Freshwater streams are critical resources that provide benefits to humans and natural systems. As climate becomes more extreme, changes to the hydrologic cycle and surface air temperatures will affect the health of aquatic ecosystems, individual biota in the system, and their relationship to human uses of freshwater. Understanding of the vulnerability of stream ecosystems to climate change is critical to ensure their continued health and protection. Therefore, the goal of this study was to develop a modeling process to assess the impacts of climate change on fish and macroinvertebrate measures of stream health, as represented by four measures: the number of Ephemeroptera, Plecoptera, and Trichoptera (EPT) taxa, Family Index of Biotic Integrity (FIBI), Hilsenhoff Biotic Index (HBI), and fish Index of Biotic Integrity (IBI). The research objectives were to: (1) identify sets of influential in-stream variables for stream health modeling, (2) develop widely applicable large-scale stream health models, but with reach-scale resolution appropriate for natural resources decision-making, and (3) identify vulnerable stream ecosystems at risk of declining stream heath due to climate change. The framework was developed for seven watersheds that encompass cold, cold transitional, cool, and warm stream thermal classes in Michigan. This process linked Soil and Water Assessment Tool (SWAT) hydrological models, selection of ecologically relevant in-stream variables, adaptive neuro-fuzzy inference systems (ANFIS) stream health models, and an ensemble of climate models and representative concentration pathways. A stream temperature model was also developed The Bayesian variable selection technique was identified as superior to Spearman’s Rank Correlation and Principal Component Analysis as the best method for selecting influential in-stream variables. A few key flow regime variables, mostly related to timing and duration of major low and high flow events, played a significant role in dictating the health of streams in the studies area. Building ANFIS stream health models based on stream thermal class improved their performance. The best stream health models were suitable for performing large-scale impacts assessments at the individual reach level necessary for site-specific decision-making. Extending the stream health models into the future, the process was repeated with a climate model ensemble from the Coupled Model Intercomparison Project Phase 5 (CMIP5) to compare a control period of 1980-2000 to 2020-2040. The overall impacts of climate change on stream health across stream thermal regimes were low in terms of magnitude of stream health decline. However, at the reach level there were many streams with high probability of declining stream health coupled with large projected declines in stream health, revealing highly vulnerable aquatic communities. By combining the probability and magnitude of declining stream health, decision-makers can target stream ecosystems that are critically at-risk due to climate change.
Read
