Can a length-based pseudo-cohort analysis (LBPA) using multiple catch length-frequencies provide insight into population status in data-poor situations?

Stock status for many mediumand small-scale fisheries is unknown due, for example, to a lack of catch data and the absence of scientific observer programs. However, length-frequency data are often available for such fisheries because they are the cheapest and easiest data to obtain. Various stock assessment methods have been developed that use length-frequency data and make equilibrium assumptions regarding both recruitment and fishing mortality. These assumptions raise questions regarding the reliability of the results, particularly when the method is applied to a single sample of length-frequency. We developed a Length-Based Pseudo-cohort Analysis (LBPA) model whose parameters can be estimated using multiple length frequencies and penalized maximum likelihood, under the assumption that using more than one length-frequency sample reduces the effects of the equilibrium conditions assumed in the model. We explored the performance of LBPA using simulations that examined the implications for estimation of the spawning potential rate (SPR) and the relative fishing mortality (F/FMSY). These simulations considered scenarios related to exploitation rate, steepness, numbers of years and sample sizes for length-frequency, and data weights. The performance of LBPA was compared to that of length based spawning potential ratio (LBSPR). LBPA performed better with additional length-frequencies than with increased sample size, irrespective of how the data were weighted, and generally outperformed LBSPR. Estimates were more accurate and less biased when exploitation rates were high. This work provides guidelines that should be considered when using length-based pseudo-cohort models for data-poor fisheries.