Temperature and flow models fitted to 1950–1993 Bonneville Dam counts did not predict 1994–2025 run timing: sockeye came about 4 days earlier and spring Chinook 5–8 days later

Question: do relations between river conditions and salmon and shad migration timing, fitted to the period that Quinn & Adams (1996) studied, predict the 32 later years? Approach: we used daily adult counts at Bonneville Dam on the Columbia River (1939–2025) and at three upstream dams. We calculated the median passage date of five runs. We fitted models of median date on river temperature and discharge (30-day window, or the date of a fixed isotherm) to 1950–1993 or 1970–1993, and we predicted 1994–2025. Because the scroll-case and USGS temperature records disagree by up to 0.8 °C in 5-year means, we repeated each test with two temperature series. Results: sockeye passed 2.7–4.3 days earlier than all 16 model variants predicted, and the error replicated at three upstream dams. Their year-to-year response to river temperature stayed weak (0.06–0.29 days per day of isotherm date). Spring Chinook passed 4.8–8.3 days later than all 16 variants predicted, although the river was warmer; the best single break is at 2005, and the error replicated at The Dalles and McNary. River conditions predicted 0–34% of the shifts of these two runs. The strong 1950–1993 coupling of shad timing to the 17 °C isotherm date (0.33–0.80 days per day) became weaker after 1993 (0.14–0.39), and the shad advance stopped. Shad had met cooler water each decade to 1993, as Quinn & Adams reported, but met warmer water after 1994. The share of sockeye that passed at 20 °C or more rose to 14–15% in 2010–2025. Fall Chinook timing did not change (negative control). Limitations: the temperature series disagree on long-term warming, so shad results depend partly on the series; counts are daylight counts in fixed run windows; we did not test causes such as stock composition, fisheries, predation, ocean conditions or evolution.

phenology · migration timing · Pacific salmon · American shad · sockeye salmon · Chinook salmon · Columbia River · Bonneville Dam · river temperature · out-of-sample test · climate change · fish passage counts