基于Bootstrap的ELEFAN方法在评估方氏云鳚群体生长参数中的应用

Evaluating the growth parameters of Pholis fangi based on the bootstrap-ELEFAN method

渔业数据有限性是小型渔业资源评估所面临的常见问题。电子体长频率分析(electroniclengthfrequency analysis,ELEFAN)常用于年龄数据难以获取或缺失的渔业,但该方法的可靠性尚待检验。本研究根据2013―2018年春、秋季共11个航次的海州湾底拖网调查数据,分别使用传统的ELEFAN与结合Bootstrap的ELEFAN方法,比较了2013―2015年与2016―2018年两个时间段内海州湾方氏云鳚(Pholis fangi)群体von-Bertalanffy生长方程中参数之间的变化。结果显示,在海州湾海域,方氏云鳚的生长参数具有显著变化, 2013―2018年,群体的极限体长变小,生长速率加快,说明海州湾方氏云鳚群体近年来呈现小型化的趋势。相比传统的ELEFAN方法,结合Bootstrap的ELEFAN方法能够给出较为稳健的参数估计,受采样随机性的影响较小,可以较好地应用于数据缺乏的小型渔业中。本研究加深了对方氏云鳚种群动态的认识,并推动了基于体长频率的生长参数估算方法在数据有限资源评估中的应用。

Fish stock assessment usually requires a wide range of supporting data, including an abundance index, production, age structure. However, some data are hardly available in many fisheries because of limited research funding and social attention. Therefore, many fisheries, particularly small-scale fisheries, often do not have suffi- cient data to support fish stock assessment and are considered data-limited or data-poor. An increasing amount of literature has been focused on the development of data-poor stock assessment methods in recent decades, among which electronic length frequency analysis (ELEFAN) is a prevalent method that uses length frequency distribu- tion data to assess the status of fisheries. One crucial application of the ELEFAN is the estimation of growth pa- rameters in the Von Bertalanffy growth function (VBGF). However, the method is based on certain optimization algorithms and cannot provide information on its precision or confidence intervals for growth parameters, which implies that the reliability of ELEFAN needs to be tested in future studies. This study used a bootstrap approach to evaluate the uncertainty of the ELEFAN method based on the survey data of Pholis fangi in Haizhou Bay. This species is one of the dominant species in Haizhou Bay and plays an important role in the food web and ecosystem of the Yellow Sea. Although the declines in fishery resources have drawn increasing attention in many regions of the world, relevant studies have commonly focused on large-scale fisheries, whereas small-scale fisheries, such as that of P. fangi, has been largely overlooked. Therefore, the biological characteristics of this species and their temporal changes is not well understood. This study was focused on the temporal changes in VBGF growth pa- rameters of P. fangi in Haizhou Bay at different survey periods. We conducted annual bottom trawl surveys in Ha- izhou Bay in the spring and autumn from 2013 to 2018, and used the ELEFAN method to estimate the VBGF growth parameters infinite length (L ) and growth parameter (K) of P. fangi. In addition, the bootstrapped ELEFAN was used to evaluate the variation in the growth parameters, and the difference was compared between 2013 2015 and 2016 2018. We analyzed the robustness of ELEFAN with respect to three aspects:(1) the effect of bin size of body length on parameter estimation,(2) the selection of different optimization algorithms (Simulated Annealing, SA;Genetic Algorithm, GA;Response Surface Analysis, RSA), and (3) the confidence intervals of parameter estimation through the bootstrap approach. The results showed that the VBGF growth parameters of P. fangi in Haizhou Bay changed significantly during 2013-2018, and the decreased infinite length (L ) and increased growth parameter (K) indicated that there was a significant trend of miniaturization. The bin size of body length significantly affected the goodness of model fit and improper bin size settings might lead to unreasonable pa- rameter estimations. Bootstrapped ELEFAN provided robust parameter estimations compared to the conventional ELEFAN approach, and the bootstrapped results were less affected by the randomness of sample data. The Genetic Algorithm could benefit from parallel computing in the TropfishR package, which significantly sped up computa- tion. This study improved the understanding of population dynamics of P. fangi. In particular, the changes of growth characteristics of this species may have a substantial impact on the Haizhou Bay ecosystem. We demon- strated that bootstrapped ELEFAN performed well and could be applied to the prevalent data-poor and small-scale fisheries.