A new species of lotic breeding salamander (Amphibia, Caudata, Hynobiidae) from Shikoku, Japan

Nomenclatural acts

The electronic version of this article in portable document format will represent a published work according to the International Commission on Zoological Nomenclature (ICZN), and hence the new names contained in the electronic version are effectively published under that Code from the electronic edition alone (see Articles 8.5–8.6 of the Code). This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank Life Science Identifiers (LSIDs) can be resolved and the associated information can be viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org/. The LSID for this publication is as follows: urn:lsid:zoobank.org:pub: 1EFFBF1B-1249-4A9F-8AB2-36BFCD6F6B9E. The online version of this work is archived and available from the following digital repositories: PeerJ, PubMed Central and CLOCKSS.

Sample collection

Using the protocol of Nishikawa et al. (2007), we collected salamanders in the field and fully anesthetized them with an acetone-chloroform solution for subsequent processes. Liver and muscle tissues were collected from the anesthetized salamanders and preserved in 99% ethanol or stored in a deep-freezer for molecular analyses. Voucher specimens were then fixed in 10% formalin, and later preserved in 70% ethanol for permanent storage at the Graduate School of Human and Environment Studies, Kyoto University (KUHE). Animal collection and experiments followed the guideline of animal experiments of the university and were approved by the animal experimentation ethics committee at the KUHE (certificate number: 29–A–7, 30–A–7, 20–A–7, 20–A–5).

Genetic sample and DNA extraction

To survey intraspecific genetic variation in H. hirosei, 116 samples covering the entire geographic range of Shikoku were employed (Fig. 1A, Table 1). Total genomic DNA was extracted from the tissues using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany).

Mitochondrial DNA analysis

The sequence data of full-length cytochrome b (cyt b; 1,141 bp) of mtDNA were obtained using the primers HYD_Cytb_F1 5′–CYAAYCCTAAAGCWGCAAAATA–3′ (Matsui et al., 2008); salamander_cytb_R_N2 5′–YTYTCAATCTTKGGYTTACAAGACC–3′ (Matsui et al., 2008); cytb_R1_cynops 5′–AARTAYGGGTGRAADGRRAYTTTRTCT–3′ (Aoki, Matsui & Nishikawa, 2013); and cytb_F2_cynops 5′– CAYTTYYTGYTMCCATTYYTAATTGCAGG–3′ (Aoki, Matsui & Nishikawa, 2013).

The detailed experimental protocols were described by Aoki, Matsui & Nishikawa (2013). Sequences were assembled and checked using the Chromas Pro 1.34 software (Technelysium Pty Ltd., South Brisbane, QLD, Australia). The resulting sequence was deposited in GenBank (accession numbers ON110825–ON110940). For comparisons, the GenBank data of H. amakusaensis from Kumamoto Prefecture (voucher number: KUHE 30338; GenBank accession number: AB921167), H. boulengeri from Nara Prefecture (KUHE 25653: AB266675), H. ikioi from Miyazaki Prefecture (KUHE 22817: AB921162), H. katoi from Shizuoka Prefecture (KUHE 37128: AB266673), H. kimurae from Shiga Prefecture (KUHE 16689: AB266674), H. naevius from Saga Prefecture (KUHE 28584: AB266672), H. osumiensis from Kagoshima Prefecture (KUHE 24797: AB921165), H. oyamai from Oita Prefecture (KUHE 32584: LC433769), H. retardatus from Hokkaido Prefecture (KUHE 13057: AB363573), H. sematonotos from Hiroshima Prefecture (KUHE 34694; LC433768), H. shinichisatoi from Oita Prefecture (KUHE 22889: AB921163), H. stejnegeri from Kumamoto Prefecture (KUHE 28011: AB921166) and Salamandrella keyserlingii from Hokkaido Prefecture (KUHE 13057: AB363573) were added. Table 1 lists the citations that contain these sequences.

The obtained sequence data were aligned using MAFFT 7 (Katoh & Standley, 2013). Before phylogenetic analysis, we selected the best substitution model using Kakusan4 (Tanabe, 2011) based on the Akaike information criterion (Akaike, 1974) for Likelihood (ML) method and Bayesian information criterion (Schwarz, 1978) for Bayesian inference (BI) method. The ML tree was estimated using RAxML 8.2 software (Stamatakis, 2014). For the confidence of the ML tree, bootstrap analysis was conducted with 1,000 replicates. Nodes with bootstrap values (BS) >70% were considered as sufficiently supported (Huelsenbeck & Hillis, 1993). The BI tree was generated using MrBayes 3.2.6 software (Ronquist et al., 2012). The BI analyses were performed using two parallel runs of four Markov chain Monte Carlo (MCMC) methods for 20 million generations. The first 25% of the generations were discarded as a burn-in, and one of every 100 of the remaining generations was sampled. The convergence of the MCMC runs was checked using TRACER 1.6 software (Rambaut et al., 2014). The uncorrected p-distance among samples was calculated using MEGA 7 software (Kumar, Stecher & Tamura, 2016).

SNP analysis

To estimate the genetic structure of H. hirosei (114 individuals from 26 populations; DDBJ accession number: DRR361514–DRR361627), we used genome-wide SNP obtained based on MIG-seq (Suyama & Matsuki, 2015) using the Illumina MiSeq system. We followed the protocol described by Suyama & Matsuki (2015), except that an annealing temperature of 38 °C (originally, 48 °C) for the first PCR and indexed forward and reverse primers for the second PCR (originally, common forward and indexed reverse primers) were employed. The sequenced data were filtered using fastp ver. 3 (Chen et al., 2018) to trim the primer regions and remove low-quality and short reads. The qualified quality phred, length required, and front/tail trimming settings were 30, 80, and 14, respectively. After filtered reads 1 and 2 were connected to one, Stacks ver. 2.3c (Catchen et al., 2011) was executed for SNP detection using the ‘denovo_map.pl’ program as follows: number of mismatches allowed between stacks within individuals (M) = 2; and number of mismatches allowed between stacks between individuals (n) = 2. Of the several stages in the program, ‘ustacks’ was configured as follows: minimum depth of coverage required to create a stack (m) = 3 and maximum distance allowed to align secondary reads to primary stacks (N) = 2. Finally, the ‘populations’ program was set as follows: minimum percentage of samples in a population (r) = 0.7; minimum number of populations in a locus (p) = 2; and specify a minimum minor allele frequency required to process a SNP (min-maf) = 0.05; and specify a maximum observed heterozygosity required to process a SNP (max-obs-het) = 0.75.

The genetic structures within H. hirosei were estimated using Bayesian clustering in STRUCTURE ver. 2. 3. 4 (Pritchard, Stephens & Donnelly, 2000), as well as EasyParallel (Zhao et al., 2020), which is multithreaded parallelization to facilitate population structure analysis. The Monte Carlo Markov chains ran for 1 million generations, including a burn-in of 100,000 generations. This calculation was repeated ten times for the number of each cluster to evaluate genetic differentiation in nuclear markers among the clusters. Multiple STRUCTURE runs were combined using CLUMPP (Jakobsson & Rosenberg, 2007).

Morphological data and analysis

Measurements to 0.1 mm were recorded using a Mitutoyo digital caliper and a stereoscopic microscope when necessary. For the morphological comparisons, 83 adult specimens of H. hirosei were used and, following 26 metric characters and five meristic characters, were employed by referring to Nishikawa et al. (2007): 1) SVL (snout-vent length); 2) HL (head length); 3) HW (head width); 4) MXHW (maximum head width); 5) SL (snout length); 6) LJL (lower jaw length); 7) IND (internarial distance); 8) IOD (interorbital distance); 9) UEW (upper eyelid width); 10) UEL (upper eyelid length); 11) AGD (axilla-groin distance); 12) TRL (trunk length); 13) TAL (tail length); 14) BTAW (basal tail width); 15) MTAW (medial tail width); 16) BTAH (basal tail height); 17) MXTAH (maximum tail height); 18) MTAH (medial tail height); 19) FLL (forelimb length); 20) HLL (hindlimb length); 21); 2FL (second finger length); 22) 3FL (third finger length); 23) 3TL (third toe length); 24) 5TL (fifth toe length); 25) VTW (vomerine tooth series width); 26) VTL (vomerine tooth series length); 27) UJTN (number of upper jaw teeth); 28) LJTN (number of lower jaw teeth); 29) VTN (number of vomerine teeth); 30) CGN (number of costal grooves following Misawa (1989)); and 31) LON (number of costal folds between adpressed limbs).

When the groups recognized in H. hirosei based on molecular analysis were compared (see Results), only adult males were used as H. hirosei is known to exhibit sexual dimorphism in most of morphological characters (Nishikawa et al., 2007) and a larger number of adult males than adult females was obtained. Sex and maturity were determined by direct gonad observations. To examine differentiation between the groups, we compared SVL using the Student’s t-test. The Mann-Whitney’s U-test was used to assess meristic characters and ratio values of metric characters to SVL. To examine the overall morphological variation between the groups, linear discriminate analysis (LDA) was conducted using log_e-transformed metric values of the total of the 26 measurements. The difference in the LDA score between recognized groups was then determined using the Mann-Whitney’s U-test.

The clutch size among the populations was compared using Tukey’s pairwise post-hoc test.

In these analyses, the TAL data for the regenerated tail were omitted. All statistical analyses were performed using Past 4.04 (Hammer, Harper & Ryan, 2001).

For larval specimens, the developmental stage was determined according to Iwasawa & Yamashita (1991) and the SVL was measured accordingly.

Mitochondrial DNA analysis

Complete sequences of the cyt b (1,141 bp) of mtDNA were obtained, which included 450 variables and 345 parsimonious informative sites. The best selected substitution models were the codon-equal rate model with the general time reversible model (GTR: Tavaré, 1986) + G for the first, second, and third in the ML analysis and the codon proportional model with SYM_Gamma (Zharkikh, 1994), HKY85_Gamma (Hasegawa, Kishino & Yano, 1985), and GTR_Gamma for the first, second, and third in the BI analysis, respectively. MCMC analysis using MrBayes did not converge; thus, only the phylogenetic tree constructed by ML (likelihood value [lnL] = −7,206.929; Fig. 1B) was presented. This tree indicated that a clade including H. amakusaensis, H. ikioi, H. osumiensis, and H. shinichisatoi, H. naevius, H. stejnegeri, the Tsurugi group in H. hirosei (BS = 100) from around the Tsurugi Mountains (localities 1–6, 10), H. oyamai, H. katoi, and a clade including H. sematonotos and the remaining samples of H. hirosei formed one clade (BS = 98), although the phylogenetic relationships within the clade were not well resolved. The remaining samples of H. hirosei were separated into the Central group (BS = 99) from the Sanuki Mountains (localities 8 and 9), the Takanawa Mountains (locality 19), and western part of the Shikoku Mountains (localities 10–18, 20–23); and a clade of the Nanyo group (BS = 100) from the Onigajo and Sasayama Mountains (localities 24–26). The Nanyo group had a sister relationship with H. sematonotos (BS = 99). Individuals of both the Tsurugi and Central groups were sympatrically found at locality 10.

The mean pairwise genetic distances (Table 2) were 9.4% between the Tsurugi and Central group, 10.3% between the Tsurugi and Nanyo group, and 6.9% between the Central and Nanyo group. The genetic distance between the Nanyo group and H. samatonotos was 4.6%.

SNP analysis

A total of 373,642 reads generated based on MIG-seq were filtered using fastp, resulting in 287,094 reads. Stacks detected 716 SNP loci in the reads, and the genotyping rate was 0.74. Based on STRUCTURE analysis, at K = 2, individuals were clustered into southeastern Shikoku (blue) vs. the remaining area of Shikoku (orange), and transitional structures were found at localities 8 and 9 (Sanuki Mountains) and 10–12 (Fig. 2). Locality 7 was included in the Central group by mitochondrial analysis, but was judged as the Tsurugi group in the STRUCTURE analysis. At K = 3, individuals from southeastern Shikoku (blue), the remaining area of Shikoku, except the Onigajo and Sasayama Mountains (orange), and the Onigajo and Sasayama Mountains (localities 24–26; green) were divided. These three groups recognized by mtDNA analysis were also detected using the SNP data; however, the Tsurugi and Central groups were not separated and clinally continued in the SNP data. Thus, we temporarily treated these two groups as one group for the morphological analyses.

Morphological analysis

To survey the morphological separation of the Nanyo group from the other groups, we conducted morphological analyses of the Nanyo and Tsurugi + Central groups. The obtained values of SVL and the ratios of the other characters to the SVL are shown in Table 3. The Nanyo group had significantly lower values for SVL (T = 2.8601, P = 0.0060), significantly larger values for RAGD (U = 80.00, P = 0.0002), RUEL (U = 146.00, P = 0.0156), and RMTAW (U = 167.00, P = 0.0448), and significantly lower values for RTAL (U = 123.00, P = 0.0061) and RIND (U = 130.50, P = 0.0065) than the remaining samples (Tsurugi + Central group).

In the LDA results, the eigenvalues of the axis accounted for 2.79. On the axis, the highest absolute magnitude of the standardized discriminant coefficients was −266.35 of SVL, followed by that of TRL (224.66), HL (72.85), AGD (−27.06), and BTAW (−10.84). The Nanyo group and the remaining samples were significantly separated (U = 0.00, P = 0.0000), with a small overlap observed (Fig. 3).

Systematics

In summary, the Nanyo group and the remaining samples were clearly separated based on the nuclear genome and external morphology, whereas the Tsusrugi and Central groups formed hybrid zones. The Nanyo group and the remaining samples were genetically isolated; thus, the Nanyo group must be an independent evolutionary lineage with sufficient genetic and morphological differentiation. This information strongly indicates that the Nanyo group is an independent species. The remaining samples (hereafter H. hirosei sensu stricto) included topotypic H. hirosei from Mt. Ishizuchi; however, the Nanyo group has not been named. Therefore, the Nanyo group was revealed to be a new species.

Hynobius oni sp. nov. urn:lsid:zoobank.org:act:D9C7A32A-7CB0-44F1-A0F2-202E7528EBEA

(Japanese name: Nan-yo-sanshouo)

(Figs. 4D–4F, 5–8)

Pachypalaminus boulengeri: Sato (1934): 464 (part);

Hynobius (Pachypalaminus) boulengeri: Nakamura & Uéno (1963): 13 (part);

Hynobius boulengeri (population 9): Nishikawa et al. (2001): 281 (part).

Holotype: KUHE 62785, an adult male from Mt. Yatsuzura, Uwajima City, Ehime Prefecture, Shikoku, Japan (N 33.18°; E 132.62°; alt. 1,000 m a.s.l.), collected by S. Kanamori on April 24, 2021.

Paratypes: A total of 11 specimens: one female (KUHE 61097) from Mt. Onigajo, Uwajima City, Ehime Prefecture (N 33.19°; E 132.61°; alt. 980 m a.s.l.), collected by M. Hibino on May 3, 2019; one male (KUHE 61098) collected by Y. Tomimori on April 29, 2019, one male (KUHE 62327) collected by Y. Onishi on October 25, 2020, and eight males (KUHE 62784, 62786–62792) collected by S. Kanamori, K. Nishikawa, and Y. Suzuki on April 24, 2021, from the type locality.

Referred specimens: One male specimen (KUHE 61096) from Sozu, Ainan Town, Ehime Prefecture (N 33.04°; E 132.57°; alt. 730 m a.s.l.), collected by I. Fukuyama on May 2, 2019; and five juvenile specimens (KUHE 62320–62324) from the type locality, collected by Y. Onishi, S. Kanamori, Y. Tomimori, and I. Fukuyama on October 24, 2020.

Etymology: The specific name is derived from the “Oni” in Japanese, which is a traditional Japanese demon. The habitats of the new species are areas where prior generations believed that the Oni and Ushi-oni, which is a type of Oni, occurred. The type locality is located in the Oni-ga-jo Mountains, which is considered to be the castle of the Oni.

Diagnosis: A large-sized species (adult SVL 73.6–87.5 mm in males) of the lotic-breeding Hynobius, breeding in montane streams; dorsum uniformly dark reddish brown and immaculate in adult; tips of fore- and hindlimbs adpressed on body scarcely meeting (overlap of −2.0 to 0.0 costal folds in males); fifth toe well developed; ova large, pigmentless; egg sacs relatively long and crescent in shape, with distinct whiptail structure on free end; larvae lack claws on their tips of fingers and toes; most similar to H. hirosei, but distinct based on its smaller body size, longer axilla-groin distance, shorter tail length, shorter internarial distance, longer upper eyelid length, and larger medial tail width. Hynobius oni is genetically closer to H. sematonotos than H. hirosei based on mtDNA; however, H. oni has no large markings on the body, in contrast to many silvery spots on H. sematonotos, and a larger SVL than H. sematonotos.

Description of holotype: Head-body moderately large and robust; head oval and moderately depressed, distinctly longer than wide; snout rounded, slightly projecting beyond lower jaw; nostril close to snout tip; labial fold absent; eye large, prominently protruded, slightly inset from edge of head in dorsal view; upper eyelid well developed, shorter than snout; gular fold distinct, curving slightly anteriorly; parotoid gland evident, extending from angle of jaw to gular fold; postorbital grooves distinct, branching posterior to angle of jaw, one short and running down to lower jaw, the other long and posteriorly to parotoid gland; vomerine tooth series wider than long, V-shaped, anterior margin distal to choanae; tongue broad, both sides free from mouth floor; fore- and hindlimbs long and thick; number of costal grooves between axilla and groin 13; depressed limbs separated by two costal folds; relative length of fingers I<IV<III<II, toes I<V<II<IV<III; fifth toe well developed; cloaca longitudinal slit; genital tubercle on anterior cloaca absent; tail moderately short and thick, cylindrical at base, increasingly compressed posteriorly, dorsal fin evident posteriorly; tip of tail rounded in lateral view.

Measurements and counts of holotype (measurements in mm): SVL 82.7; HL 18.6; HW 13.6; MXHW 14.2; SL 5.4; LJL 11.2; IND 4.6; IOD 4.2; UEW 2.9; UEL 4.9; AGD 45.0; TRL 64.1; TAL 66.3; BTAW 8.4; MTAW 6.2; BTAH 6.5; MXTAH 9.3; MTAH 8.6; FLL 19.4; HLL 24.5; 2FL 3.7; 3FL 3.5; 3TL 5.9; 5TL 2.5; VTW 4.8; VTL 3.6; UJTN 71; LJTN 67; VTN 43.

Color: Dorsum uniformly dark reddish brown without marking; underside lighter than the dorsum; underside of tail slightly ochre; iris dark brown without marking. In preservatives, dorsal coloration tends to fade and becomes gray-brown, but otherwise has no obvious changes.

Variation: Morphometric data are presented in Table 3. Individuals of the type series and referred specimens are generally similar in body size and proportion. The single paratype female (SVL = 84.6 mm) is as large as the male holotype (82.7 mm). The female has smaller upper eyelid width (3.1%SVL vs. 3.2–3.8%SVL in males [n = 12]) and lower and thinner tail (BTAH: 7.8%SVL vs. 8.7–10.2%SVL; MTAH: 5.1%SVL vs. 6.2–7.6%SVL; BTAH: 6.6%SVL vs. 7.2–9.6%SVL; MXTAH: 7.0%SVL vs. 8.4–12.0%SVL; MTAH: 6.3%SVL vs. 8.3–10.5%SVL) than males (n = 12). The dorsal ground color is usually dark reddish brown, but is sometimes less reddish and darker. Young individuals have dorsum scattered by white dots.

Eggs and egg sacs: Hynobius oni has a smaller clutch size (mean: 23.5 ± 5.7, range: 16–36, n = 21) than H. hirosei sensu stricto from Mt. Ibuki, Ehime Prefecture (mean: 34.3 ± 7.6, range: 25–59; n = 81 (Tanabe & Okayama, 2004); df = 110, F = 16.01, P = 0.0000) and from the Tsurugi Mountains, Tokushima Prefecture (mean: 31.0 ± 12.4, range: 15–52, n = 9 (Tamura, 2012); df = 110, F = 16.01, P = 0.0356). Egg sacs are crescent in shape (length 176.8 ± 23.8 mm (n = 15), width 17.0 ± 1.9 mm (n = 15)) with relatively thicker envelope than that of other congeners, but thinner than that of H. boulengeri, and similar in thickness to that of H. naevius, H. oyamai, and H. sematonotos, with a distinct whiptail structure on the free end (length 35.2 ± 23.7 mm (n = 15)). Both the animal and vegetal poles of eggs have a cream color. The eggs (oval diameter: 4.9–5.9 mm) form a single and/or double row in each egg sac.

Larvae: SVL for fully grown larvae at stage 63 (n = 3) was 21.7 ± 1.5 mm, and SVL for a larva when onset of metamorphosis in late July at stage 64 (n = 1) was 20.6 mm; head rounded in dorsal and lateral views (Fig. 6); snout short and broadly rounded; eyes slightly protruded, inset from the edge of head in dorsal view; labial fold distinct at posterior half of upper jaw; external gills developed; caudal fin higher than head; dorsal fin higher than ventral fin; origin of dorsal fin at distal half to three-fourth of trunk; ventral fin originating from vent; tail tip weakly pointed; limbs slender; claws on fingers and toes absent. In life, dorsum light brown with small dark-brown dots and blotches; venter whitish and transparent; golden dots scattered on tail fin. In preservative, the dorsal coloration tends to fade, becoming light brown and golden dots fading to white.

Comparisons: Among species of the lotic-breeding Hynobius, H. oni with uniform dorsal color resembles H. boulengeri, H. katoi, H. shinichisatoi, H. osumiensis, and H. hirosei, whereas the remaining species in the group, H. kimurae, H. fossigenus, H. ikioi, H. amakusaensis, H. naevius, H. oyamai, H. sematonotos, H. stejnegeri, H. tsurugiensis, H. guttatus, and H. kuishiensis, have markings that can be easily distinguished from the new species. Hynobius oni is distinct from H. boulengeri (mean male SVL 79.4 mm vs. 93.9 mm in H. boulengeri) and H. hirosei (85.5 mm) based on its smaller body size, and from H. katoi (58.4 mm) and H. osumiensis (68.4 mm) based on its larger body size. Hynobius oni is distinguishable from H. shinichisatoi based on its shorter tail length (mean 80.4%SVL vs. 90.4%SVL in H. shinichisatoi) and shallower vomerine teeth series (VTW/VTL: 1.6 vs. 1.2 in H. shinichisatoi). In addition to the significantly smaller body size of H. oni than that of H. hirosei, H. oni is distinguished from H. hirosei using the following ratio values: longer axilla-groin distance, shorter tail length, shorter internarial distance, longer upper eyelid length, and larger medial tail width.

Range: Known only from the Onigajo and Sasayama Mountains in Ehime and Kochi prefectures, southwestern part of Shikoku Island, western Japan (Fig. 1A).

Natural history: Hynobius oni inhabits around mountain streams with partially exposed bedrock (Fig. 8). On April 29, 2019, egg sacs, some females with egg, and many males were observed in the water. On April 24, 2021, 12 adult males and egg sacs were found under stones in the water. Thus, the breeding season is presumed to be late April. Adults in the non-breeding season and juveniles were found under stones, rotten wood, and debris near the stream. However, the larval life history of H. oni is poorly understood. Only one overwintered larva (total length: 71 mm) was once found in an open stream through three-night surveys on April 29, May 3, 2019 and April 28, 2022. In the surveys, the observed density of overwintered larvae was lower (approximately 0.25 per 1 m²) than that of H. hirosei (more than 10 per 1 m²; our personal observation) on May 8, 2021. No sympatric hynobiid salamanders (Hynobius or Onychodactylus) were observed with H. oni.

Conservation: As the known range of the new species is restricted to small areas of only the Onigajo and Sasayama Mountains, habitat destruction and over-collecting for private purposes could have negative impacts on the natural populations, as reported for Onychodactylus tsukubaensis (Yoshikawa & Sakamoto, 2016), which also occurs in small areas (the Tsukuba Mountains of Ibaraki Prefecture). Hynobius oni has a very small range and some habitats have been degraded; thus, this species should be protected as an endangered species. Furthermore, the conservation status of H. hirosei should be reconsidered immediately after this taxonomic change (presently, listed as Near Threatened (Matsui, 2014)).