Scientific inference and experiment in Ecosystem Based Fishery Management, with application to Steller sea lions in the Bering Sea and Western Gulf of Alaska

Learning about ecosystem processes and patterns is an essential component of Ecosystem Based Fishery Management and the sustainable use of natural resources. Currently, such learning is usually done through adaptive management (passive or active) or Management Strategy Evaluation, which are explained. An example of adaptive management in northwestern Australia shows the strengths and limitations of management experiments and raises the question of how to learn if an experiment is not practicable. Both adaptive management and Management Strategy Evaluation are examples of scientific inference, an idea introduced by Sir Harold Jeffreys nearly 80 years ago. With sufficient variation, even if it is not through controlled experiments, scientific inference is possible by combining mechanistic models with statistical methods; the recently proposed paradigm of 'adaptive monitoring' is another case of scientific inference. The decline of Steller sea lions in the Bering Sea and Aleutian Islands is reviewed, including the only work in which 10 hypotheses concerning the decline were simultaneously compared. It is concluded that scientific inference using mechanistic models and fine scale data at the level of the rookery can provide useful information about the interactions of fisheries, fish populations, and Steller sea lions.