emergedpopulations.Emergence of herbivores andenemies via relative sizes andshapes of emergence holes.Region of stem or trunkinfested.Removing bark or Location of boring species. Mainly collects larvae which need tosplitting shoots and Identity of boring species. be reared to adulthood (difficult) tologsidentify accurately.Extent of attack.Effects on the tree.Continued

SAMPLING INSECTS FROM TREES 115Methodology Topics addressed CommentsSticky traps, lobsterpottraps; baitedtrapsIdentifies and assessesdensities of mobile species,larvae nymphs or adults.Identifies timing of lifecyclechanges such as adultemergence and oviposition.Trap logging Estimates of population May be used to manage pests such asdensity and species richness bark and ambrosia beetles.of flying adults looking fornew host trees to colonize.Flight and Estimates of population A more sophisticated and speciespheromonetraps density and species richness specific version of trap logging.of flying adults.Early warning of pestinvasion in stands, log yardsand ports.

CHAPTER 6Insects in flightMARK YOUNGIntroductionThe ability to fly, that makes insects such interesting animals for study, also hasa great impact on the ways that they can be caught. Their high mobility bothhelps them to approach traps and helps them avoid them, so specialized trappingstrategies are needed. Essentially such traps either rely on actively attractinginsects, or they merely intercept their flight. Examples of both are includedbelow.Light trapsThe legendary attraction of insects to light has long been used in ecologicalstudies. Although it is still not certain why the attraction takes place, the resultsof standardized light trapping have produced vital ecological insights in manydifferent places and conditions.Basic principle of useEven if it is not yet clear why light sources attract insects, it is certain that the effectdepends on the degree of contrast between the light source and its surroundings,with reduced catches when contrast is low (Nag & Nath 1991). Thebasic design of a light trap therefore includes a single bright source and a methodof trapping the insects that arrive. The behavior of insects when very close tothe light may change, however, in that they tend to avoid the very close, brightlight by flying into the dark area around it. In fast-flying species, such aslarge Lepidoptera, their momentum may carry them right up to the light trap,whereas slow-flying small Diptera may avoid being trapped by diverting fromthe light into the dark adjacent area (Muirhead-Thomson 1991). Most of thevariation in trap design arises from the need to increase trap efficiency for thechosen type of insect and to achieve practicality, portability, and economy, forwhat is necessarily a cumbersome and expensive trap type.116

INSECTS IN FLIGHT 117Type of light sourceIt is believed that different types of light source are optimal for different insectorders. For each type of source, however, catch is increased as the intensity oflight is increased. Tungsten filaments, emitting largely in the visible spectrum,are apparently most attractive to many Diptera, whereas Lepidoptera andTrichoptera respond best to ultraviolet-rich sources, such as mercury-vaporbulbs, including even “black” lights, which emit only in the ultraviolet range(Southwood & Henderson 2000). However, the addition of suction traps tomercury-vapor traps increases the catch of small Diptera (Downey 1962), andthis implies that the apparent preference for visible sources by these speciesmay actually be because the extra intensity of the ultraviolet traps leads to anincreased “escape” behavior close to the trap.Waring (1980, 1990) compared the catch of moths in traps fitted with bulbsemitting different ultraviolet levels. He found that a 6 W fluorescent tube attractsonly 15–40 percent of the number of individuals and 50 percent of thenumber of species, compared with a 125 W “MB” bulb, and it is well known thaton light nights the 6 W tube is almost ineffective. Bowden and Church (1973)showed that catches of Lepidoptera may be 3–10 times greater on clear, moonlessnights than on clear nights when the moon is full, a finding confirmed byTaylor (1986). Although this may be because fewer moths fly at full moon, it ismore likely that the reduced catch reflects the lower light source contrast on amoonlit night. Bowden and Morris (1975) showed that some insects fly more ata full moon, but are caught less often then. Haufe and Burgess (1960) found lighttraps to be ineffective for mosquitoes on light Arctic nights, and substituted trap“sources” which were merely high-contrast black and white striped cylinders.Comparative catchabilityLight traps vary greatly in their effectiveness for different types of insect.Diptera, Lepidoptera, and Trichoptera are orders frequently trapped, whereasonly some species of Coleoptera and Hymenoptera appear. However, this maymerely reflect the relative abundance of night-flying representatives of thesegroups. The selectivity also applies within orders. Moth trappers know thatmale moths almost always outnumber females and that some species may beabundant at nearby sugar baits but do not enter light traps (Young 1997). Part ofthe reason may just be that some species (and females) fly less than others (andmales) but some residual selectivity does remain and interpretation of catchesmust reflect this.Influences on catch efficiency and effective catching distanceGaydecki (1984), McGeachie (1987), and others have observed that highcatches of nocturnal insects are made on relatively warm, still, dark, and humid

118 CHAPTER 6nights. Gentle rain may not reduce catches but heavy rain and moderate orstrong winds do so. Gregg et al. (1994) found reduced catches of migratorymoths when wind speed increased, as did Edwards et al. (1987) for Culicoidesmidges. Wind direction is also important, insofar as it reflects meteorologicalconditions. In Britain southwesterly winds are associated with mild, humidairstreams, and high insect catches, whereas northeasterlies are colder and drierand catches are low.Direct observation of small Diptera shows that they are often unaffected bylight traps beyond a range of 5 meters, whereas large Lepidoptera are sometimesseen to veer towards the light when up to 20 or more meters distant. Roberts(1996) released marked moths of the family Noctuidae at various distancesfrom a standard 125 W “Robinson pattern” trap and showed a rapid drop-off inrecapture, such that beyond 15 m the trap was almost completely ineffective.However, Graham et al. (1961) estimated the trapping radius of their trap as60+ m for the pink bollworm Pectinophora gossypiella in the USA. Roberts (1996)observed that flying moths often settled well short of the trap, and Hartstacket al. (1968) found a similar behavior in Heliothis species, whereas Trichoplusia nitended to fly up into the trap.Trap designsA basic design for Lepidoptera is that of the “Robinson” pattern trap. The light isheld above a downward-pointing, open-ended cone, within which are 2–4vertical baffles. Moths drop down through the cone and are trapped in the boxlikebase. The upper portion of the base is transparent, so that moths trying toescape fly to this instead of finding their way back through the narrow coneopening. This trap resembles a lobster pot, and many minor modifications exist(Southwood & Henderson 2000). It usually uses a generator or mainspoweredhigh-intensity UV source, whereas a low-wattage fluorescent tube,run from a car battery, is used in the more portable but essentially similar“Heath” trap.In the 1930s C.B. Williams designed a standardized but relatively inefficient“Rothamsted” trap for Lepidoptera at Rothamsted Experimental Station(Williams 1948) (Fig. 6.1). A high-power tungsten bulb is held under an opaquesquare top and glass baffles lead moths into a killing jar. This trap is very resistantto wet, windy weather and is designed to be used every night of the year in astandardized trapping sequence. The use of the killing jar prevents escape andwill increase catches in all trap designs. By the use of timing devices it is also possibleto separate the trap into components, so as to investigate catches in eachpart of the night.Light traps for Diptera tend to incorporate a suction device, so as to completethe catching of individuals that approach the trap. A common example ofthis type is the “New Jersey” trap (Mulhern 1942), which was designed tocatch mosquitoes, but many variants exist and some are illustrated in detail

INSECTS IN FLIGHT 119Fig. 6.1 Rothamsted Insect Survey standard design light trap. These are used in long-termmonitoring of moth populations. Picture courtesy of Ian Woiwod, Rothamsted ExperimentalStation.by Southwood and Henderson (2000). Rawlings et al. (2003) used up to 40 standardized8 W black light traps, incorporating a suction trap held below the light,scattered across all of South Africa, to sample Culicoides midges as part of a faunaland veterinary study. The lights attract the midges but very small flying insectstend not to enter the actual trap unless they are sucked into it by the suction device.

120 CHAPTER 6Examples of useThe large influence of variable weather conditions, plus the variation in contrastand exposure at different sites, makes it very difficult to compare lightcatches from different places and/or different nights. Consequently, light trapsare difficult to standardize and so their use has to be carefully controlled. Theyare excellent for surveys or general assessment of species richness and theyhave played a large part in quantifying the species richness of different tropicalforest areas (e.g. Robinson & Tuck 1993). They are also used to detect the emergenceof pest species, prior to control measures.If used over long periods, however, the nightly variation is “averaged out”and useful data accrue. The Rothamsted trap data series, for some of their arrayof traps now running for nearly 40 years, has allowed extensive analysis of ecologicalpatterns of diversity and population dynamics. Early work by Taylor et al.(1976) established the validity of measures of diversity in insect communities,and more recently Woiwod and Hanski (1992) were able to use long-term dataruns to demonstrate density-dependent effects, which had proved highly elusivein the more short-term studies. Woiwod and Harrington (1994) provide anoverview of the work of the Rothamsted Insect Survey, illustrating the crucialimportance of their light-trap series.Chapman et al. (2002) provide an interesting example where the use of alight trap has validated another method of assessing the abundance of a flyinginsect. They were interested in the abundance of Plutella xylostella as a highaltitudemigrant, and were primarily using vertically orientated radar to assessthis. Although reasonably confident that they could discriminate P. xylostella onthe basis of the size of the radar reflections and the wing beat frequency, theyused an MV light trap (and also a net deployed from a tethered balloon) toconfirm the presence and general abundance of the moth.Suction trapsSuction traps have been used very extensively to trap small flying insects,with a particular emphasis on biting Diptera and on aphids. These arenot caught easily by light traps or netting methods, although many speciesnow have pheromone or other baits available for them. Experience hasshown that suction traps can provide a very effective catching efficiency,although they may be heavily influenced by weather conditions and precisetrap placement. Some standardized traps have been running for manyyears, providing an unrivaled dataset for detecting seasonal and annual patternsin the populations of the target species. They have almost completely replaced“trawl” traps, where a net is towed through the air for a given distance, ata set speed.

INSECTS IN FLIGHT 121Basic designs and simple modificationsThe basic design of a suction trap is that a motorized fan draws or forces airthrough a filter, on which the insects are caught. Modifications affect the powerof the fan (and therefore the volume of air processed in a standard time); thedirection, placement, dimension, and precise shape of the entry port; and thenature of the filter and trapping chamber. Frequently, the suction device is associatedwith another trapping method, usually a light trap, a bait trap, or apheromone trap. The time of use, the height at which used, and the trap surroundingsalso influence the trapping efficiency.Since all of the factors that are varied have an undoubted effect on the sizeand composition of the catch, it is essential in any study to standardize these factorsand to define them very clearly in all reports.Standard Rothamsted suction trapIn 1964 the first of what is now a European series of standardized suction trapsfor aphids was set up at Rothamsted Experimental Station (Fig. 6.2). This wasdesigned by L.R. Taylor, as a modification of previously used traps at the researchstation (Taylor 1951), and is characterized by a 12-meter-high entry tube. This isdesigned to ensure that the trap catches aphids which are moving significant distancesin the air column, rather than just ones in close proximity to a host plant,and this reflects the proposed use for the traps. As well as providing informationon aphid species richness and diversity, and data on year-to-year populationchanges, the traps are used to provide early warning of the arrival of pest species.The spring movements of crop pest aphids can be plotted by the network of trapsand farmers given appropriate warning of the need to use insecticide spray. AnAphid Bulletin is published to provide this advice and is circulated within oneweek of each capture event (Woiwod & Harrington 1994).The currently used fan draws in 45 m 3 per minute through a pipe of244 mm internal diameter, resulting in high “extraction” efficiency butalso a very high wind velocity in the pipe (Muirhead-Thompson 1991). Toavoid damaging the insects, the wind velocity is reduced by having anexpansion chamber just above the filter. This results in a “capture” velocity ofjust over 1 m per s, which keeps the catch in good condition. A jar with preservativefluid is the actual receptacle in which the catch is stored. To run such apowerful fan, it is necessary to use a mains power supply, so that this design oftrap is not portable. The traps are shaped and colored so that they are neitherattractive nor repellent to insects but merely catch the individuals present in avolume of air.In 1994 15 standard Rothamsted suction traps were in operation in Britain,with 14 more in France and several others in the rest of Europe (Woiwod & Harrington1994). These provide a wide-ranging picture of the pattern of aphid

122 CHAPTER 6Fig. 6.2 Rothamsted Insect Survey standard design suction trap. These are used to providewarning of increasing aphid numbers in spring and early summer. Picture courtesy of IanWoiwod, Rothamsted Experimental Station.movement across Europe and are the model which is also being adopted worldwidefor national schemes.The use of data from these standardized, continuous catches has allowed rigoroustesting of the relationships between climatic variables and aphid abundance.For pest species, such as Aphis fabae or Myzus persicae, this has also alloweddevelopment of predictions of aphid abundance based on winter weather (Baleet al. 1992). It has been found that the temperatures recorded in January/February,which affect mortality of overwintering aphids, are well correlated withnumbers the following season, with time of migration to the host crop dependinglargely on accumulated day-degrees from midwinter. This has allowed goodprediction of the likely time of arrival, and abundance, of the pest species, sohelping farmers plan preventative spraying. Using these models it has also beenpossible to predict the likely impact of climatic warming, with alarmingly highincreases proposed for an overall 3°C rise in temperatures (Anon. 1991).The Rothamsted dataset for aphids (and also some moths from their light

INSECTS IN FLIGHT 123traps) have also been used successfully to test for the presence of “chaotic”dynamics in insect populations, a procedure only possible where long-termdata are available (Zhou et al. 1997).As well as for small Diptera, suction traps have been used for many othersmall aerial invertebrates, for example ballooning spiders (Dean & Stirling1985) and beetles (Leos Martinez et al. 1986).Modified designsSuction traps began to be used extensively in the 1950s and early 1960s anda very commonly used type of fan was the 9-inch (230 mm) diameter commerciallyavailable Vent-Axia® (Johnson & Taylor 1955). This moves around10 m 3 per min, considerably less than the Rothamsted design, but sufficient tomake large catches of small Diptera in favorable conditions. In a very commondesign (the “exposed cone” design), this fan is set at the top of a long mesh coneand the insects are gradually sieved down to the bottom of the cone, into a collectingtube. Passage past the fan blades does not usually harm small insects.Often there is a series of plates which can be dropped into the collecting tube atpre-set intervals so as to segregate the catch. This allows investigation of thetime of flight of the species being caught.More recently smaller fans, powered by batteries, have been used in trapsspecifically designed for a particular type of insect, often in association with anattractive “lure”. In this case the insects are attracted from the surroundings bythe lure and then trapped by being sucked into the collecting tube. This preventscalculations of absolute insect density in the air but maximizes numbers caught.These traps are frequently fully portable and can even be hauled up into thecanopy of a forest.It is now routine to modify the precise design of the trap to suit the study insect.Some of the slower flying mosquitoes or midges (e.g. Ceratopogonidae)need only low-powered fans but are very sensitive to the placement and shapeof the trap opening. Some are caught only in exposed locations and others onlywithin shelter; many are only caught if a CO 2or pheromone bait is also used todraw the insects to the vicinity of the trap. A further consideration is whetherthe shape and coloration of the trap is itself attractive or repulsive, for if the purposeof the trap is to produce a measure of the density of individuals in the location,then the trap itself must be neutral in attractive effect. If the trap appears tobe a vertical shape then it may be used as a cue, above which form swarms ofmale Diptera. These may then be caught in huge abundance, distorting the trueabundance of the species.Calibration of suction trap catches and the effect ofdiffering wind speedIf a suction trap is neither attractive nor repulsive, then the catch may represent

124 CHAPTER 6the true number of insects that were present in the volume of air that hasbeen filtered. This will only apply to individuals that fly so weakly that theycannot take evasive action, of course, and progressively stronger flying specieswill be caught less and less effectively. If the trap really does provide a properestimate of the numbers of insects present then it becomes possible to makefull estimates of abundance by considering the volume of air sampled. Sincewind speed both affects the efficiency of capture and also physically moves theair and its insects past the trap opening, it is also necessary to consider windspeed in any calculations. Taylor (1962) investigated the relationship betweenthese factors and derived an equation for the efficiency of capture of each of aseries of commonly used traps. He then produced a logged conversion factorfor each size of insect, at various wind speeds, so as to arrive at the followingformula:log catch per hour + conversion factor = log density◊10 6 cu ft of air (6.1)There is no doubt, however, that this formula is too simple to account fully forthe many factors that influence catches, and so results from it (and more recentmodifications) must be interpreted critically.Taylor (1962) also noted that there was some variation in the efficiency ofany one design, and so a difference of at least 6 percent in catch for a 9-inch(230 mm) fan, or 10 percent for an 18-inch (460 mm) fan, is necessary beforea real difference in the level of catch can be claimed.Factors influencing catches by suction trapsIt has been found that the precise surroundings of a suction trap can alter thecatch dramatically, as shown by a study of mosquitoes in Florida which usedmoveable mesh barriers and potted shrubs to vary the degree of exposure(Bidlingmayer 1975). Different species were caught in open, sheltered, and partiallysheltered conditions. It has also been found that the height of the trap inletis of considerable importance. Some woodland mosquitoes in the UK fly onlybelow 30 cm, and so in one study these were caught using a trap which had beensunk into the ground (Service 1971).This relationship between height of inlet and species caught has always beenconsidered (vide the 12-meter-high Rothamsted traps), and Southwood andHenderson (2000) record a procedure to develop an equation which relatescatch abundance to outlet height for any given species. In The Gambia it wassimilarly found that different species flew at different heights, although therewas also a difference associated with time of night.Four main factors influence the size and composition of the catch, aside fromheight or placement. These are wind speed, insect size, suction rate and pipe diameter,and precise shape of the inlet tube. Many years ago comparisons weremade between the efficiency of suction traps and of sticky traps and tow nets atdifferent wind speeds (Johnson 1950). In general wind speeds below 3–5 miles

INSECTS IN FLIGHT 125per hour (5–8 km◊per h) favor suction traps, whereas above 7 miles per hour (11km per h) tow nets still work but suction traps are ineffective. The relationshipalso depends on insect size, however, for large Diptera are predominantlycaught on sticky traps, rather than by suction traps.The wind speed disrupts the suction gradient, as can be detected by usingsmoke particles to visualize it, but even in still air small traps may only trap insectsthat approach within 20 cm of the inlet. It has also been found that the “exhaust”airflow can disrupt the suction gradient, so that, if more than one trap isused, care must be taken in their mutual placement (Bidlingmayer & Hem1981). The influence of wind speed is indirect, as well as direct, in that fewer individuals,especially of small species, may fly in high wind speeds.The area of the inlet diameter influences the catch, as does its precise shapeand direction, and the use of directional traps and gauze baffles have shown thatdifferent species respond differently to these factors. Often mosquitoes fly upwindin a gentle breeze and so are caught more when the inlet is directed downwind.However, the overall shape of the trap is also important, especially forbiting species, and some species of Simulium approach and are caught by a suctiontrap resembling a standing human, whereas for others a longitudinal “cow”shape is most attractive (Coupland 1990). Many large species are barely trappedat all and presumably fly against the suction action; this can be substantiated byusing radar that shows the presence near a trap of species such as medium tolarge moths, which nevertheless do not appear in the trap contents (Schaeferet al. 1985).The use of “baits” with suction trapsIt is now commonplace to use either light, or attractively colored shapes, orvolatile baits, with suction traps (Fig. 6.3). For example, researchers at theInstitute of Animal Health at Pirbright have designed and used extensively aportable combined light and suction trap for various small Diptera (the “Pirbright”trap). These traps are now widely used, as by Linton (1998) and others,who caught minute Culicoides midges in the UK and Spain. (The use of volatilebaittraps is described below.)If light traps have become an almost universal type of trap for large flying insects,then suction traps have certainly become similarly generally used forsmall flying species. As with light traps, however, they are so easily affected byprecise trapping conditions that great care must be used in interpreting theircatches.Water (or pan) trapsMany flying insects settle onto surfaces whose tone and/or color contrastclearly with the background. Such behavior is obviously related to the taxesthat lead to recognition of, and settling onto, such things as flowers or ponds

126 CHAPTER 6Fig. 6.3 Onderstepoort Veterinary Institute design combined light and suction trap. Smallinsects are attracted by the light and then sucked down into the collecting chamber.

INSECTS IN FLIGHT 127and so, as expected, different species and sexes of insects have different responses.This behavior is taken advantage of in the use of “water” or “pan” traps,which are nothing more than a contrasting surface and water to trap the insectsthat settle onto it.Basic design and useSimple water traps (Fig. 6.4) are merely shallow dishes, usually around 15 cmdiameter, which are held at a standardized height above the ground, often 1 m.They contain water, to which has been added a drop of detergent to reduce surfacetension, and sometimes an added preservative, such as formalin. The trapsare left in place for a set time, often 2–4 days, after which their catches arefiltered through a sieve and stored. Such traps catch a very wide range andFig. 6.4 Simple water trap. Insects are attracted to settle on the bright surface and becometrapped in the water.

128 CHAPTER 6abundance of flying insects and so can be used in general faunal surveys, as wellas in more focused studies. The catch is very dependent on weather conditions,for in windy or cool weather few insects fly and are caught. Nevertheless, thetrap cheapness allows easy replication and if sufficient traps are used then thecatch data can often be regarded as semi-quantitative.ModificationsA frequent modification is to use upright, mutually perpendicular baffles, setabove the trap so as to intercept flying insects (Coon & Rinicks 1962). However,the biggest influence on the selectivity and efficiency of the traps is the color ofthe pan. Harper and Story (1962) showed that different colored traps attractvarying numbers of the sugar beet fly Tetanops myopaeformis, and numerousauthors have repeated this experience, including Leong and Thorp (1999), whostudied bees that visit white-flowered Limanthes douglasii rosea. They found thatfemale Andrena were attracted to white and blue pans, whereas males werecaught more in white and less in blue and yellow ones, differences that might berelated to natural flower visiting. The effect of different colored dishes, as well asof differently patterned baffles, has also been investigated by researchers attemptingto monitor numbers of tsetse flies (Deansfield et al. 1982). Blue washighly attractive to Glossina tachinoides, whereas white was best for G. morsitans,and the catch was increased by the use of black baffles. Other authors havefound that red traps are favored by some beetles (Dafni et al. 1990), whereaswhite and yellow pans are visited most by Diptera in general (Disney et al.1982). Following such experience, most general studies have used one of thesetwo pale colors (Kirk 1984), whereas focused studies should use a preliminarycatching period to select the most efficacious color.Types of useWater traps have been used for widely different studies. The fact that most flyinginsects are attracted at least to some extent has allowed their use in faunalsurveys and comparisons. For example, Young and Armstrong (1994) usedwhite traps in various stand types of native pinewoods to make comparisonsbetween the stands. They found that the catches included some of all orders offlying insects present in the forests, but that Diptera, Coleoptera, and Hymenopterapredominated. Furthermore, the trap catches allowed easy differentiationbetween stand type. However, the catches were greatly affected byprecise location, so that sheltered traps caught best during windy episodes, althougheasily visible traps in open forest were generally to be preferred. Theyand others (e.g. Disney et al. 1982) found many “tourist” species in the catchesand these reduce the site specificity of the results and make them less suitablefor focused site comparisons than results from (for example) pitfall traps.McGeachie (1987) used a circular array of water traps at varying distances

INSECTS IN FLIGHT 129from light sources to assess the position at which insects alighted, when theyapproached the light from different directions. In this example the catches weremerely of settling insects, rather than ones attracted by the pans, but theyproved very effective.Leong and Thorp’s (1999) study was more specific. They were catching insectsthat pollinated a particular plant, and the color and size of the pans usedwere designed to act as “super-flowers”. However, as well as catching the actualpollinators, they also caught many other insects and this “wastage” may not bewidely acceptable.The catching power and simplicity of the water trap has often been combinedwith pheromone traps to produce highly effective designs. Many pest specieshave been monitored using such combined traps. For example, Thompson et al.(1987) compared the catch of the European corn borer Heliothis zea in light trapsand pheromone traps, with either a sticky surface or a water trap. At times lighttraps worked best, but water traps improved the success of the pheromonelures.In summary, water traps will provide either a wide-ranging and abundantcatch, which reflects the general composition of flying fauna of an area, or amore focused assessment of one taxon. However, they cannot achieve betterthan a semi-quantitative result and so interpretation of catch data must be cautious.Nevertheless, their simplicity and economy make them useful in manycircumstances.Sticky trapsThe basic trapping style of sticky traps is similar to that of water traps. Insectssettle onto a surface and are caught there. This process may either be passive,where the sticky surfaces merely intercept insects that are blown or fly inadvertentlyonto them, or active, where the insects choose to settle onto the surfaces.The shape, position, and color of the traps all influence the trappingefficiency, and they are often used in conjunction with other trap types.The major practical differences between water and sticky traps are twofold.Water traps have to be set facing upwards, whereas sticky traps can be angledand shaped to whatever design is required, which can be a major advantage.However, insects caught on sticky traps are frequently badly affected by theprocess. The usual practice is to use a sticky substance that can be dissolved, usuallychemically but in some cases by warmth, so that the insects can then besieved out and preserved in alcohol. However carefully this is carried out, delicateinsects often lose scales and/or appendages in the process and may also bedistorted, so reducing ease of identification. This matters less for beetles or robustwasps, but moths and small flies may be beyond use. This is a substantialproblem, which greatly reduces the usefulness of sticky traps for general purposes.Their main use is therefore in focused studies, where another attractant

130 CHAPTER 6is also used, so that only one species or taxon is being caught, and where thecondition of the catch is unimportant.Basic design and useA basic sticky trap is a surface coated with the trapping gum. Various substancesare available commercially and are of varying holding strength. For tiny insectsa rather fluid grease, or even castor oil, may suffice, and this will cause less damage,whereas for larger insects a thick resin or gum is better, and polyisobutyleneis now widely used. The gums themselves act to preserve the insects for alimited period and a standard practice is to cover the sticky surface, with itstrapped insects, in the field with “cling film,” so that sheets can be stored togetherfor later analysis. As soon as possible thereafter the gum is dissolved(Murphy 1985) and the catch transferred to fluid preservative.If an upright cylinder or flat surface is used, then the catch will include insectsthat have merely been “blown” onto the surface. In most circumstances, however,the surface is angled and shaped to produce a more selective catch, often inconjunction with a specific attractant.The cost of each sticky trap is small and so many can be included in a samplingprogram. This allows extensive replication, so that confidence can be assignedto results, and this is a major advantage. However, the trapping surface can becomefully saturated, or the gum can lose its stickiness (when cold or coatedwith dust), so that there may not be a linear catch rate over time. Catches arealso highly dependent on weather conditions, insofar as these affect flight, sothat interpretation of results must allow for this.ModificationsThe most common use of sticky traps is to provide the catching power in trapsusing attractive lures, frequently pheromones. However, the color, shape, andsize of the sticky surfaces also affect the basic efficacy and these will be discussedfirst.Different species of insect fly at different heights and respond to colors in differentpositions. Finch and Collier (1989) provided flat sticky squares, angledfrom facing directly upwards, by 45° intervals through directly downwardsback to upwards, and recorded the catch type in different agricultural fields.Syrphidae were almost exclusively caught on vertical surfaces, Psila rosea (theturnip root fly) on faces pointing 45° downwards, and Delia species on upwardsor upwardly angled plates.Vertically placed cylinders with sticky surfaces are claimed to catch freely flyinginsects without bias, but it was realized very early on (e.g. Taylor 1962) that,even if the wind speed and direction were recorded, it was not possible to relatecatch density directly to actual abundance. Furthermore, the color of the surfaceinfluences the catch. (For this reason, suction traps were developed to trap

INSECTS IN FLIGHT 131aphids in realistic numbers.) Many researchers have used flat, square surfaces,but others have mimicked the shape of the natural target of their insects. Kring(1970) compared the efficiency of catch of red spheres and yellow panels forapple maggot flies Rhagoletis pomonella and found that a red hemisphere set on ayellow background out-performed a flat red circle on the same yellow background.Later workers, such as Jones (1988), have also found that red spheresare the optimum shape for this species of pest. However, Meyerdirk et al. (1979)found no difference in catches between triangular, square, circular, and rectangularyellow surfaces for the citrus blackfly Aleurocanthus woglumi.The effect of different colors of sticky traps has been studied often, includingthe preferential choice of red spheres reported above. Katsoyannos (1987)found that the preferential sequence of colors for Mediterranean fruit fliesCeratitis capitata was yellow > orange > black > red = green > white = blue. Webbet al. (1985) showed that greenhouse whitefly Trialeurodes vaporarium showeda rather similar preference spectrum, namely yellow > green = orange > white= violet = blue = red = black. They showed that the bright yellow traps evenout-competed leaves as a landing surface.Comparative trap efficiencyThe crucial question is how well sticky trap catches compare with real abundancesof target species. Heathcote (1957) made an early attempt to comparethe real efficiency of different trap types by noting the numbers of aphids caughtby water, flat sticky, and cylindrical sticky traps, as a ratio to catches from adjacentsuction traps. He found that results were very variable, with high catchesfor some species from water traps (e.g. ratio = 3.91 for Tuberculoides annulatus),and on cylindrical sticky traps (e.g. 1.92 for Aphis fabae), but generally lowcatches on flat sticky traps (ratios from 0.01 to 0.84). In a greenhouse environmentGillespie and Quiring (1987) compared the numbers of Trialeurodesvaporariorum found on plants with those caught on yellow sticky traps. Theyfound that the density of traps influenced the answer. It was possible to swampthe system with traps, so that almost all insects were on traps, rather than onplants. At low trap densities, however, there was a reasonable relationship betweeninsects on leaves and on traps. Traps also proved to be very sensitive anddetected whiteflies at levels that were barely perceptible on plants.In field conditions Ramaswamy and Cardé (1982) compared the efficiency oftraps under different conditions for the spruce budworm Choristoneura fumiferana.Sticky surfaces, associated with pheromone lures, greatly increased thesensitivity of the traps, but only if they were changed sufficiently often to preventsurface saturation (but not so often as to disturb recently arrived moths).This last example re-emphasizes the point that sticky traps are most oftenused in association with another lure. They can be highly effective, and arecheap and easily replicated; however, they only represent real population levelsin certain closely defined circumstances.

132 CHAPTER 6Baited trapsCertain insects are attracted powerfully to volatile chemicals, and these chemicalscan be used in traps to attract and kill the insects, either so as to monitortheir numbers or to actually reduce their populations. Blood-sucking species,such as tsetse flies Glossina spp or mosquitoes (e.g. Aedes spp), for example, areattracted by host odors (as well as host shapes), whereas male moths of manyforest pest species are attracted by sex pheromones released by females.Basic trap designs and common modificationsThe elements of a bait trap are the source of attractant and a trapping surfaceor container. Although these basic elements are similar for “bait” and“pheromone” traps, there are sufficient differences for them to be discussedseparately.“Bait” trapsBait traps divide roughly into those which use host odors to attract female flieswhich are hoping to feed, and those which use carrion or dung in which the femaleflies hope to oviposit (carrion or dung traps). Vessby (2001) used naturallycollected dung placed in suspended “bucket” traps to collect samples of thedung beetles Aphodius spp, and showed clearly that different catches were madeif the dung was allowed to dry out, rather than if it was watered to keep it moist.Early carrion traps typically used liver as a bait and flies were caught if theyflew from the bait up into a conical catching chamber above the bait. These canbe highly effective for some species, such as blow-flies (e.g. Lucilia spp) thatcause “strike” in sheep. However, the traps suffer from operational problemsand are influenced by various environmental conditions (Gillies 1974).First of all the age of the liver bait influences the species caught, so that it is difficultto compare different catches. However, one response to this, to discardbaits after three days (Vogt et al. 1983), leads to the traps being very inefficientwhen used against screw-worms (Cochliomyra spp.) in the USA. It was foundthat these are only attracted efficiently by baits over 5–7 days old (Coppedge etal. 1978). Secondly many non-target species may be caught, including large anddisruptive carrion beetles and ants. To prevent this, baits are now typically presentedon poles incorporating ant “baffles” and with a coarse screen in front ofthe catching chamber. The liver bait is also usually shielded by a gauze cover, toprevent excessive egg laying and subsequent changes due to maggot feeding. Asan alternative to liver, commercially available chicken legs were used by Smithand Merrick (2001) to attract carrion-feeding Nicrophorus beetles.Inevitably more female than male flies are caught, although some males alsoarrive both to feed and to find mates. It has also been found that the proportionof males varies through the day and with weather conditions (Vogt et al. 1983).

INSECTS IN FLIGHT 133Generally catches of both males and females increase directly with temperature,up to a threshold level, but males are caught more in bright sunlight. Increasingwind speed reduces catches but some species are found mainly inexposed traps, whereas others come mainly to sheltered sites.Recently the principal development in “carrion” types has been the use ofspecific chemicals in place of meat baits. For screw-worm controls, it has beenfound that a cocktail of the volatile chemicals that are released as the liverdecomposes make a very effective attractant, originally called Swormlure(Coppedge et al. 1978). A series of improved recipes has since been used, varyingboth the chemicals used and their relative concentrations.Whole animals as baitsTraps which use “host” baits were originally literally that. Whole animals wereused to attract the biting flies (Phelps 1968, McCreadie et al. 1984). Even nowan effective way to collect individual tsetse flies is to “poot” them off the surfaceof a human or animal bait, and Coupland (1994) used a similar method to assessthe activity and behavior of Simulium in Scotland. This catches small numbers offlies in good condition, and early attempts to make larger catches involved suddenlydropping nets over tethered oxen. The practical problems of using suchtraps are easily imagined and they are not now widely used, except in conjunctionwith small bait animals such as rabbits. However a commonly used variantkeeps the whole animal as a bait but uses a series of electrocuting surfaces ornets around the animal on which the flies are killed and caught (e.g. Vale 1974,Rogers & Smith 1977). Alternatively a suction trap may be placed directly abovea small animal bait, as has been used to trap various nuisance flies in Trinidad(Davies 1978).Greater efficiency of capture for low-density biting flies is provided by the useof moving baits, often over a set transect, with stops at specified catchingstations. However, as Glasgow (1961) found, it is still only possible to detectgross population changes using such methods.The attractant nature of an animal is often a combination of its size, shape,heat output, exhaled breath components including CO 2, and body odor. It isdifficult to mimic all of these, hence the continued use of whole animals, butvarious trap designs use combinations. For example Coupland (1990) useda cow-sized, rectangular black Malaise-style trap to catch Simulium spp, butcatches were enormously enhanced by adding a slow leak of CO 2from the undersideof the trap. This was provided cheaply by using inner tires as pressurizedreservoirs of CO 2. Anderson and Yee (1995) describe similar combined devicesused for Simulium spp in America.More recent improvements have come principally from the use of improvedformulations of attractant chemicals. These range from CO 2to aldehydes, ketones,and octenols. Octenols have proved especially useful, partly because theyare only slowly volatile and so a long release is possible from an impregnated

134 CHAPTER 6lure (Hall et al. 1984). Blackwell et al. (1996) used 1-octen-3-ol, a component ofthe body odor of ruminant mammals, to attract Culicoides midges. Theycombined field trials with laboratory Y-tube preference tests and electroantennograms,to provide a full picture of the effectiveness of this chemical.Typically this chemical is attractive at low to moderate concentrations but becomesrepellent at high concentrations.In this study the chemicals were used in “Delta” traps, which are open, triangularsticky traps, but the effectiveness of different traps is discussed below.Pheromone trapsThe attractiveness of virgin female insects to males, especially in Lepidoptera,has long been known, and even in the 1930s attempts were made to use crudeextracts of female gypsy moth Lymantria dispar abdomens to attract males. Thesewere largely unsuccessful, but living female moths have often been used aslures (e.g. for attracting male spruce budworm moths Choristoneura fumiferana;Miller & McDougall 1973). Once the attractive chemicals were identified andcould be synthesized, as began in the 1960s and 1970s, it became possible tomake effective pheromone traps (Fig. 6.5). These are now used routinely toattract males to killing lures, so as to reduce pest populations; to swamp wildfemale attractiveness, so as to disrupt mating; and to monitor pest populationsby detecting males as soon as they appear (Cardé & Minks 1995).An early discovery was that the female attractants are usually a subtle blendof varying concentrations of several chemicals; only rarely, as in the gypsymoth, are single chemicals used (Cardé & Baker 1984). This complexity hasmade the production of effective lures much more difficult, and, althoughsuper-effective formulations have occasionally been produced, a wild femalestill generally outperforms the chemical lure.Typically the chemical lure is impregnated into a “plug,” which is then heldclose to sticky trap surfaces, or above a collecting box (e.g. Sanders 1988).A common problem of sticky traps is that they become saturated, so that thecatch is only proportional to numbers of males present in the local environmentwhen these numbers are low. To overcome this problem large collecting chamberswith an insecticide, or water-filled chambers, are sometimes used, butthese are more expensive and more complicated to use than simple sticky surfaces(Kendall et al. 1982).The study of Keil et al. (2001) illustrates the problem that, whereas males canbe caught at pheromone lures, the capture of females needs a different technique.They marked the orchard pest Cydia pomonella by incorporation of a reddye in the larval diet, released the marked moths and then re-caught the malesin pheromone traps and the females using light traps.The design of the trap has been very varied, for it has been found that evenminor variations can influence the catch of a species significantly. This has resultedin a plethora of commercial designs. For some pest species it is now

INSECTS IN FLIGHT 135Fig. 6.5 Simple pheromone trap. Pheromone is released from the “wicked” tube and insectsare caught on the sticky interior surface of the trap.known what design is to be preferred, but often a sampling program merely hasto make use of an easily available and cheap trap such as the Delta or Pherocon®design. The precise height and location of the trap also influences catches significantly,so preliminary trials are recommended to investigate optimal trappingconditions. Some of this variation is associated with the way that thevolatile pheromone spreads away from the trap in a plume. These plumes can bevisualized by releasing visible substances, such as smoke or soap bubbles, fromthe lure site and assuming that these behave in the same way as the pheromonechemical. The wind speed and its constancy, as well as the presence of obstacles,affects the plume and this alters the attractiveness of the lure (Elkinton et al.1987).The concentration of the attractive chemical also affects the behavior of male

136 CHAPTER 6moths, and hence whether they are ultimately trapped. Typically very low concentrationslead to increased alertness, slightly higher concentrations lead towalking activity, higher still induce flight, but atypically high concentrationsmay be repellent (Cardé & Charlton 1984). The flight tends to be upwind in thepheromone plume but erratically cross-wind if the plume is lost. The distance ofattraction can be tens of meters in favorable conditions for large moths but onlya few meters for small Diptera (David et al. 1983).The use of pheromone trapsPheromone traps are used to determine whether a pest species is present in ageographical area, to confirm whether immigration or emergence has happened,to trap and kill pests, and/or to monitor population levels. For the latteruse particularly it is essential to know how trap catches relate to actual populationlevels.Various studies have related numbers caught at pheromone lures with eitherdirect counts of larvae or pupae, or with light-trap catches. In general very goodrelationships have been found between catches and actual numbers at lowpopulation levels (Speight & Wainhouse 1989). Frequently males are caughteven when population levels are too low to be detected in any other way. However,as natural population levels increase, pheromone trap catches begin tolevel off and they become non-representative at moderate to high levels. This isbecause of a range of factors, including physical trap saturation and principallythe swamping effect of many wild females (Croft et al. 1986). At high populationlevels light traps perform significantly better than pheromone traps. Ofcourse this also applies at times when the insects are not mating, perhaps in aninitial immature period or before hibernation. Barbour (1987) compared therelationship between pheromone trap catches of the pine beauty moth Panolisflammea and absolute counts of pupae in standardized soil areas. He found thathe had to make an allowance for the patchy nature of the pupal distributionand for the mobility of the moths but that, after these adjustments, regressionanalysis yielded relatively high r 2 values (e.g. 61 percent), indicating that thepheromone traps were providing reasonable representation of the actual mothnumbers.Numerous studies have now been carried out using pheromone lures, andmany successful control programs depend on them (Ridgway et al. 1990). Wellworkedexamples include that of the pink bollworm Pectinophora gossypiella inthe USA, Egypt, and Pakistan (Campion et al. 1989) and the spruce budwormChoristoneura fumiferana in the USA and Canada (Silk & Kuenen 1988).In studies where initial detection of a pest’s presence, or the attainment ofa threshold level requiring a pest control program, is what is required, thenpheromone traps are ideal, and their importance in trapping insects cannot beover-stated. If a representation of the full range of population levels is requiredthen they are less useful.

INSECTS IN FLIGHT 137Interception trapsThere is an essential difference between those traps that are undetectedby insects and act by literally intercepting their flight and those that aredetected but act merely as a screen, trapping insects that land and crawlon them. True interception traps do not attract insects and may provide anunbiased estimate of the true population flying in an area. In this they differfundamentally from attractive traps and are more analogous to suction trapsor direct netting.Undetected trapsUndetected interception traps are usually a series of sheets of a fully transparentmaterial, these days often Perspex or acrylic, that are set out either at random,or close to a reference point. Such traps are called window traps and havebeen used to catch a wide range of insects, especially including bark beetles(Canaday 1987, Young & Armstrong 1994) and dispersing ground beetles(van Huizen 1977). The sheets are placed above a water reservoir, intowhich the insects drop and are retained, especially if a drop of detergent or oil isadded to the water. By using angled Perspex sheets, often in a cross shape, it ispossible to detect the direction of flight. Sometimes an object of investigation,such as a dead tree, may be surrounded by window traps, so allowing the relationshipbetween wind direction and insect approach to be determined (Tunsetet al. 1988).In practice raindrops and/or dust soon adhere to window traps, however,rendering them visible and so reducing their unbiased activity.Other undetected interception traps are fast moving nets, including tow netsand sweep nets. Sweep nets are very widely used to dislodge and collect insectsfrom relatively long ground vegetation. They are of limited comparative use,because of the factors discussed below, but they do provide a quick, simple, andsometimes acceptable indication of the relative abundance of some of the insectsflying close to the ground layer. For example, Banks and Brown (1962)found less than 10 percent variation in catch between replicated sets of sweepsin wheat fields. The efficacy of a sweep varies with many factors, however, includingthe height, size, and power of the netting stroke; length, nature, anddensity of the vegetation; whether the vegetation is wet; weather conditions;time of day; and the “holding” power of different insects. Despite this, sweepnettingis still widely used to provide quick estimates of abundance, and studiesto determine the influence of the confounding factors have been carried outover many years (e.g. Gray & Treloar 1933, Rudd & Jensen 1977, Cherrill &Sanderson 1994).Tow nets may be towed by planes (e.g. Reling & Taylor 1984) or trucks (e.g.Bidlingmayer 1974), or may be rotated on long arms, although the last hardlyacts as a tow net, since the airspeed is usually too low. The advantage of these

138 CHAPTER 6tow nets is that very large volumes of air are sampled, leading to relatively largecatches of even scarce insects, but the nets are too cumbersome to be restrictedto individual components of the environment. The exception is that airbornenets can provide an estimate of insects that are dispersing well above the vegetationlayer. Although studies using such nets are frequently directed at specifictypes of insects, nevertheless they do catch a wide range of species, fittingthem to faunal surveys.Visible interception trapsMost interception traps are detected by the insects they catch, but are supposedlyneutral in their effect, not being strongly attractive. This assumption iseasily challenged, however, even in the case of the most widely used of all suchtraps, the Malaise trap (Malaise 1937), and Roberts (1972) showed that thecolor, size, and placement of the trap all influence catches.The Malaise trap (see Chapter 4) is a “tent” of various types of material,arranged so that insects are led up into the trap’s inner corner, at which is placeda non-return collecting jar. The original design has been modified several timesand is often used in conjunction with an attractant bait, such as CO 2(e.g.Coupland 1990). Studies have shown that Malaise traps are far from random intheir catch, and that even different genera of flies within one family show differentresponses. Tallamy et al. (1976) found that Chrysops horse flies were notcaught easily, whereas Tabanus were frequent visitors. However, such traps arestill frequently used to make generalized catches where this is appropriate. Forexample, Petersen et al. (1999) successfully studied the dispersion of Plecopteraand Trichoptera from a stream using a series of Malaise traps.Other more directional “funnel” traps have been designed to detect movementpatterns in various insects, including mosquitoes in Africa (Gillies et al.1978).Overall, interception traps do have a role in specific, planned studies, wheretheir apparent lack of bias can be tested, and also in the production of faunallists. However, it is often difficult to interpret their catches. Recently they havefrequently been combined with an attractant lure, to produce a more focusedresult.ConclusionsThis chapter reviews some commonly used techniques, but an essential messageis that virtually every study has to use a modified technique. In a review ofthe last 30 papers published in a representative journal, namely Ecological Entomology,it was found that not one merely used a “standard” widely used method.Most incorporated elements of a general technique, such as sweep-netting,but all field work had been designed specifically for its study. In all cases, a

INSECTS IN FLIGHT 139preliminary period had been used to develop an appropriate method. Furthermore,methods are becoming ever more specific in their catches, except whenthey are deliberately chosen to be wide-ranging, and the improved chemicalformulae used in pheromone traps illustrate this trend. It seems likely that advancesin trapping techniques will come from better observation of the behavioralresponses of the target insect, rather than from more modern technology,although miniaturization of components and improved battery life are boundto be important. As Southwood and Henderson (2000) comment in their preface,most of the trapping techniques were designed years ago and have stoodthe test of time, essentially unaltered, apart from the fine-tuning needed foreach individual study.ReferencesAnderson, J.R. & Yee, W.C. (1995) Trapping blackflies (Diptera: Simuliidae) in northernCalifornia.1. Species composition and seasonal abundances on horses, host models and ininsect flight traps. Journal of Vector Ecology 20, 7–25.Anon. (1991) The Potential Effects of Climate Change in the United Kingdom. First Report of theUnited Kingdom Climate Change Impacts Review Group. HMSO, London.Bale, J.S., Harrington, R., & Howling, G.G. (1992) Aphids and winter weather. I. Aphids andclimate change. In Proceedings of the Fourth European Congress of Entomology and the XIII InternationaleSymposium für die Entomofaunistik Mitteleuropas, Volume 1, pp 139–143. HungarianNatural History Museum, Budapest.Banks, C.J. & Brown, F.S. (1962) A comparison of methods of estimating population density ofadult sunn pest, Eurygaster integriceps Put. (Hemiptera, Scutelleridae) in wheat fields. EntomologiaExperimentalis et Applicata, 5, 255–260.Barbour, D.A. (1987) Monitoring pine beauty moth by means of pheromone traps: the effect ofmoth dispersal. In Population Biology and Control of the Pine Beauty Moth (Panolis flammea) (ed.S.R. Leather, J.T. Stoakley, & H.F. Evans), pp. 49–56. Forestry Commission Bulletin 67,Edinburgh.Bidlingmayer, W.L. (1974) The influence of environmental factors and physiological stage onflight patterns of mosquitoes taken in the vehicle aspirator and truck, suction, bait and NewJersey light traps. Journal of Medical Entomology, 11, 119–146.Bidlingmayer, W.L. (1975) Mosquito flight paths in relation to the environment: effect of verticaland horizontal visual barriers. Annals of the Entomological Society of America, 68, 51–57.Bidlingmayer, W.L. & Hem, D.G. (1981) Mosquito flight paths in relation to the environment:effects of forest edge upon trap catches in the field. Mosquito News, 41, 55–59.Blackwell, A., Dyer, C., Mordue (Luntz), A.J., Wadhams, L.J., & Mordue, W. (1996) Therole of 1-octen-3-ol as a host-odour attractant for the biting midge, Culicoides impunctatusGoetghebuer, and interactions of 1-octen-3-ol with a volatile pheromone produced byparous female midges. Physiological Entomology, 21, 15–19.Bowden, J. & Church, B.M. (1973) The influence of moonlight on catches of insects in lighttraps in Africa. II. The effect of moon phase on light-trap catches. Bulletin of Entomological Research,63, 129–142.Bowden, J. & Morris, M. (1975) The influence of moonlight on catches of insects in light-trapsin Africa. III. The effective radius of a mercury-vapour light-trap rid the analysis of catchesusing effective radius. Bulletin of Entomological Research, 65, 303–348.

140 CHAPTER 6Campion, D.G., Critchley, B.R., & McVeigh, L.J. (1989) Mating disruption. In Insect Pheromonesin Plant Protection (ed. A.R. Jatsum & R.F.S. Gordon). Wiley, Chichester.Canaday, C.L. (1987) Comparison of insect fauna captured in six different trap types in aDouglas-fir forest. Canadian Entomologist, 119, 1101–1108.Cardé, R.T. & Baker, T.C. (1984) Sexual communication with pheromones. In Chemical Ecologyof Insects (ed. W.J. Bell & R.T. Cardé), pp. 355–383. Chapman & Hall, London.Cardé, R.T. & Charlton, R.E. (1984) Olfactory sexual communication in Lepidoptera. Strategy,sensitivity and selectivity. In Insect Communication (ed. T. Lewis), pp. 241–265. AcademicPress, London.Cardé, R.T. & Minks, A.K. (1995) Control of moth pests by mating disruption: successes andconstraints. Annual Reviews of Entomology, 40, 559–585.Chapman, J.W., Reynolds, D.R., Smith, A.D., Riley, J.R., Padgley, D.E., & Woiwod, I.P. (2002)High altitude migration of the diamondback moth Plutella xylostella to the UK: a study usingradar, aerial netting, and ground trapping. Ecological Entomology, 27, 641–650.Cherrill, A.J. & Sanderson, R.A. (1994) Comparison of sweep-net and pitfall trap samples ofmoorland Hemiptera: evidence for vertical stratification within vegetation. Entomologist,113, 70–81.Coon, B.F. & Rinicks, H.B. (1962) Cereal aphid capture in yellow baffle trays. Journal of EconomicEntomology, 55, 407–408.Coppedge, J.R., Ahrens, E.H., & Snow, J.W. (1978) Swormlure-2 baited traps for detection ofnative screwworm flies. Journal of Economic Entomology, 71, 573–575.Coupland, J.B. (1990) The ecology of black flies (Diptera: Simuliidae) in the Scottish Highlandsin relation to control. PhD thesis, University of Aberdeen.Coupland, J.B. (1994) Factors influencing nuisance blackfly (Diptera: Simuliidae) activity inthe Scottish Highlands. Medical and Veterinary Entomology, 8, 125–132.Croft, B.A., Knight, A.L., Flexner, J.L., & Miller, R.W. (1986) Competition between caged virginfemale Argyrotaenia citrana (Lepidoptera: Tortricidae) and pheromone traps for capture ofreleased males in semi-enclosed courtyard. Environmental Entomology, 15, 232–239.Dafni, A., Bernhardt, P., Shmida, A., Iruri, Y., Greenbaum, S., & O’Toole, C. (1990) Red bowlshapedflowers: convergence for beetle pollination in the Mediterranean region. Israel Journalof Botany, 39, 81–92.David, C.T., Kennedy, J.S., & Ludlow, A.R. (1983) Finding a sex pheromone source by gypsymoths released in the field. Nature, 303, 804–806.Davies, J.B. (1978) Attraction of Culex portesi Senevet & Abonnenc and Culex taeniopus Dyar &Knab (Diptera: Culicidae) to 20 animal species exposed in a Trinidad forest. Bulletin of EntomologicalResearch, 68, 707–719.Dean, D.A. & Sterling, W.L. (1985) Size and phenology of ballooning spiders at 2 locations ineastern Texas (USA). Journal of Arachnology, 13, 111–120.Deansfield, R.D., Brightwell, R., Onah, J., & Okolo, C.J. (1982) Population dynamics ofGlossina morsitans submorsitans Newstead, and G. tachinoides Westwood (Diptera: Glossinidae)in sub-Sudan savanna in northern Nigeria. I. Sampling methodology for adults and seasonalchanges in numbers caught in different vegetation types. Bulletin of Entomological Research,72, 175–192.Disney, R.H.L., Erzinclioglu, Y.Z., Henshaw, D.J. de C., et al. (1982) Collecting methods andthe adequacy of attempted fauna surveys, with reference to the Diptera. Field Studies, 5,607–621.Downey, J.E. (1962) Mosquito catches in New Jersey Mosquito traps and ultra-violet lighttraps. Bulletin of the Brooklyn Entomological Society, 57, 61–63.Edwards, P.B., Kettle, D.S., & Barnes, A. (1987) Factors affecting the numbers of Culicoides(Diptera: Ceratopogonidae) in traps in coastal South-East Queensland (Australia) with

INSECTS IN FLIGHT 141particular reference to collections of Culicoides subimmaculatus in light traps. Australian Journalof Zoology, 35, 469–486.Elkinton, J.S., Schal, C., Ono, T., & Cardé, R.T. (1987) Pheromone puff trajectory upwind flightof male gypsy moth in the forest. Physiological Entomology, 12, 399–406.Finch, S. & Collier, R.N. (1989) Diptera caught on sticky boards in certain vegetable crops.Entomologia Experimentalis et Applicata, 52, 23–27.Gaydecki, P.A. (1984) A quantification of the behavioural dynamics of certain Lepidoptera inresponse to light. PhD thesis, Cranfield Institute of Technology.Gillespie, D.R. & Quiring, R. (1987) Yellow sticky traps for detecting and monitoring greenhousewhitefly (Homoptera: Aleyrodidae) adults on greenhouse tomato crops. Journal ofEconomic Entomology, 80, 675–679.Gillies, M.T. (1974) Methods for assessing the density and survival of blood sucking Diptera.Annual Review of Entomology, 19, 345–362.Gillies, M.T., Jones, M.D.R., & Wilkes, T.J. (1978) Evaluation of a new technique for recordingthe direction of flight of mosquitoes (Diptera: Culcidae) in the field. Bulletin of EntomologicalResearch, 68, 145–152.Glasgow, J.P. (1961) The variability of fly-round catches in field studies of Glossina. Bulletin ofEntomological Research, 51, 781–788.Graham, H.M., Glick, F.A., & Hollingsworth, J.P. (1961) Effective range of argon glow lampsurvey traps for pink bollworm adults. Journal of Economic Entomology, 54, 788–789.Gray, H. & Treloar, A. (1993) On the enumeration of insect populations by the method of netcollection. Ecology, 14, 356–367.Gregg, P.C., Fit, G.P., Coombs M., & Henderson, G.S. (1994) Migrating moths collected intower-mounted light traps in northern New South Wales, Australia: influence of local andsynoptic weather. Bulletin of Entomological Research, 84, 17–30.Hall, D.R., Beevor, P.S., & Cork, A. (1984) 1-Octen-3-ol; a potent olfactory stimulant and attractantfor tsetse isolated from cattle odours. Insect Science and its Application, 5, 335–339.Harper, A.M. & Story, T.P. (1962) Reliability of trapping in determining the emergence periodand sex ratio of the sugar-beet root maggot Tetanops myopaeformis Röder (Diptera: Otitidae).Canadian Entomologist, 94, 268–271.Hartstack, A.W., Hollingsworth, J.P., & Lindquist, D.A. (1968) A technique for measuring trappingefficiency of electric insect traps. Journal of Economic Entomology, 61, 546–552.Haufe, W.O. & Burgess, L. (1960) Design and efficiency of mosquito traps based on visual responseto patterns. Canadian Entomologist, 92, 124–140.Heathcote, G.D. (1957) The comparison of yellow cylindrical, flat and water traps and ofJohnson suction traps, for sampling aphids. Annals of Applied Biology, 45, 133–139.Johnson, C.G. (1950) The comparison of suction trap, sticky trap and townet for the quantitativesampling of small airborne insects. Annals of Applied Biology, 37, 268–285.Johnson, C.G. & Taylor, L.R. (1955) The development of large suction traps for airborne insects.Annals of Applied Biology, 43, 51–61.Jones, V.P. (1988) Longevity of apple maggot (Diptera: Tephritidae) lures under laboratory andfield conditions in Utah. Environmental Entomology, 17, 704–708.Katsoyannos, B.L. (1987) Effect of color properties of spheres on their attractiveness forCeratitis capitata (Wiedmann) in the field. Journal of Applied Entomology, 104, 79–85.Keil, S., Gu, H., & Dorn, S. (2001) Response of Cydia pomonella to selection on mobility: laboratoryevaluation and field verification. Ecological Entomology, 26, 495–501.Kendall, D.M., Jennings, D.T., & Houseweart, M.W. (1982) A large capacity pheromonetrap for spruce budworm moths (Lepidoptera: Tortricidae). Canadian Entomologist, 114,461–463.Kirk, W.D. (1984) Ecologically selective coloured traps. Ecological Entomology, 9, 35–41.

142 CHAPTER 6Kring, J.B. (1970) Red spheres and yellow panels combined to attract apple maggot flies. Journalof Economic Entomology, 63, 466–469.Leong, J.M. & Thorp, R.W. (1999) Colour-coded sampling: the pan trap colour preferences ofoligolectic and nonoligolectic bees associated with a vernal pool plant. Ecological Entomology,24, 329–335.Leos Martinez, J., Granovsky, T., Williams, H.J., Vinson, S.B., & Burkholder, W.E. (1986) Estimationof aerial density of the lesser grain borer (Rhyzopertha dominica) (Coleoptera:Bostrichidae) in a warehouse using dominicalure traps. Journal of Econonomic Entomology, 79,1134–1138.Linton, Y.M. (1998) Characterisation of the South African Culicoides imicola (Kieffer, 1913)species complex, and its phyllogenetic status in Europe. PhD thesis, University of Aberdeen.Malaise, R. (1937) A new insect trap. Entomologisk Tidskrift, 58, 148–160.McCreadie, J.W., Colbo, M.H., & Bennett, G.F. (1984) A trap design for the collection ofhaematophagous Diptera from cattle. Mosquito News, 44, 212–216.McGeachie, W.J. (1987) The effect of air temperature, wind vectors and nocturnal illuminationon the behaviour of moths at mercury vapour light traps. PhD thesis, Cranfield Instituteof Technology.Meyerdirk, D.E., Hart, W.G., & Burnside, J. (1979) Evaluation of a trap for the citrus blackflyAleurocanthus woglumi (Homoptera: Aleyrodidae). Canadian Entomologist, 111, 1127–1129.Miller, C.K. & McDougall, G.A. (1973) Spruce budworm moth trapping using virgin females.Canadian Journal of Zoology, 51, 853–858.Muirhead-Thomson, R.C. (1991) Trap Responses of Flying Insects. Academic Press, London.Mulhern, T.D. (1942) New Jersey mechanical trap for mosquito surveys. New Jersey AgriculturalCollege Experimental Station Report, 421, 1–8.Murphy, W.L. (1985) Procedure for the removal of insect specimens from sticky-trap material.Annals of the Entomological Society of America, 78, 881.Nag, A. & Nath, P. (1991) Effect of moonlight and lunar periodicity on the light trap catches ofcutworm Agrotis ipsilon (Hufn.) moths. Journal of Applied Entomology, 111, 358–360.Petersen, I., Winterbottom, J.H., Orton, S., et al. (1999) Emergence and lateral dispersal of adultPlecoptera and Trichoptera from Broadstone Stream, UK. Freshwater Biology, 42, 401–416.Phelps, R.J. (1968) A falling cage for sampling tsetse flies (Glossina: Diptera). Rhodesia Journalof Agricultural Research, 6, 47–53.Ramaswamy, S.B. & Cardé, R.T. (1982) Nonsaturating traps and long-life attractant lures formonitoring spruce budworm males. Journal of Economic Entomology, 75, 126–129.Rawlings, P., Meiswinkel, R., Labuschange, K., Welton, N., Baylis, M., & Mellor, P.S. (2003)The distribution and species characteristics of the Culicoides biting midge fauna of SouthAfrica. Ecological Entomology, 28, 559–566.Reling, D. & Taylor, R.A.J. (1984) A collapsible tow net used for sampling arthropods by airplane.Journal of Economic Entomology, 77, 1615–1617.Ridgway, R.L., Silverstein, R.M. & Inscoe, M.N. (eds) (1990) Behaviour-Modifying Chemicals forPest Management: Applications of Pheromones and other Attractants. Dekker, New York.Roberts, I. (1996) The efficiency of light traps for moths in relation to meteorological conditions.BSc Thesis, Aberdeen University.Roberts, R.H. (1972) The effectiveness of several types of Malaise traps for the collection ofTabanidae and Culicidae. Mosquito News, 32, 542–547.Robinson, G.S. & Tuck, K.R. (1993) Diversity and faunistics of small moths (Microlepidoptera)in Bornean rainforest. Ecological Entomology, 18, 385–393.Rogers, D.J. & Smith, D.T. (1977) A new electric trap for tsetse flies. Bulletin of Entomological Research,67, 153–159.Rudd, W.G. & Jensen, R.L. (1977) Sweep net and ground cloth sampling for insects in soybeans.Journal of Economic Entomology, 70, 301–304.

INSECTS IN FLIGHT 143Sanders, C.J. (1988) Monitoring spruce budworm population density with six pheromonetraps. Canadian Entomologist, 120, 175–183.Schaefer, G.W., Bent, G.A., & Allsopp, K. (1985) Radar and opto-electronic measurements ofthe effectiveness of Rothamsted Insect Survey suction traps. Bulletin of Entomological Research,75, 701–715.Service, M.W. (1971) Flight periodicities and vertical distribution of Aedes cantans (Mg), Aedesgeniculatus (01), Anopheles plumbeus Steph, and Culex pipiens L. (Diptera: Culicidae) in southernEngland. Bulletin of Entomological Research, 60, 639–651.Silk, P.J. & Kuenen, L.P.S. (1988) Sex pheromones and behavioural biology of theconiferophagous Choristoneura. Annual Reviews of Entomology, 33, 83–101.Smith, R.J. & Merrick, M.J. (2001) Resource availability and population dynamics ofNicrophorus investigator, an obligate carrion breeder. Ecological Entomology, 26, 173–180.Southwood, T.R.E. and Henderson, P.A. (2000) Ecological Methods. 3rd edn. Blackwell, Oxford.Speight, M.R. & Wainhouse, D. (1989) Ecology and Management of Forest Insects. Oxford UniversityPress, Oxford.Tallamy, D.W., Hansens, E.J., & Denno, R.F. (1976) A comparison of Malaise trapping andserial netting for sampling a horsefly and deerfly community. Environmental Entomology,5, 788–792.Taylor, L.R. (1951) An improved suction trap for insects. Annals of Applied Biology, 38, 582–591.Taylor, L.R. (1962) The absolute efficiency of insect suction traps. Annals of Applied Biology, 50,405–421.Taylor, L.R., Kempton, R.A., & Woiwod, I.P. (1976) Diversity statistics and the log-series model.Journal of Animal Ecology. 45, 255–272.Taylor, R.A.J. (1986) Time series analysis of numbers of Lepidoptera caught at light traps inEast Africa, and the effect of moonlight on trap efficiency. Bulletin of Entomological Research,76, 593–606.Thompson, D.V., Capinera, J.L., & Pilcher, S.D. (1987) Comparison of an aerial water-panpheromone trap with traditional trapping techniques for the European cornborer(Lepidoptera: Pyralidae). Environmental Entomology, 16, 154–158.Tunset, K., Nilssen, A.C., & Anderson, J. (1988) A new trap design for primary attraction ofbark beetles and bark weevils (Coleoptera, Scolytidae and Curculionidae). Journal of AppliedEntomology, 106, 266–269.Vale, G.A. (1974) The responses of tsetse flies (Diptera: Glossinidae) to mobile and stationarybaits. Bulletin of Entomological Research, 64, 545–588.Van Huizen, T.H.P (1977) The significance of flight activity in the life cycle of Amara plebeja Gyll.(Coleoptera, Carabidae). Oecologia, 29, 27–41.Vessby, K. (2001) Habitat and weather affect reproduction and size of the dung beetle, Aphodiusfossor. Ecological Entomology, 26, 430–435.Vogt, W.G., Woodburn, T.L., Morton, R., & Ellem, B.A. (1983) The analysis and standardisationof trap catches of Lucilia cuprina (Widemann) (Diptera: Calliphoridae). Bulletin of EntomologicalResearch, 73, 609–617.Waring, P. (1980) A comparison of the Heath and Robinson M.V. moth traps. Entomologists’Record and Journal of Variation, 92, 283–289.Waring, P. (1990) Abundance and diversity of moths in woodland habitats. PhD thesis, OxfordBrookes University.Webb, R.E., Smith, F.F., & Affeld, H. (1985) Trapping greenhouse whitefly with coloured surfaces;variables affecting efficacy. Crop Protection, 4, 381–393.Williams, C.B. (1948) The Rothamsted light trap. Proceedings of the Royal Entomological Society ofLondon A, 23, 80–85.Woiwod, I.P. & Hanski, I. (1992) Patterns of density dependence in moths and aphids. Journalof Animal Ecology, 61, 619–629.

144 CHAPTER 6Woiwod, I.P. & Harrington, R. (1994) Flying in the face of change: the Rothamsted Insect Survey.In Long-term Experiments in Agricultural and Ecological Sciences (ed. R.A. Leigh & A.E. Johnston),pp. 321–342. CAB International, London.Young, M.R. (1997) The Natural History of Moths. T. & A.D. Poyser, London.Young, M.R. & Armstrong, G. (1994) The effect of age, stand density and variability on insectcommunities in native pine woodlands. In Our Pinewood Heritage (ed. J.R. Aldhous), pp.201–221. FC, RSPB, SNH, Edinburgh.Zhou, X., Perry, J.N., Woiwod, I.P., & Harrington, R. (1997) Detecting chaotic dynamics of insectpopulations from long-term survey data. Ecological Entomology, 22, 231–241.Index of methods and approachesMethodology Topics addressed CommentsLight traps Faunal surveys and general Catches only nocturnal insects.biodiversity assessment. Different light sources are attractiveAccumulation of generalized to different insect types.long-term data sets.Greater catches result from highPartly focused survey of contrast between light andselected groups.background.Catches affected by moon phase andby some meteorological factors.Catches of small insects increasedwhen suction trap is added.Catches at best semi-quantitative.Suction traps General faunal surveys, Power of fan and detailed design ofespecially of smaller insects. nozzle affects catch type and size.Partly focused surveys of Often used in conjunction with otherselected groups.traps.Standardized, often long Standardized designs have producedterm catches.well-replicated catches.Catches can be calibrated andpredicted to some extent.Catch much affected by localconditions and weather.Water (or pan) Faunal surveys and general Catch much affected by color of trap.traps biodiversity assessment. Catch much affected by localWide-ranging, easilyconditions and weather.replicated sampling designs. Difficult to design so that catch isPartly focused surveys (only focused on particular groups.by careful design of trap). Catches many “tourist” species andhas high “by-catch.”Continued

INSECTS IN FLIGHT 145Methodology Topics addressed CommentsVery easily and cheaply replicated.Very difficult to calibrate catches.Sticky traps General faunal surveys and Easily and cheaply replicated, oftenbiodiversity assessment, commercially available.especially of small species. Precise trap design affects catchPartly focused sampling, by greatly.careful trap design.Often used in association with otherFrequently aimed at pest trap types.species, aimed to reduce Catch is often damaged by stickypopulation or detectsubstance used.presence.Traps can become saturated orclogged with dust.Very difficult to calibrate catches.Baited traps Single species may be caught Species specificity possible withby use of correct pheromone correct bait (often pheromone orbait.other chemical attractant).Functional group (e.g.May be sex-specific (e.g. malescarrion feeders) caught by attracted by female pheromone).correct bait.Precise condition of bait may affectOften aimed at pest species, catch type and quantity.aimed to reduce population Baits may be whole animals.or detect presence.Often used against pests, so trapsmay be commercially available.Often used in conjunction withanother trap type to ensure efficientcatch.May catch at densities well belowdetection level of other traps.May be partially calibrated.Interception Faunal surveys or If clean and well designed may offertraps generalized biodiversity truly general catch.assessment.Transparent sheets quickly loseAt best may be unbiased effectiveness if wet or dirty.collection of all flyingMalaise traps may be biased in catch.insects.Difficult to calibrate effectively.

CHAPTER 7Techniques and methods forsampling canopy insectsCLAIRE M.P. OZANNEIntroductionThe forest canopy has been described as the last biological frontier (Erwin 1983,Lowman & Wittman 1995). Whether this designation is justified or not, muchremains to be discovered about this significant terrestrial habitat. Forest biomescurrently extend across 25 percent of the world’s land surface (FAO 1999).Tropical moist forests cover approximately 7 percent of this area yet are estimatedto support 75 to 85 percent of all insect species, described and undescribed(Hammond 1992). Recent research suggests that up to 50 percent offorest insect species can be found in the canopy, with the percentage of truecanopy specialists lying between 7 and 13 percent (Collembola: Rodgers &Kitching 1998; Coleoptera: Hammond et al. 1997; the figure is slightly higherfor mites at 17–18%: Winchester 1997, Walter et al. 1998). This stratum of theforest habitat is therefore of great importance to global biodiversity (Ozanneet al. 2003).More importantly, however, canopy insects contribute significantly to anumber of fundamental forest ecosystem processes. These processes includenutrient cycling by herbivores (Schowalter et al. 1981), decomposition byleaf-surface and suspended-soil arthropods (Nadkarni & Longino 1990),predator–prey interactions (Winchester 1997, McGeoch & Gaston 2000), andthe pollination and dispersal of forest plants and epiphytes (Aizen & Feinsinger1994, Marini-Filho 1999). Forest canopies also provide us with the opportunityto test an array of hypotheses that attempt to explain population and communitydynamics at a wide range of scales — from that of the individual leaf orneedle (centimeters), through branch and tree (meters), to plantation stand,catchment area, and whole forest (kilometers).Until Erwin’s pioneering work in the late 1970s in the tropics, Crossley andSchowalter’s work in the USA, and Southwood’s work in Europe in the early1980s, little attempt had been made to investigate canopy arthropods in a quantitativemanner (Erwin 1982, Crossley et al. 1976, Schowalter et al. 1981,Southwood et al. 1982; but see Martin 1966). The exceptions to this were a fewindividual species regarded as a threat to timber production (see Speight &Wainhouse 1989). The establishment of several canopy science networks since1994 (e.g. the International Canopy Network (ICAN), the European Science146

TECHNIQUES AND METHODS 147Foundation Tropical Canopy Research Programme, and the Global Canopy Programme)have stimulated a notable rise in research (Nadkarni & Parker 1994,Stork & Best 1994, Mitchell 2001), which is now being carried out in a range oftropical and temperate managed and planted forests. Work on canopy arthropodsin now being carried out in western Canada (Winchester & Ring 1996a),through the USA (Schowalter & Ganio 1998) and South America (Adis et al.1998b, Basset & Charles 2000), across Europe (Schubert & Ammer 1998,Ozanne 1999), Africa (Moran et al. 1994, Wagner 2000, Winchester & Behan-Pelletier 2003), and India (Devy 1998), to Japan (Watanabe 1997), SoutheastAsia (Guilbert 1998, Floren & Linsenmair 2000), Australia, and New Zealand(Lowman et al. 1996, Didham 1997, Kitching et al. 2000a, Majer et al. 2000).Much of the research published in the literature is aimed at reporting basic informationon canopy communities — addressing issues of species distribution,population densities, and community structures. Only very recently have morecomplex questions about the role of arthropods in ecosystem function, spatialdistribution, and response to human impact begun to be addressed. There arethree reasons that account for the historical lack of good data on canopy arthropods:firstly the challenge of accessing the canopy without significant disturbance,secondly the sheer species richness and complexity of many canopycommunities, and thirdly the difficulties of sampling with appropriate replicationand experimental design (partly due to accessibility difficulties, richness,and heterogeneity).Access to the canopyThere is a growing literature on canopy access techniques (Heatwole & Higgins1993, Dial & Tobin 1994, Moffet & Lowman 1995, Mitchell et al. 2002) whichoften divides the methods into “high tech” and “low tech” according to theequipment and cost (Barker 1997). Some of the sampling methods that are discussedin this chapter can be utilized from the ground and therefore do notstrictly require access (e.g. chemical knockdown and branch clipping). However,in high canopies (e.g. over 20 m) they may be made more effective by operatingthem from within the canopy. Other sampling methods described herespecifically require the operator to be in the canopy itself (Basset et al. 2003a).One of the simplest access techniques allows the researcher to climb directlyinto the canopy using a single rope and harness — the single rope technique, orSRT (Barker & Standridge 2002). This method of access has the benefit that aclimber can get into the trees with the minimum of disturbance, although it isessential that appropriate training is undertaken to ensure safety. The equipmentrequired is less expensive and more transportable than in other accesstechniques and it has the significant advantage that a greater number of treescan be used in the course of a study, thus permitting proper replication of samples.Trees can be rigged with ropes in several ways. Most methods involve

148 CHAPTER 7attaching the rope to a cord that can be used to pull the rope over a stablebranch. Fishing line that has been shot (or thrown; Dial & Tobin 1994) over thebranch pulls up the cord. Placing the line up into the right location in the canopyrequires some expertise and a great deal of practice.More permanent access techniques include platforms and walkways, towers,and cranes. Platforms and walkways have now been erected in a number oftemperate and tropical forests (Reynolds & Crossley 1995, Inoue et al. 1995,Ring & Winchester 1996) (Fig. 7.1). These can act as foci for multi-faceted researchprojects, and they are useful for detailed studies of a small area of canopyand for work in which a chronological sequence of samples is required from thesame location. Economically, walkways have the added attraction that they canbe used to combine research and eco-tourism.The most permanent of the canopy access structures is the canopy crane. Currentlythere are eight crane projects worldwide, based in the Pacific NorthwesternUSA (Wind River), Panama (two cranes, Panama City), Australia(Cape Tribulation, Queensland), Sarawak (Lambir Hills), Japan (Tomakomai,Hokkaido), Germany (Burgaue near Leipzig, Solling), and Switzerland (Hofstetten).These are large construction cranes consisting of a tower, boom, andsuspended gondola (Mitchell et al. 2002).Cranes have the advantage over some other access methods that, after the installationprocess is complete, non-destructive research can be carried out withFig. 7.1 Canopy walkway at the Forest Research Institute Malaysia (FRIM), Kepong, KualaLumpur, Malaysia. Photo: M.V. Graham.

TECHNIQUES AND METHODS 149minimal impact on canopy organisms. The cranes also allow researchers tostudy processes in situ, often requiring bulky analytical equipment which in thepast could only have been used in the laboratory (Bauerle et al. 1999). Thetower of the crane can be used to fix monitoring and trapping equipment, whilstthe gondola can move the researcher horizontally and vertically through andabove the canopy. The disadvantage of a crane lies in the fixed location, whichreduces the opportunity for replication of samples. Additionally, because thecost of purchase and installation necessitates preservation of the site for futurework, destructive or removal sampling is often discouraged, restricting thescope for entomological studies.The most dramatic method of accessing the canopy is undoubtedly via thecanopy raft (“radeau des cimes”) and its smaller companion the sledge or luge(Hallé & Blanc 1990). These large red inflatable and netting platforms (raft —hexagonal structure 580 m 2 ; luge — triangular 16 m 2 ) are carried over the forestby a dirigible and can be placed down onto the upper canopy surface. Researcherscan climb to the raft using ropes or be carried over the canopy surfaceto required points in the forest. A number of sampling techniques includingbranch clipping, Malaise traps, and sweeping can be used from the platforms,and researchers can walk about freely on the raft, enabling direct observation ofanimal–plant interactions — providing the researchers have a good head forheights!Sampling issuesThere are a number of sampling issues to be considered when designing a studythat involves collecting insects from the forest canopy. Firstly, canopy communitiesmay be composed of organisms drawn from a wide range of taxa, and theinsect assemblages may be particularly species-rich. The highly diverse insectsamples collected from tropical canopies have been used as a basis for estimatesof global species richness made initially by Erwin (1982), then May (1990), andmore recently Stork (1993). In addition, insects and other arthropods are oftenvery numerous (Table 7.1).Because of this diversity, a collecting event can yield samples that requireconsiderable time to clean out debris and to sort, even to a low taxonomic levelsuch as order. Since canopy insect communities are poorly known, samplingfrequently yields species new to science. The use of morpho-taxa (Oliver &Beattie 1996) to assist with identification is well established in tropical work,but even this approach requires considerable taxonomic expertise.The second issue is that the distribution of insects within the forest canopy isheterogeneous. Insects exhibit vertical, horizontal, and temporal variation inlocation & density (Costa & Crossley 1991, Hollier & Belshaw 1993, Springate& Basset 1996, Ozanne et al. 1997, Rodgers & Kitching 1998, Foggo et al.2001, Basset et al. 2003b). This heterogeneity may be the focus of

150 CHAPTER 7Table 7.1 Typical canopy insect densities for a range of tree species.Tree species Collection method Density ReferenceQuercus robur Pyrethrum 591.3 per m 2 Southwood et al. 1982knockdownEucalyptus Branch clipping 247.1 per kg leaf Abbott et al. 1992marginatabiomassPinus sylvestris Pyrethrum 1046.31 per m 2 Ozanne 1996knockdownAporusa Pyrethrum 220.3 per m 2 Floren and Linsenmairlagenocarpa knockdown 1997Tropical forest, Pyrethrum 117.4 per m 2 Stork 1991BruneiknockdownRain forest canopy, Branch clipping 16.56 per sample Basset et al. 1992Cameroon (approx. 0.85 m 2leaf area)investigation, but if it is not then steps need to be taken in the study design totake account of it. Thus the issues of representative sampling, adequate replication,and the avoidance of pseudoreplication need to be addressed (Hurlbert1984, Guilbert 1998).Finally, the range of taxa present and the complexity of the habitat mean thata study may require the use of several sampling techniques to collect the appropriatedata. No one technique can collect all groups of insects equally, andindeed many sampling methods (e.g. activity traps) have strong biases (Bassetet al. 1997). The most effective technique or combination of techniques must bechosen in the light of the research questions that the study is seeking to answeror the hypotheses to be tested.Chemical knockdownChemical knockdown is arguably the most effective, comprehensive, andreplicable of canopy sampling techniques (Stork & Hammond 1997, Majer et al.2000). Knockdown can be used to collect insects and other arthropods fromvegetation that spans the canopy height range from understory saplings andshrubs (Hill et al. 1990), to the upper sections of tropical emergent trees whichmay be 40–50 m from the ground (Adis et al. 1997, Paarmann & Stork 1987).Knockdown has been used successfully to collect insects from the completecanopy of individual trees (Floren & Linsenmair 1997), to investigate withincanopy variation (Kitching et al. 1993), and to study the spatial distribution of

TECHNIQUES AND METHODS 151organisms across forests (Ozanne et al. 2000). The technique can be used togather information on insect population densities and community structure,e.g. guild proportions and trophic structure (Kitching et al. 2002), and to collectlive specimens for subsequent experimental work on population dynamics andfeeding strategies (Paarmann & Kerck 1997).Knockdown is a passive sampling method and involves the delivery of a contactchemical that affects the insect nervous system — often temporarily. Themost commonly used chemicals induce repetitive axon firing, resulting in a lossof coordinated movement which causes insects to fall from the vegetation orfrom flight. Knockdown can be rapid, occurring in a matter of minutes,although differential rates of absorption of the chemical through the cuticlecan extend this time to more than an hour (Paarmann & Kerck 1997, Ozanneunpublished data).Two main techniques are employed to deliver the chemical to the canopy:fogging and misting. Both collect insects that are in flight through the canopy orare surface-dwellers on the leaves, flowers, fruit, twigs, branches, and trunk ofthe tree. Insects on the outer surfaces of epiphytes (vascular and non-vascular)may also be collected, as could those on the surface of suspended soils. Knockdownis not an entirely comprehensive sampling method. The technique willnot reliably collect insects that spin leaves together or that inhabit leaf domatiaand epiphytes or that bore into bark (Stork & Hammond 1997, Walter & Behan-Pelletier 1999). Thus, as with other techniques, its use should be appropriate tothe research questions being asked.FoggingFogging was first used for collecting arboreal arthropods by Roberts (1973).Fogging machines produce a hot cloud of chemical droplets that rises upwardsand outwards in a still air column, allowing the chemical to reach the heights requiredto sample rainforest emergents. The thermal fog is produced by allowingthe chemical to drip in a controlled manner onto a hot surface generated by theexhaust from the petrol-driven engine. There is a range of machines but themost commonly used in insect sampling are the Swingfog® and Dyna-Fog®versions.The fog can be delivered more reliably to the upper canopy by hoisting themachine into the canopy on a system of ropes and pulleys. However the foggermay be difficult to control once suspended and so this requires careful roperigging. Some research groups have developed a radio-control mechanismthat turns on the flow of chemical once the fogger has reached the appropriatelocation in the canopy (Adis et al. 1998a). Fogging is typically carried out for5–10 minutes in one location.The efficiency of collection is dependent on the environmental conditions.The inability to control the movement of the fog even in relatively calm conditionsis one of the greatest disadvantages of this technique. In turbulent air the

152 CHAPTER 7fog may not reach the canopy above the collecting trays and thus although insectsmay be knocked down they will not be collected and the wind may sweepaway falling insects. Fogging should therefore be carried out at dawn or duskwhen the air is still. If it is raining then fog will not rise and disperse in the requiredmanner and if the foliage is wet the chemical tends to pool on the leaves,reducing its effectiveness; insects stick to the foliage rather than falling.MistblowingThe second method used in knockdown sampling is mistblowing. The principlesinvolved here are quite different from those of fogging. The mistblowerconsists of a 2-stroke engine driving a fan that blows a strong air current alongthe delivery pipe. The chemical is allowed to drip into the air current at a ratecontrolled by the nozzle aperture, and as it hits the air stream the liquid issheared into small droplets and carried up into the canopy (Fig. 7.2).The height to which the mist reaches is determined in part by the power ofthe engine and fan and in part by the density of the foliage. Typically the mistreaches between 6 and 12 m (Southwood et al. 1982, Ozanne et al. 1988),although the mistblower can be hoisted up into the canopy in the same manneras a fogger to increase its range (Kitching et al. 2000b). The volume of chemicalused and the droplet size spectrum can be controlled such that mistblowers canbe set up for low-volume (LV: 20–300 l/ha) or ultra-low-volume delivery (ULV:Fig. 7.2 Hurricane Major mistblower (Cooper Pegler). Photo: M.R. Speight.

TECHNIQUES AND METHODS 1535–20 l/ha). Misting is usually carried out for only a few minutes (0.5–5 minutes)depending on the canopy volume and chemical flow rates. Machines most commonlyused are the Hurricane-Major® and the Stihl® backpack mistblowers.Similar factors to those affecting fogging influence the efficiency of thismethod. Wind or rain will reduce the effectiveness of sampling. However, ashort shower during the knockdown period after spraying can, ironically, increasethe catch as insects are washed off the foliage into the collecting trays(personal observation). The structure of the foliage influences the dispersal ofthe chemical within the canopy, affecting, for example, the amount of activeingredient reaching the upper and lower surfaces of leaves and or needles(Ozanne et al. 1988).Comparison of misting and fogging suggests that fogging results in knockdownover a much wider area, particularly downwind of the sample point. Thisis an important disadvantage in sensitive habitats and in areas where other treesor proximate locations are going to be sampled. The proportion of insects in differentgroups in the sample can vary with technique (M.R. Speight et al. unpublisheddata). This may be attributed partly to the method of chemicaldispersal and partly to the different chemicals commonly used with the differenttechniques (natural pyrethrum vs. synthetic pyrethroids). Misting seems tobe the most effective (M.R. Speight et al. unpublished data).The chemicals used in knockdown are mixtures of either natural pyrethrinsor synthetic pyrethroids. These are usually carried in an oil (e.g. kerosene), andin ULV delivery this is used undiluted, but in LV delivery an emulsion is madewith water. The chemical may be synergized by piperonyl butoxide if the aim isto kill the insects rapidly, but larger species are capable of recovering from aknockdown event.Natural pyrethrum has the advantage that it is inactivated by ultraviolet lightmore rapidly than synthetic equivalents (probably within 24–48 hours), a significantfactor when sampling in sensitive sites or when conducting recolonizationstudies which require repetitive sampling (e.g. Floren & Linsenmair 1997).Natural pyrethrum should be used if live specimens are required (Adiset al. 1997), but these are much more expensive than synthetics. In Europe andNorth America it is necessary to observe pesticide handling and application proceduresincluding health and safety regulations when using these chemicals.The time taken for insects to fall from the tree varies with their location, susceptibility,and size, but the majority of animals can be collected up after two hours(Ozanne 1991, Stork & Hammond 1997).The second element of the knockdown sampling system is the collecting trayor mat used to capture fallen insects. These have become more sophisticatedand therefore more efficient over the last 20 years, moving from large plasticsheets spread on the ground (Yamashita & Ishii 1976), through cloth traysstretched on wooden frames (Southwood et al. 1982), to conical hoops madefrom vinyl or tenting material (Ozanne 1996, Adis et al. 1998a, Kitching et al.2000b) (Fig. 7.3). Vinyl hoops work particularly well because they are robust

154 CHAPTER 7Fig. 7.3 Collecting hoops (Natural History Museum UK design). Photo: I.P. Palmer.and the surface is very shiny, allowing insects to roll down into the jar at theapex. Remaining insects can be washed or gently brushed into the jar, whichshould contain a small amount of preserving fluid (e.g. 70% ethanol).Current collecting hoops have been developed to reduce the handling ofspecimens, which are easily damaged (although the knockdown chemicalseems to produce autotomy in some long-legged insects anyway; Paarmann &Kerck 1997), and to prevent small insects and mites from being left behind.Hoops are of a standard surface area (0.5 or 1 m 2 ) to allow densities of insects tobe quantified per unit ground area, and they are hung under the canopy by clippingto branches, to a network of cords, or to a tower. Strong cord and large clipsallow the hoops to be easily handled without tangling. Problems may arise ifhoops are hung too early and catch debris from the canopy, or if the jar fills withrainwater during the collection period; some hoops have a built in storm vent.The specific placement of trays depends on the study design. In plantations theycan be set up under the canopy of several trees to reduce the effect of betweentreevariation (Ozanne 1996), while trays near to the trunk may have differentcatches from those at the crown margins. Trays can be attached to a tower at differentheights to investigate vertical distribution of species in the canopy (M.R.Speight, personal communication).Overall, knockdown compares favorably with other canopy samplingtechniques.1 Compared with beating, it collects higher densities (Fig. 7.4, Lowman et al.1996).

TECHNIQUES AND METHODS 155450400Number of insects per m 3350300250200150100500Beating Misting 1 Sweeping Misting 2Fig. 7.4 Mean densities of insects sampled in Australian sub-tropical rainforest using threecollection techniques: beating, sweeping, and pyrethrum misting (misting 1 = misting afterbeating; misting 2 = misting after sweeping) ±s.e. (from Lowman et al. 1996).2 Compared with sweeping, it collects higher densities (Fig. 7.4, Lowman et al.1996), and better estimates of collembolan and dipteran densities (Lowmanet al. 1996).3 Compared with branch clipping, it produces better estimates of the richness ofparasitic Hymenoptera (Blanton 1990); it estimates the density of large mobileinsects and cryptic insects less well (Majer & Recher 1988); it underestimatessessile insects e.g. Psyllidae (Majer & Recher 1988); it is not as good for biomassestimation (Blanton 1990).Branch bagging and clippingA viable alternative sampling strategy to chemical knockdown, and one preferredby a number of research groups, is branch bagging and clipping. Thismethod may be used at a wide range of canopy levels, limited only by the heightto which the mechanism can be operated accurately from the ground, or by thecanopy access technique used. The technique was first reported for canopysampling by Crossley et al. in 1976, and has been used in temperate forests toinvestigate vertical stratification of communities (Schowalter & Ganio 1998)and the diel movement of arthropods within the canopy (Ohmart et al. 1983),and in tropical forest to study plant–herbivore relationships (Basset & Höft1994). Branch bagging and clipping can be used to measure a number of populationand community attributes, including presence and absence of species,

156 CHAPTER 7population density, guild structure, and heterogeneity of distribution. Themethod can be used to standardize insect densities to units of plant biomass andsurface area (Basset et al. 1992, Ohmart et al. 1983).Branch clipping involves passing a mesh, cloth, or plastic bag over the endportion of a branch and then drawing the bag closed to prevent the escape ofmobile insects (although see Ohmart et al. (1983) where branches were clippedfirst and dropped into a calico bag). The branch section is then cut off and thesample brought to the ground. The bag and clippers are usually attached to longpoles or arms that can be fixed in length or telescopic, allowing them to bepushed up into the canopy from the ground (Basset & Höft 1994), alongbranches from a platform (Winchester & Ring 1996b, Winchester 1997), orfrom a cherry-picker (Majer et al. 1990). Shorter poles, affording more control,can be used when the clipping is carried out from a walkway or from the canopyraft or luge (Basset et al. 1992). The amount of branch and foliage cut down inany one sample varies from 20 to 60 leaves (Johnson 2000), through samples of2–5 g in weight (Schowalter et al. 1988), to larger samples up to 120 g in weight(Majer & Recher 1988).The selection of branches to be clipped is a key step in designing an effectiveinvestigation, and samples can be taken at random or from specified locationsto investigate particular microhabitats (Johnson 2000). Insects shaken off thefoliage into the bag can be counted live in situ (see Johnson (2000)), but frequentlythe samples are removed from the site for further study. In somestudies the bag is filled with CO 2or other chemical (e.g. pyrethroid spray orethyl acetate; Basset et al. 1992) before closure and in others the bag is chilled(Schowalter et al. 1981) to prevent escape on opening. Storage of the clippedsamples (perhaps in a cool environment to reduce mould growth) can allowinsects that are difficult to sample, e.g. dipteran larvae and pupae, to emergeas adults (I.P. Palmer, personal communication).Branch clipping is an excellent method for sampling sedentary insects onbranch and leaf surfaces. Comparative work indicates that the technique is ableto capture insects from all orders. Several studies suggest that large mobile insects,e.g. Odonata, are under-sampled (Cooper & Whitmore 1990, Johnson2000) but other groups do not have time to avoid the bag as it is drawn over thefoliage. The technique is not effective for sampling aerial components of thecanopy fauna such as midge clouds (Chironomidae) (Johnson 2000).Branch clipping has the advantage over many other canopy sampling techniquesthat the species richness or density of insects can be converted directlyto units of plant biomass and/or leaf and branch surface area (Schowalteret al. 1981, Abbott et al. 1992, Winchester & Ring 1996b). This can providevaluable data on herbivore loads and microhabitat preferences of insects in thecanopy. The technique can also be used to investigate epiphyte communities —particularly those of non-vascular epiphytes such as moss mats and lichens —and to collect insects that are leaf-spinners or that hide in deep bark crevices.The most important disadvantage is that it is the branch tips that are usually

TECHNIQUES AND METHODS 157clipped. This introduces a bias towards insects that are attracted to rapidly growingtissue often found at the apices of branches and a bias away from invertebratesinhabiting large branches and the trunk.Aerial and arboreal traps: Malaise, interception,emergence and lightActive and passive trap systems can be used to investigate insects moving withinthe forest and in the spaces in and around the canopy, such as above thecanopy surface, in gaps, or at edges. Traps that have been designed for use inother habitats (described in other chapters in this book), can be employed effectivelyabove the ground, although sampling efficiency may be affected by thelocation in which the traps are placed in the three-dimensional spaces of treecrowns. Appropriate placement will depend on the research question. Traps canbe used to answer general questions about canopy community structure or totest specific hypotheses about the use of particular strata of the canopy. For example,Compton et al. (2000) used sticky traps to sample the location andmovement of fig wasps within and above the canopy surface.With some modification, flight interception traps have also been used in thecanopy environment. Hill and Cermak (1997) describe a plastic window trapfitted with collecting trays at the base and a roof to keep out the rain. They usedthis apparatus to compare ground and canopy insect catches by hoisting sometraps up into the canopy, securing them with guy ropes to prevent twisting inthe wind. Traps were installed in locations that ensured foliage did not interferewith the capture surface. A wide range of arthropod groups was collected, withColeoptera, Diptera, and Hymenoptera dominating the samples.The most effective canopy traps can be built by combining the best features ofdifferent ground-based mechanisms. For example, combination Malaise andinterception traps have been designed for use in the canopy (Fig. 7.5). Springateand Basset (1996) used such a trap to investigate diel movement of insects withintree crowns. The apparatus consisted of a rectangular cross-panel of blacknetting (see Chapter 4 for Malaise trap design and discussion of effects of nettingcolor) with a white netting roof connected to a collecting jar. This part of the trapintercepts a range of insect groups including Diptera and Hymenoptera. A clearplastic funnel was also attached below the main body of the trap and connectedto a large collecting jar containing ethanol. This part of the apparatus acts as awindow trap, capturing those insects that close their wings and drop downwardson alighting such as the Coleoptera. In order to allow the trap to be leftout for long periods of time an overflow grid was inserted in the middle of thelower jar to cope with heavy rainfall.Combination traps are essentially activity-dependent and therefore underestimatethe contribution of sedentary and flightless arthropods to the community(Springate & Basset 1996). Their effectiveness is influenced by crown

158 CHAPTER 7Fig. 7.5 Combined Malaise and interception trap at the MASS site, Vancouver Island, BC,Canada. Photo: I.P. Palmer.structure and by their location in relation to insect flight paths. However,Behan-Pelletier and Winchester (1998) found that traps set in the canopy oftemperate rainforest in British Colombia caught significant numbers of flightlessarthropods (e.g. oribatid mites, Acarina), perhaps because they werecarried through the canopy by air currents or because they are activelymoving about within tree crowns.Light traps can also be used very effectively in the canopy to collect activelyflying insects. They are particularly efficient at sampling Lepidoptera andColeoptera, but also capture Hemiptera and other insect groups. Light trapshave been used in forests to investigate the impact of fragmentation on communities(Kitching et al. 2000b) and to investigate vertical distribution of moths(Intachat & Holloway 2000). Light traps generate three kinds of data:

TECHNIQUES AND METHODS 159presence/absence data for individual species, qualitative data for comparativework between sites, and relative estimates of population densities (Southwood& Henderson 2000). With careful calibration these relative estimates can bemade absolute. Estimates of population density are generated per unit trappingeffort, for example, per trap night (TNI: trap night index).The mostly commonly used light traps are Rothamsted tungsten-filamentand Robinson mercury-vapor light-traps (Intachat & Woiwood 1999), and aPennsylvania trap modified for wet environments and canopy suspension(Kitching et al. 2000b). The traps can be hoisted up to the required height inthe canopy using ropes and pulleys adjusted so that they can be let down to beemptied and then re-hoisted. Alternatively they can be fixed to canopy accesstowers. Efficiency of trap operation in the canopy is dependent on moonlight,weather conditions (e.g. cloudy and clear nights may produce quite differentdata sets), temperature, and vegetation density (which affects penetration ofthe light source) (Bowden 1982). Where multiple traps are used they shouldbe spaced such that the light cannot be seen from any other trap to avoid interference.Light-trap catches complement those from other sampling techniquessuch as chemical knockdown, which seems to be less effective at capturingLepidoptera. The main disadvantage of light-trap catches is the difficulty ofdetermining where the insects have come from within the forest.Insects in fruit, seeds, and silk: moss cores, suspended soils,and bark spraysMost of the collecting techniques described in this chapter have been designedto capture insects that are free-living on the surface of the vegetation, or flyingthrough the air spaces in the canopy. There are, of course, a number of insectgroups that make a significant contribution to the canopy community but livewithin the plant tissue (stems, leaves, seeds, and fruit), within epiphytes, in silkcocoons, bark fissures, and suspended soils, and that are rarely represented inmore general canopy samples.In temperate rainforest and tropical cloud forest, trees support large mossmats and a considerable quantity of suspended soil (Nadkarni & Longino 1990,Winchester & Ring 1996a). These diverse micro-/mesohabitats contributesignificantly to ecosystem processes in the canopy and support rich, diverse,and distinctive invertebrate communities. They present an interestingsampling challenge because organisms can only be collected from them byaccessing the canopy directly using one of the techniques discussed at the startof this chapter.Invertebrates are collected by taking samples of the habitat (soil, leaf litter, ormoss) in the canopy and then removing the animals in the laboratory either byactive extraction (e.g. Winkler extraction, Tullgren funnel) or by washing(Behan-Pelletier et al. 1996). Habitat samples should be of a known weight or

160 CHAPTER 7volume so that the density of animals can be standardized to habitat unit (e.g.biomass of moss). This is usually achieved by taking a core sample (e.g. mossmat cores of 3 ¥ 5 cm; Winchester and Ring 1996b, Winchester 2002). Samplesmay be collected from particular locations along branches, from differentheights in the canopy, or from the center of epiphytes (Rodgers & Kitching1998, Walter et al. 1998), depending on the research question. Core samples arevery effective since it is clear where the animals are located in the canopy, andtherefore the technique lends itself to answering questions about key ecosystemprocesses. Other micro-/mesohabitats within the canopy that canyield insects are seeds and fruit, which may be sampled by clipping from thevegetation or by collecting fallen fruit from the ground (e.g. figs; W. Paarmann,personal communication).ConclusionsThe canopy is a spatially and architecturally complex environment, supportinga host of insects. Some canopy insects are tourists (sensu Moran & Southwood1982), others are habitat generalists that move between forest strata, whilstsome are canopy specialists well adapted to the particular niches available intree crowns (e.g. leaves, bark crevices, epiphyte surfaces, and suspended soils).In order to collect data that can answer the kinds of questions entomologistsmight wish to ask about these insects, a range and often a combination of samplingtechniques is required.Sampling in the canopy is distinct from sampling ground vegetation in thechallenges it poses in terms of access, community richness, and spatial heterogeneity.Several studies mentioned in this chapter suggest that the canopyfauna may indeed have a composition that is distinct from that of the understory,ground, or soil. In order to gain a fuller understanding of insect ecologyand to conduct hypothesis testing in a range of globally representative habitats,we have to continue to rise to, and overcome, the challenges presented by thisfrontier between the biosphere and the atmosphere.ReferencesAbbott, I., Burbidge, T. Williams, M., & Van Heurck, P. (1992) Arthropod fauna of jarrah (Eucalyptusmarginata) foliage in Mediterranean forest of Western Australia: spatial and temporalvariation in abundance, biomass, guild structure and species composition. Australian Journalof Ecology, 17, 263–274.Adis, J., Paarmann, W., da Fonseca, C.R.V., & Rafael, J.A. (1997) Knockdown efficiency ofnatural pyrethrum and survival rate of living arthropods obtained by canopy fogging inCentral Amazonia. In Canopy Arthropods (ed. N. Stork, J. Adis, & R. Didham), pp. 67–81.Chapman & Hall, London.

TECHNIQUES AND METHODS 161Adis, J., Basset, Y., Floren, A., Hammond, P.M., & Linsenmair, K.E. (1998a) Canopy fogging ofan overstorey tree — recommendations for standardization. Ecotropica, 4, 93–97.Adis, J., Harada, A.Y., da Fonseca, C.R.V., Paarmann, W., & Rafael, J.A. (1998b) Arthropodsobtained from the Amazonian tree species “Cupiuba” (Goupia glabra) by repeated canopyfogging with natural pyrethrum. Acta Amazonica, 28, 273–283.Aizen, M.A. & Feinsinger, P. (1994) Habitat fragmentation, native insect pollinators and feralhoney bees in Argentine “Chacos Serrano”. Ecological Applications, 4, 378–392.Barker, M.G. (1997) An update on low-tech methods for forest canopy access and on samplingthe forest canopy. Selbyana, 18, 16–26.Barker, M. & Standridge N. (2002) Ropes as a mechanism for canopy access. In The GlobalCanopy Handbook (ed. A. Mitchell, K. Secoy, & T. Jackson), pp 13–23. GCP, Oxford.Basset, Y. & Charles, E. (2000) An annotated list of insect herbivores foraging on theseedlings of five forest trees in Guyana. Anais da Sociedade Entomologica do Brasil, 29,433–452.Basset, Y. & Höft, R. (1994) Can apparent leaf damage in tropical trees be predicted byherbivore load or host-related variables? A case study in Papua New Guinea. Selbyana, 15,3–13.Basset, Y., Aberlanc, H.P., & Delvare, G. (1992) Abundance and stratification of foliage arthropodsin a lowland rain forest of Cameroon. Ecological Entomology, 17, 310–318.Basset, Y., Springate, N.D. Aberlanc, H.P., & Delvare, G. (1997) A review of methods for samplingarthropods in tree canopies. In Canopy Arthropods (ed. N. Stork, J. Adis, & R. Didham),pp. 27–52. Chapman & Hall, London.Basset, Y., Novotny, V., Miller, S.E., & Kitching, R.L (2003a) Methodological advances and limitationsin canopy entomology. In Arthropods of Tropical Forests: Spatio-Temporal Dynamics andResource Use in the Canopy (ed. Y. Basset, V. Novotny, S.E. Miller, & R.L. Kitching), pp. 7–16.Cambridge University Press, Cambridge.Basset, Y., Hammond, P., Barrios,H., Holloway, J.D. and Miller, S.E. (2003b) Vertical stratificationof arthropod assemblages. In Arthropods of Tropical Forests: Spatio-Temporal Dynamics andResource Use in the Canopy (ed. Y. Basset, V. Novotny, S.E. Miller, & R.L. Kitching), pp. 17–27.Cambridge University Press, Cambridge.Bauerle, W.L., Hinckley, T.M., Cermak, J., Kucera, J., & Bible, K. (1999) The canopy water relationsof old-growth Douglas-fir trees. Trees (Berlin), 13 (4), 211–217.Behan-Pelletier, V. & Winchester, N.N. (1998) Arboreal oribatid mite diversity: colonising thecanopy. Applied Soil Ecology, 9, 45–51.Behan-Pelletier, V.M., Tomlin, A., Winchester, N., & Fox, C. (1996) Sampling protocols formicroarthropods. In A Workshop Report on Terrestrial Arthropod Sampling Protocols forGraminoid Ecosystems (ed. A.T. Finnamore). http://www.eman-rese.ca/eman/reports/publications/sage [accessed May 8, 2004].Blanton, C.M. (1990) Canopy arthropod sampling: a comparison of collapsible bag and foggingmethods. Journal of Agricultural Entomology, 7, 41–50.Bowden, J. (1982) An analysis of the factors affecting catches of insects in light-traps. Bulletinof Entomological Research, 72, 535–556.Compton, S.G., Ellwood, M.D.F., Davis, A.J., & Welch, K. (2000) The flight heights of chalcidwasps (Hymenoptera, Chalcidoidea) in a lowland Bornean rain forest: fig wasps are the highfliers. Biotropica, 32, 515–522.Cooper, R.J. & Whitmore, R.C. (1990) Arthropod sampling methods in ornithology. Studies inAvian Biology, 13, 29–37.Costa, J.T. & Crossley, D.A. Jr. (1991) Diel patterns of canopy arthropods associated with threetree species. Environmental Entomology, 20, 1542–1548.

162 CHAPTER 7Crossley, D.A. Jr., Callahan, J.T., Gist, C.S., Maudsley, J.R., & Waide, J.B. (1976) Compartmentalizationof arthropod communities in forest canopies at Coweeta. Journal of the GeorgiaEntomological Society, 11, 44–49.Devy, M.S. (1998) Breeding systems in bee-pollinated canopy forests of Southwestern Ghats,India. Selbyana, 19, 274.Dial, R. & Tobin, S.C. (1994) Description of arborist methods for forest canopy access andmovement. Selbyana, 15, 24–37.Didham, R.K. (1997) Dipteran tree-crown assemblages in a diverse southern temperate rainforest.In Canopy Arthropods (ed. N. Stork, J. Adis, & R. Didham), pp. 320–343. Chapman &Hall, London.Erwin, T.L. (1982) Tropical forests: their richness in Coleoptera and other arthropod species.Coleopterist’s Bulletin, 36, 74–75.Erwin, T.L. (1983) Tropical forest canopies: the last biotic frontier. Bulletin of the EntomologicalSociety of America, 29, 14–19.FAO (1999) State of the World’s Forests. FAO, Rome.Floren, A. & Linsenmair, K.E. (1997) Diversity and recolonisation dynamics of selected arthropodgroups on different tree species in a lowland rainforest in Sabah, Malaysia with specialreference to Formicidae. In Canopy Arthropods (ed. N. Stork, J. Adis, & R. Didham), pp.344–381. Chapman & Hall, London.Floren, A. & Linsenmair, K.E. (2000) Do ant mosaics exist in pristine lowland rain forests? Oecologia(Berlin), 123, 129–137.Foggo, A., Ozanne, C.M.P., Hambler, C., & Speight, M.R. (2001) Edge effects, tropical forestsand invertebrates. Plant Ecology, 153, 347–359.Guilbert, E. (1998) Studying canopy arthropods in New Caledonia: how to obtain a representativesample. Journal of Tropical Ecology, 14, 665–672.Hallé, F. & Blanc, P. (eds.) (1990) Biologie d’une Canopeé de Forêt equatoriale. Rapport de Mission.Radeau des Cimes Octobre —Novembre 1989, Guyane Française. Montpellier II et CNRS-Paris VI,Montpellier / Paris.Hammond, P.M. (1992) Species inventory. In Global Biodiversity: Status of the Earth’s LivingResources (ed. B. Groombridge), pp. 17–39. Chapman & Hall, London.Hammond, P.M., Stork, N.E., & Brendell, M.J.D. (1997) Tree-crown beetles in context: a comparisonof canopy and other ecotone assemblages in a lowland tropical forest in Sulawesi. InCanopy Arthropods (ed. N. Stork, J. Adis, & R. Didham), pp. 184–223. Chapman & Hall,London.Heatwole, H. & Higgins, W. (1993) Canopy research methods: a review. Selbyana, 14, 23.Hill, C.J. & Cermak, M. (1997) A new design and some preliminary results for a flight intercepttrap to sample forest canopy arthropods. Australian Journal of Entomology, 36, 51–55.Hill, D. Roberts, P., & Stork, N. (1990) Densities and biomass of invertebrates in stands of rotationallymanaged coppice woodlands. Biological Conservation, 51, 167–177.Hollier, J.A. & Belshaw, R.D. (1993) Stratification and phenology of a woodland Neuropteraassemblage. The Entomologist, 112, 169–175.Hurlbert, S.H. (1984) Pseudoreplication and the design of ecological field experiments. EcologicalMonographs, 54, 187–211.Inoue, T., Yumoto, T., Hamid, A.A., Seng, L.H., & Ogino, K. (1995) Construction of a canopyobservation system in a tropical rainforest of Sarawak. Selbyana, 16, 24–35.Intachat, J. & Holloway, J.D. (2000) Is there stratification in diversity or preferred flight heightof geometroid moths in Malaysian lowland tropical forest? Biodiversity and Conservation, 9,1417–1439.Intachat, J. & Woiwood, I.P. (1999) Trap design for monitoring moth biodiversity in tropicalrainforests. Bulletin of Entomological Research, 89, 153–163.

TECHNIQUES AND METHODS 163Johnson, M.D. (2000) Evaluation of an arthropod sampling technique for measuring foodavailability for forest insectivorous birds. Journal of Field Ornithology, 71, 88–109.Kitching, R.L., Bergelson, J.M., Lowman, M.D., McIntyre, S., & Carruthers, G. (1993) The biodiversityof arthropods from Australian rainforest canopies: general introduction, methods,sites and ordinal results. Australian Journal of Ecology, 18, 181–191.Kitching, R.L., Orr, A.G., Thalib, L., Mitchell, H., Hopkins, M.S., & Graham, A.W. (2000a) Mothassemblages as indicators of environmental quality in remnants of upland Australian rainforest. Journal of Applied Ecology, 37, 284–297.Kitching, R.L., Vickerman, G., Laidlaw, M., & Hurley, K. (2000b) The Comparative Assessment ofArthropod and Tree Biodiversity in Old-World Rainforests: the Rainforest CRC / Earthwatch ProtocolManual. Rainforest CRC, Cairns.Kitching, R.L., Basset, Y., Ozanne, C.M.P., & Winchester, N.N. (2002) Canopy knockdowntechniques. In The Global Canopy Handbook (ed. A. Mitchell, K. Secoy, & T. Jackson), pp.134–139. GCP, Oxford.Lowman, M.D. & Wittman, P.K. (1995) The last biological frontier? Advancements in researchon forest canopies. Endeavour (Cambridge), 19, 161–165.Lowman, M.D., Kitching, R.L., & Carruthers, G. (1996) Arthropod sampling in Australian subtropicalrain forests: how accurate are some of the more common techniques? Selbyana, 17,36–42.Majer, J.D. & Recher, H.F. (1988) Invertebrate communities on Western Australian eucalypts:a comparison of branch clipping and chemical knockdown procedures. Australian Journal ofEcology, 13, 269–278.Majer, J., Recher, H.F., Perriman, W.S., & Achuthan, N. (1990) Spatial variation of invertebrateabundance within the canopies of two Australian eucalypt forests. Studies in Avian Biology,13, 65–72.Majer, J., Recher, H.F., & Ganesh, S. (2000) Diversity patterns of eucalypt canopy arthropods ineastern and western Australia. Ecological Entomology, 25, 295–306.Marini-Filho, O.J. (1999) Distribution, composition, and dispersal of ant gardens and tendingants in three kinds of central Amazonian habitats. Tropical Zoology, 12, 289–296.Martin, J.L. (1966) The insect ecology of red pine plantations in central Ontario. IV. The crownfauna. Canadian Entomologist, 98, 10–27.May, R.M. (1990) How many species? Philosophical Transactions of the Royal Society, Series B, 330,293–304.McGeoch, M.A. & Gaston, K.J. (2000) Edge effects on the prevalence and mortality factorsof Phytomyza ilicis (Diptera, Agromyzidae) in a suburban woodland. Ecology Letters, 3,23–29.Mitchell, A. (2001) Introduction — canopy science: time to shape up. Plant Ecology, 153,5–11.Mitchell, A., Secoy, K., & Jackson, T. (eds.) (2002) The Global Canopy Handbook. GCP, Oxford.Moffet, M. & Lowman, M.D. (1995) Canopy access techniques. In Forest Canopies (ed. M.D.Lowman & N. Nadkarni), pp. 3–26. Academic Press, San Diego.Moran, V.C. & Southwood, T.R.E. (1982) The guild composition of arthropod communities intrees. Journal of Animal Ecology, 51, 289–306.Moran, V.C., Hoffmann, J.H., Impson, F.A.C., & Jenkins, J.F.G. (1994) Herbivorous insectspecies in the tree canopy of a relict South African forest. Ecological Entomology, 19, 147–154.Nadkarni, N.M. & Longino, J.T. (1990) Invertebrates in canopy and ground organic matter in aneotropical montane forest, Costa Rica. Biotropica, 22, 286–289.Nadkarni, N.M. & Parker, G.G. (1994) A profile of forest canopy science and scientists — whowe are, what we want to know and obstacles we face: results of an international survey. Selbyana,15, 38–50.

164 CHAPTER 7Ohmart, C.P., Stewart, L.G., & Thomas J.R. (1983) Phytophagous insect communities in thecanopies of three Eucalyptus forest types in south-eastern Australia. Australian Journal of Ecology,8, 395–403.Oliver, I., & Beattie, A.J. (1996) Designing a cost-effective invertebrate survey: a test of methodsfor rapid assessment of biodiversity. Ecological Applications, 6, 594–607.Ozanne, C.M.P. (1991) The arthropod fauna of coniferous plantations. D.Phil thesis, OxfordUniversity.Ozanne, C.M.P. (1996) The arthropod communities of coniferous forest trees. Selbyana, 17,43–49.Ozanne, C.M.P. (1999) A comparison of the canopy arthropods communities of coniferous andbroad-leaved trees. Selbyana, 20, 290–298.Ozanne, C.M.P., Speight, M.R., & Evans, H.F. (1988) Spray deposition and retention in thecanopies of five forest tree species. Aspects of Applied Biology, 17 (2), 245–246.Ozanne, C.M.P., Foggo, A, Hambler, C., & Speight, M.R. (1997) The significance of edge-effectsin the management of forests for invertebrate biodiversity. In Canopy Arthropods (ed. N.Stork, J. Adis, & R. Didham), pp. 534–550. Chapman & Hall, London.Ozanne, C.M.P., Speight, M.R., Hambler, C., & Evans, H.F. (2000) Isolated trees and forestpatches: patterns in canopy arthropod abundance and diversity in Pinus sylvestris (ScotsPine). Forest Ecology and Management, 137, 53–63.Ozanne, C.M.P., Anhuf, D., Boulter, S.L., et al. (2003) Biodiversity meets the atmosphere: aglobal view of forest canopies. Science, 301, 183–186.Paarmann, W. & Kerck, K. (1997) Advances in using the canopy fogging technique to collectliving arthropods from tree-crowns. In Canopy Arthropods (ed. N. Stork, J. Adis, & R.Didham), pp. 53–66. Chapman & Hall, London.Paarmann, W. & Stork, N.E. (1987) Canopy fogging, a method of collecting living insects forinvestigation of life history strategies. Journal of Natural History, 21, 563–566.Reynolds, B. & Crossley, D.A. Jr. (1995) Use of a canopy walkway for collecting arthropods andassessing leaf area removed. Selbyana, 16, 21–23.Ring, R.A. & Winchester, N.N. (1996) Coastal temperate rainforest canopy access systems inBritish Columbia, Canada. Selbyana, 17, 22–26.Roberts, H.R. (1973) Arboreal Orthoptera in the rain forests of Costa Rica collected with insecticide:a report on grasshoppers (Acrididae) including new species. Proceedings of the Academyof Natural Sciences, Philadelphia, 125, 46–66.Rodgers, D. & Kitching, R.L. (1998) Vertical stratification of rainforest collembolan (Collembola:Insecta) assemblages: description of ecological patterns and hypotheses concerningtheir generation. Ecography, 21, 392–400.Schowalter, T.D. & Ganio, L.M. (1998) Vertical and seasonal variation in canopy arthropodcommunities in an old-growth conifer forest in southwestern Washington, USA. Bulletin ofEntomological Research, 88, 633–640.Schowalter, T.D., Webb, W.J., & Crossley, D.A. Jr. (1981) Community structure and nutrientcontent of canopy arthropods in clearcut and uncut forest ecosystems. Ecology, 62,1010–1019.Schowalter, T.D., Stafford, S.G., & Slagle, R.L. (1988) Arboreal arthropod community structurein an early successional coniferous forest ecosystem in Western Oregon. Great BasinNaturalist, 48, 327–333.Shubert, H. and Ammer, U. (1998) Comparison of arthropod fauna in canopies of natural andmanaged forests of southern Germany. Selbyana, 19, 298.Southwood, T.R.E. & Henderson, P.A. (2000) Ecological Methods. 3rd edn. Blackwell Science,Oxford.

TECHNIQUES AND METHODS 165Southwood, T.R.E., Moran, V.C., & Kennedy, C.E.J. (1982) The assessment of arboreal insectfauna — comparisons of knockdown sampling and faunal lists. Ecological Entomology, 7,331–340.Speight, M.R. & Wainhouse, D. (1989) Ecology and Management of Forest Insects. Oxford UniversityPress, Oxford.Springate, N.D. & Basset, Y. (1996) Diel activity of arboreal arthropods associated with PapuaNew Guinea trees. Journal of Natural History, 30, 101–112.Stork, N. (1991) The composition of the arthropod fauna of Bornean lowland rainforest trees.Journal of Tropical Ecology, 7, 161–180.Stork, N. (1993) How many species are there? Biodiversity and Conservation, 2, 215–232.Stork, N.E. & Best, V. (1994) European Science Foundation — results of a survey of Europeancanopy research in the tropics. Selbyana, 15, 51–62.Stork, N.E. & Hammond, P. (1997) Sampling arthropods from tree-crowns by fogging withknockdown insecticides: lessons from studies of oak tree beetle assemblages in RichmondPark (UK). In Canopy Arthropods (ed. N. Stork, J. Adis, & R. Didham), pp. 3–26. Chapman &Hall, London.Wagner, T. (2000) Influence of forest type and tree species on canopy-dwelling beetles inBudongo forest, Uganda. Biotropica, 32, 502–514.Walter, D.E. & Behan-Pelletier, V. (1999) Mites in forest canopies: filling the size distributionshortfall? Annual Review of Entomology, 44, 1–19.Walter, D.E., Seeman, O., Rodgers, D., & Kitching R.L. (1998) Mites in the mist: how unique isa rainforest canopy-knockdown fauna? Australian Journal of Ecology, 23, 501–508.Watanabe, H. (1997) Estimation of arboreal and terrestrial arthropod densities in the forestcanopy as measured by insecticide smoking. In Canopy Arthropods (ed. N. Stork, J. Adis, &R. Didham), pp. 401–416. Chapman & Hall, London.Winchester, N.N. (1997) Canopy arthropods of coastal Sitka spruce trees on Vancouver island,British Colombia, Canada. In Canopy Arthropods (ed. N. Stork, J. Adis, & R. Didham), pp.151–168. Chapman & Hall, London.Winchester, N.N. (2002) Canopy micro-arthropod diversity: suspended soil exploration. InThe Global Canopy Handbook (ed. A. Mitchell, K. Secoy, & T. Jackson), pp. 140–144. GCP,Oxford.Winchester, N.N. & Behan-Pelletier, V. (2003) Fauna of suspended soils in an Ongokea gore treein Gabon. In Arthropods of Tropical Forests: Spatio-Temporal Dynamics and Resource Use in theCanopy (ed. Y. Basset, V. Novotny, S.E. Miller, & R.L. Kitching), pp. 102–109. Cambridge UniversityPress, Cambridge.Winchester, N.N. & Ring, R.A. (1996a) Northern temperate coastal Sitka spruce forests withspecial emphasis on canopies: studying arthropods in an unexplored frontier. NorthwestScience, 70, (special issue), 94–103.Winchester, N.N. & Ring, R.A. (1996b) Centinelan extinctions: extirpation of Northern Temperateold-growth rainforest arthropod communities. Selbyana, 17, 50–57.Yamashita, Z. & Ishii, T. (1976) Basic structure of the arboreal arthropod fauna in the naturalforest of Japan. Ecological studies of the arboreal arthropod fauna 1. Report of the EnvironmentalScience, Mie University, 1, 81–111.

166 CHAPTER 7Index of methods and approachesMethodology Topics addressed CommentsChemical Investigation of within- Collects insects in flight through theknockdown canopy variation in density canopy and surface dwellers on theand species richness.leaves, flowers, fruit, twigs, branches,Association of populations and trunk of the tree. Homoptera,and communities with Psocoptera, Collembola, Coleoptera.individual trees.All groups of insects collected; lesseffective for Lepidoptera.Studies of the spatialdistribution of organisms Does not reliably collect insects thatacross habitats.spin leaves together, or that inhabitleaf domatia and epiphytes, or thatAbsolute estimates ofbore into bark.population density andspecies richness.Assessment of communitystructure, e.g. guild.Collection of live specimensfor subsequent experimentalwork on population dynamicsand feeding strategies.Branch clipping Assessment of the vertical Particularly effective for sedentaryand bagging stratification of communities. insects; collects wide range of groups.Investigation of dielLarge mobile insects are undermovementwithin the forest sampled, e.g. Odonata, midge cloudscanopy.(Chironomidae).Specific questions aboutplant–herbivore relationships.Questions relating to presenceand absence of species,absolute estimates ofpopulation density, guildstructure, and heterogeneityof distribution.Studies requiring count ofinsect densities per unit ofplant biomass and surface area.Aerial and arboreal Questions about canopy Interception traps: dominant groupstraps: Malaise, community structure. Coleoptera, Diptera andinterception, Testing of specific Hymenoptera.emergence, and hypotheses about the use of Combination traps: flightlesslight particular strata of the arthropods, e.g. oribatid mites.canopy.Underestimate the contribution ofRelative estimates ofsedentary and flightless arthropodspopulation density and to the community.species richness.Light traps: Lepidoptera, Coleoptera,Continued

TECHNIQUES AND METHODS 167Methodology Topics addressed CommentsQuestions about movement Hemiptera.of insects within thecanopy space.Moss cores, Questions about specific Acarina, Araneae, Collembola,suspended soils insect–plant relationships. Psocoptera.and bark sprays Resource partitioning withinthe forest canopy.Absolute estimates ofpopulation density andspecies richness.

CHAPTER 8Sampling methods for water-filledtree holes and their artificialanaloguesS.P. YANOVIAK AND O.M. FINCKEIntroductionInsects of small aquatic habitats found in plants, called phytotelmata (plantheldwaters; Varga 1928), have attracted the attention of naturalists for thegreater part of a century (Fish 1983). For biological investigations, the relativelysmall accumulations of water occurring in bromeliads, pitcher plants, and treeholes offer several methodological advantages over lakes, streams, and othercomparatively large systems (e.g. Maguire 1971). First, phytotelmata are discreteand can be treated as individual units for sampling and faunal surveys.Second, these habitats are often abundant where they occur, permitting samplesizes appropriate for statistical analyses. Finally, the macrofauna of phytotelmatais often specialized and of manageable diversity and abundance. This is especiallytrue of the aquatic insect inhabitants (e.g. Kitching 2000). Water-filledtree holes are among the most tractable of small aquatic systems, in part becausethey are relatively persistent, and can be mimicked with plastic cups, bamboosections, or other inexpensive materials. Despite these unique features of treeholes and their specialized inhabitants, the extent to which processes affectingtheir biodiversity and community structure can be generalized to larger systemsremains to be seen.Natural tree holesWater-filled tree holes are formed by the collection of rainwater in naturalcavities occurring in the above-ground woody portions of trees (e.g. Kitching1971a). They exist in hardwood forests all over the world (Fish 1983, Kitching2000), and are the most abundant standing water systems in some tropicalforests. Tree holes occur in a variety of shapes and sizes. In the lowland moistforest of Panama, they may be superficially categorized as slit-shaped, bowlshaped(Fig. 8.1), or pan-shaped (Fig. 8.2), based on the morphology of the holeaperture and the ratio of water volume to surface area (Fincke 1992a). Temperatetree holes have been classified according to the presence or absence of a continuouslining of tree bark on the hole interior (Kitching 1971a). Although treeholes occur in the crowns of trees and may exceed 50 liters in size (Fincke168

SAMPLING METHODS FOR WATER-FILLED TREE HOLES 169Fig. 8.1 Typical cup- or bowl-shaped tree hole in Panama.1992a, Yanoviak 1999a, 1999b), most are much smaller, and many occur below2 meters, where they are easily accessible. As such, they are excellent focal habitatsfor investigations of aquatic insect behavior, population biology, and communityecology.A variety of macroorganisms use tree holes as breeding sites, and manyspecies breed exclusively in this habitat. Aquatic insects dominate the assemblagesof macrofauna in tree holes; larvae of true flies (Diptera) are generally themost common inhabitants (e.g. Snow 1949, Kitching 2000, Yanoviak 2001a).Tree holes are also the primary breeding sites for many disease vectors, includingmosquitoes (Diptera: Culicidae; Galindo et al. 1955) and biting midges(Diptera: Ceratopogonidae; Vitale 1977). Tropical tree holes have the most diversefauna and harbor an array of predators that are absent in temperate holes(e.g. odonates and tadpoles of dendrobatid frogs; Kitching 1990, Fincke 1992a,1998, Orr 1994). Aquatic insect assemblages of tree holes are sufficiently diversein terms of taxonomy and ecological function to permit theory-basedstudies, yet distinct and simple enough to be manageable for students withlimited entomological background.Here we present methods for non-destructive sampling of aquatic insects and

170 CHAPTER 8Fig. 8.2 Pan-shaped tree holes formed by the collection of rainwater in the trunk of a fallentree.other macroorganisms from water-filled tree holes based on our experience inNeotropical forests. Our goals are to describe a thorough approach to samplingtree holes, and to identify potential problems associated with data collectionand interpretation, which also apply to other types of phytotelmata. All of theconcerns we address may not be applicable to tree holes in all types of forests.For example, holes in temperate forests often lack predators, support lower insectdiversity, and are subject to stronger seasonal effects, which may influencethe frequency and timing of sampling required for a thorough inventory of treehole occupants. We conclude with some caveats that should be consideredbefore drawing general ecological or evolutionary inferences from tree holessystems. Belkin et al. (1965) and Service (1993) provide additional usefulinformation and references regarding insect sampling from tree holes, withemphasis on mosquito larvae.Sampling techniques for natural tree holesAccurate estimates of aquatic insect abundance and diversity in most waterfilledtree holes can be obtained with simple procedures and equipment(Fig. 8.3). The most common approach is removal of contents of the hole to apan for counting. Researchers have devised a variety of techniques to accomplishthis task, but reasonably complete samples are obtained by removing detritusand water from the hole, and sieving out the macroorganisms (e.g.

SAMPLING METHODS FOR WATER-FILLED TREE HOLES 171Fig. 8.3 Basic equipment used for sampling water-filled tree holes.Jenkins & Carpenter 1946, Bradshaw & Holzapfel 1983, Walker & Merritt 1988,Copeland 1989, Barrera 1996).Sub-sampling is one alternative approach to data collection from tree holes.Kitching (1971b) invented a core sampler that extracts a fraction of the holevolume with each use. This device can provide density data for populationstudies of some taxa (Kitching 1972a, 1972b), and it collects deep sediments,but the size and rigidity of the corer limit its use to a subset of holes withsufficiently large openings (Barrera 1988). Moreover, insects are often nonrandomlydistributed within and among tree holes (e.g. Barrera 1996), and itis unlikely that sub-samples collected with a corer would be useful in generalsurveys or community-level studies.Although techniques will vary according to the nature of the investigation,thorough tree hole sampling can be summarized as a five-step process:1 organisms in the undisturbed hole are noted with the aid of a flashlight, andwater chemistry parameters are measured;2 detritus and sediments are removed;3 fluid contents are removed;4 the hole is repeatedly flushed with clean water;5 the interior walls of the empty hole are inspected with a flashlight.These steps are useful for documenting the macrofauna of the most commonlyencountered tree holes: those of small to medium volume (e.g.

172 CHAPTER 8accomplished with small electronic probes. Tree hole water chemistry and temperaturevary with hole size, and fluctuate considerably over a 24-hour period(e.g. Fincke 1999), hence multiple readings are preferable.Regardless of tree hole volume or morphology, step 1 should be completedbefore a hole is disturbed. If the water is relatively clear and fauna are known tothe investigator, careful examination of hole contents can yield accurate dataon species richness and abundance for some taxa. Species are more likely tobe missed after a hole is disturbed; individuals may hide in crevices or beoverlooked if fine sediments do not settle rapidly. Steps 2–5 are sometimes unnecessary(i.e. in very small holes with minimal detritus), or excessively timeconsumingif not impossible in very large holes. In aseasonally wet forests (e.g.at La Selva, Costa Rica), tree holes typically accumulate much more sedimentthan in forests where they dry out and remain dry for some time each year. Removingall of the sediment in the former cases can be extremely tedious, and isnot necessary if the taxa of interest are macroorganisms, which typically remainabove the sediment layer.The type of equipment used for completion of steps 2–4 depends on the sizeand shape of the hole, but almost any hole can be sampled with common materials(Fig. 8.3). In small tree holes, water and soft sediments are removed toa large graduated cylinder for volume measurement using a large suctionpipette (e.g. a turkey baster). The contents are then transferred to a white plasticpan for counting. Detritus is removed by hand or with long forceps, rinsed inthe tree hole water, and set aside in another pan. A hole should always beprobed with a stick or pencil before using bare hands to remove detritus. Tropicaltree holes occasionally contain scorpions, land crabs, and ponerine ants,which, if unnoticed, can quickly ruin an otherwise productive field trip. Flushingby repeated filling (two or three times) with water collected from the holetends to dislodge most organisms remaining in the hole (e.g. Lounibos 1981). Ifadditional water is used for flushing, it should be held in a second pan to avoiddilution or contamination of chemicals and nutrients in the original hole water.Larger holes can be emptied by using a flexible garden hose to siphon waterinto a pail. Detritus is removed by hand (or by using a wooden ruler or trowel tolift small packs of leaves), rinsed in the tree hole water and set aside. Rather thancompletely refilling the hole with a large quantity of water, rinsing the wallswith a few liters of clean water is an effective way to dislodge remaining insects.While the detritus and water collected from flushing are allowed to settle intheir pans, the interior walls of the hole can be inspected with a flashlight to spotelusive organisms. Damselfly larvae commonly cling to the walls, and dragonflylarvae are often found covered with sediment at the bottom of the hole,where they can be quite cryptic (Fincke 1992a). With some experience, one caneasily recognize elusive species and they can often be counted without removal.Agitating leaves and other detritus in the collected water usually rids them ofany clinging organisms; the composition of detritus can be noted, and litter canthen be returned to the hole. After sediments settle in the pan, the clear water

SAMPLING METHODS FOR WATER-FILLED TREE HOLES 173is decanted off. This concentrates macroorganisms such as odonates, syrphids,and tadpoles, making them easier to find and count. A small flashlight, whichhelps focus the investigator’s attention to small areas, makes counting mucheasier, especially on overcast days or under dense forest canopy. A large grid(e.g. 4 ¥ 4 cm) drawn on the bottom of the pan is also helpful when insects arevery abundant. For large tree holes, very numerous insects such as mosquitolarvae can be removed in batches to small cups, which permits one to count withgreater accuracy. Sub-samples of taxa unfamiliar to the investigator can be collectedlive for rearing and identification in the laboratory. Small plastic bags(e.g. Nasco® Whirl-Paks) or vials provide the best means of transporting livespecimens. Depending on the climate, it may be necessary to transport samplesin a cooler with ice to prevent overheating. After subsamples are collected, theremaining organisms and original water can be returned to the hole, and thecollection pans are rinsed before the next hole is sampled. Following this protocol,ten or more small to mid-size tree holes in the forest understory can be thoroughlysampled in a day.Important considerations for natural tree hole experimentsAdequate sample size is a concern in the design of any field experiment (seeChapter 1), and can be problematic in long-term studies of tree holes, which aredynamic systems. Some of the largest holes form suddenly when a tree falls anddepressions in the trunk fill with water, but most of these holes do not persist formore than a season or two (depending on the tree species). Even holes in livingtrees, which often hold water for decades, can vary in volume considerablyfrom year to year, gradually filling completely with mud, or suddenly rottingthrough. On Barro Colorado Island (BCI), Panama, for example, of 44 waterfilledholes in live trees checked in 1982, 6.8 percent had rotted through twoyears later, compared with 58 percent of those in fallen trees (n = 12) and 44 percentof those in dead, upright trees (n = 9). Of 23 water-filled holes checked in1984, 28.6 percent of those in live trees (n = 21) had rotted through by the timethey were again checked 10 years later. From these data, we estimate a turnoverrate for water-filled tree holes in live trees between 2.8 and 3.4 percentper year. Thus, studies of longer than a year should always use more than theminimum number needed for sufficient statistical power in an experiment,with the percentage of additional holes depending on the proportion of studyholes in living vs. dead trees.Another problem associated with tree hole studies is the large number of variablesthat can affect community properties and interactions among residentaquatic insects. For example, diversity and abundance tend to increase withtree hole volume (Sota 1998, Fincke 1999, Yanoviak 1999b), and predator effectsmay be stronger in smaller holes (Fincke 1994). This problem is best overcomeby surveying a large number of holes several weeks before the start of theexperiment, then focusing manipulations on a subset of holes that fall within

174 CHAPTER 8an acceptable range of variation. Some fauna are found only in very large holes(e.g. Agalychnis callidryas tadpoles), whereas others may be more commonin shaded holes (e.g. Heteroptera: Veliidae), holes high in dissolved oxygen(e.g. Physalaemus pustulosus tadpoles), or holes with abundant, fruity detritus(e.g. Diptera: Syrphidae) (Fincke 1999, Yanoviak 1999c, 2001a). Therefore,biodiversity surveys should incorporate a broad range of hole types and, wherepossible, note the detritus composition. Because the fauna of tree holes is depauperaterelative to that of streams or lakes, overlooking a few species canmake a significant difference in conclusions drawn about biodiversity within orbetween forests.Tree holes located within 2 m of the ground are best for replicated experimentsdue to the time and hazards associated with canopy work. However, treehole height can affect community properties and distributions of some species(Galindo et al. 1951, 1955; Lounibos 1981). In Panama, for example, speciesrichness in tree holes generally declines with increasing height above theground (Yanoviak 1999b). Thus, diversity surveys and community-levelstudies should include tree holes from the ground to the canopy. Holes in thecrowns of trees are easier to find than to sample. Overflow stains (Snow 1949)and drinking monkeys (Yanoviak 1999b) pinpoint the locations of canopy treeholes to the ground-based observer, but usually only a small percentage are accessible.Moffett and Lowman (1995) reviewed methods for canopy access; thesingle-line climbing technique (Perry 1978) is the most effective for canopy treehole work. Once in a tree crown, the investigator can tie in to a fixed point, leavethe main rope, and move laterally along branches to sample tree holes. Thisis a slow and often difficult process, with minimal data resulting from extensivetime and energy expenditure. Despite the risk of pseudoreplication (Hurlbert1984), the most efficient strategy for canopy tree hole work is to focus climbingefforts on tree species that typically possess many holes per crown, and repeatedlysample holes that are readily accessible.For those who cannot or choose not to climb trees, cranes or canopy walkways(Moffett & Lowman 1995) provide alternative access to the canopy. However,both of these methods require the use of artificial holes that can bepositioned in accessible areas. A rope and pulley system can also be used to raiseartificial tree holes into the canopy (e.g. Loor & DeFoliart 1970), but the instabilityof the containers makes them prone to disturbance from wind and canopymammals, and may result in lost data. Sampling the colonists of artificial holessecured or suspended in tree crowns or at midstory will at least provide a list oforganisms that likely use natural tree holes at the same level (Yanoviak 1999b).Artificial tree holesMany of the problems associated with sampling natural phytotelmata for ecologicalexperiments can be overcome by using artificial analogues. Simple con-

SAMPLING METHODS FOR WATER-FILLED TREE HOLES 175tainers can be used to mimic a variety of phytotelmata, such as Heliconia sppbracts (Naeem 1988) and bromeliads (Frank 1985, 1986; Haugen 2001).A major advantage of artificial plant containers is that water volume, nutrientinput, and the initial presence or absence of some species can be standardized.Plastic analogues are generally inexpensive and can be censused completely inmuch less time than the same number of natural habitats of similar size. Mostimportantly, artificial containers generally attract the same fauna as the naturalsystems (e.g. Pimm & Kitching 1987, Fincke et al. 1997, Yanoviak 2001a) andwill even be readily defended by territorial odonates and frogs (Fincke 1992b,1998; Haugen 2001).Almost any container filled with rainwater and a small amount of leaf litterwill function as an artificial tree hole for short-term experiments. Tree hole analogueswith varying degrees of realism can be constructed from bamboo sections(e.g. Lounibos 1981), automobile tires (e.g. Juliano 1998), stone vases(e.g. Sota et al. 1994), or plastic pots (e.g. Fincke 1992a). Galindo et al. (1951,1955) described two bamboo trap designs (closed-top and open-top), and discusseddifferences in mosquito species composition between the types. Closedtoptraps with small lateral openings mimic a specific tree hole morphology thatis difficult to sample, thus they provide a useful addition for tree hole experimentsor surveys. Some containers (e.g. tires and stone vases) are weak replicasof tree holes, but attract many tree hole mosquito species and are often used invector control studies.We prefer to use plastic containers for artificial tree holes because they arereadily available, lightweight, and durable. Of the several sizes and shapes ofcontainers we use to replicate water-filled tree holes in tropical forest studies,three types seem to give the best results.Because most natural holes are less than 1.0 liter, the artificial hole we oftenuse is a 0.65-liter plastic cup (Churchill Container Corp., Shawnee, KS;Fig. 8.4). A second type is constructed from a 1.5-liter plastic funnel (DetailedDesigns/Injectron, Inc., N. FN-01, USA) in which the spout is removed and thebottom hole is closed from the inside with a rubber stopper (Fig. 8.5). A funneldesign that is flat on one side facilitates secure attachment to a tree. To mimiclarger holes, we use either a 6.65-liter oil drain pan (Koller Enterprises, Inc.,Fenton, MO) or larger (9.0 liter) brown plastic wash tub (Action Industries, Inc.,Cheswick, PA; Fig. 8.6). The latter has convenient handles contiguous with therim that make attachment easier, and its considerable depth results in proportionallyless water loss during the inevitable tipping that occurs after attachment.These types of artificial tree holes will survive years of exposure andclosely approximate the shape of similar-sized natural holes.Artificial tree holes in the form of cups and funnels may be tied to small treetrunks or branches, whereas larger pan-type holes are either secured to forkedbranches in the canopy or to the trunks of fallen trees in the understory(Fig. 8.6). Polypropylene rope (6 mm diameter) is best for securing artificialholes, but a stronger material (i.e., wire) is required if ants or termites are

176 CHAPTER 8Fig. 8.4 A cup-shaped artificial tree hole with wire cage to prevent oviposition by odonates.nesting in the tree (Yanoviak 1999b). Plastic-coated flexible wire hooks can beused to hang a cup or funnel from rope around the tree (Fig. 8.5), allowing rapidremoval and replacement when frequent sampling is planned. Pans and smallartificial holes sampled less often can be secured with rope passed once aroundthe tree and through perforations or handles in the container rim (Fig. 8.4).These methods cause no obvious harm to the tree.We fill artificial holes with rainwater and put a partially submerged pieceof tree bark or balsa wood (Novak & Peloquin 1981) in them as a perch forovipositing insects. Recently fallen leaf litter collected from the forest floor isadded as a nutrient base for the aquatic community (Fish & Carpenter 1982).An initial volume of uncompressed litter within 25–50% of the total artificialhole volume is appropriate for general studies, but the quantity of litter usedwill depend on the nature of the experiment, the type of forest, and the season.Litter fall reflects seasonal and species-specific patterns of leaf fall and fruiting(e.g. Foster 1982), resulting in variation in nutrient input over space and time.

SAMPLING METHODS FOR WATER-FILLED TREE HOLES 177Fig. 8.5 A funnel-shaped artificial tree hole.For example, 30-day litter accumulation in 0.65-liter cups (71 cm 2 opening)placed in the BCI forest ranged from 0.0 to 1.2 g dry mass (x = 0.45 ± 0.07 g s.e.;Yanoviak 2001b), and a single fruit fall can result in a pulse of superabundantnutrients (Fincke et al. 1997). To keep nutrients above some minimum forexperimental purposes, it may be necessary to periodically add small amountsof litter (e.g. 10% of hole volume) or a substitute nutrient (e.g. fish food oryeast) to some holes.Apart from providing a standardized physical environment, artificial treeholes also allow some control over potentially important biological variables,such as nutrient input or colonization by key taxa. For example, modifying anartificial tree hole by covering it with a large-mesh wire screen cage (Fig. 8.4)prevents most natural nutrient input, but allows colonization by mosquitoesand most other macroorganisms (Fincke et al. 1997). This cage design also effectivelyexcluded odonates from artificial tree holes in Panama, where they arethe most common top predators in this system (Fincke 1998). Similar screening

178 CHAPTER 8Fig. 8.6 A pan-shaped artificial tree hole.was only moderately effective at excluding odonates from natural tree holes, inpart because eggs laid in the bark prior to screening could not be detected and removed(Yanoviak 2001b). Because the screen cages exclude most falling detritus,additional leaf material must be added to experiments lasting more than afew weeks. Deciding on the quantity of additional nutrient input is not a trivialproblem, particularly if growth rates or biodiversity are being measured. Twomedium-sized leaves added bimonthly to our 0.65-liter cup-shaped holes keptthe abundance of mosquito larvae similar to controls that were open to naturalleaf fall, whereas adding 0.05 g of yeast bimonthly resulted in higher than normallevels of mosquito larvae (O.M. Fincke, unpublished). Litter that falls intoan adjacent, uncovered, but otherwise identical container could be added to theexperimental hole on a regular basis, making the nutrient input more closelyreflect natural conditions.Sampling techniques for artificial tree holesArtificial tree holes offer a big advantage over natural tree holes because theycan be easily emptied completely; time is the only limiting factor in gettingaccurate counts of the fauna. As for natural tree holes, chemical parametersshould be measured and a preliminary census of the fauna made before theartificial hole is disturbed. A cup- or funnel-type hole is then untied (or

SAMPLING METHODS FOR WATER-FILLED TREE HOLES 179unhooked) from the tree and its contents poured into a white pan. Using themethods described earlier for natural holes, one can census organisms in a 0.65-liter hole in under 30 minutes, and in a 1.5-liter hole in under 60 minutes.Large pan-type artificial holes can be left in place for sampling but may requireseveral hours to census, depending on the focus of the study.Important considerations for artificial tree hole experimentsArtificial tree holes provide an excellent means of controlling multiple variablesand increasing sample sizes for experimentation. However, because artificialtree holes are not integral parts of trees, researchers using them should be awareof four differences that might affect their results:1 artificial tree holes are typically younger than holes in living trees, and thuslack potentially relevant biological history (e.g. accumulations of feces, refractorydetritus, and sediments);2 they may receive less stemflow than natural holes in upright trees;3 their contents have no direct contact with living wood;4 their inner sides are much smoother than the creviced surface of naturaltree holes, which may provide protection for some species or life historystages.Stemflow inputs and contact with wood are potentially important becauseboth can affect nutrient dynamics and insect productivity in tree holes (e.g.Carpenter 1982, Walker et al. 1991), and stemflow contributes to washout disturbance(Washburn & Anderson 1993). Contact between tree hole water andliving wood allows the exchange of materials (e.g. tannins, sap, nitrogenouswastes) between the water and the tree, whereas this exchange and any potentialtree species effects on community structure would not occur in artificial treeholes. Abiotic conditions in artificial holes can differ significantly from naturalholes of similar size (Table 8.1). However, most tree hole inhabitants tolerate awide range of pH and dissolved oxygen (Fincke 1999, Yanoviak 1999a), andTable 8.1 Comparison of abiotic variables in 11 artificial holes and 25 natural tree holes at LaSelva, Costa Rica. Means were calculated from measurements taken three times per day (seeFincke 1998 for methods). Ranges in parentheses. Significant differences between naturaland artificial holes indicated by *p < 0.05, **p < 0.01 (t-tests).Hole type Volume (liters) Temperature pH Dissolved(°C)oxygen (ppm)Artificial 0.8 ± 0.2 27.2 ± 0.5** 5.5 ± 0.2* 2.7 ± 0.3**(0.1–2.0) (24.0–29.0) (3.4–6.2) (0.7–3.9)Natural 0.9 ± 0.1 25.0 ± 0.2 4.7 ± 0.1 1.0 ± 0.1(0.1–2.0) (24.7–29.7) (3.4–6.0) (0.3–2.2)

180 CHAPTER 8such differences should not affect colonization or survivorship of most macrofauna(although this may not be true for microorganisms). Artificial holes areparticularly good mimics of natural holes in fallen trees (Fig. 8.2), which typicallyreceive limited stem flow, do not contact living tissue, and are relativelyyoung.Some simple procedures can be used to add realism to the artificial systemif necessary. Inoculation of artificial holes with water from natural holes (e.g.during setup and occasionally thereafter) can quickly establish and maintainthe microbial assemblage, which is a critical part of tree hole food webs (e.g. Fish& Carpenter 1982, Walker et al. 1991). The rope used to secure cups to treesoften conducts stemflow to the cup interior (S.P. Yanoviak, personal observation),and additional stemflow can be directed into a hole by placing the emergentportion of bark or balsa wood against the tree trunk (Fig. 8.4) or by tackinga small piece of plastic onto the tree and allowing it to drain into the hole.Detritus composition and container color are two additional considerationsfor those using artificial tree holes in field experiments. The type of litter addedto a hole can affect insect productivity and aquatic community structure(Carpenter 1982, Fish & Carpenter 1982, Walker et al. 1997, Yanoviak 1999d),so the composition of litter in a hole (in terms of fragment size, species, age, etc.)should either be consistently haphazard or standardized. Habitat color influencesinsect colonization in artificial tree holes (Yanoviak 2001c) and othertypes of phytotelmata (Frank 1985, 1986). Although some workers use clearplastic pans to mimic tree holes in temperate forests (e.g. Srivastava & Lawton1998), we recommend black or dark brown containers. In Panama, black containersattracted more species than blue or green containers (Yanoviak 2001c).Clear plastic pots can be painted black on the outside, and tubs of any color canbe made more realistic by lining the inside with a piece of black plastic (garbagebags work well) that hangs down over the outside edge.Statistical methods for water-filled tree holesIn most cases, data gathered from replicated tree hole experiments can be analyzedusing standard statistical techniques (e.g. ANOVA). Repeated-measuresANOVAs are often used to compare treatment means when the same artificial ornatural holes are sampled multiple times (e.g. Fincke et al. 1997, Yanoviak1999b). Because a large number of ecological and physical variables can bemeasured in each tree hole, multivariate analyses may be appropriate for manyresearch questions (e.g. Barrera 1988, 1996). It is common for one tree hole tocontain zero individuals while hundreds of mosquitoes are present in another.The log(x + 1) transformation will usually normalize this extreme variation(Sokal & Rohlf 1981). Note that some holes are depauperate of both predatorsand prey species simply because of resource limitation or abiotic factors; it is importantto differentiate between those factors and low diversity resulting frombiological interactions (e.g. Fincke et al. 1997).

SAMPLING METHODS FOR WATER-FILLED TREE HOLES 181Problems in interpretation of comparative dataConsideration of spatial and temporal scale is critical when using tree holes as asystem to test ecological or evolutionary theory. Tree holes, like bromeliad phytotelmatadescribed by Picado (1913), are analogous to a “subdivided swamp”for most macrofauna using them. Because the resource is subdivided, colonizationby certain taxa may be limited with respect to volume, height above theground, or even morphology of the tree hole opening (see also Frank & Lounibos1987; Fincke 1992a). Whereas individual tree holes are discrete, replicableunits, the scale of the “swamp,” which is ecologically comparable to a lake orstream, would be all the tree holes in a forest, which is neither discrete nor easilyreplicable. For example, top predators decrease diversity within water-filledtree holes on BCI (Yanoviak 2001b). But do forests (at similar latitude) lackingmajor tree hole predators have greater diversity of tree hole species than forestswithout those predators? Answering that question requires pooling diversityacross replicate holes, with and without predators. Even then, unless thesample of tree holes is representative of the natural distribution with respect tovolume, height, and age since the last filling, conclusions may vary.Finally, most tree hole denizens represent only the larval stage of a species;adults typically are not limited to using a single hole over their reproductive lifespan, and may have species-specific dispersal distances. In evolutionary studies,for example, the scale of interest would not be simply the fitness of individualsusing a given hole, but rather the fitness derived from all the tree holes usedover an individual’s reproductive life span (e.g. Fincke & Hadrys 2001). Hence,conclusions about community or population processes may be premature withoutknowledge of the seasonality, longevity, and dispersal ability of the adults inquestion.ConclusionsAlthough there is a growing number of studies documenting the insect fauna ofwater filled tree holes around the world (Kitching 2000, Yanoviak 2001a), currentknowledge remains overwhelmingly biased towards potential disease vectors.Despite considerable interest in the ecology of this system, few studies haveaddressed the importance of microbial diversity and ecology in tree holes (e.g.Walker & Merritt 1988; Walker et al. 1991). Decomposer microbes (bacteria andfungi) form a critical link between the nutrient base (e.g. leaf litter) and secondaryconsumers (e.g. mosquito larvae) in tree holes (Fish & Carpenter 1982).Various other microorganisms, such as microcrustaceans, rotifers, and protozoans,also occur in tree holes (Kitching 2000, Yanoviak 2001a), and may functionas prey or competitors with the macrofauna. Microbial ecology has beenlargely overlooked in tropical tree holes, and several basic questions remain tobe answered for this system in general. For example, what regulates microbialdiversity and productivity in tree holes? How does the composition of detritus

182 CHAPTER 8affect decomposer assemblages? Does microbial diversity influence macroorganismdiversity or productivity? Are microbial assemblages more speciesrichin tropical tree holes? The ecology of microorganisms has been examined inother phytotelmata (e.g. Addicott 1974, Cochran-Stafira & von Ende 1998,Carrias et al. 2001), and these studies exemplify the kinds of investigations thatare needed in tree holes. Likewise, few studies have addressed the ecologicalimportance of inorganic nutrients (e.g. nitrogen and phosphorus) in tree holes(e.g. Carpenter 1982; Walker et al. 1991). Microbial and nutrient dynamicshave been described for many large freshwater systems, and some of the techniquescommonly used by stream and lake ecologists to quantify these parameterscould be transferred to tree holes.In summary, water-filled tree holes are tractable habitats for ecological andbehavioral studies; sampling their insect fauna is a relatively simple process, andthe use of artificial holes is an inexpensive way to increase sample size and controlmultiple factors for experiments. The extent to which inferences from treehole data have a more general application for freshwater systems remains to beseen. Nevertheless, given their important ecological role, these aquatic microhabitatsmerit much more attention than they have received, especially intropical forests.AcknowledgementsWe are grateful to Coral McAllister for the illustrations. Comments fromS. Stuntz and C. Ozanne, and discussions with L. P. Lounibos, improved themanuscript.ReferencesAddicott, J.F. (1974) Predation and prey community structure: an experimental study of theeffect of mosquito larvae on the protozoan communities of pitcher plants. Ecology, 55,475–492.Barrera, R. (1988) Multiple factors and their interactions on structuring the community ofaquatic insects of treeholes. PhD thesis, Pennsylvania State University.Barrera, R. (1996) Species concurrence and the structure of a community of aquatic insects intree holes. Journal of Vector Ecology, 21, 66–80.Belkin, J. N., Hogue C. L., Galindo, P., Aitken, T. H., Schick, R. X., & Powder, W. A. (1965) Mosquitostudies (Diptera, Culicidae). II. Methods for the collection, rearing and preservation ofmosquitoes. Contributions of the American Entomological Institute 1, 19–78.Bradshaw, W.E. & Holzapfel, C.M. (1983) Predator-mediated, non-equilibrium coexistence oftree-hole mosquitoes in southeastern North America. Oecologia, 57, 239–256.Carpenter, S.R. (1982) Stemflow chemistry: effects on population dynamics of detritivorousmosquitoes in tree-hole ecosystems. Oecologia, 53, 1–6.Carrias, J.-F., Cussac, M.-E., & Corbara, B. (2001) A preliminary study of freshwater protozoain tank bromeliads. Journal of Tropical Ecology, 17, 611–617.

SAMPLING METHODS FOR WATER-FILLED TREE HOLES 183Cochran-Stafira, D.L. & von Ende, C.N. (1998) Integrating bacteria into food webs: studieswith Sarracenia purpurea inquilines. Ecology, 79, 880–898.Copeland, R.S. (1989) The insects of treeholes of northern Indiana with special reference toMegaselia scalaris (Diptera: Phoridae) and Spilomyia longicornis (Diptera: Syrphidae). GreatLakes Entomologist, 22, 127–132.Fincke, O.M. (1992a) Interspecific competition for tree holes: consequences for mating systemsand coexistence in neotropical damselflies. The American Naturalist, 139, 80–101.Fincke, O.M. (1992b) Consequences of larval ecology for territoriality and reproductive successof a neotropical damselfly. Ecology, 73, 449–462.Fincke, O.M. (1994) Population regulation of a tropical damselfly in the larval stage by foodlimitation, cannibalism, intraguild predation and habitat drying. Oecologia, 100, 118–127.Fincke, O.M. (1998) The population ecology of Megaloprepus coerulatus and its effect on speciesassemblages in water-filled tree holes. In Insect Populations: in Theory and in Practice (ed. J.P.Dempster & I.F.G. McLean), pp. 391–416. Kluwer, Dordrecht.Fincke, O.M. (1999) Organization of predator assemblages in Neotropical tree holes: effects ofabiotic factors and priority. Ecological Entomology, 24, 13–23.Fincke, O.M. (unpublished ms.) Constraints on adaptive cannibalism, clutch size, and offspringsex ratios in a shared tree hole nursery.Fincke, O.M. & Hadrys, H. (2001) Unpredictable offspring survivorship in the damselfly,Megaloprepus coerulatus, shapes parental behavior, constrains sexual selection, and challengestraditional fitness estimates. Evolution, 55, 762–772.Fincke, O.M., Yanoviak, S.P., & Hanschu, R.D. (1997) Predation by odonates depresses mosquitoabundance in water-filled tree holes in Panama. Oecologia, 112, 244–253.Fish, D. (1983) Phytotelmata: flora and fauna. In Phytotelmata: Terrestrial Plants as Hosts forAquatic Insect Communities (ed. J.H. Frank & L.P. Lounibos), pp. 1–27. Plexus, Medford, NJ.Fish, D. & Carpenter, S.R. (1982) Leaf litter and larval mosquito dynamics in tree-hole ecosystems.Ecology, 63, 283–288.Foster, R.B. (1982) Seasonal rhythms of fruitfall on Barro Colorado Island. In Ecology of a TropicalForest: Seasonal Rhythms and Long-Term Changes (ed. E.G. Leigh, A.S. Rand, & D.M.Windsor), pp. 151–172. Smithsonian Institution, Washington, DC.Frank, J.H. (1985) Use of an artificial bromeliad to show the importance of color value in restrictingthe colonization of bromeliads by Aedes aegypti and Culex quinquefasciatus. Journal ofthe American Mosquito Control Association, 1, 28–32.Frank, J.H. (1986) Bromeliads as ovipositional sites for Wyeomyia mosquitoes: form and colorinfluence behavior. Florida Entomologist, 69, 728–742.Frank, J.H. & Lounibos, L.P. (1987) Phytotelmata: swamps or islands? Florida Entomologist, 70,14–20.Galindo, P., Carpenter, S.J. & Trapido, H. (1951) Ecological observations on forest mosquitoesof an endemic yellow fever area in Panama. American Journal of Tropical Medicine, 31, 98–137.Galindo, P., Carpenter, S.J., & Trapido, H. (1955) A contribution to the ecology and biology oftree hole breeding mosquitoes of Panama. Annals of the Entomological Society of America, 48,158–164.Haugen, L. (2001) Privation and uncertainty in the small nursery of Peruvian tadpoles: larvalecology shapes the parental mating system. PhD thesis, University of Oklahoma.Hurlbert, S.H. (1984) Pseudoreplication and the design of ecological field experiments. EcologicalMonographs, 54, 187–211.Jenkins, D.W. & Carpenter, S.J. (1946) Ecology of the tree hole breeding mosquitoes of nearcticNorth America. Ecological Monographs, 16, 31–47.Juliano, S.A. (1998) Species introduction and replacement among mosquitoes: interspecificresource competition or apparent competition? Ecology, 79, 255–268.

184 CHAPTER 8Kitching, R.L. (1971a) An ecological study of water-filled tree-holes and their position in thewoodland ecosystem. Journal of Animal Ecology, 40, 281–302.Kitching, R.L. (1971b) A core sampler for semi-fluid substrates. Hydrobiologia, 37, 205–209.Kitching, R.L. (1972a) The immature stages of Dasyhelea dufouri Laboulbene (Diptera: Ceratopogonidae)in water-filled tree-holes. Journal of Entomology (ser. A), 47, 109–114.Kitching, R. L. (1972b) Population studies of the immature stages of the tree-hole midge Metriocnemusmartinii Thienemann (Diptera: Chironomidae). Journal of Animal Ecology, 41, 53–62.Kitching, R. L. (1990) Foodwebs from phytotelmata in Madang, Papua New Guinea. TheEntomologist, 109, 153–164.Kitching, R. L. (2000) Food Webs and Container Habitats: the Natural History and Ecology ofPhytotelmata. Cambridge University Press, Cambridge.Loor, K.A. & DeFoliart, G.R. (1970) Field observations on the biology of Aedes triseriatus.Mosquito News, 30, 60–64.Lounibos, L.P. (1981) Habitat segregation among African treehole mosquitoes. EcologicalEntomology, 6, 129–154.Maguire, B. Jr. (1971) Phytotelmata: biota and community structure determination in plantheldwaters. Annual Review of Ecology and Systematics, 2, 439–464.Moffett, M.W. & Lowman, M.D. (1995) Canopy access techniques. In Forest Canopies (ed. M.D.Lowman & N.M. Nadkarni), pp. 3–26. Academic Press, San Diego.Naeem, S. (1988) Predator–prey interactions and community structure: chironomids, mosquitoesand copepods in Heliconia imbricata (Musaceae). Oecologia, 77, 202–209.Novak, R.J. & Peloquin, J.J. (1981) A substrate modification for the oviposition trap used fordetecting the presence of Aedes triseriatus. Mosquito News, 41, 180–181.Orr, A.G. (1994) Life histories and ecology of Odonata breeding in phytotelmata in Borneanrainforest. Odonatologica, 23, 365–377.Perry, D.R. (1978) A method of access into the crowns of emergent and canopy trees. Biotropica,10, 155–157.Picado, C. (1913) Les broméliacées épiphytes comme milieu biologique. Bulletin Scientifique dela France et de la Belgique, 47, 215–360.Pimm, S.L. & Kitching, R.L. (1987) The determinants of food chain lengths. Oikos, 50, 302–307.Service, M.W. (1993) Mosquito Ecology: Field Sampling Methods. 2nd edn. Kluwer, Dordrecht.Snow, W.E. (1949) The Arthropoda of wet tree holes. PhD thesis, University of Illinois, Urbana.Sokal, R.R. & Rohlf, F.J. (1981) Biometry. W.H. Freeman, New York.Sota, T. (1998) Microhabitat size distribution affects local difference in community structure:metazoan communities in treeholes. Researches on Population Ecology, 40, 249–255.Sota, T., Mogi, M., & Hayamizu, E. (1994) Habitat stability and the larval mosquito communityin treeholes and other containers on a temperate island. Researches on Population Ecology, 36,93–104.Srivastava, D.S. & Lawton, J.H. (1998) Why more productive sites have more species: experimentaltest of theory using tree-hole communities. The American Naturalist, 152, 510–529.Varga, L. (1928) Ein interessanter Biotop der Biocönose von Wasserorganismen. BiologischesZentralblatt, 48, 143–162.Vitale, G. (1977) Culicoides breeding sites in Panama. Mosquito News, 37, 282.Walker, E.D. & Merritt, R.W. (1988) The significance of leaf detritus to mosquito (Diptera: Culicidae)productivity from treeholes. Environmental Entomology, 17, 199–206.Walker, E.D., Lawson, D.L., Merritt, R.W., Morgan, W.T., & Klug, M.J. (1991) Nutrient dynamics,bacterial populations, and mosquito productivity in tree hole ecosystems and microcosms.Ecology, 72, 1529–1546.Walker, E.D., Kaufman, M.G., Ayres, M.P., Riedel, M.H., & Merritt, R.W. (1997) Effects of variationin quality of leaf detritus on growth of the eastern tree-hole mosquito, Aedes triseriatus(Diptera: Culicidae). Canadian Journal of Zoology, 75, 707–718.

SAMPLING METHODS FOR WATER-FILLED TREE HOLES 185Washburn, J.O. & Anderson, J.R. (1993) Habitat overflow, a source of larval mortality for Aedessierrensis (Diptera: Culicidae). Journal of Medical Entomology, 30, 802–804.Yanoviak, S.P. (1999a) Community ecology of water-filled tree holes in Panama. PhD thesis,University of Oklahoma, Norman.Yanoviak, S.P. (1999b) Community structure in water-filled tree holes of Panama: effects ofhole height and size. Selbyana, 20, 106–115.Yanoviak, S.P. (1999c) Distribution and abundance of Microvelia cavicola Polhemus (Heteroptera:Veliidae) on Barro Colorado Island, Panama. Journal of the New York EntomologicalSociety, 107, 38–45.Yanoviak, S.P. (1999d) Effects of leaf litter species on macroinvertebrate community propertiesand mosquito yield in Neotropical tree hole microcosms. Oecologia, 120, 147–155.Yanoviak, S.P. (2001a) The macrofauna of water-filled tree holes on Barro Colorado Island,Panama. Biotropica, 33, 110–120.Yanoviak, S.P. (2001b) Predation, resource availability, and community structure in Neotropicalwater-filled tree holes. Oecologia, 126, 125–133.Yanoviak, S.P. (2001c) Container color and location affect macroinvertebrate communitystructure in artificial treeholes in Panama. Florida Entomologist, 84, 265–271.Index of methods and approachesMethodology Topics addressed CommentsGeneral surveys Descriptive data on community Collections are taken from astructure.large number of holes overseveral seasons. Providesbasic natural history datafrom which further questionsand experiments aredeveloped.Quantitative sub- Distribution and abundance of a May be accomplished with asampling given species. corer or similar tools. Oftenthe only practical option forvery large tree holes.Species exclusion or Predator effects on community Exclusion methods dependaddition structure; interspecific interactions. on organism size andbehavior, and may not be100% effective.Manipulation of litter Effects of basal resources on Qualitative and quantitativeinputs community structure. characteristics of litter areimportant considerationsForest canopy access Effects of environmental gradients Ratio of effort and timeon species distributions.expenditure to quantity ofdata recovered may beprohibitive. Artificial treeholes provide a viable option

CHAPTER 9Sampling devices and samplingdesign for aquatic insectsLEON BLAUSTEIN AND MATTHEW SPENCERIntroductionIn this chapter, we consider two major problems associated with accuratelyestimating populations and community structure of aquatic insects: samplingdevices appropriate for specific questions, and errors associated with estimatesfrom sampling. Types of sampling devices for aquatic insects are numerous. Thisreflects the fact that among aquatic insects there is great diversity in mobility,behavior, and microhabitat use. Consequently, many sampling devices are onlyuseful for a specific group of species in a specific type of habitat. In this chapter,we will briefly describe a small subset of these devices. Next, we will consider errorsin estimating population size and species richness. In this “errors” section,we give prominence to the often overlooked problem of making comparisonsacross environmental conditions or experimental treatments when samplingefficiency may vary across these conditions. We also consider how many samplesare needed to estimate various parameters such as population densities orspecies richness, given a defined level of accuracy and precision. Finally, wevery briefly consider a few ethical considerations when sampling in aquatic environments.While we attempt to give some coverage to sampling the differenthabitat types, we do give particular emphasis to the habitat that we are most familiarwith — small lentic habitats.Before addressing these problems, because there is occasionally ambiguity inthe literature, we begin by defining a few terms used in this chapter: absolutedensity, relative abundance, and sampling efficiency. Absolute density refers tothe real density — number per unit area or volume. Relative abundance (or abundanceindex) refers to the number collected per sampling effort, which is quantitativeand may be used for comparative purposes but is not an estimate of thereal density. This can be the number caught in a one-meter sweep withoutknowing how many individuals escaped the net, or the number caught per lighttrap, which by itself tells us nothing about the real density but may allow us tomake comparisons. Sampling efficiency, in the case of sampling devices that measurethe number caught per unit area or volume, gives the proportion of individualsin that sampled area that are caught. Knowing sampling efficiency in suchcases allows us to convert a relative abundance to an absolute density. In othercases, sampling efficiency is a relative term that is not linked to actual densities.186

SAMPLING AQUATIC INSECTS 187For example, we might determine that a light trap is twice as efficient at trappingone family of beetles as it is at trapping a second family, even though we mightnot be able to relate numbers caught to actual densities.A survey of sampling devicesSampling devices have been categorized according to: (i) type of habitat forwhich they are suitable; (ii) whether they yield absolute density estimates versusabundance indices; (iii) whether the insect actively enters the device (e.g.light trap) or is passive but is caught (e.g. sweep net); (iv) time and cost (Merrittet al. 1996, Turner & Trexler 1997). Our survey of sampling devices is organizedlargely according to the inverse of number (iii) — i.e. we categorize according tothe activity of the collecting individual and not the activity of the insect. Foractive-operator devices, the operator moves the device to capture the insects. Ina passive-operator device, the device is stationary and insects enter on theirown or, in the case of lotic environments, are swept into the device by streamflow. In general, passive-operator devices have several advantages: they tend tohave better precision and accuracy than active-operator devices, and environmentaldisturbance is minimized. For example, two passive-operator samplingdevices, an aquatic light trap and minnow trap, cause little environmental disturbance,whereas actively sweeping with a D-net can uproot and tear vegetationin the sampled area. Disturbing the environment may not be permissible,or may be ethically questionable. Moreover, disturbing the environment mayalso be undesirable if repeated samples are necessary and earlier sampling affectsdensities and species composition in subsequent samples. A disadvantageof passive-operator devices is that the operator must come at least twice for onesample — first to set the device in place and again to collect the sample.Our survey is very brief and provides minimal instructions for use. For a morecomprehensive list and description of sampling devices, we suggest beginningwith a very useful table by Merritt et al. (1996, Table 3) that classifies devicesaccording to various factors and provides references for further information.A short written description for how to use a sampling device is a poor substitutefor observing a highly experienced operator in the field, particularly for activeoperatorsampling devices. One excellent proxy is a video cassette prepared byResh et al. (1990) in which sampling is demonstrated for approximately 30 devicesin the field.Sampling with nets and dippersNets are active-operator sampling devices, and are probably the most commongroup of devices for sampling aquatic insects. For most nets, the operator activelysweeps the net through the water. In the case of tow nets for samplingpelagic insect species such as Chaoborus, the net is attached to a line and is pulled

188 CHAPTER 9across the water. Absolute density estimates of sweeps or tows can be madein theory if both the distance that the net is moved through the water and thesampling efficiency are known. However, since sampling efficiency is generallynot known, sweeping often serves as a quantitative measure of “number persweep” or “number caught per unit volume” rather than “number per unit volume”.Nets vary in mesh sizes. There is a trade-off between mesh size and catchefficiency — smaller mesh size will catch a wider range of size classes but catchefficiency will be reduced, particularly for larger and more mobile species.A longer bag can partially remedy this problem. Sweep nets have frames of variousshapes but those used for benthic organisms are generally D-shaped. Theflat part of the frame is at the bottom to maximize the fit of substrate contourwith the frame. Both precision and efficiency in sampling benthic insects shouldbe lower in stony substrate than in fine substrate. Filamentous algae and macrophytesprobably reduce both precision and efficiency for sweep nets for insectsoccupying all levels of the water column.Typical “dippers” differ from sweep nets in that the collecting devices are solidcontainers rather than mesh and thus the volume of water sampled for a singlesample is confined to the size of the container. Dippers are generally used to collectorganisms at, or close to, the water–air interface. The most common of suchsamplers is the mosquito dipper. As implied by its name, it is used for samplingmosquito immatures though it can also be used to sample associated species inshallow aquatic habitats (e.g. Washino & Hokama 1968). An extensive literaturereview assessing dipping can be found in Service (1993, Chapter 2).Although there are many variations, the dipper usually consists of a one-pint(473 ml) or half-liter container attached to a one-meter pole. The container isgenerally white, making the dark larvae more detectable. Dippers also includesoup ladles for sampling small habitats such as water-filled tires and small rockpools. The actual technique for making the dip sample varies greatly among researchers.Resh et al. (1990) illustrate dip sampling by dragging the dipper alongthe water surface as one might sweep a net. Others, by rotating the wrist,“carve” out a volume of water. Still others place the dipper in the water, allowingsuction to fill in the volume of the dipper. Advantages of the dipper as a samplingdevice are that it is inexpensive and light. Generally, a number of dipscollected randomly, or across a transect, are concentrated together through anet to constitute a single sample. While sampling programs using dippers aregenerally considered to give relative abundance estimates, there have been attemptsto calibrate dipping in order to yield an absolute density estimate (e.g.Stewart & Schaefer 1983). Andis et al. (1983, reported in Service 1993) foundsurprisingly strong, positive correlations between numbers of mosquito larvaeper dip and number of larvae collected in a unit area sampler.Area or column samplersDespite the fact that sweep netting and dipping provide a sample of known vol-

SAMPLING AQUATIC INSECTS 189ume, they are rarely considered to provide absolute densities because there isoften great difficulty in determining sampling efficiency. Probably for this reason,these sampling devices tend to be called “semi-quantitative” samplers inthe literature (for example, see Merritt et al. 1996). There are other samplingdevices in which the operator actively samples a known area or volume ofhabitat with less error. A popular device for sampling benthos in lotic habitats(riffles) is the Surber sampler. Contributing to its popularity is that it is easy totransport (it is foldable and light) and to use. This device consists of a quadratand an attached net perpendicular to the quadrat. The quadrat is randomlyplaced on the substrata upstream from the operator. Each rock within thequadrat is slightly lifted and benthos are dislodged by rubbing the rock surfaceswith one’s hand. These dislodged individuals are then swept into the attacheddrift net by the stream flow. For higher efficiency, these rocks can also be placedinto a plastic bag or bucket and brought to shore for additional inspection. Afterthis has been done to all rocks within the quadrat, the substrate inside thequadrat can be vigorously rubbed with the operator’s hand to dislodge any remainingfauna. The device is then lifted and brought to shore where the net isinverted and its contents placed in a white pan for species identification andenumeration on the spot, or preserved and processed later in the laboratory.This device is suitable for water depths of up to 30 cm and for velocities wherethe operator and the sampler can maintain positions. If done meticulously, bothprecision and accuracy are considered to be quite high and there is probablyless variance in among-operator sampling efficiency than with many activeoperatordevices. However, rocks that lie only partially inside the quadrat mustbe dealt with in systematic matter. This becomes more and more problematic asrock size increases. Similarly, the depth of the sample into the substrate mustalso be standardized because macroinvertebrate vertical distributions vary withspecies (Rutherford & MacKay 1985).A functionally similar sampling device that also yields absolute density estimatesof stream benthos is the Hess sampler. This device consists of a cylinderthat has a mesh screen on one side to allow flow and a long attached net on theother side. The cylinder is placed on the substrate. A foam lining at the cylinderbottom makes for a good seal with the substrate. One’s hand is then placed insidethe cylinder, dislodging organisms from the substrate, and stream flowthrough the screen front results in the deposition of the dislodged organisms insidethe net.In soft substrate, core samplers can quantitatively determine abundance ofhyporheic (substrate dwelling) species. These core samples also allow for determinationof vertical distributions, though it should be noted that some preservationtechniques may warp the core sample, thus distorting the real verticaldistributions (Rutledge & Fleeger 1988). Another method for quantitatively estimatingabsolute density of hyporheic species in soft substrate is the use of grabsamplers. These devices contain a jaw-like apparatus with a rectangular openingat the bottom. The jaws plunge into the substrate when dropped from

190 CHAPTER 9above, encompassing a known area and volume of fine substrate. They eitherclose upon contact with the substrate (Petersen grab) or a weight is sent downthe attached line afterwards to trigger the closure of the grab (Ekman dredge). Ifa rock or debris prevents the complete closure of the trap, then part of the sampleis lost and the sample is unusable.Some devices simultaneously sample the entire water column. Column samplersmay be cylindrical or rectangular in cross section. They are forced downthrough the water into the substrate, thus trapping species inside the volume ofthe sampler. Such column samplers work best where there is soft substrate becausethe bottom must be sealed to prevent escape of organisms. The length ofthe device must of course exceed the water depth. The insects in the columnsample can then be collected by pumping the water out through a sieve, or bycontinually sweeping with a small net inside the column sampler until additionalsweeps capture no additional organisms. The sample can then estimatedensities of neustonic, pelagic, and benthic species. However, as is the case withmany active-operator devices, mobile insects are likely to be sampled withlower efficiency as they are more likely to escape.While most column samplers are designed for plunging them from the airdown to the bottom, Resh et al. (1990) demonstrate a “bottom-up” water columnsampler — i.e. it is pulled up from the bottom. This bottom-up or “pull-upsampler” can also be used in flexible vegetation. It consists of a pole with a sharppoint and a net attached perpendicular near the base of the pole. The pole isforced into the substrate such that the net frame lies parallel to and on the substrate.The pole can then be rotated 180 degrees so that the net lies below a relativelyundisturbed water column. After some re-equilibration time, the pole islifted up catching species within the water column. If there is vegetation, thepole can be first lifted just above the water. The vegetation that extends outsidethe frame is then snipped along the net frame. In this way, only the vegetationinside the column is collected, and this can give a density estimate of macrophytebiomass in addition to abundance of the insects.Sampling with natural and artificial substratesSome passive-operator sampling devices use artificial habitats that serve as substratesfor colonization or oviposition, and substrate samplers have become increasinglypopular for estimating density of stream benthos. Some samplers usenatural substrate. For example, rocks from streams can be collected, washed,and possibly sterilized, then placed in wire mesh baskets entrenched into thesame kind of substrate. If left in long enough, the substrates within the samplersapproach the densities and species compositions outside the samplers. Othersubstrate samplers use an artificial substrate that simulates natural substrate.For example, clay tiles can be placed in the stream to simulate the rocks(Lamberti & Resh 1985). Rutherford (1995) used artificial grass substratesanchored to natural substrate in streams to assess insect colonization and dis-

SAMPLING AQUATIC INSECTS 191persion. Immediately after removing the artificial substrates, she sprayed themwith an aerosol anaesthetic (Cytocool®) to freeze invertebrates, in order to facilitatemeasuring the spatial distribution. Still other substrate samplers do notsimulate real substrate but instead simply provide a standardized substrate forcolonization that can be compared across time or space. A commonly used one,the Hester–Dendy sampler (Hester & Dendy 1962) consists of a set of discs orplates connected and separated by a central pole, with plates spaced wideenough to allow colonization by macroinvertebrates (e.g. Caquet et al. 1996,Turner & Trexler 1997). As these artificial substrates are standardized in terms ofarchitecture and surface area, abundances derived from such samples in differentplaces left in the habitat for the same period could be compared. For an excellentand extensive critique of artificial substrates for sampling freshwaterbenthic macroinvertebrates, see Rosenberg and Resh (1982).Sampling with mesocosmsMesocosms such as outdoor artificial pools, streams, or enclosures that are openat the top, though generally thought of as bodies of water to be sampled, canserve as sampling devices in their own right (Blaustein & Schwartz 2001). Forexample, artificial pools have been used as a sampling device to examine howrisk of predation (e.g. Chesson 1984, Resetarits 2001), food level (e.g. Blaustein& Kotler 1993), and many other factors influence oviposition site selection byaquatic insects. The number of eggs, egg strings, or egg rafts in a particular mesocosmcan represent a single sample.One should keep in mind that if one is interested in comparing the relativeabundances of different species colonizing artificial pools, the physical attributesof the artificial pools can greatly influence the answer. For example, sizeof experimental aquatic mesocosms varies greatly among experiments (Petersenet al. 1999), and predator species may be much more likely to colonizelarger artificial pools than smaller ones (Pearman 1995, Wilcox 2001).In terms of how many samples are necessary to address ecological questions,a general rule of sampling is that the more samples we take, the more likely weare to statistically detect small treatment differences (we deal in depth with thisquestion later). This generality may not be the case in mesocosm experimentswhen the experiment depends on oviposition by an insect of limited populationsize (i.e. the total number of individuals colonizing mesocosms does not increaseproportionally with the number of mesocosms). Suppose, as was the casein the study by Morin et al. (1988), we wished to understand how colonizinginsect herbivores might compete with tadpoles. Morin et al. set up a number ofartificial pools containing known numbers of tadpoles. Some pools were leftopen to allow colonization by herbivorous insects while others were coveredwith screening to prevent oviposition. Tadpoles growing in the pools with opencolonization (by herbivorous insects) were adversely affected. Had they usedconsiderably more replicated pools, the density of colonizing herbivorous

192 CHAPTER 9insects per pool would likely have been lower. Had this been the case (fewerherbivorous insects per pool if more total pools), the competitive effect of herbivorousinsects on tadpoles would have been lower. Not only would the magnitudeof the effect have been lower, but as a consequence the probability of atype II error would probably have been higher. So, in such cases, the generalityof “the more sampling units the better” is not necessarily true, depending on theexperimental design of mesocosm experiments.Sampling by trapsTraps are sampling devices in which the operator is passive and the insects activelyenter the traps. In the case of drift nets used in streams (Matthaei et al.1998), the active agent is not the organism but the flowing water. If water velocityand trap efficiency are known, a quantitative estimate of the drift densitycan be calculated. However, for traps in general, it is the insect that is the activeagent and trap catch for interception devices depends on swimming speeds anddirection. If the trap actually attracts the insects, then it does not measure densitiesbut instead some number that is a relative count to the number captured inanother trap. Gee minnow traps appear to initially act as interception devices ifno bait is added. The trap is made of metal mesh and contains inverted funnelson both ends. Active insects crawl or swim through the funnel into the trap andgenerally cannot find their way back out. After the first individuals enter theminnow trap, the trap’s contents may then begin to attract or repel other individualsand species. Also, because the sample is live, predation may occur insidethe traps at high rates. Minnow traps are only effective traps for insects if meshsize is small. Turner and Trexler (1997), who used a large mesh size (6 mm),caught very few insect individuals but Blaustein (1988), who used a smallermesh size (2 mm) captured coleopterans and some hemipteran species in abundance.Minnow traps can be used to capture insects live, but if the trap is totallysubmerged, insects that require atmospheric oxygen will eventually drown.While drift nets and minnow traps serve largely as interception traps for horizontallymoving individuals, emergence traps intercept and capture individualsemerging from the water, and can estimate absolute density of emerginginsects. Emergence traps are generally pyramidal or conical in shape. The openbase is often set just below the water surface. They can be attached to floats sothat they are in the correct vertical position even if the water depth changes.Emerging insects climb the funnel-shaped trap into a collecting container at thetop.Some traps have attractants and thus cannot provide absolute density estimates(unless a mark–recapture or removal program is used), but instead providerelative indices of abundance. These attractants can be light (e.g. Washino& Hokama 1968), food (e.g. Vance et al. 1995), or some type of ovipositionattractant (e.g. Trexler et al. 1998).

SAMPLING AQUATIC INSECTS 193Visual observation and photographyInsect populations can be sampled, or in some cases complete counts can bemade, by visual counts or photography. This is most often done for surfacedwellingorganisms. Quadrats can be set up for such counts but in the case ofsmall habitats such as rock pools, the total number can be counted. This canoften be done easily for stages found at the water surface. For example, we haveused total counts of mosquito egg rafts laid in artificial or natural rock pools (e.g.Blaustein et al. 1995) and the number of chironomid pupal exuviae on thewater surface (Blaustein et al. 1996). Counts can be made for odonate emergenceby counting the exuviae left by emerging individuals on surfaces abovethe water. Pelagic or benthic species may also be counted sometimes in clear,shallow water. We have done so to estimate densities of chironomid larval cases(S.S. Schwartz et al. unpublished data). Surprisingly few studies have usedphotography as a sampling method to measure density and spatial distributionin aquatic insects. Resh et al. (1990) demonstrate use of a container with atransparent plate (similar to a diving mask) placed onto the water surface thatallows photography of benthic organisms in clear, shallow water. Digital camerasand image analyzers not only can facilitate counts, but reduce errors incounting and measuring.Sampling errorsThe problem of measurement error is the central concept in the design of anysampling program. All measurements are subject to errors, of which there aremany kinds (Rowe 1994). In particular, ecologists need to be aware of systematicerrors, random errors, and measurement interactions. Systematic errors areconsistent biases in the estimate of some variable. Examples include: when sizefractions of a population are small enough to escape through the mesh of a net;when some individuals escape a moving net (Fleminger & Clutter 1965); whensome individuals that colonized an artificial substrate sampler are lost uponmaking the collection (Rosenberg & Resh 1982); the overestimation of bodymass from an incorrectly calibrated balance. Random errors are differences betweenmeasured and true values arising from chance factors. Examples of randomerror are the number of animals from a population of given density thathappen to be caught on a given sweep of a net, or variation between estimatesof the mass of a single individual due to the variable amount of water adheringto its surface. We discuss both systematic and random errors at length below.Measurement interactions occur when the true value of a quantity is changedwhile attempting to measure it. For example, we may underestimate thedensity of a mobile insect such as a gerrid by taking a regular grid of netsweeps because the insects may flee from the sampled area in response to the

194 CHAPTER 9sampling activities. We will not have much more to say about measurement interactions,but it is worth remembering that they often trade off against randomerrors: the more intensive the sampling, the lower the random errors but thegreater the risk that the system is altered.Are systematic errors important?Classical statistics takes no account of systematic errors, because systematic errorsare rarely amenable to statistical analysis. Physics textbooks typically makethe assumption that systematic errors are small enough to be ignored (Taylor1982), and biostatistics texts (e.g. Zar 1984, Sokal & Rohlf 1995) usually confinetheir discussion of systematic errors to the desirability of ensuring that there arenone. On the contrary, analyses of historical trends in estimates of basic physicalconstants (which are probably the most reliable measurements of any naturalquantities) suggest that systematic errors are large, important, and difficultto eliminate (Shlyakhter & Kammen 1992).There are cases, usually experimental, in which the presence of systematicerrors may be unimportant (e.g. how does the number of Ephemeropterachange with nutrient concentrations?). A constant additive systematic errorcancels out in an additive model (such as an analysis of variance on untransformeddata), and a constant proportional systematic error cancels out in a multiplicativemodel (such as an analysis of variance on log-transformed data).Factors affecting sampling efficiencySampling devices that yield the number of individuals collected per unit area orvolume sampled do not give accurate (and generally give under-) estimates ofabsolute densities prior to correcting for sampling efficiency. For example, individualslocated in the area to be sampled may escape a moving net (Fleminger &Clutter 1965) or contents of artificial substrate samplers may be lost when liftingthe samplers (Rosenberg & Resh 1982). If absolute densities are necessary,it may be possible to calibrate the sampling device — i.e. to determine thesampling efficiency of the device by sampling under conditions where exactdensities are known. We deal with specifics of calibration in the next section.Calibration may not be necessary if relative densities (e.g. number per 1 msweep), and not absolute densities are sufficient. However, in both cases, samplingefficiency for a particular sampling device may vary with environmentalconditions or across species or size classes. This becomes particularly problematicwhen the researcher is comparing densities in different environments andsampling efficiency varies between environments. Similarly, it is particularlyproblematic when the researcher compares densities of different species or sizeclasses within a species and it is assumed that sampling efficiency is constantacross these categories when in fact it is not. In such cases, part or all of a differencethat might be found may not be a true difference but instead the result of

SAMPLING AQUATIC INSECTS 195differential sampling efficiency. We believe these problems to be common onesthat are often ignored. We present some examples below, drawing largely fromour own experiences.Sampling efficiency can vary among species or among size classes within a speciesIf the goal of sampling is to acquire accurate estimates of age (or size) class structureof a population, it would be ill-advised to assume that the sampling devicesamples different age classes with equal efficiency. Two age classes within aspecies that display different behavior or different swimming/crawling speedsare likely to be sampled at different efficiencies. For example, the minnow trapsamples different age classes of fish differentially (Blaustein 1989) and this isvery likely the case for aquatic insects as well. A second example is dipping,which does not sample different instars of mosquito larvae with the same efficiency(reviewed in Service 1993). In general, later instars are more likely to escapeactive-operator sampling devices than early instars.Similarly, without knowing the sampling efficiency of a specific sampling devicefor each species of interest, a very different picture of community structuremight emerge depending on the sampling device used. For example, withouttaking into consideration possible differences in sampling efficiency amongspecies, dipterans would be considered the dominant aquatic insect species inrice fields based on dipping, but coleopterans and hemipterans would be thedominant species based on aquatic light traps (Washino & Hokama 1968).Sampling efficiency can vary as a function of densitySampling efficiency with dipping may drop with increasing mosquito larvaldensity because an alarm reaction by mosquito larvae, which causes them to descend,may increase with increasing larval density (Thomas 1950, reported inService 1993). If this is the case, and if there are large true differences in the densitiesof two treatments, then the difference observed by dipping between twotreatments may be underestimated. The opposite — i.e. an overestimate of atreatment effect — is also possible. Imagine that some treatment effect, e.g. apesticide, results in truly lowering the density of odonate naiads. At low densities(the pesticide-treated ponds), most individuals may occupy the undersideof rocks, a specific habitat that is sampled at a low efficiency. At high densities(the non-treated ponds), perhaps many individuals are then relegated to a differentmicrohabitat, such as the upper side of rocks, that is sampled at a higherefficiency. In this case, without considering the differential efficiency as a functionof density, the treatment effect would be overestimated.Sampling efficiency can vary with respect to vegetation type and densitySampling devices, particularly those in which the collector actively samples,

196 CHAPTER 9have differential sampling efficiencies in open versus vegetated habitats, indifferent densities of vegetation, or in different types of vegetation (Turner &Trexler 1997). For example, sweep-netting benthos with a D-net may workquite well, yielding high sampling efficiency, in flexible submergent vegetationsuch as Chara or Najas species, but this device is pretty much useless in rigidemergent vegetation such as rice or cattail. Efficiency in passive-operatordevices across different vegetation types likely varies less than with activeoperatorsampling devices, but may still exist. For example, we would expectthat aquatic light traps should attract a smaller proportion of individuals as densityof vegetation increases. Turner and Trexler (1997) assessed invertebratespecies richness in different vegetation types using many types of sampling devices.Had they used only a stovepipe (column) sampler, they could have concludedthat invertebrate species richness was quite similar in sawgrass andspikerush habitats. Had they used only a Hester–Dendy sampler, they wouldhave concluded that species richness was higher in sawgrass.Similarly, this problem is very likely to occur when comparing forested (i.e.shaded) habitat versus habitat open to direct solar radiation. Open habitats willlikely contain high densities of filamentous algae, which in turn will affectsampling efficiency of active-operator devices such as sweep nets.Sampling efficiency can vary across substrate typeActive-operator sampling devices of benthos along a coarse (stony) substrateare likely to be less efficient than sampling in fine substrate. Based on our samplingsalamander larvae with D-frame sweep nets in pools with large stones versuspools with fine mud substrate, we might have concluded that there weremany more larvae in the fine substrate pools than in the stony-bottom pools.However, some of these pools dried shortly after sampling them with D-nets,giving us the opportunity to get a rough idea of our sampling efficiency since wecould then get a rather accurate count of the total number of larvae in thesepools. We found that larval densities were not lower in the stony pools, as oursampling indicated, but that sampling efficiency was considerably lower in thishabitat. This lesson should apply for many benthic insects as well.Another pilot study of ours illustrates the potential problem of differentiatingbetween real differences in size class distributions of a species across habitatsand differential sampling or counting efficiency across habitats. We attemptedto compare the size structure of libellulid dragonfly nymphs in two locations ina small pond: shallow water close to the shore, and deep water far from theshore. We used sweeps with an aquatic D-net in both locations and measuredthe head widths of each nymph we caught in the field. The preliminary datasuggested that larvae were, on average, larger near the shore (mean head width2.9 mm, range 1.3–6.4 mm, n = 42) than far from shore (mean head width1.5 mm, range 0.6–6.0 mm, n = 122). It is quite possible that there is a real differencein size class structure at the two locations (if, for example, most oviposi-

SAMPLING AQUATIC INSECTS 197tion occurs on vegetation near the center of the pond, young larvae tend to befound close to the oviposition site, and older larvae disperse). However, thenear-shore sample was filled with mud while the far-shore sample was muchcleaner. Efficiency to catch different size classes with a sweep net may differunder these different habitats. Even more likely, the probability of detecting thesmaller individuals in the muddy sample may have been lower when processingin the field.Sampling efficiency may be influenced by indirect effects of an interacting speciesA species that is manipulated and alters the environmental conditions can affectsampling efficiency. Suppose that we wish to assess the effect of a predator onlarval chironomid densities in rock pools. Our experimental design consists ofcontrol (no predator) pools and pools containing the predator. Perhaps wechoose to use a sweep net to estimate abundances of various invertebrate taxa.Predators, either by reducing herbivores or by nutrient recycling, can cause increasedamounts of filamentous algae in small pools (e.g. Blaustein et al. 1996).Sweep-netting through waters containing heavy mats of filamentous algae(predator pools) is likely to be differentially efficient compared with waterswithout such mats. Now suppose that we measure 50 percent fewer chironomidlarvae in predator plots than in non-predator plots based on sweep samples.Is all or part of this reduction due to the direct consumptive effects of the predatoron chironomid larvae? Alternatively, is it possible that the predator has littleor no effect on the chironomid densities but that sampling efficiency for chironomidsis simply much higher in control (low filamentous algae) pools?Sampling efficiency can be influenced by differential behavioral responsesacross treatmentsCommonly manipulated factors in aquatic studies such as sublethal effects ofchemicals and risk of predation can influence behavior. Over the past twodecades, considerable information has accumulated that many aquatic insects,in response to risk of predation, will reduce their activity (Lima 1998). The predatormay also affect the behavior of the prey species, which may in turn affectthe proportion of prey caught. For example, chironomid larvae swim in thewater column. Thus, we might be able to assess relative densities of chironomidlarvae by sweep-netting the water column or, depending on water visibility,counting swimming chironomid larvae. We did this in shallow artificial poolswhere predaceous fire salamander larvae were manipulated and before anybuild-up of filamentous algae (S.S. Schwartz et al. unpublished data). When wecompared the number of chironomid larvae per sweep sample in the presenceand absence of salamanders, the predator caused nearly a 100 percent reductionof chironomid larvae counted. With these data alone, we might concludethat this predator reduces densities of chironomid larvae by nearly 100 percent.

198 CHAPTER 9However, when we counted the number of chironomid larvae in their casesor the number of chironomid pupal exuviae on the water surface, we foundlittle or no difference between control and predator pools. Thus, it seemsthat the predators largely influence the behavior (reduced swimming) but actuallyhave little if any effect on the abundance of chironomids. Here, the treatmentfactor (the predator) altered the proportion of chironomid larvae caughtin our net.Dealing with systematic errorsIn many cases, the aim of a sampling program is to estimate the true value of aquantity (e.g. “how many dragonfly larvae are there in the pond?”). Ecologistsmay want to compare their estimates, not only with estimates from another experimentaltreatment sampled in the same way, but with published estimatesfor other species or other habitats, or with predicted values obtained from theory.In the previous section, we emphasized that knowledge of systematic samplingerror is important even for estimates of relative abundances whensampling efficiency varies across environmental conditions.In order of preference, here are some ways in which one might attempt todeal with systematic errors when sampling.1 Directly estimate and correct for systematic errorsIt may be possible to set up situations in which the true answer is known, andcalculate a calibration curve. For example, Stewart and Schaefer (1983) wantedto calibrate estimates of the density of larval mosquitoes in rice fields obtainedby sampling 1 m 2 enclosures with dippers. They set up enclosures intowhich known numbers of larvae were introduced, and estimated the meannumber of larvae per dip. The calibration problem is then simply to find a gooddescription of the relationship between the true number and the measuredvalue: in this case, a linear regression was used, but other kinds of relationships(logarithmic, quadratic, etc.) might better fit the data. To be of practical use, thecalibration curve must fit the data well because the goodness of fit determinesthe precision with which true values can be estimated. The true value should bethe predictor (because it is assumed to be without sampling error) and the measuredvalue should be the response. When subsequently applying the calibrationcurve, one needs to estimate the true value given the measured value, by rearrangingthe equation. This is known as inverse prediction (Sokal & Rohlf 1995).For example, if the calibration curve isYˆ= a+bX(9.1)where Ŷ is the predicted measurement, a is the intercept, b is the slope andX is the true value, then one should estimate true values from measurementsusing

SAMPLING AQUATIC INSECTS 199ˆX Y -=ab(9.2)where Xˆ is the inverse-predicted true value, Y is the observed measurement anda and b are the intercept and slope from Equation 9.1. It would not be correct toestimate a calibration curve by regression using the measured values as predictorsand the true values as responses, because this is contrary to the assumptionthat the predictor is without error. One last point about such calibration curvesis that the values of the parameter estimates are meaningful in themselves. Ina simple linear calibration, a non-zero intercept indicates an additive componentof systematic error and a slope that is different from one indicates a proportionalcomponent of systematic error.There are cases — e.g. a large, spatially heterogeneous lake — in which itwould be very difficult to construct a calibration curve. In these situations, thefollowing methods should be more practical.2Take test samples with different methods likely to have different kinds ofsystematic biasesIf the difference between the mean estimates obtained by different methods ismuch lower than the standard error of any of those estimates, one can be reasonablyconfident that the systematic errors are small enough to ignore, and usewhichever method is most convenient. If time and money allow, one mightcontinue to use several methods. Southwood (1978, p. 4) suggests weightingthe estimate from each method by the inverse of its variance.It is important that the methods are sufficiently different from each other sothat they are not likely to have the same kind of systematic error. For example,kick samples taken with three different sizes of net and three different durationsof sampling probably all suffer from the same kinds of bias. On the other hand,kick samples, grab samples, and artificial substrate samples likely suffer fromquite different biases, so showing that all three gave similar results would be astrong argument that the biases are small enough to ignore. How small a differenceis “small enough to ignore”? This is a matter of judgment, but one needs tothink about the absolute difference between the results from different methods,the smallest difference between two measurements which one would think ofas “important,” and the standard errors of estimates obtained by each method.A difference between methods that is large relative to the smallest “important”difference and to the standard errors of both methods is cause for concern.Southwood (1978, p. 4) suggests formal statistical tests, but common sense alsohelps.3Take samples by several sufficiently different methods (as above), estimatethe size of the systematic errors, and carry these systematic errors throughsubsequent calculationsWe illustrate this with a simple case study. To estimate the ratio of mosquito

200 CHAPTER 9pupae to larvae in a small rock pool (0.6 m long ¥ 0.3 m wide ¥ 0.09 m deep), wetried two sampling techniques. The water was clear, so we first counted all thepupae and larvae we could see in one minute (which we felt was long enough tocount all those visible at any time). We repeated this count ten times. Then weswept an aquarium net twice along the length of the pool, counted the numbersof pupae and larvae caught in the net and returned them to the pool. We alsotook ten samples in this way. For each replicate sample in each method, we calculatedthe ratio of observed pupae to larvae. We estimated the expected meanand standard error of the ratio with each number of sampling units from one toten, using a non-parametric bootstrap (Hilborn & Mangel 1997).As we would expect, the mean ratio from 1000 bootstrap replicates does notchange with the number of sampling units for each method, and the standarderror decreases as the number of sampling units increases (Fig. 9.1). However,visual counts give a consistently lower and less variable estimate of the ratio ofpupae to larvae (10 sampling units: bootstrap mean 0.69, standard error 0.05)than net sweeps (10 sampling units: bootstrap mean 0.95, standard error 0.10).The difference between the bootstrap means from 10 sampling units is 0.26,which is clearly not negligible (5.77 standard errors of the visual counts, or 2.73standard errors from the sweeps).Which, if either, is the better estimate? Our first guess might be visual countsbecause of the smaller standard error. However, it is quite possible that visual1.21.00.8Pupae/larvae0.60.40.2visualsweep002 4Number of sampling units68 10Fig. 9.1 Mean estimates of the ratio of mosquito pupae to larvae in a small rock poolobtained by visual counts (open circles) and net sweeps (filled circles), with standard errors.Means and standard errors were estimated by a non-parametric bootstrap with 1000replicates for each number of sampling units.

SAMPLING AQUATIC INSECTS 201counts are highly repeatable (precise) but biased. For example, it might havebeen harder to see pupae than larvae because of differences in behavior. Thestandard error for the sweeps might be larger because each sweep captures asmall and variable fraction of all the individuals in the pool. On the other hand,we might be overestimating the ratio of pupae to larvae by net sweeps becauselarvae are better than pupae at avoiding the net. We have no reason to believethat there is any particular relationship between the sizes of the random andsystematic components of error. For a small rock pool, with enough patience,we could probably get much closer to the true ratio of pupae to larvae by emptyingthe pool, filtering all the water through a net, and searching the materialremaining in the pool. Even so, there would still be biases. The smallest larvaeare much smaller than pupae, so we might be more likely to miss larvae thanpupae. We could also set up artificial pools with known numbers of pupae andlarvae and estimate a calibration curve for either sampling method (as describedabove). However, there are many cases in which no obviously better method isavailable, and a calibration curve cannot be constructed. The best way of expressingour ignorance is to use both methods, and carry out all subsequentanalyses using the values from each method separately. Formally, we could usethe interval [0.69, 0.95] as a way of expressing the range of possible values of thepupae : larvae ratio, and we would obtain other intervals representing the resultsof any subsequent calculations.Figure 9.2a shows the separate proportion histograms for the bootstrap estimatesof the ratio of pupae to larvae from each sampling method (with ten samplingunits in each case, and 1000 bootstrap replicates). We might be tempted toaverage them, but this would almost certainly be wrong. The result of averagingthe two distributions in Fig. 9.2a has a mean of 0.82, with 95 percent of valueslying between 0.70 and 0.92 (Fig. 9.2b). This seems to suggest that the true ratioof pupae to larvae is exactly halfway between the means obtained by each of thetwo methods, and that a true value as extreme as the mean of either of themethods alone is quite unlikely. This will only be correct if the systematic errorsin the two methods are equal and opposite, for which we have no evidence. Possibilitytheory (Dubois & Prade 1988) provides an alternative way to deal withthese uncertainties. We might reasonably decide that the median estimate fromeach method (0.69 for visual counts or 0.95 for sweeps), or any value betweenthese medians, was an “entirely possible” value, given the information currentlyavailable. We could give the interval [0.69, 0.95] the possibility level 1.The lowest bootstrap estimate of the ratio we ever obtained in 1000 replicateswas 0.53, and the highest was 1.20 (from visual and sweep data respectively).We could treat this interval as the range of values that are “just possible,” with apossibility level just above zero. Between possibility levels zero and one, thereare infinitely many other intervals, each with a different possibility level. In particular,we might be interested in the interval between the lowest observedlower 95 percent confidence limit and the highest observed upper 95 percentconfidence limit (in this case, 0.60 to 1.13). This is not itself a confidence

202 CHAPTER 91.0(a)1.0(b)Probability0.80.60.40.2visualsweepProbability0.80.60.40.2visualsweep000.50.60.70.80.91.01.11.20.50.60.70.80.91.01.11.21.0 (c)0.8Probability0.60.40.200.50.7 0.9 1.1Pupae/larvaeFig. 9.2 (a) Proportion histograms for bootstrap estimates of the ratio of mosquito pupae tolarvae in a small rock pool (open bars are visual counts, solid bars are sweeps, 10 samplingunits and 1000 bootstrap replicates in each case). The labels on the abscissa are the midpointsof every second bin. (b) Proportion histogram for the mean ratio of mosquito pupae to larvae,averaging bootstrap estimates from visual counts and sweeps as in (a) and assuming errorsare independent. The abscissa is labeled as in (a). (c) A fuzzy number representing thepossible values for the ratio of mosquito pupae to larvae, derived from the bootstrap estimatesin (a).interval, but it is an interval that encloses the 95 percent confidence intervals forthe ratio of pupae to larvae obtained by both sampling methods. Thus, weshould give this interval possibility level 0.05 (although the exact possibilitylevel is not particularly important so long as the ordering of possibility levels isright). Figure 9.2c shows the set of intervals for a range of possibility levels. Thisis known as a fuzzy number. Fuzzy numbers are often more satisfactory thanprobability theory for dealing with uncertainty (Ferson & Kuhn 1992), and canbe manipulated by a consistent set of arithmetical operations (Dubois & Prade1988).The fuzzy number we have defined here would change if we used differentsampling methods. If we used only one sampling method, we would obtain asingle best estimate. By using more sampling methods, we apparently becomeless certain about the best estimate of the ratio of pupae to larvae. If we really donot know how each method is likely to be biased, this is a reasonable reflectionof our subjective uncertainty. Of course, we have no guarantee that the truevalue lies within the range of values that we consider to be possible. Even if we

SAMPLING AQUATIC INSECTS 203used many different sampling methods, they might all be biased in the same direction.We can make this unlikely by using several methods that are likely to besubject to completely different kinds of bias. In our case study we can concludeonly that, based on our current knowledge, the ratio of pupae to larvae in therock pool we studied is possibly between 0.69 and 0.95, and that the 95 percentconfidence interval is 0.60 to 1.13.Random errors: how many sample units?Random errors, combined with the desired precision to estimate a value, determinethe number of sample units. Asking a well-defined question that includesthe desired precision will help to ensure that the sampling program is designedappropriately. We will discuss some design considerations for seven suchquestions.1 What is the average density of a species in a pond (with a standard errorno greater than 5 percent of the mean)?To answer this question, one needs to know variability among sampling units.Taking a few pilot samples can yield this information. A general formula for thenumber of samples required issn = Ê ˆË Ex ¯2(9.3)where n is the number of samples required, s is the estimated standard deviation,E is the required standard error/mean ratio and x¯ is the estimated mean(Southwood 1978). If the precision needed can be expressed in terms of thestandard error, this formula can be applied no matter what the distribution ofthe population.It is often useful to express precision as a function of the width of a confidenceinterval rather than a standard error. Calculating the number of samplesneeded to achieve a given width of confidence interval is more complicated,because the confidence interval depends on the population distribution aswell as the standard error. At the planning stage, approximate confidence intervalsbased on the t distribution (assuming a normal distribution of the sampledvariable) may be good enough, especially if working with large means:where D is the desired ratio of half the width of the 100(1 - a) percent confidenceinterval to the mean and t a,n-1is the critical value of the t distribution fora given a and n (Southwood 1978). How large is “n large”? From a set of statisti-t n sn = Ê a,-1ˆË Dx ¯2sª Ê 2 ˆË Dx ¯if n large and a = 005 .(9.4)

204 CHAPTER 9cal tables (e.g. Rohlf & Sokal 1995), it can be seen that t converges to a stablevalue as n increases, and that for a of 0.05, t is approximately 2 if n is more thanabout 30 (to the level of accuracy needed in planning a sampling program).Equation 9.4 strictly applies only to normally distributed data. However,most populations are not normally distributed. For example, counts can only bezero or positive integers. Although transformations can be used to make thedata approximately normal, one is more likely to be interested in properties ofthe untransformed data (for example, the arithmetic mean) than of transformeddata (for example, the geometric mean that results from back-transforminga log-transformed dataset). If working with count data, a formulabased on the Poisson or negative binomial distributions may be appropriate.The number of samples needed in these cases isnta2, n-1sÊ 1 1ˆD Ë x k¯ ª 4 Ê 1D Ë x + 1ˆ22k¯= ( ) +if n large and a = 005 .(9.5)where k is an estimate of the negative binomial exponent (Krebs 1989). For aPoisson distribution, k is •. With smaller values of k (corresponding to a moreaggregated population), the number of samples needed to achieve a givenwidth of confidence interval will be larger. Several methods can be used to obtaina preliminary estimate of k. Iterative solution of the following equation isthe best method if the number of pilot samples is fairly large:xN lnÊ1+ˆË k ¯ =•Ax ( )Âx = 0k+x(9.6)where A(x) is the sum of frequencies of sampling units having more than x individuals,and N is the number of sampling units in the pilot study (Crawley1993). To solve for k, a first guess can be made (the estimate of k from Equation9.7 below is reasonable) and left- and right-hand sides of Equation 9.6 are calculated.The guess is too large if the left-hand side is greater than the right-handside, and vice versa, so the value should be adjusted such that the equation balances(minimizing the difference between the two sides using a spreadsheet ormathematical software is the quickest way). However, if N is small, it may not bepossible to obtain a solution to Equation 9.6. In these cases, a rough estimate isk =s2x2- x(9.7)After trying the best estimate of k in Equation 9.5, it may be a good idea to alsotry a slightly lower value (which will give a slightly higher estimate of the numberof samples needed). If a small change in k makes a large difference in thenumber of samples needed, one should be conservative and take more samplesthan is thought necessary.

SAMPLING AQUATIC INSECTS 205What is a reasonable level of precision to aim for? The natural variabilityof the population sets a limit to the precision that can be achieved in practice.For benthic invertebrates, a 95 percent confidence interval around ±30–55 percentof the mean is often considered to be “moderate” precision, while a 95 percentconfidence interval around ±10–25 percent of the mean is “high” precision(Norris et al. 1992). It is also important to think about the size of the samplingunits. When the mean density of the organism is high, smaller sample units canbe more cost-effective, but at low densities larger sampling units avoid the problemof zero counts, which are often difficult to analyze (Norris et al. 1992).However, ecological considerations (it is not sensible to use sampling units largerthan the scale at which one is interested in estimating density) and practicalconstraints (many sampling devices are only available in a few different sizes)may constrain choices.Where should samples be taken? Deciding to estimate average densityimplies that microhabitat variations (such as shallow areas vs. deep areas)are not of particular interest. Nevertheless, to avoid any bias (and to allow theopportunity of examining small-scale patterns in density if one later decidesthis is necessary), the aim should be to sample each microhabitat in proportionto the fraction of the total habitat size that it contributes. Systematic or stratifiedrandom sampling is a good way to achieve this, although pure random samplingwill be reasonable if many sampling units are taken. One pitfall is that theappropriate arrangement of sampling locations depends on whether oneis sampling organisms that use the whole water column, or organisms thatuse only the surface or the benthos. If sampling organisms that use thewhole water column, one needs to arrange sampling units so that each kind ofmicrohabitat is represented in proportion to the fraction of habitat volumethat it contributes. If sampling organisms that use only the surface or the benthos,one needs to arrange sampling units so that each kind of microhabitat isrepresented in proportion to the fraction of habitat area that it contributes.These two alternatives are only the same if depth is constant throughout thehabitat.2 How many individuals of a species are there in a sediment sample(to within 5 percent of the estimated asymptotic number)?This objective seems similar to the previous one (estimating the averagedensity in a defined region), yet it is sometimes more efficiently approachedwith a very different sampling design. In the previous case, one would get a biasedestimate of average density if sampling effort were concentrated whereone expects to find organisms. In this case, sampling where organisms are expectedis the most efficient way time-wise. Of course it is worth checking a fewunlikely places as well, to be sure that one’s ideas about where to look were correct.Also, it should be remembered that if sampling is conducted in this way,one will not be able to convert estimates of abundance into a valid estimate ofaverage density across the whole sample. Failing to recognize this distinction is

206 CHAPTER 9one of the likely causes of the negative relationship between sampling area andestimated density that is often reported across studies (Blackburn & Gaston1999, Gaston et al. 1999, Johnson 1999).We expressed our desired precision as “to within 5 percent of the estimatedasymptotic number.” This is a sensible approach if the sample can be searchedfairly completely, but there is some constraint on total searching effort. For example,suppose live counts of chironomid larvae in sediment cores are desired,but there is concern that if trying to maintain the cores in the laboratory formore than a day, the number of larvae may change (perhaps there are predatorsin the sample). One could search the sample under a dissecting microscope, removingeach larva as it is found and recording the time. At the end of the day,one could plot the cumulative number of larvae against the time at which eachlarva was found. The cumulative number of larvae should flatten out at highsampling effort. If an asymptotic function to this curve is fitted and the estimatedasymptote is within 5 percent (or whatever value one decides issatisfactory) of the final total, the sampling method is satisfactory.3 What is the ratio of densities of two species (with a standard errorno larger than 20 percent of the mean)?First, one should ask whether it is really necessary to estimate a ratio. Ratios andother derived variables often have much higher standard errors than directlymeasurable variables (Taylor 1982, Jasienski & Bazzaz 1999). The sampling distributionsof ratios do not lend themselves to standard statistical methods(Atchley et al. 1976). The relationships between variables with common components(for example, between X/Y and Y) are mathematically constrained.This can lead to two kinds of problems. First, if the shared measurement error islarge (for example, if most of the measurement error in X/Y results from measurementerror in Y), any relationship between the variables is determinedmainly by this shared error, and is unlikely to be informative (Prairie & Bird1989). Second, variables with very strong mathematical constraints may notcontain much biological information. For example, the ratio of the number ofpredatory to non-predatory species has a more or less constant value close to 1,across many food webs (Cohen 1978). However, this is simply a mathematicalnecessity, once one realizes that most species in most food webs are both predatorsand prey by these definitions (Closs et al. 1993). Ratios should be used carefullywhen they measure quantities of genuine interest and with due regard fortheir statistical peculiarities.Ratios of dependent variables are the source of much confusion in ecologicalliterature. For example, Krebs (1989) suggests the following estimate of themean ratio Rˆ of two variables x and y:ˆR =xy(9.8)

SAMPLING AQUATIC INSECTS 207This is often misleading, as is the associated standard error suggested by Krebs(following Cochran 1977). Unless x and y are independent, the ratio of themeans (Equation 9.8) is not the same as the mean ratio. The general formula forthe mean ratio isRˆ x x= -cov Ê,(9.9)y Ë y y ˆ¯ ◊ 1 yÊ xwhere cov , is the covariance between x/y and y (Welsh et al. 1988). Welshy y ˆË ¯et al. (1988) and Kirchner (1998) give formulae for means and variances of severaluseful functions, but for many cases the formulae only apply to certain specialdistributions (e.g. normal or lognormal), or to independent variables. Ofcourse, one can calculate the mean ratio and its standard error directly from theratio in each sampling unit, which is much easier. Simple formulae for confidenceintervals are rarely available. We suggest the following:a For a rough idea of the number of samples needed, calculate the mean ratioand its standard error, and use normal approximate confidence intervals to estimatethe number of samples needed for a given precision (Equation 9.4). Becausethe normal approximation is unlikely to be very accurate, it may be a goodidea to take more samples than the formula suggests.b In cases where the level of sampling effort must be determined accurately(for example, if the cost per unit effort is high), use simulation to estimate theexpected width of confidence intervals for a given level of sampling effort. Aftertaking some pilot samples, find a parametric distribution that describes eachcomponent of the ratio reasonably well, and estimate the parameters (e.g. themean for a Poisson distribution, the mean and k for a negative binomial distribution,the mean and variance for a normal distribution). Estimate the correlationbetween the variables. For a range of proposed numbers of sampling units,generate many (perhaps 1000) random datasets with the appropriate numberof sampling units, distribution of each component variable, and correlation betweenthem. Special software is available, or one could code one of the simplealgorithms for generating correlated random variables (e.g. Nelsen 1986,1987). Estimate the confidence limits as the 100(a/2) and 100(1 -a/2) percentilesof the distribution of mean ratios over all random datasets of a givennumber of sampling units. Choose the lowest number of sampling units forwhich the confidence interval is narrow enough.4 What is the difference in density of a species between two ponds or habitats(with a 95 percent confidence interval of the difference no wider than 5individuals per m 2 )?To keep things simple, we will assume that some transformation of the

208 CHAPTER 9distribution of sample mean density estimates in each pond can make the estimatesmore or less normally distributed with similar variances. Our illustrationis based on pages 223–5 in Sokal and Rohlf (1995). Given two sample means Ȳ 1and Ȳ 2, our best estimate of the difference D between them is simply Ȳ 1- Ȳ 2. Thestandard error of the difference (s D) issD =2( ) + ( - )È n1-11stn 1 sÍÎ n1+ n2-22 2 2˘Ên1+n2ˆ˙˚Ënn 1 2¯(9.10)where s iis the sample standard deviation and n iis the number of sampling unitstaken from population i. The 100(1 -a) percent confidence interval for the differencebetween the means can then be calculated from the t distribution( Y1-Y2) ± t a[v ] s D(9.11)where the degrees of freedom (n) are n 1+ n 2- 2. The best approach is to takepilot samples in each pond and estimate Ȳ 1- Ȳ 2, s 1and s 2. Then we calculate howthe width of the 95 percent confidence interval changes as we substitute differentvalues for n 1and n 2, and choose values that are sufficient to achieve our aim.The width of 95 percent confidence interval chosen as acceptable is a way of indicatingthe range of estimates that we would be prepared to think of as more orless the same.For example, we ran a pilot study to assess the use of minnow traps to comparedensity estimates of libellulid naiads in shallow and deep water in a smallpond. We set up five traps in each habitat type and counted the number of larvaethey contained after three hours. The raw data are shown in Table 9.1.Given the size of these means, we think we would like to know the differencebetween them (9.2 - 5.4 = 3.8) with a 95 percent confidence interval no widerthan 2 individuals per trap (this is approximately ±26 percent of the difference,Table 9.1 Numbers of libellulid larvae sampled by minnow traps in three hours in twodifferent habitat types (shallow water near shore and deep water far from shore) in a smallpond.ShallowDeep9 76 1012 612 37 1Mean 9.2 5.4Variance 7.7 12.3

SAMPLING AQUATIC INSECTS 209or “high” precision for benthic invertebrates (Norris et al. 1992). As the numberof samples is small, we cannot be sure what kind of distribution would best describethe data, so we use the untransformed values. This will only give us an approximateestimate of the sample size required. Figure 9.3 shows the estimatedwidth of the 95 percent confidence interval on the difference between themeans, for a range of sample sizes. To achieve a confidence interval no widerthan 2 individuals per trap, we would need at least 80 traps per habitat. This isnot practical; given the size of the pond, we could barely fit so many traps intoeach habitat at the same time. We could either settle for a less precise estimate ofthe difference with fewer traps or find a less variable sampling method.We have chosen to emphasize the estimation of a confidence interval for thedifference between two means rather than a p value for two reasons. First, calculatingthe size of a difference and its confidence interval is much more informativethan simply stating a p value (Harlow et al. 1997). P values combinethe size of a difference and the precision with which this estimate is known intoa single number. The same p value could come from a small difference with highprecision or a large difference with low precision, yet we would interpret thesetwo results quite differently. Second, meta-analyses are increasingly importantin ecology (Osenberg et al. 1999), and are based on estimates of effect size ratherthan p values. Routinely thinking about effect size rather than p values will thusimprove our understanding of our data, and serve as the foundation for understandingmodern statistical tools. Estimating confidence intervals on measuresof effect size is closely related to power analysis. Thinking about power at the designstage is important because if we don’t design our sampling program so as tohave sufficient power, we will not only waste our time, but may be tempted todraw misleading inferences. For a readable introduction to power analysis, seeMurphy and Myors (1998). Formulae for calculating the sample size needed to30Width of 95% confidenceinterval25201510500 204060 80 100Number of traps per habitatFig. 9.3 The width of the 95% confidence interval for estimating the abundance of libellulidsper trap as a function of the number of minnow traps set.

210 CHAPTER 9achieve a specified power for a specified difference between means are given byZar (1984, pp. 133 & 193) and Sokal and Rohlf (1995, p. 263), and are discussedby Norris et al. (1992). What level of power should we aim for? Power below 0.5is worse than useless. Assuming that there is really a difference between twomeans (and the probability of there being absolutely no difference is infinitelysmall), any failure to detect such a difference is a type II error. With power lessthan 0.5, we would therefore make fewer errors by flipping a coin than by doinga survey or experiment (Murphy & Myors 1998). The minimum level of powerfor which one should aim is often suggested to be 0.8 (Murphy & Myors 1998).Even with a power of 0.8, the chance of two experiments on identical systemsyielding consistent results (either both statistically significant or both not statisticallysignificant) is only 0.68 (Gurevitch & Hedges 1999).5 How does the age structure of the species change through the season, on a timescale short enough to estimate the survival probability of age classes of 2 weekseach, with a standard error less than 0.2 in each estimate?Such a question can be addressed by mark–recapture techniques. Marking softbodiedaquatic insects is more problematic (though radioactive isotopes may beused; e.g. Croset et al. 1976) than marking hard-bodied organisms such as adultbeetles (Nürnberger 1996, Svensson 1999). Krebs (1989) suggests a number ofmethods for estimating survival from marked individuals, each with differentrequirements and different ways of obtaining standard errors. If one is not ableto mark individuals, survival probabilities can be estimated from time series ofsamples. This is difficult (Caswell & Twombly 1989, Manly 1990, 1997, Wood1997). Equal sampling intervals, the same as the width of the age classes, willmake things a bit easier. If using stages rather than age classes, the sampling intervalshould be no longer than the minimum duration of the shortest stage,and again using equal sampling intervals is desirable. To avoid wasting samplingeffort, one should first make up some plausible data, using known survival parameterswith added random error based on fairly pessimistic estimates of theamount of sampling variability one expects to encounter. Then the data shouldbe run through the estimation process one intends to use. If unable to recoverthe known parameters, the sampling program will unlikely be worthwhile.6 How many samples are needed to obtain a good estimate of the speciescomposition of a community?Most species have low relative abundances (May 1975), and it would take atremendous amount of effort to enumerate them all. If we intend to study the“real” distribution of relative abundances in the community, we should be preparedfor an exhaustive sampling program (e.g. Siemann et al. 1999). On theother hand, we might only be interested in those species that are abundantenough to be important in the community. A sensible approach is to decide on

SAMPLING AQUATIC INSECTS 211(and explicitly state!) a working definition of “important” in terms of relativeabundance, and make a rough estimate of the sampling effort needed to detectthe rarest important species. To do this, we need to know the abundance–frequency distribution in the community, and the relationship between abundanceand detection probability. To solve this problem before carrying out asampling program, we will have to make some guesses. We might also be interestedin evaluating the effectiveness of a sampling program that has alreadybeen conducted, in which case the necessary data will already exist.Abundance–frequency curves provide a way to think about the consequencesof a given decision about the lowest level of abundance we would liketo detect. For example, suppose that we want to detect the most common y percentof the species. How rare is the rarest of these y percent? Or suppose we decidethat we are not interested in detecting a species with an abundance lowerthan x, what proportion of species in the community will we be ignoring? Figure9.4 shows the abundance–frequency distribution for invertebrates (mainly insectlarvae) in snag habitats of a subtropical blackwater river (data from Benkeet al. 1984, upper snag site). For many communities, the abundance–frequencydistribution is approximately lognormal (May 1975). Fitting a lognormal1.0Cumulative proportion spp0.80.60.40.20–2 02 4ln (individuals per m 2 )6 8 10Fig. 9.4 Abundance–frequency distribution for invertebrates in snag habitats of a subtropicalblackwater river (data from Benke et al. [1984], upper snag site). The circles are the observeddata, with short dashed lines indicating the 95% Kolmogorov–Smirnov confidence intervalfor an intrinsic hypothesis (one in which we estimate the mean and variance of a fitteddistribution from the data themselves; Sokal 1995). The solid line is a fitted lognormaldistribution, assuming that the lowest observed abundance is the lowest observableabundance. The long dashed line is a fitted lognormal distribution, estimating the lowestobservable abundance from the sampling effort.

212 CHAPTER 9distribution to observed data usually requires a correction for the lowestabundance that could have been observed (described by Magurran 1988 and byKrebs 1989). In count data, the correction is usually the logarithm of 0.5 (thelower boundary of the log abundance class in which only one individual wasobserved). However, the data of Benke et al. (1984) are measured as numbersper m 2 , so we have to divide the (natural) logarithm of 0.5 by the total areasampled. A total of 96 samples were taken, with areas usually between 0.04and 0.01 m 2 , so the lowest detectable abundance might possibly have been aslow as ln(0.5 ¥ 96/0.04) =-2.04, or as high as ln(0.5 ¥ 96/0.01) =-0.65. The lowestnatural log abundance actually observed was -1.20, so we fitted lognormaldistributions with lowest detectable natural log abundances of -2.04 and-1.20. These alternatives make little difference to the predicted distribution(Fig. 9.4). Moreover, both fitted distributions lie well within the 95%Kolmogorov–Smirnov confidence interval (Sokal & Rohlf 1995) for theobserved distribution. In a retrospective analysis, we could conclude that thelognormal distribution was a reasonable description of our data.For a prospective analysis, we are unlikely to have data like those in Fig. 9.4,but we might still be able to make a plausible guess at the abundance–frequency distribution. The “canonical lognormal” distribution predicts a relationshipbetween the number of species and the variance in log abundanceS = s pÊ2 expÁË( s ln 2)22ˆ˜¯(9.12)where S is the true number of species in the community and s 2 is the variance inlog abundance. It has been forcefully argued that this relationship is a consequenceof the way in which niche space is partitioned (Sugihara 1980). If wefind Sugihara’s argument convincing, and we can guess the true number ofspecies and either the mean or the total log abundance, we can also make aguess at the abundance–frequency distribution. Hypotheses about how theworld works are very often used as the justification for selecting a particulardistribution to describe data. For example, normal or lognormal distributionsare widely used even when data are too scarce to discriminate among the set ofdistributions that might reasonably be used, because the central limit theoremprovides a theoretical basis for the normal and lognormal (Hattis & Burmaster1994).The sample actually contained 29 distinguishable taxa at the upper snag site,with a total of 33,300 individuals per m 2 . The estimated variance in natural logabundance from the fitted lognormal distributions was 7.05 to 7.51 (assumingthat the lowest detectable natural log abundance was either -2.04 or -1.20, asabove). Suppose that, based on past experience or literature surveys, we hadguessed that we might find somewhere between 20 and 40 species, and somewherebetween 20,000 and 40,000 individuals per m 2 (even if we don’t knowvery much about the habitat, we will probably be able to make guesses like

SAMPLING AQUATIC INSECTS 213these). The mean abundance might then be as low as 500 or as high as 2000 individualsper m 2 . Using Equation 9.12, we would predict a variance of 7.3 to 9.7in natural log abundance.Once we have fitted or guessed a distribution, we can answer questions like“how rare is the rarest of the most abundant y percent of species?” and “whatproportion of species will be missed if we ignore those with abundance lowerthan x?” For example, how rare is the rarest of the most abundant 95 percent ofspecies in the data of Benke et al. (1984)? This is equivalent to finding the 5thpercentile of abundance, for which we need the inverse normal distributionfunction, and the mean and variance of natural log abundance (the inverse normaldistribution function is available in most spreadsheets and statistical programs).Given the means and variances we guessed, the 5th percentile ofabundance might be as low as 2.98 or as high as 23.49 individuals per m 2 . Thecurves we fitted to the observed distribution have a 5th percentile of 0.84 to 1.05individuals per m 2 , so the abundance at the 5th percentile is actually considerablylower than our guess. As another example, suppose we decide that we willnot attempt to detect species with a density lower than 1 individual per m 2 .What proportion of species will we be ignoring? This requires the cumulativenormal distribution function, which is also readily available. From our guessedmeans and variances, we would expect to be ignoring somewhere between 0.2percent and 2 percent of the species. From the fitted curves, we would expect tobe ignoring somewhere between 5 percent and 6 percent of the species. Ourrough guesses were clearly a little overoptimistic, but they are close enough tobe useful at the planning stage. For a retrospective analysis, we could use theabundance curve fitted to the data we actually observed.Assuming we have decided on the rarest species of interest, how much effortwill we need to expend to be reasonably sure of detecting these species? Wewould need to have some idea of the probability of detecting an individual animalin a sampling unit, and the distribution of true abundances in samplingunits for a given mean abundance. In a prospective analysis, we might try to estimatethe probability of detecting an individual animal from a few preliminarytrials (in which we could make artificial samples containing known numbers ofindividuals). In a retrospective analysis, we could estimate the probability thatsome individuals were missed by repeat counts of samples. The distribution oftrue abundances in sampling units depends on how the mean and the varianceof abundance are related. Again, this can be established either from preliminarytrials or retrospectively. We can estimate the number of sampling units neededto detect the species with a given probability as( )( )ln 1 - aN =ln 1 - p(9.13)where N is the number of sampling units needed, a is the desired probability ofdetecting the species, and p is the probability of the species being present in a

214 CHAPTER 9250Number of units needed20015010050alpha = 0.7alpha = 0.8alpha = 0.9000.050.1 0.15 0.2Probability of detection per unitFig. 9.5 Estimating the number of sampling units needed to detect a species (Equation 9.13).The abscissa is the probability that a species will be detected in a single sampling unit (p), andthe ordinate is the number of sampling units needed (N). Curves are plotted for threedifferent values of the desired probability of detecting the species in the entire sample (a).Solid line, a=0.7; dashed line, a=0.8; dotted line, a=0.9.single sampling unit (McArdle 1990). Figure 9.5 shows the relationship betweenp and N for three different values of a. The number of sampling unitsneeded depends very strongly on the probability that the species will bedetected in a single sampling unit, so we will need to work much harder toachieve a desired probability of detecting a rare than a common species.7 On which prey species does a given predator feed (including all species thatmake up more than 5 percent of prey individuals)?Properties of food webs such as the average number of feeding interactionsper species are sensitive to sampling effort (Martinez et al. 1999). Yield–effortcurves like the species accumulation curves we discussed in the previous sectioncan be used to determine how many feeding interactions are likely to havebeen missed across the whole food web. In general, a very large amount of effortis needed to thoroughly document a food web. The number of possible feedinginteractions is N 2 , where N is the number of species, and many of these possibleinteractions may occur very rarely. For example, Polis (1991) illustrated ayield–effort curve for the number of prey species of a single species of scorpion.After more than 2000 person-hours of field work on the whole food web,spread over five years, more than 100 different prey species had been recordedfor the scorpion, with no sign of an asymptote.

SAMPLING AQUATIC INSECTS 215If one is interested in the overall properties of food webs, there may be noalternative but to attempt a very large sampling program. On the other hand,one might only want to know about those prey species that make up a substantialpart of the diet of a predator. One could plot the frequency with which eachprey species is recorded in the diet (over all sampling units) against the cumulativesampling effort at which the prey species was first recorded. On average,prey species that make up a large proportion of the diet will tend to be firstrecorded earlier in the sampling program than prey species that make up a smallproportion of the diet. One could stop sampling when the relationship betweenfrequency and effort at first recording falls below some threshold (which definesthe frequency below which one thinks a prey species is not important).Ethical considerationsEthical considerations should be a part of any sampling program and may includeavoidance of: (i) killing more individuals than needed even when thepopulation is not under threat; (ii) increasing the probability of extinction of aspecies; and (iii) damaging or altering an environment. We suggest a few guidelinesfor ethical sampling.Excessive killing, even when the species is not endangered, receives more attentionwhen the animals sampled are mammals and other higher animals (e.g.Puttman 1995) than for insects. To minimize the number of individuals killed, itmight be possible to collect individuals live in the field, count them and returnthem alive. Regardless of whether one feels that “excessive” killing of a commoninsect species is an ethical consideration, collecting too many individualsmay actually significantly reduce the population and thus alter results of subsequentsamples. Sampling, counting, and returning live specimens may alsoallow for more samples without the researcher actually significantly reducingpopulations through sampling.Overall, damage to the environment should be a serious consideration forall field workers. Caution should be taken not to spill preservatives such asformaldehyde in the field. Active-operator sampling devices such as benthicnets or grabs can potentially perturb the environment while passive-operatorsampling devices such as traps and colonizing substrates cause little damage.Researchers are also probably responsible for unintentional introductions ofspecies due to propagules on sampling equipment or on boots. This may givefalse impressions of the degree of isolation of water bodies and the degree ofseparation of populations in later genetic studies. Unintentional introductionsmay also include disease agents. Herpetologists have expressed particular concernabout the unintentional spreading of pathogens to anuran species whichhas prompted the DAPTF (Declining Amphibian Populations Task Force) toadopt a “fieldwork code of practice” which calls for sterilizing boots and samplingequipment with alcohol.

216 CHAPTER 9AcknowledgmentsWe thank Vince Resh for useful discussion. The writing of this chapter was supportedby US–Israel Binational Science Foundation grants 95-0035 awarded toLB and Joel E. Cohen and 98-0390 awarded to LB and Marc Mangel.ReferencesAndis, M.D., Meek, C.L., & Wright, V.L. (1983) Bionomics of Louisiana riceland mosquito larvaeI. A comparison of sampling techniques. Mosquito News, 43, 195–203.Atchley, W.R., Gaskins, C.T., & Anderson, D. (1976) Statistical properties of ratios. I. Empiricalresults. Systematic Zoology, 25, 137–148.Benke, A.C., Van Arsdall, T.C. Jr., & Gillespie, D.M. (1984) Invertebrate productivity in a subtropicalblackwater river: the importance of habitat and life history. Ecological Monographs,54, 25–63.Blackburn, T.M. & Gaston, K.J. (1999) Density, survey area, and the perfection (or otherwise)of ecologists. Oikos, 85, 570–573.Blaustein, L. (1988) Biotic interactions in rice fields: an ecological approach to mosquito control.PhD thesis, University of California, Davis.Blaustein, L. (1989) Factors affecting the efficiency of minnow traps to sample populations ofmosquitofish (Gambusia affinis) and green sunfish (Lepomis cyanellus). Journal of the AmericanMosquito Control Association, 5, 29–35.Blaustein, L. & Kotler, B.P. (1993) Oviposition habitat selection by Culiseta longiareolata: effectsof immature conspecifics, tadpoles and food levels. Ecological Entomology, 18, 104–108.Blaustein, L. & Schwartz, S.S. (2001) Why study ecology in temporary pools? Israel Journal ofZoology, 47, 303–312.Blaustein, L., Kotler, B.P., & Ward, D. (1995) Direct and indirect effects of a predatory backswimmer(Notonecta maculata) on community structure of desert temporary pools. EcologicalEntomology, 20, 311–318.Blaustein, L., Friedman, J., & Fahima, T. (1996) Larval Salamandra drive temporary pool communitydynamics: evidence from an artificial pool experiment. Oikos, 76, 392–402.Caquet, Th., Lagadic, L., Jonot, O., et al. (1996) Outdoor experimental ponds (mesocosms) designedfor long-term ecotoxicological studies in aquatic environment. Ecotoxicology and EnvironmentalSafety, 34, 125–133.Caswell, H. & Twombly, S. (1989) Estimation of stage-specific demographic parameters forzooplankton populations: methods based on stage-classified matrix projection models. InEstimation and Analysis of Insect Populations (ed. L. McDonald, B. Manly, J. Lockwood, & J.Logan), pp. 93–107. Springer, New York.Chesson, J. (1984) Effects of notonectids (Hemiptera: Notonectidae) on mosquitoes (Diptera:Culicidae): predation or selective oviposition? Environmental Entomology, 13, 531–538.Closs, G., Watterson, G.A., & Donnelly, P.J. (1993) Constant predator–prey ratios: an arithmeticalartifact? Ecology, 74, 238–243.Cochran, W.G. (1977) Sampling Techniques. Wiley, New York.Cohen, J.E. (1978) Food Webs and Niche Space. Princeton University Press, Princeton.Crawley, M.J. (1993) GLIM for Ecologists. Blackwell, Oxford.Croset, H., Papierok, B., Rioux, J.A., Gabinaud, A., Cousserans, J., & Arnaud, D. (1976)Absolute estimates of larval populations of culicid mosquitoes: comparisons of capture–recapture, removal, and dipping methods. Ecological Entomology, 1, 251–256.

SAMPLING AQUATIC INSECTS 217Dubois, D. & Prade, H. (1988) Possibility Theory: an Approach to Computerized Processing ofUncertainty. Plenum Press, New York.Ferson, S. & Kuhn, R. (1992) Propagating uncertainty in ecological risk analysis using intervaland fuzzy arithmetic. In Computer Techniques in Environmental Studies IV (ed. P. Zannetti), pp.387–401. Elsevier, London.Fleminger, A. & Clutter, R.I. (1965) Avoidance of towed nets by zooplankton. Oceanography andLimnology, 10, 96–104.Gaston, K.J., Blackburn, T.M., & Gregory, R.D. (1999) Does variation in census area confounddensity comparisons? Journal of Applied Ecology, 36, 191–204.Gurevitch, J. & Hedges, L.V. (1999) Statistical issues in ecological meta-analysis. Ecology, 80,1142–1149.Harlow, L.L., Muliak, S.A., & Steiger, J.H. (eds.) (1997) What If There Were No Significance Tests?Lawrence Erlbaum. Mahwah, NJ.Hattis, D. & Burmaster, D.E. (1994) Assessment of variability and uncertainty distributions forpractical risk analyses. Risk Analysis, 14, 713–730.Hester, F.E. & Dendy, J.S. (1962) A multiple sampler for aquatic macroinvertebrates. Transactionsof the American Fisheries Society, 91, 420–421.Hilborn, R. & Mangel, M. (1997) The Ecological Detective. Princeton University Press, Princeton.Jasienski, M. & Bazzaz, F.A. (1999) The fallacy of ratios and the testability of models in biology.Oikos, 84, 321–326.Johnson, C.N. (1999) Relationships between body size and population density of animals: theproblem of the scaling of study area in relation to body size. Oikos, 85, 565–569.Kirchner, T.B. (1998) Time, space, variability, and uncertainty. In Risk Assessment: Logic and Measurement(ed. M.C. Newman and C.L. Strojan), pp. 303–323. Ann Arbor Press, Chelsea, MI.Krebs, C.J. (1989) Ecological Methodology. HarperCollins, New York.Lamberti, G.A. & Resh, V.H. (1985) Comparability of introduced tiles and natural substrates forsampling lotic bacteria, algae and macroinvertebrates. Freshwater Biology, 15, 21–30.Lima, S.L. (1998) Nonlethal effects in the ecology of predator:prey interactions. Bioscience, 48,25–34.Magurran, A.E. (1988) Ecological Diversity and its Measurement. Croom Helm, London.Manly, B.F.J. (1990) Stage-Structured Populations: Sampling, Analysis and Simulation. Chapman &Hall, London.Manly, B.F.J. (1997) A method for the estimation of parameters for natural stage-structuredpopulations. Researches on Population Ecology, 39, 101–111.Martinez, N.D., Hawkins, B.A., Dawah, H.A., & Feifarek, B.P. (1999) Effects of sampling efforton characterization of food-web structure. Ecology, 80, 1044–1055.Matthaei, C.D., Werthmuller, D., & Frutiger, A. (1998). An update on the quantification ofstream drift. Archives of Hydrobiology, 143, 1–19.May, R.M. (1975) Patterns of species abundance and diversity. In Ecology and Evolution ofCommunities (ed. M.L. Cody & J.L. Diamond), pp. 81–120. Belknap, Harvard.McArdle, B.H. (1990) When are rare species not there? Oikos, 57, 276–277.Merritt, R.W., Resh, V.H., & Cummins, K.W. (1996) Design of aquatic insect studies: collecting,sampling, and rearing procedures. In An Introduction to the Aquatic Insects of North America (ed.R.W. Merritt & K.W. Cummins), pp. 12–28. Kendall/Hunt, Dubuque, IA.Morin, P.J., Lawler, S.P., & Johnson, E.A. (1988) Competition between aquatic insects andvertebrates: interaction strength and higher-order interactions. Ecology, 69, 1401–1409.Murphy, K.R. & Myors, B. (1998) Statistical Power Analysis: a Simple and General Model forTraditional and Modern Hypothesis Tests. Lawrence Erlbaum, Mahwah, NJ.Muscha, M.J., Zimmer, K.D., Butler, M.G., & Hanson, M.A. (2001) A comparison of horizontallyand vertically deployed aquatic invertebrate activity traps. Wetlands, 21, 301–307.

218 CHAPTER 9Nelsen, R.B. (1986) Properties of a one-parameter family of bivariate distributions with specifiedmarginals. Communications in Statistics: Theory and Methods, A15, 3277–3285.Nelsen, R.B. (1987) Discrete bivariate distributions with given marginals and correlation. Communicationsin Statistics: Simulation and Computation, B16, 199–208.New, T.R. (1998) Invertebrate Surveys for Conservation. Oxford University Press, Oxford.Norris, R.H., McElravy, E.P., & Resh, V.H. (1992) The sampling problem. In Rivers Handbook (ed.P. Calow & G.E. Petts), pp. 282–306. Blackwell, Oxford.Nürnberger, B. (1996) Local dynamics and dispersal in a structured population of the whirligigbeetle Dineutus assimilis. Oecologia, 106, 325–336.Osenberg, C.W., Sarnelle, O., & Goldberg, D.E. (1999) Meta-analysis in ecology: concepts,statistics, and applications. Ecology, 80, 1103–1104.Pearman, P.B. (1995) Effects of pond size and consequent predator density on two species oftadpoles. Oecologia, 102, 1–8.Petersen, J.E., Cornwell, J.C., & Kemp, W.M. (1999) Implicit scaling in the design of experimentalaquatic ecosystems. Oikos, 85, 3–18.Polis, G.A. (1991) Complex trophic interactions in deserts: an empirical critique of food webtheory. American Naturalist, 138, 123–155.Prairie, Y.T. & Bird, D.F. (1989) Some misconceptions about the spurious correlation problemin the ecological literature. Oecologia, 81, 285–288.Puttman, R.J. (1995) Ethical considerations and animal welfare in ecological field studies. Biodiversityand Conservation, 4, 903–915.Resetarits, W.J. (2001) Colonization under threat of predation: avoidance of fish by an aquaticbeetle, Tropisternus lateralis (Coleoptera: Hydrophilidae). Oecologia, 129, 155–160.Resh, V.H., Feminella, J.W., & McElravy, E.P. (1990) Sampling Aquatic Insects. Videotape. Officeof Media Services. University of California, Berkeley, CA.Rohlf, F.J. & Sokal, R.R. (1995) Statistical Tables. W.H. Freeman, New York.Rosenberg, D.M. & Resh, V.H. (1982) The use of artificial substrates in the study of freshwaterbenthic macroinvertebrates. In Artificial Substrates (ed. J. Cairns Jr.), pp. 175–235. Ann ArborScience Publishers/Butterworth Group, Ann Arbor, MI.Rowe, W.D. (1994) Understanding uncertainty. Risk Analysis, 14, 743–750.Rutherford, J.E. (1995) Patterns of dispersion of aquatic insects colonizing artificial substratesin a southern Ontario stream. Canadian Journal of Zoology, 73, 458–468.Rutherford, J.E. & MacKay, R.J. (1985) The vertical distribution of hydropsychid larvae andpupae (Trichoptera : Hydropsychidae) in stream substrates. Canadian Journal of Zoology, 63,1306–1355.Rutledge, P.A. & Fleeger, J.W. (1988) Laboratory studies on core sampling with application tosubtidal meiobenthos collection. Limnology and Oceanography, 33, 274–280.Service, M.W. (1993) Mosquito Ecology: Field Sampling Methods. Elsevier, London.Shlyakhter, A.I. & Kammen, D.M. (1992) Sea-level rise or fall? Nature, 357, 25.Siemann, E., Tilman, D., & Haarstad, J. (1999) Abundance, diversity and body size: patternsfrom a grassland arthropod community. Journal of Animal Ecology, 68, 824–835.Sokal, R.R. & Rohlf, F.J. (1995) Biometry. W.H. Freeman, New York.Southwood, T.R.E. (1978) Ecological Methods. Chapman & Hall, London.Stewart, R.J. & Schaefer, C.H. (1983) The relationship between dipper counts and the absolutedensity of Culex tarsalis larvae and pupae in rice fields. Mosquito News, 43, 129–135.Sugihara, G. (1980). Minimal community structure: an explanation of species abundance patterns.The American Naturalist, 116, 770–787.Svensson, B.W. (1999) Environmental heterogeneity in space and time: patch use, recruitmentand dynamics of a rock pool population of a gyrinid beetle. Oikos, 84, 227–238.Taylor, J.R. (1982) An Introduction to Error Analysis. University Science Books, Sausalito, CA.

SAMPLING AQUATIC INSECTS 219Trexler, J.D., Apperson, C.S., & Schal, C. (1998) Laboratory and field evaluations of ovipositionresponses of Aedes albopictus and Aedes triseriatus (Diptera: Culicidae) to oak leaf infusion.Journal of Medical Entomology, 35, 967–976.Turner, A.M. & Trexler, J.C. (1997) Sampling aquatic invertebrates from marshes: evaluatingthe options. Journal of the North American Benthological Society, 16, 694–709.Vance, G.M., VanDyk, J.K., & Rowley, W.A. 1995. A device for sampling aquatic insectsassociated with carrion in water. Journal of Forensic Sciences, 40, 479–482.Washino, R.K. & Hokama, Y. (1968) Quantitative sampling of aquatic insects in a shallowwaterhabitat. Annals of the Entomological Society of America, 61, 785–786.Welsh, A.H., Peterson, A.T., & Altmann, S.A. (1988) The fallacy of averages. American Naturalist,132, 277–288.Wilcox, C. 2001. Habitat size and isolation affect colonization of seasonal wetlands by predatoryaquatic insects. Israel Journal of Zoology, 47, 459–475.Wood, S.N. (1997) Inverse problems and structured-population dynamics. In Structured-Population Models in Marine, Terrestrial and Freshwater Systems (ed. S. Tuljapurkar & H.Caswell), pp. 555–586. Chapman & Hall, New York.Zar, J.H. (1984) Biostatistical Analysis. Prentice-Hall, Englewood Cliffs, NJ.Index of methods and approachesTopicSurvey of samplingdevicesKinds of sampling errorsSampling efficiencyEthical issuesMethodologyActive operatorSweep netsDippersSurber/HessCore samplesGrabsColumn samplesExamplesActive vs. passive operator, absolute vs. relative estimates.Systematic bias, random errors, measurement interactions.Absolute and relative estimates, abiotic and biotic factorsaffecting sampling efficiency, calibration curves, number ofsamples needed.Avoid unnecessary killing of individuals and damage to habitats.Sampling can spread diseases.UsesOpen water and flexible vegetation.Organisms close to surface in shallow water, e.g. immaturemosquitoes.Small habitats e.g. water-filled tires, rock pools.Benthos in lotic habitats.Soft benthos.Soft benthos.Water column above soft benthos.Continued

220 CHAPTER 9MethodologyPassive operatorArtificial substrates, e.g.rocks, tiles, artificialvegetation, Hester–DendyMesocosmsDrift netsMinnow trapsEmergence trapsLight trapsFood trapsOviposition attractanttrapsVisual counts andphotographyUsesRocky benthos, vegetated water bodies.Oviposition, dispersion.Flowing water.Active organisms.Insects emerging from and leaving water.Night.Predators.Females.Small habitats, surface-dwelling organisms, benthic species inclear shallow water.

CHAPTER 10Methods for sampling termitesDAVID T. JONES, ROBERT H.J. VERKERK,AND PAUL EGGLETONIntroductionTermites (order Isoptera) are predominantly tropical in distribution. Theirspecies richness is highest in lowland equatorial rain forests, and generally declineswith increasing latitude (Collins 1983, Eggleton et al. 1994) and altitude(Gathorne-Hardy et al. 2001). Termite survival is limited by low temperaturesand high aridity, and very few species occur beyond 45° latitude (Collins 1989).The forests of West Africa have the highest termite species richness, closely followedby South America, whereas the forests of Southeast Asia and Madagascarare considerably less diverse (Eggleton 2000, Davies et al. 2003). These regionaldiversity anomalies are also associated with significant differences in cladeand functional diversity (Davies et al. 2003).Termites are at the ecological center of many tropical ecosystems (Wilson1992), and can achieve very high population densities. For example, in theforests of southern Cameroon, termites are one of the most numerous of allarthropod groups (Watt et al. 1997) with abundances of up to 10,000 per m 2 ,and live biomass densities up to 100 g per m 2 (Eggleton et al. 1996). Termiteshave a wide range of dietary, foraging, and nesting habits, with many speciesshowing a high degree of resource specialization (Wood 1978, Collins 1989,Sleaford et al. 1996). The vast majority of species feed on dead plant material,while relatively few species feed on living plant tissue. On a humification gradient,from undecomposed dead wood and leaf-litter to humus in the soil, mostdetritiverous species consume material that occupies a relatively narrow rangeof the gradient (Donovan et al. 2001a). As the dominant arthropod detritivores,termites are important in decomposition processes (Wood & Sands 1978,Matsumoto & Abe 1979, Collins 1983) and play a central role as mediators ofnutrient and carbon fluxes (Jones 1990, Lawton et al. 1996, Bignell et al. 1997,Tayasu et al. 1997, Sugimoto et al. 2000). Termite activity, such as tunneling,soil-feeding, and mound building, helps to maintain macropore structure, redistributesorganic matter, and improves soil stability and quality (Lee & Wood1971, Lobry de Bruyn & Conacher 1990, Black & Okwakol 1997, Holt & Lepage2000, Donovan et al. 2001b). However, termites’ influence on ecosystemprocesses at any site is likely to depend on the species composition and abundanceof the local termite assemblage.221

222 CHAPTER 10Approximately 2650 species of termites have been described to date(Kambhampati & Eggleton 2000), and less than 3 percent of these cause significanteconomic damage to buildings or related manmade structures (Pearce1997). A similar proportion are serious pests of crops (Wood 1996). The termitefauna of urban environments is usually highly depauperate and characterizedby wood-feeding species, unlike natural habitats that often support muchgreater species and functional diversity. For example, 136 species have beenrecorded in a single forest site in Cameroon, of which 73 percent are soil-feeders(Jones & Eggleton 2000).Termite sampling methodologies have been discussed by Lee and Wood(1971), Baroni-Urbani et al. (1978), Nutting and Jones (1990), and Eggletonand Bignell (1995). Those reviews provide detailed results from numerous samplingstudies, and are recommended as a rich source of referenced informationon the subject. Our intention is not to duplicate those valuable reviews but toprovide an overview of sampling methods and a framework for their applications,and to draw together recent developments in sampling technologies andstrategies. In this chapter, we:1 outline the difficulties encountered when sampling termites;2 review all major sampling methods;3 describe two sampling regimes that have been designed for use in tropicalforests: one estimates the population density of local termite assemblages, theother is a rapid protocol for assessing species composition;4 review sampling and monitoring methods for subterranean termite pests ofbuildings;5 present a case study describing the methods used to monitor populations ofan infestation of a pest species in England.The difficulties of sampling termitesBeing eusocial insects, termite colonies have a fixed location, and the sterilecastes (workers and soldiers) are usually present throughout the year. Therefore,termites can be sampled directly, unlike many solitary and more mobile insects.However, effective sampling of termites presents considerable theoreticaland practical problems. These problems stem from the very patchy spatial distributionof colonies and individuals within habitats, and the cryptic nature ofmost species. Sampling difficulties are at their most severe in tropical forests,where the structural complexity of the habitat combines with high levels of termitespecies richness, thus making many species difficult to find. Figure 10.1shows schematically the complex distribution of termites in a West African forest,based on studies in southern Cameroon (Eggleton et al. 1995, 1996, Dibog1998). The spatial distribution of termites in the forests of South America andSoutheast Asia is shown schematically in Collins (1989).Termites occupy a wide array of microhabitats, distributed vertically from

CopNeo SchAcaold nestMicProAstAci, AdaAde, AlyAmi, AneAst, AteAma, CoxDup, EbuJug, LabMad, PhoThoCepTerCreMcrCubApiEucFurTubOphPrcNodold nestNasLepEutPosVerPsc OdoSphSynBas, FasFor, OrtPer, PrbPsm, UngKalotermitidaeRhinotermitidaeMacrotermitinae Apicotermitinae Termitinae NasutitermitinaeFig. 10.1 Schematic diagram showing the spatial distribution of termite nests in semi-deciduous forest of Mbalmayo, southern Cameroon, West Africa.Fifty-two genera, compiled from Eggleton et al. (1995, 1996) and Dibog (1998). Kalotermitidae: Neotermes. Rhinotermitidae: Coptotermes,Schedorhinotermes. Termitidae, Macrotermitinae: Acanthotermes (Aca), Microtermes (Mic), Odontotermes (Odo), Protermes (Pro), Pseudacanthotermes (Psc),Sphaerotermes (Sph), Synacanthotermes (Syn). Apicotermitinae: Acidnotermes (Aci), Adaiphrotermes (Ada), Aderitotermes (Ade), Alyscotermes (Aly),Amalotermes (Ama), Amicotermes (Ami), Anenteotermes (Ane), Astalotermes (Ast), Ateuchotermes (Ate), Coxotermes (Cox), Duplidentitermes (Dup),Eburnitermes (Ebu), Jugositermes (Jug), Labiotermes (Lab), Machadotermes (Mad), Phoxotermes (Pho). Termitinae: Apilitermes (Api), Basidentitermes (Bas),Cephalotermes (Cep), Crenetermes (Cre), Cubitermes (Cub), Euchilotermes (Euc), Fastigitermes (Fas), Foraminitermes (For), Furculitermes (Fur), Microcerotermes(Mcr), Noditermes (Nod), Ophiotermes (Oph), Orthotermes (Ort), Pericapritermes (Per), Proboscitermes (Prb), Procubitermes (Prc), Pseudomicrotermes (Psm),Termes (Ter), Thoracotermes (Tho), Tuberculitermes (Tub), Unguitermes (Ung). Nasutitermitinae: Eutermellus (Eut), Leptomyxotermes (Lep), Nasutitermes(Nas), Postsubulitermes (Pos), Verrucositermes (Ver).

224 CHAPTER 10deep in the soil to the crowns of emergent trees. These numerous microhabitatsrepresent real biological entities, but their exact limits can be difficult to definefor the purposes of designing rigorous surveys. In practice, when sampling inthe field, researchers have usually collected termites from one or more of fourbroad categories: mounds (epigeal nests that protrude above the surface of thesoil), soil, dead wood at ground level, and arboreal habitats. Each categoryhas distinct sampling problems and requires specific techniques (see Samplingmethods). Moreover, the effort needed to gather statistically meaningful datausing many of these methods can be very labor-intensive.The variation in termite nest design (described by Noirot 1970), from simplediffuse galleries excavated in soil or wood, to the most structurally complexedifices built in the animal kingdom, can complicate sampling. In many speciesthe nest is a single unit (called a calie) with a clearly delineated boundary, and iseasily distinguishable from the surrounding substrate. Other species are polycalic:the colony is distributed among numerous calies but remains interconnectedvia subterranean tunnels or arboreal runways. Polycalic species occur inthe lower termites, for example Hodotermes (Coaton & Sheasby 1975), Schedorhinotermes(Husseneder et al. 1998), and in wood-feeding species of Macrotermitinae,Termitinae, and Nasutitermitinae (e.g. Sands 1961, Holt & Easey 1985,Roisin & Pasteels 1986, Atkinson & Adams 1997). For practical reasons it may beimpossible to locate all parts of a polycalic nest. The physical resilience of thenest fabric can also hinder sampling. Mounds can be very hard due to a highcontent of cemented soil in the matrix. In contrast, many wood-feeding speciesbuild nests made of carton material (masticated wood) that are more fragile.Nests are not always restricted to a single microhabitat. The hypogeal (subterranean)nest of some species, for example Macrotermes malaccensis or Prohamitermesmirabilis, may sometimes protrude above the soil surface. Most incipientcolonies begin life within dead wood or in the soil but some species may eventuallydevelop a large and obvious epigeal nest. In a few cases, a single speciescan produce several nest types. For example, Microcerotermes crassus can buildnests entirely within wood, exterior arboreal nests, hypogeal nests, or epigealmounds, all within the same forest (Takematsu et al. 2003). Many species ofNasutitermitinae that produce exterior arboreal carton nests may often have alarge proportion of the nest within the tree trunk or branch to which it is attached.All wood-nesting termites, however, except a few genera of Kalotermitidaeand Termopsidae, maintain some association with the soil.It can be difficult to verify the territorial limits of a colony because somespecies extend their foraging range far beyond the nest. Some Macrotermesmaintain a complex network of subterranean tunnels (Darlington 1982), andothers forage in exposed columns on the forest floor (Sugio 1995). Hospitalitermesis an extreme case in which the colony sends out soldiers and workers inprocessional columns that extend for up to 65 m across the forest floor, beforeascending living trees to graze microepiphytes from the trunk and branches(Jones & Gathorne-Hardy 1995). Longipeditermes longipes is frequently observed

METHODS FOR SAMPLING TERMITES 225feeding on leaf-litter on the forest floor (Matsumoto & Abe 1979, Collins 1984)but it was recently recorded in insecticidal fogging samples from tree canopies(Hoare & Jones 1998). The extent to which species that are assumed to berestricted to foraging at ground level may venture into arboreal habitats isunknown.In some habitats, the size, number, and apparent dominance of mounds cangive the impression that they are the most abundant and ecologically importantpart of the local termite assemblage. As a result, a majority of ecological studiesin earlier decades were largely confined to mound-building species (Wood &Sands 1978). However, in the Mbalmayo Forest Reserve, southern Cameroon,only 12 percent of termite species in the local assemblage build epigeal or arborealnests (Eggleton et al. 1996), implying that surveys limited to conspicuousmounds and nests will lead to serious underestimates of species richness.Nonetheless, mounds can contribute to the termite species richness of an areabecause they often harbor secondary occupants (called inquilines) as well as themound-building species (Dejean & Ruelle 1995, Eggleton & Bignell 1997).Therefore, mounds should be checked carefully during diversity studies.Assemblage-level population studies based on mound sampling can also be inherentlybiased because in many habitats the mounds may not hold a significantproportion of termite abundance or biomass (Sands 1972). In Mbalmayo, lessthan 10 percent of the overall abundance was in mounds (Eggleton et al. 1995).As so many termites nest and forage in the soil, any assemblage-level studymust adequately sample the soil.Approaches to samplingMuch of the published data on termite species richness and population densityin local assemblages are not strictly comparable because previous studies haveused a variety of sampling methods and experimental designs, and different levelsof collecting effort (Eggleton & Bignell 1995). As a consequence, there werelimitations in the generalities and differences that could be inferred amongstudy sites. However, with the systematic use of standardized sampling methodsthat more accurately characterize the structure of local termite assemblages (seeTwo standardized methods for use in tropical forests, below), the detailed structure ofspatial patterns within and between regions is now being elucidated.Any sampling regime will be a compromise between the specific questionsthe research is trying to address, and the available resources (time, money,labor, equipment, and taxonomic expertise). It may be relatively simple tostudy the biology of a single species in the field, but as the research is widened toinclude more parts of the local termite assemblage the sampling regime will becomeincreasingly complicated. Sampling regimes can be designed to answerone of three distinct types of research question.1 Population density. Studies aimed at measuring termite density may seek toestimate either the population within a single colony or the total abundance of

226 CHAPTER 10all species encountered in a unit area of habitat. Included in this category arestudies aimed at estimating colony density by counting the number of moundsor nests per unit area.2 Species composition. Studies aimed at investigating the species compositionof an assemblage in a local area but without estimating population density.3 Termite activity. Studies aimed at investigating activity such as foragingrange, food preference and rates of consumption, or alate swarming.These three categories require different approaches in sampling methods. However,some of the methods can be modified to answer questions in more thanone category. Figure 10.2 offers a sequence of questions via which an appropriatesampling method can be chosen. This “decision tree” is meant only as a guidesince local conditions and logistical considerations may impose other practicaland statistical limitations when designing a sampling regime. All major samplingmethods are discussed in the following section (summarized in the Index ofmethods, page 250).Sampling methodsSampling moundsMany studies have focused on mounds (e.g. see Lee & Wood 1971, Pomeroy1977) because they are relatively easy to locate, and because they can be a dominantfeature of the landscape, particularly in savannas. Baroni-Urbani et al.(1978) reviewed the use of aerial photographs, line transects, and variousquadrat methods to measure the density of mounds. Very large intraspecificvariations in mound density across seemingly homogenous savanna systemsare often observed, as for example with Cubitermes sankurensis in central Africa(Mathot 1967). Spatial dispersion can be examined by mapping the location ofmounds and then employing a nearest-neighbor technique (e.g. see Wood &Lee 1971, Schuurman & Dangerfield 1997, Meyer et al. 1999).In the time it takes to dig into a large and strongly built mound, the disturbancecan cause much of the population to evacuate the hive. To prevent this,the population within a mound can be killed in situ by fumigation with methylbromide. The entire nest contents can then be excavated and the termites removedfrom the nest debris by flotation in water. The whole sampling process(described by Darlington 1984) is labor-intensive, and depending on the size ofthe mound it can take five laborers up to three weeks to complete. However,Darlington (1984) showed that sampling large mounds of Macrotermes withoutfumigation caused the population and biomass to be underestimated by up toan order of magnitude. Macrotermes mound parameters (both internal and externaldimensions) and nest population are approximately linearly related.Therefore, survey data on mound size and density can be used to estimate abundanceper unit area (Darlington & Dransfield 1987, Darlington 1990). Young

NoDo you want tostudy the wholelocal assemblage?YesDo you want toestimate populationdensity of the wholelocal assemblage?NoDo you want tostudy speciescomposition?YesDo you need arapid samplingmethod?YesStandardisedtransectmethodDo you want tostudy swarmingactivity?YesLighttrappingNoYesStartDo you want tostudy a singlespecies?NoDo you want tostudy all the speciesin one “compartment”?NoDo you want toestimate density?YesYesNoNoDo you want tostudy foragingbehaviour?BaitingmethodsYesYesHand-sortingof dead woodand litterfrom quadratsDo you want toestimate rates ofconsumption?NoYesDo you wantto estimatepopulationdensity?YesDo you wantto estimatepopulationdensity?NoDo you want toestimate colonydensity?YesAre the nests ormounds visuallyconspicuous?NoYesYesPlotsurveysMarkrecapturetechniquesObservation ofspecies foragingin the open airPlot surveys andhand-sorting ofnest sub-samplesPlot surveys andhand-sorting ofmound sub-samplesYesYesYesEstimatepopulationin dead woodand litterOrEstimatepopulationin arborealnestsOrEstimatepopulationin moundsOrCombine estimatesfrom the followingfour “compartments”Hand-sortingof soil coresHand-sortingof soil dugfrom pits,monoliths,or trenchesDetection ofacoustic emissionsfrom termites nestingin woodDirect counting ofpopulation in nestby hand-sorting orflotation methodsEstimatesubterraneanpopulationYesBaitingmethodsFig. 10.2 Decision tree giving a sequence of questions by which a method for sampling termites may be chosen.

228 CHAPTER 10colonies will require separate number-to-biomass conversion factors becausethey can have different caste ratios, a higher proportion of larvae, and will oftenproduce smaller individuals compared with mature colonies (Darlington1991).A quicker method of estimating population size is to take a sample of knownvolume from the mound, extract and count the termites, and multiply up to thetotal volume of the mound. Most mounds have irregular shapes, and thereforethe total volume can be difficult to estimate. To overcome this problem, moundshave been measured as simple shapes, such as hemispheres (Sands 1961) orcones (Holt & Easey 1993). The sample of nest material must be taken quickly toprevent the termites retreating further into the mound. After breaking upthe nest material, the termites are extracted either by flotation or by handsorting.Inaccuracies can arise because mound populations vary within each24-hour cycle as foragers leave and return to the nest, and the density of individualsin any given area of the mound changes due to migration within thenest (Sands 1965). Ohiagu (1979) showed that more than half of the populationsof four Trinervitermes species were in the soil rather than in the mounds atany one time.Sampling termites in soilMany species are restricted to the soil, with both nest and foraging galleriesconcealed underground without any indication above ground. Subterraneantermites can only be sampled by removing units of soil and extracting the individualsby some method. Therefore, the important questions are what depthand volume of sampling unit should be used, and how many units should becollected? Baroni-Urbani et al. (1978) discuss these questions in detail.In forests, subterranean termites characteristically occur in the organic layerof the soil profile (Collins 1989). A study in Malaysian rain forest reported thattermites were mainly found in the top 15 cm and were rare below 25 cm (Abe &Matsumoto 1979). In Nigerian riparian forest, most termites were in the top25 cm of the soil profile and showed no significant difference in abundanceacross seasons (Wood et al. 1982). However, in drier or more seasonal habitatsthe issue of sampling depth is complicated by vertical migration. Vertical distributionvaries with species, soil type, and season, and no general correction factorscan be applied (Wood & Sands 1978). In cultivated systems derived fromwoodland and savanna in west Africa, Microtermes is an abundant pest. In thewet season, Microtermes were usually concentrated in the upper 25 cm, whereasthe proportion of the population below 50 cm greatly increased in the dry seasonas they moved deeper in the soil (Wood & Johnson 1978, Black & Wood1989). This effect is probably less pronounced in tropical forests where thecanopy limits fluctuations in soil temperature, but movement of termites mayalso be linked to rainfall events. In a seasonal humid forest in Cameroon, Diboget al. (1998) found that both species richness and abundance in 10 cm deep soil

METHODS FOR SAMPLING TERMITES 229samples were generally higher in dry periods compared with wet periods. However,no significant changes in overall species composition were observed.Studies have used numerous sampling volumes, ranging from excavatingvery large pits or long, narrow trenches, to small soil cores. For example, Abeand Matsumoto (1979) dug one pit of area 1 m ¥ 2m to a depth of 25 cm, carefullyremoving and sorting the soil in smaller sub-units, whereas Collins(1979a) and Wood et al. (1982) took hundreds of soil cores of 10 cm diameter.Eggleton and Bignell (1995) outline the trade-offs involved in using differentsampling sizes. Due to the highly heterogenous spatial distribution of termites,most small to medium sized soil samples will contain relatively low numbers ofindividuals or none at all, while a very few samples may have extremely highnumbers if a nest or foraging party is encountered. Density estimates can thereforehave high variance, making it difficult to demonstrate statistically significantdifferences among sites.Hand-sorting is often recommended for extracting termites from soilsamples. Wood et al. (1977) reported that 78–92 percent of all termites in soilsamples were collected by hand-sorting, and applied a 12 percent loss factor insubsequent field studies (Wood et al. 1982). The technique is simple, requiringonly trays on which to examine the soil and forceps for removing the termites.Termites can also be extracted by flotation methods (Strickland 1944, Salt 1952,Madge 1969) but Wood et al. (1977) considered these too time-consuming to bepractical. Automatic extraction devices such as Berlese–Tullgren funnels areless suitable because the termites often die in situ as the soil dries out. In comparison,the Kempson extractor is better at removing soft-bodied invertebratesfrom soil samples because it is equipped with a thermostat that allows subtletemperature and moisture gradients to be maintained through the sample (Adis1987). After testing the Kempson extractor, Silva and Martius (2000) suggestedthat it was as effective at removing termites as hand-sorting. However, their resultswere not statistically conclusive because the sample sizes were very small.The use of the Kempson extractor is impractical at some study sites because itneeds about 15 days of continuous electricity supply.Sampling termites in woodSpecies that feed and nest within dead logs and branches can have huge populations.Individuals can be extremely difficult to dislodge from narrow galleries.For quantitative studies the only effective extraction method is to split the woodlengthways and remove the termites manually. Failure to sample larger items ofdead wood may severely underestimate termite abundance (Collins 1983,Eggleton & Bignell 1995, Eggleton et al. 1996). The population in larger items ofdead wood can be estimated by sampling sub-units by volume or weight and assuminga uniform population density throughout. However, few researchershave attempted to sample populations quantitatively in dead wood (see thestandardized population sampling regime, below, for an example). No non-

230 CHAPTER 10destructive techniques exist for estimating the size of colonies inhabiting livingtrees. Greaves (1967) describes methods for felling living trees and estimatingthe inhabiting populations of Coptotermes.Collins (1983) and Jones (1996) used a semi-quantitative counting methodto estimate the population in dead wood at forest sites in Borneo. The methodinvolved splitting open dead wood with a machete and visually estimating thenumber of individuals by counting in units of 10s, 100s, or 1000s. Although thismethod has not been calibrated against direct counts, the Collins and Jones estimatesappear reasonable when compared with densities recorded at a similarforest in Borneo using more rigorous methods (Eggleton et al. 1999).Sampling termites in arboreal habitatsArboreal nests on the trunks of trees and attached to understory vegetation canbe easily sampled up to a height of about 2 or 3 m above ground level. However,because it is difficult to gain access to nests above this height, no satisfactoryquantitative methods have been devised for sampling termites in forestcanopies. Present knowledge of arboreal termite diversity is limited, and basedmainly on casual samples of dead wood and nest material removed from treecrowns while using rope climbing techniques, canopy walkways, or after treeshave been felled. One study (Ellwood et al. 2002) has revealed that a high proportionof large, epiphytic birds’ nest ferns (Asplenium nidus complex) in thecanopy of a forest in Borneo contain nests of Hospitalitermes. Insecticidal foggingis not suitable for dislodging termites, because those affected by the insecticideusually remain inside their nests and foraging tunnels.“Dry-wood” termites (Kalotermitidae) such as Neotermes, Cryptotermes, andGlyptotermes usually nest wholly within dead branches, and their colonies numberfrom a few hundred individuals (Harris 1950) up to about ten thousand(Maki & Abe 1986). The frequency with which kalotermitid alates are caught inlight traps (Rebello & Martius 1994, Medeiros et al. 1999) suggests that theymay be a more significant component of forest assemblages than previouslythought. Several species of Coptotermes can pipe the inside of living trees andleave no external evidence of their presence. Many arboreal termites that buildexternal carton nests on trees also produce covered runways down the trunk.This allows researchers to identify the termites by scraping away the sheetingand collecting the foragers. Researchers can also find arboreal species that nestin dead wood if dead branches attached to trees at ground level are removed andexamined, or if sufficient fallen dry dead branches are collected from under treecrowns.Sampling termites using baitsCellulose baits simulate natural items of food such as fallen dead wood. The twomost commonly used baits are wooden stakes and rolls of unscented toilet

METHODS FOR SAMPLING TERMITES 231paper. Other materials have been used, such as translocated dungpats (de Souza1993) and sawdust (Abensperg-Traun 1993). Baits are often set in grid formation(so-called “graveyard” trials), and in some cases the area is first cleared ofnaturally occurring fallen wood (Haverty et al. 1975). Wooden baits are usuallycut from timber known to be susceptible to the local wood-feeding termites.However, field experiments have been ruined by neglecting to check whetherpurchased timber has been treated with insecticide. Stakes are driven into theground and the top is left protruding above the soil to facilitate monitoring.Alternatively, baits may be laid out on the soil surface or buried (Sands 1972,Lenz et al. 1992, Dawes-Gromadzki 2003). Two stakes can be installed in contactwith one another, as the interface between the stakes tends to encouragerapid colonization by subterranean termites. Researchers should not remove ordisturb baits too frequently, as this will discourage termites from foraging on thebait. Usher and Ocloo (1974) tested the effect of stake size, shape, and positionon the amount of damage caused by Macrotermitinae. They found that weightloss of wood increased as surface area increased, and that significantly moredamage was recorded when stakes were completely rather than partiallyburied. For details of how baiting can be used to monitor and control pestsspecies, see Methods for sampling subterranean termite pests of buildings, below.Baits attract foraging termites, and therefore give estimates of relative intensityof foraging activity rather than relative population density. Baiting has beenuseful in studying inter- and intraspecific foraging activity (Sands 1972, Buxton1981, Ferrar 1982, Pearce 1990, Pearce et al. 1990, Dawes-Gromadzki 2003),size of foraging territory (Haverty et al. 1975), and rates of food consumption(Haverty & Nutting 1974, Abe 1980). Baiting has also been used to estimatelocal species richness (de Souza 1993, Dangerfield & Mosugelo 1997, Tayloret al. 1998). However, this can be problematic because not all species are attractedto baits. Arboreal species that do not forage on the ground, and subterraneanspecies that do not forage near the soil surface, may be excluded. Also,food preference trials have shown that not all termite species are attracted to thesame bait materials (Haverty et al. 1976, Abensperg-Traun 1993, Dawes-Gromadzki 2003), implying that using a single bait type will under-sample thelocal species richness. While the degree of acceptance of cellulose baits dependson a variety of factors, it is notable that de Souza (1993) attracted 41 species (includingsoil-feeding species) using rolls of toilet paper when studying termitecommunity structure in Brazilian cerrado.Sampling using mark–recapture protocolsPopulation size can be estimated by mark–recapture protocols using radioisotopes(Spragg & Paton 1977, Easey & Holt 1989) or insoluble colored stains anddyes (Su et al. 1988, 1991, Evans 1997). However, when tested on two moundbuildingspecies the mark–recapture estimates varied widely within and amongcolonies, and could be 100 times larger than direct population counts (Evans

232 CHAPTER 10et al. 1998). These errors occurred because several of the assumptions inherentin the protocol were violated: the fat-stain markers faded quickly and weretransferred to unmarked individuals; marked individuals did not mix uniformlywith unmarked individuals; foragers displayed feeding site fidelity; and thelikelihood of recapture differed between castes and instars (Evans et al. 1998).Similar problems were encountered when the method was applied to subterraneannesting termites (Su et al. 1993, Forschler & Townsend 1996, Thorneet al. 1996), suggesting that mark–recapture protocols are unable to provide accuratepopulation estimates.Markers may be useful, however, in studies attempting to delineate colonyboundaries or foraging distances. Fluorescent dyes can either be incorporatedinto baits or applied to workers as a dust. Particles of the dust have beendetected after 48 h in the guts of workers of the highly destructive Australiangiant termite, Mastotermes darwiniensis, 95 m from the initial site of application(Miller 1993).Sampling alates using trapsAlates (or imagoes: the winged reproductive forms) must leave the nest to mate,and out-crossing can only be achieved if colonies of the same species synchronizethe release of alates. Swarming is often associated with annual weatherpatterns and ambient climatic conditions (Nutting 1969). However, the precisephysical and physiological factors that trigger alate release are still uncertain(Medeiros et al. 1999). Passive trapping devices such as flight interception andMalaise traps are not suitable for sampling termites because they collect veryfew alates (Rebello & Martius 1994). Light traps can be used to sample alates butthe results must be interpreted with caution because the technique has severalproblems. The position of the trap strongly influences the number of species andthe abundance of alates caught, because of the poor dispersal range of termites(Martius et al. 1996). Not all species show a clear preference for nocturnalswarming (Mill 1983), and alates released during the day may not be caught intraps run overnight. Some species produce relatively few alates, and others maynot produce alates every year.Depending on the degree of seasonality at the site, light traps may catch alatesthroughout the year (Martius et al. 1996) or only during a limited number ofmonths (Medeiros et al. 1999). Because there is little interspecific synchronicityof swarming, traps operating over short periods will fail to capture many localspecies. Medeiros et al. (1999) found that continuous trapping over one year inAtlantic rain forest in northeastern Brazil captured only 55 percent of thespecies previously recorded at the same site when collecting by hand. Incontrast, light traps may be more useful in urban habitats for species-specificstudies. For example, alates of the Formosan termite Coptotermes formosanushave been monitored in light traps in New Orleans (USA) over a seven-yearperiod (1989–95). Mean data showed a consistent increase over this time,

METHODS FOR SAMPLING TERMITES 233suggesting that the species can adapt to that specific urban environment(Henderson 1996).Recording the movement of termitesSeveral genera (including Hospitalitermes, Lacessititermes, Longipeditermes, Constrictotermes,and Macrotermes) form processional columns of soldiers and workersthat march in the open to feeding sites. In such cases, close-up photographsof the column taken at regular intervals can be used to estimate the number oftermites involved in the foraging activity (Collins 1979b, Miura & Matsumoto1998). However, this photographic technique can give large discrepanciesbetween the number of termites leaving the nest and the number returning(Collins 1979b).Hinze and Leuthold (1999) used two new techniques for detecting andrecording the movement of workers inside a laboratory colony of Macrotermesbellicosus. A metal detector monitored the movement of workers marked withsmall pieces of metal wire, while a photo detector counted both marked and unmarkedtermites entering and leaving the nest and the queen cell.Detection of acoustic emissionsRecently, handheld acoustic emission devices have been developed to detectthe feeding of hidden termite infestations in wood. This non-destructive techniquecan differentiate between the acoustic emissions of termites and otherwood-boring insects, and has been used successfully to detect pest species ofRhinotermitidae and Kalotermitidae in buildings (Weissling & Thoms 1999,Thoms 2000) and urban trees (Mankin et al. 2002). However, the accuracy withwhich acoustic emissions can be used to predict population density still has to bedemonstrated. Furthermore, the efficacy of this technology at detecting a rangeof termite species in natural environments has not been tested.Two standardized methods for use in tropical forestsA sampling regime for estimating termite assemblagepopulation densityFew researchers have tried to document the population density of an entirelocal termite assemblage in a diverse tropical habitat because of the considerableeffort involved. The following plot-based sampling regime is designed forestimating the population density of the local termite assemblage, excludingarboreal termites at more than 2 m above ground level. Eggleton et al. (1999)used this regime in Borneo, adapting a similar regime first used in Cameroon(Eggleton et al. 1996). The basic sampling area is a 0.25 ha plot (50 m ¥ 50 m),with an internal grid (10 m separation) marked with string to facilitate quadrat

234 CHAPTER 10placement and mapping. Within each plot, three sampling methods areemployed:1 Twenty quadrats (each 2 m ¥ 2 m) are placed using random coordinates.Quadrats falling on standing trees or other large obstacles are reassigned to newrandom coordinates. All dead wood and litter is removed from each quadratand hand-sorted on site just outside the plot by a team of trained assistants.Litter is searched and woody material is split open, and all termites removed.Larger items of dead wood are sub-sampled by volume.2 After removing the wood and litter, a soil pit of 30 cm ¥ 30 cm ¥ 25 cm depthis dug in the center of each quadrat and hand-sorted on site.3 A systematic survey of visible mounds and arboreal nests is carried out overthe entire area of the plot (searching up to a height of 2 m), making use of the internalgrid. Nests are mapped and destructively sampled. Nest populations areestimated by sub-sampling by weight (see Eggleton et al. 1996).The transect protocolThis transect-based protocol rapidly assesses the species composition of thelocal termite assemblage. The protocol, described by Jones and Eggleton (2000),was adapted from a similar method developed by Eggleton et al. (1996). Theprotocol has been used in many tropical forests around the world (Gathorne-Hardy et al. 2002, Davies et al. 2003).The transect is 100 m long and 2 m wide, and divided into 20 contiguous sections(each 5 m ¥ 2 m) and numbered sequentially. Two trained people sampleeach section for 30 minutes (a total of one hour of collecting per section). Tostandardize sampling effort, the collectors work steadily and continuously duringeach 30-minute period. In each section the following microhabitats aresearched for termites: 12 samples of surface soil (each 12 cm ¥ 12 cm, to 10 cmdepth); accumulations of litter and humus at the base of trees and between buttressroots; the inside of dead tree stumps, logs, branches, and twigs; the soilwithin and beneath very rotten logs; all mounds and subterranean nests encountered(checking for inquiline species); arboreal nests, carton runways, andsheeting on vegetation up to a height of 2 m above ground level. The protocol allowsthe collectors to use their experience and judgment to search for and sampleas many species in each section as time permits.Jones and Eggleton (2000) tested this protocol in three forest sites where thelocal termite fauna was already comprehensively documented. Two transectswere run at Danum Valley (Sabah, Borneo), one at Pasoh Forest Reserve(Peninsular Malaysia), and one at Mbalmayo Forest Reserve (Cameroon). Atthe three sites the transect samples contained 31 to 36 percent of the knownlocal termite species pool (Table 10.1), giving a reasonably high degree of samplingconsistency among sites. The taxonomic group composition (the proportionof species in each family, or subfamily in the case of the Termitidae) of thetransect samples did not differ significantly from that of the known local fauna

METHODS FOR SAMPLING TERMITES 235Table 10.1 The number of termite species collected from transects in three forest sites (Jones& Eggleton 2000). The total number of known species recorded from each site is based on allavailable records, from labor-intensive sampling programs to casual collecting. These totalsrepresent the best estimates of the species richness of each assemblage. Reproduced withpermission from Blackwell Publishing Ltd.Site Species sampled Total known Proportion ofin transect species total fauna intransectDanum Valley, Sabah, Borneo 29 93 31.2%(transect 1)Danum Valley, Sabah, Borneo 33 93 35.5%(transect 2)Pasoh, Malaysia 29 80 36.3%Mbalmayo, Cameroon 47 136 34.6%at each site. Similarly, the functional group composition (the proportion ofspecies in each feeding group) of the transect samples did not differ significantlyfrom that of the known local fauna. In addition, the two transects run atDanum Valley gave very similar patterns, suggesting that the protocol producesconsistent within-site results. One supervised training transect was shown to besufficient experience to ensure that collectors were sampling to the level of efficiencythat the protocol required.Comparison of the two standardized methodsAlthough the two methods were designed to address different questions, it isuseful to compare their relative merits. In tropical forests, the population samplingregime underestimates local species richness. This is because its strictly definedand prescriptive method only samples dead wood, termite nests, and alimited number of soil pits. In comparison, the transect protocol utilizes the expertiseof the collector to search a wider array of suitable microhabitats withineach section, thus increasing the likelihood of finding additional species. As aconsequence, the transect accumulates species much more rapidly than populationsampling (Fig. 10.3). Both methods avoid microhabitats above 2 m, butthe transect protocol often collects arboreal nesting species that forage atground level.The population sampling regime is labor-intensive, and estimates of samplingefficiency (Table 10.2) suggest that it takes about four to five times more effortto obtain and identify roughly the same number of species as one transect. Itshould be noted that population sampling generates more specimens than thetransect method and thus requires far greater taxonomic processing time for the

236 CHAPTER 103025Cumulative species richness201510Population samplingTransect500 5 10 15 20 25 30 35 40Collecting effort (days)Fig. 10.3 Species accumulation curves showing the cumulative richness produced bysampling one plot using the population sampling method, and one transect in primary forestat Danum Valley (Sabah, Borneo). Cumulative richness is plotted against the collecting effortmeasured in person days (transect = 4 days; population sampling = 40 days). The cumulativetotals were based on the smallest sampling unit for which species-level data were available.For the transect each sampling unit was one section, while for the population sampling thisrepresents 20 soil pits, 20 dead wood quadrats, and 9 mounds (see text for description ofmethods). The curves are the mean of 500 random sequences of these units. After Jones &Eggleton 2000; reproduced with permission from Blackwell Publishing Ltd.same number of species. Therefore, the transect protocol provides a much morerapid and cost-effective method for studying termite assemblage structure thanpopulation sampling regimes. The population sampling regime does, however,produce reliable estimates of termite population density and biomass (Eggletonet al. 1996) that can be used to quantify the impact of termites on ecosystemprocesses such as carbon fluxes (Bignell et al. 1997).Methods for sampling subterranean termite pests of buildingsIn regions of the world where termites cause major economic losses to buildingsand associated structures, knowledge of the species present is an essential

METHODS FOR SAMPLING TERMITES 237Table 10.2 A comparison of the approximate effort required to conduct transects and plotbasedpopulation sampling regimes, the cumulative number of termite species collected, andthe sampling efficiency of both methods, in forest at Danum Valley (Sabah, Borneo) andMbalmayo (Cameroon) (Jones & Eggleton 2000). Sampling efficiency is defined as thenumber of species collected per unit effort, where effort is measured as the number of persondaysrequired to collect and process the samples. Taxonomic processing is the time taken forone expert to sort and identify specimens, and in the case of the population sampling, tocount specimens. See text for description of the population sampling methods. Reproducedwith permission from Blackwell Publishing Ltd.Site Sampling Collecting Taxonomic Total effort Cumulative Samplingmethod time (days) processing (days) number of efficiency(days) species (number ofspeciescollectedper day)Danum 1 transect 4 8 12 29 2.42Valley 2 transects 8 16 24 40 1.671 plot 40 20 60 29 0.482 plots 80 40 120 38 0.323 plots 120 60 180 47 0.26Mbalmayo 1 transect 4 12 16 47 2.941 plot 20 15 35 28 0.802 plots 40 30 70 48 0.69prerequisite to any management or colony elimination program. The vastmajority of damage to buildings caused by termites worldwide is attributedto wood-feeding Rhinotermitidae from only two genera, Coptotermes andReticulitermes. Moreover, all Rhinotermitidae have a subterranean habit, meaningthat individual colonies require ground contact or a more or less continuousmoisture supply.Sampling of subterranean termites in localities with perceived or knownthreats from subterranean pest species can be undertaken for several reasons,including:1 to assess the presence or absence of termites and, if present, to make collectionsto allow taxonomic identifications;2 to undertake qualitative or quantitative assessments of termite assemblages,including ecological studies;3 to assess the extent and severity of a known infestation;4 to monitor the fate of a population following the implementation of amanagement or colony elimination program;5 to estimate the actual or relative population size of a given species within aprescribed area.

238 CHAPTER 10Basic methodologiesA wide variety of methods have been used to sample termites in the built environment.Baiting, mark–recapture, and light-trapping are reviewed above,while the methods outlined below are frequently used to sample subterraneanpests of buildings.Sampling of potential food sources and colony nest sitesInvestigation for wood-feeding termites in potential food sources (e.g. ingroundtimber and other cellulose sources) and colony nesting sites is often thestarting point of most studies or management programs (Verkerk 1990, Verkerk& Bravery 2001). A standard range of equipment is required for such surveys,including: a bright torch; mirrors (e.g. dental type) for viewing into confinedspaces; a ladder to gain access to roof voids and arboreal nesting sites; a largescrewdriver (the handle can be used for “sounding” timbers and the tip for probing);a sharp knife for cutting plasterboard, carpets, etc.; levers for lifting carpets,architraves, etc.; hammer and nails for butting up trap doors in timberfloors; vials and 70 percent ethyl alcohol, labels, a fine paintbrush, and forceps,for collecting and preserving specimens. More sophisticated devices such asacoustic monitors (Potter et al 2001, Mankin et al 2002) and endoscopes (Fuchset al 2004) have been used effectively to facilitate detection of termites or nestingsites in concealed areas or within trees.Timber stakes or dowelsOne of the most common methods for assessment of termite activity in andaround buildings is the insertion of timber stakes or dowels into the groundor into potential nesting sites. Timbers should be of a species and conditionknown to be susceptible to “pest” termite species in the given locality. Stakesare typically 500 mm in length, and 50 ¥ 25 mm in section, and are usuallycut to a point at their base to ease installation in hard ground (Verkerk & Bravery2001). Dowels may be punched directly into nests and are particularly useful todetermine if colony elimination has been successful (Peters & Fitzgerald 2003).Corrugated cardboard trapsVarious methods of sampling termites have employed corrugated cardboard(untreated with fungicide) as a feeding medium within monitoring systems insubterranean management programs. Three such methods are described.1 Reservoirs of corrugated cardboard (within timber boxes with slits in thebase, or within plastic or aluminum foil containers) set in the soil, can be used astermite traps. Dampened cardboard is placed within these reservoirs in layers so

METHODS FOR SAMPLING TERMITES 239that termites may be detected or collected during periodic inspections. Termitesmay be encouraged into the reservoirs by linking strips of corrugated paper toeach reservoir (Kirton et al. 1998).2 Lengths of ABS pipe (perforated or unperforated) can be filled with rolled,dampened corrugated cardboard and can be installed vertically in the soil, witha section remaining above ground for access (e.g. Myles 1996, Haverty et al.1999). The cardboard may be removed for the purpose of collecting termiteswithout disturbing the pipe/soil interface. The tops of the pipes should becovered adequately to stabilize the environment within each pipe.3 Lengths of PVC electrical conduit (e.g. 25 or 32 mm diameter) with holes (5–8 mm diameter) at 100–150 mm intervals can be packed with dampened, rolledcorrugated cardboard prior to being set in trenches (c.100 mm beneath the soilsurface), which are then backfilled with soil. The pipe system can be made continuousby way of angled connectors, with provision for access in cardboardreservoir traps at prescribed intervals (e.g. 25 m) to allow inspection (Verkerk1990, Verkerk & Bravery 2001). These traps may take the form of cardboardfilledperforated buckets (with lids) or other suitable containers, partially setinto the ground, which should be linked directly to the pipe system. Polystyreneand black polyethylene sheeting can be fixed over the traps to help stabilize environmentalconditions within the containers. The system is highly flexible andcan be used in a wide range of circumstances, but is particularly useful aroundthe perimeter of buildings or other structures. It can also be adapted for installationdirectly into known colony nests so that changes in activity patterns can bemonitored.Commercially available monitoring/baiting systemsSince the mid-1990s, various combined termite baiting and monitoring systemshave been marketed in many industrialized countries with subterraneantermite problems (e.g. USA, Japan, Australia, Spain, Italy, France). Examplesof such systems include Sentricon Colony Elimination System®, SentriTech®, Exterra® and Termigard®. These devices rely on individual in-groundstations which contain a termite food source. Following detection of activityin individual stations during periodic inspections (e.g. at monthly intervals),a termiticidal bait (usually a relatively slow-acting chitin synthesis or metabolicinhibitor) is added. The bait is transferred though the colony by trophallaxis,and cases of successful elimination of field colonies have been reported(e.g. Su & Scheffrahn 1996, Haagsma & Bean 1998, Peters & Fitzgerald 1999,Verkerk & Bravery 2001, Peters & Fitzgerald 2003). The devices can be used in avariety of configurations either as monitoring devices or as combined monitoringand baiting devices. The design of the monitoring station is important in influencingboth termite attack and the sustainability of activity (Lewis et al.1998).

240 CHAPTER 10Remote monitoringA remote, electronic, monitoring device which transmits a signal to a dataloggerwhen termites break a silver foil circuit painted onto polyethylene sheeting, inturn fixed to timber stakes, has shown considerable promise (Su 2002). Suchremote systems are likely to have greatest applicability in high-value heritagesites such as in ancient or historic buildings.CASE STUDY: INTENSIVE TERMITE MONITORING AND BAITINGPROGRAM: DEVON, UNITED KINGDOMAdapted from Verkerk & Bravery 2001In May 1998, an established infestation of a southern European subspecies of subterraneantermite (Reticulitermes lucifugus grassei Clement) was found approximately1000 km north of its indigenous distribution (northern Spain/southwestern France), in asemi-rural, coastal setting in Saunton, Devon, UK. Infestations of such termites generallyare based on expansive, diffuse, and interconnected (frequently “open”) colonies arisingfrom large numbers of neotenics (Clement et al. 2001), so infestations are therefore difficultto eliminate completely. There is some evidence that the termites were importedaccidentally, possibly more than 30 years previously (Jenkins et al. 2001). Surveys and monitoringrevealed a discrete, highly localized infestation extending over some 2400 m 2 .Within this zone, two timber-framed houses surrounded by paving, outbuildings, mixedwoodland, bracken, lawns, and gardens were affected. This case study briefly describesthe monitoring systems which were implemented following the launch in June 1998 of agovernment-funded, consortium-based, 12-year program, the goal of which was to eliminatethe infestation. At the time of writing, intensive monitoring has revealed the site tobe free from termite activity for three and a half years, suggesting the program’s goal mayhave been achieved. Monitoring as part of the government program will continue for10 years from cessation of known activity.Phase 1: initiationThree key activities were undertaken during the first two months of the program:General surveys and establishment of treatment, intensive monitoring andbuffer zonesThe mid-point of the north–south boundary between the two properties known to be infestedwas used to define the central point of the designated “eradication zone.” This1000 m diameter zone (Fig. 10.4) covers 29 independently owned properties concentratedon either side of the Saunton Road (traversing east–west through the center of theeradication zone). All 29 properties were surveyed using torches, probes, and timber“sounding” techniques for evidence of termites, with particular attention being paid toground-floor (and sub-floor) areas, exteriors of buildings, outbuildings, trees and stumps,and other areas where evidence of subterranean termites was most likely to be detected ifpresent (see Sampling of potential food sources and colony nest sites, above).Detailed inspections of properties within the treatment zoneA 75 m radius “treatment” zone was designated, and properties and their grounds withinthis zone were subjected to detailed examination of all timbers susceptible to sub-

METHODS FOR SAMPLING TERMITES 241cabFig. 10.4 Zoning system adopted for the UK Termite Eradication Programme (years 1–3).(a) = treatment zone (75 m radius from center point of known infestation); (b) = intensivemonitoring zone (75–200 m radius from center point); (c) = buffer zone (200–500 m radiusfrom center point). Any termite activity detected outside the treatment zone would haveresulted in appropriate enlargement of all zones. Re-evaluation of the zoning system occursannually.terranean termite attack. The termite infestation as revealed by these surveys was foundto extend over an approximately rectangular area of c.70 m length (east–west) and c.30 mwidth (north–south). Samples of termites were collected and the subspecies was identifiedand confirmed following cuticular hydrocarbon analyses by two independent laboratoriesin France. Samples were subsequently subjected to mitochondrial DNA andphylogenetic analyses (Jenkins et al 2001).Installation of monitoring devicesThe “intensive monitoring” zone was established between 75 m and 200 m radius fromthe notional center of the eradication zone (Fig. 10.4). A “buffer” zone was designated

242 CHAPTER 10from 200–500 m radius. In addition to the installation of 160 commercial monitoring/baitingdevices (Sentricon Colony Elimination System® and Sentri Tech®), four other types ofmonitoring device were installed:1 Wooden stakes. These were prepared from Scots pine Pinus sylvestris sapwood andmeasured 50 ¥ 25 mm section ¥ 500 mm length. Approximately 1000 were installed 3 mapart in more-or-less concentric rows with 10 m between rows throughout the intensivemonitoring zone. In both the intensive monitoring and buffer zones, stakes were installed(at 2–3 m intervals) around all buildings and outbuildings.2 An underground, perimeter “pipe and bucket” system. This comprised corrugatedcardboard-filled,perforated uPVC pipe (32 mm diameter) installed c.100 mm beneath thesoil surface with corrugated-cardboard-filled, perforated bucket stations at approximately25 m intervals around the perimeter of the grounds of both infested properties toprovide a continuous detection system to help determine if termites reached these boundaries.The attractiveness of the cardboard was tested and confirmed by separate tests withtermites feeding within the treatment zone.3 Timber monitoring slates. These were prepared from Pinus sylvestris sapwood andmeasured 25 ¥ 12 mm section ¥ 180 mm length. They were installed into approximately 60holes (25 mm diameter) drilled to reach the ground beneath specific areas of concrete andpaving around the two main buildings known to be infested.4 Conduit-based, grid monitoring system. On apparent collapse of the termite infestation(mid-1999) following installation of a bespoke treatment system based on impregnationof Pinus sylvestris sapwood “wafers” with hexaflumuron in an aqueous extract of afungal attractant (Gloeophyllum trabeum Pers ex Fr), a supplementary grid monitoringsystem (at 3 m centers) was installed. This vertically orientated monitoring system (1 mdepth) comprised two pairs of 500 mm length Pinus sylvestris sapwood timber slates(upper and lower pairs, each linked with cable ties), housed in heavily perforated (5 mmdiameter) conduits (32 mm diameter). The devices were set at 3 m centers and covered theentire extent of the known infestation area. The upper (open) end of each conduit wascapped flush with the ground or paving levels. The system allows for treatment of infestationsby substitution of the upper pair of devices with timber-based baits, should activitybe detected in individual monitoring devices.Phase 2: monthly monitoring and baitingMonthly monitoring began in the third month of the program, for the purpose of checkingon activity in all devices and attaching commercial baits containing hexaflumuron(chitin synthesis inhibitor) (Recruit® or Sentry®) as necessary to active bait stations orconstructional timbers. Termite abundance was assessed in relative terms according tothe number of stations showing activity, and by using an arbitrary index of activity at eachdevice where termites were present. Termite recruitment into bait stations was foundto be very satisfactory, with 17 percent and 22 percent recruitment into the Sentri Tech®stations (92 in total) in July and August 1998 (one and two months respectively followinginstallation of the stations). However, evidence of apparent avoidance behavior bythe termites was noted by September 1998 in nearly all cases where commercial baitshad been deployed, with no subsequent bait consumption. In addition, by October 1998,the area of termite activity detected by the available monitoring devices had extended toapproximately 85 m along the east–west axis.Phase 3: bait refinement phaseThe indications of bait avoidance prompted an intensive laboratory-based investigationusing cultures of R. lucifugus grassei (Devon strain collected from the field) as well as a less

METHODS FOR SAMPLING TERMITES 243sensitive laboratory strain of R. santonensis for comparative purposes. The work culminatedin the development of novel timber-based baits which were attractive to the targetstrain. Based on this work, a prophylactic baiting program was launched in early 1999.Phase 4: prophylactic baiting programBy August 1999, 153 timber-based bait devices and 256 untreated monitoring devices hadbeen installed in the treatment zone. Analysis of the extent of activity at common pointsin August 1998 and August 1999 suggested that the termite population at Saunton appearedto have been suppressed by at least 90 percent following the installation of theprophylactic system. However, to counter the possibility that apparent suppression hadbeen caused by further bait avoidance, the supplementary grid monitoring system was installedwithin the treatment zone.For two years monitoring visits continued monthly, with the exception of the “closedseason”months of December and January. By April 2000 a total of 719 devices had been installedin the treatment zone alone, 535 of these being untreated monitoring devices and177 being hexaflumuron-impregnated timber-based baits. In 2000, all treated deviceswere removed from the site and replaced with untreated devices. Ongoing monitoring inthe 27 other properties within the eradication zone has continued to reveal no evidenceof termite activity. Within the treatment zone, activity was last detected in a structural,above-ground timber in August 2000. With no activity detected in any device or substratefor three and a half years, inspection frequencies have now been reduced to two occasionsa year (early and late season respectively).AcknowledgmentsWe wish to thank the UK Office of the Deputy Prime Minister (ODPM) forfinancial and logistical support, the program Steering Group for its advice, aswell as the members of the project consortium, including: ODPM; BuildingResearch Establishment Ltd.; Imperial College London; Natural HistoryMuseum, London; Dow AgroSciences; UK Forestry Commission; and theNatural Resources Institute (University of Greenwich). Further thanks areoffered to the residents of Saunton who have cooperated with the program sosuccessfully.ReferencesAbe, T. (1980) Studies on the distribution and ecological role of termites in a lowland rain forestof West Malaysia. 4. The role of termites in the process of wood decomposition in PasohForest Reserve. Revue d’Ecologie et de Biologie du Sol, 17, 23–40.Abe, T. & Matsumoto, T. (1979) Studies on the distribution and ecological role of termites in alowland rain forest of West Malaysia. 3. Distribution and abundance of termites in PasohForest Reserve. Japanese Journal of Ecology, 29, 337–351.Abensperg-Traun, M. (1993) A comparison of 2 methods for sampling assemblages ofsubterranean, wood-eating termites (Isoptera). Australian Journal of Ecology, 18, 317–324.

244 CHAPTER 10Adis, J. (1987) Extraction of arthropods from neotropical soils with a modified Kempson apparatus.Journal of Tropical Ecology, 3, 131–138.Atkinson, L. & Adams, E.S. (1997) The origins and relatedness of multiple reproductives incolonies of the termite Nasutitermes corniger. Proceedings of the Royal Society of London Series B:Biological Sciences, 264, 1131–1136.Baroni-Urbani, C., Josens, G., & Peakin, G.J. (1978) Empirical data and demographic parameters.In Production Ecology of Ants and Termites (ed. M.V. Brian), pp. 5–44. Cambridge UniversityPress, Cambridge.Bignell, D.E., Eggleton, P., Nunes, L., & Thomas, K.L. (1997) Termites as mediators of carbonfluxes in tropical forest: budgets for carbon dioxide and methane emissions. In Forests and Insects(ed. A.D. Watt, N.E. Stork, & M.D. Hunter), pp. 109–134. Chapman & Hall, London.Black, H.I.J. & Okwakol, M.J.N. (1997) Agricultural intensification, soil biodiversity andagroecosystem function in the tropics: the role of termites. Applied Soil Ecology, 6, 37–53.Black, H.I.J. & Wood, T.G. (1989) The effects of cultivation on the vertical distribution ofMicrotermes spp. (Isoptera: Termitinae: Macrotermitinae) in soil at Mokwa, Nigeria. Sociobiology,15, 133–138.Buxton, R.D. (1981) Changes in the composition and activities of termite communities in relationto changing rainfall. Oecologia, 51, 371–378.Clement, J.L., Bagneres, A.G., Uva, P., et al. (2001) Biosystematics of Reticulitermes termites inEurope: morphological, chemical and molecular data. Insectes Sociaux, 48, 202–215.Coaton, W.G.H. & Sheasby, J.L. (1975) National survey of the Isoptera of Southern Africa. 10.The genus Hodotermes Hagen (Hodotermitidae). Cimbebasia, 3 (10), 105–138.Collins, N.M. (1979a) A comparison of the soil macrofauna of three lowland forest types inSarawak. Sarawak Museum Journal, 27, 267–282.Collins, N.M. (1979b) Observations on the foraging activity of Hospitalitermes umbrinus(Haviland), (Isoptera: Termitidae) in the Gunung Mulu National Park, Sarawak. EcologicalEntomology, 4, 231–238.Collins, N.M. (1983) Termite populations and their role in litter removal in Malaysian rainforests. In Tropical Rain Forest: Ecology and Management (ed. S.L. Sutton, T.C. Whitmore, & A.C.Chadwick), pp. 311–325. Blackwell, Oxford.Collins, N.M. (1984) The termites (Isoptera) of the Gunung Mulu National Park, with a key tothe genera known from Sarawak. Sarawak Museum Journal, 30, 65–87.Collins, N.M. (1989) Termites. In Tropical Rain Forest Ecosystems: Biogeographical and EcologicalStudies (ed. H. Lieth & M.J.A. Werger), pp. 455–471. Elsevier, Amsterdam.Dangerfield, J.M. & Mosugelo, D.K. (1997) Termite foraging on toilet roll baits in semi-aridsavanna, south-east Botswana (Isoptera: Termitidae). Sociobiology, 30, 133–143.Darlington, J.P.E.C. (1982) The underground passages and storage pits used in foraging by anest of the termite Macrotermes michaelseni in Kajiado, Kenya. Journal of Zoology (London), 198,237–247.Darlington, J.P.E.C. (1984) A method for sampling the population of large termite nests. Annalsof Applied Biology, 104, 427–436.Darlington, J.P.E.C. (1990) Populations in nests of the termite Macrotermes subhyalinus inKenya. Insectes Sociaux, 37, 158–168.Darlington, J.P.E.C. (1991) Relationship of individual weights to nest parameters in termites ofthe genus Macrotermes (Isoptera, Macrotermitinae). Sociobiology, 18, 167–176.Darlington, J.P.E.C. & Dransfield, R.D. (1987) Size relationships in the nest populations andmound parameters in the termite Macrotermes michaelseni in Kenya. Insectes Sociaux, 34,165–180.Davies, R.G., Eggleton, P., Jones, D.T., Gathorne-Hardy, F.J., & Hernández, L.M. (2003) Evolutionof termite functional diversity: analysis and synthesis of local ecological and regionalinfluences on local species richness. Journal of Biogeography, 30, 847–877.

METHODS FOR SAMPLING TERMITES 245Dawes-Gromadzki, T. (2003) Sampling subterranean termite species diversity and activityin tropical savannas: an assessment of different bait choices. Ecological Entomology, 28,397–404.Dejean, A. & Ruelle, J.E. (1995) Importance of Cubitermes termitaries as shelter for alien incipienttermite societies. Insectes Sociaux, 42, 129–136.de Souza, O.F.F. (1993) Effects of habitat fragmentation on termites in Cerrado. PhD thesis, ImperialCollege, University of London.Dibog, L. (1998) Biodiversity and ecology of termites (Isoptera) in a humid tropical forest,southern Cameroon. PhD thesis, Imperial College, University of London.Dibog, L., Eggleton, P., & Forzi, F. (1998) Seasonality of the soil termites in a humid tropical forest,Mbalmayo, southern Cameroon. Journal of Tropical Ecology, 14, 841–850.Donovan, S.E., Eggleton, P., & Bignell, D.E. (2001a) Gut content analysis and a new feedinggroup classification of termites. Ecological Entomology, 26, 356–366.Donovan, S.E., Eggleton, P., Dubbin, W., Batchelder, M., & Dibog, L. (2001b) The effect of asoil-feeding termite, Cubitermes fungifaber (Isoptera: Termitidae) on soil properties: termitesmay be an important source of soil microhabitat heterogeneity in tropical forests. Pedobiologia,45, 1–11.Easey, J.F. & Holt, J.A. (1989) Population estimation of some mound building termites(Isoptera: Termitidae) using radioisotope methods. Material und Organismen, 24, 81–91.Eggleton, P. (2000) Global patterns of termite diversity. In Termites: Evolution, Sociality, Symbiosis,Ecology (ed. T. Abe, D.E. Bignell, & M. Higashi), pp. 25–51. Kluwer, Dordrecht.Eggleton, P. & Bignell, D.E. (1995) Monitoring the response of tropical insects to changes in theenvironment: troubles with termites. In Insects in a Changing Environment (ed. R. Harrington& N.E. Stork), pp. 473–497. Academic Press, London.Eggleton, P. & Bignell, D.E. (1997) Secondary occupation of epigeal termite (Isoptera) moundsby other termites in the Mbalmayo Forest Reserve, southern Cameroon, and its biologicalsignificance. Journal of African Zoology, 111, 489–498.Eggleton, P.E., Williams, P.H., & Gaston, K.J. (1994) Explaining global termite diversity: productivityor history? Biodiversity and Conservation, 3, 318–330.Eggleton, P., Bignell, D.E., Sands, W.A., Waite, B., Wood, T.G., & Lawton, J.H. (1995) Thespecies richness of termites (Isoptera) under differing levels of forest disturbance in theMbalmayo Forest Reserve, southern Cameroon. Journal of Tropical Ecology, 11, 85–98.Eggleton, P., Bignell, D.E., Sands, W.A., et al. (1996) The diversity, abundance, and biomass oftermites under differing levels of disturbance in the Mbalmayo Forest Reserve, southernCameroon. Philosophical Transactions of the Royal Society of London, Series B, 351, 51–68.Eggleton, P., Homathevi, R., Jones, D.T., et al. (1999) Termite assemblages, forest disturbanceand greenhouse gas fluxes in Sabah, East Malaysia. Philosophical Transactions of the RoyalSociety of London, Series B, 354, 1791–1802.Ellwood, M.D.F., Jones, D.T., & Foster, W.A. (2002) Canopy ferns in lowland dipterocarp forestsupport a prolific abundance of ants, termites and other invertebrates. Biotropica, 34,575–583.Evans, T.A. (1997) Evaluation of markers for Australian subterranean termites (Isoptera:Rhinotermitidae & Termitidae). Sociobiology, 29, 277–292.Evans, T.A., Lenz, M., & Gleeson, P.V. (1998) Testing assumptions of mark–recapture protocolsfor estimating population size using Australian mound-building, subterranean termites.Ecological Entomology, 23, 139–159.Ferrar, P. (1982) Termites of a South African savanna. III. Comparative attack on toilet roll baitsin subhabitats. Oecologia, 52, 139–146.Forschler, B.T. & Townsend, M.L. (1996) Mark–recapture estimates of Reticulitermes spp.(Isoptera: Rhinotermitidae) colony foraging populations from Georgia, USA. EnvironmentalEntomology, 25, 952–962.

246 CHAPTER 10Fuchs, A., Schreyer, A., Feuerbach, S., & Korb, J. (2004) A new technique for termite monitoringusing computer tomography and endoscopy. International Journal of Pest Management, 50,63–66.Gathorne-Hardy, F., Syaukani, & Eggleton, P. (2001) The effects of altitude and rainfall on thecomposition of the termites (Isoptera) of the Leuser Ecosystem (Sumatra, Indonesia). Journalof Tropical Ecology, 17, 379–393.Gathorne-Hardy, F.J., Syaukani, Davies, R.G., Eggleton, P., & Jones, D.T. (2002) Quaternaryrainforest refugia in Southeast Asia: using termites (Isoptera) as indicators. Biological Journalof the Linnean Society, 75, 453–466.Greaves, T. (1967). Experiments to determine the populations of tree-dwelling colonies oftermites (Coptotermes acinaciformis (Froggatt) and C. frenchi Hill). In Termites of Australian ForestTrees. Division of Entomology Technical Paper No. 7. Commonwealth Scientific and IndustrialResearch Organization, Australia.Haagsma, K. & Bean, J. (1998) Evaluation of a hexaflumuron-based bait to control subterraneantermites in southern California (Isoptera: Rhinotermitidae). Sociobiology, 31,363–369.Harris, W.V. (1950) Dry-wood termites. The East African Agricultural Journal, 16, 50–52.Haverty, M.I. & Nutting, W.L. (1974) Natural wood-consumption rates and survival of a drywoodand a subterranean termite at constant temperatures. Annals of the Entomological Societyof America, 67, 153–157.Haverty, M.I., Nutting, W.L., & LaFage, J.P. (1975) Density of colonies and spatial distributionof foraging territories of the desert subterranean termite, Heterotermes aureus (Snyder). EnvironmentalEntomology, 4, 105–109.Haverty, M.I., Nutting, W.L., & LaFage, J.P. (1976) A comparison of two techniques for determiningabundance of subterranean termites in an Arizona desert grassland. Insectes Sociaux,23, 175–178.Haverty, M.I., Getty, G.M., Copren, K.A., & Lewis, V.R. (1999) Seasonal foraging and feedingbehaviour of Reticulitermes spp. (Isoptera: Rhinotermitidae) in a wildland and a residentiallocation in northern California. Environmental Entomology, 26, 1077–1084.Henderson, G. (1996) Alate production, flight phenology, and sex ratio in Coptotermesformosanus Shiraki, and introduced subterranean termite in New Orleans, Louisiana. Sociobiology,28, 319–326.Hinze, B. & Leuthold, R.H. (1999) Age related polyethism and activity rhythms in thenest of the termite Macrotermes bellicosus (Isoptera, Termitidae). Insectes Sociaux, 46, 392–397.Hoare, A. & Jones, D.T. (1998) Notes on the foraging behaviour and taxonomy of the southeastAsian termite Longipeditermes longipes (Termitidae: Nasutitermitinae). Journal of Natural History,32, 1357–1366.Holt, J.A. & Easey, J.F. (1985) Polycalic colonies of some mound building termites (Isoptera:Termitidae) in northeastern Australia. Insectes Sociaux, 32, 61–69.Holt, J.A. & Easey, J.F. (1993) Numbers and biomass of mound-building termites (Isoptera) ina semi-arid tropical woodland near Charters Towers, north Queensland, Australia. Sociobiology,21, 281–286.Holt, J.A. & Lepage, M. (2000) Termites and soil properties. In Termites: Evolution, Sociality, Symbioses,Ecology (ed. T. Abe, D.E. Bignell, & M. Higashi), pp. 389–407. Kluwer, Dordrecht.Husseneder, C., Brandl, R., Epplen, C., Epplen, J.T., & Kaib, M. (1998) Variation between andwithin colonies in the termite: morphology, genomic DNA, and behaviour. Molecular Ecology,7, 983–990.Jenkins, T.M., Dean, R.E., Verkerk, R.H.J., & Forschler B.T. (2001) Phylogenetic analyses oftwo mitochondrial genes and one nuclear intron region illuminate European subterranean

METHODS FOR SAMPLING TERMITES 247termite (Isoptera : Rhinotermitidae) gene flow, taxonomy, and introduction dynamics.Molecular Phylogenetics and Evolution, 20, 286–293.Jones, D.T. (1996) A quantitative survey of the termite assemblage and its consumption of foodin lowland mixed dipterocarp forest in Brunei Darussalam. In Tropical Rainforest Research:Current Issues (ed. D.S. Edwards, W.E. Booth, & S. Choy), pp. 297–305. Kluwer, Dordrecht.Jones, D.T. & Eggleton, P. (2000) Sampling termite assemblages in tropical forests: testing arapid biodiversity assessment protocol. Journal of Applied Ecology, 37, 191–203.Jones, D.T. & Gathorne-Hardy, F. (1995) Foraging activity of the processional termite Hospitalitermeshospitalis (Termitidae: Nasutitermitinae) in the rain forest of Brunei, north-west Borneo.Insectes Sociaux, 42, 359–369.Jones, J.A. (1990) Termites, soil fertility and carbon cycling in dry tropical Africa: a hypothesis.Journal of Tropical Ecology, 6, 291–305.Kambhampati, S. & Eggleton, P. (2000) Phylogenetics and taxonomy. In Termites: Evolution,Sociality, Symbioses, Ecology (ed. T. Abe, D.E. Bignell, & M. Higashi), pp. 1–23. Kluwer,Dordrecht.Kirton, L.G., Brown, V.K., & Azmi, M. (1998) A new method of trapping subterranean termitesof the genus Coptotermes (Isoptera: Rhinotermitidae) for field and laboratory experimentalstudies. Sociobiology, 32, 451–458.Lawton, J.H., Bignell, D.E., Bloemers, G.F., Eggleton, P., & Hodda, M.E. (1996) Carbon flux anddiversity of nematodes and termites in Cameroon forest soils. Biodiversity and Conservation, 5,261–273.Lee, K.E. & Wood, T.G. (1971) Termites and Soils. Academic Press, London.Lenz, M., Creffield, J.W., & Barrett, R.A. (1992) An improved field method for assessing the resistanceof woody and non-woody materials to attack by subterranean termites. Material undOrganismen, 27, 89–115.Lewis, V.R., Haverty, M.I., Getty, G.M., Copren, K.A., & Fouche, C. (1998) Monitoring stationfor studying populations of Reticulitermes (Isoptera: Rhinotermitidae) in California. Pan-Pacific Entomologist, 74, 121–133.Lobry de Bruyn, L.A. & Conacher, A.J. (1990) The role of termites and ants in soil modification:a review. Australian Journal of Soil Research, 28, 55–93.Madge, D.S. (1969) Litter disappearance in forest and savanna. Pedobiologia, 9, 288–299.Maki, K. & Abe, T. (1986) Proportion of soldiers in the colonies of a dry wood termite, Neotermeskoshunensis (Kalotermitidae, Isoptera). Physiology and Ecology Japan, 23, 109–117.Mankin, R.W., Osbrink, W.L., Oi, F.M., & Anderson, J.B. (2002) Acoustic detection of termiteinfestations in urban trees. Journal of Economic Entomology, 95, 981–988.Martius, C., Bandeira, A.G., & Medeiros, L.G.S. (1996) Variation in termite alate swarming inrain forests of central Amazonia. Ecotropica, 2 (1), 1–11.Mathot, G. (1967). Premier essai de détermination de facteurs écologiques correlatifs à la distributionet l’abondance de Cubitermes sankurensis (Isoptera, Termitidae, Termitinae). Proceedings,5th Congress of the International Union for the Study of Social Insects, Toulouse, 1965, pp.117–129.Matsumoto, T. & Abe, T. (1979) The role of termites in an equatorial rain forest ecosystem ofwest Malaysia. 2. Leaf litter consumption on the forest floor. Oecologia, 38, 261–274.Medeiros, L.G.S., Bandeira, A.G., & Martius, C. (1999) Termite swarming in the northeasternAtlantic rain forest of Brazil. Studies non Neotropical Fauna and Environment, 34, 76–87.Meyer, V.W., Braack, L.E.O., Biggs, H.C., & Ebersohn, C. (1999) Distribution and density of termitemounds in the northern Kruger National Park, with specific reference to those constructedby Macrotermes Holmgren (Isoptera: Termitidae). African Entomology, 7, 123–130.Mill, A.E. (1983) Observations on Brazilian termite alate swarms and some structures used inthe dispersal of reproductives (Isoptera: Termitidae). Journal of Natural History, 17, 309–320.

248 CHAPTER 10Miller, L.R. (1993) Fluorescent dyes as markers in studies of foraging biology of termitecolonies (Isoptera). Sociobiology, 23, 127–134.Miura, T. & Matsumoto, T. (1998) Foraging organization of the open-air processional lichenfeedingtermite Hospitalitermes (Isoptera, Termitidae) in Borneo. Insectes Sociaux, 45 (1),17–32.Myles, T.G. (1996) Development and evaluation of a transmissible coating for control of subterraneantermites. Sociobiology, 28, 373–457.Noirot, C. (1970) The nests of termites. In Biology of Termites, vol. 2 (ed. K. Krishna & F.M.Weesner), pp. 73–125. Academic Press, New York.Nutting, W.L. (1969) Flight and colony foundation. In Biology of Termites, vol. 1 (ed. K. Krishna& F.M. Weesner), pp. 233–282. Academic Press, New York.Nutting, W.L. & Jones, S.C. (1990) Methods for studying the ecology of subterranean termites.Sociobiology, 17, 167–189.Ohiagu, C.E. (1979) Nest and soil populations of Trinervitermes spp. with particular reference toT. geminatus (Wasmann), (Isoptera), in southern Guinea savanna near Mokwa, Nigeria. Oecologia,40, 167–178.Pearce, M.J. (1990) A new trap for collecting termites and assessing their foraging activity.Tropical Pest Management, 36, 310–311.Pearce, M.J. (1997) Termites: Biology and Pest Management, CAB International, Wallingford.Pearce, M.J., Cowie, R.H., Pack, A.S., & Reavey, D. (1990) Intraspecific aggression, colonyidentity and foraging distances in Sudanese Microtermes spp. (Isoptera, Termitidae, Macrotermitinae).Ecological Entomology, 15, 71–77.Peters, B.C. & Fitzgerald, C.J. (1999) Field evaluation of the effectiveness of three timberspecies as bait stakes and the bait toxicant hexaflumuron in eradicating Coptotermes acinaciformis(Froggatt) (Isoptera: Rhinotermitidae). Sociobiology, 33, 227–238.Peters, B.C. & Fitzgerald, C.J. (2003) Field evaluation of the bait toxicant chlorfluazuron ineliminating Coptotermes acinaciformis (Froggatt) (Isoptera : Rhinotermitidae). Journal of EconomicEntomology, 96, 1828–1831.Pomeroy, D.E. (1977) The distribution and abundance of large termite mounds in Uganda.Journal of Applied Ecology, 14, 465–475.Potter, M.F., Eliason, E.A., Davis, K., & Bessin, R.T. (2001) Managing subterranean termites(Isoptera : Rhinotermitidae) in the Midwest with a hexaflumuron bait and placement considerationsaround structures. Sociobiology, 38, 565–584.Rebello, A.M.C. & Martius, C. (1994) Dispersal flights of termites in Amazonian forests(Isoptera). Sociobiology, 24, 127–146.Roisin, Y. & Pasteels, J.M. (1986) Reproductive mechanisms in termites: polycalism and polygynyin Nasutitermes polygynus and N. costalis). Insectes Sociaux, 33, 149–167.Salt, G. (1952) The arthropod population of the soil in some East African pastures. Bulletin ofEntomological Research, 43, 203–220.Sands, W.A. (1961) Nest structure and size distribution in the genus Trinervitermes (Isoptera,Termitidae, Nasutitermitinae) in West Africa. Insectes Sociaux, 8, 177–188.Sands, W.A. (1965) Mound population movements and fluctuations in Trinervitermes ebenerianusSjöstedt. Bulletin de l’Union Internationale pour l’étude des insectes sociaux, 12, 49–58.Sands, W.A. (1972) Problems in attempting to sample tropical subterranean termite populations.Ekologia Polska, 20 (3), 23–31.Schuurman, G. & Dangerfield, J.M. (1997) Dispersion and abundance of Macrotermesmichaelseni colonies: a limited role for intraspecific competition. Journal of Tropical Ecology, 13,39–49.Silva, E.G. & Martius, C. (2000) Termite (Isoptera) sampling from soil: handsorting orKempson extraction? Sociobiology, 36, 209–216.

METHODS FOR SAMPLING TERMITES 249Sleaford, F., Bignell, D.E., & Eggleton, P. (1996) A pilot analysis of gut contents in termites fromthe Mbalmayo Forest Reserve, Cameroon. Ecological Entomology, 21, 279–288.Spragg, W.T. & Paton, R. (1977) Tracing, trophallaxis and population measurement of coloniesof subterranean termites (Isoptera) using a radioactive tracer. Annals of the Entomological Societyof America, 73, 708–714.Strickland, A.H. (1944) The arthropod fauna of some tropical soils, with notes on the techniquesapplicable to entomological soil surveys. Tropical Agriculture, 21, 107–114.Su, N.Y. (2002) Dimensionally stable sensors for a continuous monitoring program to detectsubterranean termite (Isoptera: Rhinotermitidae) activity. Journal of Economic Entomology,95, 975–980.Su, N.Y. & Scheffrahn, R.H. (1996) Fate of subterranean termite colonies (Isoptera) after baitapplications: an update and review. Sociobiology, 27, 253–275.Su, N.Y., Scheffrahn, R.H., & Ban, P.M. (1988) Retention time and toxicity of a dye marker,Sudan red 7B, on formosan and eastern subterranean termites (Isoptera: Rhinotermitidae).Journal of Entomological Science, 23, 235–239.Su, N.Y., Ban, P.M., & Scheffrahn, R.H. (1991) Evaluation of twelve dye markers for populationstudies of the eastern and Formosan subterranean termite (Isoptera: Rhinotermitidae). Sociobiology,19, 349–362.Su, N.Y., Ban, P.M., & Scheffrahn, R.H. (1993) Foraging populations and territories of the easternsubterranean termite (Isoptera: Rhinotermitidae) in southeastern Florida. EnvironmentalEntomology, 22, 1113–1117.Sugimoto, A., Bignell, D.E., & MacDonald, J.A. (2000) Global impact of termites on the carboncycle and atmospheric trace gases. In Termites: Evolution, Sociality, Symbiosis, Ecology (ed. T.Abe, D.E. Bignell, & M. Higashi), pp. 409–435. Kluwer, Dordrecht.Sugio, K. (1995) Trunk trail foraging of the fungus-growing termite Macrotermes carbonarius(Hagen) in southeastern Thailand. Tropics, 4, 211–222.Takematsu, Y., Inoue, T., Hyodo, F., Sugimoto, A., Kirtebutr, N., & Abe, T. (2003) Diversity ofnest types in Microcerotermes crassus (Termitinae, Termitidae, Isoptera) in a dry evergreen forestof Thailand. Sociobiology, 42, 587–596.Tayasu, I., Abe, T., Eggleton, P., & Bignell, D.E. (1997) Nitrogen and carbon isotope ratios in termites:an indicator of trophic habit along the gradient from wood-feeding to soil-feeding.Ecological Entomology, 22, 343–351.Taylor, H.S., MacKay, W.P., Herrick, J.E., Guerra, R.A., & Whitford, W.G. (1998) Comparison offield methods to detect termite activity in the Northern Chihuahuan Desert. Sociobiology, 32,1–15.Thoms, E.M. (2000) Use of an acoustic emissions detector and intragallery injection ofspinosad by pest control operators for remedial control of drywood termites (Isoptera:Kalotermitidae). Florida Entomologist, 83, 64–74.Thorne, B.L., Russek-Cohen, E., Forschler, B.T., Breisch, N.L., & Traniello, J.F.A. (1996) Evaluationof mark–release–recapture methods for estimating forager population-size of subterraneantermite (Isoptera: Rhinotermitidae) colonies. Environmental Entomology, 25, 938–951.Usher, M.B. & Ocloo, J.K. (1974) An investigation of stake size and shape in “graveyard” fieldtests for termite resistance. Journal of the Institute of Wood Science, 9, 32–36.Verkerk, R.H.J. (1990) Building Out Termites: an Australian Manual for Environmentally ResponsibleControl, Pluto Press, Leichhardt, Australia.Verkerk, R.H.J & Bravery, A.F. (2001) The UK termite eradication programme: justificationand implementation. Sociobiology, 37, 351–360.Watt, A.D., Stork, N.E., Eggleton, P., et al. (1997) Impact of forest loss and regeneration on insectabundance and diversity. In Forests and Insects (ed. A.D. Watt, N.E. Stork, & M.D. Hunter),pp. 273–286. Chapman & Hall, London.

250 CHAPTER 10Weissling, T.J. & Thoms, E.M. (1999) Use of an acoustic emission detector for locating Formosansubterranean termite (Isoptera: Rhinotermitidae) feeding activity when installingand inspecting aboveground termite bait stations containing hexaflumuron. FloridaEntomologist, 82, 60–71.Wilson, E.O. (1992) The effects of complex social life on evolution and biodiversity. Oikos, 63,13–18.Wood, T.G. (1978) Food and feeding habits of termites. In Production Ecology of Ants and Termites(ed. M.V. Brian), pp. 55–80. Cambridge University Press, Cambridge.Wood, T.G. (1996) The agricultural importance of termites in the tropics. Agricultural ZoologyReviews, 7, 117–155.Wood, T.G. & Johnson, R.A. (1978) Abundance and vertical distribution in soil of Microtermes(Isoptera: Termitidae) in savanna woodland and agricultural ecosystems at Mokwa, Nigeria.Memorabilia Zoolgie, 29, 203–213.Wood, T.G. & Lee, K.E. (1971) Abundance of mounds and competition among colonies of someAustralian termite species. Pedobiologia, 11, 341–366.Wood, T.G. & Sands, W.A. (1978) The role of termites in ecosystems. In Production Ecology of Antsand Termites (ed. M.V. Brian), pp. 245–292. Cambridge University Press, Cambridge.Wood, T.G., Johnson, R.A., Ohiagu, C.E., Collins, N.M., & Longhurst, C. (1977). Ecology and Importanceof Termites in Crops and Pastures in Northern Nigeria. Project report 1973–76, Centre forOverseas Pest Research, London.Wood, T.G., Johnson, R.A., Bacchus, S., Shittu, M.O., & Anderson, J.M. (1982) Abundanceand distribution of termites (Isoptera) in a riparian forest in the Southern Guinea savannazone of Nigeria. Biotropica, 14, 25–39.Index of methods and approachesMethodology Topics addressed CommentsEstimating colony densityPlot surveys Visual survey of plots to count the Only possible withnumber of mounds and/or conspicuous colonies. Aerialarboreal nests.photography can be used toRecording the colony coordinates survey larger mounds in openwithin plots also allows estimates habitats. Often difficult to seeof dispersion.arboreal nests in forest canopy,even using binoculars. Manyarboreal nests may be hiddeninside wood or large epiphytes.Detection of acoustic Hand-held device for detecting This has been used for theemissions the acoustic emissions from detection of pest species intermite colonies hidden inside buildings and urban trees. Thewood.sampling methodology of thisrelatively new technology hasnot yet been fully developed,nor has it been tested on nonpestspecies in natural habitats.Continued

METHODS FOR SAMPLING TERMITES 251Methodology Topics addressed CommentsEstimating population densityDirect sampling of Estimation of population in a Termites can be extracted fromnests single colony by destructive nest material by hand-sortingsampling of entire nest, or sub- or flotation methods. Samplingsampling of part of the nest. an entire nest can be verylabor-intensive and timeconsuming.Fumigation of nestto kill the colony beforesampling gives a much moreaccurate population estimate.Mark–recapture Estimation of colony Population estimates can betechniques population size by mark– very inaccurate compared withrecapture protocols usingdirect population countscolored stains, dyes, orbecause the assumptions of theradioisotopes to markprotocol are often violated.individual termites.Soil pits, monoliths, Estimation of subterranean Very labor-intensive and/oror trenches termite population by digging pits, time-consuming. Densitymonoliths, or trenches, and then estimates often have highremoving the termites by hand- variances due to the verysorting the soil. Digging very large patchy distribution of termitespits or long trenches can also give in soil. Less effective inestimates of hypogeal colony savannas, where termites oftendensity.migrate further down the soilprofile compared with termitesin forest soils.Soil cores Estimation of subterranean termite Collecting soil samples with apopulation by taking soil cores, sharp-edged corer is quickerand then removing the termites by than digging pits, monoliths orhand-sorting, flotation or using an trenches. Removing theautomatic extraction device. This termites by hand-sorting themethod does not give accurate soil is considered to be moreestimates of the density of efficient than using flotationhypogeal colonies.methods or extraction devicessuch as Berlese–Tullgrenfunnels.Sampling dead wood Estimation of total population by Very labor-intensive and/orsplitting open all items of dead time-consuming because allwood collected within quadrats, items of wood must be splitand hand-sorting the termites. lengthways and individualPopulations in larger items of termites extracted by hand.wood can be estimated by The semi-quantitative methodsampling sub-units of known of visual estimation has nevervolume or weight. Alternatively, been calibrated against directthe wood can be split open and counts to test its accuracy.Continued

252 CHAPTER 10Methodology Topics addressed Commentsthe number of individuals can beestimated visually by counting inunits of 10s, 100s, or 1000s.Sampling regime for A plot-based sampling regime for A standardized samplingestimating population estimating the population density regime that combines plotdensity of a local of the local assemblage in tropical surveys of mounds and neststermite assemblage forest by sampling termites in with soil pits and dead woodmounds, nests, dead wood, and from quadrats. Very laborsoil.intensive and time-consuming.Excludes arboreal termitesoccurring at more than 2 mabove ground level.Assessing species compositionTransect protocol A rapid sampling method for This protocol has been shownassessing the species composition to produce samples thatof a local assemblage in a tropical represent accurately theforest. This protocol standardizes taxonomic and functionalthe amount of sampling effort and composition of the localarea along a belt transect.assemblage. No sampling isconducted at more than 2 mabove ground level but manyarboreal-nesting species arecollected as they forage on theground.Baiting methods Estimation of local species Not all wood-feeding speciesrichness by attracting foraging are attracted to cellulose baits.termites to cellulose baits. A range Species show preferences forof materials can be used as baits, different bait material, so aincluding wood, litter bags, toilet combination of bait types isrolls, corrugated cardboard, recommended to maximize thedungpats, and sawdust.number of species attracted.Light trapping Estimation of local species Light traps fail to capturerichness by attracting alates to many species because not alllight traps at night.show a clear preference fornocturnal swarming, and thealates of many species have apoor dispersal range. Evenrunning traps throughout theyear may only capture abouthalf of the known localspecies.Studying termite activityObserving termites Estimation of the number of Only a relatively small numberthat forage in the individuals involved in foraging by of species send outopen air direct counts in real time, or from processional columns ofclose-up photographs taken of the foragers that move in the openContinued

METHODS FOR SAMPLING TERMITES 253Methodology Topics addressed Commentsforaging column at regular air. Many of these only forageintervals. The amount of food at night, making it morecollected by the colony can be difficult to quantify theirestimated if the workers return activity.with visible balls of forage held intheir mandibles. The foragingrange and frequency of individualcolonies can be studied easily.Baiting experiments Studying rates of consumption, Ensure that timber purchasedfood preference and size of as baits has not been treatedforaging territory. Can also be used with pesticide, and that toiletto study inter- and intraspecific rolls are unscented as onlyforaging activity.foragers are attracted to baits,this gives a measure of relativeintensity of foraging activityrather than relative populationdensity.Mark–recapture Studying foraging activity and Provided sufficient foragersterritory size by using colored are marked, it is easier tostains, dyes, or radioisotopes to produce reliable estimates ofmark individual termites.foraging activity and territorysize than it is to use mark–recapture techniques toestimate population density.Light trapping Studying temporal patterns of Local assemblages show littleswarming by attracting alates interspecific synchronicity ofto light traps.swarming, so traps operatingover short periods will fail tocapture many species. Evenrunning traps throughout theyear may only capture abouthalf of the known localspecies.In addition, not all speciesshow a clear preference fornocturnal swarming, and thealates of many species have apoor dispersal range.Detection of acoustic Detecting infestations of termites This has been used for theemissions hidden in wood by using a detection of pest species inhandheld device.buildings and urban trees. Thesampling methodology of thisrelatively new technology hasnot yet been fully developed,nor has it been tested on nonpestspecies in natural habitats.

CHAPTER 11Parasitoids and predatorsNICK MILLSIntroductionInsect parasitoids and predators are major contributors to the third trophic levelof terrestrial plant-based food webs, and pose some interesting and unique constraintsfor sampling. For example, as parasitoids spend a large part of their lifecycle in intimate association with a host, sampling is often based on the hostpopulation with few opportunities to estimate absolute densities of the freelivingadult stage. In contrast, predators are always free-living and predatorpopulations or communities can readily be sampled independently of prey populations.In addition, the third trophic level imposes constraints on the abundanceof parasitoid and predator populations relative to their host populations,and thus the accuracy of estimates of abundance or diversity are more dependenton large sample size. Finally, although the size of insect parasitoids and predatorsvaries from the larger rhyssine ichneumonids and carabid beetles to thesmallest mymarid egg parasitoids and coccinellids, the small size of many entomophagousspecies can make them rather more difficult to sample than otherinsects using standard sampling techniques.There are many reasons why researchers may need to sample insect parasitoidand predator populations, but the three most frequent can be categorizedas evaluating (i) the composition of the entomophagous assemblage associatedwith a particular host species, (ii) the impact of entomophagous species onthe dynamics of a particular host population, and (iii) the biodiversity of entomophagousspecies within a local or regional community. In this chapter I willdiscuss the issues that pertain to the sampling of insect parasitoids and predatorsfor each of these purposes, and indicate how they build upon the standard samplingtechniques discussed in other chapters.Composition of entomophagous insect assemblagesTremendous care needs to be exercised in the sampling of entomophagous insectassemblages, as the literature contains numerous errors in predator–preyand host–parasitoid associations (Askew & Shaw 1986, Hodek 1993, Shaw1994, 1997). Mistakes can arise as a result of either misidentification or incor-254

PARASITOIDS AND PREDATORS 255rect association. As the majority of parasitoids and predators attack juvenilestages of their host, it can be difficult to correctly identify the immature host,particularly when several host species occur together in the same location. Similarly,the majority of insect predators are predaceous during their immaturestages, and correct identification of juvenile predators can be challenging.In many cases it is necessary to rear the immature parasitoids and predatorsthrough to the adult stage for correct identification, and it is surprising how frequentlyother insects can be introduced into a rearing and result in incorrect associations.This is less of a problem for insect predators, but of particular concernfor insect parasitoids as the food plant used in rearing may support additionalsmall or cryptic species that produce parasitoids of their own. This is particularlynoticeable when parasitoids recovered from a rearing include species that belongto a taxonomic group that has absolutely no association with the host insectof interest. For example, it is not uncommon for aphid parasitoids(Braconidae: Aphidiinae) to be recorded from rearings of lepidopteran larvae(Herting 1975), due to the presence of parasitized aphids or aphid mummies onthe foliage or bark of the food plant. As aphidiine parasitoids have never beenreared from Lepidoptera, but instead have a strong and close association withthe Aphididae (Stary 1970), the error can easily be detected. However, whencontaminating hosts are more closely aligned to the host in question, such asa buprestid feeding under the bark of a log primarily infested with scolytidbeetles, or a twig-boring lepidopteran in the branches of foliage fed to leafchewingLepidoptera, the incorrect associations are far more difficult to detect.Parasitoid assemblagesThe intimacy of contact between parasitoid and host requires correct associationsto be determined by rearing. Direct observation of parasitoid speciesprobing a particular host provides an indication of potential association,but such observations cannot be relied upon as the parasitoid may either notaccept the host for oviposition (Kitt & Keller 1998), or eggs may be consistentlyencapsulated by the host to prevent parasitism (Blumberg & Van Driesche2001).Contamination of host rearings by other host insects or parasitoids can beminimized by use of the following procedures (after Shaw 1997):• Search the food plant thoroughly to remove contaminants, not only fromfoliage but also from the bark, stem, flowers, or fruit.• Never rear more than one host species together in the same rearing container,even if they are easily distinguishable.• Count the host individuals and parasitoid clutches to more accuratelyassociate host mortality with the presence of parasitoid cocoons, pupae, orpuparia.• Keep exited parasitoids as individual clutches whenever possible to accuratelyassociate adults with immature stages.

256 CHAPTER 11• Preserve the host remains and parasitoid cocoons/puparia with the adult parasitoidto provide a means of checking host identity and associating immaturestages of the parasitoids with the adult.To determine the complete parasitoid assemblage of a host species it is necessaryto sample all stages of the life cycle of the host insect, as parasitoids havevaried, and sometimes very narrow, windows of parasitism of a host (Mills1994). Larval and nymphal stages of host insects have frequently been sampledfor parasitism, but we know far less about parasitoids confined to the egg, pupal,or adult stages of their hosts. The most important factor influencing the speciesrichness of parasitoid assemblages is sample size (Hawkins 1994). The numberof hosts that must be collected from a local host population is influenced by twofactors, the probability of a host individual being parasitized, and the degreeof heterogeneity in the distribution of the parasitoid species within the hostpopulation. On average, the number of parasitoid species associated withholometabolous hosts in the UK is approximately seven, but in individual instancescan be as high as 25 (Hawkins 1988). Nonetheless, to be able to collectthe majority of species in a local parasitoid assemblage requires a sample size of1000 individuals or more (Hawkins 1994). In some cases, surveys of parasitoidassemblages for readily reared hosts can be facilitated by use of trap hosts to increasesample size (e.g. Floate et al. 1999). In contrast, the parasitoid assemblagesof hemimetabolous hosts contain far fewer species. For example, aphidssupport an average of about two parasitoid species (Porter & Hawkins 1998,Stadler 2002), although individual aphid species may support as many as eight(Muller et al. 1999). The lower richness of hemimetabolous parasitoid assemblagesshould reduce the influence of heterogeneity (Keating & Quinn 1998)and facilitate the estimation of local assemblage richness from smaller samplesizes. In this regard, the biodiversity software package EstimateS (Colwell 1997)can be used to indicate the sufficiency of sample size, and to estimate the totalspecies richness of the complete parasitoid assemblage, as used by Stiremanand Singer (2002) in studying the parasitoid assemblage of a polyphagouscaterpillar.Predator assemblagesIn contrast to parasitoid assemblages, the lack of intimacy in the interaction betweenpredators and their prey poses interesting challenges for correct identificationof the range of predators associated with a particular prey species. Oneof three approaches can be used, direct observation, indirect sampling, or gutdetection.Direct observationThis is the most effective sampling strategy as it is the only technique that unequivocallyensures the validity of the predator–prey association. Qualitative

PARASITOIDS AND PREDATORS 257sampling is accomplished for any readily observable prey species, by visualsearch, collection, and correct identification of all predators seen to be activelyfeeding on the prey species. Direct observation is time consuming, but has theadvantages that it can also be used to quantify predator abundance, and that itprovides the observer with an intimate knowledge of the activity and behaviorof the different predator species. As predators are most frequently observed intheir immature stages, correct identification is essential. This can be achieved bybuilding a reference set of close-up photographs showing the successive stagesof development as individuals are reared through to maturity. In addition,valuable keys to the immature stages of some of the more frequent taxa ofinsect predators include those for the Coccinellidae (Hodek 1973, Gordon &Vandenberg 1995, Rhoades 1996), Chrysopidae (Diaz-Aranda & Monserrat1995, Tauber et al. 2000, Monserrat et al. 2001), and Syrphidae (Rotheray 1988,1993, Rotheray & Gilbert 1989, 1999).Indirect samplingThe most frequent approach for evaluating predator assemblages is indirectsampling through use of either suction samplers (e.g. Parajulee & Slosser 1999),sweep nets (e.g. Elliott & Kieckhefer 1990), beating trays (e.g. Wyss 1996), oryellow pan traps (e.g. Bowie et al. 1999). Although these sampling techniquesprovide greater numbers of entomophagous species than direct observation,the problem of accurate association is paramount, and so caution should be exercisedwhen using indirect sampling to identify the predator assemblage of aparticular prey species. The associations should at least be tested in feeding trials,using simple laboratory arenas with predators at different stages in their development.Care must also be taken to distinguish essential prey, which fullysupport growth and reproduction of the predator, from alternative prey, whichcan sustain predators for short periods of time (Hodek 1993). Indirect samplingis best used as a technique to quantify the absolute or relative abundance ofmembers of a predator assemblage that are known to be true associates fromeither direct observation or gut detection. Both the choice of sampling methodand the time of day that samples are collected can have an important influenceon the types of predators collected. For example, Brown and Schmitt (2001)found that chrysopids, clerids, and Thysanoptera were more effectively sampledfrom apple trees by beating at night. Similarly, Costello and Daane (1997)found that drop cloths were more efficient than funnels or suction samplers formonitoring spider assemblages in vines.Gut detectionThis last approach makes use of either artificial (elemental and immunological)or natural (immunological and PCR) markers for the prey and methods toscreen predators, collected through indirect sampling, for the presence of

258 CHAPTER 11markers in their gut. Artificial markers include rubidium, which must be fed toprey via the food plant or diet (Akey & Burns 1991, Johnson & Reeves 1995),and readily available vertebrate immunoglobulins, which can either be fed toprey or applied topically (Hagler & Durand 1994). In contrast, natural markersare present naturally in all prey populations and take the form of monoclonalantibodies (currently available for only a small number of prey species; Greenstone1996) and PCR products (Chen et al. 2000). Standard techniques areavailable to detect the presence or absence of all four forms of marker in the gutsof individual predators (see individual references cited above). Two key difficultieswith gut detection are that they do not distinguish necrophagy, saprophagy,and higher-order predation from primary predation, and that detectability andreliability vary with predator species and size, with meal size and time sincefeeding, and in some cases with the size of the marker. However, the relativeease with which PCR primers can be used to develop specific markers for individualprey species is likely to lead to significant advancements in gut detectionin the near future and its more widespread use in the characterization of predatorassemblages.Regional variationFor both parasitoids and predators, the regional species richness of an assemblageis likely to be influenced by the rate of turnover or beta-diversity of parasitoidor predator species across the geographic distribution of the host. Thus itis important to consider how much sampling must be conducted a single site,and how many sites should be sampled in order to provide an accurate estimateof a regional assemblage. For example, Hawkins (1988) clearly shows howthe size of the species-rich parasitoid assemblages of Lepidoptera in the UK increaseswith spatial scale. In contrast, the size of the parasitoid assemblages ofgrass-feeding chalcid wasps remains more or less constant throughout the UK,with little evidence of regional variation (Dawah et al. 1995). The number ofsites sampled will be of greatest importance in assessing regional diversity, asshown by Gaston and Gauld (1993) for the pimpline ichneumonid fauna ofCosta Rica, but it remains unclear how large a sample is needed at each locality.Sampling to estimate the impact of entomophagous speciesThe most frequent need for sampling entomophagous insects is to estimatetheir impact on the dynamics of a particular host population (Luck et al. 1988,Sunderland 1988, Mills 1997). As both parasitoids and predators are importantsources of mortality at all stages in the life cycle of phytophagous insects weoften want to know which entomophagous species play the greatest role in reducingor regulating host abundance. The impact of entomophagous speciescan be analyzed through life table analysis (Bellows et al. 1992, Yamamura

PARASITOIDS AND PREDATORS 2591999) or through simulation modeling (Gutierrez 1996, van Lenteren & vanRoermund 1999, Kean & Barlow 2001). Both approaches require data fromthe regular sampling of both host and entomophagous insect populations andpresent many challenges to the field ecologist. As parasitoids have an intimaterelationship with their host that lasts many days, the impact of parasitoids hasfocused on sampling hosts to directly estimate percent parasitism. In contrast,as predation events are so brief in time it is extremely difficult to obtain a directestimate of predation mortality, and thus sampling has focused on monitoringthe abundance and consumption rate of predators.Percent parasitismPercent parasitism is commonly estimated from a sample of hosts as the ratio ofthe number parasitized to the total number of hosts in the sample. In general,reports of percent parasitism in the literature are based either on the peak or themean level of parasitism from a series of samples collected at intervals from thesame location. Van Driesche (1983) points out the gross inaccuracies in thiscommon practice, and Van Driesche et al. (1991) provide a number of solutionsfor more effective estimation of percent parasitism. For hosts with discrete generations,the goal is to estimate stage-specific parasitism, the percentage of hostindividuals recruited to the susceptible stage that are attacked by a particularparasitoid species, representing the generational mortality attributable to parasitism(Table 11.1). In contrast, for hosts with overlapping generations, it isTable 11.1 A summary of the methodological approaches for estimating the impact ofparasitism.Methodology Application ExamplesStage-specific generational parasitism for hosts with discrete generationsSingle sample Effective only if susceptible host stage is Hill (1988), Gouldclearly separated from host stage killed et al. (1992a)Death rate analysis Useful when there is insufficient Gould et al. (1992a)knowledge of host stages susceptible toparasitismRecruitment analysis Unbiased, but requires detailed knowledge Van Driesche &and regular sampling Bellows (1988),Gould et al. (1992a)Time-specific rate of parasitism for hosts with discrete or overlapping generationsInclusive host stages Should be applied in all cases to avoidbias caused by non-inclusive stagesRecruitment analysis Avoids bias, but requires detailed Lopez and Vanknowledge of susceptible stages Driesche (1989)Correction for differential Should be applied whenever parasitism Russell (1987)duration of development influences host development time

260 CHAPTER 11important to estimate a time-specific rate of parasitism, the percentage of hoststhat die from parasitism during a specific interval of time, and how that ratechanges over a season (Table 11.1). The susceptible stage of the host is the earliestlife stage or instar that can be attacked by the parasitoid of interest, a characteristicof either the parasitoid guild (a classification that defines the pattern ofhost utilization by parasitoids; Mills 1994) or the individual parasitoid species.In some cases, the estimation of percent parasitism also requires knowledge ofthe earliest host stage or instar killed by the parasitoid, such that the inclusivehost stages, those that support the parasitoid during its successive stages of developmentfrom oviposition to emergence from the host, can be clearly defined.Stage-specific generational parasitismThe key to estimation of generational parasitism is to devise a sampling plan thatprovides accurate input for the ratio estimate of percent parasitism. The numeratorof the ratio must be representative of the total number of host individuals inthe population parasitized by a specific parasitoid species, and the denominatormust be representative of the total number of host individuals in the populationrecruited to the susceptible stage. There are three ways in which generationalparasitism can be estimated, depending on the life history of the parasitoid:single samples, death rate analysis, and recruitment analysis.Single samplesUnder some circumstances the life histories of host and parasitoid can facilitatethe sampling of host populations for the estimation of generational parasitism.This occurs when the susceptible host stage is clearly separated from the hoststage killed by the parasitoid (Van Driesche 1983, Royama 2001). Several guildsof endoparasitoids of holometabolous hosts are known to attack early inhost life cycle and delay their development to kill much later host stages(Mills 1994), such as egg–prepupal parasitoids, early larval parasitoids, and larval–pupalparasitoids. For these parasitoids there is a point in the host life cyclewhen parasitoid attack is complete, but mortality from parasitism has yet tooccur. In other situations, host diapause can also arrest host and parasitoid developmentat a point intermediate between the susceptible stage and the stageexperiencing mortality from parasitism. In both of these cases, a single sampleof hosts collected when the population is at an intermediate life stage can providean effective estimate of generational parasitism (Van Driesche 1983). Thisapproach proved valuable for estimating percent parasitism by larval parasitoidsof both armyworm (Hill 1988) and gypsy moth larvae (Gould et al.1992a). Of course, such estimates can still be biased by any differential mortalityof parasitized and healthy hosts that might occur before collection of the hostsample or during subsequent rearing.Additionally, parasitoids often leave distinct emergence holes in the resilientcoverings of eggs, pupae and many sessile Homoptera. In these instances, it is

PARASITOIDS AND PREDATORS 261possible to monitor generational mortality directly from the percentage ofparasitized individuals in a single sample collected after hosts have developedbeyond the inclusive stage (Van Driesche 1983). This approach has often beenused for egg (e.g. Nakamura & Abbas 1987) and pupal (e.g. Doane 1971) parasitism,but can lead to underestimation if there is a significant level of predationas a contemporaneous mortality factor. In the latter case, both predation andparasitism must be estimated and marginal attack rates calculated (see below).Death rate analysisDeath rate analysis was developed by Gould et al. (1989, 1992a) to estimategenerational parasitism in populations of gypsy moth, and is based on regularmonitoring of hosts killed by a specific parasitoid species. A series of host samplesmust be collected at regular intervals and for each sample the number of individualskilled by the parasitoid is noted while the hosts are held under naturalconditions for the interval between two successive samples. The approach differsfrom the more traditional measurement of parasitism in that the death ratefrom parasitism is monitored for the interval between samples only rather thankeeping each sample until all hosts have either died from parasitism or developedbeyond the inclusive stage. The death rate is first converted to a proportionsurviving from parasitism, and generational parasitism is subsequently estimatedfrom (1 - s) * 100, where s is the product of the proportion survivingfrom each of the successive samples (Gould et al. 1992a). Death rate analysis hasthe advantage that it does not require prior knowledge of the host stages susceptibleto parasitism or those that are killed by parasitism. However, the seriesof successive samples must span the complete period during which hosts arekilled by the parasitoid, and each sample must be representative of all hoststages present at that time. It also assumes that no other fast-acting mortalityfactor could have intervened in the field during the interval between samples tokill some of the host individuals in each sample before they are killed by parasitism.As a result it is important to use a sample interval that is short enough toreduce the chances of bias due to the exclusion of contemporaneous mortalityfactors. Death rate analysis has not been used extensively, but has proved valuablefor monitoring parasitism of gypsy moth larvae (Gould et al. 1990, 1992a),and a variant of the approach is advocated by Royama (2001).Recruitment analysisIn some cases, the recruitment of hosts to the susceptible stage and of parasitoidsto the host population can be estimated directly. Van Driesche and Bellows(1988) successfully developed this approach to estimate the generational parasitismof imported cabbageworm by the larval parasitoid Cotesia glomerata.Recruitment of individuals to the susceptible host stage (young larvae) wasmeasured by removing all larvae from randomly selected collard plants andcounting the number of new first and second instar larvae 3–4 days later. Thisprocedure was repeated every 3–4 days over the complete egg-laying period of

262 CHAPTER 11the host, using new plants each time, and the counts of young larvae summed toestimate total recruitment to the susceptible stage for that generation. The recruitmentof parasitized hosts was then measured using a short marker stagemethod. This involved dissection of the host larvae removed from the collardplants each time to count the number of hosts containing parasitoid eggs (excludingthose containing parasitoid larvae), representing only those parasitoidsrecruited during the previous sampling interval. The total count of hosts withnewly recruited parasitoids from the complete set of samples provided an estimateof the sum of hosts parasitized for the generation, and generational parasitismcould be estimated from the ratio of the sum of parasitized hosts to thesum of hosts recruited to the susceptible stage.As an alternative, the recruitment of parasitized hosts can be measured usingthe trap host method, in which susceptible-stage hosts are exposed to parasitismunder natural conditions for a period equivalent to the sample interval. Thisworks best for sessile hosts, as mobile hosts often have a tendency to move awayfrom the location where released. In addition, it is critical to use trap hosts atnatural densities, place them in natural settings, and verify that their developmentrate during exposure is typical, to ensure that they are no more or less susceptibleto parasitism than wild hosts. Trap hosts have been used to monitorrecruitment of parasitized hosts for imported cabbageworm (Van Driesche1988), and for larvae (Gould et al. 1992a) and pupae (Gould et al. 1992b) ofgypsy moth.The success of recruitment analysis is dependent upon the biology of thespecies involved and the ease with which methods can be devised to effectivelymeasure host and parasitoid recruitment. If more than one parasitoid species isto be monitored, different techniques may need to be employed for each,and this can result in a very demanding and time-consuming sampling plan.Nonetheless, when possible this approach provides an unbiased estimate ofstage-specific parasitism.Time-specific rate of parasitismAlthough time-specific rates of parasitism are an effective way to monitorchanges in parasitism over a season for hosts with overlapping generations,they are also frequently used to estimate the impact of parasitism at specificpoints in time for hosts with discrete generations. The most important biases insampling for time-specific rates of parasitism in host populations with overlappinggenerations are the presence of non-inclusive host stages in the sample,and the duration of time that parasitized individuals spend in the inclusivestages relative to healthy individuals. The presence of non-inclusive host stagesin a sample leads to underestimation, and prolonged availability of parasitizedhosts leads to overestimation of the rate of parasitism.Knowledge of the inclusive stages of the host in relation to specific parasitoidscan be used to ensure that only those stages that support the parasitoid during

PARASITOIDS AND PREDATORS 263its development are sampled for dissection or rearing in the laboratory. Bylimiting sampling to the inclusive host stages, estimates of rate of parasitism canbe greatly improved, but they are still subject to bias caused by differential mortalityduring development and the intervention of other contemporaneousmortality factors. One approach to avoid these additional sources of bias is recruitmentanalysis (see above). Although recruitment analysis has been usedsuccessfully to monitor parasitism of cabbage aphid populations (Lopez & VanDriesche 1989), it has not been used extensively. Other approaches used forstage-specific parasitism are not appropriate for hosts with overlapping generationsdue to the disruptive influence of variable rates of recruitment to the susceptiblestage and of loss from the inclusive stage for both host and parasitoidpopulations.A simple correction factor can be used to improve estimates of rate of parasitismwhen parasitized and healthy hosts spend different periods of time in theinclusive host stages. Parasitized hosts often develop more slowly than healthyhosts, and as a result become over-represented in samples. Russell (1987) indicatesthat this can be corrected by elevating the number of healthy hosts in thecalculation of percent parasitism by a factor determined from the ratio of the periodof time spent in the inclusive stages by parasitized (P) and healthy (H) hosts:No. of parasitized hosts * 100% parasitism = [ No. of healthy hosts * P/H ]+ No. of parasitized hosts(11.1)This correction is most effective when sampling intervals are short relative tothe development periods of the host and parasitoid, but can still provide a usefulimprovement in the estimation of rate of parasitism when sample intervals arelonger.Rates of parasitism over specific intervals of time can also be used to estimatethe influence of various environmental factors on parasitism of hosts witheither discrete or overlapping generations. In this case, parasitism is best estimatedthrough use of trap hosts exposed to parasitism for short intervals oftime under different environmental conditions. This approach has frequentlybeen used to monitor the effectiveness of mass-released parasitoids in inundativebiological control programs (e.g. Petersen et al. 1995, Wang et al. 1999),and to assess the influence of landscape diversity on parasitism (e.g. Cappuccinoet al. 1998, Menalled et al. 1999).General sampling concernsFor all these approaches, samples of hosts must be large enough to provide reasonableaccuracy in the estimation of a percentage, must be unbiased in theirrepresentation of parasitized and healthy hosts, and must be free from crosscontaminationafter collection. One important source of bias in sampling hostpopulations in the field is the differential spatial distribution of parasitized and

264 CHAPTER 11healthy individuals. Heterogeneous parasitism can result from random oraggregated attack (Olson et al. 2000), greater parasitoid activity at the edge of ahost population (Lopez et al. 1990, Brodmann et al. 1997), or parasitized hostsmoving away from the main host population (Ryan 1985, Lopez et al. 1990).The collection of hosts from a greater number of quadrats within the samplesite, the use of transects to collect samples, and knowledge of the influenceof parasitism on host behavior can all help to minimize bias from heterogeneity.In addition, when hosts are reared to await emergence of parasitoids, considerablecare must be taken to prevent cross-contamination among hoststhrough parasitism by emerging parasitoids. Fortunately, this can readily beavoided by separating hosts into individual containers as they are collected fromthe field.Once a sample of hosts has been collected from the field they are either rearedto estimate apparent parasitism, the observed percentage of hosts killed by aparasitoid, or they are dissected to estimate the marginal attack rate, the percentageof hosts attacked by a parasitoid. This is an important distinction, asthe two measurements can provide very different estimates of parasitismfrom the same sample if there is differential mortality of parasitized and healthyhosts or interference between contemporaneous mortality factors (Waage& Mills 1992, Royama 2001). Royama (1981, 2001) and Elkinton et al.(1992) provide detailed discussions of how to separate the marginal death rates,based on rearing data, of contemporaneous mortality factors. Although dissectionof hosts, if based on use of short marker stages for parasitism, provides amore direct measure of the marginal death rate, this approach is dependentupon accurate detection of parasitoid eggs or young larvae, which can often beconcealed within host tissues. Recent studies using DNA markers, however, indicatethat species-specific parasitism can be detected 24 hours after parasitoidoviposition, offering new opportunities in the accuracy of estimating marginaldeath rates (Tilmon et al. 2000, Zhu et al. 2000, Ratcliffe et al. 2002, Zhu &Williams 2002).PredationPredation is rather more difficult to measure than parasitism for two reasons:the interaction between a predator and its prey is very brief, a matter of minutesin most cases, which severely reduces the chances of detection by an observer;and very often there are few, if any, remains of the prey that can be detectedafter predation has taken place. As a result there are two basic approaches thatcan be used to estimate predation under natural conditions. In the few caseswhere remains of prey are detectable after a predation event, percent predationprovides a direct estimate of the impact of a predator population. In most cases,however, the impact of predation must be estimated indirectly as a predationrate based on a combination of predator abundance and per capita rate ofconsumption.

PARASITOIDS AND PREDATORS 265Percent predationEggs, pupae, and the coverings of Homoptera often remain intact for a sufficientlength of time after predation has occurred to be used to directly quantify percentpredation from a single sample, as was the case for percent parasitism. Persistentstructural components of the prey can also provide distinctive evidenceof the type of predator: for example, a pair of small holes result from predationby Neuroptera, a single hole from Heteroptera, and a peppering of small holesfrom ants. Andow (1990) differentiated the attack of several different insectpredators on eggs of the European corn borer, and examples of successful measurementof predation from prey remains include whitefly nymphs (Heinz &Parrella 1994), European corn borer eggs (Andow 1992), and gypsy mothpupae (Cook et al. 1994). In addition, sessile prey, such as eggs and pupae, canreadily be placed out in the field at natural densities to monitor losses from predation,but it is important that the sentinel prey are no more or less susceptibleto predation than the wild population. For example, Andow (1992) exposedegg masses of the European corn borer to monitor predation in corn fields underdifferent tillage systems, and Cook et al. (1994, 1995) exposed freeze-driedpupae of gypsy moth in the field to monitor predation by small mammals.Mobile stages of prey, such as larvae or adults, can be exposed to predation forlimited time periods by tethering with strong sewing thread. Tethering obviouslyaffects the behavior of the prey, however, and thus predation estimatedin this way must carefully correct for the greater susceptibility of the prey (seeWeseloh [1990] for an example).Predation rateIn contrast to percent predation, predation rate (N a) is a measure of the biomass(or number) of prey eaten per unit area per unit time. Estimating predation raterequires sampling for predator abundance and for per capita rate of consumptionby the predator population. The predation rate is then given by:nÂN = PR ( p D )a i ii = 1ii(11.2)where P iis the density of predators of stage i per unit area, R iis the per capita rateof consumption of prey by an individual predator of stage i per unit time, p iis theproportion of predators of stage i that test positive for gut detection of prey, D iis the detection interval for prey in the gut of a predator of stage i, and n isthe number of predator stages. The bracketed term is included only when gutdetection is the approach used to estimate predation events (see below).Predator density (P)The abundance of predators is most commonly monitored through use ofsweep-netting, vacuum sampling, or visual counts of individuals either directly

266 CHAPTER 11in the field or from destructive plant samples brought into the laboratory forcloser examination. Predator individuals must be classified by stage of development(instar) due to change in predation potential with age, and samples mustbe taken at regular intervals to show changes in age structure of the predatorpopulation over the period of interest. In addition, a sampling plan must be devisedto optimize the accuracy of predator density estimation, based on thenumber, size, and cost of samples to be taken (Chapter 1). In general, the accuracyof estimates is enhanced by use of a greater number of smaller-sized samplesthan fewer large-sized samples, as found by Ellington and Southward(1999) for monitoring the abundance of predators in cotton.Rate of consumption (R)There are three approaches to the estimation of the per capita rate of consumptionof predators in each of their successive stages of development: laboratoryestimation, direct observation, and gut detection (Table 11.2).Laboratory estimation is based on monitoring the number of prey consumedper unit time by individual predators in artificial arenas held at constant temperature(representing the average temperatures experienced under naturalconditions) with excess prey. These estimates represent the stage-specific maximumper capita feeding rates of the predator under ideal conditions, withoutthe need to search for and subdue prey and where metabolic costs are mini-Table 11.2 A summary of the methodological approaches for estimating the rate ofconsumption by insect predators.Methodology Application ExamplesLaboratory estimation Used to estimate maximum stage- Tamaki et al. (1974), Ro &specific per capita feeding rates under Long (1998)ideal conditionsDirect observation Uses multiple point observations of Edgar (1970), van denstage-specific per capita feeding Berg et al. (1997)activity under field conditionsGut detection: Effective if heavily chitinized parts of Sunderland (1975), Breenedissection the prey are consumed et al. (1990)Gut detection: Esterase bands; easy to develop, low Jones & Morse (1995),electrophoresis specificity, moderate retention time Solomon et al. (1996)Gut detection: Monoclonal antibodies; difficult to Greenstone (1996), Haglerimmunology develop, high specificity, moderate (1998)retention timeGut detection: PCR PCR products; easy to develop, high Chen et al. (2000),specificity, short retention time Greenstone & Shufran(2003)

PARASITOIDS AND PREDATORS 267mized. This approach was developed by Bombosch (1963) and Tamaki et al.(1974) to estimate the impact of predation on aphid populations, and has subsequentlybeen used by Chambers and Adams (1986) and Ro and Long (1998).Direct observation is not only the best way to assess the range of predators attackinga particular prey population and the diet breadth of individual predatorspecies, but can also provide stage-specific estimates of rate of consumptionunder natural conditions. Edgar (1970) was among the first to use this approachto estimate predation rates of lycosid spiders from observations of the number ofhours a day that the predators are active (t a), the mean proportion of time thatan average predator spends actively feeding (f) and the mean time in hours thata predator takes to fully consume a single prey item (t c). The rate of consumption(R), or the number of prey consumed by an individual predator per day is givenby:R = t f tac(11.3)The mean proportion of time spent feeding by an average predator (f ) is estimatedfrom a series of point samples of predator activity (proportion observedfeeding) taken at intervals during the active hours of a day and throughout theperiod of interest. As the proportion of time spent feeding will vary with preydensity, point samples of predator activity should be repeated across a series ofplots representing a gradient of prey abundance. The predator activity period(t a) is estimated from direct field observation and predator consumption time(t c) can be estimated either directly in the field or from laboratory observation.The direct observation approach has subsequently been used to estimate spiderpredation in a variety of crops (Kiritani et al. 1972, Sunderland et al. 1986,Nyffeler et al. 1987) and a variant of this technique has been used to quantifyaphid predation (van den Berg et al. 1997). It can readily be applied to manytypes of predators that can be observed without disturbance under naturalconditions, and has even been used to estimate rates of parasitism of barkbeetles (Mills 1991).Gut detection requires positive identification of prey remains in the gut of fieldcollectedpredators. In early studies, detection of prey remains relied on dissectionand visual inspection (e.g. Sunderland 1975, Hildrew & Townsend 1982,Breene et al. 1990). Subsequently, electrophoretic analysis (Powell et al. 1996,Solomon et al. 1996) was used, allowing predators with extra-oral digestion aswell as chewing predators to be studied, and most recently immunological(Greenstone 1996, Hagler 1998) and PCR analysis (Chen et al. 2000,Greenstone & Shufran 2003) have been developed for prey detection. An importantcaveat associated with gut detection of prey is that a positive for detectioncan result from saprophagy, necrophagy, and higher-order predation aswell as from primary predation, and thus prior knowledge of the feeding habitsof the predator is essential. Of the newer techniques available, electrophoreticanalysis is the least sensitive, as the same esterase bands may be shared by a

268 CHAPTER 11specific prey species, the alternative foods of the predator, the prey’s food, andeven the gut wall of the predator. Recent examples, highlighting some of thesedifficulties, can be found in Giller (1986), Lister et al. (1987) and Jonesand Morse (1995). Immunological analysis is based on the development ofmonoclonal antibodies for the prey of interest and their detection in the gut ofa predator through one of several gut content immunoassays (see Hagler 1998 forspecific details of techniques). Monoclonal antibodies are very specific, eitherfor prey species or stages, and are considered better for gut detection than polyclonalantibodies. They have distinct disadvantages, however, in that they aredifficult and costly to produce (hence available for just a handful of prey species;Greenstone 1996). Most recently PCR analysis has shown promise for thedetection of prey-specific DNA fragments in the guts of insect predators (Zaidiet al. 1999, Chen et al. 2000, Hoogendorn & Heimpel 2001) and spiders (Agustiet al. 2003, Greenstone & Shufran 2003). The shorter development time,lower cost, and greater certainty of analysis is likely to make PCR analysis of gutdetection more widely applicable than immunological analysis in the future.Whichever method is used for prey detection in the guts of the predators, theresults provide an estimate of the proportion of predators (p) that show positivefor feeding on the prey species of interest. It must be remembered, however,that these detection events are qualitative rather than quantitative as a predatorwill show positive whether it has eaten one or several prey items. Thus to estimatepredation rate, firstly the detection interval (D) must be experimentallydetermined to allow the observed proportion of predators with positive responsesto be corrected for loss of detectable prey remains, and secondly the rateof consumption (R) must be experimentally determined. The latter has eitherbeen assumed to be equivalent to just a single prey item (Dempster 1960), or hasbeen estimated using the laboratory estimation approach discussed above. Recently,however, Naranjo and Hagler (2001) have recommended incorporatinga functional response experiment with gut detection to estimate the relationshipbetween R and prey density and thus to improve the estimation of predationrate.Diversity of entomophagous speciesWith the increasing need for monitoring ecosystem health and biodiversity ingeneral, predators and occasionally parasitoids have been used as indicatorgroups. The sensitivity of spider assemblages to environmental conditions hasled to their use as an indicator group in the monitoring of restoration landscapes.For example, spiders provide indicator species for restoration activitiesin redwood forests (Willett 2001), and have proved valuable in monitoring thereclamation of limestone quarries (Wheater et al. 2000). Similarly, carabidbeetle assemblages have been used as indicator species to assess managementpractices in cereal crops and grasslands (Luff 1996), landscape changes due to

PARASITOIDS AND PREDATORS 269human activity (Niemelä et al. 2000), and more generally for habitat disturbance(Rainio & Niemelä 2003). Ant assemblages have also been shown to bevaluable bioindicators of biodiversity responses to pollution (Madden & Fox1999, Andersen et al. 2002).Parasitoid assemblages have not been used as general indicators of ecosystemhealth, but braconid parasitoids have been suggested to be a valuable indicatortaxon for disturbance in forest stands (Lewis & Whitfield 1999). In addition,parasitoids have proved to be a useful group for investigating latitudinal gradientsin species diversity (Janzen 1981, Hawkins 1994, Sime & Brower 1998)and for estimation of global biodiversity (Bartlett et al. 1999, Dolphin & Quicke2001).Monitoring the species richness of entomophagous guilds in terrestrial communitieshas made use of a variety of sampling techniques. The type of habitat isprobably the most important determinant of the sampling technique to employ.For forest trees, insecticide fogging is clearly the only way to adequately sampletree canopies (Ozanne et al. 2000, Majer et al. 2001; Chapter 7). In contrast, forepigeal predators, pitfall trapping has been shown to be effective for carabids(Larsen & Williams 1999; Chapter 3), spiders (Brennan et al. 1999), and ants(Miller & New 1997). The type of habitat can also influence the efficacy of differentsampling methods for the same taxon of predators. For example, visualsearching proved to be the most efficient way to sample spiders in citrus orchards(Amalin et al. 2001), whereas pitfall traps were more effective than visualsearch for spiders in a heathland landscape (Churchill & Arthur 1999).In comparing five different sampling methods to study the diversity of parasitoidsin the forests of Sulawesi, Noyes (1989) found insecticide fogging to bethe most effective, followed by sweep-netting, Malaise traps, yellow water trapsand lastly intercept traps. Malaise traps (Chapter 4) have been used extensivelyin the sampling of parasitoid communities for biodiversity studies. Townes(1972) describes the design of a Malaise trap suitable for sampling parasitoids,which has been used effectively to monitor the larger Ichneumonidae (Gaston& Gauld 1993, Gaasch et al. 1998) and Braconidae (Lewis & Whitfield 1999), aswell as smaller taxa such as Mymaridae (Noyes 1989). Malaise traps have alsobeen widely used to monitor the diversity of syrphid predators in a variety ofagricultural landscapes (Hondelmann 1998, Salveter 1998). The advantage ofMalaise traps is that they can be left in situ for longer periods of time, as they cancollect directly into preserving materials. However, the placement of the trapscan have an important influence on the number of insects trapped (Chapter 7).Yellow pan traps (Chapter 6) have also proved valuable for sampling the diversityof parasitoids in orchard (Purcell & Messing 1996), forest (Villemant &Andrei-Ruiz 1999), and fen (Finnamore 1994) landscapes. In addition, yellowsticky traps (Chapters 5 & 6) have been used to monitor the diversity of parasitoidson Bahamian islands (Schoener et al. 1995) and of coccinellids in a seriesof agricultural landscapes (Colunga-Garcia et al. 1997). Other sampling methodsused to investigate the diversity of parasitoid and predator communities

270 CHAPTER 11include sweep-net sampling (Chapter 4) in old field successions (Siemann et al.1999) and suction sampling (Chapters 4 & 6) in grasslands (Harper et al. 2000).Noyes (1982) describes the construction of a sweep net that has been found tobe particularly suitable for sampling smaller insect parasitoids, and Noyes(2003) provides a detailed account of the full range of sampling methods usedfor collecting smaller parasitoids, particularly the Chalcidoidea.Duelli et al. (1999) and Duelli and Obrist (2003) provide useful insightsfor ways in which sampling for biodiversity of entomophagous species can beoptimized in cultivated landscapes, and Kitching et al. (2001) discuss theneed to use packages of different sampling methods to be effective in the assessmentof arthropod biodiversity. One final consideration when sampling entomophagouscommunities is that many species of parasitoids and predators arerather less common than their hosts or prey. This suggests a need for greatersampling effort when planning the number of sites, number of samples, and sizeof each sample for an inventory survey.ReferencesAgusti, N., Shayler, S.P., Harwood, J.D., Vaughan, I.P., Sunderland, K.D., & Symondson,W.O.C. (2003) Collembola as alternative prey sustaining spiders in arable ecosystems: preydetection within predators using molecular markers. Molecular Ecology, 12, 3467–3475.Akey, D.H. & Burns, D.W. (1991) Analytical consideration and methodologies for elementaldeterminations in biological samples. Southwestern Entomologist, 14 (Suppl.), 25–36.Amalin, D.M., Pena, J.E., McSorley, R., Browning, H.W., & Crane, J.H. (2001) Comparison ofdifferent sampling methods and effect of pesticide application on spider populations in limeorchards in south Florida. Environmental Entomology, 30, 1021–1027.Andersen, A.N., Hoffmann, B.D. Muller, W.J., & Griffiths A.D. (2002) Using ants as bioindicatorsin land management: simplifying assessment of ant community responses. Journal ofApplied Ecology 39, 8–17.Andow, D.A. (1990) Characterization of predation on egg masses of Ostrinia nubilalis(Lepidoptera: Pyralidae). Annals of the Entomological Society of America, 83, 482–486.Andow, D.A. (1992) Fate of eggs of first generation Ostrinia nubilalis (Lepidoptera: Pyralidae) inthree conservation tillage systems. Environmental Entomology, 21, 388–393.Askew, R.R. & Shaw, M.R. (1986) Parasitoid communities: their size, structure and development.In Insect Parasitoids (ed. J. Waage & D. Greathead), pp. 225–264. Academic Press,London.Bartlett, R., Pickering, J., Gauld, I., & Windsor, D. (1999) Estimating global biodiversity: tropicalbeetles and wasps send different signals. Ecological Entomology, 24, 118–121.Bellows, T.S., Van Driesche, R.G., & Elkinton, J.S. (1992) Life-table construction and analysisin the evaluation of natural enemies. Annual Review of Entomology, 37, 587–614.Blumberg, D. & Van Driesche, R.G. (2001) Encapsulation rates of three encyrtid parasitoids bythree mealybug species (Homoptera: Pseudococcidae) found commonly as pests in commercialgreenhouses. Biological Control, 22, 191–199.Bombosch, S. (1963) Untersuchungen zur vermehrung von Aphis fabae Scop. in Samenrübenbestandenunter besonderer Berücksichtigung der Schwebfliegen (Diptera:Syrphidae). Zeitschrift für Angewandte Entomologie, 52, 105–141.

PARASITOIDS AND PREDATORS 271Bowie, M.H., Gurr, G.M., Hossain, Z., Baggen, L.R., & Frampton, C.M. (1999) Effects ofdistance from field edge on aphidophagous insects in a wheat crop and observations on trapdesign and placement. International Journal of Pest Management, 45, 69–73.Breene, R.G., Sweet, M.H., & Olson, J.K. (1990) Analysis of the gut contents of naiads of Enallagmacivile (Odonata: Coenagrionidae) from a Texas pond. Journal of the American MosquitoControl Association, 6, 547–548.Brennan, K.E.C., Majer, J.D., & Reygaert, N. (1999) Determination of an optimal pitfall trapsize for sampling spiders in a Western Australian Jarrah forest. Journal of Insect Conservation,3, 297–307.Brodmann, P.A., Wilcox, C.V., & Harrison, S. (1997) Mobile parasitoids may restrict the spatialspread of an insect outbreak. Journal of Animal Ecology, 66, 65–72.Brown, M.W. & Schmitt, J.J. (2001) Seasonal and diurnal dynamics of beneficial insect populationsin apple orchards under different management intensity. Environmental Entomology,30, 415–424.Cappuccino, N., Lavertu, D., Bergeron, Y., & Regniere, J. (1998) Spruce budworm impact,abundance and parasitism rate in a patchy landscape. Oecologia, 114, 236–242.Chambers, R.J. & Adams, T.H.L. (1986) Quantification of the impact of hoverflies (Diptera:Syrphidae) on cereal aphids in winter wheat: an analysis of field populations. Journal ofApplied Ecology, 23, 895–904.Chen, Y., Giles, K.L., Payton, M.E., & Greenstone, M.H. (2000) Identifying key cereal aphidpredators by molecular gut analysis. Molecular Ecology, 9, 1887–1898.Churchill, T.B. & Arthur, J.M. (1999) Measuring spider richness: effects of different samplingmethods and spatial and temporal scales. Journal of Insect Conservation, 3, 287–295.Colunga-Garcia, M., Gage, S.H., & Landis, D. (1997) Response of an assemblage of Coccinellidae(Coleoptera) to a diverse agricultural landscape. Environmental Entomology, 26, 797–804.Colwell, R.K. (1997) EstimateS: Statistical estimation of species richness and shared speciesfrom samples. Version 5. User’s guide and application published at http://viceroy.eeb.uconn.edu/estimates.Cook, S.P., Hain, F.P., & Smith, H.R. (1994) Oviposition and pupal survival of gypsy moth(Lepidoptera: Lymantriidae) in Virginia and North Carolina pine–hardwood forests. EnvironmentalEntomology, 23, 360–366.Cook, S.P., Smith, H.R., Hain, F.P., & Hastings, F.L. (1995) Predation of gypsy moth(Lepidoptera: Lymantriidae) pupae by invertebrates at low small mammal populationdensities. Environmental Entomology, 24, 1234–1238.Costello, M.J. & Daane, K.M. (1997) Comparison of sampling methods used to estimate spider(Araneae) species abundance and composition in grape vineyards. Environmental Entomology,26, 142–149.Dawah, H.A., Hawkins, B.A., & Claridge, M.F. (1995) Structure of the parasitoid communitiesof grass-feeding chalcid wasps. Journal of Animal Ecology, 64, 708–720.Dempster, J.P. (1960) A quantitative study of the predators of the eggs and larvae of the broombeetle, Phytodecta olivacea Forster, using the precipitin test. Journal of Animal Ecology, 29,149–167.Diaz-Aranda, L.M. & Monserrat, V.J. (1995) Aphidophagous predator diagnosis: Key to generaof European chrysopid larvae (Neur.: Chrysopidae). Entomophaga, 40, 169–181.Doane, C.C. (1971) A high rate of parasitization by Brachymeria intermedia (Hymenoptera:Chalcididae) on the gypsy moth. Annals of the Entomological Society of America, 64, 753–754.Dolphin, K. & Quicke, D.L.J. (2001) Estimating the global species richness of an incompletelydescribed taxon: an example using parasitoid wasps (Hymenoptera: Braconidae). BiologicalJournal of the Linnean Society, 73, 279–286.

272 CHAPTER 11Duelli, P. & Obrist, M.K. (2003) Biodiversity indicators: the choice of values and measures.Agriculture Ecosystems & Environment, 98, 87–98.Duelli, P., Obrist, M.K., & Schmatz, D.R. (1999) Biodiversity evaluation in agricultural landscapes:above-ground insects. Agriculture Ecosystems & Environment, 74, 33–64.Edgar, W.D. (1970) Prey and feeding behaviour of adult females of the wolf spider Pardosaamentata (Clerk). Netherlands Journal of Zoology, 20, 487–491.Elkinton, J.S., Buonaccorsi, J.P., Bellows, T.S., & Van Driesche, R.G. (1992) Marginal attackrate, k-values and density dependence in the analysis of contemporaneous mortality factors.Researches on Population Ecology, 34, 29–44.Ellington, J. & Southward, M. (1999) Quadrat sample precision and cost with a highvacuuminsect sampling machine in cotton ecosystems. Environmental Entomology, 28,722–728.Elliott, N.C. & Kieckhefer, R.W. (1990) Dynamics of aphidophagous coccinellid assemblages insmall grain fields in eastern South Dakota. Environmental Entomology, 19, 1320–1329.Finnamore, A.T. (1994) Hymenoptera of the Wagner Natural Area, a boreal spring fen in centralAlberta. Memoirs of the Entomological Society of Canada, 169, 181–220.Floate, K., Khan, B., & Gibson, G. (1999) Hymenopterous parasitoids of filth fly (Diptera:Muscidae) pupae in cattle feedlots. Canadian Entomologist, 131, 347–362.Gaasch, C.M., Pickering, J., & Moore, C.T. (1998) Flight phenology of parasitic wasps (Hymenoptera:Ichneumonidae) in Georgia’s Piedmont. Environmental Entomology, 27,606–614.Gaston, K.J. & Gauld, I.D. (1993) How many species of pimplines (Hymenoptera: Ichneumonidae)are there in Costa Rica? Journal of Tropical Ecology, 9, 491–499.Giller, P.S. (1986) The natural diet of the Notonectidae: field trials with electrophoresis. EcologicalEntomology, 11, 163–172.Gordon, R.D. & Vandenberg, N. (1995) Larval systematics of North American Coccinella L.(Coleoptera: Coccinellidae). Entomologica Scandinavica, 26, 67–86.Gould, J.R., Van Driesche, R.G., Elkinton, J.S., & Odell, T.M. (1989) A review of techniques formeasuring the impact of parasitoids of lymantriids. In The Lymantriidae: a Comparison of Featuresof New and Old World Tussock Moths (ed. W.E. Wallner), pp. 517–531. USDA Forest ServiceGeneral Technical Report NE-123.Gould, J.R., Elkinton, J.S., & Wallner, W.E. (1990) Density-dependent suppression of experimentallycreated gypsy moth, Lymantria dispar (Lepidoptera: Lymantriidae), populations bynatural enemies. Journal of Animal Ecology, 59, 213–234.Gould, J.R., Elkinton, J.S., & Van Driesche, R.G. (1992a) Suitability of approaches for measuringparasitoid impact on Lymantria dispar (Lepidoptera: Lymantriidae) populations. EnvironmentalEntomology, 21, 1035–1045.Gould, J.R., Elkinton, J.S., & Van Driesche, R.G. (1992b) Assessment of potential methods ofmeasuring parasitism by Brachymeria intermedia (Nees) (Hymenoptera: Chalcididae) ofpupae of the gypsy moth. Environmental Entomology, 21, 394–400.Greenstone, M.H. (1996) Serological analysis of arthropod predation: past, present and future.In The Ecology of Agricultural Pests: Biochemical Approaches (ed. W.O.C. Symondson & J.E.Liddell), pp. 265–300. Chapman & Hall, London.Greenstone, M.H. & Shufran, K.A. (2003) Spider predation: species-specific identification ofgut contents by polymerase chain reaction. Journal of Arachnology, 31, 131–134.Gutierrez, A.P. (1996) Applied Population Ecology. Wiley, New York.Hagler, J.R. (1998) Variation in the efficacy of several predator gut content immunoassays.Biological Control, 12, 25–32.Hagler, J.R. & Durand, C.M. (1994) A new method for immunologically marking prey and itsuse in predation studies. Entomophaga, 39, 257–265.

PARASITOIDS AND PREDATORS 273Harper, M.G., Dietrich, C.H., Larimore, R.L., & Tessene, P.A. (2000) Effects of prescribed fire onprairie arthropods: an enclosure study. Natural Areas Journal, 20, 325–335.Hawkins, B A. (1988) Species diversity in the third and fourth trophic levels patterns andmechanisms. Journal of Animal Ecology, 57, 137–162.Hawkins, B.A. (1994) Pattern and Process in Host–Parasitoid Interactions. Cambridge UniversityPress, Cambridge.Heinz, K.M. & Parrella, M.P. (1994) Biological control of Bemisia argentifolii (Homoptera:Aleyrodidae) infesting Euphorbia pulcherrima: evaluations of releases of Encarsia luteola(Hymenoptera: Aphelinidae) and Delphastus pusillus (Coleoptera: Coccinellidae). EnvironmentalEntomology, 23, 1346–1353.Herting, B. (1975) A Catalogue of Parasites and Predators of Terrestrial Arthropods. Volume VI, Part 1.Commonwealth Agricultural Bureaux, Farnham Royal.Hildrew, A.G. & Townsend, C.R. (1982) Predators and prey in a patchy environment: a freshwaterstudy. Journal of Animal Ecology, 51, 797–815.Hill, M.G. (1988) Analysis of the biological control of Mythimna separata (Lepidoptera:Noctuidae) by Apanteles ruficrus (Braconidae: Hymenoptera) in New Zealand. Journal ofApplied Ecology, 25, 197–208.Hodek, I. (1973) Biology of Coccinellidae. Junk, The Hague.Hodek, I. (1993) Prey and habitat specificity in aphidophagous predators (a review). BiocontrolScience and Technology, 3, 91–100.Hondelmann, P. (1998) Hoverflies (Diptera: Syrphidae) of agrarian ecosystems: study of theLoess Plains in southern Lower Saxony. Drosera, 98, 113–122.Hoogendoorn, M. & Heimpel, G.E. (2001) PCR-based gut content analysis of insect predators:using ribosomal ITS-1 fragments from prey to estimate predation frequency. MolecularEcology, 10, 2059–2067.Janzen, D.H. (1981) The peak in North American ichneumonid species richness lies between38° and 42°N. Ecology, 62, 532–537.Johnson, P.C. & Reeves, R.M. (1995) Incorporation of the biological marker rubidium in gypsymoth (Lepidoptera: Lymantriidae) and its transfer to the predator Carabus nemoralis(Coleoptera: Carabidae). Environmental Entomology, 24, 46–51.Jones, S.A. & Morse, J.G. (1995) Use of isoelectric focusing electrophoresis to evaluate citrusthrips (Thysanoptera: Thripidae) predation by Euseius tularensis (Acari: Phytoseiidae). EnvironmentalEntomology, 24, 1040–1051.Kean, J.M. & Barlow, N.D. (2001) A spatial model for the successful biological control of Sitonadiscoideus by Microctonus aethiopoides. Journal of Applied Ecology, 38,162–169.Keating, K.A. & Quinn, J.F. (1998) Estimating species richness: the Michaelis–Menten modelrevisted. Oikos, 81, 411–416.Kiritani, K., Kawahara, S., Sasaba, T., & Nakasuji, F. (1972) Quantitative evaluation of predationby spiders on the green rice leafhopper, Nephotettix cincticeps, by a sight-count method.Researches on Population Ecology, 13, 187–200.Kitching, R.L., Li, D., & Stork, N.E. (2001) Assessing biodiversity “sampling packages”: howsimilar are arthropod assemblages in different tropical rainforests? Biodiversity and Conservation,10, 793–813.Kitt, J.T. & Keller, M.A. (1998) Host selection by Aphidius rosae Haliday (Hym., Braconidae)with respect to assessment of host specificity in biological control. Journal of Applied Entomology,122, 57–63.Larsen, K.J. & Williams, J.B. (1999) Influence of fire and trapping effort on ground beetles in areconstructed tallgrass prairie. Prairie Naturalist, 31, 75–86.Lewis, C.N. & Whitfield, J.B. (1999) Braconid wasp (Hymenoptera: Braconidae) diversity inforest plots under different silvicultural methods. Environmental Entomology, 28, 986–997.

274 CHAPTER 11Lister, A., Usher, M.B., & Block, W. (1987) Description and quantification of field attack ratesby predatory mites: an example using an electrophoresis method with a species of Antarcticmite. Oecologia, 72, 185–191.Lopez, E.R. & Van Driesche, R.G. (1989) Direct measurement of host and parasitoid recruitmentfor assessment of total losses due to parasitism in a continuously breeding species, thecabbage aphid, Brevicoryne brassicae (L.) (Hemiptera: Aphididae). Bulletin of EntomologicalResearch, 79, 47–60.Lopez, E.R., Van Driesche, R.G., & Elkinton, J.S. (1990) Rates of parasitism by Diaeretiella rapae(Hymenoptera: Braconidae) for cabbage aphids (Homoptera: Aphididae) in and outside ofcolonies: why do they differ? Journal of the Kansas Entomological Society, 63, 158–165.Luck, R.F., Shepard, B.M., & Kenmore, P.E. (1988) Experimental methods for evaluatingarthropod natural enemies. Annual Review of Entomology, 33, 367–391.Luff, M.L. (1996) Use of carabids as environmental indicators in grasslands and cereals. AnnalesZoologici Fennici, 33, 185–195.Madden, K.E. & Fox, B.J. (1999) Arthropods as indicators of the effects of fluoride pollution onthe succession following sand mining. Journal of Applied Ecology, 34, 1239–1256.Majer, J.D., Kitching, R.L., Heterick, B.E., Hurley, K., & Brennan, K.E.C. (2001) North–southpatterns within arboreal ant assemblages from rain forests in Eastern Australia. Biotropica,33, 643–661.Menalled, F.D., Marino, P.C., Gage, S.H., & Landis, D.A. (1999) Does agricultural landscapestructure affect parasitism and parasitoid diversity? Ecological Applications, 9, 634–641.Miller, L.J. & New, T.R. (1997) Mount Piper grasslands: pitfall trapping of ants and interpretationof habitat variability. Memoirs of the Museum of Victoria, 56, 377–381.Mills, N.J. (1991) Searching strategies and attack rates of parasitoids of the ash bark beetle(Leperisinus varius) and its relevance to biological control. Ecological Entomology, 16, 461–470.Mills, N.J. (1994) Parasitoid guilds: defining the structure of the parasitoid communities ofendopterygote insect hosts. Environmental Entomology, 23, 1066–1083.Mills, N.J. (1997) Techniques to evaluate the efficacy of natural enemies. In Methods in Ecologicaland Agricultural Entomology (ed. D.R. Dent & M.P. Walton), pp. 271–291. CAB International,Wallingford.Monserrat, V.J., Oswald, J.D., Tauber, C.A., & Diaz-Aranda, L.M. (2001) Recognition of larvalNeuroptera. In Lacewings in the Crop Environment (ed. P.K. McEwen, T.R. New, & A.E.Whittington), pp. 43–81. Cambridge University Press, Cambridge.Muller, C.B., Adriaanse, I.C.T., Belshaw, R., & Godfray, H.C.J. (1999) The structure of anaphid–parasitoid community. Journal of Animal Ecology, 68, 346–370.Nakamura, K. & Abbas, I. (1987) Preliminary life table of the spotted tortoise beetle, Aspidomorphamillaria (Coleoptera: Chrysomelidae) in Sumatra. Researches on Population Ecology,29, 229–236.Naranjo, S.E. & Hagler, J.R. (2001) Toward the quantification of predation with predator gutimmunoassays: a new approach integrating functional response behavior. Biological Control,20, 175–189.Niemelä, J., Kotze, J., Ashworth, A., et al. (2000) The search for common anthropogenicimpacts on biodiversity: a global network. Journal of Insect Conservation 4, 3–9.Noyes, J.S. (1982) Collecting and preserving chalcid wasps (Hymenoptera: Chalcidoidea).Journal of Natural History, 16, 315–334.Noyes, J.S. (1989) A study of five methods of sampling Hymenoptera (Insecta) in atropical rainforest, with special reference to the Parasitica. Journal of Natural History, 23,285–298.Noyes, J.S. (2003) Universal Chalcidoidea database: collecting and preserving chalcidoids.http://www.nhm.ac.uk/entomology/chalcidoids/collecting1.html [accessed May 6, 2004].

PARASITOIDS AND PREDATORS 275Nyffeler, M., Dean, D.A., & Sterling, W.L. (1987) Evaluation of the importance of the stripedlynx spider, Oxyopes salticus (Araneae: Oxyopidae), as a predator in Texas cotton. EnvironmentalEntomology, 16, 1114–1123.Olson, A.C., Ives, A.R., & Gross, K. (2000) Spatially aggregated parasitism on pea aphids,Acyrthosiphon pisum, caused by random foraging behavior of the parasitoid Aphidius ervi.Oikos, 91, 66–76.Ozanne, C.M.P., Speight, M.R., Hambler, C., & Evans, H.F. (2000) Isolated trees and forestpatches: patterns in canopy arthropod abundance and diversity in Pinus sylvestris (ScotsPine). Forest Ecology and Management, 137, 53–63.Parajulee, M.N. & Slosser, J.E. (1999) Evaluation of potential relay strip crops for predatorenhancement in Texas cotton. International Journal of Pest Management, 45, 275–286.Petersen, J.J., Watson, D.W., & Cawthra, J.K. (1995) Comparative effectiveness of three releaserates for a pteromalid parasitoid (Hymenoptera) of house flies (Diptera) in beef cattlefeedlots. Biological Control, 5, 561–565.Porter, E.E. & Hawkins, B.A. (1998) Patterns of diversity for aphidiine (Hymenoptera:Braconidae) parasitoid assemblages on aphids (Homoptera). Oecologia, 116, 234–242.Powell, W., Walton, M.P., & Jervis, M.A. (1996) Populations and communities. In InsectNatural Enemies. Practical Approaches to Their Study and Evaluation (ed. M. Jervis & N. Kidd),pp. 223–292. Chapman & Hall, London.Purcell, M.F. & Messing, R.H. (1996) Ripeness effects of three vegetable crops on abundance ofaugmentatively released Psyttalia fletcheri (Hym.: Braconidae): improved sampling and releasemethods. Entomophaga, 41, 105–116.Rainio, J. & Niemelä, J. (2003) Ground beetles (Coleoptera: Carabidae) as bioindicators. Biodiversityand Conservation 12, 487–506.Ratcliffe, S.T., Robertson, H.M., Jones, C.J., Bollero, G.A, & Weinzierl, R.A. (2002) Assessmentof parasitism of house fly and stable fly (Diptera: Muscidae) pupae by pteromalid(Hymenoptera: Pteromalidae) parasitoids using a polymerase chain reaction assay. Journalof Medical Entomology, 39, 52–60.Rhoades, M.H. (1996) Key to first and second instars of six species of Coccinellidae(Coleoptera) from alfalfa in southwest Virginia. Journal of the New York Entomological Society,104, 83–88.Ro, T.H. & Long, G.E. (1998) Population dynamics pattern of green peach aphid (Homoptera:Aphididae) and its predator complex in a potato system. Korean Journal of Biological Sciences,2, 217–222.Rotheray, G.E. (1988) Third stage larvae of six species of aphidophagous Syrphidae Diptera.Entomologist’s Gazette, 39, 153–159.Rotheray, G.E. (1993) Colour Guide to Hoverfly Larvae (Diptera, Syrphidae) in Britain and Europe. D.Whiteley, Sheffield.Rotheray, G.E. & Gilbert, F.S. (1989) The phylogeny and systematics of European predaciousSyrphidae (Diptera) based on larval and puparial stages. Zoological Journal of the LinneanSociety, 95, 29–70.Rotheray, G.E. & Gilbert, F.S. (1999) Phylogeny of Palaearctic Syrphidae (Diptera): evidencefrom larval stages. Zoological Journal of the Linnean Society, 127, 1–112.Royama, T. (1981) Evaluation of mortality factors in insect life table analysis. Ecological Monographs,5, 495–505.Royama, T. (2001) Measurement, analysis, and interpretation of mortality factors in insect survivorshipstudies, with reference to the spruce budworm, Choristoneura fumiferana (Clem.)(Lepidoptera: Tortricidae). Population Ecology, 43, 157–178.Russell, D.A. (1987) A simple method for improving estimates of percentage parasitism by insectparasitoids from field sampling of hosts. New Zealand Entomologist, 10, 38–40.

276 CHAPTER 11Ryan, R.B. (1985) A hypothesis for decreasing parasitization of larch casebearer (Lepidoptera:Coleophoridae) on larch foliage by Agathis pumila. Canadian Entomologist, 117, 1573–1574.Salveter, R. (1998) Habitat use of adult syrphid flies (Diptera: Syrphidae) in a highly diversifiedagricultural landscape. Mitteilungen der Schweizerischen Entomologischen Gesellschaft, 71,49–71.Schoener, T.W., Spiller, D.A., & Morrison, L.W. (1995) Variation in the hymenopteran parasitoidfraction on Bahamian islands. Acta Oecologica, 16, 103–121.Shaw, M.R. (1994) Parasitoid host ranges. In Parasitoid Community Ecology (ed. B.A. Hawkins &W. Sheehan), pp. 111–144. Oxford University Press, Oxford.Shaw, M.R. (1997) Rearing parasitic Hymenoptera. The Amateur Entomologist, 25, 1–46.Siemann, E., Haarstad, J., & Tilman, D. (1999) Dynamics of plant and arthropod diversity duringold field succession. Ecography, 22, 406–414.Sime, K.R. & Brower, A.V.Z. (1998) Explaining the latitudinal gradient anomaly in ichneumonoidspecies richness: evidence from butterflies. Journal of Animal Ecology, 67, 387–399.Solomon, M.G., Fitzgerald, J.D., & Murray, R.A. (1996) Electrophoretic approaches topredator–prey interactions. In The Ecology of Agricultural Pests: Biochemical Approaches (ed.W.O.C. Symondson & J.E. Liddell), pp. 457–468. Chapman & Hall, London.Stadler, B. (2002) Determinants of the size of aphid–parasitoid assemblages. Journal of AppliedEntomology 126, 258–264.Stary, P. (1970) Biology of Aphid Parasites (Hymenoptera: Aphidiidae) with Respect to Integrated Control.Junk, The Hague.Stireman J.O. & Singer M.C. (2002) Spatial and temporal variation in the parasitoid assemblageof an exophytic polyphagous caterpillar. Ecological Entomology, 27, 588–600.Sunderland, K.D. (1975) The diet of some predatory arthropods in cereal crops. Journal ofApplied Ecology, 12, 507–515.Sunderland, K.D. (1988) Quantitative methods for detecting invertebrate predation occurringin the field. Annals of Applied Biology, 112, 201–224.Sunderland, K.D., Fraser, A.M., & Dixon, A.F.G. (1986) Field and laboratory studies onmoney spiders (Linyphiidae) as predators of cereal aphids. Journal of Applied Ecology, 23,433–447.Tamaki, G., McGuire, J.U., & Turner, J.E. (1974) Predator power and efficiency: a model toevaluate their impact. Environmental Entomology, 3, 625–630.Tauber, C.A., De Leon, T., Penny, N.D., & Tauber, M.J. (2000) The genus Ceraeochrysa(Neuroptera: Chrysopidae) of America north of Mexico: larvae, adults, and comparativebiology. Annals of the Entomological Society of America, 93, 1195–1221.Tilmon, K.J., Danforth, B.N., Day, W.H., & Hoffmann, M.P. (2000) Determining parasitoidspecies composition in a host population: a molecular approach. Annals of the EntomologicalSociety of America, 93, 640–647.Townes, H. (1972) A light-weight Malaise trap. Entomological News, 83, 239–247.van den Berg, H., Ankash, D., Muhammad, A., et al. (1997) Evaluating the role of predation inpopulation fluctuations of the soybean aphid Aphis glycines in farmers’ fields in Indonesia.Journal of Applied Ecology, 34, 971–984.Van Driesche, R. (1983) Meaning of “percent parasitism” in studies of insect parasitoids.Environmental Entomology, 12, 1611–1622.Van Driesche, R. (1988) Field measurement of population recruitment of Apanteles glomeratus(L.) (Hymenoptera: Braconidae), a parasitoid of Pieris rapae (L.) (Lepidoptera: Pieridae), andfactors influencing adult parasitoid foraging success in kale. Bulletin of Entomological Research,78, 199–208.Van Driesche, R. & Bellows, T. (1988) Use of host and parasitoid recruitment in quantifyinglosses from parasitism in insect populations. Ecological Entomology, 13, 215–222.

PARASITOIDS AND PREDATORS 277Van Driesche, R., Bellows, T.S., Elkinton, J.S., Gould, J.R., & Ferro, D.N. (1991) The meaningof percent parasitism revisited: solutions to the problem of accurately estimating total lossesfrom parasitism. Environmental Entomology, 20, 1–7.van Lenteren, J.C. & van Roermund, J.W. (1999) Why is the parasitoid Encarsia formosa sosuccessful in controlling whiteflies. In Theoretical Approaches to Biological Control (ed. B.A.Hawkins & H.V. Cornell), pp. 116–130. Cambridge University Press, Cambridge.Villemant, C. & Andrei-Ruiz, M.-C. (1999) Diversity and spatial distribution of parasitoidhymenoptera in the green oak forest of the Fango valley (Corsica). Annales de la SociétéEntomologique de France, 35 (Suppl.), 259–262.Waage, J.K. and Mills, N.J. (1992) Understanding and measuring the impact of natural enemieson pest populations. In Biological Control Manual. Volume I. Principles and Practice of BiologicalControl (ed. R.H. Markham, A. Wodageneh, & S. Agboola), pp. 84–114. InternationalInstitute of Tropical Agriculture, Cotonou, Benin.Wang, B., Ferro, D.N., & Hosmer, D.W. (1999) Effectiveness of Trichogramma ostriniae and T.nubilale for controlling the European corn borer Ostrinia nubilalis in sweet corn. EntomologiaExperimentalis et Applicata, 91, 297–303.Weseloh, R.M. (1990) Estimation of predation rates of gypsy moth larvae by exposure of tetheredcaterpillars. Environmental Entomology, 19, 448–455.Wheater, C.P., Cullen, W.R., & Bell, J.R. (2000) Spider communities as tools in monitoring reclaimedlimestone quarry landforms. Landscape Ecology, 15, 401–406.Willett, T.R. (2001) Spiders and other arthropods as indicators in old-growth versus logged redwoodstands. Restoration Ecology, 9, 410–420.Wyss, E. (1996) The effects of artificial weed strips on diversity and abundance of the arthropodfauna in a Swiss experimental apple orchard. Agriculture Ecosystems and Environment, 60,47–59.Yamamura, K. (1999) Key-factor/key-stage analysis for life table data. Ecology, 80, 533–537.Zaidi, R.H., Jaal, Z., Hawkes, N.J., Hemingway, J., & Symondson, W.O.C. (1999) Can multiplecopysequences of prey DNA be detected amongst the gut contents of invertebrate predators?Molecular Ecology, 8, 2081–2087.Zhu, Y.C., & Williams, L. (2002) Detecting the egg parasitoid Anaphes iole (Hymenoptera: Mymaridae)in tarnished plant bug (Heteroptera: Miridae) eggs by using a molecular approach.Annals of the Entomological Society of America, 95, 359–365.Zhu, Y.C., Burd, J.D., Elliott, N.C., & Greenstone, M.H. (2000) Specific ribosomal DNA markerfor early polymerase chain reaction detection of Aphelinus hordei (Hymenoptera: Aphelinidae)and Aphidius colemani (Hymenoptera: Aphidiidae) from Diuraphis noxia (Homoptera:Aphididae). Annals of the Entomological Society of America, 93, 486–491.Index of methods and approachesMethodology Topics addressed CommentsThe composition of an entomophagous assemblageSampling for Composition of the parasitoid Minimize contamination;parasitoids complex of a specific host sample all host stages andspecies.multiple sites.Sampling for Composition of the predator Direct observation has manypredators assemblage associated with a advantages over generalspecific prey species.sampling or gut detection.

278 CHAPTER 11Methodology Topics addressed CommentsThe impact of an entomophagous speciesPercent parasitism Stage-specific generational Devise a sampling plan thatparasitism of a host withprovides accurate input for thediscrete generations.ratio of hosts parasitized tosusceptible hosts.Time-specific rate of parasitism Easily biased by inclusion offor hosts with either discrete or instars in host samples that docontinuous generations.not support parasitism or by aninfluence of parasitism on hostdevelopment time.Percent predation Stage-specific generational Best with sentinel prey thatpredation of a prey withhave persistent structuraldiscrete generations.components (eggs, pupae,scales).Predation rate Time-specific rate of prey Requires regular sampling forconsumption for prey with predator abundance andeither discrete or continuous accurate estimation of pergenerations.capita rate of prey consumptionby predator instars.The biodiversity of entomophagous speciesSampling for Indicators of ecosystem health. Ants, carabid beetles, andbiodiversityspiders can be used as sensitivebioindicators.Latitudinal gradients of species Parasitoid taxa showrichness.contrasting latitudinal gradientsof species richness.Global biodiversity.Parasitoids can be used toestimate the richness of onlypartially known taxa.

IndexNote: Page numbers in italics refer to figures and those in bold to tablesAbbas, I. 261Abbott, I. 150, 156Abe, T. 221, 225, 228–31Abensperg-Traun, M. 41, 231Abies balsamea 83absolute density 186abundanceestimation of 3indices of 192see also activity–abundance conceptabundance index see relative abundanceabundance–frequency distribution 211,211, 212Acanthotermes 223Acarina 158, 167acetic acid 18in traps 43–4Acidnotermes 223acoustic emission 233acoustic monitors 238Acrididae 78active traps 5, 66, 150activity patterns, diel 45activity–abundance concept 49–50,57Adaiphrotermes 223Adaliabipunctata 2decempunctata 2Adam, E.E. 38, 42, 44Adams, E.S. 224Adams, T.H.L. 267Addicott, J.F. 182Adelgesabietis 80–1, 81tsugae 84Adelgidae 78, 80, 83–4Aderitotermes 223Adis, J. 44, 147, 150–1, 153, 229Aedes 132Aeolus mellillus 20Africa 89, 138, 147, 226, 228Agalychnis callidryas 174Agassiz, D. 96Agromyzidae 67, 76Agusti, N. 268Aizen, M.A. 146Akey, D.H. 258alcohol 215in traps 43–4aldehydes 133Aleutherocanthus woglumi 131All, J.N. 23Allsopp, P.G. 22Alyscotermes 223Alzugaray, M.D.R. 25Amalin, D.M. 269Amalotermes 223America 96, 133see also Central America; North America;Panama; South America; UnitedStatesAmicotermes 223Ammer, U. 147ammonia 18anaesthetic, aerosol 191analysis of variance (ANOVA) 180Andersen, A.N. 269Anderson, J.R. 133, 179Andis, M.D. 188Andow, D.A. 265Andrei-Ruiz, M.-C. 269279

280 INDEXAndrena 128Andricus 79, 81Anenteotermes 223Anobiidae 78, 110Anobium 80ants 37, 41, 45, 49, 132associated with aphids 17–18as bioindicators 269ponerine 172predation by 265Aphid Bulletin 121Aphididae 16, 78, 83, 255aphids 2, 6, 11, 16, 30, 78alate 61apterous 61balsam twig 83birch 13bird cherry 4cabbage 263cereal 8, 67, 76comparative catch study on 131European movement of 121–2extraction of 28non-cereal 67, 76parasitoids of 255–6pest species 122predation of 267reproduction during sampling 27Rothamsted dataset for 122subterranean 17–18suction traps for 60–1, 120–1true 83woolly 78, 83gall woolly 79hemlock woolly 84, 85pine woolly 77spruce woolly 80Aphis fabae 122, 131Aphodius 132Apilitermes 223Aporusa lagenocarpa 150Araneae 37, 58, 167Arctica 117Armstrong, G. 45, 128, 137armyworm 260beet 11Arnold, A.J. 17arthropods 24–5, 97–8, 150, 157arboreal 151canopy 146–7diel movement of 155flightless 157–8, 166suspended-soil 146Arthur, J.M. 269Asia, South/Southeast 89, 147, 221–2Askew, R.R. 254aspirator (pooter) 69Asplenium nidus 230Astalotermes 223Atchley, W.R. 206Ateuchotermes 223Atkinson, L. 224Attfield, B. 83Ausden, M. 26Austin, A.D. 82Australia 21, 82, 89, 103, 147–8, 232, 239autoradiography (d 13 C and 32 P) 29Baars, M.A. 41–2, 44–5, 47–8Badenhausser, I. 22Baert, L. 49bagging and clipping, index of methods andapproaches 166Bahamas 269baited traps 67, 121, 132–6animals as bait 133–4designs and modifications 132index of methods and approaches 145wind effect on 135–6baits 20–1, 120hexaflumuron 242–3pheromone 106, 123potato 21for subterranean larvae 20sugar 117Baker, T.C. 134Bakke, A. 106Bale, J.S. 122Balogh, J. 62bamboo traps 175Banks, C.J. 137Banksia marginata 82Barber, H.S. 38, 45Barbour, D.A. 136bark sprays 159–60index of methods and approaches 167Barker, M. 147Barker, M.G. 147Barlow, N.D. 259Barndt, D. 45

INDEX 281Baroni-Urbani, C. 222, 226, 228Barrera, R. 171, 180barrier traps 45Bartlett, R. 269Basidentitermes 223basin traps 12Basset, Y. 15–65, 68, 147, 149, 150,155–7Bauerle, W.L. 149Baylis, J.P. 29Bazzaz, F.A. 206Bean, J. 239Beasley, V.R. 44beating trays 69, 76, 83, 257Beattie, A.J. 149bees 128beetles 4, 46, 65, 76, 110, 123adult, boring by 79–80ambrosia 78–80, 114–15bark 13, 78–9, 86–8, 114–15attack density distribution of 102baits for 107flight patterns of 109, 109flight trapping of 106, 107interception trapping of 137parasitism in 267species accumulation curves 99–100,100survival/emergence data for 104carabid 268–9carrion 132color preference by 128dead-wood 66distortion by adhesive 129dung 132eucalyptus longhorn 94five-spined engraver 103ground 37, 137jewel 78ladybird 2, 13, 83larch bark 103longhorn 78–80, 114–15egg laying by 90larvae of 86, 94popular longhorn 79tropical 101marked 210olive bark 104pine shoot 79–80, 86, 87powder-post 78, 80, 114–15predatory 11, 60, 75roundhead 79rove 37saproxylic 97spruce bark 79, 103, 107surface-active 6“woodworm” 78, 110Behan-Pelletier, V. 147, 151, 158–9Belcher, D.W. 20Belgium 103Belkin, J.N. 170Bell, J.R. 70Bellows, T.S. 258, 259, 261Belshaw, R. 63Belshaw, R.D. 149Benke, A.C. 211–13benzoic acid, in traps 44Berlese–Tullgren funnel 26, 229Blasdale version 26Bernklau, E.J. 21Best, V. 147Bidlingmayer, W.L. 124–5, 137Bignell, D.E. 221–2, 225, 229, 236Binns, M.R. 21, 23biotrons 29Bird, D.F. 206Bjostad, L.B. 21Black, H.I.J. 221, 228Blackburn, T.M. 206blackfly, citrus 131Blackwell, A. 134Blair, J.M. 26Blanc, P. 149Blank, R.H. 24Blanton, C.M. 68, 155Blasdale, P. 26Blaustein, L. 186–220Bloom, P.E. 42, 44, 47Blossey, B. 16Blumberg, D. 255bollworm, pink 118, 136Bombosch, S. 46, 267booklice 77, 78Borden, J.H. 100borers 79–80, 86European corn 129, 265lepidopteran 89mahogany shoot 89pine shoot 89teak beehole 110

282 INDEXvaricose 79wood 109–10Borneo 230, 233–4, 235, 236, 237Bostanian, N.J. 45Bowden, J. 58, 117, 159Bowie, M.H. 66–7, 257Bracken, G.K. 17Braconidae 82, 255, 269Aphidiinae 255Bradshaw, W.E. 171Bravery, A.F. 238–40Brazil 232cerrado 231Breene, R.G. 266–7Brennan, K.E.C. 269Briggs, J.B. 40, 42, 44brine (saline solution) 18pipe method 18–19in traps 43–4Briones, M.J.I. 29Britain 121, 125, 240British Columbia 158Brodmann, P.A. 264bromeliads 168, 175, 181Brower, A.V.Z. 269Brown, F.S. 137Brown, M.W. 257Brown, V.K. 16, 21–2Brunei 150Brust, G.E. 29budworm, spruce 131, 136Buffington, M.L. 59, 61–2, 70Buprestidae 78–9Burgess, L. 117Burke, D. 66Burmaster, D.E. 212Burns, D.W. 258Buxton, R.D. 231Byerly, K.F. 61–2, 68Byers, J.A. 108cabbageworm 261California 86, 110Calyptera 65, 75Cameroon 150, 221–2, 223, 225, 228,233–4, 235, 237Campion, D.G. 136Campos, L.A.O. 58Canada 95, 99–100, 136, 147, 158Canaday, C.L. 62–3, 66, 137canopy see forest canopyCappuccino, N. 263Caquet, Th. 191Carabidae 37, 41, 44, 46–8, 50–1carbon dioxide 61, 67, 123, 133, 138, 156evolution of 21cardboard traps 238–9Cardé, R.T. 131, 134, 136Carpenter, D.R. 176Carpenter, S.J. 171Carpenter, S.R. 179–82Carrias, J.-F. 182carrion traps 132Caswell, H. 210caterpillars 9, 13defoliating 11caves 38Cedrela 90Centaurea 24Central America 89Cephalotermes 223Cerambycidae 78, 86, 90, 94, 101Ceratitis capitata 131Ceratopogonidae 123, 169Cercopidae 16Cermak, M. 157chafer, garden 21Chalcidoidea 67, 76, 270Chambers, R.J. 267Chaoborus 187Chapman, J.W. 120Chapman, P.A. 45Chara 196Charles, E. 147Charlton, R.E. 136Cheiropachus quadram 104chemical knockdown 68, 83, 150–5collecting hoops 154, 154collecting trays 153–4comparative advantages of 154–5fogging 151–2, 269index of methods and approaches 166mistblowing 152–5Hurricane Major ® mistblower 152–3;Stihl ® mistblower 153pyrethrum 150Chen, Y. 258, 266, 267–8Cherrill, A.J. 137

INDEX 283Chesson, J. 191China 103Chironomidae 67, 76, 156, 166larvae of 197–8, 206chloral hydrate, in traps 43–4Choristoneura fumiferana 131, 134, 136Chrysomelidae 69, 76Chrysopidae 257Chrysops 138Church, B.M. 117Churchill, T.B. 269cicadas 16–17Cicadidae 16circle traps 96Clarke, W.H. 42, 44, 47Clements, R.O. 19, 25climatic warming 122Closs, G. 206clover 18, 29Clubionidae 37Clutter, R.I. 193–4Coaton, W.G.H. 224Coccidae 78, 84Coccinellidae 67, 76, 257Coccoidea 90Coccus pseudomagnoliarum 85Cochliomyra 132Cochran, W.G. 207Cochran-Stafira, D.L. 182Cohen, J.E. 206Coleoptera 65–6, 75–6, 78, 96, 166bark-dwelling 90, 97borers 86canopy specialists 146epigeal 37host–plant relationship 101larvae of 22, 22, 30, 88, 101night-flying 117saproxylic 97small 63species accumulation curves for 99subterranean 16–18, 20, 20, 24trapping of 128, 157–8wood-destroying 110collecting, informed 13–14collecting hoops 154, 154collecting traps (pots) 95–6Collembola 29, 58, 75–6, 97, 166–7canopy specialists 146densities of 155digging-in effects 49extraction of 26–7, 27, 30, 31rhizophagous 17trapping of 61, 65,68Collier, R.N. 130Collins, N.M. 221–2, 225, 228–30,233color traps 76spacing of 66–7column samplers 190Colunga-Garcia, M. 269Colwell, R.K. 256combined traps 129, 157, 158Compton, S.G. 157Conacher, A.J. 221Conotrachelus nenuphar 96conservation 3Constrictotermes 233Cook, S.P. 265Coon, B.F. 67, 128Cooper, R.J. 156Copeland, R.S. 171Coppedge, J.R. 132–3Coptotermes 79–80, 223, 230, 237formosanus 232Cordo, H.A. 16core samplers 189corn rootworm, southern 29Cornelius, M.L. 67Cossidae 78, 110Cossus 80cossus 79Costa, J.T. 149Costa Rica 89, 90, 172, 258Costantini, C. 67Costello, M.J. 257Cotesia glomerata 261cotton 61, 68, 266Coupland, J.B. 125, 133, 138Coxotermes 223crabs, land 172cranes 174crawl traps 96Crawley, M.J. 204Crenetermes 223crickets 78Croft, B.A. 136Croset, H. 210

284 INDEXCrossley, D.A. 26, 146, 148–9, 155Crossley, J.D.A. 38, 42, 45Cryptococcus 93fagisuga (= fagi) 77, 90, 93Cryptotermes 230Cubitermes 223sankurensis 226Culicidae 169Culicoides 118–19, 125, 134curculio, plum 96Curculio caryae 29, 96Curculionidae 24, 86, 88, 96Curtis, D.J. 43, 49Cydia pomonella 134Cynipidae 81Cytocool ® 191Daane, K.M. 257Dafni, A. 128Daktulosphaira vitifoliae 17damselflies, larvae of 172Dangerfield, J.M. 226, 231Dangerfield, P.C. 82Danum Valley 234, 235, 236, 237Darling, D.C. 65Darlington, J.P.E.C. 224, 226, 228Darwin, C. 2David, C.T. 136Davies, J.B. 133Davies, R.G. 221, 234Davis, A.E. 65Dawah, H.A. 258Dawes-Gromadzki, T. 231De Barro, P.J. 22, 24, 67de Souza, O.F.F. 231Dean, D.A. 123Deansfield, R.D. 128death rate analysis 261Declining Amphibians Populations TaskForce (DAPTF) 215DeFoliart, G.R. 174Dejean, A. 225Delia 130antiqua 24–5radicum 17Delta traps 134–5Dempster, J.P. 268Den Boer, P.J. 38, 47–51Dendroctonusfrontalis 105micans 79valens 109, 109Dendy, J.S. 191Dennehy, T.J. 17Dennis, P. 40, 44, 51Desender, K. 38, 44–5, 51deserts 38detrended correspondence analysis (DCA)97–8Devy, M.S. 147Diabrotica undecimpunctata 29Dial, R. 147–8Diaz-Aranda, L.M. 257Dibog, L. 222, 223, 228Didham, R.K. 147diesel fuel 18Dietrick, E.J. 59Dietrick vacuum sampler (D-vac) 59Digweed, S.C. 44, 47–9Dioryctria cristata 79dippers 188, 195Diptera 75–6, 78, 166biting 120color preference by 128densities 155large 125larvae of 18–19, 169rhizophagous 16sedentary 68, 76male 123night-flying 117slow-flying 116small 118, 123, 125, 136subterranean 23–4trapping of 62–3, 65–6, 97, 118, 128, 157in tree holes 174and visible light 117Disney, R.H.L. 65–7, 128distributionaggregated (clumped or contagious) 7,21, 45random 7regular (uniform) 7Dixon, A.F.G. 4Doane, C.C. 261Dodsall, L.M. 24Dolphin, K. 269Dondale, C.D. 61Donovan, S.E. 221Doty, R. 42, 44

INDEX 285Downey, J.E. 117dragonflieslarvae of 172libellulid nymphs of 196drainpipe traps 107Dransfield, R.D. 226Dreistadt, S.H. 85–6drift fences 45–6drift nets 192Dromius 51Dubois, D. 201–2Duelli, P. 270dung traps 132Duplidentitermes 223Durand, C.M. 258Dutcher, J.D. 23dyes, fluorescent 232Dyna-Fog ® 151earthworms 29sampling of 19Easey, J.F. 224, 228, 231East, R. 23Eburnitermes 223eclosion 17Ecological Entomology 138ecological studies, pitfall trapping in 37–57Edgar, W.D. 266, 267Edwards, P.B. 118Eggleton, P. 221–53, 223Egypt 136Ekman dredge 190Elateridae 20, 20electrophoretic analysis 267Elkinton, J.S. 135, 264Ellington, J. 61–2, 266Elliott, N.C. 257Ellwood, M.D.F. 230elutriation 25Elvin, M.K. 24emergence traps 17, 157–9, 192endoscopes 238Ephemeroptera 194epiphytes 159–60Epstein, M.E. 44, 46, 48Erbilgin, N. 106Ericson, D. 38, 42–4, 46, 48Ernobius mollis 110Erwin, T.L. 146, 149EstimateS software 256ethanol 109, 154, 157ethics of sampling 215ethyl acetate 156ethyl alcohol 238ethylene glycol 43–4eucalyptus 94, 94Eucalyptus marginata 150Euceraphis punctipennis 13Euchilotermes 223Eucosma 89sonomana 89Europe 1, 86, 91, 121, 146–7, 153European Science Foundation TropicalCanopy Research Programme146–7Eutermellus 223Evans, T.A. 231–2external feeders 83–6Exterra ® 239extraction methodsfield see field extraction methodslaboratory see laboratory extractionmethodsFabre, J.H. 2Faragalla, A.A. 38, 42, 44Fastigitermes 223feces see frassFeinsinger, P. 146Feoktistov, B.F. 44Fermanian, T.W. 18Ferrar, P. 231Ferson, S. 202Fichter, E. 38, 42, 44Fidgen, J.C. 80, 81field cricket, black 21field extraction methods 18–23baits 20–1behavioral 19–20chemical 18–19hand-sorting 21–3index of methods and approaches 36figs, fallen 160Finch, S. 5, 67, 130Fincke, O.M. 168–85Finland 97–8, 98, 106Finnamore, A.T. 269fir, grand 100Fish, D. 168, 176, 180–1Fitzgerlald, C.J. 238–9

286 INDEXFleeger, J.W. 189Fleminger, A. 193–4flies 4, 133, 169apple maggot 131biting 64, 67, 133blow 132fungus 78horse 138Mediterranean fruit 131small, distortion by adhesive 129sugar beet 128tsetse 128, 132–3turnip root 130Floate, K. 256Floren, A. 68, 147, 150, 153Florida 124flotation techniques 61, 228–9compared to Tullgren extraction 28for egg counts 24Fluon ® 98fluorescent bulbs, black 61Foggo, A. 149foliage bagging 68, 76Foraminitermes 223forest 38canopyaccess to 147–9; canopy crane 148;canopy raft 149; canopy sledge(luge) 149; canopy walkways148, 174aerial and arboreal traps in 157–9insect densities in 150sampling insects from 146–67sampling issues 149–50cloud 159neotropical 170plantation 6, 70rainforests 67, 70, 100, 159, 228, 232temperate 70understorydefinition of 58insects associated with 58Forest Research Institute Malaysia 148formaldehyde 215formalin 18, 127in traps 43–4Formicidae 29, 37Formosa 232Forrester, G.J. 23Forschler, B.T. 232Foster, R.B. 176Fowler, R.F. 22Fox, B.J. 269France 121, 239–41Frank, J.H. 37, 175, 180–1frass (feces) 17, 70–71, 77, 89–90, 101,110French Guiana 101frogsdendrobatid, tadpoles of 169territorial 175Fuchs, A. 238funnel traps 12, 107, 109, 138Furculitermes 223fuzzy numbers 202Gaasch, C.M. 269Gadagkar, R. 63Gair, R. 12Galindo, P. 169, 174–5gallery patterns 102species-specific 100–1galls 80–2formers of 114–15size/population density relationship of82The Gambia 124Gange, A.C. 16–36, 18–19, 21–2, 22, 27Ganio, L.M. 147, 155gasoline see petrolGaston, K.J. 146, 206, 258, 269Gathorne-Hardy, F.J. 221, 224, 234Gauld, I.D. 258, 269Gaydecki, P.A. 117Gee minnow traps 192Geiger, C.A. 83Geometridae 95larvae of 69George, K.S. 12Germany 148Geurs, M. 25Giblin-Davis, R.M. 38, 45Gilbert, F.S. 257Giller, P.S. 42, 46, 268Gillespie, D.R. 131Gillies, M.T. 132, 138Gist, C.S. 38, 42, 45Gladwin, J. 104Glasgow, J.P. 133Gleditsia triacanthos 69

INDEX 287Glen, D.M. 4Global Canopy Programme147Glossina 132morsitans 128tachinoides 128Glyptotermes 230Goldson, S.L. 17, 24Gonzalez, R. 103Good, J.A. 42, 46Gora, V. 93Gordon, R.D. 257Gould, F. 29Gould, J.R. 259, 260–2Goulet, H. 66grab samplers 189Ekman dredge 190Petersen grab 190Grace, B. 23Gradwell, G. 96Graham, H.M. 118Gramineae 67grasshoppers 78Gratwick, M. 19, 26Gray, D.R. 84Gray, H. 62, 137Great Lakes IPM 96Greaves, T. 230Greenslade, P. 41–2, 44–5Greenslade, P.J.M. 37, 40–2, 44–5,47–9Greenstone, M.H. 258, 266, 267–8Greg, P.C. 118grids 46–7, 89Grove, S. 97grubschafer 17–18, 20, 22, 30white 20, 22Gryllidae 78Guilbert, E. 147, 150Gurevitch, J. 210gut detection 257, 267–8Gutierrez, A.P. 62, 83, 259gutter traps 45Haagsma, K. 239Haarløv, N. 26habitatsseasonal changes in 4on trees 78hackberry, Chinese 85–6Hadrys, H. 181Hagler, J.R. 258, 266, 267–8Hall, D.R. 134Hall, D.W. 42–4Hallé, F. 149Halsall, N.B. 50Hambler, C. 70Hammer, M. 26Hammond, P.M. 38, 42, 44–5, 146, 150,153Hand, S.C. 61Hanski, I. 38, 44–5, 48, 120Hanula, J.L. 22, 96Harcourt, D.G. 21, 23Harlow, L.L. 209Harper, A.M. 128Harper, M.G. 270Harrington, R. 120–1Harris, W.V. 230Harrison, R.D. 29Hartstack, A.W. 118Hattis, D. 212Haufe, W.O. 117Haugen, L. 175Haverty, M.I. 231, 239Havukkala, I. 24Hawkins, B.A. 256, 258, 269Hawthorne, D.J. 17Heath trap 118Heathcote, G.D. 4, 131heather 66–7Heatwole, H. 147Hébert, C. 95–6Hedges, L.V. 210Heimpel, G.E. 268Heinz, K.M. 265Heliconia, bracts of 175Heliothis 118zea 129Hellqvist, C. 87, 87Hem, D.G. 125Hemiptera 78, 97, 167bark-dwelling 90distribution of 80, 91population density of 83–4rhizophagous larvae of 16trapping of 158Henderson, G. 233Henderson, I.F. 61

288 INDEXHenderson, P.A. 14, 17, 24–5, 62, 70,117–19, 124, 139, 159Hepialidae 78Hepialus californicus 17heptane 25Herms, D.A. 69–70Herting, B. 255Hertz, M. 38, 40, 45Hess sampler 189Hester, F.E. 191Hester–Dendy sampler 191, 196Heteropsylla cubana 83Heteroptera 69, 76, 174, 265Heydemann, B. 50Higgins, W. 147Hilborn, R. 200Hildrew, A.G. 267Hill, C.J. 157Hill, D. 150Hill, M.G. 259, 260Hinze, B. 233Hoare, A. 225Hodek, I. 254, 257Hodotermes 224Höft, R. 155–6Hokama, Y. 188, 192, 195Hollier, J.A. 149Holloway, J.D. 158Holopainen, J.K. 43–5, 47Holt, J.A. 221, 224, 228, 231Holzapfel, C.M. 171Homoptera 75–6, 166predation of 265sessile 260trapping of 61, 66, 69Hondelmann, P. 269Honêk, A. 38, 40, 42–3, 46–7Hoogendorn, M. 268Hopkin, S.P. 17Hospitalitermes 224, 230, 233House, G.J. 25Houston, W.W.K. 45hoverflies 83Howard, F.W. 89, 90Howell, R.S. 93Hurlbert, S.H. 150, 174Husseneder, C. 224Hutcheson, J. 64hydrocarbon adhesion 24–5, 30Hylastes longicollis 109, 109Hylobiusabietis 6, 11pales 96Hylurgops 103Hylurgops palliatus 102Hymenoptera 75–6, 166aculeate 61epigeal 37forest-dwelling 58, 81gall-forming 82night-flying 117parasitic 88, 104, 155predation by 29small 62, 65trapping of 63, 66, 67, 97, 128, 157Hypsipyla 89grandella 79, 90, 90Hyvis ® 95Ichneumonidae 269incidence counts 11India 147Inoue, T. 148inquilines 225insectsactivity 78evidence of 79–80arboreal habitats 78bark-boring 101canopy, sampling techniques and methods146–67cryptic 155densities 155dimorphic 2entomophagousassemblages of 254–8diversity of 268–70index of methods and approaches277–8external clues to presence on roots17–18in flight 116–45in fruit, seeds, and silk 159–60large mobile 155migratory 5of phytotelmata 168–85polymorphic 2root-feeding (rhizophagous) 16scale see scale insectssessile 155

INDEX 289Institute of Animal Health, Pirbright 125Intachat, J. 158–9interception traps 4, 63, 232, 269composite 64, 158index of methods and approaches 145undetected 137–8use in forest canopy 157–9, 158visible 138window 65International Canopy Network (ICAN) 146inverse prediction 198invertebrates, epigeal 47sampling of 37–8Ips 79, 103, 105, 106, 107, 108cembrae 102, 103grandicollis 96, 103typographus 103, 105–6, 107, 108Ishii, T. 153Isoptera 221–53Italy 239Jactel, H. 89, 91, 91–2Jaffe, K. 44–5Jakus˘, R. 101, 102, 103Janzen, D.H. 269Japan 147–8, 239Jarosík, V. 38Jasienski, M. 206Jenkins, D.W. 171Jenkins, T.M. 240–1Jensen, R.L. 137Johnson, C.N. 206Johnson, M.D. 156Johnson, P.C. 258Johnson, R.A. 228Jones, D.T. 221–53Jones, J.A. 221Jones, S.A. 266, 268Jones, S.C. 222Jones, V.P. 131Joose, E.N.G. 42, 49Jugositermes 223Juliano, S.A. 175Kaila, L. 97–8Kalotermitidae 223, 224, 230, 233Kambhampati, S. 222Kammen, D.M. 194Kapteijn, J.M. 49Katsoyannos, B.L. 131Kauffman, W.C. 87Kean, J.M. 259Keaster, A.J. 20Keating, K.A. 256Kegel, B. 48Keil, S. 134Keller, M.A. 255Kelsey, R. 104Kempson extractor 229Kendall, D.M. 134Kendrick, W.B. 26Kerck, K. 151, 154kerosene 153in traps 44Kethley, J. 25ketones 133Kharboutli, M.S. 38, 46–7Kieckhefer, R.W. 257Kirchner, T.B. 207Kiritani, K. 267Kirk, W.D. 128Kirk, W.D.J 66Kirton, L.G. 239Kitching, R.L. 17, 146–7, 149–53, 158–60,168–9, 175, 181, 270Kitt, J.T. 255Kleintjes, P.K. 83Klironomos, J.N. 26Knight, J.D. 4Kolmogorov–Smirnov confidence interval211, 212Koponen, S. 98–9Kotler, B.P. 191Kozicki, K.R. 18Kramer, E. 44Krebs, C.J. 204, 206–7, 210, 212Kring, J.B. 131Kuenen, L.P.S. 136Kuhn, R. 202Kulman, H.M. 44, 46, 48Kuschel, G. 45Labiotermes 223laboratory extraction methods 23–8, 159behavioral methods 25–8dissection of roots 23–4flotation methods 24–5elutriation 25hydrocarbon adhesion 24–5index of methods and approaches 36

290 INDEXlaboratory visualization methods 28–30index of methods and approaches 36Labuschagne, L. 17, 21Lacessititermes 233ladybirds 83Lamb, R.J. 62–3Lamberti, G.A. 190Långström, B. 87, 87Larsen, K.J. 269Lasius 29Lauenstein, G. 24Lawrence, E. 58Lawrence, K.O. 43, 45Lawson, S.A. 103Lawton, J.H. 180, 221leaf miners 11, 13leaf mines 70Leather, S.R. 1–15leatherjackets 18Lee, K.E. 221–2, 226Lemieux, J.P. 38, 43–4Lenz, M. 231Leong, J.M. 128–9Leos Martinez, J. 123Lepage, M. 221Lepidoptera 63, 75–6, 78fast-flying 116large 118larvae of 255borers 86, 89, 110rhizophagous 16male attractants in 134mimic species 2night-flying 117and moonlight 117parasitoid assemblages of 258shoot boring 88trapping of 65, 69, 95, 118, 158–9, 166and ultraviolet light 117web-spinning 13Leptomyxotermes 223Lerin, J. 22Leuthold, R.H. 233Lewis, C.N. 269Lewis, V.R. 239Libellula, larvae of 208, 208, 209life table analysis 258light traps 4, 116–21, 129, 157–9, 187,232aquatic 195–6“black” 117, 119comparative catchability 117disadvantage of 159examples of use 120index of methods and approaches 144influences on catch efficiency andeffective catching distance 117–18mercury vapor 17, 117Onderstepoort Veterinary Institute design126priciples of use 116tungsten 17, 117–18types of light source 117Lima, S.L. 197Limnanthes douglasii rosea 128Lindgren, B.S. 38, 43–4Lindgren traps 109linear transects see transectsLinsenmair, K.E. 68, 147, 150, 153Linton, Y.M. 125Lister, A. 268Lobry de Bruyn, L.A. 221lobster-pot traps 96, 118Lolium perenne 18Long, G.E. 266, 267Longino, J.T. 146, 159Longipeditermes 233longipes 224Loor, K.A. 174Lopez, E.R. 259, 263–4Loska, I. 62Lounibos, L.P. 172, 174–5, 181Lowman, M. 62Lowman, M.D. 146–7, 154–5, 174Lozano, C. 104Lucilia 132Luck, R.F. 258Luff, M.L. 40–1, 44–8, 268Lunderstädt, J. 93Lussenhop, J. 26, 29Lycosidae 37Lyctidae 78Lyctus 80Lygus 63lineolaris 69Lymantria dispar 70, 71, 134Machadotermes 223Macías-Sámano, J.E. 100Mack, T.P. 38, 46–7

INDEX 291MacKay, R.J. 189Macleod, A. 60MacMahon, J.A. 42Macrotermes 224, 226, 233bellicosus 233malaccensis 224Madagascar 221Madden, K.E. 269Madge, D.S. 229Madoffe, S.S. 106Maelfait, J.P. 38, 44–5, 49, 51maggot, sugar beet root 22–3magnesium sulfate 24–5Maguire, B. 168Magurran, A.E. 212mahogany 89–90, 90Majer, J.D. 49, 68, 147, 150, 155–6, 269Maki, K. 230Malaise, R. 64, 138Malaise traps 4, 75, 158, 232, 269aerial and arboreal use of 149, 157–9basic design for 63–5, 138Cornell design 64Coupland design 133Townes model 64Malaysia 228, 234, 235Mangel, M. 200Mankin, R.W. 233Manly, B.F.J. 210Margarodidae 91Marini-Filho, O.J. 146Maron, J.L. 17, 23Marshall, D.A. 42, 44Martikainen, P. 99, 106, 107Martin, J.L. 146Martius, C. 229–30, 232Masner, L. 66Masters, G.J. 17Mastotermes darwiniensis 232Mathot, G. 226Matsucoccus feytaudii 91, 91–2Matsumoto, T. 221, 225, 228–9, 233Matthaei, C.D. 192Matthews, J.R. 64–5Matthews, R.W. 64–5May, R.M. 149, 210–11Mbalmayo Forest Reserve 225, 234, 235,237McArdle, B.H. 214McCreadie, J.W. 133McDougall, G.A. 134McGeachie, W.J. 117, 128McGeoch, M.A. 146McSorley, R. 24, 26Meads, M.J. 96–7mealy bugs 16, 77wax-covered 24Medeiros, L.G.S. 230, 232Melanotuscommunis 20depressus 20similis 20verberans 20Melbourne, B.A. 38, 45–6, 49Menalled, F.D. 263Merrick, M.J. 132Merritt, R.W. 171, 181, 187, 189mesocosms, sampling with 191–2mesohabitats 159–60Mesostoa kerri 82, 82Messing, R.H. 269methyl bromide 226methylbutenol 108Meyer, V.W. 226Meyerdirk, D.E. 131microarthropods 26–7, 29Microcerotermes 223crassus 224microcrustaceans 181microhabitats 159–60microhymenoptera 66, 76Microtermes 223, 228midges 118–19, 134, 166biting 169clouds 156slow-flying 123Mill, A.E. 232Miller, C.K. 134Miller, L.J. 269Miller, L.R. 232Mills, N. 254–78Mindarus abietinus 83minnow traps 187, 195, 208Gee model 192Mitchell, A. 147–8Mitchell, B. 40, 42, 46mites 26, 146oribatid 158, 166Miura, T. 233Moeed, A. 96–7

292 INDEXMoffet, M. 147, 174molten agar technique 24–5Mommertz, S. 38, 45, 50Monserrat, V.J. 257Moran, V.C. 147, 160Morin, P.J. 191morpho-taxa 149Morrill, W.L. 41–2, 45, 50Morris, M. 117Morse, J.G. 266, 268mosquito dipper 188mosquitoes 67, 117–18, 123–4, 132, 138biting 125breeding sites 169egg rafts 193larvae of 170, 173, 181, 195, 198,200–1, 202pupae of 200–1, 202and trap shape 125in tree holes 175, 177, 180woodland 124moss mats 159–60cores from 159–60index of methods and approaches167Mosugelo, D.K. 231moths 110, 131distortion by adhesive 129European pine shoot 79ghost 17goat 79–80gypsy 70, 71, 134, 260–2, 265large 136larvae of 78, 86boring by 79–80migratory 118and pheromones 132pine beauty 8, 136sampling of 12pine shoot 79–80, 89release study of marked specimens 118spruce budworm 134and ultraviolet light 117winter 8–9, 95–6, 96wood 79–80, 110Muirhead-Thomson, R.C. 116, 121Muller, C.B. 256Müller, H. 24Murphy, K.R. 209–10Murphy, W.L. 130Murray, P.J. 25Mycetophilidae 78Myles, T.G. 239Mymaridae 269Myors, B. 209–10Myzus persicae 122Nadkarni, N.M. 146–7, 159Naeem, S. 175Nag, A. 116Najas 196Nakamura, K. 261Naranjo, S.E. 268Nasco ® Whirl-Paks 173Nasutitermes 223Nath, P. 116Natural History Museum, UK 154Nealis, V.G. 88necrophagy 258, 267Nelsen, R.B. 207Neotermes 223, 230Neuroptera 265New Jersey trap 118New, T.R. 269New Zealand 97, 147Newton, A.C. 89–90Nicrophorus 132Niemelä, J. 5, 38, 40–1, 43–8, 269Nigeria 228nitric acid 18nitrogen 182Noctuidae 118Noditermes 223Noirot, C. 224Norris, R.H. 205, 209–10North America 88–9, 153see also United StatesNovak, R.J. 176Noyes, J.S. 62–6, 67, 269–70Nürnberger, B. 210Nutting, W.L. 222, 231–2Nyffeler, M. 267Nyrop, J.P. 29–30oaks 13, 70, 71, 81Obeng-Ofori, D. 38, 40, 42, 50Obrist, M.K. 270Obrtel, R. 46, 48Ocloo, J.K. 2311-octen-3-ol 134

INDEX 293octenols 133Odendaal, F.J. 105, 105Odonata 156, 166, 169, 173predatory 177–8territorial 175Odontotermes 223Ohiagu, C.E. 228Ohmart, C.P. 155–6oilseed rape 66–7Okwakol, M.J.N. 221Oliver, I. 149Oliviera, M.L. 58Olson, A.C. 264onion fly 24–5Operophterabruceata 95brumata 95Ophiotermes 223Orr, A.G. 169Orthoptera 78, 97Orthotermes 223Osenberg, C.W. 209Otiorhynchusligustici 21, 23sulcatus 17, 21Otitidae 23oviposition scars 101Owen, J.A. 45Ozanne, C.M.P. 58–76, 146–67, 150, 269Paarmann, W. 150–1, 154, 160Pachypappa 27, 28Pachypappella 27, 28Packer, L. 65Pakistan 136Palmer, I.P. 156pan traps 5, 257Panama 148, 168, 169, 173–4, 177, 180Panolis flammea 12, 136Parajulee, M.N. 67, 257parasitismapparent 264death-rate analysis of 261estimating impact of 259heterogeneous 264marginal attack rate 264percent parasitism 259–64recruitment analysis 261–2sampling concerns 263–4single-sample estimation of 260–1species-specific, DNA detection of 264stage-specific generational 260–2time-specific rate of 262–3trap host method 261parasitoids 66, 104, 106, 254–78assemblages 255–6braconid 269regional variation 258Parker, G.G. 147Parmenter, R.R. 42Parrella, M.P. 265Pasoh Forest Reserve 234, 235passive traps 5Pasteels, J.M. 224Paton, R. 231PCR analysis 257–8, 268Pearce, M.J. 222, 231Pearman, P.B. 191Peck, R.W. 109, 109Peck, S.B. 65Pectinophora gossypiella 118, 136Peloquin, J.J. 176Penev, L.D. 21Pennsylvania trap 159Penny, M.M. 49Pericapritermes 223Perry, D.R. 174pest control 3pest-monitoring programs 38, 44–5Peters, B.C. 238–9Petersen grab 190Petersen, I. 138Petersen, J.E. 191Petersen, J.J. 263petrol (gasoline) 18Phelps, R.J. 133Pherocon ® trap 135pheromone traps 12, 17, 107, 108–9,134–6, 135combined 121, 129uses of 136pheromones 45, 67, 106, 108, 120, 123,130Phloeotribus scarabaeoides 104Phoracantha 79semipunctata 94, 94Phoridae 67, 76phosphorus 182photography 1phototaxis 13, 63

294 INDEXPhoxotermes 223Phyllopertha horticola 21Phyllophaga 22phylloxera, grape 17Phymatodes 80Physalaemus pustulosus 174phytotelmata 168, 170, 174, 180–2Picado, C. 181Picea sitchensis 27Pimm, S.L. 175pine scale, maritime 91, 91a-pinenes 45b-pinenes 45pines 100, 103plantations 22trees 86–7fallen shoots of 87Pineus pini 77Pinus sylvestris 150, 242pipe and bucket system 239, 242pipe traps 107, 108piperonyl butoxide 153pipette, suction 172Pirbright trap 125Pires, C.S.S. 81Pissodes 80strobi 88pitcher plants 168pitfall trapping 38in ecological studies 37–57pitfall traps 6, 37–57, 98, 128, 269baits used in 44–5design and application of 39–45, 40, 56funnels in 42, 56killing agents, preservatives, anddetergents 42–4, 43, 48, 56materials used in 40–1, 56rims of 42, 56roofs of 41–2, 56shape and size 41, 56spatial arrangement of 47specialized designs of 45, 56subterranean 45types of 39use of baits in 44–5, 56uses of 38Pityogenes 103Plantago lanceolata 63Platypodidae 78Platypus 80Plecoptera 138Plutella xylostella 120Poisson distribution 204Polis, G.A. 214Polygraphus 103polyisobutylene 130Pomeroy, D.E. 226Pontin, A.J. 27pooter (aspirator) 69populationage structure of 3density of 3dispersion (distribution) of 3distribution of 7dynamicschaotic 123prediction of 3mortality 3natality 3trend 3Porter, E.E. 256possibility theory 201Postsubulitermes 223potassium permanganate 18Potter, M.F. 238Powell, W. 267Prade, H. 201–2Prairie, Y.T. 206predation 264–8percent predation 265prey-specific DNA, detection of 268rate of 265–8predators 66, 254–78assemblages 256–8density of 265–6rate of consumption 266–8, 266regional variation 258Price, P.W. 81Proboscitermes 223Procubitermes 223Proffitt, J.R. 24Prohamitermes mirabilis 224propylene glycol 43–4Protermes 223protozoans 181Prueitt, S.C. 89Prunus padus 4Pscoptera, bark-dwelling 13Pseudacanthotermes 223Pseudococcidae 16Pseudomicrotermes 223Psila rosea 130

INDEX 295Psocoptera 65, 68, 76, 78, 166–7Psyllidae 83, 155leucaena 83Pteromalidae 104Pulvinaria 92–3regalis 77, 90–1, 93Purcell, M.F. 269Pyralidae 78, 86, 89pyrethrins 153pyrethroids 153, 156pyrethrum 153quadrats 6, 10–12, 22, 193, 264deep 21and suction sampling 61for sampling termite mounds 226,233–4vertical 92quarantine lists 5Quercus 71robur 150Quicke, D.L.J. 269Quinn, J.F. 256Quiring, R. 131radiography 29–30radioisotopes 231Raffa, K.F. 38, 45, 106rainforests 67, 70, 100, 159, 228, 232Rainio, J. 269Ramaswamy, S.B. 131ramp traps 45Ratcliffe, S.T. 264Rawlings, P. 119rearing rooms 104Rebello, A.M.C. 230, 232Recher, H.F. 68, 155–6Recruit ® 242Redak, R.A. 59, 61–2, 70Reeves, R.M. 258Reid, M.L. 104relative abundance 186Reling, D. 137Resetarits, W.J. 191Resh, V.H. 187–8, 190–1, 193–4Reticulitermes 237lucifugus grassei Clement 240, 242santonensis 243Reynolds, B. 148Rhagoletis pomonella 131Rhinotermitidae 223, 233, 237Rhoades, M.H. 257Rhopalosiphon padi 4Rhyacionia 89buoliana 79Ridgway, R.L. 70, 71, 136Rieske, L.K. 38, 41, 45Ring, R.A. 147–8, 156, 159–60Rinicks, H.B. 67, 128Ro, T.H. 266, 267Robb, T. 104Roberts, H.R. 151Roberts, I. 118Roberts, R.H. 138Roberts, R.J. 17, 23Robinson, G.S. 120Robinson pattern trap 118, 159Rodgers, D. 146, 149, 160Rogers, C.E. 23Rogers, D.J. 133Rohlf, F.J. 180, 194, 198, 204, 208, 210,212Roisin, Y. 224Romero, H. 44–5root fly, cabbage 17Rosenberg, D.M. 191, 193–4Ross, D.W. 89Rothamsted Experimental Station 121Rothamsted Insect Survey 4, 119, 120, 122Rothamsted trap 118, 119, 122, 124, 159data series 120Rotheray, G.E. 257rotifers 181Rowe, W.D. 193Royama, T. 260–1, 264rubidium 258Rudd, W.G. 137Ruelle, J.E. 225Russell, D.A. 259, 263Rutherford, J.E. 189–90Rutledge, P.A. 189Ryan, R.B 264rye grass 18, 27Saccharicoccus sacchari 24Safranyik, L. 99, 100sage scrub 61St Ives fluid 18Sakal, R.R. 180salamanders 196–7Salt, D.T. 27–8Salt, G. 229

296 INDEXSalveter, R. 269samplingaccess systems 69, 92artificial substrate 199avoidance of bias in 10bagging and clipping 69, 88, 155–7index of methods and approaches 166of bark, external surface 90–9by capturing 95–9by counting 91–4of bark/sapwood interface 99–109by bark removal and log dissection101–4by emergence trapping 104–6by entrance/emergence hole sampling104by flight trapping 106–9by hand-searching 99–100by trap-logging 100–1by beating 69, 83binomial 84sequential 22by branch clipping 69calibration of 194canopy insects, techniques and methods146–67chemical knockdown, index of methodsand approaches 166choice of techniques for 58–9climate effect on 64–5by column samplers 190concepts of 9–12by core samplers 189dampness (rain/dew) effect on 61, 63,65, 118design for aquatic insects 186–220destructive versus non-destructive 6–7devices for aquatic insects 186–220area or column samplers 188–90mesocosms 191–2natural/artificial substrates 190–1nets and dippers 187–8survey of 187–93traps 192visual observation and photography193direct habitat 5direct observation 256–7efficiency 186factors affecting 194–8of eggs 24, 92–3, 95–6of entomophagous species 258–68of epigeal invertebrates 37–8equipment for tree holes 171errors 193–215measurement interactions 193random 193, 203–15systematic 193–4, 198–203ethical considerations concerning 215experimental/controlled 2–3of external feeders 83–6forest canopy issues 149–50of galls 80–2by grab samplers 189, 199Ekman dredge 190Petersen grab 190by Hess sampler 189by Hester–Dendy sampler 191, 196index of methods and approaches 219indirect 257informed 13–14by Kempson extractor 229by kick samples 199location effect 64by mark–recapture 210and measures of unit area 12methods for forest understory vegetation58–76grasses and herbs 59–62shrubs 62–8tall vegetation, including small trees68–70passive 59, 62, 151patterns of 10programs 4of roots 16–36samples required in 9of shoots and twigs 80–6external 80–6internal 86–90by stovepipe (column) sampler 196strategy 22, 31, 45–6, 56depletion effect 48–9, 56digging-in effects 49duration and temporal pattern 47–8,56spatial arrangement 46–7, 56and surrounding vegetation structure49, 57trap number 46, 56

INDEX 297stratified 8–9suction 17, 59–62, 75, 257by Surber sampler 189techniques of 3–4for natural tree holes 170–3of termites 221–53approaches to 225–6difficulties of 222–6index of methods and approaches250–3methods of 226–33, 227in mounds 226–8population density 233–6in soil 228–9subterranean pests of buildings236–40transect protocol 234–6in trees 230using baits 230–1using mark–recapture protocols231–2using traps 232–3in wood 229–30see also termitestheory and practice of 1–15tools/techniques 4–5of trees: shoots, stems, and trunks77–115detection 77–8index of methods and approaches114–15methods of 78types of information required in 10of understory vegetationlow plants, grasses and herbs 59–62medium-height vegetation, includingshrubs 62–8tall vegetation, including small trees68–70units of 9–10criteria for 10–12estimation in 214use of secondary characteristics in 70,76by vacuum 59–62, 69, 75, 265trees and shrubs 70by water washing 83of water-filled tree holes 168–85processes of 171–3wind effect on 118Sanders, C.J. 134Sanderson, R.A. 137Sands, W.A. 221, 224–5, 228, 231sap-feeders 114–15Saperda populnea 79saprophagy 258, 267Sarawak 148savanna 226, 228scale insects 77, 78, 83–5, 90beech 77, 90, 93citriola 85–6density of cover 93–4horse chestnut 77, 90–1, 93maritime pine, vertical distribution of91, 91pine, distribution on bark types 91–2visibility of 91scarab larvae 29Scarabaeidae 17–18, 20, 22, 22, 30Schaefer, C.H. 188, 198Schaefer, G.W. 125Schedorhinotermes 223, 224Scheffrahn, R.H. 239Scheller, H.V. 41, 44, 46Schlyter, F. 108Schmitt, J.J. 257Schoener, T.W. 269Schowalter, T.D. 68, 146–7, 155–6Schubert, H. 147Schuurman, G. 226Schwartz, S.S. 191, 193, 197Scolytidae 78, 86, 96, 99, 101Scolytus 79scorpions 172screw-worms 132–3Scudder, G.G.E. 64Scuhravy, V. 44Seastedt, T.R. 21Sedcole, J.R. 17Sentri Tech ® 239, 242Sentricon Colony Elimination System ®239, 242Sentry ® 242Service, M.W. 124, 170, 188, 195Seybold, S.J. 110Shaw, M.R. 254–5Sheasby, J.L. 224Sheppard, A.W. 16Shlyakhter, A.I. 194shoot borer, mahogany 79

298 INDEXShufran, K.A. 266, 267–8Siemann, E. 210, 270Siitonen, J. 99Silk, P.J. 136Silva, E.G. 229Sime, K.R. 269Simmons, C.L. 38Simon, U. 58Simpson, J.A. 103simulation modeling 259Simulium 125, 133Singer, M.C. 256single rope technique (SRT) 147Sirex 79–80Sitobion avenae 12Sitona 24discoideus 17, 24lineatus 17Sleaford, F. 221Sleeper, E.L. 38Slosser, J.E. 67, 257Slovakia 103Smart, L.E. 17Smith, B.J. 40, 45Smith, D.T. 133Smith, R.J. 132Snodgrass, G. 69snout beetle, alfalfa 21, 23Snow, W.E. 169, 174soapy water 18sodium benzoate 66soilssampling 12sieving 22, 25suspended 159–60Sokal, R.R. 194, 198, 204, 208, 210, 211,212Solomon, M.G. 266, 267Sota, T. 173, 175South Africa 119South America 89, 147, 221–2Southward, M. 266Southwood, T.R.E. 14, 17, 24–5, 62, 70,117–19, 124, 139, 146, 150, 152–3,159–60, 199, 203Spain 104, 125, 239–40spanworm, Bruce 95Speight, M.R. 77–115, 136, 146, 153–4Spence, J.R. 38, 41, 45Spencer, M. 186–220Sphaerotermes 223spiders 269assemblages 257ballooning 123environmental sensitivity of 268lycosid 267predation by 267–8wandering 37, 49webs of 70spittlebugs 16Spodoptera exigua 11Spragg, W.T. 231Springate, N.D. 64, 149, 157spruce 101, 102, 103, 107Norway 105, 110Sitka 8–9, 27white 81Spurr, S.H. 58Srivastava, D.S. 180St-Antoine, L. 95–6Stadler, B. 256Standridge, N. 147Staphylinidae 37, 60, 75Stary, P. 255Stein, W. 46Sterling, P.H. 70Steven, D. 41Stewart, A.J.A. 59–61Stewart, R.J. 188, 198Stewart, R.M. 18sticky traps 4, 17, 129–31, 269aerial and arboreal use of 157baited, effectiveness of 107, 108basic design and use 124–5, 130comparative efficiency 131Delta pattern 134importance of shape and color 131index of methods and approaches 145modifications 130–1use on trees 85, 95Stireman, J.O. 256Stirling, W.L. 123Stone, C. 103Stork, N. 149, 150Stork, N.E. 83, 147, 150–1, 153Story, T.P. 128stovepipe (column) sampler 196Strickland, A.H. 229Strickler, J.D. 64–5Strong, D.R. 17, 23

INDEX 299Su, N.Y. 231–2, 239–40suction samplers, types of 59–60, 95, 257suction sampling 17, 59–62, 75, 95, 270suction traps 4–5, 117, 120–5, 131calibration of catches 123–4designs and modifications 121factors influencing catches by 124–5index of methods and approaches 144modified designs 123Onderstepoort Veterinary Institute design126standard Rothamsted model 121–3, 122use in indirect sampling 257use of baits 125wind speed effect on 123–4Sugihara, G. 212Sugimoto, A. 221Sugio, K. 224Sulawesi 269Sunderland, K.D. 38, 258, 266, 267Surber sampler 189suspended soils 159–60index of methods and approaches 167Sutherland, W.J. 14, 61Svensson, B.W. 210swarms 65Sweden 87, 105sweep nets 10, 62–3, 75, 187, 269–70comparative efficiency 69, 196compared to vacuum sampling 61in indirect sampling 257undetected 137sweeping 149, 154, 265Swietenia 90mahagoni 90Swingfog ® 151Switzerland 148Swormlure 133Synacanthotermes 223Syrphidae 67, 76, 257in sticky traps 130in tree holes 172, 174Tabanidae 65, 75Tabanus 138Tachyporus 60, 75tadpoles 191in tree holes 173–4Takematsu, Y. 224Tallamy, D.W. 138Tamaki, G. 266, 267Tanglefoot ® 95Tauber, C.A. 257Tavakilian, G. 100Tayasu, I. 221Taylor, H.S. 231Taylor, J.R. 194, 206Taylor, L.R. 117, 120–1, 124, 130Taylor, R.A.J. 137Teleogryllus commodus 21Terell-Nield, C. 51Termes 223Termigard ® 239Termite Eradication Programme, UK241termites 79acoustic emissions from 233alates 232index of methods and approaches250–3methods for sampling 221–53monitoring and baiting program, Devon,UK 240–3nest units of 224population density of 233–6processional columns of 233recording movements of 233sampling by transect protocol 234–5,235–6as subterranean pests of buildings236–40swarming 232wood boring 80Termitidae 223, 234Apicotermitinae 223Macrotermitinae 223, 224, 231Nasutitermitinae 223, 224Termitinae 223, 224Termopsidae 224Tetanops myopaeformis 23, 128Tetropium 79Thiele, H.-U. 37–8, 50thigmotaxis 13Thomas, C.F.G. 42Thomas, J.D.B. 38Thompson, D.V. 129Thoms, E.M. 233Thoracotermes 223Thorne, B.L. 232Thornhill, E.W. 59

300 INDEXThornhill vacuum sampler 59Thorp, R.W. 128–9Thorpe, K.W. 70, 71thrips 77Thysanoptera 257Tilmon, K.J. 264Tipulidae 18–19, 24larval behavior 19–20larval sampling 18–19, 26Tobin, S.C. 147–8Tomicus 79, 87piniperda 76, 79, 87, 87, 88Topping, C.J. 38, 50Tortricidae 86, 89tow nets 124–5, 137–8, 187Townes, H. 64, 269Townsend, C.R. 267Townsend, M.L. 232transects 188, 264linear 46–7, 226plough 23sweeping along 62termite protocol 234–5trap night index (TNI) 159traps 4–5active 5, 66, 150index of methods and approaches 219aerial and arboreal 157–9index of methods and approaches166–7baited 67, 121, 132–6animals as bait in 133–4designs and modifications 132“graveyard” trials 231index of methods and approaches145wind effect on 135–6bamboo 175barrier 45basin 12beating trays 69, 76, 83, 257Berlese–Tullgren funnel 26, 229Blasdale version 26bicolored 65cardboard (corrugated) 238–9carrion 132chemical knockdown 68, 83, 150–5collecting hoops 154, 154collecting trays 153–4comparative advantages of 154–5fogging 151–2, 269index of methods and approaches 166mistblowing 152–5; HurricaneMajor ® mistblower 152–3; Stihl ®mistblower 153; pyrethrum 150circle 96collecting hoops 154, 154collecting pots 95–6color 76spacing of 66–7combined 129, 157, 158crawl 96Delta 134–5dippers 188, 195mosquito 188drainpipe 107drift fences 45–6drift nets 192dung 132effect of moonlight on 159emergence 17, 157–9, 192flight-interception 63foliage bagging 68, 76funnel 12, 96, 99, 105–7, 109, 138, 229Gee see minnowgutter 45Heath pattern 118installation, disturbance caused by 49interception 4, 63, 232, 269composite 64, 158index of methods and approaches145undetected 137–8use in forest canopy 157–9, 158visible 138window 65light 4, 116–21, 129, 157–9, 187, 232aquatic 195–6“black” 117, 119comparative catchability 117design of 118–19disadvantage of 159examples of use 120index of methods and approaches 144influences on catch efficiency andeffective catching distance 117–18mercury vapor 17, 117Onderstepoort Veterinary Institutedesign 126principles of use 116tungsten 17, 117–18types of light source 117

INDEX 301Lindgren 109lobster-pot 96, 118Malaise 4, 75, 158, 232, 269aerial and arboreal use of 149, 157–9basic design for 63–5, 138Cornell design 64Coupland design 133Townes model 64minnow 187, 195, 208Gee model 192New Jersey pattern 118pan 5, 257passive 5index of methods and approaches 220patterns for 46–7Pennsylvania pattern 159Pherocon ® 135pheromone 12, 17, 107, 108–9, 134–6,135combined 121, 129uses of 136pipe 107, 108Pirbright pattern 125pitfall 6, 37–57, 98, 128, 269baits used in 44–5design and application 39–45, 40, 56funnels in 42, 56killing agents, preservatives, anddetergents 42–4, 43, 48, 56materials used in 40–1, 56rims of 42, 56roofs of 41–2, 56shape and size 41, 56spatial arrangement 47specialized designs of 45, 56subterranean 45types of 39use of baits in 44–5, 56uses of 38ramp 45Robinson pattern 118, 159Rothamsted pattern 118, 119, 122, 124,159data series 120sticky 4, 17, 129–31, 269aerial and arboreal use of 157baited, effectiveness of 107, 108basic design and use 124–5, 130comparative efficiency 131Delta pattern 134importance of shape and color 131index of methods and approaches 145mini-sticky 85modifications 130–1use on trees 85, 95stovepipe (column) 196suction 4–5, 117, 120–5, 131calibration of catches 123–4designs and modifications 121factors influencing catches by 124–5index of methods and approaches 144modified designs 123Onderstepoort Veterinary Institutedesign 126standard Rothamsted pattern 121–3,122use in indirect sampling 257use of baits 125wind speed effect on 123–4sweep nets 10, 62–3, 75, 187, 269–70comparative efficiency 69, 196compared to vacuum sampling 61in indirect sampling 257undetected 137time-sorting 45tow nets 124–5, 137–8, 187trawl 120trunk 96–7, 107window 96–7, 98water (pan) 4–5, 66, 67, 125–9, 127, 131basic design and use 127–8importance of pan color 128index of methods and approaches 144modifications 128types of use 128–9; in indirectsampling 257window 4, 65–6, 76, 106, 107–8, 137,157yellow water 4, 257, 269trawl traps 120tree holes 169–70abiotic variables in 179analogues of 175–6artificial 174–80, 176–8important considerations for 179–80sampling techniques for 178–9classification of 168–9comparative data, problems ofinterpretation 181index of methods and approaches 185natural 168–74experimental considerations 173–4

302 INDEXtree holes, natural (cont’d)sampling techniques for 170–3species diversity with height 174water-filledsampling of 168–85statistical methods for 180treesindex of methods and approaches114–15major stem habitats on 78sampling from shoots, stems, and trunks77–115Treloar, A. 62, 137Tretzel, E. 50Trexler, J.C. 187, 191–2, 196Trexler, J.D. 192Trialeurodes vaporarium 131Trichoplusia ni 118Trichoptera 138night-flying 117and ultraviolet light 117Trichosirocalus troglodytes 63Trifolium repens 18Trinervitermes 228Trinidad 133tropics 146trunk traps 96–7window traps 96–7, 98Trypodendron 79–80Tuberculitermes 223Tuberculoides annulatus 131Tuck, K.R. 120Tullgren extraction 27compared to flotation 28Tullgren funnel 26–8, 61, 159Tunset, K. 137Tupy, J.L. 110Turchin, P. 105, 105Turner, A.M. 187, 191–2, 196Twombly, S. 210Uetz, G.W. 37–8, 42–3, 46–7ultrasound 110Unguitermes 223United States 83, 96, 109, 118, 132, 136,146–7, 232, 239Arizona 81Missouri 20north-western 148Oregon 89, 109, 109Unzicker, J.D. 37–8, 42–3, 46–7Usher, M.B. 66–7, 231vacuum sampler 60vacuum sampling 59–62, 69, 75, 265of trees and shrubs 70Vale, G.A. 133van den Berg, H. 266, 267van den Berghe, E. 44Van Driesche, R.G. 255, 259, 259, 260–3van Huizen, T.H.P. 137van Lenteren, J.C. 259van Roermund, J.W. 259Vance, G.M. 192Vandenberg, N. 257Varga, L. 168variance–mean ratios 7–8Varis, A.L. 43–5Varley, G. 96Veliidae 174Vent-Axia ® fan 123Verkerk, R.H.J. 221–53Verrucositermes 223Vessby, K. 132video technology for in situ observation 29Villani, M.G. 29–30Villemant, C. 269visualization methods see laboratoryvisualization methodsVitale, G. 169Vlijm, L. 42Vogt, W.G. 132von Ende, C.N. 182Waage, B.E. 43Waage, J.K. 264Wagner, T. 147Wainhouse, D. 86, 93, 136, 146Walker, E.D. 64–5, 171, 179–82Walsh, G.B. 44Walter, D.E. 24–6, 146, 151Wang, B. 263Ward, R.H. 20Ward, S.A. 11Waring, P. 117Washburn, J.O. 179Washino, R.K. 188, 192, 195wasps 104chalcid 258distortion by adhesive 129

INDEX 303fig 157gall 79, 81–2wood 79–80, 114–15Watanabe, H. 147water chemistry 171–2water (pan) traps 4–5, 66, 67, 125–9, 127,131basic design and use 127–8importance of pan color 128index of methods and approaches 144modifications 128types of use 128–9in indirect sampling 257Watt, A.D. 1–15, 221Way, M.J. 94, 94Wearing, C.H. 83Webb, R.E. 95, 131weevils 63, 80, 86, 88, 96black vine 17large pine 6, 11pea and bean 17pecan 96smaller pecan 29white pine 88Weissling, T.J. 233Welsh, A.H. 207Weseloh, R.M. 265Weslien, J. 105–6West Africa 221–2, 223wet-sieving 18, 28, 30Wheater, C.P. 268White, E.G. 17White, T.C.R. 69whitefly 265greenhouse 131Whitfield, G.H. 23Whitfield, J.B. 269Whitmore, R.C. 156Whittaker, T.M. 61Wigglesworth, V.B. 63Wilcox, C. 191Willett, T.R. 268Williams, C.B. 118Williams, J.B. 269Williams, L. 264Williams, P. 21Willoughby, B.E. 23willow 13Wilson, E.O. 221Wilson, L.F. 22Wilson, L.T. 62Winchester, N.N. 64, 146–8, 156, 158–60wind-speed effect 124–5window traps 4, 65–6, 76, 106, 107–8, 137,157windthrow 65Winkler extraction 159wireworms 20, 20larval bait for 21Wise, I.L. 62–3Wittman, P.K. 146Woiwood, I.P. 120–1, 159Wood, S.N. 210Wood, T.G. 221–2, 225–6, 228–9Woodcock, B.A. 37–57“woodworm” 78, 80, 110Wratten, S.D. 50Wright, A.F. 59–61Wright, R.J. 29Wylie, F.R. 77, 89, 94Wyss, E. 257Wytham Wood study 96X-ray analysis 29, 110X-ray computed tomography 29Xylechinus 101pilosus 102Xyleutes 80ceramica 110Xyloterus 103lineatus 102Yamamura, K. 258Yamashita, Z. 153Yanoviak, S.P. 168–85Yasuda, K. 45Yeargan, K.V. 24Yee, W.C. 133yellow water traps 4, 257, 269yield–effort curves 214Young, M. 116–45Young, M.R. 117, 128, 137Zaidi, R.H. 268Zar, J.H. 194, 210Zhang, Q.H. 102, 102, 103Zhou, X. 123Zhu, Y.C. 264Zolubas, P. 108

Sampling Insects from Trees - LAC Biosafety

Create successful ePaper yourself

Delete template?

Save as template?