Scale dependency of diversity components estimated from primary biodiversity data and distribution maps

Different sources of information about biodiversity may lead to unrealistic or biased estimation of its components, with different patterns according to the scale of analysis. In this study, we analyse patterns of species richness at the local (average alpha) and regional (gamma) scales, and the relationship between them (Whittaker's beta), in central Mexico, using as a source of data for the species' distributions: (1) museum specimen occurrence data for birds, and (2) distribution maps based on ecological niche models developed and refined by experts. We performed analyses at five spatial resolutions (1/32 degrees-1/2 degrees). Scale changes (grain and extent) affected significantly the estimates of average alpha, gamma, and beta. Use of raw occurrence data vs. distribution maps yielded contrasting results, with raw data underestimating alpha and overestimating beta, as functions of area. As regards species-area relationships, our results suggest a natural decomposition of factors into an area-invariant component (related to alpha), and an area dependent factor (related to beta). Most of our results are maintained in a null model that randomizes occurrences without changing observed range-size distributions. From this result we argue that average alpha and Whittaker's beta capture little information about the spatial covariation of species distribution patterns.