Robust allocation of fire management resources towards threatened species conservation under uncertainty

Various strategies have been proposed to contend with the threat posed by increased fire frequencies to fire-adapted species such as those found in Mediterranean-type ecosystems. Biophysical models for fire-sensitive population dynamics coupled with decision-theoretic methods have presented a promising approach for guiding the allocation of available management resources. However, uncertainty is often disregarded in these models, as model parameters are fixed at baseline values assumed to represent the best, or most likely, estimates. Here, we present an approach for optimizing the spatial allocation of limited resources towards fire management to maximize the robustness to uncertainty of a desired species-specific conservation outcome. We performed a global evaluation of robustness — one which weighs possible model parameterizations according to their probability — facilitated by a Bayesian analysis of population dynamics to simulate a proactive approach to management. As a case study, we considered the obligate-seeding plant Hesperocyparis forbesii (Tecate cypress), native to Southern California, and assumed a simple, hypothetical tradeoff between area and magnitude of managed change in fire return intervals. With a target outcome of stabilizing population dynamics within a relatively small portion of the species range, we found that resource allocation was most effective over a few locations at relatively long expected fire return intervals. Conversely, with more ambitious outcomes, the most robust approach was achieved by concentrating management on locations with relatively short fire return intervals. Our analysis not only emphasizes the importance of a rigorous accounting of uncertainty to inform fire management decisions but also provides a quantitative approach generalizable to other disturbances, conservation objectives, and environmental dependencies.