Study improves prediction of menhaden reproduction, potentially aiding managers
The abundance of two small but important fish species—Atlantic and Gulf menhaden—is more predictable than once thought, according to a new study in the journal Fish and Fisheries. The finding represents a potentially significant step toward more precise management of two of the largest fisheries in the United States.
Around 650,000 metric tons of menhaden were landed in the Gulf of Mexico and on the East Coast in 2016—primarily for production of fish meal and fish oil for national and international markets. These products have a variety of uses, including livestock feed, aquaculture feed, pet food, and fish oil supplements. Menhaden is also a food source for species like striped bass, bluefish, and weakfish.
As with most major fisheries, managers conduct extensive analysis and modeling to understand menhaden populations and set catch limits. But they have not been able to accurately predict the amount of young menhaden entering the catchable population in a given year, a quantity known as recruitment.
“This study shows that menhaden recruitment is in fact predictable,” said George Sugihara, a study co-author and a professor of biological oceanography at the Scripps Institution of Oceanography. “The key is to look for patterns in the data that are not apparent from conventional methods.”
To look for those patterns, the research team used a set of techniques called empirical dynamic modeling (EDM). EDM is a forecasting method that, unlike conventional modeling, does not use equations that assume specific mathematical relationships among variables. It has its roots in physics and has been applied in several other fields of research.
Instead, EDM takes advantage of the fact that changes in a single variable over time can contain information about the dynamics of the full system. This allows researchers to reconstruct the big picture using only one-time series of data. They can then use this reconstruction as a tool for predicting changes.
For this study, the researchers used EDM to predict recruitment for Atlantic and Gulf menhaden. Recruitment had previously been regarded as unpredictable for Atlantic menhaden, and highly uncertain for Gulf menhaden.
They found that EDM improved forecast accuracy compared to current practice; in statistical terms, it explained about half the variation in recruitment of Atlantic menhaden. The current stock assessment uses the long-term average for recruitment, but still has a lot of uncertainty. The data on Gulf menhaden are less complete than for Atlantic menhaden, but the method explained about one-sixth of the variation in recruitment for that species.
The research team is now working with their co-author, Amy Schueller of the National Marine Fisheries Service, to integrate the method into the “projection analysis” for Atlantic menhaden, which managers can use to understand the likely impacts of various management options.
“We think improved predictions offer several benefits to both fishermen and the ecosystem,” said Ethan Deyle, a Scripps scientist and lead author of the paper. “For example, it could help provide timely warning of an impending drop in recruitment, which could lead to a population crash. Or it could reveal a pulse of strong recruitment, which might allow for increased harvest without harm to dependent predators.”
