Bayesian sample assignment
On average, negative log-likelihood scores for the independent
Structure runs increased from K = 2 through K =
13, before decreasing at K = 14 (Supplemental Appendix S1 Figure
S5). Based on the ∆K method implemented in
StructureHarvester, the optimal partition scheme was determined
to be K = 3 (Supplemental Appendix S1 Figure S6). However
additional peaks, potentially representing additional partition schemes,
were detected at K = 6, K = 8, and K = 13.
Individual probabilities of assignment (Q ) for major modes for
values of K = 3, K = 6, K = 8, and K = 13
(Figure 2) show the clustering of the invasive populations relative to
those in the native range. Major modes for values of K= 2 throughK =14 are presented in Supplemental Appendix S1 Figure S7. A
summary of the biogeographic patterns based on the distance matrix
analysis of the population membership coefficients (Structure
“popfiles”) for K = 6, K = 8, and K = 13 are
presented graphically in Supplemental Appendix S1 Figure S8, Figure 3,
and Supplemental Appendix S1 Figure S9, respectively.