Carbon fluxes across alpine, oasis, and desert ecosystems in northwestern China: The importance of water availability

Science of the Total Environment 697 (2019) 133978

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Carbon fluxes across alpine, oasis, and desert ecosystems in northwestern China: The importance of water availability Haibo Wang a,b, Xin Li c,d,⁎, Jingfeng Xiao b, Mingguo Ma e, Junlei Tan a, Xufeng Wang a, Liying Geng a a Key Laboratory of Remote Sensing of Gansu Province, Heihe Remote Sensing Experimental Research Station, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou 730000, China b Earth Systems Research Center, Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, NH 03824, USA c National Tibetan Plateau Data Center, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing 100101, China d CAS Center for Excellence in Tibetan Plateau Earth Sciences, Chinese Academy of Sciences, Beijing 100101, China e Chongqing Engineering Research Center for Remote Sensing Big Data Application, School of Geographical Sciences, Southwest University, Chongqing 400715, China

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• We examined the spatial patterns of carbon fluxes across alpine-desertoasis sites. • Water availability mainly controls spatial variations of carbon fluxes in drylands. • Marked differences in precipitation affect carbon fluxes within and outside of oases. • Irrigation and groundwater supply affect water availability in oases in arid regions.

a r t i c l e

i n f o

Article history: Received 3 May 2019 Received in revised form 15 August 2019 Accepted 17 August 2019 Available online 21 August 2019 Editor: Elena Paoletti Keywords: Dryland ecosystems Alpine-desert-oasis system Carbon exchange Evapotranspiration Water use efficiency Water availability

a b s t r a c t Dryland regions cover N40% of the Earth's land surface, making these ecosystems the largest biome in the world. Ecosystems in these areas play an important role in determining the interannual variability of the global terrestrial carbon sink. Examining carbon fluxes of various types of dryland ecosystems and their responses to climatic variability is essential for improving projections of the carbon cycle in these regions. In this study, we made use of observations from a regional flux tower observation network in a typical arid endorheic basin, the Heihe river basin (HRB). As a representative area of both the arid region of China and the entire region of central Asia, the HRB includes the main ecosystems in arid regions. We compared the spatial variations of carbon fluxes of five terrestrial ecosystems (i.e., grassland, cropland, desert, wetland, and forest ecosystems) and explored the responses of ecosystem carbon fluxes to climatic factors across different ecosystems. We found that our region exhibits a carbon sink ranging from 85.9 to 508.7 gC/m2/yr for different ecosystems, and the water availability is critical to the spatial variability of carbon fluxes in arid regions. Carbon fluxes across all sites exhibited weak correlations with temperature and precipitation. Marked differences in precipitation effects were observed between the sites within oases and those outside of oases. Irrigation and groundwater recharge were of great importance to the variations in carbon fluxes for the sites within oases. Evapotranspiration (ET) exhibited strong relationships with carbon fluxes, indicating that ET was a better metric of soil water availability than was precipitation in

⁎ Corresponding author at: National Tibetan Plateau Data Center, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing 100101, China. E-mail address: [email protected] (X. Li).

https://doi.org/10.1016/j.scitotenv.2019.133978 0048-9697/© 2019 Elsevier B.V. All rights reserved.

2

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

driving the spatial variability of carbon fluxes in arid regions. This study has implications for better understanding the carbon budget of terrestrial ecosystems and informing ecological management in dryland regions. © 2019 Elsevier B.V. All rights reserved.

1. Introduction Dryland regions cover N40% of the global terrestrial surface (Schimel, 2010; Bastin et al., 2017). These widely distributed drylands play an important role in the global carbon cycle (Ahlström et al., 2015). The dryland ecosystems are particularly vulnerable to environmental stresses and human activities (Puigdefábregas and Mendizábal, 1998; Wang et al., 2015), which can significantly affect the carbon balance of arid regions (Gao et al., 2012). Given the wide distribution of dryland ecosystems and their important role in the global carbon cycle, elucidating the dynamics and environmental drivers of carbon fluxes of these ecosystems can help us better understand and project the terrestrial carbon dynamics in these regions in a changing climate (Houghton, 2007). In recent years, the eddy covariance (EC) technique has offered a valuable means to continuously and reliably assess the seasonality, annual variations, and long-term trends of ecosystem carbon exchange in terrestrial ecosystems. Currently, there are N900 EC flux sites in the world (Chu et al., 2017). By joining these flux towers and regional networks such as AmeriFlux, Euroflux, AsiaFlux, and OzFlux globally (Baldocchi et al., 2001; Aubinet et al., 2000; Mizoguchi et al., 2009; Beringer et al., 2016), FLUXNET provides the largest synthesis data set of carbon and water fluxes available (Chu et al., 2017). With a large number of flux data available for us, many studies have analyzed the spatial patterns and climatic controls of carbon and water fluxes using observations from EC sites encompassing a range of ecosystems and climate types (e.g., Law et al., 2002; Yu et al., 2013; Chen et al., 2015; Lu et al., 2017; Kondo et al., 2017). Nonetheless, the EC sites of these synthesis studies were dominated by grasslands, forests, croplands and wetlands, and much less attention has been given to arid and semiarid ecosystems (Elmar and Veenendaal, 2004; Hastings et al., 2005; Biederman et al., 2017). The availability of continuous, long-term carbon dioxide (CO2) exchange measurements lagged in arid regions compared to mesic and humid regions (Biederman et al., 2017). Recent studies have synthesized carbon fluxes in a number of dryland ecosystems in Australia (Cleverly et al., 2013, 2016; Eamus et al., 2013, 2016), North America (Anderson-Teixeira et al., 2011; Biederman et al., 2016, 2017), Europe (López-Ballesteros et al., 2016), and Africa (Tagesson et al., 2016). However, few studies have examined the spatial patterns and variations of carbon fluxes across alpine, oasis, and desert ecosystems along large climate gradients in arid regions of Asia. China is among the countries with a broad distribution of drylands (Wang et al., 2004), and the arid and semiarid regions cover more than one third of China's land area (Liu et al., 2008). The alpine-desert-oasis ecosystems, which is unique and mainly distributed in the Hexi Corridor (including large areas of Gansu, Qinghai and Xinjiang Provinces) in the arid regions of northwestern China, plays an important role in maintaining the stability of the ecological structure and function of these regions and is especially beneficial to the protection against desertification and maintenance of ecological security in arid regions. With the very low rainfall and monsoon climate in these regions, the growth of vegetation is sensitive to changes in water availability and relies on shallow groundwater and irrigation (Zhao and Liu, 2010). To examine the characteristics of carbon fluxes across alpine, oasis, and desert ecosystems in the dryland regions of northwestern China, we established an intensive

EC network in the Heihe River Basin (HRB) (Li et al., 2013, 2017; Liu et al., 2018), which consists of 14 EC flux sites with large climate gradients and various landscape units in this typical arid endorheic catchment. The dataset is unique and is regarded as representative of the main ecosystems in the arid regions of western China and central Asia. Identifying the importance of the influences of climatic (e.g., temperature and precipitation) and biogeochemical (e.g., biomass, growing season length) conditions on ecosystem processes across geographical gradients remains a challenge (Michaletz et al., 2018); in particular, it is not well known to us about the influences of these conditions on dryland ecosystems due to the relatively limited availability of measurements in drylands compared to mesic and humid regions. Among dryland zones, there are two distinct seasons (i.e., winter/spring and summer) with different climate dynamics (Misson et al., 2010; Scott et al., 2012). For example, precipitation mainly falls in the warm summer season in regions dominated by the monsoon climate, while much of precipitation occurs in the cold winter season in Mediterranean regions (Rambal et al., 2003; Misson et al., 2010; Scott et al., 2012). Due to the unique seasonal dynamics of climate conditions in drylands (Anderson-Teixeira et al., 2011; Scott et al., 2012), dryland ecosystems exhibit strong linkages between carbon and water fluxes and high variability in these fluxes (Biederman et al., 2017). The influence of temperature and soil moisture on carbon exchange processes has been investigated in previous studies (e.g., Wen et al., 2010; Fei et al., 2018; Stange, 2007), while no unified formulation has quantified their effects. Previous studies have used the quadratic/ exponential regression to quantify the relationships of temperature and soil moisture with the carbon exchange (e.g., Wen et al., 2010; Fei et al., 2018). Stange (2007) also used the optimum function from O'Neill (Diekkruger et al., 1995) to simulate the climate effects on carbon fluxes, which increases quasi-exponentially at lower temperatures and decreases at temperatures beyond the optimum. Previous studies also have demonstrated that carbon fluxes in dryland ecosystems are particularly sensitive to hydroclimatic variability (Rambal et al., 2003; Huxman et al., 2004; Vargas et al., 2013) and are thus vulnerable to climatic extremes (Reichstein et al., 2002; Misson et al., 2011; Ma et al., 2015). Precipitation is a common proxy for ecosystem water availability (Sala et al., 2012), which is a primary driver for carbon fluxes and evapotranspiration (ET) in dryland regions (Cleverly et al., 2013; Cleverly et al., 2016). The effects of precipitation regimes on carbon fluxes mainly depend on the magnitude and timing of precipitation (Huxman et al., 2004), and plants can adapt to limited precipitation and management to some extent (e.g., Misson et al., 2010; Nolan et al., 2018). It is well established that drought is an important determinant in water losses and carbon fluxes in Mediterranean dryland ecosystems (e.g., Reichstein et al., 2002; Reichstein et al., 2003; Rambal et al., 2003; Misson et al., 2010). However, it remains not fully understood to us that how dryland ecosystems respond to variations in precipitation regimes (González-Megías and Menéndez, 2012; Cleverly et al., 2013), especially in the monsoon-type dryland ecosystems. Previous studies have demonstrated that the apparent responses of productivity to precipitation in certain drylands of Australia were caused by large interannual variability in precipitation (Cleverly et al., 2013, 2016), while other studies reported the ET instead of precipitation can be used as a reliable metric of available water in explaining the variability of carbon exchanges in drylands in southwestern US with complex terrain (e.g., Biederman et al., 2017). Furthermore, since precipitation

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

in arid regions is limited, dryland regions typically encompass a wide hydrological niche space (e.g., from arid or semiarid areas to oases). The hydrologic niche separation caused by irrigation and groundwater discharge can be a major component of the water balance, which is likely a critical factor driving the within-biome variability of key plant traits related to carbon fluxes (Nolan et al., 2018). However, it is not yet well understood how this hydrological niche separation affects carbon fluxes in dryland ecosystems. Here we examined the carbon fluxes across alpine, oasis, and desert ecosystems along a large precipitation gradient from MAP of 26 to 500 mm in the HRB of northwestern China to address the gaps in the current understanding of how carbon fluxes respond to water availability and other climatic patterns in the HRB. To our knowledge, our study is the first to examine the carbon fluxes of compound alpine-desertoasis ecosystems across a wide range of climate and ecosystem types. The objectives of this study are as follows: (1) to examine the magnitude, spatial patterns, and temporal variations in the carbon fluxes

3

across alpine, desert and oasis ecosystems of the same catchment; (2) to investigate the underlying drivers of the variations in carbon fluxes in these ecosystems; (3) to quantify how water availability regulates ecosystem carbon exchange in this arid region. The large magnitude and seasonal distribution gradients of precipitation and water discharges across various ecosystems provide us the opportunity to test the following hypotheses. First, we hypothesized that the carbon fluxes within dryland ecosystems of the HRB exhibit large spatial variability because of the large spatial differences in climatic conditions. Second, we hypothesized that the carbon flux patterns differ between the sites located within oases and those outside of oases because of the hydrological niche separation in the study area. Third, we hypothesized that water availability can better explain carbon flux variability in the HRB than can temperature. Our study can provide useful information for assessing the carbon budget of terrestrial ecosystems and managing the terrestrial ecosystems in the arid regions of northwestern China.

Fig. 1. The location and distribution of the eddy covariance (EC) flux sites across the Heihe River Basin (HRB) in northwestern China. The top right corner map shows the location of the HRB in China. The base map is the land use and land cover data of the HRB. The three pink regions represent the key experimental areas in different parts of the HRB. For sites codes and descriptions, see Table 1. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

4

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

2. Materials and methods 2.1. Regional flux observation network We used the flux and micrometeorological observations from a flux observation network across alpine-desert-oasis ecosystems of the same catchment in northwestern China (Fig. 1). Our study area, the Heihe River Basin (HRB) (37.7°-42.7°N, 97.1°-102.0°E), is a typical endorheic basin that originates in Qilian Mountain and terminates in Juyan Lake. The HRB is the second largest inland river basin in China, covering an area of approximately 1,432,000 km2, and is located in the middle part of the Hexi Corridor in the arid region of northwestern China (Cheng et al., 2014). This basin is characterized by distinct cold and arid landscapes distributed from the upper to the lower reaches of the basin (Li et al., 2013; Zhou et al., 2019). In the mountainous upstream area of the HRB, where typical cold region landscapes such as glacier, alpine grassland, and alpine meadow areas are distributed (Wang et al., 2013), the elevation ranges from 2640 to 5000 m, the annual precipitation is 500–800 mm, and the annual evapotranspiration (ET) is 321–600 mm (Li et al., 2018; Wang et al., 2019). The middle reaches are dominated by artificial oasis-riparian zone-wetland-desert compound ecosystems; the annual precipitation is 100–250 mm, while the annual potential evaporation (PE) is 1200–1800 mm. The downstream area is characterized by natural oases dispersed within desert areas and has an annual precipitation of b50 mm and annual PE over 3000 mm (Li et al., 2013). The HiWATER program was initiated in 2012 to improve the observability of hydrological and ecological processes and to build an observation system at the basin scale (Li et al., 2016; Li et al., 2017). HiWATER is a watershed-scale eco-hydrological experiment designed from an interdisciplinary perspective to address complex problems, such as heterogeneity, scaling, uncertainties, and closing the water cycle at the watershed scale (Li et al., 2013). According to the design of the program, three key experimental areas in different parts of the HRB (i.e., the cold region, artificial oasis, and natural oasis experimental areas) were selected. During the duration of the HiWATER program, we built a flux observation network in these experimental areas of the HRB (Fig. 1), which included three grassland sites, five desert sites, two cropland sites, three forest sites, and one wetland site. These sites encompass the major eco-regions and ecosystem types in the endorheic basin. Since different climatic and hydrological conditions are present within and outside of an oasis, we divided these sites into two groups (i.e., sites close to or within an oasis and sites far from an oasis) according to the distance of each site to the nearest oasis. Oasis sites were mainly crops in the middle stream and downstream areas of the HRB (the cropland was planted with plastic film for water-saving) and riparian trees in the downstream regions of the HRB, while the sites outside the oasis are mainly steppe vegetation in the upper stream regions with

higher elevation and the desert sites in the middle stream and downstream regions with relative lower elevation. The specific site information on these sites is shown in Table 1. There are large variations in temperature, precipitation and altitude gradients among different experimental areas in the HRB (Fig. 2). The high-elevation sites are wetter and colder than the low-elevation sites. These climatic gradients result in a heterogeneous compound alpine-desert-oasis landscape, which is typical of the arid regions in northwestern China and central Asia.

2.2. Flux and micrometeorological measurements and data processing EC flux towers can provide continuous measurements of net ecosystem exchange (NEE) and evapotranspiration (ET). NEE represents the net flux of CO2 between an ecosystem and the atmosphere. Net ecosystem productivity (NEP) is defined as the difference between the gross primary production (GPP) and ecosystem respiration (Reco), and thus NEP is equal to −NEE if nonbiological fluxes (e.g., carbonate geochemistry, photodegradation) are negligible (Chapin et al., 2006). The openpath EC (OPEC) system was used to measure carbon and water vapor fluxes. The OPEC system at each site consists of a 3D sonic anemometer (CSAT-3/Gill, Campbell Scientific Instruments Inc., USA/Gill, UK) to measure three-dimensional wind speed and temperature fluctuations and an open path infrared gas analyzer (Li-7500A, Licor Inc., USA) to measure carbon and water vapor densities. The meteorological variables including air temperature, precipitation, solar radiation, soil temperature, and soil moisture were measured simultaneously at each site. Detailed information on the meteorological measurements collected by this study is shown in Supplemental Table S1. An investigation of the energy balance closure ratio (EBR) showed that the half-hourly EBR of all the sites within the flux network was within 0.53–0.99, with an average of 0.84 (Zhou and Li, 2019). All flux data were carefully controlled for quality and processed at half-hourly time steps. The Eddypro software developed by Li-Cor (http://www.licor.com/env/products/eddy_covariance/software.html) was applied to process the raw 10 Hz data. Postprocessing included spike detection, time lag correction of H2O/CO2 relative to vertical wind components, coordinate rotation, sonic virtual temperature correction, frequency response correction (Moore, 1986) and density fluctuation corrections (Webb-Pearman-Leuning (WPL) correction) (Xu et al., 2013; Liu et al., 2011). We also removed any half-hourly flux values that (1) were collected during a sensor malfunction period, (2) were collected during a rainy period, (3) were outside the reasonable range of values, or (4) had a friction velocity (u*) at night that was less than the corresponding ecosystem type-dependent threshold friction velocity (Papale et al., 2006). Detailed information on the data processing and quality control of the flux data can be found in Xu et al. (2013). In our study, approximately 28.43% of data, on average,

Table 1 Site descriptions of the flux observation network within the Heihe River Basin (HRB). MAP: mean annual precipitation (mm); MAT: mean annual temperature (°C). Biomes types

Site Name

Time Periods

Landscape

Location of HRB

Within or out of oases

Lat (N)

Lon (E)

Elev (m)

MAT (°C)

(mm)

Grassland Grassland Grassland Desert Desert Desert Cropland Wetland Desert Desert Forest Cropland Forest Forest

Yakou (YKZ) Dashalong (DSL) A'rou (ARZ) Huazhaizi (HZZ) Shenshawo (SSW) Gebi (GBZ) Daman (DMZ) Shidi (SDZ) Huangmo (HMZ) Luodi (LDZ) Huyanglin (HYL) Nongtian (NTZ) Hunhelin (HHL) Sidaoqiao (SDQ)

2015–2016 2013–2016 2013–2016 2012–2016 2012–2016 2012–2014 2012–2016 2012–2016 2015–2016 2012–2014 2013–2015 2013–2015 2013–2016 2013–2016

Alpine meadow Alpine meadow Alpine grassland Desert steppe Desert steppe Gobi steppe Maize Reed Desert steppe Desert steppe Populus euphratica Cantaloupe Mixed forest Tamarix forest

Upstream Upstream Upstream Middle stream Middle stream Middle stream Middle stream Middle stream Downstream Downstream Downstream Downstream Downstream Downstream

Outside Outside Outside Outside Outside Outside Within Within Outside Within Within Within Within Within

38.0142 38.8399 38.0473 38.7652 38.7892 38.9150 38.8555 38.9751 42.1135 41.9993 41.9928 42.0048 41.9903 42.0012

100.2421 98.9406 100.4643 100.3186 100.4933 100.3042 100.3722 100.4464 100.9872 101.1326 101.1236 101.1338 101.1335 101.1374

4148 3739 3033 1731 1594 1562 1556 1460 1054 878 876 875 874 873

−4.7 −3.9 −0.3 9.1 8.9 9.1 6.9 9.2 10.1 12.3 10.3 9.4 10.0 10.1

500.8 314.4 444.7 166.4 139.7 102.3 135.7 119.9 36.1 24.8 26.0 35.6 35.5 37.1

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

5

Fig. 2. Seasonal variations of monthly precipitation (mm), mean temperature (°C), solar radiation (MJ m−2 mon−1) and soil moisture (cm3 cm−3) of different sites and biomes in the Heihe River Basin (HRB). The left panels depict the sites in the alpine cold region of the experimental area in the upper reaches of the basin; the middle panels depict sites in the artificial oasis experimental area in the middle reaches, a compounded desert-oasis ecosystem in the arid region of northwestern China; the right panels depict sites in the natural oasis experimental area in the lower reaches of the HRB, an extremely arid region.

were discarded. The percentages of data gaps for each site are listed in the Table S2. Because EC flux towers directly measure NEE, ecosystem respiration (Reco) and GPP were derived from NEE measurements according to a commonly used flux partitioning algorithm (Wutzler et al., 2018). We used the marginal distribution sampling method (MDS) (Reichstein et al., 2005) to fill the gaps in the half-hourly carbon fluxes. In this study, all the gap-filling, flux uncertainty estimations and partitioning methods were based on the processing procedures (REddyProc) provided by the Max Planck Institute for Biogeochemistry, Jena (Wutzler et al., 2018).

2.3. Data analysis We compiled the fluxes and meteorological data from all 14 sites from the observational network in the HRB. For each site, the annual carbon fluxes (GPP, Reco and NEP) and annual climate variables (MAT, MAP, and soil moisture) were calculated from the daily carbon flux, micrometeorological, and soil moisture data. We examined the magnitude and spatial patterns of carbon fluxes across grassland, desert, forest, cropland and wetland ecosystems. To test whether there is a significant difference in the magnitude of NEP among different sites within each vegetation type, we performed an analysis of variance (ANOVA) test using individual site years for each vegetation type. We then examined the regulatory mechanisms of annual carbon fluxes by using regression models between carbon fluxes (NEP, GPP and Reco) and micrometeorological variables. Both linear and nonlinear exponential regression models were used to quantify the relationships between micrometeorological variables and carbon fluxes at monthly and annual timescales. Following Wen et al. (2010) and Fei et al. (2018), we also used exponential regression (Eq. (1)) and two-factor regression models (Eq. (2)) to quantify the relationship of air

temperature and soil moisture with carbon fluxes at the monthly timescale: f ðxÞ ¼ a  exp ðbTÞ

ð1Þ

f ðxÞ ¼ a  exp:ðbTÞ  Wc

ð2Þ

where T is the air temperature (°C); W is the soil water content (%); f (x) represents carbon exchange (GPP, Reco or NEP). In addition to precipitation, both soil moisture and ET were applied to explore the relationship between water availability and the spatial-temporal variability of carbon fluxes in the HRB. To better investigate the influence of temperature on carbon exchange processes, we also used the optimum function from O'Neill (Diekkruger et al., 1995), which is more suitable to characterize microbiological processes at high temperatures than the monotonically increasing Arrhenius function (Stange, 2007). The formulation of the O'Neil function is described in the supporting text (ST). The satellite-derived “greenness” index (e.g., the MODIS enhanced vegetation index (EVI)) may be empirically related to in situ measurements of carbon fluxes (Sims et al., 2006). We therefore also used the MODIS EVI product to explore seasonal similarities between vegetation greenness and carbon fluxes. The MOD13Q1 product with a 16-day temporal resolution and 250-meter spatial resolution was used in this study. 3. Results 3.1. Magnitude and spatiotemporal variability of carbon fluxes across different climatic gradients Monthly time series of NEP, GPP and Reco (units: gC/m2/mo) were used to illustrate the seasonal variations in carbon fluxes across ecosystems in the HRB (Fig. 3). The precipitation was mainly concentrated in

6

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

Fig. 3. Seasonal variations in monthly carbon fluxes (monthly NEP, GPP and Reco, with units of gC m−2 mon−1) and monthly EVI across different ecosystems in the Heihe River Basin (HRB): grassland (YKZ, DSL and ARZ); cropland (DMZ and NTZ); desert (HZZ, GBZ, SSW, LDZ and HMZ); wetland (SDZ); forest (HYL, HHL and SDQ). The abbreviations of the sites can be found in Table 1. For each ecosystem, the lines with stars and the band ranges represent the means and standard deviations of the carbon fluxes or EVI, respectively.

the growing seasons, and the seasonal dynamics of carbon fluxes was also generally consistent with the variations in air temperature and precipitation (Fig. 2). The carbon fluxes had similar seasonal dynamics across different biomes although their amplitude exhibited large variations. The cropland and wetland ecosystems had the highest carbon fluxes in amplitude, followed by the grassland and forest ecosystems; the desert ecosystems had the lowest carbon fluxes. The NEP of the desert ecosystems ranged from −15 to 25 gC/m2/mo and had no clear seasonal patterns. The seasonal patterns of carbon fluxes were closely related to the variations in vegetation greenness. We also averaged the hourly instantaneous carbon exchange data (Fig. 4), which reveal the averaged diurnal variation of NEP during the whole year in the study periods. Fig. 4 shows that the NEP of different ecosystems in the HRB during the whole year presented obvious diurnal variations: the ecosystem was a carbon source during nighttime because of ecosystem respiration; during the daytime, as the plants photosynthesized, the whole ecosystem turned into a carbon sink at approximately 08:00 (local time), reached its diurnal maximum carbon assimilation typically at 12:00–13:00 (local time), and then gradually decreased in the afternoon. On the whole, at the annual scales, all the ecosystems were apparent sinks for atmospheric CO2. The cropland and wetland ecosystems had the maximum potential for carbon uptake, followed by the forest ecosystems and grassland ecosystems, while the desert ecosystems had the lowest carbon uptake. The average annual NEP values of the grassland, cropland, desert, wetland and forest ecosystems were 146.0 ± 60.9, 324.6 ± 299.1, 85.9 ± 31.6, 508.7 ± 50.6, and 216.4 ± 83.4 gC/m2/yr, respectively

(Fig. 5). Based on the results of the ANOVA, there was no statistically significant difference (p N 0.05) in the NEP among different sites for either desert or forest, which are marked with asterisks in Fig. 4. In contrast, a statistically significant difference (p b 0.05) in the NEP among sites was found for grassland and cropland. The differences between the two irrigated cropland types (DMZ and NTZ) were mainly caused by differences in the crop type and growing season length (GSL). NTZ is a cantaloupe-cropped site located in an extremely arid region, and has a GSL of approximately 3 months, while DMZ is a maize-cropped site with a GSL of approximately 5 months. The differences between the two grassland types (ARZ and DSL) were mainly caused by differences in the climate conditions. ARZ is an alpine grassland site located in a seasonally frozen region, while DSL is an alpine meadow site located in a permafrost region. In comparison, the differences in the carbon fluxes between the desert and forest ecosystems were relatively small. Large variability in annual carbon fluxes were found both within and across biomes (Fig. 6e and f). The annual NEP of the terrestrial ecosystems in the HRB ranged from 46.76 to 536.06 gC/m2/yr. All ecosystems were carbon sinks with an average of 191.01 gC/m2/yr for all sites. The largest annual NEP (536.06 gC/m2/yr) was observed for the maize site, followed by the wetland ecosystem (508.67 gC/m2/yr); the lowest annual NEP was found for the sandy desert (46.76 gC/m2/yr) and Gobi desert (59.30 gC/m2/yr) ecosystems. The annual GPP of the 14 sites ranged from 174.57 gC/m2/yr in the Gobi desert ecosystem to 1204.62 gC/m2/yr in the irrigated cropland site. The maize and wetland sites had the highest annual GPP (N1000 gC/m2/yr). The annual Reco ranged from

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

7

Fig. 4. Diurnal variations of average net ecosystem productivity (NEP) in the study period in different ecosystems in the HRB (the time step is half-hourly). a, b, c and d represents the grasslands, desert, cropland and wetland, and forest ecosystems of environmental conditions, respectively. For sites codes and descriptions, see Table 1.

105.60 gC/m2/yr at HZZ (desert ecosystem) to 694.40 gC/m2/yr at ARZ (grassland ecosystem). We examined the patterns of annual carbon fluxes across the 14 sites along the MAT (Fig. 6a), MAP (Fig. 6b), SM (Fig. 6c) and ET (Fig. 6d) gradients, these variables were ranked according to the topographic gradient in the HRB. Large spatial variability in temperature and water availability were found both within and across biomes, which in turn resulted in large spatial variability of carbon fluxes across sites in our study area (Fig. 6e and f). The MAT generally increased with decreasing elevation, while MAP decreased with decreasing elevation. However, we did not find apparent trends as changing elevation, MAT or MAP. We also ranked the annual carbon fluxes according to the

order of MAT, MAP, SM and ET separately (Figs. S2, S3, S4 and S5 in the Supporting Materials, respectively). Carbon fluxes did not generally increase with increasing MAT, MAP or SM (Figs. S2, S3 and S4). By contrast, carbon fluxes generally increased with increasing ET (Fig. S5). This finding shows that the annual carbon flux patterns across the 14 sites were not similar to the MAT, MAP or SM patterns but, rather, resembled the annual ET patterns, indicating that annual ET is a better predictor than the other variables of annual carbon fluxes in the HRB. Fig. 6 (g) also shows that the water use efficiency (WUE) of the ecosystems of the HRB changed with ET. WUE varied from 0.74 g to 7.27 gC m−2 mm−1 with different vegetation types (Fig. S5). With the increase of ET, there was no significant trend in WUE among different ecosystems. The higher WUE for desert and forest ecosystems reflects the adaptability of these ecosystems to the water-limited environment. Certain croplands such as the cantaloup-cropped NTZ had a lower WUE, while the maize-cropped DMZ had a higher WUE, which is related to the difference in water consumption and management techniques for these two crops. 3.2. Effects of climatic factors on carbon fluxes

Fig. 5. Magnitude of annual carbon fluxes across different ecosystems in the Heihe River Basin (HRB). For each ecosystem, the bar length and error bars represent the means and standard deviations of the carbon fluxes, respectively. Based on analysis of variance (ANOVA), there was no statistically significant difference (p N 0.05) in the NEP among different sites for either desert or forest (marked with asterisks), the sample sizes of ANOVA were labeled in the figure.

We calculated the relationships between carbon fluxes and temperature at both monthly and annual timescales (Fig. 7). Monthly carbon fluxes increased with increasing air temperature across all the site-years. In contrast, the annual carbon fluxes first increased and then decreased as the MAT increased. To reveal the effects of temperature on carbon fluxes at different timescales, different response functions were used. At the monthly timescale, the dependency of carbon fluxes on temperature could be generally described with an exponential function (Fig. 7 a, c, and e). We used a nonlinear regression function, the O'Neill function (Diekkruger et al., 1995), to fit the temperature response function curves of annual carbon exchange (Fig. 7 b, d, and f). The O'Neill function is

8

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

Fig. 6. Spatial variations in annual carbon fluxes in 14 sites along the mean annual temperature (MAT), mean annual precipitation (MAP), soil moisture (SM), and annual evapotranspiration (ET) gradient from upstream to midstream and downstream regions of Heihe River Basin (HRB). All sites were ranked with the topographic gradient across the HRB. The full site names and vegetation types are provided in Table 1. For each site, the bar length and error bar represent the means and standard deviations of the specific variable, respectively. No data showed (marked with the Red Cross) for SDZ, NTZ and HYL in Fig. 5 (c), since SM measurement were not available in these sites. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

suitable for characterizing the increase in carbon fluxes quasiexponentially at relatively low temperatures and the decrease in carbon fluxes at temperatures beyond the optimum (Fig. S1 a, c, and e). Despite the statistical significance of the relationships at both monthly and annual timescales, these relationships were relatively weak, indicating that temperature had little effect on the spatial patterns of carbon fluxes. Meanwhile, after we ranked the data in temperature-limited and not-limited sites, we found the relationship between temperature and carbon fluxes was stronger in certain temperature-limited sites in the upstream regions of the HRB (Supporting Test, Fig. S1 b, d, f, and Table 2), since these sites are rather cool and rarely water-limited. While for certain sites distributed in downstream and midstream reaches of the HRB that are

warmer and relatively more water-limited, the relationships between temperature and carbon fluxes were weak. Precipitation is a common proxy of water variability, which is highly important to carbon uptake in dryland ecosystems. However, our results showed that carbon fluxes were not statistically significant correlated with MAP across the site-years as a whole (the black dashed lines in Fig. 8 (a, c, e) and 9 (a, c, e) indicate the overall fit of all sites). The growth of vegetation within the oases mainly depended on irrigation or groundwater recharges due to the shortage of precipitation, while the vegetation outside the oases relied on precipitation. As we separate these sites into two groups, for both within-oasis sites and outside-oasis sites, carbon fluxes were significantly and strongly correlated with precipitation at both monthly and annual timescales (the two

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

9

Fig. 7. Effects of temperature on carbon fluxes at monthly and annual timescales: (a), (c) and (e) are the exponential relationships between the carbon fluxes and mean monthly temperature (MMT) at the monthly timescale; (b), (d) and (f) are relationships between the carbon fluxes and mean annual temperature (MAT) at the annual timescale. We fitted the relationship using a nonlinear temperature response function, the O'Neill function (more details in Supporting Text (ST)), and the R2 values were 0.12, 0.14 and 0.19 for NEP, GPP, and Reco, respectively.

groups were labeled with blue and red lines, respectively, in both Fig. 8 (a, c, e) and 9 (a, c, e)). However, the carbon fluxes of these two groups exhibited different responses to precipitation at both timescales (Figs. 8 and 9). There were marked differences in the effects of precipitation and irrigation (or groundwater supply) between these groups of sites. NEP was sensitive to precipitation in the natural ecosystems distributed outside of the oases, such as the grassland ecosystems in the alpine zones of the upper reaches of the HRB (YKZ, DSL and ARZ) and desert ecosystems on the margin of the oases (HZZ, GBZ, SMZ and HMZ). The groundwater supply had larger effects on the carbon fluxes at the sites inside oases, including croplands relying on irrigation (i.e., DMZ, NTZ), ecosystems near groundwater outlets (SDZ), and sites close to the Heihe River (HYL, HHL and SDQ), than at the sites outside oases. Similar to Biederman et al. (2017), we also used ET as a metric of ecosystem water availability in our study area. Unlike different patterns between within- and outside-oasis sites for correlations of carbon fluxes with precipitation, ET was strongly correlated with carbon fluxes across all site-years (Fig. 8 (b, d, f); Fig. 9 (b, d, f)). The average coefficient of

determination (R2) between the component fluxes and ET was 0.57, indicating that ET explained 57% of the variations in the carbon fluxes. Therefore, we also proved that ET was a better predictor of ecosystem carbon fluxes than MAP. Soil moisture at sites with irrigation depends on both precipitation and irrigation. We also compared the relationships between carbon fluxes and soil moisture. Soil moisture had moderate relationships with GPP and Rreco across all ecosystems and a weak relationship with NEP (Fig. 10), demonstrating that soil moisture could better explain the spatial variations in GPP and Rreco in arid regions than MAP or MAT. To explore the interactive effect of temperature and soil moisture on carbon fluxes, we used multiple factor regression models to examine the relationships between these factors. The results showed that the carbon fluxes generally increased with increasing air temperature and soil moisture (Fig. S6). The R2 between the carbon fluxes (i.e., GPP, Reco and NEP) and the interactions of air temperature and soil moisture were 0.36, 0.42 and 0.15, which demonstrated that the interactions between air temperature and soil moisture played an important role in

10

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

Table 2 Coefficients of determination (R2) for the relationships between monthly carbon fluxes and environmental factors in the HRB. * and ** indicate that the relationships are statistically significant at 0.05 and 0.01 level, respectively. Site name

Precipitation (Linear function)

Temperature (Ta) (Linear function)

GPP, Reco, NEP

GPP, Reco, NEP

GPP, Reco, NEP

GPP, Reco, NEP

GPP, Reco, NEP

Yakou (YKZ) Dashalong (DSL) A'rou (ARZ) Huazhaizi (HZZ) Shenshawo (SSW) Gebi (GBZ) Daman (DMZ) Shidi (SDZ) Huangmo (HMZ) Luodi (LDZ) Huyanglin (HYL) Nongtian (NTZ) Hunhelin (HHL) Sidaoqiao (SDQ)

0.76⁎⁎, 0.74⁎⁎, 0.33⁎⁎ 0.74⁎⁎, 0.73⁎⁎, 0.59⁎⁎ 0.82⁎⁎, 0.83⁎⁎, 0.63⁎⁎ 0.35⁎⁎, 0.54⁎⁎, 0.03 0.52⁎, 0.65⁎, 0.09 0.31⁎⁎, 0.53⁎⁎, 0.05 0.67⁎⁎, 0.72⁎⁎, 0.59⁎⁎ 0.56⁎⁎, 0.48⁎⁎, 0.52⁎⁎ 0.01, 0.01, 0.11 0.05, 0.13, 0.01 0.02, 0.07, 0.01 0.01, 0.01, 0.02 0.15⁎, 0.19⁎⁎, 0.07 0.10⁎, 0.16⁎⁎, 0.04

0.68⁎⁎, 0.63⁎⁎, 0.33⁎ 0.50⁎⁎, 0.69⁎⁎, 0.29⁎⁎ 0.63⁎⁎, 0.73⁎⁎, 0.41⁎⁎ 0.45⁎⁎, 0.33⁎⁎, 0.15⁎⁎ 0.68⁎, 0.43⁎, 0.39⁎⁎ 0.39⁎⁎, 0.39⁎⁎, 0.01 0.53⁎⁎, 0.65⁎⁎, 0.44⁎⁎ 0.66⁎⁎, 0.68⁎⁎, 0.54⁎⁎ 0.47⁎⁎, 0.02, 0.34⁎ 0.67⁎⁎, 0.63⁎⁎, 0.37⁎⁎ 0.74⁎⁎, 0.83⁎⁎, 0.47⁎⁎ 0.33⁎⁎, 0.47⁎⁎, 0.20⁎ 0.75⁎⁎, 0.76⁎⁎, 0.56⁎⁎ 0.65⁎⁎, 0.79⁎⁎, 0.38⁎⁎

0.56⁎⁎, 0.44⁎⁎, 0.27⁎⁎ 0.31⁎⁎, 0.47⁎⁎, 0.16⁎⁎ 0.24⁎⁎, 0.38⁎⁎, 0.07⁎ 0.39⁎⁎, 0.06⁎⁎, 0.30⁎⁎ 0.32⁎⁎, 0.45⁎⁎, 0.04⁎ 0.39⁎⁎, 0.60⁎⁎, 0.04⁎ 0.80⁎⁎, 0.76⁎⁎, 0.73⁎⁎

0.67⁎⁎, 0.52⁎⁎, 0.47⁎⁎ 0.62⁎⁎, 0.69⁎⁎, 0.43⁎⁎ 0.81⁎⁎, 0.82⁎⁎, 0.63⁎⁎ 0.34⁎⁎, 0.79⁎⁎, 0.01 0.67⁎⁎, 0.85⁎⁎, 0.11⁎ 0.43⁎⁎, 0.52⁎⁎, 0.01 0.73⁎⁎, 0.70⁎⁎, 0.66⁎⁎ 0.60⁎⁎, 0.57⁎⁎, 0.53⁎⁎

0.71⁎⁎, 0.48⁎⁎, 0.56⁎⁎ 0.81⁎⁎, 0.77⁎⁎, 0.67⁎⁎ 0.90⁎⁎, 0.86⁎⁎, 0.76⁎⁎ 0.33⁎⁎, 0.58⁎⁎, 0.02 0.59⁎⁎, 0.59⁎⁎, 0.18⁎ 0.47⁎⁎, 0.48⁎⁎, 0.01 0.87⁎⁎, 0.78⁎⁎, 0.82⁎⁎ 0.91⁎⁎, 0.77⁎⁎, 0.86⁎⁎

0.02, 0.01, 0.05 0.71⁎⁎, 0.58⁎⁎, 0.48⁎⁎ 0.88⁎⁎, 0.84⁎⁎, 0.69⁎⁎ 0.70⁎⁎, 0.63⁎⁎, 0.61⁎⁎ 0.79⁎⁎, 0.70⁎⁎, 0.70⁎⁎ 0.88⁎⁎, 0.67⁎⁎, 0.77⁎⁎

0.06, 0.01, 0.05 0.41⁎⁎, 0.38⁎⁎, 0.23⁎⁎ 0.78⁎⁎, 0.81⁎⁎, 0.55⁎⁎ 0.53⁎⁎, 0.59⁎⁎, 0.40⁎⁎ 0.89⁎⁎, 0.78⁎⁎, 0.78⁎⁎ 0.89⁎⁎, 0.81⁎⁎, 0.69⁎⁎

controlling the spatial variations in the annual carbon fluxes of the ecosystems in the HRB. We also investigated the effects of seasonal variations in climatic factors on carbon exchange at the site level (Table 2). The results showed that temperature and water availability had different effects on carbon fluxes across sites in the HRB. In the arid and extremely arid ecosystems in the middle and lower reaches of the HRB, temperature controlled seasonal variations in carbon fluxes. In contrast, in ecosystems in the upper reaches of the HRB, temperature and soil moisture jointly controlled variations in carbon fluxes. In addition, we also observed ET played a greater role in controlling spatial patterns of carbon fluxes than other climatic factors. Meanwhile, significant correlations between EVI and carbon fluxes across sites were found in Table 2, suggesting that the seasonal variations in EVI were also consistent with those in carbon fluxes. 4. Discussion 4.1. Spatial-temporal variability of carbon fluxes in arid regions This study analyzed the spatial and temporal patterns of carbon fluxes in alpine-desert-oasis ecosystems in the HRB. Our results provide useful information for assessing the carbon budget of dryland ecosystems. The alpine-desert-oasis ecosystems in this region act as carbon sinks, confirming previous studies reporting that the Northern Hemisphere mid-latitudes have a strong carbon sequestration capacity (Ciais et al., 2000; Yu et al., 2013). Compared with other carbon flux synthesis results in the southwestern US with complex terrain (Biederman et al., 2017), our sites were generally relatively strong carbon sinks for arid regions. This is partly because northwestern China and the southwestern US are characterized by different climates. The HRB is dominated by a monsoon climate with large topographic gradients that feature the synchronization of precipitation and heat. By contrast, the southwestern US is characterized by both Mediterranean and monsoon climates, and therefore a wide range of carbon sink and source function across sites was observed (Biederman et al., 2017). Additionally, compared with other studies in the mesic region, the average annual GPP and Reco in the HRB were slightly lower than those in China (Xiao et al., 2013) and Asia (Chen et al., 2013). However, the average annual NEP in the HRB was slightly higher than that in Asia but lower than that in China. Few sites in the dryland region were included in these studies in China and Asia (e.g. Xiao et al., 2013; Yu et al., 2013; Chen et al., 2013), and most of the sites were located in mesic climatic or coastal areas. The average precipitation of the HRB (especially the downstream and midstream reaches) is much lower than that of China and Asia as a whole, leading to the difference in the magnitudes of carbon fluxes among these studies. The grasslands of the HRB are

Soil moisture (SM) (exponential function)

– 0.05⁎, 0.01⁎, 0.09⁎ – – – 0.19⁎⁎, 0.20⁎⁎, 0.14⁎⁎ 0.26⁎⁎, 0.33⁎⁎, 0.14⁎⁎

ET (Linear function)

EVI (Linear function)

mainly located in the alpine region of high latitudes with semi-humid climates, while the grasslands of the other studies (e.g., Xiao et al., 2013 and Chen et al., 2013) are mainly located in the semi-arid region. The annual precipitation of the grassland sites in the HRB (428.97 mm) was higher than that of China (375.56 mm), and therefore the carbon fluxes in the grasslands of the HRB were slightly higher than those of China and Asia. The wetland carbon fluxes reported for the study sites were higher than those reported for coastland wetland ecosystems by Xiao et al. (2013), indicating that the carbon fluxes of these inland wetlands are higher than those of coastal wetlands. For croplands, overall, the differences between the carbon fluxes in the HRB and Asia were small. Specifically, the average annual GPP and Reco in the HRB were slightly higher than those in China but lower than in Asia, while the NEP in the HRB was higher and lower than that in Asia and China, respectively. For forests, the carbon fluxes in the HRB were lower than those in China and Asia mainly because our study area is within a typical arid region, and the average annual precipitation of forest sites in the HRB (31.12 mm) was much lower than that of China as a whole (360.9 mm). Carbon flux synthesis studies of drylands can help reveal the importance of dryland ecosystems in the terrestrial carbon cycle. These synthesis studies also have limitations. Instrument malfunction and poor weather such as precipitation will force investigators to reject a proportion of the data (Falge et al., 2001), while the gap filling methods applied will introduce uncertainties to the flux data. Previous studies showed that abiotic fluxes (e.g., photo-degradation, geochemistry) could also contribute to carbon losses in certain dryland ecosystems (e.g., Rutledge et al., 2010; Serrano-Ortiz et al., 2010; Adair et al., 2017). These nonbiological processes may in some cases also play an important role in carbon exchanges in drylands, and the role of the abiotic fluxes in carbon budgets in our study area should be quantified in the future. 4.2. Responses of carbon fluxes to environmental variables in dryland ecosystems Previous synthesis studies of ecosystem carbon fluxes in ecosystems identified with strong temperature and precipitation controls on the spatial variability of carbon flux components (NEP, GPP and Reco) (e.g., Kondo et al., 2017; Law et al., 2002; Yu et al., 2013; Chen et al., 2015). However, the annual carbon fluxes in the HRB showed weak relationships with MAT, demonstrating that MAT is not the main factor controlling the spatial variations in the HRB carbon fluxes, which is similar to other dryland studies (e.g., Anderson-Teixeira et al., 2011; Biederman et al., 2017). For example, Biederman et al. (2017) also found a negative relationship between the mean annual GPP and MAT

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

11

Fig. 8. Effects of mean monthly precipitation (MMP) and mean monthly ET on carbon fluxes. (a), (c) and (e) show linear relationships between monthly carbon fluxes and MMP; the black dashed lines indicate the overall fit of all sites; the sites are grouped into inside-oasis (labeled in red, including DMZ, SDZ, NTZ, HYL, HHL, SDQ and LDZ) and outside-oasis (labeled in blue) sites. (b), (d) and (f) show the linear relationships between monthly carbon fluxes and monthly ET. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

and an insignificant relationship between Reco and MAT. Unlike annual carbon fluxes, monthly carbon fluxes (NEP, GPP and Reco) showed stronger positive correlations with air temperature, which is similar to the results for other ecosystems in mesic regions (e.g., Wen et al., 2010; Fei et al., 2018), and revealed that the seasonal variations of temperature explained the monthly variability in carbon fluxes well. Precipitation is recognized as a common proxy of water availability and generally plays an important role in controlling dryland carbon fluxes (Reichstein et al., 2002; Rambal et al., 2003; Misson et al., 2010). Both the magnitude and the seasonal distribution of precipitation are critical factors in the carbon uptake in arid and semiarid regions (Mielnick et al., 2005). However, the relationships between the carbon component fluxes (NEP, GPP and Reco) and precipitation are complex at both monthly and annual timescales. Unlike previous studies in dryland ecosystems and other regions (Biederman et al., 2017; Yu et al., 2013), our results showed markedly different linear relationships between carbon fluxes and precipitation among different ecosystems in the HRB. This difference is mainly due to the differences in water

availability between sites within oases and sites outside of oases in the alpine-desert-oasis system. From the view of water budget of basin hydrology, with the scarcity of precipitation in the arid region, water supply for vegetation from surface flows is generally limited; the desert ecosystems distributed at the margins of oases were more sensitive to precipitation than those within oases. For plants within the artificial and natural oasis regions (e.g. cropland and wetland ecosystems), the precipitation is much lower than ET (Li et al., 2018), irrigation and groundwater thus become important sources of water availability that greatly affect the spatial and temporal distributions of soil moisture. Due to the irrigation and groundwater recharges, these ecosystems had high levels of productivity, although they are distributed in a water-limited region. Therefore, annual precipitation, although a common proxy for annual ecosystem water availability (Sala et al., 2012), is not an effective metric for the available water in our study region. In addition to precipitation, soil moisture is commonly recognized as a primary control of both photosynthesis and respiration in dryland ecosystems. However, soil moisture is often measured at different depths at

12

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

Fig. 9. Effects of mean annual precipitation (MAP) and mean annual ET on annual carbon fluxes. (a), (c) and (e) show the linear relationships between annual carbon fluxes and MAP; the dashed black lines indicate the overall fit of all sites; the sites are grouped into inside-oasis (labeled in red, including DMZ, SDZ, NTZ, HYL, HHL, SDQ and LDZ) and outside-oasis (labeled in blue) sites. (b), (d) and (f) show the linear relationships between annual carbon fluxes and annual ET. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 10. Effects of soil moisture on annual carbon fluxes across sites in the Heihe River Basin (HRB). An exponential regression model was used to explore the relationships between the carbon fluxes and soil moisture.

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

different sites. Soil moisture may also be unsynchronized with other ecological constraints (e.g., phenology and energy), which may confound in annual-scale analyses (Biederman et al., 2016). Therefore, we found that soil moisture explained 20–30% of the variance in annual GPP and Reco but had a weak relationship with annual NEP. Temporal variations in soil moisture affected seasonal carbon fluxes. Soil moisture had a significant impact on seasonal carbon fluxes at certain sites located in the upper reaches of the HRB. In addition, our analysis of the interactive effects of soil moisture and temperature on temporal variations in carbon fluxes showed that soil moisture and temperature had a relatively strong joint control on variations in carbon fluxes in arid regions. ET represents the efflux of soil moisture and has long been recognized as a reliable metric of the water available to ecosystems in dryland regions (Schwalm et al., 2010; Ponce-Campos et al., 2013; Biederman et al., 2016). Due to the scarcity of precipitation in dryland regions, the hydrologic niche separation caused by irrigation and groundwater discharge can be a major component of the water balance, leading to a divergence in plant functional traits (Nolan et al., 2018). The remaining precipitation becomes soil moisture and is partly transferred into the atmosphere by ET. Annual ET can account for up to 95% of all water inputs in arid zones (Wilcox and Thurow, 2006). Our study demonstrated that ET could be a better proxy for water availability that drives ecosystem CO2 exchange in the dryland region of northwestern China and is a stronger predictor of ecosystem carbon fluxes than precipitation since ET data accounts for the influence of all types of ecosystem soil water access (Thompson et al., 2011). This finding is consistent with the previous findings on the water-limited ecosystems in North America (Biederman et al., 2017). Both the HRB and the southwestern US have large climatic and topographic gradients, and precipitation is not the only source of the water availability. However, our finding is based on a number of sites encompassing a larger climate gradient from extremely arid to arid to semi-arid or semi-humid regions in northwestern China (and central Asia) and demonstrates that this finding is also applicable to other water-limited regions beyond the southwestern US. Moreover, these two places had different climates. The HRB is dominated by a monsoon climate with large topographic gradients, while the southwestern US is characterized by both Mediterranean and monsoon climates. By contrast, in certain dryland areas in the Southern Hemisphere, carbon fluxes are mainly driven by precipitation (e.g., Cleverly et al., 2013, 2016). Ecosystems in these regions are mainly rain-fed, and the water availability is less influenced by groundwater recharge or irrigation.

4.3. Potential effects of climate change on carbon sequestration in dryland ecosystems Climate change projections have predicted rising MAT in dryland regions (Huang et al., 2016). The projected climatic warming may lead to accelerated dryland expansion (Huang et al., 2016), which would lead to reduced carbon sequestration. With climate change, the semi-arid ecosystems may play a more important role in driving the interannual variability of global carbon cycle than tropical rainforests in the future (Poulter et al., 2014). Our study area has experienced a strong increasing trend in temperature (approximately 0.04 °C/year) and a weak increasing trend in precipitation (approximately 0.16 mm/year) since the 1970s (Li et al., 2018). These increases in temperature and the frequency of extreme rainfall events may induce a decrease in the annual carbon assimilation of the arid region. Meanwhile, human activities (e.g., agricultural management) have had strong effects on the vegetation dynamics of the HRB (Wang and Ma, 2016). With the implementation of ecological rehabilitation and water-saving and water allocation projects in the HRB, a gradual restoration of wetlands and the ecological environment has occurred in the HRB, and the wetland area has increased by 729.3 ha since 2011 (Wang and Ma, 2016). These activities

13

will affect the regional carbon budget and improve the ecological stability in these arid regions. 4.4. Possible implications for ecological sustainability development in arid regions Large climate and topography gradients form unique desert-oasisalpine ecosystems in the arid region of China that plays an important role in maintaining the stability of the ecological structure and functions in arid regions (Zhao and Liu, 2010). Our results revealed significant carbon sinks in the dryland region of northwestern China, the water availability is critical to the spatial variability of carbon fluxes in arid regions. Meanwhile, desert and forest ecosystems had a higher WUE, indicating that these ecosystems gained more carbon by consuming less water. This pattern reflects the ecological adaption of these ecosystems to the water-deficient environment. The relatively higher WUE at the maizecropped site indicates that the planting of maize with plastic film is an effective water-saving technique. These results can inform ecological management in response to climate change in the studied arid region. Our results suggested that water availability is a key control of ecosystem carbon fluxes in arid regions. In arid regions, annual rainfall is much lower than annual potential ET (Newman et al., 2006), and the water supply for vegetation from surface flows (e.g., runoff and rivers) is generally limited. Therefore, groundwater is an important source of water consumption that greatly affects the redistribution of soil moisture in arid regions (Lowry et al., 2011). As the world's largest freshwater resource, groundwater is vital to irrigated agriculture and consequently global food security (Aeschbach-Hertig and Gleeson, 2012). Approximately 90% of the global consumptive water use is for irrigation (Scanlon et al., 2007), and the use of groundwater in irrigation accounts for approximately 43% of the total consumptive irrigation water use (Siebert et al., 2010). In arid regions, groundwater is often the only available water resource to support or expand agricultural production (Aeschbach-Hertig and Gleeson, 2012). With the arid climate conditions, the groundwater level in some places of the HRB that heavily relies on groundwater has dropped significantly, which results in a series of ecological and environmental problems such as desertification (Ma et al., 2015). Consequently, we need to sustainably manage agriculture and reconsider the ecological costs when we choose plant species with a high water demand in the arid region. Measures should be taken to change the planting structure in this arid region. For example, limiting planting of high water-consuming plants and encouraging planting certain desert species that are adapted to water-limited environments. Meanwhile, the government has adopted various environmental protection policies, such as “returning farmland to grassland and wetland”, which can reduce the cropland area and increase the wetland and grassland area, thereby increasing carbon sequestration in the arid regions. In addition, ecological engineering projects such as the “Three-North Shelter Forest Program” (Qiu et al., 2017) can also increase the carbon sequestration in arid terrestrial ecosystems. 5. Conclusions Based on observation data from the regional flux observation network in the HRB, we analyzed the spatial variability and controlling factors of the carbon fluxes across alpine, oasis, and desert ecosystems encompassing a large climate gradient in the dryland region of northwestern China. These ecosystems include both natural ecosystems and managed ecosystems, which showed an important carbon sink behavior through the region. Our results suggest that water availability plays a more important role than temperature in controlling spatial variability in carbon fluxes across alpine-desert-oasis ecosystems in the dryland regions. Due to the irrigation and groundwater recharges, marked differences in precipitation effects were observed between the sites distributed within oases and those outside of oases. ET exhibited stronger relationships with carbon fluxes than precipitation, indicating that ET

14

H. Wang et al. / Science of the Total Environment 697 (2019) 133978

is a better proxy of water availability than is precipitation. Our results have implications for ecological management in response to climate change in the dryland regions. Under climate change, we need to take actions such as limiting planting of high water-consuming plants and encouraging planting certain desert species that are adapted to waterlimited environments. Additionally, the implementation of ecological rehabilitation and water-saving projects can increase carbon sequestration in dryland ecosystems. Acknowledgments This work was supported by the Strategic Priority Research Program of the Chinese Academy of Sciences: Grant no. XDA20100104; the National Natural Science Foundation of China: Grant Nos. 41871250, 41771453 and 91425303; and the China Postdoctoral Science Foundation funded project: Grant No. 2016T90960. J. Xiao was supported by the National Aeronautics and Space Administration (NASA) (Carbon Cycle Science Program: Grant No. NNX14AJ18G; Climate Indicators and Data Products for Future National Climate Assessments: Grant No. NNX16AG61G). We thank all the scientists and engineers who participated in the HiWATER field campaigns, particularly Dr. Shaomin Liu and Ziwei Xu from Beijing Normal University for flux data collection and processing. All the data of this work are available at the data center of the “Integrated research on the eco-hydrological process of the Heihe River Basin” (https://heihe.tpdc.ac.cn). We thank the anonymous reviewers for their constructive comments on an earlier version of the manuscript. Appendix A. Supplementary data. Supplementary data to this article can be found online at https://doi. org/10.1016/j.scitotenv.2019.133978. References Adair, E.C., Parton, W.J., King, J.Y., Brandt, L.A., Lin, Y., 2017. Accounting for photodegradation dramatically improves prediction of carbon losses in dryland systems. Ecosphere 8 (7), e01892. Aeschbach-Hertig, W., Gleeson, T., 2012. Regional strategies for the accelerating global problem of groundwater depletion. Nat. Geosci. 5 (12), 853. Ahlström, A., Raupach, M.R., Schurgers, G., Smith, B., Arneth, A., Jung, M., Reichstein, M., Canadell, J.G., Friedlingstein, P., Jain, A.K., Kato, E., 2015. The dominant role of semiarid ecosystems in the trend and variability of the land CO2 sink. Science 348 (6237), 895–899. Anderson-Teixeira, K.J., Delong, J.P., Fox, A.M., Brese, D.A., Litvak, M.E., 2011. Differential responses of production and respiration to temperature and moisture drive the carbon balance across a climatic gradient in New Mexico. Glob. Change Biol. 17 (1), 410–424. Aubinet, M., Grelle, A., Ibrom, A., Rannik, Ü., Moncrieff, J., Foken, T., Kowalski, A.S., Martin, P.H., Berbigier, P., Bernhofer, C., Clement, R., Elbers, J., Granier, A., Grünwald, T., Morgenstern, K., Pilegaard, K., Rebmann, C., Snijders, W., Valentini, R., Vesala, T., 2000. Estimates of the annual net carbon and water exchange of forests: the EUROFLUX methodology. Adv. Ecol. Res. 30, 113–175. Baldocchi, D., Falge, E., Gu, L., Olson, R., Hollinger, D., Running, S., Anthoni, P., Bernhofer, Ch., Davis, K., Evans, R., Fuentes, J., Goldstein, A., Katul, G., Law, B., Lee, X., Malhi, Y., Fuentes, J., Munger, W., Oechel, W., Paw, U.K.T., Pilegaard, K., Schmid, H.P., Valentini, R., Verma, S., Vesala, T., Wilson, K., Wofsy, S., 2001. FLUXNET: a new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities. Bull. Am. Meteorol. Soc. 82 (11), 2415–2434. Bastin, J.F., Berrahmouni, N., Grainger, A., Maniatis, D., Mollicone, D., Moore, R., Patriarca, C., Picard, N., Sparrow, B., Abraham, E.M., Aloui, K., 2017. The extent of forest in dryland biomes. Science 356 (6338), 635–638. Beringer, J., Hutley, L.B., McHugh, I., Arndt, S.K., Campbell, D., Cleugh, H.A., Cleverly, J., Dios, V.R., Eamus, D., Evans, B., Ewenz, C., Grace, P., Griebel, A., Haverd, V., Hinko-Najera, N., Huete, A., Isaac, P., Kanniah, K., Leuning, R., Liddell, M.J., Macfarlane, C., Meyer, W., Moore, C., Pendall, E., Phillips, A., Phillips, R.L., Prober, S.M., Restrepo-Coupe, N., Rutledge, S., Schroder, I., Silberstein, R., Southall, P., Yee, M.S., Tapper, N.J., van Gorsel, E., Vote, C., Walker, J., Wardlaw, T., 2016. An introduction to the Australian and New Zealand flux tower network – OzFlux. Biogeosciences 13, 5895–5916. Biederman, J.A., Scott, R.L., Goulden, M.L., Vargas, R., Litvak, M.E., Kolb, T.E., Yepez, E.A., Oechel, W.C., Blanken, P.D., Bell, T.W., Garatuza-Payan, J., 2016. Terrestrial carbon balance in a drier world: the effects of water availability in southwestern North America. Glob. Change Biol. 22 (5), 1867–1879. Biederman, J.A., Scott, R.L., Bell, T.W., Bowling, D.R., Dore, S., Garatuza-Payan, J., Kolb, T.E., Krishnan, P., Krofcheck, D.J., Litvak, M.E., Maurer, G.E., 2017. CO2 exchange and evapo-

transpiration across dryland ecosystems of southwestern North America. Glob. Change Biol. 23 (10), 4204–4221. Chapin, F.S., Woodwell, G.M., Randerson, J.T., Rastetter, E.B., Lovett, G.M., Baldocchi, D.D., Clark, D.A., Harmon, M.E., Schimel, D.S., Valentini, R., Wirth, C., Aber, J.D., Cole, J.J., Goulden, M.L., Harden, J.W., Heimann, M., Howarth, R.W., Matson, P.A., McGuire, A.D., Melillo, J.M., Mooney, H.A., Neff, J.C., Houghton, R.A., Pace, M.L., Ryan, M.G., Running, S.W., Sala, O.E., Schlesinger, W.H., Schulze, E.D., 2006. Reconciling carboncycle concepts, terminology, and methods. Ecosystems 9, 1041–1050. Chen, Z., Yu, G.R., Ge, J.P., Sun, X.M., Hirano, T., Saigusa, N., Wang, Q.F., Zhu, X.J., Zhang, Y.P., Zhang, J.H., Yan, J.H., Wang, H.M., Zhao, L., Wang, Y.F., Shi, P.L., Zhao, F.H., 2013. Temperature and precipitation control of the spatial variation of terrestrial ecosystem carbon exchange in the Asian region. Agric. For. Meteorol. 182, 266–276. Chen, Z., Yu, G., Zhu, X., Wang, Q., Niu, S., Hu, Z., 2015. Covariation between gross primary production and ecosystem respiration across space and the underlying mechanisms: a global synthesis. Agric. For. Meteorol. 203, 180–190. Cheng, G., Li, X., Zhao, W., Xu, Z., Feng, Q., Xiao, S., Xiao, H., 2014. Integrated study of the water–ecosystem–economy in the Heihe River Basin. Natl. Sci. Rev. 1 (3), 413–428. Chu, H., Baldocchi, D.D., John, R., Wolf, S., Reichstein, M., 2017. Fluxes all of the time? A primer on the temporal representativeness of FLUXNET. J. Geophys. Res. Biogeosci. 122 (2), 289–307. Ciais, P., Peylin, P., Bousquet, P., 2000. Regional biospheric carbon fluxes as inferred from atmospheric CO2 measurements. Ecol. Appl. 10 (6), 1574–1589. Cleverly, J., Boulain, N., Villalobos-Vega, R., Grant, N., Faux, R., Wood, C., Cook, P.G., Yu, Q., Leigh, A., Eamus, D., 2013. Dynamics of component carbon fluxes in a semi-arid Acacia woodland, central Australia. J. Geophys. Res. Biogeosci. 118, 1168–1185. Cleverly, J., Eamus, D., Coupe, N.R., Chen, C., Maes, W., Li, L., Faux, R., Santini, N.S., Rumman, R., Yu, Q., Huete, A., 2016. Soil moisture controls on phenology and productivity in a semi-arid critical zone. Sci. Total Environ. 568, 1227–1237. Diekkruger, B., Nörtersheuser, P., Richter, O., 1995. Modeling pesticide dynamics of a loam site using HERBSIM and SIMULAT. Ecol. Model. 81, 111–119. Eamus, D., Cleverly, J., Boulain, N., Grant, N., Faux, R., Villalobos-Vega, R., 2013. Carbon and water fluxes in an arid-zone Acacia savanna woodland: an analyses of seasonal patterns and responses to rainfall events. Agric. For. Meteorol. 182, 225–238. Eamus, D., Huete, A., Cleverly, J., Nolan, R.H., Ma, X., Tarin, T., Santini, N.S., 2016. Mulga, a major tropical dry open forest of Australia: recent insights to carbon and water fluxes. Environ. Res. Lett. 11, 125011. Elmar, M., Veenendaal, O., 2004. Seasonal variation in energy fluxes and carbon dioxide exchange for a broadleaved semi-arid savanna (mopane woodland) in southern Africa. Glob. Change Biol. 10, 318–328. Falge, E., Baldocchi, D.D., Olson, R.J., Anthoni, P.M., Aubinet, M., Aubinet, M., Bernhofer, C., Burba, G., Ceulemans, R., Clement, R., Dolman, H., Granier, A., Gross, P., Grunwald, T., Hollinger, D.Y., Jensen, N.O., Katul, G.G., Keronen, P., Kowalski, A.S., Lai, C.T., Law, B.E., Meyers, T.P., Moncrieff, H., Moors, E.J., Munger, J.W., Pilegaard, K., Rannik, U., Rebmann, C., Suyker, A.E., Tenhunen, J., Tu, K.P., Verma, S.B., Vesala, T., Wilson, K.B., Wofsy, S.C., 2001. Gap filling strategies for defensible annual sums of net ecosystem exchange. Agric. For. Meteorol. 107 (1), 43–69. Fei, X., Song, Q., Zhang, Y., Liu, Y., Sha, L., Yu, G., Zhang, L., Duan, C., Deng, Y., Wu, C., Lu, Z., Luo, K., Chen, A., Xu, K., Liu, W., Huang, H., Jin, Y., Zhou, R., Grace, J., 2018. Carbon exchanges and their responses to temperature and precipitation in forest ecosystems in Yunnan, Southwest China. Sci. Total Environ. 616, 824–840. Gao, Y., Li, X., Liu, L., Jia, R., Yang, H., Li, G., Wei, Y., 2012. Seasonal variation of carbon exchange from a revegetation area in a Chinese desert. Agric. For. Meteorol. 156, 134–142. González-Megías, A., Menéndez, R., 2012. Climate change effects on above- and belowground interactions in a dryland ecosystem. Phil. Trans. R. Soc. B. 367, 3115–3124. Hastings, S.J., Oechel, W.C., Muhlia-Melo, A., 2005. Diurnal, seasonal and annual variation in the net ecosystem CO2 exchange of a desert shrub community (Sarcocaulescent) in Baja California, Mexico. Glob. Change Biol. 11, 927–939. Houghton, R.A., 2007. Balancing the global carbon budget. Annu. Rev. Earth Pl. Sc. 35, 313–347. Huang, J., Yu, H., Guan, X., Wang, G., Guo, R., 2016. Accelerated dryland expansion under climate change. Nat. Clim. Chang. 6 (2), 166. Huxman, T.E., Smith, M.D., Fay, P.A., Knapp, A.K., Shaw, M.R., Loik, M.E., Smith, S.D., Tissue, D.T., Zak, J.C., Weltzin, J.F., Pockman, W.T., Sala, O.E., Haddad, B.M., Harte, J., Koch, G.W., Schwinning, S., Small, E.E., Williams, D.G., 2004. Convergence across biomes to a common rain-use efficiency. Nature 429 (6992), 651. Kondo, M., Saitoh, T.M., Sato, H., Ichii, K., 2017. Comprehensive synthesis of spatial variability in carbon flux across monsoon Asian forests. Agric. For. Meteorol. 232, 623–634. Law, B.E., Falge, E., Gu, L., Baldocchi, D.D., Bakwin, P., Berbigier, P., Davis, K., Dolman, A.J., Falk, M., Fuentes, J.D., Goldstein, A., Granier, A., Grelle, A., Hollinger, D., Janssens, I.A., Jarvis, P., Jensen, N.O., Katul, G., Mahli, Y., Matteucci, G., Meyers, T., Monson, R., Munger, W., Oechel, W., Olson, R., Pilegaard, K., Paw, U., K.T., Thorgeirsson, H., Valentini, R., Verma, S., Vesala, T., Wilson, K., Wofsy, S.C., 2002. Environmental controls over carbon dioxide and water vapor exchange of terrestrial vegetation. Agric. For. Meteorol.. 113(1), 97–120. Li, X., Cheng, G.D., Liu, S.M., Xiao, Q., Ma, M.G., Jin, R., Che, T., Liu, Q.H., Wang, W.Z., Qi, Y., Wen, J.G., Li, H.Y., Zhu, G.F., Guo, J.W., Ran, Y.H., Wang, S.G., Zhu, Z.L., Zhou, J., Hu, X.L., Xu, Z.W., 2013. Heihe watershed allied telemetry experimental research (HiWATER): scientific objectives and experimental design. Bull. Am. Meteorol. Soc. 94 (8), 1145–1160. Li, X., Yang, K., Zhou, Y., 2016. Progress in the study of oasis-desert interactions. Agric. For. Meteorol. 230–231, 1–7. Li, X., Liu, S.M., Xiao, Q., Ma, M.G., Jin, R., Che, T., Wang, W.Z., Hu, X.L., Xu, Z.W., Wen, J.G., Wang, L.X., 2017. A multiscale dataset for understanding complex eco-hydrological processes in a heterogeneous oasis system. Scientific Data 4, 170083.

H. Wang et al. / Science of the Total Environment 697 (2019) 133978 Li, X., Cheng, G., Ge, Y., Li, H., Han, F., Hu, X., Tian, W., Tian, Y., Pan, X., Nian, Y., Zhang, Y., Ran, Y., Zheng, Y., Gao, B., Yang, D., Zheng, C., Wang, X., Liu, S., Cai, X., 2018. Hydrological cycle in the Heihe River basin and its implication for water resource management in endorheic basins. J. Geophys. Res. Atmos. 123 (2), 890–914. Liu, H., Tu, G., Fu, C., 2008. Three-year variations of water, energy and CO2 fluxes of cropland and degraded grassland surfaces in a semi-arid area of Northeastern China. Adv. Atmos. Sci. 25, 1009–1020. Liu, S.M., Xu, Z.W., Wang, W.Z., Bai, J., Jia, Z., Zhu, M., Wang, J.M., 2011. A comparison of eddy-covariance and large aperture scintillometer measurements with respect to the energy balance closure problem. Hydrol. Earth Syst. Sci. 15 (4), 1291–1306. Liu, S.M., Li, X., Xu, Z.W., Xiao, Q., Ma, M.G., Liu, Q.H., Jin, R., Guo, J.W., Wang, L.X., Wang, W.Z., Qi, Y., Li, H.Y., Xu, T.R., Ran, Y.H., Hu, X.L., Shi, S.J., Zhu, Z.L., Tan, J.L., Zhang, Y., Ren, Z., 2018. The Heihe integrated observatory network: a basin-scale land surface processes observatory in China. Vadose Zone J. 17 (1), 180072. López-Ballesteros, A., Serrano-Ortiz, P., Sánchez-Cañete, E.P., Oyonarte, C., Kowalski, A.S., Pérez-Priego, Ó., Domingo, F., 2016. Enhancement of the net CO2 release of a semiarid grassland in SE Spain by rain pulses. J. Geophys. Res. Biogeosci. 121, 52–66. Lowry, C.S., Loheide, S.P., Moore, C.E., Lundquist, J.D., 2011. Groundwater controls on vegetation composition and patterning in mountain meadows. Water Resour. Res. 47, W00J11. Lu, W., Xiao, J., Liu, F., Zhang, Y., Liu, C., Lin, G., 2017. Contrasting ecosystem CO2 fluxes of inland and coastal wetlands: a meta-analysis of eddy covariance data. Glob. Change Biol. 23, 1180–1198. Ma, X., Huete, A., Moran, S., Ponce-Campos, G., Eamus, D., 2015. Abrupt shifts in phenology and vegetation productivity under climate extremes. J. Geophys. Res. Biogeosci. 120, 2036–2052. Michaletz, S.T., Kerkhoff, A.J., Enquist, B.J., 2018. Drivers of terrestrial plant production across broad geographical gradients. Glob. Ecol. Biogeogr. 27 (2), 166–174. Mielnick, P., Dugas, W.A., Mitchell, K., Havstad, K., 2005. Long-term measurements of CO2 flux and evapotranspiration in a Chihuahuan desert grassland. J. Arid Environ. 60, 423–436. Misson, L., Rocheteau, A., Rambal, S., Ourcival, J.M., Limousin, J.M., Rodriguez, R., 2010. Functional changes in the control of carbon fluxes after 3 years of increased drought in a Mediterranean evergreen forest? Glob. Chang. Biol. 6, 2461–2475. Misson, L., Degueldre, D., Collin, C., Rodriguez, R., Rocheteau, A., Ourcival, J.M., Rambal, S., 2011. Phenological responses to extreme droughts in a Mediterranean forest. Glob. Chang. Biol. 17 (2), 1036–1048. Mizoguchi, Y., Miyata, A., Ohtani, Y., Hirata, R., Yuta, S., 2009. A review of tower flux observation sites in Asia. J. For. Res. 14, 1–9. Moore, C.J., 1986. Frequency response corrections for eddy correlation systems. Bound.Layer Meteorol. 37 (1–2), 17–35. Newman, B.D., Wilcox, B.P., Archer, S.R., Breshears, D.D., Dahm, C.N., Duffy, C.J., McDowell, N.G., Phillips, F.M., Scanlon, B.R., Vivoni, E.R., 2006. Ecohydrology of water-limited environments: a scientific vision. Water Resour. Res. 42 (6). Nolan, R.H., Tarin, T., Rumman, R., Cleverly, J., Fairweather, K.A., Zolfaghar, S., Santini, N.S., O'Grady, A.P., Eamus, D., 2018. Contrasting ecophysiology of two widespread arid zone tree species with differing access to water resources. J. Arid Environ. 153, 1–10. Papale, D., Reichstein, M., Aubinet, M., Canfora, E., Bernhofer, C., Kutsch。, W., Longdoz, B., Rambal, S., Valentini, R., Vesala, T., Yakir, D., 2006. Towards a standardized processing of net ecosystem exchange measured with eddy covariance technique: algorithms and uncertainty estimation. Biogeosciences 3, 571–583. Ponce-Campos, G.E., Moran, M.S., Huete, A., Zhang, Y., Bresloff, C., Huxman, T.E., Eamus, D., Bosch, D.D., Buda, A.R., Gunter, S.A., Scalley, T.H., Kitchen, S.G., McClaran, M.P., McNab, H., Montoya, D.S., Morgan, J.A., Peters, D.P.C., Sadler, E.J., Seyfried, M.S., Starks, P.J., 2013. Ecosystem resilience despite large-scale altered hydroclimatic conditions. Nature 494 (7437), 349. Poulter, B., Frank, D., Ciais, P., Myneni, R.B., Andela, N., Bi, J., Broquet, G., Canadell, J.G., Chevallier, F., Liu, Y.Y., Running, S.W., 2014. Contribution of semi-arid ecosystems to interannual variability of the global carbon cycle, Nature. 509 (7502), 600. Puigdefábregas, J., Mendizábal, T., 1998. Perspectives on desertification: western Mediterranean. J. Arid Environ. 39, 209–224. Qiu, B., Chen, G., Tang, Z., Lu, D., Wang, Z., Chen, C., 2017. Assessing the Three-North Shelter Forest program in China by a novel framework for characterizing vegetation changes. ISPRS J. Photogramm. 133, 75–88. Rambal, S., Ourcival, J.M., Joffre, R., Mouillot, F., Nouvellon, Y., Reichstein, M., Rocheteau, A., 2003. Drought controls over conductance and assimilation of a Mediterranean evergreen ecosystem: scaling from leaf to canopy. Glob. Chang. Biol. 9, 1813–1824. Reichstein, M., Tenhunen, J.D., Roupsard, O., Ourcival, J.M., Rambal, S., Miglietta, F., Peressotti, A., Pecchiari, M., Tirone, G., Valentini, R., 2002. Severe drought effects on ecosystem CO2 and H2O fluxes at three Mediterranean evergreen sites: revision of current hypotheses? Glob. Change Biol. 8, 999–1017. Reichstein, M., Tenhunen, J., Roupsard, O., Ourcival, J.-M., Rambal, S., Miglietta, F., Peressotti, A., Pecchiari, M., Tirone, G., Valentini, R., 2003. Inverse modeling of seasonal drought effects on canopy CO2/H2O exchange in three Mediterranean ecosystems. J. Geophys. Res. 108 (D23), 4726. Reichstein, M., Falge, E., Baldocchi, D., Papale, D., Aubinet, M., Berbigier, P., Bernhofer, C., Buchmann, N., Gilmanov, T., Granier, A., Grünwald, T., Havrankova, K., Ilvesniemi, H., Janous, D., Knohl, A., Laurila, T., Lohila, A., Loustau, D., Matteucci, G., Meyers, T., Miglietta, F., Ourcival, J.M., Pumpanen, J., Rambal, S., Rotenberg, E., Sanz, M., Tenhunen, J., Seufert, G., Vaccari, F., Vesala, T., Yakir, D., Valentini, R., 2005. On the separation of net ecosystem exchange into assimilation and ecosystem respiration: review and improved algorithm. Glob. Change Biol. 11, 1424–1439.

15

Rutledge, S., Campbell, D.I., Baldocchi, D.D., Schipper, L.A., 2010. Photodegradation leads to increased carbon dioxide losses from terrestrial organic matter. Glob Change Biol 16 (11), 3065–3074. Sala, O.E., Gherardi, L.A., Reichmann, L., Jobbagy, E., Peters, D., 2012. Legacies of precipitation fluctuations on primary production: theory and data synthesis. Phil. Trans. R. Soc. B. 367, 3135–3144. Scanlon, B.R., Jolly, I., Sophocleous, M., Zhang, L., 2007. Global impacts of conversions from natural to agricultural ecosystems on water resources: quantity versus quality. Water Resour. Res. 43 (3). Schimel, D.S., 2010. Drylands in the earth system. Science 327 (5964), 418–419. Schwalm, C.R., Williams, C.A., Schaefer, K., Arneth, A., Bonal, D., Buchmann, N., Chen, J., Law, B.E., Lindroth, A., Luyssaert, S., Reichstein, M., Richardson, A.D., 2010. Assimilation exceeds respiration sensitivity to drought: a FLUXNET synthesis. Glob. Change Biol. 16 (2), 657–670. Scott, R.L., Serranoortiz, P., Domingo, F., Hamerlynck, E.P., Kowalski, A.S., 2012. Commonalities of carbon dioxide exchange in semiarid regions with monsoon and Mediterranean climates. J. Arid Environ. 84, 71–79. Serrano-Ortiz, P., Roland, M., Sanchez-Moral, S., Janssens, I.A., Domingo, F., Godderis, Y., Kowalski, A.S., 2010. Hidden, abiotic CO2 flows and gaseous reservoirs in the terrestrial carbon cycle: review and perspectives. Agric. For. Meteorol. 150, 321–329. Siebert, S., Burke, J., Faures, J.M., Frenken, K., Hoogeveen, J., Doll, P., Portmann, F.T., 2010. Groundwater use for irrigation—a global inventory. Hydrol. Earth Syst. Sci. 14, 1863–1880. Sims, D.A., Rahman, A.F., Cordova, V.D., El-Masri, B.Z., Baldocchi, D.D., Flanagan, L.B., Goldstein, A.H., Hollinger, D.Y., Misson, L., Monson, R.K., Oechel, W.C., 2006. On the use of MODIS EVI to assess gross primary productivity of North American ecosystems. J. Geophys. Res-Biogeo. 111 (G4). Stange, C.F., 2007. A novel approach to combine response functions in ecological process modelling. Ecol. Model. 204 (3–4), 547–552. Tagesson, T., Fensholt, R., Cappelaere, B., Mougin, E., Horion, S., Kergoat, L., Nieto, H., Mbow, C., Ehammer, A., Demarty, J., Ardö, J., 2016. Spatiotemporal variability in carbon exchange fluxes across the Sahel. Agric. For. Meteorol. 226, 108–118. Thompson, S.E., Harman, C.J., Konings, A.G., Sivapalan, M., Neal, A., Troch, P.A., 2011. Comparative hydrology across AmeriFlux sites: the variable roles of climate, vegetation, and groundwater. Water Resour. Res. 47, W00J07. Vargas, R., Sonnentag, O., Abramowitz, G., Carrara, A., Chen, J.M., Ciais, P., Correia, A., Keenan, T.F., Kobayashi, H., Ourcival, J., Papale, D., Pearson, D., Pereira, J.S., Piao, S.L., Rambal, S., Baldocchi, D.D., 2013. Drought influences the accuracy of simulated ecosystem fluxes: a model-data meta-analysis for Mediterranean oak woodlands. Ecosystems 16 (5), 749–764. Wang, H., Ma, M., 2016. Impacts of climate change and anthropogenic activities on the ecological restoration of wetlands in the arid regions of China. Energies 9 (3), 166. Wang, X.P., Kang, E.S., Zhang, J.G., Li, X.R., 2004. Evapotranspiration of Artemisia ordosia vegetation in stabilized arid desert dune in Shapotou, China. Arid Land Res. Manage. 18, 63–76. Wang, H., Ma, M., Wang, X., Yuan, W., Song, Y., Tan, J., Huang, G., 2013. Seasonal variation of vegetation productivity over an alpine meadow in the Qinghai–Tibet Plateau in China: modeling the interactions of vegetation productivity, phenology, and the soil freeze–thaw process. Ecol. Res. 28 (2), 271–282. Wang, H., Ma, M., Geng, L., 2015. Monitoring the recent trend of aeolian desertification using Landsat TM and Landsat 8 imagery on the north-east Qinghai–Tibet Plateau in the Qinghai Lake basin. Nat. Hazards 79 (3), 1753–1772. Wang, H., Li, X., Ma, M., Geng, L., 2019. Improving estimation of Gross primary production in dryland ecosystems by a model-data fusion approach. Remote Sens. 11, 225. Wen, X.F., Wang, H.M., Wang, J.L., Yu, G.R., Sun, X.M., 2010. Ecosystem carbon exchanges of a subtropical evergreen coniferous plantation subjected to seasonal drought, 2003–2007. Biogeosciences 7 (1), 357–369. Wilcox, B.P., Thurow, T.L., 2006. Emerging issues in rangeland ecohydrology: vegetation change and the water cycle. Rangeland Ecol. Manage. 59 (2), 220–224. Wutzler, T., Lucas-Moffat, A., Migliavacca, M., Knauer, J., Sickel, K., Šigut, L., Menzer, O., Reichstein, M., 2018. Basic and extensible post-processing of eddy covariance flux data with REddyProc. Biogeosciences 15, 5015–5030. Xiao, J.F., Sun, G., Chen, J., Chen, H., Chen, S., Dong, G., Gao, S., Guo, H., Guo, J., Han, S., Kato, T., Li, Y., Lin, G., Lu, W., Ma, M., McNulty, S., Shao, C., Wang, X., Xie, X., Zhang, X., Zhang, Z., Zhao, B., Zhou, G., Zhou, J., 2013. Carbon fluxes, evapotranspiration, and water use efficiency of terrestrial ecosystems in China. Agric. For. Meteorol. 182-183, 76–90. Xu, Z.W., Liu, S.M., Li, X., Shi, S.J., Wang, J.M., Zhu, Z.L., Xu, T.R., Wang, W.Z., Ma, M.G., 2013. Intercomparison of surface energy flux measurement systems used during the HiWATER-MUSOEXE. J. Geophys. Res. Atmos. 118, 13140–13157. Yu, G.R., Zhu, X.J., Fu, Y.L., He, H.L., Wang, Q.F., Wen, X.F., Li, X.R., Zhang, L.M., Zhang, L., Su, W., Li, S.G., Sun, X.M., Zhang, Y.P., Zhang, J.H., Yan, J.H., Wang, H.M., Zhou, G.S., Jia, B.R., Xiang, W.H., Li, Y.N., Zhao, L., Wang, Y.F., Shi, P.L., Chen, S.P., Xin, X.P., Zhao, F.H., Wang, Y.Y., Tong, C.L., 2013. Spatial patterns and climate drivers of carbon fluxes in terrestrial ecosystems of China. Glob. Change Biol. 19 (3), 798–810. Zhao, W.Z., Liu, B., 2010. The response of sap flow in shrubs to rainfall pulses in the desert region of China. Agric. For. Meteorol. 150, 1297–1306. Zhou, Y.Z., Li, X., 2019. Energy balance closures in diverse ecosystems of an endorheic river basin. Agric. For. Meteorol. 274, 118–131. Zhou, Y.Z., Li, D., Li, X., 2019. The effects of surface heterogeneity scale on the flux imbalance under free convection. J. Geophys. Res. Atmos. 124 (2018JD029550).