Carbon Flux and Growth in Mature Deciduous Forest Trees Exposed to Elevated CO2

See allHide authors and affiliations

Science  26 Aug 2005:
Vol. 309, Issue 5739, pp. 1360-1362
DOI: 10.1126/science.1113977


Whether rising atmospheric carbon dioxide (CO2) concentrations will cause forests to grow faster and store more carbon is an open question. Using free air CO2 release in combination with a canopy crane, we found an immediate and sustained enhancement of carbon flux through 35-meter-tall temperate forest trees when exposed to elevated CO2. However, there was no overall stimulation in stem growth and leaf litter production after 4 years. Photosynthetic capacity was not reduced, leaf chemistry changes were minor, and tree species differed in their responses. Although growing vigorously, these trees did not accrete more biomass carbon in stems in response to elevated CO2, thus challenging projections of growth responses derived from tests with smaller trees.

How forest trees, the largest biomass carbon (C) pool on Earth, will respond to the continued rise in atmospheric CO2 is unknown (1). Is there a potential for more growth, and perhaps more C storage, as a result of CO2 fertilization (2)? Are trees in natural forests already carbon saturated, given that CO2 concentrations have already reached twice the glacial minimum concentration (3)?

Experimental ecology has made important advances in recent years answering such questions, but unlike grass and shrub vegetation, adult forest trees do not fit any conventional test system with elevated CO2. Free air CO2 enrichment (FACE) is currently applied to fast-growing plantations (47), but to date, tall trees in a natural forest have not been studied because of overwhelming technical difficulties. We solved this problem with a technique called web-FACE (8) that releases pure CO2 through a fine web of tubes woven into tree canopies with the help of a construction crane (Fig. 1) (9). Here, we present responses of 32- to 35-m-tall trees in a near-natural deciduous forest in Switzerland (9) to a 530 parts per million (ppm) CO2 atmosphere over 4 years. Given the size and species diversity of the study trees, this project inevitably is tree- and not ecosystem-oriented, in contrast to other FACE projects, which use smaller trees. Because plant responses to CO2 are species-specific (10, 11), insight from single-species approaches remains limited, no matter how large the test scale. The statistical power of our approach is limited at the species level because of reduced intraspecific replication, but tree-ring analysis helps to account for much of the variation in tree vigor. The ultimate effect of rising CO2 will remain concealed within our limited time scales, yet knowing the dynamics of tree responses over a number of years helps to estimate the nonlinear, longer-term trends. The project is thus a compromise between realism and precision, given that there is no way to maximize both (12).

Fig. 1.

Canopy CO2 release by a web of porous tubes (left), with “iso-meters” (front) and gas sampling lines (back) for control, all operated from a 45-m-high crane (right).

Web-FACE uses CO2 gas with a constant carbon isotope composition (δ13C) of –29.7 ± 0.3‰ (mean ± SE of four annual means). Mixture with ambient-air CO213C = –8‰) results in a 13C tracer signal in photoassimilates that was monitored at ∼50 canopy positions with “iso-meters” (small containers planted with a grass using the C4 photosynthetic pathway), which yielded a 4-year mean of 5.8 ± 0.5‰ 13C tissue depletion. Once assimilated by leaves, the signal penetrates the various biomass compartments and allows one to track the fate of carbon. During the first season, leaves accreted 40% (Quercus), 65% (Fagus), 67% (Acer), 79% (Carpinus), and 100% (Tilia) of new carbon as compared with the iso-meter mean (9). By the end of the first season, new assimilates contributed 71% and 73% to newly formed leaf and wood tissue across all trees (Fig. 2), suggesting very close coupling between leaf and tree-ring construction (13). Tree rings of Quercus and Tilia consisted almost completely of new carbon after the first year, whereas in Fagus, the tree-ring signals were weaker and lagged. By the end of the fourth year, 99% of all leaf mass and 91% of all wood mass consisted of new C. After 4 years the new <1-mm fine root fraction (9) reached 40% (2.3‰ 13C depletion in 2004), suggesting a ∼10-year fine root turnover, similar to the data for other deciduous forests (14, 15).

Fig. 2.

New carbon in leaves, tree rings, and fine roots as assessed by 13C tracer signals (means ± SE). The maximum steady-state carbon isotope difference of 5.8‰ between ambient and elevated CO2 is represented by the iso-meter reference value (C4 grass growing in the canopy, straight dashed line).

Soil CO2 sampled in 170 gas-wells at 3- to 11-cm soil depth across the crane area was 1‰ more depleted after 7 weeks in the treatment zone and 1.8‰ after 4 months, and remained at 2.2 ± 0.2‰ during the remaining test period, indicating a near steady-state efflux of new C (38% of total CO2 emitted). Soil CO2 concentrations were 25% higher in the elevated CO2 area as compared with the control area (P = 0.03) in June 2001 and reached +44% in 2002. Mycorrhizal fungi carried a 64% new carbon signal already by the end of the first year (16). CO2 enrichment thus rapidly produced isotopic fingerprints in the whole leaf-soil continuum. The strong soil-air signal indicates a rapid flux of new carbon through this system, and increased soil CO2 concentrations provide clear evidence for enhanced metabolic activity in soils under CO2-enriched trees (16), as was found in smaller test systems (1719) and in four forest FACE studies (4, 20).

We found no downward adjustment of photosynthetic capacity in response to growth in elevated CO2 (21), similar to findings for other forest FACE experiments (2224) but contrasting with those for some smaller test systems (25). Hence, it is not surprising that leaf nitrogen concentrations in the canopy were little affected (Table 1). Only Carpinus showed a significant mean reduction in leaf N by 15%, and across all trees the mean depletion was not significant when the dilution by nonstructural carbohydrates (NSC) was accounted for. Quercus and the subdominant taxa showed a significant +22% to +25% (P = 0.005) increase of NSC + in leaves + in elevated CO2, similar to that for sweetgum in the Oak Ridge FACE (23). Across all trees, NSC increased by 12% (Table 1). NSC concentrations in branch wood were not significantly affected. The slight reductions in N and increases in NSC across trees were largely driven by the presence of two taxa, whereas the 5 to 8% reduction in specific leaf area (SLA) was not species-specific. Fagus, the only species with a periodic growth response (see below), did not show a CO2 response in any of these traits.

Table 1.

Leaf nitrogen (N), nonstructural carbohydrates (NSC), and specific leaf area (SLA) across taxa for 13 trees growing in elevated CO2 versus 16 control trees [repeated measures analysis of variance (ANOVA), with pretreatment values for each individual tree as covariable]; data are for 2001 to 2004, and means for early and late summer are pooled. Dry matter was used as a common reference, either including or excluding NSC.

Leaf trait Difference (%) P-value Main driver
Leaf NSC +12 0.007 Quercus, TAPView inline
N, total -10 0.027 Carpinus, TAP
N, NSC-free -7 0.136 Carpinus
SLA, total -8 0.002 Nonspecific
SLA, NSC-free -5 0.025 Nonspecific
  • View inline* TAP is a mix of rare taxa of the genera Tilia, Acer, and Prunus.

  • Foliage per unit land area (leaf area index, LAI) and leaf duration did not exhibit consistent changes. Elevated CO2 increased mean leaf duration for 2002 to 2004 in Carpinus and Fagus by 5 to 6 days and reduced it by 5 days in Quercus (marginally significant for the species × CO2 interaction for Quercus and Fagus; P = 0.068). Annual litter production did not respond to CO2 (Fig. 3). It indicates a constant LAI of ∼5, in line with data for younger monocultures (7, 1719, 26).

    Fig. 3.

    Leaf litter production standardized (divided) by the mean for two pretreatment years and the year 2001 (9) for each litter trap (218 ± 9 g m–2 for 56 traps) to account for a priori spatial variability (means ± SE).

    Total C and N in fresh fallen litter did not change significantly (9), but NSC concentrations (varying from 1.8 to 5.7%) rose, suggesting an overflow of photoassimilates channeled to the decomposer pathway (Table 2). Lignin concentrations ranging from 6.5 to 15% across species in ambient CO2 were significantly reduced in elevated CO2, largely driven by Fagus and the subdominant taxa. Quercus litter was hardly affected. These results indicate a shift in carbon fractions from recalcitrant to more labile compounds. Consequently, litter mass loss was enhanced after 220 days of in situ decay (largely driven by Carpinus and the subdominant taxa). In summary, litter became richer in starch and sugar, poorer in lignin, and decomposed faster, but the effect was species-specific. Ecosystem consequences will thus reflect species presence and abundance.

    Table 2.

    Change in litter composition (% dry matter) as well as 220-day decomposition of fresh leaf litter collected in the canopy (two-factorial ANOVA with CO2 and species as fixed factors).

    Litter trait Difference (%) P-value Main driver
    Total C +2 0.03 Nonspecific
    Total N -1 0.73
    NSCView inline +21 0.002 Fagus, TAPView inline
    LigninView inline -11 0.007 Fagus, TAP
    Litter mass loss +14 0.001 Carpinus, TAP
  • View inline* Nonstructural carbohydrates (sugar, starch).

  • View inline TAP is a mix of rare taxa of the genera Tilia, Acer, and Prunus.

  • View inline Correcting for NSC-free litter mass changed the relative difference in lignin between CO2 treatments from -10.7 to -10.0% (P = 0.01) but had no influence on the relative difference in nitrogen concentration.

  • Using pretreatment radial growth rates of trees (tree-ring analysis) (9), we standardized basal stem area increments during the treatment period derived from girth tapes. Fagus showed a sharp growth stimulation in the first year (+ 92%, P = 0.026), whereas no effects were observed in the other taxa at any time (Fig. 4). The CO2 response in Fagus became insignificant in 2002, but recovered in 2003 (P = 0.028) during a centennial heat wave during which elevated CO2 improved plant water relations (27). The cumulative 4-year response of Fagus was only marginally significant at P = 0.07. We performed a second analysis for trees irrespective of species that yielded no significant basal area effect, because the Fagus signal was diluted by other species signals and, hence, the cumulative CO2 signal for all trees in 2004 was zero (P = 0.79), suggesting C saturation (3, 28). Tree-ring chronologies illustrate that the four treatment seasons included a pair of best and worst years over 9 years of growth data, partly explaining the greater variation. Tagged and revisited leading shoots in the top canopy showed no consistent length growth response across species and years (mean annual increment ∼15 cm). Fruiting was unaffected except for a gain in a single masting year in Carpinus (Fig. 5).

    Fig. 4.

    Stem basal area (BA) growth of trees, standardized by the mean pretreatment BA increment for each tree. The two asterisks in Fagus indicate the only significant stimulation; 2003 was a drought summer; no significant CO2 × year interaction in any species). The bottom diagram is for all trees that showed measurable growth, i.e., 10 in elevated CO2 versus 29 controls (means ± SE).

    Fig. 5.

    Fruit debris collected with litter traps for three different tree species in years of mass fruiting. Pretreatment-to-treatment ratios for years with mass fruiting were used to account for spatial variability (numbers within bars for n traps). For Fagus we show the two latest mass fruiting episodes. The height of the bars denotes the degree of mass fruiting in the respective years (P = 0.047 for Carpinus in 2004; mean ± SE).

    The data suggest no lasting growth stimulations by CO2 enrichment in these mature trees after 4 years. It should be emphasized that the dominant study trees have reached only half of their average life span, have reached about two-thirds of their maximum size, and are growing vigorously by forestry standards (9). Other forest FACE experiments with younger trees have either shown a continuous growth stimulation (Pinus taeda) (4) or a transitory response, very similar to the one shown here for Fagus (Liquidambar styraciflua) (5). A lack of a response in stem growth or leaf litter production does not preclude a faster rate of below-ground (root) production, as was shown for the Oak Ridge FACE (15), but these are transitory C pools, which could translate into more recalcitrant forms of carbon in soil humus, mineral nutrients permitting.

    Initially we asked whether forest trees will reduce their C uptake or trap more carbon in a CO2-rich world. Our data suggest that they instead “pump” more carbon through their body. The few secondary effects found varied with species. Had we studied only single species, each would have led us to draw different conclusions: one species reducing its protein concentration, another consistently increasing its leaf carbohydrates, litter being affected in one group of species and not in others, and one species showing a transitory growth stimulation and others not. The Swiss forest FACE study thus points at the crucial role of tree species identity (29) and so far does not support expectations of greater carbon binding in tree biomass in such deciduous trees. The lack of a growth response or the transient response in one species is unlikely to be associated with N shortage (30), given the overabundance of N in this region (9). Broader stoichiometric constraints, soil microbial feedback (18, 31), or counteracting ambient ozone (6, 32) may hold answers. In summary, we find no evidence that current CO2 concentrations are limiting tree growth in this tallest forest studied so far.

    Supporting Online Material

    Materials and Methods

    Fig. S1

    References and Notes

    View Abstract

    Navigate This Article