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.