Evaluating the Impact of Systematic Error on the Estimation Performance of a Length‐Based Spawning Potential Ratio

Abstract

AbstractDuring recent years, there has been growing efforts in methodological development for length‐based approaches to meet an increasing demand for science‐based management of data‐limited fisheries. Numerous simulation‐estimation analyses have been conducted to evaluate and compare the performance of length‐based data‐limited methods under various conditions, but none of them have evaluated the impact of observation error on the estimation performance. In this study, a simulation‐estimation analysis was conducted to evaluate the performance of a length‐based data‐limited method for length‐based spawning potential ratio (LBSPR) across a variety of cases in relation to systematic observation errors, life history types, and selectivities. The results showed that the estimation performance of LBSPR was sensitive to all types of the cases above. In particular, a disproportionately high frequency of small fish in the length composition could cause larger deviations in estimates of stock status compared with cases in which large fish are overrepresented. Partial dependence plots from the general linear model show how factors such as systematic changes in bias in length composition data affect the estimation of LBSPR. Our study suggests that the effectiveness of length‐based data‐limited stock assessment should be carefully evaluated on a species‐by‐species basis when observation error, such as systematic error in length‐frequency data, is present.