Assessing the nonlinear decay of community similarity: Permutation and site‐block resampling significance tests

Aim Modelling how community similarity decays with spatial distance is a key tool for the study of processes behind variation (beta diversity). Distance-decay models are computed from pairwise metrics (i.e. and between localities) hence suffer dependence in data, precluding use standard significance tests. Besides, distance-decay patterns inherently nonlinear because bounded 1 0. However, only method to assess model under dependency Mantel test, which considers linear model. To allow assessment patterns, we introduce here test combining pseudo-R2 statistic either permutations or block-site resampling replacement. Location Global. Taxon Applicable any taxon. Methods performance type I error statistical power), have applied this exponential power-law generalized simulated data. We compared its one test) illustrated real Results The has generally good power, even highly relationships. shows an adequate empirical approximation error, but presents lower power when relationship departs linearity. Main conclusions proposed should be preferred over test. Furthermore, could also used other distance-based regressions. novel site-block questions related that cannot tackled using permutations, such as estimating variance parameters.