Global diversity and genetic landscape of natural populations and hatchery stocks of largemouth bass micropterus salmoides across American and Asian regions

Although largemouth bass Micropterus salmoides has shown its extremely economic, ecological, and aquacultural significances throughout the North American and Asian continents, systematic evaluation of genetic variation and structure of wild and cultured populations of the species is yet to be documented. In this study, we investigated the genetic structure of M. salmoides from 20 wild populations and five cultured stocks across the United States and China using eight microsatellite loci, which are standard genetic markers for population genetic analysis. Our major findings are as follows: (1) the result of Fst showed largemouth bass had high genetic differentiation, and the gene flow indicated the genetic exchange among wild populations is difficult; (2) AMOVA showed that 14.05% of the variation was among populations, and 85.95% of the variation was within populations; (3) The majority of largemouth bass populations had a significant heterozygosity excess, which is likely to indicate a previous population bottleneck; (4) Allelic richness was lower among cultured populations than among wild populations; (5) Effective population size in hatcheries could promote high levels of genetic variation among individuals and minimize loss of genetic diversity; China’s largemouth bass originated from northern largemouth bass of USA. The information provides valuable basis for development of appropriate conservation policies for fisheries and aquaculture genetic breeding programs in largemouth bass.

population structure. The UPGMA tree built from the matrix of pairwise allele-sharing distance among 25 populations (Fig. 1) and the assignment tests revealed the multilocus microsatellite genotypes to discriminate populations of largemouth bass. The phylogeny based on eight microsatellites revealed a clear distinction between northern and southern populations, although samples from 25 populations were different.
The software STRUCTURE was used to examine how the sampled populations clustered based on the genetic data. The preliminary STRUCTURE run indicated that the most likely number of clusters was two; thus, the second and more robust simulation was run with k = 26. However, in the latter analysis, the results suggested that the most likely number of clusters was four (Fig. 2). population bottlenecks. The majority of largemouth bass populations had a significant heterozygosity excess, which is likely to indicate a previous population bottleneck. However, under the SMM and TPM models, there is no bottleneck signature; under the IAM model, PLWI (P = 0.01) and LKTX (P = 0.008) suggested that there had been a recent population bottleneck. 13 Table 4) are indicated along the X-axis. Each vertical line represents one individual, and Y-coordinate denotes each individual's percentage assignment to each of these seven genetic clusters.

Discussions
This study is the first to use microsatellite DNA markers to examine the genetic variation within and among wild and hatchery populations of largemouth bass across America and China. Other freshwater fish species using microsatellites revealed thenumber of alleles per locus (5.4) 39 was lower than we observed (18.1), which may have been due to the geographic we sampled, the number of populations we sampled, or the loci we selected. Some loci failed to be amplified in cultured populations, which might be due to high sensitivity of these loci under environmental pressure and selective breeding. The highest values of the mean number of alleles and private alleles were detected in the RBRMS population. In the past, hatchery procedures often had overlooked genetic differences between populations, as well as the consequences of reducing heterogeneity through inbreeding 16 . In the present study, the value of observed heterozygosities per population was detected from 0.0213 to 1. The expected heterozygosities were from 0.1158 to 0.9905. Thus, the populations we collected can provide a wide range of selection to avoid reducing heterogeneity resulting from inbreeding. The inspection of genetic information provides fishery managers with a means for understanding and predicting stocking population variation for regulations and other management programs in largemouth bass fisheries. Genetic diversity is increasingly declining as a result of various factors, including direct human impact and structural alteration of ecosystems resultingfrom changes in human life styles, bottleneck experienced, inbreeding, etc 40 . Genetic diversity values are a better indicator of the genetic polymorphism within a population, because estimates of the mean number of alleles are influenced by sample size. In the present study, the majority of largemouth bass populations had a significant heterozygosity excess, which is likely to indicate a previous population bottleneck. However, comparing Southern and Northern populations of largemouth bass, the South Carolina populations showed higher values of gene diversity and allele richness. The populations have maintained their genetic integrity and been only minimally affected. Nevertheless, the high genetic diversity of invasive species may be caused by hybridization and variation. In this study, the results showed that the largemouth bass had high genetic differentiation. In addition, according to AMOVA analysis, overall genetic differentiation among largemouth bass from the 25 populations was significant (P < 0.05), suggesting significant genetic differentiation among localities or populations. However, overall, the genetic variations within populations are higher than those among populations. In the natural environment, the long-term success of stocked largemouth bass might be the greatest for cohorts with the highest levels of genetic variability 16 . Microsatellite allelic frequencies and Fst analyses showed that all 25 populations were genetically unique. These genetic differences have resulted from long-term exposure to natural selective pressures and provide thespecies with enough variation to deal with the Barge range of different environmental conditions in which it exists 41 .  www.nature.com/scientificreports www.nature.com/scientificreports/ The largemouth bass has two subspecies, the Florida largemouth bass M. s. floridanus and the northern largemouth bass M. s. salmoides. They occupy their native habitats from peninsular Florida to throughout northeastern Mexico, southeastern Canada, and the U.S. corridor in between 15 . The genetic difference is particularly pronounced for the two recognized subspecies, but pertains to populations within each subspecies as well. The UPGMA tree built from the matrix of pairwise allele-sharing distance among 25 populations across America and the assignment tests revealed the multilocus microsatellite genotypes to discriminate populations of largemouth bass. The phylogeny based on eight microsatellites revealed a clear distinction between northern and southern populations. The China population and Minneapolis populations were clustered in a group, indicating that China's largemouth bass originated fromnorthern largemouth bass, since China introduced largemouth bassonly one time from California, USA before our sampling time.
In conclusion, there is relatively high genetic diversity of wild largemouth bass populations in North America, and significant genetic differentiation among localities or populations in the largemouth bass, but obviously lower genetic diversity in hatchery stocks. The study suggested that China's largemouth bass originated from northern largemouth bass of USA. This study provided valuable information for the fishery management and conservation implication of largemouth bass, and indicated that the differentiation in largemouth bass was caused by the founder effect. Besides the founder effect, geographicaldistance might also be responsible for the significant genetic differentiation among wild populations. These analyses clearly reveal substantial genetic differences among populations in the United States. This study demonstrates the need for incorporating genetic information and principles into current and future fisheries management programs. However, the effectiveness of the stocking populations should be evaluated and the long-term effects on the genetic composition of specified largemouth bass populations should be monitored.

Methods populations studied.
A total of 1,045 individuals were collected by trapping and/or electro-fishing from 25 different localities and coordinated by geographic sampling coordinates throughout North America and China ( Fig. 1 and Supplementary Table S1). Twenty wild populations and five hatchery populations were analyzed. Details of the samples are given in Table 4, and the approximate geographical location of the sampled populations is indicated in Fig. 3.
Fin clips were the sources of nuclear DNA. They were collected from the freshly caught fish and immediately preserved in 95% ethanol.
Extraction of DNA and microsatellite amplification. Total DNA was extracted from an optimum volume of fin tissue following a modified version of the pure gene protocol for extraction from fish tissue 37 . Amplification of microsatellite loci were performed using eight primers selected from those isolated and  Table 4.
www.nature.com/scientificreports www.nature.com/scientificreports/ characterized by Lutz-Carrillo (Table 5) 16,37,38 . Each locus was amplified with a three-primer system in which only the M13 and CAG primers were fluorescently labeled with FAM, HEX, or NED.
Polymerase chain reaction of 6 μL contains 3 μL of JumpStart RedMix (Sigma), 1.5 pmol of both non-tailed and labelled primers, and 0.1 pmol of the tailed primer, 25 ng DNA, in the presence of 100 μm spermidine. Amplification was be conducted in PTC-100 thermal cyclers (MJ Research) using an initial denaturation at 94 °C for 2 min, followed by 35 cycles of 30 s denaturation at 94 °C, 30 s annealing at a locus-specific temperature (Table 5), 30 s extension at 72 °C, and a final 5-min extension at 72 °C. Amplification products were separated using an ABI 3130 Prism DNA genetic analyzer.

Data analyses. Analyses of genetic variation and genetic differentiation.
Deviations from Hardy-Weinberg equilibrium (HWE) and linkage disequilibrium (LD) were being tested for all locus-population and locus-locus combinations using ARLEQUIN version 2.0 42 . Significance levels were be modulated for multiple comparisons using the sequential Bonferroni method 43 . Genetic variation was evaluated according to observed heterozygosity and the number of alleles per locus within populations and regions. Private alleles among populations were computed using GenAlEx 6 44 .
Analyses of molecular variance. Analyses of molecular variance (AMOVA) were conducted with Arlequin. One AMOVA was performed under the null hypothesis of no genetic structure. AMOVAS will examine population structure after combining populations in various ways to test for geographical structure.
Assignment tests. Genetic structure among populations was analyzed based on a chord distance according to Cavalli-Sforza and Edwards 45 . A UPGMAtree was constructed using the programs NEIGHBOR and CONSENSE in PHYLIP version 3.6 in terms of Felsenstein 46 . Genetic structure among individuals was assessed using the model-based Bayesian clustering method according to Pritchard 47 and Lutz-Carrillo 37 . Assignment tests were performed with geneclass version 2.0 48 . The program STRUCTURE was used to identify genetic clusters without using any prior information of the sampling location of the individuals 47 . Two separate analyses were carried out in STRCTURE, both times with allele frequencies as correlated and the admixture model. The first analysis was used to infer the most likely range for K. Here, we used a burn-in of 10000 and a MCMC length of 5000 iterations and the simulated number of populations from K = 1-25. The upper limit of 25 was chosen as this corresponds to the number of sampled populations. Twelve independent simulations were performed of each K to check for consistency across runs. The preliminary results were assessed using the Evanno method where the most likely K was determined by the distribution of △K.
Bottlenecks analyses. Allelic richness was calculated with FSTAT version 2.9.3.2 49 . All these tests were adjusted for multiple simultaneous comparisons using a sequential Bonferroni correction. Means are reported as ±SE. We used bottleneck version 1.2.02 50 to test for evidence consistent with recent population bottlenecks or expansions by recognizing significant heterozygosity excess or deficiency for each population using a Wilcoxon signrank test. We conducted the tests using 1,000 iterations using the SMM of microsatellite evolution, as well as a two-phased mutation model (TPM). Allele frequency data was tested for evidence of a "heterozygosity excess" (HE) using the program bottleneck 48,50 . Three statistical tests (sign test, standardized differences test, and Wilcoxen signed-ranks test) were conducted in order to determine whether there is significant HE, which may indicate that a recent bottleneck has occurred.
Ethics. This study and all experimental procedures involving animals were performed according to the protocol approved by the Ohio State University Institutional Animal Care and Use Committee. Permissions for field studies were granted by:  Table 5. Characteristics of largemouth bass microsatellite markers.