Effects On Spatial Interpolation Methods Of Environmental Factors Of The Distribution Prediction Of Coilia Nasus In The Yangtze River Estuary

With the increasing impact of human activities and global climate change on the marine environment,the marine ecosystem have been seriously damaged.Spatial-temporal distribution of marine fishes is strongly influenced by environmental factors.Spatial-temporal distribution of marine fishes is strongly influenced by environmental factors.At present,the acquisition of marine environmental data mainly includes satellite remote sensing data,field observation,numerical simulation and acoustic detection.Field observation is the main survey means for nearshore marine environment investigation and fishery resources monitoring because of its most direct and high accuracy.However,field observations are usually conducted in accordance with the stationary sampling designs,so continuous environmental data cannot be obtained directly.The spatial interpolation methods(SIMs)are one of the effective methods to obtain the continuous distribution of environmental factors in this point-based data acquisition method.SIMs can reconstruct the missing values of the data and evaluate the attribute information of the unsampled region to obtain continuous data.It has been widely used in the spatial continuous distribution of marine environmental factors.However,there are many factors that affect the results of interpolation.How to choose an appropriate method for a given data set is the focus of current research.The Yangtze River Estuary(YRE)is the largest estuary in China,and its complex ecological environment affects the growth,reproduction and larval process of fishes in this area.Coilia nasus is an important traditional economic fish in the YRE.As a result of increasing fishing activities,the resource of Coilia nasus has been overfished.Monitoring the change of marine environmental accurately can evaluate the changes of its resources better in the YRE.In the study of the relationship between spatial distribution of fishes and environmental changes in the YRE,the potential impacts of different SIMs on environmental factors and abundance prediction should be taken into account.Determining the best combination of spatial interpolation methods can reduce the uncertainty of the prediction results of spatial and temporal distribution of fishes.In this study,we use different SIMs including Inverse Distance Weighted(IDW)interpolation,Ordinary Kriging(OK)(semivariogram model: exponential(OKE),gaussian(OKG)and spherical(OKS))and Radial Basis Function(RBF)(Regularized Spline Function,RS and Tension Spline Function,TS),to obtain continuous environmental variables(water depth,water temperature,salinity,p H and chlorophyll-a)in the YRE.The accuracy and effect of SIMs were cross-validated,and two-stage Generalized additive model(GAM)was used to predict the distribution of Coilia nasus from 2012 to 2014 in YRE.The results show that:(1)Based on VIFs analysis,the values of DO and salinity were larger than 3.However,salinity was reconsidered in subsequent analysis since it is an important variable influencing changes in the spatial distribution of fish communities in most estuaries.Moreover,salinity had the highest correlation coefficient between environmental variables,particularly with COD.Because of this,DO and COD were then removed from subsequent analysis because of collinearity.The remaining variables used in our next steps were longitude,latitude,water depth,water temperature,salinity,p H and chlorophyll-a,which had VIF values of 1.74,2.38,1.24,1.21,2.57,1.31 and 1.26 respectively.After implementing the AICc approach,the model explained of first stage GAM model deviation accounted for 30.1% of the variance,and had an AUC value of 0.85,which better predicted the probabilistic presence of Coilia nasus.The second stage GAM model deviation accounted for69.0% of the variance,with RMSE of 324.50.(2)The values of environmental variables estimated in YRE by various SIMs varied significantly.The cross-validation of the interpolations for each environmental factor showed that the optimal SIMs applicable to each variable differed.IDW was the most suitable method for water depth and chlorophyll-a in the YRE.Meanwhile,the method that well suited water temperature and salinity was RS,and the applicable method for p H was OKG.The mean value of each variable were close to the measured data.Although the estimated values of environmental variables obtained by various SIMs differed,SIMs generally still did not change the overall distribution trend of environmental variables.When small amounts of variation were observed in the raw data of the environmental variables,such as water depth,water temperature,p H,and chlorophyll-a,the interpolated values from different SIMs were similar.For the estimated values of salinity in YRE,the spatial interpolation methods showed significant differences,in terms of the characteristics of high homogeneity and the distribution of specific sampling sites.(3)Different SIMs obtained different estimated values of the environmental variables,which in turn will affect the predicted values of the abundance of Coilia nasus.The mean value of the predicted abundance of Coilia nasus by each method were similar,but the maximum value varied significantly.Results showed that the spatial trend of the predicted abundance of Coilia nasus based on the estimated environmental values obtained by SIMs were similar,and the abundance on the inside of the estuary was significantly larger than outside.However,there were differences in the spatial distribution and abundances trends,The semivariogram models of OK also showed differences,while the abundances obtained using the two RBFs had similar spatial distribution maps of predicted values.The values and distribution trends of the predicted abundance values based on the most suitable SIMs of each environmental factor were between the maximum and minimum values of the other methods.