Characterization of CO₂ flux in three Kobresia meadows differing in dominant species

Abstract

INTRODUCTION

Terrestrial ecosystems of the northern hemisphere are considered to be net sinks for atmospheric CO₂ (Ciais et al. 1995; Fan et al. 1998; Janssens et al. 2003; Pacala et al. 2001). However, the extent of spatial variation in sink strength remains uncertain (Pacala et al. 2001; Schimel et al. 2001). The sink strength differs in different vegetations or biomes. Almost all models dealing with terrestrial CO₂ budget have, thus, focused on the vegetation type and/or biomes (e.g. Cramer et al. 2001; Ito and Oikawa 2002). However, even within the same vegetation, plant communities with different dominant species may vary in CO₂ sink strength. The variation of CO₂ fluxes at the plant community level has received less attention so far. Characterizing the spatial variation of CO₂ sink strength at plant community level would be needed if we were to accurately estimate and precisely predict carbon fluxes in terrestrial ecosystem.

To quantify sink strength, much attention has been focused on forests, such as those in North America (Barford et al. 2001; Chen et al. 1999; Law et al. 2001) and Eurasia (Kolari et al. 2004; Kowalski et al. 2004; Schulze et al. 1999; Valentini et al. 2000). Unlike in forest ecosystems, however, CO₂ exchange in grasslands has received relatively less attention if we take into account the fact that grasslands cover ∼40.5% of the Earth's terrestrial area, excluding permanent ice cover (Adams et al. 1990; White et al. 2000). CO₂ fluxes have been observed in some grasslands (e.g. Flanagan et al. 2002; Frank and Karn 2003; Ham and Knapp 1998; Hunt et al. 2004; Meyers 2001; Novick et al. 2004; Sims and Bradford 2001; Suyker et al. 2003; Xu and Baldocchi 2004). These studies show that grassland can act as either a sink or a source of atmospheric CO₂. Most of these studies focused on prairies and pastures and there is insufficient knowledge concerning with ecosystem CO₂ exchange in alpine meadows such as those on the vast Qinghai-Tibetan Plateau (Cao et al. 2004; Gu et al. 2003; Kato et al. 2006).

The Qinghai-Tibetan Plateau has one of the most extensive alpine meadows in the world. The alpine meadow covering above 50 × 10⁴ km on the plateau has been considered identical in carbon sink strength in most global and regional models. From a large-scale survey, we found, however, a high spatial heterogeneity of soil organic carbon within the alpine meadow (Yang et al. 2008). The spatial heterogeneity of soil organic carbon may be partly ascribed to heterogeneity in habitat, which often results in different plant communities. However, little information is currently available for our understanding how carbon sink strength differs in different plant communities.

Kobresia meadow is the major grassland vegetation on the Qinghai-Tibetan Plateau (Wang et al. 2006; Yang et al. 2008). Within the Kobresia meadow, three plant communities dominated by Kobresia pygmaea, Kobresia humilis and Kobresia tibetica are most widely distributed (Miehe et al. 2008; Wang et al. 2006, 2008; Zhang et al. 2009). The three communities often coexist within a local region but have distinct habitat segregation in soil moisture. Kobresia pygmaea occurs at driest but K. tibetica at wettest habitat. The carbon sink strength may differ among the three communities because of the difference in their plant species composition and soil water content. We were to test the hypothesis by (i) examining the seasonal variations of CO₂ fluxes and (ii) assessing the major biotic and abiotic factors controlling the CO₂ fluxes for the three communities of alpine meadow.

MATERIALS AND METHODS

Study site

The study site is located in a large valley plain within the Qilian Mountains on the northeast of Qinghai-Tibetan Plateau (lat 37°36′N, long 101°18′E; 3 250 m above sea level). The annual average temperature and precipitation for 1981–2000 were −1.7°C and 561 mm. At the study site, there are three common types of Kobresia meadow; they are dominated by K. pygmaea, K. humilis or K. tibetica.

In this study, we designed two experiments that (i) line-transect survey for examining the spatial distribution and its limiting factor of three alpine meadows and (ii) fence survey for examining the CO₂ flux and its controlling factors in the three alpine meadows.

Line-transect survey.

We set three 350-m parallel line transects, which started from a hill top and ended near to a river and were 10 m apart from each other. Vegetation and soil were observed along the three line transects between 9 and 15 August 2005 when the AGB reached its maximum.

A quadrat of 20 × 20 cm was placed at 10-m intervals along each of the line transects. At the center of each quadrat, we measured the soil volumetric water content with time-domain reflectometry (TDR) soil moisture sensor (CS620; Campbell Scientific, Thuringowa, Queensland, Australia). AGB was measured within a 21-cm diameter circle area. The cut aboveground parts were then dried to a constant weight at 80°C.

Fence survey.

We fenced a plot of 10 × 10 m in each of the three Kobresia meadow types in August 2004. In each plot, five circular collar bases were buried at ∼5-cm depth in the soil at 2-m intervals along a line transect randomly determined.

To keep the vegetation similar to that of the winter grazing grasslands in the area, all three meadows were grazed during winter, at which time, the livestock removed almost all the AGB. The carbon of the biomass removed by the livestock was assumed to all be released into the atmosphere as CO₂ from the respiration of the livestock and from the decomposition of their excrement. We ignored the carbon that might have been stored in the livestock because in this grazing system, the livestock do not increase in weight during the winter. Other factors such as respiration from insects and mammals other than livestock were ignored in the data analysis.

AGB measurement

To assess the AGB, we set six quadrats of 25 × 25 cm randomly within each plot. We then measured the coverage, mean height and maximum height of the vegetation. The aboveground part was cut every 2 or 3 weeks during the growing season in 2006. The cut parts were then dried to a constant weight at 80°C for the AGB.

CO₂ flux and parameter measurements

The CO₂ flux (net ecosystem exchange [NEE]) measurement system consisted of a chamber body, a polyvinyl chloride (PVC) collar, a data acquisition unit, a fan for mixing air and a power supply. We had two types of chamber: a transparent (acrylic) chamber for ecosystem photosynthesis and a dark (PVC) chamber for ecosystem respiration (ER). Within the transparent chamber, we installed a cooling unit to keep the within-chamber temperature close to the environmental temperature.

The transparent chamber was a cylinder 30 cm high and 30 cm in diameter. During measurement, the chamber was mounted on the rim on the upper side of the buried collar. The collars had a diameter of 30 cm and a height of 7 cm; they were installed in the soil at 5-cm depth in August 2004. The cooling unit included an electronic cooler (FTA951; Ferrotec Corp., Tokyo, Japan) with a proportional–integral–derivative controller (TTM1520; Toho Electronics Inc., Tokyo, Japan). The air temperature within the chamber could be controlled within a range of ±2.0°C of the outside air temperature during flux measurement. However, the temperature within the dark chamber used for ER measurement showed less change during measurement without the cooling unit. The data acquisition unit consisted of a CO₂ probe (GMP343; Vaisala, Helsinki, Finland) installed at the center of the top of the chamber to monitor CO₂ concentration, a photon sensor (IKS-27; Koito Industries, Kanagawa, Japan) for measuring photosynthetic photon flux density (PPFD) within the chamber and two copper-constantan thermocouples for air temperature inside and outside the chamber. The CO₂ probe was calibrated by the manufactory of Vaisala before our use in the experiment. All electric signals from these sensors were recorded by a data logger (Thermic Model 2300A; EtoDenki Ltd, Tokyo, Japan). The power supply unit consisted of car batteries.

We measured NEE and the environmental parameters 1 day once every 2 or 3 weeks during the growing season and once in early March 2006. During each day, we first measured NEE and then ER for five points at 1-h intervals from 7:00 to 22:00 (local time). Once mounting the chamber on the collar, we recorded CO₂, temperature and PPFD within the chamber at 1-s intervals. At the end, we mounted a dark chamber to measure ER. Each measurement (measurement of NEE or ER) for one point took ∼3 min. During the measurements of NEE and ER, we measured the soil moisture from 0- to 12-cm depth with TDR soil moisture sensor (CS620; Campbell Scientific), soil temperatures at 5- and 10-cm depths with platinum resistance thermometers (IT-2100; As One Corporation, Osaka, Japan) and the soil surface temperature with an infrared thermometer (IT-420; As One Corporation). We obtained one measurement of NEE or ER by averaging the data from five points and then a regression curve by fitting these measurements for 1 day. The regression curves for each period were then used for further data processing and simulation as follows.

Data processing

NEP (μmol CO₂ m⁻² s⁻¹) as CO₂ uptake (i.e. net ecosystem photosynthesis) and ER (μmol CO₂ m⁻² s⁻¹) were calculated by the following equation:

(1)

where S is the linear slope of the CO₂ concentration (μmol CO₂ mol⁻¹ air) within the chamber against time, V is the chamber volume (m³), D is air density (mol m⁻³) and A is the ground surface area (m⁻²) within the chamber. The slope of the CO₂ concentration was obtained by fitting the linear model C = St + b to the sampling data during a period of 60 s from 30 to 90 s after the chamber had been mounted on the collar. In the linear model, C is the instantaneous CO₂ concentration at time t and b is the intercept.

Gross primary production (GPP) was calculated as the sum of NEP and ER.

Model simulation of carbon fluxes

We used the following two models (Equation 2; Falge et al. 2001, and Equation 3; Lloyd and Taylor 1994) to estimate the ecosystem carbon budget for the three meadows.

(2)

where α is the initial slope of the PPFD versus GPP curve (μmol CO₂ [μmol photon]⁻¹) and can be considered as the quantum yield of ecosystem photosynthesis. GPP_max is the GPP at light saturation (μmol m⁻² s⁻¹).

(3)

where ER is the ecosystem respiration rate (μmol CO₂ m⁻² s⁻¹), R_ref is the ecosystem respiration rate (μmol CO₂ m⁻² s⁻¹) at the reference temperature T_ref (K) and Ea is the activation energy, a parameter determining the temperature sensitivity of ER (J mol⁻¹). These latter two parameters are site specific. R_k is the universal gas constant (8.134 J K⁻¹ mol⁻¹) and T_soil is the soil temperature at 5-cm depth. T_ref was chosen as R₁₀, the respiration rate at T_ref equal to 283.16 K (10°C) and evaluated for every month. Ea was evaluated from the regression of all ER data points against T_soil as a constant value throughout the whole year.

The parameters of Equations (2) and (3) were assumed to be site specific for the three Kobresia meadows and used to obtain the GPP, ER and NEP as follows (Table 1). We first obtained the half-hourly GPP for each of the three meadows by substituting half-hourly measured data of PPFD observed from the micrometeorological tower in a K. humilis meadow into Equation (2). We then estimated ER for each site from Equation (3) by substituting the soil temperature at 5-cm depth observed by TidBit temperature sensor (Onset Computer Corp., Bourne, MA) for the respective meadows during the 2006. Half-hourly NEP at the three Kobresia meadows was calculated by the following equation:

(4)

Half-hourly estimated values of GPP, NEP and ER values were integrated daily, monthly and annually.

We estimated the daily GPP, ER and NEP for each period, usually 2 or 3 weeks, by using the measured parameters for that period during the growing season from day of year (DOY) of 121 to DOY of 277. We used the parameters from the measurements of early March in 2006 to estimate winter GPP, ER and NEP from DOY of 278 to DOY of 120 the next year.

RESULTS

AGB and soil water content

The patterns of seasonal variation in AGB were similar among the three Kobresia meadows (Fig. 1). AGB started to increase from May, peaked in August and declined to near to zero in late September (Fig. 1). The maximum AGB differed substantially among the three meadows, increasing in the order 365, 402 and 434 g dry weight m⁻² in K. pygmaea, K. humilis and K. tibetica meadow. The mean AGB was significantly higher in K. tibetica meadow than the other two meadows (Table 2).

Figure 1:

seasonal variations in AGB in Kobresia pygmaea, Kobresia humilis and Kobresia tibetica meadows, 2006.

Open in new tabDownload slide

We observed the AGB in relation to soil water content in the three Kobresia meadows in August 2005 (Fig. 2). Kobresia pygmaea appeared only in dry soil environments where the soil water content ranged from 18 to 40% and its biomass decreased with increasing soil water content. Kobresia humilis was widely distributed over a soil water content of 18 to ∼70%, but the AGB showed little correlation with soil water content. Kobresia tibetica became the dominant species when the soil water content was higher than 60%; its AGB increased with soil water content.

Figure 2:

relationship between soil water content and AGB in the Kobresia pygmaea, Kobresia humilis and Kobresia tibetica meadows, measured in August 2005.

Open in new tabDownload slide

Ecosystem photosynthesis and ER

To facilitate across-site comparisons of GPP and ER, we standardized the measured CO₂ fluxes in relation to the controlling factors. Gross ecosystem photosynthesis (GEP) increased with increasing PPFD and showed no apparent light saturation under the observed PPFD (Fig. 3). In the K. pygmaea, K. humilis and K. tibetica meadows, the standardization model of Equation (2) explained 87, 81 and 79%, respectively, of the total variation in GEP associated with PPFD. The initial slope of the fitted curve (α), indicating the light utilization efficiency of canopy photosynthesis was substantially higher in K. humilis meadow than in the other two meadows, which had almost the same α as each other. The light-saturated GEP (GEP_max) estimated from the model increased in the order of K. pygmaea, K. humilis and K. tibetica meadow (Table 1).

Figure 3:

relationships between GEP and PPFD in the Kobresia pygmaea, Kobresia humilis and Kobresia tibetica meadows. Each data point represents one temporal sampling for a mean of five chambers. A temporal sampling was taken at 1-h intervals during 1 day every 2 or 3 weeks in the growing season, 2006. The model equation is GEP = αGEP_max×PPFD/(GEP_max+ αPPFD), and the estimated parameters are α= 0.0246, GEP_max= 22.2, R²= 0.87, P<0.001, for K. pygmaea; α= 0.0335, GEP_max= 29.9, R²= 0.81, P<0.001, for K. humilis; and α= 0.0223, GEP_max= 37.8, R²= 0.79, P<0.001, for K. tibetica.

Open in new tabDownload slide

ER was strongly associated with change of soil temperature (T_soil) at 5-cm depth in all three meadows (Fig. 4). The fitted model for ER, Equation (3), showed that soil temperature accounted for 69, 74 and 80% of the variation in ER in the K. pygmaea, K. humilis and K. tibetica meadows, respectively. The base rate of ER at 10°C (R_ref) was site specific, ranging from 3.38 μmol m⁻² s⁻¹ (K. pygmaea meadow) to 5.81 μmol m⁻² s⁻¹ (K. tibetica meadow). The activation energy (Ea) was the lowest in the K. pygmaea meadow and highest in the K. tibetica meadow.

Figure 4:

relationships between ER and soil temperature at 5-cm depth (T_soil) in the Kobresia pygmaea, Kobresia humilis and Kobresia tibetica meadows. Each data point represents one temporal sampling for a mean of five chambers. A temporal sampling was taken at 1-h intervals during 1 day every 2 or 3 weeks in the growing season and one time in early March 2006. The fitted model is ER = R_refexp[(Ea/R_k) (1/283.16 –1/T_soil)] with R_ref= 3.38, Ea = 54 691, R²= 0.69, P<0.01, for the K. pygmaea meadow; R_ref= 4.99, Ea = 63 729, R²= 0.74, P<0.01, for the K. humilis meadow; and R_ref= 5.81, Ea = 64 716, R²= 0.80, P<0.01, for the K. tibetica meadow.

Open in new tabDownload slide

The measured GPP and ER for the three meadows were significantly different among the three meadows (Table 2). The K. pygmaea meadow showed always significantly lowest fluxes, including GEP and NEP among all the meadows.

Estimated GPP, ER and NEP

Seasonal variations in GPP and NEP were obtained by using the measured PPFD, soil temperature and parameters shown in Table 1 and Fig. 5. The estimated GPP and NEP in all three Kobresia meadows started to increase from May, peaked in August and declined to close to zero in September (Fig. 6). Annual GPP differed significantly among the three meadows and increased in the order of the K. pygmaea, K. humilis and K. tibetica meadows, at 798, 1 064 and 1 158 g m⁻² year⁻¹, respectively (Table 2). The estimated NEP (= GPP – ER) was positive (CO₂ net sink) when GPP exceeded ER, whereas it was negative (CO₂ source) when ER exceeded GPP. A positive NEP occurred only during the growing season, from May to August (Fig. 6). Annual NEP differed among the three meadows at 74, 164 and 128 g C m⁻² year⁻¹ for the K. pygmaea, K. humilis and K. tibetica meadows, respectively (Table 2). Overall, the three Kobresia meadows were assessed as sinks of CO₂ in 2006.

Figure 5:

seasonal variations in PPFD, air temperature (T_air) observed from the micrometeorological tower in a Kobresia humilis meadow and soil temperature at 5-cm depth observed by TidBit temperature sensor in the Kobresia pygmaea, K. humilis and Kobresia tibetica meadows during 2006.

Open in new tabDownload slide

Figure 6:

seasonal variations in GPP and NEP estimated in the three Kobresia meadows: Kp, Kobresia pygmaea; Kh, Kobresia humilis; Kt, Kobresia tibetica.

Open in new tabDownload slide

DISCUSSION

Variations of CO₂ sink strength and the potential mechanisms

Our study showed a distinct difference of CO₂ sink strength among the three alpine meadows with different dominant species. The result indicated that the measured mean daily NEP was ∼1.5 times lower in the K. pygmaea meadow (Table 2). Since the three meadows always distribute within the same climate conditions, almost all the models dealing with carbon sinks considered Kobresia meadow as an identical vegetation type and, therefore, would have ignored the differences among the communities. This study suggests that such difference among meadows should be taken into account in future model estimation and/or prediction of ecosystem carbon budget if we are to achieve an accurate and precise result.

Our current data could not provide a detailed assessment for the mechanism controlling the sink strength revealed in the study. However, two factors, the AGB and soil water content, might potentially contribute to the NEP difference among the three meadows. First, the lowest AGB may contribute the lowest NEP in the K. pygmaea. This is because the AGB in the Kobresia meadows was almost all resulted from the leaves, i.e. a high AGB means almost always a high leaf area index in these meadows (Kato et al. 2004). The leaf area index is the major determinant of GPP (e.g. Reichstein et al. 2007). When ER variation is small, a high GPP would result in a high NEP. In addition to the amount of AGB, the biological activity including photosynthesis and respiration rate may play an important role to the difference of NEP among the three meadows. The ratio of GEP_max to AGB (GEP_max/AGB) in August was 0.0311, 0.0604 and 0.0398 in the K. pygmaea, K. humilis and K. tibetica meadow, respectively. This result indicates that K. pygmaea might have lower photosynthetic activity and/or a high proportion of leaf biomass in AGB as compared with the other two meadows. On the other hand, the ratios of annual ER to soil carbon content (ER/SC) were 0.0144, 0.0175 and 0.0185, in K. pygmaea, K. humilis and K. tibetica meadow, respectively, indicting also a lower soil respiration activity per unit soil carbon in the K. pygmaea meadow.

Second, the low soil water content in the K. pygmaea might also limit NEP observed in the study. Soil moisture content differed markedly among the three Kobresia meadows that K. pygmaea appeared only in dry soil environments, where the soil water content ranged from 18 to 40%, while K. tibetica appeared in the wettest soil environment such higher than 60% (Fig. 2). On the Qinghai-Tibetan plateau, soil moisture is the major determinant of spatial variation in soil organic carbon content (Yang et al. 2008). Our current study indicated that, under similar precipitation conditions, a high soil water content may result in a high carbon sink strength. The importance of soil hydrological properties in regulating grassland carbon dynamics has been stressed in many studies (e.g. Li et al. 2004; Luo et al. 2008). In comparison with NEP, it seems that GPP and ER have even stronger correlation with soil water contents, as suggested by the evidences from different ecosystems (Fig. 7).

Figure 7:

(a) relationship between annual precipitation and carbon fluxes (GPP, ER and NEP) in different grasslands of the world. (b) Relationship between soil water content and carbon fluxes (GPP, ER and NEP) in different grasslands of the world (data for Kobresia tibetica were not included). All data and their references are shown in Table 3. The values of this study were showed by the blacked circle or square or triangle.

Open in new tabDownload slide

Despite the large variation of NEP, all the meadows in the study showed a net CO₂ sink (Figs 6 and 7). The annual CO₂ sink strength in grassland ecosystems in general showed much higher variability than forest ecosystems. In the Netherlands, Jacobs et al. (2003) found a net C uptake of 680 g C m⁻² year⁻¹ in a grassland dominated by Loliumperenne and Poa trivialis. Schapendonk et al. (1998) reported an NEP of 300 g C m⁻² year⁻¹. Miscanthus sinensis grassland, however, was assessed to be CO₂ resource −56 g C m⁻² in 2000 and −100 gC m⁻² in 2001 (Yazaki et al. 2004). Long-term observation is needed to determine the annual variability of the sink strength found in the study in the future.

Temporal variation of ecosystem CO₂ fluxes

We examined the seasonal dynamics of CO₂ fluxes and AGB. The seasonal pattern was similar among the three meadows (Fig. 1), which is due to the controlling factors including PPFD; temperatures and water availability should follow the same seasonal pattern.

In terms of temporal variability of the ecosystem CO₂ fluxes, PPFD explained most diurnal and seasonal variations in the ecosystem photosynthetic rate in the three alpine meadows (Fig. 3 and Table 1), while temperature explained most variations of ER in the three meadows (Fig. 4). The temperature effect on ER was well described by an exponential function in each meadow of the study, as has already been frequently reported (e.g. Fang and Moncrieff 2001; Lloyd and Taylor 1994). Temporal variation in ER is often determined by the soil temperature and moisture content (Franzluebbers et al. 2002; Leiros et al. 1999; Orchard and Cook 1983).

Comparison of GPP, ER and NEP between chamber and micrometeorological measurements

Annual GPP and ER based on the chamber measurements were much higher than (about double) those reported by Kato et al. (2006) using the eddy covariance method in similar K. humilis-dominated meadow vegetation, but the annual NEP was similar (Tables 2 and 3). The higher ER may largely due to the sampling interval in the current study.

We were not able to measure ER during the winter. We therefore estimated the winter ER by assuming that the flux should be constant during the whole winter and the value would be the same as that we obtained from the measurements in the early March 2006 (Table 1 and Fig. 6). For the K. humilis meadow, the eddy covariance method estimated ER as 6.4, 15.6 and 19.4 g C m⁻² for 2002, 2003 and 2004, respectively during the winter period from November to March of the next year (Kato et al. 2006). For the same period, we estimated ER as 38.1 g C m⁻² for the same meadow. However, the NEP estimates were similar: Kato et al. (2006) estimated NEP as −22.0, −17.6, −5.2 g C m⁻² for 2002, 2003 and 2004, respectively, while our estimate was −11.1 g C m⁻² for the same period. This was due to that our estimate for GPP was also ∼2 times of that from the eddy covariance method. In conclusion, our currently simulation seems to result in an overestimation in the winter time ER but the winter NEP is not likely to result in any significant bias in the estimation of the annual NEP because of a similar overestimation existed in the estimated GPP during the winter period.

Yearly variation in climatic factors—including radiation, temperature and precipitation—may have contributed to this difference. Differences in estimates of ecosystem CO₂ flux among different measuring approaches have been frequently reported and discussed but no widely acceptable conclusion has yet been reached. Wohlfahrt et al. (2005) observed that the agreement among several approaches, including eddy covariance, ecosystem chamber and scaled-up leaf and soil chamber measurements, in quantifying ER was within 35% (=the level of uncertainty) in mountain grasslands. Studies of forest ecosystems indicate that respiration estimated from chamber measurements is higher than that estimated by the eddy covariance method (Drewitt et al. 2002; Goulden et al. 1996; Griffis et al. 2004; Lavigne et al. 1997; Law et al. 1999). However, in a Norway spruce forest, there was good agreement between short-term comparisons of respiration made by chamber and eddy covariance methods (Wallin et al. 2001).

FUNDING

Integrated Study for Terrestrial Carbon Management of Asia in the 21st Century Based on Scientific Advancements; Early Detection and Prediction of Climate Warming Based on the Long-Term Monitoring of Alpine Ecosystems on the Tibetan Plateau.

The authors thank Drs Xiaoyong Cui, Song Gu, and Yingnian Li for their valuable comments and assistance in our study.

Conflict of interest statement. None declared.

References