Plant nutrient responses to 4 years of CO 2 enrichment were investigated in situ in calcareous grassland. Beginning in year 2, plant aboveground C:N ratios were increased by 9% to 22% at elevated CO 2 (P < 0.01), depending on year. Total amounts of N removed in biomass harvests during the ®rst 4 years were not aected by elevated CO 2 (19.9 1.3 and 21.1 1.3 g N m A2 at ambient and elevated CO 2 ), indicating that the observed plant biomass increases were solely attained by dilution of nutrients. Total aboveground P and tissue N:P ratios also were not altered by CO 2 enrichment (12.5 2 g N g A1 P in both treatments). In contrast to non-legumes (>98% of community aboveground biomass), legume C/N was not reduced at elevated CO 2 and legume N:P was slightly increased. We attribute the less reduced N concentration in legumes at elevated CO 2 to the fact that virtually all legume N originated from symbiotic N 2 ®xation (%N dfa % 90%), and thus legume growth was not limited by soil N. While total plant N was not aected by elevated CO 2 , microbial N pools increased by +18% under CO 2 enrichment (P 0.04) and plant available soil N decreased. Hence, there was a net increase in the overall biotic N pool, largely due increases in the microbial N pool. In order to assess the eects of legumes for ecosystem CO 2 responses and to estimate the degree to which plant growth was P-limited, two greenhouse experiments were conducted, using ®rstly undisturbed grassland monoliths from the ®eld site, and secondly designed`microcosm' communities on natural soil. Half the microcosms were planted with legumes and half were planted without. Both monoliths and microcosms were exposed to elevated CO 2 and P fertilization in a factored design. After two seasons, plant N pools in both unfertilized monoliths and microcosm communities were unaected by CO 2 enrichment, similar to what was found in the ®eld. However, when P was added total plant N pools increased at elevated CO 2 . This community-level eect originated almost solely from legume stimulation. The results suggest a complex interaction between atmospheric CO 2 concentrations, N and P supply. Overall ecosystem productivity is N-limited, whereas CO 2 eects on legume growth and their N 2 ®xation are limited by P.
Key words Dinitrogen ®xation á Plant functional types á legumes á Nutrient limitation á Phosphorus
IntroductionMost studies of vascular plant responses to atmospheric CO 2 enrichment show an enhancement of growth when carried out under sucient nutrient supply. In natural environments, plant productivity is often limited by the availability of mineral nutrients such as nitrogen and phosphorus (Vitousek and Howarth 1991) which in turn may limit the stimulation of plant growth by CO 2 enrichment. This underlines the importance of considering plant nutrient status in understanding their growth responses to increased CO 2 concentrations (Woodward et al. 1991;Overdieck 1993;KoÈ rner 1995b