Abstract
Variation partitioning is an important tool to investigate the importance of spatial structure to species distribution in communities, but its use is lacking in marine ecosystems. The influence of rare species and the level of taxonomic resolution are important features to be considered when planning sampling designs and efforts for community studies. We tested the significance of spatial and environmental components to species distribution of a polychaete community, considering how the results varied among models with increasingly rare species (and contrasting metrics), and among models with distinct taxonomic levels. For all models evaluated, both spatial and environmental components significantly explained the structure of species distribution. However, spatial processes explained a higher percentage of variance than environmental ones in almost all models. The inclusion of rare species did not influence the outcome in most models, with the exception of a Chi‐square distance model with all species included. Family taxonomic level was sufficient to detect the relative contribution of both components, but the number of environmental variables included in the model was higher for the genus level. The influence of rare species was related to the metric applied to the analysis. If weight to rare species is not desirable, the use of a few dominant species coupled with the Hellinger function can provide a good surrogate to the whole sampled community in variation partition analysis. The genus level is suggested as an appropriate taxonomic level with which to evaluate the relevance of spatial and environmental components in polychaete community studies.