• buffalo grass;
  • clover;
  • elevated CO2;
  • LAI;
  • legumes;
  • nitrogen;
  • pasture;
  • phosphorus;
  • production;
  • Stenotaphrum secundatum;
  • Trifolium repens


Thirty-six mesocosms, each containing a two-species community of Trifolium repens (C3 legume) and Stenotaphrum secundatum (C4 grass), were grown in sand with three nutrient regimes, zero N low P, zero N high P and supplied N high P, under ambient (aCO2) and twice ambient CO2 (eCO2) for 15 months in two greenhouses. Aboveground annual production in the P limited mesocosms did not respond to eCO2 and was reduced by 50% relative to mesocosms with an adequate P supply, where dry-matter production was increased by 12–24% under eCO2. The stimulation of production by eCO2 occurred throughout the year despite a clear seasonality in growth. There was no effect of eCO2 on leaf area index (LAI), which was larger under high P than low P. Live root mass at the end of the experiment was higher under eCO2 in all nutrient treatments, but the response of total belowground C (root+soil) to eCO2 depended on P treatment. Under limiting P, belowground C was not significantly changed by eCO2 (2–2.3 t belowground C ha−1). Under high P supply, both root and soil C pools increased under eCO2. Under aCO2, low P supply increased belowground C by 0.7–1 t C ha−1 above that added by the high P treatment. P is commonly limiting in Australian ecosystems and the majority of ecosystem N input is provided by biological N fixation. Consequently, the response of legumes to eCO2 is of particular importance. These results demonstrate that at low P availability, there is likely to be only a limited response of biomass production by T. repens to eCO2, which in turn may constrain any ecosystem response.