Shifting Vital Rate Correlations Alter Predicted Population Responses to Increasingly Variable Environments

Time series of vital rates are often used to construct "environment-blind" stochastic population projections and calculate the elasticity of population growth to increased temporal variance in vital rates. Here, we show that the utility of this widely used demographic tool is greatly limited by shifts in vital rate correlations that occur as environmental drivers become increasingly variable. The direction and magnitude of these shifts are unpredictable without environmentally explicit models. Shifting vital rate correlations had the largest fitness effects on life histories with short to medium generation times, potentially hampering comparative analyses based on elasticities to vital rate variance for a wide range of species. Shifts in vital rate correlations are likely ubiquitous in increasingly variable environments, and further research should empirically evaluate the life histories for which detailed mechanistic relationships between vital rates and environmental drivers are required for making reliable predictions versus those for which summarized demographic data are sufficient.