It is commonly asserted in the ecological and economic literature that habitat loss is the main cause of loss of imperiled species. The evidence clearly shows that habitat loss is a common contributing factor, but there is little evidence that it is the most important factor. Studies that have focused on the mechanisms of species loss have failed to produce models capable of predicting patterns of loss as a function of human activities. I propose that this is because ecologists have employed an unrealistic conceptual model of the functioning of natural systems. Karl Popper's construct of the propensities of natural systems provides a more realistic view, and better potential to yield predictive models. I provide two examples of patterns of biodiversity and species loss in Canada where mechanistic reasoning is inconsistent with the observed propensities of species loss.