Soil moisture control on sap-flow response to biophysical factors in a desert-shrub 1 species , Artemisia ordosica 2

1. Yanchi Research Station, School of Soil and Water Conservation, Beijing Forestry 5 University, Beijing 100083, China 6 2. Beijing Vocational College of Agriculture, Beijing 102442, China 7 3. Key Laboratory of State Forestry Administration on Soil and Water Conservation, 8 Beijing Forestry University, Beijing, China 9 4. Faculty of Forestry and Environmental Management, 28 Dineen Drive, PO Box 4400, 10 University of New Brunswick, New Brunswick, E3B5A3, Canada 11 5. Faculty of Science and Forestry, School of Forest Sciences, University of Eastern 12 Finland, Joensuu, FI-80101, Finland 13


Introduction
This study provides a significant contribution to the understanding of acclimation processes in desert-shrub species to drought-associated stress in dryland ecosystems.
Due to the low amount of precipitation and high potential evapotranspiration in desert ecosystems, low soil water availability limits both plant water-and gas-exchange and, as a consequence, limits vegetation productivity (Razzaghi et al., 2011).Shrub and semi-shrub species are replacing grass species in arid and semi-arid lands in response to ongoing aridification of the land surface (H.Huang et al., 2011).This progression is predicted to continue under a changing climate (Houghton et al., 1999;Pacala et al., 2001;Asner et al., 2003).Studies have shown that desert shrubs are able to adapt to hot, dry environments as a result of their small plant surface area, thick epidermal hairs, and large root-to-shoot Published by Copernicus Publications on behalf of the European Geosciences Union.
T. Zha et al.: Sap flow in Artemisia ordosica ratios (Eberbach and Burrows, 2006;Forner et al., 2014).Plant traits related to water use are likely to adapt differentially with species and habitat type (Brouillette et al., 2014).Plants may select water-acquisition or water-conservation strategies in response to water limitations (Brouillette et al., 2014).Knowledge of physiological acclimation of changing species to water shortages in deserts, particularly with respect to transpiration, is inadequate and, in the context of plant adaptation to changing climatic conditions, is of immense interest (Jacobsen et al., 2007;H. Huang et al., 2011).Transpiration maintains ecosystem balance through the soilplant-atmosphere continuum and its magnitude and timing are related to the prevailing biophysical factors (Jarvis, 1976;Jarvis and McNaughton, 1986).
Sap flow can be used to reflect species-specific water consumption by plants (Ewers et al., 2002;Baldocchi, 2005;Naithani et al., 2012).Sap flow can also be used to continuously monitor canopy conductance and its response to environmental variables (Ewers et al., 2007;Naithani et al., 2012).Biotic and abiotic effects on sap flow and transpiration are often interactive and confounded.The decoupling coefficient ( ) was used to examine the relative contribution of plant control through stomatal regulation of transpiration (Jarvis and McNaughton, 1986).Stomatal regulation becomes stronger as approaches zero.Stomatal conductance (g s ) on the plant scale exerts a large biotic control on transpiration particularly during dry conditions (Jarvis, 1976;Jarvis and McNaughton, 1986).Stomatal conductance couples photosynthesis and transpiration (Cowan and Farquhar, 1977), making this parameter an important component of climate models in quantifying biospheric-atmospheric interactions (Baldocchi et al., 2002).
Studies have shown that xylem hydraulic conductivity was closely correlated with drought resistance (Cochard et al., 2008(Cochard et al., , 2010;;Ennajeh et al., 2008).With increasing aridity, trees can progressively lessen their stomatal conductance, resulting in lower transpiration (McAdam et al., 2016).Generally, desert shrubs can close their stomata, reducing stomatal conductance, and reduce their water consumption when exposed to dehydration stresses.However, differences exist among shrub species in terms of their stomatal response to changes in air and soil moisture deficits (Pacala et al., 2001).
In Elaeagnus angustifolia, transpiration is observed to peak at noon, i.e., just before stomatal closure under waterdeficit conditions (Liu et al., 2011), peaking earlier than radiation, temperature, and water vapor pressure deficit (VPD).This response lag or hysteresis effect has been widely noticed in dryland species (Du et al., 2011;Naithani et al., 2012), but its function is not completely understood.Transpiration in Hedysarum scoparium peaks multiple times during the day.During dry periods of the year, sap flow in Artemisia ordosica has been observed to be controlled by volumetric soil water content (VWC) at about a 30 cm depth in the soil (Li et al., 2014).For other shrubs, sap flow has been observed to decrease rapidly when the VWC is lower than the water loss through evapotranspiration (Buzkova et al., 2015).Sap flow in Caragana korshinskii and Hippophae rhamnoides have been found to increase with increasing rainfall intensity (Jian et al., 2016); whereas in Haloxylon ammodendron, it was found to vary in response to rainfall, from an immediate decline after a heavy rainfall to no observable change after a small rainfall event (Zheng and Wang, 2014).Droughtinsensitive shrubs have relatively strong stomatal regulation and, therefore, tend to be insensitive to soil water deficits and rainfall, unlike their drought-sensitive counterparts (Du et al., 2011).Support for the relationship between sap flow in desert shrubs and prevailing environmental factors is decidedly variable (McDowell et al., 2013;Sus et al., 2014), potentially varying with plant habitat and species (Liu et al., 2011).
Artemisia ordosica, a shallow-rooted desert shrub, is the dominant species in the Mu Us Desert of northwestern China.It plays an important role in combating desertification and in stabilizing sand dunes (Li et al., 2010).Increases in air temperature, precipitation variability, and associated shorter wet and longer dry periods are expected to ensue under the influence of climate change (Lioubimtseva and Henebry, 2009).However, our understanding of the mechanisms of desertshrub acclimation during periods of water shortage remains incomplete.Questions that need answers from our research include (1) how do changes in sap flow relate to changes in biotic and abiotic factors, and (2) whether Artemisia ordosica selects a strategy of water-conservation or water-acquisition under conditions of drought?To attend to these questions, we continuously measured stem sap flow in Artemisia ordosica and associated environmental variables in situ throughout the growing seasons of 2013 and 2014 (May-September period of each year).Our findings lead to insights concerning the main environmental factors affecting transpiration in Artemisia ordosica, e.g., optimal temperature, VPD, and VWC.This understanding can lead to improving phytoremediation practices in desert-shrub ecosystems.

Experimental site
Continuous sap-flow measurements were made at the Yanchi Research Station (37 • 42 31 N, 107 • 13 47 E; 1530 m above mean sea level), Ningxia, northwestern China.The research station is located between the arid and semi-arid climatic zones along the southern edge of the Mu Us Desert.The sandy soil in the upper 10 cm of the soil profile has a bulk density of 1.54 ± 0.08 g cm −3 (mean ± standard deviation, n = 16).Mean annual precipitation in the region is about 287 mm, of which 62 % falls between July and September.Mean annual potential evapotranspiration and air temperature are about 2024 mm and 8.1 • C, respectively, based on meteorological data  from the Yanchi County weather station.Normally, shrub leaf-expansion,  (Chen et al., 2015).
Half-hourly data were recorded by data logger (CR3000 data logger, Campbell Scientific Inc., USA).VWC at a 30 cm depth were measured using three ECH 2 O-5TE soil moisture probes (Decagon Devices, USA).In the analysis, we used half-hourly averages of VWC from the three soil moisture probes.VPD (in kPa) was calculated from recorded RH and T .

Measurements of sap flow, leaf area, and stomatal conductance
The experimental plot (10×10 m) was located on the western side of Yanchi Research Station in an Artemisia ordosicadominated area.The mean age of the Artemisia ordosica was 10-years old.Maximum monthly mean leaf area index (LAI) for plant specimens with full leaf expansion was about 0.1 m 2 m −2 (Table 1).Over 60 % of their roots were distributed in the first 60 cm of the soil complex (Zhao et al., 2010;Jia et al., 2016).Five stems of Artemisia ordosica were randomly selected within the plot as replicates for sapflow measurement.Mean height and sapwood area of sampled shrubs were 84 cm and 0.17 cm 2 , respectively.Sampled stems represented the average size of stems in the plot.A heat-balance sensor (Flow32-1K, Dynamax Inc., Houston, USA) was installed at about 15 cm above the ground surface on each of the five stems (Dynamax, 2005).Sap-flow measurements from each stem were taken once per minute.
Half-hourly data were recorded by a Campbell CR1000 data logger from 1 May to 30 September, for both 2013 and 2014 (Campbell Scientific, Logan, UT, USA).Leaf area was estimated for each stem every 7-10 days by sampling about 50-70 leaves from five randomly sampled neighboring shrubs with similar characteristics to the shrubs being used for sap-flow measurements.Leaf area was measured immediately at the station laboratory with a portable leaf-area meter (LI-3000, LI-COR Inc., Lincoln, NE, USA).LAI was measured on a weekly basis from a 4 × 4 grid of 16 quadrats (10 × 10 m each) within a 100 × 100 m plot centered on a flux tower using measurements of sampled leaves and allometric equations (Jia et al., 2014).Stomatal conductance (g s ) was measured in situ for three to four leaves on each of the sampled shrubs with a LI-6400 portable photosynthesis analyzer (LI-COR Inc., Lincoln, NE, USA).The g s measurements were made every 2 h from 07:00 to 19:00 (all times are local time) every 10 days from May to September 2013 and 2014.
The degree of coupling between the ecosystem surface and the atmospheric boundary layer was estimated with the decoupling coefficient ( ).The decoupling coefficient varies from zero (i.e., leaf transpiration is mostly controlled by g s ) to one (i.e., leaf transpiration is mostly controlled by radiation).The was calculated as described by Jarvis and Mc-Naughton (1986) where is the rate of change of saturation vapor pressure vs. temperature (kPa K −1 ), γ is the psychrometric constant (kPa K −1 ), and g a is the aerodynamic conductance (m s −1 ; Monteith and Unsworth, 1990), where u is the wind speed (m s −1 ) at 6 m above the ground, and u * is the friction velocity (m s −1 ), measured by a nearby eddy covariance system (Jia et al., 2014).

Data analysis
In our analysis, drought days were defined as those days with daily mean VWC < 0.1 m 3 m −3 .This is based on a VWC threshold of 0.1 m 3 m −3 for the sap flow per leaf area J s (Fig. 1), with J s increasing as VWC increased, saturating at a VWC of 0.1 m 3 m −3 , and decreasing as VWC continued to increase.The VWC threshold of 0.1 m 3 m −3 is equivalent to a relative extractable soil water (REW) of 0.4 for drought conditions (Granier et al., 1999(Granier et al., , 2007;;Zeppel et al., 2004;Fig. 2d, e).Duration and severity of "drought" were defined based on a VWC threshold and REW of 0.4.REW was calculated with where VWC is the specific daily soil water content (m 3 m −3 ), and VWC min and VWC max are the minimum and maximum VWC during the measurement period in each year, respectively.Sap-flow analysis was conducted using mean data from five sensors.Sap flow per leaf area (J s ) was calculated according to where J s is the sap flow per leaf area (kg m −2 h −1 or kg m −2 d −1 ), E is the measured sap flow of a stem (g h −1 ), A l is the leaf area of the sap-flow stem, and n is the number of stems sampled (e.g., n = 5).
Transpiration per ground area (T r ) was estimated in this study according to: where T r is transpiration per ground area (mm d −1 ).
Linear and non-linear regressions were used to analyze abiotic control on sap flow.In order to minimize the effects of different phenophases and rainfall, we only used data from the mid-growing season, non-rainy days, and daytime hours from 08:00 to 20:00, i.e., from 1 June to 31 August, with hourly shortwave radiation greater than 10 W m −2 .Relations between mean sap flow at specific times over a period of 08:00-20:00 and corresponding environmental factors from 1 June to 31 August were derived from linear regression (p < 0.05; Fig. 3).Regression slopes were used as indicators of sap-flow sensitivity (degree of response) to the various environmental variables (see Zha et al., 2013).All statistical analyses were performed with SPSS version 17.0 for Windows software (SPSS Inc., USA).Significance level was set at 0.05.

Sap-flow response to environmental factors
In summer, J s increased with increasing VWC, R s , T , and VPD (Figs. 2d, f, and 3).Sap flow increased more rapidly with increases in R s , T , and VPD under high VWC (i.e., VWC > 0.1 m 3 m −3 in both 2013 and 2014; Fig. 4) compared with periods with lower VWC (i.e., VWC < 0.1 m 3 m −3 in both 2013 and 2014; Fig. 4).Sap flow was more sensitive to R s , T , and VPD under high VWC (Fig. 4), which coincided with a steeper regression slope under high VWC conditions.Sensitivity of J s to environmental variables (in particular, R s , T , VPD, and VWC) varied depending on time of day (Fig. 5).Regression slopes for the relations of J s -R s , J s -T , and J s -VPD were greater in the morning before 11:00, and lower during mid-day and early afternoon (12:00-16:00).In contrast, regression slopes of the relation of J s -VWC were lower in the morning (Fig. 5), increasing thereafter, peaking at ∼ 13:00, and subsequently decreasing in late afternoon.Regression slopes of the response of J s to R s , T , and VPD in 2014 were steeper than those in 2013.

Diurnal changes and hysteresis between sap flow and environmental factors
Diurnal patterns of J s were similar in both years (Fig. 6), initiating at 07:00 and increasing thereafter, peaking before noon (12:00), and subsequently decreasing thereafter and remaining near zero from 20:00 to 06:00.Diurnal changes in g s were similar to J s , but peaking about 2 and 1 h earlier than J s in July and August, respectively (Fig. 6).There were pronounced time lags between J s and R s over the 2 years (Fig. 7), J s peaking earlier than R s and, thus, earlier than either VPD or T .These time lags differed seasonally.For example, mean time lag between J s and R s was 2 h during July, 5 h during May, and 3 h during June, August, and September of 2013.However, the time lags in 2014 were generally shorter than those observed in 2013 (Table 2).
Clockwise hysteresis loops between J s and R s during the growing period were observed (Fig. 7).As R s increased in the morning, J s increased until it peaked at ∼ 10:00.Sap flow www.biogeosciences.net/14/4533/2017/Biogeosciences, 14, 4533-4544, 2017 declined with decreasing R s during the afternoon.Sap flow (J s ) was higher in the morning than in the afternoon.Diurnal time lag in the relation of J s -R s were influenced by VWC (Figs. 8,9).For example, J s peaked about 2 h earlier than R s on days with low VWC (Fig. 8a), 1 h earlier than R s on days with moderate VWC (Fig. 8b), and at the same time as R s on days with high VWC (Fig. 8c).Lag hours between J s and R s over the growing season were negatively and linearly related to VWC (Fig. 9: lag (h) = −133.5 × VWC + 12.24, R 2 = 0.41).The effect of VWC on time lags between J s and R s was smaller in 2014, with evenly distributed rainfall during the growing season, than in 2013, with a pronounced summer drought (Fig. 9).Variables g s and showed a significantly increasing trend with increasing VWC in 2013 and 2014 (Fig. 10).This trend was more obvious in the dry year of 2013 than in 2014.

Sap-flow response to environmental factors
Drought tolerance of some plants may be related to lower overall sensitivity of plant physiological attributes to envi-ronmental stress and/or stomatal regulation (Y.Huang et al., 2011;Naithani et al., 2012).In this study, steep regression slopes between J s and the environmental variables (R s , VPD, and T ) in the morning indicated that sap flow was less sensitive to variations in R s , VPD, and T during the drier and hotter part of the day (Fig. 5).The lower sensitivity combined with lower stomatal conductances led to lower sap flow, and, thus, lower transpiration (water consumption) during hot mid-day summer hours, pointing to a water-conservation strategy in plant acclimation during dry and hot conditions.When R s peaked during mid-day (13:00-14:00) in summer, there was often insufficient soil water to meet the atmospheric demand, causing g s to be limited by available soil moisture and making J s more responsive to VWC at noon, but less responsive to R s and T .Similarly, sap flow in Hedysarum mongolicum and some other shrubs in a nearby region were positively correlated with VWC at noon (Qian et al., 2015).For instance, sap flow in Picea crassifolia peaked at noon (12:00 and 14:00) and then decreased, heightening by increasing R s , T , and VPD, when R s < 800 W m −2 , T < 18.0 • C, and VPD < 1.4 kPa (Chang et al., 2014); sap flow in Caragana korshinskii was significantly lower during the stress period, its conductance decreasing linearly after the wilting point (She et al., 2013).The fact that J s was less sensitive to meteorological variables when VWC < 0.10 m 3 m −3 , highlights the water-conservation strategy taken by drought-afflicted Artemisia ordosica.The positive linear relationship between g s and VWC in this study further supports this conclusion.
Precipitation, being the most important source of soil moisture and, thus, VWC, affected transpiration directly.Frequent small rainfall events (< 5 mm) are crucially important to the survival and growth of desert plants (Zhao and Liu, 2010).Variations in J s were clearly associated with the intermittent supply of water to the soil during rainfall events (see Fig. 2d, f).Reduced J s during rainy days can be largely explained by a reduction in incident R s and liquid waterinduced saturation of the leaf surface, which led to a decrease in leaf turgor and stomatal closure.After each rainfall event, J s increased quickly when soil moisture was replenished.Schwinning and Sala (2004) have previously shown that VWC contributed the most to the post-rainfall response in plant transpiration at similar sites.The study shows that Artemisia ordosica responded differently to wet and dry conditions.In the mid-growing season, high J s in July was related to rainfall-fed soil moisture, which increased the rate of transpiration.However, dry soil conditions combined with high T and R s led to a reduction in J s in August of 2013 (Fig. 2).In some deep-rooting desert shrubs, groundwater may replenish water lost by transpiration (Yin et al., 2014).Artemisia ordosica roots are generally distributed in the upper 60 cm of the soil (Zhao et al., 2010), and as a result the plant usually depends on water directly supplied by precipitation.This is because groundwater levels in drylands can often be below the rooting zone of many shrub species, typ-  ically at depths greater than 10 m as witnessed at our site.Similar findings regarding the role of rainfall and VWC in desert vegetation are reported by Wang et al. (2017).

Hysteresis between sap flow and environmental factors
Diurnal patterns in J s corresponded with those of R s from sunrise until diverging later in the day (Fig. 7), suggesting that R s was a primary controlling factor of diurnal s .As an initial energy source, R s also can force T and VPD to increase, causing a phase difference in time lags among the relations of J s -R s , J s -T , and J s -VPD.
The hysteresis effect reflects plant acclimation to water limitations, due to stomatal conductance being inherently dependent on plant hydrodynamics (Matheny et al., 2014).The large g s in the morning promoted higher rates of transpiration (Figs. 6,7), while lower g s in the afternoon reduced transpiration.Therefore, diurnal curves (hysteresis) were mainly caused by a g s -induced hydraulic process (Fig. 7).The finding that hysteresis varied seasonally, decreasing with increasing VWC, further reflects the acclimation to water limitation causing J s to peak in advance of the environmental factors.At our site, dry soils accompanied with high VPD in summer, led to a decreased g s and a more significant control of the stomata on J s relative to the environmental factors.
www.biogeosciences.net/14/4533/2017/Biogeosciences, 14, 4533-4544, 2017   The result that g s increased with increasing VWC (Fig. 10a), along with the synchronization of J s and g s , suggests that J s is more sensitive to g s in low VWC and less so to R s .Due to the incidence of small rainfall events in drylands, soil water supplied by rainfall pulses was largely insufficient to meet the transpiration demand under high mid-day R s , resulting in clockwise loops.Lower values (< 0.4) at our site also support the idea that g s has a greater control on transpiration than R s under situations of water limitation (Fig. 10).Altogether, stomatal control on the diurnal evolution of J s by reducing g s combined with lower sensitivity to meteorological variables during the mid-day dry hours help to reduce water consumption in Artemisia ordosica.Seasonally, plant-moderated reductions in g s and increased hystere-  sis, leads to reduced J s and acclimation to drought conditions.It is suggested here that water limitation invokes a water-conservation strategy in Artemisia ordosica.Contrary to our findings, counterclockwise hysteresis has been observed to occur between J s and R s in tropical and temperate forests (Meinzer et al., 1997;O'Brien et al., 2004;Zeppel et al., 2004), which is reported to be consistent with the capacitance of the particular soil-plant-atmosphere system being considered.Unlike short-statured vegetation, it usually takes more time for water to move up and expand vascular elements in tree stems during the transition from night to day.

Conclusions
The relative influence of R s , T , and VPD on J s in Artemisia ordosica was modified by soil water, indicating J s 's lessened sensitivity to the environmental variables during dry periods.Sap flow was constrained by soil water deficits, causing J s to peak several hours prior to the peaking of R s .Diurnal hysteresis between J s and R s varied seasonally and was mainly controlled by hydraulic stresses.Soil moisture controlled sap-flow response in Artemisia ordosica to meteorological factors.This species escaped and acclimated to water limitations by invoking a water-conservation strategy through the hysteresis effect and stomatal regulation.Our findings add to our understanding of acclimation in desert-shrub species under stress of dehydration.The information advanced here can assist in modeling desert-shrub-ecosystem functioning under changing climatic conditions.Author contributions.DQ and TZ contributed equally to the design and implementation of the field experiment, data collection and analysis, and writing the first draft of the manuscript.XJ gave helpful suggestions concerning the analysis of the field data and contributed to the scientific revision and editing of the manuscript.BW contributed to the design of the experiment.CP-AB and HP contributed to the scientific revision and editing of the manuscript.YB, WF, and YT were involved in the implementation of the experiment and in the revision of the manuscript.
Competing interests.The authors declare that they have no conflict of interest.
Special issue statement.This article is part of the special issue "Ecosystem processes and functioning across current and future dryness gradients in arid and semi-arid lands".It is not associated with a conference.

Figure 1 .
Figure 1.Sap flow per leaf area (J s ) as a function of soil water content (VWC) at 30 cm depth in non-rainy, daytime hours during the mid-growing period from 1 June to 31 August 2013 and 2014.Data points are binned values from pooled data over 2 years at a VWC increment of 0.003 m 3 m −3 .The dotted line represents the VWC threshold for J s .
), respectively, annual means being 224.8 W m −2 , 17.7 • C, 1.03 kPa, and 0.08 m 3 m −3 .The corresponding range of daily means for 2014 were 31.0-369.9W m −2 , 7.1-25.8• C, 0.08-2.5 kPa, and 0.06-0.16m 3 m −3 (Fig. 2a-d), respectively, annual means being 234.9 W m −2 , 17.2 • C, 1.05 kPa, and 0.09 m 3 m −3 .Total precipitation and number of days with rainfall events during the 2013 measurement period (257.2 and 46 days) were about 5.6 and 9.8 % lower than those during 2014 (272.4 mm and 51 days; Fig.2d), respectively.More irregular rainfall events occurred in 2013 than in 2014, with 45.2 % of rainfall falling in July and 8.8 % in August.Drought mainly occurred in May, June, and August of 2013 and in May and June of 2014 (shaded sections in Fig.2d, e).Both years had dry springs.Over a 1-month period of summer drought occurred in 2013.The range of daily J s during the growing season was 0.01-4.36kg m −2 d −1 in 2013 and 0.01-2.91kg m −2 d −1 in 2014 (Fig.2f), with annual means of 0.89 kg m −2 d −1 in 2013 and 1.31 kg m −2 d −1 in 2014.Mean daily J s over the growing season of 2013 was 32 % lower than that of 2014.Mean daily T r were 0.05 and 0.07 mm d −1 over the growing season of 2013 and 2014 (Fig.2f), respectively, being 34 % lower in 2013 than in 2014.The total T r over the growing season (1 May-30 September) of 2013 and 2014 were 7.3 and 10.9 mm, respectively.Seasonal fluctuations in J s and T r corresponded with seasonal patterns in VWC (Fig.2d, f).Daily mean J s and T r decreased or remained nearly constant during dry-soil periods (Fig.2d, f), with the lowest J s and T r observed in spring and mid-summer (August) of 2013.

Figure 2 .
Figure 2. Seasonal changes in daily (24 h) mean shortwave radiation (R s ; a), air temperature (T ; b), water vapor pressure deficit (VPD; c), volumetric soil water content (VWC; d), relative extractable water (REW; e), daily total precipitation (PPT; d), and daily sap flow per leaf area (J s ; f), and daily transpiration (T r , mm d −1 ; f) from May to September for both 2013 and 2014.The horizontal dashed lines in (d, e) represent VWC and REW thresholds of 0.1 m 3 m −3 and 0.4, respectively.Shaded bands indicate periods of drought.

Figure 3 .
Figure 3. Relationships between sap flow per leaf area (J s ) and environmental factors (shortwave radiation, R s ; air temperature, T ; water vapor pressure deficit, VPD; and soil water content at 30 cm depth, VWC) in non-rainy days between 08:00 and 20:00 during the mid-growing season of 1 June to 31 August for 2013 and 2014.Data points are binned values from pooled data over 2 years at increments of 40 W m −2 , 1.2 • C, 0.3 kPa, and 0.005 m 3 m −3 for R s , T , VPD, and VWC, respectively.

Figure 4 .
Figure 4. Sap flow per leaf area (J s ) in non-rainy, daytime hours during the mid-growing season of 1 June to 31 August for both 2013 and 2014 as a function of shortwave radiation (R s ), air temperature (T ), and vapor pressure deficit (VPD) under high volumetric soil water content (VWC > 0.10 m 3 m −3 both in 2013 and 2014) and low VWC (< 0.10 m 3 m −3 , 2013 and 2014).J s is given as binned averages according to R s , T , and VPD based on increments of 100 W m −2 , 1 • C, and 0.2 kPa, respectively.Bars indicate standard error.

Figure 5 .
Figure 5. Regression slopes of linear fits between sap flow per leaf area (J s ) in non-rainy days and shortwave radiation (R s ), vapor pressure deficit (VPD), air temperature (T ), and volumetric soil water content (VWC) between 08:00 and 20:00 during the midgrowing season of 1 June to 31 August for 2013 and 2014.

Figure 6 .
Figure 6.Mean monthly diurnal changes in sap flow per leaf area (J s ) and stomatal conductance (g s ) in Artemisia ordosica during the growing season (May-September) for both 2013 and 2014.Each point is given as the mean at specific times during each month.

Figure 7 .
Figure 7. Seasonal variation in hysteresis loops between sap flow per leaf area (J s ) and shortwave radiation (R s ) using normalized plots for both 2013 and 2014.The y axis represents the proportion of maximum J s (dimensionless), and the x axis represents the proportion of maximum R s (dimensionless).The curved arrows indicate the clockwise direction of response during the day.

Figure 9 .
Figure 9.Time lag between sap flow per leaf area (J s ) and short wave radiation (R s ) in relation to volumetric soil water content (VWC).Hourly data in non-rainy days during the mid-growing season of 1 June to 31 August 2013 and 2014.The lag hours were calculated by a cross-correlation analysis using a 3-day moving window with a 1-day time step.Rainy days were excluded.The solid line is based on an exponential regression (p < 0.05).

Figure 10 .
Figure 10.Relationship between volumetric soil water content (VWC) and (a) stomatal conductance (g s ) in Artemisia ordosica, and (b) decoupling coefficient ( ) for 2013 and 2014.Hourly values are given as binned averages based on a VWC-increment of 0.005 m 3 m −3 .Bars indicate standard error.Only statistically significant regressions (with p values < 0.05) are shown.
Data availability.Data of environmental factors and sap-flow rate in 2013 and 2014 and data of stomatal conductance (g s ) in 2013 and 2014 are available from the following link: http://pan.baidu.com/s/1sl8wgVz.

Table 1 .
Seasonal changes in transpiration (T r ) per month, leaf area index (LAI), and stomatal conductance (g s ) in Artemisia ordosica during the growing seasons (May-September period) of 2013 and 2014.

Table 2 .
Mean monthly diurnal cycles of sap flow (J s ) response to shortwave radiation (R s ), air temperature (T ), and water vapor pressure deficit (VPD), including lag times (h) as a function of R s , T , and VPD.