Root-derived C inputs
Root-derived C accumulation decreased with increasing ECM dominance at the plot-level (Fig. 1a; R2 = 0.09, P = 0.013), and ECM dominance and site together predict root-derived C (R2 = 0.30, P = 0.004). There was no significant relationship between root-derived C accumulation and root production (Fig. 1b). Moreover, we found no significant relationship between metrics of soil N cycling and root-derived C (data not shown).
Comparing the three most AM-dominated and three most ECM-dominated plots only (i.e. ‘end-member plots’), root-derived C was greater in AM plots compared to ECM plots across sites (F1,29 = 5.33,P = 0.028) with no statistically significant difference among sites (Fig. 2a). The highest root-derived C inputs were observed in AM plots at HF (418 ± 151 g C m-2 yr-1) while the lowest C inputs were observed in ECM plots at TRC (114 ± 28.3 g C m-2 yr-1). Annual root-derived C generally exceeded aboveground litter mass C inputs (Fig. 3; SI Fig. 2). The ratio of annual root-derived C: litter mass C was significantly predicted by site (F5,29 = 3.60, P =0.012) and plot mycorrhizal type (F1,29 = 7.20,P = 0.012), with no significant interaction between site and mycorrhizal type. Root-derived C did not mirror plot ANPP estimates, as indicated by variable root-derived C: ANPP ratios across plots (SI Fig. 1). The amount of root-derived C recovered in MAOM pools was greater in AM compared to ECM plots (P = 0.041; Fig. 4). There was no site or site × plot type effect on root-derived MAOM-C.