Population Genetic Analysis of the Wood Turtle from Maine to Virginia

In 2015, eight Northeastern States began a cooperative project for Conservation Planning for the Wood Turtle (Glyptemys insculpta) under a Competitive State Wildlife Grant (CSWG). This portion of the study uses genetic data to identify genetic diversity across the study area (Maine to Virginia), identify the number of populations in the study area, and determine the success of genetic assignment of individuals to sites of origin. Tissue samples were collected as blood, tail tips, toenails and shell shavings or scutes from 1,895 Wood Turtles. Most tissue samples were collected in 2015 and 2016; however, some collectors submitted tissue samples from tissue archived from previous collections with the earliest collection dated 2005. Tissue samples were genotyped at 16 microsatellite markers for 1,244 individuals. Genetic data were analyzed for genetic diversity (using HP-RARE, GENEPOP and GENALEX), allele frequency exact test (using GENEPOP), genetic clustering (using STRUCTURE), full siblings (using COLONY), and genetic assignment (using GENECLASS). Samples sizes ranged from 5 to 50 individuals (average n=17.4) collected from 62 sites. Unbiased allelic richness ranged from 3.4 to 6.2 (average 5.1), private alleles ranged from 0 to 0.3 (average 0.05), unbiased expected heterozygosity ranged from 0.5 to 0.7 (average = 0.6) and FIS ranged from -0.21 to 0.14 (average =0). FST ranged from 0 to 0.23 (average 0.07). Allele frequency exact tests identified significant pairwise differences between 91% of the sites. The Bayesian genetic clustering analysis indicated that there are likely 3 to 5 clusters with 4 clusters providing the most optimal clustering pattern in the data set. The major population groups identified were northern ME, Potomac, coastal MA and NJ/NY. Sites in PA and NH showed admixture with the neighboring clusters. The results indicate that clear genetic differences among populations (or subpopulations) are detectable across the study area. The Bayesian clustering analysis indicate that an island stepping-stone model describes the population genetic structure where sites are exchanging individuals with neighboring sites creating a gradation of genetic structure over the study area. Isolation by distance was significant for 2 of 3 clusters tested in Potomac and Maine/NH (p<0.01). The northern Maine cluster showed a similar pattern but was not significant for isolation by distance (p=0.17). Tests for full sibling families indicated a maximum distance between family members of 50 km. Genetic assignments indicated that 52% of individuals in the data set assigned correctly to the collection site. The genetic assignment was moderately successful with some sites providing relatively high (>75%) correct genetic assignment; however, assignment success using these markers varied across the sites/populations and, at some sites, correct assignment was relatively low (<50%) limiting the application of this method for management and enforcement for Wood Turtles confiscated from illegal harvest.