Observations and model simulations link stomatal inhibition to impaired hydraulic conductance following ozone exposure in cotton


  • David A. Grantz,

    1. University of California at Riverside, Department of Botany and Plant Sciences and Air Pollution Research Center, Kearney Agricultural Center, Parlier, CA 93648, USA, and
    Search for more papers by this author
  • Xijie Zhang,

    1. University of California at Riverside, Department of Botany and Plant Sciences and Air Pollution Research Center, Kearney Agricultural Center, Parlier, CA 93648, USA, and
    Search for more papers by this author
  • Toby Carlson

    1. Department of Meteorology, The Pennsylvania State University, University Park, PA 16802, USA
    Search for more papers by this author

Correspondence: David A.Grantz fax: +1 559 646-6593; e-mail: david@uckac.edu


Ozone (O3) inhibits plant gas exchange and productivity. Vapour phase (gs) and liquid or hydraulic phase (K) conductances to water flux are often correlated as both change with environmental parameters. Exposure of cotton plants to tropospheric O3 reduces gs through reversible short-term mechanisms and by irreversible long-term disruption of biomass allocation to roots which reduces K. We hypothesize that chronic effects of O3 on gas exchange can be mediated by effects on K without a direct effect of O3 on gs or carbon assimilation (A). Experimental observations from diverse field and exposure chamber studies, and simulations with a model of mass and energy transport, support this hypothesis. O3 inhibition of K leads to realistic simulated diurnal courses of gs that reproduce observations at low ambient O3 concentration and maintain the positive correlation between midday gs and K observed experimentally at higher O3 concentrations. Effects mediated by reduced K may interact with more rapid responses of gs and A to yield the observed suite of oxidant impacts on vegetation. The model extends these physiological impacts to the extensive canopy scale. Simulated magnitudes and diurnal time courses of canopy-scale fluxes of H2O and O3 match observations under low ambient concentrations of O3. With greater simulated concentrations of O3 during plant development, the model suggests potential reductions of canopy-scale water fluxes and O3 deposition. This could represent a potentially unfavourable positive feedback on tropospheric O3 concentrations associated with biosphere–atmosphere exchange.


Root response to O3

Exposure of plants to ambient concentrations of tropospheric ozone (O3) reduces biomass production and alters biomass allocation within the plant. Root to shoot biomass ratio is frequently reduced ( Cooley & Manning 1987; Oshima et al. 1979 ), often in an allometric fashion ( Reiling & Davison 1992), although in some cases the root to shoot ratio is unchanged or increased by O3 ( Reiling & Davison 1992).

Ozone reduces yields of adapted upland (Gossypium hirsutum L.) cotton cultivars by about 20% in the San Joaquin Valley of California ( Grantz & McCool 1992; Olszyk et al. 1993 ; Oshima et al. 1979 ; Temple et al. 1988 ), despite many cycles of yield selection in this O3-impacted environment. Yield and productivity of Pima (G. barbadense L.) cotton cultivars, selected in low-O3 environments, are even more sensitive ( Grantz & McCool 1992; Olszyk et al. 1993 ). In this irrigated region of high evaporative demand and high agronomic inputs of mineral nutrients, root function is a critical determinant of yield and biomass production. Exposure to O3 has been shown to reduce both vapour phase (gs) ( Grantz & Yang 1995, 1996a; Temple 1986, 1990) and liquid or hydraulic phase (K) conductances to water flux ( Grantz & Yang 1995, 1996a, b) in cotton.

Parallel declines in gs and K may reflect independent responses to O3. However, typically conserved allometric relationships between root and shoot biomass (e.g. Farrar & Gunn 1996) and reduced productivity of cotton plants with restricted root development ( Browning et al. 1975 ) suggest that O3-inhibited root system development could mediate reductions in long-term shoot gas exchange. Differential yield sensitivity to O3 of two cultivars of sweet corn (Zea mays) has been linked to differential responses of root system development and hydraulic properties ( Harris & Heath 1981). Mechanistic relationships between stomatal function and root hydraulic properties have been suggested previously ( Meinzer & Grantz 1990). This could be mediated by root tip metabolism or water relations, through altered synthesis or transport to the shoot of phytohormones, mineral nutrients or other substances (e.g. Zhang & Davies 1990; Dodd et al. 1996 ; Muller et al. 1996 ; Puliga et al. 1996 ). Alternatively, the interaction could be mediated by effects of K on soil water acquisition and resulting water status of the leaf mesophyll (ψm) or epidermal (ψe) tissue in which the stomata are embedded ( Fuchs & Livingston 1996; Shackel & Brinckmann 1985; Bunce 1996). Differences between ψe and ψm have been documented ( Shackel & Brinckmann 1985), although associated stomatal control mechanisms and sensitivity to O3 exposure have not been assessed.

Rapid reductions in shoot gas exchange following exposure to O3 may be associated with altered phloem loading (e.g. McCool & Menge 1983; McLaughlin & McConathy 1983; Mortensen & Engvild 1995; Grantz & Farrar 1999) and end-product inhibition of carbon metabolism. Reduced root system capacity may also lead to long-term reductions in carboxylase activity and gas exchange performance. Feedback from root to shoot following chronic exposure to O3 would necessarily be long-term and slowly reversed, and would be expressed as a limit on maximal stomatal response to other environmental parameters. The potential mediation of chronic phytotoxic effects of O3 on gas exchange by such a mechanism has not been investigated.

Rapid shoot response to O3

In contrast, rapid gas exchange responses to O3 have been well documented. In bean plants (Phaseolus vulgaris L.), stomatal conductance (gs) was reduced within 6–12 min by 0·3–0·5 ppm O3 ( Moldau et al. 1990 ). While gs declined by about 50% within 3 h, the mesophyll conductance to CO2 was unaffected. Similar apparent direct effects of O3 on gs have been identified by Hill & Littlefield (1969) and by Amundson et al. (1987) .

Light-saturated carbon assimilation declined by about 25% in winter wheat (Triticum aestivum L.; Farage et al. 1991 ) and 40% in pea and oak (Pisum sativum L. and Quercus robur L.; Farage & Long 1995), while gs declined by about 40% (wheat, pea) and about 80% (oak) following 4 h of exposure to 0·4 ppm O3. This was associated with minimal effect on photosynthetic light reactions assayed as variable chlorophyll fluorescence and as quantum yield. Carboxylation capacity was reduced by about 30% (wheat, oak) to 45% (pea), consistent with O3-induced reductions in Rubisco protein and activity ( Eckard & Pell 1995; Lehnherr et al. 1987 ; Pell & Pearson 1983), attributed in part to oxidative modification of sulphhydryl residues ( Eckard & Pell 1995). O3 did not reduce synthesis of Rubisco ( Brendley & Pell 1998), but accelerated proteolysis in older leaves. Similar effects on gas exchange mediated by mesophyll responses to O3 have been identified by Lehnherr et al. (1988) and Myhre et al. (1988) .

In cotton we observed rapid stomatal and mesophyll responses to O3 in older leaves that required several days to reverse in O3-free air (unpublished observations) while young leaves exhibited substantially enhanced gs and A. This potentially compensatory gas exchange in young leaves exposed to O3 ameliorates whole-plant impacts of O3 on gas exchange (e.g. Pell et al. 1994 ) and reflects nitrogen remobilization from older leaves undergoing accelerated senescence (e.g. Brendley & Pell 1998). These short-term, often reversible (e.g. Guidi et al. 1997 ), effects of O3 on gas exchange may reflect parallel impacts on both gs and mesophyll function, with reduced Rubisco activity restricted to older leaves and longer time frames (e.g. in wheat; Grandjean Grimm & Fuhrer 1992).

Present approach

Use of model simulations may reveal mechanistic relationships when experimental manipulations have proven inconclusive, and may elucidate effects at larger scales of biological organization than can be readily manipulated. Here we use a comprehensive simulation model of stomatal conductance ( Lynn & Carlson 1990; Olioso et al. 1996 ; Taconet et al. 1986 ) to explore the role of O3-inhibited root system development on gas exchange performance at leaf and canopy scales. The model as modified for this study describes single leaf gs and canopy-scale fluxes of energy, water and O3 between the bulk soil, the plant, and the atmospheric mixed layer. The single-day simulations invoke a partial dependence of gs on epidermal water status, but exclude chemical root signals, direct effects of O3 on photosynthesis or guard cell metabolism, and longer-term feedbacks on canopy transport associated with O3-reduced growth and leaf area development. The model incorporates a direct O3 effect on K as the only physiological impact of O3. Secondary effects of K on gs, leaf water relations, and energy and mass balances at the canopy scale are simulated from the single physiological impact on K. This is in contrast to previous simulations of growth and gas exchange responses to O3 (e.g. Constable & Taylor 1997) in which a direct impact of O3 on leaf gas exchange (maximal carboxylation capacity) is an input parameter.

We investigate the gas exchange and water relations behaviour of mature, individual cotton plants embedded in an extensive cotton canopy, following chronic exposure to O3 during plant development. We begin with experimental observations obtained under a variety of conditions and attempt to reproduce them using the mechanistic simulation model. The simulated values reproduce observed gas exchange behaviour without invoking any direct effect of O3 on leaf gas exchange.


Experimental observations

Field exposure chambers

Cotton plants (Gossypium hirsutum; cv. Acala SJ-2 and G. barbadense; cv. Pima S-6) were grown in 1992 from seed in closed-top field exposure chambers as described by Musselman et al. (1986) located in Riverside, California. Seeds were planted in rows 1 m apart with approximately 13 seeds/m. Plants received drip irrigation daily and complete fertilization monthly. Yield and stomatal responses to O3 of similar plants in 1991 were substantial ( Grantz & McCool 1992). Data from the 1992 season have not previously been published.

Stomatal conductance (gs; mol m−2 s−1) was determined on both surfaces simultaneously of the youngest fully expanded leaf, at midday using a transient gas exchange system (LI 6200; LiCor Inc., Lincoln, Nebraska, USA) and expressed relative to projected leaf area. Measurements were obtained on four replicate plants per chamber under natural sunlight and as near as practical to ambient humidity. The water potential of these leaves (ψm; MPa) was determined at midday with a pressure chamber. Leaves were sealed in foil-covered zip-lock bags prior to excision at the base of the petiole. Sealed bags were stored in an insulated, dark container prior to measurement (< 30 min).

Hydraulic conductance (K; mol m−2 MPa−1 s−1) was determined using the gradient in water potential between the soil and the transpiring leaf, water flux during periods of high transpiration, and leaf area of the plant. Soil water potential was determined as pre-dawn leaf water potential, which was always approximately 0 MPa. These values were confirmed using the water potential of de-topped stems as described by Grantz & Yang (1996a). Water flux was approximated as the product of gs and total plant leaf area, and leaf to air vapoor pressure difference (V; defined in the within-canopy air space and taken as a representative value of 1·5 kPa inside the chambers). Alternative methods of determining K in these studies have yielded similar values, as described by Yang & Grantz (1996).

Commercial field

Acala cotton plants (Gossypium hirsutum L.; cv. Delta Pine 6166) were grown in the field in 1991 (as described by Grantz et al. 1997 ) and again in the same field in 1994, under commercial conditions in the San Joaquin Valley near Firebaugh, California (36°48′50′′N, 120°40′38′′W).

Data were obtained in 1994 during a lengthy period including 23 August (day of year (DOY) 235, the date of our simulations). The canopy was 1 m in height and fully covered the ground, with a leaf area index of 2·52. The roughness length was determined experimentally to be 0·13 m. The field was furrow-irrigated, with the interval between irrigations lengthened to induce early reproductive development. This periodic soil water deficit allowed measurements to be obtained at different soil water contents and resulting soil–root hydraulic conductances.

Stomatal conductance of both surfaces of representative leaves at each of six insertion levels on four plants was determined near midday with a steady state porometer (LI 1600; LiCor Inc.). Hydraulic conductance was determined as in the field exposure chambers with water flux measured directly using sap flow gauges (Dynagage, Inc., Houston. Texas, USA). These were installed above the ground surface, insulated with closed-cell foam and plastic bubble wrap, and shielded from radiation with aluminium foil, to avoid the radiation-induced errors encountered previously ( Shackel et al. 1992 ) in this environment. The resulting data agreed (± 12%) with canopy-scale measurements of water vapour flux at midday obtained with the Bowen ratio energy balance technique (not shown).

Maximal ozone concentration at 50 m measured at this site in 1991 was about 0·08 ppm ( Delany et al. 1992 ), and that at the ground surface measured in 1994 was about 0 ppmv. Ozone deposition (F) was measured at 5 m using an eddy covariance protocol ( Delany et al. 1992 ).

Model simulations

Model background

The simulation model used in this study is a further development of the soil/plant/canopy/atmosphere model described by Taconet et al. (1986) , Lynn & Carlson (1990) and Olioso et al. (1996) . Following modification for the current analyses, this model simulates daily courses of atmospheric, soil and plant parameters including single leaf stomatal conductance and leaf water potential, and canopy fluxes of momentum, heat, water vapour, carbon dioxide and O3. The modelling domain extends from the bulk soil through the rhizosphere, plant canopy, surface layer, and well into the atmospheric mixed layer, the height of which develops with model iteration. The model is exercised over single diurnal periods when net radiation is positive, from initial conditions in the early morning.

Stomatal conductance (gs), and bulk leaf mesophyll and epidermal water potentials (ψm, ψe) are interactive functions of modelled and input local environmental parameters including root hydraulic conductance (K), an input variable in the original model ( Lynn & Carlson 1990), but here a function of input [O3] during simulated plant development, as:


where [O3] is the 12 h mean O3 concentration. This relationship is derived from observations in field and greenhouse exposure chambers (previously unpublished results; Fig. 1a; and Grantz & Yang 1996a). A similar O3-induced reduction in K has been observed in red spruce (Picea rubens (Sarg.)) by Lee et al. (1990) . While many environmental variables may alter root development and hydraulic properties (e.g. nitrogen; Grantz & Yang 1996b) and others may affect gs (e.g. humidity; Aphalo & Jarvis 1991; Grantz 1990), these are not explicitly considered except as they vary interactively over the single day of simulation. Over longer time periods such factors will alter plant growth and development and would require parameterization in the model.

Figure 1.

. Measured individual relationships for two species of cotton, between 12 h seasonal mean O3 concentration imposed in field exposure chambers and (a) leaf area-specific plant hydraulic conductance (K), (b) water potential of exposed, transpiring leaves (Ψm), and (c) stomatal conductance to water vapour of exposed, transpiring leaves (gs). Data are means of four plants obtained near solar noon on one representative day. Curves are fit by non-linear regression of the form y = a + b/x.

Epidermal water potential (ψe) is calculated as a function of the edaphic, atmospheric and physiological inputs embedded in the canopy-scale model (see Lynn & Carlson 1990 for details) and linked to bulk leaf mesophyll water potential (ψm) as


where the constant parameter, β = 0·02 MPa kPa−1. This relates ψe to transpiration (T), through its interdependence with V, and to ψm through the efficiency of hydraulic connections between mesophyll and epidermis incorporated in β. It also incorporates an implicit stomatal response to V through the interdependence of gs, ψe and V incorporated in eqns 2–5, rather than the explicit response of gs to V in many other treatments (e.g. Jarvis 1976; Grantz et al. 1987 ). These relationships capture the feed-forward behaviour of T and ψm with respect to V, with both increasing and then decreasing in absolute magnitude as V increases with changing leaf temperature, air humidity and transpiration. They also capture the uncoupling of ψe from ψm, predicting changes in opposite directions under some conditions. This has been considered a potential mechanistic control feature underlying the feed-forward stomatal responses to humidity ( Aphalo & Jarvis 1991; Bunce 1996; Grantz 1990; Maier-Maercker 1983), although this concept may require reconsideration in light of recent evidence ( Franks et al. 1997 ; Monteith 1995; Mott & Parkhurst 1991).

Stomatal responses to interacting environmental variables are treated as multiplicative functions (simplified after Jarvis 1976; Grantz et al. 1987 ), so that gs is related to photosynthetically active photon flux density (PPFD) and ψe as


where a = 2·94 and b = 1000. The hyperbolic stomatal response to PPFD is that derived for similar field-grown cotton by Grantz et al. (1997) . The function f(ψe) is a linear discontinuous stomatal response to epidermal water status ( Lynn & Carlson 1990) as


where b1 = –0·0001, b2 = 0·3, and ψc = –1·6 MPa is the critical epidermal water potential.

Canopy transpiration (T) and O3 flux (F) are calculated (e.g. for T) as


and the total flux from the surface, including that from the soil (e.g. for water vapour, Eg) as


where rl is leaf resistance and raf is leaf boundary layer resistance (to water vapour), ρ is air density, P is atmospheric pressure, and Le is the latent heat of vaporization of water. Leaf resistance to water vapour (rl) is composed of the stomatal resistance (rs) and the cuticular resistance (rc = 10 s cm−1), in parallel. Eg is the evaporative flux from the ground, a function of temperature and moisture of both the soil and the air near the surface ( Lynn & Carlson 1990).

Ozone and water vapour fluxes follow similar paths, in opposing directions. The O3 concentration in the canopy (Cf) is calculated as


where Ca is O3 concentration at 50 m, modelled as a diurnal course increasing from 0·02 ppmv at dawn to 0·08 ppm at 16:00 Pacific Daylight Time (PDT) followed by a plateau in accordance with field observations at this site in 1991, ra is the aerodynamic resistance above the canopy, and rtot is the sum of parallel resistances for cuticular, stomatal and soil fluxes, expressed as


where rag is the aerodynamic resistance to water vapour between the ground and the top of the canopy, and LAI is leaf area index corrected by a shelter factor ( Lynn & Carlson 1990; Taconet et al. 1986 ; Olioso et al. 1996 ). Resistances to water vapour are converted to resistances to O3 by the factors 1·32 and 1·66, for non-stomatal and stomatal resistances, respectively ( Jones 1992).


The model was initialized with canopy characteristics, meteorological soundings, ozone concentrations, and approximate soil type (loamy sand) and crop species (cotton) observed at this site on or around DOY 235 during the two years of intensive measurements (1991 and 1994). Day length and incoming shortwave radiation were calculated as functions of geographic coordinates and date. Shortwave radiation was used to calculate PPFD using an empirical regression developed at this site. Simulated values of PPFD agreed well with measurements (not shown).

Data presentation

The model was run in Microsoft Fortran PowerStation (version 4·0; Microsoft Inc., Redland, Washington. USA) with output exported to Sigma Plot (SPSS Inc.; Chicago, Illinois, USA) for statistical analyses and preparation of figures. All measured and simulated conductances are converted to molar units using appropriate temperature and pressure, for compatibility with the plant physiology literature.


Experimental data

Effects of ozone

Mature cotton plants grown in field exposure chambers under chronic exposure to realistic concentrations of O3 exhibited a progressive reduction in plant hydraulic conductance with increasing O3 concentration. This was the case on a per plant basis (not shown) and when expressed relative to transpiring leaf area as leaf area-specific hydraulic conductance (K;Fig. 1a). The functional parameter, K ( Grantz & Yang 1996a; Yang & Tyree 1993) overcomes the confounding effects of differing plant size between ozone exposures, experiments and locations, and relates root capacity to the water and nutrient requirements of the foliage.

The response of K to O3 was similar in Acala and Pima cottons ( Fig. 1a) although sensitivity was higher in Acala. Similar effects of O3 on K have been reported for younger Pima cotton plants in greenhouse exposure chambers ( Grantz & Yang 1996a) and for seedlings of red spruce ( Lee et al. 1990 ). This reduction in K ( Fig. 1a) could result in a substantial reduction of the mesophyll water potential of transpiring leaves (ψm), particularly during the midday period when evaporative demand and transpirational fluxes are maximal. This has been observed (e.g. Heggestad et al. 1985 ; Roberts & Cannon 1992) but is not generally the case. Often ψm is unaffected or improved slightly (e.g. Grantz & Yang 1996a; Lee et al. 1990 ; Temple 1986, 1990; Temple et al. 1988 ) reflecting a tight coordination of stomatal and hydraulic conductance. In citrus trees, both positive and negative effects of O3 on ψm have been observed on different days in the same study ( Olszyk et al. 1991 ). In cotton, ψm exhibits little consistent response to ozone exposure in mature, field chamber-grown plants ( Fig. 1b). Similar results were obtained with other field chamber-grown mature plants ( Temple 1986, 1990), and with greenhouse chamber-grown seedlings ( Grantz & Yang 1995, 1996a, b).

This homeostasis of leaf water status ( Fig. 1b) despite degraded root ( Grantz & Yang 1996a) and resulting plant ( Fig. 1a) hydraulic capacity indicates that water loss is reduced by declining gs with increasing O3 concentration. Reductions in gs parallel to the declines in K were observed in the present study ( Fig. 1c), and in previous studies with Pima cotton seedlings ( Grantz & Yang 1996b). In general, a reduction in gs is observed in response to ozone exposure, although changes in K are infrequently evaluated.

Relationship between stomatal and hydraulic properties

A potential functional relationship between vapour phase (stomatal; gs) and liquid phase (hydraulic; K) conductances is suggested by the similarity in the responses to O3 concentration shown in Fig. 1(a,c). Similar relationships were also observed in sugarcane (Saccharum spp. hybrid; Meinzer & Grantz 1990) as K and gs varied with plant age and soil moisture. A strong linear relationship between gs and K was apparent in sugarcane ( Meinzer & Grantz 1990). A similar linear relationship between gs and K was observed in cotton ( Fig. 2) as both declined with increasing exposure to O3.

Figure 2.

. Measured combined relationship for two species of cotton between midday stomatal (gs; vapour phase) and hydraulic (K; liquid phase) conductances to water transport. Data are taken from Fig. 1(a,c). The line is obtained by regression.

Exposure to O3 may reduce stomatal conductance in the short term. This may involve direct impacts on guard cell metabolism and indirect effects mediated by inhibition of mesophyll photosynthetic function and resulting increases in intercellular CO2 concentration ( Farage & Long 1995; Farage et al. 1991 ; Moldau et al. 1990 ). However, O3 may also induce accelerated senescence and regrowth of new leaves, which exhibit enhanced rates of stomatal gas exchange. Such compensatory assimilation over longer time scales may offset the impact of short-term effects of O3 on stomatal conductance and carbon assimilation ( Pell et al. 1994 ). In the case of chronic exposure to O3, the correlation between gs and K may reflect a longer-term physiological integration. Reduced K could indirectly reduce gs and thus mediate deleterious effects on shoot gas exchange, limiting carbon acquisition and biomass productivity independently of direct effects of O3 on stomata. Recent evidence supports the occurrence of both a hydraulic linkage between edaphic conditions and stomatal response (e.g. Fuchs & Livingston 1996), and chemical communication from roots to shoot (e.g. Dodd et al. 1996 ). During plant development in the presence of chronic O3 exposure, disrupted allocation of biomass to roots could directly reduce root function and indirectly mediate the observed reductions in gs. Experimentally, this has proven difficult to demonstrate.

Model simulations

Stomatal and hydraulic conductances

We use a simulation model to test the hypothesis that O3-induced reduction of K is sufficient to mediate dose-dependent reductions in gs, reproducing the linear relationship between gs and K. The model excludes any direct effect of ozone on leaf gas exchange, but links K to O3 exposure, ψm and ψe to K, and gs to ψe.

The observed diurnal course of gs in the field ( Fig. 3; open squares) was accurately reproduced by the model under conditions similar to those reported by Grantz et al. (1997) of low ambient ozone concentration and non-limiting K ( Fig. 3; outer line; K > 9·4 mmol m−2 MPa−1 s−1). Agreement was particularly good during the midday period, when gs is frequently characterized, although under some conditions of soil moisture and V (as well as O3, see below) this may correspond to a period of partial midday stomatal closure ( Tenhunen et al. 1984 ).

Figure 3.

. Representative diurnal course of stomatal conductance (gs) measured in a well-irrigated, commercial field of Acala cotton (□) and modelled with non-limiting hydraulic conductance (K > 9·4 mmol m−2 MPa−1 s−1) corresponding to low ambient exposure to O3 during plant development (outer, bell-shaped curve). Increasing simulated exposures to O3 (successively lower K; from eqn 1) progressively reduces modelled maximum gs through increasingly severe midday stomatal closure (successively lower lines).

Increasing concentrations of O3 resulted in corresponding reductions in hydraulic conductance ( Fig. 1a). This physiological impact of O3 on K (eqn 1) is the only direct effect of O3 incorporated in the model ( Fig. 4a), yielding a decline in simulated K with increasing O3 exposure. Simulated midday leaf water potential (ψm) exhibited a slight reduction with increasing O3 concentration ( Fig. 4b), consistent with reduced root hydraulic capacity and reproducing qualitatively the experimental observations obtained in Pima cotton ( Fig. 1b; solid line).

Figure 4.

. Modelled relationships between 12 h seasonal mean concentration of O3 during plant development and (a) leaf area-specific plant hydraulic conductance (K; from eqn 1), (b) midday (14:00–14:30 PDT) water potential of exposed, transpiring leaves (Ψm), and (c) midday stomatal conductance to water vapour of exposed, transpiring leaves (gs).

The simulated diurnal time courses of gs exhibited increasingly severe midday depressions with increasing O3 exposure ( Fig. 3; successively lower lines and smaller K). The simulated midday stomatal closure began earlier in the day and persisted longer with decreasing K ( Fig. 3). The resulting midday values of simulated gs ( Fig. 4c) described an O3 dose–response relationship similar to the observed relationship (cf. Fig. 1c). The near homeostasis of simulated ψm ( Fig. 4b) thus reflects the parallel declines of simulated midday gs and K (cf. Fig. 4a,c), much as actual ψm ( Fig. 1b) reflects the parallel declines in actual gs and K (cf. Fig. 1a,c).

A strong correlation ( Fig. 5; solid crosses, solid line) was apparent between these simulated midday values of gs ( Fig. 4c) and K ( Fig. 4a), similar to that observed (cf. Fig. 2). The data obtained in field exposure chambers (triangles; from Fig. 2) were well described by the model regression ( Fig. 5, triangles). The data obtained from a commercial cotton field with contrasting soil moisture were also well-described by the relationship ( Fig. 5; squares), particularly at low values of K and gs, although gs was higher than predicted immediately following irrigation ( Fig. 5; upper squares). Deviations of experimental observations from the model relationship ( Fig. 5; cf. triangles and squares, solid line) are well within expected levels of experimental variability (cf. Fig. 2). Yet this simulated relationship developed from model iteration in the absence of any direct modelled impact of O3 on leaf gas exchange.

Figure 5.

. Modelled and measured relationships between midday stomatal (gs; vapour phase) and hydraulic (K; liquid phase) conductances to water transport. Data are from field exposure chambers (▵, ▿; from Fig. 2) or from a commercial Acala field in which hydraulic conductance varied with soil moisture (□). Model output values (+) are obtained as in Fig. 3 (means 14:00–14:30 PDT) with discrete input values of O3 concentration during plant development. The line is obtained by regression through the model output only.

The model ( Lynn & Carlson 1990) was developed independently of the plant species (cotton) and region (high radiation and evaporative demand; Grantz et al. 1997 ) to which we have applied it. The correspondence between simulated and measured relationships at the single leaf level supports the hypothesis that chronic O3-induced reductions in gs could be mediated by reduced carbon allocation to roots, reduced K, and a decline in midday ψe. A similar relationship between historic air pollution and K observed in Scots pine (Pinus sylvestris L.; Rust et al. 1995 ) was also accompanied by a parallel decline in gas exchange.

Canopy fluxes of water vapour and ozone

It is difficult to observe effects of ambient ozone on extensive canopies, yet these are hypothesized to lead to ecosystem-scale effects on productivity, species diversity, crop yield and forest decline ( MacKenzie & El-Ashry 1989). Simulations from the present model over a range of O3 exposures allowed these effects to be investigated.

Midday values of leaf to air vapour pressure difference in the canopy (V;Fig. 6a) increased with increasing O3 concentration and decreasing K. The peak value of V at 14:00 PDT nearly doubled from 1·75 to 3·25 kPa when K decreased from its non-limiting value of 9·4 to 2·2 mmol m−2 MPa−1 s−1 in response to simulated exposure to elevated O3 during plant development. The increase in V reflected reductions in both evaporative cooling of leaves and in canopy humidification, as gs declined with K. The substantial midday stomatal closure at the single leaf scale ( Fig. 3) led to a substantial reduction in daily water vapour flux at the canopy scale (E; successively lower lines with lower K, Fig. 6b), although not to a corresponding midday depression of simulated E (e.g. Tenhunen et al. 1984 ) even at very small values of K. This uncoupling of midday E from gs reflects the occurrence of feedback between gs and V in the canopy that is not observed at the single leaf level ( Jarvis & McNaughton 1986).

Figure 6.

. Modelled diurnal courses in a commercial cotton field of (a) leaf to air vapour pressure difference (V) and (b) evapotranspirational water vapour flux (E) under non-limiting hydraulic conductance (inner line in (a); outer line in (b)) and under increasing simulated exposure to O3 (lower K; as in Fig. 3).

The O3 concentration within the canopy air space (O3,c) increased with decreasing K ( Fig. 7a) and consequent decreasing gs ( Fig. 3). While K was modelled as a function of O3 concentration during simulated plant development, the regional values of tropospheric O3 (Ca; eqn 5a) over the simulated measurement day were independent of these values and varied diurnally with the same time course and magnitude for all values of K. The increasing values of O3,c within the canopy with decreasing K reflect the reduced uptake of ambient ozone by individual leaves within the canopy associated with reduced gs and the reduced sink strength of the cotton canopy for O3.

Figure 7.

. Modelled diurnal courses in a commercial cotton field of (a) canopy-level concentration of O3 and (b) canopy ozone depositional flux (F) under non-limiting hydraulic conductance (inner line in (a); outer line in (b)) and under increasing simulated exposure to O3 (lower K; as in Fig. 3). The squares in (b) are measurements obtained in a well-irrigated commercial field as in Fig. 3.

The ozone flux (F) predicted by the model under conditions of high hydraulic conductance (0·032 μmol m−2 s−1 near midday; Fig. 7b; outermost curve with K > 9·4) agreed well with field observations (0·031 μmol m−2 s−1 near midday; squares, Fig. 7b; Delany et al. 1992 ) obtained with the eddy covariance technique. This deposition of O3 to cotton, and to other crop surfaces, makes a substantial contribution to O3 removal from the atmosphere in this environment ( Grantz et al. 1994 ).

Canopy O3 flux ( Fig. 7b), including plant and soil components, declined with decreasing K despite the greater driving force for plant uptake represented by the greater O3 concentration near the leaves ( Fig. 7a). This represents a negative physiological feedback at the canopy scale that complicates efforts to predict ozone phytotoxicity in the field from exposures in well-stirred chambers, or from flux–response relationships obtained with individual leaves. This interaction also suggests an additional undesirable consequence of potential climate change. Tropospheric O3 concentrations may exhibit positive feedback, as increasing O3 concentration leads to reduced vegetative removal of O3 from the atmospheric mixed layer.


Experimental exposure to O3 caused a reduction in K, restricting water availability to transpiring leaves. This reduction in whole-plant K is attributed to reduced biomass allocation to roots and consequent reduction in root hydraulic capacity. Only minimal reduction in transpiring leaf ψm was observed in spite of the reduced K because of a concomitant reduction in gs. The resulting correlation between liquid phase (gs) and vapour phase (K) conductances to water transport was highly significant.

Application of a soil/plant/canopy/atmosphere flux model that reduced K realistically with O3 exposure, and parameterized gs partially as a function of leaf epidermal water potential (ψe), itself a partial function of K, reproduced these systemic effects of ozone exposure on gas exchange performance. Simulated diurnal time courses of gs agreed well with observations in a low O3 field environment. Increasing exposure to O3 during simulated plant development led to reduced simulated K and increasingly severe simulated midday stomatal closure. The model reproduced the significant linear correlation observed experimentally between midday gs and K, maintaining the homeostasis of simulated ψm. The model reproduced canopy-scale fluxes of O3 under low O3 field conditions, indicating that these simulations scale appropriately to the extensive canopy level where quantification of O3 effects has proven difficult. The canopy simulations indicate that deposition of O3 to vegetated surfaces may decline with increasing concentration of tropospheric O3, an unwelcome positive feedback in biosphere–atmosphere exchange.

We conclude that observed effects of chronic O3 exposure on leaf and canopy gas exchange do not require postulation of a direct physiological impact of O3 on stomatal or mesophyll photosynthetic function. Rather, O3-induced reduction of root hydraulic function mediated by altered carbohydrate allocation at the whole-plant scale is sufficient to mediate these chronic oxidant effects.


The authors thank Dr Pat McCool for many helpful discussions regarding the field chamber experiments, Dr Kel Wilson for many helpful discussions regarding model development, Mr David Vaughn for help with analysis and preparation of graphics, and the US Department of Commerce, National Oceanic and Atmospheric Administration (NOAA), National Weather Service in Hanford and Fresno, Callifornia, for providing atmospheric sounding data. Support to D.A.G. from the San Joaquin Valley Air Pollution Study Agency through Award No. 91-11, the California Environmental Protection Agency Air Resources Board through Contract No. 93-731, and the US Department of Agriculture NRICGP through Agreement No. 96-35100-3841 is gratefully acknowledged.