Response of stream ecosystems to climate change (I): Linking invertebrate biodiversity to hydrologic and thermal alteration

Background/Question/Methods

Climate change is projected to significantly alter the biodiversity of most ecosystems. In this project, we linked climate, temperature, hydrologic, and species distribution models to examine how stream invertebrate biodiversity in the conterminous USA might respond to climate change (A2 emission scenario) by the end of the 21^st century. We first downscaled current (2000) and future (2100) climate predictions to an ecologically relevant spatial resolution (4 km) and then developed Random Forest models to predict local stream temperature and hydrologic regime from climate and watershed attributes. The predictions of these models and channel slope were then used in a Random Forest, multi-taxon niche model (MNM) to predict how site-specific probabilities of capture (pc) of 539 benthic invertebrate taxa would change at 1197 stream reaches by 2100.

Results/Conclusions

Current spatial variation in taxa composition was associated with variability in daily stream flow, mean summer stream temperature (MSST), date of 50 percent annual flow (50PCFlowDate), flow predictability, and channel slope. 354 taxa were predicted to increase in frequency of occurrence, and 185 taxa were predicted to decrease. On average, assemblages were predicted to become slightly more similar to one another (2000 and 2100 Bray-Curtis dissimilarities (BCD) = 0.42 and 0.36). However, some sites were predicted to change markedly in assemblage composition (maximum change in BCD = 0.61), whereas other sites were predicted to experience no change in composition. Some sites were predicted to lose taxa (up to 18%), and other sites were predicted to gain taxa (up to 17%). Only 5 taxa were predicted to go regionally extinct. Sites differed in their vulnerability to predicted climate change, but vulnerability differed depending on how likely taxa were predicted to occur at sites. When vulnerability was based on core taxa (pc > 0.5), taxa loss occurred and was most severe at sites with initially low flow variability, low temperatures, high 50PCFlowDate, and high flow predictability. When all taxa were considered, assemblages tended to gain taxa under these same initial conditions.