A Watershed Management Decision-Support Tool to Predict Juvenile Herring Out-Migration Loss Using a Process-Based Hydrologic Model.

King, Katherine*¹, James Knighton¹, Eric Shultz², Michael Burgess², ¹University of Connecticut, Natural Resources and the Environment, Storrs, CT, ²University of Connecticut, Ecology and Evolutionary Biology, Storrs, CT

River herring serve valuable ecological and economic roles in freshwater streams, estuaries, and ocean ecosystems. The migration between fresh and saltwater is a key life stage of river herring controlled by environmental and physiological factors. Past studies demonstrate stream discharge is important in determining the timing and magnitude of downstream migration by juveniles; however, there is little exploration into the influence of weather variables on flow regimes in small watersheds where river herring reside. We apply SWAT+ models of stream discharge and a distributed random forest (DRF) model to understand how weather variables impact low flows in river herring habitat. Using a flow threshold below which juvenile outmigration is not successful, we created a tool to predict the probability of downstream migration loss during the migration season (April - November) based on year-to-date weather data in our study watershed. We calibrated a SWAT+ model of stream flow using observed daily discharge. We used calibrated parameters to simulate 9,999 years of daily using SWAT+ weather generator data. Using these outputs, we implemented a DRF model to determine weather variables most predictive of migration loss and found reservoir depth and preceding monthly precipitation were among the most predictive. We used this DRF model to predict the probability of migration loss in the year based on year-to-date weather data. With the goal of maintaining migration throughout the season, water resource managers will use this tool to make choices about impoundment releases and water withdrawals, particularly in years with a high risk of migration loss.