Age structure augments the predictive power of time series for fisheries and conservation

Abstract

Ecological forecasts are potentially of great value for managing fisheries and for stakeholders dependent on their long-term sustainability. Yet, most forecasting approaches are data-intensive, requiring information not just on the focal species but also on ecological interactions and the physical environment. Empirical dynamic modeling (EDM) is an equation-free approach to forecasting species’ abundance using only data on past abundance, but the time series required for this approach must be long enough to reconstruct the dynamics of the system. This requirement is rarely met, especially for long-lived species. Here we used simulations and empirical data to demonstrate that incorporating time series from multiple age classes can improve our ability to forecast abundance compared to a single age class or index of total abundance. Including data from multiple age classes produced the greatest gains in forecast accuracy when time series were the shortest. Overall, our results show that the incorporation of age structure could allow EDM to be applied to many species for which relatively short time series would have previously been a limiting factor.