Differences in ecosystem CO2 exchange under
different grazing intensities
During the 2020 growing season, we found that NEE, ER, GEP, and SR
showed strong seasonal dynamics consistent with the monthly variation in
precipitation (Fig. S2). According to the results of repeated measures
ANOVA, NEE, ER, GEP, and SR varied significantly between months. NEE and
GEP also varied significantly between grazing intensities and the
interaction between month and grazing intensity, but ER and SR did not
differ significantly between grazing intensities or the interaction
between month and grazing intensity (Table 2). During July, NEE was
positive indicating release as a carbon source (Fig. 4a); it was
negative, indicating a carbon sink for the rest of the growing season.
Both NEE (Fig. 4a) and GEP (Fig. 4c) were lowest in August, while ER
(Fig. 4b) and SR (Fig. 4d) were highest in August. When we compared
grazing treatments, we found that the rates of NEE (Fig. 4a), ER (Fig.
4b), GEP (Fig. 4c) and SR (Fig. 4d) were all significantly lower than
the control plots, with the heavy grazing treatment often having lowest
(or highest) values.
plant factors and soil
factors on ecosystem CO2 exchange
We used RDA model to examine the relationship between the explanatory
variable (plant and soil factors, blue lines with arrows) and response
variable (ecosystem carbon exchange and soil respiration, red lines with
arrows) in Fig. 6. We found that plant factors (e.g., above and below
ground biomass, plant carbon and nitrogen nutrients) explained 98.10%
of the variance of ecosystem CO2 exchange and soil
respiration (Axis 1 explained 71.49 % of the total variance while Axis
2 explained 26.61%; Fig. 6a). Soil factors (e.g., Soil nutrient index)
explained 98.20 % of the variance of ecosystem CO2exchange and soil respiration (Axis 1 explained 73.50 % of the total
variance while Axis 2 explained 24.70 %; Fig. 6b). For plant and soil
factors, SS (R2 = 0.36) contributed the highest degree
of variance to NEE, and next highest was AGB (R2=
0.21, Fig. 5c); AGB (R2 = 0.28) contributed the
highest degree of variance to GEP, and next highest was SS
(R2 = 0.22, Fig. 5E); BGB (R2 =
0.25, R2 = 0.23) contributed the highest degree of
variance to ER and SR (Fig. 5d, Fig. 5f);
Based on the results of the redundancy and GLM analyses, we developed
structural equation models to better explain the driving mechanisms of
ecosystem carbon exchange and soil respiration. Our SEM analysis showed
that grazing had a direct negative effect on NEE and GEP. Specifically,
grazing reduced NEE and GEP by reducing aboveground biomass, especially
through the indirect reduction of NEE due to lower shrub and semi-shrub
biomass (Fig. 6a and c). However, the lower soil nutrient content in the
grazing treatment was not associated with NEE and GEP (Fig. 6e and g).
In contrast, grazing and aboveground biomass did not directly affect ER
and SR (Fig. 6b and d), but they did directly affect belowground biomass
and indirectly reduce belowground biomass by decreasing ammonium N. This
came to affect the rate of SR as well (Fig. 7f and h).