Integrating hydrodynamic and predictive ecological modelling to forecast the effects of climate change on urban...

Urban streams are exposed to the combined effects of climate change and persistent anthropogenic pressures, yet predicting ecological responses under future conditions remains challenging. This study evaluated how projected climate scenarios may alter macroinvertebrate communities in urban streams by coupling hydrodynamic simulations with a predictive machine learning model. A regional dataset from central Portugal was used to develop a HYDRA model predicting macroinvertebrate taxon occurrence based on environmental variables. Hydrodynamic models were used to simulate flow variables under four climate change scenarios representing medium-term (2041-2060) and long-term (2081-2100) projections under contrasting emission pathways. The models were then applied to forecast changes in macroinvertebrate assemblages at 14 urban stream sites in Coimbra, Portugal. Climate projections indicated consistent warming and intensification of extreme rainfall events, resulting in higher simulated spring baseflows. Predicted community responses varied among sites and were influenced by baseline urbanisation levels. Less urbanised streams, which supported richer and more sensitive assemblages, experienced greater declines in family richness, IBMWP scores, and the proportion of EPT taxa under higher emission scenarios. In contrast, highly urbanised streams showed weaker or neutral responses, likely constrained by prior biological simplification. Taxonomic changes were characterised by losses of cold-adapted and flow-sensitive taxa and persistence or colonisation of tolerant and generalist taxa. Overall, although less urbanised streams showed stronger ecological declines, they still maintained higher biological quality than more degraded systems, highlighting the importance of conserving and restoring urban streams to enhance their resilience to climate change.