We have recently proposed to use partial canonical ordinations to partition the variation of species abundance data into four additive components: environmental at a local scale, the spatial component of the environmental influence, pure spatial, and an undetermined fraction. By means of an example, we show how to use the information contained in these fractions to provide better insight into the data. In particular, the interpretation is assisted by separately mapping the various canonical axes and relating them to possible generating processes. We derive a general framework for the causal interpretation of the various fractions of this partition, which includes the environmental and the biotic control models, as well as historical dynamics.
biotic control canonical correspondence analysis cryptostigmatic mites environmental control historical dynamics mapping modeling ecological relationships spatial patterns variation partitioning