Seasonal/Interannual Variations of Carbon Sequestration and Carbon Emission in a Warm-Season Perennial Grassland

Carbon sequestration and carbon emission are processes of ecosystem carbon cycling that can be affected while land area converted to grassland resulting in increased soil carbon storage and below-ground respiration. Discerning the importance of carbon cycle in grassland, we aimed to estimate carbon sequestration in photosynthesis and carbon emission in respiration from soil, root, and microbes, for four consecutive years (2007–2010) in a warm-season perennial grassland, Japan. Soil carbon emission increased with increasing growing season temperature which ranged from 438 to 1642 mg CO 2 m −2 h −1 . Four years’ average soil carbon emission for growing season, nongrowing season, and annual emission was 1123, 364, and 1488 gCm −2 , respectively. Nongrowing and snow covered season soil carbon emission contributed 23–25% and 14–17% to the annual emission. Above-ground biomass varied seasonally and variation in green biomass affected soil carbon emission with increasing temperature and precipitation. Temperature effect on root carbon emission contributed about 1/4th of the total soil carbon emission. Variation in soil and root carbon emission is affected by below-ground biomass. Long-term estimation concluded that seasonal and interannual variations in carbon sequestration and emission are very common in grassland ecosystem.


Introduction
Carbon sequestration and emission can be affected while the ecosystem land area converted to the grassland [1][2][3], resulting in increased soil carbon storage [4,5], belowground respiration, and carbon cycling [6]. Land area converted to grassland has been reported to affect ecosystem carbon balance in short term through increased variations in ecosystem carbon cycling [7][8][9]. Moreover, one of the major long-term effects of such land conversion is the reduced carbon sink strength of grassland ecosystems [10]. Grasslands are able to store 10-30% of global soil organic carbon due to its large coverage (40%) of the earth's terrestrial surface and are considered to sequester large amounts of carbon from the atmosphere and hence acquire significant sink/source capacity of the ecosystem and play a major role in the global carbon balance [11][12][13].
Soil as a whole is a large pool of atmospheric carbon. Globally, it accumulates 1500-2000 petagrams (Pg carbon, 1 Pg = 10 15 g) in the form of soil organic carbon in the upper 1 m of soil [14,15] or even greater [16], which is at least three times larger than atmospheric carbon. Small variation in carbon storage in grassland soil will greatly affect the concentration of atmospheric carbon and the trend of regional or global climate change [17,18], although grassland acts as a sink/source of atmospheric carbon that depends on its grazing intensity, land use, management, and climate [19]. The biogeochemical process (carbon cycling) that controls the exchange of carbon between soil and atmosphere is not yet well understood [20]. The amount of carbon during sequestration and emission is not the highest for grasslands compared to forest, but their large land mass plays an important role in global carbon storage and cycling [21]. On a global scale, grasslands cover approximately the same area as forests [22], but very less attention has been given to the grassland carbon cycling which may have a significant role in balancing the global carbon budget [23,24].
Soils are considered to sequester large amounts of carbon from the atmosphere via plant photosynthesis and emit carbon via respiration by plant and microorganisms and their balance determining whether the ecosystem is a carbon sink/source [25][26][27]. Thus this process has been extensively 2 Journal of Ecosystems investigated to improve the understanding of carbon cycling in terrestrial ecosystems. Soil carbon emission is very sensitive to environmental conditions importantly influenced by soil temperature [28] and availability of soil water [29] that have potential role either to amplify global warming [25,30] or to mitigate climate change due to enhanced soil carbon sequestration and reduced emission [31,32]. Therefore, it is essential to improve our understanding of the variations in carbon sequestration and carbon emission in ecosystem for better management of these potentials. Because of high seasonal [33][34][35] and interannual [36][37][38][39][40][41][42] variations of carbon sequestration and carbon emission, continuous and long-term studies in local scale are necessary for evaluating the carbon budget on the global scale [12].
Considering a major emission between the atmosphere and land, soil carbon emission is estimated to be 50-78 Pg C y −1 and accounted 25% of the global carbon exchange [43][44][45]. Accurate estimation of soil carbon emission thus becomes one of the important issues in the global carbon cycle. The important factors determining seasonal and interannual variations in soil carbon emission are temperature and soil moisture in grasslands [36,[46][47][48]. Generally, soil carbon emissions increase through respiration with temperature; exponential [47,[49][50][51][52] power [50] and linear functions [53] are used to describe the temperature function of soil carbon emission. Similarly, carbon sequestration increases through photosynthesis via biomasses with increasing soil moisture in growing season [28,[54][55][56][57][58]. In carbon cycling, carbon sequestration and emission between ecosystem and atmosphere in grassland can be roughly estimated using environmental variables for short period and for the confounding result, long-term continued measurements and estimation for years are required. So far, very few literatures are available regarding information on perennial grassland ecosystem relating to long-term estimation of carbon sequestration (uptake), emission (release), and its dependence on climate. Hence long-term measurements and estimation for years holds significant scientific importance to have quantitative study on carbon cycling between soil and atmosphere as well as prediction on its future possible changes [59]. The most important link in grassland carbon cycling is soil carbon emissions to atmosphere through respiration and the amount of soil carbon determines directly soil carbon turnover rate [60].
Knowing the importance of carbon cycling in grassland and its implications on present climate change, we aimed to measure and estimate the carbon sequestration during photosynthesis from plant biomass and carbon emission during respiration from soil including root and microbial respiration for four consecutive years in a warm-season perennial grassland. This perennial grassland sequesters carbon only during the growing season that is half of the year (May-October) and emits carbon throughout the year, but not in same amount. Except early/late growing season (April/November) and growing season (May-October), the grassland is covered with snow in winter season (December-March). Thus, overall goal of this study was to estimate and determine the seasonal/interannual variation of carbon sequestration and carbon emission in the grassland with less climatic variability (i.e., temperature and precipitation) between the years from 2007 to 2010.

Site Description.
This study was conducted in permanent grassland located near the Takayama Experimental Field Station (36 ∘ 08 N, 137 ∘ 25 E, elevation 1,342 m a.s.l.) from 2007 to 2010. The site is about 15 km east of Takayama City in central Japan and it was established in 1913 which occupies 17.9 ha including small land area covered by forest to provide grazing land for cows. Every year, cattle enter in June and leave in October in the site; hence grazing period occurs for five months in a year. It has cool-temperate climate with high diurnal temperature and plentiful rainfall in summer. The mean air temperature was 7.2 ∘ C and annual precipitation was 2073.5 mm from 2001 to 2010 recorded at the Takayama Experimental Field Station (100 m from the study site). Temperature and total monthly precipitation increases with the progress of growing season starting from May and decreases from the end of October.
The site is dominated by a warm-season perennial C 4 grass Zoysia japonica and other grass species such as Ranunculus japonicus, Trifolium repens, and Poa spp. are also common. Z. japonica begins to grow in early May and reaches its seasonal peak in summer between mid-July and September. It is very active during the period of high temperature and precipitation, whereas growth of the other grass species peaks in June [55].

Soil Carbon Emission Measurement.
The measurement of soil carbon emission (CO 2 ) was conducted monthly using a portable photosynthesis system (LI-6400; Li-Cor, Lincoln, NE) fitted with a soil respiration chamber (LI-6400-09) from May to October, 2007, in the study site. The chamber (12 cm diameter and 15 cm height) was fitted to a soil collar ( = 20) made of polythene, which was inserted into the soil one day before the measurements in order to remove some installation effects. All the measurements were made between 11:00 and 12:00 AM in each month to avoid diurnal fluctuations. Three consecutive measurements were conducted on each collar without removing the chamber and average of three measurements was taken as the representative value of each collar. Soil temperatures between 3 and 5 cm depth were measured during soil carbon emission measurements in each time. Volumetric soil water content (5 cm soil depth) was also measured by the time-domain reflectometry method (TRIME, IMKO, Ettlingen, Germany) within the soil collar at the time of soil carbon emission measurements. Continuous measurements of temperature between 3 and 5 cm soil depth were also recorded in the site at 1 h intervals over the 4 years of our study period by using Stowaway TidbiT Temperature Data Logger, Onset Computer, MA.
The relationship between soil carbon emission and soil temperature was analyzed by regression analysis. The sensitivity of soil carbon emission to soil temperature was assessed by fitting exponential functions to the data as follows: where [ ( )] is the predicted soil carbon emission rate (mg CO 2 m −2 h −1 ) at ST (T, ∘ C) at a depth of 3-5 cm, represents the intercept of soil carbon emission rate when ST is zero, and represents the temperature sensitivity of soil carbon emission. The value was used to calculate a coefficient of temperature sensitivity (respiration quotient, 10 ), which describes the change in soil carbon emission over a 10 ∘ C increase in soil temperature by Values of the parameters (i.e., , , and 10 ) derived from the seasonal data sets reflect the effects of temperature and other covarying factors on soil carbon emission rate. Thus, an apparent 10 is used to denote the derived temperature sensitivity of soil carbon emission.

Root Carbon Emission
Measurement. An open-flow gasexchange system was used to measure root carbon emission in the laboratory. The measurements were conducted by using roots sampled from field during midgrowing season in August to determine the temperature effect on root carbon emission. Initially, sampled roots ( = 3) from soil block were washed carefully to remove the soil and separated into fine roots and rhizomes. Measurements were made within 1 h of root sorting from the soil. Roots were placed in a cylindrical chamber (polythene pipe; 3 cm in diameter, 25 cm in length) connected to the respiration measurement system, and ambient air at a rate of 0.5 L min −1 was flowed into the system. The temperature inside the chamber was monitored using a copper-constantan thermocouple and controlled by keeping it in a water bath regulated by a condenser (Coolnit C80R, Taiten, Japan) to maintain the roots at five different temperatures: 7, 12, 17, 22, and 27 ∘ C. The CO 2 concentrations of the reference gas and a mixture of sample and reference gas were measured using a CO 2 /H 2 O gas analyzer (LI-7000, Li-Cor) to determine the CO 2 concentration of the sample gas. After the root carbon emission measurements, the root samples were oven-dried and weighed to calculate the rate of respiration per unit of dry root biomass.

Plant Biomass Measurement.
Plant biomass was measured monthly during the growing season between May and October from 2007 to 2010 by removing the aboveground plant biomass at ground level and below-ground plant biomass up to 20 cm soil depth in the grazing area. The aboveground plant parts were sampled in five 20 × 20 cm quadrants, oven-dried at 70 ∘ C for 72 h, and weighed to calculate the dry weight. Similarly, below-ground parts were sampled in five 20 × 10 cm soil blocks, washed with water, and then hand-sorted into fine root and rhizome. These samples were oven-dried at 70 ∘ C for 72 h and weighed separately.   Table 1. The values show that growing season and annual soil carbon emission in consecutive years are variable and not exactly the same but nongrowing season soil carbon emission in years were nearly stable and equal. Contribution of nongrowing season and snow covered season (Nov.-Feb.) soil carbon emission to the annual soil carbon emission was 23 to 25% and 14 to 17%, respectively. We estimated that four consecutive years' (2007 to 2010) average soil carbon emissions of the grassland were 1123, 364, and 1488 g C m −2 for growing season, nongrowing season, and annual emission, respectively.

Plant Biomass.
Carbon sequestration in ecosystem followed accumulation of plant above-ground and belowground biomass through growth. The above-ground biomass varied seasonally (one-way repeated-measures ANOVA; <  0.0001) as in Figure 5. The green biomass starts to grow from 49.5, 108.5, 152.4, and 57.9 g dw m −2 (dead parts from previous year were not included) and reached its (dead + green) peak at 711. 8 varied seasonally with the variations of growing season and it ranged from 1074.5 to 1235.0 g dw m −2 in consecutive four years. Figure 6 represents the variations of soil carbon emission with the changes in above-ground green biomass during growing season in 2007, 2008, 2009, and 2010 and shows that green biomass was increased with increasing growing season. In the meantime, some amount of green biomass was removed from grassland in grazing by cattle. Therefore, soil carbon emission was influenced with the change in growing season; it increases with increasing above-ground green biomass and soil temperature (Figure 3(a)). And plant biomass increases with increasing soil water content and temperature; the unclear effect of soil water content on soil carbon emission (Figure 3(b)) in this study is not much far from its positive relation. However, higher precipitation suppresses soil carbon emission due to soil capillary effect and higher temperature with removing water availability in soil affects plant growth and soil carbon emission. The above-ground green biomass was peaked at 322.9, 285.9, and 240.2 g dw m −2 in September 2007, 2009, and 2010, respectively, and it was peaked at 411.0 g dw m −2 in August 2008.

Root Carbon Emission.
The temperature effect on root carbon emission was explained by Figures 7(a) and 7(b). Measurements of fine root (Figure 7(a)) and rhizome (Figure 7(b)) were performed separately to estimate their contribution to total root carbon emission and their seasonal variations. Temperature effect on fine root and rhizome was very significant which was explained by 2 = 0.96 and 2 = 0.97, respectively. The coefficient of temperature sensitivity ( 10 ) of carbon emission from fine root and rhizome was 2.3 and 1.9, respectively. Higher 10 of fine root was obtained compared to the rhizome because fine roots are more active and sensitive than the rhizomes during growing period in the temperate perennial grassland.
The regression equations obtained from the temperature effect on root carbon emission were used to analyze seasonal variations in carbon emissions from root during the study period as represented in Figure 8. Root carbon emission was increased with increasing temperature during growing season and it was peaked at 208.4 mg CO 2 m −2 h −1 for fine root and 149.6 mg CO 2 m −2 h −1 for rhizome in August from 2007 to 2010. Fine root contributed higher than the rhizome to the total root carbon emission; the contribution of it was increased in mid-and late growing season (Figure 8).
Below-ground biomass was increased with progressing growing season. The variations of soil carbon emission and root carbon emission with the changes in below-ground biomass from 2007 to 2010 were explained in Figure 9. Both soil carbon emission and root carbon emission increased with increasing growing season temperature and were peaked in August (midgrowing season) and decreased with decreasing temperature towards the end of growing season. The difference between soil carbon emission and root carbon emission is the microbial or heterotrophic carbon emission. Contribution of root carbon emission to total soil carbon emission ranged from 24 to 30%. Therefore, root carbon emission contributed about 1/4th of the total soil carbon emission and remaining higher contribution was made by heterotrophic carbon emission.

Discussion
Common ecosystem variables, light, temperature, precipitation, and growth, can effectively predict carbon sequestration and carbon emission in grassland during growing period which has been reported by Flanagan et al. [61], Hui et al. [62], and Nagy et al. [38], and temperature is the best confounding abiotic factor determining soil carbon emission that is proven by significant ( 2 = 0.83) positive exponential relationship between soil carbon emission and soil temperature (Figure 3(a)) obtained in this study. Such relation is consistent with the studies [50,63] in grasslands, [51,64] in forest, and [65] in desert shrubland. The higher range (438 to 1642 mg CO 2 m −2 h −1 ) of soil carbon emission obtained during growing season of this study was comparable with the range (790 to 1828 mg CO 2 m −2 h −1 ) obtained in C 3 /C 4 grassland in Japan and also followed similar trend of lowest soil carbon emission in early growing season (April-May) and highest in midgrowing season in August [66]. The similar range (261 and 1361 mg CO 2 m −2 h −1 ) of soil carbon emission was also reported by Yan et al. [67] in Chinese poplar plantation comparable to our study. The reason for higher range of soil carbon emission in growing season of this study was owing to the lowest soil temperature in May and that was the period after 4/5 months of long snowy winter so as vegetation of the grassland was almost dead and the highest soil temperature was reached in August after the period of plentiful rainfall, and sufficient quantity of plant biomass was reached with grass growth in July. Other reports (47 to 745 mg CO 2 m −2 h −1 ) in perennial Mongolian grassland [68] during growing season and (24 and 730 mg CO 2 m −2 h −1 ) in northern Chinese forest [69] (63 and 306 mg CO 2 m −2 h −1 ) in arid shrubland in Spain [35], and (73 and 225 mg CO 2 m −2 h −1 ) in different subtropical Chinese forest [51] in growing and nongrowing season have lower range of soil carbon emission than this study; however all the values were not exactly the same and varying each other. McCulley et al. [70] suggested that mean soil organic carbon in forest can be as much as two times larger compared to increased soil carbon emission in grasslands, although soil carbon emissions were comparable and lower/higher in both forest and grassland. Therefore, soil carbon emission is determined by environmental variables, vegetation, land use change, and geographical locations which are mostly not the same and variable. 10 temperature response of soil carbon emission value 2.1 of this study (Figure 3) [64] in forests. The range (3.4-5.6) of 10 values reported by Davidson et al. [71] in a temperate mixed hardwood forest was much higher than some previous studies, and it was attributed by contribution of root to the soil carbon emission and variations due to temperature, precipitation, and soil microclimate.
The temperature effect of soil carbon emission was not affected ( 2 = 0.018) by the soil water content in our measurement period (Figure 3(b)), but Inoue and Koizumi [75] observed a positive linear relationship ( 2 = 0.55) only in midgrowing season (July-September); however soil moisture is the determinant abiotic factor of soil carbon emission [71,72,76,77]. Zhou et al. [78] observed both aboveground biomass and soil carbon emission increased linearly with mean annual precipitation in Southern Great Plains, USA, and not only summer rainfall but also winter snow accumulation in grassland affects carbon cycling processes in winter and subsequent summer [79]. Greater soil water facilitates microbial decomposition of organic matter and increases heterotrophic carbon emission [80] and increases above-ground and below-ground plant and associated vegetation carbon emission [81,82]. The unclear effect of soil water content on soil carbon emission of our study might be owing to the variations in measurement point, effect of removing grass inside the soil collar and single measurement in a month. Although effect of soil moisture on soil carbon emission is more complex and depends on site specific soil parameters [83], hence long-term continuous measurement might be useful to confound the relation.
Significant seasonal variations in soil carbon emission were observed in four years (Figure 4), and increasing temperature, precipitation, and plant growth are the fundamental factors that result in increasing soil carbon emission during growing season. Maximum values of monthly soil carbon emissions in 2007 (1285 mg CO 2 m −2 h −1 ), 2008 (1305 mg CO 2 m −2 h −1 ), and 2009 (1285 mg CO 2 m −2 h −1 ) were nearly equal and comparable with each other, but it is slightly higher (1414 mg CO 2 m −2 h −1 ) in 2010 ( Figure 4) which was owing to higher soil temperature and precipitation (Figure 1) in July and August than the previous years. However, the range of soil carbon emissions in each month for all years was nearly equal and they all followed similar trend of seasonal variations so that the values were comparable to each other. Due to the seasonal variations of soil carbon emission, estimation of annual soil carbon emission is effective and accurate, if it is separately made during growing season and nongrowing season followed by the separate measurement made in each month and it was supported by Wei et al. [69] and Yan et al. [67]. Our growing season soil carbon emissions (1082, 1134, 1090, and 1187 g C m −2 ) in four years were consistent with the values (997, 1043, and 1358 g C m −2 ) reported by Li et al. [50]. The annual soil carbon emissions (1448, 1500, 1459, and 1544 g C m −2 ) of our study were comparatively higher than the values (395 to 973 g C m −2 ) reported by Wei et al. [69] in different Chinese grasslands and (755, 719, and 1037 g C m −2 in 2007, 2008, and 2009) by Inoue and Koizumi [75] in the same study site. The reason behind variations in soil carbon emission was owing to the variations in study area, use of separate equipments, and size of the chamber; the smaller the size of chamber, the higher the value of emission [75] and measurements time. Comparable annual soil carbon emissions (1151 and 1161 g C m −2 ) were reported by Li et al. [50] in grasslands and (1070 and 1246 g C m −2 ) Kang et al. [84] and (1115 g C m −2 ) Li et al. [50] in different forest ecosystems.
Comparatively, the nongrowing season soil carbon emissions (366, 366, 369, and 357 g C m −2 ) were lower than the growing season emission, but they were very close to each other and not much different due to similar climatic years without much varying soil temperature and precipitation between the years. A research conducted in the field has convincingly demonstrated that soil microbial activity occurs at freezing temperatures [85,86] and soil could release CO 2 when incubated in the laboratory at freezing temperatures down to −39 ∘ C [86]. Therefore, comparing measurements of soil carbon emission taken annually with those taken during growing season is important for accurately estimating annual carbon budgets, modeling the effects of climate changes on soil carbon storage and release to the atmosphere [87] and calculating below-ground carbon allocation by plants [26].
Contribution of nongrowing season (Nov.-Apr., 23 to 25%) and snow covered season (Dec.-Mar., 14 to 17%) to annual soil carbon emission of this study is well within the range (8-50%) reported in alpine ecosystems [88,89] and higher than the range (3.48-7.30%) reported in meadow grasslands, North China [69]. Snow covered season (Nov.-Feb.) soil carbon emission should be included while estimating annual soil carbon emission because soil surface continuously emits carbon during nongrowing season and even from the snow covered surface when the emission remains low and constant [90,91] and the inclusion of wintertime emission into annual carbon budgets could increase the annual carbon efflux of arctic tundra ecosystems by 17% and change some ecosystems from net annual sinks to net sources of carbon to the atmosphere [92]. Brooks et al. [93] reported that annual carbon sequestration is overestimated by 72% in deciduous forests and 111% in coniferous forests when winter soil carbon emission is not included. The one-fourth (1/4th) of our annual soil carbon emission was contributed by nongrowing season emission; therefore including nongrowing season measurements of respiration is needed to attain a more precise estimate of annual ecosystem carbon balance [47].
The annual soil carbon emissions 1448, 1500, 1459, and 1544 g C m −2 in consecutive years (Table 1) are slightly variable and the variations derived from increased/decreased growing season soil carbon emissions are owing to altering abiotic (climatic and growth) and biotic factors (root and microbial). Our annual soil carbon emissions are well within the range (698 to 1861 g C m −2 ) obtained in different land use change sites in south-east China [51]. Four consecutive years' average soil carbon emissions in growing season (1123 g C m −2 ), nongrowing season (364 g C m −2 ), and annual emissions (1488 g C m −2 ) of this study might be higher because these are quite site specific and include higher role of soil temperature and precipitation on microbial activities Journal of Ecosystems 9 [29,94]. More recently, estimates of global terrestrial soil carbon emission to the atmosphere have improved substantially, both in accuracy and number, especially in relation to different biomass [95]. The interannual variability of soil carbon emissions is very common; values are not exactly equal even in noncontrasting years resulting in that the rate of emission is controlled by abiotic and biotic environment and not accurately estimated but can be ranged after a long run of observations. Atmospheric carbon sequestered during photosynthesis is accumulated in plant biomass which is not consistent in different seasons and years. Hence, seasonal and interannual variation in plant productivity is very common in grasslands [38,96] due to variations in amount [97,98] and pattern [99,100] of annual precipitation or both [101,102]. IPCC 2007 [10] reported that total precipitation (rainfall) amounts and frequency of extreme rainfall, drought events have projected, which may have greater impacts on ecosystem carbon cycling than the singular or combined effects of rising CO 2 and temperature [103]. Higher above-ground biomass production reported in August 2008 of this study was owing to increased precipitation from March to June and constant but higher air temperature in July and August (Figure 1) that might be favorable to increase plant growth. In grassland ecosystems, variation in production at the seasonal scale (month) may be more sensitive and complex than response to climatic change at the annual scale. Therefore, Piao et al. [104] reported that the influence of climate on the vegetation activity of temperate grasslands in China varied between the growing seasons, and increased vegetation productivity was closely coupled with climate warming. The warm-season grasses grow with increasing temperature and precipitation in some limit. The variable temperature, precipitation, and grazing might be cause of comparatively lower productivity in growing season 2010. However, grazing directly affects growth in grasses and it facilitates rapid growth resulting in increase in plant biomass [105]. This is common in Z. japonica grasslands, although growth declines when grazing intensity exceeds the grasses' regrowth rate [106][107][108]. Similarly, seasonal and interannual variations of below-ground compartment which represents half of the total grassland production [109] are also critical for ecosystem carbon cycling. The seasonal variations of below-ground biomass in this study were not much higher which is consistent with the report in temperate grasslands of Oklahoma [78].
Carbon emission increases with increasing plant biomass due to higher rate of respiration from large respiratory surface ( Figure 6) and the above-ground plant biomass is the result (gross primary productivity, GPP) obtained from photosynthesis and respiration. Mechanism of exchanging carbon during photosynthesis and autotrophic (plant)/heterotrophic (microbial) respiration between atmosphere and ecosystem is a process of carbon cycling. The autotrophic respiration is limited by photosynthesis and biomass maintenance costs which is sensitive to environmental factors commonly temperature and precipitation. Rate of carbon emission either autotrophic or heterotrophic is dependent on temperature, precipitation, and growth, but more sensitive to the changing temperature. However, growth is more sensitive to precipitation than temperature in warm-season grassland; higher precipitation suppresses soil carbon emission even at high temperature due to capillary effect on soil after some hours of rainfall and snow melt. In the same way, higher soil temperature suppresses soil carbon emission by removing soil water availability and declines plant growth, and plant productivity benefit of higher GPP is offset to some extent by the loss of higher autotrophic respiration [110]. Therefore, increased above-ground biomass in August 2008 was owing to favorable higher temperature and precipitation at the time of grass growth.
The significant (fine root, 2 = 0.96, and rhizome, 2 = 0.97) temperature effect on root carbon emission (Figure 7) at pick growing season showed higher seasonal variations of root carbon emissions from below-ground surface (Figure 8). The microbial carbon emission was higher than the root carbon emission in four years, although it is difficult to measure microbial carbon emission directly; the difference between soil and root carbon emission is the microbial carbon emission and was calculated. Combining microbial and root carbon emission, large amount of carbon had been estimated which continuously released from the grassland. Contribution of root carbon emission to the total soil carbon emission ranging from 24 to 30% in this study is well within the range 15-70% ( [111][112][113]; 22-53%); higher range (10-90%) of contribution was also reported by Hanson et al. [114] and <70% was reported by Hunt et al. [12] in dry grassland. The recent reviews have suggested that contribution of autotrophic and microbial carbon emission accounts for approximately half of the total soil carbon emission [114][115][116]. However, these estimates showed considerable variation because of ecosystem diversity and potential biases in the different techniques and time scales used [28] and also with the variations in microbial carbon emissions in grazing grassland that are strongly site-specific.

Conclusion
Mechanisms of carbon cycling comprise carbon sequestration and carbon emission during photosynthesis and respiration which is expressed by plant productivity and carbon loss by root and microbial respiration. Seasonal and interannual variations in soil carbon emission were directed by soil temperature, but the role of precipitation for the variation in soil carbon emission was not established in this study. Hence, this indicated that soil temperature exerts a stronger control than soil water on soil carbon emission in grassland; however previous works have reported soil water as major control affecting soil carbon emission. Four consecutive years' average soil carbon emissions for growing season, nongrowing season, and annual emission were 1123, 364, and 1488 g C m −2 , respectively. Above-/below-ground biomass and soil carbon emission varied with variations in growing season temperature and precipitation. Root carbon emission was peaked in midgrowing season and about 1/4th of soil carbon emission was contributed by microbial or heterotrophic carbon emission. Grazing intensity directly affects above-ground biomass compared to the below-ground biomass. Our long-term estimation of carbon cycling concluded that seasonal and interannual variations are very common in carbon sequestration and emission in grassland ecosystem.