3.4 Evolutionary consequences of transition to homostyly
Mating system in homostyles - Outcrossing rates estimated from progeny arrays were more variable and significantly lower in homostylous than heterostylous families (single locus [t s]: mean ± SE; 0.24 ± 0.10 [homostylous] and 0.83 ± 0.05 [heterostylous]; multi-locus [tm ]: 0.14 ± 0.06 and 0.98 ± 0.02, respectively; Wilcoxon signed-rank test W = 63, P = 0.002 and 0.001, respectively; Figure 5). As expected, the frequency of homostyles within populations was positively correlated with population-level estimates of selfing rate (pseudo-R² = 0.439,P = 0.002; pseudo-R² = 0.288, P = 0.007;pseudo-R² = 0.447, P < 0.001; for s(g2) ,s(ML) and s(BES) , respectively; Table 1, Figures 6A and S2). Seed germination rate was significantly lower for homostyles than heterostyles (0.11 ± 0.03 vs. 0.26 ± 0.05; W = 308, P = 0.04) resulting in an estimated inbreeding depression (δ) of 0.58.
Patterns of genetic diversity – 18 out of 22 populations (except for D03, D06, D08 and D10) had significant deviations from Hardy-Weinberg equilibrium due to a heterozygote deficit (Supplementary Table S4). Population-level estimates of allelic richness (Ar ) varied from 1.58 to 3.78 (mean ± SE, 3.19 ± 0.09), expected heterozygosity (He ) from 0.14 to 0.53 (0.42 ± 0.01), and observed heterozygosity (Ho ) from 0.11 to 0.46 (0.37 ± 0.02). Inbreeding coefficient (Fis ) ranged from -0.08 to 0.39 among populations (0.13 ± 0.02; Table 1) and was positively correlated with the frequency of homostyly ( = 0.405, P = 0.03), but not with population size (P = 0.27; Figure 6B).
Mean values of genetic differentiation among populations were 0.083 (0.074–0.093) and 0.164 (0.147–0.181) for Fst and G’st , respectively (Table S4), and were strongly correlated with each other (pseudo-R² = 0.974, P = 0.001). Pairwise population values of Fst and G’st are shown in Supplementary Material Table S5. As expected, pairwise G’st was positively correlated with pairwise increase of frequency of homostyles (Mantel test: = 0.481, P < 0.001).