Computing species compositional turnover across the environmental gradient
To estimate the magnitude of species turnover across the precipitation gradient and evaluate whether there was an influence of island age on the magnitude of this turnover from drier to wetter sites, we used a metric of beta diversity that captures the degree to which species are non-randomly distributed across space using individual- and coverage-based rarefaction and extrapolation (called betaC sensuEngel et al., 2021). We used pairwise comparisons among plots for each island, comparing the betaC for a pair of plots with the absolute difference in mean annual precipitation (hereafter ‘precipitation difference’) among that pair. If the precipitation difference is large, plots are quite distinct in their precipitation, whereas when the difference is small, plots are in similar precipitation habitats. Because plot sizes varied considerably in this compiled dataset, we used a rarefaction approach to control for the plot size differences by randomly sampling 13 individuals (the lowest number of individuals observed in a plot) from each plot of a pair to compute betaC. We used sample completeness of 60% to compute betaC, which was the maximum possible coverage given the number of individuals (13) used in the randomization. Each randomization was repeated 100 times and we estimated the mean value of betaC for each pair of plots. Finally, tested for the effect of a different level of coverage on the betaC estimates using a lower level of sample completeness, i.e., 30%. We calculated betaC using the function ‘betaC’ from (Engel et al., 2021).