Recent studies at the macro-scale have demonstrated that geographic gradients in the richness of plants, in particular of woody plants such as trees and shrubs, can be viewed as by-products of water-energy dynamics. According to this view, they are climatic rather than latitudinal/longitudinal gradients, relating to coincident and predictable variations in planetary surface-atmosphere thermal dynamics and consequent patterns in biological activity. Previous analyses have shown that a two-variable model capturing the dynamic relationship between energy (heat/light) and water (rainfall) accounts for most of the variation in woody plant richness across southern Africa at species, genus, and family levels. Here we move towards a more complete explanation, while demonstrating how geographic analysis of residuals can be used to identify the type and sequence of additional variables for inclusion, either at the same or at more discrete scales of analysis. Residual geographic variation in richness from the two-variable model displays a geographic pattern unrelated to longitude and latitude. Regional clusters of under- and over-prediction point to macro-scale variation in topographic relief as a significant factor. When topographic relief is added as a third variable, the explanatory power (R2) increases by 7 to 12%, and the subsequent pattern of variation in residuals becomes even more unpredictable. What clustering remains points to other macro-, and meso- or micro-scale variables that need to be considered. Such a top-down, trans-scalar approach permits systematic and objective development of more complete explanations, while the three-variable macro-scale model developed herein is the basis for a powerful research tool for ecologists, biogeographers, conservationists and bio-climatologists alike. [ABSTRACT FROM AUTHOR]