Assessing the Impacts of Climate Change on the Distribution of Trout Species in the Sierra Nevada Region of California Using Output from a Landscape Scale Hydrological Model

Abstract:

Species distribution models (SDMs) are statistical tools that combine data on the locations where a species is known to occur with data on the environment. The resulting species-environment relationships are then used to identify regions a species could occupy and are often used to make predictions about how a species will respond to climate change. The data used in SDMs generally exhibit a pattern known as spatial autocorrelation (SAC), which is the positive association between the proximity of sample locations. The presence of SAC violates the statistical assumptions that must be met for unbiased estimation of species-environment relationships and can cause traditional SDMs to exhibit low predictive accuracy. To determine the effects of SAC on predicting species’ responses to climate change, the distributions of three species of trout were predicted throughout the Sierra Nevada mountain range in California. Measures of contemporary streamflow, water temperature, dissolved oxygen, and sediment concentration are based on model outputs from the Soil and Water Assessment Tool (SWAT) landscape scale hydrological model. SWAT-derived future hydrological conditions driven by downscaled General Circulation Models (GCMs) indicate that increasing temperatures and changes in precipitation will alter these variables across the region, all of which have the ability to negatively impact the distribution of trout species. Trout distribution data along with data on contemporary stream flow, water temperature, dissolved oxygen and sediment concentration from the years 1990 to 1999 were used to develop both spatial and non-spatial SDMs for each species. These species-environment relationships along with data on future environmental conditions for the years 2050-2059 derived from three separate GCM scenarios were then used to predict the future distribution of each species. Predictions from spatial and non-spatial models will be discussed with a focus on the difference in area and the amount of overlap between current and future distributions of each species.