2.6 Influence of host plants and geographic distance on genetic divergence
We analyzed the influence of geographic distance and host plants on genetic divergence among H. pungens complex populations from the native range. Non-native populations in these analyses were excluded because samples collected in these locations reached the areas by human mediated dispersal and would distort the analyses. We used a combination of correlation tests and multiple regression analyses of distance matrices to analyze the importance of geographic distance and host plant use on genetic divergence.
In these analyses, we used pairwise Fst linearized values at population level based either on nuclear SNPs or mtDNA datasets to construct the matrix of genetic distances. To construct the geographic distance matrix, we used the coordinates (latitude and longitude) of each population (Table S1). Host plant distance matrix was built considering host plant families where mealybugs were collected: Cactaceae, Portulacaceae, and Amaranthaceae. Thus, we constructed a binary matrix with three categories (the host plant families) considering patterns of host plant use.
To analyze the influence of geographic distance and host plant use on genetic divergence we employed two Mantel tests (Mantel, 1967), first considering geographic and genetic matrices and second host plant and genetic distance matrices. We then conducted a partial Mantel test (Smouse, Jeffrey, Long, Robert, & Sokal, 1986) to evaluate the influence of host plants on genetic divergence, using pairwise Fst and host plant distance matrix, while controlling for geographic distance (Legendre & Fortin, 2010). For all correlation analyses we used the R package vegan v2.4.4 (Oksanen et al., 2007) and 9,999 permutations to assess statistical significance. We further tested the influence of geographic distance and host plant use on genetic divergence by performing matrix regressions using the function MRM from the package ecodist v2.0.1 (Goslee & Urban, 2007), and ran 9,999 permutation to assess the significance of regression coefficients.