Submitted to: Journal of Vegetation Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/11/2007
Publication Date: 9/22/2007
Citation: Goslee, S.C., Urban, D.L. 2007. The ecodist Package for Dissimilarity-based Analysis of Ecological Data. Journal of Statistical Software. 22(7):1-19.
Interpretive Summary: The presence of spatial structure provides new insight into ecological questions, but challenges scientists to identify and interpret it. This paper reviews dissimilarity-based methods for analysis of spatial structure, and presents a new method that overcomes many of the problems of the existing methods,. The partial Mantel correlogram makes it possible to understand complex nonlinear and linear spatial structures. We illustrate this method with data from the Sequoia National Park in California.
Technical Abstract: Ecologists are concerned with the relationships between species composition and environmental factors, and with spatial structure within those relationships. A dissimilarity-based framework incorporating space explicitly is an extremely flexible tool for answering these questions. Although the partial Mantel test is often used to account for the effects of space, the assumption of linearity greatly reduces its effectiveness for complex spatial patterns. We introduce a modification of the Mantel correlogram designed to overcome this restriction and allow consideration of complex nonlinear structures. This extension of the method makes it possible to use partial multivariate correlograms and to test relationship between variables at different spatial scales. We also review the standard Mantel methods and make some recommendations for their use. Reanalysis of vegetation and environmental data from Sequoia National Park using partial Mantel correlograms demonstrates that for this site, relationships of compositional dissimilarity and environmental dissimilarity with space are linear, but also shows the existence of an unmeasured spatial process acting at scales less than 1000 m.