Phenological Adaptations in Ficus tikoua Exhibit Convergence with Unrelated Extra-Tropical Fig Trees

Flowering phenology is central to the ecology and evolution of most flowering plants. In highly-specific nursery pollination systems, such as that involving fig trees (Ficus species) and fig wasps (Agaonidae), any mismatch in timing has serious consequences because the plants must balance seed production with maintenance of their pollinator populations. Most fig trees are found in tropical or subtropical habitats, but the dioecious Chinese Ficus tikoua has a more northerly distribution. We monitored how its fruiting phenology has adapted in response to a highly seasonal environment. Male trees (where fig wasps reproduce) had one to three crops annually, whereas many seed-producing female trees produced only one fig crop. The timing of release of Ceratosolen fig wasps from male figs in late May and June was synchronized with the presence of receptive figs on female trees, at a time when there were few receptive figs on male trees, thereby ensuring seed set while allowing remnant pollinator populations to persist. F. tikoua phenology has converged with those of other (unrelated) northern Ficus species, but there are differences. Unlike F. carica in Europe, all F. tikoua male figs contain male flowers, and unlike F. pumila in China, but like F. carica, it is the second annual generation of adult wasps that pollinate female figs. The phenologies of all three temperate fig trees generate annual bottlenecks in the size of pollinator populations and for female F. tikoua also a shortage of fig wasps that results in many figs failing to be pollinated.


Introduction
The times of year when plants flower and set seed are not random, even among plant species growing in relatively aseasonal tropical environments [1]. Flowering phenology is subject to selection from a combination of abiotic, biotic and intrinsic factors linked to life history, and also to the plant's phylogeny [2,3]. Abiotic factors include constraints imposed by physiological responses to temperatures, day lengths and other climatic variables, while biological factors include the availability of pollinators and seed dispersal agents and competition with other plants flowering at the same time [4,5]. In temperate latitudes, strong climatic seasonality provides particular constraints, with threshold temperatures limiting both insect pollinator activity and the length of the period when floral and seed development can continue. This has led to widespread convergence in flowering times, most noticeably with a spring-time concentration of flowering [6,7], despite potential competition for pollinators among animal-pollinated plants and increased likelihood of receipt of heterospecific pollen from other species flowering at the same time. Variation in flowering times also has evolutionary implications, potentially contributing to reproductive isolation and speciation [8,9,10].
The time of year when flowers are available to be pollinated is especially important for plants which depend on one or a small number of insect species for pollination [11]. In the case of nursery pollination systems, where the reward provided by the plant is a place for the insects to breed, any mismatch in timing has serious consequences for the population dynamics of both partners in the mutualism [12].  [13]. The significance of figs for vertebrates results from their structure, which makes them easy to eat, their abundance in a variety of habitats and what is often an all-year round fruiting phenology that makes figs available at times of the year when the fruits of other plants are absent. The all-year fruiting phenology of many Ficus species may be linked to their unique pollination system, because it helps maintain populations of their pollinator fig wasps, with which they have an obligate association [14]. Highly reciprocal adaptive traits and co-evolutionary dynamics are ubiquitous between fig hosts and their pollinators, and continue to stimulate ecological and evolutionary questions [15].
Adult female pollinating wasps do not feed and only survive for one or two days after emerging from their natal figs [16]. As a consequence, the synchronization of pollinator release with the production of receptive figs is critical for both the maintenance of pollinator populations and pollination of the figs [17].  [18,19].
Latitudinal trends in the flowering times of plants are well documented, with selection for example tending to favour earlier flowering among plants growing at higher latitudes [20,21]. The vast majority of Ficus species have tropical or subtropical distributions [14], a latitudinal range that appears to be related to global temperature patterns, because during a warmer period of Earth history they were also present in northern Europe [22]. A small number of fig tree species currently extend to higher latitudes [23], where they have evolved atypical fruiting phenologies in response to the strong seasonality of their environments, in particular the long winter periods that are too cold for fig wasps to be dispersing between trees.
The fruiting phenologies of three species of dioecious fig trees with largely extra-tropical distributions have been described, F. erecta and F. pumila in China including Taiwan, and F. carica in Europe [24,25,26]. All three species belong to Ficus subgenus Ficus [27] and are passively pollinated by species of Blastophaga and Wiebesia belonging to Agaonidae, Subfamily Agaoninae [25,[28][29][30][31]. The three species produce relatively synchronized crops, population wide, at set periods each year. The resulting precise matches between the phenologies of male and female trees facilitate the pollination of female figs while at the same time maintaining pollinator populations [25,30]. In Europe, the pollinator of F. carica has larvae that overwinter in one crop of male figs. They become adults in the spring, a time when there are many receptive male figs available to lay their eggs in, often on the same plants. This allows pollinator numbers to increase, but when the next generation of pollinators emerges in the summer there are very few receptive male figs available, but many receptive female figs. Seed set is ensured, but at the expense of an annual bottleneck in pollinator populations [24,32]. Not all individuals exhibit precisely the same phenology, and an extra generation of figs is developed in some plants [33]. In most parts of its range, F. erecta has a flowering phenology that is largely the same as that of F. carica [25]. This changes when the plant is grown under warmer conditions, outside its native range, where fig production on male trees becomes asynchronous [34]. The phenology of a second Asian species, F. pumila, is also similar to those of F. carica and F. erecta, but with one major difference. Like the other species, male trees produce two major crops each year, and female trees produce a single major crop. However, whereas F. carica and F. erecta build up pollinator numbers in the spring by fitting in a post-winter generation in male figs, this is not the case in F. pumila, where it is the generation of fig wasps that has overwintered as larvae that emerges at the same time as female figs are receptive, and so contributes to seed set. Based on the phylogenetic relationships of the plants (and also their pollinators), the three plants are not closely related species [35,36] and their atypical phenologies represent convergent responses to selection pressures generated by the seasonality of their environments.
F. tikoua has been assigned to Ficus Subgenus Ficus [27], but molecular evidence [36] and also its morphology (F. Kjellberg, Pers. Comm.) show that it should be placed in Ficus Subgenus Sycomorus, which contains both monoecious and dioecious species. It is pollinated by an undescribed species of Ceratosolen  [37]. In passively pollinated fig trees the insect makes no direct effort to collect or transport pollen, and pollination is dependent on pollen grains that were transported on the body of the insect. This form of pollen transfer is less efficient, and requires male plants to produce more pollen. This relative inefficiency is reflected in the much larger number of male flowers present in figs of passively pollinated fig trees [28].
Each year, only one of the generations of fig wasps that emerges from male figs of F. carica, F. erecta and F. pumila has a high probability of entering female rather than male figs and thereby contribute to the reproductive success of the male plants from which they emerged. Pollen carried by the pollinators released at other times of the year represents a metabolic cost for which there is no direct reward to the plants. Reflecting this, figs on male trees of F. carica and F. pumila that are produced at other times of the year do not contain functional male flowers, so the pollinators that emerge from them carry no pollen [24,38]. These species are passively pollinated, and their pollinators can clearly develop successfully in male figs that receive no pollen. It is unclear whether F. erecta is similar.
Unlike the other northern species, F. tikoua has an active pollinator, which may limit its ability to produce pollen-free male figs if pollen aids pollinator fecundity [39]. Here, we studied the fruit phenology of F. tikoua within its native range in China and address the following questions: (1)

Ethics Statement
Our sampling site was not in a national park or protected area. The studies species, Ficus tikoua, is not an endangered or protected species, so specific permission was not required. The specific location of the sampling site is 31.45˚N, 104.60˚E.

Ficus tikoua and its fig wasps
The natural distribution of F. tikoua Bureau covers Southwest and Central China and montane areas of Northeast India, Laos and North Vietnam, where it is found in wastelands, grassy banks, rocky areas and open woodland [40]. It is a prostrate shrub that does not reach a height of more than about 30 centimeters, with figs located at the leaf axils. The figs are often partially buried in the soil and for this reason it is called ''di-guo'' in Chinese, meaning 'fruit from soil'. The figs are small, flattened ovoid, reaching 10-20 mm in diameter at maturity. Both male and female figs remain yellow-brown when ripe, suggesting that terrestrial mammals may contribute to seed dispersal [13].
Genetic differentiation between adjacent F. tikoua populations suggests that its Ceratosolen sp. pollinator disperses less widely than the pollinators of most Ficus species [41]. This may be a consequence of the plant's small crops and the cryptic location of its figs, all of which reduce the 'apparency' of F. tikoua to pollinators, and makes long distance detection of suitable figs more difficult.

Locality and methods
Our study population was located on a small hill with sparse deciduous forest located in Mianyang, Sichuan Province, China (31.45˚N, 104.60˚E), which is towards the northern edge of the natural distribution of F. tikoua. The region has a subtropical monsoon climate with four distinct seasons. Summers are long, hot and humid, and winters are relatively short and mild, but with some snow. Mean minimum temperatures for the coldest month (January) are about 3˚C, and mean maximum temperatures for the hottest months (July and August) are about 30˚C (http://www.chinaweatherguide.com/sichuan/mianyang-weather.htm).
The creeping growth form of F. tikoua makes it difficult to distinguish between individuals and to identify which figs are produced by each individual. We therefore established sampling points with dense F. tikoua foliage that were at least 30 meters apart between each other that were assumed to represent 32 different plants. One meter square areas were marked at each point. The numbers and developmental stages of the figs in each square were generally recorded every seven to 10 days between 30 November 2012 and 2 March 2014, but recorded monthly during the winter periods. Fig developmental phases were assigned based on the scheme of Galil [42]. Seasonal differences in male and female flower numbers and fig wasp contents in male figs were tested using Generalized Linear Models (GLMs) assuming quasiPoisson distribution of residuals. Pair-wise comparisons were carried out using multiple tests with Bonferroni correction. All analyses were carried out using R [43].

Results
No sampling squares were recorded to produce both male and female figs. Among 32 sampling squares, 20 produced male and 8 produced female figs, so they were regarded as male and female plants respectively. The rest four sampling squares/ plants produced no figs within our observation period.
The phenology of F. tikoua   Table S1 in File S1). The receptive (phase B) periods of both crops of female figs corresponded closely with the periods when fig wasps were being released from male figs (phase D figs, Figures 2 & 3). The much larger early summer female crop also corresponded with the larger numbers of male figs releasing wasps at that time, and also will have benefitted from a virtual absence of competition for pollinators from B phase male figs (Figure 3). This contrasts with the smaller supplementary second crops of female figs, which was produced at a time when many of the pollinators emerging from male figs had the opportunity to enter receptive figs on the same plants.

The development, pollination and abortion of male and female figs
Relationships between fig size and weight were recorded for 201 male and 100 female figs from outside the demarcated areas. A power-function relationship was present between their fresh weights and diameters (R 2 50.91 and 0.99 for male and female figs respectively, Figure S2 in File S1). The female figs were a little heavier than male figs of similar diameters, especially after they had been pollinated (from the beginning of C-phase). However, no significant differences were found between the regression equations of each sex (likelihood ratio test, x 2 50.298, p50.585). The smallest female figs that had been entered by pollinators had a diameter of 6.09 mm, compared with 7.11 mm for male figs ( Figure S2 in File S1), but mature female figs (E phase) were considerably larger than male figs at the time that they released pollinators (D phase) (Female figs have no equivalent to D phase and pass directly from C to E phase).
Few figs could be recovered after they became detached from the plant, but repeated measurements of the same figs allowed us to decide which figs had aborted without being pollinated, based on their diameters on the last occasion when they were still attached to the plants. Figs with smaller diameters than the observed maximum diameter of un-pollinated figs were assumed to have aborted without being entered by fig wasps. Abortions were frequent among both male and female figs and 'replacement' figs often themselves had to be replaced after they also aborted (Table S1, S2). Overall, 62% of the initially-marked male figs (436/700) and 33% of the female figs (67/205) aborted before they matured (reached D phase if male figs, or E phase if female figs). Most of these abortions (including those from some replaced figs) took place before the figs reached the size when pollinators entered ( Figure 4) and an estimated 86% (589/683) of the  (Table S1 in File S1). Substantially different abortion rates were also found among female crops, with most spring crop figs reaching maturity (86%, 138/160), whereas no summer crop figs reached maturity (Table S2 in File S1). There were also large between-crop differences in the sizes of the figs when they aborted. Most spring and summer crop abortions among male figs occurred when the figs were small, indicating a shortage of pollinators, whereas the high abortion rates among autumn male crop figs resulted from a combination of early abortions among un-entered small figs and losses of larger, pollinated figs, through the winter period ( Figure 4). Female figs showed a similar pattern (Figure 4).   [45]. If they enter figs on other male plants then there is no benefit accruing from pollen production.

Contents of the figs
The evolution of a dioecious breeding system provided many potential advantages for fig trees relative to their ancestral monoecious breeding system, including avoidance of self-pollination, the potential to chemically defend ovules from non-pollinating fig wasps and the partial decoupling of flowering times of male and female individuals [46,47]. Tropical dioecious fig trees exhibit a wide range of phenologies, including seasonal concentrations of flowering that allow peaks in pollinator release from male plants to coincide with peaks in the numbers of female figs waiting to be pollinated [14,23,48,49]. The flowering phenologies of extra-tropical dioecious species can be seen as extensions of these to cope with a cold winter period.
The  influence rates of seed development and germination, and the ability of seedlings to establish. The timing of mature seed production in F. tikoua may simply be as early in the year as can be achieved given pollination constraints, but their fruiting phenology has a clear benefit in that it avoids the need for figs containing developing seeds to be retained on female plants through the winter period. F. tikoua is the only one of the four temperate dioecious fig trees to benefit from active pollination of its flowers. Inflorescence structure in figs from different crops of male F. erecta has not been compared, but in F. carica and F. pumila only one crop of male figs each year contains male flowers -the crop that is synchronized with the availability of female figs to pollinate. Pollen production is more costly for these passively-pollinated species than in the actively-pollinated F. tikoua, because they need to produce more pollen to achieve adequate fertilization. Passive pollinators haphazardly distribute pollen within the male and female figs they enter, but fertilization of female flowers in male figs is inhibited [50]. Because F. tikoua is actively pollinated, its male figs contain fewer male flowers than would be required for passive pollination [28], so the cost of retaining male flowers is lower and selection for their loss in figs produced at times of year when no female figs are available is likely to be less than in passively-pollinated species. Furthermore, if some second crop female figs do manage to survive to maturity, even at very low frequencies, then benefits would accrue to male plants that are releasing pollen-carrying pollinators in late summer.
Results  [50]. Experiments where pollinator foundresses that lack pollen are introduced into figs suggest that actively-pollinating species often suffer reduced reproductive success, due to increased larval mortalities, whereas passively pollinating species appear not to benefit from pollination [51,52]. The presence of male flowers in male figs of F. tikoua throughout the year may therefore also reflect selection on the trees acting via pollinator fecundity.
Reproduction by fig trees is often limited by the number of figs entered by fig wasps [53,54], but abortion rates in the autumn and spring crops of male and late summer crop of female F. tikoua seem particularly high. This resulted from a combination of a lack of pollinators and over-wintering losses among autumncrop figs and a shortage of pollinators that had survived the winter and became available to pollinate the spring male crop. In contrast, abortion rates among the main late-spring crops of female figs were much lower, and emphasize that the plant's phenology delivers effective pollination and seed production despite the seasonal lows in pollinator populations that it generates.
F. tikoua is a short creeping plant with rather small crops of small, inconspicuous figs. As such, the plant and its figs have a low 'apparency' to insects and they are likely to be hard to find from long distances [55,56]. Perhaps reflecting this, its pollinators rarely disperse far [41]. Our study was carried out in an area with a large, dense population of F. tikoua that was clearly able to maintain a resident population of pollinators. Small founder populations may not be able to do so, and vegetative reproduction may prove to be a significant component of the plant's overall reproductive strategy.

Supporting Information
File S1. Figure S1. Changes in densities of phase AB figs over time in marked areas on six male Ficus tikoua individuals in Mianyang that produced three crops during the sampling period. Figure S2. The relationship between fresh weights and diameters of male and female figs of Ficus tikoua in Mianyang. Figure S3. The numbers of flowers in male Ficus tikoua figs from Mianyang. Table S1. Abortion rates of figs from one metre square sections of twenty male Ficus tikoua individuals in Mianyang.