鸢乌贼（Sthenoteuthis oualaniensis）作为暖水性较强的大洋性头足类，生命周期短、繁殖力强，可承受高强度的捕捞压力，是南海最具开发潜力的种类之一。渔情预报是海洋渔场学研究的重要内容，不仅可以缩短寻找渔场的时间，对南海外海渔业资源的优化管理及生产也提供了科学的技术指导。近年来，随着近海渔业资源的日益枯竭以及远洋渔业的发展，对其开发成为研究工作重点。本文主要研究了南海外海鸢乌贼渔场分布及渔情预报模型的构建。其中主要工作概括如下：

1.采用卫星遥感获取的海表面温度（SST）、叶绿素a浓度（chla）、海洋初级生产力（NPP）、海表面高度异常（SSHA）、海表面盐度（SSS）和表层海流速度（SSC）等海洋环境数据及渔捞日志生产数据，运用广义线性模型（GLM）和加入空间自相关的广义线性模型（S-GLM）对鸢乌贼CPUE数据进行标准化。最小AIC指标值揭示S-GLM模型标准化结果优于GLM模型，其中基于指数空间自相关的S-GLM模型最佳。标准化后的CPUE均低于或者接近于名义CPUE。

2.运用广义加性模型（GAM）探讨了海洋环境因子与渔场的关系。结果显示，除chla外，其它因子对CPUE均有显著性影响。除SSC以外，与CPUE均呈非线性关系。渔场主要集中在9.75~11°N，113.75~115.75°E区域和7~9.25°N，111.75~114.75°E区域附近，影响因子重要性依次为long、lon×lat、NPP、SST、SSHA、SSC和SSS。

3.基于标准化CPUE数据，利用Yield Density函数建立各季节的单因子适宜性指数（SI）模型，并作出相应的适宜性曲线。研究结果显示，鸢乌贼作业渔场最适的SST在夏季最高，冬季最低；SSHA在秋季最高，冬季最低；NPP在冬季最高，其它季节基本相同；四个季节最适的SSS基本相同；SSC在冬季最高，秋季达到低谷。

4.运用主成分分析法（PCA）确定各栖息因子的权重，从而构建栖息地适宜性指数（HSI）模型。结果显示，春季、冬季SSS的权重最大，夏季、秋季NPP的权重最大。模型验证结果表明，四个季节的作业渔区预报准确率均在60%以上，CPUE总体相对误差平均值为17.87%。

5.运用BP神经网络模型预测鸢乌贼中心渔场。研究结果显示，春季、夏季和秋季的最优BP模型结构均为8-6-1模型，冬季的最优模型结构为8-7-1模型。利用2019年数据对中心渔场进行验证发现，CPUE总体相对误差平均值为2.655%。与HSI模型进行比较发现，BP神经网络模型优于HSI模型。

As an oceanic cephalopod with strong warm water, the sthenoteuthis oualaniensis has a short life cycle, strong reproduction ability, and can withstand high-intensity fishing pressure. it is one of the most potential species in the South China Sea. The fishery forecast is an important content of marine fishery research, it can not only shorten the time to find fishing grounds, but also provide scientific and technical guidance for the optimal management and production of fishery resources in the South China Sea. In recent years, with the increasing depletion of offshore fishery resources and the development of  pelagic fishery, its development has become the focus of research work. This paper mainly studies the fishing ground distribution of Sthenoteuthis oualaniensis and the construction of fishing forecast model in open South China Sea. The main work is summarized as follows:

 

1.Sea surface temperature, chlorophyll a concentration, net primary productivity, sea surface height anomaly, sea surface salinity and surface current velocity obtained by satellite remote sensing and fishing log production data, the generalized linear model and spatial autocorrelation GLM were selected to standardize the CPUE data of sthenoteuthis oualaniensis. According to the minimum Akaike Information Criterion, it is revealed that the standardized results of the S-GLM are better than the GLM, and the S-GLM based on exponential spatial autocorrelation is the best. The standardized CPUE is lower than or close to the nominal CPUE.

 

2.The generalized additive model is used to analyze the relationship between marine environmental factors and distribution of fishing grounds. It is found that except for the chlorophyll aconcentration, other factors have significant effects on CPUE. Except for surface current velocity, other factors has a non-linear relationship with CPUE. The fishing grounds are mainly concentrated around 9.75~11°N, 113.75~115.75°E and 7~9.25°N, 111.75~114.75°E. impact factors sorted by importance are as follows: longitude and latitude, net primary productivity, sea surface temperature, sea surface height anomaly, surface current velocity, sea surface salinity.

 

3.Based on standardized CPUE data, the Yield Density function is used to establish a single factor suitability index model for each season, and the corresponding suitability curve is drawn. The results of the study showed that the optimal sea surface temperature for the fishing grounds of sthenoteuthis oualaniensis officinalis is highest in summer and lowest in winter; sea surface height anomaly is highest in autumn and lowest in winter; net primary productivity is highest in winter and and it is basically the same in other seasons; The optimal sea surface salinity for the four seasons is basically the same; sea surface salinity is highest in winter and reaches a trough in autumn.

 

4.Principal components analysis is used to determine the weight of each habitat factor, thereby constructing a habitat suitability indes model. It is found that the weight of sea surface salinity in spring and winter is the largest, and the weight of net primary productivity in summer and autumn is the largest. The model verification results show that the forecast accuracy rates of the operating fishing areas in the four seasons were all above 60%, and the overall average relative error of the CPUE is 17.87%.

 

5.The BP artificial neural network model is used to predict the central fishing ground of the sthenoteuthis oualaniensis.  It is found that the optimal error back propagation artificial neural network models structure in spring, summer and autumn is the 8-6-1 model, and the optimal model structure in winter is the 8-7-1 model. Using 2019 data to verify the central fishing grounds, it is found that the average overall relative error of the CPUE is 2.655%. Compared with the HSI model, it is found that the BP neural network models is better than the HSI model.

[1]邱永松, 曾晓光, 陈涛等. 南海渔业资源与渔业管理[M]. 北京: 海洋出版社, 2008.
[2]陈作志, 邱永松. 南海区海洋渔业资源现状和可持续利用对策[J]. 湖北农学院学报, 2002(06):507-510.
[3]车斌, 熊涛. 南海争端对我国南海渔业的影响和对策[J]. 农业现代化研究, 2009,30(4):415-418.
[4]黎祖福, 吕慎杰. 南海渔业产业现状及与东盟周边国家合作机制探讨[J]. 海洋与渔业,2013(12):42-45.
[5]张俊, 邱永松, 陈作志等. 南海外海大洋性渔业资源调查评估进展[J]. 南方水产科学, 2018,14(6):118-127.
[6]龚玉艳, 詹凤娉, 杨玉滔等.南海鸢乌贼摄食习性的初步研究[J]. 南方水产科学, 2016,12(4):80-87.
[7]冯波, 颜云榕, 张宇美等. 南海鸢乌贼(Sthenoteuthis oualaniensis)资源评估的新方法[J]. 渔业科学进展,2014(4):1-6.
[8]范江涛, 冯雪, 邱永松等. 南海鸢乌贼生物学研究进展[J]. 广东农业科学, 2013(23):122-128.
[9]陈新军. 渔业资源与渔场学[M]. 北京: 海洋出版社, 2004: 86-215.
[10]Wang X H, Qiu Y S, Zhang P, et al. Natural mortality estimation and rational exploitation of purpleback flying squid sthenoteuthis oualaniensis in the southern South China Sea[J]. Chinese Journal of Oceanology and Limnology, 2017, 35(4): 902-911.
[11]杨吝, 张旭丰, 张鹏等. 南海区海洋小型渔具渔法[M]. 广州: 广东科技出版社, 2007.
[12]陈文河, 宋利明. 应用灯光罩网开发三沙海域渔业资源[J]. 海洋开发与管理, 2013(S1):68-70.
[13]陈森, 张鹏, 晏磊等. 南海新建钢质罩网渔船渔获组成及渔场分析[J]. 南方水产科学, 2015,11(5):125-131.
[14]杨权, 李永振, 张鹏等. 基于灯光罩网法的南海鸢乌贼声学评估技术研究[J]. 水产学报, 2013,37(7):1032-1039.
[15]刘必林, 陈新军, 钟俊生. 采用耳石研究印度洋西北海域鸢乌贼的年龄生长和种群结构[J]. 大连水产学院学报, 2009,24(3):206-212.
[16]张溢卓. 日本渔业产业的发展历程及其变化要因分析[J]. 世界农业, 2017(10):43-47.
[17]宇田道隆. 海洋渔场学[M]. 东京: 恒星社厚生阁发行所, 1963: 20-22.
[18]仉天宇, 邵荃琴, 周成虎. 卫星测高数据在渔情分析中的应用探索[J]. 水产科, 2001,20(6):4-8.
[19]Yamanaka I, Niwa k, Tanake R, et al. The fisheries forecasting system in Japan for coastal pelagic fish[J]. FA0 Fisheries Technical Paper, 1988, 301:72-80.
[20]Laurs R M. Fishery-advisory information available to tropical Pacific tuna fleet via radio facsimile broadcast[J]. Marine Fisheries Review, 1971,33(4): 40-42.
[21]施益强, 陈玲. 遥感技术在环境科学工程应用中的进展[J]. 科技导报, 2002(12):25-29.
[22]张震东, 杨森. 中国海洋渔业简史[M]. 北京: 海洋出版社, 1983: 5-28.
[23]Hawkes, Elden W. NOAA/NMFS Approves the Mid-Atlantic ACL/AM Omnibus Amendment[J]. Fisheries, 2011,36(11):530-531.
[24]MYERS, D.G, HICK, et al. An application of satellite-derived sea surface temperature data to the Australian fishing industry in near real-time[J]. International Journal of Remote Sensing, 1990,11(11):2103-2112.
[25]李雪渡. 海水温度与渔场之间的关系[J]. 海洋学报, 1982,4(1):103-l12.
[26]刘树勋, 韩士鑫, 魏永康. 判别分析在渔情预报中应用的研究[J]. 海洋通报, 1988,7(1):63-70.
[27]刘树勋, 韩士鑫, 魏永康. 东海西北部水团分析及与渔场的关系[J]. 水产学报, 1984,8(2):125-133.
[28]陈新军, 钱卫国. 印度洋西北部海域鸢乌贼资源密度分布的初步分析[J]. 上海水产大学学报, 2004(03):218-223.
[29]杨晓明, 陈新军, 周应祺等. 基于海洋遥感的西北印度洋鸢乌贼渔场形成机制的初步分析[J]. 水产学报, 2006(05): 669-675.
[30]余为, 陈新军. 印度洋西北海域鸢乌贼9-10月栖息地适宜指数研究[J]. 广东海洋大学学报, 2012(06): 74-80.
[31]范江涛, 陈作志, 张俊等. 基于海洋环境因子和不同权重系数的南海中沙西沙海域鸢乌渔场分析[J]. 南方水产科学, 2016(04): 57-63.
[32]余景, 胡启伟, 李纯厚等. 西沙-中沙海域春季鸢乌贼资源与海洋环境的关系[J]. 海洋学报, 2017,39(6):62-73.
[33]Maunder M N, Punt A E. Standardizing catch and effort data: a review of recent approaches[J]. Fisheries Research,2004,70(2-3):141-159.
[34]Guan W J, Tian S Q,Wang X F,et al.A review of methods and model selection for standardizing CPUE[J]. Journal of Fishery Sciences of China, 2014, 21(4): 852-862.
[35]Mourato B L, Hazin H G, Wor C, et al. Environmental and spatial effects on the size distribution of sailfish in the Atlantic Ocean[J]. Ciencias Marinas, 2010, 36(3):225-236.
[36]Bigelow K A, Boggs C H, Xi H E. Environmental effects on swordfish and blue shark catch rates in the US North Pacific longline fishery[J]. Fisheries Oceanography, 2010, 8(3):178-198.
[37]Bishop J, Venables W N, Wang Y G. Analysing commercial catch and effort data from a Penaeid trawl fishery: A comparison of linear models, mixed models, and generalised estimating equations approaches[J]. Fisheries Research, 2004, 69(2-3):179-193.
[38]Xu L, Chen X, Guan W, et al. The impact of spatial autocorrelation on CPUE standardization between two different fisheries[J]. Journal of Oceanology and Limnology, 2018, 36(3):973-980.
[39]Dormann C F, Mcpherson J M, Miguel B Araújo, et al. Methods to account for spatial autocorrelation in the analysis of species distributional data:a review[J]. Ecography, 2010, 30(5): 609-628.
[40]《渔业船舶管理概论》编委会. 渔业船舶管理概论[M]. 上海: 上海交通大学出版社, 2015:20-50.
[41]Venables W N, Dichmont C M. GLMs, GAMs and GLMMs: An overview of theory for applications in fisheries research[J]. Fisheries Research, 2004, 70(2-3):315-333.
[42]武妍戎. 正态分布及其衍生分布的性质统计意义研究[J]. 科学家, 2017, 5(1):20-22.
[43]林清泉. 计量经济学[M]. 北京: 中国人民大学出版社, 2013:192-245.
[44]Campbell R A. CPUE standardisation and the construction of indices of stock abundance in a spatially varying fishery using general linear models[J]. Fisheries Research, 2004, 70(2-3):0-227.
[45]戴小杰, 马超, 田思泉. 印度洋中国大眼金枪鱼延绳钓渔业CPUE标准化[J]. 上海海洋大学学报, 2011,20(2):275-283.
[46]方文东, 郭忠信, 黄羽庭. 南海南部海区的环流观测研究[J]. 科学通报, 1997(21):26-33.
[47]王新星, 于杰, 李永振等. 南海主要上升流及其与渔场的关系[J]. 海洋科学, 2015,39(09):131-137.
[48]邹建伟, 陈立峰, 林蒋进等. 南海外海灯光罩网主要渔场分布及变动研究-基于广西渔船的生产监测统计[J]. 南方水产科学, 2014,10(4):78-84.
[49]陈新军, 邵锋. 印度洋西北部公海鸢乌贼资源特征及其与海况的关系[J]. 中国海洋大学学报(自然科学版), 2006,36(4):101-106.
[50]范江涛, 张俊, 冯雪等. 南沙海域鸢乌贼渔场与海洋环境因子的关系[J]. 上海海洋大学学报, 2019,28(3):419-426.
[51]Shipley B, Douma J C. Generalized AIC and chi-squared statistics for path models consistent with directed acyclic graphs[J]. Ecology, 2020, 101(3): 1-7.
[52]Golub G H, Heath M, Wahba G. Generalised cross-validation as a method for choosing a good ridge parameter [J]. Technometrics, 1979, 21(2):215-223.
[53]高姗. 基于遥感的南海初级生产力时空变化特征与环境影响因素研究[D]. 北京: 中国气象科学研究院, 2008:28-37.
[54]张波. 中国近海食物网及鱼类营养动力学关键过程的初步研究[D]. 山东: 中国海洋大学, 2005:15-21.
[55]徐红云. 南海外海鸢乌贼栖息地分布与关键环境因子分析[D]. 上海: 上海海洋大学, 2017:31-32.
[56]余为, 陈新军, 易倩. 西北太平洋海洋净初级生产力与柔鱼资源量变动关系的研究[J]. 海洋学报, 2016,38(2):64-72.
[57]田永青, 黄洪辉, 巩秀玉等. 2013年南沙群岛海域温跃层的季节变化及形成机理[J]. 海洋学报, 2017,39(12):20-31.
[58]晏磊, 张鹏, 杨炳忠等. 南海鸢乌贼产量与表温及水温垂直结构的关系[J]. 中国水产科学, 2016,23(2):469-477.
[59]李杰, 张鹏, 晏磊等. 南海中南部海域鸢乌贼CPUE影响因素的GAM分析[J]. 中国水产科学,2020,27(8):906-915.
[60]邵锋,陈新军.印度洋西北海域鸢乌贼渔场分布与海面高度的关系[J].海洋科学, 2008,32(11):88-92.
[61]宋婷婷, 樊伟, 伍玉梅. 卫星遥感海面高度数据在渔场分析中的应用综述[J]. 海洋通报, 2013,32(4):474-480.
[62]郭吉鸽. 由Argo浮标推断全球大洋表层海流并验证[D]. .北京: 中国科学院研究生院(大气物理研究所), 2007:20-23.
[63]唐峰华, 史赟荣, 朱金鑫等. 海洋环境因子对日本海太平洋褶柔鱼渔场时空分布的影响[J]. 中国水产科学, 2015(5):190-197.
[64]杨胜龙, 范秀梅, 张忭忭等. 阿拉伯海域围网渔场时空分布及其与海表环境因子的关系[J]. 中国农业科技导报, 2019,21(9):149-158.
[65]CHEN XR, LIU Z H, WANG HY, et al. Significant salinity increase in subsurface waters of the South China Sea during 2016–2017[J]. Acta Oceanol Sin, 2019, 38(11): 51-61.
[66]方水美. 福建沿海8种作业捕捞"综合能力"的比较研究[J]. 南方水产科学, 2006(2):29-36.
[67]U.S. Fish and wildlife Service. Standards for the development of habitat suitability index models[R]. U.S. Fish and wildlife Service, 1981:1-81.
[68]赵海龙, 陈新军, 方学燕. 基于栖息地指数的东太平洋黄鳍金枪鱼渔场预报[J]. 生态学报, 2016,36(3):778-785.
[69]姜启源. 数学模型（第二版）[M]. .北京: 高等教育出版社,1993.
[70]陈诚, 廖桂平, 李锦卫等. 基于GAM、BP神经网络和多元回归的集成算法研究[J]. 计算技术与自动化, 2011,30(2):89-95.
[71]RUMELHART D E. Learning representation by back-propagating errors[J]. Nature, 1986, 323 (6088): 533-536.
[72]毛江美, 陈新军, 余景. 基于神经网络的南太平洋长鳍金枪鱼渔场预报[J]. 海洋学报, 2016,38(10):34-43.
[73]魏联, 陈新军, 雷林等. 西北太平洋柔鱼BP神经网络渔场预报模型比较研究[J]. 上海海洋大学学报, 2017,26(003):450-457.
[74]Laurent D, Michel P, Stretta J M. Simulation of large scale tropical tuna movements in relation with daily remote sensing data: the artificial life approach[J]. Biosystems, 1997, 44: 167-180.
[75]Iglesias A, Dafonte C, Arcay B, et al. Integration of remote sensing techniques and connectionist models for decision support in fishing catches[J]. Environ modell softw, 2007, 22: 862-870.
[76]邵帼英. 基于神经网络的东南太平洋公海智利竹筴鱼渔情预报的研究[D]. 上海: 上海水产大学, 2007:1-73.