Summer droughts projected to increase in central Europe due to climate changes strongly influence the carbon cycle of ecosystems. Persistent respiration activities during drought periods are responsible for a significant carbon loss, which may turn the ecosystem from a sink into a source of carbon. There are still gaps in our knowledge regarding the characteristic changes taking place in the respiration of the different components of the ecosystem in response to drought events. In the present study, we combined a physical separation of soil respiration components with continuous measurements of soil CO₂ efflux and its isotopic (¹³C) signals at a dry grassland site in Hungary. The physical separation of soil respiration components was performed by means of inox meshes and tubes inserted into the soil. The root-excluded and root- and mycorrhiza-excluded treatments served to measure the isotopic signals of the rhizospheric, mycorrhizal fungi and heterotrophic components, respectively. In the dry grassland investigated in the study the three components of the soil CO₂ efflux decreased at different rates under drought conditions. During drought the contribution made by the heterotrophic components was the highest (54 ± 8 %; mean ±SE). Rhizospheric component was the most sensitive to soil drying with its relative contribution to the total soil respiration dropping from 66 ± 7 (non-stressed) to 35 ± 17 % (mean ±SE) under drought conditions. According to our results the heterotrophic component of soil respiration is the major contributor to the respiration activities during drought events in the dry grassland ecosystem studied.

Interannual variation of carbon fluxes of grasslands on sandy (5 years data) and heavy clay soils (4 years data) have been analysed. The sandy grassland was carbon sink in 3 (2004, 2005, 2006) out of the investigated 5 years. Its annual C-balance is precipitation limited, the relation seems strongly conservative, with r 2 of 0.83. More than half of the net source activity fell to the summer droughts. The heavy clay grassland was net source of carbon in one year (2007) only with no whole year record from 2003, a drought and heat wave year. Dependence of the C-balance on precipitation was somewhat weaker (r 2 =0.57) than in the sandy grassland. Length of growing period showed less variation here compared to the sandy grassland. Recovery of sink activity after rains was much slower for the heavy clay grassland than for the sandy grassland. The reason behind is that the amount of water required to reach optimal soil water content for plant functioning is several times larger for the mountain grassland. This f...