Assessing species richness and diversity at the community level : methodological background

Most difficulties in biodiversity assessment arise from the very fact that it is essentially a matter of measuring a variability and not an average : counting the number of species which are different from each other rather than the average number of stems per ha. As a consequence : the usual sampling strategies which aim at estimating an average and its precision are not necessarily adapted; much more than for other resources, assessing biodiversity is scale- and level-dependent. Another important methodological point for regional studies and across-ecosystems studies is that the assessment of species diversity and richness of ecosystems is the dual viewpoint of the study of the ecological niche and geographic distribution of species. In this context, this lecture reviews the following topics : - General statistical background : sampling design and estimators; accuracy and precision; random, systematic, stratified and cluster sampling. - Assessment of species richness : number of observed species and species list; why the natural estimator obtained as the number of observed species is biased (it always underestimates the true richness); the cumulative species-area and species-individual curves and the correction method designed by Sanders and Hurlbert to compare samples of different size; the relative efficiency of various sampling designs and estimators : random, systematic and stratified sampling, fixed-area vs. fixed-number of individual plots, non parametric estimators. - Assessment of species diversity : diversity as a combination of species richness and evenness and the role of rare species; statistical properties of the usual estimators of Simpson's and Shannon's indexes of diversity : accuracy and precision. (Résumé d'auteur)