Science Advisory Report 2020/004

Development of a Species Distribution Modelling Framework and its Application to Twelve Species on Canada's Pacific Coast

Summary

• Species distribution models (SDMs) are relevant to several national objectives related to marine spatial planning, vulnerability assessments, emergency response, and stock assessment. Benefits to Fisheries and Oceans Canada (DFO) of effectively applying SDM methods include explicit consideration of ecological and management contexts, consistent preparation of available data, and application of appropriate analytical methods to construct and evaluate models.
• DFO Pacific Science Branch has proposed a framework that includes a set of guidelines for the development of consistent, interpretable, and defensible SDMs.  The framework can be implemented with purpose-built scripts written in the R statistical programming language (R Core Team 2018).The framework is intended to support practitioners by following current best practices and providing guidance on key aspects of data preparation, model fitting and evaluation, uncertainty estimation, and interpretation of results.
• An application of the framework is illustrated by applying three modelling methods of increasing complexity to twelve benthic species, for two study areas on Canada's Pacific coast (Figure 1). The resulting predictions are evaluated using standardized performance metrics, and diagnostic plots and maps, including the relative importance and marginal effects of contributing predictors. The application demonstrates the importance of a consistent data model, building multiple models including a knowledge-based model, and uncertainty estimation.
• SDMs for the twelve species are intended to support oil-spill response planning and preparedness. The models generally fit the data well as indicated by the performance metrics. The appropriateness of these model predictions for uses beyond oil spill response will need to be assessed by the user, based on clearly identified model objectives defined by their particular application.
• Sources of uncertainty arise from methodological choices, and the availability and quality of data. Uncertainty in the model predictions is assessed with four different methods that reduce or measure uncertainty spatially across a study area. Additional sources of uncertainty are considered by following best practices for data selection where practical.
• The framework is a scientifically defensible, transparent, and reproducible method for producing SDMs. It can be used to produce additional species models, or adapted for other modelling extents.
• Several recommendations for the successful development and application of SDMs arose from the framework.

This Science Advisory Report is from the June 11-12, 2019 regional peer review on Habitat Suitability Modelling Best Practices for Canada's Pacific Ocean. Additional publications from this meeting will be posted on the Fisheries and Oceans Canada (DFO) Science Advisory Schedule as they become available.

Science Advisory Report 2020/004 (PDF, 2,130 KB)