New Phytol. (1996), 134, 487-493 Root hemiparasitism in a West African rainforest tree Okoubaka aubrevillei (Santalaceae) BY E. M . V E E N E N D A A L ^ * , I. K. ABEBRESE^ M. E. WALSH^ AND M.D. SWAINE^ ^ Okavango Research Centre, University of Botswana, P. Bag 0022, Gaborone, Botswana ^Forestry Research Institute Ghana, University P.O. Box 63, Kumasi, Ghana ^Department of Plant and Soil Science, University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen AB9 2UD, UK {Received 22 February 1996; accepted 1 July 1996) SUMMARY We Studied hemiparasitism in Okoubaka aubrevillei Pellegr. & Normand (Santalaceae), an uncommon West African rainforest tree of widespread distribution (Ivory Coast to Zaire) which attains heights of up to 40 m. It has a very large seed (mean fresh mass of 101 g), and slow accumulation of biomass characteristic of seedlings of shadetolerant rainforest trees. O. aubrevillei seedlings became hemiparasitic within 6 months when grown next to seedlings of the tree species Entandrophragma angolense (Wei.) D C , Pericopsis elata (Harms) Van Meeuwen, Pterygota macrocarpa K. Schum., and Tieghemella heckelii Pierre ex Chev. P. macrocarpa, and the nitrogenfixing legume, P. elata were most infected. Characteristically for hemiparasites, midday leaf water potentials in O. aubrevillei were at — 2 MPa lower than those of the host seedlings. Stomatal conductance, however, was low, with a maximum of 111 mmol m^ s~^ After 1 yr, no significant effects of the hemiparasitism were observed on the grow^th of O. aubrevillei or on its nutrient status as measured by foliar element concentrations. By contrast, the two most infected host species showed increased mortality and/or reduced growth. Foliar element concentrations were not, however, affected in host plants of the surviving species. Although the resources of a large seed might enable seedlings of O. aubrevillei to grow^ independently from hosts for prolonged periods, their simultaneous strong effect on the mortality and growth of host seedlings might point to an alternative competitive function of hemiparasitism. Key words: Foliar nutrients, hemiparasite, leaf water potential, rain forest tree, stomatal conductance. INTRODUCTION Woody angiosperms which are hemiparasitic on the other plants generally belong to the famihes Santalaceae and Olaceae (Rao, 1942; Kuijt, 1969). Most of these do not grow beyond the size of small bushes or trees but a few larger examples exist. The best studied of these are the Sandalwoods {Santalum spp.), which are found in India, the Far East, Australia and the Pacific. A tree-sized root hemiparasite, Acanthosyris A. paulo alvinii Barr. has also been reported from South America (Alvim & Seeschaaf, 1968). Possibly the largest species in the Santalaceae, Okoubaka aubrevillei is a forest tree which attains heights up to 40 m and occurs in the West African rainforest (Hawthorne, 1995). Throughout its range * To whom correspondence should be addressed. E-mail: Okavango@noka.ub.bw the species is locally reputed to kill other trees (Keay, Onochie & Stanfield, 1964) and is often found in openings in the forests. The presence of haustoria on the roots of O. aubrevillei has been reported (Swaine & Hall, 1986). O. aubrevillei produces large fruits containing a single, very large seed. These fruits and the bark are highly sought after for medical and magical purposes (Kreutzkampf & Jurcic, 1985; Abbiw, 1990). The tree is uncommon in the forest, and regenerating seedlings are very difficult to find (Hawthorne, 1995 and personal observations). Perhaps for this reason, there is concern about its regeneration, particularly as the importance of parasitism for O. aubrevillei during the regeneration phase is unknown. After germination, root hemiparasites can undergo a pre-parasitic phase lasting from several hours to several months before attachment to host plants occurs (Pate, Kuo & Davidson, 1990a), and fac- 488 E. M. Veenendaal and others ultative hemiparasites are also known (Press, Smith & Stewart, 1991). In Santalum album haustoria-free specimens have been observed in up to 9-month-oId saplings (Nagaveni & Srimathi, 1985). There is thus a large variability in the dependence on hosts in the early growth of hemiparasites. Although some hemiparasites only infect a narrow range of hosts, man^^ infect a large number of species (Kuijt, 1969). The degree of infection of host species differs (Rama Rao, 1918; Gibson & Watkinson, 1989). Nitrogen-fixing hosts in particular appear to be parasitized more heavily (Gibson & Watkinson, 1991) and haustoria directly attached to nitrogenfixing nodules have been observed in Santalum album (Subbarao et al., 1990). The degree to which root hemiparasites benefit in growth, photosynthesis and nutrient uptake from their attachment might depend on the nutrient status of the host so that attachment to nitrogen-fixing hosts may benefit the hemiparasite, particularly in nitrogen uptake and growth (Seel, Cooper & Press, 1993; Seel & Press, 1993). In woody root hemiparasites, however, competition between host and parasite may also play a role. In Santalum spicatum for example, reduced growth was observed in 9-month-old saplings attached to Acacia mangium (Struthers et al., 1986). For effective solute transfer fro the host's xylem, hemiparasites must maintain lower leaf water potentials than their hosts. These and associated higher stomatal conductances are characteristic of hemiparasites (Pate et al., 19906; Press et al., 1991). In this paper, we attempt to answer the following questions about root hemiparasitism in O. aubrevillei seedlings by means of shadehouse experiments: {a) When do young seedlings of O. aubrevillei become parasitic? {b) Does O. aubrevillei show host specificity ? (c) To what extent is O. aubrevillei dependent on its host(s) for further growth ? {d) Are water relations in O. aubrevillei seedlings similar to those of other hemiparasites? {e) Does O. aubrevillei infiuence the growth of the host plant ? MATERIALS AND METHODS Growth experiment Kumasi. In the shadehouse, irradiances of 25-35 % of ambient irradiance (measured over weekly periods with Didcot DRP 02 integrating PAR sensors, Didcot Instruments, Oxford, UK) were maintained, at this irradiance, relative grown rate (RGR) is maximal for most tree species and shows little variation with variation in irradiance (Veenendaal et al., 1996). The wooden cases were filled with a sandy clay loam and planted with {a) O. aubrevillei and one seedling of each of four host species (total number of monoliths planted = 11). (b) O. aubrevillei alone (total number of monoliths planted = 4) and (c) one seedling of each host plant but no O. aubrevillei (total number of monoliths planted = 4). Six-month-old seedlings of four Ghanaian tree species representing a range of families encountered in natural forest were used in the experiment (Table 1). Host plants and hemiparasite treatments were randomly distributed amongst the monoliths. The degree of infection by O. aubrevillei of host plants was assessed at 6 months and 1 yr. At 6 months, casings of six monoliths were carefully removed and the soil gently washed out from the roots in situ. Haustorial connections were carefully traced back to the host plants. Dry weights of hosts and hemiparasites were determined after oven drying to constant weight at 80 °C. At 6 months, host plant selectivity was assessed statistically by testing whether haustoria were equally distributed among host plants. For this purpose, haustoria numbers were adjusted proportionally for the number of infected host plants of each host species, on the assumption that the initial infection process could well have been random. At each encounter between host and hemi-parasite the root biomass of parasite and host were also taken into account, as a larger root biomass may automatically result in more haustorial connections. Thus, on each occasion the number of haustoria was adjusted proportionally for root biomass of the host and parasite encountered. After 1 yr the remaining boxes were harvested and haustoria and biomass determined as before. The root systems of P. elata seedlings, which had died by this time, had decayed. Haustoria found on dead roots in the vicinity of dead P. elata were considered to have been attached to this host. As a result of the decay of the roots of P. elata, the haustoria distribution could not be adjusted as at 6 months. Effects of parasitism on growth and architecture of both host and parasite was assessed at the end of the experiment by measuring plant height and biomass (oven d. wt) of roots, stems and leaves. Leaf area was measured with a Panasonic® digital scanner. Seeds were collected from a mature O. aubrevillei tree in Asenanyo Forest Reserve in central Ghana. Fresh seed mass of individual seeds was determined. After removal of the pericarp in 12 randomly sampled seeds, weight of the remaining embryo and endosperm was determined after drying to constant weight at 80 °C. The other seeds were planted, and after slow and prolonged germination (2—4 months), 15 seedlings of even size were obtained for the experiment. Plant water relations Monoliths with a wooden casing (60 x 60 x At 1 yr, leaf water potential was measured in hosts 100 cm) were constructed inside a shadehouse located at the Forestry Research Institute Ghana in and parasite with a pressure bomb (Tyree & Root hemiparasitism in O. aubreviJJei 489 Table 1. Ecological guilds of host species Species Family Ecological guild* Entandrophragma angolense (Wei.) Meliaceae non-pioneer light-demander Papilionaceae Pioneer/non-pioneer light-demander (N-fixer) Non-pioneer light-demander Non-pioneer light-demander/shadebearer DC. Pericopsis elata (Harms) Van Meeuwen Pterygota macrocarpa K. Schum. Tieghemella hecketii Pierre ex Chev. Sterculiaceae Sapotaceae * Source: Agyeman, 1994; Hawthorne, 1995; Veenendaal et at, 1996. Nomenclature follows Hawthorne, 1995. Hammel, 1972). Measurements were made at dawn (0600-0800 hours) and at midday (1230-1400 hours) on either the youngest fully expanded leaves (T.heckelii), or the apical parts of the branch (O. aubreviUei) or distal leaflets (E. angolense). P. elata was not measured, as by this time all but one had died, while the leaves of P. macrocarpa hosts were too big to be inserted in the pressure chamber. Stomatal conductance {g^) was measured at the same time with a transit-time diffusion porometer (model 75 95 115 135 Fresh seed mass (g) A.P.4, Delta T. Devices, Cambridge, UK) (Beadle, Ludlow & Honeysett, 1993). For each individual Figure 1. Distribution of fresh seed mass in O. aubrevillei plant, g^ at dawn and midday was taken as the mean (n — 173), Values on the abscissa are mid-class points. of three consecutive readings taken from the (a) (adjusted) youngest fully expanded leaves. Foliar and seed element analysis 35% The youngest fully expanded leaves of hosts and seeds (endosperm and embryo only) of O. aubrevillei were oven-dried at 80 °C, finely ground and used for further analysis. N was determined using a Carlo Erba auto-analyser (Kirsten, 1979). For P, K, Ca, Mg and Mn, plant material was wet-ashed with concentrated H2O2/H2SO4 solution (Allen et al., 1989). P was then determined colorimetrically, K with flame photometry and Mg,, Ca and Mn with atomic absorption spectrophotometry (Marr & Cresser, 1983). B E. angolense D R elata p. macrocarpa 7; heckelii Statistical analysis Data were analysed with ANOVA (Systat for Apple Macintosh) using Tukey's HSD test, and correlation (Sokal & Rohlf, 1981). Mineral element concentrations were log-transformed to ensure normal distribution (Slob, 1986). Figure 2. Percentage distribution of haustoria among host plants and number of host plants infected (in brackets) at 6 months (a) and 1 yr [b). {a, adjusted) are the same data as {a) but adjusted for the number of encounters and root mass of parasite and hosts upon each encounter, (Number of monoliths sampled at 6 m = 6, at 1 yr = 5). RESULTS closely correlated (r^ = 0'92, n = 12) so that average dry seed mass can be estimated to be 43 g. Seed mass Fresh seed mass (embryo, endosperm and pericarp, but fruit flesh removed) of O. aubrevillei varied between 52 and 168 g with a mean mass of 101 g (Fig 1). The dry mass of endosperm and embryo was 43 % of fresh seed mass on average and both were Infection of hosts At 6 months, O. aubrevillei seedlings had made contact on average with 2-5 hosts in each box (range; 2—3) and formed 24 haustoria (range; 4—40) per plant (Fig. 2a). P. elata and P. macrocarpa were more 490 E. M. Veenendaal and others 500 250 EA PE PM TH OA PM TH OA PM TH OA o 03 05 o o E JC CO CC O O CC s 20- EA PE PM EA TH PE Species Figure 3. Height, d.wt LAR and root:shoot ratio in control treatment seedlings ( D , n — 4) and in seedlings exposed to the hemiparasite ( • , n = S). Abbreviations: host plants: EA, E. angolense; PE, P. elata; PM, P. macrocarpa; T H , T. heckelii; hemiparasite: OA, O. auhrevillei. All but one infected P. elata seedling died and thus no data are available. Columns sharing superscripts are not significantly different at a = 0-05 with Tukey's H S D test. often infected, with six and five infected plants, than E. angolense and T. heckelii with one and two, respectively. Haustoria were found in larger numbers on P. elata and P. macrocarpa. When the distribution of haustoria was adjusted for number of individuals of a species encountered and for root biomass of parasite and host in each monolith, the distribution of haustoria appeared more even between host species (Fig 2 a, adj) and although P. elata still had more haustoria per plant than the other species, this difference was not statistically significant (chi-squared test). The degree of infection, although heavily skewed towards P. macrocarpa and P. elata, was thus not necessarily the result of a preferential mechanism in O. aubrevillei. No haustoria were observed in direct contact with nodules of P. elata. However, two P. elata host plants had died at 6 months, while the others appeared smaller than the control plants. After 12 months the remaining five O. aubrevillei had infected 2*8 hosts per monolith (range: 1-4) and formed 22 haustoria per plant (12-50; Fig. 2b). Five P. elata, five P. macrocarpa, two E. angolense and two T. heckelii were infected. P. elata and P. macrocarpa were thus more often infected than the other two hosts throughout the experiment. Effects of the infection on growth of host and parasite At 1 yr all but one P. elata host seedling had died, again suggesting selection by the parasite of this species. In the other highly infected host species, P. macrocarpa, height and d.wt were significantly reduced in infected plants (Fig. 3). Leaf area ratio appeared also reduced, but this was not statisticall significant, owing to the high variability of the dat£ In the less infected host species, E. angolense an i T. heckelii, the differences in height, d.wt. and lein area ratio (LAR) between plants exposed to th; hemiparasite and control plants were not significan • The root:shoot ratio in infected P. macrocarpa wi ' significantly higher than in the control plant , because of a reduction in shoot growth in infecte 1 plants. The root:shoot ratio was not significant' affected in the less infected host plants, E. angolen e and T. heckelii, compared with their controls. Root hemiparasitism in O. aubrevillei 491 Table 2. Mean element concentrations in leaves of control and infected host seedlings and in the leaves and seed of O. aubrevillei {n = 3-5) Element Species E. angolense Control Infected P. elata Control Infectedf P. macrocarpa Control Infected T. heckelii Control Infected O. aubrevillei Control Infecting Seed N (mg g-i) P (mg g-i) 23-6 bc* 24-3 bc 1-3 abc 1-7 bed K (mg g-') 7-6 a 13-3 a Ca (mg g"') Mg Mn (mg g'^) (/^gl 33-1 b 30-3 b 1-4 ab 2-1 b r') 4-4 a 4-6 a 32-5 c 1-7 bed 10-3 a 14-5 bc — — 12-3 ab — 1-2 ab — — — 32-9 c 32-4 c 2-3 d 2-0 cd 22-8 b 25-1 b 36-9 b 46-2 b 2-7 b 2-5 b 27-9 c 16-8 bc 14-8 a 17-4 ab M ab 1-0 a 11-3 a 8-8 a 23-2 b 14-7 ab 1-6 ab 1-2 ab 9-4 ab 4-3 a 31-2 c 33-9 c 22-6 abc 1-5 abc 1-4 abc 1-7 bed 19-7 b 20-3 b 8-4 a 2M b 34-6 b 1-1 a 2-2 b 4-8 a 5-3 a 4-5 a 2-8 b 0-9 a * Values in each column sharing a letter do not differ significantly at a = 0-05 Tukey's HSD test on log-transformed data, t All but one plant died Parasitic O. aubrevillei seedlings connected to hosts were, however, very similar in height, d.wt, LAR and root:shoot ratio to control plants not in contact with hosts. There was thus no significant effect of the parasitic association on growth in the hemiparasite. host seedlings exposed to O. aubrevillei and their controls, nor in parasitic and non-parasitic O. aubrevillei (two-way analysis of variance). Pooled data (Table 3) of the morning and midday showed higher mean^g (143-180 mmol m""^- s~^) for seedlings of the light-demanding host species, P . elata, P. macrocarpa and E. agolense, than for the more shadetolerant T. heckelii { < 100 mmol m~^ s^^). The value Foliar and seed element concentrations of ^g in O. aubrevillei was similar to that of T. heckelii With the exception of Ca, the highest foliar element and both species were mostly significantly different concentrations were found in P. elata and P. macro- from the other species (Tukey's H S D procedure carpa and the lowest in T. heckelii (Table 2). The a = 0-05). differences between species were, however, not Leaf water potential (T leaf) could not be always statistically significant (Tukey's H S D pro- measured in P. macrocarpa owing to its large leaf cedure OL = 0-05). In none of the seedlings of the host size. Early-morning W leaf differed between T. species did foliar concentrations differ significantly heckelii (-0-35 MPa) and O. aubrevillei (-0-9 MPa) between control plants and plants exposed to control plants (Table 3),but the difference between O. aubrevillei, although at times higher values for species or treatments was not statistically significant Mn were found in the control plants. (Tukey's H S D procedure a = 0-005). At midday Foliar element concentrations in the seedlings of there was still no difference in T leaf between host O. aubrevillei were similar in plants attached to hosts plants growing with and without O. aubrevillei, but and in the control plants. Element concentrations in varied between T. heckelii (control and exposed) the seeds of O. aubrevillei were significantly lower (-0-9 MPa) and E. angolense (control) ( - 1 - 4 MPa). for K, Ca, and Mg compared with the leaves of the The control and exposed seedlings of O. aubrevillei seedlings. The other elements did not differ sig- reached midday ^ least values near — 2 MPa, significantly lower than those in control and exposed. nificantly. T. heckelii, but not those in E. angolense. Plant water relations DISCUSSION Stomatal conductance {g^ was measured on an overcast day with maximum irradiance inside the shadehouse reaching 300-350/^mol m"^ s"^ and maximum vapour pressure deficits of 1-3 kPa. Values for g^ were not significantly different during the In many aspects, O. aubrevillei is a unique hemiparasite. The dry seed mass reported in this study is, at 43 g, the largest reported for a hemiparasite and at least 200 times larger than seeds of Santalum album dawn and midday measurement periods, between (Nagaveni & Anantha Padmanabha, 1986). Such a 492 E. M. Veenendaal and others Table 3. Dawn, midday and mean stomatal conductance {gj and leaf water potential {W leaf) in control and infected host seedlings and in control and infecting O. aubrevillei {n = 4) gg ( m m o l m~^ s~^) Species Dawn E. angolense 193 Control 152 Infected P. elata 151 Control — Infectedf P. macrocarpa 161 Control 163 Infected T. heckelii 86 Control Infected 56 O. aubrevillei 111 Control 98 Infecting ^,ea Midday Mean Dawn Midday 151 134 172 a* 143 ab 0-59 a 0-51 a 1-38 ab 1-00 ab 160 — 156 a 200 156 7S 95 59 57 n.m. n.m. — — 181 a 160 a n.m. n.m. n.m. n.m. 85 b 71 b 0-37 a 0-49 a 0-88 a 0-90 a 85 b 77 b 0-97 ab 0-62 a 1.94 b 1-99 b * Values in g^ or ^^^^j sharing a letter do not differ significantly at a = 0-05 with Tukey's HSD test, f All but one plant died, n.m., not measured. large seed mass makes the species dependent for its dispersal on large animals such as forest elephants (Hawthorne, 1995). Limited dispersal possibilities and a smaller number of large seeds are likely to decrease the number of encounters with potential hosts and a selective strategy would therefore seem less advantageous in the species-rich rainforest. Nevertheless in our study O. aubrevillei mainly infected two of the four host species. This could have been the result of chance of encounter, as illustrated by adjusting the distribution of haustoria among hosts for factors such as plant size, but the possibility of a foraging pattern of the roots for nutrients similar to the one observed in non-hemiparasitic plant species should not be excluded (Hutchings & De Kroon, 1994). Contact with hosts normally leads to a significant increase in growth in both herbaceous and woody hemiparasites, particularly if the host is a nitrogenfixing legume (Anantha Padmanabha, Nagaveni & Rai, 1988; Subbarao et al., 1990; Seel, Cooper & Press, 1993; Seel & Press 1993). However, seedlings of O. aubrevillei were hardly larger, at 50 g d.wt, than the original seed mass of 43 g, even though the nitrogen-fixing host P. elata was strongly infected. A large seed mass and very slow growth is characteristic of shade-tolerant trees, which naturally regenerate in conditions of low irradiance (Boot, 1996; Veenendaal et al., 1996). After 1 yr, O. aubrevillei seedlings did not show significant differences in foliar nutrient concentrations in the presence or absence of hosts. Even though concentrations of K, Ca, and Mg were lower in the seed than in foliar material of O. aubrevillei, the large seed mass and N and P resources probably enable slow growth of O. aubrevillei independent of host, at least until plant biomass has increased several fold beyond the original dry seed mass. The initial reliance on the seed resources could also explain why the average number of haustoria per hemiparasite did not increase between 6 months and 1 yr. The midday ^leaf ^^ O. aubrevillei tended to be lower than in host plants, which is characteristic for hemiparasites, but g^ was similar to that found in shade-tolerant tree seedlings and lower than that of its main hosts in this experiment (Pate et al., 1990&; Riddoch et al., 1991; Press et al., 1993). In plant water relations, O. aubrevillei thus again shows characteristics of both a shade-tolerant tree and a hemiparasite. In sharp contrast to the apparent lack of effect of the hemiparasitism on the growth of O. aubrevillei, a strong negative effect was observed on the growth of the most infected host species, P. elata (high mortality) and P. macrocarpa (reduced growth). This raises questions about an alternative function of the parasitic relationship. Possibly the ability to kill or reduce growth in neighbouring seedlings might enhance the competitiveness of the slow-germinating slow-growing O. aubrevillei in an environment where light may be the limiting factor. Further study of the regeneration niche of this species in the field will be needed to test this hypothesis. The low rate of ^g in O. aubrevillei compared with other hemiparasites and hosts also invites further study, particularly as g^ is considered to be the driving force behind solute transfer (Press et al., 1993). The trophic aspect of the host-parasite interaction should be further studied using stable isotope techniques (Govier, Nelson & Pate 1967; Tennakoon & Pate 1996). Finally, for the conservation of O. aubrevillei, we Root hemiparasitism in O. aubrevillei suggest the raising of seedlings from freshly collected seeds for subsequent establishment in protected areas. Exposure to host plants appears not to be necessary, as is demonstrated in this study, at least in the initial phase after germination. 493 Kuijt J. 1969. The biology of parasitic fiowering plants. Davis, CA, USA: University of California Press. Marr LL, Cresser MS. 1983. Environmental chemical and analysis. Glasgow: Blackie & Sons. Nagaveni HC, Anantha Padmanabha HS. 1986. Seed polymorphism and germination in Santalum album. Van- Vigyan 24: 25-28. Nagaveni HC, Srimathi RA. 1985. A note on haustoria-less sandal plants. Indian Eorester 111: 161-163. Pate JS, Kuo J, Davidson NJ. 1990a. Morphology and anatomy ACKNOWLEDGEMENTS of the haustorium of the root hemiparasite Olax phyllantii The authors thank P. Amoakoh and R. T. Lecha for their (Olaceae), with special reference to the haustorical interface. Annals of Botany 65: 425^36. help with the growth experiments and D. Mackinnon for help with the nutrient analyses. We are also grateful for Pate JS, Davidson NJ, Kuo J, Milburn JA. 19906. Water relations of the root hemiparasite Olax phyllantii {l^ahiW) R.Br. advice and critical comments by F. Martin, W. Seel, A. (Olaceae), and its multiple hosts. Oecologia 84: 186-93. Taylor, and two anonymous referees. This work was Press MC, Parsons AN, Mackay AW, Vincent, CA, Cochrane funded by ODA/OFI FRP Research grant R 4740. V, Seel WE. 1993. Gas exchange characteristics and nitrogen relations of two Mediterranean root hemiparasites. Oecologia 95: 145-151. Press MC, Smith S, Stewart, GR. 1991. Carbon acquisition and REFERENCES assimilation in parasitic plants. Functional Ecology 5: 278-283. Rama Rao M. 1918. Field investigation of 'spike' disease in Abbiw DK. 1990. Useful plants of Ghana. London: Intermediate sandal in the Kolimatai Hills. India Eorester 44: 58-65. Technology Publications. Rao LN. 1942. Parasitism in the Santalaceae. Annals of Botany, 6: Agyeman VK. 1994. Environment influences on tropical tree 131-149. seedling growth. Ph. D. thesis, University of Aberdeen, UK. Riddoch I, Grace J, Fasehun FE, Riddoch B, Ladipo DO. Alvim P de T, Seeschaaf KW. 1968. Dieback and death of Cacao 1991. Photosynthesis and successional status of seedlings in a trees caused by a new species of parasitic tree. Nature 28: tropical semi-deciduous rain forest in Nigeria. Journal of 1386--1387. Ecology 79: 491-503. Allen SE, Grimshaw HM, Parkinson JA, Quarmby C. 1989. Seel WE, Cooper RE, Press MC. 1993. Growth, gas exchange Chemical analysis of ecological materials, 2nd edition. Oxford: and water use efficiency of the facultative hemiparasite RhinBlackwel! Scientific Publications. antus minor associated with hosts differing in foliar nitrogen Anantha Padmanabha HS, Nagaveni HC, Nagaveni HC, concentration. Physiologia Plantarum 89: 64—70. Rai SN. 1988. Influence of host plant on growth of Sandal. Seel WE, Press MC. 1993. Infiuence of the host on three subMyforest 24: 154-160. Arctic annual facultative root hemiparasites. New Phytologist Beadle CL, Ludlow M, Honeysett JL. 1993. Water relations. 125: 131-138. In: Hall DO, Scurlock JMO, Bolhar-Nordenkampf HR, Slob W. 1986. Strategies in applying statistics in ecological research. Leegood RC, Long SP, eds. Photosynthesis and Production in a Amsterdam: Free University Press. Changing Environment London: Chapman & Hall, 103—109. Sokal RR, Rohlf FJ. 1981. Biometry, 2nd edition. New York: W. Boot R. 1996. The significantly of seedling size and growth rate of H. Freeman & Co. tropical rain forest tree seedlings for regeneration in canopy Struthers R, Lamont BB, Fox JED, Wijesuriya S, Crossland, openings. In: Swaine MD. ed. Ecology of Tropical Eorest Tree T. 1986. Mineral nutrition of sandalwood. Journal of ExSeedlittgs. Pans/Carnforth: UNESCO/Parthenon. perimental Botany 37: 1274-1284. Govier RN, Denise Nelson M, Pate JS. 1967. Hemiparasitic Subbarao NS, Yadav D, Anantha Padmanabha HS, Naganutrition in angiosperms. I. The transfer of organic compounds veni HC, Singh CS, Kavimandan SK. 1990. Nodule haustoria from host to Odontites Verna (Bell.) Dum. (Scrophulariaceae). and microbial features of Cajanus and Pongamia parasitized by New Phytologist 66, 285-297. sandal (Sandalwood). Platit and Soil 128: 249-256. Gibson CC, Watkinson AR. 1989. The host range and selectivity Svt^aine MD & Hall JB. 1986. Forest structure and dj'namics. In: of a parasitic plant: Rhinanthus minor L. Oecologia 78: 01-406. Lawson GW. ed. Plant Ecology in West Africa. Chichester: Gibson CC, Watkinson AR. 1991. Host selectivity and the John Wiley & Sons. mediation of competition by the root hemiparasite Rhinanthus Tennakoon KU, Pate JS. 1996. Heterotrophic gain of carbon minor L. Oecologia 86: 81-87. from hosts by the xj-lem-tapping root hemiparasite Olax Hawthorne WD. 1995. Ecological profiles of Ghanaian forest trees. phyllanthi (Olacaceae). Oecologia 105: 369-376. Tyree MT, Hammel HT. 1972. The measurement of turgor Tropical Forest Papers 29. OFI/ODA, Oxford. pressure and the water relations of plants by the pressure bomb Hutchings MJ, De Kroon H. 1994. Foraging in plants: the role technique. Journal of Experimental Botany 23: 267-282. of morphological plasticity in resource acquisition. Advances in Veenendaal EM, Swaine MD, Lecha RT, Walsh MF, Ecological Research 25: 159-238. Abebrese IK, O>vusu-Afriyi K. 1996. Growth responses of Keay RWJ, Onochie CFA, Stanfield DP. 1964. Nigerian trees. West African forest tree seedlings to irradiance and soil fertility. Ibadan: Nigerian National Press. Eunctional Ecology (in press). Kirsten WJ. 1979. Automatic methods for the simultaneous determination of carbon, hydrogen and sulphur and for sulphur Wagner H, Kreutzkamp B, Jurcic K. 1985. Inhaltstoffe und Pharmakologie der Okoubaka aubreviUei-rmde. Planta-Medica alone in organic and inorganic materials. Analytical Chemistry, 51: 404-407. 51: 1173-1175. 

Root Hemiparasitism in a West African Rainforest Tree Okoubaka aubrevillei (Santalaceae) Author(s): E. M. Veenendaal, I. K. Abebrese, M. F. Walsh, M. D. Swaine Source: New Phytologist, Vol. 134, No. 3 (Nov., 1996), pp. 487-493 Published by: Blackwell Publishing on behalf of the New Phytologist Trust Stable URL: http://www.jstor.org/stable/2558566 . Accessed: 01/09/2011 05:16 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. Blackwell Publishing and New Phytologist Trust are collaborating with JSTOR to digitize, preserve and extend access to New Phytologist. http://www.jstor.org New Phytol.(1996), 134, 487-493 Root hemiparasitism in a West African rainforesttree Okoubaka aubrevillei ( antalaceae) BY E. M. VEENENDAAL1*, AND M. D. SWAINE3 I. K. ABEBRESE2, M. F. WALSH3 1 Okavango Research Centre, Universityof Botswana, P. Bag 0022, Gaborone, Botswana Institute Ghana, UniversityP.O. Box 63, Kumasi, Ghana Research 2Forestry 3Department of Plant and Soil Science, Universityof Aberdeen, CruickshankBuilding, St Machar Drive, Aberdeen AB9 2UD, UK (Received 22 February 1996; accepted 1 July 1996) SUMMARY We studied hemiparasitism in Okoubaka aubrevillei Pellegr. & Normand (Santalaceae), an uncommon West African rainforesttree of widespread distribution (Ivory Coast to Zaire) which attains heights of up to 40 m. It has a very large seed (mean fresh mass of 101 g), and slow accumulation of biomass characteristic of seedlings of shadetolerant rainforest trees. 0. aubrevillei seedlings became hemiparasitic within 6 months when grown next to seedlings of the tree species Entandrophragma angolense (Wel.) DC., Pericopsis elata (Harms) Van Meeuwen, Pterygota macrocarpa K. Schum., and Tieghemella heckelii Pierre ex Chev. P. macrocarpa, and the nitrogenfixing legume, P. elata were most infected. Characteristically for hemiparasites, midday leaf water potentials in 0. aubrevillei were at -2 MPa lower than those of the host seedlings. Stomatal conductance, however, was low, with a maximum of 111 mmol m2 s-1. After1 yr,no significanteffectsof the hemiparasitism were observed on the growth of 0. aubrevillei or on its nutrient status as measured by foliar element concentrations. By contrast, the two most infected host species showed increased mortality and/or reduced growth. Foliar element concentrations were not, however, affected in host plants of the surviving species. Although the resources of a large seed might enable seedlings of 0. aubrevillei to grow independently from hosts for prolonged periods, their simultaneous strong effect on the mortality and growth of host seedlings might point to an alternative competitive function of hemiparasitism. Key words: Foliar nutrients, hemiparasite, leaf water potential, rain forest tree, stomatal conductance. INTRODUCTION Woody angiosperms which are hemiparasitic on the other plants generally belong to the families Santalaceae and Olaceae (Rao, 1942; Kuijt, 1969). Most of these do not grow beyond the size of small bushes or trees but a few larger examples exist. The best studied of these are the Sandalwoods (Santalum spp.), which are found in India, the Far East, Australia and the Pacific. A tree-sized root hemiparasite, AcanthosyrisA. paulo alvinii Barr. has also been reported from South America (Alvim & Seeschaaf, 1968). Possibly the largest species in the Santalaceae, Okoubaka aubrevillei is a forest tree which attains heights up to 40 m and occurs in the West African rainforest(Hawthorne, 1995). Throughout its range * To whom correspondenceshould be addressed. E-mail: Okavango(noka. ub . bw the species is locally reputed to kill other trees (Keay, Onochie & Stanfield, 1964) and is often found in openings in the forests. The presence of haustoria on the roots of 0. aubrevillei has been reported (Swaine & Hall, 1986). 0. aubrevillei produces large fruits containing a single, very large seed. These fruitsand the bark are highly sought afterfor medical and magical purposes (Kreutzkampf & Jurcic, 1985; Abbiw, 1990). The tree is uncommon in the forest, and regenerating seedlings are very difficultto find (Hawthorne, 1995 and personal observations). Perhaps for this reason, there is concern about its regeneration, particularly as the importance of parasitism for 0. aubrevillei during the regeneration phase is unknown. Aftergermination, root hemiparasites can undergo a pre-parasitic phase lasting from several hours to several months before attachment to host plants occurs (Pate, Kuo & Davidson, 1990a), and fac- 488 E. M. Veenendaal and others Kumasi. In the shadehouse, irradiances of 25-35 % of ambient irradiance (measured over weekly periods with Didcot DRP 02 integrating PAR sensors, Didcot Instruments, Oxford, UK) were maintained. at this irradiance, relative grown rate (RGR) is maximal for most tree species and shows little variation with variation in irradiance (Veenendaal et al., 1996). The wooden cases were filledwith a sandy clay loam and planted with (a) 0. aubrevillei and one seedling of each of four host species (total number of monoliths planted = 11). (b) 0. aubrevillei alone (total number of monoliths planted = 4) and (c) one seedling of each host plant but no 0. aubrevillei (total number of monoliths planted = 4). Six-month-old seedlings of four Ghanaian tree species representing a range of families encountered in natural forestwere used in the experiment (Table 1). Host plants and hemiparasite treatments were randomly distributed amongst the monoliths. The degree of infection by 0. aubrevillei of host plants was assessed at 6 months and 1 yr. At 6 months, casings of six monoliths were carefully removed and the soil gently washed out from the roots in situ. Haustorial connections were carefully traced back to the host plants. Dry weights of hosts and hemiparasites were determined afteroven drying to constant weight at 80 ?C. At 6 months, host plant selectivity was assessed statistically by testing whether haustoria were equally distributed among host plants. For this purpose, haustoria numbers were adjusted proportionally for the number of infected host plants of each host species, on the assumption that the initial infection process could well have been random. At each encounter between host and hemi-parasite the root biomass of parasite and host were also taken into account, as a larger root biomass may automatically result in more haustorial connections. Thus, on each occasion the number of haustoria was adjusted proportionally for root biomass of the host and parasite encountered. After 1 yr the remaining boxes were harvested and haustoria and biomass determined as before. The MATERIALS AND METHODS root systems of P. elata seedlings, which had died by this time, had decayed. Haustoria found on dead Growth experiment roots in the vicinity of dead P. elata were considered Seeds were collected from a mature 0. aubrevillei to have been attached to this host. As a result of the tree in Asenanyo Forest Reserve in central Ghana. decay of the roots of P. elata, the haustoria disFresh seed mass of individual seeds was determined. tribution could not be adjusted as at 6 months. Effectsof parasitism on growth and architecture of After removal of the pericarp in 12 randomly sampled seeds, weight of the remaining embryo and both host and parasite was assessed at the end of the endosperm was determined after drying to constant experiment by measuring plant height and biomass weight at 80 'C. The other seeds were planted, and (oven d. wt) of roots, stems and leaves. Leaf area was afterslow and prolonged germination (2-4 months), measured with a Panasonic? digital scanner. 15 seedlings of even size were obtained for the experiment. Plant water relations Monoliths with a wooden casing (60 x 60 x At 1 yr, leaf water potential was measured in hosts 100 cm) were constructed inside a shadehouse located at the Forestry Research Institute Ghana in and parasite with a pressure bomb (Tyree & ultative hemiparasites are also known (Press, Smith & Stewart, 1991). In Santalum album haustoria-free specimens have been observed in up to 9-month-old saplings (Nagaveni & Srimathi, 1985). There is thus a large variability in the dependence on hosts in the early growth of hemiparasites. Although some hemiparasites only infect a narrow range of hosts, many infecta large number of species (Kuijt, 1969). The degree of infection of host species differs (Rama Rao, 1918; Gibson & Watkinson, 1989). Nitrogen-fixing hosts in particular appear to be parasitized more heavily (Gibson & Watkinson, 1991) and haustoria directly attached to nitrogenfixing nodules have been observed in Santalum album (Subbarao et al., 1990). The degree to which root hemiparasites benefit in growth, photosynthesis and nutrient uptake from their attachment might depend on the nutrient status of the host so that attachment to nitrogen-fixinghosts may benefit the hemiparasite, particularly in nitrogen uptake and growth (Seel, Cooper & Press, 1993; Seel & Press, 1993). In woody root hemiparasites, however, competition between host and parasite may also play a role. In Santalum spicatum for example, reduced growth was observed in 9-month-old saplings attached to Acacia mangium (Struthers et al., 1986). For effectivesolute transferfro the host's xylem, hemiparasites must maintain lower leaf water potentials than their hosts. These and associated higher stomatal conductances are characteristic of hemiparasites (Pate et al., 1990b; Press et al., 1991). In this paper, we attempt to answer the following questions about root hemiparasitism in 0. aubrevillei seedlings by means of shadehouse experiments: (a) When do young seedlings of 0. aubrevillei become parasitic? (b) Does 0. aubrevillei show host specificity?(c) To what extent is 0. aubrevillei dependent on its host(s) for further growth? (d) Are water relations in 0. aubrevillei seedlings similar to those of other hemiparasites? (e) Does 0. aubrevillei influence the growth of the host plant? 489 Root hemiparasitismin 0. aubrevillei Table 1. Ecological guilds of host species * Species Family Ecological guild* angolense(Wel.) Entandrophragma DC. Pericopsiselata (Harms) Van Meeuwen PterygotamacrocarpaK. Schum. TieghemellaheckeliiPierreex Chev. Meliaceae non-pioneerlight-demander Papilionaceae light-demander Pioneer/non-pioneer (N-fixer) Non-pioneerlight-demander Non-pioneerlight-demander/shadebearer Sterculiaceae Sapotaceae Source: Agyeman,1994; Hawthorne,1995; Veenendaal et al., 1996. NomenclaturefollowsHawthorne,1995. Hammel, 1972). Measurements were made at dawn (0600-0800 hours) and at midday (1230-1400 hours) on either the youngest fully expanded leaves (T. heckelii), or the apical parts of the branch (0. aubrevillei) or distal leaflets (E. angolense). P. elata was not measured, as by this time all but one had died, while the leaves of P. macrocarpa hosts were too big to be inserted in the pressure chamber. Stomatal conductance (g,) was measured at the same time with a transit-time diffusionporometer (model A.P.4, Delta T. Devices, Cambridge, UK) (Beadle, Ludlow & Honeysett, 1993). For each individual plant, gs at dawn and midday was taken as the mean of three consecutive readings taken from the youngest fully expanded leaves. Foliar and seed elementanalysis The youngest fully expanded leaves of hosts and seeds (endosperm and embryo only) of 0. aubrevillei were oven-dried at 80 ?C, finelyground and used for further analysis. N was determined using a Carlo Erba auto-analyser (Kirsten, 1979). For P, K, Ca, Mg and Mn, plant material was wet-ashed with concentrated H202/H2SO4 solution (Allen et al., 1989). P was then determined colorimetrically, K with flame photometry and Mg, Ca and Mn with atomic absorption spectrophotometry (Marr & Cresser, 1983). 25 200 1510 _ O50- 55 Data were analysed with ANOVA (Systat for Apple Macintosh) using Tukey's HSD test, and correlation (Sokal & Rohlf, 1981). Mineral element concentrations were log-transformed to ensure normal distribution (Slob, 1986). RESULTS 95 115 135 155 Freshseed mass (g) Figure 1. Distributionof freshseed mass in 0. aubrevillei (n = 173). Values on the abscissa are mid-classpoints. (a) (adjusted) (a) 1% (1) 41% 35% K) 46% I ~~~~~(6) 18% 12% (2) 10% (3) (b) 26% (5) E6 / 60 Statistical analysis 75 0 E. angolense P elata P macrocarpa 4%) (2) * T heckelii (5) Figure 2. Percentagedistributionofhaustoriaamonghost plants and numberof host plants infected(in brackets)at 6 months(a) and 1 yr (b). (a, adjusted) are the same data as (a) but adjusted forthe numberof encountersand root mass of parasiteand hostsupon each encounter.(Number of monoliths sampled at 6 m = 6, at 1 yr = 5). closely correlated (r2 = 0-92, n = 12) so that average dry seed mass can be estimated to be 43 g. Seed mass Fresh seed mass (embryo, endosperm and pericarp, but fruit flesh removed) of 0. aubrevillei varied between 52 and 168 g with a mean mass of 101 g (Fig 1). The dry mass of endosperm and embryo was 43 % of fresh seed mass on average and both were Infection of hosts At 6 months, 0. aubrevillei seedlings had made contact on average with 2-5 hosts in each box (range: 2-3) and formed 24 haustoria (range: 4-40) per plant (Fig. 2a). P. elata and P. macrocarpa were more 490 E. M. Veenendaal and others 250 500 a E 200 -400 -a -~~~~ ~~QQ 0 - a ~a 150 b 0~~~~~~~~~~~~~~~~~~~~ b ~ b b c b T3 100 C C20 50 -100 0 EA PE PM TH 0 QA PE PM TH QA 1.5- 80 60- EA a b a b a a a b c E 40 -dbc d < c d d ~~~~~d 20 b b c C 0~~~~~~~~~~~~~~~~ a 0ab a b b bb b ~~~~~~d0.5 0 0 20 EA PE PM TH QA Species EA PE PM TH QA Figure 3. Height, d. WtLAR and root:shoot ratioin controltreatmentseedlings(LI, n = 4) and in seedlings exposed to the hemiparasite(N, n = 5). Abbreviations:host plants: EA, F. angolense;PE, P. elata; PM, P. macrocarpa;TH, T. heckelii;hemiparasite:OA, 0. aubrevillei.All but one infectedP. elata seedling died different at ac= 0 05 with and thus no data are available. Columns sharingsuperscriptsare not significantly Tukey's HSD test. ofteninfected,withsix and fiveinfectedplants,than E. angolenseand T. heckelii with one and two, respectively. Haustoria were found in larger numberson P. elata and P. macrocarpa. When the distributionof haustoriawas adjusted fornumberof individualsof a species encountered and for root biomass of parasite and host in each monolith, the distributionof haustoria appeared more even between host species (Fig 2a, adj) and althoughP. elata stillhad more haustoriaper plant than the other species, this differencewas not statistically significant (chi-squaredtest).The degree of infection, although heavily skewed towards P. macrocarpaand P. elata, was thus not necessarily the resultof a preferentialmechanismin 0. aubrevillei. No haustoriawere observedin directcontact withnodules ofP. elata. However,two P. elata host plants had died at 6 months, while the others appeared smallerthan the controlplants. After12 monthsthe remainingfive0. aubrevillei had infected2 8 hostsper monolith(range: 1-4) and formed22 haustoriaper plant(12-50; Fig. 2b). Five P. elata, five P. macrocarpa,two E. angolenseand two T. heckeliiwere infected.P. elata and P. macrocarpa were thus more ofteninfectedthan the other two hosts throughoutthe experiment. on growthof hostand parasite Effectsof theinfection At 1 yr all but one P. elata host seedlinghad died, again suggestingselection by the parasite of this species. In the other highlyinfectedhost species, P. macrocarpa,height and d.wt were significantly reduced in infectedplants (Fig. 3). Leaf area ratio appeared also reduced,but this was not statistically significant, owingto the highvariabilityof the data. In the less infectedhost species, E. angolenseand in height,d.wt. and leaf T. heckelii,the differences area ratio (LAR) between plants exposed to the hemiparasiteand controlplantswerenot significant. The root:shoot ratio in infectedP. macrocarpawas significantlyhigher than in the control plants, because of a reductionin shoot growthin infected plants. The root:shoot ratio was not significantly affectedin the less infectedhost plants,E. angolense and T. heckelii,comparedwith theircontrols. 491 Root hemiparasitismin 0. aubrevillei Table 2. Mean elementconcentrationsin leaves of controland infectedhost seedlingsand in the leaves and seed of 0. aubrevillei (n = 3-5) Element Species E. angolense Control Infected P. elata Control Infectedt P. macrocarpa Control Infected T. heckelii Control Infected 0. aubrevillei Control Infecting Seed Mn (ag g-) N (mg g1) P (mg g1) K (mg g1) Ca (mg g1) Mg (mg g1) 236 bc* 243 bc 13 abc 1-7bcd 76 a 133 a 33-1b 303 b 1 4 ab 21 b 44 a 46 a 325 c 1-7bcd 103 a 123 ab 1 2ab 145 bc 32-9c 324 c 23d 2-0cd 22-8b 25 1 b 36-9b 462 b 27 b 2-5b 27-9c 168 bc 148 a 174ab 1 1 ab 10a 113 a 8-8a 23 2 b 147ab 1-6ab 12ab 94 ab 43a 31-2c 33 9 c 226 abc 1-5abc 1 4 abc 1-7bcd 19-7b 20 3 b 84 a 21 1 b 34-6b 77 a 22b 2-8b 09 a 4-8a 53 a 4-5a * Values in each column sharinga letterdo not differ at a = 0 05 Tukey's HSD teston log-transformed significantly data. t All but one plant died host seedlings exposed to 0. aubrevillei and their controls, nor in parasitic and non-parasitic 0. aubrevillei (two-way analysis of variance). Pooled data (Table 3) of the morning and midday showed higher mean gs (143-180 mmol m-2- s-1) for seedlings of the light-demanding host species, P. elata, P. macrocarpa and E. agolense, than for the more shadetolerant T. heckelii( < 100 mmol m-2 s-1). The value Foliar and seed elementconcentrations ofg, in 0. aubrevillei was similar to that of T. heckeili With the exception of Ca, the highest foliar element and both species were mostly significantlydifferent concentrations were found in P. elata and P. macro- from the other species (Tukey's HSD procedure carpa and the lowest in T. heckelii (Table 2). The Z = 0 05). differences between species were, however, not Leaf water potential (T leaf) could not be always statistically significant (Tukey's HSD pro- measured in P. macrocarpa owing to its large leaf cedure ct= 0 05). In none of the seedlings of the host size. Early-morning T leaf differed between T. species did foliar concentrations differsignificantly heckelii (-0 35 MPa) and 0. aubrevillei (-0 9 MPa) between control plants and plants exposed to control plants (Table 3),but the differencebetween 0. aubrevillei, although at times higher values for species or treatments was not statistically significant Mn were found in the control plants. (Tukey's HSD procedure ct= 0005). At midday Foliar element concentrations in the seedlings of there was still no differencein T leaf between host 0. aubrevillei were similar in plants attached to hosts plants growing with and without 0. aubrevillei, but and in the control plants. Element concentrations in varied between T. heckelii (control and exposed) the seeds of 0. aubrevillei were significantly lower (- 09 MPa) and E. angolense (control) (-1 4 MPa). for K, Ca, and Mg compared with the leaves of the The control and exposed seedlings of 0. aubrevillei seedlings. The other elements did not differ sig- reached midday P least values near -2 MPa, nificantly. significantlylower than those in control and exposed. T. heckelii, but not those in E. angolense. Parasitic 0. aubrevillei seedlings connected to hosts were, however, very similar in height, d.wt, LAR and root: shoot ratio to control plants not in contact with hosts. There was thus no significant effect of the parasitic association on growth in the hemiparasite. Plant water relations Stomatal conductance (g,) was measured on an overcast day with maximum irradiance inside the shadehouse reaching 300-350 ,umol m-2 s-' and maximum vapour pressure deficitsof 1-3 kPa. Values for g, were not significantly different during the dawn and midday measurement periods, between DISCUSSION In many aspects, 0. aubrevillei is a unique hemiparasite. The dry seed mass reported in this study is, at 43 g, the largest reported for a hemiparasite and at least 200 times larger than seeds of Santalum album (Nagaveni & Anantha Padmanabha, 1986). Such a 492 E. M. Veenendaal and others Table 3. Dawn, midday and mean stomatal conductance (g,) and leaf water potential (T leaf) in control and infectedhost seedlings and in control and infecting0. aubrevillei (n = 4) g; (mmol m-2s-1) Species Dawn E. angolense Control 193 Infected 152 P. elata Control 151 Infectedt P. macrocarpa Control 161 Infected 163 T. heckelii 86 Control Infected 56 0. aubrevillei 111 Control Infecting 98 Tleaf (MPa) Midday Mean Dawn Midday 151 134 172 a* 143 ab 059 a 051 a 1 38 ab 1 00 ab 160 156 a n.m. n.m. 200 156 181 a 160 a n.m. n.m. n.m. n.m. 75 95 85 b 71 b 037 a 049 a 088 a 090 a 59 57 85 b 77 b 0 97 ab 062 a 1.94 b 1 99 b * Values in g5or Tleaf sharinga letterdo not differ at ca= 0 05 withTukey's HSD test.t All but one plant significantly died. n.m., not measured. large seed mass makes the species dependentforits dispersalon large animals such as forestelephants (Hawthorne, 1995). Limited dispersal possibilities and a smaller number of large seeds are likely to decrease the number of encounterswith potential hosts and a selectivestrategywould thereforeseem less advantageous in the species-rich rainforest. Nevertheless in our study 0. aubrevillei mainly infectedtwoofthefourhostspecies.This could have been the resultof chance of encounter,as illustrated by adjusting the distributionof haustoria among hostsforfactorssuch as plantsize, but thepossibility ofa foragingpatternoftherootsfornutrientssimilar to the one observed in non-hemiparasiticplant species should not be excluded (Hutchings & De Kroon, 1994). Contactwithhosts normallyleads to a significant increase in growthin both herbaceous and woody hemiparasites,particularlyif the host is a nitrogenfixinglegume (Anantha Padmanabha, Nagaveni & Rai, 1988; Subbarao et al., 1990; Seel, Cooper & Press, 1993; Seel & Press 1993). However,seedlings of 0. aubrevilleiwere hardly larger,at 50 g d. wt, thanthe originalseed mass of 43 g, even thoughthe nitrogen-fixing hostP. elata was stronglyinfected.A largeseed mass and veryslow growthis characteristic ofshade-tolerant trees,whichnaturallyregeneratein conditionsoflow irradiance(Boot, 1996; Veenendaal et al., 1996). After1 yr, 0. aubrevilleiseedlingsdid not show significantdifferencesin foliar nutrient concentrationsin the presence or absence of hosts. Even thoughconcentrations of K, Ca, and Mg were lower in the seed than in foliar material of 0. thelargeseed mass and N and P resources aubrevillei, probablyenable slow growthof 0. aubrevilleiindependent of host, at least until plant biomass has increasedseveral fold beyond the originaldry seed mass. The initialrelianceon theseed resourcescould also explainwhytheaveragenumberofhaustoriaper hemiparasitedid not increasebetween6 monthsand 1 yr. The midday Tleaf in 0. aubrevilleitended to be lowerthan in host plants,whichis characteristic for hemiparasites,but gs was similar to that found in shade-toleranttreeseedlingsand lower than thatof its main hostsin thisexperiment(Pate et al., 1990b; Riddoch et al., 1991; Press et al., 1993). In plant water relations, 0. aubrevillei thus again shows characteristicsof both a shade-toleranttree and a hemiparasite. In sharp contrastto the apparentlack of effectof thehemiparasitismon thegrowthof 0. aubrevillei,a strongnegativeeffectwas observedon the growthof the most infectedhost species, P. elata (high mortality) and P. macrocarpa(reduced growth). This raises questionsabout an alternativefunctionof the parasiticrelationship.Possibly the abilityto kill or reduce growth in neighbouring seedlings might oftheslow-germinating enhancethecompetitiveness where slow-growing0. aubrevilleiin an environment light may be the limitingfactor.Furtherstudy of the regenerationniche of this species in the field will be needed to testthis hypothesis.The low rate of gs in 0. aubrevilleicompared with other hemistudy,particuparasitesand hostsalso invitesfurther larlyas gsis consideredto be thedrivingforcebehind solute transfer(Press et al., 1993). The trophic aspect of the host-parasiteinteractionshould be furtherstudied using stable isotope techniques (Govier, Nelson & Pate 1967; Tennakoon & Pate 1996). Finally,forthe conservationof 0. aubrevillei,we in 0. aubrevillei Root hemiparasitism suggest the raising of seedlings fromfreshlycollected seeds for subsequent establishment in protected areas. Exposure to host plants appears not to be necessary, as is demonstrated in this study, at least in the initial phase after germination. ACKNOWLEDGEMENTS The authorsthankP. Amoakohand R. T. Lecha fortheir help withthe growthexperimentsand D. Mackinnonfor help with the nutrientanalyses. We are also gratefulfor advice and criticalcommentsby F. Martin, W. Seel, A. Taylor, and two anonymous referees.This work was fundedby ODA/OFI FRP Research grantR 4740. REFERENCES Abbiw DK. 1990. UsefulplantsofGhana. London: Intermediate Technology Publications. Agyeman VK. 1994. Environmentinfluenceson tropical tree seedlinggrowth.Ph. D. thesis,Universityof Aberdeen,UK. Alvim P de T, Seeschaaf KW. 1968. Dieback and deathofCacao trees caused by a new species of parasitic tree. Nature 28: 1386-1387. Allen SE, Grimshaw HM, Parkinson JA, Quarmby C. 1989. Chemicalanalysisof ecologicalmaterials,2nd edition. Oxford: BlackwellScientificPublications. Anantha Padmanabha HS, Nagaveni HC, Nagaveni HC, Rai SN. 1988. Influenceof host plant on growthof Sandal. Myforest24: 154-160. Beadle CL, Ludlow M, Honeysett JL. 1993. Water relations. In: Hall DO, Scurlock JMO, Bolhar-NordenkampfHR, Leegood RC, Long SP, eds. Photosynthesis and Productionin a ChangingEnvironment London: Chapman & Hall, 103-109. Boot R. 1996. The significantly ofseedlingsize and growthrateof tropical rain foresttree seedlings for regenerationin canopy openings.In: Swaine MD. ed. Ecologyof TropicalForest Tree Seedlings.Paris/Carnforth:UNESCO/Parthenon. Govier RN, Denise Nelson M, Pate JS. 1967. Hemiparasitic nutritionin angiosperms.I. The transferoforganiccompounds fromhost to OdontitesVerna(Bell.) Dum. (Scrophulariaceae). New Phytologist 66, 285-297. Gibson CC, Watkinson AR. 1989. The hostrangeand selectivity of a parasiticplant: RhinanthusminorL. Oecologia78: 01-406. Gibson CC, Watkinson AR. 1991. Host selectivityand the mediationof competitionby the root hemiparasiteRhinanthus minorL. Oecologia86: 81-87. Hawthorne WD. 1995. Ecologicalprofiles ofGhanaianforesttrees. Tropical Forest Papers 29. OFI/ODA, Oxford. Hutchings MJ,De Kroon H. 1994. Foragingin plants: the role of morphologicalplasticityin resourceacquisition.Advancesin EcologicalResearch25: 159-238. Keay RWJ,Onochie CFA, Stanfield DP. 1964. Nigeriantrees. Ibadan: NigerianNational Press. Kirsten WJ. 1979. Automatic methods for the simultaneous determination ofcarbon,hydrogenand sulphurand forsulphur alone in organicand inorganicmaterials.AnalyticalChemistry, 51: 1173-1175. 493 Kuijt J. 1969. The biologyofparasiticflowering plants.Davis, CA, USA: Universityof CaliforniaPress. Marr LL, Cresser MS. 1983. Environmentalchemical and analysis.Glasgow: Blackie & Sons. Nagaveni HC, Anantha Padmanabha HS. 1986. Seed polymorphismand germinationin Santalumalbum.Van-Vigyan24: 25-28. Nagaveni HC, Srimathi RA. 1985. A note on haustoria-less sandal plants. Indian Forester111: 161-163. Pate JS, Kuo J,Davidson NJ. 1990a. Morphologyand anatomy of the haustoriumof the root hemiparasiteOlax phyllantii (Olaceae), with special referenceto the haustoricalinterface. Annals of Botany65: 425-436. Pate JS, Davidson NJ, Kuo J, Milburn JA. 1990b. Water relationsofthe roothemiparasiteOlax phyllantii(Labill) R.Br. (Olaceae), and its multiplehosts. Oecologia84: 186-93. Press MC, Parsons AN, Mackay AW, Vincent, CA, Cochrane V, Seel WE. 1993. Gas exchangecharacteristicsand nitrogen relationsof two Mediterraneanroot hemiparasites.Oecologia 95: 145-151. Press MC, Smith S, Stewart, GR. 1991. Carbon acquisitionand assimilationin parasiticplants.FunctionalEcology5: 278-283. Rama Rao M. 1918. Field investigationof 'spike' disease in sandal in the Kolimatai Hills. India Forester44: 58-65. Rao LN. 1942. Parasitismin the Santalaceae.AnnalsofBotany,6: 131-149. Riddoch I, Grace J, Fasehun FE, Riddoch B, Ladipo DO. 1991. Photosynthesisand successional statusof seedlingsin a tropical semi-deciduous rain forest in Nigeria. Journal of Ecology79: 491-503. Seel WE, Cooper RE, Press MC. 1993. Growth,gas exchange and water use efficiency of the facultativehemiparasiteRhinantus minorassociated with hosts differingin foliarnitrogen concentration.PhysiologiaPlantarum89: 64-70. Seel WE, Press MC. 1993. Influenceof the host on threesubArctic annual facultativeroot hemiparasites.New Phytologist 125: 131-138. Slob W. 1986. Strategiesin applyingstatistics in ecologicalresearch. Amsterdam:Free UniversityPress. Sokal RR, Rohlf FJ. 1981. Biometry, 2nd edition.New York: W. H. Freeman & Co. Struthers R, Lamont BB, Fox JED, Wijesuriya S, Crossland, T. 1986. Mineral nutritionof sandalwood. Journal of ExperimentalBotany37: 1274-1284. Subbarao NS, Yadav D, Anantha Padmanabha HS, Nagaveni HC, Singh CS, Kavimandan SK. 1990. Nodule haustoria and microbialfeaturesof Cajanus and Pongamiaparasitizedby sandal (Sandalwood). Plant and Soil 128: 249-256. Swaine MD & Hall JB. 1986. Foreststructureand dynamics.In: Lawson GW. ed. Plant Ecology in West Africa. Chichester: JohnWiley & Sons. Tennakoon KU, Pate JS. 1996. Heterotrophicgain of carbon from hosts by the xylem-tappingroot hemiparasite Olax phyllanthi(Olacaceae). Oecologia 105: 369-376. Tyree MT, Hammel HT. 1972. The measurementof turgor pressureand thewaterrelationsofplantsby thepressurebomb technique.JournalofExperimentalBotany23: 267-282. Veenendaal EM, Swaine MD, Lecha RT, Walsh MF, Abebrese IK, Owusu-Afriyi K. 1996. Growth responses of WestAfricanforesttreeseedlingsto irradianceand soil fertility. FunctionalEcology(in press). und Wagner H, Kreutzkamp B, Jurcic K. 1985. Inhaltstoffe Planta-Medica Pharmakologieder Okoubaka aubrevillei-rinde. 51: 404-407.