REVIEWS Mistletoes as parasites: host specificity and speciation David A. Norton and Margaret A. Carpenter t has been estimated that c. 1% Recent research on parasite evolution has a single host species, and such instances may reflect a poor highlighted the importance of host of all angiosperm species are knowledge of the parasite's range specialization in speciation, either parasitic and that about 40% of rather than specialization on one through host-switching or cospeciation. plant parasites are shoot paraMany parasites show common patterns of host 11. The usual pattern among sites, parasitizing the abovespecialists is a single common host specificity, with higher host ground parts of their host plants, host, and a number of other hosts specificity where host abundance is high while the other 60% are root paraless frequently parasitized. The sites 1. Although much of the para- and reliable, phylogenetically conservative site literature focuses on animal host specificity, and formation of races on common host is required to sustain the parasite population, as or in different host species. Recent parasites, plant parasites are ecothe sporadic hosts alone are not advances in our understanding of host logically and economically signifisufficient for this although they specificity and speciation patterns in a cant 1-3 and share many features in may be important in the maintevariety of animal parasites provides common with animal parasites. nance of genetic variation within valuable insights into the evolutionary Parasites can be defined as organthe parasite population 12.Generalbiology of mistletoes. isms that complete a whole stage ists, which infect a large number of their life associated with a single of host species, tend not to be host individual in a relationship David Norton and Margaret Carpenter are in the totally unrestricted in their host that is beneficial to the parasite Conservation Research Group, School of Forestry, range and show preference for but not to the host 4,5. This definiUniversity of Canterbury, Private Bag 4800, some host species above others. tion includes all plant parasites, Christchurch, New Zealand Unlike most angiosperm root viruses, some phytophagous (d.norton@fore.canterbury.ac.nz). parasites, which typically have insects, parasitoids, and ecto- and broad host ranges 1, mistletoes endoparasites of animals such as show a variety of patterns of host lice and liver flukes. specificity from generalist to speMistletoes (Box 1), the predomicialist 13 in a similar manner to many animal parasites. For nant group of angiosperm shoot parasites, are a fascinating example, in the New Zealand loranthaceous mistletoe flora, and diverse group of plants found in a wide range of ecosystems including boreal forests, tropical rainforests and arid Alepis flavida, Peraxilla colensoi and P. tetrapetala are spewoodlands. While seed of many plant parasites germinates cialists, primarily parasitizing different species of Nothofagus only in response to chemical signals from host plants 1, (southern beech), although they have occasionally been recorded as parasites on other plant species 14. In contrast, mistletoe seeds germinate readily in almost all situations. lleostylus micranthus (Fig. lb) and Tupeia antarctica are genHowever, the key limiting step in a mistletoe's life cycle is establishment, which is dependent on an appropriate diseralists and have been recorded parasitizing a large number of host species TM. Similar variation in host specificity can be perser, deposition on a suitable sized branch, and mistletoe-host compatibility6,7. In having such tight establishseen in other mistletoe flora such as in Australia 15. ment requirements, mistletoes have much in common with An emerging trend in host specificity in several unremany animal parasites. Like other plant and animal paralated parasite groups follows the latitudinal climate gradisites, mistletoes also live in an intimate association with ent from temperate to tropical regions. Data from different their hosts and derive nutrition from the host, and, of types of parasite (e.g. digenean trematodes 9 and parasitoid course, share a life-long association with a single host Ichneumenidae 16) indicate greater host specificity in temindividual. perate regions. This pattern is also clearly evident in loranMistletoes, as with other plant parasites 8 have recently thaceous mistletoes, which show low host specificity in been described as both agricultural pests and as threatened species in different parts of the world 7. If we are to manage these species appropriately, it is important that we underBox 1. Mistletoes - a polyphyletic group of shrubby aerial stem parasites stand such basic aspects of their biology as patterns of host specificity and the ways in which they speciate. These asThe Loranthaceae (Figs la,b) are the largest family within the mistletoes with c. 950 species, while the Viscaceae (F~g. lc) contains c. 365 species6. The Loranthpects of mistletoe biology are, however, only beginning to aceae are predominantly southern or tropical in their distribution, suggesting a be understood. Recent advances in our understanding of Gondwanan ancestry, while the Viscaceae are more common in northern latihost specificity and speciation patterns in a variety of anitudes 4o. The two groups have probably been derived independently from nonmal parasites provide valuable insights into similar patterns mistletoe ancestors. All mistletoes derive water and nutrients by tapping the host in plant parasites and especially in mistletoes, but also highxylem but differ in their dependence on the host for carbon. Dwarf mistletoes (Arceuthobium and Korthalsella, Viscaceae) are considered to be primarily heterolight key areas where our knowledge is still limited for these trophic, tapping the host phloem for carbon compounds2L Other mistletoes are groups. I Patterns of host specificity Parasites vary in their host specificity. Groups of parasites tend to include a spectrum from highly host-specific through to host generalistsg, 10.Few are known to infect only T R E E v o L 13, n o . 3 M a r c h 1998 usually regarded as autotrophic, depending on their host for water and inorganic nutrients only, although even in these mistletoes there is evidence for carbon uptake from the host 6. While dwarf mistletoes have explosively dispersed seeds, most mistletoes are reliant on birds for dispersal and have developed close asscciations with particular bird groups for dispersal 6. Copyright © 1998, Elsevier Science Ltd. All rights reserved. 0169-5347/98/$19.00 PII: S0169-5347(97)01243-3 101 REVIEWS heterogeneous tropical rainforests (e.g. in New Guinea) and high host specificity in relatively species poor temperate forests (e.g. in eastern Australia and New Zealand13). An interesting exception to the latitudinal pattern of host specificity is also an exception to the latitudinal pattern of species richness: tropical mangrove forests contain few host species which mistletoes can parasitize, and in these areas mistletoes show higher host specificity L7. This latitudinal pattern can be explained in terms of host abundance. As species richness in many groups is greater in the tropics, most tropical areas do not have a high relative abundance of any one potential host species, making it difficult for a parasitic species to use one host exclusivelyTM. In contrast, many temperate ecosystems contain plant and animal communities in which usually only a few species dominate, whose abundance makes specialization a more viable option. Host specificity of phytophagous insects also appears to be related to abundance and reliability of host plants TM. The relationship between abundance and specificity has also been illustrated in a recent community study involving five trophic levels (plant, phytophage, parasite, parasitoid and hyperparasite) in which specialization was greater at the lower trophic levels correlating with a greater relative abundance of host species~9. It has been suggested that as host species become increasingly scarce, they are unable to support specialized parasites (resource fragmentation hypothesis1% For each parasite species there must be a threshold of absolute host abundance required to support the parasite population. This host population will be made up of a few species if their density is sufficient, otherwise a wider range of hosts will be used to make up the required host numbers. Of course, a parasite can only specialize on a host it is able to use in the first place, so host specificity is in fact related to the abundance of suitable potential hosts, as determined by biochemical and physiological constraints. Host specialization is thought to be favoured by the advantages of adapting to interact more profitably with a frequently encountered host. Given sufficient abundance of such a host, the benefits of specializing on that host outweigh the disadvantages of interacting less well with other potential hosts~8,20. Probably the most significant aspects of host specialization are those that increase the parasite's efficiency in capturing resources from the host 20,21. Also important is the ability to overcome host defenses 9. Parasites may also specialize on a host where they are less susceptible to predation, providing the impetus for evolution of cryptic mimicry of the host 13,22. For some parasites, host specialization may facilitate the location of potential mates or the potential for pollination and dispersal by making them easier to find 9,2°. Being a host generalist can also be advantageous, especially in a heterogeneous community, as it allows a parasite to grow successfully in or on many of the potential hosts encountered. If host populations are unpredictable and ephemeral, generalist parasites are also more likely to occur 4. The degree of host specificity can be seen as an equilibrium of two opposing drives: (1) to use a maximum number of the hosts encountered, and (2) to make best use of the most frequently encountered hosts. However, as the number of hosts increases, the probability that a specialist parasite can locate a suitable host decreases, reducing the advantages gained through specialization. Thus, relative host abundance is the key to host specificity. As relative host abundance is variable, host specificity must be seen as dynamic, variable in both space and time, and dependent on "102 the opportunities available in any one time or place 4.~3.Not only does the level of specificity change but also the identity of the host species used. The changeable nature of host specificity has been demonstrated repeatedly by parasites which become pests on introduced species, especially crops grown in extensive monocultures 2,4.Even among mistletoes, some species have been able to widen their host range with the arrival of new potential host species into a country. For example, the New Zealand mistletoes lleostylus micranthus and Tupeia antarctica parasitize a large number of host trees that have been introduced into New Zealand since European settlement 150 years ago TM. In the case of Tupeia, the introduced tree Chamaecytisus palmensis is now the main host in several parts of New Zealand as habitat loss has removed much of the indigenous vegetation. In determining the host range of a parasite, dispersal is an important influence on the potential hosts which the parasite contacts and therefore has the opportunity to use as hosts. Mistletoes, with the exception of the explosively dispersed dwarf mistletoes, are almost totally dependent on avian dispersers 6 and can only establish and grow on the host plants where their avian dispersers deposit seeds. Mistletoes are therefore only likely to develop specificity to hosts on which they are frequently deposited ~2. Monteiro et al. 24 demonstrated the importance of an avian disperser for the loranthaceous mistletoe Psittacanthus robustus which occurred on a secondary host only when present under the crown of the primary host. Aspects of the dispersers' behaviour, such as the way they use different potential host plants and the length of time between ingestion and egestion of a mistletoe seed, influence the probability of dispersal to the same or a different host 25,26. There is a tendency among parasites that infect more than one host species to infect closely related hosts - that is, species within the same genus or family. For example, many species of the dwarf mistletoe Arceuthobium parasitize only one of the two subgenera of Pinus 27. Closely related parasites also tend to infect closely related host species. For example, six of the eight species of Lysiana, an endemic Australian loranthaceous genus, parasitize Acacia species ~5. Similarly, closely related Australian species within the large loranthaceous genus Amyema parasitize hosts in the same genus ~7. Many parasites described as host generalists have, on closer inspection, been shown to actually be specialists at a local level; that is, over the whole of the parasite species range they might use a large number of host species, but are specific to only a few hosts in any one area. This has been shown for North American mistletoes inArceuthobium 27and Phoradendron 28, and recently for the Australian mistletoe Amyema miquelii 2~..The occurrence of local host specificity suggests that differences in host utilization may be genetically based as differential success of individuals of one mistletoe population when grown on a different host species has been shown. For example, Clay et al. 28 showed significant differences in haustorial disk formation when Phoradendron tomentosum individuals were swapped between host species, with greatest success when they were grown on the source host species. These differences lead to the formation of different races of the parasite that may appear morphologically similar but are distinct in their host utilization. Mechanisms of parasite speciation Recent reviews of parasite evolution have discussed the role of host specificity in parasite speciation4, lo.u. Both the formation of distinct parasite races and the tendency for T R E E vol. 13, no. 3 M a r c h 1998 REVIEWS p f " s , Fig. 1. (a) Peraxilla tetrapetala (Loranthaceae), a large, leafy host-specialist mistletoe with fleshy bird-dispersed fruit. This mistletoe most commonly parasitizes tall Nothofagus solandri trees in cool temperate New Zealand rainforests, although it has been recorded from 16 other host taxa14. (b) Ileostylus mieranthus (Loranthaceae), a leafy host-generalist mistletoe parasitizing a wide range of shrub and tree genera throughout New Zealand, and also on Norfolk Island, photographed here parasitizing Coprosmapropinqua. This species has been recorded from 209 different host taxa14, although it primarily parasitizes trees and shrubs in the genera Coprosma, Podocarpusand Pittosporum. (c) Korthalsellaclavata(Viscaceae), a primarily heterotrophic mistletoe that parasitizes a range of shrub genera in New Zealand, photographed here parasitizing Coprosmapropinqua.The dwarf mistletoes differ from most other mistletoes in that their seeds are released from the pericarp either explosively (Arceuthobium)or spontaneously (Korthalsella), and birds are only involved in subsequent seed dispersal. closely related parasites to infect closely related hosts appear to be important in speciation by specialization on different hosts. The formation of parasite races involves genetic changes which are part of adaptation to improved growth on a particular host. Parasite races are likely to form when the gene flow between parasite populations is diminished by factors such as distance, limited range of dispersal, and patchy host populations. This limited gene flow over a substantial period is likely to lead to allopatric speciation. However, speciation through parasite race formation is really only part of another process, that of host-switching. A host-switch is a two-step process. First, the parasite must include a new host in its range. This requires preadaptation to that host, either through phylogenetic similarity to its existing host or through a chemical or ecological similarity which fortuitously allows the parasite to use the new host. Second, the parasite has to adapt to the new host in a way that restricts itself from previous hosts (parasite race formation). Thus, host-switching involves both an initial decrease in host specificity and a subsequent increase in host specificity as specialization to the new host occurs. Without both these steps, the change is not a host-switch but rather a contraction or expansion of the existing host range• As host-switching involves both expansion and contraction of the host range, it can occur without producing any general trend towards either high or low host specificity. A host-switch could also be seen as a change in the most frequently used host, such that a previously minor host becomes the major one. This is likely to happen if the relative abundance of the hosts changes (e.g. as a result of climatic or other environmental changes) or if the parasite expands its range into an area where relative host abundances differ to those normally encountered. In this way, selection for specialization on the most abundant host will have different outcomes, initially producing two parasite races and potentially producing distinct species with time. Parasite speciation can also occur in response to host speciation. Closely related parasites tend to use closely related hosts, which can lead to the parasites and their hosts having congruent phylogenies30.This tendency (known as Farenholz rule) has received considerable recent attention from parasitologists, through use of parasite phylogenies to help establish host phylogenies, and vice versaaL The rule also has important consequences concerning parasite speciation, as it suggests that parasites and their hosts cospeciate 4. TREE vol. 13, no..3'March 1998 Investigations of the Farenholz rule have revealed that cospeciation does o c c u r 31,32, although host-switching events can also be common 33,34.Two studies have suggested that both events are approximately equally common among parasite groups u,3s. It has been suggested that the frequency of cospeciation tends to be higher if host specificity is also high, because host-specific parasites are frequently phylogenetically conservative in their host choice 9. Cospeciation seems less likely to occur in a generalist parasite, as speciation of one host would not be a strong influence on a parasite using several other hosts. However, if populations of both the parasite and its hosts became geographically isolated, cospeciation through allopatric speciation might occur, irrespective of levels of host specificity. The prevalence of cospeciation probably varies considerably between different groups of parasites, but may be influenced by aspects of the parasite's ecology such as host specificity and mechanisms of dispersal to new hosts 36. There are a few problems associated with testing the Farenholz rule. One is the difficulty in distinguishing a cospeciation event from a host-switch to a closely related host, especially as this is likely to be the kind of host-switch which would occur most readily due to similarities between closely related hosts. A second difficulty is that the hostparasite association does not always reflect the phylogenetic relationships even in the absence of host-switching. This can occur if some host speciation events are not matched by speciation in the parasite 30. Parasite speciation resulting from changes regarding the host can therefore occur in two ways: parasite cospeciation with their hosts and parasite speciation by changing host specificity. In addition, parasite speciation can also occur without changes in the host. Each of these modes of parasite speciation could occur following the allopatric, peripheral isolates or sympatric models of speciation11, 37. Speciation patterns in mistletoes Speciation in mistletoes is likely to follow the same patterns as in other parasites - cospeciation with hosts and speciation by host-switching. Comparison of mistletoe and host phylogenies has the potential to describe the histories of host associations for particular mistletoe groups and tell us about the speciation patterns in mistletoes, particularly regarding the relative importance of cospeciation and hostswitching. There are, however, only limited data available on mistletoe and host phylogenies to do this. The dwarf 103 REVIEWS Juniperus Abies D D H H 51 Picea Acknowledgements D We t h a n k Hazel C h a p m a n , P e t e r d e Lange, A a r o n Liston, Dan Nickrent, J a k e O v e r t o n , Adrian P a t e r s o n , A s h l e y S p a r r o w a n d t h r e e a n o n y m o u s r e f e r e e s for v a l u a b l e c o m m e n t s on this paper. This r e s e a r c h was funded by New Zealand F o u n d a t i o n for R e s e a r c h , S c i e n c e a n d T e c h n o l o g y grant UOC510. Pseudotsuga 6 Abies 7 8 9 H D H, Picea D D D D D D m 14 D H D Fig. 2. Mapping of the genus or subgenus of principal hosts27 on to the phylogenetic tree of Arceuthobium38with minimum host-switches. H = Pinus subgenus Haploxylon. D = Pinus subgenus Diploxylon. Numbers represent host-switches, Phylogenyfrom Ref. 38, with permission. mistletoes of the genus Arceuthobium are probably the bestknown group, and a molecular phylogeny has been recently produced including some members of this genus 38. Most of the hosts used by Arceuthobium are in the genus Pinus, although a few other genera are also used 27. A phylogeny of the hard pines (Pinus subgenus Diploxylon), the subgenus which includes the majority of Arceuthobium hosts, has recently been published by Krupkin et al. 39 Using the principal host associations described by Hawksworth and Wiens 27, the Arceuthobium and hard pine phylogenies can be compared; this comparison shows little evidence for cospeciation, suggesting that host-switching is the more common of the two processes among Arceuthobium. Mapping of the genus or subgenus of the principal host (as stated by Hawksworth and Wiens 27) on to the phylogeny of Arceuthobium 38allows estimation of the evolution of host associations of Arceuthobium (Fig. 2). Arceuthobium has probably evolved primarily as a parasite of Pinus, particularly of the subgenus Diploxylon. The most parsimonious arrangement involves nine switches between host genus or subgenus. Conclusions Patterns of host specificity and the ways in which mistletoes spe~iate certainly warrant further research. The role of relative host abundance in determining host specificity could be evaluated by examining the abundance of host species in regions where different races of mistletoes occur, to see if host preference is related to host species abundance. This might also allow estimation of the threshold of absolute host abundance required to sustain a mistletoe population. Analysis of the genetic variation within and between mistletoe races, compared to that occurring between mistletoe species, might give us indications of incipient speciation 104 and allow estimation of how far two populations have diverged along the road to speciation. A biogeographic analysis would determine whether genetic similarity was related to geographic proximity or host species. Phylogenetic comparison of mistletoes and their hosts would reveal the relative importance of cospeciation and host-switching events in mistletoe speciation. References 1 Musselman, L.J. and Press, M.C. (1995) Introduction to parasitic plants, in Parasitic Plants (Press, M.C. and Graves, J.D., eds), pp. 1-13, Chapman & Hall 2 Parker, C. and Riches, C.R. (1993) Parasitic Weeds of the World. Biology and Control, CAB International 3 Pennings, S.C. and Calloway, R.M. (1996) Impact of a parasitic plant on the structure and dynamics of salt marsh vegetation, Ecology 77, 1410-1419 4 Thompson, J.N.(1994) TheCoevolutionaryProcess, Universityof Chicago Press 5 Douglas, A.E. (1994) Symbiotic Interactions, Oxford University Press 6 Reid, N., Stafford Smith, M. and Yah, Z. (1995) Ecology and population biology of mistletoes, in Forest Canopies (Lawman, M.D. and Nadkarni, N.M., eds), pp. 285-310, Academic Press 7 Norton, D.A. and Reid, N. (1997) Lessons in ecosystem management from management of threatened and pest loranthaceous mistletoes in New Zealand and Australia, Canserv. Biol. 11, 759-769 8 Marvier, M.A.and Smith, D.L. (1997) Conservation implications of host use for rare parasitic plants, Conserv. Biol. 1I, 839-848 9 Rohde, K. (1993) Ecology of Marine Parasites, CAB International 10 Shaw, M.R. (1994) Parasitoid host ranges, in Parasitoid Community Ecology (Hawkins, B.A. and Sheehan, W., eds), pp. 111-144, Oxford University Press 11 Brooks, D.R. and McLennan, D.A. (1993) Parascript: Parasites and the Language of Evolution, Smithsonian Institution Press 12 Atsatt, P.R. (1983) Host-parasite interactions in higher plants, in Physiological Plant Ecology 1I~ Responses to the Chemical and Biological Environment (Lauge, O.L. et aL, eds), pp. 519-535, Springer-Verlag 13 Barlow, B.A. and Wiens, D. (1977) Host-parasite resemblance in Australian mistletoes: the case for cryptic mimicry, Evolution 31, 69 -84 14 de Lange, P.J., Norton, D.A. and Molloy, B.P.1 (1997) Checklist of New Zealand loranthaeeous hosts, in New Zealand's Loranthaceous Mistletoes (de Lange, P.J. and Norton, D.A., eds), pp. 83-104, New Zealand Department of Conservation 15 Barlow, B.A. (1984) Loranthaceae, in Flora of Australia VoL 22 (George, A.S., ed.), pp. 68-131, Australian Government Printing Service 16 Janzen, D.H. (1981) The peak in North American lchneumonid species richness lies between 38 ° and 42 ° N, Ecology 62,532-537 17 Barlow, B.A. (1992) Conspectus of the genus Amyema Tieghem (Loranthaceae), Blumea 36, 293-381 18 Bernys, E.A. and Chapman, R.F. (1994)Host-PlantSelection by Phytophagous Insects, Chapman & Hall 19 Dawah, H.A., Hawkins, B. and Claridge, M.F. (1995) Structure of parusitoid communities of grass-feeding chalcid wasps, 1 Anita. Ecol. 64, 708-720 20 Jaenike, J. (1990) Host specialization in phytophagous iusects, Annu. Rev. Ecol. Syst. 21,243-273 21 Kearn, G.C. (1994) Evolutionary expansion of the monogenea, Int. J. Parasitol. 24, 1227-1271 22 Bernys, E. and Graham, M. (1988) On the evolution of host specificity in phytophagous arthropods, Ecology 69. 886-892 TREE vol. 13, no. 3 March 1998 REVIEWS 23 Kennedy, C.R. (1975) Ecological Animal Parasitology, Blackwell 24 Monteiro, R.F., Martins, R.P. and Yamamoto, K. (1992) Host specificity and seed dispersal of Psittacanthus robnstus (Loranthaceae) in south-east Brazil, I Trop. Ecol. 8, 307-314 25 Murphy, S.R. et al. (1993) Differential passage time of mistletoe fruits through the gut of honeyeaters and flowerpeckers: effects on seedling establishment, Oecologia 93, 171-176 26 Overton, J.McC. (1994) Dispersal and infection in mistletoe metapopulations, J. Ecol. 82, 711-723 27 Hawksworth, F.G. and Wiens, D. (1996) Dwarf Mistletoes: Biology, Pathology, and Systematics, US Department of Agriculture 28 Clay, K., Dement, D. and Rejmanek, M. (1985) Experimental evidence for host races in mistletoe (Phoradendron tomentosum), Am. J. Bot. 72, 1225-1231 29 Norton, D.A., Hobbs, R.J. and Atkins, L. (1995) Fragmentation, disturbance, and plant distribution: mistletoes in woodland remnants in the Western Australian wheatbelt, Conserv. Biol. 9, 426-438 30 Hennig, W. (1966) Phylogenetic Systematics, University of Illinois Press 31 Hafner, M.S. and Nadler, S.A. (1988) Phylogenetic trees support coevolution of parasites and their hosts, Nature 332,258-259 32 Paterson, A.M, Gray, R.D. and Wallis, G.P. (1993) Parasites, petrels and penguins; does louse presence reflect seabird phylogeny? Int. i Parasitol. 23, 515-526 33 Brown, JM. et al. (1994) Phylogeny of Greys (Lepidoptera: Prodoxidae), based on nucleotide sequence variation in 34 35 36 37 38 39 40 mitochondrial cytochrome oxidase I and II: congruence with morphological data, Mol. Biol. Evol. 11,128-141 Briese, D.T., Espiau, C. and Pouchet-Lermans, A. (199(;) Micro-evolution in the weevil genus Larinus: the formation of host biotypes and speciation, Mol. Ecol. 5, 531-545 Humphries, C.J., Cox, J.M. and Nielsen, E.S. (1986) Nothofagus and its parasites: a cladistic approach to coevolution, in Coevolution andSystematics (Stone, A.R. and Hawksworth, D.L., eds), pp. 55-76, Clarendon Press Paterson, AM. and Gray, R.D. (1997) Host-parasite cospeciation, host-switching and missing the boat, in Host-Parasite Evolution" General Principles and Avian Models (Clayton, D.H. and Moore, J.. eds), pp. 236-250, Oxford University Press Price, P.W. (1980) Evolutionary Biology of Parasites, Princeton University Press Nickrent, D.L, Schuette, K.P. and Starr, EM. (1994) A molecular phylogeny of Arceuthobium (Viscaceae) based on nuclear ribosomal DNA internal transcribed spacer sequences, Am. J. Bot. 81, 1149-1160 Krupkin, A.B., Liston, A. and Strauss, S.H. (1996) Phylogenetic analysis of the hard pines (Pinus subgenus Pinus, Pinaceae) from chloroplast DNA restriction site analysis, Am. J Bot. 83, 489-498 Barlow, B.A. (1983) Biogeography of Loranthaceae and Viscaceae, in The Biology of Mistletoes (Calder, M. and Bernhardt, P., eds), pp. 19-46, Academic Press Phylogenetic supertrees: assembling the trees of life Michael J. Sanderson, Andy Purvis and Chris Henze marks in cells where taxa from one Systematists and comparative biologists espite the recent exploanalysis have not been scored in commonly want to make statements sive growth in phylogenanother 4 (Fig. 1). This 'superabout relationships among taxa that have etics, the number of never been collectively included in any matrix' approach has the advanspecies included in physingle phylogenetic analysis. Construction tage that the information retained logenies to date is still an insignifiin individual characters can help of phylogenetic 'supertrees' provides one cant fraction of biodiversity. Moresort out relative strengths and solution. Supertrees are estimates of over, most individual studies phylogeny assembled from sets of smaller weaknesses within and between sample only a few taxa (usually the different data sets 5, which is estimates (source trees) sharing some under 50), so that our current philosophically in keeping with but not necessarily all their taxa in understanding of the tree of life is common. If certain conditions are met, the so-called 'total evidence' fragmentary. More inclusive physupertrees can retain all or most of the approach to combining phylologenetic hypotheses are highly desirable: systematists wish to be information from the source trees and also genetic information 6. make novel statements about However, as a long-term stratas comprehensive as possible in relationships of taxa that do not co-occur egy for assembling ever larger making statements about phylogeon any one source tree. Supertrees have phylogenies, reliance on the connetic relationships, and comparacommonly been constructed using struction of a character supertive biologists often study sets of subjective and informal approaches, but matrix is untenable. If only a few taxa that do not correspond neatly taxa are common between data several explicit approaches have recently to sets found on available phybeen proposed. sets, most of the newly combined logenies, so are forced to cobble data matrix will be scored as questogether disparate phylogenies 1-3 tion marks. Gathering the new into a single tree. Any such tree Michael Sanderson and Chris Henze are at the data needed to fill in the gaps containing all the taxa from a colSection of Evolution and Ecology, would require exorbitant inlection of trees is a 'supertree' in University of California, Davis, CA 95616, USA; the broad sense (Box 1). An ideal Andy Purvis is at the Dept of Biology, Imperial College, vestments of resources. Other drawbacks of the supermatrix supertree, which we term a 'strict Ascot, Berkshire, UK SL5 7PY. approach are that some data sets supertree', is one that agrees with (e.g. from DNA-DNA hyall the trees from which it was bridization) cannot be included, derived. and that construction of initial The obvious approach for hypotheses of homology and/or alignment becomes ever combining analyses is simply to combine the original data more difficult as the matrix grows. matrices into a single larger matrix, inserting question D TREE vol. 13, no. 3 March 1998 Copyright © 1998, Elsevier Science Ltd. All rights reserved. 016!}-5347/98/$19.00 PII: S0169-5347(97)01242-1 1 0 5