Pollination of Ficus elastica: India rubber re-establishes sexual reproduction in Singapore

Ficus elastica, otherwise known as India Rubber (although its geographical origins are unclear), was an important source of latex in the early 19th century and was widely cultivated in tropical Asia. Like all figs, F. elastica is dependent on tiny, highly specific wasps for pollination, and detailed studies based out of Singapore in the 1930s suggested that through the loss of its pollinator F. elastica was extinct in the wild. However, around 2005 wild seedlings of F. elastica began appearing in Singapore. We identified the pollinator as Platyscapa clavigera, which was originally described from F. elastica in Bogor in 1885. A visit to Bogor Botanical Gardens revealed that not only was F. elastica being pollinated by P. clavigera in the gardens, but there was clear evidence it had been reproducing naturally there over many decades. Although Singapore has a native fig flora of over 50 species, F. elastica went unpollinated for at least 70 years and probably from the time it was introduced during the 19th century. These observations illustrate the extraordinary specificity of this interaction and, through the fig’s ability to wait for its pollinators, demonstrates one way in which such highly specific interactions can be evolutionarily stable.

enabling the female fig wasps to enter, although they lose their antennae and wings in the process. Once inside, the wasps pollinate the uniovulate female flowers and lay eggs in some of them. An ovule that receives a wasp egg develops into a gall, which protects and nourishes the larva within. Ovules of pollinated flowers that do not receive a wasp egg develop into seeds in the normal way. The system is an example of seed predator pollination, whereby the plant gives up a proportion of its seeds to feed developing pollinator larvae 10 . Approximately one month later, the adult male wasps chew their way out of their galls and mate with the gall-enclosed females. The females then emerge, collecting pollen on the way, and disperse in search of a receptive fig of the same species in which to lay their eggs. The female wasps use wind assisted dispersal [11][12][13] , sometimes traveling huge distances 14 , before locating a tree with receptive figs of the correct species by identifying the specific floral bouquet 15 [16][17][18] , sympatric pollination by two or more pollinator species 19,20 and the existence of cheaters (co-"pollinating" fig wasps that do not pollinate) 21 have been described. The genus is estimated to be slightly over 75 million years old and phylogenetic studies indicate a general (but imperfect) pattern of co-speciation down the fig and pollinator lineages 22 . As mentioned above, because F. elastica was never observed to produce seeds (and wild seedlings were also never observed), Corner assumed that the pollinator must be extinct and that, therefore, F. elastica was extinct in the wild 6 .
Around 2005 in Singapore, young plants of F. elastica began appearing, including some in odd places such as on top of a bus shelter, indicating a production of wild seedlings. A quick investigation revealed the presence of seeds inside the figs of  3) The original pollinator of F. elastica was in fact not extinct, but had only recently colonised Singapore. We set out to examine these three hypotheses and, in the process, shed some light on the biology of F. elastica.

Results
Microsatellite analyses demonstrated that wild seedlings of F. elastica in Singapore were not clones (Figs S1 and S2). Seedlings inherited some alleles from a nearby adult, presumably the maternal parent, but other alleles came from other non-identified individuals. This demonstrated that F. elastica is breeding in Singapore, and hence we could reject our first hypothesis.
To decide between the second and third hypotheses, we collected emerging fig wasps and conducted a morphological examination of the pollinator. A pollinator was described from F. elastica in Bogor in 1885 23,24 . Unfortunately, the type material was lost and hence we had to rely on the description, including drawings. Nevertheless, a comparison between the wasps collected by us in Singapore and the description shows they are remarkably similar (Figs 1 and 2). Key characters for identifying female fig pollinators include the antennae (Fig. 1a), the mandibular appendage (a teeth-like structure attached to the underside of the mandibles, which helps the wasps pull themselves through the bracts when entering the fig) (Fig. 1b) and the shape of the stigmal vein on the fore wings (Fig. 1c). Male fig pollinators have very reduced traits making them difficult to identify. However, again in the shape and relative size of the antennae, the dorsal view of the head and the dorsal view of the mesosoma (Fig. 2b), the pollinators collected from Singapore are similar to the 1885 description of the pollinator of F. elastica from Bogor. Hence, we tentatively identified the pollinators collected in Singapore as Platyscapa clavigera (Mayr 1885) (it was originally named as Blastophaga clavigera) 24 .
To check on this identification we visited Kebun Raya Bogor (Bogor Botanic Gardens) in 2012 and inspected the F. elastica growing there. We found several individuals bearing pollinated figs and inspection of the pollinating wasps found them to be the same species as those collected from Singapore. Moreover, the presence of several large stranglers indicated that F. elastica had been reproducing naturally in the gardens over many years (Figs 3 and 4). Stranglers are hemi-epiphytes, meaning they begin life as an epiphytic seedling in the fork of a large tree and then send roots down the host truck to connect with the ground 25 . In mature individuals it is possible to deduce the position on the host tree of the original epiphytic seedling, as this is the point at which the aerial roots connect to the trunk and branches (Fig. 3). Several of the F. elastica in the Kebun Raya Bogor had aerial roots extending over 15 m up the trunk of their host, making it inconceivable that someone had planted them. Moreover, the size of some individuals ( Fig. 4) suggested they were several decades, possibly a century or more, in age.

Discussion
We conclude that the pollinators of F. elastica collected in Singapore are P. clavigera 23,24 . Moreover, based on our observations of large stranglers in the Kebun Raya Bogor we suggest that P. clavigera has persisted in Bogor for a long time. Since making these observations we discovered recent reports of natural reproduction in F. elastica in Ujong Kulong in the extreme west of Java 26 and Thung Yai Naresuan East Wildlife Sanctuary in Umphang district, Thailand 27 . However, neither report established the identity of the pollinators. Both these sites have calcareous soils lending support to the suggestion that F. elastica is a limestone specialist in its natural condition. As mentioned earlier, in 1874 Strettell also found both large individuals and epiphytic seedlings of Ficus elastica in a limestone forest in north Burma, whereas elsewhere he reports finding only planted individuals 2 . We have inspected dozens of F. elastica individuals bearing figs in urban areas in Peninsular Malaysia, Thailand and Yunnan but we have yet to observe any that have been pollinated. Hence, we suggest that, in line with Corner's suggestions 4 , F. elastica probably originated as a limestone specialist and spread to other areas through cultivation. If we assume this to be correct, its known natural distribution includes western Java, southern Sumatra, Peninsular Malaysia, western Thailand and north Burma 2, 4, 26, 27 . The alternative hypothesis that P. clavigera is a lowland moist forest species that became extinct throughout a native range centered on NE India and Burma 5 but persisted in Bogor and isolated limestone areas seems much less likely. Although local, temporary extinctions of some fig pollinators have been recorded in extreme climatic events 28 , these did not involve Platyscapa and related genera, which are capable of long distance wind-assisted dispersal [11][12][13][14] 14 . They have also been recorded several tens of kilometers offshore at ships lights and have been quick to colonise exploded volcanic islands 31,32 , indicating that sea crossings would not appear to be a barrier to dispersal. Hence, it seems surprising that P. clavigera was present in the region, but failed to reach Singapore until 2005. It is around 70 years since Corner's observations, but it seems likely that the pollinator has been absent from the time F. elastica was first introduced to the island during the 19 th century. Nevertheless, to reach Singapore P. clavigera would have had to disperse across extensive lowland moist forests that harbour few or no F. elastica. It is also quite possible that habitat specialisation presented a barrier to successful colonisation of Singapore by P. clavigera until recently. Throughout a large part of East Africa, Ficus sur is pollinated by two wasp species, one of which occurs in forested areas and one in savannas 33 .
The fact that F. elastica was present in Singapore but went unpollinated for at least 70 years but probably over 150 years, also illustrates the extreme specificity of the interaction. There are about 50 native species of fig in Singapore 8 , including several species pollinated by Platyscapa and related genera. Elsewhere, it has been found that in the absence of the legitimate pollinator a low proportion of figs may be entered by the pollinators of other species, indicating a breakdown in specificity 30,34 . Fig species with multiple sympatric pollinators have also been discovered 19,20 . However, the specificity of the F. elastica -P. clavigera interaction was maintained over many decades, despite ample opportunity for breakdown. It would be interesting to investigate the degree of specificity across a large sample of fig -fig pollinator interactions and examine the ecological and evolutionary correlates. Natural populations of F. elastica and P. clavigera would also appear to be isolated in widely separate patches of limestone forest and therefore would make an interesting topic for a population genetic study 18 .
Last, the F. elastica -P. clavigera story illustrates how such extremely specialised interactions can also be evolutionarily stable. Although F. elastica in Singapore were cultivated, individual trees can live at least a century and possibly much longer. The growth rates of large stranglers are not well understood 25,35 , but it seems unlikely that an individual as colossal in size as the one by the pond in the Kebun Raya Bogor (Fig. 4) could be anything less than several centuries old (the oldest fig of known planting date is a Bodhi tree (F. religiosa) which was planted in Sri Lanka in 255 BC) 36 . Although almost certainly reaching Singapore by artificial means, F. elastica has been able to consolidate this range extension by waiting for its pollinator to catch up. There are also some examples of In these cases, presumably the pollinators were also transported via the horticultural trade, as larvae in developing figs. The ability of figs to wait for their specific pollinator to catch up over decades, and possibly centuries, has likely contributed to the stability of the interaction over evolutionary time-scales as different lineages dispersed around the globe 22 . Thus, after an extended period of abstinence, India Rubber has rediscovered sex in Singapore.

Methods
Microsatellite analyses. The leaves of F. elastica were collected on 8 June 2011 from two large adult trees and two spontaneously growing seedlings found in the vicinity of the Chinese Heritage Centre (1.343820°N, 103.684023°E) of at the Nanyang Technological University campus, Singapore. Leaf samples were surface-sterilized with 30% bleach then autoclaved milliQ water for 10 minutes. The sterilized leaf samples were then stored in a −80 °C freezer on the same day.
Prior to DNA extraction, the four leaf samples of F. elastica were cut into pieces. Then, 0.5 g of the cut leaves of each sample was ground with liquid nitrogen using a mortar and pestle. The frozen and ground leaf tissue was subsequently used for DNA extraction, following standard protocols 40,41 with minor modifications. We added 600 μL of extraction buffer to the ground tissue and incubated for 60 °C for 25 min. We then added 720 μL of chloroform-octanol and inverted the tubes 20 times. The emulsion was spun at 13,000 rpm for 15 min, and then 400 μL of the aqueous phase was transferred to a new 1.5 mL tube. To this we added 200 μL of 5 M NaCl, 800 μL of −20 °C 95% ethanol, and placed the tube in ice for 1 h. The tube was then spun at 13,000 rpm for 10 min and the supernatant was poured out. The remaining pellet was washed with 50 μL of 0 °C ethanol and spun again at 13,000 rpm for 10 min. The ethanol was removed and the pellet dried by leaving it uncovered for 20 min at 37 °C. Finally, the pellet was dissolved with 30 μL of milliQ water. Eight microsatellite primer pairs developed from Ficus carica 42 were used to amplify DNA extracted from these leaves. The Polymerase Chain Reaction (PCR) amplification protocol also followed that of 34

Fig wasp morphological observations.
To produce high quality images, some specimens were point-mounted on grey card in order to avoid loss of contrast caused by white background. Images were produced with an EntoVision Premium Portable Imaging System, comprising a Leica M16 zoom lens, a JVC KY-75U 3CCD digital camera and a portable computer workstation running EntoVision Imaging Suite software (GT Vision, Hagerstown, MD U.S.A.). Cartograph v5.6.0 (Microvision, Evry, France) software was subsequently used to merge an image series (representing about ten to twenty focal planes), producing a single image with increased depth of field. Illumination was achieved using a "quadrant" setup, with four fibre optic light guides stemming from two individual light "sources (Leica CLS 150 X). Images were edited using Adobe Photoshop CS4 software.