Study On CPUE Standardization Of Indian Ocean Swordfish(Xiphias Gladius) Based On Observer Data

Xiphias gladius is a top predator widely distributed in the three oceans.It is also the main catch of tuna longline fishery in China.Catch per unit effort(CPUE)is a key statistical index to express the status of fish stocks.Accurate and reliable CPUE plays an important role in the evaluation of swordfish resources in the Indian Ocean.In tuna longline fisheries,many previous CPUE studies have shown that CPUE is affected by many factors and difficult to standardize,such as resource abundance,fishing efficiency,environmental effects,fishing gear renewal and conservation and management treaties.In the research on CPUE standardization of longline fishery,previous studies mainly focused on the comparative study of different models,parameter construction,target strategy effect,the impact of data spatial scale and so on.This paper starts to analyze the difficulties in the standardization process of Indian Ocean swordfish CPUE and selects three angles to study the standardization process.The observer data includes the date of each operation of the relevant voyage of the fishing vessel,the longitude and latitude of the operation position,the number of production hooks,the start and end time of hook release,the start and end time of hook lifting,the type of hook,bait composition,the type,weight,and mantissa of catch,etc.Therefore,this paper integrates the fishing data from 2012 to 2019 and the environmental data with monthly time resolution based on remote sensing images,and uses the data of Chinese tuna longline fishery observers from 2012 to 2019,With the help of the excellent accuracy and data richness of observer data:(i)Comparison of generalized linear model(GLM)and generalized additive model(GAM)in CPUE standardization of swordfish in the Indian Ocean;(ii)the differences in the standardization process of four nominal CPUE of different fishing efforts(number of hooks,hook soaking time)and the combination of two catches(mantissa per thousand hooks and weight per thousand hooks);(iii)Environmental factors(sea surface temperature,SST,sea surface salinity,SSS,current velocity,CV,mixed layer depth,MLD,sea surface height and sea surface chlorophyll a concentration,SSC)and fishing gear operation factors(hooks between floats,Hb F,bait type,hook type,whether fluorescent rod is used or not).The main results are as follows:(1)When comparing the differences of different statistical methods in the standardization of the Indian Ocean swordfish CPUE,the significance of the GLM model and the GAM model both show that the significance of the temporal and spatial factors is far greater than the influence of environmental factors and fishing gear operating parameters.more significant than environmental factors.(2)In this study,GAM explained the average deviation of CPUE data more effectively than GLM,the deviation explanation rate of GAM was 52.8%,which was higher than that of GLM 49.7%;the adjusted R2 of GAM was 0.428,which was also high 0.453 of GLM;AICc(Corrected Akaike information criterion)of GAM is smaller than GLM model.In this study,GAM outperformed GLM.Therefore,GAM is likely to be more suitable for CPUE normalization of Indian Ocean swordfish than GLM.In future standardization of CPUE,the nonlinear relationship between predictors and CPUE should be prioritized.(3)In describing the nominal CPUE of tuna longline fisheries,catch is often expressed in terms of catch weight or number of catches,and effort is expressed in number of hooks deployed or hook soaking time.In different nominal CPUE models,the adjusted R2 range is 0.429-0.463,and the model’s deviation explanation rate is between 50.1% and 54%;The coefficients range from 0.941 to 0.993.The results reflect that the difference between the normalized CPUE of the number of hooks and soaking time is very small,and the difference between the normalized CPUE of the catch weight and the number of catches is small,but the difference is small.Compared with counting the number of individuals to express the population,the catch weight is more scientifically meaningful in counting the biomass of the population and expressing the difference between individuals.Through the comparative analysis of this study,we suggest that under the condition of limited data collection budget,fishing logs should give priority to ensuring accurate records of swordfish catches and hooks.(4)It is a hot topic to explore the relationship between environmental factors and species resources.However,whether to add environmental factors to the CPUE standardization process has always been an important topic discussed in the academic circles.The results of this paper show that the standardized CPUE of the environmental model and the comprehensive model of the fishing gear model are significantly different,with a correlation of 0.50-0.53 for the annual trend,while the correlation between the fishing gear model and the comprehensive model is 0.99.The comprehensive model has the best performance in the comparative study in this paper,with an R2 of 0.459 and a deviation explanation rate of 52.6%;the final comprehensive model explanatory variables include year,month,longitude,latitude,number of hooks between floats,hook type,and bait type.,sea surface salinity and velocity.(5)Judging from the annual standardized CPUE,the standardized CPUE of Indian Ocean swordfish showed a downward trend from 2012 to 2015,and a steady upward trend from 2015 to 2019.Judging from the monthly normalized CPUE,the normalized CPUE of the Indian Ocean Swordfish showed a stable and fluctuating trend except for the decline in some months from 2012 to 2013.