Test that the hypothesis, H0 = observed loss in PD is not different from random loss (random extinction probabilities).

We used the IUCN50 model of species’ extinction probabilities (see Table S2 for other models). The three orders with only one species in our dataset were removed from the table. The order Notoryctemorphia, with 2 data-deficient species, was also discarded. n = number of species that are not classified as data-deficient, and the number of data-deficient species is given in brackets; relative loss = PDloss. Significance tests with α = 5% are highlighted in bold; a sign in brackets indicates whether PDloss is higher (+) or lower (−) than expected randomly. Orders are presented in increasing order regarding PDloss when data-deficient species are discarded.