Effects of water temperature and flow on adult salmon migration swim speed and delay

The effects of temperature and flow on the migration of adult Chinook salmon Oncorhynchus tshawytscha and steelhead O. mykiss through the Columbia River hydrosystem were determined with a novel technique that fits a broken linear model of swim speed versus temperature and flow by partitioning data into speed ranks. Using the migration times of passive integrated transponder (PIT)-tagged adult Chinook salmon upstream between Bonneville and Lower Granite dams (462 km) over the years 1998-2002, we found that a maximum swim speed of about 1 body length/s occurred at 16.3 degrees C. Speed was less above and below this optimum temperature. For PIT-tagged steelhead, migration speed uniformly decreased with increasing temperature, suggesting that the fish migrated at temperatures above the optimum. Migration delay was also a unimodal function of temperature, the minimum delay occurring around 16-17 degrees C. The broken linear model was compared with seven alternative models of unimodal and monotonic speed versus temperature and flow. The unimodal models fit the data better than the monotonic models (when ranked by the Akaike information criterion), and the broken linear model fit the data best. Flow was insignificant in all of the monotonic models and only marginally significant in the unimodal models. The findings of this study have significance in evaluating the effects of hydrosystem operations and climate change on salmon and steelhead fitness.