Phylogenetic assessment of endangered and look‐alike Pigtoe species in a freshwater mussel diversity hotspot

Abstract The Green River in Kentucky in the eastern United States is a freshwater mussel biodiversity hotspot, with 71 known species. Among them, the endangered Pleurobema plenum coexists with other morphologically similar species in the genera Fusconaia and Pleurobema, known colloquially as “pigtoes.” Identification of species in these genera is challenging even for mussel experts familiar with them. In our study, the correct identification of these species by experts ranged from 57% to 83%. We delineated taxonomic boundaries among seven species and tested for cryptic biodiversity among these look‐alike mussels utilizing mitochondrial and nuclear DNA sequence variation. Phylogenetic analysis of combined (1215 bp) mitochondrial DNA cytochrome oxidase I (COI) and NADH dehydrogenase 1 (ND1) genes showed five well‐diverged groups that included F. flava, F. subrotunda, P. cordatum, and P. plenum as distinct clades, with P. sintoxia and P. rubrum grouped into a single clade. While our mitochondrial DNA analyses did not distinguish P. sintoxia and P. rubrum as phylogenetically distinct species, the typical shell forms of these two nominal taxa are very distinct. Further phylogenetic analysis using nuclear ribosomal transcribed spacer region subunit I (ITS1) DNA sequences also showed that P. sintoxia and P. rubrum were not distinct lineages. No cryptic species were detected in the Fusconaia and Pleurobema samples analyzed from the Green River. The highest haplotype diversity (h), average number of nucleotide differences (k), and nucleotide diversity (π) were observed for F. subrotunda at both the COI (h = 0.896, k = 3.805, π = 0.00808) and ND1 (h = 0.984, k = 6.595, π = 0.00886) markers, with similarly high genetic diversity in the other taxa. Our results give managers confidence that cryptic taxa do not occur within or among these morphologically similar species in the Green River, and populations appear genetically diverse, indicative of large and healthy populations.


| INTRODUC TI ON
Freshwater mussels of the Family Unionidae comprise 679 species, with a global hotspot of diversity in North America and other areas of high diversity in South America and southeast Asia (Lopes-Lima et al., 2018). In particular, the southeastern United States is an area of high species richness, endemism, and imperilment (Elkins et al., 2019). Understanding of the systematics of unionids mussels is still in flux (Graf & Cummings, 2007), and cryptic, species-level variation is still becoming recognized; for example, Schilling (2015) found evidence of several cryptic species in the genus Pleurobema and Pleuronia in the upper Tennessee River basin. Hence, it is important to investigate and recognize species limits in order to inform conservation actions. The so-called pigtoe mussels (Tribe Pleurobemini; genera Fusconaia, Pleurobema, and Pleuronaia), which are broadly distributed in river systems in central and eastern North America (Campbell & Lydeard, 2012b;Heard & Guckert, 1970), are such a group of freshwater mussels (Campbell & Lydeard, 2012aInoue et al., 2018;Morrison et al., 2021;Schilling, 2015), and those of the Ohio River basin are yet to receive sufficient phylogenetic characterization.
The Green River, Kentucky (KY) has one of the most diverse assemblages of mussels in the United States, including seven species of Fusconaia and Pleurobema. All seven pigtoe species are of conservation management concern (Master et al., 1998), with IUCN Red List status as: critically endangered for rough pigtoe (P. plenum; Bogan, 1996a) and Clubshell (P. clava; Bogan, 1996b), vulnerable for long-solid (F. subrotunda; Cummings & Cordeiro, 2012a), nearthreatened for Ohio pigtoe (P. cordatum; Bogan & Seddon, 1996) and pink pigtoe (P. rubrum; Bogan, 1996c), and least-concern for Wabash pigtoe (F. flava; Cummings & Cordeiro, 2011) and round pigtoe (P. sintoxia; Cummings & Cordeiro, 2012b). Complicating their management, the shell phenotypes of these species are very similar ( Figure 1). Notably, five mussel experts, familiar specifically with pigtoes, identified several pigtoes individuals collected from the Green River. Individuals of F. flava were the easiest to identify by the experts (Figure 2). Fusconaia subrotunda had a high rate of misidentification and species in the genus Pleurobema were the most challenging to identify and frequently confused among each other. Such results show that pigtoes in the Green River are particularly difficult to distinguish morphologically even by experts.
Previous studies describing these species focused on a suite of shell and soft-body morphological characters. For example, some of the principal differences among species of Fusconaia and Pleurobema are the number of gills charged when gravid (four for Fusconaia and two for Pleurobema), conglutinate morphology (leaflike for Pleurobema and subcylindrical for Fusconaia), and foot color (white for Pleurobema and generally orange for Fusconaia). However, these shell and soft-body characters can overlap, depending on environmental and genetic factors, mussel age and size, and phenotypic plasticity (Inoue et al., 2014), and are difficult to characterize on living individuals or are not expressed in all seasons. In addition, similar phenotypes could be the result of all these Pleurobemini species being closely related phylogenetically, resulting in these species sharing characters.
Recent advances in development of molecular markers and the widening application of phylogenetic analyses have led to more reliable identification of freshwater mussels and to development of more scientifically defensible management plans.
Phylogenetic analysis of freshwater mussels in North America has focused increasingly on application of DNA sequence variation, including variation at the mitochondrial DNA (mtDNA) 16S rRNA, cytochrome oxidase subunit I (COI), cytochrome b (Cyt-b), and NADH dehydrogenase 1 (ND1) genes, as well as at nuclear genes, including large ribosomal subunit 28S rDNA and the ribosomal internal transcribed spacer region subunit 1 (ITS1) (Campbell et al., 2008;Campbell & Lydeard, 2012a;Graf & Cummings, 2007;Jones et al., 2015). Several recent studies have characterized phylogenetic relationships among species in the genera Fusconaia and Pleurobema using these markers. For example, F. flava, F. cerina, and F. askewi were shown to be the same phylogenetic species based on COI and ND1 haplotypes (Campbell & Lydeard, 2012a).
Further, species in the genus Fusconaia tend to show low intrapopulation (F. subrotunda) and low interpopulation (F. flava/cerina) variation, i.e., mtDNA divergence is low both within and among species in this genus (Campbell & Lydeard, 2012a). For mussel species in the Green River, phylogenetic relationships among F. flava, F. subrotunda, P. cordatum, P. clava, P. plenum, P. sintoxia, and P. rubrum were recently assessed using CO1, ND1, and ITS1 markers (Campbell et al., 2008;Campbell & Lydeard, 2012b;Inoue, 2018;Jones et al., 2015). Early studies suggested the existence of a P. cordatum group which included P. cordatum, P. plenum, P. rubrum, and P. sintoxia (Campbell et al., 2008;Campbell & Lydeard, 2012b), and more extensive sampling by Inoue et al. (2018) andJones et al. (2015) validated the interpretation that P. cordatum and P. plenum are indeed different species. Utilizing ND1 and COI markers, these authors showed that only a few nucleotide differences exist between individuals of P. sintoxia and P. rubrum and found that further phylogenetic assessment was needed to delineate these two nominal taxa. While the shells of P. sintoxia generally are morphologically distinctive, they can occasionally be mistaken for P. rubrum and vice-versa ( Figure 3). However, the umbos of P. rubrum are pointed and pronounced, resembling a pyramid, hence the common name "Pyramid pigtoe." Further, this species typically has a well-defined sulcus traversing the middle of each valve, especially in larger and older mussel specimens (Miller et al., 2008). Another relevant example of closely related species is P. clava and P. oviforme, species endemic to the Tennessee and Cumberland River watersheds, for which Campbell et al. (2008) showed few molecular differences at mtDNA genes between these taxa, although when assessed at nuclear ITS1, differences were observed. More recently, Morrison et al. (2021) conducted a rang-wide assessment of mitochondrial (mtDNA) and nuclear microsatellite DNA and showed minimal mtDNA genetic divergence and even some haplotype sharing over wide geographic areas between the two taxa throughout the Ohio River basin, but very high divergence at microsatellite markers and distinctive morphology for a population occurring in the extreme headwaters of the upper Tennessee River basin. Therefore, utilization of nuclear as well as mitochondrial DNA sequences is critical for assessing phylogenetic differentiation among closely related species.
Phylogenetic relationships among species in the genera Fusconaia and Pleurobema have been assessed across various geographic regions in North America using a suite of molecular markers. While these comparisons have been made with a large number of specimens from various species belonging to the Tribe Pleurobemini (Campbell et al., 2008;Campbell & Lydeard, 2012aGraf & Cummings, 2007;Jones et al., 2015), molecular data are sparse for these species in the Green River, Kentucky.
Rigorous phylogenetic assessment of morphologically similar species in these two genera would be vital to support development of a probabilistic dichotomous key for the Green River and the regional Ohio River mussel faunas. Further, an in-depth phylogenetic identification of mussels in the genera Fusconaia and Pleurobema from the Green River utilizing large sample sizes, would help determine whether any cryptic species occur in the river, and assist with the design and implementation of appropriate management plans for imperiled and critically endangered species, especially for P. plenum.

| Sample collection
A total of 258 mussel specimens belonging to species in the gen-

| Polymerase chain reaction
We extracted DNA using an Isohelix (Harrietsham) DNA Isolation Kit.
Concentration and purity of the double-stranded DNA were measured using a μLite PC spectrophotometer (Biodrop), and DNA was diluted to 10-30 ng/μl. All polymerase chair reactions (PCRs) were performed in either a T100™ or MyCycler™ thermocycler (both from Bio-Rad). PCR products were sent to the Fralin Life Sciences Institute (Blacksburg, VA) for Sanger sequencing. For ND1, we used two different pairs of primers to obtain amplified sequences for all species (Table 1). Amplification products were obtained for most mussel specimens of F. flava, F. subrotunda, P. cordatum, and P. plenum using primers Leu-uurF and LoGlyR . For some mussel specimens (later identified as P. sintoxia and P. rubrum), it was necessary to use primers nadh1-F and nadh1-R (Buhay et al., 2002;. Details regarding PCR amplification appear in Supplmental Material S1. Two slightly different forward primers were used to amplify COI sequences for all species (Table 1). Sequences for COI were obtained for mussel specimens of F. flava, F. subrotunda, P. cordatum, and P. rubrum and P. sintoxia using the primers LCO1490 (Folmer et al., 1994) and HCO700dy2 (Walker et al., 2006). In the case of P. plenum, most sequences were obtained by using primers COIF (Campbell et al., 2005) and HCO700dy2. Details regarding PCR amplification appear in Supplemental Material S1.
We amplified the nuclear ribosomal internal transcribed spacer region subunit 1 (ITS1) sequence for mussel specimens molecularly identified as P. rubrum and P. sintoxia (with mitochondrial COI and/or ND1 markers) from the Green River (40 mussel specimens). Sequences used as outgroups included P. sintoxia or P. rubrum from the Clinch River (four sequences) and Tennessee River (three sequences). Finally, two mussel specimens of F. flava, three F. subrotunda, two P. cordatum, and three P. plenum were sequenced and used as outgroups. Details regarding PCR amplification appear in Supplemental Material S1.

F I G U R E 2 Bar graph showing percentages of correctly identified and misidentified individuals for each species of Fusconaia and
Pleurobema by five experts. On average, the experts were able to correctly identify the mussels 70% of the time. Mussel specimens were collected in 2015 from Pool 4 (GPS coordinates = 37.18286, −86.6296; river mile = 149) in the Green River, Kentucky.
(GeneStudio, Inc.). DNA sequence variation metrics-such as polymorphic nucleotide sites, number of haplotypes, nucleotide diversity, and haplotype diversity-were calculated using DnaSP 5.10 (Rozas et al., 2009). Pairwise difference values within and between species were estimated using p-distances, and the most likely model of nucleotide substitution was identified using MEGA6 (Tamura et al., 2013).
For construction of phylogenetic trees and networks, the most appropriate model of nucleotide substitution was selected using MrModeltest 2 (Nylander, 2008), which works in an interface with PAUP 4.0 (Swofford, 1998 (Rambaut, 2014). To analyze the MCMC runs resulting from MrBayes, we used Tracer v 1.6.0 (Rambaut et al., 2009). In this software, the effective sample size (ESS) was >200 for all the trees.
Phylogenetic trees were constructed for each mitochondrial marker COI and ND1, using sequences from all the mussel specimens collected, and an additional tree was constructed using all markers combined. In addition, a phylogenetic tree for P. sintoxia and P. rubrum was constructed for the nuclear marker ITS1. These analyses incorporated four MCMC chains with trees sampled every 1000 F I G U R E 3 Pie chart showing experts' field identification for mussel specimens that where morphologically identified to the clade Pleurobema sintoxia/rubrum. Two representative shells for (a) Pleurobema rubrum and (b) Pleurobema sintoxia shell forms. These mussel specimens were identified consistently as these two shell forms by all the experts. Mussel specimens were collected in 2015 from Pool 4 (GPS coordinates = 37.18286, −86.6296; river mile = 149) in the Green River, Kentucky.
generations ND1 and combined COI + ND1 and every 250 generations for COI. Finally, the ITS1 trees were sampled every 100 generations. Species differentiation was assessed using the Automatic Barcode Gap Discovery (ABGD) method (Puillandre et al., 2012) using ND1, COI, and combined COI + ND1 sequences. For COI, ND1, and combined COI + ND1 sequences, species delimitation was assessed by using ABGD. To assign mussel specimens to species, the Kimura (1980) two-parameter (K2P) distance model was used, where the minimum intraspecific genetic distance (P min ) was set to 0.001 and the maximum intraspecific genetic distance (P max ) was set to 0.1.

| Amplification of molecular markers
We were able to PCR-amplify COI using one reverse primer (HCO700dy2), but we needed two different forward primers (Table 1). The first forward primer, LCO1490 (Folmer et al., 1994), developed for a borad range of invertebrates, amplified COI sequences for individuals belonging to F. flava, F. subrotunda, P. cordatum, and P. sintoxia/rubrum. To amplify COI sequences for individuals subsequently identified as P. plenum, we used forward primer COIF (Campbell et al., 2005) that was developed for species in the tribe Pleurobemini (Burlakova et al., 2012;Campbell et al., 2005;Campbell & Lydeard, 2012a. We used two pairs of primers to amplify a 744-bp region of ND1. The first primer set, Leu-uurF and LoGlyR , which amplified sequences for individuals of F. flava, F. subrotunda, P. cordatum, and P. plenum, has been used successfully for a wide range of freshwater mussels (Schilling, 2015;Smith et al., 2018). However, we needed additional primers to amplify ND1 sequences of P. sintoxia and P. rubrum, nadh1-F, and nadh1-F (Buhay et al., 2002;, which have been used in other studies of mussels of Tribe Pleurobemini (Burlakova et al., 2012;Campbell & Lydeard, 2012aCampbell et al., 2008).
Nuclear ITS1 sequences were amplified using primers 18S and 5.8S (King et al., 1999), which worked for most individuals of P. sintoxia/rubrum and for outgroups F. flava, F. subrotunda, P. cordatum, and P. plenum. A problem observed when amplifying these sequences was gaps among aligned sequences, which many times were due to an artifact of sequence quality. To ensure the quality of the sequences, we re-sequenced those samples that presented extra bases and those not of high quality (<80% GC). Like Schilling (2015), we did not encounter length differences among sequences from the same mussel specimens. Hence, length differences were not quantified, as in Schilling (2015). The ITS1 sequences used in this study included only one ITS1 sequence for all individuals analyzed.

| Genetic diversity
For F. flava, F. subrotunda, and P. cordatum observed haplotype and nucleotide diversities were higher for the ND1 DNA sequences compared to the COI DNA sequences (  (Olivera-Hyde, 2021), was consistent with that of Inoue et al. (2018).

| Phylogenetic analysis
In addition, haplotypes for each species collected from the Green River consistently grouped into the same species clades as haplotypes from specimens collected in the Clinch and Tennessee rivers, including one F. subrotunda (collected from the Clinch River), 14 P. plenum (12 from the Clinch River and two from the Tennessee River), and seven P. sintoxia and P. rubrum (four from the Clinch River and three from the Tennessee River). The additional sequences of Arkansas. In addition, our mussel specimens of P. sintoxia and P. rubrum grouped together in the same clade with those of the same species reported by Inoue et al. (2018); this is particularly interesting, as the authors added sequences from several locations where P. rubrum and P. sintoxia occur. The P. sintoxia/rubrum clade was paraphyletic with Pleurobema riddelli, which seems to be closely related.
In the ND1 tree, sequences from our specimens grouped together with those from Bertram (2015) The phylogenetic tree constructed from combined COI and ND1 sequences resulted in five well-defined clades that included F. flava, F. subrotunda, P. cordatum, P. plenum, and P. sintoxia/rubrum ( Figure 5).
As in the separate trees for COI and ND1, the tree of combined sequences showed little evidence to support these clades as different species for P. rubrum and P. sintoxia. There are four individuals in the P. sintoxia/rubrum clade that were separated in two additional clades when COI and ND1 were combined and species delimitation was assessed. Two of these individuals shared the ITS1_PSR03 (tag # WE779 from Pool 4, and tag # BLU009 from the Clinch River) sequence. The other two individuals, one from Pool 4 (tag # WG591) and another from the Clinch River (tag#RubClinch) have ITS1_ PSR04 and ITS1_PSR01, respectively. Only one individual was identified as P. rubrum by the five mussel identification experts.

| Split networks
Topologies of the split networks resulting from analysis of COI, ND1, and COI + ND1 sequences were consistent and showed five distinct clades, F. flava, F. subrotunda, P. cordatum, P. plenum, and P. sintoxia/rubrum. In the COI split networks (Figure 6a), we added sequences of P. sintoxia and P. rubrum from Inoue et al. (2018), which clustered within our P. sintoxia/rubrum clade. The same results were observed when ND1 sequences of P. sintoxia and P. rubrum from Jones et al. (2015) were included (Figure 6b). Split networks constructed from ITS1 sequences included both P. sintoxia and P. rubrum, as well as outgroups from other species, such as F. flava, F. subrotunda, P. cordatum, and P. plenum. For both ITS1 alignments, one of the haplotypes of P. sintoxia/rubrum (ITS1_Psr05) seemed to be particularly distinct from the other haplotypes in this clade. However, specimens from the same outgroup species fell into different clades, making inference of species-level differentiation using ITS1 sequences unreliable (Figure 7).

| Pairwise differentiation
In the case of the mitochondrial COI gene, pairwise differentiation values using the Tamura −86.1154; river mile = 197) in the Green River, Kentucky, and additional mussel specimens were collected from the Clinch River (CL) and Tennessee River (TN).
Intraspecific pairwise p-distances for mussel specimens of P. cordatum ranged between 0.4% for COI and 0.5% for ND1 (Table 3). These results were similar to those reported by Jones et al. (2015), in which ND1 intraspecific distances for P. cordatum ranged between 0.4% and 0.7%. Results for interspecific pairwise distances showed that the highest differences were between P. cordatum and P. sintoxia/rubrum, with 6.9% for COI and 9.8% for ND1. The second-most differentiated species from P. cordatum was P. plenum, with 6.2% p-distance for COI and 7.8% for ND1.
Intraspecific variation among mussel specimens of P. plenum ranged between 0.4% and 0.6% for COI and ND1 sequences, respectively. These results were similar to those reported by Jones et al. (2015), who reported ND1 intraspecific pairwise differences between 0.6% (Green River, Kentucky) and 0.8% (Tennessee River, Tennessee) for P. plenum. In this study, intraspecific pairwise differences were highest between P. plenum and P. sintoxia/rubrum, which were between 8.3% and 8.8% for COI and ND1 sequences, respectively. Finally, intraspecific pairwise differences for P. sintoxia/rubrum were 0.4% for both COI and ND1 sequences, comparable to those of Jones et al. (2015), who reported ND1 intraspecific pairwise differences of 0.1% for P. sintoxia and 0.8% for P. rubrum.

| DISCUSS ION
Most mussel species show substantial differences in shell morphol-

| Molecular markers
The concept of using DNA sequences to "barcode" species relies on intraspecific variation being clearly lower than interspecific variation for mitochondrial (COI, ND1) and nuclear markers (ITS1). The most-used marker for DNA barcoding in eukaryotes is the mitochondrial COI gene (Bleidorn, 2017). The principal practical reasons for use of mitochondrial markers for barcoding are the availability of "universal" PCR primers, larger numbers of mtDNA copies per cell relative to nuclear DNA, and high interspecific variation that gives rise to so-called barcoding "gaps" (Puillandre et al., 2012). However, the use of only mitochondrial markers in phylogenetic studies has been criticized due to their solely maternal mode of inheritance, inconsistent mutation rate, limited power to detect introgression, low effective population size, low information content among closely related species, and heteroplasmy or pseudogenization (Bleidorn, 2017). In this study, for COI and ND1 primers, target DNA from some specimens of P. plenum and P. sintoxia/ rubrum may have amplified or not with one primer-pair combination due to DNA sequence variation at the primer-binding site.
The principal limitations for the ITS1 marker were low nucleaotide variability and too few fixed nucleotide mutational states to distinguish some of the study species. Species delimitation using AGBD was not possible for ITS1 sequences. The ITS1 marker split the sequences into too many groups, erratically mixing species that were well defined with the mitochondrial markers; e.g., over-splitting occurred for sequences of F. flava and P. cordatum, as well as P. sintoxia/ rubrum. Over-splitting could be due to higher intraspecific variation  (2015), who observed that some estimates for interspecific variation were lower than those for intraspecific variation.

| Phylogenetic assessment
Delineation of species in the genera Fusconaia and Pleurobema is supported by a suite of morphological and life-history traits (Olivera-Hyde, 2021). In the Green River, a principal morphological difference which generally is reliable is foot color. Mussels in the genus Pleurobema typically have a white foot, whereas mussels in the genus

| Pyramid and round pigtoes
In this study, we collected a relatively large number of individuals identified morphologically as P. sintoxia and P. rubrum from the Green River, KY to enhance the probability of delineating these two nominal taxa and for detecting any cryptic species which could have small populations and prove similar in appearance to these two species. By using a large sample size, any intraspecific and interspecific nucleotide differences are more detectable and better characterized, enabling identification of any species-level differences among taxa. Further, we added DNA sequences of P. sintoxia and P. rubrum from other studies, to include COI (Inoue et al., 2018) and ND1 (Jones et al., 2015), and all these sequences grouped F I G U R E 6 Split phylogenetic network using mitochondrial DNA COI gene (a) and ND1 gene sequences (b). For both Split networks, distances were calculated using GTR with a gamma rates model and a proportion of invariable sites estimated with splits Tree4 (Huson & Bryant, 2006 Bertram et al. 2015;Burlakova et al., 2012;Jones et al., 2015;Marshall et al., 2018;Schilling, 2015 with associated accession numbers available in Table S1. The outgroups are Pleuronaia dolabelloides COI sequence (GenBank accession number: MF962140) and ND1 sequence (GenBank accession number: KT118034).

F I G U R E 7
Phylogenetic trees constructed using ITS1 sequences with clustal alignment and Bayesian consensus trees that were constructed using MrBayes. The most appropriate model of nucleotide substitution using the Akaike information criterion (AIC) was the symmetrical model (SYM + G) with gamma rates. The analysis was run with 400,000 generations, and trees were sampled every 100 generations, which generated a total of 1785 trees. The final standard deviation of split frequencies was lower than 0.01 with a −ln likelihood of −982.06. The outgroups that were collected from the Green River and included F. flava (ITS1_Ffla), F. subrotunda (ITS1_Fsub), P. cordatum, (ITS1_Pcor), and P. plenum (ITS1_Pple  Figure 1). For P. rubrum, the shell shape is typically a scalene triangle with beaks facing forward and with a very marked sulcus that traverses the shell from near the umbo to the ventral margin, whereas for P. sintoxia, individuals are much more rounded in shape with beaks facing each other, and without a well-defined sulcus. Separation of these two putative species seems to be supported by morphological differences in their glochidia (Culp et al., 2009). Thus, it is important to investigate quantitatively whether glochidial differences exist between mussels expressing the P. rubrum and P. sintoxia shell forms.

| Genetic diversity
Haplotype diversities in our study were high, suggesting that this high contemporary genetic diversity could be due to these species historically occurring in much larger, interconnected populations that were linked to those in the mainstem of the Ohio River and TA B L E 3 Estimates of evolutionary divergence based on analysis of mitochondrial DNA COI and ND1 sequence pairs between and within species of Fusconaia and Pleurobema using p-distances (lower diagonal) and the Tamura-Nei with invariable sites model of nucleotide substitution (upper diagonal). Recruitment failure in these populations would be catastrophic, as the extirpation of these populations from the Green River would have serious consequences for the long-term conservation of these species. Fortunately, there is evidence of recruitment for all five investigated species in the Green River, which is one of the best refuge strongholds for these and many other species in the Ohio River system (Haag & Cicerello, 2016). Periodic monitoring to assess the abundance, recruitment, and genetic diversity of the Green River mussel fauna will be critical for managing the viabilty of these species.

| Management implications
The IUCN Red List status for F. flava is "least concern," although this species showed the lowest haplotype and nucleotide diversity among our study species. The effective population size (N e ) has not been estimated for this species, mainly due to a lack of PCR primers for DNA microsatellites specifically designed for this species or even for a closely related Fusconaia species. The Fusconaia subrotunda clade was well supported phylogenetically and was the clade with the highest nucleotide and haplotype diversities in the study.
Principal concerns regarding management of these species include continued demographic declines as large-to medium-sized freeflowing riverine habitats are lost (Haag & Cicerello, 2016). Future efforts are needed to develop nuclear DNA genetic markers to estimate N e and quantify genetic diversity of this species in the Green River.
Pleurobema cordatum numbers have declined range-wide, likely due to the reduction of large river habitats. However, P. cordatum seems more tolerant of impoundments than F. subrotunda and other Pleurobema species (Haag & Cicerello, 2016). This species showed higher nucleotide diversity (π) and smaller haplotype diversities (h) than other P. cordatum populations reported for the Green and the Tennessee Rivers (Jones et al., 2015).
The federally protected Pleurobema plenum has been listed under the U.S. Endangered Species Act as endangered since 1976 and its recovery plan was approved in 1984 (U.S. Fish andWildlife Service, 1976, 1984). Pleurobema plenum is not very tolerant of impoundments and has been extirpated from most of its historical range. Because P. plenum is sensitive to habitat modification, its critical habitat (medium-to large-sized rivers) must be protected.
However, high haplotype diversity suggests that the P. plenum population is healthy and reasonably abundant in the Green River, KY.
Recruitment and abundance of this species needs to be regularly monitored to ensure these values are not indicative of an aging, potentially nonrecruiting population.
While both P. sintoxia and P. rubrum appear to belong to only one phylogenetic clade based primarily on mtDNA, these two nominal taxa should be treated as separate species until additional morphological and nuclear DNA marker-based studies have been completed.
Similar to most of the species collected from the Green River, specimens belonging to the P. sintoxia/rubrum clade seem to be marginally tolerant to even intolerant of impounded riverine conditions (Haag & Cicerello, 2016). Ongoing studies assessing morphological differentiation between these two putative species are being conducted by Dr. Monte McGregor at the Center for Mollusk Conservation of the Kentucky Wildlife Resources Agency, who is currently assessing glochidial morphological differences.
Finally, in contrast to Schilling (2015), who performed a similar molecular marker-and morphology-based study of mussels in the Tennessee River basin and found evidence of several cryptic species in the genus Pleurobema and Pleuronia, we did not find cryptic species in the Green River. An additional species that has been reported for the Green River, the endangered clubshell (P. clava) (Haag & Cicerello, 2016), still occurs in the river upstream of the sampling sites, but was not found during the field collections in Pool 4 and MCNP. This species was reported from the Green River (Kentucky) in Hart and Taylor counties by Watters (1994).
However, he did not find live mussels, but only fresh-dead shells.
Thus, future studies are needed to monitor the recruitment, abundance, and genetic diversity of this species there to determine population status.

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
GenBank accession numbers for the haplotype sequences are listed in Table S1.