Why do several small patches hold more species than few large patches?

Background The principle that a single large habitat patch should hold more species than several small patches totalling the same area (SL > SS) is used by conservation agencies to favour protection of large, contiguous areas. Previous reviews of empirical studies have found the opposite, SS > SL, creating the single large or several small (SLOSS) debate. Aims Review the empirical and theoretical SLOSS literature; identify potential mechanisms underlying the SS > SL pattern; evaluate these where possible. Location Global. Time period 1976-2018. Major taxa Plants, invertebrates, vertebrates. Methods Literature review. Results Like previous reviews, I found that SS > SL dominates empirical findings. This pattern remained, although it was somewhat weakened, in studies where sampling intensity was proportional to patch size. I found six classes of theory, and conducted five preliminary evaluations of theory. None of the predictions was supported. The SS > SL pattern held for specialist species groups, suggesting it does not result from incursion by generalists into small patches. I found no evidence for the prediction that the reverse pattern (SL > SS) becomes more common over time since patch creation, through gradual species losses from SS. I found no difference between results for natural and anthropogenic patches. There was also no evidence for predictions that SL > SS is more common when the matrix is more hostile, or for stable than ephemeral patches. Main conclusions Most empirical comparisons find SS > SL. While there are several potential causes, more empirical work is needed to identify those at play. Meanwhile, conservation practitioners should understand that there is no ecological evidence supporting a general principle to preserve large, contiguous habitat areas rather than multiple small areas of the same total size.