The assessment of the impacts of climate change on freshwater biodiversity lags far behind those on terrestrial and marine systems despite the vulnerability and importance of freshwaters to humans. Here we introduce a new project aiming at combining statistical hydrological and ecological models with empirical approaches to produce a comprehensive assessment of the risks of climate warming to freshwater biodiversity across representative river networks in Japan. Currently in its first year, we focus our presentation on the initial phase of the project consisting on the establishment of air and water temperature monitoring networks in target catchments and the subsequent development of statistical models for prediction of stream temperatures based on air temperatures and landscape predictors related to air-water heat exchange processes. Preliminary results from simple linear models show already good performance in reproducing daily average temperatures in monitored streams, with over 83.2% of variance explained and a reasonable out-of-sample prediction accuracy with mean standard error of 1.32 °C. These results suggest a large scope for improvement in the following years of the project as more data become available and we test more complex models. |