Ethics statement

Our experiments were associated with the HiWATER-MUSOEXE. We used seven sites to conduct our study, which encompassed a vegetable field, orchard, wetland, Gobi, sandy desert, desert steppe and a maize cropland. No specific permissions were required to our sites. The studied areas did not involve endangered or protected species. The specific locations of our study are as follows: Vegetable field (100.35816 E, 38.89322 N), Orchard (100.36974 E, 38.84512 N), Maize cropland (100.37225 E, 38.85557 N), Gobi (100.30420 E, 38.91496 N), Sandy desert (100.49330 E, 38.78917 N), Desert steppe (100.31860 E, 38.76519 N), Wetland (100.44640 E, 38.97514 N).

Sites description

The study area, Zhangye oasis irrigation area, is located in Gansu province, northwest China, which is in the middle of the Hexi Corridor (Fig. 1A). The Heihe River runs through the entire territory, forming a unique desert oasis (Fig. 1A). The annual mean air temperature is approximately 6°C, and the warmest month is July. The annual mean precipitation is approximately 114.9 mm; more than 70% of this precipitation is accumulated during the growing season from June to September. The annual frost-free period lasts approximately 169 days [14].

The Zhangye oasis irrigation area is surrounded by sandy desert, wetland, Gobi and desert steppe ecosystems and is a typical irrigated agricultural area [13]. Fig. 1A shows the location of the focus area for this study. Seven sites, with orchard, vegetable field, wetland, desert steppe, Gobi, sandy desert, and maize surfaces, were included in this study. Fig. 1B shows the surfaces at different sites. Each site is fairly open and flat and the fetch is approximately 100 m- 200 m in all directions. The prevailing wind direction is northwesterly. The area of the orchard (EC location) is approximately 53 hm², and consists mainly of planted apple trees. Chilies are the main vegetable in the vegetable field, which has an area of 21.6 hm². Phragmites australis are the dominant species in the wetland, which has a total area of nearly 1733 hm². The dominant vegetation in the desert steppe is Salsola passerina, while in the Gobi, the vegetation is sparse and dominated by Compositae and Salsola passerina. In contrast, there is slightly more vegetation in the sandy desert ecosystem, which is dominated by drought-tolerant vegetation, such as Peganum harmala and Tribulus terrestris. Maize is the main crop in the irrigation area. The superstation is located in a maize cropland field with an area of 200 hm². During the studied timeframe, irrigation occurred once a month in the orchard, vegetable field and maize cropland.

Measurements and data acquisition

The fluxes of CO₂ were continuously measured using EC systems. The data used in the study spanned from June 10 in 2012 to September 14 in 2012, except for in the wetland where the data spanned from June 26 in 2012 to September 14 in 2012. The EC system used consists of a three-dimensional sonic anemometer (CSAT-3/Gill, Campbell Scientific Instruments Inc., USA/Gill, UK) and an open-path infrared gas analyzer (LI-7500A, Licor Inc., USA), which measure the three components of wind speed, sonic virtual temperature, and the concentrations of water vapor and CO₂. The data were recorded with a data logger (CR1000/CR3000/CR5000, Campbell Scientific Instruments Inc.) and CF card at a frequency of 10 Hz. Air temperature and relative humidity (HMP45C, Vaisala Inc., Helsinki, Finland) were measured 5 and 10 m above the ground. Photosynthetically Active Radiation (PAR) (LI-190SA, LI-COR Inc.) was measured above the canopy at a height of 12 m. Rainfall was measured using a tipping bucket rain gauge (TE525MM, Campbell Scientific Instruments Inc.). The soil temperature (Campbell-107, Campbell Scientific Instruments Inc.) and volumetric soil water content (TRIMEEZ/IT, IMKO, Ettlingen, Germany) were measured at 0.02, 0.04, 0.10, 0.20, 0.40, 0.60 and 1.0 m depths in all plots. In the analysis, we use the soil temperature and water content measurements from a depth of 0.04 m and the air temperature and relative humidity from a height of 5 m. Because of the different vegetation conditions and instrument types, the heights of the EC systems are not the same among the different ecosystems. The heights for the EC systems were 7 m, 2.85 m, 3.8 m, 4.6 m, 4.6 m, 5.2 m, and 4.5 m in the orchard, desert steppe, vegetable field, Gobi, sandy desert, wetland, and maize cropland ecosystems, respectively.

Data processing and gap filling

Post-processing calculations were performed using the Edire software package (http://www.geos.ed.ac.uk/abs/research/micromet/EdiRe) and included outlier exclusion, time delay corrections, density fluctuations, spectral loss, secondary coordinate rotation and sonic virtual temperature conversion [15–18].

Due to the atmospheric stability of the weather and the physical constraints of the impact apparatus, the data retained implausible points after this processing procedure. Therefore, we used the following procedure to further process the data.

(1) Removed the data from one hour before and after rainy periods [19]. (2) Excluded the negative data collected during the night [20, 21]. (3) Removed the data that were outside the instruments’ measurement range or outside the reasonable range of values [20, 21]. The range for the different ecosystems was not the same. (4) Removed the data for which the friction velocity (u) was less than the threshold friction velocity in night [20, 21]. The threshold friction velocity is different between the ecosystems, with 0.15 m·s⁻¹ for the orchard, 0.14 m·s⁻¹ for the maize cropland, 0.13 m·s⁻¹ for the wetland and sandy desert, 0.10 m·s⁻¹ for the vegetable field, Gobi and steppe desert.

NEE is calculated from the sum of the EC flux (F_c) and the storage term (F_s). F_s is the flux associated with the CO₂ stored in the layer below the level in which CO₂ flux measurements are taken. The value of F_s is obtained by integrating the change in CO₂ concentration in the air layer up to the height of the EC [22]. Because the flux measurement systems were not high at the studied sites, the storage term F_s was usually smaller than F_c and the daily values tended to be zero, so F_s is neglected in the calculation of NEE. GPP is the difference between the estimated ecosystem respiration (R_eco) and observed NEE: (1) GPP (mg CO₂·m⁻²·s⁻¹) represents CO₂ assimilated by photosynthesis and R_eco (mg CO₂·m⁻²·s⁻¹) consists of respiratory CO₂ released from the soil and vegetation.

Nighttime NEE values are equal to R_eco because the GPP equals zero at night. Daytime estimates of R_eco were obtained from the nighttime NEE-temperature relationship. The relationship between nighttime NEE and air temperature can be described by the Vant Hoff equation (Eq. 2) [23], (2) where a and b are regression parameters and T is the air temperature (°C). The temperature sensitivity coefficient (Q₁₀) was determined from the following equation: (3)

Additionally, missing nighttime data were also estimated by an exponential regression, using Eq. 2, of the measured nighttime R_eco and air temperature. In this study, we used the air temperature as the fitting parameter because it yielded better results (higher R² values) than the soil temperature.

There are five basic models used for describing the partial dependence of the NEE on the PAR [24]. The Michaelis–Menten rectangular hyperbola model (Eq. 4) [25, 26] is typically used. (4) where PAR (μmol· m⁻²·s⁻¹) is the incident photosynthetically active radiation, P_max (mg CO₂· m⁻²·s⁻¹) is the maximum CO₂ flux (with infinite light) and K_m is the level of PAR at which NEE is one half of P_max. R_eco (mg CO₂· m⁻²·s⁻¹) is the average daytime respiration. The apparent quantum yield (α) can be calculated as follows [27]: (5)

The missing or bad data for the daytime were gap-filled using the ‘look-up’ table method, which is based on the empirical relationships created for each site between NEE, air temperature and incident PAR when meteorological data are available [19]. Previous studies have shown that the daily change of CO₂ flux is associated to the temperature, soil moisture and vegetation phenology [28]. To make the gap-fill more rational, we interpolated the missing data by the month in the orchard, wetland, and vegetable field ecosystems and by the growth stage in the maize cropland. Four different growth stages were defined according to the maize phenology: seeding stage (June 10–19), jointing stage (June 20-July 20), filling stage (July 21-August 5) and mature stage (August 6-September 15). Considering that seven leaves were growing at the beginning of our observation, it was understood that the seeding stage in our study did not include the prior period.

Diurnal variations in NEE

Based on the observations from the study period (i.e., June 10—September 14, and, in wetland, June 26—September 14), the diurnal variations of NEE were calculated for different ecosystems. Fig. 2 shows the diurnal variations of NEE for different ecosystems, which were calculated using the average flux during the observation period (positive values represent CO₂ emissions; negative values represent CO₂ absorption). As seen in Fig. 2, the diurnal variations in NEE from the different ecosystems show a single peak. The values of NEE are negative in the daytime and positive at nighttime, and the minimum value occurs around noon. The negative daytime values of NEE indicate that CO₂ assimilation via photosynthesis is greater than CO₂ release from the respiratory processes of ecosystems. During the nighttime, however, CO₂ is released to the atmosphere via respiratory processes in the soil and plants, and the NEE values are then positive.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Diurnal variations of NEE in different ecosystems.

https://doi.org/10.1371/journal.pone.0120660.g002

However, there were differences among the ecosystems. CO₂ absorption and emission in sandy desert, desert steppe, and Gobi ecosystems were very weak, and the values were close to zero, meaning the diurnal variations of NEE were not significant. Conversely, the diurnal variations of NEE were significant in the vegetable field, orchard, maize cropland, and wetland ecosystems. There was an extremely small CO₂ flux in the sandy desert, desert steppe and Gobi ecosystems; the maximum net CO₂ absorptions were 0.081, 0.036, 0.038 mg CO₂·m⁻²·s⁻¹, respectively. Carbon absorption during the daytime was similar in desert steppe and Gobi, but the nighttime carbon emission was higher in the desert steppe than in the Gobi. This is because the Gobi surface contains more gravel than the desert steppe, resulting in relatively low soil respiration at night. There was a relatively large fluctuation in the maize cropland, orchard, wetland and vegetable field ecosystems; the minimum values were −1.15, −0.71, −0.64 and −0.46 mg CO₂·m⁻²·s⁻¹, respectively. The peak CO₂ uptake for the maize cropland was approximately 1.6 times the value observed in the orchard, 1.8 times the wetland value and 2.5 times the vegetable field value.

Daily variation in GPP, R_eco and NEE

Values of NEE, R_eco, and GPP for different ecosystems over the observation period are shown in Fig. 3. Because sparse and drought-tolerant vegetation dominate the sandy desert, desert steppe and Gobi ecosystems, the values of NEE, GPP, and R_eco were much lower in these ecosystems than in other ecosystems. Meanwhile, daily NEE, GPP, and R_eco variations were not evident. However, there were significant daily variations in NEE, GPP, and R_eco in the vegetable field, orchard, maize cropland, and wetland ecosystems, which are closely related to the growth and phenology of the vegetation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Daily variations in NEE, R_eco and GPP in different ecosystems.

https://doi.org/10.1371/journal.pone.0120660.g003

The minimum daily NEE values were −3.11, −2.40, and −3.82 g CO₂·m⁻²·d⁻¹ in the desert steppe, Gobi and sandy desert ecosystems, respectively, and occurred in July in these three ecosystems (Table 1). The maximum values of R_eco were 2.90, 1.14, and 3.37 g CO₂·m⁻²·d⁻¹, while the maximum values of GPP were 5.41, 3.41, and 5.68 g CO₂·m⁻²·d⁻¹ in the desert steppe, Gobi and sandy desert ecosystems, respectively (Table 1). The total NEE, R_eco and GPP during the observation period in the desert steppe, Gobi and sandy desert ecosystems are shown in Table 2. The total NEE were −67.9, −129.6 and −17.8 g CO₂·m² in the Gobi, sandy desert and desert steppe, respectively. Thus, these three ecosystems acted as carbon sinks during the observation period. The GPP/R_eco ratio in these three ecosystems is greater than 1 (Table 2), which also illustrates their carbon sink nature. These results are consistent with previous studies because many studies also found that the desert ecosystem is a carbon sink and the value of total NEE in our study is within the observed range from previous studies. For example, Liu et al. [29] found that the saline desert ecosystem in western China was a significant carbon sink and the values of NEE were −236.2 g CO₂·m⁻², −63.1 g CO₂·m⁻², and −92.0 g CO₂·m⁻² during the growing seasons of 2004, 2005, and 2006 respectively. Wohlfahrt et al. [30] measured the Mojave Desert ecosystem in the USA and found that it was a CO₂ sink, taking up 102±67 and 110±70 g C·m⁻² in 2005 and 2006, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Maximum GPP, R_eco, and Minimum NEE for different ecosystems during the observation period (June 10—September 14, and, in wetland, June 26—September 14).

https://doi.org/10.1371/journal.pone.0120660.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Total GPP, R_eco, and NEE for different ecosystems during the observation period (June 10—September 14, and, in wetland, June 26—September 14)

https://doi.org/10.1371/journal.pone.0120660.t002

In comparison, the fluctuations of NEE in the maize cropland, orchard, wetland, and vegetable field ecosystems were large during the observation period (Fig. 3). On rainy days, the ecosystems become sources of CO₂ because the NEE is primarily composed of soil and plant respiration (Fig. 3). The NEE values were negative throughout the observation period in the orchard, maize cropland and wetland ecosystems. However, at the beginning of the observation, NEE was positive in vegetable field, but gradually decreased and became slightly negative by the end of June. This is likely explained by the fact that the chilies in the vegetable field had been recently transplanted and grew very slowly at the beginning of our observation. The minimum daily NEE was observed in July, at which point the values were −45.3, −28.4, −26.8 and −19.2 g CO₂·m⁻²·d⁻¹ in the maize cropland, orchard, wetland and vegetable field ecosystems, respectively (Table 1). The maximum GPP values were 70.8, 59.8, 41.7, 41.8 g CO₂·m⁻²·d⁻¹ and the maximum R_eco values were 27.5, 32.3, 19.4, 28.7 g CO₂·m⁻²·d⁻¹ in the maize cropland, orchard, wetland and vegetable field ecosystems, respectively (Table 1). The total NEE, R_eco and GPP for the maize cropland, orchard, wetland, and vegetable field ecosystems throughout the entire observation period are shown in Table 2. During the observation period, the GPP/R_eco ratios were higher than 1 (Table 2), which implies that NEE is dominated by GPP and that PAR is a dominant factor controlling NEE. During the observation period, the total NEE for the maize cropland, orchard, wetland and vegetable field were −2342.6, −1050.3, −1085.1, and −468.1 g CO₂·m², respectively (Table 2). Obviously, these ecosystems are characterized by CO₂ sinks. The carbon absorption capacities of various ecosystems are significantly different. The integrated order of the carbon absorption capacity in each ecosystem is: maize cropland > wetland > orchard > vegetable field> sandy desert> Gobi> desert steppe.

The maximum net CO₂ uptake reported from our site was compared with other ecosystems. There are major differences regarding the daily maximum net CO₂ uptake among different crops. In our study, CO₂ uptake in the maize cropland was slight greater than 38.47 g CO₂·m⁻²·d⁻¹, the value measured in an oasis field maize ecosystem in Xinjiang by Zhao [31], and was similar to the uptake in a maize ecosystem in the North China Plain (45.8 g CO₂·m⁻²·d⁻¹) [32]. However, the value observed in this study is lower than that obtained by Baker and Griffis [33] in a maize/soybean ecosystem in the USA (51.3 g CO₂·m⁻²·d⁻¹) and is also lower than the value (67.8 g CO₂·m⁻²·d⁻¹) measured by Hollinger in a maize and soybean rotation agriculture ecosystem in the USA [34]. Precipitation may be one of the reasons for the carbon uptake difference. The annual mean precipitation of the two sites in the USA is more than 570 mm, which is significantly higher than in our studied area. Meanwhile, different maize species and tillage practices may also result in differences in the CO₂ flux.

There is less research being conducted regarding CO₂ flux in vegetable fields. The net CO₂ uptake in the vegetable field is higher than in wheat (15.62 g CO₂·m⁻²·d⁻¹) and cotton ecosystems (8.96 g CO₂·m⁻²·d⁻¹) in Xinjiang according to Zhao [31], but lower than in the wheat ecosystem (30.0 g CO₂·m⁻²·d⁻¹) in the North China Plain [32]. The daily maximum net CO₂ uptake in the apple orchard in our study is similar to that from a vineyard ecosystem (28.9 g CO₂·m⁻²·d⁻¹) in northwestern China [35]. However, the value is higher than in a broad-leaved Korean pine forest (22.3 g CO₂·m⁻²·d⁻¹) and a Dahurian larch forest (15.8 g CO₂·m⁻²·d⁻¹) in northeast China [36]. This is because the environmental conditions in artificial ecosystems are better than those in a natural ecosystem.

The daily maximum net CO₂ absorption in the wetland was higher than 14.3 g CO₂·m⁻²·d⁻¹ measured in an alpine wetland ecosystem on the Qinghai-Tibetan Plateau [37]. The value is within the observed range from wetlands on Chongming Island of Shanghai (18.3 g CO₂·m⁻²·d⁻¹–44.0 g CO₂·m⁻²·d⁻¹) [38]. During the observation period, the wetland was a strong CO₂ sink, and the total NEE was −1085.1 g CO₂·m⁻². Liao et al. [39] reported that the carbon absorption capacities of Spartina alterniflora, Phragmites australis, and Scirpus mariqueter were 5096, 3410 and 770 g CO₂·yr⁻¹, respectively. Clearly, the CO₂ absorption capacity of Phragmites australis is very high.

Response of R_eco to temperature

The exponential function given in Eq. 2 is used to describe the relationship between R_eco and air temperature. Table 3 shows the relationship between nighttime R_eco and air temperature in the various ecosystems. Overall, nighttime R_eco increases exponentially with the air temperature. The relationship between R_eco and the air temperature varies across ecosystems and month-to-month (i.e., growth stages) in an ecosystem. Overall, the values of the coefficient of determination (R²) were not high in all of the ecosystems. This is caused by the uncertainties in the measured nighttime CO₂ fluxes. The nighttime CO₂ flux data are not very effective or reliable because of the low turbulence and stable atmospheric stratification [20, 21]. Compared to other ecosystems, the R² values were lower in sandy desert, desert steppe, and Gobi ecosystems. There was little variations of the nighttime R_eco because commonly few vegetation living in these ecosystems and also because of the low moisture and organic content. In addition the night temperature changed rapidly in large gradient. All the above reasons are coupled to properly explain why the fitting of R_eco and temperature is not satisfactory occasionally. For the vegetable field, orchard and wetland ecosystems, the results fit best in September and most poorly in July. For the maize cropland, the best fit occurred during the filling stage and the poorest fit occurred during jointing. This shows that the effect of temperature on R_eco in the strong growth month or stage is higher than other months or stages.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. The relationship between nighttime ecosystem respiration and air temperature for different ecosystems.

https://doi.org/10.1371/journal.pone.0120660.t003

The temperature sensitivities of the ecosystem (Q₁₀) also have different values. Using Eq. 3, the Q₁₀ values were estimated to be 1.26, 1.54, and 1.77 in the desert steppe, Gobi, and sandy desert ecosystems, respectively, during the observation period. The Q₁₀ value ranged from 1.40 to 2.05 for the vegetable field, 1.38 to 1.77 for the orchard, 1.49 to 1.55 for the wetland and 1.48 to 2.05 for the maize cropland ecosystem. These values are similar to the Q₁₀ value (1.5) based on air temperature at a global scale (excluding the wetlands) [40]. High Q₁₀ values are observed in June for the vegetable field and orchard and in August for the wetland. Low Q₁₀ values occurred in July for the vegetable field and wetland and in September for the orchard. For the maize cropland, high Q₁₀ values were observed in the seedling stage, and low Q₁₀ values in the jointing stage.

Thus, responses of R_eco to the environment vary depending on the ecosystems [41–43]. R_eco is regulated by numerous components, such as the respiration in the stem [44, 45], branch [46], root [47, 48] and foliage [46], microorganisms [49], decomposition of litter [50, 51], soil organic carbon [47, 48] and coarse wood debris [52]. Most researchers divide R_eco into respiration from the vegetation and from the soil [32, 53]. Given the low vegetation coverage in the desert, desert steppe and Gobi ecosystems, soil respiration dominated these ecosystems’ respiration. When vegetation coverage is relatively low in June (or the seedling stage), soil respiration plays an important role in R_eco and leads to high Q₁₀ values. Low Q₁₀ values occur in July and during the jointing stage when vegetation grows vigorously, which increases the proportion of R_eco derived from vegetation. In the orchard ecosystem, little change was observed in the apple trees during the observation period, and low Q₁₀ values occurred in September with the apple harvest.

Response of NEE to PAR

PAR is the most important factor for regulating daytime ecosystem CO₂ exchanges during the growing season [53]. We used Eq. 4 to describe daytime NEE in response to PAR. The data were analyzed by month in the orchard, wetland, and vegetable field ecosystems and by growth stage in the maize cropland.

Fig. 4 shows the relationship between daytime NEE and PAR in the vegetable field, orchard, wetland and maize cropland ecosystems. This figure shows that the CO₂ flux gradually decreases with increases in PAR. Furthermore, the CO₂ flux decreases linearly with low solar radiation (PAR<500 μmol·m⁻²·s⁻¹) in all months or growth stages in all ecosystems; the rate of decrease gradually decreases until the CO₂ flux stabilizes. The light saturation point varies with different vegetation. When PAR is greater than 1000 μmol·m⁻²·s⁻¹, the CO₂ flux became more stable in the vegetable field and wetland ecosystems. The PAR value was greater than 1600 μmol·m⁻²·s⁻¹ and the carbon absorption rate remained stable in the orchard. However, as PAR values increased toward their maximum of 2000 μmol·m⁻²·s⁻¹, carbon assimilation continued to increase in the maize cropland.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Response of NEE to PAR in different ecosystems.

https://doi.org/10.1371/journal.pone.0120660.g004

Table 4 shows the daily variation of the light response parameters α, P_max and R_eco in different ecosystems. In general, the seasonal patterns of these parameters closely followed the vegetation phenology. Additionally, there were some differences between the light response parameters of the different species. Previous studies found that P_max first increased and then decreased following the processes of crop growth, development and senescence [32]. We find similar results in our experiment. During the observation period, P_max shows seasonal trends in parallel with the vegetable growth. The value increased with the growth of leaves and reached a maximum in July; it then began to decrease as fruit ripening or foliage turned brown in the vegetable field, wetland, and orchard ecosystems. However, P_max continuously increased during the observation period in the maize cropland. This is because the maize was in the growth stage during the entire observation period. The P_max values ranged from 0.337 to 0.991 mg CO₂·m⁻²·s⁻¹ for the vegetable field; 0.977 to 1.584 mg CO₂·m⁻²·s⁻¹ for the orchard; 0.877 to 1.282 mg CO₂·m⁻²·s⁻¹ in the wetland; and 1.121 to 4.357 mg CO₂·m⁻²·s⁻¹ in the maize ecosystem (Table 4). These results clearly demonstrate that the peak P_max varies across ecosystems. The peak P_max was highest in maize cropland, followed in decreasing order by the orchard, wetland, and vegetable field ecosystems.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Seasonal dynamics of light-response parameters for different ecosystems.

https://doi.org/10.1371/journal.pone.0120660.t004

The apparent quantum yield (α), calculated via Eq. 5, is a basic parameter of photosynthetic CO₂ absorption, light utilization and material productivity. In general, the α value changes with the growth stage for all vegetation types, and a higher α value is presented during the months or stages in which plants grow rapidly. During the observation period, α first increased, reached a maximum in July or August for all vegetation types, and then decreased.

The α reported from our site is comparable to that found in other studies. The α value for the vegetable field ranged from approximately 0.0010 to 0.0024 mg CO₂·μmol⁻¹ and is similar to the value for dry croplands, which is between approximately 0.0009 and 0.0022 mg CO₂·μmol⁻¹ [54]. Flanagan et al. [55] reported α for a beech forest ranging from 0.0009 to 0.0013 mg CO₂·μmol⁻¹; this value is lower than that reported for the orchard in this study. Adequate water and fertilizer conditions may have led to the high α value. Zhao et al. [37] found the α value of an alpine wetland to be 0.0056 μmol·m⁻²·s⁻¹ and 0.0082 μmol·m⁻²·s⁻¹ in July and August, respectively. The value of the α of the wetland in this study is significantly higher than the value of the alpine wetland reported by Zhao et al., which could be due to the difference in hydrothermal conditions. The annual mean air temperature recorded at the alpine wetland (−1.7°C) is lower than in our study. Regarding the maize cropland ecosystem, the α value ranged from 0.0010 to 0.0022 mg CO₂·m⁻²·s⁻¹ and is similar to the value reported in previous studies. Suyker et al. [25] reported an α value of maize ranging from 0.02 to 0.04 μmol·m⁻²·s⁻¹(0.0009–0.0018 mg CO₂·m⁻²·s⁻¹). Lei & Yang [53] reported a value of approximately 0.005 μmol·m⁻²·s⁻¹ (0.0002 mg CO₂·m⁻²·s⁻¹) during the early growth stage and 0.065 μmol·m⁻²·s⁻¹(0.0029 mg CO₂·m⁻²·s⁻¹) during the rapid growth stage for maize.

The variation in R_eco is similar to the variation in α. R_eco gradually increased and reached a maximum in July and then began to decrease in vegetable field, wetland, and orchard ecosystems in parallel with the temperature. For the maize cropland, the maximum R_eco was reached during the jointing stage. At this stage, the temperature is at its highest and maize is in its vigorous growth stage. The temperature is an important factor for R_eco, and R_eco increases with temperature [53]. The pattern of R_eco is similar to the daytime R_eco derived from the light-response curve.

The R² between NEE and PAR are significant in the main growth season of the vegetable field, orchard, wetland and maize cropland (Table 4). This means that the PAR and NEE fit well. The lowest R² was 0.319 in June in the vegetable field. The transplanted vegetation were very small in June, and the site was nearly a bare field; also, the utilization of light was very low. These facts caused the poor fit in June in the vegetable field. In other ecosystems, the R² values were high, especially in the maize cropland, which had a R² greater than 0.670 for the entire growth stage.