Influence Of Spatial Scale Selection Of Environmental Factors On The Prediction Of Distribution Of Coilia Nasus In Changjiang River Estuary

Due to the influence of human activities,climate change and other aspects,the Marine environment and ecosystem have undergone great changes,and suffered a lot of damage.As the junction of river and ocean,estuary is an important part of coastal zone,with special biological characteristics and a variety of ecological functions.The ecological characteristics of the river inlet section in the estuary are close to those of the river,while the coastal area outside the estuary is dominated by Marine characteristics.Therefore,the estuary area is a complex environment of fresh-brackish water-brackish water transition,so some Marine environmental elements will undergo drastic changes in a small range.It is important to understand the influence of spatial scale selection on distribution prediction of estuarine species.As the largest estuary in China,the Yangtze River Estuary is located at the junction of the end of the Yangtze River and the East China Sea,which has an important ecological status.It is not only the spawning ground,feeding ground and migration channel of many species,but also the habitat of a large number of young and young fish and migratory fish.In addition,it was once a famous fishing ground with rich water quality,which provided a favorable living environment for a variety of economic fish and shrimp,such as Lateolabrax Maculatus and Palaemon Gravieri.It is also home to a variety of rare species,such as Acipenser sinensis,NeopHocaena Asiaeorientalis and Anguilla marmorata.In our field survey,Coilia nasus was the most common catch in the trawl survey,and Coilia nasus is not only one of the main economic fish in the Yangtze Estuary,but also a famous migratory species.It is a common research to use environmental factors to model and then use the model to predict fishery resources.GAM model is one of the multiple regression model,which can be used to describe the nonlinear relationship between the dependent variable and multiple independent variables,the space-time distribution has better response variables prediction ability,in the statistical analysis of data of high flexibility,so on the analysis of the fishery resources density distribution and its relationship with environmental factors has a good performance,It is widely used in the prediction of fishery resources.Spatial interpolation is an important approach to obtain small regional Marine environmental elements,the interpolation process may choose different space scale characteristics for the forecast of some environmental factors,the change of environment will potentially affect the biological distribution prediction result,so the different interpolation space scale environmental factors on biological effects of time and space distribution pattern is very important.In this study,we took Coilia nasus,an important economic fish species in the Yangtze Estuary,as the research subjects,and applied a second-order Generalized additive model(GAM)to investigate the spatial distribution of Coilia nasus at different spatial scales(1 ’×1’,2 ’×2’,3 ’×3’,4 ’×4’,5 ’×5’)in response to environmental factors.Our results show that :(1)according to variance inflation factor(VIF)analysis,the values of all environmental variables are less than 3,so there is no need to delete any environmental factors before backward stepwise regression.(2)The optimal model of first-order GAM contains six variables: Year,Month,latitude(Lat),water temperature(Tem),salinity(Sal)and chemical oxygen demand(COD),and the optimal model of second-order GAM contains four variables: Year,latitude,pH and chloropHyll A concentration(Chla).The two variables of year and latitude were retained,indicating that these two spatio-temporal factors could affect both the presence and abundance of Coilia nasus.(3)At different spatial scales,the mean predicted values of 3 ’×3’ Chla were significantly lower than those of other spatial scales,and the mean predicted values of 5 ’ × 5’ Sal were higher than those of other spatial scales,indicating that different spatial scales had a significant influence on the final interpolation results.(4)For environmental factors with little variation in estuaries,such as ph,the interpolation results on a small scale can see the subtle variation.But for dramatic changes in the estuary area,the highest and the lowest differences of environmental factors,such as salinity,distribution characteristics of all scales are similar,This shows that the sensitivity of different environmental factors to spatial scale is different.(5)In terms of spatial distribution of abundance,in most cases,the predicted results of various spatial scales during the same period were similar,but not all the predicted patterns of Coilia nasus distribution at all scales were very similar.The predicted distribution patterns of Coilia nasus at 5 ’×5’ in winter of 2012 were significantly different from those at other scales.Therefore,we think that the choice of time and space scale the interpolation results of environmental factors,especially for some scale is more sensitive to the space environment factors,different environmental factors of interpolation results after the substitution model,and further to predict the spatial distribution of species caused certain effect,but the specific reasons and forms of the impact is not yet clear,We therefore suggest that future studies need to continue to assess the potential effects at the spatio-temporal scale.