SOIL CARBON POOLS IN TWO NATURAL GRASSLANDS OF SERBIAN HIGHLANDS

: Grasslands are a major player in the global carbon cycle, although carbon stocks in grasslands are influenced by human activities and natural disturbances. The aim of this study is to determine differences in carbon stock on two test areas of grassland ecosystem in the highlands of Stara Planina and Zlatar Mountains (Serbia). The investigated sites are natural mountain grasslands of the same vegetation community ( Agrostietum capillaris Pavl. 1955) and soil type ( Umbric Leptosol (Dystric) and Haplic Cambisol (Dystric)), but with different grazing intensity. Aboveground and belowground biomasses were measured in each sample plot, and soil was sampled at fixed depths of 0-10, 10-20 and 20-40 cm. The estimation of C stock and the rate of soil C accumulation were determined by the Tier 2 method IPCC (2003). Carbon mineralization potentials were determined via sequential incubation procedure in the labora tory conditions. According to the obtained results, the greater amount of precipitation on Mt. Stara Planina resulted in a greater accumulation of aboveground biomass, which was subjected to a greater decomposition in situ, thus showing a lower amount of PMC in vitro . In addition, potentially mineralizable carbon (PMC) among the sample plots from both sites indicates that the mineralization of soil organic matter was more influenced by the factors related to the soil characteristics, climatic conditions and grazing.


SOIL CARBON POOLS IN TWO NATURAL GRASSLANDS OF SERBIAN INTRODUCTION
The largest carbon exchange takes place between the atmosphere and plants, therefore the terrestrial ecosystems play a key role in the global carbon cycle. The way of managing grassland ecosystems in order to accumulate carbon and increase the yield, reduction in sensitivity to nitrogen inputs, make the whole ecosystem more resistant to climatе change (Lal, 2009;FAO, 2011).
Biomass in grassland systems, being predominantly herbaceous, is a small, transient carbon pool (compared to forest) and hence soils constitute the dominant carbon stock (C o n a nt , 2010). Global estimates of the relative amounts of C in different vegetation types suggest that grasslands probably contribute >10% of the total biosphere storage (Eswaran et al., 1993;Nosberger et al., 2000). The aboveground biomass of grassland ecosystems can accumulate relatively small carbon stock in comparison to that in the soil. Soils store at least three times as much carbon (in SOM) as is found in either the atmosphere or in living plants (Fischin et al., 2007). Grasslands with their belowground carbon storage are a major player in the global carbon cycle, although carbon stocks, productivity and turnover time are subject to considerable uncertainty (S cu rl o ck & H al l , 1998).
An unprotected C pool in SOM consists of both the light fraction and/or the particulate organic matter (POM) fraction, as well as microbial debris (Jones & Donnely, 2004) that are highly labile organic matter pools and the variation in the light fraction pool is the best indicator of management-induced changes in SOM as shown by many researchers (Jenkinson, 1988;Gregorich et al., 1996; H ay n e s , 2000, S a l j n i ko v, 2004;L u o & Zhou, 2006;F u n a kawa et al., 2010). Because almost all the labile C in grassland ecosystems is in the soil, and our entire understanding of processes occurring here is still relatively weak, the emphasis was made in identifying the total carbon stock and labile carbon sources via soil respiration and mineralization of organic matter.
Soil respiration is the second largest carbon (C) flux between terrestrial ecosystems and the atmosphere in the global C cycle (Jia & Zhou, 2009;Thomey et al., 2011), and plays an important role in regulating the soil carbon pool and ecosystem C-cycling (C ox et al., 2000;S a i z et al., 2006;Wang et al., 2016).
The paper presents the results of a pilot project intended to estimate and compare carbon stock on two test areas of grassland ecosystem in the highlands of Stara Planina and Zlatar Mountains in Serbia, determined by measuring the aboveground and belowground biomass and carbon stock, total and labile carbon of SOM. The main aim of the research was to find out the influence of land use on carbon storage and carbon allocation in grassland carbon pools, in the two natural mountain grassland sites, under similar climatic and pedological conditions and different grazing intensity, and the soil factors that are responsible for those differences.

Study area
The study was conducted in highland grasslands of Mt. Zlatar and Mt. Stara planina Mt. Zlatar is located in southwest Serbia and it belongs to the mountain range of the Dinarides (Fig. 1). The average height of Mt. Zlatar is 1200-1400 m asl. Transitional climate type of Mt. Zlatar results from the interaction, both of maritime and continental air masses. Mean annual air temperature is 7.10C and total annual precipitation is 751.5 mm based on the observations of the 1990-2010 period (RHOS, 2010 -MMS Sjenica). The massif of Mt. Zlatar mainly consists of limestone, sandstone, shales and flint (Ć i r i ć et al., 1977) and Umbric Leptosol (Dystric) and Haplic Cambisol (Dystric) (WRB, 2006). The vegetation of Mt. Zlatar is conditioned by the climate and characterized by the presence of climatogenic vegetation communities such as frigoriphilic spruce and mesophilic beech and beech-fir forest types. Pastures are spread over an area of 24.31% and meadows on 15.45% of the area of Mt. Zlatar (D ragovi ć et al., 2009). Grasslands of Mt. Zlatar are characterized by a high-mountain community of Agrostietum capillaris Pavl. 1955 which alternates with other types of grassland communities in the zone of beech, beech-fir and spruce forests.
Mt. Stara planina is located in southeast Serbia on the border between Serbia and Bulgaria as the extension of the Carpathian mountain range (Fig.  1). The area has a mountain climate with mean annual air temperature of 6.10C and mean annual precipitation of 1090 mm . Mt. Stara planina is overlaid with conglomerates, sandstones, argillaceous schist and limestone (Krst i ć et al., 1970). Unlike Mt. Zlatar, there is altitudinal zonation of forest vegetation in Mt. Stara planina. The most common forest type at altitudes 1200 to 1550 m asl. is Montane beech forest (Fagetum moesiacae montanum Jov. 1953). Grassland communities have been developed in accordance with the height gradient -from mountain meadows and pastures over the mountain and sub-alpine to Alpine mountain pastures (M i š i ć et al., 1978). The dominant grassland community is Agrostietum capillaris Pavl. 1955.

Data collection
In total eight sample plots were established in the highland pastures in the community of Agrostietum capillarisin Mt. Zlatar (Z1, Z2, Z3 and Z4) and Mt. Stara planina (SP1, SP2, SP3 and SP4) ( Figure  1). The altitude of the sample plots varied from 1296 to 1479 m asl. (Table 1). The investigated soil types were Leptosol and Cambisol (WRB, 2006) formed on flint and schists. Average bulk densities of the investigated soils were 0.95g cm -3 (Mt. Zlatar) and 0.80g cm -3 (Mt. Stara planina). In investigated pastures of Mt. Zlatar moderate free grazing by goats has occasionally been present, while investigated pastures on Mt. Stara planina were undisturbed by grazing or other activities such as clipping and mowing.

Biomass and C stock
Aboveground biomass (net primary production ANPP) was determined in July 2015 according to LULUCF Guidelines (IPCC, 2003) applying the destructive "clip and weight" method (SOP2034: 1994). For each site four 0.5 m 2 quadrates were sampled. Aboveground biomass within the quadrates was harvested to the ground level (living aboveground biomass, standing litter and ground litter), then the living aboveground biomass was separated from the standing litter of the previous year and ground litter, pooled by plot and weighted. Harvested biomass was oven-dried at 700C to constant weight.
Belowground biomass (net primary production BNPP) was measured in each sample plot using the soil core method (Ravindranath & Ostwald, 2008) by 2 soil cores of 10.3-cm diameter in 10-cm depth layers to 30 cm within each quadrate. Belowground biomass (roots, rhizomes etc.) contained within the excavated soil was separated with water through a 0.3-mm mesh sieve. The samples of belowground biomass were oven dried at 70°C to constant weight.
Aboveground and belowground biomass total C was measured with an elemental CNS analyzer, Vario model EL III (ELEMENTAR Analysensysteme GmbH, Hanau, Germany; N e l s o n & S o m m e rs , 1996).

Soil properties
Four soil profiles were opened on each sample plot (a total of eight profiles), with soil sampling carried out under the Soil Sampling Protocol to Certify the Changes of Organic Carbon Stock in Mineral Soils (S to l b o vo y et al., 2005), at fixed depths of 0-10 cm, 10-20 cm and 20-40 cm. The main physical and chemical soil properties were determined on air-dried samples using the following methods: the pipette method was used for  ' -43°24'27'' N and 19°48'36'' -19°48'4'' E 8.08 1275-1428 W-SW particle size analysis (ISO 11277:1998); bulk density (BD) was measured by drying the cores at 1050C to a constant weight (ISO 11272:1993); particle density (ISO 11508:2002); soil pH was determined using a glass electrode in a 1:5 (volume fraction) suspension of soil in water (pH in H 2 O) (ISO 10390:2007); the hydrolytic acidity (cmol kg -1 ) (extraction by CH 3 COONa, titration with 0.1M NaOH) and the sum of exchangeable basis (S)(cmol kg -1 ) (extraction by 0.1M HCl, titration with 0.1M NaOH) was determined using Kappen's method (Kappen, 1929), and the total capacity of cation adsorption (T)(cmol kg -1 ) and degree of base saturation (V%) -were calculated (Hissink, 1925). Soil organic carbon (C) was measured by the Tjurin method (Nelson & Sommers, 1996) and total nitrogen (N) by the Kjeldahl method (ISO 11261:1995). After extraction, the available P and K were determined by the Al-method of Egner-Riehm (E gn er & Ri eh m, 1958). To ensure reliable results all the analyses were performed in 2 replications.

Soil respiration
Carbon mineralization potentials were determined via sequential incubation procedure in laboratory conditions under the controlled moisture and temperature environment for 2-, 4-, 6-, 8-, 10-weeks as described by Janzen (1987). Soil respiration was measured after trapping the evolved CO 2 by 1М NaOH and by titration of the remained NaOH by HCl.
Potentially mineralizable carbon (C 0 ) was obtained after fitting the data of mineralized C to the first order kinetic model (SPSS Inc., 2007): C min =C 0 *(1-exp(-k*t)), where, C min is an experimental value of mineralized C at a given time (t), C 0 is the value of potentially mineralizable C that was calculated after fitting the curve, k is the mineralization rate constant.
Annual litter input to soil, excluding root exudation and turnover, can be determined as aboveground plant production plus standing root mass. Belowground respiration includes total carbon lost through decomposition of soil organic matter and root respiration (H u n ga t e et al., 1997). Тhe loss of carbon through respiration measured repeatedly throughout the 90 days in a laboratory incubation assessment.

RESULTS
The main soil chemical properties are given in Table 2. The studied soils from both sites showed moderate to very low pH values with low base saturation. Soil acidity in water solution was not statistically different between the studied sites. Generally, cation absorption capacity and content of humus in 0-10 and 10-20 cm soil layer was higher in Stara Planina than in Zlatar. The content of soil total C and N were significantly higher in Stara Planina than in Zlatar. The contents of plant available P and K were not statistically different in two locations, where the content of P was low, and K content is medium in both locations.
By mechanical composition, the soil from Mt. Zlatar showed a higher content of clay fraction than the soil from Mt. Stara Planina (Table 2). Mean particle density in the 0-10 cm soil layer was 64% in the Mt. Zlatar soils, and 66% in the Mt. Stara planina soils.
The aboveground biomass in the Mt. Zlatar sites ranged from 3.09 (plot 3) to 4.11 Mg ha -1 (plot 1). The aboveground biomass in the Mt. Stara Planina sites ranged from 3.93 (plot 4) to 5.88 Mg ha -1 (plot 1). A lower amount of aboveground biomass was recorded in the Mt. Zlatar sites in the plots with grazing (Table 3).
Soil respiration and the amount of potentially mineralizable carbon (PMC) in the two sites did not statistically differ. The rate of organic carbon mineralization was higher in the Mt. Zlatar soils compared to the Mt. Stara Planina soils. The highest amounts of PMC were recorded in site 3 on Mt. Stara Planina and site 4 on Mt. Zlatar, while the highest rates of mineralization were recorded in sites 1 and 3 on Mt. Zlatar and site 1 on Mt. Stara Planina (Table 3).
The total carbon stock in biomass was higher in the Mt. Zlatar sites than in the Mt. Stara Planina soils, but that was not statistically confirmed, while carbon stock in the soils from the two sites significantly differed, showing higher C reserves in the soils of Mt. Stara Planina (Table 4). This difference determined the overall significantly higher total carbon stock in the Mt. Stara Planina soils than in the Mt. Zlatar soils.

DISCUSSIONS
The differences in the content of labile carbon in the Mt. Stara Planina and Mt. Zlatar soils can be attributed to the differences in the amount of aboveground and belowground biomass, which resulted in the different composition of the light fraction OM. Grasslands contain a much higher belowground biomass than aboveground biomass, but as a rule the composition of the aboveground grass biomass is a much easier source for microbial attack after their deposition into the soil when it becomes a part of soil organic matter. Soils from Mt. Stara Planina showed generally higher respiration under the controlled conditions, which might indicate higher microbial activity in these soils. In addition, more intense respiration resulted in a greater amount of measured labile carbon in the Stara Planina soils. Decomposition of SOM is a biological process. Therefore, changes in below-ground carbon or substrate availability may alter microbial community composition and activity that in turn alters the decomposition rate and the type of organic matter that are decomposed . This assumption is also confirmed by the soil C stock that was higher in the Stara Planina soils, and the total biomass C stock that was higher in the Mt. Zlatar soils (Table  4). This indicates that in the Mt. Zlatar soils, the soil organic matter mineralization processes was going on more intensively in situ, thus depleting the light fraction OM, which was reflected in the laboratory incubation measurement of labile C, where lower PMC and a higher mineralization rate constant were recorded.
Despite greater belowground biomass the content of total C was much lower in the Mt. Zlatar soils than in the Mt. Stara Planina soils. The accumulation and decomposition of soil organic matter primarily depend on moisture and temperature and available substrate for soil microorganisms, where the labile carbon accumulation gradient increases from wet to dry and from cold to hot climate. In contrast, the gradient of accumulation of soil organic matter increases from dry to wet climate and from hot to cold climate (S a l j n i kov et al., 2009). This is confirmed by the data from the two studied sites, of which Mt. Zlatar is characterized by higher temperatures and lower precipitation (7.10C and 751.5 mm, respectively) compared to Mt. Stara Planina (6.10C and 1090 mm). The higher amount of precipitation on Mt. Stara Planina resulted in a greater accumulation of aboveground biomass, which was subject to greater decomposition in situ, thus showing a lower amount of PMC in vitro. This is in accordance with the results of Guntiñas et al. (2013), who reported that the sensitivity of soil organic matter to temperature is higher at low soil moisture contents. Ko n g & S i x (2010) and R a s s e (2005) also reported that root-originated carbon resist in soil much longer than above-ground biomass and at the same time, root-derived substrate contains much of rhizosphere that actively decomposes the soil organic matter (S ch mi d t et al., 2011), which might also aid the greater mineralization of SOM of the soils from Mt. Zlatar, when temperature and moisture conditions were favorable under laboratory incubation. In our research despite the same vegetation and soil type of each site, the distribution of labile carbon, measured as potentially mineralizable carbon (PMC) among the sampled plots from both sites indicate that mineralization of soil organic matter was more influenced by the factors related to soil characteristics and grazing impact.
The correlations between biological soil parameters confirm the hypothesis that different factors influenced the amount of PMC and soil respiration (Table 5). In the Mt. Zlatar soils, where the belowground biomass was significantly higher than in the Mt. Stara Planina soils, significant positive correlations were found between belowground biomass and soil respiration and between belowground biomass and PMC, while in the Mt. Stara Planina soils, there was a correlation recorded between soil total carbon and PMC. This implies that the sources of substrate for microbiological decomposition were different between the sites.
Significant changes in the large pool of total and labile soil carbon are difficult to be detected over the short period of this experiment. Small increases in soil carbon can, however, lead to large increases in soil respiration if the carbon is delivered to one or more highly labile fractions in the soil (T h o mps o n et al., 1996). In addition to that, according to H af n e r et al. (2012) the larger belowground C allocation of plant biomass, the larger the amount of recently assimilated C remaining in the soil, which is the result of a positive effect of moderate grazing on soil C stock and C sequestration, which is similar to the Zlatar site. According to Fa n et al. (2013) fenced grasslands in order to exclude grazing, affected total carbon in aboveground biomass to be significantly higher than in grazed grasslands, and they also suggest this method as an alternative approach to sequester C to the soil in alpine meadow systems.

CONCLUSION
The soil C stock was higher in the Stara Planina soils, and the total biomass C stock was higher in the Zlatar soils, which could be the result of more intensive processes of organic matter mineralization in the Zlatar soils in situ, thus depleting the light fraction organic matter. In the Zlatar soils, despite greater belowground biomass the content of total C was much lower than in the Stara Planina soils. The main conclusion of the preliminary study implies that the amount of labile carbon stock was not statistically different, while the sources of labile carbon differ between the studied sites. The distribution of labile carbon, measured as potentially mineralizable carbon (PMC) from both sites indicate that the mineralization of soil organic matter was more influenced by the factors related to soil properties and grazing impact.
The obtained results suggest further detailed studies of organic carbon characteristics and dynamics, as well as the processes of transformation of organic carbon fractions for a better understanding of carbon transformations in natural mountain meadow associations.
bia" reference no. 404-02-216/10/2015-15 financially supported by the Ministry of Agriculture and Environmental Protection of the Republic of Serbiaand the project "The Climate Change and Its Impact on the Environment -Monitoring, Adaptation and Mitigation" with reference number 043007 financially supported by Ministry of Education and Science of the Republic of Serbia.