Paper 209 Disc Ecology Letters, (2001) 4 : 151±158 REPORT Emmanuelle Jousselin1,2* and Finn Kjellberg1 1 CNRS-CEFE (Centre d'Ecologie Fonctionnelle et Evolutive), 1919 route de Mende, 34293 Montpellier CeÂdex 5, France 2 Biology Department, Universiti Brunei Darussalam, Tungku Link, BE 1410, Brunei Darussalam, S.E. Asia *Correspondence: E-mail: Jousselin@cefe.cnrs-mop.fr The functional implications of active and passive pollination in dioecious figs Abstract Fig-pollinating wasps lay their eggs in fig flowers. Some species of fig-pollinating wasps are active pollinators, while others passively transfer pollen. In dioecious fig species, the ovules of male figs produce wasps but no seeds. By observations and experiments on four dioecious Ficus species we show that (i) passive pollinators distribute pollen haphazardly within figs, but fertilization of female flowers in male figs is inhibited. Consequently, wasp larvae will develop in nonfertilized ovules: they cannot benefit from pollination; (ii) active pollinators efficiently fertilize flowers in which they oviposit. Lack of pollination increases larval mortality. Hence, fig pollinators are not obligate seed eaters but ovule gallers. Active pollination has probably evolved as a way to improve progeny nourishment. Comparison of pollination and oviposition process in male and female figs, suggests that stigma shape and function have coevolved with pollination behaviour, in relation to constraints linked with dioecy. Keywords Ficus, agaonid wasps, mutualism, coevolution, dioecy, pollination, active pollination. Ahed Bhed Ched Ecology Letters (2001) 4 : 151±158 Dhed Ref marker Fig marker Table marActive pollination has been documented in three INTRODUCTION ker specialized plant±pollinator associations: the Yucca/yucca Ref end moth interaction (Aker & Udovic 1981; Pellmyr et al. Oviposition site often determines the amount and Ref start 1996), the Senita cactus/senita moth association (Fleming quality of resources available for offspring development. & Holland 1998) and the fig/agaonid wasp association Hence, any maternal behaviour increasing the quality (Galil & Eisikowitch 1969; Janzen 1979). The fig/agaonid of a given oviposition site will be favoured by selection. wasp mutualism is unique among these interactions in that Insects, such as gall formers, manipulate the growth of some agaonid species actively pollinate while others their host tissue during oviposition, to improve larval passively transfer pollen. nourishment (Price et al. 1986; Hartley 1998). In other Figs are closed inflorescences composed of uniovulate cases, insects use a biological agent. Some parasitoids female flowers and male flowers. When figs are receptive, inject viruses to control their host immune system agaonid wasps lay their eggs in some of the flowers and (Whithfield 1990) and some insects inoculate fungi on pollinate them in the process. When figs are mature, which their larvae will feed (Roy 1994). In a number of offspring female wasps get loaded with pollen and leave cases, insects have evolved morphological and betheir natal fig. Half of all Ficus species are monoecious, havioural adaptations aimed at collecting and depositwith both wasps and seeds produced in the same ing the biological agent that will benefit their offspring. inflorescence, while the remaining Ficus species are Fungus-feeding insects such as gall midges and ambrosia functionally dioecious. In these species, development of beetles have specialized `pockets' aimed at storing insects and seeds occurs on different trees. Female trees fungal spores (Rohfritsch 1997; see Pellmyr 1997 for bear figs with long-styled uniovulate flowers. Pollinators a review). Similarly, some insects whose larvae develop cannot lay their eggs in these figs because their ovipositor at the expense of seeds have specialized morphois shorter than the length of the styles and thus cannot logical and behavioural traits which allow them to reach the ovule (Galil 1973; Valdeyron & Lloyd 1979). pollinate their host (Pellmyr 1997). Such a syndrome is Male trees bear figs with male flowers and short-styled called active pollination. #2001 Blackwell Science Ltd/CNRS Paper 209 Disc 152 E. Jousselin and F. Kjellberg female flowers in which agaonid wasps can lay their eggs. Only wasps and pollen are produced in these figs: their female flowers either produce a wasp or remain empty. As wasps carry pollen from their natal fig, they contribute to the fig's male function. Hence male trees are morphologically hermaphroditic but functionally male. When pollination is active, wasps store pollen from their natal fig in thoracic structures called pollen pockets (Galil & Eisikowitch 1969). Each time a wasp lays an egg in an ovule, it also deposits some pollen grains on the stigmas (Galil & Eisikowitch 1969; Frank 1984). When pollination is passive, the pollinator does not show any pollen collection and deposition behaviour. Associated figs of these species produce large quantities of pollen so that the wasps become covered with pollen within their natal fig (Galil & Neeman 1977; Galil & Meiri 1981). Molecular phylogenies of fig-pollinating wasps (Machado et al., 2001) suggest that active pollination has evolved once and has reverted many times towards passive pollination. These multiple reversions suggest that the selective forces favouring active pollination can be relaxed and that there is a cost to active pollination. Why do some agaonid wasps pollinate actively? According to classical descriptions of the mutualism, fertilized flowers provide a better feeding substrate for wasp larvae (Verkerke 1989; Anstett et al. 1997; Herre 1999). Active pollination behaviour would have been selected to allow wasps to pollinate flowers in which they oviposit. The occurrence of active pollination in dioecious figs supports this hypothesis. In dioecious figs, wasps can only benefit from pollination of figs in which they reproduce, i.e. male figs. For pollination to provide some benefit to the wasp, it has to result in ovule fertilization. As male figs do not produce any seeds, it can thus be predicted that wasp offspring benefit from developing in fertilized flowers (Kjellberg et al. 1987) and that wasps only pollinate flowers in which an egg is laid, otherwise there would be many seeds formed. The situation in passively pollinated figs is likely to be very different. Passively pollinating wasps cannot control where pollen is deposited. It can therefore be predicted that pollen should be scattered among flowers independent of oviposition sites. If this prediction were to be verified, then one could ask what prevents seed production in these figs? The purpose of this paper is to test hypotheses about the benefit of pollination for the wasp and examine the consequences of pollen deposition pattern in both passively and actively pollinated dioecious fig species. To do so, we examined whether (i) flowers in which wasps attempt oviposition are more likely to receive pollen in male and female figs; (ii) pollination results in ovule fertilization in male figs and (iii) wasps develop normally in unpollinated figs. #2001 Blackwell Science Ltd/CNRS MATERIALS AND METHODS Species and study sites We studied two actively pollinated species (Ficus fulva and Ficus condensa) and two passively pollinated species (Ficus carica and Ficus deltoidea). Ficus condensa (section Sycocarpus, subgenus Sycomorus) and F. fulva (section Ficus, subgenus Ficus), belong to the two distinct lineages of dioecious figs (Weiblen 2000). Ficus deltoidea and F. carica both belong to subgenus Ficus but are pollinated by different wasp genera. All observations and experiments on F. condensa, F. deltoidea and F. fulva were performed in Brunei Darussalam (North Borneo). Experiments on F. carica were conducted in Montpellier (South of France). Do fig wasps pollinate flowers in which they attempt oviposition? We observed whether pollen grains were mainly deposited on the stigma of flowers in which pollinators laid an egg or attempted to lay an egg, i.e. flowers in which they had inserted their ovipositor. Figs were picked on F. condensa, F. fulva and F. deltoidea trees when pollinators were seen on the crop entering the ostiole. Figs were cut open and those in which a wasp was still alive were rejected, since foundresses might not have finished depositing the pollen they transported. This was done on one male tree and one female tree for each species. Controlled pollination experiments were performed on F. carica. To obtain receptive male figs at the same time as receptive female figs, we induced the early development of male figs by removing the apical bud of several branches. On the same day, we introduced one foundress per fig into male and female figs (method as in Khadari et al. 1995). Foundresses were collected from a single male tree. Figs were collected 24 h after pollinator introduction. Flowers were observed on the day of fig collection. In all species studied, the styles of flowers in which pollinators had inserted their ovipositor presented a conspicuous bruise. For each fig, we removed a set of flowers haphazardly chosen within the inflorescence. For each flower, we first assessed from the darkening of the style whether pollinators had attempted to oviposit in it. On 10 randomly chosen ``oviposited'' flowers and 10 randomly chosen ``virgin'' flowers, we assessed the presence/absence of pollen grains on stigmas. Flowers were stained with 0.01% aniline blue in 0.1 M K3PO4. Then, the stigma of each flower was removed from the ovule and squashed on a slide in order to visualize pollen grains under an epifluorescence microscope (Kearns & Inouye 1993). Ahed Bhed Ched Dhed Ref marker Fig marker Table marker Ref end Ref start Paper 209 Disc Active and passive pollination in figs 153 Do wasp larvae develop in fertilized ovules? On one male tree of F. condensa, we made controlled pollinator introductions. Figs were enclosed before pollination in a fine-mesh nylon bag to prevent wasp and parasite visits. When figs became receptive, we introduced one foundress per fig in 10 figs. Foundresses were collected from a single donor tree. Figs were collected 2 days after pollinator introduction and were preserved in FAA (formalin, acetic acid, ethanol). For F. deltoidea and F. fulva, figs were collected when pollinators were seen entering through the ostiole on a single male tree for each species. To let pollen tubes grow, collected figs were then kept for two days in a plastic vial with wet cotton inside. After two days, figs were put into FAA. For each species, 7±11 figs were embedded in paraffin blocks. A portion of each fig was cut into 15 mm transverse serial sections. Slices were then stained with 0.01% aniline blue in 0.1 M K3PO4 and observed under an epifluorescence microscope. For each fig, the fate of the ovules viewed on the median section of the portion of the fig cut (15±30 ovules per fig), was determined by compiling the information from 20 to 30 slides. For each ovule we noted (i) presence or absence of a pollinator egg and (ii) presence or absence of pollen tubes penetrating the ovule. Can fig wasp larvae develop in unpollinated figs? In F. condensa, we introduced wasps without pollen. To prevent wasps from collecting pollen, we removed the stamens of figs in which mated female wasps were about to emerge from their gall. Wasps were then allowed to emerge in fine-mesh bags. A subsample of 20 females used in each treatment were squashed on a slide and the presence of pollen grains in their pollen pockets was checked under a light microscope. The 20 from the sample of wasps used in the control experiment carried numerous pollen grains in their pockets (4 400 pollen grains per wasp). Two wasps out of 20 of the sample used in the `pollen free' experiment carried pollen grains. On one male tree, figs at the prepollination stage were enclosed in a fine-mesh nylon bag. On the day figs became receptive we either introduced a pollen-free wasp or a normally pollen loaded wasp into randomly chosen fig. All foundresses emerged from figs from the same male tree. A drop of nontoxic glue was applied to the ostiole after pollinator introduction to prevent foundress exit from a fig and re-entry into another fig (see Gibernau et al. 1996). Mesh bags were then replaced around figs to prevent attacks by parasitoids. Figs were collected 4 weeks later just before maturation. In each, the numbers of male wasps, female wasps, undeveloped flowers and bladders (swollen but empty flowers) were counted. In a few cases, wasps had already emerged from the fig prior to its collection. The number of galls that had been exited by wasps was then counted to assess total wasp production in these figs. Statistical analysis Test of independence between pollination and oviposition Data were classified in a contingency table according to three factors: fig, oviposition (presence/absence), pollen tube or pollen grains (presence/absence). We then fitted a log-linear model to the data assuming a Poisson error distribution and a log link function in the Glim statistical package (Glim 3.7i, 1985). To assess the significance of association between oviposition and pollination, we removed the two-way interaction term from the full model. The difference in deviance between two nested models follows a w2 distribution, with number of degrees of freedom equal to the difference in number of parameters between the two models (Crawley 1993). Pollen free experiment The effect of lack of pollen on fig production was tested with an analysis of covariance with number of flowers in a fig as a covariate and treatment as a fixed effect. We used the GLM procedure from the SAS statistical package (SAS 1992). RESULTS Do fig wasps pollinate flowers in which they attempt oviposition? Actively pollinated figs In both F. condensa and F. fulva, in male figs, most flowers in which pollinators had inserted their ovipositor were pollinated while those in which pollinators had not inserted their ovipositor were mostly not pollinated (Fig. 1). Pollen deposition and oviposition attempt were strongly associated (Table 1). Furthermore, most of the pollen grains had started to germinate (80% of all pollen grains observed for F. condensa and F. fulva). In female figs, about 60% of the ovules in which the wasps had attempted to oviposit were pollinated, compared to 50% of those that had not been probed. Nevertheless, oviposition attempt and pollination were dependent (Table 1). In female figs of F. fulva and F. condensa, stigmas stick together forming a synstigma (Fig. 2a) and we frequently observed for both species, pollen grains that landed on one stigma growing into a neighbouring style (Fig. 2b). In male figs of both species, stigmas were well separated and no such phenomenon was observed (Fig. 2c). #2001 Blackwell Science Ltd/CNRS Ahed Bhed Ched Dhed Ref marker Fig marker Table marker Ref end Ref start Paper 209 Disc 154 E. Jousselin and F. Kjellberg Figure 1 Pollen grain distribution among flowers of male and female figs of actively pollinated species. Table 1 Values of w2 for two-way inter- Data set Number of figs (number of flowers) w2 (1 d.f.) Test Actively pollinated figs F. condensa, male F. condensa, female F. fulva, male F. fulva, female 29 16 14 14 (560) (320) (280) (280) 236 4.51 184 5.58 P 55 1074 P = 0.03 P 55 1074 P = 0.02 Passively pollinated figs F. carica, male F. carica, female F. deltoidea, male F. deltoidea, female 6 7 13 10 (120) (140) (260) (60) 0.37 0.73 0.10 all flowers pollinated and probed P = 0.54 P = 0.39 P = 0.74 Passively pollinated figs The probability of a flower receiving pollen grains was the same whether or not wasps had inserted their ovipositor in the style of the flower for male and female figs of F. carica and male figs of F. deltoidea (Fig. 3, Table 1). In F. deltoidea male figs, only 20% of all pollen grains observed had germinated. Moreover, 70% of the resulting pollen tubes were blocked by a callose plug. In female figs of F. deltoidea, there are very few flowers (from 8 to 15) and all styles were darkened and had received numerous pollen grains that were germinating normally. For F. carica, although the number of female flowers in male figs and female figs is similar (& 1200; F. Kjellberg unpublished data), the percentage of pollinated flowers was higher for female figs (65 + 30%, 20 flowers per fig, 7 figs) than for male figs (18 + 8%, 20 flowers per fig, 6 figs) (w21df = 16.29; P 5 1074). The number of pollen grains per pollinated flower was significantly higher in female figs (mean = 3.3, SD = 2.5) than in male figs (mean = 1.4, SD = 0.5) (w21df = 7.29; P 5 1072). In female figs of F. deltoidea and F. carica, the stigmas do not #2001 Blackwell Science Ltd/CNRS action between pollen grain presence and oviposition attempt Ahed Bhed Ched Dhed Ref marker Fig marker Table marker Ref end stick together (Fig. 2d), pollen germination from one style Ref start into another is thus morphologically impossible. In male figs, stigmas are well individualized and many pollen grains had fallen between the flowers, on the sepals. Do fig wasp larvae develop in fertilized ovaries? In male figs of F. fulva and F. condensa, about 90% of the flowers in which an egg was laid were fertilized and few fertilized flowers did not contain a wasp egg (5% of all flowers) (Fig. 4). The presence of a pollen tube and the presence of a wasp egg in an ovule were closely associated (Table 2). In contrast, for F. deltoidea, among the 123 flowers analysed, only two were fertilized (Fig. 3). Do actively pollinating fig wasps develop normally in unpollinated figs? In F. condensa, male figs with and without pollen developed normally. Rates of fig abortion were identical in both treatments (3 figs out of 16). Unpollinated figs produced Paper 209 Disc Active and passive pollination in figs 155 Figure 2 (a) Synstigma in female figs of F. condensa: the stigmas are fused together. (b) Lateral pollen tube: growth on flowers of F. fulva female figs: pt, pollen tube; sty, style; stg, stigma. (c) Female flowers of male figs of F. condensa: stigmas are individualised. (d) Female fig of F. carica: stigmas are elongate and individualised. Ahed Bhed Ched Dhed Ref marker Fig marker Table marker Ref end Figure 3 Pollen grain distribution among flowers of male and female flowers of passively pollinated species. Ref start significantly (F1,23 = 5.10; P = 0.03) fewer agaonid offspring (mean = 67, S.D. = 57) than pollinated ones (mean = 110, S.D. = 55). The number of bladders was similar in pollinated and unpollinated figs (respectively, mean = 66, S.D. = 23; mean = 44, S.D. = 38; F1,23 = 0.94, P = 0.34). The number of flowers had no effect on fig production and did not need to be considered as a covariate. The proportion of males in the broods did not differ between treatments (unpollinated figs: mean = 0.30; S.D. = 0.11 n = 13; pollinated figs: mean = 0.22, S.D. = 0.09, n = 7; Mann±Whitney U-test, P = 0.69). If some foundresses considered as `pollen free' had carried some pollen then our results would give an underestimate of the cost of lack of pollination. DISCUSSION In dioecious fig species, as summarized in Table 3, passive and active pollination result in a strikingly different functioning of the fig/fig wasp mutualism. Active pollination and functional dioecy In the two actively pollinated fig species studied, in male figs, wasps mainly pollinate the flowers in which they oviposit. As a result, most flowers into which an egg has been laid are fertilized. In F. condensa, some wasps manage to develop in unpollinated figs, hence in unfertilized #2001 Blackwell Science Ltd/CNRS Paper 209 Disc 156 E. Jousselin and F. Kjellberg behaviour is similar in male and female figs (Galil 1973). We suggest that the tight packing of stigmas in female figs may limit the precision of pollen deposition. flowers, but brood size is strongly reduced by lack of pollination. Together, these observations imply that wasp offspring benefit from pollination and that this benefit is probably linked to the fertilization of the ovule in which the egg has been deposited. Fertilized ovules may provide a better feeding substrate for the developing larvae. Our observations also partly explain why there are no seed produced in actively pollinated male figs: the strong association between pollination and oviposition in male figs ensures that nearly every fertilized flower contains a wasp egg; thus, all fertilized flowers will develop as a gall. Interestingly, in female figs, much more pollen was deposited on flowers that had not been probed by the wasps than in male figs, despite the fact that wasp Passive pollination and functional dioecy In passively pollinated fig species, pollen is homogeneously distributed between probed and unprobed flowers in male and female figs. However, in male figs, flower fertilization is limited by poor adherence of pollen to the stigmas (data on F. carica) and limited pollen tube growth (F. deltoidea our data, F. carica, Beck & Lord 1988). This implies that most pollinators of F. deltoidea and F. carica will develop in unfertilized flowers. Consequently, they are true ovule gallers. The observation that ovule fertilization is partly prevented in passively pollinated male figs explains the lack of seed production and hence functional dioecy. Evolution of pollination mode in dioecious figs Actively pollinating wasps are probably selected to pollinate flowers in which they lay an egg in order to enhance larval nutrition. Nevertheless, oviposition alone has the capacity to induce the development of ovule tissue necessary to support the development of some wasps. This suggests that there is no absolute barrier to the loss of active pollination behaviour. This is consistent with the observation that reversions to passive pollination are numerous (Machado et al., in press). Although, the pathway leading to the loss of active pollination is still unknown, we can hypothesize that costs associated with pollination behaviour (time spent collecting and depositing pollen) sometimes exceed costs associated with development in unfertilized flowers. Figure 4 Pollen tube and wasp egg distribution across ovaries of male figs of F. condensa, F. fulva, F. deltoidea. Table 2 Values of w2 for two way Ficus species Number of figs (number of flowers) w2 (1 d.d.f.) F. condensa F. fulva F. deltoidea 9 (215) 11 (230) 7 (120) 174.4 24.7 only two flowers were fertilized Test P 55 1074 P 55 1074 interaction between pollen tube presence and wasp egg presence in male figs Table 3 Summary of results on pollen dispersion, flowers fertilization and benefit of pollen transport for the wasp Mode of pollination Fig sexual function Pollen dispersion Flower fertilization Benefit of pollination for the wasps Active Male Female Male Female Depends strongly of oviposition site Depends slightly of oviposition site Haphazardly distributed Haphazardly distributed Yes + no lateral pollen tube growth Yes + lateral pollen tube growth Rare Yes + no lateral pollen tube growth Inc.brood size None None None Passive #2001 Blackwell Science Ltd/CNRS Ahed Bhed Ched Dhed Ref marker Fig marker Table marker Ref end Ref start Paper 209 Disc Active and passive pollination in figs 157 The observation that, in passively-pollinated male figs, pollen germination is prevented, suggests that active pollination cannot re-evolve from passive pollination in dioecious figs; in such figs, wasps will simply never benefit from developing in fertilized flowers. This prediction is in agreement with the phylogenetic reconstruction of mode of pollination in fig wasps and the phylogenetic reconstruction of the reproductive system of the host fig (Machado et al., in press). Active pollination has evolved in the monoecious ancestors of dioecious figs, and it has been lost several times (in monoecious and dioecious figs). So far there is no evidence of reacquisition of active pollination in dioecious figs. flowers in which they oviposit while the host fig would be selected to ensure seed production. Indeed, actively pollinated monoecious figs present a synstigmatic structure (Jousselin and Kjellberg, unpublished observations on sections Americana, Conosycea, Galoglychia, Urostigma) and this structure is lacking in lineages of monoecious figs that have evolved passive pollination independently (Jousselin and Kjellberg, unpublished observations on sections Pharmacosycea and Conosycea). Hence, investigations of pollen distribution by actively pollinating wasps on monoecious figs will be necessary to test our hypothesis that synstigmas evolve in response to precise pollen deposition by the wasps. Coadaptation of wasp pollination behaviour and stigma morphology CONCLUSIONS In all male figs we have observed, including representatives of all dioecious Ficus sections, the stigmas are well separated and in all female figs of actively pollinated species we have observed, stigmas are cohesive (observations on sections Sycocarpus, Neomorphe, Ficus, Rhizocladus and Sycidium). In all female figs of passively pollinated figs we have observed, the stigmas are non cohesive (observations on sections Ficus, Rhizocladus and Kalosyce). Hence, according to available data, the style shapes described for actively and passively pollinated fig species are representative of dioecious figs. The difference in stigma structure between male figs and female figs, suggests that cohesive stigmas enhance female function (i.e. seed production). Furthermore, the lack of cohesive stigmas in passively pollinated female figs suggests that this structure is maintained by selection for increasing pollen dispersion when wasps actively pollinate, i.e. when they control where the pollen is deposited. Such a morphology would have no selective value in passively pollinated species, in which pollen is homogeneously distributed among flowers. Formal comparative analysis will be necessary to conclude on the coadaptation between stigma shape and wasp behaviour. The phylogeny of Ficus (Weiblen 2000) is not yet sufficiently resolved to establish whether passive pollination evolved several times in dioecious figs. Thus, we cannot test whether independent losses of active pollination lead to the loss of synstigma in female figs. Hence, no general conclusion may be drawn from the sole investigation of dioecious fig species. Hypothesized coadaptation of stigma arrangement with mode of pollination may, however, be tested by investigating monoecious species which produce both seeds and wasps in the same figs. The selective pressures in monoecious figs could be convergent with those evidenced for female figs. Actively pollinating wasps may be selected to only fertilize the The evolution of active pollination in the fig/fig wasp mutualism does not follow the same pathway as that in the Yucca/yucca moth and the Senita/senita moth interactions. In these latter, pollinator larvae feed on several seeds produced in their natal flower. Hence, in the absence of copollinators, pollination of the host by the mother is essential for larval survivorship. In contrast, agaonid wasp larvae complete their development in a single ovule of the fig inflorescence, and response to selection for increasing the quality of their oviposition site includes fertilization of the ovule but also manipulation of the flowers. Studies on other actively pollinating insect lineages had concluded that seed eating was a prerequisite for the evolution of active pollination (Pellmyr et al. 1996). Active pollination in the fig/agaonid wasp mutualism is more likely to have evolved from ovule parasitism. ACKNOWLEDGEMENTS This research was supported by a Brunei Shell Environmental Studies Fellowship. The authors are grateful to Dr D. Edwards and to the Biology Department staff at Universiti Brunei Darussalam for their help. They also acknowledge the Director of the Brunei Museum and the Forestry Department for allowing them to export specimens. Many thanks to C. Maycock and all the staff at the Kuala Belalong Field Studies Center for field assistance, and to C. Brouat, M. Hossaert-Mckey, Doyle Mckey, John Thompson, for critically reading earlier versions of this manuscript. REFERENCES Aker, C.L. & Udovic, D. (1981) Oviposition and pollination behavior of the yucca moth, Tegeticula maculata, (Lepidoptera: Prodoxidae) and its relation to the reproductive biology of Yucca whipplei (Agavacae). Oecologia, 49, 96±101. #2001 Blackwell Science Ltd/CNRS Ahed Bhed Ched Dhed Ref marker Fig marker Table marker Ref end Ref start Paper 209 Disc 158 E. Jousselin and F. Kjellberg Anstett, M.-C., Hossaert-McKey, M. & Kjellberg, F. (1997) Figs and fig pollinators: evolutionary conflicts in a coevolved mutualism. Trends Ecol. Evol., 12, 94±99. Beck, N.G. & Lord, E.M. (1988) Breeding system in Ficus carica, the common fig. II Pollination events. Am. J. Botany, 75, 1913±1922. Crawley, M.J. (1993) GLIM for Ecologists. Blackwell Scientific Publications, Oxford, UK. Fleming, T.H. & Holland, J.N. (1998) The evolution of obligate pollination mutualisms: Senita cactus and Senita moth. Oecologia, 114, 368±375. Frank, S.A. (1984) The behaviour and morphology of the fig wasps Pegoscapus assuetus and P. jimenezi: descriptions and suggested behavioral characters for phylogenetic studies. Psyche, 91, 289±308. Galil, J. (1973) Pollination in dioecious figs. Pollination of Ficus fistulosa by Ceratosolen hewitii. Gardens' Bulletin, 26, 303±311. Galil, J. & Eisikowitch, D. (1969) Further studies on the pollination ecology of Ficus sycomorus L. (Hymenoptera, Chalcidoidea, Agaonidae). Tijdschrift Voor Entomologie, 112, 1±13. Galil, J. & Meiri, L. (1981) Number and structure of anthers in fig syconia in relation to behaviour of the pollen vectors. New Phytologist, 88, 83±87. Galil, J. & Neeman, G. (1977) Pollen transfer and pollination in the common fig (Ficus carica L.). New Phytologist, 79, 163±171. Gibernau, M., Hossaert-McKey, M., Anstett, M.-C. & Kjellberg, F. (1996) Consequences of protecting flowers in a fig: a one-way trip for pollinators? J. Biogeogr., 23, 425±432. Hartley, S.E. (1998) The chemical composition of plant gals: are levels of nutrients and secondary compounds controlled by the gall-former? Oecologia, 113, 492±501. Herre, E.A. (1999) Laws governing species interactions? Encouragements and caution from figs and their associates. Levels of Selection (by Keller, L.). Princeton University Press, Princeton, NJ, USA. Janzen, D.H. (1979) How to be a fig. Ann. Rev. Ecol. Syst, 10, 13±51. Kearns, C.A. & Inouye, D.W. (1993) Techniques for Pollination Biologists. University Press of Colorado, Niwot, CO. Khadari, B., Gibernau, M., Anstett, M.C., Kjellberg, F. & Hossaert-McKey, M. (1995) When figs wait for pollinators: the length of fig receptivity. Am. J. Botany, 82, 992±999. Kjellberg, F., Michaloud, G. & Valdeyron, G. (1987) The Ficus± pollinator mutualism: how can it be evolutionary stable? Insect #2001 Blackwell Science Ltd/CNRS Plants (eds Labeyrie, V., Fabres, G. & Lachaise, D.), pp. 335± 340. W. Surck Publishers, Dordrecht, The Netherlands. Machado, C., Jousselin, E., Kjellberg, F., Compton, S.G. & Herre, E.A. (2001) Phylogenetic relationships, historical biogeography and character evolution of fig pollinating wasps. Proc. Roy. Soc. Lond. B., 268, 1±10. Pellmyr, O. (1997) Pollinating seed eaters: why is active pollination so rare? Ecology, 78, 1655±1660. Pellmyr, O., Thompson, J.N., Brown, J.M. & Harrison, R.G. (1996) Evolution of pollination and mutualism in the yucca moth lineage. Am. Nat., 148, 828±845. Price, P.W., Waring, G.L. & Fernandes, G.W. (1986) Hypotheses on the adaptive nature of galls. Proc. Entomol. Sic. Wash., 88, 361±363. Rohfritsch, O. (1997) Morphological and behavioural adaptations of the gall midge Lasioptera arundinis (Schiner) (Diptera, Cecidomyiidae) to collect and transport conidia of its fungal symbiont. Tijdschrift Voor Entomologie, 140, 59±66. Roy, B.A. (1994) The use and abuse of pollinators by fungi. Trend Ecol. Evolut., 9, 335±339. SAS (1992) SAS User's Guide. SAS Institute Inc, Cary, NC, USA. Valdeyron, G. & Lloyd, D.G. (1979) Sex differences and flowering phenology of the common fig, Ficus carica L. Evolution, 33, 673±685. Verkerke, W. (1989) Structure and function of the fig. Experientia, 45, 612±622. Weiblen, G.D. (2000) Phylogenetic relationships of functinally dioecious Ficus (Moraceae) based on ribosomal DNA sequences and morphology. Am. J. Botany, 87, 1342±1357. Whithfield, J.B. (1990) Parasitoids, polydnaviruses and endosymbiosis. Parasitol. Today, 6, 381±384. BIOSKETCH Emmanuelle Jousselin's main research interest is the evolutionary ecology of mutualistic interactions. Her research involves experimental ecology but also phylogeny and comparative analysis. Editor, M. Parker Manuscript received 25 October 2000 First decision made 7 December 2000 Manuscript accepted 9 January 2001 Ahed Bhed Ched Dhed Ref marker Fig marker Table marker Ref end Ref start