The root system of permanent grasslands is of outstanding importance for resource acquisition. Particularly under semi-arid conditions, the acquisition of water and nutrients is highly variable during the vegetation growth period and between years. Additionally, grazing is repeatedly disturbing the functional equilibrium between the root system and the transpiring leaf canopy. However, very few data is available considering grazing effects on belowground net primary productivity (BNPP) and root-shoot dry mass allocation in natural grassland systems. We hypothesise that grazing significantly reduces BNPP due to carbon reallocation to shoot growth. Root biomass and BNPP were estimated by soil coring in 2004, 2005 and 2006 and from ingrowth cores in 2005 and 2006 at one site which has been protected from grazing since 1979 (UG79), at one winter grazing (WG), and one heavily grazed (HG) site. BNPP was estimated from the summation of significant increments of total and live root biomass and from accumulated root biomass of ingrowth cores. Belowground biomass varied from 1,490-2,670 g m −2 and was significantly lower under heavy grazing than at site UG79. Root turnover varied from 0.23 to 0.33 year −1 and was not significantly different between sites. Heavy grazing significantly decreased live root biomass and BNPP compared to site UG79. Taking BNPP estimates from live root biomass dynamics and ingrowth cores as the most Plant Soil (2008) 307:41-50 reliable values, the portion of dry mass allocated belowground relative to total net primary productivity (BNPP/NPP) varied between 0.50-0.66 and was reduced under heavy grazing in 2005, but not in 2006. The positive correlation between cumulative root length density of ingrowth cores and leaf dry matter suggests that the ingrowth core method is suitable for studying BNPP in this semi-arid steppe system. Grazing effects on BNPP and BNPP/NPP should be considered in regional carbon models and estimates of belowground nutrient cycling.

Fluxes of nitrous oxide (N 2 O), methane (CH 4 ) and carbon dioxide (CO 2 ) were followed at winter-grazed (WG) and ungrazed steppe (UG99) in Inner Mongolia during the winter-spring transition of 2006. Mean fluxes during the period March 12-May 11 were 8.2±0.5 (UG99) and 1.5±0.2 μg N 2 O-N m −2 h −1 (WG) for N 2 O, 7.2±0.2 (UG99) and 3.0± 0.1 mg CO 2 -C m −2 h −1 (WG) for CO 2 and −42.5±0.9 (UG99) and −14.1±0.3 μg CH 4 -C m −2 h −1 (WG) for CH 4 . Our data show that N 2 O emissions from semiarid steppe are strongly affected by freeze-thawing. N 2 O emissions reached values of up to 75 μg N 2 O-N m −2 h −1 at the UG99 site, but were considerably lower at the WG site. The observed differences in N 2 O, CH 4 and CO 2 fluxes between the ungrazed and grazed sites were ascribed to the reduced plant biomass at the grazed site, and-most important-to a reduction in soil moisture, due to reduced snow capturing during winter. Thus, winter-grazing significantly reduced N 2 O emission but on the other hand also reduced the uptake of atmospheric CH 4 . To finally evaluate which of the both effects is most important for the non-CO 2 greenhouse gas balance measurements covering an entire year are needed.

Productivity of semiarid grasslands is affected by soil water and nutrient availability, with water controlling net primary production under dry conditions and soil nutrients constraining biomass production under wet conditions. In order to investigate limitations on plants by the response of root-shoot biomass allocation to water and nitrogen (N) availability, a field experiment, on restoration plots with rainfed, unfertilized control plots, fertilized plots receiving N (25 kg urea-N ha(-1)) and water (irrigation simulating a wet season), was conducted at two sites with different grazing histories: moderate (MG) and heavy (HG) grazing. Irrigation and N addition had no effect on belowground biomass. Irrigation increased aboveground (ANPP) and belowground net primary production (BNPP) and rain-use efficiency based on ANPP (RUE(ANPP)), whereas N addition on rainfed plots had no effect on any of the measured parameters. N fertilizer application on irrigated plots increased ANPP and RUE(ANPP) and reduced the root fraction (RF: root dry matter/total dry matter), resulting in smaller N effects on total net primary production (NPP) and rain-use efficiency based on NPP. This suggests that BNPP should be included in evaluating ecosystem responses to resource availability from the whole-plant perspective. N effects on all measured parameters were similar on both sites. However, site HG responded to irrigation with higher ANPP and a lower RF when compared to site MG, indicating that species composition had a pronounced effect on carbon allocation pattern due to below- and aboveground niche complementarity.