Data Analysis
After ensuring data met normality and homogeneity of variance assumptions using the Shapiro-Wilk test, we evaluated the influence of grazing treatment on aboveground and belowground biomass, plant nitrogen and carbon content, plant functional groups and several soil chemical variables, as well as the ecosystem CO2 exchange and soil respiration. To do so, we used repeated measures ANOVA to test the effects of grazing intensity and sampling month on the aboveground biomass, plant functional group biomass, ecosystem CO2exchange and soil respiration. We used one-way ANOVA followed by a Duncan test for pairwise comparison to test the effects of grazing intensity on the belowground biomass, plant total carbon, plant total nitrogen and soil nutrient content. A P < 0.05 indicated significance in the treatment effects.
We correlated several abiotic factors with ecosystem carbon exchange, including temperature, precipitation, soil temperature, and soil moisture in each treatment using regression analysis.
To investigate the influence of soil and plant factors on ecosystem carbon exchange, we used redundancy analysis to rank the impact of the factors on carbon exchange. Furthermore, we used a generalized linear model (GLM) and structural equation model (SEM) to determine the effects of plant and soil factors on ecosystem CO2 exchange and soil respiration. To do so, we first calculated the contribution of the plant and soil factors to the ecosystem CO2 exchange and soil respiration using the GLM, and then we removed insignificant pathways and simplified the SEM model based on the GLM results. We obtained path coefficients using a maximum likelihood estimation technique.
We performed ANOVA, repeated measures ANOVA and the GLM analyses in version R 4.0.3. The SEM analyses were performed using the “piecewise SEM” package (Lefcheck, 2016) in R version 4.0.3. We performed regression and redundancy analyses in Origin 2023 software.