Figure legends
Figure 1 Three structure equation models (SEM) with different assumptions on direct and indirect relationships between environmental factors and topsoil Δ14C are compared based on the available datasets for alpine grasslands provided by Chen et al. (2021). (a) Model-A includes direct effect of plant carbon input and indirect effect of precipitation on Δ14C. (b) Model-B adds direct effect of precipitation on Δ14C. (c) Model-C adds direct effect of precipitation and mineral protection and indirect effect of plant carbon input on Δ14C. Fit indices, including degree of freedom (DF ), Chi-squareis (χ2 ), probability level (P ), Akaike information criterion (AIC ), comparative fit index (CFI ), goodness of fit (GFI ), root mean square residual (RMR ) and root mean square error of approximation (RMSEA ), are listed on the left panel of each model. Similar to Chen et al. (2021), precipitation is the first principal component (PC1) of mean annual precipitation (MAP), precipitation of the wettest month (PWM), and precipitation of the wettest quarter (PWQ); Plant C input is the PC1 of plant carbon input in topsoil, normalized difference vegetation index (NDVI), enhanced vegetation index (EVI) and leaf area index (LAI); mineral protection is the PC1 of molar ratios of dithionite- and oxalate-extractable Fe/Al oxides to SOC (Fed + Ald and Feo + Alo), and the molar ratios of exchangeable Ca2+(Caexe) and Mg2+(Mgexe) to SOC. Logarithm transformation is performed for the four variables of mineral protection before principal component analysis (PCA).