Seasonality of Leaf and Fig Production in Ficus squamosa, a Fig Tree with Seeds Dispersed by Water

The phenology of plants reflects selection generated by seasonal climatic factors and interactions with other plants and animals, within constraints imposed by their phylogenetic history. Fig trees (Ficus) need to produce figs year-round to support their short-lived fig wasp pollinators, but this requirement is partially de-coupled in dioecious species, where female trees only develop seeds, not pollinator offspring. This allows female trees to concentrate seed production at more favorable times of the year. Ficus squamosa is a riparian species whose dispersal is mainly by water, rather than animals. Seeds can float and travel in long distances. We recorded the leaf and reproductive phenology of 174 individuals for three years in Chiang Mai, Northern Thailand. New leaves were produced throughout the year. Fig production occurred year-round, but with large seasonal variations that correlated with temperature and rainfall. Female and male trees initiated maximal fig crops at different times, with production in female trees confined mainly to the rainy season and male figs concentrating fig production in the preceding months, but also often bearing figs continually. Ficus squamosa concentrates seed production by female plants at times when water levels are high, favouring dispersal by water, and asynchronous flowering within male trees allow fig wasps to cycle there, providing them with potential benefits by maintaining pollinators for times when female figs become available to pollinate.


Introduction
The phenology of plants reflects selection generated by seasonal climatic factors, interactions with other plants and animals and constraints imposed by their phylogenetic history [1,2]. Phenology is the study of the timing of life-history events associated with the course of the seasons and the factors that lead to its occurrence [3]. It influences evolution and adaptation at individual and community scales [4]. Studies of plant phenology have examined the timing of occurrence of biological events such as bud-burst, leaf-expansion, abscission, flowering, fertilization, seed set, fruiting, seed dispersal and germination [5,6]. In tropical regions, climatic variations are less pronounced than elsewhere, leading to different plant responses in relation to climate. Seasons are often marked by differences in rainfall, with life-history events occurring in response to water availability [7]. The timing of biological often responds fluctuations of climate [8][9][10]. In general, the more specialized the relationship between species, the more vulnerable each of them is likely to be to the phenological effects of future climate change [4]. There is currently widespread concern that climate-driven changes in the timing of seasonal events may disrupt important ecological interactions such as pollination or cause temporal mismatches between critical periods in animal life cycles and food availability. More specifically, an extreme specialist such as fig trees of the genus of Ficus (Moraceae) have complex reproductive phenologies and exhibit one of the obligate species-specific mutualisms, with pollination performed by fig wasps (Agaonidae) [11][12][13][14][15][16][17][18][19][20]. Both the plants and their associated insects require the other for survival, to accomplish their life cycles. Together they form an obligate and specific mutualistic relationship and changes in the abundance of one of the partners can drastically reduce their fitness of the other [21]. The specific fig-fig wasp mutualism often follows locally a one-to-one rule (one Agaonidae species for one Ficus species) [22]. However, recent studies have shown that Ficus species often have more than one pollinator species in different geographical, and sometimes in sympathy [23][24][25][26][27][28]. Nevertheless, the fig and fig wasp association remains highly specific [29].
Seasonality in fig production has often been noted in phenological studies that report asynchronous flowering with continual fig production at the population-level, combined with flowering synchrony and sub-annual fruiting at the individual level [22,30]. In general, the seasonal timing of flowering and fruiting, i.e. the reproductive phenology of plants, is determines the resources available to animals and consequently has effects on pollination and seed dispersal [31]. Flowering patterns may be related to the abundance of pollinators or the optimal time for pollination, and the ripening of fruits tends to peak during the best time for dispersal and germination [32]. Genus Ficus, with around 800 described species is divided reasonably evenly between monoecious and functionally dioecious species [33]. The breeding system of a fig species influences its reproductive phenology [34]. In monoecious Ficus species where each plant has figs that support seed production and the development of pollinator larvae each individual tree typically produces a few highly synchronous crops of figs each year, but asynchrony among trees allows year-round survival of the pollinating fig wasps populations [16].
In dioecious Ficus species, several crops of figs a year can be produced, but crops are often less synchronized than among monoecious species [17,35,36]. Relatively few studies have described the reproductive phenology of dioecious fig species in tropical region [34][35][36][37][38], but considerable variation in phenology is recorded [4,[35][36][37] with patterns often reflecting seasonal variation in climate [33]. Under subtropical to temperate conditions, male trees seem to produce a major annual crop of figs that releases pollinating wasps at the time when a major annual crop of figs on female trees is receptive, together with minor male crops that allow yearround survival of pollinating wasps [4]. For example, with Ficus carica L. in southern France, pollinators disperse between figs only twice yearly, in May and August [39]. Dispersal of the pollinator, Blastophaga psenes L. during the winter is thus avoided. In general, males of dioecious Ficus species produce figs all year round, more earlier [40]. Dioecy may be seen as a trait enabling male trees to support pollinators populations while allowing female figs to concentrate at times of optimal conditions, and at the same time avoiding self-fertilization [41].
In general, female fruit ripening peaks are expected to occur prior to favorable conditions for the germination of seeds and the development of seedlings [31]. Fruit development and seed maturation should be timed to match with the seasonal availability of legitimate dispersal agents and suitable environmental conditions for dispersal and plant establishment [42]. For these reasons, phenological synchronization among individuals of a population is directly related to the reproductive success of the species [43].
Thies and Kalko [44] have suggested that differences in the flowering phenology of tropical forests are primarily caused by abiotic climatic factors such as water and light, whereas differences in fruiting phenology are mostly influenced by biotic factors (the presence of seed dispersers). The flowering and fruiting phenologies of several dioecious Ficus species are related with seasonal conditions, such as: Ficus auriculata, F. fulva, F. hispida, F. oligodon, F. semicordata, F. triloba and F. variegata, in Thailand [45], F. fulva in Malaysia [38] and F. variegata in Australia [34].
Berg et al. [46] noted that F. squamosa has unusually long persistent styles with retrorse hairs which they regarded as adaptations to anchor the seeds to the substrate, allowing seeds to adhere to substrates on the soil or rock and preventing their washing-away into unfavorable sites. How this species disperse is unknown, but with regard to the seed morphological evidences and seed dispersal of rheophytes, water is expected to play a major role. Our others hypothesize that seeds of F. squamosa can float when it is ripening and falling into the water at times of high water level or floods. In several rheophytes, buoyant seeds have been observed, and some seeds have special adaptations to hydrochory (i.e. plant dispersal by surface water) [42]. Such adaptations can be: cork-like, air-filled tissue, resulting in a low relative density or a hydrophobic seed coat and fiber or hair covered seeds [42]. For example, seeds have outer layer with hooked fibers or hairs and may aid in dispersal by allowing the fruits to functions as trap air bubbles that float to the water flow and attached to seeds may be as anchors on the substrate (i.e. Pinanga rivularis, a rheophytic palm of Borneo [47] and Euterpe endocarps, a hydrophilic palm of Brazil [48]).
The natural hydrological conditions in rivers and streams are a prerequisite for facilitating the spatial movement of seeds especially during high floods. For example, Boedeltje et al. [49] suggested that water is a vector for transport of several type of seed plants and the majority of seeds was dispersed before late autumn and/or winter that floods may have transported seeds from streams into the canal at the Twentekanaal in the eastern part of the Netherlands. Besides Schneider and Sharitz [50] reported that the number of species and seeds dispersed were positively related to water levels in a swamp forest. The information about temporal dispersal pattern of F. squamosa in relation to discharge patterns carrying seeds is still poorly understood. It is, however, likely that surface water is an important vector for seed dispersal but it is not the only one dispersal vector in freshwater wetlands [42]. Although, there is no detailed information is available of the other dispersal vector of F. squamosa seeds, but wind and animal dispersal have an additional value to riverine plant seeds, besides hydrochory, whereas water dispersal in streams is unidirectional and takes place in downstream direction only, wind and animal may enable inter-catchment dispersal and upstream dispersal [42].
The seasonality of phenological events in riparian areas seems to be less pronounced. Whereas in some riparian forests the phenological events of the woody species are uniform along the year [7], in others they seem to be also associated with rainfall, with peaks in the wet season [51]. The present study aims to describe the reproductive phenology of riparian fig species and correlate them with climatic variables (i.e. air temperature, relative humidity, sunshine hours and rainfall).
We examined whether the riparian habitat of F. squamosa, and in particular whether its use of moving water to disperse its seeds, is reflected in its phenology of leaf and flower/fruit production. We addressed the following specific questions: (1) Do leaf and fig production vary seasonally in relation to climatic variables such as rainfall? (2) Does the fruiting phenology of F. squamosa reflect seasonal variability in water levels in its riparian habitat? and (3) Does the fruiting phenology of male and female plants differ? The results can help inform future management of water flows in SE Asian rivers, with implications for the conservation of riparian Ficus diversity [42].  [52], but their phenological responses to seasonal variability are poorly understood. Ficus squamosa is a functionally dioecious species distributed along watercourses in tropical forests. It is strictly riparian species and rheophytic, restricted to streamsides, and is adapted to living and establishment in (rapidly) flowing water [46,52]. It grows in riparian areas that are occasionally flooded, but may be dry for periods of the year [53]. In Northern Thailand, the water levels of streams in the rainy season can reach about 1-1.5 meters above these in the dry season (P. Pothasin, personal obs.). Plants such as F. squamosa are exposed to the highest water levels where become mostly submerged, and can even be destroyed by extreme floods. Normally, F. squamosa will hardly ever be submerged for such a long period of time (short or long heavy tropical rainstorms in the upper reaches of a stream may cause the water level to rise for several hours to a few days or during a longer rainy spell for a few weeks, but not usually for several months. When the flash floods pass and water levels go down again, the plants are re-exposed. Conversely, in the dry season, they are often growing in seemingly dry habitats such as cracks of rock, sandbars or sandy shores (P. Pothasin, personal obs.).

Ficus squamosa and its fig wasps
Ficus squamosa Roxb. (subgenus Sycomorus, section Sycocarpus) is a dioecious shrub with rooting stolon-like branchlets that grows approximately 1 to 2 m tall. It was placed in subsection Macrostyla by Berg et al. [54], together with a species from Sarawak, because they share flowers in their female figs that have very long persistent styles with deflexed hairs. The species are rheophytes and their unusual styles may be due to convergent adaptations to anchor waterborne fruits to river substrates [46]. The natural distribution of F. squamosa covers Thailand through to Laos, Myanmar, China (Yunnan), NE india, Sikkim, Bhutan and Nepal [55], where it is mostly found in the rocky beds of quick running streams, and no more than 10 meters the water's edge [52]. Its figs are located along the branches and branchlets (Fig 1). The figs are rhombic-ovoid, reaching 3-4 cm in diameter at maturity. Both male and female figs remain yellow-green when ripe, reflecting their use of water currents (and possibly terrestrial mammals) for seed dispersal (P. Pothasin, unpublished data).

Study sites
Our study populations were located along perennial streams in the Ping River basin, Chiang Mai Province, Thailand (18°51´N, 98°54´E). The region is mountainous with some lowlands along river tributaries, with about 60 percent of the land area having an elevation above 500 m. Average annual rainfall, relative humidity, and annual temperature are 1073 mm, 70% percent, and 25.9°C, respectively [56]. The climate of Chiang Mai is influenced by tropical monsoons and falls into three main seasons, the rainy, the cool dry and the hot dry seasons (S1 and S2 Figs). Rainfall is frequent from June to late October. This is followed by the cool dry season until February, and the hot dry season running from March to June. Seasonality flooding is a regular feature of the monsoon climate. Heavy downpours are common in the rainy season or in September, and can cause flash floods. Floods are common in north Thailand because this region experiences the highest seasonality in rainfall distribution within Thailand and flash floods were reported for some area in Northern Thailand when monsoon rainfall completed with heavy rainfall [57].
Our four study sites (Fig 1) were in fragmented riverside forest [58][59][60]. The streams were fast flowing and bedrock-lined, with large boulders, (S1 Table). The streams flow all year and generally carry only small amounts of silt and sand. Runoff events and rapid temporary increases in flow rates occur in the rainy season, and result in water colour changing to reddish-brown. Two sites, Mae Ka (MK) and Pang Dang Nai (PDN) were located in the northern, Chiang Dao District of the Upper-Ping watershed and had mixed limestone and igneous substrates. The MK stream flows through Doi Chiang Dao's wildlife Research Station, where some water emanates from a small natural hot spring. MK flows through hill evergreen forest, deciduous dipterocarp forest and agricultural land. The site, at an altitude of 480 m above sea level. The width of the stream was between about 3.5 and 12 m. The other two streams had mainly igneous substrates and widths ranging from approximately 5-20 m. The PDN stream flows through Ban Pang Dang Nai village and through maize fields and mixed deciduous forest with bamboos and Teak (Tectona grandis L.f.). This site, at an altitude of 500 m above sea level and a concrete check dam is present. Streams Mae Sa and Mae Sa Noi (MS and MSN) were located in the Mae Rim District and had several waterfalls. The sites, at an altitude of 350 and 570 m above sea level, respectively. Over the last decade much of their surrounding area has become peri-urbanized, with various tourist attractions and extensive stream-side encroachment.

Meteorological data
Simultaneous meteorological data, including daily average temperature (in degrees Centigrade) and relative humidity (as a percentage), and sunshine (in hours of daylight) were obtained from the documentation of the Northern Meteorological Center, Thai Meteorological Department. Monthly rainfall (in millimeter) were obtained from the Hydrology and Water Management Centre for Upper Northern Region, Royal Irrigation Department, Thailand database. The weather data were measured at a fixed-site station located close to each site. The diameter and number of stems on each plant was recorded. For plants with more than one stem, a combined stem diameter was calculated from the sum of the basal areas of each stem. The height of each tree, its maximum crown width and the number of stems were also recorded. Leaf condition was summarized according to the following index: 0 (no leaves), 1 (leaf buds), 2 (young leaves), 3 (mixed mature and young leaves), 4 (mixed mature and senescent leaves) and 5 (senescent leaves).
On each occasion, we examined individuals for the presence of figs and recorded the numbers of figs present. Fig diameters were measured at their equator to the nearest 0.1 mm with Vernier calipers. We recorded the phase of development for the most abundant age class of figs on the tree, but also recorded any other phases present. Crop sizes were often small. We recorded crop sizes as the maximum number of figs present on a particular date and calculated only for crops, which finished development.

Data analysis
Most statistical analyses were carried out in R (Version 3.2.0) [61]. As most data was not normally distributed and there was considerable variation in sample sizes, the non-parametric Mann-Whitney U-test was applied [62]. Similarity, for the analysis of distributions of the variance in the distributions of development phases among different groups, Kruskall-Wallis (K-W) non-parametric one way variance tests were used. Separate analyses were carried out for male and female trees growing at each site. We compared the size of plant (stem diameter, height and crown width) between male and female trees at each site. We also compared measures of fig abundance (fig production) and asynchrony using the pooled data. The presence or absence of asynchronous fig crops within individual trees was recorded with for example, an asynchronous crop in two or more development stage having figs within the same tree. Spearman rank correlations were performed between phenological and meteorological factors. Abiotic factors including sunshine hours, average mean temperature, average relative humidity and average rainfall were averaged over the 30 preceding days.
The reproductive phenological patterns of the plant and all circular statistical analyses were examined using ORIANA V.4 [63,64]. We converted the dates of the phenological events to angles, from 0°(January) to 330°(December) at intervals of 30°per month [63,65]. We estimated the mean angle (a), mean date (time converted from mean angle), vector length (r, concentration around mean angle) and circular standard deviation for each study site. Mean angle (a), and mean date, indicate average dates of the fruiting peak in the populations. Vector r has no units and indicates the degree of aggregation (seasonality) among individual plants fruit traps or synchrony of reproductive activity. High r-values generally indicate aggregated phenological behavior. We also performed a single-sample-distribution Rayleigh's test (z) to determine whether the circular distribution of phenological activity was significantly non-random, or if there was seasonality. The Rayleigh test assessed the occurrence (or absence) of seasonal patterns within the four study sites. Mardia-Watson-Wheeler Multisampling Tests compared whether the variables had the same distributions throughout the year.
Both male and female F. squamosa were evergreen at our study sites, with small-scale leaf loss and gains of new leaves occurring almost throughout the year. Newly opened leaves took about one month to mature. Based on synchronized flashed of leaves, plants that produced well. The average longevity of the leaves varied between 6-8 months on different plants. The fewest leaves were present on the plants in the dry season (January to April) and late September, with peak numbers of new leaves in May, after rains had ended the dry season (Fig 3).
New leaf production on a few plant was a little bimodal, about 6 months apart, but most plants revealed a single seasonal peak in new leaf initiation (Fig 4). Rayleigh's Z values for seasonality of leaf initiation were mostly highly significant (Rayleigh tests, P < 0.001, S2 Table). Z values were consistently lower among males than female plants, and at two sites compared with the other sites. There was no significant seasonality in the occurrence of young leaves among male plants. In female trees, leaf initiation was concentrated in late May at MK and PDN, in the middle of June at MS and at the end of September at MS. In male trees, leaf initiation peaks occurred from May to July at the different sites (Fig 4). New leaf production of both sexes was significant positively correlated with higher temperatures and negatively correlated with relative humidity ( Table 1). The proportion of trees with young leaves present was mostly significantly correlated with the proportion of fig-bearing trees and new leaves were sometimes significant positively correlated with numbers of figs initiated (Table 1).

Reproductive phenology
The plants were assessed at approximately two-weekly intervals from April 2009 to March 2012. Eighteen plants that produced no figs were excluded from further analysis. Eleven of the initial 174 plants were lost through flash floods and associated erosion of the stream banks during the course of observation. Mature figs were retained on female plants until they were carried away by rising water levels during the rainy season and did not become noticeably softer or changes colour before being carried away. Fruiting patterns were broadly similar at the four sites. A total of 442 crops were recorded from 120 female plants and 389 crops from 54 males. Individual female trees produced up to 6 completed crops and averaged 3.68 ± 1.12 crops (mean ± SD) crops during the period of the 36-month (n = 120 trees), equivalent to 1.43 crops per year. Male figs produced 3-9 completed crops with an average of 7.20 ± 1.38 (mean ± SD) crops (n = 54 trees), equivalent to 2.82 crops per year. There were no significant differences between sites in the number of crops produced by each sex over the study period (Kruskal Wallis test, female trees: x 2 = 6.330, P = 0.097, n = 120; male trees: x 2 = 8.867, P = 0.031, n = 54).
The average crop sizes were much higher on male trees at the four sites (female trees: mean ±SD = 32.27 ±18.45, n = 442 crops; male trees: mean ±SD = 37.83 ±18.45, n = 389 crops). However, there were no significant differences in the average crop sizes between sexs (Mann-Whitney U-test, U = 0.918, P = 0.359). Further analysis of site by site comparisons of reproductive phenology showed that the average crop sizes were significant differences between sites in female trees, but not in male trees (Kruskal Wallis test, female trees: x 2 = 12.87, P = 0.005, n = 120; male trees: x 2 = 2.03, P = 0.566, n = 54.) Tree diameter (d.b.h.), height of tree and crown diameter were all not significantly correlated with the number of crops produced over the study period by each sex (Spearman rank correlation tests; Table 2). Crop size of male and female trees were both significantly correlated with d.b.h., but there were not significantly correlated with height and crown diameter except in MK site, where crop size was significantly correlate with height of tree (Spearman rank correlation tests; Table 2).
F. squamosa individuals initiated fig crops at different times throughout the year, with obvious seasons when few or many figs were present. Sexual differentiation in phenological patterns was apparent. More crops of male trees were asyschronous than female trees (define what asynchronous was = B and D+E stages) (Mann-Whitney U-test, P < 0.001; Table 3). F. squamosa producing figs all year-round, more or less continuously on male trees than female trees (As an example at MK site, female; S3 Fig, Male; S4 Fig). Some crops on both male and female trees were aborted without having been entered by pollinators. Crop abortion was more common on female than male trees (e.g. female tree; numbers 9, 10, 13-15 and 35). A few female trees (numbers 1, 2, 3, 13, 16 and 37) died because of the flooding after the rainy season in  Table,

Correlation with meteorological factors
The numbers of A-phase (recently initiated) figs and the total numbers of figs on the plants were correlated with meteorological data. Generally, A-phase figs and total figs were positively correlated with mean temperatures over the previous 30 days, but figs on male trees at MK had a significant negative correlation with temperatures (Table 4) (Table 4).

Discussion
Studies conducted in riparian forests point out a phenological pattern of F. squamosa which is one of the dominant fig species in stream habitat in the northern part of Thailand [52]. Ficus squamosa is a shrub with creeping stem found in rocky beds of fast-flowing streams that are exposed to two alternate environments. During summer period the water level is low, the species are exposed to dry air and high temperatures. On the other hand, during rainy season, water level rises and flash floods occur frequently, so these plants are exposed to severe flooding periods, with bank erosion and burial sediment making it inaccessible. The species that occupied these habitats developed unique characteristics as well, due to the fact that they must stand the flood period. Nomura et al. [66] noted in general that the narrow shape of the leaves or rheophytes is the adaptation not for the light intensity but for the flood resistance, through an increased mechanical toughness. F. squamosa has figs that originate on branches near to the water level or even below. Their fruits have brushy long, hairy stigmas, which persist when the figs are ripe. Seeds can also take advantage of water currents to float their offspring away. This is an adaptation for growing in running streams near fast-flowing rivers [54].
Ficus was unique in that the association between Ficus (Moraceae) and their pollinators, the  fig wasps (Agaonidae, Hymenoptera), involves a species-specific and unique pollination system  [11,67]. Phenology of flowering production at the population level must ensure survival of the pollinators if their obligate mutualistic relationship is to be maintained. This requirement may bring about the typical phenological pattern of Ficus found in tropical regions, which is annual or supra-annual flowering at the individual level integrated into a continuous pattern at the population level [33].

Leaf phenology
The production of leaves on F. squamosa trees in northern Thailand varied among individuals, but there was a clear response to the individuals of F. squamosa showed evergreen behavior. F. squamosa never appears leafless during the entire year. Although some new leaves were produced in all months of the year, they were more abundant in the beginning wet period (May), greening the canopy well in advance of the onset of the monsoon. Leaf exchange occurred in small amounts throughout the year, leaf senescence and leaf fall was maximal in the dry season (January to April) and no differences between the sexes. F. squamosa in southwestern China by Liu et al. [68] exhibited similar leaf phenology in our study. There was little seasonal variation in the leaf states, with small quantities of new and senescing leaves present more or less continuously (see citations in figure 2 of Liu et al. [68]). Similar observation has also been reported in other species by kuaraksa et al., reported that F. semicordata and F. hispida were evergreen in dry deciduous forest to hill evergreen forest in northern of Thailand [45]. The pattern of leaf phenology was strongly influenced by drought. Drought caused a significant increase in leaf shedding which was then followed by new leaf production with the renewal of rain. Harrison et al. reported a sudden increase in the deciduousness of fig species that was associated with severe drought linked to the El Niño event of 1997-1998, thus suggesting that the pattern of leaf phenology in figs was strongly influenced by drought and major rainfall oscillations may substantially affect phenology [38]. In seasonal environments, deciduous species may or may not initiate growth before evergreen species, but they complete it before the evergreens. They shed all their leaves prior to dry winter, which may allow them to conserve and store water and survive during cold dry winters [69,70] (i.e. F. variegata [34,45], F. obtusifolia [71], F. racemosa [72], F. fulva [38,45], F. auriculata, F. oligodon and F. triloba [45]). In addition to the degree of drought to which trees are exposed varies widely, depending on temperature and availability of soil water, and tree characteristics such as rooting depth [31]. Accordance with the reports from other monsoon forests in Thailand, deciduous species form Spearman's rho (rank correlation tests) between numbers of A-phase figs (A) or total numbers of figs (Tot.) and different abiotic factors averaged over the 10, 20, 30, 60 preceding days, and with a 10-day delay were calculated. "+" indicates that some correlations over different durations were positive and significance (P <0.05). "-" indicates they were some negative and significance (P <0.05) and "NS" indicates they were all non-significance. We never obtained a mix of significant positive and negative correlations. The full results are available in the supplementary data (S4 Table). new leaves one to two months before the first monsoon rains in April/May, during the hottest and driest part of the year around the spring equinox. While evergreen species are restricted to relatives moist and there were almost no discernable seasonality of vegetative development [73]. Moreover, most of evergreen species shows deep root system, enabling access to water [74]. As rheophytes, F. squamosa shows stolon-root like stem system, so they were maintained a high water potential during the dry season. We found that the proportion of trees with young leaf presents was mostly significantly positively correlated with temperature in both sexes, but neither between young leaf present and rainfall in both sexes. In tropical trees initiation of leaf flush (leaf bud), marking the termination of deciduousness duration, has been reported to be triggered by several factors such as increasing day length and/or temperature, significant amounts of the first rain, and photoperiod [74,75]. According to Borchert [76], the increased photoperiod with rising temperature may cause conversion of starch into sugar in the roots and stem and osmotic adjustment in bud tissues which may induced bud busting by increasing water absorption and availability of sugars in summer flushing trees. However, there was a slight negative correlation with leaf flushing and relative humidity. In F. squamosa, drought caused a significant decrease in leaf flushing which proportion of trees without young leaves visibly matched with dry periods.
Correlations between vegetative and reproductive phenologies occurred in some figs, such as F. fulva in lowland dipterocarp forest at Sarawak, leaf flushing correlate with syconia initiation almost certainly reflects a physiologically based constraint on species, which produce their syconia at the tips of the twigs [38] and F. variegata in a seasonal wet tropical forest at caper tribulation, Australia [34]. In  Most of the study trees did not produce figs (10% of all trees observed have never produced figs), although all individuals were apparently mature (mean stem diameter >3 cm). The population-level phenology of F. squamosa is similar to the phenological pattern for dioecious figs proposed by kjellberg and Maurice [41], patterns that were also seen at the population level [36] and parallels the patterns of F. carica (subg. Ficus, sect. Ficus) in southern France [4], F. variegata (subg. Sycomorus, sect. Sycomorus) in a seasonal wet tropical forest at Cape tribulation, Australia [34], F. hispida (subg. Sycomorus, sect. Sycocarpus) and F.semicordata (subg. Sycomorus, sect. Hemicardia) in Doi Suthep-Pui National Park, Northern Thailand [45]. Kjellberg and Maurice proposed that dioecious figs respond differently to seasonal variability than monoecious figs, due to the separation of seed and wasp production between trees [41]. In dioecious Ficus, each gender specializes in making ether seeds or wasps, but wasp production translates directly into male reproductive function since wasps are the pollen vectors. The female figs would be produced during the wet season, presumably under the most favorable conditions for seed production, germination and establishment. The crop development period was approximately twice as long on female trees. On the other hand, male trees maintain a local wasp population by producing figs during adverse conditions. This pattern benefits male individuals (not simply the population as a whole) by increasing their reproductive success through the donation of pollen to females [16] -fig wasp mutualism [19]. However, our results pointed out seasonality in fig abundance that was significantly correlated with climatic seasonality (S3 Table). It is important to stress that riparian areas in northern of Thailand have a typical moonsoon climate and presents temperature and rainfall seasonality [73]. There are occasionally flooded during rainy season and exposed to dry air and very high temperatures when the water is low in hot season [77]. Male trees have shown sub-annual flowering patterns and produced larger productions of figs during hot season (April-May) in all sites. Female trees have an annual flowering pattern during late hot season to rainy season (May-July). We suggest that continuous production of figs is an adaptation to the stream habitat conditions. In addition, we observed fig abundance on female trees was significantly positively correlated with rainfall, but for male trees fig abundance showed a significant negative relationship with rainfall.
F. squamosa demonstrated considerable sexual specialization. At the individual-level, male F. squamosa trees produce a greater proportion of asynchronous crops than females. We observed the initiation of fig production showed seasonality in female trees, but mostly male trees appeared to be random or aseasonal in the initiation of fig production (S3 Table). Nonseasonality could be an adaptive feature, as Kjellberg and Maurice [41] have demonstrated with simulations that the number of plants needed to support the pollinator increase with seasonality reproduction.
We observed a peak in male fig production preceding the wet season (April-May) can increase the pollinator population, releasing pollen-carrying wasps at a time when the majority of female figs are receptive to pollinators. We may compare these results with results obtained on F.hispida [45], F. fistulosa [78], and F. septica [79], three species that belong to the same species group as F. squamosa within section Sycocarpus [54]. While fig production is rather continuous throughout the year, the curve of increasing male fig production preceded the curve of increasing female fig production, suggesting adjustment to pollinate female trees.
In addition to the tendency for males to produce more crops per year than females, which produces three to nine crops of wasps and one to six crops of seed figs per year. F. squamosa have also tended to demonstrate a longer development time of syconia on female trees than on male trees. Patel and Mckey [36] suggested that the longer development times for female figs were because of the greater number of ovaries and fleshier figs on female trees requiring greater investment of resources. This pattern differs from that of dioecious F. auriculata (subg. Sycomorus, sect. Sycomorus), F. fulva (subg. Ficus, sect. Eriosycea) and F. triloba (subg. Ficus, sect. Eriosycea) in Northern Thailand [45]. In fact, tropical phenological patterns can vary widely over the geographic range of species [32]. Even the phenological pattern of F. squamosa is similar to F. variegata in Australia [34], but differs from that F. variegata in Northern Thailand [45]. It would therefore be instructive to compare the phenological patterns of dioecious in seasonal wet tropical forest and the dry monsoon forest parts of its range. Kuasaksa [45] reported that F. variegata in mixed forest (evergreen and deciduous) in Thailand was synchronous within-trees of both sexes. However, there was displayed asynchronous fig production at the population level in both areas [34,45].

Conclusion
In Northern Thailand, F. squamosa is typically a species of riparian habitats, where is exposed to two alternate environments; i.e. the dry air and very high temperatures during dry period when the water level is low and the flooding period during rainy season. The present study of the phenology of F. squamosa demonstrated considerable sexual specialization and adaptive to riparian habitat. At the population level, male F. squamosa tree bears figs continually to maintain the pollinator fig-wasps population, while female tree produces one to several fig crops during the optimal season for seed production, dispersal and establishment (rainy season). The timing of dispersal of F. squamosa is shaped by the arrival of the rainy season and water vectors seem to be the main factor of seed dispersal their local habitats.