Quantifying belowground C inputs
We quantified root production as the total root mass recovered within a given core after two years. Fine roots of temperate trees typically turn-over in one year (McCormack et al. 2015), and thus roots recovered in the ingrowth cores likely resulted from one and not two years of production. Consequently, we did not divide our estimates of fine root production by two. However, we acknowledge our root production values reflect the balance between root production and root turnover over a two-year period. We quantified root-derived C accumulation into each core using a two-pool mixing model following Panzacchi et al. (2016). Broadly, as C3 rhizodeposits are incorporated into the C4 soil core, the δ13C signature of the C4 soil becomes more deplete over time, i.e. the δ13C signature becomes more similar to that of the C3 soil. This change in δ13C of the ingrowth core soil can be used to calculate total root-derived C inputs into the core.
First, the fraction of soil C derived from root inputs (Frd; unitless) was calculated using a two-end member mixing model:
Frd = (δ13Cingrowth- δ13Ccontrol) / (δ13Croot – δ13Ccontrol)
where δ13Cingrowth is the δ13C of C4 soil collected from an individual ingrowth cores after two years in the field and δ13Ccontrol is the average δ13C of C4 soil collected from two PVC control cores from the same plot as the ingrowth core after two years in the field. We estimated root δ13C for AM/ECM ‘mixed’ plots as the mean of site-specific AM and ECM δ13C values. For any given plot, δ13Crootwas estimated as the mean δ13C for a given mycorrhizal plot-type at a given site. The net root-derived C inputs (Crd-net; g C m-2) into surface soils (0-15 cm) was calculated as:
Crd-net = ⍴ * [C] * Frd * 75000
where ⍴ is the initial C4 soil bulk density (g cm-3), [C] is the C content (%) of the core after two years in the field, and 75000 is the conversion factor to transform % C to g C m-2 to a depth of 15 cm. Bulk density (1.21 g mL-1) was measured on the initial C4 soil: sand mixture and is thus constant across all plots. To estimate an annual net flux (for comparison to annual aboveground net primary production), we divided root-derived C by the number of years cores were in the field (i.e. two years at all sites).
Several assumptions are worth noting with this approach. First, the δ13C signature is assumed to be uniform throughout the soil core, such that various C fractions (e.g. mineral-associated vs. particulate C) of the C4 soil which may turnover at different rates are assumed to have identical δ13C values. While recent crop rotations between C4 and C3 plants where the ingrowth C4 soil was harvested could theoretically create heterogeneity in soil δ13C, any such differences are likely small given the long-term dominance of C4 corn production at the site. Second, diffusion of dissolved organic C either laterally or vertically through the ingrowth cores could also deplete δ13C signatures, resulting in artificially high estimates of root-derived soil C. However, vertical diffusion should be similar between ingrowth and control cores, and previous estimates of δ13C depletion within ingrowth cores due to lateral diffusion suggest the effect is negligible (Phillips, R.P. unpublished data ).