The pollination system of the widely distributed mammal‐pollinated Mucuna macrocarpa (Fabaceae) in the tropics

Abstract Although the pollinators of some plant species differ across regions, only a few mammal‐pollinated plant species have regional pollinator differences in Asia. Mucuna macrocarpa (Fabaceae) is pollinated by squirrels, flying foxes, and macaques in subtropical and temperate islands. In this study, the pollination system of M. macrocarpa was identified in tropical Asia, where the genus originally diversified. This species requires “explosive opening” of the flower, where the wing petals must be pressed down and the banner petal pushed upward to fully expose the stamens and pistil. A bagging experiment showed that fruits did not develop in inflorescences (n = 66) with unopened flowers, whereas fruits developed in 68.7% of inflorescences (n = 131) with opened flowers. This indicated that the explosive opening is needed for the species to reproduce. Four potential pollinator mammals were identified by a video camera‐trap survey, and 78.3% and 60.1% of monitored inflorescences (n = 138) were opened by gray‐bellied squirrels (Callosciurus caniceps) and Finlayson's squirrels (C. finlaysonii), respectively, even though more than 10 mammal species visited flowers. Nectar was surrounded by the calyx, and the volume and sugar concentration of secreted nectar did not change during the day. This nectar secretion pattern is similar to those reported by previous studies in other regions. These results showed that the main pollinators of M. macrocarpa in the tropics are squirrels. However, the species' nectar secretion pattern is not specifically adapted to this particular pollinator. Pollinators of M. macrocarpa differ throughout the distribution range based on the fauna present, but there might not have been no distinctive changes in the attractive traits that accompanied these changes in pollinators.


| INTRODUC TI ON
Plants pollinated by specific pollinators attract and limit them by specific floral traits, such as flower shape, color, and odor (Córdoba & Cocucci, 2011;Gómez et al., 2008;Hirota et al., 2012;Johnson, Burgoyne, Harder, & Dötterl, 2011). However, some widely distributed plant species with pollinator limitation seldom have the same pollinator species throughout their distribution range, because the fauna differ across the plant species' range (Boberg et al., 2014;Inoue & Amano, 1986;Johnson & Steiner, 1997;Sun, Gross, & Schiestl, 2014). In other words, plants that can be pollinated by various pollinators can increase their distribution range.
When pollinators of plants with pollinator limitation differ regionally, plants may accept alternative pollinators within the same taxon. For example, effective pollinators comprise several moth species for the orchid Platanthera bifolia (Boberg et al., 2014) and various bee species for Campanula punctata (Campanulaceae) (Nagano et al., 2014). In these examples, the pollinator species differed among the regions, but the activity time and basic shape of the pollinators were similar. On the other hand, there are some examples where the taxon and activity time of pollinators differ among regions.
The pollinators of Carnegiea gigantea (Cactaceae) have shifted from nectar bats to birds (Fleming, Sahley, Nolland, Nason, & Hamrick, 2001). Behavior of bats differs from birds. While bats can freely use their forelimbs for feeding, birds cannot use their anatomically equivalent wings in the same manner. These shifts in pollinators are examples of regional differences in pollinators of the same plant species. Moreover, flower shape and flowering timing might also differ among regions. There are also examples of changes in the attractive traits of nectar and volatile components associated with pollinator differences (Breitkopf et al., 2013;Perret, Chautems, Spichiger, Peixoto, & Savolainen, 2001;Wester, Johnson, & Pauw, 2019). Such differences in pollination systems among different region are important to understand the speciation process. To understand the speciation process caused by shifts in pollinators, study sites should be representative of the distribution areas, although it is difficult to identify the pollination system of widely distributed plants.
Once a flower explosively opens, the stamens and pistil are never covered by the carina petals. In at least two sites in Japan, this species needs explosive opening to bear fruit, because unopened flowers do not bear fruit, as experimentally in both bagged and unbagged treatments (Kobayashi, 2017). Thus, a flower-opening animal (the "explosive opener") is necessary to the reproduction of the plant species, making explosive openers effective pollinators.
Although there are different openers, in each region, flower shape and nectar characteristics are not clearly different among Kyushu, Okinawa, and Taiwan (Kobayashi, Denda, Liao, Lin, Wu, et al., 2018).
However, these previous studies have been conducted on subtropical and temperate islands. Because the number of mammalian species on the islands was small in almost all cases in general (Brown, 1978;Fox & Fox, 2000), the main pollinator taxon did not occur on the other islands. However, all pollinator taxa (fruit bats, macaques, and squirrels) occur sympatrically in continental tropical Southeast Asia (Duckworth, Salter, & Khounboline, 1999;Lekagul & McNeely, 1988). The aim of this study was to identify the pollination system of M. macrocarpa in continental tropical Southeast Asia and to compare this system among previous study sites. Accordingly, we tested the following hypotheses: were observed in the present study, the flower-visiting pattern should be different among species, because Yumoto, Momose, and Nagamasu (2000) showed that the pollinators of four squirrel species visited flowers at different times. Thus, we also aimed to determine the flower-visiting pattern of each species. and Hydnocarpus ilicifolius (Achariaceae) occur frequently (Lamotte, Gajaseni, & Malaisse, 1998).

| Observation of flower visitors
Video camera traps (Ltl-5210A, Ltl-5210A940, and Ltl-6210MC; Shenzhen Ltl Acorn Electronics Co., Ltd.) were used to monitor flower visitors. In total, 35 cameras were set up to monitor 138 inflorescences (2,198 flowers) in six plants throughout the day at various heights. Monitored plants grew within 2 ha, and three of them grew close together, while the others were at least 30 m apart. The height of inflorescences was measured to 1-cm accuracy by a laser distance meter (Leica DISTO™ X310; Leica). When we set up the cameras, 42 flowers from 14 inflorescences had been opened by animals. Each camera was kept in place until all the flowers on a monitored inflorescence disappeared. The recording mode was set for 30-s video clips with no interruption between clips, and sensitivity was set to normal (Kobayashi, Denda, et al., 2017;Kobayashi et al., 2015).
We calculated the flower visitation rate (VR) and explosive opening rate (EOR) of the inflorescence as shown in Equations (1) and (2): The terms "VR" and "EOR" were used instead of the absolute number of flowers because flowers in an inflorescence matured individually at different times, making it difficult to determine a single flowering period using video camera traps.
The behavior of flower-visiting animals was divided into six categories based on their effects on flowers (Kobayashi et al., 2015):

| Bagging experiments
In

| Nectar survey
In total, 4-6 flowers of two plants were collected every 3 hr during

| Data analysis
To examine the statistical significance of any differences, Fisher's exact test was conducted to compare the fruit set rate, the chisquared test to examine the height above ground level of the visit by the explosive openers, and the Mann-Whitney U test to compare sugar levels in the nectar. All statistical analyses were performed using R ver. 3.5.0 (R Core Team, 2018).

| Flower visitors and their behavior toward flowers
At least 10 species of mammalian flower visitors were recorded by the camera traps (Table 1; Figure 3). Gray-bellied squirrels

| Fruit set rates
No fruits were observed in the bagging experiment, and no flow-

| Nectar characteristics
Nectar volume increased with increased calyx width (Figure 7).
When a flower matured, nectar was stored throughout the day.
The calyx width of flowers opened by animals was 14.13 ± 0.68 mm. The relationship between nectar volume and calyx width (Figure 7) revealed that the nectar content in flowers opened by animals did not exceed 100 µl, and these flowers were therefore opened before all the nectar was fully stored.

| Pollination system of M. macrocarpa in northeastern Thailand
No fruits were observed in the bagging experiment in the present study, as reported in previous studies with similar experiments in different sites (Kobayashi, 2017 (Kobayashi, 2017). Thus, M. macrocarpa showed a clear requirement of explosive opening for fruit setting. In addition, explosive openers removed high quantities of pollen grains, and the stigma made contact with their lower jaw. Although some insects visited opened flowers, most pollen grains of these flowers had already been removed by the explosive openers. Therefore, the explosive openers were likely to be the effective pollinators in this study area and elsewhere (Kobayashi, Denda, Liao, Lin, Liu, et al., 2018;Kobayashi, Denda, et al., 2017;Kobayashi et al., 2015). are pollinators of Madhuca sp., and they segregate flower-visiting times (Yumoto et al., 2000). In the present study, the visiting time  This study showed that all explosive openers visited M. macrocarpa flowers during the daytime. Several Mucuna species have become highly specialized to diurnal visitor species. For example, M. japira, which is pollinated by diurnal birds, stores nectar during the day, whereas the nocturnal bat-pollinated M. urens blooms and secretes nectar only at night (Agostini, Sazima, & Galetto, 2011).
However, M. macrocarpa stores nectar throughout the day in both bat-and squirrel-pollinated regions, even though pollinators are reported to differ between geographic regions (Kobayashi, Denda, Liao, Lin, Wu, et al., 2018). Assuming that M. macrocarpa is a squirrel-pollinated species, this nectar secretion pattern is the characteristic responsible for attracting diurnal animals, such as squirrels.
As for sugar composition, a sucrose-dominant nectar is a common

| Comparison of pollination system of M. macrocarpa among distribution ranges
Mucuna macrocarpa is pollinated by flying foxes and macaques in subtropical Okinawa and temperate Kyushu, respectively (Kobayashi, Denda, Liao, Lin, Liu, et al., 2018;Kobayashi et al., 2015;Toyama et al., 2012). Squirrels occur on neither island, flying foxes do not occur in Kyushu, and macaques are not on Okinawa (Ohdachi, Ishibashi, Iwasa, Fukui, & Saitoh, 2015). A different species of Callosciurus squirrel, C. erythraeus, is a pollinator in Taiwan, where flying foxes are absent (Kobayashi, Denda, et al., 2017). Regarding explosive opening behavior, only Japanese macaques in Kyushu, the northern limit of its range, used both hands for opening, and the others opened using their snout by holding a flower in their forelimb (  ; therefore, this species might have originated in Southeast Asia. The present study revealed that squirrels are the main pollinator When the pollinator shift is observed within a plant species, some plants adapt their floral traits to pollinators in each region (Boberg et al., 2014;Johnson & Steiner, 1997;Nagano et al., 2014;Wester et al., 2019). In M. macrocarpa, flowers were smaller in Thailand than in other regions (Table 5) (Table 5). In addition to flower shape and nectar characteristics, floral color and odor are also important for attracting mammalian pollinators (Faegri & van der Pijl, 1979;Johnson et al., 2011;Knudsen & Tollsten, 1995;Wester et al., 2019). The flowers of M. macrocarpa have pale green and purple petals in all regions. Pale green is one of the characteristics of plants pollinated by nocturnal animals, but purple color is frequently found in both nocturnal and diurnal animal-pollinated plants (Willmer, 2011). Thus, we could not estimate the pollinator based on floral color. As for the odor, it emits strong smell (Toyama et al., 2012), but it is unclear whether the odor regionally differs. Therefore, further studies are needed on attractive traits. To conclude the floral traits, although external characteristics perhaps adapt to the main pollinator in each region, attractive traits such as nectar and flower color suggest that this species may attract a variety of mammals.

| CON CLUS ION
Our results show that the main pollinators of M. macrocarpa in the tropics are likely to be two Callosciurus squirrels and that they divide

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
SK, TD, MI, SW, CA, and SP designed the study. All authors conducted field survey and SK mainly analyzed the data. All authors contributed to writing the paper.

DATA AVA I L A B I L I T Y S TAT E M E N T
Raw data are available in the Dryad Digital Repository: https://doi. org/10.5061/dryad.nt71r49.