Quantifying factors for understanding why several small patches host more species than a single large patch

Several small habitat patches typically contain more species than a single large patch of comparable area. This pattern has fueled controversy over reserve design (the SLOSS debate); yet its causes remain poorly understood, limiting discussion of any broader biodiversity implications. Using 44 published faunal metacommunity datasets from fragments, islands and habitat islands, we test four hypotheses to explain differences in species richness (SRd) between several small patches and a single large patch: (i) beta diversity (BD); (ii) evenness of species abundance distributions (EV); (iii) size bias in sampling efficiency (SE); and (iv) taxonomy. We used individual-based null models and confirmatory path analysis to compare direct and indirect support for these non-exclusive hypotheses. Broad taxonomic grouping (invertebrates > birds = non-volant vertebrates) accounted for most of the variation in SRd, while BD, EV and SE had smaller, but comparable, direct effects (standardized coefficients = 0.42, 0.31 and −0.26 respectively). Within a taxonomic group, larger SRd occurred when both BD and EV increase. Although sampling bias probably overstates the difference in SRd, it does not explain the pattern. Beyond reserve design, the SLOSS pattern implies potential principles for managing biodiversity in fragmented landscapes. Whether this pattern should inform conservation requires greater understanding. We confirm beta diversity is the major contributor, but it is mediated by increases and decreases in evenness among small patches. Attention to how sub-division alters relative abundances (e.g., by biotic relaxation or altering competitive outcomes) might help clarify how the SLOSS pattern arises.