Scaling ozone effects from seedlings to forest trees


Author for correspondence: Lisa SamuelsonTel: +1 334 8441040 Fax: +1334 8441084


Biospheric ozone has become a widely distributed air pollutant, and a growing body of research indicates that ozone impacts forest health and productivity. Ozone effects are mediated by the ozone concentration present in the external environment and the movement of ozone into the leaf via the stoma. The cumulative dose received by the plant is, in the simplest terms, a function of ambient ozone concentration and stomatal conductance to water vapor. This relationship is important in understanding ozone flux into the leaf and subsequent ozone response in plants. Here, current progress in understanding ozone uptake in juvenile and mature trees is examined. Through an analysis of two long-term case studies, the significant uncertainty in assessing ozone effects on forests is pinpointed to be the scaling of ozone sensitivity from controlled seedling studies to large forest trees. A rigorous statistical and monitoring approach, which includes ozone uptake as a cause variable, may provide the missing information on processes that are known to be important to risk assessment of ozone impacts on forest trees.


  • Summary 21

  • I. Introduction 22

    • 1. Background 22

    • 2. Characterization of ozone exposure 22

    • 3. The need for scaling 23

  • II. Scaling from seedling to tree, evidence from a Quercus rubra case study 24

    • 1. Study background 24

    • 2. Facilities and measurements 24

    • 3. Ozone exposure dynamics 24

    • 4. Above-ground processes 25

    • 5. Below-ground processes 26

    • 6. A process modelling exercise 27

    • 7. Conclusions 28

  • III. Scaling from chamber to forest, evidence from a field case study 29

    • 1. Study background 29

    • 2. Field sites and measurements 30

    • 3. Ozone exposure dynamics 30

    • 4. Stomatal conductance and ozone uptake in forest trees 30

    • 5. Conclusions 32

    • IV. Evidence from a scaling exercise 33

  • V. Concluding remarks 36

  • Acknowledgements 37

  • References 37

I. Introduction

1. Background

Changes in the chemical climate of North America and Europe became an issue of growing concern in the late 1970s and early 1980s, as the development of precipitation monitoring networks began to indicate that the chemistry of rainfall was being modified by anthropogenic emissions associated with the combustion of fossil fuels. Air pollution, once thought to be largely an urban problem, began to take on new dimensions as it became clear that significant amounts of pollutants were being transported considerable distances downwind from both major point sources, as well as from significant area sources. One outcome of this realization was the deployment of more extensive monitoring networks combined with intensive studies of pollutant transport. A major discovery evolving from these evaluations was the confirmation that ozone was no longer an urban pollutant confined to major metropolitan areas, but a widely distributed pollutant often occurring at higher concentrations in rural locations than in cities, in both North America and Europe (National Research Council, 1991; Chameides & Cowling, 1995; Fuhrer et al., 1997; Stockwell et al., 1997).

A major conclusion derived from extensive evaluations undertaken in the USA as part of the National Acid Precipitation Assessment Program (NAPAP) was the identification of ozone as the pollutant of greatest concern with respect to potential regional scale impacts on forests (Barnard et al., 1990). Similar conclusions were reached in Europe as a result of United Nations Economic Commission for Europe (UNECE) analyses (Fuhrer et al., 1997). The acid rain era also saw a shift in research on air-pollution effects away from the impacts of acute exposures to studies related to the long-term chronic impacts of pollutants, and the impact that ozone might have on the health and productivity of forest systems (Skelly et al., 1997).

Many studies focusing on ozone effects have been conducted since the phytotoxicity of ozone was first described in the 1950s (Middleton, 1956; Richards et al., 1958). Much of the work on trees has been summarized by Pye (1988); Barnard et al. (1990); Sandermann et al. (1997); and Chappelka and Samuelson (1998). Lefohn et al. (1997), summarizing the results of several studies, observed that the effects of ozone on individual trees were complex, involving processes that could vary both within and among forest species in response to a host of environmental factors, the most important being availability of water during the growing season, especially during periods of elevated ozone. Thus, the risk ozone poses to the forest is dependent on both the variation in ozone exposure across a particular region, and the variation in key plant and environmental variables during the same time frame (Hogsett et al., 1997).

2. Characterization of ozone exposure

In order to cause injury, ozone must enter the plant through the leaf stomata, dissolve in the aqueous layer lining the cell walls, diffuse through the cellular membrane, and react with cellular components and metabolic processes. However, because the magnitude of stomatal conductonee to water vapor (gH2O) varies for each individual plant, and even among leaves on the same plant, a direct measure of ozone uptake is difficult (US Environmental Protection Agency, 1996). This situation is made more complicated by the fact that at least some plants have the ability to mitigate at least a portion of the potential ozone impact through natural defence processes (Pell et al., 1994).

Although the amount of work conducted on tree species is limited compared with the body of literature available on crops, fairly consistent patterns are beginning to emerge. Based on an extensive review of the available literature on juvenile trees, Chappelka and Samuelson (1998) concluded that a consensus is developing that ozone stress can reduce carbon (C) fixation; alter rates of leaf and root respiration; shift the partitioning of C into different chemical forms; and disrupt C and nutrient allocation patterns. Similar conclusions were drawn for crops, but the degree to which crop responses would describe ozone effects in long-lived trees was unclear. Reich & Amundson (1985) and Skärby et al. (1998), among others, suggest that ozone concentrations typical of moderately polluted areas in the USA and Europe are sufficient to begin to induce these changes over the long-term, although the majority of significant growth reductions due to ozone reported in the literature occurred in seedlings at concentrations above ambient.

Historically, rather than attempting to assess uptake and establish a dose-based standard for the protection of crops and forests, peak values based on 1 h averages were used in the standard-setting process. These values in the USA are based largely on human health concerns (primary standard) rather than public welfare considerations (secondary standard). The secondary standard in the USA, in effect from 1979 to 1997, was set to equal the 1 h primary ozone standard of 120 ppb. Because many of the early studies of biological effects on ozone focused more on visible injury in crops, this approach seemed reasonable at the time. A simple time-averaged concentration-based standard also has the advantage of being easier to establish and enforce as a nationwide criterion. The secondary standard has been replaced by a new 8 h primary standard of 80 ppb, which is currently under litigation (Anon., 1997).

As emission control strategies shifted in focus from local to more regional exposures, research efforts on biological effects began to investigate the influence of chronic exposures on longer-lived species, such as trees. Much of the ensuing work, as summarized by US Environmental Protection Agency (1996), demonstrated that at chronic levels of exposure trees tend to respond to the accumulated dose of ozone above certain threshold levels. One of the first attempts to define an exposure threshold parameter for adverse effects on agricultural crops was proposed by Lefohn & Benedict (1982), who selected a threshold level of 100 ppb. As early as 1988 the European Community agreed to adopt critical load concepts as a means to evaluate the adverse effects of ozone (Bull, 1991), and by 1994 had agreed on an accumulated exposure over a threshold of 40 ppb (AOT40) as an index (Fuhrer et al., 1997).

Following the European lead, the US Environmental Protection Agency conducted a recent re-evaluation of the secondary National Ambient Air Quality Standard for ozone, and both threshold (SUM60) and sigmoidally weighted (W126) cumulative indices were evaluated. Most cumulative forms attempt to take into account, either directly or indirectly, evidence that peaks produce a disproportionate response relative to lower concentrations, by selecting a threshold value below which the ozone concentrations are not accumulated (US Environmental Protection Agency, 1996). Hourly values which fall above the threshold are then added together to give a cumulative exposure for the ozone season. This differs slightly from the AOT40 approach, where only the mathematical difference between the threshold value and the hourly mean value is accumulated to give a total exposure for the ozone season. The AOT approach is thought to place more emphasis on that portion of the exposure which is biologically effective (Fuhrer et al., 1997). In formulating any cumulative index, consideration must also be given to the number of h d−1 used for cumulative purposes (generally 24 vs 12 h), as well as the arbitrarily assigned length of the ozone season (generally 90 vs 150 d). Recent work has explored the use of dose-based rather exposure-based standards for protecting vegetation from ozone (Rubin et al., 1996; Massman et al., 2000).

While it is not our intention to debate the relative merits of various regulatory approaches or the indices used to assess them, it is important to acknowledge the role that regulatory issues have played in the design and conduct of biological effects research.

3. The need for scaling

Despite decades of research on ozone sensitivity in tree seedlings, forest responses to ozone are still poorly understood. Development of ozone standards designed to protect forest resources from ecological damage must rely on scaling ‘individual-based mechanisms to patterns observed at higher scales’ (Levin, 1993), notwithstanding the uncertainty in predicting the ozone sensitivity of mature trees. Pye (1988) was among the first to note that most of our insights to date on the potential response of trees and, by extrapolation, of forests to ozone have been derived from short-term studies on seedlings. A series of long-term observational studies of ozone injury in mature Pinus ponderosa and Pinus jeffreyi conducted in California by Miller et al. (1997) found that variation in productivity, carbohydrate reserves, and stemwood vs root allocation patterns between seedlings and mature trees made extrapolation of seedling studies to forest trees difficult. Kelly et al. (1995) also began to question the efficacy of using seedling response, even from long-term studies, to represent mature tree response to ozone, as no studies were available at that time to provide mature tree functions or to test the extrapolation of seedling responses to mature trees. Subsequently, the 1990s saw an increase in longer-term studies, as well as studies of mature trees in both controlled and uncontrolled environments (Sandermann et al., 1997; Chappelka & Samuelson, 1998). Although limited in number, studies intended to assess the seedling–tree relationship were initiated (Grulke & Miller, 1994; Laurence et al., 1994; Thornton et al., 1994; Kelly et al., 1995; Laurence et al., 1997).

In addition, mechanistic carbon flux models of the type developed by Weinstein et al. (1991, 1992) were proposed as a means to take physiological insights from comparative seedling and mature tree studies, and to scale ozone responses from seedlings to mature trees. Model predictions were compared with empirical data on Picea rubens (Laurence et al., 1993) and Quercus rubra (Weinstein et al., 1998). Carbon flux models provided a means to integrate and scale plant physiological responses from seedlings to trees, and have recently been linked to stand-level models (Laurence et al., 2000). The EPA developed a strategy that made use of regionally representative tree species as a surrogate for the forest in that region (Hogsett et al., 1993). Others, such as McLaughlin & Downing (1995), proposed alternative, geostatistically based approaches that make use of observed measurements of tree growth from the region, and the known relationships between ozone response and soil water availability, as a means of integrating the variety of physical and biological factors affecting ozone response at the forest-stand level.

Paralleling the need to scale from seedlings to trees was the need to develop ozone exposure indices for various landscape units, so that dose–response functions derived from individual species and extrapolated with carbon flux models could be projected across the landscape. Geostatistical approaches were proposed as a means to solve this problem and provide an estimate of ozone exposures across a region, convert the estimated exposure to a probable dose, and then equate that dose to a stand response (Hogsett et al., 1997; Lefohn et al., 1997).

As noted earlier, one of the outcomes of the NAPAP analysis was the identification of uncertainties in extrapolating ozone response from seedlings to individual trees, to forest stands, and to the landscape or regional level. Ongoing discussions in the research community during the 1980s and early 1990s emphasized the need to develop a continuum of understanding that would allow extrapolations to be made through a meshing of known physiological responses and geographical information systems (GIS). Hogsett et al. (1993), focusing on the US Environmental Protection Agency’s needs, proposed a hierarchy to scale ozone exposure from the response of individual plants to landscape units, and made extensive use of a single-tree process model to provide the initial scaling from seedling to tree, and a stand model to convert individual tree response to the forest level. Response functions would be developed at all three levels, which could then be used with GIS procedures to provide spatial risk characterization (Hogsett et al., 1993). However, a significant void in this approach was, and still is, a lack of comparable information on seedlings and mature trees with which to test the validity of using data obtained from seedlings to represent mature tree responses.

In our review, we demonstrate that one of the most significant uncertainties in assessing ozone effects on forests is the scaling of ozone sensitivity from controlled seedling studies to large forest trees. We propose that the use of gH2O derived from studies on potted seedlings may underestimate ozone uptake by large hardwood trees on average by 47%, and overestimate ozone uptake of large coniferous species on average by 26%. The challenge in scaling is in determining what processes control the system at each scale (Baldocchi, 1993). We suggest that gH2O is an important underlying factor controlling variation in ozone sensitivity between seedlings and mature trees. We support our arguments with empirical and modelling research from two long-term case studies, initiated to provide scaling linkages between seedlings and mature forest trees, and a literature-based scaling exercise. The outcomes of scaling quantitative mechanisms from small scales to synthetic assessments over large scales have been described as ‘grand expressions of scientific confidence or sobering warnings that information is still missing’ (Field & Ehleringer, 1993). Our analysis issues a ‘sobering warning that information is still missing’, and, thus, conservatism is necessary when extrapolating ozone response from seedling to mature tree.

II. Scaling from seedling to tree, evidence from a Quercus rubra case study

1. Study background

A pilot study was initiated in 1990 (Edwards et al., 1994), and a full scale project in 1992 (Samuelson & Edwards, 1993), to evaluate the response of 30-yr-old trees and 2-yr-old seedlings of Quercus rubra to three different levels of ozone exposure. Emphasis was placed on making, to the greatest extent possible, comparable measurements over several growing seasons on both trees and seedlings. Measurement comparisons focused on the general areas of growth foliar physiological response, and below-ground processes.

2. Facilities and measurements

The facilities used to conduct this study are described in detail by Samuelson & Edwards (1993); Edwards et al. (1994); and Samuelson et al. (1996). In brief, the study was conducted during the period from 1992 to 1994 in an open-top chamber facility located in a Q. rubra seed orchard on the Tennessee Valley Authority Reservation at Norris, TN, USA. Nine reproductively mature trees, approx. 30 years in age at the start of the study, were individually enclosed in large open-top chambers. The large chambers were 9.2 m high and 4.6 m in diameter. Two-yr-old seedlings grown from seed collected from orchard trees were fumigated with ozone in nine smaller 2.4 × 3.0 m open-top chambers, with each chamber containing 30 potted seedlings. Variation in ozone concentration between seedling and mature tree chambers was, on average, < 5% (Samuelson & Edwards, 1993). To minimize any potential interactive effects of plant water stress with ozone exposure, supplemental water was applied when soil water potential fell below −0.2 to −0.5 MPa.

Seasonal diameter growth was measured on all trees, along with periodic height and diameter measurements on seedlings. Total biomass values were determined annually for seedlings through destructive harvests conducted on a subset of the seedlings in each chamber. Total leaf biomass was measured on each mature tree by collecting annual leaf fall. Specific leaf weights, bud weights, and fine root production and turnover were also determined. Carboxylation efficiency, quantum yield, chlorophyll fluorescence, dark respiration and water potential were examined in leaves of mature trees and seedlings, along with seasonal patterns of light-saturated photosynthesis and gH2O. Carbon partitioning and allocation was also investigated for both seedlings and trees during the final year of study.

The experimental unit in most data analyses was the chamber. Within an age grouping and date, responses were analysed by a randomized block design with blocks and ozone treatments as the main effects. Repeated-measures analyses were also conducted, with ozone treatment as the between-subjects main effect and seasonal sampling dates representing repeated trials. Regression analyses were used to determine significant ozone-treatment effects on growth and nutrition. No statistical comparisons were performed between seedlings and mature trees because the mean square error differed between age classes (Samuelson & Edwards, 1993).

3. Ozone exposure dynamics

Subambient (charcoal-filtered, on average 60% lower that ambient concentrations based on 7 h means); ambient; and twice ambient ozone treatments were applied continuously from mid-April through to late September during the 1992, 1993 and 1994 growing seasons. Mean monthly 7 h ozone concentrations, and monthly SUM00, SUM60 and AOT40 indices for each treatment and growing season, are summarized in Table 1. Ambient ozone levels were highest during the 1993 growing season, with a peak 1 h-value in the ambient chambers of 105 ppb. The 120 ppb US ozone standard was never exceeded in ambient air during the three growing seasons of the study. In the twice ambient chambers, peak 1 h mean values for each of the three study years were 159, 225 and 198 ppb All three of these mean values exceeded the 120 ppb threshold of the US ozone standard. Cumulative SUM00 values averaged over the three growing seasons were 31, 85 and 169 ppm h−1, respectively, for the subambient, ambient and twice ambient treatments (Samuelson et al., 1996). Mean AOT40 indices for the same period were 0, 7 and 67 ppm h−1 for the subambient, ambient and twice ambient treatments.

Table 1.  Monthly 7 h means and 12 h SUM00, 12 h SUM60, and 24 h AOT40 indices for ozone treatments delivered to Quercus rubra seedlings and mature trees during the 1992, 1993, and 1994 growing seasons at Norris, TN, USA
Ozone3 treatmentAprMayJunJulAugSepTotal
  • *

    na, not applicable. ppb, parts per billion; ppm, parts per million.

 7 h mean (ppb)1615 16141515na*
 SUM00 (ppm h−1) 8 6  6 5 4 5 34
 SUM60 (ppm h−1) 1 0  0 0 0 0  0
 AOT40 (ppm h−1) 0 0  0 0 0 0  0
 7 h mean (ppb)3938 38353534na
 SUM00 (ppm h−1)1217 15131512 84
 SUM60 (ppm h−1) 2 1  1 0 0 1  5
 AOT40 (ppm h−1)<1 1  1<1<1<1  5
Twice Ambient       
 7 h mean (ppb)7278 72666369na
 SUM00 (ppm h−1)1733 32251926152
 SUM60 (ppm h−1) 823 19131013 86
 AOT40 (ppm h−1) 416 13 8 5 8 54
 7 h mean (ppb)2920 19191916na
 SUM00 (ppm h−1) 8 8  7 6 6 3 38
 SUM60 (ppm h−1) 0 0  0 0 0 0  0
 AOT40 (ppm h−1) 0 0  0 0 0 0  0
 7 h mean (ppb)4950 53534637na
 SUM00 (ppm h−1)1720 201815 6 96
 SUM60 (ppm h−1) 3 4  4 5 2 1 19
 AOT40 (ppm h−1) 2 3  3 3 2<1 13
Twice Ambient       
 7 h mean (ppb)9298106949478na
 SUM00 (ppm h−1)3038 41343214189
 SUM60 (ppm h−1)2126 312523 9135
 AOT40 (ppm h−1)1418 211714 5 88
 7 h mean (ppb)1516 11 8 910na
 SUM00 (ppm h−1) 1 6  4 3 3 4 22
 SUM60 (ppm h−1) 0 0  0 0 0 0  0
 AOT40(ppm h−1) 0 0  0 0 0 0  0
 7 h mean (ppb)3945 41353735na
 SUM00 (ppm h−1) 221 14141212 75
 SUM60(ppm h−1) 0 2  1 1 1 0  5
 AOT40(ppm h−1)<1 2  1<1<1<1  4
Twice Ambient       
 7 h mean (ppb)6098 62798274na
 SUM00 (ppm h−1) 347 23333129166
 SUM60 (ppm h−1) 235 11202116105
 AOT40 (ppm h−1) 124  71010 7 59

4. Above-ground processes

The Q. rubra case study documented a linear reduction in light-saturated net photosynthesis and a maximum reduction in net photosynthesis of 60% in leaves of mature trees, in response to increasing internal ozone exposure (Fig. 1) (Hanson et al., 1994). By contrast, a maximum reduction of 10% in net photosynthesis was detected in seedling leaves. Since gH2O of seedlings was, on average, approximately half that of trees (Samuelson & Edwards, 1993; Hanson et al., 1994; Samuelson, 1994a), seedlings never accumulated the same seasonal internal ozone dose as the leaves of older trees (Fig. 1). Leaf C gain in trees was reduced by stomatal and nonstomatal factors because gH2O, carboxylation efficiency, and apparent quantum yield were reduced by ozone treatment (Samuelson & Edwards, 1993; Samuelson, 1994a).

Figure 1.

Relative reduction in light-saturated net photosynthesis of mature trees (closed circles) and seedlings (open circles) Quercus rubra in response to internal ozone uptake. Redrawn from Hanson et al., 1994.

The influence of ozone treatment on the translocation and partitioning of C was examined in seedlings and mature trees late in the growing season, when leaves were fully mature and roots probably acted as a strong sink for C (Adams et al., 1990). After 3 yrs of ozone treatment, ozone had no significant influence on C retention in seedling leaves (Fig. 2), but linear reductions in soluble carbohydrate concentrations in coarse and fine root tissues indicated an influence of ozone on C partitioning in seedling roots, and potential declines in sucrose translocation to below-ground tissues (Samuelson & Kelly, 1996). Despite lower soluble carbohydrate concentrations in roots in response to increasing ozone exposure, root mass, total biomass, tissue allocation, diameter and height of seedlings were not significantly affected by ozone exposure for three growing seasons (Samuelson et al., 1996).

Figure 2.

Foliar retention (percentage of initial fixation) of 14C in leaves 1, 5 or 7 d after labelling shoots of (a) seedling and (b) mature trees of Quercus rubra in response to subambient, ambient or twice ambient ozone concentration. Initial fixation of 14C was measured on 1 September 1994 during the third season of ozone exposure. Standard deviations are indicated by bars. An asterisk indicates a significant (P < 0.10) linear response to ozone. Redrawn from Samuelson & Kelly, 1996.

Declines in C assimilation in leaves of mature trees resulted in increased C retention and reduced starch concentration in leaves, and increased starch concentration in branches (Fig. 2; Table 2). In addition, nitrogen concentrations in coarse and fine roots of mature trees were increased in response to increasing ozone concentration (Samuelson et al., 1996), possibly as a result of reduced C translocation to roots and lower cumulative root production in mature trees (Kelting et al., 1995). Other researchers (Meier et al., 1990; Paynter et al., 1991; Buckner & Ballach, 1992; Friend et al., 1992; Smeulders et al., 1995) report retention of C in leaves and decreased foliar starch concentration in response to ozone for a variety of tree species. Because Wullschleger et al. (1996) observed no influence of ozone on construction or maintenance respiration in leaves of mature trees during the third year of exposure, elevated foliar respiration was unlikely to be the source of reductions in starch partitioning.

Table 2.  Ozone effects on net fine root production, gross fine root production, and below-ground carbon-use efficiency during the 1993 growing season, and starch concentration in foliage and branches, leaf mass, and 1-yr diameter increment in the lower stem and upper crown at the end of the 1994 growing season in mature Quercus rubra trees exposed to different ozone treatments
 Ozone treatment 
ComponentSubambientAmbientTwice ambientP > F
  1. Different letters indicate differences between means;*denotes a linear response. Data from Kelting et al. (1995); Samuelson & Kelly (1996); Samuelson et al. (1996).

Net fine root production (g m−2) 96a166b122a<0.10
Gross fine root production (g m−2)154b230a209ab<0.10
Below-ground carbon use efficiency (g m−2)  0.62a  0.73b  0.58a<0.10
Leaf starch (µmol g−1) 54.6 38.1 31.8<0.10*
Branch starch (µmol g−1)258.9243.2318.6<0.10*
Leaf mass (kg) 11.2  8.2  8.2 0.190
Lower stem diameter increment (cm)  2.9  3.5  2.3 0.149
Within-crown stem diameter increment (cm)  1.5  1.6  1.0 0.280

5. Below-ground processes

Fine root production of mature trees and seedlings was measured only during the second and third seasons of ozone exposure. At the end of the second season, cumulative production and fine root turnover of mature trees in the twice ambient treatment had declined to approx. 62% and 33%, respectively, of the level observed in the subambient treatment (Kelting et al., 1995). In contrast, elevated ozone treatment did not influence fine root dynamics of seedlings. After three seasons of ozone treatment, net (but not gross) fine root production of mature trees was lower in the twice ambient than in the ambient treatment (Table 2). While the data are more equivocal than might be desired, they do suggest reductions in below-ground carbon-use efficiency as a function of elevated ozone (Table 2). In this context it is also useful to note that ozone did induce retention of C in leaves and branches of mature trees, but not seedlings (Samuelson & Kelly, 1996). Evidence of starch accumulation and subsequent crushing of phloem cells (Wellburn & Wellburn, 1994) is consistent with reports that ozone restricts phloem loading, and, thus, assimilate translocation, from shoots to roots (Okano et al., 1984). The outcome is a change in balance between root and shoot growth, leading to reduced root : shoot ratios and potential predisposition of ozone exposed trees to drought (Mansfield, 1988; Davidson et al., 1992) and winter desiccation (Chappelka et al., 1990).

6. A process modelling exercise

To further explore the importance of C dynamics in understanding differences in ozone sensitivity between seedling and mature Q. rubra, a modelling exercise using the TREGRO model was conducted to determine how differences in photosynthetic response to ozone between seedlings and mature trees would influence C allocation patterns (Weinstein et al., 1998). The hypothesis evaluated was that the stored C pool would buffer mature Q. rubra against the effects of a stress such as elevated ozone.

The TREGRO model developed by Weinstein et al. (1991) was parameterized to reflect the processes and rates associated with C allocation in Q. rubra (Weinstein et al., 1998). The model simulates the rates of C fixation and C accumulation in the reserve supply in the tree. The tree is divided into leaf, branch, stem and root compartments, and C may be stored as living tissue, wood or total nonstructural carbohydrate (TNC) as appropriate (Weinstein & Yanai, 1994). The demands for C by leaf, branch, stem and root growth are calculated, with different C sinks being allocated varying portions of their demand, based on the phenology of growth through the year and the availability of other essential resources (Weinstein & Yanai, 1994). Data generated as part of the seedling and mature tree studies discussed previously were used to guide the parameterization of the Q. rubra model. The model was then used to examine the impact of ozone-induced reductions in photosynthesis on C allocation in seedlings and mature trees over a 2 yr period. For the purpose of these evaluations, the model was parameterized using measured and calculated rates, and pool sizes derived from the seedlings and mature trees grown in the subambient ozone treatment.

Based on these simulations, the mature tree was much more ozone-sensitive than the seedling, because of greater ozone uptake. At ambient and twice ambient ozone concentrations, simulated photosynthesis in mature trees declined by 27% and 63%, respectively, similar to reductions observed in the field experiment (Weinstein et al., 1998). These reductions in simulated photosynthesis had the greatest impacts on TNC storage pools and fine root replacement (Fig. 3). Other tissue compartments also experienced negative growth during the 2 yrs simulation (Fig. 3). Seedling response, in contrast, showed little change in growth of major components in either ambient or twice ambient simulations (Weinstein et al., 1998). Although field observations of seedlings and mature trees exposed to subambient ozone concentrations agreed with simulation values, field observations (Samuelson et al., 1996) did not support decreases in mature tree branch and stem growth in response to twice ambient ozone concentration predicted by the model (Fig. 3). Root growth of mature trees did exhibit a declining trend with increasing ozone exposure, dropping by 33% (Kelting et al., 1995) in response to the twice ambient ozone treatment, but the observed value was substantially below the reduction predicted by the model.

Figure 3.

TREGRO simulated effect of ozone exposure on carbon allocated to different components over 2 yrs in mature tree Quercus rubra. TNC, total nonstructural carbohydrate. Open bar, No ozone; closed bar, Ambient; dotted bar, twice ambient. Redrawn from Weinstein et al., 1998.

Weinstein et al. (1998) proposed that discrepancies between model predictions and field observations for mature trees may be the result of our inability to describe the effects of ozone on basic rates of C utilization, such as maintenance respiration or root senescence. It is possible that ozone stress reduced maintenance or growth respiration of woody tissues in response to reduced substrate supply or injury to metabolic processes (Wullschleger et al., 1996). Rakonczay (1997) observed lower stem respiration rates and less diameter growth in Prunus serotina and Acer rubrum seedlings in response to ozone exposure, but maintenance respiration was not separated from growth respiration.

Results from the TREGRO simulation of Q. rubra demonstrate the need for better understanding of ozone effects on C allocation in mature trees. Likewise, Constable & Taylor (1997), using TREGRO, simulated growth of reproductively mature P. ponderosa in response to different ozone levels, and determined that ozone sensitivity was strongly affected by genetic variation in gH2O and C allocation to fine root biomass. A 10 yrs simulation of ozone effects on 160-yr-old Acer saccharum indicated reductions in C storage pools, coarse root mass, and total fine root production (Retzlaff et al., 1996). However, the simulations were based on parameterization of fine root production and senescence, and photosynthetic response to ozone from seedling data. The Q. rubra TREGRO simulation, when viewed in conjunction with our experimental results, identified gH2O and C allocation as the key physiological processes in need of further investigation to facilitate effective scaling of ozone response to large trees using process models.

7. Conclusions

Greater leaf physiological sensitivity to ozone in mature trees was probably a result of a number of interacting factors, the most significant being the more than doubling of gH2O in leaves of mature trees relative to seedlings. In addition, seedlings had the ability to produce multiple growth flushes, and the potential to compensate for, or overcome, ozone damage to the first flush. Greater gH2O, coupled with the lack of recurrent flushing in mature trees, resulted in a significantly larger ozone dose per unit leaf area in mature tree leaves. The inability of mature Q. rubra trees to maintain photosynthetic rates when exposed to higher ozone concentrations indicates that, at the foliar level, trees were unable to compensate fully for ozone damage to photosynthetic processes. When considering above-ground growth, changes in C allocation and partitioning in branches and leaves of mature trees indicate the existence of compensatory responses that may postpone reductions in above-ground growth in response to ozone stress. At the whole-plant level, reductions in C allocation to cumulative fine root production indicate a cost to maintaining above-ground growth.

No statistically significant reductions in foliar biomass and stem diameter growth below and within the crown of mature trees were detected after three seasons of ozone treatment (Table 3). The statistical power of the study was limited due to low replication (n = 3). An analysis of the power of the F-test (Kirk, 1982), using a power (1-β) of 0.80, indicated that at least 10 replications would be needed to detect significant differences at α = 0.05 in diameter increment between the ambient and twice ambient treatments. Thirty chambers is an unrealistic sample size in light of the significant costs associated with large tree fumigations. Alternatively, a 69% reduction in diameter increment at α = 0.10 would be needed to detect a significant difference between the ambient and twice ambient treatments with three replications, an unrealistic reduction in growth given the long life span of trees and relatively short fumigation period.

Table 3.  Monthly 7 h means and 12 h SUM00, 12 h SUM60, and 24 h AOT40 indices for ambient ozone exposures during the 1994–97 growing seasons at three canopy heights in the Great Smoky Mountains National Park
Ozone indexCanopy height (m)JuneJulyAugSeptJuneJulyAugSeptJuneJulyAugSeptJuneJulyAugSept
Cove Mountain                 
7 h mean (ppb) 2273741404343414540413531414146
SUM00 (ppm h−1) 2 919241215161516161613 8161617
  71122261317181618181715 9171718
SUM60 (ppm h−1) 2 0 2 4 1 5 6 4 4 2 1 0 0 2 4 6
  7 0 4 7 2 7 9 6 6 4 2 1 0 4 5 8
 12 0 6 8 3 8 9 7 7 5 3 1 0 5 5 9
AOT40 (ppm h−1) 2 1 5 5 4 8 8 6 8 6 6 3 1 6 6 8
  7 2 7 6 61111 812 8 8 5 2 8 810
 12 3 8 7 71211 91310 9 6 3 9 911
Twin Creeks                 
7 h mean (ppb) 227161824263023232727262129363734
SUM00 (ppm h−1) 2 8 5 6 8 8 9 6 7 8 9 8 7 9111110
  812 8 91111131110151211 912151513
 1414 91012121412111614131013171614
SUM60 (ppm h−1) 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1
  8 2 0 1 0 1 2 1 1 3 1 0 0 1 3 3 3
 14 3 1 1 1 1 3 2 1 4 2 1 0 2 5 4 4
AOT40 (ppm h−1) 2 0 0 0 0 0 1 0 0 0 0 0 0 0 1 1 1
  8 2 0 1 1 1 2 1 1 4 2 1 1 1 3 3 3
 14 2 0 1 1 1 3 2 1 4 2 1 1 2 4 4 3

It is unclear if the photosynthetic dose–response function depicted in Fig. 1 is similar between seedlings and trees, because seedlings never accumulated the same maximum dose as trees, and the rate of ozone accumulation within the leaf varied between seedlings and trees. Unless we subjected seedlings and trees to different ozone exposures to compensate for lower gH2O and recurrent flushing in seedlings, seedlings would never accumulate the same seasonal dose as large trees. How would the slope of the seedling response function be affected if a dose of 35 mmol m−2 was accumulated by the end of the season? If seedling leaves never accumulate the same internal dose, comparison of dose–response functions between seedlings and trees may be statistically biased. Thus, it is difficult to weight the contributions of other mechanisms such as antioxidant potential (Pell et al., 1999) in explaining age-related divergence in ozone sensitivity when variation in ozone uptake is so overwhelming. Brendley (1996) proposed that insensitivity of foliar Rubisco protein to ozone in seedling and mature Q. rubra may be a result of antioxidant compounds which are synthesized in Q. rubra leaves regardless of tree age.

Variation in gH2O between seedling and mature Q. rubra indicates that gH2O is useful in scaling ozone response among trees of different sizes, and in describing how internal dose is accumulated across a growing season (Samuelson & Kelly, 1997). Understanding the source of higher gH2O in mature trees in the Q. rubra case study is central to scaling gH2O and ozone uptake between seedlings and mature trees of other species. Variation in canopy microclimate between seedlings and mature Q. rubra did not play a dominant role in determining foliar ozone sensitivity, because seedlings grown in the canopies of mature trees and exposed to ozone responded as did seedlings exposed to ozone in the open environment (Samuelson, 1994a). Photosynthetic capacity and foliar nitrogen (N) concentration are often positively correlated (Field & Mooney, 1986), and foliar N may explain, to some degree, differences in leaf gas-exchange rates between seedling and mature trees. Foliar N concentrations were 27% higher at the end of the third growing season in mature trees than in seedlings, although variation in foliar N concentration between seedlings and mature trees was inconsistent within and between growing seasons, and no differences in soil nutrient concentrations between seedlings and mature trees were detected (Samuelson et al., 1996).

Empirical data and the modelling exercise indicated that an increase in demand for carbohydrates may underlie higher foliar gas-exchange rates in mature Q. rubra. A greater relative reduction in foliar C retention (Fig. 2) and a lower starch concentration (Table 2) in leaves of mature trees than in seedlings exposed to subambient ozone concentrations suggest that C demands are proportionately greater and foliar C reserves are proportionally smaller in large northern red oak trees. Mature tree characteristics, such as a large respiratory mass, flowering and fruiting, and complex root and canopy systems, may increase the number and strength of C sinks. For instance, mature trees produced acorns in each of the three growing seasons of the study.

In summary, it appeared as if age- and size-related differences in C allocation and partitioning in Q. rubra were important in understanding higher net photosynthesis and gH2O in leaves of mature trees compared with seedlings. We speculate that because of a deciduous habit, sensitivity of root production of mature trees to ozone, and reports of continued reduction in root production after the cessation of ozone exposure (Andersen et al., 1997), mature Q. rubra would be unable to recover maximum rates of net canopy C gain with continued ozone stress or even with brief respites from elevated exposure.

III. Scaling from chamber to forest, evidence from a field case study

1. Study background

To determine how well results from the Q. rubra case study applied to forest trees, and to better quantify physiological differences between small and large forest trees, we initiated a long-term study of leaf physiology and ozone uptake rates in understorey and overstorey Q. rubra, Prunus serotina and Acer rubrum. Our objectives were to compare seasonal patterns in gH2O and ozone uptake of deciduous forest trees in different canopy locations with rates from controlled studies, and to assess the potential for adverse ozone effects on forest trees using dose–response functions derived from controlled exposure studies. While gH2O has been relatively well characterized in seedlings, knowledge of temporal and spatial variation in gH2O of large forest trees is limited, and subsequently estimates of gH2O from young trees may lead to inaccurate estimation of CO2 and ozone fluxes at the tree and ecosystem level (Ryan, 1991). Furthermore, assessments based on process models of forest tree responses to varying environmental scenarios, such as global warming and elevated atmospheric CO2 concentrations, rely on estimates of gH2O to calculate transpirational water loss, net photosynthesis, water-use efficiency and pollutant uptake at different spatial scales (Reich et al., 1990; Amthor et al., 1994; Williams et al., 1996; Baldocchi, 1997; Ollinger et al., 1997; Weinstein et al., 1998).

2. Field sites and measurements

The study was conducted in the Great Smoky Mountains National Park, designated as an International Biosphere Reserve and a World Heritage Site. The park, of > 200 000 ha, is located in eastern Tennessee and western North Carolina, contains half of the old-growth forests of the eastern USA, and hosts more visitors than any other national park. Air quality issues, particularly ozone effects on forests within the park, are of increasing public concern (Shaver et al., 1994).

Stomatal conductance and ozone uptake were examined in understorey saplings (2–8 m high) and in the upper canopy of overstorey trees (10–23 m high) at Twin Creeks; a low elevation (600 m) site located in a cove; and Cove Mountain, a high elevation (1240 m) site located on a south-west facing ridge in the park. The two sites were described previously by Samuelson & Kelly (1997). Acer rubrum and P. serotina were studied at both sites, with the addition of Q. rubra at Cove Mountain.

Ozone concentrations were continuously monitored at three heights in the canopy (approx. 2, 8 and 13 m) at each site. Leaf gas exchange was monitored over four growing seasons (1994–97) in three understorey saplings and three mature trees of each species at Twin Creeks, and two to three understorey saplings and two mature trees of each species at Cove Mountain. Midday measures of in situgH2O were conducted on up to 10 leaves per tree and four leaves per sapling, at least once per month each season at ambient or light-saturating photosynthetic photon flux rate (PPFR) using portable gas-exchange systems (LI-6200, LI-1600, Li-Cor, Lincoln, NE, USA).

Cumulative leaf ozone uptake from June 1 to September 30 of each year was calculated using the product of midday in situgH2O, interpolated between hours and measurement sessions, and average hourly ozone concentrations from 11:00 to 17:00 hours at the appropriate height. Soil moisture to a 30 cm depth was measured at approx. 09:00 hours during leaf gas-exchange sessions using time domain reflectometry and stainless steel rods permanently installed near the base of each sapling and tree.

3. Ozone exposure dynamics

Over the 4 yr study, average seasonal 7 h ozone concentrations in the upper canopy ranged from 46 to 48 ppb at Cove Mountain, and from 36 to 47 ppb at Twin Creeks from June to September (Table 3). Twelve hours SUM00 and SUM60 indices and 24 h AOT40 indices were generally higher at Cove Mountain, the high elevation site, as a result of low diurnal variability in ozone concentration (Fig. 4), typical of high elevation sites (Taylor et al., 1992). Cumulative seasonal AOT40 and SUM00 indices in the upper canopy ranged from 4 to 13 and 45–60 ppm h−1, respectively, at Twin Creeks, and from 32 to 39 and 65–71 ppm h−1, respectively, at Cove Mountain. Average seasonal exposure indices indicated greater ozone exposure in 1997 than in preceding years at Twin Creeks, whereas at Cove Mountain exposures were more similar among years.

Figure 4.

Average hourly ozone concentrations 14 and 12 m above the forest floor in low elevation (Twin Creeks) and high elevation (Cove Mountain) forests, respectively, in the Great Smoky Mountains National Park, June–September 1997. Redrawn from Patterson et al., 2000.

Biological sinks for ozone in the upper canopy and lack of transport of ozone into the understorey (Fuentes et al., 1992) were probably responsible for the average 36% reduction in ozone concentration from the upper canopy to forest floor at Twin Creeks, the more sheltered site. The 14% difference in ozone concentration between the canopy and forest floor at Cove Mountain was similar to a 13% reduction in average ozone concentration from the canopy to forest floor observed in a ridge-top deciduous forest in central Pennsylvania (Skelly et al., 1996).

4. Stomatal conductance and ozone uptake in forest trees

Sun leaves of overstorey trees exhibited higher leaf N concentration, higher specific leaf weight, and higher rates of light-saturated and in situ net photosynthesis and gH2O than shade leaves of understorey saplings (Samuelson & Kelly, 1997). Nitrogen concentration expressed on a leaf area basis accounted for 47–80% of the variation in average seasonal gH2O between upper canopy leaves of mature trees and shade leaves of understorey saplings (Fig. 5) (Samuelson & Kelly, 1997).

Figure 5.

Relationship between average light-saturated net photosynthesis (Pn) or stomatal conductance (gH2O) and N concentration per unit leaf area for trees of different size: (a) black cherry (b) red maple, and (c) northern red oak. Each observation is a seasonal average for a seedling or mature canopy tree. Redrawn from Samuelson & Kelly, 1997.

Higher gas exchange rates (Figs 6–8) and, to a lesser extent, higher ozone concentrations resulted in greater internal ozone doses in the upper canopy than in the understorey, particularly in P. serotina and Q. rubra (Fig. 9). Understorey saplings exhibited low diurnal (Patterson et al., 2000), seasonal (Samuelson & Kelly, 1997), and yearly variability in gH2O and cumulative ozone uptake relative to upper canopy leaves of overstorey trees. Variability within a growing season in gH2O of overstorey trees was only weakly correlated with leaf microclimate (Samuelson & Kelly, 1997; Patterson et al., 2000). However, during the 1997 growing season positive correlations between soil moisture and gH2O were observed with soil moisture explaining > 50% of the seasonal variability in daily maximum gH2O of overstorey Q. rubra and P. serotina and understorey A. rubrum at the high elevation site (Patterson et al., 2000). It is unclear if higher ozone concentrations would have inhibited stomatal closure during drought, as reported for Fagus sylvatica (Pearson & Mansfield, 1992).

Figure 6.

Yearly average in situ stomatal conductance (gH2O) of overstorey (a) and understorey (b) black cherry at Cove Mountain (open columns) and Twin Creeks (closed columns) in the Great Smoky Mountains National Park. Standard deviations indicate variation between understorey or overstorey replicates.

Figure 7.

Yearly average in situ stomatal conductance (gH2O) of overstorey (a) and understorey (b) red maple at Cove Mountain (closed columns) and Twin Creeks (open columns) in the Great Smoky Mountains National Park. Standard deviations indicate variation between understorey or overstorey replicates.

Figure 8.

Yearly average in situ stomatal conductance (gH2O) of overstorey (open columns) and understorey (closed columns) northern red oak at Cove Mountain in the Great Smoky Mountains National Park. Standard deviations indicate variation between understorey or overstorey replicates.

Figure 9.

Yearly cumulative ozone uptake (gO3) by overstorey and understorey black cherry, red maple and northern red oak at Cove Mountain and Twin Creeks in the Great Smoky Mountains National Park. Standard deviations indicate variation between understorey or overstorey replicates.

Correlation between average seasonal gH2O (Figs 6–8) and leaf microclimate or soil moisture was examined using Pearson’s correlation coefficient (r). A significant (P < 0.10) positive correlation was detected between gH2O and PPFR for understorey black cherry at both sites (r = 0.77 and 0.70), overstorey black cherry at Twin Creeks (r = 0.54), understorey red maple at both sites (r = 0.62 and 0.80), and overstorey red maple at both sites (r = 0.72 and 0.86). Thus, from 29% to 74% of the yearly variation in average gH2O of red maple and black cherry could be predicted by PPFR during leaf measurement.

5. Conclusions

How applicable were leaf physiological rates from our Q. rubra case study to forest trees? Foliar N concentration, light-saturated photosynthesis, gH2O and midday leaf water potential of Q. rubra overstorey trees at Cove Mountain were similar to values observed for mature trees in the Q. rubra case study. For example, averaged from June to September over two growing seasons, gH2O of trees exposed to ambient ozone concentrations in the controlled study was 233 mmol m−2 s−1. Over four growing seasons, average gH2O was 225 mmol m−2 s−1 in mature Q. rubra forest trees. In addition, seasonal variation in light-saturated net photosynthesis was similar between the two case studies, and average seasonal maximum gH2O was 250 and 300 mmol m−2 s−1, respectively, in overstorey forest trees and mature trees fumigated with ambient ozone concentrations. Based on a linear response function derived from controlled ozone exposures in the Q. rubra case study (Hanson et al., 1994), the cumulative internal dose of 30 mmol m−2 to upper canopy leaves of overstorey forest trees in 1996 resulted in a 50% reduction in light-saturated net photosynthesis in the controlled experiment. Cumulative uptake of forest trees in 1996 was equivalent to uptake by trees exposed to twice ambient concentrations of ozone in the case study. While many factors hinder extrapolation from controlled to uncontrolled environments, the potential for ozone-induced reductions in photosynthesis in the upper canopy of Q. rubra at Cove Mountain is indicated.

In overstorey P. serotina trees at both sites, gH2O in July and August (Samuelson & Kelly, 1997) was surprisingly similar to gH2O of leaves in the upper canopy of forest P. serotina trees in central Pennsylvania during the same months (Fredericksen et al., 1995). With the exception of 1996, average cumulative uptake of 23 mmol m−2 by overstorey P. serotina trees at both sites was somewhat higher than seasonal uptake (16 mmol m−2) reported for trees in Pennsylvania because of higher gH2O in June at the southern sites. Based on 7 h averages, ozone exposures were similar to those reported to induce visible adaxial foliar stipple in P. serotina forest trees (Lee et al., 1999). Visible injury on P. serotina leaves was observed late in the growing season at both sites, but stipple was inconsistent within tree canopies and between years. Because the relationship between visible foliar injury and physiological function is still tenuous, it is unclear if the ambient ozone concentration was high enough to induce stress in overstorey P. serotina.

Because of lower gH2O in forest understorey saplings than in potted saplings, cumulative ozone uptake by understorey saplings of all three species over the 4 yr was below levels that induced detectable reductions in net photosynthesis and biomass accumulation in fumigation studies with potted, well watered and fertilized P. serotina, A. rubrum and Q. rubra seedlings grown under a 75% reduction in ambient light (Samuelson, 1994a,1994b). However, dose–response functions estimated for seedlings grown under optimal experimental conditions may underestimate ozone damage to understorey trees because of the low ratio of photosynthesis to ozone uptake in more light-limited forest environments (Fredericksen et al., 1996; Volin et al., 1998).

In summary, based on observations from both case studies, we suggest that controlled studies of open-grown seedlings have limited application to understorey and overstorey forest trees. Foliar N concentration expressed on a leaf area basis is a useful mechanistic linkage in predicting the large variation in average gH2O between canopy locations. Less uncertainty was associated with the prediction of within-season and between-season variability in gH2O of juvenile understorey trees than of the upper canopy of overstorey trees. Greater variability in gH2O and weaker relationships between gH2O and leaf microclimate indicate that prediction of ozone uptake in overstorey trees requires a greater understanding of factors controlling gH2O in large forest trees.

IV. Evidence from a scaling exercise

To determine how applicable our results from the Q. rubra case study were to other species, and to stimulate further discussion on scaling issues, we conducted a scaling exercise by summarizing results from direct experimental comparison of gH2O between younger and older trees, and by examining data from previously published studies on seasonal average gH2O for small and large trees of 13 species (Table 4).

Table 4.  Summary of direct experimental or literature-based comparison of stomatal conductance to water vapor, gH2O (mmol m−2 s−1), between younger and older trees
SpeciesAge (yr)gH2OCommentsReference
  1. In situ denotes seedlings or saplings planted in the ground or naturally regenerated.*Conversion from cm s−1 (Farquhar et al., 1978). Rates of gH2O for Pinus are expressed on a total area basis; rates for Picea and Pseudotsuga are expressed on a projected area basis.

Experimental comparison    
Prunus serotina2250In situ, planted, 1 season Fredericksen et al. (1995)
 80233Upper crown of 21 m forest trees, 1 season 
Quercus rubra2120Potted seedlings, open-top chambers, 2 seasons; upper crown of orchard trees, open-top chambers, 2 seasonsHanson et al. (1994)
 177Shaded seedlings in tree crown, late season measures (Aug–Sep), open-top chambers; lower crown of orchard trees, shade foliage, late season measures (Aug–Sep), open-top chamberSamuelson (1994a)
Picea rubensSeedlings1.13 seasons, current year foliage, rates expressed in mol kg−1 s−1, forest seedlings, saplings and trees at 2 mountain sites, upper canopy of trees Thornton et al. (1994)
 Juvenile260Grafted mature and juvenile scions, potted, open-top chamber, 1 season, current year foliageRebbeck et al. (1993)
Pinus ponderosa5440Middle canopy of 10 m trees, 1 d midday average, 1 yr foliageYoder et al. (1994)
 22925Middle canopy 32 m trees, 1 d midday average, 1 yr foliage 
 40120Upper canopy of 12 m trees, 3 d midday average, 1 yr foliageHubbard et al. (1999)
 23060Upper canopy of 33 m trees, 3 d midday average, 1 yr foliage 
 370In situ, open-top chamber, 1 season, 1 yr foliageMomen et al. (1997)
 1262Upper canopy of seed orchard trees, 1 season, 1 yr foliage 
Pinus radiata162Maximum gH2O of current year and 1 year foliage on one d in irrigated in situ seedlings and 6 m treesAttiwill et al. (1982)
Pinus taedaSeedlings120Midday comparison between in situ seedlings and trees on one d1989Cregg et al. (1989)
Pseudotsuga menziesii5–10 38Upper canopy of 56–65 m, old growth trees and open-grown saplings, 1 season, average of measures at predawn and noon, trees and saplings not measured on the same dBauerle et al. (1999)
Sequoiadendron giganteum2200Naturally regenerated stands, current year foliage, seasonal maximaGrulke & Miller (1994)
Literature comparison    
Acer rubrum3–4309Potted, outdoors, 1 seasonCroker et al. (1998)
 Juvenile152In situ, 2 m tall, 1 seasonKubiske et al. (1996)
 Mature80*Upper canopy of 15 m trees, 3 seasonsJurik (1986)
 Mature125Upper canopy of 18 m trees, 2 seasonsSamuelson & Kelly (1997)
Acer saccharum146Potted, open-top chamber, 1 seasonRebbeck & Loats (1997)
 2–3131In situ, 2 seasonsKruger & Reich (1997)
 Juvenile70In situ, 0.4–1 m tall, 1 seasonEllsworth & Reich (1992)
 35117Upper canopy of 15 m trees, 1 seasonTjoelker et al. (1995)
Fagus sylvatica2145Potted, Solardomes, 2 seasonsHeath and Kerstiens (1997)
 2150Potted, open-top chamber, 1 seasonTaylor & Dobson (1989)
 9217In situ, open-top chamber, 1 seasonDixon et al. (1998)
 59200Upper canopy of 21 m trees, 1 seasonRoberts & Rosier (1994)
Prunus serotina1150Potted, CSTR, measured over a 3 wk periodLoats & Rebbeck (1999)
 Juvenile400In situ, 1 seasonHarrington et al. (1989)
 Juvenile381In situ, 1 seasonAbrams & Mostoller (1995)
 Juvenile280In situ, 2 m tall, 1 seasonKubiske et al. (1996)
 Mature306Upper canopy of 23 m trees, 2 seasonsSamuelson & Kelly (1997)
Quercus alba1–2106*Potted, CSTR, 2 seasonsFoster et al. (1990)
 1–3160In situ, open-top chamber, 3 seasonsGunderson et al. (1993)
 Mature240*Upper canopy of 20 m trees, 2 seasonsLoewenstein & Pallardy (1998)
Quercus robur2183Potted, Solardomes, 1 seasonHeath & Kersteins (1997)
 31243Upper canopy of 16 m trees, 1 season, irrigated treesEpron & Dreyer (1993)
Quercus rubra1115Potted, greenhouse, 1 seasonKleiner et al. (1992)
 1–2146*In situ, 2 seasonsCrunkilton et al. (1992)
 Juvenile275In situ, 1 m tall, 1 seasonKruger & Reich (1993)
 1–2368In situ, 2 seasonsKruger & Reich (1997)
 Mature120*Upper canopy of 15 m trees, 3 seasonsJurik (1986)
 Mature225Upper canopy of 7 m trees, 1 season average of daily maximaReich & Hinckley (1989)
 Mature190Upper canopy of 18 m trees, 2 seasonsSamuelson & Kelly (1997)
Picea abies1200Potted, open-top chamber, 1 season, current year foliageKarlsson et al. (1997)
 4116Potted, growth chambers, 1 season, current year foliageKronfuβet al. (1998)
 5107Potted, open-top chamber, 2 seasons, current year foliageKarlsson et al. (1995)
 7166In situ, open-top chamber, 1 season, current year foliageWallin et al. (1992)
 357810–14 m trees, 1 season, current year foliageZimmerman et al. (1988)
 60–651043 seasons, current year foliageWieser & Havranek (1994)
Picea sitchensis3205Potted, outdoors, 1 season, current year foliageJackson et al. (1995a)
 161402 seasons, current year foliageBarton et al. (1993)
Pinus ponderosa393*Potted, open-top chamber, 1 season, 1 yr foliageTakemoto et al. (1997)
 14–2085*1 season, 5–10 m trees, 1 yr foliage Coyne & Bingham (1982)
 25601 season, one year foliage, high soil moistureCregg (1993)
Pinus sylvestris3187Potted, grown outside, 1 season, current year foliageJackson et al. (1995b)
 4062Upper canopy of 14 m trees, 2 seasons, current year foliageJackson et al. (1995a)
 Mature55Upper canopy of 17 m trees, 1 season, current year foliageBeadle et al. (1985)
Pinus taeda197Potted, greenhouse, 1 season, current year foliageSamuelson (2000)
 2–482In situ, open-top chamber, 3 seasons, current year foliageTissue et al. (1997)
 21001 season, current year foliageSamuelson (1998)
 9114Midcanopy, 1 season, current year foliage, irrigatedMurthy et al. (1997)
 1095Upper canopy of 10 m trees, thinned plantation, 1 season, current year foliageGinn et al. (1991)
 1274Upper canopy, 1 season, current year foliageGravatt et al. (1997)
 13105Upper canopy, 1 season, current year foliageTang et al. (1999)
Pseudotsuga menziesii299Potted seedlings from 25 populations, 1 seasonZhang & Marshall (1995)
 15936 m seed orchard trees from 25 populations, 1 seasonZhang et al. (1993)

The selection of species for the literature scaling exercise was based on the availability of physiological information and a species’ importance either commercially or ecologically. Cultural activities, genetic material, sampling methodologies and plant environment varied by study, and the number of studies varied by species and age. Average gH2O was calculated for open-grown seedlings and saplings in pots, planted directly in the ground or naturally regenerated (in situ), and sun leaves of mature trees. Specific requirements for the use of published data in calculating a seasonal average included specification of foliage age; indication as to whether gH2O was expressed on a projected or total area leaf area basis (conifers); seedlings and saplings at least 1-yr-old and  2 m in height; mature trees at least 7 m in height; measurement of gH2O at midday at saturating PPFR through at least one full growing season; no evidence of moisture stress; no rehydration of cut branches; and growth of potted seedlings near ambient PPFR. An exception to the above criteria was the study by Loats and Rebbeck (1999), which measured gH2O of P. serotina over a 3-wk period in the middle of the growing season.

With the exception of A. rubrum, average seasonal gH2O calculated from the literature was 25–61% lower in potted juvenile trees of angiosperm species relative to mature trees (Fig. 10). In the Q. rubra case study, average seasonal gH2O was 60% lower in sun leaves of potted seedlings compared with mature trees (Hanson et al., 1994). Similarly, Bassow and Bazzaz (1994) observed a three-fold difference in the decrease in light-saturated net photosynthesis of potted vs. mature Q. rubra. Based on the physiology of potted seedlings, one might conclude that gH2O of angiosperm species is lower in juvenile than in mature trees. With the exclusion of A. rubrum, when the literature-based average gH2O of seedlings grown in situ was compared with mature trees, differences in gH2O between tree sizes were either diminished or, as in the case of Q. rubra and Fagus sylvatica, shifted to higher gH2O in seedlings than in mature trees. It is interesting to note that in the study of potted A. rubrum, which reported high rates of gH2O, individuals were planted in 22-l pots and were well watered to maintain a high soil matric potential. This analysis offers compelling evidence that size related variation in gH2O may be confounded by pot restrictions on root growth that may result in imbalances between sources and sinks (Arp, 1991) or soil water limitations. Rapid soil drying typical in potted plants may reduce gH2O either with (Will & Teskey, 1997) or without (Croker et al., 1998) concomitant changes in leaf water potential.

Figure 10.

Direct experimental (D) or literature-based comparisons of average stomatal conductance to water vapor (gH2O) measured at light saturation over at least one growing season in leaves of open-grown seedlings and the upper canopy of mature trees. In situ seedlings were planted or naturally regenerated. Values for Pinus are expressed on a total leaf area basis. Values for Pseudotsuga and Picea are expressed on a projected area basis. Values for the first P. rubens comparison are in cmol g−1 s−1. All rates were based on current-year foliage, except those for P. ponderosa which were based on 1-yr-old foliage. Hatched columns, juvenile : pot; open columns, juvenile : in situ; closed columns, mature tree.

Experiments that directly compared gH2O of between in situ juvenile trees and mature trees of Pinus ponderosa, Pinus radiata, Pinus taeda, Pseudotsuga menziesii, Picea rubens and Sequoiadendron giganteum reported rates of gH2O that were as much 40% lower in older trees. Average seasonal gH2O of potted seedlings calculated from published studies was 26% greater than gH2O of mature trees, and variation in gH2O between potted and in situ juvenile trees was less than in angiosperm species. Work by Rebbeck et al. (1993) and Grulke and Miller (1994) established that greater ozone sensitivity in juvenile than mature P. rubens and S. giganteum was primarily a function of gH2O. However, because of variation in shoot phenology of Pinus species between juvenile and mature trees, the degree to which higher ozone uptake rates in seedling leaves are offset by the production of new foliage is unclear (Clark et al., 1995). Based on results from a TREGRO simulation of ozone response in mature P. taeda, Constable and Retzlaff (1997) concluded that sensitivity of leaf C gain to peak ozone episodes was also dependent on shoot phenology.

The potential influence of pot restrictions raises questions concerning the relevance of applying size-related differences in gH2O and ozone response in the Q. rubra case study to other species. It must be kept in mind that the majority of ozone studies have been conducted with potted seedlings, and evidence of size-related variation in the gH2O of Q. rubra indicates that physiology and ozone responses are different between mature trees and potted seedlings. In addition, because Q. rubra oak seedlings used in the case study were transplanted into increasingly larger pots each season to prevent pot binding, the degree to which roots were restricted in growth or had limited water access is unclear.

In coniferous species, gH2O appears to be lower in mature trees than in situ seedlings and saplings, but the magnitude of difference in gH2O between small and large coniferous trees may vary diurnally, which is important in understanding the interaction between ozone dynamics and gH2O (Cregg et al., 1989; Yoder et al., 1994). While maturation may explain, to some extent, higher gH2O in younger tissue as seen in the P. rubens grafting study (Table 4), hydraulic limitation due to height, branching complexity and root hydraulic resistance may to some degree account for size-related variation in gH2O of conifers (Yoder et al., 1994; Momen et al., 1997; Bauerle et al., 1999; Hubbard et al., 1999; Kolb & Stone, 2000).

Results from the Q. rubra case study and scaling exercise indicate that studies of ozone response in Q. rubra, Quercus alba, Quercus robur, P. serotina, Acersaccharum and Fagus sylvatica, using potted seedlings, may underestimate ozone uptake by mature trees on average by 47%. In contrast, studies of potted conifer seedling may overestimate ozone uptake of mature conifers on average by 26%. The scaling exercise leads us to conclude that the confounding influence of pot effects greatly complicates scaling of gH2O and ozone response from seedling studies to forest trees. Dose–response functions derived from studies of seedlings grown in situ are more applicable to mature trees than studies with potted seedlings.

V. Concluding remarks

The influence of ozone on individual trees is a complex process that can vary in response to a host of environmental factors. Understanding of physiological age or size-based variation in gH2O of forest trees is additionally confounded by seedling/sapling studies that examined trees grown in pots, and single-factor experiments that did not address the interaction of ozone with other environmental factors such as light, nutrition and water stress. If in situ rather than potted seedlings are compared with mature trees, we concur with Kolb et al. (1997), who concluded that physiological response to ozone in small vs. large trees could be predicted by ranking avoidance of ozone uptake, and that for most species large trees have lower gH2O than seedlings, when differences in height exceed 4 m. We offer compelling evidence that gH2O rates derived from studies on potted seedlings may underestimate ozone uptake by large hardwood trees and overestimate ozone uptake by large coniferous species. We have demonstrated that in scaling ozone response from potted seedlings to forest trees, pot size and water availability should be considered when evaluating seedling studies.

Our work with forest trees indicates that the most significant environmental factor influencing seasonal variability in ozone uptake of large P. serotina and Q. rubra forest trees was the availability of soil water. The compensatory response to ozone observed in the Q. rubra case study illustrates that ozone exposure of sufficient magnitude may reduce fine root production and predispose large Q. rubra forest trees to drought. The substantial overestimation of ozone effects on simulated C allocation to stem, branch and fine root growth by TREGRO emphasizes our lack of understanding of whole-tree C dynamics, and raises the question as to whether process models are accurately parameterized to predict mature tree responses to stress.

So how might we go forward in ecological risk assessment for ozone, given that we have issued the ‘sobering warning that information is still missing’? Large uncertainties exist in: ozone exposure dynamics over forested areas; ozone effects on different life stages of trees and chronic effects in long-lived species; definition of population; and processes controlling the system at larger scales. Despite these uncertainties, we believe that gH2O is an underlying parameter that will facilitate extrapolation of ozone sensitivity of forest trees across temporal and spatial scales by way of process based modelling and probabilistic approaches. Model-based upscaling of gH2O and ozone uptake from the leaf to the canopy is under development (Baldocchi, 1993; Jarvis, 1995). For example, big leaf and layered models have been used to estimate hourly ozone uptake by a mixed deciduous oak–maple stand (Amthor et al., 1994), and to simulate the interactive effects of ozone uptake and drought on annual net primary production of hardwood forests in the north-eastern USA (Ollinger et al., 1997). Deposition models have been used to predict ozone fluxes above a canopy (Walton et al., 1997) and are being developed to consider dose response and plant defences (Massman et al., 2000). A probabilistic modelling approach has been used to estimate both uncertainty and the effects of climate change and ozone on growth of P. taeda on a regional basis (Woodbury et al., 1998). Assessment of ozone effects on forest resources and ecosystems requires continued development of predictive models that include definition of alternative outcomes and expression of uncertainties, as well as research focus on cause–effect relationships at different scales.

Given the formidable costs of free-air and mature-tree fumigation, and the number of replications needed to detect small biological changes over short-term experimentation, we recommend that research on cause–effect relationships in forest trees should focus on the rigorous statistical and monitoring protocol developed by Schreuder & Thomas (1991). Dose–response relationships could be developed for ozone-sensitive variables such as stem circumference (McLaughlin & Downing, 1995); leaf area retention (Miller et al., 1997); carbohydrate partitioning in leaf and root tissues (Samuelson & Kelly, 1996); and fine root production (Kelting et al., 1995) from long-term observational studies, if both cause and effect variables are measured accurately, estimated properly, and based on a sufficiently large sample size (Schreuder & Thomas, 1991). This approach may provide the missing information on the processes that we now know are important to a more realistic representation and projection of ozone effects on forest trees.


We thank Dr Art Chappelka and Dr John Laurence for providing comments on earlier versions of the manuscript. Funding for the case study projects described in this review was provided by the Tennessee Valley Authority, the Electric Power Research Institute, and the US Environmental Protection Agency. The authors express their appreciation to Dr Robert Goldstein and Dr William Hogsett for their vision and insight in providing the financial support for these projects. The authors also acknowledge with greatest gratitude the dedicated contributions over the past decade of Alan Mays, Jason Scarbrough, and Cassandra Wylie. This review is a joint project of the Alabama and Iowa Agricultural Experiment Stations, and was supported by McIntire-Stennis as well as State of Alabama and Iowa funds. Journal Paper No. J-18933 of the Iowa Agriculture and Home Economics Experiment Stations.