Fishery Forecast Research Of Sthenoteuthis Oualaniensis In The Open South China Sea Based On Habitat Suitability Index Model

The South China Sea,located in the south of mainland China and west of the Pacific Ocean,is the largest and deepest of China’s three marginal seas,with a natural sea area of about 3.5 million square kilometers,vast waters and rich natural resources,and is an important fishing ground for China.In recent years,offshore fishery resources are increasingly depleted due to overfishing,in order to effectively curb the situation,reduce the intensity of offshore fishing,the development of pelagic fisheries is particularly important.There are two traditional fishing grounds in China in the open South China Sea,which have rich resources such as tuna and purpleback flying squid,however,the annual catch of these resources in China is very little and the development efforts are far from enough.Therefore,it plays a very critical role to carry out the research on fishery forecast of Sthenoteuthis oualaniensis in the open South China Sea to guide the scientific production of pelagic fisheries and optimize the management of fisheries resources.This article explored the influence of the marine environment on the habitat of the open South China Sea Sthenoteuthis oualaniensis by using the CUPE with five environmental factors:sea surface temperature（SST）,sea surface salinity（SSS）,sea surface height（SSH）,photosynthetically active radiation（PAR）and chlorophyll-a（Chla）from January to April 2019-2021,based on the generalized additive model（GAM）and geographically weighted regression（GWR）model.The result shows that:the optimal GAM model consists of SST,SSS,SSH,PAR and Chla.The total deviance explained of the model was 42.3%,and the deviance explained of each factor is SST（29.40%）,PAR（20.10%）,SSH（3.10%）,SSS（1.10%）and Chla（1.00%）.The optimal GWR model consists of SST,SSS,SSH and PAR,the local regression coefficients of all factors are positively correlated with CPUE of purpleback flying squid.Among them,SST is 92.23%positively correlated with CPUE,SSS is 81.55%positively correlated with CPUE,SSH is 95.15%positively correlated with CPUE,PAR is99.03%positively correlated with CPUE.The spatial non-smoothness is weaker in the sea area with dense islands and reefs in the open South China Sea,and stronger in the sea area with more dispersed islands and reefs.Comparing the Akaike information criterion（AIC）and the regression coefficient R²,the GWR model is significantly better than the GAM model,and the environmental factors affecting the distribution of Sthenoteuthis oualaniensis are SST,SSS,SSH and PAR.In order to better understand and sustainably exploit Sthenoteuthis oualaniensis resource,this article used the fishing data of S.oualaniensis in the open South China Sea（110—118°E,6—-13°N）during the fishing season January-April from2019—2021 and combined with sea surface temperature（SST）,sea surface salinity（SSS）,sea surface height（SSH）and photosynthetically active radiation（PAR）,the habitat suitability index model（HSI）was established based on four types of marine environmental factors by using weighted mean model（WMM）,and the optimal weighting ratio of each factor was derived by non-linear regression.The model prediction results were validated based on the 2022 fishing data and the corresponding marine environmental data,and the changes of the central fishing ground in each month were compared and analyzed.The results show that the forecast accuracy of each month is above 80%,including 86.48%in January,80.64%in February,85.92%in March and 82.14 in April,with an average forecast accuracy of 83.93%.The average forecast accuracy of HSI grade is 82.79%,including 100%for grade 1,100%for grade 2,83.91%for grade 3,67.41%for grade 4 and 62.62%for grade 5.The average forecast accuracy of grade 1 is 100%,grade 2 is 100%,grade 3 is 83.91%,grade 4 is 67.41%,and grade 5 is 62.62%,the average CPUE weights under each class were 5.41%,14.27%,23.77%,32.25%,and 24.30%.The average forecast accuracy of HSI grade 4 and grade 5(I_HSI>0.6)are above 60%,and the sum of the average weight of both CPUEs reach 56.55%,indicating that the HSI model based on different weights can better predict the distribution of S.oualaniensis fishery.