2.4 Statistical analyses
To evaluate the impacts of inter-annual precipitation fluctuation and experimental treatments on plant, soil and AM fungal parameters, repeated measures analysis of variance (ANOVA) was used with N, P and fungicide application as between-subject factors, and year as a within-subject factor. In addition, three-way ANOVA (N, P and fungicide application) was used to test the impacts of experimental treatments on these parameters. Statistical analyses were conducted via SPSS software (version 21.0; IBM, Armonk, NY, USA).
To evaluate the direct and indirect impacts of the experimental treatments on ANPP, we performed structural equation modelling (SEM) in Amos (version 18.0, IBM, USA). In our SEM model, we supposed that precipitation, N and P addition and AM fungi can impact ANPP both directly and indirectly by altering the aboveground biomass of main plant species. We examined the model’s fitness using the χ2 test, root mean square error of approximation and Akaike information criterion.χ2 values and P -values > 0.05 suggested that the covariance of the chosen data was no significantly differ with which we expected (Grace, 2006).