Multi-level habitat selection of boreal breeding mallards

Canada's western boreal forest is a vital breeding habitat for North American duck populations. This region has experienced considerable demand for its valuable natural resources (e.g., oil and gas, forestry) resulting in extensive industrial development (e.g., infrastructure), which is predicted to continue. The potential effects of industrial development on breeding ducks in the western boreal forest, however, remains largely unexplored. We used backpack harness global positioning system (GPS) transmitters to document habitat selection by breeding female mallards (Anas platyrhynchos) across a gradient of industrial development in the western boreal forest of Alberta, Canada. We modeled breeding home range (second order) selection and habitat selection within the home range (third order) using resource selection functions, and spatially predicted our models across the landscape to highlight important breeding habitat. Contrary to our predictions, breeding female mallards did not avoid all industrial development at the second and third orders. Females established home ranges (second order) with greater proportions of marsh, graminoid fen, and well pads, and lower proportions of forest. Within their home range (third order), females selected shrub swamp, graminoid fen, marsh, well pads, and borrow pits, and avoided open water, swamp, treed peatland, forest, forest harvest areas, and industrials (e.g., buildings). Females also selected habitat that was close to pipelines and roads. Our results suggest that the magnitude and direction of breeding season habitat selection by female mallards varies depending on the scale and landscape features, but current levels of industrial development within our study area still allowed for the establishment of breeding season home ranges. Our spatially predicted maps contribute to the increasing body of work surrounding boreal waterfowl ecology by highlighting important breeding habitat for mallards in Canada's western boreal forest.