Although scientists can successfully assess many fisheries, some populations are harder to model. Short-lived species in variable environments such as market squid and the Pacific sardine can vary wildly from year to year in both location and reproduction levels, making it difficult to manage them with traditional modeling approaches. In order to accurately predict population dynamics and identify optimal harvest limits for these species, recent studies have suggested that another type of modeling known as empirical dynamic modeling may be a reasonable alternative to conventional models.
In this project, Johnson will evaluate the feasibility of using empirical dynamic modeling for the assessment and management of short-lived marine species in the California Current. She aims to compare the prediction accuracy of conventional models and empirical dynamic modeling for six commercially valuable short-lived stocks and use empirical dynamic programming to derive near-optimal harvest control rules for each stock. She will also improve the forecast performance by including spatial information in empirical dynamic modeling algorithms.
This research will help improve our ability to accurately forecast the dynamics of economically valuable, highly variable species. This research will also advance fisheries modeling more generally.