3.3 Population structure
A comparison of BIC values from the genus-level DAPC analyses suggested a total of five genetic clusters, with two populations in T. blandingii and three in T. pulverulenta , matching the clades identified in the phylogenetic analyses (Fig. S3). Our admixture-based method, LEA, identified two distinct genetic clusters at the genus level, corresponding to the two Toxicodryas species, and the same two populations for T. blandingii and three populations forT. pulverulenta as suggested by DAPC (Fig. 4). A low amount of admixture was identified in the Cameroonian sample of T. blandingii , and varying levels of admixture were suggested for the Gabonese samples of T. pulverulenta (Fig. 4). The population assignment of individuals between the two clustering methods was identical; however, admixture between populations was not detected by DAPC. Similarly, STRUCTURE suggested two populations at the genus level, and two in T. blandingii , but combined the Central African clades and suggested two populations, instead of three, for T. pulverulenta . Three populations were supported as the second highest ΔK and showed identical admixture proportions to those from LEA. We used five populations for our remaining analyses because multivariate-based analyses such as LEA and DAPC do not make assumptions about Hardy-Weinberg equilibrium and may be preferable over Bayesian methods such as STRUCTURE when sample sizes are small or uneven (Puechmaille, 2016).