Abstract
摘要:漁業捕撈針對體型較大和較年長的個體,這樣會導致年齡截斷效應和生活史演化效應等。這些效應會對魚群產生許多負面影響,包括總生物量減少,種群波動加大,恢復能力降低等。因此,漁業評估管理應該不僅要預測魚群總生物量的變化,還要能正確的預測各體長或年齡群的動態變化。
傳統漁業評估模型雖然已經認知到年齡體長動態的重要性,但是其分析方法仍然基於線型矩陣模型和假定的魚量補充關係。此模式有很大的侷限。首先,近年來許多新的統計分析方法發現魚群的動態有非線性。其次,單一固定的魚量補充關係這一假設並不符合實際。事實上魚量補充關係是動態變化的,與氣候變遷有強烈的關聯。
為了能將年齡體長動態整合進入漁業管理,我們計劃運用經驗動態模型。經驗動態模型是基於狀態空間重構的非線性分析方法。在這一架構下,商業標的魚群的各個年齡體長的數量動態能耦合在一起。我們計劃用多元嵌入法同時結合各個年齡體長來構建動態系統。這一靈活的無母數分析方法不需要對模型結構做出假設,所以非常適用於非線性耦合體系。本方法能夠分析外源因子對於動態系統的影響,所以可以將環境因子或者漁澲死亡率加入分析。
這一方法雖然在分析許多漁業和環境因子的關係中得到了成功運用,但是卻從來沒有被運用於年齡體長的數量分析當中,也沒有與傳統漁業分析方法有系統地比較其優劣之處。
本研究將:一、比較傳統的漁業評估分析模型和經驗動態模型分析的優缺點。二、確認環境變遷和人為捕撈因素對魚群結構的相對作用及相互關係。三、進行情景模擬以研究氣候變遷和漁業捕撈對魚群結構的潛在作用。本研究的成果將顯著改善漁業資源管理方法。
Abstract: Fishing is highly size selective. Selective removal of bigger and older fishes in a population results in age truncation and fishing-induced evolution. These in turn generate various negative effects, such as reduced spawning biomass, increased population variability, and decreased resilience. Therefore, effective sustainable management requires good predictive ability to understand the dynamics of size classes in addition to the total biomass of the population.
Traditional stock assessment models, though recognizing the importance of size/age specific dynamics, are rooted entirely in the linear matrix-modeling framework coupled with an assumed stock-recruitment relationship (such as virtual population analysis or stock synthesis). This classic framework contains several limitations. First, it has been shown that the dynamics of fish population are largely nonlinear. Second, the assumed “fixed” stock-recruitment relationship is too rigid and unrealistic; indeed, the stock-recruitment relationship is likely to be dynamic and can vary according to climate.
Here, to realistically incorporating information of size/age specific dynamics into fisheries management, we propose to employ the Empirical Dynamic Modeling (EDM) framework we recently developed. EDM is a nonlinear method based on State Space Reconstruction (SSR). Under this EDM framework, we view the size/age specific abundances (or biomasses) of the exploited stock as a coupled dynamic system. We use the multivariate embedding method to reconstruct the system dynamics; that is, we use time series of size/age specific abundances for system reconstruction and then forecasting. This method is specifically suitable for nonlinear coupling systems. This method is nonparametric and adaptive, and requires no assumption of model structure. This method can easily incorporate external variables by considering environmental variables or fishing mortality as additional time series in the multivariate embedding.
This method has been applied successfully to various systems to understand the predictability of fish populations and their relationship with the environmental variables. However, this method has never been used to incorporate multivariate time series of size/age specific abundances for forecasting fisheries stocks. Moreover, there is no systematic comparison between the traditional stock assessment models and EDM so far.
Our objectives are 1) to compare the predictive power of existing matrix-based models to EDM, 2) to identify the relative importance of fishing and other environmental variables including climate factors that govern the dynamics of size classes, and 3) to develop scenario exploration to investigate potential effects of climate changes or fishing on exploited stocks. The knowledge gained from this project can be used effectively for improving fisheries management.
Keyword(s)
非線性預測
狀態空間重構
漁業評估
年齡體長分佈
全球變遷
過度捕撈
Nonlinear forecasting
state space reconstruction
stock assessment
size/age structure
climate change
overfishing