A total of 18 Karstsinnectes specimens were collected during field surveys from 2019 to 2024, from Guangxi and Guizhou in southwestern China (Fig. 1). The treatment of experimental animals in this study was consistent with the Chinese animal welfare laws (GB/T 35892–2018). All of the specimens used for morphological studies were fixed in a 10% formalin buffer and then transferred to 75% ethanol and stored at the Animal Ecology Laboratory of Guizhou Normal University (GZNU), Guiyang City, Guizhou, China. All of the molecular samples were stored at −80 °C in a refrigerator.

In total, 10 specimens of new species and K. anophthalmus were analyzed and measured, following Zhao et al. (2024) who measured 27 kinds of morphological data using digital calipers with an accuracy of 0.1 mm. All of the measurements were taken on the left side of each fish specimen. Morphological data were also collected from the literature for K. acridorsalis, K. hyalinus, K. longzhouensis, and K. parvus so that morphological comparisons could be made (Ge et al. 2024).

Given the morphological similarity between the new species and K. anophthalmus, statistical analysis was performed of the morphometric data for both species. In order to reduce the impact of allometry, a size-corrected value from the ratio of each character to standard length was calculated for the following morphometric analyses. Principal component analyses with eigenvalues greater than one, the maximum variance method, and simple bivariate scatter plots were used to explore and characterize the morphometric differences (Xu et al. 2023) between the new species and closely related species. One-way analysis of variance was conducted to determine the significance of differences in morphometric characters between the new species and the aforementioned similar species. All of the statistical analyses were performed using SPSS v.21.0 (SPSS, Inc., Chicago, IL, USA), and differences were considered statistically significant at P < 0.05.

Skeletal X-ray scanning was performed at the Key Laboratory of Vertebrate Evolution and Human Origins, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, using micro-computed tomography (Siemens Somatom Definition X-ray machine). The skull images were exported from a virtual 3D model which was reconstructed using Volume Graphics Studio 3.0 software.

Total genomic DNA was extracted from three samples of three species from 95% ethanol-preserved tissues using the cetyltrimethylammonium bromide method. The process of DNA library construction and detection followed Luo et al. (2023) and DNA was sequenced by TSINGKE Biotechnology Co., Ltd. (Chengdu, Sichuan, China) on an Illumina NovaSeq 6000 platform. A total of 36 Gb of raw bases were generated, with each sample generating approximately 12 Gb of raw data. These sequencing data were completed using MitoFinder v.1.4.2 (Allio et al. 2020) for genome assembly and annotation of the mitogenome using Oreonectes platycephalus as the reference (accession number: NC_031579). All of the sequences have been deposited in GenBank (Table 1).

In this study, three mitochondrial genomes were newly sequenced while 26 mitochondrial genomes and three Cyt b were downloaded from the US National Center for Biotechnology Information. Two datasets were constructed for phylogenetic analysis in this study, where datasets 1 and 2 included the mitogenome and only the Cytochrome b (Cyt b), respectively. Here, the mitogenome was extracted using PhyloSuite v.1.2.3 (Zhang et al. 2020) for 13 protein coding genes, two rRNAs, and 22 tRNAs for subsequent genetic analysis.

Multiple sequence alignment was performed using MAFFT v.7.4 (Katoh and Standley 2013) within PhyloSuite v.1.2.3 (Zhang et al. 2020) and checked using MEGA v.7.0 (Kumar et al. 2016) to rule out possible errors. In addition, Partitionfinder v.2.1.1 (Lanfear et al. 2017) was used to select the best-fit partitioning and nucleotide substitution model for the two datasets, based on the Bayesian information criterion. In dataset 1, each gene fragment was pre-set as an independent partition. Luo et al. (2023) was referred to for selecting species of Homaloptera parclitella as an outgroup (Table 1).

Phylogenetic trees were reconstructed using Bayesian inference (BI) and maximum likelihood (ML) methods based on best-fit partitioning and nucleotide substitution models. Bayesian inference analysis was performed using MrBayes v.3.2.1 (Ronquist et al. 2012). Each BI analysis was run independently using four Markov Chain Monte Carlo chains (three heated chains and one cold chain) starting with a random tree; each chain was run for 2 × 10⁷ generations and sampled every 1000 generations. Convergence of the data runs was confirmed when the average standard deviation of split frequencies was less than 0.01. The ML analysis was performed using IQ-tree v.2.0.4 (Nguyen et al. 2015) based on the best-fit model with 10,000 ultrafast bootstrap (UFB) replicates (Hoang et al. 2018). The ML analysis was performed until a correlation coefficient of at least 0.99 was achieved. Nodes were considered highly supported when the Bayesian posterior probability (BPP) value for the BI analysis was greater than 0.95 and the UFB value for the ML analysis was greater than 95%. Based on dataset 2, genetic distances were calculated using the uncorrected p-distance model and 1000 bootstrap replications, in MEGA v.7.0 (Kumar et al. 2016).

Molecular dating and time tree reconstruction was conducted in BEAST v.2.4.7 (Bouckaert et al. 2014), using dataset 1. Considering that the genus Karstsinnectes has no known fossil record, two calibration nodes were set up with reference to recent studies (Wang et al. 2016; Hirt et al. 2017): (1) the origin of the family Nemacheilidae dates to 45.5 million years ago (Mya) (95% confidence interval (CI): 36.7–55.8 Mya) (Hirt et al. 2017); (2) the divergence of the genera Triplophysa and Barbatula occurred at 23.5 Mya (95% CI: 20.5–26.1 Mya) (Wang et al. 2016). The BEAST analysis used an uncorrelated lognormal relaxation clock and a Yule tree prior. The BEAST analysis was run for 200 million generations and sampled every 1000 generations. All of the calibrations used a normal prior, monophyly, and standard deviation values of 2.2. Convergence of the run parameters was checked using Tracer v.1.7.1 (Rambaut et al. 2018) to ensure that the effective sample size of all of the parameters was greater than 200. Three runs were made in total, and the trees were finally merged using LogCombiner v.2.4.7. A maximum clade credibility tree was generated using TreeAnnotator v.2.4.1 (implemented in BEAST) by applying a burn-in of 25%.

Species of the genus Karstsinnectes are currently known to be distributed mainly in the Pearl River drainage in southern China (Fig. 1), showing a top-down distribution and can be mapped onto phylogenetic topologies (Zhao et al. 2024). To estimate the origin and dispersal of species in the genus Karstsinnectes, the present study conducted biogeographic analyses using the R package BioGeoBEARS (Matzke 2013). Prior to the analysis, six models were estimated using BioGeoBEARS, Dispersal-Extinction-Cladogenesis (DEC), a ML version of Dispersal-Vicariance Analysis (DIVALIKE), Bayesian biogeographical inference model (BAYAREALIKE), each with and without founder-event speciation (+J).

In this study, 29 mitogenomes and three Cyt b sequences were collected for phylogenetic and genetic analysis. The total length of dataset 1 included 15,664 bases, 9377 conserved sites, 6263 variable sites, and 4993 parsim-informative sites. The total length of dataset 2 was 1140 bp, 871 conserved sites, 6269 variable sites, and 197 parsim-informative sites. The predefined 16 partitions of dataset 1 were suggested by Partitionfinder to be divided into five partitions and the corresponding evolutionary models were GTR+I+G, GTR+I+G, TVM+I+G, TVM+I+G, and HKY+I+G (Table 2).

Phylogenetic analyses highly supported the monophyly of Karstsinnectes (BPP/UFB = 1.00/100) and as the sister clade to Guinemachilus, Micronemacheilus, and Oreonectes (Fig. 2, Suppl. material 1). Karstsinnectes contains two clades (BPP/UFB = 1.00/100), where Clade I contains K. acridorsalis and K. cehengensis while Clade II contains K. parva, K. longzhouensis, K. anophthalmus, and a group of four samples Karstsinnectes sp. from Daxin County, Guangxi (Fig. 2). Divergence among species of the genus Karstsinnectes is highly supported (Suppl. material 1).

Figure 2. 

Chronogram and phylogeny of the genus Karstsinnectes. Circles at each node indicate Bayesian posterior probabilities (BPP). Colored squares at the tip of Karstsinnectes branches indicate species distributions, and the three major clades are marked using Roman numerals (I–II). Divergence times for major clades are marked in parentheses with 95% confidence intervals. Arrows show dispersal events. Species photos from Zhu (1989).

Genetic differences within the genus Karstsinnectes were assessed using Cyt b and ranged from 5.8 to 13.1% (Table 3). Genetic differences between samples from Daxin County, Guangxi, and species within the genus Karstsinnectes ranged from 6.9 (vs. K. anophthalmus) to 12.8% (vs. K. acridorsalis), with the smallest genetic distances being greater than the recognized interspecies difference of 5.8% (K. longzhouensis vs. K. parvus) (Table 3). Thus, the genetic evidence supports that the geographic population of Daxin County, Guangxi, should be treated as a new undescribed phylogenetic species.

The time tree in Fig. 2 shows that the genus Karstsinnectes originated at the Oligocene/Miocene boundary, ca. 22.37 Mya (95% CI = 17.17–29.04 Mya), with the most recent common ancestor occurring in the early Miocene, ca. 18.87 Mya (95% CI = 13.81–24.96 Mya). Interspecific divergence occurred from the middle of the Miocene to the early Pliocene (5.06–15.96 Mya), and the divergence of the Daxin population of Guangxi from its sister species K. anophthalmus occurred in the late Miocene, ca. 6.71 Mya (95% CI = 4.16–10.28 Mya). Additionally, the time tree provides the origins of other genera in the family Nemacheilidae. For example, the split between Paranemachilus and Troglonectes occurred ca. 14.39 Mya (95% CI = 10.25–19.27 Mya); the split between Yunnanilus and Eonemachilus occurred ca. 9.89 Mya (95% CI = 6.63–13.80 Mya); Oreonectes originated ca. 20.27 Mya (95% CI = 15.47–26.58 Mya); and the split between Guinemachilus and Micronemacheilus occurred between about 16.62 Mya (95% CI = 11.94–22.52 Mya) (Fig. 2).

The results of model comparisons used for ancestral area reconstruction within BioGeoBEARS are provided in Table 4. Based on the DEC+J model (Fig. 2, Suppl. material 2, 3), dispersal events may have shaped the current distribution of Karstsinnectes. Biogeographical inference based on the best model, DEC+J (AIC weight = 0.53; Table 4), suggests that the most recent common ancestor of extant Karstsinnectes likely inhabited the Hongshui River basin and then dispersed southwestward into the Zuojiang-Yujiang and Beipanjiang river basins (Fig. 2). After colonizing the Zuojiang-Yujiang River basin, the genus dispersed eastward into the Youjiang River basin. The BAYAREALIKE+J model (AIC weight = 0.47; Table 4; Suppl. material 4) was second-best and yielded similar results.

Principal component analysis of the new species and K. anophthalmus using 18 measured characters showed that a total of five principal component (PC) factors were extracted based on eigenvalues greater than one. The first three components explained 69.45% of the total variance, with PC1 accounting for 37.97%, PC2 for 17.75%, and PC3 for 13.74% (Table 5). In the scatterplot of PC1 versus PC2, K. daxinensis sp. nov. and K. anophthalmus formed distinct clusters that were separated on the PC1 axis (Fig. 3). Body depth, caudal peduncle depth, head depth, head width, mouth width, and maxillary barbel length were mainly loading on PC1 (Table 5). Analysis of variance indicated that 40% of the 20 morphometric characters shared by K. daxinensis sp. nov. and K. anophthalmus were significantly different (Table 6). When compared with K. anophthalmus, K. daxinensis sp. nov. has a smaller predorsal length (15.1 ± 2.0 vs. 18.9 ± 2.6), prepelvic length (13.8 ± 1.2 vs. 18.1 ± 2.7), head length (5.4 ± 0.5 vs. 7.3 ± 0.9), head depth (2.1 ± 0.4 vs. 3.1 ± 0.5), head width (3.4 ± 0.7 vs. 4.8 ± 0.8), mouth width (1.7 ± 0.2 vs. 2.7 ± 0.4), but for inrostral barbel length, K. daxinensis sp. nov. is larger than K. anophthalmus (2.2 ± 0.5 vs. 1.3 ± 0.1) (Table 6).

Figure 3. 

Scatter plots of the 1^st and 2^nd principal components for Karstsinnectes daxinensis sp. nov. and K. anophthalmus.