View article

We compared two Monte-Carlo randomization tests as neutral models, for measuring interspecific spatial associations. Complete randomization randomly re-allocates the individuals of each species into the spatial sample with the same abundances as in the studied sample. Thus, this separates two components of association:(1) textural, ie related to the species abundance distribution, and (2) structural, ie related to the spatial distribution of the species. Random shifts keep the spatial patterns of individual species, but change their relative positions. Thus, this method separates (1) the interspecific components of associations from (2) the intraspecific ones, randomizing the first. We also applied a new algorithm for determining the signs of pairwise associations (ie positive or negative). This technique is based on the comparison of the observed and simulated joint distributions of a species pair. We tested the neutral models and sign determination algorithms on multispecies circular transect ('trainsect') data. All statistics were calculated across a range of scales, ie the calculations were repeated at a series of sampling unit lengths. One field transect and four simulated patterns were examined. The simulated patterns, created by spatially explicit individual based model, represent known assembly rules. We concluded that circular transects are appropriate in field studies for detecting fine scale community patterns because their topology allows us to apply more types of randomizations. The separation of textural, intraspecific and interspecific effects provided us more opportunity to detect and interpret the spatial associations. The new sign determination …