Complex introgression among three diverged largemouth bass lineages

Abstract Hybrid zones between diverged lineages offer a unique opportunity to study evolutionary processes related to speciation. Natural and anthropogenic hybridization in the black basses (Micropterus spp.) is well documented, including an extensive intergrade zone between the widespread northern Largemouth Bass (M. salmoides) and the Florida Bass (M. floridanus). Phenotypic surveys have identified an estuarine population of Largemouth Bass (M. salmoides) in the Mobile‐Tensaw Delta, with larger relative weight and smaller adult size compared to inland populations, suggesting a potential third lineage of largemouth bass. To determine the evolutionary relationships among these Mobile Delta bass populations, M. salmoides and M. floridanus, putative pure and intergrade populations of all three groups were sampled across the eastern United States. Phylogenetic analyses of 8582 nuclear SNPs derived from genotype‐by‐sequencing and the ND2 mitochondrial gene determined that Delta bass populations stem from a recently diverged lineage of Largemouth Bass. Using a novel quantitative pipeline, a panel of 73 diagnostic SNPs was developed for the three lineages, evaluated for accuracy, and then used to screen 881 samples from 52 sites for genetic integrity and hybridization on the Agena MassARRAY platform. These results strongly support a redrawing of native ranges for both the intergrade zone and M. floridanus, which has significant implications for current fisheries management. Furthermore, Delta bass ancestry was shown to contribute significantly to the previously described intergrade zone between northern Largemouth Bass and Florida Bass, suggesting a more complex pattern of secondary contact and introgression among these diverged Micropterus lineages.


| INTRODUC TI ON
Hybrid zones are geographical areas of ongoing hybridization between diverged lineages and therefore offer a unique opportunity to study evolutionary processes related to speciation (Barton & Hewitt, 1985;Gompert et al., 2017). The most accepted process by which hybrid zones arise is secondary contact of diverged lineages that have not evolved complete reproductive isolation. This contact may occur due to natural shifts in geological or environmental barriers, or anthropogenic introductions of non-native species (Largiadèr, 2007;Woodruff, 1973). Nearly all studied hybrid zones are those involving two parental taxa; however, more complex hybrid zones involving multiple taxa are known to occur in nature (Chhatre et al., 2018;Keck & Near, 2009;Nevado et al., 2011;Peñaloza-Ramírez et al., 2010). When hybridization is thought to be the result of anthropogenic translocations of non-native species, it can be of particular concern due to loss of biodiversity (McDonald et al., 2008;Rhymer & Simberloff, 1996;Viard et al., 2020).
Interspecific hybridization is relatively common among freshwater fishes. Of over 150 described pairs of hybridizing species in the United States, 20% are found within Centrarchidae, including the black basses (Micropterus spp.; Bolnick, 2009). Natural and anthropogenic hybridization in black basses is well documented, using either morphometrics (Bailey & Hubbs, 1949;Baker et al., 2013) or diagnostic genetic markers (Bagley et al., 2011;Bangs et al., 2018;Barthel et al., 2010;Li et al., 2015;Lutz-Carrillo et al., 2006;Philipp et al., 1983;Thongda et al., 2020). One of the most notable cases of hybridization in Micropterus is an extensive intergrade zone between the widespread northern Largemouth Bass (M. salmoides) and the Florida Bass (M. floridanus), 1 of which pure populations are thought to be restricted to peninsular Florida. This hybrid zone was initially described using morphology and meristics and was thought to extend through northern Florida, Georgia, and small parts of Alabama and South Carolina (Bailey & Hubbs, 1949). Thirty years later, Philipp et al. (1983) evaluated allele frequencies at two diagnostic allozymes which extended the hybrid zone west to Mississippi and north to Virginia. Their result is unsurprising, for disconnects between morphological and genetic descriptions of hybrid zones are common, as phenotype can often be inaccurate for identifying later-generation hybrids (Lutz-Carrillo et al., 2018;Meyer et al., 2017;Taylor et al., 2018). While Philipp et al. (1983) acknowledged the inaccuracy of meristics for detecting hybrids, they also suggested that the apparent hybrid zone expansion was due to stockings of Florida Bass outside their native range for recreational fishing. This latter view has persisted, with a recent review portraying the geographic extent of natural hybridization the same as was first described by Bailey and Hubbs (Taylor et al., 2019). Modern genetic tools are required to accurately characterize the hybrid zone of M. salmoides-M. floridanus and determine whether this zone is primarily shaped by stocking or historical hybridization.
In addition to the high diversity of Micropterus species, the southeastern United States has the greatest aquatic diversity in North America (Warren et al., 1997), which is exemplified by the numerous freshwater and brackish water fish species supported by the Mobile-Tensaw River Delta in Alabama (Swift et al., 1986;Swingle & Bland, 1974;Swingle et al., 1966). With the confluence of the Tombigbee and Alabama rivers at its northernmost point and Mobile Bay to the south, the Mobile Delta is the second largest delta in the contiguous United States. Due to a rich history of geologic shifts, glaciation, and sea-level fluctuations, freshwater fishes of the southeastern United States have experienced repeated bouts of habitat isolation and reconnection (Swift et al., 1986). These dramatic changes in recent evolutionary history are likely responsible for the region's high endemicity and hybridization observed in numerous freshwater fish, such as mosquitofishes (Wilk & Horth, 2016), crappies (Travnichek et al., 1996), sunfish (Avise & Saunders, 1984), and bass (Near et al., 2003). Such sea-level fluctuations in the Late Pliocene are thought to have isolated M. floridanus from M. salmoides on the Florida peninsula (Near & Kim, 2021), resulting in genetically and phenotypically diverged sister taxa.
In the Mobile Delta exists another phenotypically distinct population of largemouth bass. These "Delta" bass have a smaller adult size and higher relative weight compared to other M. salmoides populations, as well as physiological differences that may facilitate their survival in water with elevated salinity (up to 13 ppt;DeVries et al., 2015;Glover et al., 2012Glover et al., , 2013Tucker, 1985). Preliminary genetic analysis using isozymes and microsatellites showed that Delta bass were genetically similar to M. salmoides, but were unable to conclusively determine whether they were a distinct genetic lineage (DeVries et al., 2015;Hallerman et al., 1986

| Sample collection and GBS library preparation
Sample collection and genotype-by-sequencing (GBS) using the PstI restriction enzyme was previously described in Thongda et al. (2020) (Li et al., 2015;Zhao et al., 2018). We also sampled an additional 786 largemouth bass individuals across 52 populations for extensive population genetic structure analysis and hybrid classification using the diagnostic SNP panel developed in this study (Figure 1; also refer to File S2). For the 881 total samples genotyped with the diagnostic SNP panel (including some of the samples used for GBS), DNA was extracted from fin clips using a simple sodium hydroxide (NaOH) and hydrochloric acid (HCl) tissue digestion (Truett et al., 2000). For mitochondrial gene sequencing, DNA was extracted from fin clips or whole blood samples using E.Z.N.A. DNA Kits (Omega Bio-Tek).
GBS reads were trimmed and demultiplexed using the process_radtags program in STACKS, while removing reads with an uncalled base (-c) and low-quality scores (-q). For reference-based SNP calling, we mapped the demultiplexed reads to the M. floridanus genome using the BWA-MEM algorithm from BWA (Burrows-Wheeler Aligner) v0.7.17 with default settings ). The mapped reads were sorted using the "sort" function in SAMTOOLS v1.6  and then used to call SNPs with the "gstacks" pipeline and default SNP model ("marukilow"), which uses a Bayesian approach to identify SNPs across all samples for each GBS locus and then genotypes each individual at each SNP. The alpha thresholds for SNP discovery and calling genotypes were left at the defaults of 0.05, and the maximum soft-clipping level was increased to 0.3 (-max-clipped 0.3). The "populations" program was used to merge loci that were produced from the same restriction enzyme cut sites (-merge-sites), perform initial SNP filtering, and export a variant call format (VCF) file. Stringent additional SNP filtering was done with VCFtools v0.1.17 (Danecek et al., 2011), custom Python code, and code adapted from Jon Puritz's laboratory (Puritz et al., 2014) to remove potentially erroneous genotypes due to low coverage, sequencing error, or paralogs.
For population genetic analyses, biallelic SNPs were retained if they had a minor allele frequency greater than 0.05, had a minor allele count greater than 4, were genotyped in at least three population groups (NLB, DLB, FLB, or ILB), and had <15% missing data across all individuals. SNPs that departed Hardy-Weinberg equilibrium with a p-value cutoff of 0.05 in two or more sampling sites were also removed. SNPs in linkage disequilibrium (r 2 > 0.2) or within the same 1000-bp window were pruned using BCFtools +prune plugin

. For phylogenetic analyses that included outgroup
Micropterus species, variable sites were retained if they were genotyped in at least 4 species and 70% of all individuals, were heterozygous in fewer than 70% of samples (-max-obs-het), and had a minor allele count greater than 3. The VCF output of stacks was converted to a phylip file using vcf2phylip v2.0 (Ortiz, 2019).

| ND2 sequencing
DNA from 58 samples was amplified with previously published primers targeting the mitochondrial NADH subunit 2 (ND2) gene (Kocher et al., 1995). Amplification of the ND2 gene used 25 ml reaction volumes and included 1× Kapa HiFi HotStart ReadyMix (2×; Kapa Biosystems), 0.2 mM forward primer, 0.2 mM reverse primer, and 2 μl of template DNA. All PCRs included negative control reactions (no DNA template). Thermal cycler reactions were conducted with a heated lid and the following conditions: initial denaturation at 95°C for 3 min, followed by 5 cycles at 94°C for 30 s, annealing at 57°C for 30 s, and 72°C for 75 s; an additional 5 cycles of annealing at 56°C for 30 s and 5 cycles with annealing at 55°C for 30 s; followed by 20 cycles at 94°C for 30 s, annealing at 54°C for 30 s and 72°C for 75 s, and a final extension at 72°C for 10 min, followed by a final hold at 4°C (Baker et al., 2013). PCR products were visualized on a 1% agarose gel to check the amplification success and that a single band was observed. Amplicons were then cleaned using ExoSAP-IT (Thermo Fisher Scientific). Quantity and quality of PCR products were determined with a NanoDrop Spectrophotometer (Thermo Fisher Scientific).
Sanger cycle sequencing of all products was performed by Genewiz, Inc. using the PCR primers and internal ND2 primer MET-F (Kocher et al., 1995). Consensus sequences of ND2 were obtained by analyzing the forward sequencing trace files using CodonCode Aligner (2019.2.1) and then trimmed to the first 598 bp. The reads were inspected manually for quality and for any discrepancies in base calls. All sequences were aligned in MEGA-X using MUSCLE to determine the number of unique haplotypes present for subsequent analyses (Edgar, 2004;Kumar et al., 2018). A minimum spanning network for haplotypes was calculated and visualized in the R package pegas v0.14 (Paradis, 2010).

| Phylogenetics and population genetics
Maximum likelihood phylogenetic analyses of 100 individuals from pure NLB, FLB, and DLB populations and 17 outgroup Micropterus samples were conducted using 18,492 concatenated variable GBS sites with IQ-TREE v1.6.12 (Nguyen et al., 2015). ModelFinder as implemented in IQ-TREE used BIC and determined the best-fit substitution model that included ascertainment bias correction to be TVM + F + ASC + R3 (Kalyaanamoorthy et al., 2017). Branch supports were assessed with both 1000 ultrafast bootstrap approximation replicates and 1000 bootstrap replicates for the SH-like approximate likelihood ratio test (Guindon et al., 2010;Hoang et al., 2018). The resulting tree was rooted with the five outgroup bass species. The same analysis was performed on 18,641 SNPs with ILB individuals included. The resulting best trees were plotted with the R package ggtree (Yu et al., 2017).
Population genetic summary statistics were calculated on GBS SNPs for population groups (NLB, DLB, FLB, ILB) and individual sampling sites with at least 5 individuals. Based on equations from Nei and Chesser (1983), observed heterozygosity (H o ), expected heterozygosity (H e ), overall F ST , and F IS were calculated using the basic.stats function in the R package hierfstat v0.5.7 (Goudet & Jombart, 2015). Confidence intervals for population-specific F IS were determined using the boot.ppfis function in hierfstat with 1000 bootstrap replicates. Pairwise F ST following Weir and F I G U R E 1 Sampling sites of largemouth bass assayed with 73 SNP panel. Populations are labeled as in Table S2 and represented by pie graphs showing the mean estimated ancestry proportions for three bass lineages based on 73 diagnostic SNPs and the population genetic program STRUCTURE (K = 3). Delta Largemouth Bass (DLB) are pink, northern Largemouth Bass (NLB) are blue, and Florida Bass (FLB) are green. Native ranges of NLB and FLB are shown, as well as the NLB-FLB hybrid zone as described by Bailey and Hubbs (1949). Map data from ESRI; FLB and NLB native ranges from Taylor et al. (2019) Cockerham (1984 was calculated using the genet.dist function in hierfstat (Weir & Cockerham, 1984). Variable sites were determined to be "fixed" between groups if one group had an allele frequency ≥0.98 in one group and <0.02 in the other (based on allele frequencies reported by STACKS). This cutoff was chosen to allow for singleton alleles observed in a lineage that may be due to genotyping error.
The model-based Bayesian clustering method STRUCTURE v2.2.4 (Pickrell & Pritchard, 2012) was used to determine the number of distinct genetic clusters (K), using the admixture model with correlated allele frequencies, no prior location information, and a burn-in period of 50,000 repetitions followed by 200,000 repetitions. Five replicate analyses were performed on the thinned GBS SNP dataset with values of K = 1-10. Replicates were summarized and visualized using the CLUMPAK server (Kopelman et al., 2015).
The ∆K method implemented in STRUCTURE HARVESTER was used to determine an optimal K (Earl & vonHoldt, 2012). Principal component analysis (PCA) was implemented in the R package adegenet v2.1.1 (Jombart & Ahmed, 2011) and visualized with the R package PCAviz v0.3.29 (Novembre et al., 2018). Missing data were filled by randomly drawing an allele based on the overall allele frequency across all individuals in one of the prespecified groups (NLB, DLB, FLB, ILB) using custom R code. As other methods support significant differentiation among these groups, this approach is more appropriate than filling in missing data using the mean or overall allele frequency.

| Panel development, testing, and genotyping
Our goal when developing a SNP assay for accurate identification of largemouth bass lineages and hybrids was to find a subset of SNPs that best matched the ancestry results determined by STRUCTURE when using the full GBS SNP dataset. We first used custom Python scripts and the populations.sumstats.tsv output from Stacks to identify 2809 SNPs that were diagnostic between DLB, NLB, and FLB, with the criteria that they had >90% or <10% allele frequency in at least two of the three largemouth bass lineages. PCA was then performed on these diagnostic SNPs for 100 "pure" largemouth bass samples (excluding intergrades) as previously described. To identify SNPs that had the greatest contributions to the PCA and therefore were the most diagnostic, we extracted 549 SNPs in the 90% quantile of loadings for PC1 and PC2. We then randomly sampled 80-

| Genotype simulation and assignment
To evaluate the accuracy of our ancestry assignment using the 73 SNP panel and STRUCTURE, we employed a simulation approach developed by Vähä and Primmer (2006). Individuals used as the pure reference populations in our STRUCTURE runs (14 FLM, 24 DLB, 24 NLB) were used to simulate genotypes from random mating and hybridization with the hybridize R function in adegenet. One hundred genotypes were simulated for each pure lineage, 50 for each type of F1, 150 for each type of F2 through F4 generations assuming neutral admixture, and 300 triple hybrids (considered the product of crossing an F1 from a pair of species with a pure individual of the third species or a backcross between a triple hybrid and a pure species). These simulated genotypes were analyzed using STRUCTURE with the same settings and reference genotypes. The performance of STRUCTURE for hybrid and pure individuals was evaluated based on efficiency (number of individuals in a simulated group that are correctly assigned), accuracy (proportion of an identified group that truly belongs to that group), and performance (efficiency multiplied by accuracy). Finally, the optimal threshold values of Q to assign individuals to the different genotypic categories were determined.

| Geographical patterns of Florida Bass introgression
To determine whether patterns of FLB genetic ancestry could be were designated as either "River" or "Reservoir," where reservoirs were natural or artificial bodies of water with known stocking history (Alford & Jackson, 2009;Bunch et al., 2017;Hargrove et al., 2019). Reservoirs or lakes that had no known history of largemouth bass stocking (e.g., Lake Mattoon, IL) were coded as "River." Ancestry proportions were those calculated using the 73 SNP diagnostic panel described below. This model was compared with a model based only on distance [glm(FLB ancestry ~ distance, family = "quasibinomial")], by comparing the reduction in deviance with a chi-squared test.

| Population genetics
A PCA of 144 individuals using 8582 SNPs clearly differentiated the three lineages across PCs 1 and 2, with PC1 representing 30% of SNP variation and PC2 representing 11% variation (Figure 2). We  Table 2). There were 532 unlinked SNPs that were fixed between FLB-NLB but not between DLB-FLB, suggesting possible introgression between DLB-FLB.
When comparing summary stats between sampling sites, Sugar Lake, MN, had the lowest observed heterozygosity, followed by FLB samples from a captive hatchery population (Table S1). Pairwise F ST between sites within a lineage group was generally low (mean pairwise F ST 0.051-0.124), with the highest within-lineage F ST observed between Sugar Lake, MN, and Hatchery NLB (0.239; Table 1).

| Phylogeny and inferred ancestry
Maximum likelihood phylogenetic inference using 18,492 GBS SNPs supported DLB as a distinct lineage that is sister to NLB, with 100% support from both the SH-like approximate likelihood ratio test (SH-aLRT) and ultrafast bootstrapping (Figure 3). When samples from intergrade populations were included, SH-aLRT support for DLB as a unique lineage was 99.9% and ultrafast bootstrap support was 96% ( Figure S1). For the first 598 bp of the ND2 mitochondrial gene, there were two Delta haplotype, ten FLB haplotypes, and ten NLB haplotypes (Figure 4). There were two fixed ND2 SNPs proportions of at least 6% in two or more lineages, were distributed across both the GBS phylogenetic tree and across mitochondrial haplotypes ( Figure S1 and Figure 4). Individuals from some sampling sites, including Sugar Lake, MN, and Demopolis Lake, AL, sequenced well on the MassARRAY platform but did not amplify or sequence well with available ND2 primers (Kocher et al., 1995), suggesting a polymorphism in the ND2 primer recognition sites. Some intergrade individuals appeared to be heterozygous at ND2 sites that were diagnostic between lineages, which may be a sign of heteroplasmy (Piganeau et al., 2004) or sequencing of homologous regions in the nuclear genome (nuMTs; Simone et al., 2011). This included two individuals from West Point Reservoir, GA, one from Lake Eufaula, AL, one from Hatchie River, TN, and one from Sutton Lake, NC. All of these sites have known recent stocking history.  Table 3).

| Panel development
The genetic assignment of simulated pure individuals from the three lineages showed high levels of performance (98%-100%). For hybrid classes involving two lineages and fewer than four generations of backcrossing, performance values ranged from 99% to 93%. While individuals were classified as NLB-FLB hybrids with 100% accuracy, 6.5% of the simulated NLB-FLB hybrids were misclassified as triple hybrids, showing a small observed DLB Q-value (<0.1). Triple hybrids had >90% performance. When fourth-generation backcrossed hybrids (Bx4) were included, efficiency decreased to 82.8%-92.4% for hybrids and accuracy decreased to 58%-78.7% for pure species, as many Bx4 hybrids were assigned as pure (Table S2). while intergrade populations between FLB and NLB were also found outside of the intergrade zone as described by Bailey and Hubbs (1949). Twenty-six intergrade populations with >6% mean ancestry from all three lineages were identified, primarily in Alabama, Georgia, Tennessee, and Louisiana. Compared to previous estimates of FLB-NLB ancestry using a 35 SNP panel, this 73 SNP panel appears to more accurately represent FLB ancestry in sites with high DLB ancestry. For example, while the 35 SNP panel found that populations from the Mobile Delta were 27%-30% FLB and 70%-73% NLB, the 73 SNP panel identifies them as 95%-98% DLB ( Figure S3, File S2).

| Geographical patterns of ancestry and hybridization
To evaluate the relative roles of geography and stocking history for determining FLB genetic ancestry, we implemented fractional logistic regression models based on the distance of a site from our reference FLB population and the stocking history of a site (coded as either River or Reservoir). The proportion of FLB ancestry in an individual was best described by a model that included stocking: FLB ancestry ~ distance * stocking (chi-squared test of reduction in deviance, p-value = 7.43e-12), where separate geographic clines were fit for River and Reservoir sites (Figure 6). Based on values predicted by  (Nei & Chesser, 1983). Zhao et al., 2020). Our results contribute to a growing scientific consensus around the validity of M. floridanus as a distinct species (Kassler et al., 2002;Near et al., 2003;Near & Kim, 2021  perch (Stepien et al., 2015), suggesting that the timing of population divergence in this geographic locale can vary across fish taxa.

TA B L E 2 Genetic differentiation of largemouth bass lineages
Genome-wide genetic diversity was higher in pure DLB populations compared to pure populations of FLB or NLB (Table 1 and   Table S1). As we only had two to four wild populations of each lineage, this may reflect our limited geographical sampling. This result contrasts with walleye and perch diversity, where their Mobile Delta populations had lower genetic diversity than those in the north (Stepien et al., 2015). Elevated genetic diversity is encouraging, as it suggests that the DLB populations have not experienced dramatic population declines and bottlenecks (Nei et al., 1975). Historical or recent introgression with FLB is one potential explanation for higher genetic diversity in DLB populations. Genetic divergence between DLB-FLB is lower than between NLB-FLB (  (Armstrong et al., 2000;Rainer, 2010). Sugar Lake, MN, had the lowest genetic diversity out of all sampled sites, which may be due to a recent genetic bottleneck and range expansion after the last glacial maximum (Dyke & Prest, 1987). Sugar Lake is also the only NLB population with >99% mean inferred NLB ancestry based on STRUCTURE. We hypothesize that the low levels of DLB ancestry (<5%) observed in other "pure" NLB populations, such as those in Illinois, are likely due to ancestral polymorphisms and incomplete lineage sorting. The potential genetic bottleneck experienced by Sugar Lake may have removed some of these shared variants between DLB and NLB from the population.

| Multilineage hybrid zone
The hybrid zone between NLB and FLB was first described by Bailey and Hubbs in 1949 using meristics and found to extend only through Georgia and a small part of Alabama (Bailey & Hubbs, 1949).
Although subsequent studies using larger sample sizes and biochemical or genetic assays have determined the geographic extent of hybridization to be much larger, these results are often discounted as merely the product of M. floridanus recreational stocking outside of Florida. Using our diagnostic 73 SNP panel, we assayed 52 populations across the east and southeast, representing both water bodies with known stocking history and unaltered sites (Figure 1) (Keck & Near, 2009;Littrell et al., 2006;McDonald et al., 2008;Nevado et al., 2011).
Whether the result of natural introgression, anthropogenic introductions, or a combination of the two, the large hybrid zone observed between these three diverged lineages provides an excellent opportunity to study the speciation process (Barton & Hewitt, 1985;Gompert et al., 2017). When hybrid zones act as semipermeable barriers to gene flow between parental species, patterns of introgression can vary across the genome (Harrison & Larson, 2014). Differential introgression among loci can be due to stochastic processes (i.e., drift) or selection associated with genotype-byenvironment interactions or reproductive isolation (Fitzpatrick et al., 2009;Gompert et al., 2012;Taylor et al., 2015). With additional genomic sequencing and a careful geographic sampling design aided by the diagnostic panel produced in this study, largemouth bass can be developed into a new case study for genomic patterns of introgression when multiple lineages are involved.

| SNP assay for hybridization and species identification
Characterizing genomic contributions of hybrids is critical for understanding hybrid zone dynamics and informing fisheries management (Fitzpatrick et al., 2015). Empirical and simulation-based studies have determined that at least 50 ancestry-informative genetic markers may be required for accurately classifying F2 hybrids and advanced-generation backcrossed individuals (Fitzpatrick, 2012;Malde et al., 2017). Previous work has resulted in panels of 18 microsatellites or 25-38 SNPs for assessing integrity and hybridization between M. salmoides/M. floridanus; however, these panels did not include samples from the Mobile-Tensaw Delta during development and therefore may not accurately describe patterns of introgression (Li et al., 2015;Seyoum et al., 2013;Zhao et al., 2018

| Management implications for largemouth bass
As highly popular sportfish species, management of NLB and FLB involves balancing angler-desirable traits (e.g., size, catchability), stable fisheries, and biodiversity concerns (Young et al., 2006). The phenotypic and demographic impacts of anthropogenic hybridization between the two species have been studied extensively, with some finding negative fitness consequences due to outbreeding depression (Cooke et al., 2001;Cooke & Philipp, 2006;Philipp & Claussen, 1995) and others seeing a potential to enhance managed fisheries (Maceina & Murphy, 1992;Maceina et al., 1988). Many of these early studies failed to take genetic ancestry into account or evaluated hybrids between geographically disparate sites, making it difficult to extrapolate results to naturally occurring intergrades.
The discovery of a genetically distinct lineage of Largemouth Bass that may differ in fishery-relevant traits (e.g., smaller size, slower growth rate, and environmental tolerances) further strengthens the need for accurate genetic tools to determine purity and hybridization. To better inform stocking success, future performance assays should assess genetic ancestry and leverage natural intergrades to provide a complete picture of fitness variation and environmental adaptation across largemouth bass lineages.
Current diagnostic markers in largemouth bass include a set of 35 SNPs designed for the MassARRAY Sequenom (Li et al., 2015;Zhao et al., 2018) and a panel of 18 microsatellites (Barthel et al., 2010;Seyoum et al., 2013). While we have not evaluated our new panel against these microsatellites, they likely face the same issues as the 35 SNP panel for mischaracterizing ancestry. Given the numerous populations with Delta bass ancestry observed in the southeast, genetic markers used by state agencies and evolutionary biologists need to be updated to better characterize the geographic extent of pure and hybrid populations of all three lineages. Resources.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.  Kassler et al. (2002) and Near et al. (2003), although the American Fisheries Society's Names of Fishes Committee continues to use the subspecies designation Florida Largemouth Bass (Micropterus salmoides floridanus) (Page et al., 2013).