2.1 Sample collection

We collected 66 fin tissue samples from six geographic populations of O. stewartii in the Yarlung Zangbo river, with the populations denoted by the region of the river section: Zhongba (ZB; 29°46′ N, 84°01′ E), Saga (SG; 29°33′ N, 85°23′ E), Nyingchi (LZ; 29°09′ N, 87°63′ E), Xietongmen (XTM; 29°43′ N, 88°26′ E), Xigaze (RKZ; 29°26′ N, 88°88′ E) and Medog (MT; 29°32′ N, 94°33′ E) (Figure 1). The samples from ZB (n = 5), SG (n = 23) and LZ (n = 5) were collected in 2015, those from XTM (n = 10) were collected in 2016, and those from RKZ (n = 7) and MT (n = 16) were collected in 2017. Fin clips from the sampled fish were stored in 75% ethanol at −20°C. We extracted genomic DNA from the fin tissue using a standard proteinase K phenol-chloroform extraction protocol (Javadi et al., 2014).

2.2 Primer screening

We sequenced the transcriptome of O. stewartii using 10 organs (skin, muscle, liver, spleen, kidney, heart, intestine, eyes, brain and blood) and identified approximately 20,000 microsatellites (data unpublished), among which 26 microsatellite markers were chosen to analyse the genetic diversity. The primer sequences are listed in Table 1, and the optimal annealing temperature for each pair of primers was 60°C.

2.3 Polymerase chain reaction amplification and gel-electrophoretic detection

The DNA was amplified by the polymerase chain reaction (PCR) in a total reaction volume of 10 μL containing 5 μL of 2× Dream Taq Green PCR Master Mix (Thermo Scientific, CA, USA), 1 μL of 50 ng/μL genomic DNA, 0.5 μL of 10 μM each of the forward and reverse primers and 3 μL of ddH₂O. The PCR was setup as follows: initial denaturation at 95°C for 3 min followed by 25 cycles of denaturation at 95°C for 30 s, annealing at 60°C for 30 s, extension at 72°C for 30 s and final extension at 72°C for 5 min.

The PCR amplification products were detected using 10% non-denaturing polypropylene gel (acrylamide: bisacrylamide = 29:1) electrophoresis (220 V, 4°C for 2-3 h). After electrophoresis, the gel was silver stained (0.1% silver nitrate), then was rinsed slowly with distilled water and coloured with a colour-developing solution (2% sodium hydroxide, 0.04% anhydrous sodium carbonate, 0.4% formaldehyde). The sliver-stained gel was scanned using Bio-Rad GS-900 scanner (Bio-Rad, Hercules, CA, USA).

2.4 Data analysis

We assigned genotypes for each sample based on analyses of microsatellite fingerprint maps using GEL-PRO version 4.5 (Media Cybernetics, Rockville, MD, USA). We tested for Hardy–Weinberg equilibrium (HWE) and genotypic linkage disequilibrium using GENEPOP version 1.2 (Raymond & Rousset, 1995). We used POPGENE version 1.32 (Yeh, 1997) to analyse genetic variation, including the gene frequency, allele numbers (observed alleles, Ao; effective alleles, Ae), heterozygosity (observed heterozygosity, Ho; effective heterozygosity, He), genetic distance (D) among populations and fixation index (FIS). We conducted Analysis of molecular variance (AMOVA) tests in Arlequin version 3.1 to analyse the degree of genetic differentiation (F_st) and the global F-statistic among populations (Excoffier et al., 1992). An analysis of polymorphism information content (PIC) for each locus was conducted following the method proposed by Botstein et al. (1980). We estimated the ancestral lineage and interbreeding of the populations using STRUCTURE version 2.3.1 (Falush et al., 2003), setting a cluster K value from 1 to 6, run 10 times, with the parameter of 50,000 burn-in iterations and 100,000 Markov chain Monte Carlo (MCMC) length. STRUCTURE HARVESTER was used to determine the optimum K (Earl & von Holdt, 2012). BayesAss version 3.0.4 software was used to estimate the short-term migration rates between populations (Wilson & Rannala, 2003), and the linkage-disequilibrium method was applied to estimate the effective population sizes (Waples & Do, 2008) using the software LDNE version 1.31.

3.1 Genetic diversity

Twenty-six microsatellite markers were successfully amplified from six populations of O. stewartii, with a rate of 0.96 polymorphism; excluding locus OstC010, 25 polymorphic markers were obtained. Among the polymorphic loci, the number of amplified alleles (Ao) ranged from 2 to 9, with an average of 3.2692, and the number of expected alleles (Ae) ranged from 1.0981 to 5.1393, with an average of 2.4052. Observed heterozygosity (Ho) ranged from 0.0156 to 0.9844, expected heterozygosity (He) from 0.0894 to 0.8054 and PIC from 0.0854 to 0.7787, with an average of 0.4210 (Table 2). Among the six wild populations, population LZ displayed the lowest genetic diversity (Table 3), with the lowest values of He and PIC. The highest genetic diversity was observed in population MT. Three populations (RKZ, MT and XTM) showed a significant level of intrapopulation relatedness (inbreeding) (p < 0.05), and population SG showed a highly significant level of inbreeding (p < 0.01). Based on the genetic parameters (Ao, Ae, Ho, He and PIC), the levels of genetic diversity among the populations ranked them as MT > SG > RKZ > XTM > ZB > LZ. The overall results of the analysis indicated a low level of genetic diversity in these six populations.

3.2 Hardy–Weinberg equilibrium

Tests of global HWE (Table 2) showed significant deviation in HWE at 15 of the 25 polymorphic loci; of these, OSTC0038, OSTC0077, OSTC0041 and OSTC0091 showed heterozygote deficiency, whereas OstC007, OstC0066, OstC0072, OstC0046, OstC0075, OstC0078, OstC0080, OstC0084, OstC0050, OstC0059 and OstC0068 showed heterozygosity excess. Significant deviations were detected in four populations (MT, SG, RKZ and XTM) and not in populations ZB and LZ. Population SG showed the highest level of deviation, with deviation at 20% of the total loci; populations MT and XTM showed approximately 12% frequency in site deviation.

3.3 Genetic differentiation of O. stewartii populations

Analysis of the genetic structure with AMOVA revealed that 7.00% of the genetic variance was contributed by among populations (p < 0.001) and 93.00% was contributed by among individuals within populations (Table 4). The pairwise F_st values ranged from −0.00264 to 0.18755 (Table 5), indicating moderate genetic differentiation among the six populations. Apart from population RKZ, there was a significant level of genetic differentiation among the other five populations (p < 0.05).

3.4 Ancestral lineage and phylogenetic analyses

Ancestral lineage analysis using STRUCTURE software showed that the best cluster was two (K = 2; Figure 2), and individuals were clearly divided into two groups (Figure 3), which indicated that these O. stewartii populations comprise two ancestral lineages. The populations RKZ and MT were clustered together (represented by RT), whereas the other four populations (LZ, SG, XTM and ZB) were clustered as another genetic group (represented by SLZX). To investigate whether significant difference occurred between the two lineages, an extra AMOVA analysis was performed, and the result showed the genetic variance between SLZX and RT contributed to 3.07% of the total genetic variance with a significant level (p < 0.0001), which indicated that there was a significant genetic differentiation between SLZX and RT.

Phylogenetic analysis of the six populations (Figure 4) showed that the RKZ and MT populations clustered into one group, while populations LZ, SG, XTM and ZB clustered into another group, which was consistent with the results of the ancestral lineage analysis.

3.5 Effective population estimation and gene-flow analysis

When LDNE was used to estimate effective population size (N_e), it could not be calculated for populations RKZ, LZ and ZB because of small sample sizes. Among the other three populations, SG had the largest effective population with a value of 221.5 while MT had the smallest with a value of 91.9 (Table 6).

We used BayesAss version 3.0.4 to detect recent migration rates among populations. The migration rates between the geographic populations of O. stewartii were found to be small and asymmetrical. The migration rates from MT to RKZ, and from SG to LZ, were highest (Table 7). Interpopulation genetic differentiation correlated with interpopulation migration (Table 4). The results indicated significant genetic differentiation (p < 0.05) among five populations (including MT, LZ, SG, ZB and XTM), and the migration rates between them were the lowest. Notably, there was no significant difference in genetic differentiation between SG and LZ populations due to the high mobility between the two populations. Low genetic differentiation was detected between populations RKZ and LZ, yet the estimated migration rate between them was the highest.

To evaluate whether migrate rates among populations were significant correlated to geographical distances, the correlation analysis between geographical distance and migration rate was conducted, and the result showed that there was no significant correlation between these (p = 0.38).

4.1 Genetic diversity analysis of O. stewartii

The PIC value is a measure of a microsatellite marker’s usefulness to detect polymorphisms. Botstein et al. (1980) proposed a PIC index to evaluate the level of gene variation: when PIC > 0.5, the site shows high polymorphism; 0.25 < PIC < 0.5, the site shows moderate polymorphism; and when PIC < 0.25, the site shows low polymorphism. The genetic diversity analysis showed nine highly polymorphic loci (OstC0075, OstC0050, OstC0078, OstC0066, OstC0084, OstC0080, OstC007, OstC0017 and OstC0059), 11 moderately polymorphic sites (OstC0046, OstC0068, OstC0089, OstC0091, OstC0041, OstC0038, OstC0083, OstC0072, OstC0077, OstC0022 and OstC0065) and six sites of low polymorphism. The average PIC signified moderate polymorphism in the six populations (Table 2). These results revealed that the microsatellite markers used in this study could be used to effectively test the genetic diversity of the O. stewartii populations in the YZRB.

Heterozygosity is a suitable parameter to measure genetic variation in natural populations. The average heterozygosity values approximately reflect the genetic variation among the populations. The level of heterozygosity is directly proportional to the degree of genetic diversity in the population: for instance, high heterozygosity indicates high genetic diversity (Li et al., 2006). Here, the population MT had the highest value of H_o at 0.4375, whereas the lowest value of 0.3154 was observed in population LZ. The average H_o in the six populations ranged from 0.3154 to 0.4375, which was consistent with the study of Takezaki and Nei (1996) who speculated that heterozygosity determined from microsatellites is in the range of 0.3–0.8. Statistical analysis of the heterozygosity data showed moderate genetic diversity in the six populations.

4.2 Population genetic differentiation analysis

Thorpe (1982) suggested that the genetic distance among populations of the same species is between 0.03 and 0.2; consistent with this, the genetic distances of the six populations of O. stewartii in our study were between 0.04318 and 0.16926. Though the fish from the six different geographic locations could be considered the same population based on genetic distance analysis, there were significant differences in the values of F_st (p < 0.001) and pairwise F_st (p < 0.05) among the populations, indicating significant genetic differentiation among the populations. The phylogenetic analysis showed that the six populations were clustered into two groups, which was consistent with the analysis of ancestral lineages (i.e. two ancestral lineages were detected among the six populations). From the combined results of the genetic differentiation, phylogenetic and ancestral lineage analyses, we deduce that the populations of O. stewartii comprise two ancestral lineages: one includes populations MT and RKZ, and the other lineage includes populations SG, LZ, XTM and ZB, which was proved by the result of genetic differentiation between these two lineages from AMOVA. that is significant genetic differentiation (p < 0.0001) occurred between these two linages. The biological characteristics of Schizothorax species, such as slow growth, long lifespan, late sexual maturity, and low fecundity, make their populations fragile and sensitive to environmental changes and human disturbances (Huo et al., 2012; Ma et al., 2012; Zhou et al., 2015). The survey results of Li et al. (2019) showed that the population size and density of O. stewartii presented a negative correlation with the distance from cities, such as Lhasa, Xigaze, Shannan, and Nyingchi and other large cities and regions with high demographic density, and the natural population sizes of fishes declined more severely closer to cities. Overfishing is presumed to be a main factor affecting the resources of species of Schizothorax (also members of the Schizothoracinae), and Li et al. (2019) showed that overfishing likewise has a considerable impact on the O. stewartii resource. Under the combined influence of natural and human factors affecting fishes, the genetic structure of O. stewartii has possibly changed. The genetic structure of O. stewartii might also be related to elevation. The geographical distributions of the populations should also be considered; the average elevation of the Xigaze region (inclusive of populations RKZ, SG, ZB, LZ and XTM) is above 4000 m, thus the five geographic populations in that region might be logically grouped, whereas Medog county (inclusive of MT, the sixth population) has an average elevation of 1200 m and the O. stewartii population there might be separately grouped. This aligns with the results of the genetic structure analysis (excepting population RKZ) and the correlation analysis between migrate rates and geographic distances of six populations, which suggest that the human disturbances and the change of environment rather than the geographical distribution might influence the population genetic structure.

4.3 Effective population size and gene-flow analysis

Although LNDE is a biased estimator when the sample size is very smaller, the results still can reflect a species’ biological characteristics (Beerli & Felsenstein, 2001; Waples, 2006). The estimated effective population sizes of the six populations of O. stewartii in the YZRB were small yet were positively correlated with the population resources. A decrease in the whole population will lead to a decreased effective population size (Shrimpton & Heath, 2003). In recent years, some researchers have proved that the population of O. stewartii in the YZRB is gradually shrinking, which would correspondingly lead to a smaller effective population size. The effective population size is an important concern in conservation biology as is a main driving force of population evolution (Leberg, 1992; Li et al., 2019; Wang et al., 2004). Effective population size is closely related to indices of genetic diversity because small populations tend to be vulnerable to genetic drift, which causes a loss of genes and gradual losses of population genetic diversity, followed by less ability to respond to environmental changes. Therefore, the population of O. stewartii in the YZRB must be strengthened in its size to protect the species from ongoing human disturbances, and this would also help to avoid inbreeding and maintain the genetic characteristics of the population.

The analysis of inbreeding relatedness (hybridization) with STRUCTURE software showed that all populations were genetically mixed, which indicates gene flow among the populations. Although Palstra et al. (2007) believed that the BayesAss calculation of gene flow had some deviation, it could reflect the actual characteristics of gene flow among populations. The results show that larger populations were the main source of migration, and the gene flow from larger to smaller populations indicates that these populations are in urgent need of protection (Hansen et al., 2007; Kuang et al., 2010). Gene flow plays an indispensable role in maintaining the stability of population genetic structure. For small populations, gene flow can increase population genetic variation, reduce the pressure to inbreed, and enhance a population’s ability to adapt to environmental changes. However, population size is a key factor for species protection (Lande, 1988; Palstra et al., 2007), and reduced population resources will cause obstacles to gene exchange, thus it is vital to protect the O. stewartii populations and so enhance their gene exchange.