Asymmetric Diurnal and Monthly Responses of Ecosystem Carbon Fluxes to Experimental Warming

2.1 Study site

This study was conducted at the Haibei Grassland Ecological Monitoring Station of the China, Meteorological Administration in the meadow grasslands of the Xihai Town, Haiyan County, Haibei Prefecture, Qinghai Province, China (100°51′E, 36°57′N, 3140 m above sea level). Located deep in the interior of Eurasia, the study area has a typical continental plateau climate. The average annual precipitation is 398.2 mm from 1976 to 2010 and 85% of this fell during the GS from April to October. The annual average air temperature is 0.8°C, with the monthly mean air temperature ranging from −14.2°C in January to 13.4°C in July (Haibei Grassland Ecological Monitoring Station of the China Meteorological Administration). The study site has a sandy loam soil texture and is classified as “mountain brown” according to the Chinese soil classification system (cambisols in the Food and Agriculture Organization classification). The vegetation is typical meadow grassland dominated by Stipa sareptana var. krylovii, Aristida purpurea, Koeleriacristata, Cryptantha crymophila, Kobresia humilis, Artemisia scoparia, and Aster tataricus. The GS lasts from 20 April to 20 October (175 days), based on observations of phenology and plant growth traits 18, 27.

2.2 Experimental design

The study site (200 × 400 m²) was selected in 2008 and fenced to provide a relatively stable environment (Supporting Information Fig. S1). Before fencing it was freely grazed as a winter pasture, after fencing all livestock grazing was completely excluded. There were 10 m wide buffer strips on all four sides of the study site. The study site was divided into six replicate blocks (about 180 × 60 m² each), and each block was divided into two sub-blocks (about 90 × 60 m² each), one for warming treatments and one for control. In August 2010, six OTCs were randomly installed in the six warming sub-blocks. The OTCs were made of 6 mm thick solar transmitting material that was conical in shape and 40 cm in height, covering 2.05 m² at the ground (Supporting Information Fig. S1). These OTCs were modified according to the methods of the International Tundra Experiment 28 and have successfully been used in previous studies 29, 30.

2.3 Measurements

2.3.2 Plant biomass

Aboveground biomass (AGB) was measured once each month inside the aluminum frames of both the control and warming treatments using a non-destructive method in 2012 33, 34. In brief, in each control sub-block three 0.5 × 0.5 m² plots adjacent to the aluminum frame were selected in each month. Before clipping, the coverage and height were recorded by using a 0.5 × 0.5 m² quadrat (with 25 squares). The clipped aboveground plant material was oven dried at 65°C for 72 h and weighed. These data sets (90 in total) were used to construct the equations to calculate AGB

(1)

where C and H indicate coverage and height, respectively, R² = 0.947, p < 0.001). Equation 1 was used to calculate the AGB of vegetation in each test and control plot, using height and cover measurements taken by the same quadrant technique as for the clipped plots. More detailed information can be found in the literature 33, 34.

Belowground biomass (BGB) was measured by soil core (3.5 cm in diameter) from depths of 0–40 cm with six replicates both in the warming and control sub-blocks 18. Roots were first washed and then oven dried at 65°C for 72 h. AGB, BGB coverage, and height were measured in each month during the GS.

2.4 Data analysis

The whole GS was divided into three stages (start of GS [May to June], peak of GS [July to August], and end of GS [September to October]) based on the long-term observations from the eddy covariance technique and the long-term observations of phenology and meteorology 13-15, 35, 36. Diurnal and monthly mean values were calculated from the diurnal measurements for each replicate. Monthly mean values were calculated using all of the diurnal values from each of the two measurement days in each month. Repeated measure ANOVA analysis was used to examine the effects of sampling date, warming, and their interactive effects on NEE, ER, GPP, AGB, and BGB. Repeated measure ANOVA was applied both for each date measurement (eight groups of repeated measurements) and for the whole GS. For the whole GS, ecosystem carbon fluxes were calculated from the average values from each month, which were first calculated from the diurnal measurements for each replicate. Significant differences were evaluated at the level p < 0.05. Regression analysis was used to evaluate the relationships of monthly variations of soil temperature, soil moisture, and the monthly variations of NEE, ER, GPP. Multiple regression analysis was adopted to test warming-induced changes in soil temperature, soil moisture, AGB, and BGB and the corresponding changes in NEE, ER, and GPP, respectively. Warming-induced variations were calculated as the difference between the warming and control plots. The goodness of fit relative to the number of model parameters was evaluated by the Akaike information criterion (AIC) for each model. The model with the smallest AIC value was selected as the best regression model 37. Pearson correlation analysis was used to evaluate monthly variations of soil temperature, soil moisture, and Q₁₀.

Temperature sensitivity of ER was evaluated by fitting an exponential function to the data from individual treatments during the GS 12. ER was fitted by the function:

(2)

where T is the soil temperature, α is the intercept of respiration when the temperature is 0°C, and b is a constant that was used to calculate Q₁₀ by 12:

(3)

Both diurnal and monthly variations of Q₁₀ for ER were calculated. Monthly calculations were based on the diurnal measurements in each month, and diurnal calculations were based on all data from day and night.

3.1 Variations in microclimate, biotic, and abiotic factors

Precipitation in the study region falls mostly during May to October, and about 60% of the GS precipitation fell in July and August in 2012 (Fig. 1A). Monthly air temperatures co-varied with precipitation, with peak values of 13.29°C in July (Fig. 1B). The OTCs resulted in an increase in soil temperature but a decrease in soil moisture. The mean soil temperatures were 12.86 and 13.84°C for control and warming sub-blocks, respectively, whereas the mean soil moistures were 13.22 and 8.97% (Fig. 1 and Supporting Information S2). The effects of OTCs on soil temperature and soil moisture were greater during the day than during the night (Supporting Information Fig. S3).

AGB and BGB increased rapidly from early May to late July, when peak values occurred (Fig. 2). Compared with control treatments, experimental warming significantly increased AGB and BGB over the whole GS by 12.3 and 7.1%, respectively. When comparing growth during each month interval, significant responses of AGB and BGB were only observed during the peak of GS. Experimental warming significantly increased AGB by 12.7, 16.9, and 14.5%; and significantly increased BGB by 7.7, 8.8, and 6.8% in July, August, and September, respectively (Fig. 2).

3.2 Diurnal and monthly variations of NEE, ER, and GPP

The results of the repeated measure ANOVA showed that there were significant differences between warm and control plots and between different dates for NEE, ER, and GPP, while the significant interactive effects of warming and measuring date effects were observed only for NEE and ER over the entire GS (Tab. 1).

Monthly NEE, ER, and GPP all peaked in late July or early August (Fig. 3), which was in line with the monthly variations in soil temperature and soil moisture. When compared within each month, warming significantly increased NEE and GPP during the start (May to June), and peak (July to August) of the GS, while warming had no effects on NEE and GPP during the end (September to October) of the GS. In contrast, warming had no effect on ER during the start of the GS, but warming significantly increased ER during the peak and end of the GS (Tab. 1). In addition, there was more pronounced positive response of ER to warming during the night than the day (Fig. 4). Warming significantly increased ER by 17.8 and 33.6% during day and night, respectively.

3.3 Factors affecting NEE, ER, and GPP

NEE, ER, and GPP significantly increased with higher soil temperature and soil moisture (Supporting Information Fig. S4). The results of multiple regression analysis showed that warming-induced variations in NEE, ER, and GPP were closely associated with warming-induced changes in soil temperature, soil moisture, AGB, and BGB (Supporting Information Tab. S1). Biotic factors (AGB and BGB) together explained 50.4, 78.1, and 49.5% variation of NEE, ER, and GPP, respectively, while abiotic factors (soil temperature and soil moisture) accounted 42.4, 51.5, and 41.3% of variations. Biotic factors or abiotic factors together can explain more variations than any single factor for NEE, ER, and GPP.

3.4 Temperature sensitivity of ER

The values of Q₁₀ in the current study varied from 1.92 to 4.37 (Tab. 2). During the whole GS, Q₁₀ was 3.20 and 2.88 for the control and warming sub-blocks, respectively. Experimental warming significantly decreased Q₁₀ by 8.2 and 10.0% for the daytime and the whole GS, respectively, but warming significantly increased Q₁₀ by 18.2% during the nighttime. Q₁₀ was relatively higher during the start and end of GS than during the peak of GS, and during the night than during the day (Tab. 2). Monthly average Q₁₀ values were negatively correlated with soil temperature and soil moisture, in which soil temperature and soil moisture explained 42.1 and 68.9% of the monthly variations in Q₁₀ (Fig. 5).

4.1 Warming effects on microclimate and plant productivity

The OTCs in the current study resulted in a relatively warmer and drier microclimate (Fig. 1 and Supporting information S2), which was consistent with previous reports from other study sites 38, 39. The warming magnitudes by OTCs were also similar to the mean global temperature increase (0.76°C) since the Industrial Revolution 40, validating the use of OTCs to mimic realistic climate change 28. However, the OTCs were more effective at increasing daytime temperatures than night time temperatures (Supporting information Fig. S3), which has been reported in previous studies 38, 41. The difference in daytime versus nighttime temperature mediation can be explained by the OTC design, which magnifies solar warming in daylight but provides little insulation to maintain higher temperatures through the night 28. OTCs also led to significant decreases in soil moisture, which can be attributed to a combination of reduced precipitation within the OTCs due to their conical shape, the elevated temperature-induced increases in internal evapotranspiration, and the elevated evapotranspiration associated with the increased plant biomass 24.

Experimental warming significantly increased AGB and BGB over the whole GS (Fig. 2), which is consistent with recent results from the Tibetan Plateau and other regions 42-44. Vegetation growth on the Tibetan Plateau is typically constrained by the low temperatures, thus it was not surprising that the relatively warmer climate would significantly increase plant growth 45-47. Because temperature and precipitation share the same seasonal pattern, it is difficult to assess the relative influence of each on plant growth. However, the fact that the OTCs promoted AGB and BGB despite relatively lower soil moisture suggests that soil moisture may not be a key limiting factor at this range of temperature and precipitation, though the situation might be different in a drought 48, 49. Other possible explanations for the warming-induced increases in AGB and BGB might include warming-induced changes in community composition 50, prolonged GS [36], and enhanced nutrient availability 45.

4.2 Asymmetric effects of warming on ecosystem C fluxes

The responses of ecosystem C fluxes to warming may depend on timescales, since they were promoted by different phenological and ecophysiological processes 51, 52. Warming significantly increased C uptake (GPP) and sequestration (NEE) during the start but not the end of GS (Tab. 1). OTC-induced advanced green up and rapid plant development 53 have been suggested as the most likely mechanisms during that period 54-57. In contrast, warming had no effects on NEE and GPP during the end of GS (Tab. 1), even though warming has been reported to delay leaf senescence in other ecosystems 58. It is likely that photosynthetic activity was repressed by other factors during this period, such as lower soil temperature, soil moisture, soil nitrogen availability 59, or phonological patterns tied to photoperiod rather than temperature 60.

Contrary to NEE and GPP, warming significantly increased ER during the end but not the start of GS (Tab. 1). Given that respiration from microbes accounts for more than half of ER 61, 62, the warming effects on ER may be largely driven by availability of nutrients for microbes 20, 63. During the end of GS, a warming-induced higher input of fresh litter and greater amount of C transfer from aboveground to belowground enriched the soil with large amounts of available biological food for microbes 64, 65. This fresh C input would also stimulate soil organic matter decomposition through priming effects 65. These results provide clear evidence that warming had asymmetrical effects on ecosystem C fluxes during one GS, in which warming stimulated C uptake and sequestration during the start of GS, while warming enhanced C respiration during the end of GS.

Warming had more pronounced effects on ER during the night than during the day (Fig. 4). This was unexpected, since warming by OTCs had smaller effects on soil temperature during the night than during the day (Supporting Information Fig. S3). One possible explanation might be that warming during the day stimulated plant photosynthesis and C transfer to belowground, which would supply the plants and microorganisms with substrates for respiration during night 66, 67. In addition, ER was more temperature sensitive during the night than during the day (Tab. 2), meaning that even small increases in temperature would have comparatively large effects on ER during the night. This finding is particularly important in light of the fact that nights are warming faster than days 1. These results suggest that asymmetric diurnal responses of ER to warming should be taken into consideration when studying or modeling ecosystem C fluxes, particularly across regions with large diurnal temperature swings.

4.3 Warming effects on the temperature sensitivity of ER

Q₁₀ is among one of the most important parameters in the current terrestrial biosphere models, even though the variations and the factors affecting Q₁₀ are still under debate. The results show that warming decreased Q₁₀ through the entire GS; the reductions in Q₁₀ and the range of Q₁₀ in the current study are also comparable to other related studies on the Tibetan Plateau and other regions. From a mechanistic perspective, warming-induced reductions in Q₁₀ can be ascribed to decreased soil moisture 68, depletion of labile substrates 69, or acclimation of both plant and microbial respiration 70. Plants and microbes on the cold Tibetan Plateau have long been constrained by the lower temperatures, thus it is possible that they may respond positively to even small increase in temperatures, at least for the current short-term warming experiment. Therefore, warming-induced reductions in soil moisture and depletion of soil substrate would contribute the most to the observed reductions in Q₁₀, yet further studies are needed to clarify the soil moisture and substrate regulation of Q₁₀. To some extent, Q₁₀ was treated as a constant value in ESMs 71, 72, the results of which tended to show a major positive feedback loop of respiration to reinforce global warming 73. Warming-induced decreases of Q₁₀ in the current study might therefore weaken the predicted positive ecosystem C climate feedback, but this finding also hides important variability.

Relative higher Q₁₀ is observed during the edges of GS than during the peak of GS, contrasting with the monthly variations of soil temperature and soil moisture during one GS (Tab. 2). This may suggest that Q₁₀ would vary temporally and soil temperature and soil moisture may affect the monthly variations of Q₁₀. Plants and microorganisms are typically more metabolically sensitive to temperature changes at the upper or lower ends of their optimal ranges 18, 27, 74, thus it is likely that they would respond abruptly to even small changes in soil temperature and moisture. This might be one of the reasons for the large temperature sensitivity during the edges of GS. ER is primarily limited by low temperatures on the cold Tibetan Plateau, and appears to rise in a logarithmic curve once the limiting floor temperature is exceeded (Supporting Information Figs. S4 and S5). During the peak of GS, temperature is relatively high and mostly above the range of logistic variation, so Q₁₀ is relatively low compared to the colder months during the start and end of the GS. This trend was consistent with further regression analysis showing that Q₁₀ was inversely related to monthly averaged temperatures (Fig. 5). The relationship between temperature and Q₁₀ can also account for the higher Q₁₀ values observed during the night than during the day.

What then explains the positive responses of Q₁₀ to warming during the edges of GS and during the night (Tab. 2), when temperatures are relatively low? ER is driven largely by microbial activities, which is known to be relatively lower in spring and fall in part due to thermal constraint by low temperatures 61. When ambient temperature is close to the microbial metabolic low threshold, as is likely during the cold Tibetan springs, autumns, and nights, microbes would be highly sensitive to even small increases in temperature 20, 22. Experimental or climatic warming during these time periods would then allow for greater microbial activity and thus elevated ER and Q₁₀. Other studies have also reported that warming-induced elevated input of root exudates, fresh litter, and the higher amount of C transfer from aboveground to belowground would also stimulate microbial activities, leading to relatively higher Q₁₀ 64, 65, 75. While warming resulted in a lower average Q₁₀ over the entire GS, the average effect may be less informative or relevant than the relative timing of climate warming and Q₁₀ fluctuations.

4.4 Uncertainties

A previous publication showed that warming significantly enhanced NEE and GPP, but had no impacts on ER in the annual scale 61. The enhanced NEE and GPP might be largely associated with the warming-induced changes in plant functional types and plant biomass, while the non-significant response of ER resulted from the contrasting responses of its components. However, the effects of warming on ecosystem C fluxes also varied on timescales, since these processes were largely dependent on the phenological and ecophysiological stages 51, 52. Therefore, understanding the responses of ecosystem C fluxes at different timescales, such as the diurnal and monthly responses would contribute greatly to the understanding of the feedback between ecosystem C fluxes and global warming 3, 4. Because of the difficulty associated with taking the diurnal measurements on the Tibetan Plateau (ecosystem C fluxes were measured at 3 h intervals during day and night over the whole GS), the diurnal variations of ecosystem C fluxes in the current study were only measured in one GS in 2012. Long-term observations of the diurnal and monthly variations of ecosystem C fluxes would make the patterns and underlying mechanisms more clear. Although the one GS observations may partially weaken the conclusion, the results provide clear evidence that warming has uneven effects on the diurnal and monthly variations of ecosystem C fluxes, as well as on the Q₁₀. These processes are not well captured in current ESMs.

Each possible experimental warming method will have some unexpected impacts, making the control plots differ in more aspects than temperature alone 76, 77. Although the OTC is one of the most widely used warming methods, it should also be noted that there are some inevitable shortcomings for the warming experiment by OTC 61, 78, 79. First, warming by OTC may have larger effects on air temperature than soil temperature, making the plants may suffer extremely high temperatures and relatively low air humidity 61, 80. Second, OTC would reduce soil moisture due to enhanced soil evaporation and plant transpiration as also observed in the current study 81, 82. Third, OTC may also reduce the input of photosynthetically active radiation 83, but the design of OTC (material, height, and angle) in the current study closely followed the methods developed by International Tundra Experiment for grassland, and they are not large enough to affect the solar input 84, 85. Nevertheless, warming by OTC also has many advantages, particularly for the remote Tibetan Plateau regions, such as: inexpensive to operate, no supervision required, and easy to conduct and transport. This information suggests that warming by OTC may be one of the most suitable methods to study the climate warming effects in this remote meadow grassland.