Dynamics of Soil Respiration in Alpine Wetland Meadows Exposed to Different Levels of Degradation in the Qinghai-Tibet Plateau, China

The effects of degradation of alpine wetland meadow on soil respiration (Rs) and the sensitivity of Rs to temperature (Q10) were measured in the Napa Lake region of Shangri-La on the southeastern edge of the Qinghai-Tibet Plateau. Rs was measured for 24 h during each of three different stages of the growing season on four different degraded levels. The results showed: (1) peak Rs occurred at around 5:00 p.m., regardless of the degree of degradation and growing season stage, with the maximum Rs reaching 10.05 μmol·m−2·s−1 in non-degraded meadows rather than other meadows; (2) the daily mean Rs value was 7.14–7.86 μmol·m−2·s−1 during the mid growing season in non-degraded meadows, and declined by 48.4–62.6% when degradation increased to the severely degraded level; (3) Q10 ranged from 7.1–11.3 in non-degraded meadows during the mid growing season, 5.5–8.0 and 6.2–8.2 during the early and late growing seasons, respectively, and show a decline of about 50% from the non-degraded meadows to severely degraded meadows; (4) Rs was correlated significantly with soil temperature at a depth of 0–5 cm (p < 0.05) on the diurnal scale, but not at the seasonal scale; (5) significant correlations were found between Rs and soil organic carbon (SOC), between biomass and SOC, and between Q10 and Rs (p < 0.05), which indicates that biomass and SOC potentially impact Q10. The results suggest that vegetation degradation impact both Rs and Q10 significantly. Also, we speculated that Q10 of alpine wetland meadow is probable greater at the boundary region than inner region of the Qinghai-Tibet Plateau, and shoule be a more sensitive indicator in the studying of climate change in this zone.

Research indicates that atmospheric CO 2 concentrations rose from 280 ppm in 1975 to 397 ppm in 2014 1 , and will potentially rise to 500-1000 ppm by 2100 if no corrective actions are taken 2,3 . Atmospheric CO 2 concentrations are strongly influenced by carbon flux in terrestrial ecosystems, especially by soil respiration (Rs) processes, which can emit ~120 Pg of carbon to the atmosphere per year 4 . This rate is higher than carbon emissions from anthropogenic fossil fuel combustion 5,6 . In terrestrial ecosystems, the amount of carbon emitted from Rs processes is second only to the amount of carbon fixed by gross primary productivity (GPP) and is even more than the carbon uptake by net primary productivity (NPP) in certain situations [7][8][9] .
Rs is a key component of carbon flux in the global carbon cycle and a potential indicator of ecosystem metabolism 10,11 . It can also be used to estimate belowground carbon allocation 12 , and to reveal the processes and mechanisms of carbon sources and sinks on regional and global scales. More precisely, Rs can be used to predict future atmospheric CO 2 concentrations and the degree and rate of climate change 4,13 . However, due to the high temporal and spatial heterogeneity of Rs, it can only be accurately measured directly within each specific location, which makes it difficult to simulate, predict, and assess the spatial and temporal dynamics of Rs at global and regional scales 14 , and to identify how these dynamics contribute to climate change 15,16 . Therefore, quantitative field measurement for Rs in various ecosystem types is still urgently needed in research of global carbon cycle, which will contribute to reducing uncertainly when quantifying ecosystem carbon sequestration 17 .

Results
Thirty-year changes in temperature and precipitation in the study area. Figure 1 shows that the average annual temperature in the study region rose from 5.9 °C in 1981 to 7.5 °C in 2015, for a total increase of 1.6 °C. The average temperature increase between 1990 and 2000 was 0.37 °C greater than the average temperature increase between [1981][1982][1983][1984][1985][1986][1987][1988][1989][1990], and the average increase between 2000 and 2010 was 0.60 °C greater than that between 1990 and 2000, suggesting that the size of the temperature increase has grown over time. www.nature.com/scientificreports www.nature.com/scientificreports/ Precipitation presents a decreasing trend (p < 0.01) (Fig. 1). The precipitation decline mainly occurred after 2005. Average annual precipitation between 2006 and 2015 was only 542.4 mm, which is much lower than the annual averages between 1981 and 1990 (628.5 mm) and between 1990 and 2000 (696.6 mm) (Fig. 1).
Vegetation conditions in alpine wetland meadows impacted by different levels of degradation. Table 1 shows that vegetation coverage, aboveground biomass, and LAI were significantly lower in the SDM than in the NDM (p < 0.05); vegetation coverage and biomass were 50% lower and LAI was 80% lower. Vegetation coverage and aboveground biomass were about 40% and 75% lower in the SDM than the NDM. This suggests that alpine wetland meadow degradation results in a significant decrease in the condition of aboveground vegetation (p < 0.05). Table 2 shows that the carbon content of vegetation in alpine wetland meadows decreased significantly (p < 0.05) with increasing degradation. For example, carbon content was approximately 15.8% lower in the SDM than in the NDM.

SOC in alpine wetland meadows impacted by different levels of degradation.
SOC content in the 0-30 cm soil layer declined significantly between the NDM and SDM levels of degradation (p < 0.05). Specifically, SOC declined by 56.4% in the 0-10 cm soil layer, 61.2% in the 10-20 cm layer, and 64.1% in the 20-30 cm layer. But there was no significant difference in SOC content (p > 0.05) between the NDM and LDM levels in the 30-40 cm soil layer, nor among any of the degradation levels in the 40-50 cm soil layer (p > 0.05) ( Table 2).    www.nature.com/scientificreports www.nature.com/scientificreports/ the NDM. During the same stages of the growing season in the SDM, peak Rs was lower by 39.7% during the EGS, 57.6% during the MGS, and 63.1% during the LGS.

Diurnal and seasonal
Peak Rs was highest during the MGS. Peak Rs was lower, but similar, during the EGS and LGS. Peak Rs values during the MGS were about 44.0% higher than during the other two growing season stages in the NDM and about 35.0% higher in the LDM, but only 26.0% higher in the MDM and 24.2% higher in the SDM (Fig. 4). Figure  The daily mean Rs value during the MGS was approximately 40.0% higher than during the other two growing season stages in the NDM, about 28.8% in the LDM, but only 23.6% and 29.3% in the MDM and SDM, respectively (p < 0.05) (Fig. 5). Figures 6 and 7 show the power exponential curve relationship between Rs and Ts at the 0-5 cm soil depth within different levels of meadow degradation and during different stages of the growing season in 2014 and 2015. Almost all correlation coefficients (R 2 ) were above 0.5, and most of them were above 0.6. All power exponents passed the significance test (p < 0.01). Figure 8 shows the variation characteristics of Q 10 in different levels of degradation during different stages of the growing season. The Q 10 in the NDM reached 5.5, 7.1, and 6.2 during the EGS, MGS, and LGS, respectively, in 2014, but decreased from the NDM levels by more than 30%, 40%, and 50% in the LDM, MDM, and SDM, respectively. Q 10 was 8.0, 11.3, and 8.2 in the NDM during the EGS, MGS, and LGS, respectively, in 2015; these values were slightly higher than in 2014, but the variation between the two years was very similar.

Q 10 values for different levels of degradation.
The Q 10 values during the MSG were higher than during the ESG and LSG. There was no significant difference between the ESG and LSG Q 10 values. www.nature.com/scientificreports www.nature.com/scientificreports/ Correlations among vegetation biomass, SOC, Rs, and Q 10 . Table 3 shows the correlations among aboveground biomass, SOC, daily mean Rs value, and Q 10 in alpine wet meadows impacted by four different levels of degradation. Aboveground biomass correlated significantly with SOC at the 0-10 cm and 30-40 cm soil depths (p < 0.05), but not at other soil depth (p > 0.05), and with Q 10 (p < 0.05).

Discussion
The impacts of degradation on Rs in a Napa Lake alpine wetland meadow. Aboveground vegetation degradation has an important effect on Rs 54,55 . In this study, as the severity of degradation in an alpine wetland meadow increased from NDM to SDM, the Rs rate decreased significantly to about 50% of its original rate (Fig. 5). In addition, aboveground biomass, LAI, and SOC also declined significantly (Tables 1, 2).
Related studies have shown that aboveground biomass is an important modulating factor of Rs [56][57][58] . In this study, the positive correlation between Rs (in 2015) and aboveground biomass was significant (Table 3), and the positive correlation between Rs and SOC was significant (Table 2). Moreover, the SOC of the top soil layers correlated directly with aboveground biomass (Table 3). Degradation decreases aboveground biomass, which in turn reduces the activities of biological processes of roots and soil 59,60 , and SOC content was observed to decrease synchronously (Tables 2, 3). These direct and indirect effects can ultimately decrease the Rs rate.
Alpine wetland meadows that are perennially exposed to water are typical ecological systems in the Napa Lake region of Shangri-La. Pervasive and long-term disturbances from grazing have caused most of these alpine wetland meadows to become degraded. Grazing has heavily impacted Rs rates 61 by affecting soil nutrients 62,63 , Ts 64,65 , and aboveground vegetation 66,67 . Meanwhile, other study conduced in this region found that grazing did not significantly soil respiration 68 . Different findings maybe leaded by different vegetation types here. Anyhow, in recent years, frequent trampling from tourism activities in the region has become a more important factor leading to severe degradation of alpine wetland meadows. Together, these human activities have caused serious degradation of alpine wetland meadows, which has disrupted regional carbon balances.

Sensitivity of Rs to temperature, and variation of Q 10 . Ts is one of the most important factors govern-
ing Rs processes on different spatial-temporal scales [69][70][71][72][73][74][75][76] , but much uncertainty remains regarding the influence of other factors on Rs 77-80 . In our study, Rs displayed a significant exponential correlation with Ts on the scale of diurnal variation (Figs 6 and 7), but no significant correlation at the seasonal scale. In contrast, seasonal fluctuations in Rs correlated consistently with seasonal fluctuations in vegetation biomass at every level of degradation in this study (Table 1, Fig. 5). Thus, some researchers have concluded that temperature does not adequately account for all Rs variations 76,81,82 , and that vegetation is also key factor influencing Rs on a seasonal scale 83,84 .
The Q 10 value, which is the amount that Rs increases with each 10 °C rise in temperature, has commonly been used to assess the sensitivity of Rs to temperature across a variety of ecosystem types and climatic zones 72,[85][86][87] . In this study, we found that the Q 10 of Rs declined as degradation severity increased in an alpine wetland meadow system (Fig. 8), and that the Q 10 showed a significant direct correlation with SOC (p < 0.05) and with aboveground biomass (Table 3). These results, which appeared in different seasons and years, are similar to those of other works 11,88 who studied the alpine meadows of Haibei in the QTP. These results suggest that vegetation degradation can directly reduce the sensitivity of Rs to Ts in the alpine wetland meadows of Napa Lake in Shangri-La.
More and more evidence shows that Q 10 represents a combination of several influencing factors 89,90 , including biotic and abiotic factors 89,[91][92][93] . In this study, the Q 10 value was higher during the MGS than during the EGS or LGS (Fig. 8), which is consistent with the seasonal dynamics of aboveground biomass and Rs (Table 1, Fig. 5). A similar seasonal variation pattern in Q 10 was observed by 94 in their study of an alpine meadow in Haibei, QTP.
Overall, whether on a time scale of different seasons or different severities of vegetation degradation, the aboveground vegetation condition exerts a significant and decisive influence on Q 10 in this study.

Values of Rs and Q 10 based on transverse comparison with other studies in Shangri-La.
Rs is the rate of CO 2 release from the soil to the atmosphere, and Q 10 is the sensitivity of Rs to temperature changes. Comparatively high Rs rates indicate relatively high vegetation activity 54,83 , decomposition rates of soil organic matter 65 www.nature.com/scientificreports www.nature.com/scientificreports/ However, these factors, which are known to increase Rs, are sensitive to shifts in climate conditions, such as rising temperatures 13,85 . Meanwhile, Q 10 is a major source of uncertainty in assessing carbon budgets using carbon cycle models 99,100 , because differences in Q 10 among different ecosystems have been left out of many terrestrial carbon models 101,102 . Therefore, it is important to identify the Q 10 of different ecosystems.
In the current study, SOC content is higher within the southeastern boundary of the QTP than in the Haibei alpine meadow located in the inner QTP 51,88 , and a little higher than in alpine grasslands with an altitude of over 4500 m located in the Tibet 103-105 , but is lower than in the Zoigê alpine wetland located at the eastern edge of the QTP 106,107 . The Rs rate in the alpine wetland meadows in this study is roughly similar to that of degraded grassland located at the northeastern edge of the QTP 108 , but it is higher than in the Haibei alpine meadow 36,51,88,94 and the inner QTP 11,57,75,76,98 .
The Q 10 value of the alpine wetland meadow in this study is higher than in the Haibei alpine meadow and other alpine regions (range 1.3-5.6) 36,51,88,94,[109][110][111] , the Zoigê alpine wetland 106,107 , the inner QTP (range 1.05-2.81) 11,75,98 , and the global average (range 1.3-3.3) 85,112 . Many studies have suggested that Q 10 declines with increasing temperature 51,[113][114][115] . It is worth noting, however, that both the mean temperature and the Q 10 are higher in this study area than in the Haibei region mentioned above. Furthermore, Q 10 correlates significantly with SOC in our study (Table 3), but the Q 10 is higher than in the Zoigê alpine wetland because of the higher SOC content at our study site.
Together, these results suggest that Rs sensitivity to temperature is greater in alpine wetland meadow ecosystems located in the boundary region of the QTP than in other zones. Also, we speculated that Q 10 of alpine wetland meadow is probable greater at the boundary region than inner region of the Qinghai-Tibet Plateau, and should be a more sensitive indicator in the studying of climate change in this zone.

Conclusions
To the best of our knowledge, this study is the first to observe Rs on diurnal and seasonal time scales, and to quantitatively analyze Rs and Q 10 at four different levels of alpine wetland meadow degradation in the Napa Lake region of Shangri-La, at the southeastern edge of the QTP.
In summary, we found that vegetation degradation markedly altered the Rs of the alpine wetland meadow. Rs decreased by more than 50% when degradation intensity increased from NDM to SDM. On the scale of diurnal variation, Rs correlated significantly with Ts at the 0-5 cm soil depth (p < 0.05), but not at the seasonal scale. The Q 10 value of Rs decreased significantly with an increase in degradation from NDM to SDM during in every season. Rs and Q 10 were higher during the MGS than during the EGS and LGS at every level of degradation. These results indicate that vegetation condition plays an important role in controlling Rs and Q 10 .

Materials and Methods
Site description. This study was performed at the Napa Lake basin in Shangri-La County (N27°49′-27°55′, E99°37′-99°40′; mean altitude 3350 m), which lies at the southeastern edge of the QTP in northwestern Yunnan province, China (Fig. 9). Napa Lake is a typical plateau lake found on the Yungui plateau. It is situated in a graben basin in the alpine and gorge region of the Hengduan Mountains.
The study region has a cold and moist subtropical southwestern monsoon climate that is influenced by the region's high altitude and plateau landscape. Mean annual temperature is 6.4 °C, mean monthly minimum and maximum temperatures are −3.6 °C in January and 13.2 °C in July, mean annual precipitation is ~632.4 mm.
The annual range in temperature is small, but the daily range in temperature range is large. The rainy season lasts from June to October and the dry season lasts from November to May. The growing season lasts from about May to September. Soil types in the region are mainly swamp soil, peat soil, and alpine meadow soil.
Vast areas of alpine wetland meadows are distributed around Napa Lake, with dominant plant species including Blysmus sinocompressus, Carex muliensis, Poa szechuensis, Pedicularis longiflora, Kobresia bellardii, and Potentilla anserina. Villages are located far from the lakeside, while alpine meadows and farmland planted with Hordeum vulgare are close to the lakeside. As altitude increases, vegetation succeeds gradually from hard-leaf evergreen and broad-leaved forests, to alpine shrubs, to alpine pine forest, and to spruce and fir forests. Dominant species include Crataegus oresbia, Populus rotundifolia, Sabina squamata, Pinus densata, Picea asperata Mast., and Abies forrestii.

Plot surveys and Rs measurements. Study plots.
We classified alpine wetland meadows within the study area into four levels of degradation based on the presence of fencing, grazing activity, tourism disturbance, and aboveground biomass and vegetation cover: non-degraded meadow (NDM), lightly-degraded meadow (LDM), moderately-degraded meadow (MDM), and severely-degraded meadow (SDM) ( Table 1).
Each of the four levels of alpine wetland meadow degradation severity contained three study plots (100 m × 100 m). All plots are located adjacent to a lake, but are exposed to water year-round, and do not experience periodic flooding. By correlating the degree of degradation in the plots with vegetation cover, biomass, and species composition, we can determine the effects of grazing and tourism on the meadows.
Vegetation surveys and soil organic carbon (SOC) measurements. To further characterize and verify the degree of degradation in the different plots, we sampled aboveground biomass, vegetation cover, and species composition within one randomly-placed 1 m × 1 m frame on every plot every season. The leaf area index (LAI) within the sampling frames was determined using a plant canopy analyzer (LI-COR LAI-2200 Plant Canopy Analyzer, Li-Cor, Lincoln, Nebraska, USA).
Plant samples from each sampling frame were dried in an oven at 65  www.nature.com/scientificreports www.nature.com/scientificreports/ One soil profiles with a depth of 50 cm were collected from each plot during the experimental period, and the SOC content of the soil at different depths (0-10 cm, 10-20 cm, 20-30 cm, 30-40 cm, and 40-50 cm) was analyzed for every degradation level.
Rs measurement. Rs was measured in each plot using an automated CO 2 efflux system (Li-8100, LI-COR Inc., Lincoln, NE, USA) 57,68,75 in May (early growing season, EGS), July (mid growing season, MGS), and September (late growing season, LGS) over the course of the full growing season (May to September) in 2014 and 2015.
CO 2 measurements were collected from each of the four plots for a period of 24 h using a Li-8100 automated soil CO 2 flux system with a No. 103 chamber (Li-Cor Inc., Lincoln, NE, USA) to determine diurnal Rs during the growing season in the two years of the study. During the measurements, all chambers were placed on collars with a 20 cm inside diameter and a 10 cm height. The collars had been inserted 5 cm into the soil at least three days prior to measurement. Aboveground vegetation was clipped from the soil surface inside the collars before measuring Rs. All collars were left at the plots during the entire experimental period. Rs include respiration from plant roots and microbes.
Diurnal variations in Rs were recorded automatically every half hour from 7:00 am on the first day to 7:00 am on the next day. The duration of each automatic measurement was about 3 min, which included 15 s dead band, 45 s pre-purge, 45 s post-purge, and 90 s observation. The linear increase in CO 2 concentration within the chamber was used to estimate Rs. We simultaneously measured Ts and soil moisture at 5 cm soil depth near each collar using the temperature and moisture sensors of the Li-8100 System. Statistical analysis. An exponential equation 69 was used to describe the relationship between Rs and Ts: = Rs ae (1) bTs where, Rs is soil respiration rate (μmol·m −2 ·s −1 ), Ts is soil temperature (°C) at 5 cm depth, and a and b are fitted parameters. The sensitivity of Rs to Ts can be defined as the increase in the Rs rate that results from each 10 °C increase in Ts. This sensitivity (Q 10 ) can be calculated as follows: 10 10b www.nature.com/scientificreports www.nature.com/scientificreports/ One-way ANOVA was used to compare vegetation condition, SOC, and Rs at different levels of degradation. Exponential regression was used to evaluate the relationship between Rs and Ts in every plot. Linear regression was used to correlate vegetation, SOC, Rs, and Q 10 . All statistical analyses were performed using SPSS 13.0 software (SPSS for Windows, Chicago, IL, USA). Differences were considered significant when p < 0.05.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.