Phylogeography of the Japanese greater horseshoe bat Rhinolophus nippon (Mammalia: Chiroptera) in Northeast Asia: New insight into the monophyly of the Japanese populations

Abstract The Japanese greater horseshoe bat (Rhinolophus nippon) is distributed widely in East Asia. Within the species, R. nippon in Northeast Asia is regarded as the lineage that diverged most recently. However, the monophyly of the Japanese populations is unclear due to insufficient data about phylogenetic relationship of the western Japanese populations. To test the monophyly of the Japanese populations of R. nippon, we sampled R. nippon from western Japan and performed a phylogeographic analysis based on mitochondrial DNA cytochrome b and the D‐loop. The Northeast Asian lineage consisted of three main clades in eastern Japan (clade I), western Japan (clade II), and the continent as well as the Kumamoto population in westernmost Japan (clade III). The results of this study do not support the monophyly of the Japanese population. The findings suggest the “reverse colonization” of R. nippon from the Japanese Archipelago to the Eurasian continent, and provide important insight into the role of the island system in creation and supply of diversity to the continent.


| INTRODUC TI ON
The bat order Chiroptera (Mammalia) is the second most speciose order of mammals, consisting of at least 21 families, 230 genera, and 1401 species (Wilson & Mittermeier, 2019). It is distributed worldwide, including on various islands, but not in the polar regions.
Evolutionary biological studies of organisms on islands-island biological studies-have focused on birds that evolved dramatically as the result of isolation and adaptive radiation to the island environment as represented by Darwin's finches (Burns et al., 2002;Harvey et al., 2021;Lack, 1945;Lamichhaney et al., 2016;Sato et al., 1999;Weir et al., 2009). Although bats are regular members of island ecosystems and adapted to island environments, they have been rarely focused in island biological research.
The greater horseshoe bat Rhinolophus ferrumequinum complex (Chiroptera: Rhinolophidae), which occurs throughout the Palearctic region, including many islands, had been considered a single species until recently (Csorba et al., 2003;Huston et al., 2019;Jo et al., 2018;Sano, 2015;Smith, 2008;Yoshiyuki, 1989). Based on molecular studies (Flanders et al., 2009(Flanders et al., , 2011Koh et al., 2014;Rossiter et al., 2007), the European greater horseshoe bat R. ferrumequinum (Schreber, 1774) in the western Palearctic and the Japanese greater horseshoe bat R. nippon Temminck, 1835 in the eastern Palearctic became recognized as separate species (Burgin, 2019), which were subsequently diagnosed from each other and redescribed based on skull morphological characters (Ikeda, Jiang, et al., 2020). However, unresolved taxonomic problems remain for R. nippon populations in Northeast Asia, which includes northeastern China (Jilin and Liaoning provinces), the Korean Peninsula, the Japanese Archipelago, and peripheral islands.
The Japanese Archipelago is a biodiversity hotspot with many endemic species of various mammals including bats (Motokawa, 2015). In contrast, the distributions of several species of bats, including R. nippon, extend to the Eurasian continent. Many Japanese terrestrial animals are considered to have origin in the Eurasian continent and to have migrated through the Korean Peninsula (e.g., Tamate, 2015). When discussing the origins of Japanese terrestrial animals, phylogeographic patterns among populations in Japan and Northeast Asia are necessary to be clarified. Flanders et al. (2011) reported that R. nippon populations in Northeast Asia (from Jilin Province and eastern Japan) diverged deeply from the parapatric populations in East China (Henan Province) and form a monophyletic group based on 1098 bp of the mitochondrial ND2 gene and 13 microsatellite loci. These data suggest that the Northeast Asian lineage diverged 400,000-600,000 years ago. Then, Liu et al. (2016) revealed that the Japanese populations form a monophyletic clade that diverged from the continental populations (from Jilin and Liaoning provinces and South Korea) more recently, based on mtDNA cytochrome b data. These two studies included Japanese samples only from eastern Honshu, and did not examine the western Japanese samples (western Honshu, Kyushu, and Shikoku islands). Honshu is the largest island in East Asia and several species showed unexpected divergence within the island (Motokawa, 2017). In fact, terrestrial animals in the Japanese Archipelago tend to be diverged two or more lineages in east and west. Therefore, we conducted molecular phylogeographic analyses of Northeast Asian R. nippon, including the western Japanese populations, to verify the monophyly of the Japanese populations. Hokkaido is suitable for intraspecific comparison. Thirty-five sequences from eastern Japan (cytochrome b alleles by Sakai et al. [2003]), South

| Specimens and sampling
Korea, and China deposited in GenBank and 20 sequences from Jilin (Ji'an and Liuhe) and Liaoning (Benxi) provinces reported by Liu et al. (2016) were included in the phylogenetic analyses (Table 1, Figure 1).
In haplotype and nucleotide diversities and neutrality tests, the sequence data of Sakai et al. (2003) were treated as the actual number of individuals.

| DNA extraction, polymerase chain reaction, and sequencing
Genomic DNA was extracted from bat liver tissues preserved in 99% ethanol using the DNeasy Blood & Tissue Kit (Qiagen). MtDNA fragments from the cytochrome b and the D-loop regions were amplified by polymerase chain reaction (PCR) using the newly designed primer pairs RhGluL1 (5'-AAT CAC CGT TGT ATT TCA AC-3') and RhThrH1 (5'-CTT TTC TGG TTT ACA AGA CC-3') for cytochrome b, and the universal primer pairs P and E for the D-loop (Wilkinson & Chapman, 1991 for 7 min. The PCR products were purified using ExoSAP-IT Express PCR Product Cleanup Reagent (Thermo Fisher Scientific K.K.), and the purified products were sequenced by Macrogen Japan Co.
The sequences were edited and trimmed using GAP 4 (Staden et al., 2003), and aligned using ClustalW (Thompson et al., 1994) in MEGA X version 10.1.8 (Kumar et al., 2018). The sequences were then assembled by eye. with PartitionFinder version 2.1.1 (Lanfear et al., 2017). BI was run with four Markov Chain Monte Carlo analyses with 100,000,000

| Phylogenetic analysis
iterations and sampling every 50,000 states. Substitution rates of 1.3% per million years for cytochrome b (Liu et al., 2016) and 20% for the D-loop (Kimprasit et al., 2021;Petit et al., 1999), and a generation time of 2 years (Flanders et al., 2009)  To seek evidence of population growth based on the cytochrome b sequences, Tajima's D (Tajima, 1989) and Fu's F s (Fu, 1997) were tested with 10,000 coalescent simulations in Arlequin version 3.5.2.2 (Schneider et al., 2000). In addition, a median-joining method (Bandelt et al., 1999) was implemented in NETWORK ver-

| Sequencing and haplotypes
First, 1140 bp of cytochrome b and 465 bp of the D-loop were sequenced for all 33 R. nippon and one R. cornutus samples. The sequence data were deposited in GenBank (accession nos. LC605914-LC605946 for cytochrome b, LC605947-LC605979 for D-loop). We incorporated the sequence data from GenBank and a previous study, resulting in the analysis of data from a total of 88 sequences (Table 1) *1 was listed in Table 2.

| Haplotype network and demographic analysis in Northeast Asia
The obtained median-joining networks of the cytochrome b The haplotype and nucleotide diversities were used to interpret the population's demographic history (Grant & Bowen, 1998).
Haplotype and nucleotide diversities based on the D-loop were high in every clade (h > 0.5, π > 0.005) because of the high evolutionary rate. By contrast, the haplotype and nucleotide diversities based on cytochrome b differed among clades (Table 3). Northeast Asia lineage, clade I, and clade II showed high haplotype diversity and low nucleotide diversity (h = 0.827, π = 0.00191 for Northeast Asia; h = 0.582, π = 0.00070 for clade I; h = 0.667, π = 0.00117 for clade II), suggesting rapid growth and the accumulation of mutations after a bottleneck (Grant & Bowen, 1998). Clade III had low haplotype and nucleotide diversities (h = 0.474, π = 0.00068), suggesting the recent occurrence of a bottleneck or founder event (Grant & Bowen, 1998).

F I G U R E 2 A time-calibrated phylogenetic tree constructed using Bayesian inference (BI) method based on cytochrome b and the D-loop.
A number along each branch is posterior probability based on BI. Blue horizontal bars on nodes indicate 95% HPD intervals for node heights. Branches with posterior probabilities >0.95 are shown as bold lines. Identified haplotypes are listed in Table 1 2  (Table 3).

| DISCUSS ION
Our findings support the monophyly of the Northeast Asian lineage of R. nippon, and it is consistent with previous studies. Moreover, our The Northeast Asian lineage is the most recently diverged lineage of R. nippon (Flanders et al., 2011;Liu et al., 2016;Rossiter et al., 2007) in the middle Pleistocene (about 430,000 years ago). Rossiter et al. (2007) suggested that the Japanese population experienced genetic isolation and/or founder effects associated with island effects. The results of genetic diversity and the neutrality tests of the Northeast Asian lineage support the occurrence F I G U R E 3 Median-joining networks based on the mitochondrial cytochrome b (upper left) and the D-loop (middle). Circle size represents haplotype frequency. Identified haplotypes are listed in Table 1 :   When continental lineages are imbedded within clades restricted to islands or archipelagos, one can infer "reverse colonization" as the most likely scenario (Bellemain & Ricklefs, 2008). We found that continental clade III is imbedded within Japanese Archipelago clades I and II, suggesting reverse colonization from the Japanese Archipelago to the Eurasian continent. Reverse colonization generates biodiversity and promotes the assembly of continental biota (Patiño et al., 2017). In the Japanese Archipelago, the alpine plant refugia support reverse colonization of bats from the Japanese Archipelago to the continent via the Korean Peninsula (Figure 4). It is certain that the R. nippon bats flew across between the Japanese Archipelago and the continent. After the last emergence of the land bridge between the Korean Peninsula and the Japanese Archipelago in the middle Pleistocene, the level of the Japan Sea in the glacial periods (MISs 2.2 and 6.2) was 90-100 m lower than in the present, and the Korean Strait width was about one-third of the present 120 km (Oba & Irino, 2012). The sister species R. ferrumequinum can fly up to 30 km between summer roosts and hibernacula , and R. nippon in Kyushu recorded the distance 130 km (Sano, 2015).
Therefore, R. nippon seems to be able to fly over the narrow, shallow Korean Strait during a glacial period.
We estimated that clade I in eastern Honshu and clade II in western Honshu bordered in Kinki District (Figure 1). Such a phylogeographic pattern of divergence between eastern and western Honshu is also seen in some Japanese middle and large mammals, such as the sika deer Cervus nippon (Ba et al., 2015;Nagata et al., 1999;Yamada et al., 2007), the Asian black bear Ursus thibetanus (Ohnishi et al., 2009), and the Japanese macaque Macaca fuscata (Kawamoto et al., 2007). Two hypotheses have been proposed to explain the distribution boundary for Japanese middle and large mammals in Kinki district: the multiple-colonization hypothesis and the refugia hypotheses (Tamate, 2015). Rhinolophus Nippon is not distributed along the coast of Far East Russia, the multiple-colonization hypothesis assuming multiple migrations into the Japanese Archipelago via Korean Peninsula and Sakhalin Island is not applicable. In the refugia hypothesis, organisms evacuated to relatively warm refugia in glacial periods (Bennett & Provan, 2008;Haffer, 1969). In Japan, the southern coasts of Honshu and Shikoku, southern Kyushu, and the southern peripheral islands are considered to have been refugia for plants and animals (Iwasaki et al., 2012;Tamate, 2015;Yamada et al., 2021;Figure 4). We propose that the common ancestor of the Northeast Asian R. nippon was separated into two or more refugia located in the Japanese Archipelago (Figure 4), suggesting that the refugia hypothesis is applicable to R. nippon. Moreover, flight dispersal is a crucial difference between bats and other mammals. This difference might enable the reverse colonization from the Japanese Archipelago to the continent. Many terrestrial animals colonized Japan across the land bridge during the glacial periods MIS 16 and 12 (Aimi, 2002;Ba et al., 2015;Dobson & Kawamura, 1998;Millien-Parra & Jaeger, 1999), and became endemic species. After that, the land bridge had not reappeared, and the Korean Strait prevents the swimming dispersal of terrestrial animals. In contrary, we suggest that bats were able to fly over the strait.
In this study, we targeted the mtDNA cytochrome b and D- Our result suggests that the most likely scenario for Northeast Asian R. nippon involves reverse colonization from the Japanese Archipelago to the continent. The population transition of R. nippon within the Japanese Archipelago is consistent with patterns observed for middle and large terrestrial mammals.

ACK N OWLED G M ENTS
The authors are grateful to T. Okamoto and T. Nakano for their sup-

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.