Specimens collection

Since the dispersal of Salamandrella genus depends on water, samples in the same basin and geographically distributed in the same village were used as a sampling point. We detected 21, 14 and 30 individuals in Greater Khingan Ridge (HM, Huma, Heilongjiang Province), Lesser Khingan Ridge (TL, Tieli, Heilongjiang Province), and four populations from Changbai Mountain (SZ, Shangzhi-zhuziying, Heilongjiang Province: 10; SC, Shangzhi-cuijia, Heilongjiang Province: 8; HL, Hailin, Heilongjiang Province: 3; and BS, Baishan, Jilin Province: 9) (see in S1 Fig and S1 Table). Material was collected during the springs and summers of 2018–2019. Specimens (cryopreservation or 95% ethanol-preserved) used are stored at the College of Life Science and Technology, Harbin Normal University (HNU), China.

DNA extraction, PCR amplification and sequencing

Genomic DNA of frozen or 95% ethanol-preserved muscular tissue samples was extracted using the SanPrEP column animal genomic DNA extraction kit of Sangon Biotech (Shanghai) Co., Ltd. The complete sequences of the mitochondrial COI gene were amplified via polymerase chain reaction (PCR) using two pair of primers. Primer 1 (Forward: 5’-AATACACTACGAGGCTTGAT-3’; Reverse: 5’-GTAAGTGACAGAGTGGTTATG-3’; Annealing Temperature: 50°C) and Primer 2 (Forward: 5’-CTTCATGAAAGGGGCTCTACAACCCTTCATGTGGTT-3’; Reverse: 5’-AACTTGAAATTAACCTATGTGGGT-3’; Annealing Temperature: 55°C). This study utilized a newly designed primer set for COI of Hynobiid salamanders. DNA was bi-directional sequenced in the ABI 3730XL genetic analyzer by Sangon Biotech (Shanghai) Co., Ltd. Experimental protocols were approved and performed in agreement with the Institutional Animal Care and Use Committee (IACUC) of the Harbin Normal University (No.: HNUARIA2017001). All methods were carried out in accordance with relevant guidelines and regulations.

Calculation of genetic polymorphism, distances, and differentiation

Successful sequence peak maps were analysed using Chromas 2.6.5 software.

And Blast homologous alignments (https://blast.ncbi.nlm.nih.gov/Blast.cgi) were conducted on the NCBI website to confirm the PCR product’s identity as the target gene. The sequence alignment used Clustalx 2.1 with default parameters [22]. The sequence alignment map was run with DNAMAN 9.0 and the predicted amino acid sequence was translated using Open Reading Frame Finder (ORF) (https://www.ncbi.nlm.nih.gov/orffinder). The motif distributions of the proteins were analyzed using the MEME (Motif-based sequence analysis tools-Multiple Emfor Motif Elicitation, http://meme-suite.org/tool/meme) [23, 24]. The secondary structure was predicted using the RNA fold webserver (http://rna.tbi.univie.ac.at/cgi-bin/RNAWebSuite/RNAfold.cgi) [25]. And the 3D model was predicted using SWISS-Model (http://www.swissmodel.expasy.org/) [26]. Haplotype diversity (H_d), nucleotide polymorphism (π), average nucleotide variation (K), and variable sites (S) were calculated by using Dnasp [27]. Based on Kimura-2-Parameter (K2P), the genetic distances within and between populations were calculated by MEGA 11.0 software [28, 29]. Then, the mantel test between genetic distances and geographical distance using PC-ORD Version6 [30]. Gene flow (N_m) was calculated by equation [N_m = (1/F_st-1)/ 2] [31]. The genetic differentiation index (F_st) between the two populations was calculated using AMOVA analysis in Arlequin 3.5 software, with 1000 repeated samples [28, 29, 32, 33]

Phylogenetic relation and population historical dynamics

Phylogenetic trees and branches for both nucleotide and amino acid sequences were constructed using MEGA 11.0 and PhyloSuite softwares, based on the Poisson-based model (Neighbor-joining, NJ; Maximum parsimony, MP; Maximum likelihood, ML; Bayesian; Timetree). The node confidence (BP) was acquired by 1000 repeated sampling using Bootstrap [29, 34–39]. For outgroups, we used sequences of Hynobiids family that occur on Northeastern China with S. keyserlingii and S. tridactyla. Based on the Median-Joining algorithm in Network 4.6.1.1 [40], the haplotype evolutionary relationship was presented. Based on the mismatch distribution of pairwise haplotypes genetic differences in the COI gene, Tajima’s D and Fu’s Fs were used to estimate the population historical dynamics of Salamandrella species [41–43].

Genetic diversity

The 65 complete sequences of mitochondrial COI gene obtained from six different geographic populations have a nucleotide length of 1550 to 1551 bp. Populations in Greater Khingan Ridge (HM) and Lesser Khingan Ridge (TL) belong to S. keyserlingii, while populations in Changbai Mountain (SZ, SC, HL, and BS) belong to S. tridactyla. Sixty-five individuals contain a total of 81 nucleotide variable sites (Table 1). Base composition analysis showed that GC content ranged from 37 to 38% (S1 Table), showing a significant AT preference and anti-G bias. This is consistent with the composition of Cyt b nucleotides in other vertebrates[44]. In the analysis of haplotype diversity (H_d), nucleotide polymorphism (π), and average nucleotide variation (K), a total of 35 haplotypes (named SK1–SK15 and ST1–ST20, under accession numbers OP050070−OP050104) were found in the COI gene. The TL population had the highest number of haplotypes (11), while the HL population had the lowest number (2). All populations showed medium-high haplotype diversity (0.607–0.978) and low nucleotide polymorphism (0.00129–0.00546) (Table 1). The motif distributions of the 35 COI proteins were analyzed utilizing the MEME. And a total of 6 conserved motifs were identified. Based on this, their secondary structure and 3D model are predicted respectively. The results suggest that all haplotypes are relatively conserved and can be divided into two branches, S. keyserlingii and S. tridactyla, based on species. Additionally, they have distinct protein structures (S2 Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Genetic diversity index of 6 populations of Salamandrella genus.

https://doi.org/10.1371/journal.pone.0298221.t001

Genetic distances and differentiation

Using DANMAN to describe the genetic distances of populations, it was discovered that these distances subdivided the six populations into two categories. HM and TL had a low genetic distance of 0.003, and therefore were clustered together. Other populations (SZ, SC, HL, and BS) cluster together (0.005–0.008). However, the genetic distance between the two classes was large (0.097–0.098). This result is almost consistent with the geographical distance of the population (r = 0.589, P = 0.006) (S3A Fig). The F_st is a measure of population differentiation due to genetic structure, and its value ranges from 0 to 1. In the present study, high genetic differentiation was observed between populations spatially isolated by a long distance. F_st > 0.25 in all populations, indicating greater genetic differentiation. Except for the Nm values (1 < N_m< 4) of SZ and SC, the Nm values of other four population were all greater than 1. This suggests that gene exchange between populations is limited and weak (S3B Fig). Population molecular variation analysis revealed that, when all populations were grouped together, the variation among COI populations accounted for the majority (F_st = 0.94974, P < 0.01). However, the AMOVA analysis of COI genes in geographically close populations showed that most of the variations were found within populations (TL & HM: F_st = 0.32707, P < 0.01; SZ, SC, HL & BS: F_st = 0.44236, P < 0.01) (Table 2).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Analysis of molecular variance of Salamandrella genus (AMOVA).

https://doi.org/10.1371/journal.pone.0298221.t002

Phylogenetic relation and population historical dynamics

The results of the phylogenetic analysis indicated a high degree of similarity between the Bayesian tree of haplotypes and the NJ, MP, ML tree. Additionally, the monophyly of the genus was strongly supported (100%) (Fig 1). All samples of Salamandrella species are categorized into two lineages, in line with the mountain distribution in northeast China. The findings indicate that the haplotypes of S. keyserlingii populations (HM and TL) in Khingan Range constituted a different clade from haplotypes collected from S. tridactyla populations (SZ, SC, HL and BS) in Changbai Mountains (Fig 1). Haplotype relationships were represented in a median-joining network for COI genes, which separated into two segments. Specifically, two populations from the Khingan Range (HM, TL) were linked together, while the remaining four groups from the Changbai Mountains (SZ, SC, HL, and BS) were linked together (Fig 2). Each mountain population had its own unique branch, with no shared haplotypes between populations from different mountains (Fig 2). The phylogenetic tree analysis is further supported by the haplotype network. The haplotype network was used to visualize the distribution of haplotypes in different populations. Circles represent haplotypes and the size of the ring represents the frequency of haplotypes. Similarly, it shows differences in the proportion of haplotypes in different populations (Fig 2). In the present study, using COI genes, the oldest haplotype is not found. And there was no sharing of haplotypes between the two lineages. The haplotypes are interlinked by one or more step mutations. The median vectors (Unmarked dot; Fig 2) present in the network may represent extinct or unsampled sequences in the population. By analysing both datasets, a typical star-shaped haplotype network was observed, indicating a recent population expansion following a bottleneck event [45].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1.

Bayesian phylogenetic tree of Salamandrella genus (a) and Hynobiidae family (b) based on GenBank mitochondrial COI gene. The values on the branch are bootstrap of NJ tree, bootstrap of MP tree, bootstrap of ML tree, bootstrap of Bayes tree, and time (Mya) of species differentiation from left to right. Bootstrap values from 1000 iterations were indicated above the branches.

https://doi.org/10.1371/journal.pone.0298221.g001

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. The median-joining haplotype network and haplotype distribution obtained from sampled populations of Salamandrella genus based on the COI gene.

https://doi.org/10.1371/journal.pone.0298221.g002

The Neutral test (Table 3) and the mismatch distribution analysis (S4 Fig) of the six populations. The results showed that the values of Fu’s Fs and Tajima’s D for the TL population were negative, and the statistical results were significant (P < 0.05). The mismatch distribution pattern was unimodal, indicating a population expansion during the Last glacial period (between 64,000 and 210,000 years ago) for the TL population. The nucleotide mismatch distribution of the other 5 populations showed multi-modal distribution. However, both Tajima’s D and Fu’s Fs tests were not significant (all P > 0.05), and no expansion of population was detected (S4 Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Statistics of neutrality test of mtDNA COI gene of Salamandrella genus.

https://doi.org/10.1371/journal.pone.0298221.t003

Genetic diversity and differentiation

Population genetic diversity is a crucial component of biodiversity as it provides insights into species origin, evolution, and their ability to adapt to different environments. It also helps in understanding the level of genetic diversity, the mechanisms of its formation, and its distribution patterns. Genetic diversity is the guarantee of evolutionary potential and the basis for the conservation work. Important indices such as haplotype diversity (H_d), nucleotide polymorphism (π), and average nucleotide variation (K) are commonly used to measure population variation [46]. In this study, we observed significant geographical differences in genetic diversity among different Salamandrella populations. The haplotype diversity of the six geographic groups was found to be medium-high (all H_d > 0.6), while the nucleotide diversity was low (all π < 0.006). This phenomenon is often observed in aquatic animals and is attributed to the expansion of a small effective population following a period of stability. It can cause lower levels of genetic diversity [47–50]. Rapid population expansion enhances the ability to maintain new mutations but results in low nucleotide diversity [51]. The low nucleotide diversity in Salamandrella species may be linked to geological and climate changes in the Khingan Range and Changbai Mountains, resulting in a "Bottleneck effect" and subsequent "Founder effect" in the species. This effect led to the reduction of group size and loss of genotypes, with genotype legacy only increasing after group size. Additionally, long-term geographical isolation, late sexual maturation, and limited mobility may also contribute to the extremely low level of genetic diversity [52, 53].

Studies have shown that geographical isolation or different habitats within the same water area can restrict gene exchange, leading to population differentiation in aquatic animals [54]. The coefficient of genetic differentiation (F_st) is an important index for assessing the genetic structure of populations [55]. Large F_st (all > 0.25) and small N_m (all < 1.1) between populations indicate significant genetic differentiation. And the F_st (0.94974) between populations significantly affected by geographical isolation indicates that there is a high degree of genetic differentiation among Salamandrella species populations in Khingan Range (HM and TL populations) and Changbai Mountains (SZ, SC, HL and BS populations). It is consistent with the results of phylogenetic relationships (S3B Fig). Molecular variation analysis using AMOVA demonstrated that 94.97% of the total variation was inter-population, with the majority of variation (between 67.29%, 55.76%) being within populations from the same mountain range. This finding is consistent with the results of the haplotype phylogenetic tree (S2 Fig), population genetic distance (S3 Fig), and haplotype evolutionary network (Fig 2). These findings collectively indicate that Salamandrella species does not exhibit significant population differentiation within the same mountain range. It may be due to adjacent gene exchange between geographic populations that suppresses the genetic drift caused by the genetic differentiation [56], which is related to the river system connected by the Mountains.

Phylogenetic relation and population historical dynamics

The ancestors of Salamandrella species separated from other Hynobiids between the Oligocene and mid-Miocene and spread around. However, interspecies competition, climatic conditions, and geographical isolation prevented the expansion of the cold-adapted ancestor of Salamandrella species [1, 4]. The divergence of Salamandrella species occurred from the end of the Tertiary Period to the early Pliocene epoch (6–8 Mya, ~7.70 Mya) [1]. Molecular data have shown that the Schrenck salamander (S. keyserlingii) is much more diverse and older than Siberian salamander (S. tridactyla) (2.5 and 0.49 Myr, respectively). The phylogenetic branch of the Schrenck newt in Northeastern China represents the earliest divergence [11]. Compared with S. keyserlingii, S. tridactyla prefers warmer climate conditions, closer to the ocean, and a smaller inland stretch to the west. When S. keyserlingii invades to the Changbai Mountains, it already exists in the region of S. tridactyla block. The population history was inferred using neutral tests and mismatch distributions. Negative values and statistically significant criteria for Fu’s Fs and Tajima’s D suggest that the sequences exhibit more nucleotide site changes than expected under the neutral evolution model, indicating a history of population expansion [42, 57]. The unimodal shape of the mismatch distribution curve, following a Poisson distribution, indicates the presence of a bottleneck effect or population expansion [58]. The neutral detection results showed that the values of Fu’s Fs and Tajima’s D of TL population were negative and deviated significantly from the neutral test (Table 3). Meanwhile, the distribution pattern showed obvious unimodal (S4 Fig). It indicated that the TL population had expansion of history during the Last glacial period, which is consistent with the results obtained by genetic diversity.

Geographical isolation and strong seismic activity may have impeded expansion and influenced species distribution. Studies confirm that amphibian migration ability is limited, and they are often affected by mountains and rivers barrier. Changes in landform and climate have affected the genetic structure of Hyla species [59–61]. The other example is that Heterixalus madagascariensis and H. alboguttatus exhibit genetic discontinuity due to Mangoro River [62]. In northeast China, the Yilan–Yitong fault zone extends from Liaodong Bay in the South, through Shangzhi and Yilan of the Heilongjiang Province, and finally reaches Luobei. The late Quaternary period saw significant tectonic activity in this area, which is the primary reason for population differentiation in species such as Bombina orientalis [59, 63, 64]. Our study identified two species, S. keyserlingii and S. tridactyla, with the Yilan–Yitong fault zone and Songhua River acting as geographical barriers (S1 Fig). Previous research has confirmed that S. keyserlingii mainly occupies the western side of the fault zone and Songhua River in the Khingan Range [52], while S. tridactyla is distributed in the Changbai Mountain region on the eastern side of the fault zone and Songhua River [2, 9, 11, 13]. This finding is consistent not only with our study but also with observations in Primorye Territory, Russia. However, Maoer Mountain, located at the intersection of the fault zone and Songhua River, is a unique location where both Salamandrella species coexist. Moreover, the distance separating S. keyserlingii (Harbin) from S. tridactyla (Bin) is less than 50 kilometers. Based on sequence variation of nuclear genome genes, we expect to find interspecific hybridization in this sympatric area similar to Russian populations (Jewish autonomous oblast) [7]. The area contains one of the range borders between the two Salamandrella species. However, through the analysis of various phylogenetic trees, the two species in this study belong to two branches and are completely separate. Therefore, S. tridactyla complex was not found in this study. Nevertheless, the high coefficient of differentiation of various groups within the same species may imply the existence of other possible scenarios (e.g., species complex, cryptic species, etc.). So, the Songnen Plain and Sanjiang Plain in Northeast China are the key areas worthy of study. Furthermore, Salamandrella species have also been recorded in the Ussuri River, Wanda Mountains and Khanka Lake. However, their specific conditions require further confirmation. Finally, the sample size in this study is inconsistent due to population size and so on, may be further studied by new techniques such as eDNA [65].

Resources and conservation

Many species of Chinese amphibians are facing a survival crisis due to habitat loss, pollution, invasive species, and overutilization. Some of them have been listed as internationally and/or nationally protected species [66]. In China, certain species of tailed amphibians have been listed in the CITES appendix. Salamandrella species are endangered due to habitat disturbance caused by human activities and decreased habitat quality. Furthermore, the lack of diversity within each geographic group of Salamandrella species and the impact of natural environmental changes and the construction of water projects may contribute to a decline in genetic diversity and a reduced ability to adapt to complex environments. Thus this further threatens the survival of Salamandrella species [67, 68]. This is the main reason why S. keyserlingii was listed as a Category II National Key Protected Wild Animal in China in 2021. The low nucleotide diversity observed in this study underscores the urgent need for the preservation of its genetic diversity [69]. Conversely, S. tridactyla has not yet been listed due to insufficient evaluation of its taxonomic status in China. The results of population genetic structure and geographical pattern of pedigree system in this study can provide a scientific basis for the development of Salamandrella species conservation strategies [17, 59, 62, 70]. Therefore, it is of great significance to the discovery and protection of the cryptic species.

Currently, effective management and conservation measures for Salamandrella species have been implemented in reserves such as Khanka Lake Nature Reserve on the China-Russian border and Changbai Mountain Nature Reserve on the China-Korean border. According to the results of this study, S. tridactyla has been researched about reproductive ecology, population reintroduction, spawning habitats in these area [71–73]. However, neither this study nor previous studies on the Siberian and Primorye-Territory populations have found intermediate haplotypes of the two Salamandrella species [2]. The assessment of cryptic species in China has not been thoroughly evaluated, particularly in biodiversity evaluations. Furthermore, the mechanisms of morphological conservatism in cryptic species requires further investigation [74].