Detecting stabilizing, directional, and disruptive patterns of anthropogenic species loss with general models of nonrandom extinction

The selective landscape that gave rise to Earth’s species has shifted in the Anthropocene. Humans have accelerated extinction pressures, making efforts to detect general non-random patterns of extinction increasingly important. Much research has focused on detecting which traits make some species more likely to go extinct, such as large body size and slow reproductive rate in animals, limited dispersal in vascular plants, and narrow habitat requirements in cacti. However, general models for such non-random extinction are lacking. Here, we adapt the three general models of natural selection to classify non-random extinction as directional, disruptive, or stabilizing extinction. We develop a quantitative method for testing which general extinction model best describes observed data and apply it to the Caribbean lizard genus Leiocephalus as a case study. We surveyed the literature for recorded last occurrence for extinct and threat status for extant species. Eight species have gone extinct and ten are predicted to go extinct soon. Past extinctions in Leiocephalus showed directional extinction of large bodied species, while future-predicted extinctions exhibited a more complex extinction model similar to both random and stabilizing extinction with respect to body size. Similarly, future-predicted extinctions exhibited stabilizing extinction with respect to limb and tail lengths. Lizards with either very long or very short appendages are most likely to go extinct in the future. This shift from directional to stabilizing extinction for Leiocephalus is consistent with hunting, introduced predators, and habitat loss that first increased extinction pressure on the largest species and then extinction pressure on species that deviate from an adaptive peak centered on a generalist ground-lizard body plan. As adaptive optima shift in the Anthropocene, general models of non-random extinction are essential to developing a mature strategy for future successful conservation efforts. ### Competing Interest Statement The authors have declared no competing interest.