Genetic and morphological characterization of the freshwater mussel clubshell species complex ( and ) to inform conservation planning.

The shell morphologies of the freshwater mussel species (federally endangered) and (species of concern) are similar, causing considerable taxonomic confusion between the two species over the last 100 years. While was historically widespread throughout the Ohio River basin and tributaries to the lower Laurentian Great Lakes, was confined to the Tennessee and the upper Cumberland River basins. We used two mitochondrial DNA (mtDNA) genes, 13 novel nuclear DNA microsatellite markers, and shell morphometrics to help resolve this taxonomic confusion. Evidence for a single species was apparent in phylogenetic analyses of each mtDNA gene, revealing monophyletic relationships with minimal differentiation and shared haplotypes. Analyses of microsatellites showed significant genetic structuring, with four main genetic clusters detected, respectively, in the upper Ohio River basin, the lower Ohio River and Great Lakes, and upper Tennessee River basin, and a fourth genetic cluster, which included geographically intermediate populations in the Ohio and Tennessee river basins. While principal components analysis (PCA) of morphometric variables (i.e., length, height, width, and weight) showed significant differences in shell shape, only 3% of the variance in shell shape was explained by nominal species. Using Linear Discriminant and Random Forest (RF) analyses, correct classification rates for the two species' shell forms were 65.5% and 83.2%, respectively. Random Forest classification rates for some populations were higher; for example, for North Fork Holston (HOLS), it was >90%. While nuclear DNA and shell morphology indicate that the HOLS population is strongly differentiated, perhaps indicative of cryptic biodiversity, we consider the presence of a single widespread species the most likely biological scenario for many of the investigated populations based on our mtDNA dataset. However, additional sampling of populations at nuclear loci is needed to corroborate this finding.