You are here:
Assessment of Spatial Autocorrelation in Empirical Models in Ecology
Citation:
Cablk, M. E., R D. White, AND A. R. Kiester. Assessment of Spatial Autocorrelation in Empirical Models in Ecology. Chapter 37, Scott, M; Heglund, P; Morrison, M; Rafael, M; Wall, B; Hoffer, J (ed.), Predicting Species Occurrences: Issues of Accuracy and Scale. Island Press, Washington, DC, , 429-440, (2002).
Description:
Statistically assessing ecological models is inherently difficult because data are autocorrelated and this autocorrelation varies in an unknown fashion. At a simple level, the linking of a single species to a habitat type is a straightforward analysis. With some investigation into the assumptions of statistics and an understanding of error generation and propagation in spatial databases, however, we find that such analyses are not as straightforward as they may appear. When the analysis is further complicated by addressing multiple species and pattern-process relationships, standard statistical methods fall short. Furthermore, methods for validating spatial statistical models do not exist. We examine the issues of estimating error and error propagation in ecological models of vertebrate diversity with the state of Oregon as a case study. We propose a method for modeling vertebrate richness at a landscape scale by taxa (i.e. many species rather than a single species) which allows for the inherent and unknown spatial structure of the data to be expressed and quantified. We also propose a method for evaluating the goodness-of-fit of these models which incorporates both standard and spatial statistical methods. We found that exploratory data analysis was useful for modeling vertebrate diversity with continuous data and that the proposed method for evaluating the models was robust given spatial autocorrelation.