Plant and mycorrhizal trait effects on soil carbon dynamics across six temperate forests

Understanding how species shifts affect biogeochemical cycles and ecosystem functioning is a fundamental goal of ecosystem ecology. Tree species effects on carbon (C) cycling are largely driven by the dominant species' functional traits, reflecting both the plant's evolutionary history and its plastic responses to the environment. Yet, most research to date has focused on aboveground plant traits and far less is known about the importance of root interspecific variation in driving soil C cycling. We used root ingrowth cores to quantify fine root production (FRP) and root-derived carbon (RDC) inputs to soil across 54 forest plots varying in their abundances of ectomycorrhizal-(ECM) vs. arbsucular mycorrhizal-(AM) tree species, nested within 6 sites. At each site, we quantified both FRP and RDC using cores filled with C₄corn-field soil carrying an enriched δ ¹³C signature compared to C₃plant-derived material.

We predicted greater FRP and RDC with increasing ECM-dominance within a site, both due to plant traits and soil characteristics associated with these distinct mycorrhizal groups. We found strong plant trait by site interactions in belowground C dynamics, with FRP increasing across the ECM gradient in only half of our sites. Surprisingly, the plots with the greatest FRP were not those with the highest RDC, which tended to decrease with increasing ECM-dominance. Belowground C inputs to soil were positively correlated with mineral-associated forms of soil C in ECM plots, while the opposite was observed in AM-dominated plots. Our results suggest that variation in aboveground and belowground traits likely contribute to variation in soil C stocks in temperate forests. Additionally, tree species appear to both respond to and reflect their soil environment. As global change induces broad-scale species shifts, incorporating such plant trait variation in ecosystem models will be important to accurately predict terrestrial carbon-climate feedbacks.