Population density estimates for terrestrial mammal species

Abstract

AbstractAimPopulation density is a key parameter in ecology and conservation, and estimates of population density are required for a wide variety of applications in fundamental and applied ecology. Yet, in terrestrial mammals these data are available for only a minority of species, and their availability is taxonomically and geographically biased. Here, we provide the most plausible predictions of average population density, their natural variability and statistical uncertainty for 4,925 terrestrial mammal species.LocationGlobal.Time period1970–2021.Major taxa studiedTerrestrial mammals.MethodsWe fitted an additive mixed‐effect model accounting for spatial and phylogenetic autocorrelation on a dataset including 5,412 average population density estimates for 737 species. Average density was modelled as a function of body mass, diet, locomotor habits and environmental conditions. We validated the model using spatial and taxonomic block cross‐validation and used the estimated error to quantify the uncertainty around statistical predictions of population density for 4,925 mammal species.ResultsSmall body size, fossorial behaviour and herbivorous diets were associated with the highest population densities, whereas large size, aerial behaviour and carnivorous diets were related to the lowest densities. Species in non‐seasonal environments yielded higher densities than species in environments with high precipitation seasonality. Empirical estimates of population density vary by about four times on average within the same species, and statistically independent predictions for the majority of species deviate by about five times from observed values, indicating that prediction errors are similar to the natural variability in population densities.Main conclusionsOur predictions and uncertainty estimates of average population densities open up a number of applications in macroecology and conservation biogeography, including biomass estimation, setting population targets in conservation assessments and planning, and supporting Red List assessments. The methodology can be replicated easily for other taxonomic groups with a representative sample of georeferenced density estimates.