Changes in the environment, such as those resulting from climate, land use and atmospheric nitrogen deposition, affect the performance and distribution of populations and species from local to global scales and disrupt ecosystem services provided by natural capital. In order to manage and adapt to the effects of environmental disruptions, we need to develop general predictions of the responses of populations to the environment from local to global scales. The state of the art for predicting population responses to new environmental conditions is the use of Species Distribution Models, with future scenarios projected into geographic space. SDMs can then be coupled with demographic and dispersal models with population performance assumed to be correlated with predicted probability of occupancy from the SDM. We use demographic data collected in the field for 38 plant species from the COMPADRE plant matrix database together with predicted probabilities of occupancy from Species Distribution Models to test the key assumption that population performance can be predicted from probability of occupancy. We use a larger data set of 210 species to test whether population performance can be predicted from climatic conditions, spatial proximity and phylogenetic relationships.
Results/Conclusions
Population growth rates were not correlated with climatic suitability, but variation in population growth rate was significantly positively correlated with climatic suitability. Elasticity of fecundity was negatively correlated with climatic suitability and elasticity of survival was positively correlated with habitat suitability indicating the importance of fecundity for population growth rate in relatively unsuitable climates and the importance of survival transitions in highly suitable climates. These elasticity patterns were consistent with increasing importance of competition in highly suitable climates.
Geographic proximity and phylogeny explained 5-40% of variation in four key metrics of population performance. These metrics could be extrapolated from species that diverged ≤ 10-100mya and populations closer than 35 km. Extrapolation is therefore limited to geographic scales smaller than those at which landscape scale threats typically occur. Over 50% of populations of species with >1 population were studied within 2km of each other and 95% of populations were within 100km of each other. We therefore lack data on population performance at the species range scale. Data collection and population modeling at large spatial extents is underway but more effort is needed in order to fill this critical data gap.