Variation in Hydraulic Architecture and Gas-Exchange in Two Desert Sub-Shrubs, Hymenoclea Salsola (T

Oecologia (2000) 125:1Ð10 © Springer-Verlag 2000

Jonathan P. Comstock Variation in hydraulic architecture and gas-exchange in two desert sub-shrubs, Hymenoclea salsola (T. & G.) and Ambrosia dumosa (Payne)

Received: 13 December 1999 / Accepted: 31 March 2000

Abstract Adjustment of hydraulic architecture in re- per unit photosynthetic canopy by increasing allocation sponse to environmental conditions was studied in two to an organ which simultaneously performs photosyn- warm-desert sub-shrubs, Hymenoclea salsola and Am- thetic, support, and transport functions. brosia dumosa, both at the level of genetic adaptation along a climatic gradient and plastic response to immedi- Key words Climatic ecotypes á Hydraulic limitation á ate growth conditions. Individuals of both species origi- Hydraulic signaling á Plant morphology á Allocation nating from southern populations developed higher leaf- specific hydraulic conductance in the common greenhouse than individuals from northern populations. Hy- Introduction draulic conductance was higher in plants grown at high temperature, but did not vary as a function of growth rel- Relationships between leaf-specific hydraulic ative humidity. Hydraulic conductance was not correlat- conductance (LSC) and transpiration (E) ed within species with individual variation in vessel diameter, cavitation vulnerability, or root:shoot ratio, but Hydraulic architecture is increasingly studied with re- was strongly, negatively correlated with the fraction of spect to the limitations placed by within-plant water total plant biomass allocated to leaves. For both species, transport on plant productivity. Stomata play a well- stomatal conductance (gs) at high leaf-to-air vapor pres- understood key role as the control point for integrating ν sure difference ( ) was tightly correlated with variability the often conflicting needs to capture CO2 from the at- in hydraulic conductance, as was the sensitivity of sto- mosphere for photosynthesis and growth, and to limit matal closure to increasing ν. Experimentally increasing water loss from plant tissues to avoid dehydration shoot water potential by soil pressurization, under condi- (Cowan 1977). A substantial body of empirical data and tions where high ν had already caused stomatal closure, theory has been produced over the past few decades on led to substantial stomatal reopening in both species, but the degree of limitation actually imposed on photosyn- recovery was significantly higher in H. salsola. Hydrau- thesis and growth by stomatal closure (Farquhar and lic conductance was higher in H. salsola than A. dumosa. Sharkey 1982), and how stomata should behave in re- H.salsola also differed from A. dumosa by being a repre- sponse to environmental variability to maximize carbon sentative of a highly specialised group of desert shrubs gain over a range of different time-scales (Aphalo and which use the twigs as a major photosynthetic organ. Jarvis 1993; Ball et al. 1987; Cowan and Farquhar 1977; The southern population of H. salsola produced far few- Leuning 1995; Monteith 1995). In recent years, this area er leaves and relied much more heavily on twig photo- has seen considerable development in two related areas: synthesis than the northern population. At the whole- (1) the potential role of hormones, especially those plant level, increased reliance on twig photosynthesis borne in the transpiration stream itself, to make stomata was associated with higher leaf-specific hydraulic con- responsive to root:shoot communications (Dodd et al. ductance, but equivalent whole-plant photosynthesis on 1996; Schurr and Schulze 1996; Tardieu 1996; White- Ð1 either a dry weight (µmol CO2 g ) or nitrogen basis head 1998), and (2) the role of hydraulic signals which Ð1 (µmol CO2 g )). This suggests that twig photosynthesis could be transduced either in the root as a function of might be one way of increasing hydraulic conductance root and soil water status, or at the photosynthetic organ where, at the end of the transpiration stream, plant water J. Comstock (✉) Boyce Thompson Institute, Tower Road, Ithaca, NY 14853, USA potentials will be lowest (Comstock and Mencuccini e-mail: [email protected] 1998; Fuchs and Livingston 1996; Saliendra et al. Tel.: +1-607-2541214, +1-607-2541242 1995). 2 Transduction of a water-status signal at the photosyn- Methods thetic organ could be especially important if the hydraulic conductance of the plant, and the consequent water po- Plant material tential gradients between root and shoot, are variable and potentially limiting to plant function. Considerable evi- Seed of H. salsola (T. & G.) and Ambrosia dumosa (Payne), both subshrubs of the Mojave and Sonoran deserts of western North dence exists that hydraulic conductance is indeed partial- America, was collected from natural populations and grown in the ly limiting. Strong correlations have been observed be- greenhouse at the Boyce Thompson Institute for Plant Research in tween stomatal and hydraulic conductances in both crop Ithaca, NY (300 m). In the wild, H. salsola generally occurs in (Meinzer et al. 1990; Mencuccini and Comstock 1999; deep sandy or gravelly soils, especially intermittent stream beds Sober 1997; Sohan et al. 1999; Sperry and Pockman (desert washes), while A. dumosa is broadly distributed on slopes and flats, often with thin soil and minimal water-holding capacity. 1993) and wild plant species (Bond and Kavanagh 1999; For both species, seed collection sites were chosen from both the Irvine et al. 1998; Meinzer et al. 1999; Ryan and Yoder southern and northern extremes of the natural range. The northern 1997). Several studies have also reported that short-term collection site was dominated by Larrea scrub and Joshua tree manipulations of shoot water status can be directly linked woodland just south of the Beaverdam mountains at 945 m elevation and 37.00¡N latitude. The southern seed-collection site was lo- to changes in stomatal aperture (see previous paragraph). cated in similar topography in the Organ Pipe National Monument on the Arizona-Mexico border at 512 m elevation and 31.90 N latitude. These sites had strongly contrasting conditions during the Relationship to climate growing season, driven largely by the seasonality of precipitation (Mencuccini and Comstock 1997). The northern site had a strong unimodal precipitation pattern with maximums in the winter The expected relationship between plant hydraulic con- months, and most plant growth in the spring as temperatures ductance and productivity should be very sensitive to cli- warmed. The southern site had milder winters and a bimodal pre- matic factors that affect photosynthetic water-use effi- cipitation pattern. In the south, both spring and summer growing ciency. In this context, the intermountain west of North periods could regularly support activity by the study species, and the growing season was warmer. America presents some excellent experimental gradients. Due, in part, to the presence of north-south-running mountain chains bounding it on both the east and west, a Cultural conditions continuous belt of aridland ecosystems runs from north- The plants were grown in 30-dm3 pots in a soil mix of 3:1:1 fritted ern Mexico to southern Canada. Although mostly classi- clay (Turface):silica sand:pasteurized topsoil, and were watered fied as arid or semi-arid, these ecosystems vary greatly daily with nutrient solution containing 55:18:55 ppm N:P:K from in mean annual temperature and the seasonality of pre- Peter’s Excel. Photoperiod from combined artificial (an alternating cipitation, both of which strongly affect the leaf-to-air bank of 1000-W high-pressure Na vapor, 1000-W Super Metal Ha- ν lide, and 150-Watt incandescent floodlights) and natural lighting vapor pressure difference ( ) driving water-loss during was 12 h with a total irradiance (400Ð700 nm) of 44 mol mÐ2 dayÐ1 the growing season. Comstock and Ehleringer (1992) re- in all treatments. All treatments were set up in well-ventilated ported that some warm-desert species occupying the greenhouses with internal fans for stirring foliage. CO2 concentra- southern half of this gradient experience a more than tions were monitored continuously by a single infra-red gas-ana- ν lyser (IRGA) (Horiba, model PIR-2000, Irvine, Calif., USA) which 2-fold difference in growing-season among different cycled continuously between air sampling lines from each of the populations, and, for Hymenoclea salsola, showed a cor- three greenhouses and an outside reference line. Mean CO2 was relation between an index of growing-season ν and inter- 375/390 µmol molÐ1 (day/night). Although variations in mean daily Ð1 population variation in both tissue-level intrinsic water {CO2} of up to 15 µmol mol were seen as a function of different weather patterns, all the greenhouse bays and outside air had the use efficiency (i.e., intracellular [CO2] as indicated by same daily mean values±1.0 µmol molÐ1. carbon isotope discrimination), and also canopy architec- Plants from both populations were grown under three contrast- ture (i.e., relative contribution of twigs to the photosyn- ing conditions of temperature and humidity to test whether tem- thetic canopy). perature itself or ν during growth had a greater effect on the plastic development of hydraulic conductance. These growth treat- The geographic variation in these traits suggests that ments included a hot environment (33/20¡C day/night) at low hu- population-level adaptation has occurred in H. salsola to midity, 26% daytime relative humidity (RH), a similar hot envi- adjust intrinsic plant factors in a manner that may com- ronment at high humidity (67% RH), and a cool environment pensate for climatic factors influencing plant water use (23/20¡C day/night) at low RH (37%), referred to as hot-dry (h), and status. Direct measurements of hydraulic conduc- hot-humid (hh) and cool-dry (c), respectively. The hh and c had the same ν. The species had photosynthetic temperature optima tance, however, were not previously made on plants very near 29¡C (Comstock and Ehleringer 1988), and so photo- from this gradient. New studies were therefore undertak- synthetic capacities at 23 and 33¡C were expected to be very simi- en to test (1) the hypothesis that plants operating under lar. much higher ν would develop greater hydraulic conduc- Cuvette measurements were spread out over several months and successive plantings where made to reduce variation in age at tance per unit of photosynthetic area supported, (2) that time of measurement. Seedling cohorts were started in the late this variation is important in determining stomatal con- summer through fall at monthly intervals. Seed was sown directly ductance (gs), (3) whether both genetic adaptation and into the 30-l pots and there was no transplanting. Measurement oc- plastic response to growth environment were important, curred in late winter and spring from February till May. At the time of measurement, all plants were 5Ð6 months old, and the and (4) what aspects of plant anatomy and/or allocation main stems had extensive secondary growth. The growth periods among organs would be most important in determining were timed to occur in winter, because summers in Ithaca were variation in hydraulic conductance. moderately humid, and the h and especially c environmental con- 3 ditions were attainable in the greenhouse only in the winter season several canopy positions: (1) a twig from the plant caudex that when ambient humidity was low. was stripped of leaves and covered externally with grease to elimi- nate any water loss, which was taken to be a probe of the water- potential relations at the root-shoot transition at soil level; (2) ter- Gas exchange minal leafy twigs in lower, middle and upper crown positions; and (3) the terminal twig of the original main leader of the plant. Water flow rates through the plant, as well as stomatal responses Immediately after the balance-point was determined for the to environmental influences and leaf water potential, were mea- last time (see measurement sequence below) the cuvette lid was sured using steady-state gas-exchange techniques. Gas exchange opened and leaf samples taken for Scholander pressure chamber was measured in a whole-plant cuvette system described in Com- determination. The leaves were taken from twigs currently being stock and Mencuccini (1998). Gas-exchange calculations were held at 0 MPa xylem water potential by the balance technique, made following von Caemmerer and Farquhar (1981) and stomatal loosely wrapped during transfer in a damp paper towel, and sealed ratios treated as described in Comstock and Ehleringer (1993). within 60 s into a Scholander chamber also lined with damp tow- Gas-exchange of all plants was measured under a single set of cu- els to minimize continued water loss during measurement. The vette conditions regardless of growth temperature. Ambient CO2 Scholander readings therefore reflected water potential differences in the cuvette was 360±5 µmol molÐ1, leaf temperature was from the twig xylem to the associated transpiring leaf tissues. 30±1¡C, and irradiance was 1.8 mmol mÐ2 sÐ1 (400Ð700 nm).

Measurement sequence Hydraulic conductance All plants were subjected to a consistent set of cuvette measure- Leaf-specific hydraulic conductance (LSC, mmol mÐ2 sÐ1 MPaÐ1) ments from which both LSC and the sensitivity of stomata to leaf was calculated as the slope of the relationship between shoot wa- water potential were determined. Plants were first allowed to equi- ter potential and transpiration (E, mmol mÐ2 sÐ1) (Passioura 1988). librate under high light and reach a maximum level of gas ex- It is important to note that the term “leaf-specific” is somewhat change with a low leaf-to-air humidity difference (ν) of 10 mmol misleading for H. salsola, and is kept here only for consistency molÐ1. The soil compartment was first at normal ambient pressure. with other studies and to avoid coining new terminology. Not leaf The humidity was reduced in several steps until ν reached area per se, but total photosynthetic surface area including both 35 mmol molÐ1. Sufficient time was given (about 30 min) for leaf and green (i.e., fully photosynthetic) twig area was used in plants to reach new steady-state gas-exchange values at each hu- this calculation. A. dumosa lacked photosynthetic twigs, and only midity. Finally, three additional points were taken while pressuriz- leaf area was used to calculate LSC. ing the soil compartment as needed until a cut twig in the canopy There was also an apparent offset in pressure observed at just began to exude xylem sap. This balancing pressure (P, MPa) zero transpiration which could not be fully explained by bulk soil was measured first at ν=35 mmol molÐ1, but only on the greased water potential. This offset was seen to shift slightly on a diurnal twig from the caudex; ν was then lowered to 10 mmol molÐ1 for basis, becoming more negative in the afternoon than the morning measurement of all crown positions described above, and, finally, (Comstock and Mencuccini 1998). Data were corrected for inter- the caudex probe was measured again at 35 mmol molÐ1. cept drift prior to calculation of LSC by regression.

Natural abundance of stable isotopes Pressurization of soil compartment The abundance ratio, R, of the stable isotopes of carbon, 13C:12C, Soil pressurization was used both to measure the total water- was measured as a long-term index of intracellular CO2 concentra- potential difference through the plant for calculation of LSC, and tions (ci) during growth. Leaf samples were initially dried and to observe stomatal responses to changes in leaf water potential. ground to 40mesh. The data were expressed relative to the PDB The root system was enclosed in a pressure-chamber with a split standard as: lid allowing the intact stem to leave the pressure chamber and en- ()*RR− 1000 δ = sample standard , ‰ ter a shoot gas-exchange cuvette at ambient air pressure. This per- sample R (2) mitted experiments in which the shoot water potential was manip- standard ulated directly by pressurization of the soil compartment. The the- The analyses were performed at the SIRFER facility at the Univer- ory of how soil pressurization affects shoot water potential has sity of Utah, and at CoBSIL, the Cornell and Boyce Thompson been discussed in several previous papers (e.g., Comstock and Stable Isotope Laboratory in Ithaca, New York. Reported isotope Mencuccini 1998; Passioura 1980). The expected water potential data are presented as discrimination (∆, ‰) between leaf and air of the foliage is: carbon pools (Farquhar et al. 1989). ∆ is theoretically related to photosynthetic carbon uptake as: ΨΨ=−EP/ LSC + (1) c leaf soil ∆ =+aba() − i (3) c where E is transpiration (mmol mÐ2 sÐ1), and P is soil compart- a ment pressure (MPa). where a and b are the isotopic discrimination constants associated with diffusion and net carboxylation, respectively, and ca and ci are the ambient and intracellular CO concentrations, respectively. ∆ 2 Water potential gradients throughout the crown was calculated from isotopic data on air and plant tissues as (Farquhar et al. 1989): Total differences in water potential between the soil and shoot δδ− ∆ = air plant were measured primarily using a balance-point method. The soil δ (4) compartment was pressurized as needed until a cut twig in the 1+ plant canopy just began to exude xylem sap. At this time, the pressure 1000 δ reading on the soil compartment was considered equal to the total air was not directly measured, but was assumed to be equal to the difference needed to support the current transpiration rate. Balance global mean value of –8.3‰ when CO2 was equal to 361 µmol Ð1 δ points were measured on small twigs which had had their tips mol . Variation in air for different plant cohorts was estimated us- trimmed back a few cm and directed out through the cuvette wall ing the measured greenhouse CO2 concentration and assuming via small ports drilled for this purpose. Sap exudation was deter- that isotopic ratio of the air conformed to a standard Keeling plot mined visually with a hand lens. Balance points were measured at with intercept at –27 ‰. Ambient CO2 in the greenhouse was very 4 constant throughout the winter, varying only 2 or 3 µmol molÐ1 among months and less than 1 µmol molÐ1 among the three well- ventilated greenhouses. Although plants were 5 months old at measurement, due to geometric growth rates most tissue sampled for ∆ was produced during the last month and only that period was δ δ averaged for estimating air. This made the air correction vary less that 0.1‰ among treatments or greenhouses.

Harvest data

After the gas-exchange measurements, each plant was subjected to a total biomass harvest to permit an evaluation of how LSC and stomatal behavior were related to the distribution of standing biomass. Measurements included basal diameter of the main stem (mesuredwith callipers), total length of stems and green twigs, stem surface area, and mean diameter of different age classes measured as total projected area divided by length, dry weight broken down into categories of fine root, coarse root, tap root, main stem, other woody stems, green leaf-bearing twigs, and leaves. All tissues were dried under forced convection at 60¡C until weight loss ceased. This was as little as 2 days for most leaf samples, but up to several days for stems larger than 1 cm in diameter. Fig. 1 Illustration of balance-point data. The balance-point is de- Projected area of leaves and stems was measured with a leaf fined as the gas-phase pressure needed in the root chamber to area meter (model LI-3200, LI-COR Instruments, Lincoln, Neb., bring a wet meniscus to the surface of a cut twig-tip in the canopy. USA) calibrated with a paper comb (Comstock and Ehleringer Each plant in the study was subjected to similar measurements, 1990). and the hydraulic conductance was calculated from the slopes of Fine roots (less than 1 mm) were harvested by washing the soil the indicated relationships in an elutriation chamber with water and air jets stirring the heavi- er material from below. Roots were captured on screens in the ef- fluent. Vessel radius was measured on cross-sections of each main tern of crown branching. The original leaders were the stem using a light microscope, drawing tube, and digital input oldest tissues in the upper crown, and were no longer ac- board. tively growing. For subsequent analyses relating LSC to other physiological and harvest data, LSCroot refers to the slope of the caudex twig, and LSCcrown refers to an aver- Results age value of the three crown heights (dotted lines in Fig. 1) excluding the leader. LSCleader was excluded from Hydraulic architecture and variation in LSC LSCcrown because (1) it was measured in only a limited number of cases, (2) it was not representative of any LSC was calculated as the slope of a linear relationship large fraction of the total crown, and (3) it was highly er- between E and water potential difference (Comstock and ratic, sometimes being similar to other upper canopy Mencuccini 1998; Passioura 1988). A typical dataset sites and sometimes far lower. Above-ground resistance used in this calculation is shown for one individual of A. from the caudex to the leader averaged 86% greater than dumosa in Fig. 1. The caudex probe allowed for a two- an average upper canopy twig position in A. dumosa, and point regression estimating both slope and intercept. A 153% greater in H. salsola. correction for intercept drift was calculated from a sec- Leaf water potentials were measured on leaves, while ond high ν balance point (not shown). The non-zero in- the twigs were simultaneously being held at their balance Ψ tercept was attributed to below-ground portions of the points soil pressurization. Mean leaf under these condi- pathway and taken to be constant for all crown positions tions averaged 0.47 and 0.48 MPa for A. dumosa and H. (Comstock and Mencuccini 1998; Reiger and Litvin salsola, respectively. Because leaf water potentials were 1999; Stirzaker and Passioura 1996). Measurement of all only measured on about half of the plants, calculated ν twigs at high was not undertaken, both because of the LSCcrown values compared with other parameters do not time required to measure so many balance points, and include this part of the pathway. also because this avoided extreme overpressurization of the lower canopy while reaching balance points for the upper canopy positions at high E. The measured slope Variation in gas-exchange and relationship to LSC varied substantially with canopy position, being greatest for the defoliated probe of the caudex water potential Both desert species showed a strong stomatal response to and least for the terminal leader. This decrease in LSC decreasing ambient humidity, such that transpiration with plant height was as expected from simple path- rates at higher ν (Comstock and Mencuccini 1998) were length considerations. Somewhat unexpected was the limited by stomatal closure (Figs. 2,3,4). For plants large difference between average upper crown twigs and grown in the h environment, the two species showed the leader (Fig. 1). These positions had similar path- very similar, linear relationships between gs and LSC, length and differed only in age and the ontogenetic pat- which included both northern and southern populations 5

Fig. 2 Relationship of stomatal conductance (gs) to LSCcrown Fig. 3 Relationship between stomatal sensitivity to low humidity when transpiration was low due to low leaf-to-air vapor pressure and LSCcrown. Sensitivity is defined as the percent reduction in gs difference (ν). Although the overall regression was weakly signifi- after humidity in the measurement cuvette had been dropped and ν cant at P<0.05, neither species nor geographic races were signifi- increased from 10 to 35 mbar barÐ1. Symbols as in Fig. 2. Each cantly different in mean gs. Symbol codes: plant species: A.d Am- point represents one plant brosia dumosa, H.s. Hymenoclea salsola; seed sources: N northern, S southern; growing conditions: h hot (33¡C, 26% RH), hh hot humid (33¡C, 67% RH), and c cool (23¡C, 37% RH). All Ð1 plants were measured at 30¡C leaf temperature. Each point repre- 35 mmol mol , gs showed substantial (though not sents one plant complete) recovery towards its original high value at ν=10 mmol molÐ1. This relative recovery was greater in H. salsola than in A .dumosa, but did not differ with geo- of both species. When transpiration was low because of graphic origin (Table 1). ν ∆ low , gs was high and only weakly related to LSCcrown A strong correlation was found between measured ν ν (Fig. 2). As increased, however, plants with high on bulk leaf tissue and the ci value measured at of Ð1 LSCcrown proved less sensitive to high transpiration rates 35 mmol mol (Fig. 5). However, only for A. dumosa ∆ and maintained high gs, while plants with low LSCcrown was significantly correlated with LSC. In H. salsola, showed strong stomatal closure (Fig. 3). As a conse- although plants were healthy and growing vigorously, quence, there was a very strong relationship between gs net photosynthetic rates (A) were somewhat lower than ν and LSCcrown at high (Fig. 4). expected, and more variable. This variability in A appar- Although all populations appeared to follow the same ently decoupled the expected link between LSC, gs, and ∆ ∆ fundamental relationship between gs and LSC, they were in H. salsola. Differences in between northern and distinguished by different mean values of LSCcrown southern populations, seen in previous common garden (Figs. 2, 3, 4, Table 1). The slope species, A .dumosa, studies (Comstock and Ehleringer 1992), were not seen had lower LSCcrown than the wash species, H. salsola. in this dataset. For both species, populations originating from the south had higher LSCcrown than the populations originating from the north of the species range. The significant dif- Plastic response to growth environment ferences in LSCcrown among species and populations ν were mirrored by differences in mean gs at high , but When the northern population of H. salsola was grown ν ν not at low ; gs at low was not significantly different at contrasting humidities but the same high temperature, among species or geographic origins. there was no change in LSC, or gs measured under cu- During pressurization of the soil compartment for bal- vette conditions (cuvette measurement involved a gs vs. ance points, water potentials in the shoot were raised. In ν response at the same, intermediate temperature for all response to this manipulation, although ν remained at growth treatments) (Figs. 2,3,4, Table 2). Humid grown 6

∆ Fig. 4 Strong dependency of gs on leaf-specific hydraulic conduc- Fig. 5 Correlation of carbon isotope discrimination of leaves ( ) tance (LSCcrown) when stomatal opening was associated with high with intracellular CO2 concentration (ci) measured in the cuvette. transpiration rates due to high ν. Plants were all light-saturated This shows a consistency of stomatal behavior during growth and and near the leaf temperature optimum of 30¡C. Symbols as in during response to instantaneous cuvette measurements at high ν Fig. 2. Each point represents one plant (35 mbar barÐ1). Symbols as in Fig. 2. Each point represents one plant plants had higher ∆, however, indicating that in response served the dual function of transport and organ of photo- to contrasting growth environment conditions they may synthesis, had a higher allocation to twigs. This was have maintained higher gs in the greenhouse. Despite most notable in the southern population, for which twigs ν contrasting , this difference in gs would have made E were, in fact, the most important photosynthetic surface. during growth more similar than expected, and may ex- Percent allocation to leaves was higher in A. dumosa plain the similar development of LSCcrown. In contrast, than H. Salsola, and higher in northern than southern plants grown at lower temperature had significantly low- populations of both species (Table 1). Percent allocation er LSCcrown, lower gs under the uniform cuvette condi- to leaves was the only harvest variable to be strongly ∆ tions, but a similar to plants grown at higher tempera- correlated with LSCcrown (Fig. 7), which decreased as al- ture but with similar low RH. This suggests that gs under location to leaves increased. LSCcrown was highest in the growth conditions followed variation in RH treatments, southern population of H. salsola, where twigs contribut- but development of LSCcrown was, in contrast, more related over half of the photosynthetic surface and the high ed to growth temperature than RH. allocation to twig tissues served a dual function.

Variation in biomass allocation patterns Gas exchange and plant architecture

Harvest data on standing biomass at the date of measure- The differences in allocation resulted in strong differ- ment shows several interesting contrasts (Fig. 6). No ences in total photosynthetic surface area per gram of difference was seen between species or populations plant among both species and populations (Table 1). in relative allocation to roots (Table 1). Consistent with However, the high LSCcrown of plants with low total area their preferred microhabitats and growth habits, H. promoted greater stomatal opening, and higher photo- salsola had a higher allocation to the taproot, while A. synthesis per unit area. Neither photosynthesis per unit dumosa had higher allocation to woody branches. The biomass nor per unit nitrogen varied among species or most dramatic difference, and one with a significant populations (Table 1). × species environment interaction in Table 1 was the allo- No correlation was found between LSCcrown and xy- cation to leaf-bearing twigs. H. salsola, in which twigs lem vessel diameter. The plants in this study were also 7 Table 1 Comparison of gas-exchange and harvest data across spe- sizes were 5 plants for the northern seed sources and 6 plants for cies and geographic origin. All plants were grown in a common the southern seed sources for each species (gs stomatal conduc- greenhouse in Ithaca, New York. Mean values and the re-sults of a tance, LSC leaf-specific hydraulic conductance, PS photosynthe- two-way factorial ANOVA are given for each measure. Sample tic)

Variable Mean values Error mean Species Geographi- Species× square cal origin site Hymenoclea Ambrosia (site) salsola dumosa

North South North South

Leaf 13C discrimination, ‰ 23.0 22.9 22.5 23.1 6.15×10Ð1 0.640 0.490 0.287 Net photosynthesis, µmol mÐ2 sÐ1 24.3 30.7 19.2 21.1 1.84×101 0.001 0.037 0.238 Net photosynthesis, nmol gÐ2 sÐ1 182 182 178 172 1.08×103 0.632 0.801 0.832 Ð1 Ð1 Net photosynthesis, µmol gN s 5.91 6.00 5.32 5.13 1.06 0.114 0.918 0.753 ν Ð1 Ð2 Ð1 × Ð3 gs at =35 mmol mol , mol m s 0.507 0.608 0.330 0.423 7.25 10 0.000 0.016 0.905 ν Ð1 Ð2 Ð1 × Ð2 gs at =10 mmol mol , mol m s 0.681 0.814 0.675 0.704 1.26 10 0.244 0.107 0.292 ν × 1 Decrease in gs (as increases from 10 25.5 24.6 51.4 40.4 4.96 10 0.000 0.065 0.111 to 35 mmol molÐ1), % × 2 gs (recovery with soil press.), % 79.4 80.6 52.6 51.0 3.41 10 0.002 0.978 0.859 Ð2 Ð1 Ð1 × 1 LSCcrown (soil:twig), mmol m s MPa 24.6 32.7 10.8 18.6 1.96 10 0.000 0.001 0.937 Ð2 Ð1 Ð1 × 2 LSCstem (caudex:twig), mmol m s MPa 87.0 137.1 30.9 48.5 5.54 10 0.000 0.003 0.124 Ð2 Ð1 Ð1 × 2 LSCroot (soil:caudex), mmol m s MPa 35.2 44.5 17.9 34.7 1.25 10 0.011 0.014 0.441 Leaf area: mass ratio, cmÐ2 gÐ1 114 97 155 134 6.66×102 0.002 0.099 0.872 Twig area: mass ratio, cmÐ2 gÐ1 31.3 44.7 36.4 35.9 5.84×101 0.577 0.064 0.046 Twig diameter, mm 1.36 0.99 1.31 1.43 5.22×10Ð2 0.059 0.209 0.022 Twig length: mass ratio, m gÐ1 2.37 4.58 2.81 2.55 4.27×10Ð1 0.011 0.003 0.000 PS area per total mass, mÐ2 kgÐ1 4.12 3.78 5.43 4.11 6.07×10Ð1 0.025 0.023 0.158 Fine root, % of total biomass 15.2 11.5 14.0 15.4 2.90×101 0.578 0.633 0.282 Coarse root, % of total biomass 7.0 7.3 8.2 9.9 1.36×101 0.244 0.541 0.634 Tap root, % of total biomass 4.5 2.8 1.9 2.3 2.02 0.018 0.320 0.092 Main stem, % of total biomass 16.8 10.0 7.5 12.7 1.88×101 0.095 0.692 0.005 Woody branch, % of total biomass 2.0 5.7 10.4 7.8 1.98×101 0.012 0.786 0.115 Leaf, % of total biomass 29.2 19.4 35.1 31.3 1.76×101 0.000 0.001 0.109 PS tissue, % of total biomass 54.5 62.7 35.1 31.3 2.60×101 0.000 0.322 0.014

Table 2 Comparison of northern seed source of H. salsola grown 26%; c day temperature 23¡C, RH 37%). Mean values for growth under different environmental conditions (hh day temperature conditions with different superscripts are significantly different 33¡C, relative humidity RH 67%; h day temperature 33¡C, RH from each other (Tukey LSD); n=5 plants per treatment

Variable Error mean square P Means

hh h c

Leaf 13C discrimination, ‰ 3.76×10Ð1 0.001 24.3a 23.0b 22.5b ν Ð1 Ð2 Ð1 × Ð3 a a b gs at =35 mmol mol , mol m s 2.46 10 0.001 0.509 0.507 0.374 LSC crown (soil to twig-tip), mmol mÐ2 sÐ1 MPaÐ1 1.55×101 0.033 25.8a 24.6a 18.7b

measured individually for cavitation vulnerability. Al- 1995) where LSCcrown increased along a climatic gradi- though strong differences between species and small dif- ent of increased ν because of increased allocation to ferences between populations were found (Mencuccini branchwood. This trend permits the maintenance of high and Comstock 1997), no correlation was found within gs throughout the climatic gradient, but altered allocation species between vulnerability to cavitation and LSC. patterns could still lead to reduced total leaf area and lower growth rates. Increased allocation to non-productive branch tissues Discussion in P. sylvestris could have reduced intrinsic growth rate. In the desert sub-shrubs, LSCcrown was very strongly cor- Variation in LSCcrown and effects on gas-exchange related with percent allocation to leaves (Fig. 7). This shift in allocation was most extreme for the southern These results supported the hypothesis that genetic sel- populations of H. salsola. This did not result in a reduc- ection along a climatic gradient would help match tion in whole-canopy gas-exchange rate per unit canopy LSCcrown with the evaporative demand of the environ- biomass, however, because the twig itself became ment (Table 1). Similar results were found in Pinus syl- the primary organ of photosynthesis in these plants vestris (Berninger et al. 1995; Mencuccini and Grace (Table 1). Comstock and Ehleringer (1988) made a de- 8 synthesizing organ. This was similar to the findings for other species (Nilsen 1992; Osmond et al. 1987). The data measured here at the whole-plant level, however, show that this apparent low efficiency is indeed, as was suggested, apparently offset by the multiple functions being performed by twigs (i.e., support, transport, and photosynthesis), and that twig canopies can be just as ef- ficient as leaf canopies at the whole-plant level of inte- gration (Fig. 7, Table 1). Plastic responses of LSCcrown to temperature and evaporative demand were also observed in this study, consistent with the assertion that high transpiration rates during growth promoted increased development of hydraulic capacity (Fig. 4, Table 2). Other reports in the lit- Fig. 6 Distribution of standing biomass following whole-plant erature support the notion that this plastic response is harvest at 5 months of age. Particularly prominent is the very high allocation to photosynthetic twigs characterizing the specialized mediated through transpiration rate and water-potential crown morphology of the southern H. salsola. Each bar represents gradient, since LSCcrown was reduced when transpiration the mean of 5 plants (+SE) was low due to stomatal closure under elevated CO2 (Bunce and Ziska 1998; Heath et al. 1997) even though temperature and ν were held constant. Stomatal conductance (gs)was highly correlated with LSCcrown (Figs. 3, 4) and very sensitive to direct manipulation of leaf water potential (Table 1). A large number of studies in the literature reveal a strong correlation between gs and LSCcrown, while LSCcrown varies due to causes as variable as crown architecture (Hubbard et al. 1999; Mencuccini and Grace 1996; Ryan and Yoder 1997), drought (Irvine et al. 1998), salinity stress (Loustau et al. 1995; Sohan et al. 1999), ozone stress (Grantz and Yang 1996), or site irradiance (Maherali et al. 1997). The stomatal reopening response to soil pressurization and consequent elevation of leaf water potential agrees with previous reports (Comstock and Men- cuccini 1998; Fuchs and Livingston 1996; Saliendra et al. 1995) and indicates that at least some part of this correlation is due to direct feedback control between leaf water potential and gs.

Hydraulic architecture

Consistent, perhaps, with an herbaceous evolutionary past and current growth habits intermediate between large herbaceous perennials and woody shrubs, rather steep water potential differences were observed in the Fig. 7 Dependence of leaf-specific hydraulic conductance (LSCcrown) 40Ð60 cm of shoot height for both study species, and yet on biomass allocation. Plants with the highest allocation to leaves total belowground hydraulic resistance was still consid- have the lowest hydraulic support per unit photosynthetic area. Al- though LSC as used in this paper includes all photosynthetic sur- erably greater than total shoot axial resistance (Table 1). faces and not just leaves per se, “leaves” in the context of %allo- Gradients of this magnitude per unit height would proba- cation refers strictly to leaves and not to photosynthetic twigs, bly not be sustainable in plants of large stature. A very which also serve a transport function. Symbols as in Fig. 2. Each large, abrupt water potential difference was observed be- point represents one plant tween the twigs and leaves. It is not possible to tell from this dataset whether this gradient was primarily in the petioles, leaf veins, or in symplastic portions of the path- tailed comparison of photosynthetic behavior of leaves way associated with movement from veins to evapora- and twigs in H. salsola, and found that individual twigs tive sites. All of these have been identified as sites of un- have much lower photosynthetic rates (compared to usually high hydraulic resistance in past studies (Tyree leaves on the same respective plants) if expressed as up- et al. 1993; Yang and Tyree 1993; Zimmermann 1983). take per unit biomass or nitrogen content of the photo- In general, stomatal behavior was much better correlated 9 with total resistance rather than the resistance of any spe- Comstock JP, Ehleringer JR (1988) Contrasting photosynthetic be- cific subportion of the pathway, but a limited dataset on havior in leaves and twigs of Hymenoclea salsola, a green- twigged, warm desert shrub. Am J Bot 75:1360Ð1370 the subpathway from twigs to leaves prevented its inclu- Comstock J, Ehleringer J (1990) Effect of variations in leaf size sion in the final estimate of whole-plant LSC. on morphology and photosynthetic rate of twigs. Funct Ecol 4:209Ð221 Comstock JP, Ehleringer JR (1992) Correlating genetic variation in carbon isotopic composition with complex climatic gradi- Lack of apical dominance ents. Proc Natl Acad Sci USA 89:7747Ð7751 Comstock J, Ehleringer J (1993) Stomatal response to humidity in Studies of trees with strong apical dominance have re- common bean (Phaseolus vulgaris): implications for maxi- vealed a higher hydraulic conductance to the dominant mum transpiration rate, water-use efficiency and productivity. Aust J Plant Physiol 20:669Ð691 apex than subdominant or suppressed apices (Zimmer- Comstock JC, Mencuccini MM (1998) Control of stomatal mann 1983). Replacement of a lost tree leader is associ- conductance by leaf water potential in Hymenoclea salsola ated with reestablishment of a very high LSCcrown for the (T. & G.), a desert subshrub. Plant Cell Environ 21:1029Ð1238 leader tissue (Spicer and Gartner 1998). In contrast, both Cowan IR (1977) Stomatal behaviour and environment. Adv Bot Res 4:117Ð223 of the species in this study were desert subshrubs with Cowan IR, Farquhar GD (1977) Stomatal function in relation to no prolonged apical dominance, limited branch lifespans, leaf metabolism and environment. Symp Soc Exp Biol 31: and frequent canopy renewal via basal suckering. At 5 or 471Ð505 6 months of age, the original apex of the first vigorous Dodd IC, Stikic R, Davies WJ (1996) Chemical regulation of gas- exchange and growth of plants in drying soil in the field. J Exp shoot was still present on most of the study plants. In a Bot 47:1475Ð1490 very few cases it had already died, and in a few others, it Farquhar GD, Sharkey TD (1982) Stomatal conductance and pho- was visibly senescent while vigorous growth at the tosynthesis. Annu Rev Plant Physiol 33:317Ð345 whole-crown level continued at other loci. For most Farquhar GD, Ehleringer JR, Hubick KT (1989) Carbon isotope plants, including all the plants where the balance pres- discrimination and photosynthesis. Annu Rev Plant Physiol Plant Mol Biol 40:503Ð537 sure of the leader was measured, visible signs of senes- Fuchs EE, Livingston NJ (1996) Hydraulic control of stomatal cence were not apparent and the original leader was conductance in Douglas fir [Pseudotsuga menziesii (Mirb) identifiable only by tracing the sometimes complex Franco] and alder [Alnus rubra (Bong)] seedlings. Plant Cell branching pattern. Nonetheless, hydraulic conductance Environ 19:1091Ð1098 Grantz DA, Yang S (1996) Effect of O3 on hydraulic architecture to these canopy regions was consistently much lower in pima cotton. Plant Physiol 112:1649Ð1657 than the average for the whole crown (e.g., Fig. 1). Heath J, Kerstiens G, Tyree MT (1997) Stem hydraulic conduc- Since, regardless of the outward health of the original tance of European beech (Fagus sylvatica L.) and pedunculate leader tissue, active growth had in all cases shifted to oak (Quercus robur L.) grown in elevated CO2. J Exp Bot 48:1487Ð1498 major sidebranches from lower stems and caudex, this Hubbard RM, Bond BJ, Ryan MG (1999) Evidence that hydraulic observation is consistent with the supposition that high conductance limits photosynthesis in old Pinus ponderosa hydraulic conductance is preferentially maintained at trees. Tree Physiol 19:165Ð172 strongly growing meristems, and that a loss of it may be Irvine J, Perks MP, Magnani F, Grace J (1998) The response of an early sign of senescence. Pinus sylvestris to drought: stomatal control of transpiration and hydraulic conductance. Tree Physiol 18:393Ð402 Leuning R (1995) A critical appraisal of a combined stomatal- Acknowledgements I wish to thank Jim Brewster for his careful photosynthesis model for C-3 plants. Plant Cell Environ work in measuring plant balance-points, and Maurizio Mencuccini 18:339Ð355 for many helpful discussions on hydraulic architecture. The work Loustau D, Crepeau S, Guye MG, Sartore M, Saur E (1995) was supported by NSF grant 9496093. Growth and water relations of three geographically separate origins of maritime pine (Pinus pinaster) under saline conditions. Tree Physiol 15:569Ð576 Maherali H, Delucia EH, Sipe TW (1997) Hydraulic adjustment of References maple saplings to canopy gap formation. Oecologia 112:472Ð 480 Aphalo PJ, Jarvis PG (1993) An analysis of Ball’s empirical Meinzer FC, Goldstein G, Grantz DA (1990) Carbon isotope dis- model of stomatal conductance. Ann Bot 72:321Ð327 crimination in coffee genotypes grown under limited water Ball JT, Woodrow IE, Berry JA (1987) A model predicting stoma- supply. Plant Physiol 92:130Ð135 tal conductance and its contribution to the control of photosyn- Meinzer FC, Goldstein G, Franco AC, Bustamante M (1999) thesis under different environmental conditions. Prog Photo- Atmospheric and hydraulic limitations on transpiration in Bra- synth Res 4:221Ð228 zilian cerrado woody species. Funct Ecol 273Ð282 Berninger F, Mencuccini M, Nikinmaa E, Grace J, Hari P (1995) Mencuccini M, Comstock J (1997) Vulnerability to cavitation in Evaporative demand determines branchiness of Scots pine. populations of two desert species, Hymenoclea salsola and Oecologia 102:164Ð168 Ambrosia dumosa, from different climatic regions. J Exp Bot Bond B, Kavanagh KL (1999) Stomatal behavior of four woody 48:1323Ð1334 species in relation to leaf-specific hydraulic conductance and Mencuccini M, Comstock J (1999) Variability in hydraulic archi- threshold water potential. Tree Physiol 19:503Ð510 tecture and gas exchange of common bean (Phaseolus vul- Bunce JA, Ziska LH (1998) Decreased hydraulic conductance in garis) cultivars under well-watered conditions: interactions plants at elevated carbon dioxide. Plant Cell Environ 22:121Ð with leaf size. Aust J Plant Physiol 26:115Ð124 126 Mencuccini M, Grace J (1995) Climate influences the leaf Caemmerer S von, Farquhar GD (1981) Some relationships be- area/sapwood area ratio in Scots pine. Tree Physiol 15:1Ð10 tween the biochemistry of photosynthesis and the gas ex- Mencuccini M, Grace J (1996) Hydraulic conductance, light inter- change of leaves. Planta 153:376Ð387 ception and needle nutrient concentration in Scots pine stands 10 and their relations with net primary productivity. Tree Physiol Sober A (1997) Hydraulic conductance, stomatal conductance, 16:459Ð468 and maximal photosynthetic rate in bean leaves. Photosynthe- Monteith JL (1995) A reinterpretation of stomatal responses to hu- tica 34:599Ð603 midity. Plant Cell Environ 18:357Ð364 Sohan D, Jasoni R, Zajicek J (1999) Plant-water relations of NaCl Nilsen ET (1992) The influence of water stress on leaf and stem and calcium-treated sunflower plants. Environ Exp Bot 42: photosynthesis in Spartium junceum L. Plant Cell Environ 105Ð111 15:455Ð461 Sperry JS, Pockman WT (1993) Limitation of transpiration by hy- Osmond CB, Smith SD, Gui-Ying B, Sharkey TD (1987) Stem draulic conductance and xylem cavitation in Betula occidenta- photosynthesis in a desert ephemeral, Eriogonum inflatum: lis. Plant Cell Environ 16:279Ð287 characterization of leaf and stem CO2 fixation and H2O vapor Spicer R, Gartner BL (1998) How does a gymosperm branch exchange under controlled conditions. Oecologia 72:542Ð (Pseudotsuga menziesii) assume the hydraulic status of a main 549 stem when it takes over as leader? Plant Cell Environ 21: Passioura JB (1980) The transport of water from soil to shoot in 1063Ð1070 wheat seedlings. J Exp Bot 31:333Ð345 Stirzaker RJ, Passioura JB (1996) The water relations of the root- Passioura JB (1988) Water transport in and to roots. Annu Rev soil interface. Plant Cell Environ 19:201Ð208 Plant Physiol Plant Mol Biol 39:245Ð265 Tardieu F (1996) Drought perception by plants: do cells of Reiger M, Litvin P (1999) Root system hydraulic conductivity in droughted plants experience water stress? Plant Growth Regul species with contrasting root anatomy. J Exp Bot 50:201Ð 20:93Ð104 209 Tyree MT, Cochard H, Cruziat P, Sinclair B, Ameglio T (1993) Ryan MG, Yoder BJ (1997) Hydraulic limits to tree height and Drought-induced leaf shedding in walnut: evidence for vulner- tree growth. Bioscience 47:235Ð242 ability segmentation. Plant Cell Environ 16:879Ð882 Saliendra NZ, Sperry JS, Comstock JP (1995) Influence of leaf Whitehead D (1998) Regulation of stomatal conductance and tran- water status on stomatal response to humidity, hydraulic con- spiration in forest canopies. Tree Physiol 18:633Ð644 ductance, and soil drought in Betula occidentalis. Planta 196: Yang S, Tyree MT (1993) Hydraulic resistance in Acer saccharum 357Ð366 shoots and its influence of leaf water potential and transpira- Schurr U, Schulze E-D (1996) Effects of drought on nutrient and tion. Tree Physiol 12:231Ð242 ABA transport in Ricinus communis. Plant Cell Environ 19: Zimmermann MH (1983) Xylem structure and the ascent of sap. 665Ð674 Springer, Berlin Heidelberg New York