SITE DESCRIPTION

The study site is situated in the source region of the Yangtze River and in the middle of the QTP (Fig.1, part A, 92°56′E, 34°49′N) with mean altitude of 4635 m and typical alpine climate (Fig.1, part B). Mean annual temperature is -3.8 °C with a minimum of -27.9 °C in January and a maximum of 19.2 °C in July. Mean annual precipitation is 290.9 mm with 95% falling from May to October. Mean potential annual evaporation is 1316.9 mm, mean annual relative humidity is 57%, and mean annual wind velocity is 4.1 m s^-1 (Lu et al., 2006). The study site is a winter-grazed range, dominated by alpine meadow species: Kobresia capillifolia, Kobresia pygmaea, Carex moorcroftii, with mean plant height of 5–10 cm. Plant roots are mainly at 0–20 cm soil depth with an average soil organic C of 1.5%. Soil development is weak and belongs to the alpine meadow soil (soil taxonomy in China, and cryosols in World Reference Base taxonomy) with a mattic epipedon at approximately 0–10 cm depth and organic-rich layer at the depth of 20–30 cm (Wang et al., 2007). The parent soil material is of fluvial-glacial origin and sand (>0.05 mm) contents reach to 95%. Permafrost thickness near the experimental site is 60–200 m and the active layer is 2.0–3.2 m (Pang et al., 2009), which has been increased at the rate of 3.1 cm yr^-1 from 1995 to 2000 (Wu and Liu, 2004). The experimental field was on a mountain slope with a mean inclination of 5°. Detailed information about soil properties in the 0–20 cm layer and plant features in dry and wet conditions are in Table 1. Species composition was similar but with higher coverage and plant height in wet than in dry conditions. Average elevation difference between plots in wet and dry conditions was about 1 m (Fig. 1).

EXPERIMENTAL DESIGN AND MEASUREMENT PROTOCOL

MEASUREMENT PROTOCOL

A polyvinyl chloride (PVC) collar (80 cm² in area and 5 cm in depth) was permanently inserted 2–3 cm into soil at the center of each plot for measuring R_s. Small living plants were removed at the soil surface at least one day before the R_s measurement to eliminate the effect of above-ground biomass respiration (Zhou et al., 2007). A deep PVC tube (80 cm² in area and 50 cm in depth) was inserted into the soil in each plot near the shallow collar in July 2010 for measuring R_h. The deep PVC tube cuts off old plant roots and prevents new roots from growing inside the tube. Carbon dioxide efflux measured above deep tubes was used to represent R_h after three months of deep collar insertion. The R_a value was calculated as the difference between R_s and R_h, which was measured once or twice a month between 10:00 and 15:00 hours (local time), using a Li-COR 6400 portable photosynthesis system attached to a soil CO₂ flux chamber (Li-COR, Lincoln, Nebraska, U.S.A.).

Soil temperature was monitored by thermo-probes (Model 109, Campbell Scientific, Utah, U.S.A.) installed at 5.0 cm depth in the center of each plot. Volumetric SWC (v v%^-1) was measured based on frequency domain reflectometry (FDR; Sentek Pty, Stepney, Australia) at 0–10, 10–20, 20–40, 40–60, and 60–100 cm depths in each plot. The daily average data of soil temperature and moisture were recorded in a CR 1000 data logger (Campbell Scientific, U.S.A.).

Above-ground biomass (AGB) in each plot was obtained indirectly from a step-wised multiple linear regression (AGB = 22.76 × plant height + 308.26 × Coverage - 121.80, R² = 0.74, P < 0.01, n = 100). In each month of the growing season, we took eight repeated measurements of plant coverage and forty measurements of plant height in each plot. The average plant coverage and height were substituted into the linear function to get the monthly AGB data of each plot. Root biomass (RB) was obtained from soil samples at 0–10, 10–20, 20–30, 30–40, and 40–50 cm depths. The soil samples were air-dried for one week and passed through a 2-mm-diameter sieve to remove large particles. Procedure of separation of roots from soil and the separation of living roots from dead roots can be found in Yang et al., 2009a.

DATA ANALYSIS

Soil respiration and its components were fitted exponentially and linearly with soil temperature. Determinant coefficient was used to examine which model was better for deriving their temperature sensitivity (Q₁₀). If the linear fitting performed better than did exponential model, Q₁₀ would be obtained by multiplying the slope of linear fitting with 10.

Daily soil temperature and SWC data were used and analyzed with a two-way ANOVA analysis. The measured R_s, R_h, R_a, AGB, and RB in each replicate were averaged to get the monthly data, and the monthly data were analyzed by variance analysis of split-plot design in SPSS 16.0 when investigating the effects of warming, soil moisture condition, and their interaction on these parameters.

Fourteen measurements of R_s and R_h, and the calculated R_a were averaged. Mean R_s, R_a, and R_h were multiplied by growing season length to estimate the total C release. AGB and RB change in growing season was the sum of difference in five months between control and warming plots (May–September).

MICROCLIMATE

Mean soil temperature was higher in dry (0.32 °C) than in wet sites (-0.16 °C; Fig. 2, part A) in control plots, which significantly increased by warming in both conditions (P < 0.01; Fig. 2, part A). The increasing magnitude had no difference between the dry and wet sites (1.75 and 1.76 °C in wet and dry). Soil water content was higher in wet than in dry sites in control plots (Fig. 2, part B), which significantly decreased by 0.89% and 2.0% in 0–10 cm but increased by 2.62% and 2.99% (v v%^-1) in 60–100 cm layer in wet and dry conditions (Fig. 2, part B), respectively.

FIGURE 1.

(A) Location, (B) warming equipment installation, and (C)plots distribution of the experimental site.

TABLE 1

Soil pH, CaCO₃ (g kg^-2), soil organic carbon (SOC, g kg^-2), total nitrogen (TN, g kg^-2), inorganic nitrogen (IN, mg kg^-2), and bulk density (BD, g cm^-3) in the 0–20 cm, and plant coverage (Cov., %) and height (H, cm) in the dry and wet conditions

RELATIONSHIP OF SOIL RESPIRATION AND ITS COMPONENTS WITH ABIOTIC FACTORS

Soil temperature was positively correlated with R_s and its components in wet conditions, whereas it only correlated with R_s and R_h in dry conditions (Fig. 3, parts A, C, and E). Q₁₀ of R_s was higher in wet than in dry conditions (4.67 vs. 3.55), but that of R_h was higher in dry than in wet conditions (2.87 vs. 3.44). No obvious relationship was observed between SWC and R_s and its components in wet conditions but a positive correlation occurred between SWC and R_s and R_a in dry conditions (Fig. 3, parts B, D, and F).

RESPONSES OF SOIL RESPIRATION AND ITS COMPONENTS TO WARMING

Soil respiration and its components were higher in summer and lower in winter (Fig. 4). Soil respiration and R_h had no significant differences between the two sites (Table 2), while R_a was marginally higher in wet than in dry conditions (Table 2). Warming significantly increased R_s and R_h but had no effect on R_a (Table 2). On average R_s and R_h increased by 9.9% and 12.7% in both wet and dry conditions. The interaction between warming and SWC had a significant effect on R_s and R_a but had no effect on R_h (Fig. 5; Table 2). Both R_s and R_a decreased in dry but increased in wet condition. On average R_s and R_a decreased by 5.8% and 36.3% in dry but increased by 23.5% and 47.7% in wet. The estimated increase in soil C emission in growing season was 0.65 kg m^-2 in wet condition, of which R_h increase contributed to 0.33 kg C m^-2. The estimated reduction in soil C emission was 0.16 kg C m^-2 in dry condition, among which R_a decrease accounted for 0.17 kg C m^-2.

Contribution of R_h to R_s (R_h:R_s) was significantly different between dry and wet (Table 2). On average R_h:R_s was 76% ± 3% and 63% ± 3% in dry and wet, respectively. Interaction between warming and SWC significantly affected the R_h/R_s (Table 2). Warming increased R_h:R_s (70% ± 4% vs. 82% ± 4% in control and warming, respectively) in dry but decreased it in wet condition (67% ± 4% vs. 60% ± 4% in control and warming).

FIGURE 2.

(A) Mean soil temperature (°C, 5 cm) and (B) soil moisture (v/v%, 0–10 and 60–100 cmb) from 1 October 2010 to 18 July 2013 in dry and wet conditions. Different labels above columns represent significant difference between control and warming. (P < 0.05).

FIGURE 3.

Relationship of soil respiration (R_s µ mol m^-2 s^-1) and its components (R_a and R_h) with soil temperature (°C, 5 cm) and soil moisture (v/v%, 10 cm) in wet (filled circles and solid lines) and dry conditions (open circles and dashed lines).

RESPONSES OF ABOVE-GROUND AND ROOT BIOMASS TO WARMING

Soil water content and warming had no significant effects on AGB and RB (Table 2). However, their interaction significantly affected AGB. On average, warming increased monthly average AGB by 0.02 kg m^-2 in wet but decreased it by 0.02 kg m^-2 in dry conditions (Fig. 6). The cumulative AGB increase was 0.1 kg m^-2 in wet and the reduction was 0.1 kg m^-2 in dry condition in growing season.

The seasonal variation in Rs and its components were largely (87%, 75%, and 18%) explained by AGB (Fig. 7). R_s and R_h were more sensitive to AGB than was R_a (Fig. 7).

RESPONSES OF R_S AND ITS COMPONENTS TO WARMING

In most field experiments, warming manipulation increases R_s (Rustad et al., 2001; Wu et al., 2011), but the negative warming effect on R_s is also reported (Saleska et al., 1999). We found that warming effects on R_s depend on SWC, which was stimulated in wet but suppressed in dry condition (Fig. 5). The opposite responses of R_s in our study are consistent with results from a semi-arid steppe ecosystem (Mauritz and Lipson, 2013) and an old-field ecosystem dominated by grasses and forbs (Suseela and Dukes, 2013). In a semi-arid area, R_s and its components are inhibited when SWC is less than 10% or more than 15% (Mauritz and Lipson, 2013). The positive relationship between R_s and SWC in dry conditions (Fig. 3, part B) implies the SWC limitation on R_s in dry conditions, therefore R_s reduction is likely the result of lower SWC in dry sites (6.8% in control and decreases under warming). Nevertheless, results in our study are also different with Arctic tundra ecosystem, in which ecosystem respiration is suppressed in wet and moist but stimulated in dry tundra (Oberbauer et al., 2007). Soil water content is generally higher than the field capacity and there is even flowing surface water in the growing season in tundra ecosystem (Shaver et al., 1998). Lower water table resulting from warming would lead to saturation stress for ecosystem respiration in wet tundra (Oechel et al., 1998). Saturation stress does not exist in our study because SWC is lower than the field capacity even in the wet site and this led to the R_s increase.

FIGURE 4.

Seasonal dynamic of soil respiration (R_s) and its components (R_a and R_h) in control and warming plots in dry and wet conditions. Data are averaged from replicates, n = 14. Dates are given in the form m/dd/yyyy.

TABLE 2

Results (F-values) of split-plot variance analysis on the effects of warming, soil moisture condition, and their interaction on soil respiration (R_s) and its components (R_h and R_a), aboveground biomass (AGB), and root biomass (RB).

FIGURE 5.

Seasonal and average (inserted panels, mean ± SE, n = 13) warming effects on (A) soil respiration (R_s), (B) heterotrophic respiration (R_h), and (C) autotrophic respiration (R_a) in wet (dark columns) and dry (white columns) conditions. Asterisks denote significance between warming and control treatments (** P < 0.01; * P < 0.05).

Autotrophic respiration and R_h could both positively or negatively (Schindlbacher et al., 2009; Zhou et al., 2007) or conversely respond to warming (Li et al., 2013). The consistent positive in wet and negative responses in dry of R_a and R_h (Fig. 5) in our study could be caused by the corresponding changes in SWC and biomass. Positive correlation between R_h and soil temperature but no obvious relationship between SWC and R_h either in dry or in wet (Fig. 3, part D) indicates that soil temperature is more important than SWC in regulating soil C decomposition in alpine meadow. R_h is affected by total detritus input or AGB (Bond-Lamberty et al., 2004), and R_a is determined by RB (Zhou et al., 2007) and temperature dependence of specific root respiration (Boone et al., 1998). Possible elevated detritus input and the associated AGB increase in wet (Fig. 6, part A) therefore would be responsible for the R_h increase because of the positive relationship between AGB and R_h (Fig. 7). The effect of AGB decrease on R_h could be compensated by the increase in RB, thus resulting in non-significant change in R_h in dry condition. R_a increase in wet is likely due to the soil temperature increase because of the positive correlation between them (Fig. 3, part E). Soil moisture deficit constrains R_a through limitation on annual productivity (Li et al., 2013), root growth (Zhou et al., 2007), and specific R_a rate (Michele and Douglas, 2009). Non-significant change in RB under warming in dry condition suggests the change in specific R_a rate. We have no direct data on the specific Ra, but the species composition change in our study site (Xu et al., 2014) might change the RB quality, thus resulting in the specific R_a change under warming.

FIGURE 6.

Warming effects on above-ground biomass (AGB) and root biomass (RB, 0–50 cm soil depth) in wet and dry conditions. Values are the average of four replicates in dry and six replicates in wet from May to September 2012. Different labels above columns indicate the significant difference at the P < 0.05 level.

FIGURE 7.

Relationship of soil respiration (R_s µ mol m^-2 s^-1) and its components (R_a and R_h) with above-ground biomass (AGB) in all plots.

The response of R_a to warmer soil can affect ecosystem C allocation and the strength of positive feedback of soil CO₂ efflux to climate warming. Both R_h and R_a increased with larger relative increase in R_a in wet (53% vs. 33% in R_a and R_h), which suggests that the effect of warming-induced change in biomass on R_s is larger than direct temperature effect on R_s.

Higher Q₁₀ of R_h in dry than in wet conditions is similar to the result in an incubation study where Q₁₀ of R_h decreases as SWC increases (Guntiñas et al., 2013). However, higher Q₁₀ of R_s in wet than in dry is observed in a desert shrub ecosystem (Wang et al., 2013) and in our study. The higher Q₁₀ of R_h in dry may be due to the switch in C pool of labile substrate to recalcitrant substrates (Reichstein et al., 2002) because recalcitrant C is more sensitive to temperature change than is labile C (Knorr et al., 2005). Root always exerts a strong influence on Q₁₀ of R_s (Boone et al., 1998; Zhou et al., 2007). The lower Q₁₀ of R_s in dry is probably the result of lower RB biomass (Fig. 6, part B). The contrast performance of Q₁₀ of R_s and R_h in wet and dry in our study suggests that R_a has a large effect on apparent Q₁₀ of R_s and indicates apparent Q₁₀ obtained in field studies without exclusion of R_a may overestimate the warming effect on soil C output.

BIOMASS RESPONSES TO WARMING

In an experimental warming study covering an elevation gradient from 4300–4600 m in alpine meadow, AGB significantly decreased at the 4300 m study site, but had no significant change in the 4600 m site (Fu et al., 2013). Optimum air temperature (T_a) for AGB is about 5.8 °C (Wang et al., 2012), and AGB increases with increasing water availability in the alpine meadow of Tibet (Yang et al., 2009b). Annual mean T_a in the present study is -3.8 °C, which is much lower than the optimum temperature. The elevated temperature is probably the reason for AGB increase in wet condition. The minimum SWC for meadow growth is found to be 11.8% (Ma et al., 2004), which is much higher than the mean SWC in dry condition. Although soil temperature significantly increased in dry condition (Fig. 2, part B), lower SWC and its decrease could result in the AGB decrease (Fig. 6, part A), which suggests that AGB is more sensitive to SWC than temperature change when SWC is lower than a threshold.

IMPLICATION FOR THE FEEDBACK OF ECOSYSTEM CARBON BALANCE TO WARMING

The growing season AGB increase in wet (0.1 kg m^-2) and decrease in dry (0.1 kg m^-2) is much lower than the soil C release change under warming (0.65 kg m^-2 in wet and -0.16 kg m^-2 in dry). The balance between AGB and R_s is 0.55 and -0.06 kg C m^-2 in wet and dry conditions, respectively. The results suggest net C release in alpine meadow under warming climate when SWC maintains greater than 10%. Decrease in R_s especially in R_a could make the alpine meadow ecosystem serve as a C sink since reduction in R_s is higher than the decline in C uptake when SWC is less than 10%.