Author for correspondence: John S. King Tel: +1 919 513 7855 Fax: +1 919 515 3169 Email: email@example.com
• Concentrations of atmospheric CO2 and tropospheric ozone (O3) are rising concurrently in the atmosphere, with potentially antagonistic effects on forest net primary production (NPP) and implications for terrestrial carbon sequestration.
• Using free-air CO2 enrichment (FACE) technology, we exposed north-temperate forest communities to concentrations of CO2 and O3 predicted for the year 2050 for the first 7 yr of stand development. Site-specific allometric equations were applied to annual nondestructive growth measurements to estimate above- and below-ground biomass and NPP for each year of the experiment.
• Relative to the control, elevated CO2 increased total biomass 25, 45 and 60% in the aspen, aspen–birch and aspen–maple communities, respectively. Tropospheric O3 caused 23, 13 and 14% reductions in total biomass relative to the control in the respective communities. Combined fumigation resulted in total biomass response of −7.8, +8.4 and +24.3% relative to the control in the aspen, aspen–birch and aspen–sugar maple communities, respectively.
• These results indicate that exposure to even moderate levels of O3 significantly reduce the capacity of NPP to respond to elevated CO2 in some forests.
Decades of experimental evidence show that small forest trees experience an average stimulation of 16–31% in biomass production under elevated atmospheric CO2, but these responses can be constrained by soil nutrient availability or other environmental factors (Strain & Cure, 1994; McGuire et al., 1995; Gebauer et al., 1996; Curtis & Wang, 1998; Johnson, 1999; Zak et al., 2000; Oren et al., 2001). Evidence from long-term forest FACE (free-air CO2 enrichment) experiments corroborates this finding at larger spatial and temporal scales. Net primary production of a 13-yr-old loblolly pine (Pinus taeda) ecosystem was stimulated 26% under CO2 enrichment, and this response persisted for 4 yr (DeLucia et al., 1999; Hamilton et al., 2002). A 10-yr-old sweetgum (Liquidambar styraciflua) ecosystem increased NPP an average 21% under CO2 enrichment (Norby et al., 2002). Three species of Populus grown in short rotation culture increased woody biomass by 15–27% under elevated CO2 (Calfapietra et al., 2003). Responses of mature forests to elevated atmospheric CO2 are still poorly understood.
Most elevated CO2 studies have not considered the phytotoxic effects of simultaneous exposure to elevated tropospheric O3, which has been shown to decrease tree seedling growth from −2 to −69% (Pye, 1988). Pre-industrial concentrations of tropospheric O3 are estimated to have been less than 10 nl l−1, and have risen to 30–40 nl l−1 background levels today (Levy et al., 1997). Tropospheric O3 concentrations are expected to exceed 60 nl l−1 over large portions (50%) of the global forested land surface by the year 2100 (Felzer et al., 2004). The decrease in agricultural productivity caused by ambient O3 toxicity has been estimated at US$1.0–5.8 billion annually (1990 dollars) in the USA (Kopp et al., 1985; Adams et al., 1986; Murphy et al., 1999), but economic losses to wood production are unknown (Krupa et al., 2000). Biogeochemical modeling estimates of reductions in annual terrestrial C sequestration in the USA, caused by ambient O3 pollution during the late 1980s to early 1990s, range from 18 to 38 Tg C yr−1, which must be accounted for in future calculations of the global C budget (Felzer et al., 2004). Therefore a key to understanding the role of forests in mitigating the build-up of atmospheric CO2 is determining how NPP will respond to the interactive effects of elevated atmospheric CO2 and tropospheric O3.
Here we report on the NPP of intact experimental forest communities dominated by the most widespread tree taxa in North America, in response to the interactive effects of elevated atmospheric CO2 and tropospheric O3. The study was performed at the Aspen FACE project in Rhinelander, WI, USA (Dickson et al., 2000). Communities of pure trembling aspen (Populus tremuloides Michx.), and competitive mixes of trembling aspen–paper birch (Betula papyrifera Marsh.) and trembling aspen–sugar maple (Acer saccharum Marsh.) were exposed for 7 yr to concentrations of atmospheric CO2 and tropospheric O3 predicted for the year 2050. Net primary production was estimated with species-specific allometric regressions developed at the site, applied to annual nondestructive measurements of all trees in the experimental plots. We hypothesized that (i) elevated atmospheric CO2 would provide sustained enhancement of NPP in all three forest communities; and (ii) elevated tropospheric O3 would decrease it. Because it has been postulated that elevated CO2 may decrease O3 uptake into the plant because of decreased stomatal conductance (Allen, 1990; Wustman et al., 2003), our third hypothesis was that simultaneous exposure to both elevated CO2 and elevated tropospheric O3 would result in growth similar to the control.
Materials and Methods
The Aspen FACE project is a randomized complete block design of atmospheric CO2 and tropospheric O3 treatments, with species composition (aspen, aspen–birch, and aspen–maple) as a split-plot factor (n = 3). The 12 30-m-diameter plots are fumigated using free-air technology of the Brookhaven National Laboratory's design (Hendrey et al., 1999) to maintain atmospheric targets of 560 ppm CO2 and 1.5× ambient O3 (Table 1). Fumigation is during daylight hours only, and begins at bud break in the spring and ends at leaf senescence in the fall, with an average growing season of 145 d (Table 1). Trees were planted at 1 × 1-m spacing, and fumigation began in 1997. Over the life of the experiment, system performance has been within 20% of the target 93% of the time for CO2, and within 20% of the target 80% of the time for O3 (http://www.aspenface.mtu.edu).
Table 1. Summary of Aspen FACE project control data for atmospheric fumigation treatments, 1998–2003. Values are mean daily concentrations or sums (O3 exposure) for the entire growing season of each year
Ambient CO2 (control, +O3) (ppm)
Elevated CO2 (+CO2, +CO2 + O3) (ppm)
Ambient O3 (control, +CO2) (ppb)
Elevated O3 (+O3, +CO2 + O3) (ppb)
Ambient O3 exposure (control, +CO2) (ppmh)
Elevated O3 exposure (+O3, +CO2 + O3) (ppmh)
Exposure duration (d)
Tree growth is monitored by annual measurement of total height and diameter at 10 cm above-ground of all trees within the central core area of the plots (n = 3684). The core area in each plot measures 166, 76 and 65 m2 for the aspen, aspen–birch and aspen–maple sections, respectively. The core area in which nondestructive studies are carried out is buffered from possible edge effects by five rows of trees on the outer edge of the plots. In 2000 and 2002, complete above- and below-ground harvests were performed on a total of 196 trees at the edge of the central core area to develop species-specific allometric biomass equations. Trees were harvested in midsummer at peak leaf area by severing the stem 3 cm above ground, separating foliage from wood (stem + branches). The tree heart root system was sampled by driving a 25.4-cm internal diameter corer to a depth of 25 cm centered on the severed tree stem. Between-tree coarse and fine (<1 mm diameter) root biomass was estimated with 10 randomly located cores, 15 cm diameter by 25 cm depth, within the subplots. All roots were removed from the soil by washing over a fine-mesh screen, and were sorted into coarse (heart + all roots >1 mm diameter) and fine (all roots <1 mm diameter). Only live root data are presented in the current study. All plant parts were dried to constant mass at 65°C and subsamples were combusted at 500°C for 7 h to correct for mineral content.
Estimating stand-level biomass and NPP
Allometric biomass equations (Table 2) were developed by regressing the logarithm of plant dry weight (g; foliage, wood, heart root system) against the logarithm of stem diameter (cm) of the harvested trees for each species (aspen, paper birch, sugar maple). Regressions were tested for significant effects of the experimental treatments on model intercept and slope, as well as for suitability of transformations and independent variables. Models of the form ln(dependent variable) = B0 + B1 ln(diameter) proved to explain most of the variation in the data, with statistically insignificant additional predictive power added by height. Allometric analyses of the biomass data from the two harvests (King et al., 1996, 1999) indicated that biomass partitioning between foliage, wood and heart roots was unaffected by the treatments, allowing for use of a common model for each species. Model R2 values ranged from 0.84 to 0.99, with seven of the nine models having R2 > 0.90 (Table 2).
Table 2. Allometric regressions used to predict tree component biomass of young aspen, paper birch and sugar maple at the Aspen FACE project in Rhinelander, WI, USA
Parameter estimate (P)
MSE, mean square error.
Models were developed from trees harvested destructively within FACE plots in 2000 and 2002. All models had the form log(y) = m log(x) + b, where y = biomass component (g) and x = diameter (cm). Baskerville's (1972) adjustment to the antilogarithm was applied when calculating absolute data from the log–log models.
Aspen heart root
Paper birch foliage
Paper birch wood
Paper birch heart root
Sugar maple foliage
Sugar maple wood
Sugar maple heart root
The species-specific regressions were applied to the annual diameter measurements of all trees in the core area of the plots for each year of the experiment, correcting for slight underestimation during back-transformation of biomass estimates using the method of Baskerville (1972). Biomass estimates of foliage, wood and heart roots of all trees in the subplots (g) were summed and divided by subplot area to arrive at stand-level estimates of tree biomass (g m−2). Between-tree coarse and fine root biomass (g m−2) for each year of the experiment was estimated by assuming that the partitioning of total root biomass to heart, coarse and fine root fractions determined by destructive harvest in 2000 and 2002 did not change over time. We then applied the respective root fractions to the allometrically determined estimates of heart root biomass for each year, to arrive at total root biomass. Annual fine root production cannot be determined by this method; however, a published record of annual peak fine root biomass will be useful when fine root turnover rates become available. Additionally, presentation of fine root biomass data allows comparison of the relative size of the fine root pool relative to other plant parts. Annual net wood and coarse root production was determined by subtracting previous year biomass from subsequent year biomass (g m−2 yr−1). Tree mortality was not quantified explicitly in the current study; however, individual trees contributed to stand-level biomass and production estimates only as long as they were alive. At the point when trees were recorded as ‘dead’ they were removed from subsequent allometric modeling, thereby implicitly accounting for mortality.
Stand-level estimates of total and component biomass and production were tested for main effects (CO2, O3) and split-plot effects (community, time) using split-plot anova appropriate for the Aspen FACE experimental design (King et al., 2001) using the statistical analysis system software (SAS Institute, Cary, North Carolina, USA). Tree size in 1997 (diameter2 × height) was used as a covariate in the anova to account for initial differences in plant size.
Annual biomass production
In 1997, the newly planted stands averaged 22, 15 and 11 g biomass m−2 in the aspen, aspen–birch and aspen–maple communities, respectively (Table 3; Fig. 1). Annual production of foliage, wood (stem + branches), and coarse roots in control plots increased from 1998 to 2003 in all communities (time P < 0.000). Over this time, annual foliage production averaged 250, 213 and 127 g m−2 in the aspen, aspen–birch and aspen–maple communities, respectively. Wood production averaged 372, 285 and 190 g m−2, while coarse root production averaged 90, 76 and 52 g m−2 in the aspen, aspen–birch and aspen–maple communities, respectively. Annual fine root production was not quantified in the current study; however, July fine root biomass averaged 24, 20 and 24 g m−2 in the aspen, aspen–birch and aspen–maple control plots, respectively, from 1998 to 2003.
Table 3. Beginning (1997) and end (2003) biomass pools (g biomass m−2) and yearly biomass production (g biomass m−2 yr−1) by community and atmospheric treatment for foliage, wood (stems + branches), and coarse roots (tap root + all roots >1.0 mm diameter)
Fine root (<1.0 mm diameter) data are standing crop at height of growing season (mid-July). Values are means (n = 3) with standard error in parentheses. Community indicators: AA, pure aspen; AB, aspen–birch mix; AM, aspen–maple mix.
Wood and coarse root pool size estimates for 2003 were determined using allometric biomass equations and may differ slightly from estimates arrived at by summing annual production caused by rounding. Differences are generally <3% and within the standard error of the allometric estimates.
Pool size, 1997
Foliage, wood and coarse root production, and fine root biomass, 1998
Foliage, wood and coarse root production, and fine root biomass, 1999
Foliage, wood and coarse root production, and fine root biomass, 2000
Foliage, wood and coarse root production, and fine root biomass, 2001
Foliage, wood and coarse root production, and fine root biomass, 2002
Foliage, wood and coarse root production, and fine root biomass, 2003
Annual biomass production of foliage, wood and coarse roots varied in response to the treatments over time, with significant CO2 × time, O3 × time and CO2 × community × time interactions (Tables 3, 4). This variation in community responses to the treatments is not surprising given the interannual variability of weather, host-specific pathogens, etc. in such a long-term experiment. On average, elevated CO2 increased foliage, wood and coarse root production from 31 to 71% relative to the control, with the aspen–birch and aspen–maple communities showing larger relative responses than pure aspen (Table 5). Elevated O3 decreased annual production of foliage, wood and coarse roots relative to the control by 9–29% on average, with the aspen community generally showing the largest decrease in growth. Concurrent exposure to elevated CO2 and O3 caused small reductions in aspen foliage, wood and coarse root annual production relative to the control, and 15–30% increases in the aspen–birch and aspen–maple communities (Tables 4, 5).
Table 4. Statistical significance of atmospheric CO2 (CO2), tropospheric O3 (O3) and community experimental factors on stand-level foliage, wood and coarse root production from 1998 to 2003 at the Aspen FACE project in Rhinelander, WI, USA
Wood, stem + branches; coarse roots, roots >1 mm diameter; ns, not statistically significant (P > 0.05).
Fine root production was not determined in this study.
CO2 × O3
CO2 × community
O3 × community
CO2 × O3 × community
CO2 × time
O3 × time
CO2 × O3 × time
Community × time
CO2 × community × time
O3 × community × time
CO2 × O3 × community × time
Table 5. Average relative response (percentage change relative to control) of foliage, wood (stems + branches), and coarse root mean annual production, and mean fine root biomass from 1998 to 2003, to experimental treatments at the Aspen FACE project in Rhinelander, WI, USA
+CO2 + O3
Values calculated as (treatment mean − control mean)/control mean × 100.
Community indicators: AA, pure aspen; AB, aspen–birch mix; AM, aspen–maple mix.
Relative to the control, average July fine root biomass increased 45, 64 and 29% under elevated CO2 in the aspen, aspen–birch and aspen maple communities, respectively, Tables 3, 5, 6). Tropospheric ozone caused an average decrease in fine root biomass relative to the control of 15 and 19% in the aspen–birch and aspen–maple communities (Tables 3, 5, 6), respectively, but curiously increased it 44% in the aspen community. In the +CO2, +O3 treatment, mean July fine root biomass increased relative to the control 107, 46 and 34% in the aspen, aspen–birch and aspen–maple communities, respectively (Tables 3, 5, 6).
Table 6. Statistical significance of atmospheric CO2, tropospheric O3 and community experimental factors on stand-level biomass from 1997 to 2003 at the Aspen FACE project in Rhinelander, WI, USA
Data were analyzed using a repeated-measures split-plot anova after King et al. (2001).
Wood, stems + branches; coarse roots, all roots >1 mm diameter; fine roots, roots <1 mm diameter; ns, not statistically significant (P > 0.05).
CO2 × O3
CO2 × community
O3 × community
CO2 × O3 × community
CO2 × time
O3 × time
CO2 × O3 × time
Community × time
CO2 × community × time
O3 × community × time
CO2 × O3 × community × time
Changes in annual production caused by the experimental treatments during the 7-yr experimental period had cumulative impacts on standing biomass in the aspen, aspen–birch and aspen–maple communities (Table 6; Fig. 1). By 2003, total tree biomass in control plots averaged 3349, 2626 and 1717 g m−2 in the aspen, aspen–birch and aspen–maple communities, respectively (Table 3; Fig. 1). Biomass was distributed on average as 13.8% foliage, 67.3% wood (stems + branches), 17.6% coarse roots, and 1.5% in fine roots, with minor variations between communities. Allometric analyses (King et al., 1996) showed that shifts between the major biomass fractions because of the treatments were statistically insignificant (data not shown).
Relative to the control, elevated CO2 increased total biomass by 24.9, 45.6 and 60.3% averaged over all years in aspen, aspen–birch and aspen–maple communities, respectively (Fig. 2). The degree of stimulation increased over time, with a significant CO2 × time interaction (Table 6; Fig. 2). Tropospheric O3 decreased standing biomass relative to the control by an average 22, 13 and 14% in the aspen, aspen–birch and aspen–maple communities, respectively (Table 6; Fig. 2). The aspen community sustained the largest decline, whereas growth in the aspen–birch and aspen–maple communities became less sensitive to tropospheric O3 over time (Table 6; Fig. 2). Concurrent exposure to elevated CO2 provided some protection against the negative effects of O3 (Figs 1, 2). Standing biomass in the aspen community under the combined treatment (+CO2, +O3) declined on average 8% relative to the control, but by the fourth year of treatment had recovered to near ambient levels. Biomass production in aspen–birch and aspen–maple communities under concurrent exposure was comparable with, or greater than, that in control plots, with a stimulation of 8 and 24%, respectively, averaged over the 7-yr period (Fig. 2).
Growth of the experimental forest communities of the Aspen FACE project compares well with growth of stands of similar age reported from sites in north-central Wisconsin, Minnesota and Alaska. After 7 yr growth, control stands of pure aspen at our site had total above-ground (wood + foliage) and below-ground (coarse + fine roots) biomass of 2758 and 591 g m−2, respectively. Ruark & Bockheim (1988) reported above- and below-ground biomass in naturally regenerated 8-yr-old aspen stands in north-central Wisconsin of 2500 and 1380 g m−2, respectively. The discrepancy in below-ground biomass was probably caused by residual root biomass from the previous stand in the study of Ruark & Bockheim (1988), whereas in our study cuttings were planted in root-free soil.
Alban & Perala (1990) reported average above-ground biomass of 2376 g m−2 from a series of sites 5 yr after harvest in northern Michigan and Minnesota. Paré & Van Cleve (1993) reported above-ground biomass of naturally regenerated aspen 14 yr after harvest near Fairbanks, Alaska of c. 5000 g m−2, roughly twice the age and biomass of our study. The agreement between our study and these published reports gives confidence that the Aspen FACE stands are representative of natural forests, and supports our allometric approach to estimating biomass.
Applying species- and site-specific biomass regressions to annual nondestructive measurements of all trees in the experimental plots at Aspen FACE will be valid for estimating biomass and NPP of wood and coarse roots for some time to come (perhaps with some additional destructive harvesting). However, the utility of the allometric approach for fine roots may be limited. There has been difficulty in developing stand-level scaling relationships between above-ground plant parts and soil core estimates of fine root biomass (Kurz et al., 1996; King et al., 1999). This is because of high spatial variation in root distributions, and extremely plastic fine root responses to differences in environmental conditions (Nadelhoffer, 2000; King et al., 2002; Pregitzer, 2002). In the current study, we partitioned a fraction of heart root biomass (determined allometrically) to fine roots based on measured root biomass partitioning from the destructive harvests in 2000 and 2002, of which the fine root sampling was much more rigorous in 2002. The fraction of total root biomass partitioned to fine roots varies with seasonal changes in fine root standing crop (Hendrick & Pregitzer, 1996; King et al., 2002) and stage of stand development (King et al., 1999). Hence, fine root biomass determined from our ‘static’ partitioning may or may not accurately reflect standing fine root biomass at a given point in time, and does not capture fine root production and turnover.
Similarly, allometric modeling of foliage biomass (NPP) at Aspen FACE has been valid for the early stage of stand development, but its utility will be limited in the future. Comparison of allometric estimates of stand-level foliage production (this study) with litter-trap estimates of litter production showed good agreement (R2 = 0.89) for the years 2001–3 (C. P. Giardina, unpublished). However, foliage biomass becomes ‘uncoupled’ from stem growth after canopy closure in young stands, and is inversely related to it in older forests (Ovington, 1957; Ford, 1984; Cannell, 1985; Gower et al., 1994; Miller, 1995). Collection of annual foliar litter production suggests the aspen and aspen–birch stands are approaching canopy closure; that is, the annual increment in litter production is decreasing (C.P.G., unpublished data). Therefore future stem growth may not be accompanied by proportional increments in foliage biomass, compromising the allometric approach.
The large, sustained CO2 enhancement of NPP at Aspen FACE could have several causes. Because of the relatively high latitude of the site (45°40′ N), the soil is of recent glacial origin, with good chemical and physical properties for forest growth (Dickson et al., 2000). Analysis of soil N cycling from 1999 to 2003 (Holmes et al., 2003, 2005) suggests that soil N availability is not constraining growth responses to elevated CO2 at the Aspen FACE experiment. The north-temperate climate is mesic, with favorable site water balance for most of the year, because of low evaporative demand (calculated by King et al., 2001). A large fraction of the global forest C sink occurs in recently glaciated north-temperate and boreal forest ecosystems, where long-term C storage in soils is especially important (Schlesinger, 1997; Myneni et al., 2001). Therefore the nutrient limitation to sustained CO2 enhancement of forest NPP, as reported from low-latitude forests on highly weathered soils (Oren et al., 2001), may be less of a constraint at higher latitudes.
Additionally, the experimental stands at Aspen FACE are dominated by early successional species in the early stage of stand development, which confers greater growing space (less intertree competition) and greater productivity relative to older stands (Pregitzer & Euskirchen, 2004). This could provide the capacity for greater stimulation of NPP and other ecosystem properties in response to elevated CO2 relative to older, closed-canopy forests (King et al., 2004).
The stimulation of total biomass production at Aspen FACE was caused by proportional increases in all plant parts: roots, wood and foliage. Averaged across community type from 1997 to 2003, elevated CO2 caused 42, 45 and 41% increases in foliage, wood and coarse root biomass, respectively. Allometric analyses on an individual tree basis, using the harvest data from 2000 and 2002 and at the stand level, showed that elevated CO2 did not change biomass partitioning among plant parts (data not shown). This is consistent with our understanding of tree biomass partitioning responses to elevated atmospheric CO2 (Gebauer et al., 1996; King et al., 1996; Norby et al., 1999).
There were important differences in the magnitude of CO2 enhancement of component and total plant biomass production between communities, however. The order of relative response was generally pure aspen < aspen–paper birch < aspen–sugar maple; however, the order of absolute stand-level biomass production has been pure aspen > aspen–paper birch > aspen–sugar maple. In the pure aspen community, it is possible that intraspecific competition has constrained the potential stand-level relative growth enhancement in response to elevated atmospheric CO2. Interspecific competition in the aspen–birch community could possibly have allowed a greater overall growth response to elevated CO2. The aspen–maple community started out with smaller trees and therefore less intertree competition (intra- and interspecific competition was reduced), hence there was a greater capacity to respond to elevated atmospheric CO2.
McDonald et al. (2002) provide evidence that competitively advantaged trees in the pure aspen community at Aspen FACE show a greater relative growth response to elevated CO2 compared with competitively disadvantaged trees in an autoregressive manner (‘the big get bigger faster’). These results apparently scale to the level of the stand. It will be interesting to see how relative growth responses to the treatments change as the stands proceed through canopy closure, and intertree competition and mortality become more significant. More growing space in the young stands could contribute to the greater relative CO2 growth enhancement at Aspen FACE compared with the Duke (DeLucia et al., 1999; Hamilton et al., 2002) and Oak Ridge National Laboratory (Norby et al., 2002) experiments, which both have older, closed canopy forests. This is consistent with a recent synthesis of soil respiration results from the four forest FACE experiments, which found that the relative stimulation of soil respiration caused by elevated CO2 was greater in young, open-canopy forests compared with older, closed-canopy forests (King et al., 2004).
Decreased forest NPP from tropospheric O3
Our second hypothesis was that elevated tropospheric O3 (c. 1.5 × ) would decrease forest NPP, which was again supported by this analysis of 7 yr of growth data from the Aspen FACE experiment. This result is largely consistent with the literature but, importantly, we feel provides realistic quantification of the magnitude of the response for an important forest type in north-temperate and boreal forest ecosystems. Our understanding of O3 effects on vegetation is largely based on studies of crops or small trees grown in highly controlled environments for short periods (reviewed by Heck et al., 1984; Pye, 1988; Samuelson & Kelly, 2001; Andersen, 2003). These studies show that, in a wide range of plant species, tropospheric O3 causes almost universal reductions in crop yield or biomass production, but the magnitude of response has been highly dependent on experimental conditions. High variation in experimental results and uncertain correlation between visible foliar injury and yield reduction have led to considerable efforts to compare seedling responses with those of mature trees to determine appropriate factors for scaling O3 responses to the landscape (Chappelka & Samuelson, 1998; Matyssek & Innes, 1999; Samuelson & Kelly, 2001).
At Aspen FACE, chronic exposure to moderately elevated tropospheric O3 (c. 1.5×) has resulted in an average reduction in biomass production of 22, 12 and 16% in the pure aspen, aspen–birch and aspen–maple communities, respectively. These results are comparable with the average 23% decrease in tree seedling growth reported in the review of Pye (1988), but higher than the 2.6–6.8% decrease in annual NPP in the USA during the late 1980s to early 1990s from the modeling study of Felzer et al. (2004). Importantly, the response to O3 was modified by both community composition and time (significant O3 × community × time interaction). The pure aspen community was the most sensitive to O3 and maintained this sensitivity over time. The aspen–birch and aspen maple communities, however, appear to be losing sensitivity to O3 relative to the control.
Differences in community response could be caused by compensatory growth of less-O3-sensitive species (Pye, 1988; Broadmeadow & Jackson, 2000) in the mixed communities, or changes in O3 responsiveness induced by competition (McDonald et al., 2002; Liu et al., 2004). In the aspen–maple community, sugar maple comprises c. 9% of wood biomass, whereas in the aspen–birch community the two species are more evenly represented with no clear dominance of one over the other (data not shown). Hence compensatory growth of less-O3-sensitive species is unlikely to be the cause of the increased productivity over time. In a 2-yr phytotron study, Liu et al. (2004) observed that European beech experienced no reduction in total biomass production caused by elevated O3 when grown in monoculture. However, when grown in mixed culture with Norway spruce O3 caused a 32% reduction in beech biomass, and the spruce benefited (+13%) from the weak performance of its competitor. These results underscore the important fact that monospecific responses to O3 are not simply additive, and more realistic experimental designs are required to determine long-term ecosystem responses to the changing atmosphere. An important aspect of the Aspen FACE experiment will be to see if the mixed communities fully regain productive capacity under elevated O3.
As with elevated CO2, growth under tropospheric O3 does not appear to have altered biomass partitioning among the major plant parts, as there were no statistically significant shifts in root : shoot, foliage : branch or wood : coarse root ratios (data not shown). This finding apparently contradicts many earlier studies that show relative decreases in root growth under O3, which is thought to aid in the repair of damaged photosynthetic structures by increased C allocation above ground (Karnosky et al., 1996; Andersen, 2003). In our study, all plant parts became proportionally smaller under elevated O3. The exception to this is fine roots in the pure aspen section, which showed an average 44% stimulation in biomass. Possible causes include (i) spurious values among the three replicates for each treatment; (ii) confounding by abundant fine roots from herbaceous species that proliferated under the open canopies of the elevated O3 treatment; or (iii) it is a real effect. Fine root (<1-mm-diameter) biomass values for each replicate of the 2002 harvest upon which the static partitioning was based were 50.4, 48.1 and 34.0 g m−2, compared with an average 31.7 g m−2 for control plots at that time. If herbaceous roots were accidentally included in our estimates, the static partitioning used here would propagate the error through each year of biomass estimation. This is unlikely as all communities were harvested and processed at the same time using the same method. Sampling error could also have biased towards high root biomass estimates, but this is unlikely as 10 cores of 15 cm diameter × 25 cm deep were used in each split-plot FACE ring section. The only other fine root harvest at the site performed in 1999 did not detect significant effects of O3 on fine root biomass (King et al., 2001). In any case, this finding is highly counterintuitive, and requires further study before we can conclude that elevated O3 increases fine root biomass in young aspen ecosystems.
Tropospheric O3 compromises stimulation of NPP caused by elevated CO2
Our final hypothesis was that chronic exposure to elevated CO2 and elevated tropospheric O3 (+CO2, +O3) would result in forest NPP similar to that of the control. A putative ‘protective effect’ of elevated CO2 has been discussed (Allen, 1990; Wustman et al., 2003), in that decreased stomatal conductance under elevated CO2 might decrease the flux of O3 into the plant; there may be other protective mechanisms, such as responses of antioxidant enzymes (Rao et al., 1995). Our analysis partially supports this hypothesis. For total and component biomass production, the interaction between CO2 and O3 was never statistically significant. Thus elevated CO2 provided comparable stimulation to NPP at both levels of the O3 treatment in all communities over time. Because of the sensitivity of the aspen community to O3, however, total biomass production in this community was depressed for the first 3 yr of growth, after which it did not differ from the control. The aspen–birch and aspen–maple communities exhibited an average stimulation of total biomass production of 8 and 24%, respectively, under combined fumigation. Thus the large stimulation in biomass production all three communities experienced in response to elevated CO2 was completely annulled or greatly reduced by concurrent exposure to moderate levels of tropospheric O3.
Experiments using long-term exposure of trees to combined CO2 and O3 fumigation are beginning to show that responses to both gases are variable, depending on species/clone and provenance. However, the antagonistic effects on growth of elevated CO2 and tropospheric O3 have generally been observed in these experiments. Dickson et al. (1998) exposed five hybrid poplar genotypes to factorial treatments of CO2 and O3 in open-top chambers (OTC) for 1 yr, and found that plants exposed to combined fumigation (+CO2, +O3) had biomass similar to the control. Broadmeadow & Jackson (2000) grew seedlings of oak, ash and pine under factorial CO2 and O3 treatments in OTC for 3 yr. Elevated CO2 enhanced growth; O3 decreased it; and combined fumigation provided some protection from O3 in the order of species responsiveness: oak > pine > ash. In a 5-yr OTC experiment, Rebbeck & Scherzer (2002) found that yellow poplar growth was insensitive to O3 alone, but increased 60% with combined fumigation (+CO2, +O3) relative to the control, but not until the fifth season. Similarly, Riikonen et al. (2004) found that growth of two clones of silver birch responded negatively to O3, but only at ambient CO2. Tree growth increased under elevated CO2 and combined fumigation (+CO2, +O3) treatments. Together with our results these studies show that, in the long run, elevated CO2 provides some protection from exposure to phytotoxic concentrations of tropospheric O3 for a variety of forest tree species. However, this also means that gains in NPP that could be achieved under elevated CO2 are being compromised by tropospheric O3 pollution, and this has had continent-scale implications for C sequestration for some time (Felzer et al., 2004).
Because CO2 is chemically inert in the atmosphere, and human population growth and fossil energy consumption continue to increase, the concentration of atmospheric CO2 will continue to rise for the foreseeable future. Tropospheric O3 is highly reactive and was historically considered a regional pollutant. However, it is becoming apparent that the cumulative impact of industrialization around the globe is also raising the background concentration of this pollutant along with CO2. Our analysis of 7 yr of growth data at the Aspen FACE project, and decades of earlier research, indicate that the concentration of atmospheric CO2 expected for the year 2050 has the capacity to stimulate forest NPP. At least for young northern forests on glacial soils, this response does not appear to be constrained by nutrient or water limitations. However, the concurrent global rise in tropospheric O3 is damaging forest physiology and growth to the point that potential gains in terrestrial C sequestration caused by rising CO2 are partially or completely annulled. We conclude, therefore, that global monitoring of ambient O3 exposure of vegetation should become an important part of government environmental protection programs. Moving forward with technologies that remove important anthropogenic precursors to photochemical O3 formation (mainly oxidized forms of nitrogen) from automobile and industrial emissions would help to decrease concentrations of tropospheric O3 because of its short half-life in the atmosphere, decreasing at least one constraint on the capacity of forest ecosystems to sequester atmospheric C.
Aspen FACE is principally supported by the Office of Science (BER), US Department of Energy, Grant No. DE-FG02-95ER62125 to Michigan Technological University, and Contract No. DE-AC02-98CH10886 to Brookhaven National Laboratory, the US Forest Service Northern Global Change Program and North-central Research Station, Michigan Technological University, and Natural Resources Canada – Canadian Forest Service. Additional support for research at Aspen FACE was supported by DOE Program of Ecosystem Research (PER) grant number DE-FG02-93ER6166, the USDA Forest Service Northern Global Change Program, and USDA NRI Competitive Grants Program nos. 2001-35107-11262 and 2004-35102-14782. The authors thank Jaak Sober and Wendy Jones for their assistance in operating the Aspen FACE facility.