Electronic Scientific Paper Archive

Soil CO2 Flux in Hövsgöl National Park, Northern Mongolia

Avirmed Otgonsuren1*, Clyde. E. Goulden2, Ingrid C. Burke3 and Baatar Bulgan1
1Geoecology Institute, Mongolian Academy of Sciences, Baruun Selbe-13, Ulaanbaatar 211238, Mongolia,
2Institute for Mongolian Biodiversity and Ecological Studies, Academy of Natural Sciences, Philadelphia, PA 19103, USA
3Department of Forest, Rangeland, and Watershed Stewardship, Colorado State University, Fort Collins, Colorado 80523, USA


We investigated soil CO2 fl ux and bare soil respiration in grasslands that are located at the southern edge of the Siberian boreal forest in Northern Mongolia. The study area has warmed by almost 1.8oC over the last 40 years, and the soil and vegetation covers have been changed due to intense nomadic grazing pressure. Bare soil respiration is decreased with increasing grazing pressure, but there was no consistent pattern of total soil CO2 fl ux under three distinct grazing levels. Bare soil respiration and soil CO2 fl ux were higher on north-facing slopes than on south-facing slopes, due to high organic matter accumulation and the presence of permafrost. Both bare soil respiration and soil CO2 fl ux were signifi cantly higher in riparian areas compared with the lower and upper portions of the south-facing slope. Topography has a stronger effect on variability of soil CO2 fl ux and bare soil respiration than variability induced by grazing. Inter-annual variability in soil CO2 fl ux and bare soil respiration was very high, because of high variability in climate conditions.


Models of global climate predict that the greatest temperature increases due to greenhouse gas concentrations will be between latitudes 45°N and 65°N (Tans et al., 1990). The northern boreal forest ecosystem (taiga) lies along these latitudes, and the southern boundary of the Siberian taiga lies adjacent to a temperate grassland ecosystem that is of great interest with respect to carbon, since there are about 181.1 Mt x 109 C in these grassland soils to a depth of 1 m (Schlesinger, 1977). Around 70-80% of Mongolia is grassland and shrubland, with slightly more grassland than shrubland (Tsogtbaatar, 2000). Grasslands have been overgrazed in many parts of Mongolia. Soil respiration is one of the primary pathways by which terrestrial C is returned to the atmosphere. The effl ux of CO2 from soils is the product of both respiratory activities of roots and soil heterotrophs (Raich & Schlesinger, 1992). Thus, one may expect higher soil CO2 fl ux in places with high soil organic C or high root biomass. Environmental variables that infl uence soil respiration may be altered with both grazing pressure and topography. Soil temperature is a major factor controlling soil respiration rates in plot scale measurements in other ecosystem types (Kang et al., 2003). In addition, soil moisture, soil substrate quality, and vegetation cover infl uence soil respiration (Kang et al., 2003). In many semiarid grassland or shrubland landscapes, toeslopes are characterized by having higher vegetation cover, fi ner textured soils, and greater soil organic matter (Burke et al., 1999). Since soil respiration is a major component of ecosystem C fl ux, and the microclimates of Mongolian landscapes vary so strongly, and there is strong potential for high topographic variability in soil respiration. Published information on how grazing infl uences soil respiration is inconsistent. In belowground dominant short-grass steppe regions, stable soil organic matter pools are not substantially affected by long term cattle grazing, but are highly variable with respect to topography and microsite (Burke et al., 1999). Our goal was to evaluate soil respiration in response to grazing across a complex landscape of Northern Mongolia, and to seek possible explanations for the variability in soil CO2 fl ux. Our fi rst objective was to determine how soil respiration varies across three grazing treatments (lightly grazed, moderately grazed and heavily grazed). We expected less soil respiration with increasing grazing intensity.
• Second, we wished to elucidate how soil respiration varies with slope aspect and position. Since north-facing slopes have a higher soil C than south-facing slopes in the top 50 cm, we hypothesized higher soil respiration in the northfacing slopes. However, this effect could be ameliorated or even non-existent due to the colder soil temperatures of these landscape positions compared with south-facing slopes and the presence of permafrost.
• Third, we wished to understand the variability in soil respiration in three different topographical locations (upper, middle and lower slope areas of the watersheds) of the south-facing slopes. Our hypothesis was that soil respiration rates decrease with increasing altitude, possibly due to lower soil organic matter and coarser soil texture.
• Finally, we wished to evaluate which of the factors (grazing vs. topographic location) play the most important role in determining the annual and inter-annual variability of soil respiration.

Material and Methods

Study area. We conducted this study within the framework of the Long Term Ecological Research program at the Lake Hцvsgцl National Park. Mean annual precipitation here is 350 mm and mean annual temperature is 4.5өC. Over the past four decades, mean annual air temperature has increased by 1.8oC (Nandintsetseg, 2004). The soil cover of the eastern part of the Lake Hцvsgцl area is complex due to the mountainous features of the area, exposure differences, the presence of permafrost, and soil parent material. Generally, the study area has sandy loam soils. The valley bottoms contain alluvial peat soils. Prior results indicate that soil microclimate is altered due to grazing, with surface soil temperature increasing and soil moisture content decreasing as a result of heavy grazing (Batkhishig, 2004). The north-facing slopes are usually more cold and wet, and have permafrost, compared with the south-facing slopes (Sharkhuu & Anarmaa, 2004). One of the site characteristics is that the nongrazed and lightly grazed watersheds (please see study design below) have a thick litter layer that is lacking in the grazed watersheds, and this layer protects the soil from evaporation, leading to higher soil moisture (Ariuntsetseg, 2004). Climate conditions. There was high interannual variability in weather for the duration of this study (Fig. 1). The 2003 growing season was considerably cooler (7.8oC), and wetter (127.5 mm precipitation) than 2004 or 2005. This variability was clearly refl ected in the respiration data we collected (Nandintsetseg, 2004).

Soil moisture was higher in 2003 (71.86% and 31.29% in the north- and south-facing slopes, respectively) compared to that of 2004 (44.64% and 17.14% in the north- and south-facing slopes, respectively) in both aspects and the north-facing slope always had signifi cantly higher soil moisture compared to the south facing slopes (Ariuntsetseg, 2004; Batkhishig, 2004). Soil temperature was lower in 2003 (13.4oC) compared with 2004 (17.5oC), but on the northfacing slope, soil temperatures were almost same for these two years. Soil temperature was always lower in the north-facing slope compared with the south-facing slope (Ariuntsetseg, 2004; Batkhishig, 2004). Design and measurements. We selected six tributary stream valleys occupied by different numbers of herder families. The most northern two valleys had more than 3000 livestock (converted into sheep units following the Law on Livestock Tax of Mongolia) in each valley; these are considered to be heavily grazed valleys. The most southern two valleys had less than 500 livestock, and the study area was only infrequently grazed, with non-grazed parts at all during this study. The two middle valleys were considered to have moderate grazing since they had an intermediate number of livestock that graze permanently (439 and 453 livestock in 2003 and 2004, respectively?). Carrying capacity estimate suggests that the two northern valleys are over-grazed, the middle two valleys, though grazed, animals do not exceed beyond the carrying capacity, and the southern two valleys have little to no grazing (Bayasgalan, 2005). We set transect lines 100 m wide across each valley, extending from the summit of the southfacing slope to the summit of the north-facing slope in each of the valleys. The grassland portion of each transect was divided into south-facing upper, south-facing lower slope, riparian and north-facing lower slopes. The boreal larch forest, which we did not sample, lies above north-facing lower slopes. We measured soil CO2 fl uxes and bare soil respiration using a PP System Closed soil respiration system (www.ppsystems.com). All measurements were taken between 10-12 AM, and 2-6 PM. Measurements were made in two ways. First, CO2 release from the soil was measured by placing the soil respiration chamber above both the green biomass and the litter layer. We call these measurements “soil CO2 fl uxes”. Second, bare soil respiration measurements were made by placing the soil respiration chamber over bare ground, where all aboveground biomass and litter had been removed. In all cases, we performed measurements of CO2 release in fi ve randomly selected spots within the 100 m wide transect area in each of the strata, the south-facing upper, the south-facing lower and the north-facing lower slopes, and the riparian areas.To address the fi rst question on the relationship between grazing and soil respiration, we measured soil CO2 fl ux once during the growing season in 2003, 2004 and 2005 in the south-facing lower slope landscape position of all of the study valleys. We measured bare soil respiration once during the summer months in 2003 and 2005 in the same south-facing lower slope landscape position at the same time as when we made soil CO2 fl ux measurements. To address our second question on the aspect and soil respiration relationships, we compared data from the north-facing lower slopes and the south-facing lower slopes. Soil CO2 fl ux measurements of the south- and north facing slopes were made at the same time as measurements to answer the above question of grazing, once during the growing season of 2003 and in 2004, while bare soil respiration measurements of the southand north-facing slopes were made only in 2003. We used measurements from the southfacing upper slope, south-facing lower slope, and riparian zones to answer our third question, which addressed relationships of topography and soil respiration. Data from different landscape positions were collected in the same years as for the previous question that focused on the relationship between soil respiration and aspect (Table 1).

To address the fourth question about the importance of grazing vs. topographic location on soil respiration variability, we compared soil CO2 fl ux data of the south-facing upper, south facing lower and the north-facing lower landscape positions from 2003 and 2004 of all study watersheds. For bare soil respiration we could only make comparisons of different topographical locations and grazing interaction for 2003. Statistical analysis. We tested the effects of grazing across the three grazing treatments (lightly grazed, moderately grazed, and heavily grazed treatments) using JMP IN 5.1 software (SAS Institute Inc (2004)). For the three grazing treatments, we used a two-way analysis of variance (ANOVA) design including three years (2003, 2004 and 2005) as a random effect in the analysis of the soil CO2 fl ux. For bare soil respiration, we used the same analysis with only two years (2003, 2005) as a random effect. For the effects of two different aspects and of three different topographic locations, respectfully, we used data only from the lightly grazed treatments and implemented a one-way ANOVA with bare soil respiration data collected in 2003, and two-way ANOVA with the soil CO2 fl ux measurements including the two years (2003, 2004) as a random effect. For testing the topography and grazing interactions of soil CO2 fl ux, we included data from four topographic locations (south-facing upper, south-facing lower, riparian and north-facing lower) along the three grazing treatments (lightly grazed, moderately grazed, and heavily grazed treatments), in two years (2003, 2004). Bare soil respiration variance along the four topographical locations and three grazing treatments were analyzed from one year’s data as a two-way ANOVA test. Further, we implemented regression analysis for grazing and topography effects on the soil CO2 fl ux and bare soil respiration. The purpose of the latter regression analysis was to determine how much of the variance is explained by grazing or topography.


Grazing Soil
CO2 fl ux. The average soil CO2 fl ux, which incorporated both aboveground leaf processes and belowground respiration, is increased with increasing grazing pressure (Fig. 2). This likely occurred as a result of decreased leaf uptake of CO2 with increased grazing. However, if we evaluate the data on a yearly basis, soil CO2 fl ux in lightly grazed areas increased dramatically in 2004 after wet cold of 2003, but in 2005 it decreased back to the level of 2003 (Fig. 2). In contrast, in heavily grazed areas, soil CO2 fl ux was higher in both wet cold 2003 and dry warm 2004, but decreased in 2005 that had similar climate conditions as 2004. In other words, soil respiration signifi cantly responds to inter annual climate variability depending on level of grazing. In lightly grazed areas, increase of soil CO2 fl ux could happen as a result of higher temperature conditions followed after the wet (2003) year while in heavily grazed dry and warm conditions that lasted for 2 years could result signifi cant drop of soil respiration. The soil CO2 fl ux of moderately grazed watersheds had no signifi cant changes among the three years. Generally, we measured lowest soil CO2 fl ux in 2005 in all three grazing treatments. Bare soil respiration. Compared with the soil CO2 fl ux, bare soil respiration showed a consistent pattern of decrease with increased grazing pressure (Fig. 2). Recall that the bare soil respiration measurements did not include leaf processes of either respiration or photosynthesis, so that this decrease must result from decreases in either belowground heterotrophic respiration resulting from lower labile organic matter, or reductions in autotrophic respiration associated with plant grazing responses, or both.

Bare soil respiration was lower in 2005 than in wet cold year of 2003. This could be due to the lower precipitation and thus low soil moisture of 2005. Slope aspect
Soil CO2 flux. In the lightly grazed areas, soil CO2 fl ux was higher on south-facing slopes than north-facing slopes during 2003 (Fig. 3). This likely occurred because low soil temperature was limiting CO2 fl ux on the north-facing slope, and temperatures were warmer on the south-facing slopes, especially in wet cold of 2003. In 2004 dry warm growing season, soil CO2 fl ux was higher than in previous years on both aspects, and especially so on the north-facing slope; differences between aspects dissipated. In the moderately and heavily grazed areas in 2003, we saw similar pattern of soil CO2 fl ux as of 2003 lightly grazed area (Fig. 3).

However, in these grazing levels, in 2004, we measured signifi cantly lower soil CO2 fl ux in south facing slopes. This 2004 drop in south facing moderately and heavily grazed slopes could directly associate with limitation of soil moisture or labile organic materials. Major differences between two slopes indicated by soil temperature, soil moisture and soil organic matter amount (labile fraction of soil organic matter). One may expect lower soil temperature, higher soil moisture and more labile materials in north facing slopes.
Bare soil respiration. In lightly grazed watersheds, soil respiration rates, measured without litter and green biomass, were higher in the north-facing slopes than the south-facing slope in 2003 (Fig. 3) and in 2005. In the wet and cold 2003, bare soil respiration was higher than during dry, warm 2005 year, likely due to low soil moisture in the dry warm 2005. As before, bare soil respiration was consistently lower in dry areas (heavy grazed areas) compare with wet areas (not grazed areas).
Topography The riparian (R) area almost always had signifi cantly higher soil CO2 fl ux than both the upper south-facing and lower south-facing slopes (Fig. 4).

This may have occurred either because of the high level of soil organic matter that concentrates in lowlands or higher soil moisture. The upper south-facing slope had signifi cantly lower soil CO2 fl ux compared with the other landscape positions. Bare soil respiration showed the same pattern as soil CO2 fl ux among the different topographic locations. It was higher in the riparian zone than of the upper south-facing and the lower southfacing slopes (Fig. 4).


Grazing, slope aspect, and topography are all had signifi cant effects on soil CO2 fl ux and bare soil respiration. However, the spatial variability conferred by topography explained a great deal more of the variation than grazing, for both soil CO2 fl ux and bare soil respiration (Table 2). These results are similar to those found by Burke et al. (1999) in short grass steppe, where topography had a more important infl uence on biogeochemical dynamics than grazing. In this Mongolian system, soil microclimate and substrate quantity are likely the major proximal controls over respiration; temperature appears to be the major limitation to respiration for much of the landscape. Similarly, though the infl uence of grazing was small, the effects may be interpreted as the response to changes in soil microclimate, with higher soil temperatures under heavy grazing that resulted in higher respiration rates during the coolest year.

Though we were not able to statistically test the effect of climatic variability, it is clear from the inter-annual variability and its infl uence on spatial variability that climate is the most important factor infl uencing CO2 fl ux in this system. Locations with suffi cient substrate quality and moisture responded to increases in temperature with increased soil CO2 fl uxes, or with increased sensitivity to moisture content. These results are important, given the documented increases in air temperature for this region, and the predictions of continued temperature increases.


We greatly appreciate support of Global Environment Facility, The World Bank and a trust fund of the Government of the Netherlands for providing guidance and fi nancial support for the research and for international travel. We especially thank Dr. Anthony Whitten and Mrs. Giovanna Dore of the World Bank for their encouragement and support. We also would like to thank all Project Implementation Unit and staff of the Hцvsgцl GEF Project and all researchers of the Short Grass Steppe Long-term Ecological Research site, Colorado State University (USA) for their advice and help. Biostatistical analysis was carried with the help and advice of Professor Peter Petraites, The University of Pennsylvania and Ms. Cao, Colorado State University.


  1. Ariuntsetseg L. 2004. Impacts of Nomadic Livestock on a semi-arid Boreal steppe plant community of Northern Mongolia. In C. E. Goulden, J. Tsogtbaatar & B. Mendsaikhan (Eds.): The Dynamics of Biodiversity Loss and Permafrost Melt in Lake Hцvsgцl National Park, Mongolia. Institute of Geoecology, Mongolian Academy of Sciences, Ulaanbaatar, pp 135-165.
  2. Batkhishig O. 2004. Soil morphology and Chemistry. In C. E. Goulden, J. Tsogtbaatar & B. Mendsaikhan (Eds.): The Dynamics of Biodiversity Loss and Permafrost Melt in Lake Hцvsgцl National Park, Mongolia. Institute of Geoecology, Mongolian Academy of Sciences, Ulaanbaatar, pp 45-100.
  3. Bayasgalan A. 2005. Nomad land use and livestock grazing impact. In C. E. Goulden, J. Tsogtbaatar & B. Mendsaikhan (Eds.): The Dynamics of Biodiversity Loss and Permafrost Melt in Lake Hцvsgцl National Park, Mongolia. Institute of Geoecology, Mongolian Academy of Sciences, Ulaanbaatar, pp 53-80.
  4. Burke I. C., Lauenroth W. K., Riggle R., Brannen P., Madigan B. & Beard S. 1999. Spatial variability of soil properties in the shortgrass steppe: The relative importance of topography, grazing, microsite, and plant species in controlling spatial patterns. Ecosystems 2, 422-438.
  5. Kang S. Y., Doh S., Lee D., Lee D., Jin V. L. & Kimball J. S. 2003. Topographic and climatic controls on soil respiration in six temperate mixed-hardwood forest slopes, Korea. Glob. Change Biol. 9, 1427-1437.
  6. Raich J. W. & Schlesinger W. H. 1992. The Global Carbon-Dioxide Flux in Soil Respiration and Its Relationship to Vegetation and Climate. Tellus B. 44, 81-99.
  7. Sharkhuu N. & Anarmaa S. 2004. Permafrost study. In The dynamics of biodiversity loss and permafrost melt in Lake Hцvsgцl National Park, Mongolia, Eds C E Goulden, J Tsogtbaatar and B Mendsaikhan. pp 101-133. Geoecology Institute of Mongolian Academy of Sciences, Ulaanbaatar.
  8. Tans P. P., Fung I. Y. & Takahashi T. 1990. Observational Constraints on the Global Atmospheric Co2 Budget. Science 247, 1431- 1438. Tsogtbaatar J. 2000. Forest policy development in Mongolia. In (Ed.): Forest policy development in Mongolia. Ulaanbaatar, pp. 60-69.