PLoS ONE (Jan 2022)
Demographic and life history traits explain patterns in species vulnerability to extinction.
Abstract
As ecosystems face disruption of community dynamics and habitat loss, the idea of determining ahead of time which species can become extinct is an important subject in conservation biology. A species' vulnerability to extinction is dependent upon both intrinsic (life-history strategies, genetics) and extrinsic factors (environment, anthropogenic threats). Studies linking intrinsic traits to extinction risk have shown variable results, and to our knowledge, there has not been a systematic analysis looking at how demographic patterns in stage-specific survival and reproductive rates correlate to extinction risk. We used matrix projection models from the COMPADRE and COMADRE matrix databases and IUCN Red List status as our proxy of extinction risk to investigate if some demographic patterns are more vulnerable to extinction than others. We obtained data on demographic rates, phylogeny, and IUCN status for 159 species of herbaceous plants, trees, mammals, and birds. We calculated 14 demographic metrics related to different aspects of life history and elasticity values and analyzed whether they differ based on IUCN categories using conditional random forest analysis and phylogenetic generalized least square regressions. We mapped all species within the database, both with IUCN assessment and without, and overlaid them with biodiversity hotspots to investigate if there is bias within the assessed species and how many of the non-assessed species could use the demographic information recorded in COMPADRE and COMADRE for future IUCN assessments. We found that herbaceous perennials are more vulnerable when they mature early and have high juvenile survival rates; birds are more vulnerable with high progressive growth and reproduction; mammals are more vulnerable when they have longer generation times. These patterns may be used to assess relative vulnerability across species when lacking abundance or trend data.