Despite the importance of forests in the global carbon (C) cycle, estimates of coarse root mass in forest ecosystems remain uncertain because of tremendous variability in reported root: shoot (R:S) ratios. Recent meta-analyses find negligible ecological influences on variation in forest R:S ratios, even though experiments and ecological theory suggest otherwise. We argue that a major source of variation in R:S ratios is methodology, which obscures detection of ecological drivers of R:S ratios and leads to underestimates of C storage and sequestration. We compiled 174 R:S ratio samples from 58 published studies on forests and evaluated each study with respect to four methodological criteria: 1) recovery of roots from soil; 2) volumetric extent of excavation; 3) accounting for spatial structure in root biomass; and 4) inclusion of the root crown and tap root.
Satisfaction of each criterion increased geometric mean R:S ratio and studies that met all four criteria had a geometric mean R:S ratio of 0.36 versus 0.24 for studies that failed >1 criteria. Based on studies that met criteria, mean R:S ratio was higher under dry versus mesic soil conditions, marginally higher in natural forests than plantations, and did not differ among boreal, temperate, and tropical forests. Using this refined mean R:S ratio of 0.36 to estimate the C inventory of US forests resulted in increases of 81% in forest root biomass (~9000 Tg CO2 equivalents), 13% in total forest biomass, and 12% in annual C sequestration compared to current estimates that assume a R:S ratio of 0.20. Our refined estimates of R:S ratios should aid in developing a more accurate national carbon inventory.