Predicting extinction rates in space and time: A macroecological approach

Background/Question/Methods

Estimation of species extinction rates due to habitat loss and range contraction remains a difficult but important task. Many widely used methods, such as population and species distribution modeling, require data on species life history or occurrence patterns that may not be available for poorly studied taxa and novel habitats. Here I demonstrate how a macroecological framework can provide first-pass estimates of likely extinction rates when such data are lacking. This framework is based on the combination of two well-known macroecological patterns: the species abundance distribution (the probability that a species in a community has a total abundance n₀) and the species-level spatial abundance distribution (the probability that a species with a total abundance n₀ in a large landscape has n individuals in a subplot within that landscape). In a spatial context, I combine these distributions to create an extended species-area relationship that predicts extinction risk for individual species, provides a confidence interval around the predicted total number of extinctions, and explicitly incorporates a minimum viable population size parameter. In a temporal context, I generate predictions of the likely magnitude of extinction debt, the number of time-delayed extinctions that occur subsequent to initial habitat loss, for a variety of community types.

Results/Conclusions

The spatial analysis demonstrates that a 95% confidence interval on the extinction predictions of a species-area analysis is roughly +/- 10-20% for many real landscapes, with the largest uncertainties occurring when a community contains relatively few species. The minimum viable population size required for a species to be considered protected has a significant influence on extinction predictions, with an increase in the viable population size from one (as is assumed in the classic species area relationship) to ten implying that an order of magnitude more habitat must be protected to achieve the same level of species protection. The temporal analysis demonstrates that, across a range of models and parameters, extinction debts are generally largest for large communities, communities that follow a canonical lognormal abundance distribution, communities in which species exhibit low spatial aggregation, and landscapes experiencing relatively low levels of habitat loss. This analysis also identifies the potential for an "extinction credit", in which the diversity of a landscape rises to a higher equilibrium following an initial extinction event, that is predicted when species are highly aggregated in space.