比較三種泛線性模式標準化印度洋大目鮪族群豐度指標之適合度

Comparison of standardized abundance index of bigeye tuna (Thunnus abesus) in the Indian Ocean by three general linear models

馬慧珊(Hui-Shan Ma)

指導教授 ： 許建宗

國立臺灣大學/理學院/海洋研究所/碩士(2007年)

https://doi.org/10.6342/NTU.2007.03136

摘要

印度洋大目鮪系群，主要是由台灣、日本等國家以鮪延繩釣船和西班牙及法國以圍網船進行開發利用。印度洋大目鮪從1993年的平均產量100,000公噸增加到1995年的150,000公噸；2000年到2004年之間的平均產量也大約維持在119,000公噸。其中，台灣的捕獲量約佔所有生產量的30%，為印度洋大目鮪主要利用國之一。為有效永續利用此系群，資源評估和管理是必要的手段。在進行系群評估時，標準化單位努力漁獲量是常被用來反應資源豐度的指標之一。由於單位努力漁獲量受到時空環境因子和漁業策略等的影響，需要藉由標準化單位努力漁獲量，將各種可能影響漁獲率的變數排除，才能獲得較符合真實狀況的資源量變動趨勢。因此，本研究採用三種不同模式：泛線性模式、泛線性混合性模式及泛線性加乘模式對相同之漁獲資料進行單位努力漁獲量的標準化。並利用Akaike information criterion (AIC)、Bayesian information criterion (BIC)及Consistent Akaike's Information Criterion (CAIC)等三種統計值，作三種模式選擇之評估比較。標準化後的結果顯示，以泛線性加乘模式所得之模式統計值最小，故推斷此一模式為最適合做印度洋大目鮪系群資源量指標標準化模式。經標準化之單位努力漁獲量趨勢，自1980年代初期下降到1990年代初期的最低水準，再上升到1990年代中期的最高水準後，呈下降趨勢到2000年約為1990年之水準，後又再度上升至2002年之近十年來的次高水準。其後，三種模式有不同的資源量指標解讀，以泛線性模式和泛線性混合性模式標準化者，呈急速下降趨勢到2004年的歷年最低水準，而以泛線性加乘模式標準化者，則呈緩慢下降。

關鍵字

印度洋大目鮪 ； 泛線性模式 ； 泛線性混合模式 ； 泛線性加乘模式 ； Akaike information criterion ； Bayesian information criterion

並列摘要

Indian bigeye tuna are mainly exploited by Taiwanese and Japanese longline and French and Spanish purse seiners. Total annual catch in creased from about 100,000 mt in 1993 to about 150,000 in 1995 and stayed at around 119,000 mt averaged between 2000 and 2004. Among the catch, Taiwan fleets took about 30%, which is one of the major fishing countries. In order to exploit the stock sustainably, standardized catch-per-unit-effort (CPUE) is the most common method used as an index to reflect stock abundance. To obtain the actual abundance trend, possible factors influencing the catch rate need to be removed by using standardized because CPUE may differ across time, space, and fishing stratege…etc. In this study, these statistical models, generalized linear models (GLM), generalized linear mixed models (GLMM), and generalized additive models (GAM), were applied to standaedize the common fishery catch effort data. Furthermore, model selection and comparison were conducted among the three models using Akaike information criterion (AIC), Bayesian information criterion (BIC), and Consistant Akaike’s information criterion (CAIC). These results indicated that the GAM with the smallest information criterion was selected as the best model to standardize CPUE for Indian bigeue tuna. The standardized CPUE decreased from the early 1980s to the lowest level in the early 1990s, increased to the peak level in the mid of 1990s, then declined to the 1990 level in 2000, and increased again in 2002, the second high level over past decade. After that, the trends of standardized CPUE were different. CPUE estimated by GLM and GLMM sharply dropped to the historical low level in 2004, but it tended to slowly decline by GAM.

並列關鍵字

Indian bigeye tuna ； Generalized linear model ； Generalized linear mixed model ； Generalized additive model ； AIC ； BIC ； CAIC

參考文獻

Satoh, K., H. Okamoto and N. Miyabe. 2002.Abundance indices of Atlantic bigeye caught by the Japanese longline fishery and related information updated as of 2002. ICCAT, SCRS/2002/151.

Okamoto, H., S. K. Chang, Y. M. Yeh, C. C. Hsu. 2004.Standardized Taiwanese longline CPUE for bigeye tuna in the Indain Ocean up to 2002 applying targeting index in the model. IOTC/WPTT/10, 1-23.

Akaike, H. 1973."Information Theory and an Extension of the Maximum Likelihood Principle," in Petrov and Csaki, eds.,"Proceedings of the Second International Symposium on Information Theory," 267--281.

Bozdogan, H. 1987.Model selection and Akaike’s information criterion (AIC): the general theory and its analytical extensions. Psychometrika, 52(3), 345-370.

Hastie, T., R. Tibshirani. 1990.Generalized Additive Models.Statistical Science. 1(3),297-318.