The rate of global biodiversity loss is accelerating, making efforts to predict extinctions important. Good predictions are possible when extinction is non-random (NRE), but a general framework for predicting NRE effects on biodiversity is lacking. Natural selection is described with the adaptive landscape metaphor—a multidimensional surface that describes relationships between traits and fitness, where peaks are optimal trait combinations that maximize fitness for a population. As selective pressures change, adaptive optima shift, and individuals with maladaptive traits are less fit. NRE can be considered a response to changing adaptive optima, and like a population of species, clades can exhibit directional, disruptive, and stabilizing NRE analogous to classic selection patterns. Here, we introduce the NRE adaptive landscape framework and demonstrate it on an imperiled Caribbean lizard clade, Leiocephalus (curly-tailed lizards). We ask if loss associated with extinction is non-random, and if so, whether it explains changes to the Leiocephalus adaptive landscape that predict future extinctions. We develop methods to test for types of NRE for functional (external morphology and interactions), phylogenetic, environmental, and geographic estimates of biodiversity.
Results/Conclusions
For our analysis of Leiocephalus, we found evidence of random and non-random diversity loss. For external morphology, future and previous Leiocephalus extinctions are non-random but exhibit different patterns of loss: extinct species were markedly larger than extant species are, indicating directional NRE, but future predicted extinctions demonstrate stabilizing NRE of morphological diversity as more extreme trait values disappear. Predicted loss to environmental tolerance exhibited directional NRE. In contrast, changes in species interactions and phylogenetic relatedness were random. Our results indicate that changes in adaptive optima are not necessarily the same across concurrently evaluated traits and measures of clade biodiversity. Furthermore, our analyses suggest that shifts in adaptive optima for a clade may exhibit contrasting patterns of NRE through time. The NRE adaptive landscape framework provides additional insight into empirical extinction patterns and emphasizes the importance of evaluating biodiversity across multiple traits to prioritize conservation efforts.