Assessing biological dissimilarities between five forest communities

Background: Dissimilarity in community composition is one of the most fundamental and conspicuous features by which different forest ecosystems may be distinguished. Traditional estimates of community dissimilarity are based on differences in species incidence or abundance (e.g. the Jaccard, Sørensen, and Bray-Curtis dissimilarity indices). However, community dissimilarity is not only affected by differences in species incidence or abundance, but also by biological heterogeneities among species. Methods: The objective of this study is to present a new measure of dissimilarity involving the biological heterogeneity among species. The “discriminating Avalanche” introduced in this study, is based on the taxonomic dissimilarity between tree species. The application is demonstrated using observations from five stem-mapped forest plots in China and Mexico. We compared three traditional community dissimilarity indices (Jaccard, Sørensen, and Bray-Curtis) with the “discriminating Avalanche” index, which incorporates information, not only about species frequencies, but also about their taxonomic hierarchies. Results: Different patterns emerged for different measures of community dissimilarity. Compared with the traditional approaches, the discriminating Avalanche values showed a more realistic estimate of community dissimilarities, indicating a greater similarity among communities when species were closely related. Conclusions: Traditional approaches for assessing community dissimilarity disregard the taxonomic hierarchy. In the traditional analysis, the dissimilarity between Pinus cooperi and Pinus durangensis would be the same as the dissimilarity between P. cooperi and Arbutus arizonica. The dissimilarity Avalanche dissimilarity between P. cooperi and P. durangensis is considerably lower than the dissimilarity between P. cooperi and A. arizonica, because the taxonomic hierarchies are incorporated. Therefore, the discriminating Avalanche is a more realistic measure of community dissimilarity. This main result of our study may contribute to improved characterization of community dissimilarities.