Terrestrial ecosystems remove about 30% of the CO 2 emitted by human activities each year 1 , yet the persistence of this carbon sink partly depends on how plant biomass and soil carbon stocks respond to future increases in atmospheric CO 2 2,3 . While plant biomass often increases in elevated CO 2 (eCO 2 ) experiments 4-6 , soil carbon has been observed to increase, remain unchanged, or even decline 7 . The mechanisms driving this variation across experiments remain poorly understood, creating uncertainty in climate projections 8,9 . Here, we synthesized data from 108 eCO 2 experiments and found that the effect of eCO 2 on soil carbon stocks is best explained by a negative relationship with plant biomass: when plant biomass is strongly stimulated by eCO 2 , soil carbon accrual declines; conversely, when biomass is weakly stimulated, soil carbon accumulates. This trade-off appears related to plant nutrient acquisition, whereby enhanced biomass requires mining the soil for nutrients, which decreases soil carbon accrual. We found an increase in soil carbon stocks with eCO 2 in grasslands (8±2%) and no increase in forests (0±2%), even though plant biomass in grassland responded less strongly (9±3%) than in forest (23±2%). Ecosystem models do not reproduce this trade-off, which implies that projections of soil carbon may need to be revised.

No abstract

The introduction of nonnative plant species may decrease ecosystem stability by altering the availability of nitrogen (N) for plant growth. Invasive species can impact N availability by changing litter quantity and quality, rates of N 2 -fixation, or rates of N loss. We quantified the effects of invasion by the annual grass Bromus tectorum on N cycling in an arid grassland on the Colorado Plateau (USA). The invasion occurred in 1994 in two community types in an undisturbed grassland. This natural experiment allowed us to measure the immediate responses following invasion without the confounding effects of previous disturbance. Litter biomass and the C:N and lignin:N ratios were measured to determine the effects on litter dynamics. Long-term soil incubations (415 d) were used to measure potential microbial respiration and net N mineralization. Plant-available N was quantified for two years in situ with ion-exchange resin bags, and potential changes in rates of gaseous N loss were estimated by measuring denitrification enzyme activity. Bromus invasion significantly increased litter biomass, and Bromus litter had significantly greater C:N and lignin:N ratios than did native species. The change in litter quantity and chemistry decreased potential rates of net N mineralization in sites with Bromus by decreasing nitrogen available for microbial activity. Inorganic N was 50% lower on Hilaria sites with Bromus during the spring of 1997, but no differences were observed during 1998. The contrasting differences between years are likely due to moisture availability; spring precipitation was 15% greater than average during 1997, but 52% below average during spring of 1998. Bromus may cause a short-term decrease in N loss by decreasing substrate availability and denitrification enzyme activity, but N loss is likely to be greater in invaded sites in the long term because of increased fire frequency and greater N volatilization during fire. We hypothesize that the introduction of Bromus in conjunction with land-use change has established a series of positive feedbacks that will decrease N availability and alter species composition.

The introduction of nonnative plant species may decrease ecosystem stability by altering the availability of nitrogen (N) for plant growth. Invasive species can impact N availability by changing litter quantity and quality, rates of N2‐fixation, or rates of N loss. We quantified the effects of invasion by the annual grass Bromus tectorum on N cycling in an arid grassland on the Colorado Plateau (USA). The invasion occurred in 1994 in two community types in an undisturbed grassland. This natural experiment allowed us to measure the immediate responses following invasion without the confounding effects of previous disturbance. Litter biomass and the C:N and lignin:N ratios were measured to determine the effects on litter dynamics. Long‐term soil incubations (415 d) were used to measure potential microbial respiration and net N mineralization. Plant‐available N was quantified for two years in situ with ion‐exchange resin bags, and potential changes in rates of gaseous N loss were estimated by measuring denitrification enzyme activity. Bromus invasion significantly increased litter biomass, and Bromus litter had significantly greater C:N and lignin:N ratios than did native species. The change in litter quantity and chemistry decreased potential rates of net N mineralization in sites with Bromus by decreasing nitrogen available for microbial activity. Inorganic N was 50% lower on Hilaria sites with Bromus during the spring of 1997, but no differences were observed during 1998. The contrasting differences between years are likely due to moisture availability; spring precipitation was 15% greater than average during 1997, but 52% below average during spring of 1998. Bromus may cause a short‐term decrease in N loss by decreasing substrate availability and denitrification enzyme activity, but N loss is likely to be greater in invaded sites in the long term because of increased fire frequency and greater N volatilization during fire. We hypothesize that the introduction of Bromus in conjunction with land‐use change has established a series of positive feedbacks that will decrease N availability and alter species composition. For reprints of this Invited Feature, see footnote 1, p. 1259.