Testing for efficacy in four measures of demographic buffering

Understanding population responses to variable environments is central to much of current research in population ecology and conservation biology. Environmental variability, a key component of global climate change, increases the extinction risk of species across the tree of life. Therefore, quantifying the sensitivity of populations to environmental variability is timely in the face of global climate change. A common approach to measure the impact of environmental variability on a population is by quantifying the population’s capacity towards demographic buffering specifically, the population’s ability to reduce the impact of environmental variability on its own growth rate. This line of work has, over the past 25 years, resulted in multiple, heterogeneous methods to quantify demographic buffering. To date, we lack clarity on which method is most appropriate, and under what conditions. To identify the best method to quantify demographic buffering, we test four methods – one correlational method, two methods using terms from Tuljapurkar’s approximation and the summation of stochastic elasticities of variance – for their efficacy to inform conservation strategies. We compare and contrast these methods via three different tests to determine the efficacy of the methods across four integral projection models for plants representing different life histories. In the first test, we determine if the measures, structured by ontogeny, are similar or distinct by analyzing their covariance structure across the four species. In the second and third tests, we perform two counterfactual simulations to test if the measures offer insights about the populations’ responses to variable environments that are better than chance. We find that the four methods significantly differ in their ability to identify and quantify demographic buffering. Furthermore, our simulations identify the summation of stochastic elasticity of variance as the most effective method to quantify demographic buffering. This work represents a clear example of how and why to test the metrics we infer from structured systems prior to their applications in systems of interest (e.g., endangered populations). In addition, our finding that commonly used approaches to quantify demographic buffering are ineffective has broad implications for our current understanding of how natural populations are responding to climate change, and thus for effective conservation practices.