中国科学院海洋研究所机构知识库

本研究以生态系统能流网络为视角，探索长江口及其邻近海域生态系统渔业资源开发现状、捕食结构、历史网络更新、环境要素以及不同渔业参考点对生态系统的驱动效应。应用基于有限数据的体长频率贝叶斯评估方法（Length-based Bayesian Biomass estimator，简称为LBB），结合生长模式和生长表现指数，评估了长江口及其邻近海域七种鱼类群体（小黄鱼Larimichthys polyactis、赤鼻棱鳀Thryssa kammalensis、凤鲚Coilia mystus、黄鲫Setipinna tenuifilis、龙头鱼Harpadon nehereus、细条天竺鲷Jaydia lineata、镰鲳Pampus echinogaster）的资源开发状态及重要管理参考点；根据2019-2020年长江口及其邻近海域渔业资源稳定碳氮同位素数据，分析了该生态系统营养结构特征，并综合胃含物分析数据、文献追踪数据先验信息，利用分层贝叶斯混合模型（EcoDiet）构建具有代表性的食性组成矩阵，将不确定性引入食性拓扑结构；利用基于分层贝叶斯混合模型获得的食性矩阵，构建1985年长江口及其邻近海域生态系统能流网络模型（YE1985updates），将主要资源物种细化为功能组，并更新初级生产量输入，模型构建经过严格的检验步骤和营养关系合理化检验；基于YE1985updates生态系统模型回溯1985-2020年长江口及其邻近海域主要资源群体的营养结构时间动态，模型拟合基于15种渔业资源群体生物量时间序列、2种生物量强迫序列和2种渔获量强迫序列，基于主要资源群体生物量时间序列和主要环境因子时间序列开展相关性分析，分析主要环境驱动要素，评估基于不同单一物种评估方法获取的小黄鱼、镰鲳和龙头鱼三种资源群体的管理参考点的管理效果和生态系统影响。

（1）生长模式图（auximetric plots）和生长表现指数（Φ’）可以在一定程度上与资源评估模型相结合，为相关管理措施的制定提供指导。只有两个LBB模型显示其对应的渔业资源状态健康且捕捞压力较小（F/M<1），它们分别是2018年的黄鲫和细条天竺鲷，表征了渔业资源捕捞压力较大问题的普遍性。极限体长评估值的不确定性相对于B/B₀来说较小，赤鼻棱鳀、小黄鱼和镰鲳生物量衰减较严重。镰鲳和细条天竺鲷的极限体长随着时间的推移有所减小，相反的是，他们的B/B₀平均值均出现小幅度的增加，并且两个时期镰鲳B/B₀的置信区间重叠部分较小，更有说服力。同一年中，黄鲫相对资源量B/B₀接近0.5，比凤鲚的资源状态要好。细条天竺鲷和赤鼻棱鳀倾向于较小的体型和较快的生长速度，他们的极限体长接近历史记录的中位数。长江口的凤鲚和黄鲫的极限体长要高于FishBase中的记录值。细条天竺鲷是唯一一个生长表现参数低于历史记录中位数的物种，而镰鲳的生长表现参数高于历史记录的中位数。龙头鱼的生长表现参数在七个物种中最大，其次是镰鲳，细条天竺鲷最小。

（2）长江口及其邻近海域稳定氮同位素δ¹⁵N均值在6.99‰~13.49‰，δ¹³C均值在-19.74‰~-12.43‰，其中鮸鱼（Miichthys miiuy）（>350mm）个体的δ¹⁵N最大，说明其偏向于高营养级，黄鮟鱇（Lophius litulon）（<100mm）的δ¹⁵N最小，说明其营养级与其他物种相比较低，红线黎明蟹（Matuta planipes）的δ¹³C最大，带纹条鳎（Zebrias zebra）（>200mm）的δ¹³C在鱼类中最大，说明他们倾向于底栖食物或主要分布在底层水体的食物，棘头梅童鱼（Collichthys lucidus）（100~200mm）的δ¹³C最小，说明这个体长组的棘头梅童鱼倾向于较上层的食物或活动水体偏上的食物。渔业资源生物营养级在2.141-3.583之间。黄鮟鱇体长组的营养级跨度最大。研究区域先验胃含物数据与文献数据具有很好的互补性，结合稳定同位素数据，减少了因为忽略某一捕食关系而导致的营养级联效应的错误概率。

（3）YE1985updates模型显示1985年长江口及其邻近海域生态系统营养结构中，功能组相对重要性指数显示，浮游动物是最关键的功能组，按照营养级划分，第二营养级中浮游动物最关键，第三营养级中虾虎鱼类最关键，高营养级中海洋哺乳动物较关键，其次是黄鮟鱇。浮游植物向第二营养级输送的能流比碎屑要大，浮游植物进入碎屑的比重也最大。初级生产量和碎屑的数量级和生态系统能流输出的数量级相差三倍，系统能流主要进入第二营养级，而第二营养级向高营养级输出的能流比重较低，较大一部分通过呼吸消耗或者进入碎屑。与源模型相比，模型YE1985updates显示生态系统能流规模较高，净初级生产量较高，总消费量和呼吸量出现小幅下降，生态系统成熟度有所降低，杂食性指数增高。渔获物平均营养级降低明显，表明新模型评估的渔业资源开发强度比原模型预估的要高，高营养级捕食者占比更低，冗余占比显示新模型评估的系统可发展空间较高。

（4）长江口年度输沙量与小黄鱼、镰鲳、三疣梭子蟹（Portunus trituberculatus）、赤鼻棱鳀、刀鲚（Coilia nasus）、海鳗（Muraenesox cinereus）、黄鲫、棘头梅童鱼、鳓（Ilisha elongata）、皮氏叫姑鱼（Johnius belangerii）均有不同程度的正相关关系，年度径流量与赤鼻棱鳀、黄鲫、龙头鱼有正相关关系，海洋表面温度异常三年平均和PDO都与黄鮟鱇有正相关关系，黄鲫与暖池（WP）有负相关关系，冷舌（CT）与龙头鱼和鳀（Engraulis japonicus）都有正相关关系。1985-2020年，生态系统总体生物量未出现较大幅度的变动，鱼类生物量下降近1.5t/km²，且接近2020年时有反弹迹象，无脊椎动物与鱼类生物量的比重有所增加，中下层生物资源量总体维持在1.5t/km²，中上层生物生物量在2000年下降至2.0t/km²，降幅1t/km²；中上层生物量比重下降，1995年前后，捕食者生物量下降明显，Kempton’s Q指数和香农指数均显示1985-2020年，生态系统整体生物多样性降低。基于Beverton-Holt模型得出的F_max参考点对生态系统的影响较小，这个管理参考点可以提升镰鲳和小黄鱼资源量和整个生态系统生物多样性，同时不影响生态系统总渔获量。不同的管理参考点将会引起生物多样性和渔获物平均营养级的不同程度和不同阶段的提升，大幅度降低渔业资源开发强度可能会在一定时期内牺牲掉一部分生态系统总渔获量。

This study explores the current situation of fishery resource development, predation structure, historical network update, and the driving effects of environmental factors and different fishery reference points on the Yangtze Estuary and its adjacent waters ecosystem from the perspective of energy flow network. This study applied the date-limited length-based Bayesian biomass estimation (LBB) method, combined with growth patterns and growth performance indices, evaluated the resource exploitation status and essential management reference points for seven fish stocks. Based on the stable carbon and nitrogen isotope data of fishery resources in the study area from 2019 to 2020, this study analyzed the trophic structure characteristics of the estuary, constructed a representative diet composition matrix using the hierarchical Bayesian integrated model (EcoDiet), and introduced uncertainties into the diet topology. With the diet composition matrix obtained from the EcoDiet model, this study constructed the new historic mass-balanced model YE1985updates. This new model refined the mainly fishery functional groups into species levels and updated the input of primary production. It had undergone strict check steps and rationalization of trophic relationships. Based on the YE1985updates ecosystem model, this study conducted a retrospective analysis of the temporal dynamics for the trophic structure of the study area from 1985 to 2020. The model had fitted biomass reference time series for 15 fishery resource groups, 2 forcing biomass time series, and 2 forcing catch time series. Correlation analysis was carried out between the biomass time series of main resource groups and the time series of main environmental factors, the main environmental driving factors were analyzed, and the management reference points and their effects on ecosystem structure based on different single species assessment methods were evaluated for three resource groups, i.e., small yellow croaker (Larimichthys polyactis), sickle pomfret (Pampus echinogaster), and Bombay-duck (Harpadon nehereus).

(1) Studies have shown that growth pattern plots (auximetric plots) and growth performance index (Φ') can be combined with resource assessment models to a certain extent to provide guidance on resource recovery potential and related management measures. The setipinna tenuifilis and jaydia lineata in 2018 showed a healthy resource status with their relatively small fishing pressures (F/M<1), indicating the prevalence of the problem of high fishing pressure on fishery resources. The uncertainties of infinite length estimates were smaller than those of B/B₀ values, and the biomasses of thryssa kammalensis, larimichthys polyactis and pampus echinogaster were depleted seriously. The infinite lengths of jaydia lineata and pampus echinogaster decreased with time, on the contrary, their average values of B/B₀ showed small increases, and the overlap between confidence intervals of the B/B₀ values for pampus echinogaster in the two periods were less and convincing than those of jaydia lineata. In the same year, the B/B₀ value of setipinna tenuifilis was close to 0.5, which indicated a better stock status than coilia mystus. The Jaydia lineata and thryssa kammalensis tend to be smaller in size and faster growth rate and their infinite lengths were close to the historical records’ medians in FishBase. The infinite lengths of coilia mystus and setipinna tenuifilis were bigger than the historical records’ medians in FishBase. The jaydia lineata was the only stock with its growth performance index (Φ') below the historical records’ median in FishBase, while this index of pampus echinogaster were above the historical records’ median in FishBase. The harpadon nehereus had the largest growth performance index (Φ') among the seven stocks, followed by the pampus echinogaster, and that of jaydia lineata was the smallest.

(2) The average value of stable nitrogen isotope δ¹⁵N in the Yangtze Estuary and its adjacent waters was 6.99‰~13.49‰, and the average value of δ¹³C was -19.74‰~-12.43‰. The miichthys miiuy (>350mm) had the largest δ¹⁵N value, indicating that it was at a higher trophic level. The δ¹⁵N value of lophius litulon (<100mm) was the smallest, indicating that its trophic level was lower than others, the δ¹³C of matuta planipes was the largest, and the δ¹³C value of zebrias zebra (>200mm) was the largest among fish species, indicating that they tended to be benthic food or feed mainly in bottom water, and Collichthys lucidus (100~200mm) had the smallest δ¹³C, indicating that it tended to pelagic food or feed on upper water. The trophic level of fishery resources is between 2.141-3.583. The trophic level span of lophius litulon in different body length groups was the largest. The prior of stomach content data in the study area was well complemented with those of literature data. Combined with stable isotope data, the error probability of trophic cascade effects caused by ignoring a certain predation relationship is reduced.

(3) The YE1985updates model showed that in the trophic structure of the ecosystem in the Yangtze Estuary and its adjacent waters in 1985, the relative importance index of functional groups showed that zooplankton was the key functional group in second trophic level. Gobies was the key functional group in the third trophic level, while marine mammals were the key functional group in the high trophic level, followed by lophius litulon. The energy flow of phytoplankton to the second trophic level was larger than that of detritus, and the proportion of phytoplankton entering the detritus was also the largest. The magnitude of primary production and detritus differed by three times from the magnitude of output energy flow. The system energy flow mainly entered the second trophic level, while the energy flow output from the second trophic level to the high trophic levels had a lower proportion and a larger part of this output consumed by respirations or entering to debris. Compared with the source model, the model YE1985updates showed a higher scale of ecosystem energy flow, a higher net primary production, a slight decrease in total consumption and respiration, a decrease in ecosystem maturity, and an increase in the omnivorous index. The average trophic level of the catches had decreased significantly, indicating that the fishery resources development intensity assessed by the new model was higher than that estimated by the original model, the proportion of predators with high trophic level was lower, and the redundant proportion showed that the system assessed by the new model had more room for higher development.

(4) The annual sediment discharge in the Yangtze Estuary has different degrees of positive correlation with Larimichthys polyactis, pampus echinogaster, Portunus trituberculatus, thryssa kammalensis, Coilia nasus, Muraenesox cinereus, setipinna tenuifilis, Collichthys lucidus, Ilisha elongata, Johnius belangerii. Annual runoff is positively correlated with thryssa kammalensis, harpadon nehereus, setipinna tenuifilis. The three-year average of sea surface temperature anomalies and PDO have positive correlation with harpadon nehereus. The setipinna tenuifilis has negative correlation with WP. CT has positive correlation with harpadon nehereus and Engraulis japonicus. From 1985 to 2020, the overall biomass of the ecosystem did not change significantly. The biomass of fish dropped by nearly 1.5t/km², and there are signs of rebound when approaching 2020. The proportion of invertebrates and fish biomass has increased. The demersal biomass remained at 1.5t/km² in general. The pelagic biomass decreased about 1t/km² to 2.0t/km² in 2000. The proportion of the pelagic biomass decreased. Around 1995, the biomass of predators declined obviously. The Kempton's Q index and the Shannon index showed that the overall biodiversity of the ecosystem decreased from 1985 to 2020. The Fmax reference point based on the Beverton-Holt model has little impact on the ecosystem. This management reference point can improve the biomass of pampus echinogaster and Larimichthys polyactis, and the biodiversity of the ecosystem, with little impact on the total fishery output of the ecosystem. Different management reference points will lead to different degrees and stages of improvement in biodiversity and the average trophic level of catches. Significantly reducing the fishing intensity of fishery resources may sacrifice some of the total ecosystem fishery catches within a certain period of time.

摘 要. I

Abstract V

第一章 绪论. 1

1.1 研究海域渔业资源... 1

1.1.1 渔业资源历史开发状态... 1

1.1.2 渔业资源优势种变化... 2

1.1.3 渔业资源物种数波动... 3

1.1.4 渔业资源开发环境... 3

1.2 研究海域渔业资源与环境要素... 4

1.2.1 鱼类与环境要素... 4

1.2.2 无脊椎动物与环境要素... 5

1.3 Ecopath and Ecosim（EwE）模型研究... 5

1.3.1 生态系统营养结构... 5

1.3.2 生态系统动力学研究... 8

1.3.3 EwE模型应用阶段... 10

1.4 研究目的及意义... 11

1.4.1 研究目的... 11

1.4.2 研究意义... 12

第二章 基于LBB模型评估七种渔业资源种群的资源状态. 13

2.1 前言... 13

2.2 材料与方法... 14

2.2.1 调查海域... 14

2.2.2 采样方法... 15

2.2.3 生长模式... 16

2.2.4 基于体长频率的贝叶斯种群评估方法... 17

2.3 结果... 19

2.4 讨论... 25

2.4.1 资源状态评价标准... 25

2.4.2 饵料生物... 27

2.4.3 经济鱼类... 28

2.5 小结... 29

第三章 基于分层贝叶斯整合食性方法构建食物网营养结构. 31

3.1 前言... 31

3.2 材料与方法... 32

3.2.1 调查海域... 32

3.2.2 碳氮稳定同位素采集与测样... 32

3.2.3 胃含物数据与文献补充数据... 33

3.2.4 分层贝叶斯建模框架估算捕食关系... 34

3.3 结果... 34

3.3.1 稳定碳氮同位素特征... 34

3.3.2 整合食物网结构构建... 38

3.4 讨论... 45

3.5 小结... 47

第四章 基于整合食性数据构建营养网络历史背景模型. 49

4.1 前言... 49

4.2 材料与方法... 51

4.2.1 调查海域... 51

4.2.2 模型基本原理... 51

4.2.3 YE1985. 51

4.2.4 YE1985updates. 51

4.2.5 模型初始参数... 51

4.2.6 模型平衡... 55

4.3 结果... 56

4.3.1 模型初始化参数... 56

4.3.2 生态系统特征指标... 60

4.3.3 关键功能组与营养相互作用... 61

4.4 讨论... 64

4.5 小结... 66

第五章 整合种群和生态系统模型评估长江口渔业资源动态. 67

5.1 前言... 67

5.2 材料与方法... 68

5.2.1 Ecosim时间动态模型... 68

5.2.2 单一种群评估方法... 69

5.2.3 相关时间序列... 69

5.2.4 情景设置... 70

5.3 结果... 70

5.3.1 饵料生物、资源群体与环境驱动因子... 70

5.3.2 渔业资源种群历史变动回溯... 74

5.3.3 长江口径流量与输沙量生态系统结构响应... 78

5.3.4 不同渔业生产压力下渔业资源响应... 81

5.4 讨论... 84

5.5 小结... 86

第六章 结论与展望. 87

6.1 结论... 87

6.1.1 基于单一物种资源评估... 87

6.1.2 整合食性数据营养结构构建... 87

6.1.3 构建研究海域历史背景能流网络模型... 88

6.1.4 生态系统时间动态模型构建与管理参考点评估... 88

6.2 展望... 89

参考文献. 91

致谢. 109

作者简历及攻读学位期间发表的学术论文与研究成果   111