Structure, Demographics, and Phenotype
Results from DAPC and sNMF are highly concordant in estimating three
major lineages (Fig. 3). These correspond primarily to populations in
the Piedmont (‘CHR’), Cohutta Mountains (‘COH’) in north-central
Georgia, and Nantahala mountains in southwestern North Carolina and
adjacent north-east Georgia (‘NTA’). In contrast, conStruct estimates
two distinct lineages (CHR and NTA+COH) when accounting for spatial
distance, with the second layer accounting for ~80% of
allelic covariance (Fig. S7). Estimates accounting for admixture in sNMF
and conStruct suggest widespread introgression at the contact zones
between these lineages (montane/Piedmont in the latter), with
substantial genomic ancestry (>10%) from each lineage
inferred for multiple individuals from multiple populations across most
of the species’ range.
The GADMA estimates converged on a model with relatively small ancestral
population sizes (~240k) originating in the
mid-Pleistocene (~1.2Ma), with recent exponential
increases (~3–19x) during the latest glacial cycles
~630Ka–present (Fig. 3). Migration was very high
between the two ancestral lineages (estimated at the upper bound of 2Nm
= 20) and decreased towards the present; remaining >1 for
NTA<–>COH (1.24 and 1.85),
CHR–>NTA (1.40), and CHR–>COH (1.70).
Populations are not phenotypically differentiated based on the 17 linear
morphometric measurements taken here (Fig. 2). Specimens from all 3
lineages overlap in linear-discriminant space, where the
heaviest-loading size-corrected variable of trunk length (AG;
axilla-groin) does not differ significantly between them. Overall, these
results are congruent with the hypothesis of a single deeply structured
species, wherein the constituent phylogeographic lineages are not
phenotypically differentiated and maintain migration sufficient to
overcome drift, local adaptation, and divergent ecological selection.