基于鱼类体长频率分布和平均体长的采样站位数优化

鱼类体长数据较易获得，常用于数据有限的渔业资源评估和管理，且基于鱼类体长的指标可以作为渔业生态系统的生态监测指标。本文根据山东半岛南部海域底拖网季度调查收集的方氏云鳚体长数据，应用计算机模拟重采样方法，研究了调查站位数对估计方氏云鳚（Enedrias fangi）体长频率分布和平均体长的影响。结果表明，在各个季节，方氏云鳚体长频率分布离散指数（DI）变化范围为0~0.91，大部分站位的DI值为0.75~0.90，说明调查站位间体长频率分布较为相似，可以对调查站位数进行优化，提高采样效率；调查站位越多，体长频率分布和平均体长估计精确度越高；由于对不同体长数据指标估计的精度不同，用以估计体长频率分布和平均体长的样本量存在差异；20个和30个调查站位分别为估计方氏云鳚体长频率分布和平均体长的可接受样本量。本研究结果可为基于不同体长指标的渔业资源调查设计提供一定的理论支撑，有助于对基于体长指标有效站位数的初步了解，为山东半岛南部海域的鱼类资源科学调查和科学管理提供参考。

Optimization of number of sampling stations in fishery-independent surveys for estimating fish length frequency distribution and mean length

Length frequency distribution and mean length data are often used to estimate parameters of growth and mortality of fish stocks, as they are relatively easy to obtain from fishery-independent surveys for most fish species and can provide insights into the dynamics of fish populations, especially for data-poor or -limited fisheries. A lack of smaller size classes of fish populations may represent recruitment deficiencies, whereas a lack of larger size classes might indicate high mortality of adult individuals of fish stock. Precise and accurate estimates of population parameters are critical to ensure rational fish stock assessment and management. The performance of a survey sampling design might vary with different size structure indices, including length frequency distribution and mean length. To provide high quality length data for supporting fishery resource assessment and scientific management, the performances of the number of survey stations for estimating Enedrias fangi length frequency distribution and mean length were evaluated using computer simulations based on the original data collected from bottom trawl surveys in October 2016, and January, May, and August 2017 in the southern waters off Shandong Peninsula. The dispersion index (DI) was calculated to reflect E. fangi size composition variability within survey stations to identify the factors that influence sampling efficiency. Large DI values implied that length samples were evenly distributed among all size bins, whereas a DI close to zero indicated all observations were skewed into relatively few size bins. The sum of absolute difference (SAD) and relative estimation error (REE) were used to measure the performances (accuracy and precision) of different numbers of stations in estimating the target size structure indices. The results showed that the E. fangi length frequency distribution DI ranged from 0 to 0.91 in four seasons, whereas the DI of most stations was between 0.75 to 0.91. The number of sampling stations could be reduced so that the sampling efficiency could be improved. In general, SAD and REE showed similar trends, decreasing with sample sizes initially and then becoming stable after certain sample sizes during all four seasons, which implied that the length data estimation precision improved with an increased number of sampling stations. The optimal number of sampling stations varied with target indices. The optimal sample size was 20 stations for estimating length frequency distribution and 30 stations for estimating mean length. This study identified the accepted optimal sample sizes to estimate length frequency distribution and mean length and could provide a technical reference for fishery-independent surveys to estimate the size structure index.