2.7 Statistical analyses
Descriptive statistical analysis of the changes in root and foliar concentrations of N and P and N:P ratios were performed by SPSS 17.0. Coefficient of variation (CV) across three study areas was calculated. For plant nutrient data were tested for normality of distribution using Lilliefors and Shapiro-Wilk tests, homogeneity of variance was tested using F and Levene tests. Multiple comparisons of plant nutrient variables mean were carried out using LSD‘s test for unequal sample sizes with 95% confidence intervals.
The “Varpart” function in the “Vegan” package was used to partition the variation of root and foliar nutrition traits (N and P contents, and N:P ratio, into components due to three categories of predictors (i.e. soil chemistry factors, climatic factors, and ECM trait factors) in RStudio.
By loading the ”lavaan()” function in RStudio, we performed a structural equation models (SEMs) among soil factors, ECM traits and plant nutrients (N and P content, and N:P ratio). Firstly, we identified key soil factors and ECM traits by principal component analysis (PCA), and the first principal components and second principal components of soil factors (named Soil-PC1, Soil-PC2 respectively) and ECM traits (named ECM-PC1, ECM-PC2 respectively) were selected for SEMs analyses. Then we constructed SEMs for the effects of soil and ECM PC axes and climate factors (MAT, MAP) on root and foliar nutrients (i.e. concentrations of N and P and N:P ratio). Besides, Redundancy analysis (RDA) was used to determine the relationships of plant nutrient traits and ECM root traits across all sites and elevations. Hellinger or standardize transformation was used to transform the data of plant nutrient traits and ECM root traits for RDA.
Curve Estimation models were used to estimate the relationships of root and foliar nutrient variables with the colonization ratio of the soil exploration types (n =9). Of which, data of the colonization ratio of MDE and root P were transformed by SQRT for the regression analysis.