Parallel declines in species and genetic diversity in tropical forest fragments Struebig et al. 2011. Ecology Letters Supplementary Information legends Figure S1 Contribution of multiple fragment combinations to regional species richness of forest specialists. Bat species accumulation was calculated by combining site-wise data from 15 fragments in a random order, while retaining information on the number of fragments, species and cumulative area. Species richness estimates are shown for 10,000 random fragment combinations (grey points), and for fragments added together in order of ascending (i.e. smallest first, red/black points) or descending (i.e. largest first, blue points) area. Data are presented as plots of cumulative forest area (a), cumulative number of fragments (b), and both (c). Figure S2 Regional rarefaction curves for species (a,b) and alleles of the three study species (c-e) in continuous forest (black lines) and fragments (coloured lines). Dotted lines represent the 95 % confidence limits of the mean richness values for continuous forest sites - values that fall outside of these confidence limits are significantly higher/lower. Rarefaction analyses were undertaken on all sites combined for continuous forest and forest fragments utilising the same data used in other analyses. Table S1 Mantel test correlation coefficients between genetic differentiation measured by Fst/1-Fst and the following pairwise ecological or geographic distances for the three bat species: (i) bat assemblage dissimilarity (Morita-Horn index), (ii) Euclidean geographic distance, and (iii) least-cost distance. Mantel tests were calculated for these pairwise distances in Genalex v6.3 (Peakel and Smouse, 2006). Probabilities of the correlation coefficients, calculated using 999 permutations, are shown in parentheses.