1 1. Assessing the degree to which at-risk species are regulated by density dependent versus 2 density independent factors is often complicated by incomplete or biased information. If not 3 addressed in an appropriate manner, errors in the data can affect estimates of population 4 demographics, which may obfuscate the anticipated response of the population to a specific 5 action. 6 2. We developed a Bayesian integrated population model that accounts explicitly for interannual 7 variability in the number of reproducing adults and their age structure, harvest, and 8 environmental conditions. We apply the model to 41 years of data for a population of threatened 9 steelhead trout Oncorhynchus mykiss using freshwater flows, ocean indices, and releases of 10 hatchery-born conspecifics as covariates. 11 3. We found compelling evidence that the population is under strong density dependence, despite 12 being well below its historical population size. In the freshwater portion of the lifecycle, we 13 found a negative relationship between productivity (offspring per parent) and peak winter flows, 14 and a positive relationship with summer flows. We also found a negative relationship between 15 productivity and releases of hatchery conspecifics. In the marine portion of the lifecycle, we 16 found a positive correlation between productivity and the North Pacific Gyre Oscillation. 17 Furthermore, harvest rates on wild fish have been sufficiently low to ensure very little risk of 18 overfishing.19 4. Synthesis and applications. The evidence for density dependent population regulation, 20combined with the substantial loss of juvenile rearing habitat in this river basin, suggests that 21 habitat restoration could benefit this population of at-risk steelhead. Our results also imply that 22 hatchery programs for steelhead need to be considered carefully with respect to habitat 23 3 availability and recovery goals for wild steelhead. If releases of hatchery steelhead have indeed 24 limited the production potential of wild steelhead, there are likely significant tradeoffs between 25 providing harvest opportunities via hatchery steelhead production, and achieving wild steelhead 26 recovery goals. 27 28 Managing at-risk species requires an understanding of the degree to which population 29 dynamics are self-regulated versus driven by external factors. However, the data used to identify 30 potentially important density-dependent and population-environment relationships are rarely, if 31 ever, fully comprehensive or error free. Rather, imperfect detection, misidentification, and non-32 exhaustive sampling all lead to a somewhat distorted view of the true state of nature. For 33 example, when not addressed in an appropriate manner, errors in population censuses may cause 34 underestimates of recruitment (Sanz-Aguilar et al. 2016) or overestimates of the strength of 35 density dependence (Knape & de Valpine 2012). Similarly, imprecision in the estimated age 36 composition of the population also biases the estimated strength of density depe...