Detecting functional diversity loss under directional, stabilizing, and disruptive models of nonrandom anthropogenic extinction of species

Humans alter extinction pressures in ways that favor some species over
others, causing nonrandom loss of functional diversity. While the three
classic models of natural selection within species—directional,
stabilizing, and disruptive—are widely used to explain trait evolution, it
remains unclear whether they can also capture human-caused nonrandom
functional diversity loss among species. Here we develop a framework to
test for nonrandom extinction by tracking how trait means and variances
change as species with different functional traits are sequentially lost.
Applying this approach to the Caribbean lizard genus Leiocephalus, which
has already lost 8 of its 32 species and has 10 more threatened with
future extinction, we found that past extinctions were directional with
respect to ecomorphological traits, as large-bodied species were
disproportionately eliminated, likely due to hunting and introduced
predators. In contrast, predicted-future extinctions are best explained by
stabilizing processes, with species exhibiting extreme appendage
morphologies most at risk and species with intermediate appendage lengths
least likely to go extinct in the future, possibly because this phenotype
is better adapted to the deforested habitats that dominate Caribbean
islands today. Such a shift from directional to stabilizing nonrandom
extinction is expected when natural extinction pressures are replaced by
anthropogenic pressures that directionally shift trait distributions to
new adaptive optima. By linking trait-based extinction sequences to
classic evolutionary models, our approach provides a generalizable
framework for detecting and comparing nonrandom extinction across traits,
clades, and ecosystems.