Understanding future fish stock recruitment has long been a challenge for fisheries management - and with little success: today, recruitment is widely viewed as “unpredictable". Accurate predictions could be highly valuable aids in the sustainable management of marine fish stocks, especially for short-lived species. However, recruitment forecasting also lacks a clear definition of what success would look like. How would we know when the update of a forecast system is justified from both scientific and economic perspectives?
The latest Editor's Choice article in ICES Journal of Marine Science, presents a framework to assess recruitment models, allowing the skill and value of recruitment forecasts to be assessed in a manner that is relevant to their potential use in an operational setting.
The framework is demonstrated using predictions generated by a multi-model ensemble of generalized additive models that use multiple demographic and environmental factors and model variable selection based on a multi-model inference approach.
Predictions of both continuous and categorical (high, medium, or low) recruitment strengths are evaluated using retrospective forecasting. This is similar to techniques used in the atmospheric sciences and reflects the utility of the system in a realistic setting. Validation metrics are selected to highlight the different forecast needs of potential end-users and are always presented relative to a baseline forecast.
A quantitative value assessment is performed using an economic cost-loss decision model, providing insight into the actual monetary value of the forecast product.
The framework is highly flexible and can be applied to other analytically managed stocks, models, or data sources, providing an easy tool for comparing and producing recruitment prediction.
The authors applied the framework to lesser sandeel (Ammodytes marinus), an ecologically and economically valuable species in the North Sea. Predictability is seen in two of the four stocks, with both continuous and probabilistic validations performing on par or better than reference models. The value assessment of management area 1r (see map) shows positive value gains in all tercile categories, indicating a potential of long-term value gains from following the forecasts.
These results shows how operational recruitment forecasts can be assessed and their potential value as a tool for both managers and industrial uses.