Mon, Aug 15, 2022: 5:00 PM-6:30 PM
ESA Exhibit Hall
Background/Question/MethodsThe integration of bird conservation objectives into forest management planning can benefit from the use of categorical models of bird density for use in forest optimization models. A barrier to this can be the time-consuming and costly need for data collection and modelling of management areas of interest. Here, we aimed to produce and assess a piecewise smoothing method of building categorical models of bird density for a given study area using pre-existing, national-scale models of bird density from the Boreal Avian Modelling (BAM) Project. Categorical models of bird density were built for a single factor, (forest/land cover class) or for two factors (age and forest class). The method was tested in two study areas, and built within an open-source and readily reproducible framework that can be applied to any managed forests in Canada.
Results/ConclusionsThe residuals between the BAM national models and maps produced by the method’s predictions were overall small, and the general patterns of higher and lower density across the study landscape were replicated. However, the extremes of bird densities were not captured. Few of the classes met assumptions of unimodality, suggesting that other spatially varying factors were not accounted for. Adding the second factor (i.e., 10-year age classes) led to small but significant improvements over our suite of test statistics. This may reflect relatively homogeneous age structure within the study region. Over an appropriate scale of study area with carefully selected classes, the piecewise smoothing method shows promise in providing a rapid, cost-effective and accessible way to predict and plan for bird conservation outcomes in forest management anywhere in Canada. Future work should further investigate variation in parameters for model improvement, and demonstrate a workflow for the models in optimization models for integrated forest management and bird conservation.
Results/ConclusionsThe residuals between the BAM national models and maps produced by the method’s predictions were overall small, and the general patterns of higher and lower density across the study landscape were replicated. However, the extremes of bird densities were not captured. Few of the classes met assumptions of unimodality, suggesting that other spatially varying factors were not accounted for. Adding the second factor (i.e., 10-year age classes) led to small but significant improvements over our suite of test statistics. This may reflect relatively homogeneous age structure within the study region. Over an appropriate scale of study area with carefully selected classes, the piecewise smoothing method shows promise in providing a rapid, cost-effective and accessible way to predict and plan for bird conservation outcomes in forest management anywhere in Canada. Future work should further investigate variation in parameters for model improvement, and demonstrate a workflow for the models in optimization models for integrated forest management and bird conservation.