Introduction
Plants allocate a substantial amount of carbon (C) belowground (Gill & Finzi 2016), with important consequences for soil C storage. Root-derived inputs have been hypothesized to control soil organic matter dynamics by promoting soil C formation (Rasse et al. 2005; Clemmensen et al. 2013), stabilization (Jackson et al.2017), and turnover (Cheng & Kuzyakov 2005). Thus, roots have the potential to both increase and decrease soil C stocks. Despite this, we know remarkably little about the magnitude of these fluxes, their controls, and the consequences of belowground inputs for soil C stabilization owing to difficulties of tracking belowground inputs. Given the importance of soil C storage in regulating global C cycling and mitigating the effects of rising atmospheric CO2, it is critical to constrain estimates of belowground C supply and understand the fate of root-derived C fluxes in heterogeneous forests (Schmidt et al. 2011; Iversen et al. 2017).
While estimates of plant-derived C inputs to soil remain sparse, the total amount of C allocated belowground by plants is more commonly studied and can vary across climates, edaphic conditions and species. Previous work has revealed broad latitudinal patterns in total belowground C allocation, with 65% of GPP allocated belowground in boreal forests compared to only 30% in tropical forests (Gill & Finzi 2016). These patterns mirrored soil fertility gradients, such that partitioning of C belowground was inversely related to soil N:P. In contrast, Vicca et al. (2012) found that partitioning to belowground C was greater in nutrient-rich forests, likely due to underestimation of belowground C fluxes in nutrient-poor forests (e.g. allocation to mycorrhizal partners and root exudates which can be extremely difficult to quantify). Within a single site, there can be considerable interspecific variation in root traits (Valverde-Barrantes et al.2013), which may drive species-specific variation in belowground C allocation and nutrient uptake (Keller & Phillips 2019b). Moreover, there is little evidence of a direct, linear relationship between plant productivity and soil C accumulation, reflecting our poor understanding of the fate of belowground C allocation (Jackson et al. 2017). The degree to which belowground C allocation predicts C accumulation in soil has not been tested but is critical to understanding plant community effects on ecosystem C cycling and feedbacks to climate.
Belowground C fluxes from the plant to the soil are comprised of root and root-associated fungal turnover as well as rhizodeposition (e.g. sloughed root cells and passive and active exudation), making accurate quantitative estimates of this flux challenging (Pausch et al.2013). Traditionally, root turnover has been estimated either using sequential coring or minirhizotron approaches, with the remaining rhizosphere C flux approximated using a mass balance approach (Faheyet al. 1999, 2005; Hendricks et al. 2006). Root exudation can be directly measured from roots in the field using the cuvette method, whereby living roots are excavated from the soil and exudates are captured in situ (Phillips et al. 2008). However, these methods are both time and resource intensive, necessitating the need for simpler, time-integrated methods for estimating root-derived soil C inputs. The isotopic ingrowth core method takes advantage of the difference in δ13C signatures between C3 and C4 plants (and consequently, the soils they are growing in), providing a quantitative estimate of root-derived soil C accumulation over the course of the study methods (Hoosbeek et al. 2004; Cotrufo et al. 2011; Martinezet al. 2016). As such, the isotopic ingrowth core method may provide a better estimate of root-derived C inputs compared to traditional methods.
Tree mycorrhizal association has been shown to be an integrative plant functional trait that links plant and soil properties (Phillips et al. 2013). In temperate forests, trees associate almost exclusively with one of two groups of mycorrhizal fungi: arbuscular mycorrhizal (AM) and ectomycorrhizal (ECM) fungi. Theory predicts that divergent plant and fungal traits between AM and ECM species should lead to distinct patterns of belowground C inputs between mycorrhizal types. Previous work in temperate forests shows AM trees tend to preferentially allocate C to roots whereas ECM trees rely more heavily on hyphal proliferation to acquire scarce nutrients (Chen et al. 2016; Cheng et al. 2016). Such mycorrhizal type differences in the ratio of root production to total belowground C allocation may have consequences for soil C storage. For example, AM but not ECM plants promoted soil C destabilization in a pot experiment, with larger soil C losses when both roots and hyphae (rather than hyphae alone) were present (Wurzburger and Brookshire 2017). However, field and modeling studies suggest greater rhizodeposition in ECM compared to AM stands, which may explain greater soil C losses in many ECM-dominated forests (Yin et al. 2014, Brzostek et al. 2015, Sulman et al. 2017, but see Averill and Hawkes 2016). Thus, it remains unresolved how plant mycorrhizal type may influence patterns of root-derived soil C as well as the extent to which total belowground soil C inputs mirror root production patterns.
There is increasing evidence that plant mycorrhizal type is also predictive of soil organic matter formation patterns. Labile tissues with fast decay promote microbial activity, including microbial production and turnover. In turn, these products generated during microbial decay are thought to contribute disproportionately to slow-cycling soil organic C pools by forming associations with reactive silts and clays (Grandy & Neff 2008; Schmidt et al. 2011; Kallenbach et al. 2016). Accordingly, greater mineral-associated organic matter (MAOM) pools have been measured in AM systems (characterized by fast decay) compared to ECM systems (Craig et al. 2018; Cotrufo et al. 2019). This pattern suggests plant mycorrhizal type may be critical in determining stabilization of soil C. However, most of the empirical and theoretical work documenting these patterns (Sulman et al. 2017; Zhu et al. 2018; Jo et al. 2019) are premised on the idea of leaf litter quality differences between AM and ECM trees (Keller & Phillips 2019a). This is in contrast to studies showing roots may be a primary source of slow-cycling soil C (Sokol & Bradford 2019). Thus, there is a critical need to investigate whether tree mycorrhizal dominance affects belowground C inputs, and whether such differences contribute to MAOM formation patterns.
To this end, we measured total root-derived soil C accumulation and root production across nine-plot gradients of increasing ECM tree dominance within six temperate forests. We asked 1) what is the magnitude of root-derived soil C accumulation? 2) how does root-derived soil C accumulation vary across gradients of ECM tree dominance and 3) to what extent do belowground C inputs influence the formation of mineral-associated soil C? We hypothesized that root-derived inputs from ECM-dominated plots would be greater than those from AM-dominated plots, and that soils receiving the greatest inputs would promote greater C accumulation in MAOM pools. Using an isotopic ingrowth core technique, we found annual root-derived C accumulation was comparable in magnitude to annual aboveground net primary productivity, with greater root-derived C accumulation in AM-dominated compared to ECM-dominated plots. Root-derived C accumulation in MAOM-C pools was also greater in AM compared to ECM plots, providing evidence that mycorrhizal type differences in belowground C inputs can affect soil C stabilization patterns.