As part of global climate change, variation in precipitation in arid ecosystems is leading to plant adaptation in water-use strategies; significant interspecific differences in response will change the plant composition of desert communities. This integrated study on the ecophysiological and individual morphological scale investigated the response, acclimation and adaptation of two desert shrubs, with different water-use strategies, to variations in water conditions. The experiments were carried out on two native dominant desert shrubs, Tamarix ramosissima and Haloxylon ammodendron, under three precipitation treatments (natural, double and no precipitation, respectively), in their original habitats on the southern periphery of Gurbantonggut Desert, Central Asia, during the growing season in 2005. Changes in photosynthesis, transpiration, leaf water potential, water-use efficiency, above-ground biomass accumulation and root distribution of the two species were examined and compared under the contrasting precipitation treatments. There were significant interspecific differences in water-use strategy and maintenance of photosynthesis under variation in precipitation. For the phreatophyte T. ramosissima, physiological activity and biomass accumulation rely on the stable groundwater, which shields it from fluctuation in the water status of the upper soil layers caused by precipitation. For the non-phreatophyte H. ammodendron, efficient morphological adjustment, combined with strong stomatal control, contributes to its acclimation to variation in precipitation. On account of its positive responses to increased precipitation, H. ammodendron is predicted to succeed in interspecific competition in a future, moister habitat.
Precipitation is one of the main determinants of the types and distribution of global vegetations. It influences various physiological and ecological processes of plants at different scales. In addition, its seasonality influences the composition, structure and production of plant communities, as well as patterns of water use of dominant species. According to Schwinning & Ehleringer (2001), annual precipitation patterns play a crucial role in shaping plant adaptation in water-use strategies, and in determining the compositions of plant communities in arid environments. Patterns in water use are generally linked with plant phenotypes (especially the architecture of the root system), soil characteristics and specific sites (Flanagan, Ehleringer & Marshall 1992).
Because of its biological and ecological importance, variation in precipitation, as one of the main aspects of global climate change, will influence the subsistence of plants that exist in territorial ecosystems (Watson, Zinyowera & Moss 1996; Walther et al. 2002; Weltzin et al. 2003). However, variation in precipitation has received far less attention than elevated atmospheric CO2 and global warming in the investigation of plant responses to global climate changes. Although plant–water relations have been an important part of ecophysiological research from the outset, the influence of variation in precipitation on plants is little understood, mainly because of the unpredictable nature of precipitation and the complexity of its function. The limited research data have shown that the effects of variation in precipitation on plant growth and ecosystem productivity may be complicated and extensive, and may differ widely among plants and ecosystems (Vitousek 1994; McCarty 2001).
Arid ecosystems are predicted to be one of the ecosystems most responsive to global climate change (Smith et al. 2000). Because of the overall severely restricted water supply in arid ecosystems, precipitation is the crucial limiting factors for plant recruitment, photosynthesis, growth, nutrient dynamics and net ecosystem productivity (Brown, Valone & Curtin 1997; Weltzin & McPherson 2000). Hence, an arid ecosystem tends to show rapidly the effects of variation in precipitation.
In a typical desert environment, characterized by extreme fluctuations in precipitation and other environmental conditions, desert plants have evolved special physiological and morphological traits in the process of adapting to the frequent aridity, torridity and other environmental stresses (Smith, Monson & Anderson 1997), and they exhibit a much higher tolerance to water shortage than other plants (Lawlor & Cornic 2002). However, their metabolism could still be vulnerable to slight variation in water availability (Horton, Kolb & Hart 2001), because the main physiological parameters of desert plants, such as stomatal conductance, photosynthetic rate and transpiration rate (Tr), are variably inhibited during times of water shortage (Gibson 1996). Thus, it can be inferred that variation in precipitation will influence desert plants by affecting the water supply from the upper soil and groundwater. This influence could be a complex combination of physiological activity, individual morphology and long-term adaptive strategy (Ogle & Reynolds 2004).
Some desert shrubs possess physiological and morphological characteristics that are well suited to such integrated research (Lin, Phillips & Ehleringer 1996). It is suggested that the response of desert shrubs to variation in precipitation is a function of the complex interactions between the species of different functional types and the prevailing environmental conditions. Generally, the characteristics of the root system are a main feature in classifying functional types of desert shrubs (Rundell & Nobel 1991), because its morphology or architecture is one of the most important determinants of the availability of soil water and thus is closely related with plant–water relations and photosynthesis (Sperry & Hacke 2002).
Within the current background of global climate change, significant increases in precipitation have been recorded in the arid region of Central Asia over the past 50 years (Folland et al. 2001; Qian & Zhu 2001). In addition, in some transition regions between deserts and oases, the groundwater table has fallen significantly as a result of the overexploitation of groundwater. To understand the effects of these widely acknowledged changes in water conditions, some ecophysiological studies have analysed the water relations of several dominant species in these regions, such as Tamarix ramosissima Ledeb. Fl.Alt., Haloxylon ammodendron (C.A.Mey.) Bunge and Populus euphratica Oliv. (Pyankov et al. 1999; Gries et al. 2003).
An earlier study on T. ramosissima and H. ammodendron showed that their physiological response to precipitation is closely related to water-use strategies coupled with the functional type of the root system (Xu & Li 2006). Neither species showed significant photosynthetic response to rain pulse or depleted upper-soil water. No ecophysiological explanation was found for this lack of response. Thus, we speculated that the maintenance of the water balance of these plants was achieved mainly by morphological adjustment and/or acclimation. Hence, the current study was designed to investigate how adaptive water-use strategies of plants might affect their physiological response and morphological adjustment and/or acclimation to variation in precipitation.
The relation between individual morphology of desert shrubs, especially their root-system distribution and foliage growth, and environmental water availability is crucial for plant survival. However, because of the difficulty of field experiments, the relevant theories remain largely untested, especially in Central Asian deserts. In addition, research integrating ecophysiology with morphology of desert shrubs in their natural habitats remains inadequate. This represents a gap in our knowledge of the adaptive strategies of desert plants to water shortages, and of the inconsistencies among plant responses to a variation in water conditions. Because of the increasing environmental changes resulting from global climate change and human activity, such studies are urgently required in order to predict the future of the arid ecosystem.
MATERIALS AND METHODS
Habitats, species and precipitation treatments
During the whole growing season of the studied species in 2005 (from the 125th until the 252nd Julian Day), the experiments were conducted in the vicinity of the Fukang Station of Desert Ecology, Chinese Academy of Sciences, in the hinterland of the Eurasian continent (44°17′N, 87°56′E, 475 m a.s.l.). The station is 8 km from the southern edge of the Gurbantonggut Desert and 72 km north of the highest peak of the eastern Tianshan Mountains. The plain area of this region has a continental arid temperate climate, with a hot, dry summer and cold winter; annual mean temperature is 6.6 °C, annual mean precipitation is ≈ 160 mm, and ≈ 40% distributes in the growing season (from May to September); pan evaporation (E) is ≈ 2000 mm. The plain area in the vicinity of the station is typical temperate desert with varying soil salinity. In areas of high salinity and high groundwater table (< 5 m), native vegetation is usually dominated by T. ramosissima. In areas of low salinity and deep groundwater table (> 5 m), H. ammodendron is usually dominant in the shrub community. Companion species are rare in all plant communities of this region.
The experiments were carried out in two desert shrub communities dominated by T. ramosissima and H. ammodendron, respectively. The distance between the two sites is ≈ 5 km. The two sites are of similar climatic characteristics, and soil at both sites is heavily textured saline–alkali gault of moderate salinity. In the habitat of T. ramosissima, the highest recorded salinity within surface layers was ≈ 2.5%, and the depth to the groundwater table was 3.0–3.3 m. In the habitat of H. ammodendron, the highest recorded salinity within surface layers was ≈ 1.3%, and the depth to the groundwater table was ≈ 5.2 m.
Three precipitation treatments were applied in each habitat, representing natural precipitation (a 20 × 20 m2 section, no treatment was carried out on the shrubs inside), double precipitation (after each rainfall event, irrigation coordinated with the precipitation was imposed evenly over a 20 × 20 m2 ground surface, with the studied shrubs in the center of the section) and no precipitation (the shrubs were sheltered from rainfall by a 20 × 20 m2 awning). Under each treatment, five plants in favourable growth status were selected for physiological monitoring and biomass observation. The 15 individual shrubs of each species were all at about the same age, with approximately the same canopy size and in similar physiological conditions at the beginning of the treatments.
In 2005, precipitation during the growing season approximated the average for growing seasons in the region (≈ 70 mm). As usual, July was the wettest month (Fig. 1). The difference in precipitation and evaporation between the two habitats was negligible. The meteorological data were obtained by the Campbell automatic weather station (Campbell Scientific, Logan, Utah, USA).
Measurement of leaf water potential (ψl), transpiration and photosynthesis
All experiments were carried out on clear days throughout the growing season. A model 3005 Pressure Chamber (PMS Instrument Company, Albany, NY, USA) was used to measure the ψl of each species under the three precipitation treatments during the growing season. Pre-dawn leaf water potential (ψpd) was measured 20 min before sunrise, and midday leaf water potential (ψm) was measured at solar noon. Small branches with sufficient leaves were selected, and the sampling was repeated twice per individual plant. Thus, for each species, 10 replicates were taken to determine the average value of ψl under each precipitation treatment at a given time during the growing season.
The Tr was measured by a compensated heat-pulse system developed by Cohen et al. (Cohen, Fuchs & Cohen 1983). Sixty heat-pulse probes (10 for each treatment of each species) were installed on branches with a diameter of 8–15 mm, and the sap flow rate for each branch (Tr of the branches) was recorded every 0.5 h. To overcome the effect of variation in branch size, the Tr value was normalized on a leaf-area basis. To quantify the leaf area of each branch, all the foliage on each selected branch was photographed every two weeks with a 6 × 108 pixel digital camera (Canon 300D, Canon Inc., Tokyo, Japan). The leaf surface area of each branch was calculated from the photographs using CI-400 CIAS software (Computer Imaging Analysis Software, CID Co., Logan, UT, USA). The Tr value was then converted to a leaf-specific value according to the measured leaf surface area of each branch.
The photosynthetic light-response curves of the species were measured by a Li-6400 portable photosynthesis system (Li-Cor, Lincoln, NE, USA). In-chamber photosynthetic photon flux density (PPFDi) was controlled by use of a 20 × 30 mm2 leaf chamber with a light source (red + blue 6400-02B). The PPFDi gradient was set from 0 to 2200 µmol m−2 s−1, with intervals of 100 µmol m−2 s−1. Gas flow rate was set at 400 µmol s−1 to maintain a reference relative humidity at 20–30%, which is close to ambient humidity. Chamber temperature was controlled at 30 °C. A CO2-injecting device was attached to the system to control reference CO2 concentration [(CO2R)] at 400 µmol mol−1. The measurements were taken from 1000 to 1600 h, local time, on the day of the experiment. For each species under each precipitation treatment, two sets of leaves were measured from each of the five sample plants. The leaves measured were healthy and were the youngest mature ones. The detailed procedure has been described previously (Xu & Li 2006).
Advanced regression analysis of the non-linear curve showed that the relationship between net photosynthetic rate (Pn) and PPFDi was best fitted by exponential MnMolecular function y = A[1 – e–k(x–xc)], in which x is PPFDi, y is net photosynthetic rate, parameter A is net photosynthetic rate at light saturation point (Ps), xc is PPFDi at light compensation point (Ic) and k*A is apparent quantum efficiency of photosynthesis (Φ). From a light-response curve fitted to the average data from each set of reduplicate photosynthesis experiments, the values of Ps, Ic and Φ were calculated to indicate the photosynthetic capacity in a certain condition. The denotation of each parameter follows Sage (1994).
The photosynthetic water-use efficiency [WUE (µmol CO2·mmol−1 H2O), Pn (µmol CO2 m−2 s−1)/Tr (mmol m−2 s−1)] was calculated from Pn and Tr (obtained from sap flow measurement) at PPFD = 1600 µmol photon m−2 s−1.
Measurement of leaf area and branch biomass
Above-ground biomass accumulation of each species was observed throughout the growing season. Under each treatment, 10 well-growing branches were selected and labelled on each of the five sample plants. All the foliage on each branch was photographed with a 6 × 108 pixel digital camera (Canon Inc.), at intervals of two weeks, from the beginning to the end of the experimental period. The total leaf surface area of each branch was calculated from the photographs using CI-400 CIAS software (CID Co.). Change in the stem surface area of the labelled branch was traced in the same way. At the end of the experiment, the labelled branches were cut. Fresh and dry mass of leaves and stems on each branch was weighed to calculate tissue water content and define the relation between surface area and dry mass of the foliage/branch of each species. On the basis of this relation, the surface area of each branch was converted into its dry mass, and foliage dry mass of a plant was converted into total leaf area. Seasonal change in leaf area per branch was recorded. After the data from the first measurement were standardized among three treatments, branch growth rate during the period between every two sequential measurements was calculated. The data from 50 reduplicates were averaged to represent the general condition of foliage area expansion and branch biomass accumulation under each precipitation treatment.
The data for branch biomass and branch leaf area were standardized (transformed) on the basis of the measurement at the beginning of the experiment, in order to eliminate the difference among individuals.
Investigation of soil water status and root-system distribution
The investigation of soil water status and root-system distribution was carried out at the end of the treatments (in the early September, from the 248th until the 252nd Julian Day). Soil moisture content and salinity under each precipitation treatment were measured. Soil was sampled by auger from 0 to 3 m depth at intervals of 0.25 m, and repeated in five random positions. Soil moisture content was determined by a conventional weighting method. Salinity data were obtained from ionic analysis. The average data showed an approximation of the effect of precipitation treatments on soil water status in each habitat.
The intact root systems of the two species were excavated to investigate their functional types and water-use strategies. For each treatment, three plants with similar canopy size were randomly selected for excavation. The inner diameter of root excavation ditch was ≈12 m for H. ammodendron, and 18 m for T. ramosissima, much larger than canopy diameter in both cases. For the detailed procedure, see Xu & Li (2006). For each plant, the surface area of fresh feeder roots was calculated for each 0.1 m of the depth profile, in order to derive an overview of the vertical root distribution.
Dry mass of the roots, foliage and stems was recorded.
Data analysis and charting used the statistics software Origin7.0 (OriginLab Corp., Northampton, MA, USA.). Descriptive statistics was used to calculate means and standard deviations from each set of reduplicates. One-way analysis of variance (anova) was used to test for significance in differences of contrasting treatments.
Ecophysiological responses to variation in precipitation
During the growing season, the ψl of T. ramosissima largely maintained a constant and relatively high level (Fig. 2a), which indicated a generally superior water status attributable to a stable water source. In addition, there was no significant difference in either ψpd, ψm or Tr among the three treatments (P = 0.01), which showed that ψl and Tr of T. ramosissima did not respond to precipitation during the entire season (Fig. 2a,b). Tr and Ps showed a similar pattern during the course of the season (Figs 2b & 3a). Both Φ and Ic of T. ramosissima were maintained at a constant value over time (Fig. 3b,c), which shows that the photosynthetic capacity of T. ramosissima was not inhibited by water deficiency. Neither Ps nor Φ nor Ic showed any significant difference among treatments (P = 0.01). Hence, the leaf-scale carbon assimilation of T. ramosissima was not influenced by precipitation.
The photosynthetic WUE of T. ramosissima fluctuated slightly over time, without significant difference among treatments (P = 0.01, Fig. 4a).
Compared with T. ramosissima, the ψl of H. ammodendron was more season-dependent and increased after rainfall events under natural precipitation (Fig. 2c). In the absence of precipitation, ψpd, ψm and Tr continually decreased over time in response to the lack of water. The values of ψl and Tr under double precipitation were significantly higher than those under natural precipitation (P = 0.05, Fig. 2c,d).
Although the values for ψ indicated that the plant water status of H. ammodendron responded to variation in precipitation (Fig. 2c), Tr under natural precipitation showed no significant seasonal change (Fig. 2d). The distinct seasonal change in Ps showed that the growth climax of H. ammodendron was in mid-June, and was not closely related with rainfall events (Fig. 3d). In addition, the data for the three photosynthetic parameters were not significantly different among treatments (P = 0.01, Fig. 3d–f), indicating that the leaf-scale photosynthetic capacity of H. ammodendron was not affected by precipitation.
Under natural precipitation, the photosynthetic WUE of H. ammodendron decreased slightly in the wettest months (Fig. 4b). WUE differed significantly among treatments (P = 0.05), indicating that variation in precipitation affected the photosynthetic WUE of H. ammodendron. Under double precipitation, the mean was lower than the value under natural precipitation. In the absence of precipitation, WUE increased at times of depleted soil water (Fig. 4b) and was much higher than the values under the other two treatments.
Above-ground biomass accumulation and root distribution
Variation in precipitation did not significantly influence the seasonal patterns of branch biomass and leaf area per branch in T. ramosissima (P = 0.01, Fig. 5a,b). In addition, there was no significant difference among treatments in the tissue water content at the end of the experiment (P = 0.01, Table 1). The climax of branch growth rate occurred in late July (Fig. 5b), and it showed a strong temporal relationship with photosynthesis (Fig. 3a) and transpiration (Fig. 2b). Leaf area per branch increased over time, and the leaf area increment in July accounted for nearly half the total increment (Fig. 5a).
Table 1. Tissue water content of Tamarix ramosissima and Haloxylon ammodendron under three treatments at the end of the growing season
Values are mean ± SE. Values in the same row labelled with the same letter are not significantly different at P = 0.01, and values with different letters are significantly different at P = 0.05.
59.2 ± 0.5a
59.8 ± 0.8a
60.9 ± 1.0a
37.5 ± 0.6a
36.6 ± 1.8a
36.9 ± 1.1a
78.3 ± 1.6a
87.5 ± 1.9b
77.5 ± 1.5a
32.7 ± 2.5a
39.6 ± 0.6b
33.1 ± 2.3a
The biomass accumulation per branch in H. ammodendron significantly differed among three treatments (P = 0.05, Fig. 5c,d). Under natural precipitation, the climax of branch growth rate occurred in early July (Fig. 5d), correlated with photosynthesis; biomass accumulation per branch gradually decreased and ceased in mid-August; and leaf area per branch increased steadily over time (Fig. 5c). Under double precipitation, branch growth rate was higher than under the other two treatments, especially in the middle and at the end of the growing period (Fig. 5d); the mean leaf area per branch was larger at any time during the growing season (Fig. 5c). However, the difference was not significant between natural and double precipitations at P = 0.01. At the end of the experiment, the tissue water content under double precipitation was significantly higher than under the other two treatments (Table 1). Because of its specialized mesophyll, leaf water content of H. ammodendron was higher than that of T. ramosissima. However, the ψl of H. ammodendron was much lower than in T. ramosissima, because the bound water in leaves does not usually contribute to the alleviation of moisture stress.
Absence of precipitation severely inhibited branch biomass accumulation of H. ammodendron (Fig. 5c). Branch growth and leaf expansion slowed at times of depleted soil water. The branches and leaves began to fall off in late June, although the defoliation rate gradually decreased over time.
In both habitats, soil moisture content in the 0–1.5 m layer was significantly influenced by precipitation (P = 0.05; Fig. 6a,c).
In the habitat of T. ramosissima, within the 0–0.5 m layer, the average soil moisture content measured under the no-precipitation treatment was < 4.1%, compared with 8.8% under natural precipitation and 10.9% under double precipitation. The soil moisture content of the 1.5–3.1 m layer remained high and was not significantly different among treatments, because of the stable groundwater at a depth of 3.1 m. The higher salinity near the soil surface resulted from salt excreted from the litter fall and then leached into the upper soil by rainfall (Fig. 6b); below 1.5 m, soil salinity was lower and stable.
Under natural precipitation, the main root of T. ramosissima extended to an average depth of 2.79 m, close to the 3.1 m groundwater table. More than 50% of the feeder roots (in terms of surface area) were distributed in the depth interval of 2.3–3.1 m; no active absorbing roots were found in the soil above 0.3 m (Fig. 7a). The density of feeder roots increased with depth, and abundant feeder roots germinated and extended horizontally at depths close to the groundwater table, indicating that shrubs of this species are able to effectively use groundwater.
Under double precipitation the root system showed a similar distribution. With no precipitation, there were fewer feeder roots distributed in the soil layer above 1.5 m than in the other two treatments, the area of feeder roots was significantly greater in the layer below 2 m, and more feeder roots germinated just above the groundwater.
In the habitat of H. ammodendron, within the 0–1 m layer, the average soil moisture content in the absence of precipitation was < 4%, compared with 6.6% under natural precipitation and 9.2% under double precipitation. Below 2.5 m, the soil moisture content remained stable and was not significantly different among treatments (Fig. 6c). The difference in soil salinity among treatments was also caused by rainfall leaching (Fig. 6d).
Under natural precipitation, the main root of H. ammodendron extended to an average depth of 3.32 m, far above the groundwater at 5.2 m depth (Fig. 7d). More than 90% of the feeder roots were in the 0–0.9 m soil layer, and no roots were found below 3.5 m. Such a shallow root system indicated that this species depended on summer precipitation as its primary water source.
Under double precipitation, the root system of H. ammodendron showed a similar distribution below 0.5 m, whereas in the layer above 0.5 m the number of feeder roots increased significantly, with ≈ 80% of the feeder roots distributed in the 0–0.5 m soil layer.
In the absence of precipitation, no active absorbing roots were found in the dry soil above 0.7 m (Fig. 7f); ≈ 85% of the feeder roots were in the 1.6–2.7 m soil layer; and feeder roots tended to extend downwards into the 3.5–4.5 m layer in order to obtain water from the deeper soil. This indicated that this species might be able to use groundwater with the aid of capillary rise in the heavy-textured soil at times of extreme drought. In addition, the shoot : root ratio in dry mass and leaf : root ratio in surface area were significantly lower than under the other two treatments (P = 0.05, Table 2).
Table 2. Biomass allocation pattern and leaf : root area ratio in the individual Haloxylon ammodendron plant, under three treatments
Root dry mass (g)
Root surface area (m2)
Shoot dry mass (g)
Leaf surface area (m2)
Shoot: root (dry mass)
Leaf: root (surface area)
Values are mean ± SE. Root dry mass, root surface area and shoot dry mass are averages of original data from measurement at the end of the growing season. Leaf surface area was calculated from the original data on leaf dry mass. Values in the same column labelled with the same letter are not significantly different at P = 0.01, and values with different letters are significantly different at P = 0.05.
2242 ± 261a
1.02 ± 0.07a
3180 ± 376a
0.42 ± 0.03a
2573 ± 305b
1.38 ± 0.11b
4014 ± 575b
0.55 ± 0.03b
1809 ± 194c
0.55 ± 0.06c
1845 ± 142c
0.10 ± 0.01c
Relationship between ecophysiological responses and morphological adjustment
During the growing season, the Tr of T. ramosissima was positively and closely correlated with E on the corresponding day (Fig. 8a), while the photosynthetic WUE was negatively correlated with it (Fig. 8b). The seasonal change in Ps showed a trend similar to that in Tr (Fig. 8c). The branch growth rate increased exponentially with the increase in Ps (Fig. 8c), as did the increase in branch leaf area (Fig. 8d).
For H. ammodendron, the correlation between Tr and ψm depended on the value of ψm (Fig. 9a): Tr and ψm were highly correlated (R2 = 0.9) when ψm was lower than –4 MPa, indicating the influence of ψm on stomatal conductance. The correlation declined significantly (R2 = 0.4) when ψm remained relatively high. During the growing season, the correlation between photosynthesis and ψm was not statistically significant (R2 < 0.1); the correlation between WUE and ψm was negative and significant (Fig. 9b), indicating that the WUE of H. ammodendron increased under water stress. The above-ground biomass accumulation was positively correlated with ψm (Fig. 9c).
The earlier experiment showed that photosynthesis in the two species did not respond to sustained drought or a pulse of heavy rain (Xu & Li 2006). The current study revealed that this photosynthetic consistency was achieved either by strategic adaptation in water use (in T. ramosissima), or by integrated physiological and morphological regulations (in H. ammodendron).
Water use strategy of T. ramosissima
Throughout the growing season of T. ramosissima, there was no significant difference in water relation or branch biomass accumulation among the precipitation treatments. That is, variation in shallow water status does not affect its physiological activity or growth, indicating that it can efficiently avoid the effects of water deficiency in upper soil with its phreatophytic root system.
The comparison among treatments shows that the feeder roots in the upper soil responded to water stress, but the overall performance of the plants was not affected by a rain pulse. There were few feeder roots in the 0–0.5 m upper soil even after the double precipitation treatment. With no precipitation, the number of feeder roots decreased significantly in the soil layer above 1.5 m. However, by developing feeder roots at depths close to the groundwater table (Fig. 7a–c), the plant could obtain enough water to sufficiently offset the loss in the upper soil layer. This agrees with the mechanism proposed by Donovan & Ehleringer (1994). During the experiment, the groundwater enabled T. ramosissima to maintain a constant water supply to the foliage and a stable ψpd over time (Fig. 2a). Thus, it can be inferred that the variation in precipitation will not easily disturb the water and carbon balance of T. ramosissima.
Schwinning & Ehleringer (2001) proposed that in arid habitats, plants can adapt phenotypically to maximize use of deeper soil water by having a large root : shoot ratio, a predominantly deep root system, and lower leaf conductance with low stomatal sensitivity to water availability. The special behaviour of stomata is seen as one of the essential mechanisms in the water-use strategy of Tamarix spp. (Anderson 1982) and desert plants (Ehleringer 1995).
The growth climax of T. ramosissima occurred in mid to late July (Figs 2b & 3a). The close correlation between Tr and E (Fig. 8a) indicated that the seasonal variation of Tr resulted mainly from the seasonal change in atmospheric demand. The opposite seasonal trends in Tr and WUE (Figs 2b & 4a), and the exponential decline in WUE with the increase in E (Fig. 8b), indicated a stomatal surplus and thus the loose stomatal control of T. ramosissima under strong evaporative demand. The accordant seasonal changes in Tr and Ps (Fig. 8c) showed that greater stomatal conductance might contribute to higher carbon assimilation, but at the cost of lower photosynthetic WUE. This characteristic supports the prediction of Schwinning and Ehleringer regarding plants that use deep waters in arid ecosystems (Schwinning & Ehleringer 2001). It indicates that as a result of trade-off, T. ramosissima tends to maximize its carbon gain at the cost of higher water consumption, attributed to sufficient groundwater supply. Even under increased precipitation, T. ramosissima will probably maintain its water-use pattern in the foreseeable future.
Adaptation of H. ammodendron to variation in precipitation
Stomatal control and photosynthetic characteristics
The ψl, Tr and WUE of H. ammodendron all depended on precipitation. The lower WUE after rain events and under double precipitation suggested a lack of tight stomatal control under favourable water conditions (Fig. 4b). In the absence of precipitation, Tr decreased at times of depleted soil water (Fig. 2d), indicating efficient stomatal control under drought. However, its transpiration was always lower than 2.7 mmol H2O m−2 s−1, and not significantly enhanced by a seasonal increase in atmospheric evaporation demand (R2 between maximal Tr and daily E is less than 0.002), indicating the ability of its assimilative organ to conserve moisture; this is characteristic of strongly drought-tolerant plants.
The negative correlation between WUE and ψm (R2 = 0.6, Fig. 9b) showed that high sensitivity to water availability ensured an increase in WUE under water stress. A close correlation (R2 = 0.9) between Tr and ψl when ψl was lower than –4 MPa (Fig. 9a) demonstrates the immediate impact of ψl on stomatal conductance in severe drought. The correlation declined significantly (R2 = 0.4) when ψl remained relatively high, indicating that under moderate water conditions, stomatal conductance was directly affected by factors other than ψl. Root-derived abscisic acid may play a crucial role in stomatal closure and defoliation, even preceding the drawdown of ψl; furthermore, defoliation and stomatal closure may delay the drawdown of ψl (Zhang & Davies 1990; Buckley 2005). However, it can be inferred that if the root system cannot reach the groundwater soon enough, the available water will be depleted; the ψl of all the surviving leaves will decrease beyond the lower limit. By then, the photosynthesis and survival of H. ammodendron will be threatened.
Compared with T. ramosissima, H. ammodendron showed some photosynthetic characteristics of typical C4 species, such as a lower Φ at low PPFD, higher Ic (Fig. 3e,f) and higher WUE (Fig. 4b), which all contributed to its drought tolerance. The consistent values of Ic and Φ over time (Fig. 3e,f) showed that its photosynthetic activities could be maintained at a stable high level within a wide range of ψl from –4.7 to –1.5 MPa (Figs 2c & 9b), partly as a result of efficient osmotic adjustment in mesophyll cells (Pyankov et al. 1999).
Under natural precipitation, the seasonal fluctuation in Ps showed the growth climax in late June (Fig. 3d). It did not appear to be caused by increased stomatal conductance resulting from the precipitation because, unlike T. ramosissima, Ps in H. ammodendron did not change along with Tr seasonally or among treatments (Fig. 2d). As the values of Ic and Φ remained consistent over time (Fig. 3e,f), the most probable reason for the seasonal dependence of Ps could be seasonal changes in the number, volume or composition of chloroplasts (Oguchi, Hikosaka & Hirose 2005; Yamori, Noguchi & Terashima 2005).
Photosynthesis showed a similar seasonal pattern under the other two treatments (Fig. 3d). The clearly synchronous change in photosynthetic capacity among treatments indirectly demonstrated the efficiency in morphological adjustment.
Adjustment in feeder roots and leaves
The balance between water supply and demand in a plant can be maintained by morphological adjustment of the plant root and shoot systems, as has been found in some shrub species (Donovan & Ehleringer 1994). On the one hand, the absorbing roots in the upper soil may continue to adjust efficiently during progressive drought or after rain; on the other hand, shoot size may shrink by partial defoliation or increase by leaf emergence/expansion in response to change in soil water within the root zone. There are indications that desert shrubs may adjust their root systems towards the optimal phenotype that can maximize water availability (Schwinning & Ehleringer 2001). In addition, according to Eagleson's ecological optimality theory (Eagleson 1982), desert shrubs tend to optimize the density of their assimilative organs to adapt to environmental water availability.
In the current study, the ψl fluctuated along with rainfall events throughout the growing season of H. ammodendron (Fig. 2c), and differed significantly among the three treatments (P = 0.05), indicating that its water status was directly affected by water availability in the upper soil. Such dependence of ψl on precipitation was caused by the shallow distribution of its root system (Fig. 7d–f); the water uptake from the upper soil provided the bulk of its water consumption. Increased precipitation enabled more feeder roots to develop in the upper soil to support more assimilating organs with normal photosynthesis (Fig. 5c, Table 2).
Under double precipitation, carbon assimilation increased significantly at the individual plant scale (Fig. 5c) without any change in the photosynthetic capacity at the leaf scale (Fig. 3d). More photosynthetic products were allocated to the shoot system (Table 2). However, the ratio between leaf area and root area was similar with that under natural precipitation (P = 0.01, Table 2). The ratio 0.4 can be regarded as an eigenvalue of H. ammodendron in moderate water conditions and may remain constant over a certain range of water availability, being derived from the area of leaf that the unit-area root can supply with the requisite water.
The decrease in summer precipitation restricted the physiological activity and growth of H. ammodendron and shortened its growth period significantly (Fig. 5c). Severe water stress resulted in a significant decrease in the ratio of leaf area to root area (P = 0.05, Table 2), indicating a decrease in the area of the assimilative organ that the root system could support. A much lower leaf : root area ratio indicated that its original water balance had been disrupted. Although feeder roots developed in deeper soil layers in order to increase the chance of access to water before the severe drought imperiled the photosynthesis of the residual leaves (Fig. 7f), the increase could not compensate the loss in the upper soil (Table 2). However, such a trend in morphological response of the root system might suggest that H. ammodendron is a facultative phreatophyte. It might occur only in the site with a deeper water table, as H. ammodendron may be competitively excluded by the salt-excreting T. ramosissima in higher-water-table sites (Fig. 6b).
The decrease in area and water supply capacity of the feeder roots was accompanied by a decrease in leaf area through defoliation (Fig. 5c). This morphological adjustment maintained the balance between the water supply to the root system and the water demand of the shoot system, and ensured a consistent photosynthetic capacity of the surviving leaves (Fig. 3c). However, the carbon assimilation of the individual plant was significantly reduced by the absence of precipitation (Table 2). In addition, the decreased shoot : root ratio indicated that more photosynthetic products were allocated to the root system.
Defoliation and the higher quota for the root system are two sides of the morphological trade-off that ensures normal photosynthesis and subsistence for H. ammodendron while at the same time the plant approaches a new water source. In Fig. 9c, the branch growth rate and leaf area were positively correlated well with the ψm during the corresponding period. When ψl was lower than –4.3 MPa, the assimilative organ defoliated (Fig. 9c, where leaf area increment < 0). The sensitivity of branch growth and leaf area increment to ψl indicated an effective morphological adjustment towards variation in soil water. In addition, based on the calculated leaf area of the whole plant and the measured leaf area per branch, the number of total branches in the plant was calculated, and the standardized average was 110, 118 and 100 under natural precipitation, double precipitation and no precipitation, respectively; this indicated that new branches were initiated under double precipitation, and old ones fell off under the no-precipitation treatment. Such morphological adjustment contributed greatly to its strong drought tolerance and ensured its normal photosynthesis and survival at the physiological scale.
The comparison among treatments indicates that variation in precipitation significantly affects the individual-scale carbon gain and biomass allocation pattern of H. ammodendron. Variation in precipitation will change the existing water-use strategy of the root system and, consequently, the architecture of the whole plant.
The integrated study of detailed ecophysiological and morphological responses furthers our understanding of the morphological adaptation of desert shrubs under variation in precipitation. The functional type of the root system directly determines the response and acclimation to the predicted changes in future water conditions. The phreatophyte T. ramosissima is not significantly influenced by precipitation, but the decline in groundwater will threaten its survival. In contrast, H. ammodendron has strong positive responses to increased precipitation; and the efficiency in morphological adjustment, which is based on its physiological characteristics, will benefit its survival and competition in the future moister habitats.
We thank all staff of the Fukang Station of Desert Ecology for their excellent field and laboratory assistance; special thanks to Ms. Xie Jing-Xia for her help in carrying out biomass observations; thanks also to Ms. Zhou Bin for her work on soil analysis. This work was financially supported by grants from the National Natural Science Foundation of China (grants No. 30570286 and No. 40471048).