Edinburgh Research Explorer Genetic diversity and population structure in the Ryukyu flying fox inferred from remote sampling in the Yaeyama archipelago

The Ryukyu flying fox ( Pteropus dasymallus ) is distributed throughout the island chain spanning across southern Japan, Taiwan, and possibly the Philippines. Although P . dasymallus is listed as VU (vulnerable) in the IUCN Red List, only few genetic works have been conducted to support its conservation. In this study we analyzed 19 markers (mtDNA haplotypes and 18 microsatellite markers) to evaluate genetic diversity and investigate the genetic structure of this species. mtDNA analysis was conducted with 142 DNA remote samples, mostly from faeces, and wing tissues collected on eight islands (Miyako, Ishigaki, Kohama, Kuroshima, Hateruma, Taketomi, Iriomote, Yonaguni). 39 haplotypes were identified in 526bp of the control region, and haplotype network showed no clear genetic structure. Microsatellite analysis was also conducted with 155 samples collected on six islands (Miyako, Ishigaki, Kohama, Taketomi, Iriomote, Yonaguni). It showed that the Yonaguni population exhibits low genetic diversity, high inbreeding, and clear genetic differentiation from other populations. Gene flow between Ishigaki and Miyako through small stepstone islands might be preventing inbreeding of the Miyako population. We provide for the first time indirect proof of long-distance inter-island dispersal in the Ryukyu flying fox and revealed genetic diversity, gene flow and genetic differentiation among the archipelago’s populations. These results will be useful for delineating conservation units and designing specific conservation policies for each island based on metapopulation genetic structure.


Introduction
The Chiroptera represent a speciose and highly threatened group of mammals [1]. In Japan, 37 of the 122 mammal species are bats, yet academic research on this order has been lacking and misdirected towards the least threatened species. The country features significantly high endemicity with endemic bats most endangered while at the same time most poorly studied [2]. Flying foxes, defined hereafter as paleotropical fruit bats belonging to the Pteropus genus, serve important ecosystem functions as pollinators [3] and seed dispersers [4], and play a disproportionally large ecological role in maintaining forest structure and biodiversity [5]. Their seed dispersing capacity relies on their ability to carry large seeds across remarkably large foraging ranges. For instance, the Mauritian flying fox Pteropus niger was reported to fly up to 92 km in a single night [6], while some other large Pteropus species have demonstrated an ability to migrate between islands [7]. This is particularly critical as many paleotropical islands have lost their megafauna [8], leaving only flying foxes to perform large seed dispersal [9]. Nonetheless, their service to forested insular ecosystems is easily disrupted by demographic declines, as flying foxes cease to function as seed dispersers long before they become rare [9]. This reckoning is all the more alarming as flying foxes are arguably the most endangered group of bats worldwide and are most threatened on islands [10]. The intense pressure that they face is best epitomized by the regular conflicts and mass fatalities observed in Australia [11][12][13][14] and the notorious mass culling campaigns inflicted to the remaining population of P. niger by the Mauritian government [10,[15][16][17].
The Ryukyu flying fox (Pteropus dasymallus) is distributed across the Ryukyu archipelago in Japan, two small islands of Taiwan, and possibly in the Philippines [18,19]. P. dasymallus mainly eats fruits, nectar, and sometimes leaves, and plays an important role in pollination and seed dispersal [20][21][22][23]. P. dasymallus is listed as Vulnerable in the IUCN Red List [24] and has been the object of conservation concerns, due to rampant habitat degradation as well as to the recent discovery of previously unreported threat factors, such as predation by feral animals, powerline electrocutions, climatic stress (esp. due to typhoons and heat-induced dehydration), and most importantly conflicts with farmers and culling which have resulted in important fatality [18,25,26]. Low likeability and lack of support for conservation further warrant increased attention [27].
Considering the largely insular distribution range of these bats, genetic information such as diversity, gene flow and genetic differentiation between islands is important for their conservation. Indeed, genetic analyses on another species, Pteropus mariannus, for instance revealed genetic structure and gene flow between islands, and suggested new subspecies classification and conservation units [28]. Yet, tissue sample collection is tedious in the case of solitary, treedwelling fruit bats. On the other hand, non-invasive DNA extraction from faeces also potentially poses technical challenges due to the small size and consistency of the faeces coupled with heavy concentration in polyphenols-a known PCR inhibitor [29]. Hence, to this date, few genetic studies have been conducted on P. dasymallus and no solid knowledge exists on long-range dispersal capabilities and population structure. In this study, we try to remedy this situation by focusing on the Yaeyama archipelago, which hosts the most extensive part of the Ryukyu flying fox population and is inhabited by one allopatric subspecies, the Yaeyama flying fox P.d. yayeyamae. mtDNA haplotype and microsatellite analysis of P. d. yaeyamae were conducted based on remote DNA sampling using mostly faeces and feeding marks. This work provided the first indirect evidence of gene flow between islands and revealed the genetic diversity and structure among the eight major islands of the chain.

Samples
DNA was recovered from various direct and indirect sources, namely faeces, tissue, feeding marks of fruits or flowers, blood and hair. Faeces and fruit samples were collected from 2012 to 2019 (Fig 1) by selection of appropriate sites (mostly foraging sites, but also a few roosts) based on experience. We avoided picking several samples from the same sites, except on islands on which population density was so low that only few sites were available at the time of collection. Faeces and fruit samples were stored in either silica beads or 99% EtOH. Tissue samples, on the other hand, were obtained on Ishigaki island using the standard procedure for bats, which involved capturing P. dasymallus individuals at night using mistnets and taking one 3mm wing biopsy from the plagiopatagium or between the 4 th and 5 th finger in the distal part of the chiropatagium, depending on wing condition (see detailed protocol used in [29,30]). In total, 135 samples from 8 islands (faeces: n = 95, tissue: n = 17, fruits: n = 23) were used for mtDNA sequencing while 155 samples collected in 6 islands (faeces: n = 99, tissue: n = 8, feeding marks: n = 32, hair: n = 15, blood: n = 1) were used for microsatellite analysis. The number of analyzed individuals for each island is shown in S1 Table. DNA extraction was conducted with QIAGEN DNeasy Blood and Tissue Kit (QIAGEN) for tissue samples and QIAamp DNA Stool Mini Kit (QIAGEN) for fecal samples.
Genetic diversity analysis and Principal Coordinate Analysis (PCoA) were conducted by GenAlEx (version) [35,36]. Also, we investigated genetic structure with STRUCUTE [37]. Finally, Fst was calculated by GENEPOP on the Web [38,39] to check genetic differentiation and geneflow between islands. Taketomi was included in only PCoA because the number of analyzed individuals was not enough for the other analyses.

mtDNA haplotype analysis
45 SNPs were detected in 526 bp sequences (including 35 bp which is a part of neighboring translated gene) of 135 samples, and 39 haplotypes (RFF1~RFF39, Accession number: LC528174-LC528212) were defined. The SNPs of each haplotype are shown in S1 Fig, and the number of samples for each haplotype on each island is shown in S2 Table. Haplotype network suggested that haplotypes from each island did not form any clusters and no clear genetic structure was detected ( S2 Fig). However, in the phylogenetic tree by maximum likelihood method, Miyako population seemed closer to Ishigaki population (S3 Fig). Haplotype distribution among islands revealed that 14 haplotypes out of 39 were shared between multiple islands (Fig 1). While some haplotypes (RFF14, RFF36) were widely shared from the east end to the west end of the distribution, others were only found in certain areas. RFF16 and RFF21 were only found in Iriomote and Yonaguni (the west area), while RFF7, RFF24, RFF30, RFF35, and RFF38 were only found in Miyako and Ishigaki (the east area).
The number of haplotypes (Nh), the number of unique haplotypes (unique), haplotype diversity (h), and haplotype richness (hr) of 4 islands are shown in Table 1. The h of Ishigaki and Iriomote were 0.96 and 1.00, while that of Miyako and Yonaguni were 0.69 and 0.61, respectively. Also, the hr of Ishigaki and Iriomote were 7.51 and 9.00, while that of Miyako and Yonaguni were 3.77 and 2.79, respectively. Genetic diversity was higher in Ishigaki and Iriomote than Miyako, and the diversity of Yonaguni was the lowest.

Microsatellite analysis
The number of analyzed individuals (n), observed heterozygosity (Ho), expected heterozygosity (He), and inbreeding coefficient (F) of each island are shown in Table 2. Ho of Ishigaki, Kohama, Iriomote, Miyako were 0.666, 0.650, 0.631, 0.512, respectively. On the other hand, Ho of Yonaguni was 0.388 and F was 0.355. Yonaguni population had low genetic diversity and showed high level of inbreeding compared to the other islands.
In STRUCTURE, LnP(D) suggested that K = 4 provides the best model (S4 Fig), and the latter is shown in S5 Fig. Ishigaki, Kohama, Iriomote and a part of Miyako populations had the same genetic composition, while the rest of Miyako and Yonaguni populations had each different genetic structure.
Fst between Yonaguni and the other islands were generally high (Table 3), and it shows Yonaguni population is genetically differentiated from the other islands' populations.

Genetic diversity and its differentiations
Genetic diversity was high in populations of the central area (Ishigaki, Taketomi, Kohama, and Iriomote) and the lowest in Yonaguni population. Miyako population exhibited a medium level of genetic diversity. Comparing the Ho of each island ( Table 2) to that of wild population of the notoriously endangered Pteropus rodricensis (Ho = 0.5718) [40], Yonaguni (Ho = 0.388) had much lower value, while that of Miyako (Ho = 0.512) was slightly lower. The other islands' were, however, higher (Ho = 0.631~0.666). Also, inbreeding coefficient was high in Yonaguni population. This indicates that genetic diversity has been getting low due to inbreeding depression in Yonaguni population. In mtDNA haplotype analysis, no clear genetic differentiations between islands was detected. However, in microsatellite analysis, populations of 6 islands were genetically divided into 3 groups: Miyako group (Miyako), the Central group (Ishigaki, Taketomi, Kohama, and Iriomote), and Yonaguni group (Yonaguni). Populations in the central group had the same genetic composition, and Miyako group was partly differentiated from them. Yonaguni group was highly differentiated from the other 2 groups.

Gene flow between islands
The main habitats of the Yaeyama flying fox are Ishigaki island and Iriomote island (Central group) and populations in these two have higher genetic diversity than other islands. Other islands of the Central group might be keeping their diversity because of the gene flow linking them with the Ishigaki and Iriomote populations.
The Miyako group is geographically far from the Central group, but there are some islands such as Tarama between Ishigaki and Miyako. There could be gene flow between the two groups through these islands, and that might be why Miyako group was not differentiated completely from the Central group (Fig 2, Table 3) and had a middle level of genetic diversity. Also, Miyako and Ishigaki had the lowest Fst value (0.0221). This could be just because of active gene flow between Ishigaki and Miyako through islands between them. However, it might also be explained by the establishment of the Miyako population through recent migration from Ishigaki, which would explain the high ratio of RFF30 in Miyako in mtDNA haplotype analysis, assuming that the founders had that haplotype. This is also supported by the As regards Yonaguni, there is no steppingstone between this island and the Central group. In a research on the endangered Bonin flying fox Pteropus pselaphon, the other Japanese Pteropus species, genetic diversity of an isolated island was similar to that of Yonaguni [41]. Since Yonaguni lies far away from its closest islands, it can be considered geographically isolated and migration can safely be assumed to happen only on special occasions such as typhoons, which regularly sweep the region westwards. This is supported by our results, which show that Yonaguni is genetically differentiated from the other populations, and that is causing low genetic diversity.

Conservation implications
The observation of gene flow between islands provides the first indirect evidence that long-distance inter-island dispersal is occurring in the Ryukyu flying fox. This indicates that this species is capable of long flights and thus that individual activity may not be assumed to be limited to a small range. This confirms results of ongoing VHF and GPS tracking studies (Vincenot C.E., unpub. data) and casts doubt on the validity of general population trends deduced on the basis of counts conducted in only a small set of geographically localized foraging sites. Furthermore, our results show that island populations may be connected through flight capability and thus generate demographic and genetic interdependence, which need to be resolved to inform conservation policies. This interconnection into a metapopulation may also explain seasonal changes in perceived abundance, which so far have been considered signs of increased activity only. For instance, local inhabitants of Hateruma island, among others, have consistently reported spotting significantly more individuals in summer than the rest of the year, a phenomenon that they attributed to seasonal migrations from Iriomote island (Vincenot C.E., pers. obs.). Our findings, although inconclusive as regards Hateruma island, give some scientific credit to such local claims, especially on island featuring high gene flow.
Considering the results of our genetic research, the Yaeyama subspecies of the Ryukyu flying fox could be divided into 3 conservation units: Miyako unit (populations in Miyako islands), the central unit (populations in the center of distribution including Ishigaki, Taketomi, Kohama, and Iriomote populations), and Yonaguni unit (Yonaguni islands) (Fig 3). The central unit has many individuals and high genetic diversity, and gene flow between the central unit and Miyako unit might be preventing extinction of the Miyako unit. However, the Yonaguni unit is isolated from other populations and could be highly endangered. Recovery of genetic diversity is crucial for population viability of the Yonaguni unit. Particular care should be exerted to reduce pressure on this population by reducing threat factors (see [18,25]).
In conclusion, our analysis supports that island populations, especially the ones forming the backbone of the archipelago, should be considered as variably interconnected through exchanges of individuals. This observation may allow future population monitoring to better comprehend changes in local population sizes and, on the longer term, in genetic diversity by integrating the knowledge gained on gene flow among the different islands. We hope that the present results will be integrated into the design of conservation strategies, which are regrettably still lacking. We propose in particular that outer populations, such as the one on Yonaguni, be the object of focused attention and protection measures, as they seem most at risk of stochastic demographic collapse and inbreeding depression. https://doi.org/10.1371/journal.pone.0248672.g003 S1 Table. (TIFF) S2 Table. (TIFF)