O3 impacts on plant development: a meta-analysis of root/shoot allocation and growth

Authors

  • D. A. GRANTZ,

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

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

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

David A. Grantz. Fax: +1 559 646 6593; e-mail: dgrantz@uckac.edu

ABSTRACT

The mechanism of O3 action on plants remains poorly characterized. Symptoms include visible lesions on the leaf surface, reduced growth and a hypothesized reduction in allocation of carbohydrate to roots. The generality of this latter phenomenon has not been demonstrated. Here, a meta-analysis is performed of all available experimental data, to test the hypotheses that O3 exposure of the shoot inhibits biomass allocation below ground (the root/shoot allometric coefficient, k) and inhibits whole-plant growth rate [relative growth rate (RGR)]. Both k and RGR were significantly reduced by O3 (5.6 and 8.2%, respectively). Variability in k was greater than in RGR, and both exhibited some positive as well as mostly negative responses. The effects on k were distinct from the effects on RGR. In some cases, k was reduced while RGR was unaffected. Slow-growing plants (small RGR) exhibited the largest declines in k. These observations may have mechanistic implications regarding O3 phytotoxicity. There were no effects of type of exposure chamber on sensitivity to O3. The analyses indicate that the O3 inhibition of allocation to roots is real and general, but variable. Further experiments are needed for under-represented plant groups, to characterize exceptions to this generalization and to evaluate O3–environment interactions.

INTRODUCTION

Current concentrations of O3 in the troposphere pose significant risks to vegetation (Heck, Taylor & Tingey 1988; Lefohn 1992; Fowler et al. 1999; Emberson, Ashmore & Murray 2003). The regional distribution of O3 makes it the most damaging air pollutant globally, for both agricultural and native plant populations. Despite substantial investment in regulatory initiatives and mitigation technologies, O3 exceeds phytotoxic levels in many rural and urban areas, and in both developed and developing countries (Fowler et al. 1999). While O3 is declining in some urban areas, tropospheric O3 is still increasing globally and is projected to do so by up to 1% year−1 for the next half century (Thompson 1992; Stevenson et al. 2000).

Acute O3 exposure (i.e. to relatively high O3) typically leads to visible symptoms, including bronze, deep purple or brown lesions on the adaxial surfaces of exposed leaves (Sandermann 1996; Flagler 1998), as well as to other biological impacts. Chronic exposure to lower concentrations may not lead to visible symptoms, but induces other responses that span the range of scales and levels of biological organization that have been examined. These include changes in gene expression (Sharma & Davis 1994; Miller & McBride 1999; Grimmig et al. 2003), altered growth rates (Laurence et al. 1994), degraded ecosystem function and reduced gene frequency and biodiversity (Taylor, Pitelka & Clegg 1991; Miller & McBride 1999). Visible symptoms have not provided a reliable predictor of these O3 impacts. In loblolly pine (Pinus taeda), for example, independent segregation was observed for O3 tolerance expressed as visible injury or as growth reduction (Taylor 1994) with little correlation between the populations.

Despite decades of investigation (Long & Naidu 2002), neither the biochemical nor the physiological target of O3 attack, the mechanism of ensuing O3 phytotoxicity, nor the key components of plant resistance to O3, have been adequately characterized. This has impeded efforts to predict or model O3 impacts (Massman 2004) and to protect vulnerable crop plants and native vegetation.

Early research into the mechanism of O3 impacts on plants demonstrated that membranes were vulnerable to oxidant attack, with emphasis on plasmalemma and chloroplast membranes and implications for impaired photosynthetic carbon acquisition (Heath 1980). During the same period, evidence accumulated that exposure of foliage to O3 resulted in an accumulation of carbohydrate in the source leaves and reduced translocation to distant sinks (e.g. Hanson & Stewart 1970; Koziol & Jordan 1978). More recently, a compartmental efflux analysis of cotton (Gossypium barbadense) leaves following an acute exposure to O3 indicated that transfer of sugars from the cytoplasm into the phloem was interrupted, providing a potential mechanism for these early observations (Grantz & Farrar 2000).

Reduced allocation of carbohydrate to roots is widely reported, following both experimental exposure to O3 (Oshima, Bennett & Braegelmann 1978; Kasana & Mansfield 1986; Cooley & Manning 1987; Miller 1988; Kostka-Rick & Manning 1992; Laurence et al. 1994; Pell et al. 1994) and across natural exposure gradients (Taylor & Davies 1990; Grulke & Balduman 1999). However, a disturbing level of variability is observed in available literature, even between different O3 exposure experiments using the same plant species. Much of this variability may be related to contrasting environmental conditions and genetic diversity, and some to experimental protocols. Characterizing these effects remains an important goal of future research. A few studies (particularly Reiling & Davison 1992a; Davison & Barnes 1998) have directed attention to reports of unchanged or even increased allocation of carbohydrate to roots following foliar exposure to O3. At present, the generality of O3 impacts on allocation below the ground remains unconfirmed, and is subject to further testing. The physiological and ecological implications of such a general O3 impact are substantial, with significant public policy ramifications. These include quantitative evaluation of carbon sequestration below ground in natural and managed ecosystems, growth and yield forecasts in agricultural and forested systems, predictions of the trajectories of endangered species and development of prescribed management practices for weeds and invasive species that are consistent with changing competitive interactions.

Many reports indicate that O3 exposure alters a specific measure of allocation, the root/shoot biomass ratio, R/S (Wang, Karnosky & Bormann 1986; Cooley & Manning 1987; Chappelka & Chevone 1988; Chappelka, Chevone & Burk 1988; Darrall 1989; Reinert & Ho 1995; Grantz & Yang 1996; Reinert et al. 1996; Dickson et al. 1997; Landolt et al. 1997, 2000; Olszyk & Wise 1997; Chappelka & Samuelson 1998; Franzaring et al. 2000). Unfortunately, R/S and similar allometric relationships such as leaf area per plant mass [leaf area ratio (LAR)] change predictably with plant development (i.e. ontogenetic drift) (Farrar & Gunn 1996; den Hartog et al. 1996) as well as with altered aerial or edaphic environmental conditions (van Noordwijk et al. 1998). O3 may alter the rate of plant development. Thus, accurately determined differences in R/S, sampled at discrete time points, may reflect different points on the growth curves of O3-treated and control plants. This represents a rescaling of the time dimension (van Noordwijk et al. 1998), rather than an altered biological programme of biomass allocation. Exposure of plants to elevated CO2, for example, altered R/S at synchronous harvests without altering allocation (Farrar & Gunn 1996; Poorter & Nagel 2000). A significant treatment effect on R/S neither establishes nor precludes a significant effect on allocation.

The ambiguity of R/S as a measure of allocation may be overcome experimentally in a number of ways (van Noordwijk et al. 1998). Sampling of plants exposed to a range of treatments can be conducted at equivalent developmental stages (i.e. at the same plant size or biomass), rather than at the same plant age (i.e. time point or period after sowing) (Poorter & van der Werf 1998). This approach is less effective if the treatment causes substantial changes in biomass allocation as well as in biomass accumulation.

A more general alternative is offered by an evaluation of an allometric coefficient (k; after Troughton 1955), derived from the exponential growth equation. The planned contrast of k between two environmental treatments explicitly compares the relative growth rates (RGRs) of competing plant parts (Farrar & Gunn 1998; Gunn, Bailey & Farrar 1999). Unlike R/S, allometric coefficients such as k exhibit a nearly constant magnitude over the period of rapid development in plants exposed to unperturbed environments (Troughton 1955; Farrar & Gunn 1998; Gunn et al. 1999; Gunn & Farrar 1999). This enhances the diagnostic value, relaxes the rigid temporal constraints on measurements of biomass ratios and normalizes for treatment differences in growth rates caused by environmental perturbations such as exposure to O3. On a limited basis, an allometric analysis has been applied to the problem of O3 impacts (tabulated data in Table 1). While O3 has been reported to alter R/S in many studies (Cooley & Manning 1987), changes in k have been documented in only a subset (Reiling & Davison 1992a). A rigorous evaluation of the O3 impact on allocation between roots and shoots is required.

Table 1.  Details of the O3 exposures in the original studies providing data for the meta-analysis
PlantStage of growth at initial exposureGrowth/exposure conditions O3Exposure systemLength of exposureReference
ControlElevated
  • a

    k obtained from tabulated data.

  • b

    k calculated from data presented.

  • c

    Four independent observations of k were obtained from the two clones investigated.

  • d

    One independent observation of k was obtained from the five cultivars investigated.

  • CF, charcoal-filtered air; CEC, controlled environment chamber; GH, glasshouse; OTC, open-top field chambers.

Thirty-one speciesCotyledon/first leaf stage< 5 p.p.b.70 p.p.b.CEC7 h d−1 for 2 weeksReiling & Davison (1992a)a
Five species4 weeks16 p.p.b.71 p.p.b.CEC7 h d−1/5 d week−1 for 21 dWarwick & Taylor (1995)a
Three species5 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 43 dGimeno et al. (2004)a
Three speciesSeedlings55 p.p.b.100 p.p.b.OTCMaximum 4 h d−1/5 months year−1 for 3 yearsBroadmeadow & Jackson (2000)a
Anthyllis cornicina L.5 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 76 dGimeno et al. (2004)a
Aegilops geniculata Roth5 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 64 dGimeno et al. (2004)a
Aegilops triuncialis L.5 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 65 dGimeno et al. (2004)a
Avena sterilis L.4 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 49 dGimeno et al. (2004)a
Biserrula pelecinus L.7 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 67 dGimeno et al. (2004)a
Briza maxima L.7 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 63 dGimeno et al. (2004)a
Bromus hordeaceus L.5 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 69 dGimeno et al. (2004)a
Brassica napus ssp. oleifera var. biennis (oilseed rape), five cultivarsSeedlingsCFCF + 75 p.p.b.CEC6.5 h d−1 for 16 dOllerenshaw, Lyons & Barnes (1999)a
Calluna vulgaris (summer growth)CuttingsCF70 p.p.b.OTC8 h d−1, 5 d week−1 for 24 weeksFoot et al. (1996)a
Citrullus lanatus (watermelon), two cultivars22-day-old< 8 p.p.b.70 p.p.b.CEC6 h d−1 for 21 dFernandez–Bayon et al. (1993)a
Cucumis melo (muskmelon), two cultivars22-day-old< 8 p.p.b.70 p.p.b.CEC6 h d−1 for 21 dFernandez–Bayon et al. (1993)a
Cynosurus echinatus L.6 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 68 dGimeno et al. (2004)a
Eucalyptus globulusSeedlings4.6 p.p.b.52.3 p.p.b.CEC7 h d−1 for 37 dPearson (1995)b
Medicago sativa (alfalfa)4 weeksCF60 p.p.b.GH6 h d−1/5 d week−1 for 56 dCooley & Manning (1988)b
Ornithopus compressus L.5 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 47 dGimeno et al. (2004)a
Picea abies (Norway spruce), two clonesSeedlingsCFNFOTC24 h d−1 for 106 dKarlsson et al. (1997)b,c
Plantago major plants from 22 sites6-day-old seedlings< 5 p.p.b.70 p.p.b.CEC7 h d−1 for 14 dLyons, Barnes & Davison (1997)a
P. major6-day-old seedlings< 5 p.p.b.70 p.p.b.CEC7 h d−1 for 14 dLyons & Barnes (1998)b
P. major, 28 British populationsCotyledon< 10 p.p.b.70 p.p.b.CEC7 h d−1 for 14 dReiling & Davison (1992b)a
Triticum aestivum (spring wheat)8 d after emergence< 5 p.p.b.75 p.p.b.CEC7 h d−1 for 30 dBalaguer et al. (1995)b
T. aestivum, two cultivars of spring wheat; three cultivars of winter wheatTwo leaf stage< 5 p.p.b.75 p.p.b.CECMaximum 4 h d−1 for 41 dBarnes, Ollerenshaw & Whitfield (1995)a,d
T. aestivum L. cv. Giza 6325 d after sowingCF50 p.p.b.OTC10 h d−1/5 d week−1 for 69 dHassan (2004)a
Trifolium cherleri L.5 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 62 dGimeno et al. (2004)a
Trifolium glomeratum L.5 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 61 dGimeno et al. (2004)a
Trifolium striatum L.6 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 54 dGimeno et al. (2004)a
Trifolium subterraneum L.5 weeksCF40 p.p.b.OTC10 h d−1/5 d week−1 for 66 dGimeno et al. (2004)a

Here, the allocation coefficient, k, and meta-analytical techniques (e.g. Curtis & Wang 1998) are used to evaluate the mean (magnitude) and uncertainty (significance) of O3 impacts on allocation, using information obtained from all available experimental observations. Meta-analytical techniques are becoming frequently used in ecological analyses, most notably for synthesis of the impacts of elevated atmospheric CO2 (e.g. Curtis & Wang 1998; Hedges, Gurevich & Curtis 1999; Wand et al. 1999; Kerstiens 2001; Medlyn et al. 2001; Ainsworth et al. 2002; Wang & Curtis 2002), including a meta-analysis of the impact of elevated CO2 on allocation (Poorter & Nagel 2000). Recently, meta-analytical approaches have been applied to O3 impacts (e.g. on photosynthetic processes in the single species Glycine max) (Morgan, Ainsworth & Long 2003) and on growth responses of tree species (Wittig et al. 2005). The present study provides the first aggregated analysis of available literature on O3 impacts on allocation. The extent to which generalizations regarding O3 impacts on k may be made and the reasons that such generalizations remain problematic are evaluated. A contrast of O3 impacts on k with O3 impacts on plant RGR is provided, as these parameters similarly integrate many O3 effects and similarly facilitate a comparison of O3 effects across diverse species, growth habits, environments and sampling periods. The effects of O3 on k and RGR, documented here by meta-analysis, are contrasted with the effects of shading on k and RGR, from the analysis of Hunt & Cornelissen (1997). Finally, within the constraints of the limited data, O3 effects are contrasted between broad groups of plants and between diverse exposure technologies, to test for effects on the sensitivity of k or RGR to O3.

METHODS

Meta-analysis

All relevant literature known to the authors was evaluated, incorporating the full range of plant species, experimental designs and exposure protocols encountered. The review of the literature was conducted in 2001 using Current Contents and updated in late 2005, using Web of Science. All published accounts of O3 exposure experiments were selected, if they provided graphical or tabular values of k (the root/shoot allometric coefficient) or provided sufficient data to allow the calculation of k. This yielded 16 published studies (Tables 1 & 2), many with multiple species. These studies provided 125 independent experimental observations, comprising 60 species, that were suitable for this analysis. The O3 effects on plant growth (RGR) were analysed using a subset (114) of these observations, previously selected on the basis of k, that provided whole-plant RGR or sufficient data from which RGR could be determined. No further screening was imposed. Specifically, no consideration was given to the results of O3 exposure on k nor RGR, whether positive or negative, statistically significant or non-significant.

Table 2.  Plant species in the available literature, with abbreviations used in the figures, and references to the original studies providing data for the meta-analysis
CodeBinomialReference
  • a

    Calculated values.

Herbaceous dicotyledonous species
 acAnthyllis cornicina L.Gimeno et al. (2004)
 avAnthyllis vulnerariaWarwick & Taylor (1995)
 bnBrassica napus ssp. oleiferaOllerenshaw et al. (1999)
 bpBiserrula pelecinus L.Gimeno et al. (2004)
 c.Chenopodium albumReiling & Davison (1992a)
 cacCirsium acauleWarwick & Taylor (1995)
 cfCerastium fontanumReiling & Davison (1992a)
 clCitrullus lanatusFernandez-Bayon et al. (1993)
 cmCucumis meloFernandez-Bayon et al. (1993)
 cvCalluna vulgaris (summer growth)Foot et al. (1996)
  d.f.Deschampsia flexuosaReiling & Davison (1992a)
 ehEpilobium hirsutumReiling & Davison (1992a)
 loLotus corniculatusWarwick & Taylor (1995)
 mmMedicago minima (L.) Bartal.Gimeno et al. (2004)
 msMedicago sativaCooley & Manning (1988)a
 ntNicotiana tabacum Bel W3Reiling & Davison (1992a)
 ocOrnithopus compressus L.Gimeno et al. (2004)
 pcPlantago coronopusReiling & Davison (1992a)
 plPlantago lanceolataReiling & Davison (1992a)
 pmPlantago majorReiling & Davison (1992a)
Lyons & Barnes (1998)a
Reiling & Davison (1992b)
Lyons et al. (1997)
 pmaPlantago maritimaReiling & Davison (1992a)
 pmePlantago mediaReiling & Davison (1992a)
 poPilosella officinarumWarwick & Taylor (1995)
 psPisum sativumReiling & Davison (1992a)
 raRumex acetosaReiling & Davison (1992a)
 racRumex acetosellaReiling & Davison (1992a)
 roRumex obtusifoliusReiling & Davison (1992a)
 tsTeucrium scorodoniaReiling & Davison (1992a)
 traTrifolium angustifolium L.Gimeno et al. (2004)
 trcTrifolium cherleri L.Gimeno et al. (2004)
 trgTrifolium glomeratum L.Gimeno et al. (2004)
 trsTrifolium striatum L.Gimeno et al. (2004)
 trsuTrifolium subterraneum L.Gimeno et al. (2004)
 udUrtica dioicaReiling & Davison (1992a)
Herbaceous monocotyledonous species
 agAegilops geniculata RothGimeno et al. (2004)
 atAegilops triuncialis L.Gimeno et al. (2004)
 aeArrhenatherum elatiusReiling & Davison (1992a)
 afAvena fatuaReiling & Davison (1992a)
 asAvena sterilis L.Gimeno et al. (2004)
 beBromus erectusReiling & Davison (1992a)
 bmBriza maxima L.Gimeno et al. (2004)
 bhBromus hordeaceus L.Gimeno et al. (2004)
 bpiBrachypodium pinnatumReiling & Davison (1992a)
 bsBromus sterilisReiling & Davison (1992a)
 ceCynosurus echinatus L.Gimeno et al. (2004)
 DrDesmazeria rigidaReiling & Davison (1992a)
 foFestuca ovinaWarwick & Taylor (1995)
Reiling & Davison (1992a)
 hlHolcus lanatusReiling & Davison (1992a)
 hmHordeum marinumReiling & Davison (1992a)
 kmKoeleria macranthaReiling & Davison (1992a)
 lpLolium perenneReiling & Davison (1992a)
 lrLolium rigidum GaudinGimeno et al. (2004)
 paPoa annuaReiling & Davison (1992a)
 ptPoa trivialisReiling & Davison (1992a)
 taTriticum aestivumBarnes et al. (1995)
Balaguer et al. (1995)a
Hassan (2004)
Tree species
 egEucalyptus globulusPearson (1995)
 feFraxinus excelsiorBroadmeadow & Jackson (2000)
 paPicea abiesKarlsson et al. (1997)a
 psPinus sylvestrisBroadmeadow & Jackson (2000)
 qpQuercus petraeaBroadmeadow & Jackson (2000)

These studies included a range of acute and chronic O3 exposure regimes, and no attempt was made to distinguish between these protocols. Most of the O3 exposures were of relatively short periods of a few weeks, and of relatively young plants. Exposures were generally to environmentally relevant concentrations of O3, usually below 100 p.p.b. Experiments with pulse applications of higher concentrations were not excluded. Table 1 documents the conditions under which O3 exposures were imposed in each study. Table 2 documents the studies included, the plant species involved and the abbreviations used for convenience in the figures.

A quantitative synthesis of the often disparate results was obtained using meta-analytical techniques (Curtis & Wang 1998; Gurevitch & Hedges 1999; Rosenberg, Adams & Gurevitch 2000). The approach used here was similar to that reported by Ainsworth et al. (2002) and Morgan et al. (2003). The O3 effects were quantified as response ratios (RRs) and transformed as the natural logarithm of the RR (LRR) (Curtis & Wang 1998; Hedges et al. 1999) to provide a near-normal distribution, centred on the true mean effect, that could be analysed using simple parametric statistics. A statistical analysis was performed using a mixed-model approach with a commercially available software package (MetaWin version 2.1.4; Sinauer Assoc., Sunderland, MA, USA) (Rosenberg et al. 2000). The following null hypotheses were tested:

  • 1[H1]: O3 does not alter k or RGR;
  • 2[H2]: O3 impacts do not differ between plant groups; and
  • 3[H3]: O3 impacts do not differ between exposure technologies.

Many publications provided insufficient detail to allow calculation of the dispersion of the original data (variance, SD or SE plus sample size). Therefore, resampling was used to develop a variance of the effect size. An unweighted meta-analysis was performed with significance determined from 95% confidence intervals (CIs), which were calculated using bootstrap methods (Adams, Gurevitch & Rosenberg 1997; Rosenberg et al. 2000). This approach is less powerful than a weighted analysis (Gurevitch & Hedges 1999), but is commonly required for ecological studies (Adams et al. 1997; Morgan et al. 2003). Significant effects were assigned at P ≤ 0.05 if CIs did not overlap. This is a conservative test of significance (Waltz 2006).

Response parameters

The RGR of the whole plant was defined as:

RGR = (ln B2 − ln B1)/(t2 − t1),(1)

where B1 and B2 are the shoot, root or whole-plant biomass measured at times t1 and t2. The magnitude of the O3 effect on whole-plant RGR was expressed as the LRR (Hedges et al. 1999; Rosenberg et al. 2000), as:

LRRRGR = ln(RGRozone/RGRcontrol),(2)

and the relative response of growth was expressed as the mean percentage change (PC) in RRRGR (PCRGR), as:

PCRGR = [(exp LRRRGR) − 1] × 100,(3)

with PCRGR calculated from mean LRRRGR determined using meta-analytical techniques from the aggregated data. In the figures, the analogous parameter is calculated for individual observations.

The root/shoot allocation coefficient (k) was defined as:

k = RGRroot/RGRshoot.(4)

The magnitude of the O3 effect on k was expressed as for LRRRGR as:

LRRk = ln(kozone/kcontrol),(5)

and the relative response of allocation was expressed as the mean PC in RRk (PCk) as:

PCk = [(exp LRRk) − 1] × 100,(6)

with PCk calculated from the mean LRRk determined using meta-analytical techniques from the aggregated data. In the figures, the analogous parameter is calculated for individual observations.

All observations were pooled, intentionally aggregating diverse species, fumigation protocols and exposure systems. As categorical differences were suggested by the data, the observations were classified into the broad plant groupings of herbaceous dicotyledonous and monocotyledonous species and tree species (dicots, monocots and trees, respectively) and into the broadly defined exposure systems of controlled environment (CE), glasshouse (GH) and open-top field exposure chambers (OTCs). The original observations were not equally distributed among these categories.

RESULTS

Root/shoot allocation

The meta-analysis of all available observations disproved the first null hypothesis [H1] for k. Exposure of shoots to O3 was associated with a general, significant reduction in allocation of biomass to roots. The mean RR, RRk, was reduced (P ≤ 0.05) by 5.6% (Table 3).

Table 3.  Results of the meta-analysis of O3 exposure on the root/shoot allometric coefficient (k) and on whole-plant relative growth rate (RGR), where k = RGRroot/RGRshoot and RGR = [(ln B2 − ln B1)/(t2 − t1)]
Plant groupParameterkRGR
  • Values are the mean effect on the response ratios (RRs) (LRRk, LRRRGR) with indication of statistical significance and confidence intervals (CIs) in ln transformed units, and the relative change and sample size in native units.

  • *

    Change caused by O3 exposure differs significantly from 0.0 at P < 0.05.

  • a

    Mean effect size [natural logarithm of the RR (LRR)] caused by O3 exposure.

  • b

    Bootstrap 95% CI of the mean effect size.

  • c

    Mean percentage change (PC) caused by O3 exposure (back transformed from the mean effect size as PCk, PCRGR).

  • d

    Number of independent published observations.

  • NS, not significant.

Herbaceous dicotyledonousEffect sizea−0.0313*−0.0859*
CIb(−0.0525, −0.0113)(−0.1017, −0.0708)
PC (%)c−3.1−8.2
nd9384
Herbaceous monocotyledonousEffect sizea−0.0851*−0.0458*
CIb(−0.1725, −0.0039)(−0.0679, −0.0273)b
PC (%)c−8.2−4.5
nd2422
TreeEffect sizea−0.2727, NS−0.1976*
CIb(−0.6172, 0.0605)(−0.3367, −0.0900)
PC (%)c−23.9−17.9
nd88
AllEffect sizea−0.0571*−0.0860*
CIb(−0.0913, −0.0256)(−0.1031, −0.0708)
PC (%)c−5.6−8.2
nd125114

There was considerable diversity among herbaceous dicots, monocots and trees. The second null hypothesis [H2] was also disproved for k, as the O3 effect in trees was significantly larger than that in dicots (Table 3). The effect in monocots did not differ from that in trees or dicots. RRk declined significantly in both dicots (3.1%) and monocots (8.2%) (Table 3). The effect was not significant in trees (23.9%), even though this was the largest of the observed effects, because of the smaller number and greater variability of observations in trees (Fig. 2) relative to the other plant groups.

Figure 2.

The effect of ozone exposure on the percentage change (PC) in the root/shoot allometric coefficient (k), (PCk), of herbaceous dicotyledonous (a), herbaceous monocotyledonous (b) and tree (c) species, where k = (RGRroot/RGRshoot) and PCk = [(kozone/kcontrol − 1) × 100], modified from Eqn 6 for individual observations. For key to species abbreviations and original references, see Table 2. The solid symbols indicate that the effect of ozone was statistically significant in the original publication.

Of the 125 observations from which O3 effects on k could be obtained, approximately 25% reported a significant effect in the original study (Tables 1 & 2). These were distributed fairly uniformly among the plant groups (22, 38 and 25% for dicots, monocots and trees, respectively) (Fig. 2; Table 4). Of these significant observations, 74% reflected a decline (Table 4) and only 6% reflected a significant increase (Table 4; Fig. 2, note the location of solid symbols).

Table 4.  The distribution of the two biological end points, the root/shoot allometric coefficient (k) and relative growth rate (RGR), following O3 exposure in the original studies providing data for the meta-analysis
End pointPlant groupSignificant observationsTotal O3 exposures
nSignificant effects (% total observations)Negative changes (% significant observations)nNegative changes (% total exposures)Positive changes (% total exposures)
  1. Data are presented for all observations and separately for each of the major plant groups. The number and percentage of total observations that led to significant changes and the percentage of these significant observations that were reductions in k or RGR, are shown along with the number of total observations and percentage of all observations that yielded positive or negative changes (remainder yielding no measurable change).

kDicotyledonous2022 80 93 5333
Monocotyledonous 938 56 24 5838
Tree 225100  8 7513
All3125 74125 5533
RGRDicotyledonous5363100 84 89 6
Monocotyledonous 418100 22 8614
Tree 0 0  8100 0
All5750100114 89 7

Over all observations of changes in k, whether significant or non-significant in the original studies, 55% yielded a decrease in k. These were also distributed fairly uniformly (53, 58 and 75% in the three plant groups) (Fig. 2; Table 4). About a third of all observations yielded an increase in k (33, 38 and 13%).

There were 60 species represented in these studies. A significant reduction in k was observed in 50% (44% of 25 dicots, 64% of 14 monocots and 40% of 5 trees; calculated by re-sorting the data in Table 4).

Plant growth

The meta-analysis of all observations also disproved [H1] for RGR, reflecting a significant O3-induced reduction in whole-plant RGR. The mean RR, RRRGR, declined by 8.2% (Table 3). [H2] was also disproved for RGR, as the O3 impact on growth rate was significantly less for monocots than for dicots or trees. The effect in dicots did not differ from that in trees. The O3-induced reduction of RGR was significant in each group of plants (8.2, 4.5 and 17.9% for dicots, monocots and trees, respectively).

O3 caused a significant change in RGR in 50% of the 114 original observations (Table 4). All significant changes were decreases (Fig. 1; Table 4). These significant observations were distributed highly non-uniformly among the plant groups. Herbaceous dicots exhibited significant declines in 63% of observations, compared with only 18 and 0% in monocots and trees, respectively (Fig. 1a–c; Table 4). None of the original observations in trees was able to resolve significant impacts of O3 on RGR (Fig. 1c; Table 4), even though the mean effect size was largest in trees.

Figure 1.

The effect of ozone exposure on the percentage change (PC) in whole-plant relative growth rate (RGR), (PCRGR), of herbaceous dicotyledonous (a), herbaceous monocotyledonous (b) and tree (c) species, where RGR = [(ln B2 − ln B1)/(t2 − t1)] and PCRGR = [(RGRozone/RGRcontrol − 1) × 100], modified from Eqn 5 for individual observations. For key to species abbreviations and original references, see Table 2. The solid symbols indicate that the effect of ozone was statistically significant in the original publication. In (c) the circle within the triangle indicates superimposed data points.

Over all observations of changes in RGR, combining those yielding significant and non-significant changes, 89% yielded a reduction in RGR (Table 4; Fig. 1a–c). These reductions were relatively uniformly distributed among the plant groups (89, 86 and 100%) in contrast to those observations that were significant. Only 7% of all observations yielded increases in RGR and all were non-significant (Fig. 1, note the location of solid symbols).

Of the 38 species represented in these observations of RGR, 50% exhibited significant reductions in RGR (79, 28 and 0% of dicots, monocots and trees, respectively; calculated by re-sorting the data in Table 4).

Growth and allocation

O3 impacts on growth were larger and less variable than O3 impacts on allocation (Table 3). This is apparent in the greater clustering near the origin of relative change data for k than for RGR (cf. Fig. 3a & b).

Figure 3.

Frequency analysis of the distribution of the effects of ozone exposure on the percentage change (PC) (as in Figs 1 & 2) in relative growth rate (RGR) (a) and the root/shoot allometric coefficient (k) (b).

Significant responses were more than twice as frequently observed for RGR than for k in the original studies (50 versus 25% of observations; Table 4). All significant observations of RGR were reductions, whereas about three-fourths of significant observations of k were reductions. Similarly, 89% of all observations of RGR were reductions, compared to 55% for k. Many observations of changes in k were small, with 13% exhibiting an effect of less than ± 0.03 in LRRk (chosen as ± 3% of the range of effects). In contrast, approximately 25% of observed O3 impacts on RGR were within ± 3% of the range of observed effects. In both cases, these responses are arguably too small to be of biological significance for either RGR or k. While the responses of both parameters were convincingly negative, the impacts of O3 on RGR were much clearer than the impacts on k.

The sensitivity of allocation to O3 (PCk) was significantly and positively related (P = 0.02; Fig. 4) to plant growth rate in the absence of O3 exposure (RGRcontrol). The largest negative O3 impacts on k were thus observed in slow-growing plants with small RGRcontrol. In contrast, the sensitivity of growth to O3 (PCRGR) was not related to RGRcontrol (P = 0.07).

Figure 4.

The relationship between the percentage change (PC) in the root/shoot allometric coefficient (k) induced by O3 exposure and relative growth rate (RGR) of control plants (r2 = 0.046; P = 0.02). Two data points (0.4, +86; 0.1, −71) are included in the regression analysis, but are not shown for clarity.

There was a strong, positive relationship between kozone and RGRozone (k = 0.74 + 0.089 RGR; P = 0.0007; not shown). This demonstrated a similar sensitivity of allocation to roots and whole plant growth in the presence of O3. This may correspond to two distinct aspects of O3 resistance, as there was no relationship between kcontrol and RGRcontrol in these species (P = 0.42; not shown), nor between PCk and PCRGR (P = 0.20; not shown).

Effect of exposure technology

The meta-analysis was used to test the null hypothesis [H3] that O3 exposure technology imposed no bias in the direction or magnitude of O3 impacts on allocation. Over all observations and plant groups, the hypothesis was not disproved. No significant effects of exposure technology on the sensitivity of k to O3 could be demonstrated from the available data (Table 6). Both CEC and OTC exposures yielded significant negative O3 impacts on k in herbaceous dicots (Table 6). No significant O3 effects could be resolved in monocots or trees because of small sample sizes. There were too few observations under GH conditions to allow a meaningful analysis (Table 6).

Table 6.  The effect of plant group and O3 exposure system on the responses of allocation (k) among the plant groups, where k = RGRroot/RGRshoot
End pointPlant groupExposure system
ParameterCECGHOTC
  • Values are the mean effect on the response ratio (RR) (LRRk) with indication of statistical significance and confidence intervals (CIs) in ln transformed units, and the relative change and sample size in native units. The non-uniform distribution of plant groups precluded an analysis of all species within each exposure system. No differences were found between exposure technologies.

  • *

    Change caused by O3 exposure differs significantly from 0.0 at P < 0.05.

  • a

    Mean effect size [natural logarithm of the RR (LRR)] caused by O3 exposure.

  • b

    Bootstrap 95% CI of the mean effect size.

  • c

    Mean percentage change (PC) caused by O3 exposure (back transformed from the mean effect size as PCk).

  • d

    Number of independent published observations.

  • CEC, controlled environment chamber; GH, glasshouse; OTC, open-top field exposure chamber; n/a, not applicable; NS, not significant.

kHerbaceous dicotyledonousEffect sizea−0.0276*−0.0344 NS−0.0592*
CIb(−0.0500, −0.0066)(−0.0914, 0.0230)(−0.1467, −0.0150)
PC (%)c−2.72−3.38−5.75
nd78510
Herbaceous monocotyledonousEffect sizea−0.0770 NS−0.3449−0.0677 NS
CIb(−0.2063, 0.0389)n/a(−0.1604, 0.0088)
PC (%)c−7.41−29.17−6.55
nd1518
TreeEffect sizea0.0000n/a−0.3117 NS
CIbn/a (−0.6930, 0.0737)
PC (%)c0.00 −26.78
nd1 7
AllEffect sizea−0.0351*n/a−0.1326*
CIb(−0.0624, −0.0092) (−0.2627, −0.0173)
PC (%)c−3.45 −12.42
nd94 25

The meta-analysis also supported [H3] for RGR. Over all observations and plant groups, there was no significant effect of exposure technology on the sensitivity of RGR to O3 (Table 5). Both the CEC and OTC exposure systems yielded significant reductions in RGR, in both herbaceous dicots and monocots (Table 5). The greater impact on dicots than on monocots (Table 3) was confirmed when the analysis was restricted to O3 exposures conducted in CE chambers (CECs) (Table 5).

Table 5.  The effect of plant group and O3 exposure system on the responses of the relative growth rate (RGR) among the plant groups, where (RGR = [(ln B2 − ln B1)/(t2 − t1)]
End pointPlant groupExposure system
ParameterCECGHOTC
  • Values are the mean effect on the response ratio (RR) (LRRRGR) with indication of statistical significance and confidence intervals (CIs) in ln transformed units, and the relative change and sample size in native units. The non-uniform distribution of plant groups precluded an analysis of all species within each exposure system. No differences were found between exposure technologies.

  • *

    Change caused by O3 exposure differs significantly from 0.0 at P < 0.05.

  • a

    Mean effect size [natural logarithm of the RR (LRR)] caused by O3 exposure.

  • b

    Bootstrap 95% CI of the mean effect size.

  • c

    Mean percentage change (PC) caused by O3 exposure (back transformed from the mean effect size as PCRGR).

  • d

    Number of independent published observations.

  • CEC, controlled environment chamber; GH, glasshouse; OTC, open-top field exposure chamber; n/a, not applicable.

RGRHerbaceous dicotyledonousEffect sizea−0.0775*−0. 1824−0.1442*
CIb(−0.0927, −0.0629)n/a(−0.1987, −0.0800)
PC (%)c−7.46−16.67−13.43
nd7419
Herbaceous monocotyledonousEffect sizea−0.0446*n/a−0.0477*
CIb(−0.0593, −0.0294) (−0.1017, −0.0066)
PC (%)c−4.36 −4.66
nd14 8
TreeEffect sizea−0.0990n/a−0.2117*
CIbn/a (−0.3674, −0.0901)
PC (%)c−9.43 −19.08
nd1 7
AllEffect sizea−0.0726*n/a−0.1317*
CIb(−0.0859, −0.0600) (−0.1922, −0.0805)
PC (%)c−7.00 −12.34
nd89 24

DISCUSSION

O3 effects on k and RGR

The meta-analysis allowed a quantitative re-evaluation of studies that may have suggested biologically meaningful effects of O3 on k or RGR, but lacked the statistical power to document significant differences (Bailar 1997). In the data available for the present analysis, only about 25% of observations and 50% of individual plant species exhibited a significant effect of O3 on k in the original studies. Of these, 74% showed that k declined. When all available data were pooled, 55% of observations reflected a reduction in k.

The meta-analysis demonstrated that mean k declined significantly by 5.6% (P < 0.05) following O3 exposure. The herbaceous dicots, mostly juvenile, that dominated the available data (74% of observations) exhibited a 2.6-fold smaller response than monocots, and a 7.7-fold smaller response than trees. The modest response of this over-represented plant group reduced the overall mean effect size determined by the meta-analysis.

The response in trees differed significantly from the response in dicots but, despite its large magnitude, was not itself significantly different from zero using the conservative, non-overlapping CI method (Waltz 2006). In the context of the meta-analysis, this may simply reflect the small number of observations for trees (a third of those available for monocots and less than a tenth of those for dicots). Yet, even among the original studies, there were few significant observations of O3 impacts on k. This must reflect a large variability relative to a small sample size in these studies, and may reflect the inherent difficulties of resolving the effects on growth and allocation in trees, perhaps because of longevity, woodiness or large plant size. In trembling aspen (Populus tremuloides), the sensitivity of gas exchange parameters to O3 varied with clone, leaf age and canopy position of same-aged leaves (Karnosky et al. 2005), and in sugar maple (Acer saccharum) with tree age, shading and (tree age × shading) (Chappelka & Samuelson 1998). In the latter species, shading enhanced O3 sensitivity in mature trees (Tjoelker et al. 1995), but not young trees (Laurence et al. 1996).

Despite the conservative estimate of O3 impact provided by the dominant herbaceous dicots, and the conservative test used to document this significant effect, the data do not support uncritical generalization nor prediction of O3 effects on allocation in specific instances. The diversity among plant groups was large and many changes in k were small, with 13% less than ± 0.03 in LRRk. This may reflect the relatively moderate O3 imposed in most available observations (Table 1), because a recent meta-analysis suggested that root growth was mostly inhibited when O3 exceeded 120 p.p.b. (Wittig et al. 2005).

The available data indicate a significant, but variable, reduction of k caused by O3. The present analysis did not isolate effects caused by plant age, growth habit or environmental conditions on O3 sensitivity of k. These, along with methods of scaling O3 impacts from seedlings to mature trees, remain important research questions.

Plant growth rate was also significantly reduced by O3, by 8.2% over all observations. This larger mean effect on RGR than on k was consistent with the detection of significant reductions in RGR in 50% of observations and 50% of species. The reductions were found in 89% of all observations. Observations that resolved O3-induced reductions were thus more common for RGR than for k.

The O3 impact on RGR, as for k, was largest in trees. The effects in trees and dicots were significantly greater than in monocots, but did not differ from each other. The large effects observed in trees suggest the potential for biologically meaningful O3 effects under ambient conditions. Even small O3 effects on growth or allocation in perennial species may impose large cumulative constraints on growth over several seasons (Retzlaff, Williams & DeJong 1997b). This is supported by simulation studies of long-lived trees (Retzlaff, Williams & DeJong 1992; Retzlaff et al. 2000). These O3 impacts on growth and allocation below ground may influence predictions of the biological effects of climatic change and the efficacy of proposed carbon sequestration schemes.

Although O3 impacts on RGR were larger and more consistent than on k, the significant observations were surprisingly much less uniformly distributed among the plant groups. These ranged from 0 to 63% of observations among the plant groups for RGR, compared with 22–38% for k. The herbaceous dicots exhibited a larger proportion of significant reductions and a larger mean effect than monocots, the reverse of the case for k. About 25% of all observations of changes in RGR were within ± 3% of the range, compared with only 13% in k. The meta-analysis confirmed the well-accepted impact of O3 on plant growth, but even in this case, generalizations may prove problematic because of variability among plant groups and experiments. Predictions of O3 impacts in specific cases may be more reliable for RGR than for k.

Carbohydrate allocation has some inherent limitations as a biological end point that may contribute to the difficulty of generalizing O3 impacts. The determination of root mass is confounded experimentally by unavoidable losses during separation from soil or growth medium. Relative losses may be large in the seedling experiments that constitute the bulk of available observations, because the young shoot dominates plant biomass and calculations of RGR. Furthermore, allometric relationships such as k are strongly conserved (Causton & Venus 1981; Hunt 1990; Farrar & Gunn 1998; Gunn et al. 1999) and may be more resistant than growth to perturbation by O3. Finally, whole-plant acclimation to O3 may obscure the effects on both RGR and k, through compensatory leaf initiation and enhanced gas exchange of young leaves (Pell, Eckardt & Enyedi 1992; Farage & Long 1995; McCrady & Anderson 2000). These limitations of allocation as an end point are not limited to documentation of O3 impacts. In studies of both root–shoot competition (Cahill 2002) and of shading (Hunt & Cornelissen 1997), it has proven more difficult to document the effects on allocation than on growth.

Shading and O3 exposure are analogous environmental perturbations. Both interact directly with plants only at the shoot and affect assimilation and source strength as well as carbohydrate allocation to sink tissues. The effects of O3 documented here are contrasted with the effects of shading reported for 28 species of seedlings by Hunt & Cornelissen (1997). No attempt is made to review the literature on shading. Shading [photosynthetic photon flux density (PPFD) of 125–135 versus 250 µmol m−2 s−1] significantly reduced k in 83% of observations with herbaceous dicots and 63% in herbaceous monocots (Hunt & Cornelissen 1997). The mean effects of shading in dicots were 8.4% for k and 20.3% for RGR, larger than the effects in monocots of 7.4% for k and 14.5% for RGR (calculated from data presented by Hunt & Cornelissen 1997). In addition to the larger mean effect on RGR, a smaller proportion of observations of RGR than of k yielded no detectable response (there were none for RGR). The mean effects of shading, as for O3, were larger, more frequent and more easily documented for RGR than for k.

Other impediments to documenting O3 effects on allocation are more specific to studies of O3 impacts. Some variability is related to experimental variability in O3 exposures. For example, exposures that track stochastic ambient concentrations will vary between experiments. Uncertainties in the rates of O3 transport to the leaf interior and in the rates of biochemical detoxification of O3 make it difficult to quantify the effective biological dose of O3, even when reproducible O3 exposures can be imposed (Lefohn 1992; Massman 2004). Furthermore, these physiological factors are modified by environmental conditions in poorly characterized ways.

In a survey of non-managed vegetation in Ohio (Showman 1991), less O3 injury was observed in a year with high ambient O3 than in a similar but rainier year with lower O3. This probably reflected a reduced biological dose, as stomatal uptake of O3 would be expected to decline with lower soil moisture and humidity in the drier year. Temperature increases of 1–3 °C, commonly associated with OTCs and climatic change, altered CO2 effects on the yield of wheat (Triticum aestivum) (Van Oijen et al. 1999) and may be shown in the future to interact with O3. Increases in root : shoot ratio of radish (Raphanus sativus), induced by root chilling from 18 to 13 °C, were reduced by exposure to O3 (Kleier, Farnsworth & Winner 2001). Factors such as soil type and pot size have not been found to influence O3 effects (Taylor et al. 1986; Whitfield, Davison & Ashenden 1996; Booker et al. 2005), but merit further scrutiny.

The uncertainty in resolving O3 impacts on biomass allocation has been associated with experimental and environmental variabilities, and with real and reproducible biological differences in the sensitivity of k to O3. The nearly eightfold range of sensitivities observed among the plant groups in the present analysis probably reflects all of these. A more robust generalization of O3 impacts on biomass allocation below the ground will require experiments over a broad range of plant types and environmental conditions, specifically designed to characterize this heterogeneity.

Towards the mechanism of O3 action

The mechanism by which biomass allocation in plants is regulated remains unknown, in general, and in response to environmental perturbations such as shading, soil water deficit and mineral deficiency, as well as exposure to O3. The diversion of biomass to shoot growth demonstrated here is believed to facilitate plant defense against reactive oxygen species generated from aqueous O3 (Lee & Bennett 1982; Mehlhorn et al. 1986; Sandermann 1996) and ultimately to enable repair of O3-induced foliar wounding (Barnes 1972; McLaughlin & Shriner 1980; Amthor & Cumming 1988; Sandermann 1996; Grulke & Balduman 1999). These responses of allocation are also consistent with a relaxed interpretation of the ‘theory of functional equilibrium’ (Brouwer 1983; van Noordwijk et al. 1998; Poorter & Nagel 2000). O3 induces compensatory shoot growth in response to reduced CO2 availability to the shoot, caused by a reduced ability to harvest CO2, even though absolute resource availability remains unchanged. Whatever the mechanism, O3 induces shoot growth at the expense of root growth, despite an unchanged resource availability in the rhizosphere.

The often distinct responses of k and RGR to O3 provide some insights into the possible mechanisms of O3 action in the whole plant. For example, the sensitivities of k and of RGR to O3 were not correlated. In 27% of all observations, O3 altered k without significantly affecting RGR. This implies that O3 altered allocation to roots without suppressing the productivity or the photosynthetic substrate availability. Such a separation of effects appears to be consistent with the inhibition of translocation of photosynthetic products rather than with substrate limitation caused by the inhibition of photosynthesis. This could reflect a direct inhibition of translocation, as previously suggested (Grantz & Farrar 2000), or a programmed restoration of functional equilibrium mediated by yet unknown controls.

The sensitivity of k to O3 was significantly related to RGR in the absence of O3 exposure. Reductions in k caused by O3 were most pronounced in slow-growing plants, while rapidly growing plants exhibited less negative or even positive changes in k. This decreasing inhibition of k by O3 with increasing RGRcontrol suggests that the sensitivity of k to O3 was not mediated by the stomatal flux of O3, as stomatal conductance and carbon assimilation would likely be correlated with the rate of plant growth. There was a strong positive relationship between k and RGR, when both were determined in the presence of O3, but no significant relationship under control conditions. In the shading studies of Hunt & Cornelissen (1997), herbaceous monocots and trees (woody species) exhibited consistent positive relationships between k and RGR, but the sensitivity of k to shading increased with increasing RGRcontrol. The resource ratio model (Tilman 1988), predicting that faster-growing plants would invest resources in photosynthetic tissues at the expense of roots, was not supported by the analyses of O3 (here) or shading (Hunt & Cornelissen 1997). Maintenance of RGR in the presence of O3 is a clear manifestation of O3 resistance, and the strong correlation between kozone and RGRozone implies that such resistance is also reflected in the maintenance of allocation to roots.

The current meta-analysis indicated that the sensitivity of RGR to O3 was not significantly related to RGRcontrol. In contrast, the sensitivities of RGR to elevated CO2 and to shading were both positively related to RGRcontrol (measured at ambient CO2) (van Noordwijk et al. 1998) or at higher PPFD (Hunt & Cornelissen 1997). A similar positive relationship has been observed previously for O3 (e.g. in mixed pasture species) (Bungener et al. 1999) and in mixed forest species (Karnosky et al. 2005). In the latter case, rapidly growing pioneer species (trembling aspen; paper birch, Betula papyrifera) were more responsive than slower-growing species (sugar maple, A. saccharum).

Both shading and O3 exposure could impact biomass allocation through source limitation. Carbon assimilation is reduced by shading, and by both chronic O3 exposure (Reiling & Davison 1994; Wiese & Pell 1997) and acute O3 exposure (Forberg, Aarnes & Nilsen 1987; Darrall 1989; Farage et al. 1991; Guidi et al. 1993, 1997; Farage & Long 1995; Grantz & Farrar 1999; Guidi, Tonini & Soldatini 2000; Zheng, Lyons & Barnes 2000). Simulations using the TREe GROwth: Response of Plants to Interacting Stresses (TREGRO) model (Retzlaff et al. 1997a) linked O3 inhibition of assimilation in sugar maple with reduced allocation to roots. In soy bean (G. max), a meta-analysis found that O3 exposure reduced leaf soluble carbohydrates more than plant growth (Morgan et al. 2003). These data imply that the export or consumption of recent photosynthate exceeded its rate of synthesis, and suggest that photosynthesis may be a primary target of O3 attack.

On the other hand, in cotton, leaf soluble carbohydrates increased following O3 exposure (Grantz & Farrar 2000), and short exposures (0.75 h) to O3 inhibited phloem loading and carbohydrate export more severely than photosynthesis (Grantz & Farrar 1999). Longer exposures altered allocation differently than did source limitation induced by partial defoliation (Grantz & Yang 2000). These data are consistent with a primary effect on phloem loading and secondary feedback inhibition of assimilation, perhaps as one of multiple targets of O3 attack.

Exposure to O3 has been shown to reduce carbohydrate export from source leaves of Plantago major (Zheng et al. 2000), to reduce the amount of 14C transported to roots of clover (Trifolium repens) (Blum, Mrozek & Johnson 1983), and at higher concentrations to nearly abolish the export of recent photosynthate from source leaves of cotton (Grantz & Farrar 1999). Compartmental efflux analysis indicated that phloem loading from the cytoplasmic transport pool was inhibited, whereas exchange with the vacuolar storage pool was unaffected (Grantz & Farrar 2000). The velocity of the transport of sugars within the phloem, in contrast, was unaffected in both loblolly pine (Spence, Rykeil & Sharpe 1990) and wheat (Mortensen & Engvild 1995).

O3 impacts on k do not appear to be obligately linked to O3 effects on RGR or on photosynthesis at the leaf or whole-plant scale. Inhibition of photosynthetic processes may be inadequate to fully explain the range of responses to O3 observed in whole plants and communities.

The lower leaves act as preferential sources of assimilate for roots, and the upper leaves for shoots (Wardlaw 1968; Thorpe, Walsh & Minchin 1998). Both O3 and shading have greater effects on older, lower leaves than on younger, more exposed leaves (Nie, Tomasevic & Baker 1993; Soja & Soja 1995; Mulholland et al. 1997). Accelerated senescence and reduced phloem loading in the lower canopy could each contribute to reduced allocation to roots. A simple model of phloem transport (Minchin, Thorpe & Farrar 1993) could account for the effects of shading and O3 on allocation. The flux of carbohydrate along two paths, and thus, the allocation of carbohydrate between two non-equivalent sinks, is shown to be altered by changes occurring only in source strength. If root and shoot are non-equivalent sinks, changes in k following both shading and O3 exposure could be caused by a reduced flux of carbohydrate from the source leaves. However, the mechanisms of source strength reduction could differ.

Effect of exposure technology

A concern in discussions of O3 effects on plants is the potential impact of exposure technologies on the apparent sensitivity to O3. The effects of enclosing plants in various exposure chambers on microclimate, atmospheric transport characteristics and effective O3 dose are well documented (Heck et al. 1988; Holt 1988; Manning & Krupa 1992; Fuhrer 1994; Elagoz & Manning 2002), particularly for OTCs (Heagle, Philbeck & Heck 1973; Unsworth, Heagle & Heck 1984a,b; Heagle et al. 1988; Mills 2002).

The current meta-analysis found no significant effects of the various exposure technologies represented in the available data on the O3 sensitivity of either k or RGR. Both CECs and OTCs yielded significant reductions of both k and RGR caused by O3. The available data were not ideally distributed among the exposure categories. Most of the observations involved herbaceous dicotyledonous species exposed in CECs of various designs. Fewer observations were available under GH or OTC conditions and none under non-chamber field conditions.

In trembling aspen (P. tremuloides), a comparison of OTC exposures with open air, ambient O3 gradient exposures and with chamberless [free air CO2 enrichment (FACE)] exposures (Coleman et al. 1995, 1996; Dickson et al. 1998; Karnosky et al. 1999; Isebrands et al. 2001) found no effect on the ranking of O3 sensitivity among clones. OTCs have been found to increase plant height relative to adjacent ambient plots (Olszyk, Tibbitts & Hertzberg 1980; Heagle et al. 1988; Albaugh, Mowry & Kress 1992), but relative sensitivity to O3 has been less affected (Olszyk et al. 1986, 1992; Heagle et al. 1988). The 1–3 °C elevation in temperature common in OTCs enhanced plant response to CO2 (Van Oijen et al. 1999), but has not been demonstrated to do so for responses to O3.

While the relationship between effective dose and imposed O3 exposure depends in part on chamber characteristics, there is no indication from these somewhat limited data that O3 sensitivity is a function of exposure technology. Nevertheless, further experimentation may resolve subtle effects of exposure technology and improve current methods of scaling from exposure experiments to extensive field environments.

CONCLUSIONS

The present meta-analysis has evaluated all available published observations of O3-induced changes in biomass allocation in plants. In a convincing majority of these observations, O3 exposure led to a decrease in allocation to the roots compared to the shoots, with a mean reduction by 5.6% in k. Few exposures led to increased allocation to roots and none of these yielded statistically significant results. Many reported experiments did not resolve significant changes in k, and the variability in responses of k to O3 was quite large. Some individual plant species exhibited both increases and decreases in k, in different experiments, following exposure to O3. The sensitivity of k to O3 in individual species was not predicted by their growth rate (RGR) under control conditions. In contrast to k, the effects of O3 on growth rate were larger and more frequently observed, with a mean reduction by 8.2% in RGR. Much of the uncertainty regarding the generality of O3 effects on allocation is because of this greater variability, associated with inherent limitations of biomass allocation as an experimental and biological end point.

The available evidence is heavily weighted towards dicotyledonous herbaceous species, exposed in early stages of growth. Trees exhibited the largest responses of k to O3 (23.9%), but the mean effect was not significant because of a small sample size. These results suggest a clear need for further primary studies among monocotyledonous and tree species. Despite the clear demonstration of a general impact of O3 on biomass allocation below ground, the possibility remains that important and reproducible differences between plant species will emerge as further investigations are conducted, that older plants exposed under more natural conditions will exhibit quantitatively different responses and that the combinations of environmental perturbations expected with global change will fundamentally alter the impact of O3 on k. These remain important areas of uncertainty. A meaningful interpretation of such interactions will depend on a deeper mechanistic understanding of plant responses to O3 and to these interacting factors.

ACKNOWLEDGMENTS

We gratefully acknowledge support from the United States Department of Agriculture, National Research Initiative Competitive Grants Program, through award number 96-35100-3841 to D.A.G., during the preparation of this review. D.A.G. acknowledges support from the Centro de Estudios Academicos sobre Contaminacion Ambiental (CEACA), Autonomous University of Queretaro, during the final manuscript preparation.

Ancillary

Advertisement