Use of local climate to improve predictions of plant population viability in future climates

Background/Question/Methods

Changes in global climate could cause shifts in species growth, fecundity, and survival, and therefore affect the long-term dynamics of rare and endangered plant populations. Linking climate change models to demographic vital rates may provide useful insights into the potential effects of environmental changes on rare plants, and further aid in their current and future conservation. Our study goal was to determine predictors of plant population vital rates for five rare, native Oregon perennials, and to use these predictors to simulate future population size from forecasted climate conditions. Population vital rates were calculated from demographic data sets of seven to ten years per species with populations at one to five sites. We developed multiple one-predictor models to predict population vital rates from local climate conditions using nonlinear, nonparametric regression. Local climate predictors included total precipitation, and average minimum, maximum, and dew point temperature during specific growing seasons at each site. We then simulated future population trajectories based on 1) GCM-predicted climate conditions and 2) IID conditions of our data sets without climate predictors.

Results/Conclusions

Climate was a strong predictor of most species population vital rates; however this was confounded by site variation as predictors were not equally strong across all populations among species. Average maximum and minimum temperature explained more variation in population growth rates for species located in seasonally high precipitation habitats, while total precipitation and average dew point temperature explained more variation in precipitation-limited habitats. However, vital rates for these five species were temporally auto-correlated across years, which occasionally explained more variation in vital rates. Forecasts of population growth under climate change scenarios are more pessimistic in some plant species than forecasts based on past conditions or IID, suggesting negative effects of climate change on these rare species. Inclusion of correlated climate conditions in demographic population models increased accuracy and precision around observed population sizes and may be useful in stochastic demographic models in general, especially given the risk of current population viability in uncertain future climates.