Predicting the effects of restoring linear features on woodland caribou populations

[Display omitted] •We present a model of apparent competition, parametrized using empirical data, to predict habitat restoration efficacy.•Habitat restoration has the potential to increase caribou densities by up to 2.51-fold.•Increases in caribou density can occur with minimal changes to wolf or moose density.•Incorporating the numerical feedback into simulations of foraging efficiency increased our understanding of the total response.•Simulated predator reductions increased caribou densities by more than twice the total response of restoration, suggesting restoration should be coupled with other direct management actions. Predator-prey dynamics are increasingly being modified by the alteration of natural habitats. Such alteration has led to increased predation rates and local extirpation of woodland caribou (Rangifer tarandus caribou) in western Canada. Linear features such as roads or seismic lines (narrow corridors used for petroleum exploration that are cleared of vegetation) increase predation rates on caribou by increasing wolf (Canis lupus) movement rates and by facilitating access into caribou habitat. Linear feature restoration is therefore hypothesised to help reverse caribou declines. However, with the high financial cost to restore approximately 350,000 km of seismic lines within western Canada’s boreal forests, theoretical predictions can clarify the efficacy of such actions. We use a mathematical model based on coupled ordinary differential equations representing predator-prey dynamics to estimate equilibrium densities of caribou, moose (Alces alces) and wolves under various parameter scenarios. Changes in equilibrium density serve as a proxy for the expected effect of linear feature restoration on population densities. Our model captures dynamical feedbacks between caribou and wolf densities, and also includes moose, which are the wolf’s primary prey species. With our best estimates of parameter values, caribou density increased 2.51-fold if all linear features were restored and 1.61-fold if only seismic lines were restored. As a comparison, simulated predator reductions increased caribou densities 3.92-fold, nearly twice the total response of linear feature restoration. The effect of restoration was increased if caribou group size was smaller, yet was less pronounced if carrying capacity for ungulates was higher. By varying the parameter values and fixing population densities, our approach allowed us to partition the caribou populations’ r