Spatial Variability Of Abundance Index For Ommastrephes Bartramii In The Northwest Pacific Ocean Based On Geostatistical Methods

Ommastrephes bartramii was an important fishing target for Chinese squidjigging fishery in the northwest Pacific Ocean. Its main fishing ground was located inthe water of150°-160°E and38°N-48°N. Its population structure and spatialdistribution all had much difference. And its fishing ground changed every year withthe variation of sea conditions, so its fishing season was also different. Therefore,further study for the spatial distribution, differences and change mechanism of O.bartramii resources, understanding the migration of fishing grounds and forecastingthe resource of O. bartramii were very important for scientific managing andsustainable developing O. bartramii resources.In this paper, geostatistics method was used inclduingspatial correlation, spatialinterpolation or estimates, variogram analysis and mutation analysis of quantitativeregionalized variablesto understand the spatial variability of abundance index forOmmastrephes bartramii and explore the relationship between the variability of O.bartramii resources and marine environmental factors. The main conclusions were asfollows:(1) Spatial scale analysis on abundance index of O. bartramii in the NorthwestPacific. This study took the statistical data of squid catches for example with dividedinto12different scales (10′×10′、20′×20′……120′×120′), the geostatistics method wasused to calculate the spatial distribution pattern of squid abundance. The resultsshowed that the trends of the nugget and sill value at different scales are basicallyconsistent, but other parameters of performance existed differences. Semivariogramwas fitted by the exponential model and the spherical model in the meso-scale andsmall-scale, the range was linear distribution, and the spatial autocorrelation wasrelatively strong. However, in the large-scale the semivariogram was fitted byGaussian model, the change process of range showed the trend of fluctuations and thespatial autocorrelation was relatively weak. (2) Spatial variability of small and medium scales’ resource abundance of O.bartramii in Northwest Pacific. The production data of O. bartramii in the waters of150°-160°E and38°-48°N in the North Pacific from August to October in2011wasused and one fishing vessel per day (CPUE) was considered as the abundance index,the geostatistics were used to analyze the spatial distribution characteristics of CPUEto explore the spatial variability of abundance index of O. bartramii under the sevenspatial scales, i.e. latitudes and longitudes for10′×10′,20′×20′,30′×30′,40′×40′,50′×50′,60′×60′and70′×70. The results showed that the best spatial autocorrelationheterogeneity fitting model was exponential model during August to October. Underthe small spatial scales (10′×10′,20′×20′and30′×30′), the spatial autocorrelationheterogeneity showed the middle level and above, while under the medium scales(40′×40′,50′×50′,60′×60′and70′×70′), the spatial autocorrelation heterogeneityshowed the low level. The spatial structure of O. bartramii shows anisotropic, themonthly angle directions during August to September were northwest-southeast andnortheast-southwest respectively, while it had great change of angle direction inOctober which may be affected by the marine environment and the beginning ofsouthwards-migration of O. bartramii owing to maturation. And the suitable scales forstudying the spatial heterogeneity of CPUE were small scales and30’×30’ showedthe most stable one.(3) Central fishing ground analysis of O. bartramii in the Northwest Pacificbased on different interpolation methods. Based on the resource abundance index(CPUE) from July to October in2012, the statistical interpolation methods are used toanalyze the fishing ground. In this paper, five kinds of interpolation methods (inversedistance weighted, radial basis function, global polynomial, local polynomial,ordinary Kriging) are used to establish models and evaluate them byroot-mean-square. The result shows that ordinary Kriging is the best one. Taking theoptimal method of ordinary kriging as an example, it uses the CPUE from July toOctober to simulate and predict, and then get the map of central fishing grounddistribution which indicates the migration of monthly central fishing ground.Comparing with actual operating positions and CPUE, it appears that there has littledifference between measured and predicted CPUE with the error less than0.15t/d,mean CPUE are from0.51to1.95t/d, but the measured and predicted maximumCPUE vary greatly in July, the difference is0.79t/d, and the measured and predictedminimum CPUE also has a big difference of0.22t/d in October. (4) The spatial variability precision comparison of O. bartramii resourceabundance based on Co-kriging interpolation method (COK) and ordinary kriging(OK). In this paper, fitting accuracy was compared between COK (CPUE as a majorfactor, Chlorophyll a and sea surface salinity as two assisting factors) and OK (CPUEas a single variable). The study suggested that the accuracy of ME(Mean Error) andMSE(Mean Standard Error) values of COK were less than that of OK, and DABS(Theabsolute value of the difference between the Root-Mean-Square Error and AverageStandard Error) value was also lower than OK. Thus, the fitting accuracy of COK wasbetter than OK, and the prediction accuracy of two different assist factors wereimproved2.21%and2.45%respectively. The two figures indicated that interpolationsurface of COK was continuous and smooth than that of OK. However, the spatialdistribution of CPUE that the figures both indicated were about the same.