Prediction And Analysis Of The Spread Trend Of Three Invasive Plants Of Compositae In China Based On The Combination Model (EM)

Biological invasion is an important ecological security problem facing the world.The prevention and management of invasive plants has become an important research direction to maintain ecosystem stability and biodiversity in China.China is one of the countries with the most serious invasion of exotic species.At present,the number of invasive plants in southern China is far more than that in other provinces,and the proportion of invasive plants in Compositae is more than half.Ageratina adenophora,Eupatorium odoratum,and Mikania micrantha are three destructive invasive plants of Compositae in southern China.They invade China mainly through a large number of seeds that are transported by wind,rivers,animal transportation.Since the proliferation trend of invasive plants has an obvious response to climate change,the species distribution models(SDMs)are used to simulate and predict the invasive distribution and trend of alien plants under climate change.Then,the feasible suggestions were put forward to achieve the purpose of preventing invasive plants.In this study,an ensemble model(EM)constructed by 4 single-algorithm models was selected as the final predictive model.In the modeling process,the TSS and AUC values are used as the evaluation indicators of model accuracy,and the single algorithm model is screened twice according to the size of the TSS value.Random forest(RF),gradient boosting model(GBM),artificial neural network(ANN)and flexible discriminant analysis(FDA)with TSS average value greater than 0.8 were selected for the first time.After 10 times of modeling of every single model,a single model with a TSS value greater than 0.9 was chosen to construct EM.EM effectively reduces the uncertainty of the SDMs single algorithm model and improves the fitting accuracy of the model.Using the sampling point data of the three-time periods before 2000,2000-2010,and after 2010,combined with the data set containing only 19 bioclimatic variables(Set?),and the 22 environmental variable data sets(Set ?),which including screened climate,soil,topography,and human activities,to simulate the current invasion range of three invasive plants in Compositae.The vegetation canopy closure and the prediction results of Set ? are superimposed(Set ?)to analyze the limitation of vegetation canopy closure on the invasion of three kinds of plants.Then,using Set ? after removing human activities,predicted the proliferation trend of three invasive plants of Compositae under three different emission scenarios(RCP2.6,RCP4.5 and RCP8.5)during the 2050s and 2070s.Ipomoea batatas have the inhibitory effect on the invasion of A.adenophorum and M.micrantha,and Vitex negundo has a strong competitive effect on E.odoratum.They could become an important alternative cash crop for ecological control of the invasion of A.adenophorum,E.odoratum and M.micrantha.The paper simulated the suitable planting areas of I.batatas and V.negundo in southern China,and compared them with the proliferation area of the invasion risk area.Then,analyzed the feasibility of planting I.batatas and V,negundo in a large range to prevent the continuous spread of three exotic plants.By predicting the spread trend of invasive plants and simulating suitable areas for competitive crops that can block the invasion,the paper will provide practical references and suggestions for the management,prevention and control of plant invasions.The main conclusions are as follows:(1)In ANN,FDA,GBM,RM and EM,the accuracy of EM is higher than that of the four single algorithm models.Although the fitting degree of the model using Set ? and Set? is not obviously different,the latter is slightly higher,indicating that the collinearity of variables limits the fitting accuracy of the model,and the collinearity between data can be reduced after variable screening.(2)The invasion areas of the three exotic plants have spread rapidly in the past 30 years,and their high-risk areas are mainly distributed in the water areas along the rivers in the southern provinces and cities in China.A.adenophorum mainly expended from Yunnan to the north and the east.E.odoratum first invaded eastward from Yunnan to Guangdong and then spread northward a lot.M.micrantha was rarely found before 2000.Then,it invaded the coast of Guangdong and Guangxi rapidly to the north and inland after 2010.(3)In the future scenario,the three invasive plants mainly invade Yunnan,Sichuan,Guizhou,Jiangxi and Fujian.Both A.adenophorum and E.odoratum have the most obvious invasion trend and will reache peak state in the RCP4.5 scenario of the 2050s.M.micrantha has the largest high-risk invasion area in the RCP8.5 scenario of the 2070s.(4)A.adenophorum has low requirements for precipitation(Bio17?249mm),while the precipitation(Bio17?629mm)of E.odoratum in the invasion area is richer but still lower than the demand of M.micrantha(160mm ?Bio16 ?2133mm).The growth of A.adenophorum and E.odoratum also requires a lower temperature difference,while the growth of M.micrantha grows in a warmer and moist environment.In addition,the wind speed and direction are also consistent with the diffusion of the three invasion plants,and the vegetation canopy closure is an important variable to limit the expansion of invasive plants.Climate is the most important variable that affects the spread of the invasion of three exotic plants,and terrain,soil,and human activities are all restrictive factors for the invasion plants.(5)I.batatas and V.negundo are suitable for growing as cash crops in a wide range.In Sichuan,Chongqing,Yunnan,Guizhou,Guangxi,Guangdong,Jiangxi with serious invasion trends,the planting suitability of I.batatas and V.negundo is higher.Therefore,I.batatas and V.negundo as the competitive crops of the three invasive plants,can play a protective role when planted on the diffusion trend line.