Taxonomic Reassessment of the Izumo Lineage of Hynobius utsunomiyaorum: Description of a New Species from Chugoku, Japan

Simple Summary In the study, we investigated the true taxonomic composition of Hynobius utsunomiyaorum from Chugoku District, Japan. Our detailed morphological and molecular comparisons showed that H. utsunomiyaorum is clearly divided into two species: the true H. utsunomiyaorum and the Izumo Lineage of H. utsunomiyaorum, i.e., Hynobius kunibiki sp. nov. This new species does not satisfy the diagnosis of H. utsunomiyaorum, while these two species do not form a monophyletic group based on molecular data. Hynobius setoi is morphologically more similar to the new species than is H. utsunomiyaorum. Previous studies have suggested that H. setoi is distributed across the San’in region, i.e., the northern part of Chugoku District, Japan. However, our research revealed that H. setoi is not distributed in the northern part of Shimane Prefecture, located in the western part of the San’in region in Chugoku District. Rather than H. setoi, we found that the new species was distributed in the northern part of Shimane Prefecture. Additionally, results of phylogenetic analyses using all valid Japanese Hynobius species are provided in our study. Based on these results, we show that Japanese Hynobius species included in the subgenus Hynobius may be divided into four genetic clades. The information in our study will be vital for developing conservation management strategies and policies for these species. Abstract Here, we describe a new species of the genus Hynobius from Chugoku, Japan. In populations from central to eastern Shimane Prefecture, the Izumo Lineage of Hynobius utsunomiyaorum was clearly distinguished from the true Hynobius utsunomiyaorum based on morphological and molecular evidence. Thus, we describe the former lineage as a new species, Hynobius kunibiki sp. nov. Morphological comparisons revealed that H. utsunomiyaorum lacks a distinct yellow line on the ventral side of its tail, whereas the new species possesses this yellow line; most H. utsunomiyaorum individuals have distinct white spots on the lateral sides of their body and lack a fifth toe, whereas the new species largely lacks these spots, and all examined individuals had a fifth toe. The two species also differed significantly by several other morphological characteristics. The lentic species Hynobius setoi is morphologically similar to H. kunibiki sp. nov., but they differ significantly by various morphological characteristics. Despite their partial morphological similarity, these two species differed substantially in terms of their genetics. Finally, we show, in a phylogenetic tree including all Japanese Hynobius species, that the subgenus Hynobius can be divided into four genetic clades. Overall, this information will help develop conservation management strategies and policies for these species.


Introduction
In the family Hynobiidae (the Asiatic salamanders), the most diverse genus Hynobius includes 55 species, 37 (67.3%) of which are endemic to Japan [1]. The identification of Hynobius species can be challenging without locality and genetic data because they have few distinguishing morphological characteristics [2]. Previous studies have revealed that cryptic species of lotic or lentic Hynobius exist in Japan [3,4]; that are yet to be described [3,5].
The Hiba salamander, Hynobius utsunomiyaorum, was described from Rokunohara, Saijocho Yuki, Shobara City, Hiroshima Prefecture, and has been found in the Chugoku Mountains in eastern Chugoku District, Japan [3]. According to Matsui et al. [3], this species is not distributed in the northern part of Shimane Prefecture; instead, the typical pond breeding species, Hynobius setoi, is found in this area [3]. However, genetic evidence that H. setoi is distributed in northern Shimane Prefecture was not provided by Matsui et al. [3]; such evidence came only from morphological data. On the other hand, Hayashi and Ooi [5] sampled widely from the northern part of Shimane Prefecture and then performed mitochondrial DNA analyses on these samples. Results showed that haplotypes of H. utsunomiyaorum but not H. setoi were found in the study area [5]. However, populations from the northern part of Shimane Prefecture are clearly distinguishable from true H. utsunomiyaorum in terms of their morphology and ecology; thus, Hayashi and Ooi [5] named these populations the Izumo Lineage of H. utsunomiyaorum and suggested that they may be a new species. They concluded that H. setoi is not distributed in the northern part of Shimane Prefecture [5]; however, detailed morphological evidence for these populations is currently lacking.
The Izumo Lineage suggested by Hayashi and Ooi [5] corresponds to the B2b3 clade suggested by Matsui et al. [3]. Based on mitochondrial DNA and allozyme analyses, this lineage paraphyletic with respect to H. utsunomiyaorum [3,6]; therefore, based on the phylogenetic species concept, it should be considered a new species. When a new species is described, essential features must be directly derived from characteristics observed in specimens rather than indirectly inferred [7]; hence, the morphological characteristics that distinguish a species from its close relatives should be described in a "diagnosis." If the Izumo Lineage does not satisfy the diagnosis conditions of H. utsunomiyaorum, the two should be considered different species based on the morphological species concept. In herpetology, Frost and Hillis [8] recommend the use of the phylogenetic and evolutionary species concepts. If the Izumo Lineage does not form a monophyly with H. utsunomiyaorum after DNA analysis of samples from the entire distribution range of H. utsunomiyaorum, the two should be considered different species based on the phylogenetic and evolutionary species concepts. Although Matsui et al. [3] concluded that H. setoi is distributed in the northern part of Shimane Prefecture based on its morphology, the phylogenetic analysis of Hayashi and Ooi [5] suggests that H. setoi from the northern part of Shimane Prefecture was most likely mistaken with the Izumo Lineage because the two share a similar morphology. Thus, morphological comparisons between H. setoi and the Izumo Lineage of H. utsunomiyaorum are also required.
In the present study, we used additional DNA sequence data to reconstruct the phylogeny of Hynobius collected from Chugoku. Furthermore, we statistically analyzed morphological characteristics to compare two species and one candidate species, i.e., H. utsunomiyaorum, the Izumo Lineage of H. utsunomiyaorum, and H. setoi. We also evaluated the species validity of the Izumo Lineage of H. utsunomiyaorum using the phylogenetic, evolutionary, and morphological species concepts. We revealed the detailed distribution ranges and borders of H. utsunomiyaorum, the Izumo Lineage of H. utsunomiyaorum, and H. setoi by the sampling from their complete distribution ranges; additionally, we investigated the distribution ranges of Hynobius iwami in the eastern part of Shimane Prefecture, which are found near the distribution ranges of the Izumo Lineage of H. utsunomiyaorum. Finally, we constructed a phylogenetic tree including all Japanese Hynobius species and estimated the phylogenetic relationships in this genus.

Molecular Analysis
For phylogenetic analysis, we sampled 70 individuals (from February 2007 to January 2021) representing the Izumo Lineage of H. utsunomiyaorum (27 individuals), H. setoi (24 individuals), H. utsunomiyaorum (13 individuals), H. iwami (three individuals), Hynobius abuensis (one individual from the type locality), Hynobius tsurugiensis (one individual from the type locality), Hynobius kuishiensis (one individual from the type locality), and Hynobius guttatus (one individual not from the type locality) from 70 localities (Table S1; Figure 1). For samples taken during the breeding season, we removed a single tailbud embryo from each paired egg sac and preserved it in 99.5% ethanol. We also obtained tissue samples by clipping the caudal extremities from larvae preserved in 99.5% ethanol. Compared with other groups of Japanese Hynobius, it is more difficult to collect eggs and larvae from the Hynobius stejnegeri group, including H. tsurugiensis, H. kuishiensis, and H. guttatus; therefore, we obtained tissue samples from the caudal extremities of adults in these three species. We extracted total genomic DNA from the tail tips or single tailbud embryos and preserved it in 99.5% ethanol using a DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany). For all individual Hynobius, we amplified a 630-bp fragment of the cytochrome b gene using Ex Taq ® (TaKaRa, Tokyo, Japan) with primers L14010 (5 -TAHGGWGAHGGATTWGAWGCMACWGC-3 ) and H14778 (5 -AARTAYGGGTGRAADGRRAYTTTRTCT-3 ) [9]. The reaction mix (with a total volume of 10 µL) contained 1.0 µL of 10× Ex Taq Buffer, 0.8 µL of 25-mM dNTP mix, 0.5 µL of each of the forward and reverse primers (10 pM), 0.05 µL of Taq polymerase, 6.15 µL of distilled deionized water, and 1.0 µL of template DNA. A T100™ thermal cycler (Bio-Rad, Hercules, CA, USA), was utilized with the following conditions: an initial 3-min denaturing step at 94 • C; 40 cycles of 30 s at 94 • C, 45 s at 56 • C, and 90 s at 72 • C; with a final 10-min extension at 72 • C. We purified PCR products with Illustra™ ExoStar™ 1-Step (GE Healthcare, Buckinghamshire, UK) and sequenced them using BigDye ® Terminator ver. 3.1 (Applied Biosystems, Foster City, CA, USA) on an ABI 3130xl Genetic Analyzer (Applied Biosystems). We deposited the acquired sequences into the DNA Data Bank of Japan (Table S1). We aligned the DNA sequences using MEGA X [10], and then performed phylogenetic analyses of the aligned sequences from all recognized Japanese Hynobius species, with "Salamandrella keyserlingii" used as the outgroup (Table S1), using Bayesian inferences (BIs) and maximum likelihood (ML) estimates. We estimated the best-fit nucleotide substitution model using jModelTest 2 [11] based on the corrected Akaike's information criterion [12] and Bayesian information criterion [13]. We selected the Hasegawa-Kishino-Yano model (gamma distribution with invariant sites) based on both criteria. We constructed Bayesian and ML trees using MrBayes 3.2 [14] and MEGA X [10], respectively. For the Bayesian analysis, we performed two independent MCMC runs for 5,000,000 generations with a sample frequency of 100. We assessed the statistical support for monophyly with posterior probabilities and examined the stationarity of the likelihood scores of sampled trees using Tracer version 1.7 (http://tree.bio.ed.ac.uk/software/tracer/) (accessed on 17 February 2021). The first 25% of generations were discarded as burn-in. We regarded a group as monophyletic when a node had a posterior probability ≥0.95 according to the criterion of Huelsenbeck and Rannala [15]. For ML analysis, we estimated the statistical support for monophyly with 1000 bootstrap replicates; a group was regarded as monophyletic when a node had a bootstrap value ≥80% according to the criterion of Regier et al. [16].     From a conservation perspective, the collected specimens were measured under anesthesia and subsequently returned to their site of capture except for the candidate individuals of type specimens (holotype and two paratypes of both sexes). Before specimens were returned, we captured images of the dorsal, ventral, and lateral sides of all individuals on a black background. We also obtained tissue samples (preserved in 99.9% ethanol) from the tail tips of all individuals as evidence that they had been collected. Salamanders were anesthetized with ethyl 3-aminobenzoate methanesulfonate salt (Sigma-Aldrich, St. Louis, MO, USA) diluted 1000-fold with water [17]. Using a digital caliper, 22 measurements were made to the nearest 0.1 mm on anesthetized specimens, including the following: snout-vent length (SVL), trunk length (TRL), axilla-groin distance (AGD), head length Prior to performing morphological comparisons among the three groups, we tested for normality using a Shapiro-Wilk test. When data were found to follow a normal distribution, we tested for homoscedasticity using Bartlett's test. When the variances among the populations were found to be equal, we performed Tukey-Kramer tests; when variances were found not to be equal, we performed Games-Howel tests. When data did not follow a normal distribution and variances among populations were not equal, we performed Steel-Dwass tests. To examine the overall morphological variation among the three groups, we performed canonical discriminant analysis using standardized values for the 22 measurements. We also performed morphological comparisons between males and females for the three groups. Before the analyses, we tested for normality using a Shapiro-Wilk test. When data followed a normal distribution, we tested for homoscedasticity using an F test. When the variances among the groups were equal, we performed Student's t tests; when variances were not equal, we performed Welch's t test. When data did not follow a normal distribution and variances among groups were not equal, we performed Brunner-Munzel tests. All statistical analyses were performed in R with α = 0.05 as the significance level [18]. Additionally, we used R to calculate the Euclidean distances among the three groups based on the 22 morphological characters in both sexes.
Comparisons. The new species differs statistically from H. utsunomiyaorum in the following length measurements: SVL, RTAL, RMTAH, RVTL, R3FL, R5TL, and RSL in males; RAGD, RMTAH, R3TL, and R5TL in females; the lengths of these measurements, except for R3TL in females, are significantly longer in H. kunibiki sp. nov. than in H. utsunomiyaorum. The ventral edge of the tail of H. kunibiki sp. nov. has a distinct yellow stripe in both sexes (100%), whereas both sexes of H. utsunomiyaorum do not have this yellow stripe (100%). The dorsal edge of the tail of H. kunibiki sp. nov. has a distinct and continuous yellow stripe in both sexes (100%), whereas H. utsunomiyaorum males mostly do not have this yellow stripe (86.8%). Hynobius kunibiki sp. nov. mostly has no distinct white spots on venter in males (95.5%), but H. utsunomiyaorum males largely have these spots (78.9%). Hynobius kunibiki sp. nov. lacks distinct white spots on the lateral sides of the body in most males (95.5%) and all females (100%), whereas H. utsunomiyaorum males (89.5%) and females (90%) largely have these spots. Hynobius kunibiki sp. nov. mostly has DGM in males (81.8%), but males of H. utsunomiyaorum largely lack this gular mottling (76.3%). Hynobius kunibiki sp. nov. has the fifth toe on the hindlimbs in both sexes (100% in males and females), whereas most individuals of H. utsunomiyaorum do not have this fifth toe (89.5% in males, 70.0% in females). The new species is more morphologically similar to H. setoi than it is to H. utsunomiyaorum (Table 2; Figure 3), but it differs statistically from H. setoi in the following length measurements: RTRL, RHL, RTAL, RIND, RIOD, RSL, and RUEL in males; RVTL, RFLL, RIOD, and RSL in females; the lengths of these measurements, except for RTRL in males, were significantly longer in H. kunibiki sp. nov. than they were in H. setoi. Two lentic Hynobius, H. akiensis and H. iwami, are closely distributed with H. kunibiki sp. nov. The new species differs from H. akiensis by the presence of distinct yellow lines on the dorsal and ventral sides of the tail; it differs from H. iwami by the presence of a fifth toe [3]. Results of molecular analyses (Figure 2) [3] reveal that H. hidamontanus is closely related although its distribution area is largely separated from the habitat of the new species. H. hidamontanus differs from the new species by the absence of the fifth toe [3]. The new species and H. sematonotos are nearly distributed and have relatively closer relationship ( Figure 2) in all Japanese Hynobius; however, H. sematonotos differs from the new species by the presence of distinct markings on the dorsum [4].
Coloration. Dorsum is uniformly blackish brown or yellowish brown without distinct black spots (rarely present); venter is lighter than dorsum with no distinct white spots in adult males (occasionally present in females); lateral side of head to tail has no distinct white spots (rarely present); iris is dark brown. When preserved, dorsal coloration tended to fade to dark gray.
Etymology. The specific name is derived from "Kunibiki", with a mythology regarding the formation history of the Shimane Peninsula at which the type locality of the new species is located. The suggested common name in Japanese is Izumo-sanshouo.
Distribution. The new species is endemic to Shimane Prefecture and is known from Oda (including the former Oda City), Yasugi (including the former Hirose Town), Matsue  Remarks. The new species is genetically closest to H. akiensis and is not a monophyly with H. utsunomiyaorum based on previous [3] and present results.

Discussion
Based on our morphological surveys, H. kunibiki sp. nov. clearly did not satisfy the diagnosis suggested by Matsui et al. [3] in four ways: the fifth toe was always present, clear yellow stripes were always found on the dorsal and ventral sides of the tail, SVL was significantly larger (p < 0.0001), and RTAL was significantly longer (p < 0.0001). In particular, the presence or absence of the distinct yellow line on the ventral side of the tail was absolute and applicable to all individuals of both sexes based on our data (Table 2). Genetically, H. utsunomiyaorum and H. kunibiki sp. nov. are clearly distinguishable and do not form a monophyletic group based both on our data ( Figure 2) and that of Matsui et al. [3]. Similarly, the results of phylogenetic analyses using allozyme data based on neighbor-joining and ML methods did not support the monophyly of H. utsunomiyaorum and H. kunibiki sp. nov. [6]. If H. kunibiki sp. nov. is a synonym of H. utsunomiyaorum, polyphyletic species will become valid species. According to Matsui et al. [3], H. kunibiki sp. nov. could not be distinguished from the clade of H. utsunomiyaorum based on small nucleotide polymorphism (SNP) data; hence, Matsui et al. [3] concluded that these two clades were the same species. However, this SNP-based classification is lacking Hynobius hidamontanus despite this species being in the same clade as H. utsunomiyaorum. In addition, this classification was not based on the phylogenetic and morphological species concepts; polyphyletic groups were recognized without detailed morphological comparisons. In contrast, allozyme and mitochondrial data confirm that H. utsunomiyaorum and H. kunibiki sp. nov. are two well-differentiated species [3,6]. Furthermore, H. kunibiki sp. nov. does not satisfy the diagnosis of H. utsunomiyaorum while the two species are clearly distinguishable based on our morphological data ( Figure 3; Table 2). Indeed, our data suggest that this new species is more morphologically similar to H. setoi than to H. utsunomiyaorum (especially in males). Thus, H. utsunomiyaorum and H. kunibiki sp. nov. should be distinct species based on the three species concepts. Individuals from Pop. 31 of Iinan Town (H15) collected by Matsui et al. [3] did not include the haplotype of H. utsunomiyaorum but did contain the haplotype of H. kunibiki sp. nov. (Figures 1 and 2). However, this population is distantly isolated from other populations of H. kunibiki sp. nov. (Figure 1) and our phylogenetic analyses, including samples from Iinan Town (Pops. 106 and 107), did not support the conclusion that the new species was distributed in Iinan Town (Figure 2). Furthermore, detailed information for the voucher specimen from Iinan Town has not been provided here or elsewhere [3]. Thus, the new species we have described may not be distributed in Iinan Town; rather, the distribution range of this species may be limited to the northeastern part of Shimane Prefecture (Figure 1). Hynobius kunibiki sp. nov. is parapatrically distributed with H. setoi and H. iwami (Figure 1), but analyses of nuclear and mitochondrial DNA suggest that H. kunibiki sp. nov. is clearly separated from these two species genetically (Figure 2) [3]. Additionally, H. iwami has no fifth toe; thus, it is clearly separated from H. kunibiki sp. nov. both genetically and morphologically [3].
According to Matsui et al. [3], H. setoi is found at the lowland areas along the Japan Sea from northwestern Hyogo Prefecture to the northeastern part of Shimane Prefecture. Based on our analyses, haplotypes of H. setoi were found from the westernmost part of Hyogo, Tottori, from west to east, and the easternmost part of Shimane Prefectures; however, these haplotypes were not detected from around the northeastern part of Shimane Prefecture (Figures 1 and 2). Similar to our findings, Hayashi and Ooi [5] could not find the haplotypes of H. setoi in the northeastern part of Shimane Prefecture, so individuals of H. setoi from Daitocho Sannoji, Shinobuchi, and Hirosecho Shinoyamasa surveyed by Matsui et al. [3] are unlikely to be H. setoi; thus, the analyses of Matsui et al. [3] may have combined two different species. Consequently, the diagnosis of H. setoi should be carefully applied when attempting to identify this species. Actually, the mean SVL, FLL/SVL, and HLL/SVL suggested by Matsui et al. [3] did not differ significantly between H. setoi and H. kunibiki sp. nov., and both species possessed distinct yellow stripes on the dorsal and ventral sides of their tails (Table 2); therefore, these two species cannot be separated according to the morphology described in the diagnosis of H. setoi by Matsui et al. [3]. In fact, male H. kunibiki sp. nov. were closer to H. setoi than to H. utsunomiyaorum in terms of Euclidean distance, despite H. kunibiki sp. nov. having a closer genetic relationship with H. utsunomiyaorum than it has with H. setoi (Figure 2). Males of H. kunibiki sp. nov. and H. setoi differed significantly in terms of RTRL, RHL, RTAL, RIND, RIOD, RSL, and RUEL; hence, these morphological characteristics will be a useful reference by which to distinguish the two species.
Consistent with the results of Matsui et al. [3], our phylogenetic analyses suggested that H. hidamontanus is a synonym of H. utsunomiyaorum ( Figure 2). However, we could not collect samples of H. hidamontanus because it is strictly protected by law in Japan. Therefore, a morphological survey of this species will be required for its synonymization. In the present study, the entire distribution ranges of H. kunibiki sp. nov. and H. setoi were revealed in detail (Table S1; Figure 1), whereas the distribution ranges of H. utsunomiyaorum from the central and eastern part of Hiroshima Prefecture and H. iwami from the western part of Shimane Prefecture were not revealed in detail. Thus, additional research will be required to determine the distribution areas of the latter two species.
We conducted phylogenetic analyses using all Japanese Hynobius species (Figure 2). According to Dubois and Raffaëlli [19], Japanese Hynobius are divided into five supraspecies (i.e., hidamontanus, lichenatus, naevius, nebulosus, and stejnegeri); however, supraspecies hidamontanus and naevius cannot be distinguished genetically (Figure 2 [19] used the term supraspecies, but the common morphological characteristics of each supraspecies have not yet been evaluated to the best of our knowledge; hence, this term should not be used without an initial morphological definition. If the supraspecies are valid, further morphological studies among the groups are necessary.
Following this description, the number of Japanese Hynobius spp. is 38, but some taxonomic problems have been left unsolved. Matsui et al. [3] suggested that H. akiensis and H. nebulosus are not monophyletic species, and they may be divided into two species. Further, the study only described male morphology; thus, female morphology of many Hynobius species remains unidentified. The task of clarifying the distribution ranges of lentic Hynobius also needs to be undertaken as detailed distribution ranges of H. setouchi, H. akiensis, and H. vandenburghi are still unknown. Future studies involving taxonomic reexamination and field surveys to clarify the distribution ranges of both sexes of the Japanese Hynobius species are required.
The localities of the four species used in this study have been provided in detail, which will support conservation activities. In particular, H. kunibiki sp. nov. and H. setoi are morphologically similar, so definitive identification of these species will be difficult where DNA data is unavailable. The information in this study is expected to be useful for confirming the distribution ranges of these two species. However, it is possible that overcollection could decimate the Hynobius species following the release of detailed locality information. Indeed, overfishing of H. utsunomiyaorum has already been recognized [20]. Therefore, the conservation status of H. kunibiki sp. nov. must be reassessed, and management plans for its conservation and regulation by national or regional administrations are immediately required to prevent overcollection and possible extinction.

Conclusions
The monophyly of the Izumo Lineage of Hynobius utsunomiyaorum and H. utsunomiyaorum was rejected by the molecular evidence. Besides the genetic evidence, the Izumo Lineage of H. utsunomiyaorum did not satisfy the diagnosis from the original description of H. utsunomiyaorum in morphology. Furthermore, significant morphological differences between them were detected in both sexes, so we described the Izumo Lineage of H. utsunomiyaorum as H. kunibiki based on morphological, phylogenetic, and evolutionary species concepts. This new species is limited to the northern part of Shimane Prefecture, and the conservation status should be reassessed for appropriate conservation activities.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/ani11082187/s1, Table S1. List of specimens using molecular analyses. Population number corresponds to the localities (in Figure 1) where the salamanders were collected. Asterisks after the sampling localities show the type locality of each species.