Improved estimation and forecasts of stock maturities using generalised linear mixed models with auto‐correlated random effects

Estimation of proportion mature‐at‐age is an important component of the estimation of stock productivity. Current methods estimate maturity‐at‐age in each year or cohort independently. However, cohorts that are alive in the population at the same time are expected to experience similar conditions and therefore have similar maturity trajectories. Methods to take advantage of this information using binomial time series models with over‐dispersion are presented to improve the estimation of maturity‐at‐age, with applications to example populations of Atlantic cod, Gadus morhua Linnaeus, and American plaice, Hippoglossoides platessoides (Fabricius). An auto‐correlated beta‐binomial model fit the data well compared to other models investigated and improved forecasts of maturities for both populations. Use of an auto‐correlated beta‐binomial model should lead to improvements in projections of stock status and, hence, improvements in fisheries management advice.