Null-Hypothesis Significance Testing and the Critical Weight Range for Australian Mammals

Abstract:  The critical weight range hypothesis for Australian terrestrial mammals states that species in the intermediate size range 35–5500 g are particularly susceptible to extinction. In a 2001 study Cardillo and Bromham found no statistically significant evidence for this hypothesis and suggested that research should instead focus on why small species are resistant to extinction. We used a similar data set of body sizes of Australian mammals grouped by conservation classification, but we used test statistics (mean deviation above and below the median body size) that are more statistically powerful than those of Cardillo and Bromham (quartiles, maxima, and minima of body size distributions). We found strong evidence in favor of the critical weight range hypothesis: the body size distribution of threatened species was more clustered toward the median body size from above and below. This pattern was statistically significant at the continental scale and in the arid zone, but not in the mesic zone. Confusing statistical significance with evidence of no effect, as Cardillo and Bromham did, can have negative implications for conservation biology because it can result in failure to act when action is warranted or provision of incorrect advice that affects policy and planning decisions.