Although broad-scale diversity gradients have been investigated since the 18th century, there has been an increasing interest in this research area in the last 15 years. This resurging interest can be explained by many factors, including the conceptual framework provided by the new field of macroecology, the demand for understanding broad-scale patterns to minimize biodiversity losses in the face of the environmental crisis, the new possibilities of getting broad-scale data of species distribution and environmental variables, and the new analytical tools available. Our paper deals with one of these statistical tools, spatial autocorrelation analysis. We showed how it could be applied to broad-scale patterns in diversity, and how it can be used to understand which environmental factors drive species diversity at different spatial scales. Although we used the spatial variation of bird species richness in the western Palearctic, as an example, the broader methodological appeal of the paper can explain why it have been highly cited, since it may be of interest to all ecologists working with biodiversity patterns all around the world.

Our paper explains how spatial autocorrelation analysis can be used to understand broad-scale patterns in diversity. Spatial autocorrelation measures the similarity between samples for a variable (i.e., species richness) as a function of spatial distance and is a unifying concept in spatial statistics. Although the idea that spatial autocorrelation is important in ecological data analysis has been increasingly recognized since the early 1990s, only recently have researchers started to effectively understand the statistical issues involved and incorporate the methods in their work. In our paper, we showed how the methods of spatial autocorrelation can be used as exploratory tools to describe patterns in diversity and how environmental factors drive these patterns. We also compared the effect of incorporating spatial autocorrelation structure in multiple regression models, in a generalized least-squares approach, and our results support the hierarchical framework that is currently accepted to understand how diversity gradients are generated at different spatial scales.

Luis Mauricio Bini and I have worked on autocorrelation methods applied to ecology, genetics and evolutionary biology since the beginning of the 1990’s. In 2001, my research group started to work together with Brad Hawkins, from the University of California at Irvine, who called our attention to the lack of knowledge about how spatial autocorrelation should be measured and interpreted when applied to studies dealing with broad-scale diversity gradients. Since then, we have published many papers together, including this one.

It is now widely recognized that it is quite important to understand broad-scale patterns of biodiversity and how environmental and evolutionary processes regulate them. At the same time, the large amount of data available today requires that more sophisticated statistical and mathematical tools are applied to better understand these processes. Our paper deals with one of these new tools, called spatial autocorrelation analysis, and tries to explain to other scientists interested in ecology and biogeography how it can be used and how to interpret the results obtained.