Phylogenetically and ecologically delimited species pool affect the inference of age-dependent extinction

According to Van Valens seminal work (1973), extinction occurs at a constantly stochastic rate within ecologically homogeneous groups or adaptive zones, giving long and short-lived species equal chances of extinction. Van Valen highlighted the difficulty in defining and identifying the species pool within an adaptive zone, but fundamentally viewed it through ecological factors. Most studies have used taxonomically or phylogenetically defined species pools to test the "Law of Constant Extinction." Here, we investigate how different species pools defined by phylogeny or ecology influence the inference of age-independent extinction. Using the Canidae fossil record and a Bayesian framework, we show that species pools defined by phylogeny or ecology exhibit different age-dependent extinction dynamics. The age-dependent extinction (ADE) signal varies depending on the species pool choice, time window, and taxonomic level. Within phylogenetic species pools, we observe mixed evidence for ADE, with both positive--older species more likely to go extinct (Hesperocyoninae and Borophaginae)--and negative--younger species more likely to go extinct (Caninae)--trends. Combining subfamilies into a family-level analysis yields weak evidence for ADE or strong support for age-independent extinction, depending on the period analyzed. Within ecologically defined species pools, hypercarnivores show strong evidence for positive ADE, whereas non-hypercarnivores show signals akin to age-independent extinction. Phylogenetic pools with more hypercarnivores tended to show evidence of positive ADE, while those with fewer tended toward negative ADE. These findings emphasize that the choice of species pool significantly influences observed age-dependent extinction dynamics and that ecology impacts the regime of age-dependent extinction.