Predicting population abundance across a species range

Background/Question/Methods Linking vital rates to the environment is an essential step to understanding the contribution of environmental variation to population dynamics. Ehrlén and Morris (2015) propose extending this approach by incorporating density-dependence into models to jointly predict abundance (equilibrium density) and distribution (locations where low-density population growth rates are positive). Population models that make predictions in terms of population size or distribution can be validated with empirical data. We take advantage of spatial and temporal variation in data on the climate at and demography of populations of the annual plant Clarkia xantiana ssp. xantiana (Onagraceae) to build, validate, and test a driver-dependent population model for abundance across the geographic range of the subspecies in the southern Sierra Nevada mountains.

Results/Conclusions We model the statistical relationship between multiple vital rates (seed survival, germination, seedling survival to fruiting, seed set) and the abiotic environment, climate, and density. We show that vital rates depend on covariates for the abiotic environment (slope, azimuth, and rock type), climate (temperature and precipitation), and density (seedling numbers). We describe the effect of spatial variation in the abiotic environment and temporal variation in climate and density on vital rates. We use these analyses to obtain parameter estimates and construct a population model. For the focal sites, we use site-specific covariates and site-specific parameter estimates to predict abundance of fruiting plants. We validate the model by comparing predicted abundance and observed fruiting plant abundance at focal sites. For additional out-of-sample sites for which we have ten years of abundance data, we use site-specific covariates and mean parameter estimates to predict abundance. We test the predictive ability of the model by comparing predicted abundance and observed fruiting plant abundance at out-of-sample sites. The model should provide a tool for predicting how future environmental change will influence the abundance and distribution of our study species.