Predicting the impact of fire on a vulnerable multi-species community using a dynamic vegetation model

Abstract

Conservation management under human-induced changes to disturbance requires tools that can balance the needs of multiple species with different life histories and habitat requirements. Despite this urgent conservation need, landscape management typically focuses on single species and rarely includes the influence of disturbance-dependent vegetation transitions on multiple target species. In this paper, we describe a simulation model that achieves these goals, ranking possible fire management strategies from the viewpoint of protecting endangered coastal Southern Californian wildlife. The model involves the direct and indirect effects of fire on four animal species of conservation concern (coastal cactus wren, California gnatcatcher, Stephens’ kangaroo rat, and Pacific pocket mouse) and five vegetation types (grass, coastal sage scrub, obligate seeding and resprouting chaparral, resprouting-only chaparral, and woodlands). Using historical fire records for the region, we predicted spatially-explicit fire frequencies and ignition probabilities. For these predictions, we simulated the location and extent of fires. Combining fire history and vegetation transition data from 1933 to 2003, we specified vegetation change probabilities under simulated fire regimes. Fire occurrence in a location altered habitat suitability, directly for each of the animal species and indirectly by changing the vegetative community. For some open-habitat species, such as the Stephens’ kangaroo rat and Pacific pocket mouse, fairly frequent fire is required to reduce the density of invasive grasses and herbs. For other species, such as the coastal cactus wren and California gnatcatcher, frequent fire destroys the mature coastal sage scrub on which these species depend. The model includes a management component, allowing us to rank fire management actions. Over a 50-year time horizon, we find that populations of California gnatcatchers and Pacific pocket mouse are highly variable, and the pocket mouse is particularly prone to decline, despite prescribed burns designed to boost population viability. California gnatcatchers were also likely to be extirpated in the model, with relatively small extirpation risks for the cactus wren and Stephens’ kangaroo rat. Despite conflicting requirements with respect to fire and differing life history traits among the four animals, we identified a beneficial strategy for our four target species, namely, controlling fire in coastal sage scrub.