Balancing evidence and reducing uncertainty in the evaluation of reintroduction outcomes in ‘alalā, the Hawaiian crow

Over the course of conservation translocations, evaluating release outcomes can be critical for shaping future plans. However, reintroduction programs are often constrained in the number of animals they can release, which reduces the potential for experimental insight. Recent releases of the extinct-in-the-wild ‘alalā (Corvus hawaiiensis) highlight this dilemma. As the last remaining corvid species of the Hawaiian Islands, ‘alalā are critical seed dispersers, yet currently survive only in a small population within conservation breeding facilities. In our efforts to understand the factors contributing to the successes and failures of the 2016–2020 ‘alalā release effort, we explored various datasets and analyses to evaluate which management actions and ecological factors contributed to post-release mortality outside of an experimental framework. Specifically, we explored competing hypotheses surrounding the influence of social competition, supplemental food reduction, drawbacks of supplemental feeders, and predation on the outcome of recent releases. Using a multifaceted post-release monitoring dataset, we were able to reduce some of the uncertainty surrounding where and why ‘alalā died and what may have made them more vulnerable to mortality. We found that mortality risk was highest during times of supplemental food reduction and increased as more ‘alalā transitioned to the territorial behavior characteristic of adulthood. However, the mechanism for either risk factor remains uncertain, because low body weight, and changes to space use or food consumption did not explain why many birds ultimately died from predation, were not recovered, or suffered from multifactorial deaths (e.g. poor condition or infection). Nevertheless, these analyses highlight that future changes to the supplemental food reduction and release strategy could improve outcomes. In the context of passive adaptive management, release efforts can still learn from management actions, given sufficient monitoring of small release cohorts.