AbstractObjectiveMixed‐data models incorporating remote antenna detections from PIT‐tagged fish together with physical capture data (mixed data) can improve precision of mark–recapture abundance estimates, particularly for spawning fish. However, if the entire population is not available for physical capture during the mark–recapture sampling period, abundance estimates will be biased low. Our objectives were to examine bias and precision of two modeling approaches and develop a simple diagnostic to determine whether the entire population was sampled.MethodsWe use a simulation modeling approach to compare specified abundances with abundance estimates from closed‐capture models using mixed data (the mixed‐data approach) and using only marked (e.g., PIT‐tagged) individuals divided by the proportion of individuals that are marked (proportional approach). We use the difference in bias between the two models as our diagnostic of whether a population was randomly sampled. We then applied our diagnostic to two case studies: a spawning population of adfluvial June Sucker Chasmistes liorus in Utah Lake, Utah, 2008–2020, and a spawning population of Bull Trout Salvelinus confluentus in the Imnaha River, a tributary of the Snake River, Oregon, 2013–2020.ResultSimulation experiments revealed that the mixed‐data approach became increasingly negatively biased as the proportion of the population that was unavailable for physical capture increased, yet the proportional approach remained unbiased. Abundance estimates from the proportional approach averaged approximately 6× greater for June Sucker and 2× greater for Bull Trout compared to the mixed‐data approach, suggesting a proportion of the population was not available for physical capture in both case studies.ConclusionUnderstanding the magnitude of bias in abundance estimates is particularly important for the management of imperiled species that are subject to recovery plans. Comparing estimates using the unbiased proportional approach that we described here with those from a mixed‐data approach when PIT tag detection data are used to estimate abundance is a straightforward method to evaluate bias in these estimates.