RESULTS

When we analyzed patterns of alpha diversity including only relatively uninvaded plots and native species, we found a positive relationship between precipitation and all three facets of diversity on O’ahu, the oldest island in our study, but no relationship with precipitation on the two younger islands, Maui Nui and Hawai’i (Figure 2 a-f and Figure S4 a and c). This pattern was true regardless of the weighting of common versus rare species (q=0, 1, 2). When we included plots that were highly invaded, and measured diversity with both native and alien species, the pattern differed (Figure 2 g-l). Specifically, we found a positive relationship between precipitation and taxonomic, functional and phylogenetic diversity for O’ahu and Hawai’i, but a negative relationship between precipitation and all three types of diversity on Maui Nui (Figure 2 g-l; see model estimates and both 85% and 95% confidence intervals being negative for Maui Nui in Figure S4b). On O’ahu, the positive relationship between phylogenetic diversity and precipitation was stronger for Hill number q=0 than q=2 (see model estimates for phylogenetic q=0 being greater than those of q=2 for O’ahu in Figure S4 b and d). In all cases, the interaction term between island age and mean annual precipitation was significant (Table S1).
When we only considered native species in relatively less invaded plots, we found that species compositional turnover increases with precipitation differences among plots on O’ahu and Maui Nui (Figure 3a and Figure S5 a and c). However, on Hawai’i, the compositional turnover of native species was somewhat negatively related to precipitation differences among plots (Fig 3a Figure S5 a and c). When we included plots that were highly invaded, we found that the relationship between species compositional turnover and precipitation differences among plots was positive for Hawai’i and O’ahu, but negative for Maui Nui (Figure 3b and Figure S5 b and d). In all cases, the interaction term between island age and mean annual precipitation was significant (Table S1)
When we truncated mean annual precipitation to a common range among the three islands, we found qualitatively similar results (Figure S6). Furthermore, the significant interaction between precipitation and island age on diversity remained (and even strengthened) (Figure S7). Finally, when we redid analyses using a lower level of sample completeness (30%), we found the relationship of species compositional turnover with precipitation across islands was different, but qualitatively showed a similar pattern that turnover among different precipitation levels was higher on O’ahu) (Figure S8).