应用栖息地指数对印度洋大眼金枪鱼分布模式的研究

Study on distribution of Thunnuns obesus in the Indian Ocean based on habitat suitability index

运用分位数回归方法对温度、温差、氧差与印度洋大眼金枪鱼延绳钓钓获率进行二次回归分析，找出最佳上界方程，以最佳上界方程拟合的数值来建立栖息地指数（HSI）模型，从而揭示印度洋大眼金枪鱼栖息地的分布模式。研究表明，温度、温差、氧差与印度洋大眼金枪鱼延绳钓钓获率的最佳上界分位数回归方程分别为HRT0.9=-44．803＋7．685T0.9-0．255T0.9^2，HRd70.9=6．234＋0．953dT0.9-0．026dT0.9^29和HRdO0.88=7．422＋4．25dO0.88-0．727dO0.88^2。10°N-10°S间印度洋海域大眼金枪鱼HSI指数达到0．9以上；10°N以北的波斯湾及10°S～15°S海域的HSI指数为0．8～0．9；15°S～40°S之间海域HSI指数介于0．7～0．8，其中50°E～90°E、15°S～25°S间存在一片季节性HSI指数〈0．7的区域；40°S以南的海域HSI指数〈0．6。

Bigeye tuna Thunnuns obesus is one of key species caught by Chinese tuna longline fleets. Its spatial distribution in relation to environmental factors is highlighted by international organizations and researchers. The aim of this study is to present the analysis of correlation between hooking rates of bigeye tuna longline and environmental factors in the Indian Ocean. Three environmental variables including temperature, temperature difference and dissolved oxygen difference are used to fit habitat suitability model in order to explain the distribution pattern of bigeye tuna in the Indian Ocean. Function expressions of hooking rates and environmental factors are estimated by quantile regression. Data predicted by optimum upper boundary quantile curves are fitted to the habitat suitability model to display quarterly distribution of bigeye tuna via visualization of Surfer 8.0. The optimum upper boundary quantile curves for temperature （T）-hooking rate （HR）, temperature difference （dT）-HR and dissolved oxygen difference （dO2）-HR are in the following HRT0.9 = -44. 803 ＋ 7. 685T0.9 -0. 255T0.9^2, HRdT0.9 = 6. 234 ＋ 0. 953dT0.9 -0. 026dT0.9^2 and HRdO0.88 = 7. 422 ＋ 4. 25dO0. 88 -0. 727dO0.88^2 as, respectively. Habitat suitability index is above 0.9 within 10°N-10°S, 0.8-0.9 in the north of 10°N and within 10°S- 15°S, 0.7 -0.8 within 15°S-40°S, and below 0.6 in the south of 40°S. However, a mass of waters occurs seasonally within 50°E-90°E,15°S - 25°S where habitat suitability index is less than 0.7. The habitat suitability model indicates the reliable results and could be improved by integrating more interactive variables. It is proved in this study that quantile regression is a useful way to investigate correlation between organism and limiting ecological factors.