Integrated Habitat Index Of Blue Shark(Prionace Glauca) In Waters Near Marshall Islands

Blue shark (Prionace glauca) is one of the most important bycatch species in longline fisheries. Studies on the habitat of blue shark will benefit to the conservation of its resources and stabilize the marine ecosystem, and it also satisfies the requirements of international organizations on the reduction of shark bycatch management, it's an important part of responsible fisheries.This study is based on a survey carried out by the longliner"Shengliancheng 719"in waters near Marshall Islands from Oct, 2006 to May, 2007. Based on the survey data collected by"Shengliancheng 719"at 43 sampling stations, the vertical profile data of temperature, salinity, dissolved oxygen concentration and the catch rate data of blue shark were applied to develop the"Integrated Habitat Index (IHI)"models by the quantile regression method and GLM model. Models were developed for six water strata from 40m to 280m (40m each) and the entire water column to understand the blue shark's spatial distribution.Environmental factors which input into two kinds of model, which were the integrated habitat index models of various water layers based on quantile regression (IHIQRj) and GLM (IHIGLMi) respectively, were same. These environmental factors were the arithmetical average temperature, salinity and dissolved oxygen in various water layers, and their interaction. IHI model of the entire water column ( IHI QR, IHI GLM) was developed using the weighted average temperature, salinity, dissolved oxygen concentration and the thermocline parameters, such as, up limited depth, low limited depth, up limited temperature, low limited temperature, intensity, thickness, temperature difference.Different IHI models were compared and verificated. The paired two sample t-test was used to calculate the Poisson correlation coefficients between predicted IHI and observed CPUE of five strata and the entire water column. The Poisson correlation coefficients were assumed to indicate the predictive power of the IHI model. The trend line of the arithmetic average about the predicted IHI of five strata was compared with those of the catch rates in the specific depth stratum. The measured environmental variables at the other 18 sampling stations were used to validate the models' predictive power. These data were input into the CPUE models to predict the CPUE of five strata and the entire water column. A comparison between the predicted and observed catch rates at five strata and the entire water column was conducted by a paired two sample t-test to validate the IHI models. The results showed:(1) the key environmental parameters in the IHI models differed among the depth strata. This might be resulted from the blue shark's specific biology characteristic, the different ecological factor, and the different life stage of blue shark;(2) the highest value of IHIi is at 80~120m water layer, it indicates that blue shark prefer to swimming in this water layer. The values of IHIQRij and IHIGLMij were compared. The result showed that, IHIQRij index value were higher than IHIGLMij index value in the same water layer;(3) the comparison of IHI QRbased on Quantile Regression with the observed CPUE of blue shark shows that the predicted CPUE is greater than that of the observed CPUE, Pearson correlation coefficient is 0.75(P<0.05), the models'predictive power was good.(4) IHI GLM model based on GLM correlated significantly with blue shark's CPUE(P<0.05), and the correlation coefficient is 0.88. The distribution of IHI GLM and IHI QR are very similar, and both of the indexes are higher in the area of 9oN～12oN,172oE～176oE. In general, IHI QR index is higher than that of IHI GLM.(5) The blue sharks' IHIs in the areas of 172°E~176°E, 9°N~12°N and 166°E~169°E, 3°N~6°N were relative higher. We suggested setting the baited hooks below 120m in two areas where IHI are high to reduce the bycatch of blue shark.(6) The measured environmental variables at the other 18 sampling stations were used to validate the models' predictive power. These data were input into the CPUE models to predict the CPUE of five strata and the entire water column. A comparison between the predicted and observed catch rates at five strata and the entire water column was conducted by a paired two sample t-test to validate the IHI models. The results showed: the predicted and observed catch rates at five strata and the entire water column have remarkable correlation, and have no difference except 200~240m water strata. IHI models can be used to pridect the abundance distribution of blue shark in waters near Marshall Islands.The IHI QR model is suggested to be used for the prediction of abundance of potential blue shark resources in waters near Marshall Islands. The IHI GLM model is suggested to be used for the study of real abundance of blue shark resources in waters near Marshall Islands. We suggest to using quantile regression to develop the integrated habitat index models for various water layers.