Evaluating forest restoration strategies after herbivore overbrowsing

Abstract

The continued degradation of forests due to natural and anthropogenic disturbance has led to loss of ecosystem function, biodiversity change and in extreme cases, the extinction of endemic species. Within degraded landscapes, forests are failing to regenerate naturally, resulting in a loss of key species and producing persistent open canopied patches within previously closed canopied systems. Abundant ungulates can act as a disturbance and hinder natural forest regeneration when they severely overbrowse vegetation, and it is unclear how to restore natural forest communities in these cases. In Newfoundland (Canada), abundant non-native moose following natural disturbance have suppressed balsam fir advanced regeneration producing alternate stable states termed “spruce moose savannas”. Our objective was to evaluate several scenarios for restoring moose impacted boreal forests. To achieve this objective, we integrated data from field observations and experiments, aerial photographs and drone imagery to parameterize mathematical models of boreal understory and canopy regeneration in balsam fir dominated forest in eastern Newfoundland. We used simulations to evaluate the effect of reduced browsing pressure and seedling planting on boreal forest regeneration. Model outcomes suggest active restoration via planting birch and balsam fir seedlings is required to restore the understory and canopy vegetation to its natural state in large canopy gaps (>5 ha), and any planting should be done under low moose browsing pressure or within moose exclosures. In contrast, in medium canopy gaps (<5 ha), passive restoration via moose reduction is sufficient to restore balsam fir but birch seedling planting is the most effective for restoring birch in all canopy gap sizes. We propose that Markov models parameterized by aggregate data with simulated herbivory be used to support experimental studies and strengthen evidence for forest restoration planning. Our novel modelling approach can be used in other herbivore-disturbed systems or to simulate other disturbances to predict restoration success under various disturbance regimes.