Journal of Animal Ecology 2008, 77 , 927���935 doi: 10.1111/j.1365-2656.2008.01416.x �� 2008 The Authors. Journal compilation �� 2008 British Ecological Society Blackwell Publishing Ltd Longevity, early emergence and body size in a pollinating fig wasp ��� implications for stability in a fig���pollinator mutualism Derek W. Dunn 1,2,3 *, Douglas W. Yu 2 , Jo Ridley 2 and James M. Cook 1,3 1 Division of Biology, Imperial College London, Silwood Park, SL5 7PY, UK 2 School of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, UK and 3 School of Biological Sciences, University of Reading, Reading RG6 6AH, UK Summary 1. Fig trees ( Ficus ) are pollinated only by agaonid wasps, whose larvae also gall fig ovules. Each ovule develops into either a seed (when pollinated) or a wasp (when an egg is also laid inside) but not both. 2. Ovipositing wasps (foundresses) favour ovules near the centre of the enclosed inflorescence (syconium or ���fig���), leaving ovules near the outer wall to develop into seeds. This spatial stratification of wasps and seeds ensures reproduction in both partners, and thereby enables mutualism persistence. However, the mechanism(s) responsible remain(s) unknown. 3. Theory shows that foundresses will search for increasingly rare inner ovules and ignore outer ovules, as long as ovipositing in outer ovules is sufficiently slow and/or if inner ovules confer greater fitness to wasps. The fig���pollinator mutualism can therefore be stabilized by strong time constraints on foundresses and by offspring fitness gradients over variation in ovule position. 4. Female fig wasps cannot leave their galls without male assistance. We found that females in outer ovules were unlikely to be released. Inner ovules thus have added value to foundresses, because their female offspring are more likely to mate and disperse. 5. For those offspring that did emerge, gall position (inner/outer) and body size did not influence the order in which female pollinators exited syconia, nor did early emerging wasps enjoy increased life spans. 6. We also found that the life spans of female wasps nearly doubled when given access to moisture. We suggest that conflict resolution in the fig���pollinator mutualism may thus be influenced by tropical seasonality, because wasps may be less able to over-exploit ovules in dry periods due to time constraints. Key-words: cooperation, Ficus , fig wasp, mutualism, optimal foraging, Pleistodontes imperialis Introduction Mutualisms are interspecific interactions in which both partners benefit (Janzen 1985 Herre et al. 1999 West, Griffin & Gardner 2007a) and are important contributors to global biodiversity at multiple trophic levels (e.g. Christian 2001 Brundrett 2004 Harrison 2006). Although hosts and symbionts both profit from their association, benefits are usually obtained by imposing costs onto the other partner. Why, then, does one partner not destabilize the mutualism by exploiting the other unsustainably (Herre et al . 1999 Anstett 2001 Yu 2001 Frank 2003 Foster & Wenseleers 2006)? The diversity of mutualisms has meant that no all-encompassing explanation for their maintenance has been forthcoming until recently (see Foster & Wenseleers 2006 West, Griffin & Gardner 2007b). In some mutualisms hosts can often favour those symbionts that are cooperative and impose sanctions on cheats (e.g. Kiers, Rousseau, West & Denison 2003 Edwards et al. 2006). More generally, it has been shown that a high benefit : cost ratio of cooperation, high within-species relatedness and high between-species fidelity are important factors contributing to mutualism stability (Foster & Wenseleers 2006). Fig trees ( Ficus spp.) are important in maintaining diversity in tropical and subtropical forests (Janzen 1979 Cook & West 2006 Harrison 2006). Their association with pollinating wasps (Hymenoptera: Agaonidae) is an obligate mutualism *Correspondence author: E-mail: d.w.dunn@reading.ac.uk

928 D. W. Dunn et al. �� 2008 The Authors. Journal compilation �� 2008 British Ecological Society, Journal of Animal Ecology , 77 , 927���935 because neither partner can reproduce without the other (Cook & Rasplus 2003). Most of the 750 + Ficus species have their pollination requirements fulfilled by a single species- specific wasp species (but see Machado et al. 2005), and thus show considerable between-species fidelity. The enclosed Ficus inflorescences (syconia or ���figs���) provide food for the wasps��� larvae. Approximately 50% of Ficus species are dioecious, in which wasps develop only in ���male��� trees and thus do not inflict a significant reproductive cost to the plant (Cook & Rasplus 2003). However, in monoecious Ficus , both wasps and seeds develop in the same syconium. Ovipositing females (foundresses) lay only one egg per female flower and they do this by inserting their ovipositor down the style (Jousselin, Hossaert-McKey, Vernet & Kjellberg 2001). One wasp off- spring thus costs the tree one seed, and this highlights the inherent conflict in the mutualism. To reproduce, monoecious Ficus need to produce both wasps (to disperse pollen) and seeds from the same syconium, but the wasps gain in the short term only by exploiting the tree���s ovules. Selection should favour increased ovule exploitation by wasps, which will eventually destabilize the mutualism (Anstett 2001). Despite nearly 40 years of research, however, the mechanism(s) preventing over-exploitation of monoecious figs by their pollinators remains unidentified (Jousselin, Kjellberg & Herre 2004 Yu et al. 2004). The style lengths of flowers in the receptive syconia of monoecious Ficus are highly variable (Janzen 1979 Ganeshaiah et al. 1995 Ganeshaiah, Kathuria & Shaankar 1999). Long- styled flowers (referred to hereafter as ���outer ovules���) develop into wasp galls or seeds at the mature syconium���s outer wall, while short-styled flowers (referred to hereafter as ���inner ovules���) become galls or seeds near to the syconium���s central cavity (the ���lumen���) (Bronstein 1988 Nedft & Compton 1996). Most outer ovules remain unexploited by pollinator wasps and develop into seeds, whereas inner ovules are used by ovipositing wasps and most develop into galls, which suggests that foundresses favour inner ovules for oviposition (Herre 1989 West & Herre 1994 Nedft & Compton 1996 Jousselin et al . 2001, 2004). The four main hypotheses explaining why pollinators do not over-exploit figs are based on this spatial stratification of wasps and seeds: (1) unbeatable seeds ��� some (outer) ovules are protected physically or biochemically against wasp destruction (West & Herre 1994). However, no empirical data support this notion. (2) Short ovipositors ��� pollinators��� ovipositors are too short to reach outer ovules (Galil & Eisikowitch 1968 Ganeshaiah et al . 1999 Jousselin et al . 2004) however, this is not so in several wasp species (Bronstein 1988 Nedft & Compton 1996 Herre 1999). (3) Insufficient eggs ��� because wasps disperse long distances, too few foundresses enter receptive syconia to exploit all ovules (Nedft & Compton 1996) moreover, the ostiole (used by wasps to enter the receptive syconium) may close to prevent excess foundresses from entering (Herre 1989 Nedft & Compton 1996). However, in many Ficus more foundresses enter receptive syconia than are required to exploit all ovules, yet seeds are still set and the mutualism is maintained (Anstett, Bronstein & Hossaert-McKey 1996 Herre 1999). Yu et al . (2004) provided a new hypothesis to resolve the fig���pollinator problem by focusing upon how foundresses ���optimally forage��� for oviposition sites. If inner ovules are more profitable to ovipositing foundresses, through shorter handling times or because they convey higher fitness to wasp offspring, then foundresses should focus their oviposition initially on inner ovules. As search times for egg-free inner ovules increase foundresses will, at some point, expand their preference and also begin to utilize outer ovules. The more profitable that inner ovules are compared with outer ovules, the rarer egg-free inner ovules must be before foundresses begin to exploit outer ovules. One potential cause of variation in ovule profitability is variation in handling time caused by variation in style lengths. Outer ovules have long, narrow styles (Jousselin et al . 2001, 2004), so foundresses should take longer to deposit an egg compared with inner ovules (Yu et al . 2004). The resulting gradient in profitability has an important consequence for the stability of fig-wasp mutualisms. Achieved fitness is predicted to decrease with each succeeding foundress because early foundresses are able to lay in inner ovules, which take less time, leaving mainly outer ovules for late foundresses. As a result, the number of ovules that receive an egg is less than a straight-line extrapolation of the first foundress���s fecundity (Yu et al . 2004). Moreover, if a gradient in ovule profitability is caused by differences in offspring fitness, such that eggs laid in inner ovules have the highest value, foundress fecundity can become disconnected from foundress fitness. In the extreme, late-arriving foundresses can be selected to devote much of their short life spans (Kjellberg, Doumesche & Bronstein 1988) to searching for unexploited inner ovules, therefore lay- ing fewer total eggs than if they had oviposited randomly (Yu et al . 2004). The fig host thus ���controls��� its symbiont indirectly by preventing resource over-exploitation (Yu 2001 Kiers et al . 2003 Edwards et al . 2006). It does this by providing ovules of highly variable profitability to foundresses, which decreases the benefits to non-cooperative foundresses, i.e. those that exploit outer ovules (see Foster & Wenseleers 2006). However, there have been few attempts (Anstett 2001 Dunn et al. 2008) to measure the relative value of inner and outer ovules to foundresses. There are also few data on pollinating fig wasp longevity, with only two published studies (Kjellberg et al . 1988 Compton, Rasplus & Ware 1994). Longevity may have added importance if the time available to pollinators to lay eggs in the syconium varies between the tropical wet and dry seasons, because moisture availability often affects insect longevity (e.g. Hicks, Hagenbuch & Meffert 2004). This paper has two aims: (1) to measure the effects of humidity on female pollinator longevity and (2) to examine how ovule position affects two aspects of offspring fitness: (i) the likelihood of female offspring remaining trapped in the syconium and (ii) rapid emergence from the syconium. The highly specialized nature of fig-wasp behaviour within the syconium also suggests that female body size may be associated with rapid emergence. We therefore conducted an additional investigation