Reply to Haeseker: Value of controlled scientific experiments to resolve critical uncertainties regarding Snake River salmon survival.

In our report (1), we set out to explicitly control for the ecological differences Haeseker (2) cites so that we could assess the effect of a critical policy issue: whether Snake River dam passage results in poorer early marine survival of juvenile Snake River spring Chinook salmon. Thus, we selected smolts of common size and manipulated release times to ensure smolts from the two populations were as similar as possible, apart from the number of dams that they passed (1). We agree with Haeseker that ecological differences between the populations used in our study existed and may have influenced ocean survival; however, their net effect needed to be a 3.4-fold difference in survival to result in the nearly identical rates of apparent survival that we found. We have since repeated the experiment on salmon collected and tagged at Snake and Columbia River dams and compared their postrelease survival (3). The findings were consistent with the results reported in our article in PNAS (1): Snake River spring Chinook salmon ≥130-mm fork length did not have lower survival relative to salmon originating elsewhere. (It is now technically possible to repeat these tests on smaller wild smolts if policy makers deem it sufficiently important). Haeseker’s (2) claim concerning the ocean distribution of salmon smolts is likely unfounded: long-term ocean surveys have consistently captured juvenile Columbia River spring Chinook almost exclusively on the continental shelf north of the Columbia River (4). Furthermore, the cross-shelf distribution plots we report (figure S2 in ref. 1) demonstrate that both of the populations used in our study were shelf-limited at Lippy Point. The survival models we use thus accounted for individuals temporarily carried south or off-shelf in the Columbia River plume. Finally, because our study estimates relative survival, precise estimation of detection probability is not critical unless enough of the Yakima population migrated offshore to reduce the number detected to equal that of the Snake River smolts. As Hilborn noted in his commentary on our report (5), no amount of data are likely to resolve the gulf between ecologists arguing for a major delayed effect of Columbia River dams on ocean survival and those who do not. Many in the Columbia River Basin blame poor ocean survival on prior exposure to dams in freshwater; however, Chinook populations from undammed areas in British Columbia and Alaska have declined in recent years as well (1). Psychological studies repeatedly show that individuals and like-minded groups preferentially select those facts favoring their prior prejudices when presented with complex data capable of multiple interpretations (6), such as those in the correlation analyses cited by Haeseker (2). Without carefully designed scientific experiments that test specific variables, it may not be possible to break out of this dilemma. In other scientific fields, formal experimental tests of theories historically resulted in very rapid scientific progress. The stakes are high in the Columbia River region; the window for resolving the salmon conservation problem is likely closing fast, given the large predicted changes in future climate and poor ocean survival of salmon that will likely ensue.