Search

Landscape complexity can determine the population dynamics of interacting predators and prey. Yet, management plans are commonly developed from aspatial predictive models. This oversight may result in unexpected outcomes or the loss of opportunities to make spatial interventions that would increase a plan's effectiveness. The management of the threatened woodland caribou (Rangifer tarandus caribou), boreal population, provides an example of such shortcomings when using an aspatial approach. Currently, the most influential management recommendation is to maintain at least 65% of undisturbed forests in areas occupied by caribou populations, regardless of the spatial configuration of the forest cover. Using a spatially explicit individual‐based model (IBM), we evaluated the effects of the spatial configuration of cuts and roads on the mortality of boreal caribou living in sympatry with wolves (Canis lupus) and moose (Alces alces), an apparent competitor. Starting with a real forest landscape, we created forest management scenarios of the specific spatial distribution of cuts (mosaic, small, or large agglomeration) with increasing disturbance levels. We then ran the IBM with simulated agents, representing individuals of the three species, moving according to movement rules determined from radio‐collared individuals. We found that movement responses to land cover types and roads differed among species. For example, caribou and moose generally avoided areas close to roads, contrary to wolves. Those differences influenced the mortality of caribou agents, which not only depended on the levels of disturbance but also depended on the spatial distribution of cuts and roads. After controlling for disturbance level, wolves were more successful when forest management required an extensive road network resulting in relatively high habitat fragmentation. Caribou agents experienced lower mortality in landscapes with low densities of road and disturbance‐related edges. The effect remained much stronger, however, for the level than the spatial configuration of human disturbances. Still, our IBM demonstrated how landscape management could be used to manipulate species interactions, with the intent of either increasing or decreasing predation rates on specific populations, depending on management goals.

Primary production can determine the outcome of management actions on ecosystem properties, thereby defining sustainable management. Yet human agencies commonly overlook spatio-temporal variations in productivity by recommending fixed resource extraction thresholds. We studied the influence of forest productivity on habitat disturbance levels that boreal caribou – a threatened, late-seral ungulate under top-down control – should be able to withstand. Based on 10 years of boreal caribou monitoring, we found that adult survival and recruitment to populations decreased with landscape disturbance, but increased with forest productivity. This benefit of productivity reflected the net outcome of an increase in resources for apparent competitors and predators of caribou, and a more rapid return to the safety of mature conifer forests. We estimated 3-fold differences in forest harvesting levels that caribou populations could withstand due to variations in forest productivity. The adjustment of ecosystem provisioning services to local forest productivity should provide strong conservation and socio-economic advantages.

Conditional logistic regression (CLR) is widely used to analyze habitat selection and movement of animals when resource availability changes over space and time. Observations used for these analyses are typically autocorrelated, which biases model-based variance estimation of CLR parameters. This bias can be corrected using generalized estimating equations (GEE), an approach that requires partitioning the data into independent clusters. Here we establish the link between clustering rules in GEE and their effectiveness to remove statistical biases in variance estimation of CLR parameters. The current lack of guidelines is such that broad variation in clustering rules can be found among studies (e.g., 14-450 clusters) with unknown consequences on the robustness of statistical inference. We simulated datasets reflecting conditions typical of field studies. Longitudinal data were generated based on several parameters of habitat selection with varying strength of autocorrelation and some individuals having more observations than others. We then evaluated how changing the number of clusters impacted the effectiveness of variance estimators. Simulations revealed that 30 clusters were sufficient to get unbiased and relatively precise estimates of variance of parameter estimates. The use of destructive sampling to increase the number of independent clusters was successful at removing statistical bias, but only when observations were temporally autocorrelated and the strength of inter-individual heterogeneity was weak. GEE also provided robust estimates of variance for different magnitudes of unbalanced datasets. Our simulations demonstrate that GEE should be estimated by assigning each individual to a cluster when at least 30 animals are followed, or by using destructive sampling for studies with fewer individuals having intermediate level of behavioural plasticity in selection and temporally autocorrelated observations. The simulations provide valuable information to build reliable habitat selection and movement models that allow for robustness of statistical inference without removing excessive amounts of ecological information.