Development And Application Of A Transfer Function Between Plant Species And PH In The Mire Surface Water In The Changbai Mountains,Northeast China

At present,paleoenvironmental reconstructions tends to be quantitative,highresolution and high-precision.The transfer function model is widely used in the paleoenvironmental reconstructions as an effective quantitative method.The pH gradient is one of the most important environment factors which affect the vegetation types,species composition in peatland.The establishment of plant macrofossil-pH transfer function model providing an extra method in quantitative paleoenvironmental reconstructions based on peat deposits in the future.During May to October from 2014 to 2016,174 samples were collected along the hydrological gradient from five peatlands namely Hani,Jinchuan,Hanlongwan,Xinglong and Dongfanghong,in the Changbai Mountains.All sampling spots were covered different habitats including the hollow,the top,middle and low part of hummock.The depth to water table(DWT),pH value,conductivity,TN,TP,Ca2+,Mg2+ were measured on spot.Ordination analysis show that pH is one of the principal factors which determine the composition of vegetation in peatlands.Thereby,construct a transfer function between plant species and pH in the mire surface water in the Changbai Mountains,and subsequently applied for a 50-cm profile with sphagnum of Hani peatland(HNS1).The results are as follows:(1)Detrended Correspondence Analysis(DCA)showed that the maximum gradient length of the sorting axis was 6.79,indicating that Canonical Correlation Analysis(CCA)performed more better in explore the relationships between vegetation composition and 7 environmental variables.CCA and RDA analyses revealed that the pH individually explained significantly(Monte Carlo significance tests p<0.001)2.3% and 3.1% respectively,suggesting that pH is the principal factors after DWT.Therefore,pH can be used as the variable to construct the plant species-pH transfer function in the Changbai Mountains.(2)The weighted average model(WA)and weighted averaging partial least squares model(WA-PLS)were used to develop the transfer function.When full of the samples are used in training set,the prediction performances are poor.With –outliers,20% data filtered,the performance of models were gradually improved.Compare the regression value r2 jack and the root mean squared error of prediction(RMSEPjack),We found the WA.inv.tol of the WA was the best model(RMSEPjack = 0.3344,r2 jack = 0.3126).(3)Optima and tolerances from 47 species of helophytes were obtained through the plant species-pH transfer function based on the WA model.Mosses were located on the low value side of the entire pH gradient in general,especially the sphagnum.Vascular plants had a broad distribution on the gradient,while the optima and tolerances were species-specific.(4)WA.inv.tol and WA-PLS Comp01 were used to reconstruct the pH value of mire surface water in HNS1 profile(1957-2006).The results showed that the reconstruction results of two models were similar and WA.inv.tol possesses smaller bias,verifying that WA.inv.tol was the best model.(5)Correlation analyses between the reconstructed pH value of the mire surface water and the pH value of peat soil,the concentrations of Ca2+,Mg2+,humic acid,humification(measured in the profile)showed that the reconstructed pH was positively correlated with the concentration of Ca2+,while in negative proportional to humic acid and humification.Comparison between the pH value of mire surface water and peat soil revealed that the pH value of surface water was higher than that of peat soil above 20 cm,while turns to opposite situation below 20 cm,which may be relevant to the composition of materials in profile and the status of water supplement,resulting in reciprocal transformation of organic acid and based edaphic ions happened in watervegetation-peat soil sphere.