Abstract：The generalized linear models(GLM)and generalized additive models(GAM)were established to evaluate the relationship between spatial distribution of the fishery resources including chub mackerel(Scomber japonicus)and the environmental factors such as the temperature, salinity, and primary productivity in the East China Sea(ECS)by the data of temperature, salinity, primary productivity and observed fishing catch in the ECS on March 1998,July 1999, October 1997 and January 2000. The impacts of temperature, salinity and primary productivity on the abundance of chub mackerel were analyzed using the GLM and the GAM. The model outputs were compared with interpolated outputs of observed patterns of distribution obtained through independent-sample T-test and error analysis. In the independent-sample T-test, P value of both GLM and GAM were over 0.05, indicating hat GLM had the advantage of expressing the trend of fish spatial distribution. In four surveys, the separate degree of the analogue results of GAM was higher than that of GLM.The results showed that the predicting results of GAM was more accurate than those of GLM. The predicting results of GAM were similar to the measured data in four surveys. It gives the rules of chub mackerel spatial distribution with the help of using GLM and GAM in some extent. GAM displays an advantage in predicting the stock assessment of chub mackerel in relating to the fish habitat to the surface temperature and salinity in the ECS, because more peak values occurred in the GAM outputs.
收稿日期: 2009-12-25
引用本文:
倪一卓,程和琴,江红,韩震,刘瑜,傅文,王远飞. 鱼类栖息地模拟的比较研究-以东海鲐鱼为例[J]. 水产科学, 2009, 28(12): 726-732.
NI Yi-zhuo,CHENG He-qin,JIANG Hong,HAN Zhen,LIU Yu,FU Wen,WANG Yuan-fei. Fish Habitat Modeling in the East China Sea:A Case of Chub Mackerel. Fisheries Science, 2009, 28(12): 726-732.
