Plant transpiration plays a key role in sand-binding zones, but obtaining accurate estimates at an integrated leaf-individual-canopy scale is difficult. In this study, transpiration rates of two typical xerophytic shrubs, Caragana korshinskii and Artemisia ordosica, were investigated during the growing season (April–October) from 2008 to 2012 in the Tengger Desert, a revegetated desert area in China. Gas exchange techniques, sap flow measurements, and the crop evapotranspiration minus micro-lysimeter method were used to evaluate plant transpiration. Transpiration data were subsequently compared with the dynamical normalized leaf area-based extrapolation. The results indicated that at leaf level, the transpiration rates of C. korshinskii and A. ordosica were 2.67 and 4.51 mmol H2O m−2s−1, respectively. The sap flow rates were 0.071 and 0.086 g h−1cm−2 at the tree level, and the transpiration rates were 0.42 and 0.35 mm d−1 at the stand level. The total seasonal transpiration of the two xerophytic shrubs reached 71.79 and 55.62 mm, representing approximately 48.4 and 37.5% of the total rainfall over this period. Direct measurements of plant transpiration and upscaling transpiration from leaf level to the stand level exhibited good correspondence, which verified that leaf area was a reliable representation of scaled transpiration, especially in arid desert regions.

INTRODUCTION

Artificial vegetation restoration is considered to be one of the most effective ways to combat desertification and land degradation in arid desert areas (Wang 2004). Xerophytic shrubs, such as Caragana korshinskii and Artemisia ordosica, have been planted at the southeastern fringe of the Tengger Desert in Western China since 1956 (Li et al. 2004). The implementation of this strategy has progressed remarkably over the past years and has played a large role in the control of soil erosion, land degradation, and desertification, as well as in grassland reconstruction (Li et al. 2007). However, restoration implementation still faces a number of problems, such as poor understanding of water requirements of desert-living plants. Furthermore, inappropriate tree selection and planting could lead to a reduction in groundwater and the death of sand-binding vegetation in certain regions (Li 2005; Ford et al. 2007; Xia et al. 2008). Scientists have sought to solve this problem and have developed reliable techniques for estimating the total amount of water transpired by plants (Kelliher et al. 1992; Granier et al. 1996; Kumagai et al. 2004; Dierick & Holscher 2009). Wang et al. (2004a, b) have studied the evapotranspiration of C. korshinskii and A. ordosica using the auto-weighing lysimeter method, and the results have demonstrated that cumulative precipitation and cumulative evapotranspiration were linearly correlated. Huang et al. (2010) have studied the sap flow rate of A. ordosica and its correlation with meteorological factors. Such studies are, however, still scattered and unsystematic and, generally, do not involve scaling problems. Evaluating transpiration over a range of temporal and spatial scales remains a challenge (Tanaka et al. 2008; Du et al. 2011), especially in desert areas, primarily because of the dispersed patchy vegetation distribution and the spatially heterogeneous terrain (Domingo et al. 1999).

At present, a number of techniques are available for using field measurements to estimate daily stand transpiration rates through appropriate scaling methodologies (Chabot et al. 2005). Transpiration in potted plants (Dugas et al. 1993) and herbaceous species (Senock & Ham 1995) is often estimated using leaf-level gas exchange techniques, and stand water use in forests is usually scaled from the sap flow of individual trees by diameter or sapwood area (Hatton & Wu 1995; Granier et al. 1996). However, the suitability of these methods in revegetated desert areas remains unknown, and few studies have considered the scaling issues of transpiration estimates from an integrated leaf-individual-canopy scaling perspective. Furthermore, when scaling up an individual plant or tree transpiration to area-average estimates of stand water usage, various scaling factors have been proposed, ranging from tree basal area (Teskey & Sheriff 1996) and stem diameter at breast height (DBH) (Vertessy et al. 1997) to leaf area (Hatton & Wu 1995) and sapwood area (Dunn & Connor 1993). Hatton & Wu (1995) suggested that the xylem transfusion cross-sectional area, leaf area, DBH and the area occupied by a single tree are all reliable spatially derived scaling factors, of which the leaf area is the most credible variable for the conversion of water consumption from a single tree to the scale of a forest stand (Oren et al. 1999: Granier et al. 2000), especially in arid desert areas (Li et al. 2013). Furthermore, the deviation in the upscaling transpiration was reduced by almost half using leaf area as a basis compared with the above-mentioned other scaling factors (Ham et al. 1990; Nicolas et al. 2005). Leaf area is, however, a dynamic variable that depends on the situation and temporal aspects, which have often been ignored in previous studies (Chabot et al. 2005; Delzon & Loustau 2005).

This study is based on long-term measurements from April 2008 to October 2012. During this time period, gas exchange techniques with a portable photosynthesis system, sap flow measurements, and the crop evapotranspiration minus micro-lysimeter method were used to evaluate the transpiration of sand-binding plants at different scales. The specific objectives of the study were: (1) to quantify the daily, seasonal, and annual transpiration patterns of C. korshinskii and A. ordosica; and (2) to scale up transpiration from the leaf level to the stand level with dynamical leaf area-based extrapolation. We expected that the leaf area would be a standard scaling factor for upscaling transpiration in arid desert regions.

MATERIAL AND METHODS

Study area

The study was conducted in the Soil Water Balance Experimental Fields of the Shapotou Desert Experimental Research Station, Chinese Academy of Sciences. The research station borders the Tengger Desert, the fourth largest desert in China, located in the central part of West China (37°27′ N, 104°57′ E). The climate at the site is characterized by abundant sunshine and low relative humidity. The minimum average monthly relative humidity is 33% in April, and the maximum is 54.9% in August. The elevation is 1,330 m above sea level, with a mean annual precipitation of 188.2 mm (according to meteorological records from 1956 to 2009), falling predominantly from June to September. The mean annual temperature is 9.6 °C, and the mean monthly temperatures are −6.9 °C in January and 24.3 °C in July. The potential evapotranspiration during the growing season (May–September) is from 2,300 to 2,500 mm (observed by evaporation pan of type E-601). The windy season lasts from September to April, with an average wind velocity of 2.6 m s−1. The soil surface is loose, poorly aggregated mobile sand with a moisture content of 2 to 3%.

To protect the Baotou–Lanzhou railway line from sand burial, a non-irrigated vegetation protection system was established in 1956. After 50 years, the initial shrub plantings (among them C. korshinskii, Artemisia ordosica, Hedysarum scoparum, C. intermedia, Calligonum arborescens, and Atraphaxis bracteata) were gradually replaced with dominant shrubs, such as A. ordosica, C. korshinskii, Hedysarum scoparium, and herbaceous plants (Li et al. 2004). Two 10 m × 10 m sample plots in the revegetation system were selected in 2008 to investigate basic plant characteristics, such as species, age, density, height, and crown-width (Table 1).

Table 1

Results of agronomic measurements carried out on 17 May 2008

Vegetation Age (year) Density (individuals/100 m2Height (m) Crown-width (m2Basal stem diameter (cm) 
C. korshinskii 25 1.43 ± 0.496 1.6 ± 0.836 4.47 ± 1.349 
A. ordosica 123 0.68 ± 0.154 0.9 ± 0.261 2.15 ± 0.827 
Vegetation Age (year) Density (individuals/100 m2Height (m) Crown-width (m2Basal stem diameter (cm) 
C. korshinskii 25 1.43 ± 0.496 1.6 ± 0.836 4.47 ± 1.349 
A. ordosica 123 0.68 ± 0.154 0.9 ± 0.261 2.15 ± 0.827 

Plant transpiration measurement

Six C. korshinskii and six A. ordosica shrubs in our plot were selected for long-term monitoring and local observation. The stem heat balance (SHB) technique was used to measure sap flow rate and to determine whole-plant transpiration (Sakuratani et al. 1999; Chabot et al. 2005). Twelve sap flow gauges, ranging from SGA2 to SGB19 (Dynamax Inc., Houston, TX, USA), were placed at the base of the stem of each plant to continuously measure transpiration (Table 2). Thirty-minute averages of the flow rate were recorded by a data logger (CR1000; Campbell Scientific, Inc., Logan, UT, USA). During the growing season (April–October) from 2008 to 2012, diurnal transpiration rates were measured monthly from each sensor-wrapped stem using an LI-6400 photosynthesis system (Li-Cor, Lincoln, NE, USA) with a 6 cm2 clamp-on leaf cuvette. Two to four leaves adjacent to the quantum sensors were selected and all leaves were from a south-facing canopy position at a height of 0.5 to 1 m. Each diurnal measurement began at 8:00 and ended at 20:00 (Beijing local time), with a 1 h interval. A total of 34 diurnal measurements were made during the 5-year study period.

Table 2

Selected sensor type and stem area

C. korshinskii Sensors Wrapped stem area (cm2Artemisia ordosica Sensors Wrapped stem area (cm2
No. 1 SGA13 1.3287 No. 7 SGA5 0.2229 
No. 2 SGB19 3.1212 No. 8 SGB5 0.1877 
No. 3 SGB16 2.2981 No. 9 SGA13 1.3369 
No. 4 SGA9 0.8507 No. 10 SGA10 1.2099 
No. 5 SGB19 1.5628 No. 11 SGA2 0.0669 
No. 6 SGA10 0.94985 No. 12 SGA3 0.05723 
C. korshinskii Sensors Wrapped stem area (cm2Artemisia ordosica Sensors Wrapped stem area (cm2
No. 1 SGA13 1.3287 No. 7 SGA5 0.2229 
No. 2 SGB19 3.1212 No. 8 SGB5 0.1877 
No. 3 SGB16 2.2981 No. 9 SGA13 1.3369 
No. 4 SGA9 0.8507 No. 10 SGA10 1.2099 
No. 5 SGB19 1.5628 No. 11 SGA2 0.0669 
No. 6 SGA10 0.94985 No. 12 SGA3 0.05723 

Leaf area index (LAI) measurement

The leaf areas of the sensor-wrapped plants (Nos. 1–12) were measured twice monthly during the growing season using image analysis. With a standard scale board, the images were obtained using a digital camera (Canon Power Shot A550), as seen in Figure 1. The digital images were processed using Matlab 7.0 (MathWorks, Natick, MA, USA) software, and the actual measured areas were obtained with the proportional calibrated figure area; for more details refer to Rico-García et al. (2009). The whole plant LAI (m2/m2) was measured using LAI-2000 Plant Canopy Analyzer (PCA; Li-Cor). PCA measurements were performed for each plant during our experimental period using the same unit. In each measurement, a measurement cycle consisted of a reference measurement and four below-canopy readings. The reference measurements were collected in large clearings or open areas near the experimental plots at the beginning of each cycle. The below-canopy measurements were carried out for each plant at a fixed position and were taken at the same level as the litter traps. The fish-eye lens of the instrument was covered by a view cap with a 90° opening to ensure that the reference measurements were not influenced by the trees surrounding the clearings or by the operator. Before the measurements were taken, the LAI 2000 was calibrated directly by collecting the whole leaves of plants. The LiCor PCA was used after the sun set every evening to avoid confusing brightly sunlit leaves for gaps, and the data were recomputed with the FV2000 LI-COR software built into the instrument. Finally, the actual LAI value was the difference between measurements during the growing season and those taken before leaf emergence (branch area index).

Figure 1

Image capture of the sensor-wrapped stems of C. korshinskii (a) and A. ordosica (b).

Figure 1

Image capture of the sensor-wrapped stems of C. korshinskii (a) and A. ordosica (b).

Meteorological and soil evaporation measurements

Precipitation was recorded using a tipping bucket-type rain gauge (Casella). Meteorological factors, such as net radiation (CNR-1; Kipp & Zonen, Delft, The Netherlands) at 2 m height, soil heat flux density (HFP-01; Hukseflux Thermal Sensors, Delft, The Netherlands) at 0.05 m below the soil surface, air temperature and relative humidity (HMP45C; Vaisala Inc., Helsinki, Finland) at 2 m height, and wind speed (CSAT3; Campbell Scientific, Inc.) at 2 m height were measured at the micrometeorological tower, which was located approximately 30 m from the sample plot. The data were measured every 10 min and averaged every 30 min, and recorded on Campbell CR3000 data-loggers. Soil evaporation was measured with micro-lysimeters (Zhang et al. 2008), which were made using PVC pipes that were 30 cm high and 10 cm in internal diameter. The bottom of each micro-lysimeter was capped with steel grids permitting the free drainage of water. Three micro-lysimeters were buried in three directions (120°) from each C. korshinskii and A. ordosica plant. Thus, mean soil evaporation was determined using the variation of average soil column weights. During the no-rainfall period, continuous measurements were taken every day during the growing season of C. korshinskii and A. ordosica, but during rainy days or missed data, measurements were calculated by interpolation, based on the data before and after the period of missing data.

Simulated canopy transpiration

The reference transpiration (ET) of the plant canopy was calculated following the FAO Penman–Monteith equation, according to standard FAO procedure (Allen et al. 1998) 
formula
1
where ET0 is the reference ET (mm d−1), Rn is the net radiation at the plant surface (MJ m−2day−1), G is the soil heat flux density (MJ m−2day-1), T is the air temperature at 2 m height (°C), u2 is the wind speed at 2 m height (m s−1), es is the saturation vapor pressure (kPa), ea is the actual vapor pressure (kPa), esea is the saturation vapor pressure deficit (VPD; kPa), Δ is the slope vapor pressure curve (kPa °C−1), and γ is the psychrometric constant (kPa °C−1). ET was calculated following the classical method: 
formula
2
where Kc is the crop coefficient. Kc was deduced using the auto-weighing lysimeter method to determine the average evapotranspiration rates of C. korshinskii, A. ordosica, and sand (1.09, 1.01 and 0.68 mm d−1, respectively) for the 6-year intermittent study period (1990–1995). Thus the Kc was 0.38 for C. korshinskii and 0.32 for A. ordosica in this study (Li et al. 2013). Next, the mean transpiration of the plants canopy was calculated as: 
formula
3
where Es (mm d−1) is the mean soil evaporation measured using micro-lysimeters.

Evapotranspiration calculation processes

Sap flow data were analyzed using the DGSF software (Dynamax Inc., USA) to determine hourly sap flow rates and cumulative sap flux. Daily average transpiration per unit leaf area was calculated according to the cumulative sap flux and leaf area. Given the following data processing schedule, the results were then compared to the measurements from the LI-6400 photosynthesis system.

The transpiration rate (Tr) measured using the LI-6400 photosynthesis system was in mmol H2O m−2s−1. Based on unit conversion, the hourly average transpiration per unit leaf area (Trhr, g m−2h−1) was Trhr = 0.015 Tr. Then, the diurnal variations in transpiration rate were fitted as the following polynomial: 
formula
4
where A is a constant, Bi is the coefficient of the polynomial, and t is the time (8 ≤ t ≤ 20). Finally, the diurnal transpiration rate per unit leaf area (Tl) was obtained as: 
formula
5
Stem flow measurements from the individual plants were converted to the transpiration rate of whole trees per unit leaf area using the following expression: 
formula
6
where Td is the plant canopy transpiration (g h−1m−2), fi is the sap flow measured in the stem i (g h−1), Ai, j is the leaf area of the stem i (m2) at time j (month), n is the number of equipped stems, and LAIj is the LAI (m2/m2) measured above at time j. The daily transpiration rates (ETd; g m−2d−1) were estimated from the cumulative Td values over 24 h. The equivalent mass of a kilogram of water in one square meter of ground area was subsequently expressed as depth in millimeters (mm) such that transpiration could be given in mm d−1. Next, the direct measurements in Equation (6) were compared with simulation results in Equations (1)–(3).

Scaling model quality evaluation

To evaluate the quality of the scaling-up transpiration simulations, graphs measured against the simulated values were drawn together with the linear regression, the correlation coefficient, and the 1:1 line. Assuming there is no model error, the measured and simulated values should be identical and all points should lie on the 1:1 line. The points of good quality simulations should lie close to the 1:1 line, the slope of the linear regression should be close to one, and the correlation coefficient should be close to one. A widely used measure of agreement between measured and simulated values is the root mean squared error (RMSE) (Wallach 2006) 
formula
where is observed values and is simulated values at time i, and n is the sample number.

RESULTS

Daily transpiration rate using the photosynthesis system and sap flow measurements

The diurnal cycle of the transpiration rates of C. korshinskii and A. ordosica (averages of 34 diurnal transpiration measurements from April 2008 to October 2012) are shown in Figure 2(a). Both species exhibited a single-peak curve, with the peak occurring at 09:00 and 10:00 a.m., respectively. The diurnal variations of A. ordosica were slightly different in that the transpiration rate increased gradually at 12:00 a.m. and later decreased after 14:00 p.m. Although it showed a small peak of transpiration, the later peak appears statistically indistinguishable from the three previous hours, which indicates the ‘noon break’ phenomenon or midday depression (Li et al. 2008). Mean transpiration rates of C. korshinskii and A. ordosica, as seen in Figure 2(a), were 2.67 and 4.51 mmol H2O m−2 s−1, respectively.

Figure 2

Diurnal course of transpiration rate Tr (a) and the sap flow rate (b) in C. korshinskii and A. ordosica during the experimental period (April 2008–October 2012).

Figure 2

Diurnal course of transpiration rate Tr (a) and the sap flow rate (b) in C. korshinskii and A. ordosica during the experimental period (April 2008–October 2012).

The sap velocity of C. korshinskii and A. ordosica had distinct diurnal variations during sunny days from April 2008 to October 2012 (approximately 700 days). As shown in Figure 2(b), the sap flow rate of C. korshinskii increased immediately at 6:00 a.m., peaked at 12:00 p.m., and gradually decreased to a minimal sap velocity at night. By contrast, the sap flow rate of A. ordosica gradually increased at 8:00 a.m., and peaked at 14:00 to 16:00 p.m., 2 h later than in C. korshinskii, and then immediately decreased to zero after 20:00 p.m. No sap flow was observed at night or before dawn, which indicates that no transpiration occurred or that the transpiration was too small to measure. The average sap flow rate in C. korshinskii was 0.071 g h−1cm−2, and that in A. ordosica was 0.086 g h−1cm−2, which were mainly determined by Rs, VPD, and RH, respectively (Huang et al. 2010).

LAI variations

The monthly variations in LAI and precipitation during the experimental period are illustrated in Figure 3. LAI varied with precipitation as seen in Figures 3(a) and 3(b), especially in 2010, during which the precipitation was only 113.9 mm, resulting in a significant decrease in LAI. However, the opposite was true in 2011 and 2012, when the precipitation was 178.5 and 195.3 mm, respectively, and the LAI showed a correspondingly significant increase. During each year, the buds started to burst and the LAI increased in April and peaked in June. Then, the decreased precipitation and higher air temperatures in July caused a decrease in LAI; by August or September, C. korshinskii and A. ordosica often went through a second growing stage, even with little precipitation. Thereafter, the leaves began to fall and LAI gradually decreased to a low level. Figures 3(c) and 3(d) show that the mean LAI during the growing season (April–October) had a significant positive linear correlation with annual precipitation. During study years of heavy rainfall, such as in 2008 and 2009, the total amount of annual precipitation was approximately the same (125.2 and 127.4 mm), although the number of rain days was 23 in 2008 and 29 in 2009; results have shown that the LAI of C. korshinskii was 0.55 in 2008 and 0.64 in 2009, which indicates that rain frequency had a positive effect on LAI. On the other hand, rain frequency had little effect on the LAI of A. ordosica, which was 0.39 in 2008 and 0.38 in 2009. Thus, we could hypothesize that under certain situations, such as among different plant species or rainfall amounts, rainfall intermittency and frequency could cause differences in the LAI dynamics; this means that temporal rainfall intermittency could result in relatively higher LAIs and longer plant-growing periods compared with constant precipitation.

Figure 3

Precipitation (a), LAI dynamics (b) and its corresponding linear regression relationship with annual precipitation in C. korshinskii (c) and A. ordosica (d) during the experimental period (April 2008–October 2012).

Figure 3

Precipitation (a), LAI dynamics (b) and its corresponding linear regression relationship with annual precipitation in C. korshinskii (c) and A. ordosica (d) during the experimental period (April 2008–October 2012).

Scaling up transpiration from leaf to stand

To compare different plants, the sap flow values from SHB measurements were divided by leaf area to obtain transpiration per leaf unit. Average daily transpiration rates of C. korshinskii and A. ordosica with the sap flow measurements during our experimental period were 0.81 and 1.45 g cm−2 d−1, and the simulation results with Equations (4) and (5) were 0.64 and 1.38 g cm−2 d−1, respectively. The observed data with sap flow measurement versus simulated values from the Li-6400 extrapolation transpiration rate during the 34 diurnal measurements are shown in Figure 4. The result indicates that the model can estimate transpiration with reasonable accuracy (R2 = 0.67 for C. korshinskii and R2 = 0.77 for A. ordosica), although the simulated values were generally lower than measured values. The statistical analysis of the results showed that both the relative error was <10%, and the RMSE was 0.38 and 0.31 g cm−2 d−1 for C. korshinskii and A. ordosica, respectively. Furthermore, simulated versus observed transpiration values fell around the 1:1 line for combined data, which means that the extrapolation model was accurate and reasonable.

Figure 4

Comparisons of C. korshinskii and A. ordosica daily transpiration rates by Li-6400 extrapolation and SHB measurement during the experimental period (April 2008–October 2012).

Figure 4

Comparisons of C. korshinskii and A. ordosica daily transpiration rates by Li-6400 extrapolation and SHB measurement during the experimental period (April 2008–October 2012).

Plant canopy transpiration on a per unit ground area basis was calculated using Equation (6), and then compared with the actual ET minus soil evaporation as calculated by Equations (1)–(3). During our experimental period from April 2008 to October 2012 (909 days), mean soil evaporation in the C. korshinskii and A. ordosica community was 0.72 and 0.83 mm d−1, respectively. Furthermore, soil evaporation of the two shrubs in wet years (such as 195.3 mm rainfall in 2012) was 0.86 and 0.97 mm d−1, which was significantly greater than 0.41 and 0.59 mm d−1 in the dry years (such as 113.9 mm rainfall in 2010). The daily transpiration rates of C. korshinskii and A. ordosica communities were 0.36 and 0.28 mm d−1 with the sap flow extrapolation, respectively, whereas direct measurements from the crop evapotranspiration method were 0.42 and 0.35 mm d−1. Comparisons between the two methods are presented in Figure 5. The linearly fitted regression equation also indicated a notably good upscaling process. The value of the coefficient of determination (R2 = 0.560 for C. korshinskii and R2 = 0.647 for A. ordosica) indicated that a good correlation exists between the simulated and observed stand transpiration. The RMSE was 0.16 and 0.15 mm d−1 for C. korshinskii and A. ordosica, respectively. Stand transpiration values of the two methods were distributed around the 1:1 line for combined data, which also means that the sap flow extrapolation model is accurate and reasonable. The transpiration of C. korshinskii was higher than A. ordosica and total seasonal transpiration (April–October) amounted to 71.79 and 55.62 mm, respectively, representing approximately 48.4 and 37.5% of the total rainfall over this period.

Figure 5

Comparisons of C. korshinskii and A. ordosica daily transpiration rates with SHB extrapolation and the crop evapotranspiration minus micro-lysimeter method during the experimental period (April 2008–October 2012).

Figure 5

Comparisons of C. korshinskii and A. ordosica daily transpiration rates with SHB extrapolation and the crop evapotranspiration minus micro-lysimeter method during the experimental period (April 2008–October 2012).

DISCUSSION

Estimating plant transpiration is highly complex due to the unique, sparse vegetation cover in arid and semiarid zones (Noilhan et al. 1997; Ramirez et al. 2006). Compared with the continuous coverage observed in forests and on cropland, the soil heterogeneity and vegetation boundary differences in the desert make upscaling difficult. Researchers have their own scaling methods at different scales, but the results of their research are not always comparable. In this study, leaf area was used as a unified scaling factor that shifted from one scale to another, mainly because of the biological features of desert plants, such as low biomass or growth in the stems. Our scaled measurements have proven successful for scaling up transpiration from the leaf to the stand level. The LAI dynamics of C. korshinskii and A. ordosica are illustrated in Figure 3. LAI was significantly dependent on the precipitation. In the wet years, the LAI of C. korshinskii and A. ordosica was 0.85 and 0.39, larger than in the dry years at 0.48 and 0.32, respectively. However, the sensitivity of LAI to decreasing precipitation was greater in C. korshinskii than in A. ordosica, which may indicate that A. ordosica communities display a greater drought tolerance in arid desert regions. During rainfall years, rainfall intermittency and frequency caused differences in the LAI dynamics. This phenomenon also agrees with other studies showing that plants from desert ecosystems are favored by less frequent rainfall events for the same total seasonal rainfall volume (Sher et al. 2004), as a minimal amount of water is in fact needed to activate biological processes such as reproduction or seed germination. If precipitation is concentrated in time, this minimum amount of soil water is intermittently exceeded and vegetation can persist. If rainfall is more uniformly spread across the growing season, soil moisture never exceeds the threshold needed to activate plant reproduction and germination, and the vegetation tends to disappear (Baudena et al. 2007).

Considering that transpiration is a fundamental part of ET, the sap flow-scaled method is the main method used for upscaling transpiration from the leaf level to the canopy level. Plant transpiration within a short period can be accurately measured using a Li-6400 photosynthesis system, but the total amount of long-term transpiration is difficult to determine continuously because of the influence of microclimate factors. The difference between the average simulated daily transpiration rate and the direct sap flow measurement was readily discernible through the polynomial fit and integration of data from the experimental period, but the results of the simulation were slightly lower than the direct measurements, as indicated by the regression coefficients (both are <1). This discrepancy may be due to the different measuring instruments: the photosynthesis system only measures leaf transpiration, whereas the sap flow sensors measure both leaf transpiration and stem transpiration. The discrepancy may also be because leaf transpiration was only measured from 8:00 a.m. to 20:00 p.m., but sap flow was measured for 24 h. Determining actual transpiration based on a combination of the Penman–Monteith reference ET (ET0) and the crop coefficient (Kc) method is a simple and effective method that has gained wide acceptance in agricultural water management (Suyker & Verma 2009). In this study, the Kc values were obtained from other publications as a constant; however, these values varied significantly during the growing season (Zhang et al. 2012). Thus, more accurate parameterizations using new techniques should be developed to estimate the stand transpiration. The regression coefficients for the leaf level to tree level and the tree level to stand level of C. korshinskii were both lower than those of A. ordosica, which indicates that the leaf area-based extrapolation method is more appropriate for A. ordosica shrubs. Based on Figure 5, the sap flow extrapolation was also less than that from the crop evapotranspiration minus the lysimeter results, especially during the growing seasons. This differs from the laboratory experiments in that the sap flow sensors overestimated 13.5% of the total transpiration (Chabot et al. 2002). This discrepancy is mainly observed because certain annual and perennial herbs inhabit and reproduce on the soil surface after the re-establishment of vegetation, and their contribution to transpiration cannot be distinguished from those observed under the crop evapotranspiration method.

The transpiration values of the C. korshinskii stand were higher than those of the A. ordosica stand during the growing season because of the large difference in tree size. The seasonal (April–October) total transpiration reached 71.79 and 55.62 mm, respectively, representing approximately 48.4 and 37.5% of the total rainfall over this period. Species differences in water use per unit of leaf area were smaller, but sometimes statistically significant, especially during rainy days. A small amount of precipitation increased the sap flow and transpiration; however, large amounts of precipitation decreased sap flow and transpiration (Huang et al. 2010). A. ordosica is more likely to be affected by this phenomenon because its roots are found primarily in the uppermost soil layers, especially in the case of coarse roots, which are concentrated in the 0.2 m soil layer and are more easily influenced by rainfall. However, for C. korshinskii, the roots are primarily concentrated in the 0.6 to 1 m soil layer (Zhang et al. 2009), and the influence is relatively smaller as the water is redistributed in the soil. Further analysis showed that the percentage of soil evaporation was very high after precipitation, and it was also different at different experimental plots, although in most cases (61%) there was no significant difference (p > 0.05). However, during the rainy days, the transpiration of C. korshinskii, A. ordosica and soil evaporation was 0.72, 0.75 and 0.61 mm d−1, respectively, but during the drier days, the daily transpiration and soil evaporation were 0.12, 0.13 and 0.24 mm d−1. There was a significant difference (p < 0.01) between them. The ratio of transpiration to ET decreased as annual precipitation declined, while the ratio of evaporation to ET increased. This would mean that drought stress decreased transpiration and ET and increased precipitation loss due to evaporation, which, in turn, led to a large decrease in available water for plant growth, and so less drought-tolerant plants, such as C. korshinskii, would disappear or degrade with future vegetation succession.

CONCLUSIONS

A long-term transpiration study was conducted during the growing seasons from April 2008 to October 2012 using two xerophytic shrubs typically used in revegetation: C. korshinskii and A. ordosica. Gas exchange techniques (using an infrared gas analyzer), sap flow measurements, and the crop evapotranspiration minus lysimeter method were used to evaluate plant transpiration at the leaf level, tree level, and stand level. Simultaneous measurements using the three methods were made diurnally and were compared with the normalized leaf area-based extrapolation. The results show that leaf area is a reliable scaling factor for upscaling transpiration from the leaf level to the stand level in desert regions. At the leaf level, the transpiration rates of C. korshinskii and A. ordosica were 2.67 and 4.51 mmol H2O m−2 s−1, respectively, showing a single-peak curve with meteorological factors as the key limiting factors. The sap flow rates were 0.071 and 0.086 g h−1 cm−2 at the tree level, and the transpiration rates were 0.42 and 0.35 mm d−1 at the stand level with precipitation and LAI as the main limiting factors. Direct measurement and extrapolation during non-rainfall periods showed very good correlations. The results provide a key link for designing management strategies to improve water use efficiency and to determine the appropriate plant density in revegetated areas with limited soil water capacity.

ACKNOWLEDGMENTS

This work was supported by the National Key Basic Research program (2013CB429905-03) and Chinese National Natural Scientific Foundation (41201084; 31170385).

REFERENCES

REFERENCES
Allen
R. G.
Pereira
L. S.
Raes
D.
Smith
M.
1998
Crop evapotranspiration-guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56
.
Food and Agricultural Organization of the United Nations
,
Rome, Italy
.
Baudena
M.
Boni
G.
Ferraris
L.
von Hardenbergb
J.
Provenzaleb
A.
2007
Vegetation response to rainfall intermittency in drylands: results from a simple ecohydrological box model
.
Adv. Water Resour.
30
,
1320
1328
.
Chabot
R.
Bouarfa
S.
Zimmer
D.
Chaumont
C.
Duprez
C.
2002
Sugarcane transpiration with shallow water-table: sap flow measurements and modeling
.
Agric. Water Manage.
54
,
17
36
.
Domingo
F.
Villagarcia
L.
Brenner
A.
Puigdefábregas
J.
1999
Evapotranspiration model for semi-arid shrublands testes against data from SE Spain
.
Agric. Forest Meteorol.
95
,
67
84
.
Du
S.
Wang
Y. L.
Kume
T.
Zhang
J. G.
Otsuki
K.
Yamanakac
N.
Liu
G. B.
2011
Sap flow characteristics and climatic responses in three forest species in the semiarid Loess Plateau region of China
.
Agric. Forest Meteorol.
151
(
1
),
1
10
.
Dugas
W. A.
Wallace
J. S.
Allen
S. J.
Roberts
J. M.
1993
Heat balance, porometer, and deuterium estimates of transpiration from potted trees
.
Agric. Forest Meteorol.
64
,
47
62
.
Ford
C. R.
Hubbard
R. M.
Kloeppel
B. D.
Vose
J. M.
2007
A comparison of sap flux-based evapotranspiration estimates with catchment-scale water balance
.
Agric. Forest Meteorol.
145
,
176
185
.
Granier
A.
Huc
R.
Barigah
S. T.
1996
Transpiration of natural rain forest and its dependence on climatic factors
.
Agric. Forest Meteorol.
78
,
19
29
.
Kelliher
F. M.
Kostner
B. M. M.
Hollinger
D. Y.
Byers
J. N.
Hunt
J. E.
McSeveny
T. M.
Meserth
R.
Weir
P. L.
Schulze
E. D.
1992
Evaporation, xylem sap flow, and tree transpiration in a New Zealand broad-leaved forest
.
Agric. Forest Meteorol.
62
,
53
73
.
Li
X. R.
2005
Influence of variation of soil spatial heterogeneity on vegetation restoration
.
Science in China Series D: Earth Sciences
48
,
2020
2031
.
Li
Q. H.
Zhao
Y. M.
Liu
J. F.
Zhao
X. L.
Jiang
Z. P.
2008
The differences of growth and photosynthetic traits of five shrubs species from two origins in the northeastern Ulanbuh Desert
.
Forest Res.
3
,
357
361
.
Nicolas
E. A.
Torrecillas
M. F.
Domingo
O. R.
Alarcon
J. J.
2005
Evaluation of transpiration in adult apricot trees from sap flow measurements
.
Agric. Water Manage.
72
,
131
145
.
Rico-García
E.
Hernández-Hernández
F.
Soto-Zarazúa
G.
Herrera-Ruiz
G.
2009
Two new methods for the estimation of leaf area using digital photography
.
Int. J. Agric. Biol.
11
,
397
400
.
Suyker
A. E.
Verma
S. B.
2009
Evapotranspiration of irrigated and rainfed maize-soybean cropping systems
.
Agric. Forest Meteorol.
149
,
443
452
.
Tanaka
N.
Kume
T.
Yoshifuji
N.
Tanaka
K.
Takizawa
H.
Shiraki
K.
Tantasirin
C.
Tangtham
N.
Suzuki
M.
2008
A review of evapotranspiration estimates from tropical forests in Thailand and adjacent regions
.
Agric. Forest Meteorol.
148
,
807
819
.
Teskey
R. O.
Sheriff
D. W.
1996
Water use by Pinus radiata trees in a plantation
.
Tree Physiol.
16
,
273
279
.
Wallach
D.
2006
Evaluating crop models.
In: Working with Dynamic Crop Models: Evaluation, Analysis, Parameterization, and Applications. Elsevier, Amsterdam, The Netherlands, pp. 11–50.
Wang
X. P.
Kang
E. S.
Zhang
J. G.
Li
X. R.
2004a
Evapotranspiration of Artemisia ordosia vegetation in stabilized arid desert dune in Shapotou, China
.
Arid Land Res. Manag.
18
,
63
76
.
Wang
X. P.
Brown-Mitic
C. M.
Kang
E. S.
Zhang
J. G.
Li
X. R.
2004b
Evapotranspiration of Caragana korshinskii communities in a revegetated desert area: Tengger Desert, China
.
Hydrol. Process.
18
,
3293
3303
.
Zhang
Z. S.
Liu
L. C.
Li
X. R.
Zheng
J. G.
He
M. Z.
Tan
H. J.
2008
Evaporation properties of a revegetated area of the Tengger Desert North China
.
J. Arid Environ.
72
,
964
973
.