1. Instantaneous leaf photosynthesis and land area-based net ecosystem CO2 exchange (NEC) are almost universally increased at elevated CO2 concentrations, at least in the short term and under high light conditions. This raises the possibility that terrestrial ecosystems sequester extra C in response to elevated CO2, and it has been hypothesized that part of this extra C is stored in soils.2. Attempts to quantify ecosystem C sequestration experimentally are based on (i) ecosystem CO2 exchange measurements; (ii) C isotope tracking; and (iii) direct C stock measurements.3. Because direct C stock measurements are insensitive to increases in ecosystem C storage in the range expected, and C isotope tracking is methodologically difficult because of the need to account for new and old soil C pools, NEC measurements were considered to be a more direct and unbiased method to estimate net ecosystem production and C sequestration at elevated CO2.4. Here we present a case study in calcareous grassland under long-term CO2 enrichment in which we demonstrate that calculated C balances are extremely sensitive to systematic experimental biases inherent in any CO2 flux-based study at elevated CO2. A sensitivity analysis demonstrates that these systematic errors tend to result in severe overestimation of ecosystem C accretion under elevated CO2.5. Carbon isotope data and soil pool C and N measurements from the same study add to the evidence that the C balance derived from CO2 flux measurements at elevated CO2 is overestimated.6. Based on this evidence, and the wide attention given to C sequestration at elevated CO2, we suggest a critical reconsideration of the appropriateness of NEC measurement in CO2-enriched ecosystems as a basis for calculating ecosystem C balances at elevated CO2.

The effects of elevated CO 2 on plant biomass and community structure have been studied for four seasons in a calcareous grassland in northwest Switzerland. This highly diverse, semi-natural plant community is dominated by the perennial grass Bromus erectus and is mown twice a year to maintain species composition. Plots of 1.3 m 2 were exposed to ambient or elevated CO 2 concentrations ( n = 8) using a novel CO 2 exposure technique, screen-aided CO 2 control (SACC) starting in March 1994. In the 1st year of treatment, the annual harvested biomass (sum of aboveground biomass from mowings in June and October) was not significantly affected by elevated CO 2 . However, biomass increased significantly at elevated CO 2 in the 2nd (+20%, P = 0.05), 3rd (+21%, P = 0.02) and 4th years (+29%, P = 0.02). There were no detectable differences in root biomass in the top 8 cm of soil between CO 2 treatments on eight out of nine sampling dates. There were significant differences in CO 2 responsiveness between functional groups (legumes, non-leguminous forbs, graminoids) in the 2nd ( P = 0.07) and 3rd ( P < 0.001) years of the study. The order of CO 2 responsiveness among functional groups changed substantially from the 2nd to the 3rd year; for example, non-leguminous forbs had the smallest relative response in the 2nd year and the largest in the 3rd year. By the 3rd year of CO 2 exposure, large species-specific differences in CO 2 response had developed. For five important species or genera the order of responsiveness was Lotus corniculatus (+271%), Carex flacca (+249%), Bromus erectus (+33%), Sanguisorba minor (no significant CO 2 effect), and six Trifolium species (a negative response that was not significant). The positive CO 2 responses in Bromus and Carex were most closely related to increases in tiller number. Species richness was not affected by CO 2 treatment, but species evenness increased under elevated CO 2 (modified Hill ratio; P = 0.03) in June of the 3rd year, resulting in a marginally significant increase in species diversity (Simpson's index; P = 0.09). This and other experiments with calcareous grassland plants show that elevated atmospheric CO 2 concentrations can substantially alter the structure of calcareous grassland communities and may increase plant community biomass.

The relationship between plant species diversity and ecosystem CO2 and water vapour fluxes was investigated for planted calcareous grassland communities composed of 5, 12, or 32 species assembled from the native plant species pool. These diversity manipulations were done in factorial combination with a CO2 enrichment experiment in order to investigate the degree to which ecosystem responses to elevated CO2 are altered by a loss of plant diversity. Ecosystem CO2 and H2O fluxes were measured over several 24‐h periods during the 1994 and 1995 growing seasons. Ecosystem CO2 assimilation on a ground area basis decreased with decreasing plant diversity in the first year and this was related to a decline in above‐ground plant biomass. In the second year, however, CO2 assimilation was not affected by diversity, and this corresponded to the disappearance of a diversity effect on above‐ground biomass. Irrespective of diversity treatment, CO2 assimilation on a ground area basis was linearly related to peak above‐ground biomass in both years. Elevated CO2 significantly increased ecosystem CO2 assimilation in both years with no interaction between diversity and CO2 treatment, and no corresponding increase in above‐ground biomass. There were no significant effects of diversity on water vapour flux, which was measured only in the second year. There were indications of a small CO2 effect on water vapour flux (3–9% lower at elevated CO2 depending on the light level). Our findings suggest that decreasing plant species diversity may substantially decrease ecosystem CO2 assimilation during the establishment of such planted calcareous grassland communities, but also suggest that this effect may not persist. In addition, we find no evidence that plant species diversity alters the response of ecosystem CO2 assimilation to elevated CO2.

1. As part of a long‐term study of the effects of elevated CO2 on biodiversity and ecosystem function in a calcareous grassland, we measured ecosystem carbon dioxide and water‐vapour fluxes over 24‐h periods during the 1994 and 1995 growing seasons. Data were used to derive CO2 and H2O gas‐exchange response functions to quantum flux density (QFD). 2. The relative increase in net ecosystem CO2 flux (NEC) owing to CO2 enrichment increased as QFD rose. Daytime NEC at high QFD under elevated CO2 increased by 25% to 60%, with the greatest increases in the spring and after mowing in June when above‐ground biomass was lowest. There was much less stimulation of NEC in early June and again in October when the canopy was fully developed. Night‐time NEC was not significantly altered under elevated CO2. 3. Short‐term reversal of CO2 concentrations between treatments after two seasons of CO2 exposure provided evidence for a 50% downward adjustment of NEC expressed per unit above‐ground plant dry weight. However, when expressed on a land area basis, this difference disappeared because of a c. 20% increase in above‐ground biomass under elevated CO2. 4. Ecosystem evapotranspiration (ET) was not significantly altered by elevated CO2 when averaged over all measurement dates and positions. However, ET was reduced 3–18% at high QFD in plots at the top of the slope at our study site. In summary, CO2 enrichment resulted in a large stimulation of ecosystem CO2 capture, especially during periods of a large demand of carbon in relationship to its supply, and resulted in a relatively small and variable effect on ecosystem water consumption.

We have developed a novel CO2 exposure system for natural vegetation that is a middle ground between Free Air CO2 Enrichment (FACE) and traditional open-top chambers (OTC). Screen-Aided CO2 Control (SACC) technology uses much less CO2 per experiment and per replicate than FACE and is superior to OTCs in terms of its effects on microclimate. A SACC unit consists of a thin metal frame, a clear plastic ''screen'', and a pipe at the base of the screen through which CO2 enriched jets of air are directed into the unit. There is a gap between the ground and the bottom of the pipe and the screen is relatively short in comparison to the maximum height of the vegetation. Our SACC units are hexagonal and enclose a ground area of 1.27 m(2). SACC works in the following way: 1) the screen breaks the wind and creates turbulent mixing within the unit, 2) the mixing of the outside air with the CO2 enriched jets of air,generates relatively uniform CO2 concentrations within the screened-in vegetation, and 3) a fully automated system monitors CO2 concentrations and adjusts CO2 injection rates for each unit every ca. IO minutes to maintain preset CO2 concentrations.Twenty-four hour means of CO2 concentrations in the middle of a unit are typically maintained within 1 mu l l(-1) of their set points. Spatial variation and short-term fluctuations in CO2 concentration are similar to those in OTCs and FACE. CO2 consumption at our site is 5 kg CO2 day(-1) replicate(-1) for a total of ca. 30 tons per year for 20 elevated CO2 SACC units. Compared to OTCs, SACC units have reduced temperature peaks at full sunlight, minimal effects on solar radiation, reduced rainfall interception by chamber walls, and freer access of small animals to experimental plots. We believe that SACC is the best method for exposing short stature vegetation to elevated CO2 when financial constraints do not allow for a properly replicated FACE experiment.

In 1994 we initiated a long term study on the effects of elevated CO{sub 2} on ecosystem function and biodiversity in a species rich grassland near Basel, Switzerland. Natural vegetation was exposed to ambient and elevated (600{mu}l 1{sup -1}) CO{sub 2} using open-top chambers. We measured ecosystem carbon and water fluxes for several 24hr. periods during the growing season by partially closing the top of the chambers. During the first 3 months of CO{sub 2} enrichment, mean net ecosystem carbon uptake at high photon flux density increased by 37%. The relative photosynthetic gain was diminished as photon flux density decreased. Dark respiration, and midday soil CO{sub 2}-evolution did not differ between treatments. Midday evapotranspiration decreased by 17%, probably due to stomatal responses, since LAI did not change. Since the extra carbon has not yet appeared in either aboveground, belowground, or microbial biomass, we hypothesize that a major part of the extra carbon was sequestered into the soil compartment.