Modeling the distributions of shrub species in Pacific northwest forests

In Pacific Northwest forests, understory species are important elements of forest structure and composition, providing habitat and forage for wildlife and contributing to much of the floral diversity. Knowledge of the factors that control the distribution and abundance of understory species is central to understanding the ecology and dynamics of natural forests and how they will respond to management. Generalized linear models of the binomial family were used, at two spatial scales, to estimate the probability of occurrence of shrub species as a function of environmental variables and the cover of mature and regenerating tree species. Graphical analysis and resampling techniques were used to: (1) examine whether species show unimodal responses to environmental gradients; (2) explore the scale-dependence of model forms; and (3) predict areas of high probability of species occurrence within a multi-dimensional environmental space. Results suggest that the concept of unimodal response to environmental gradients can be used to improve predictions about species' distributions, and to strengthen the biological interpretation of statistical models. They also demonstrate the scale-dependence of model equations and the likelihood that key processes influencing plant distributions may change, or appear different, when viewed at different scales. (C) 1999 Elsevier Science B.V. All rights reserved.