Influence of Carbonic Anhydrase Activity in Terrestrial Vegetation on the 18O Content of Atmospheric CO2

See allHide authors and affiliations

Science  30 Mar 2001:
Vol. 291, Issue 5513, pp. 2584-2587
DOI: 10.1126/science.1056374


The oxygen-18 (18O) content of atmospheric carbon dioxide (CO2) is an important indicator of CO2 uptake on land. It has generally been assumed that during photosynthesis, oxygen in CO2 reaches isotopic equilibrium with oxygen in 18O-enriched water in leaves. We show, however, large differences in the activity of carbonic anhydrase (which catalyzes CO2 hydration and 18O exchange in leaves) among major plant groups that cause variations in the extent of 18O equilibrium (θeq). A clear distinction in θeq between C3 trees and shrubs, and C4 grasses makes atmospheric C18OO a potentially sensitive indicator to changes in C3 and C4 productivity. We estimate a global mean θeq value of ∼0.8, which reasonably reconciles inconsistencies between 18O budgets of atmospheric O2 (Dole effect) and CO2.

The rate of increase of the concentration of atmospheric CO2 is, on average, only about half of that expected on the basis of rates of fossil fuel emissions (1, 2). The ocean and the land biosphere must absorb the CO2 not accumulated in the atmosphere. Using13C in CO2 (3) and atmospheric O2/N2 ratios (4), direct estimates of the respective land and ocean sinks have been produced. On land, better understanding of CO2 sinks and sources requires the ability to distinguish between CO2 uptake in photosynthesis and release in respiration. The 18O content of atmospheric CO2 was shown to be a powerful tracer in this respect (5–8). The use of 18O relies on the dissolution of CO2 in water, allowing CO2-H2O oxygen exchange to occur. Water in leaves is highly enriched in 18O, relative to soil water, owing to evaporative fractionation (9, 10). Consequently, CO2-H2O exchange in leaves (associated with photosynthesis) or in soil (associated with soil respiration) produces contrasting 18O signals in the CO2 that is released to the atmosphere (5–8). Because the enzyme carbonic anhydrase (CA) is present in all plant leaves and rapidly catalyzes CO2 hydration and isotopic exchange, in spite of the short residence time of CO2 in leaves, it has generally been assumed that CO2 involved in photosynthesis is nearly completely relabeled by 18O-enriched leaf water. We directly tested this primary assumption and show that large variations in CA activity among plants result in a pervasive disequilibrium between leaf water and atmospheric CO2 that must be considered in global 18O budgets of atmospheric CO2.

The 18O content of atmospheric CO2 is usually considered in the context of the global atmospheric C18OO budget (5, 6, 8):Embedded Image Embedded Image Embedded Image(1)where c a is the concentration of atmospheric CO2, and F oa,F ao, F R,F f, and F A are the gross fluxes of CO2 between ocean and atmosphere, soil respiration, anthropogenic emission (due mainly to fossil fuel and biomass burning and land-use changes), and plant assimilation [whereF A = GPP (gross primary productivity)], with their respective isotopic composition (δx) and the associated kinetic isotopic fractionationsa w, a eff across the ocean and soil surfaces. F I is a soil invasion term due to diffusion of atmospheric CO2 into and out of soils, allowing for exchange with soil water with no net CO2 flux (11). ΔA is the 18O discrimination during plant assimilation [(6) and see below], which is highly sensitive to the isotopic equilibrium assumption tested here. Notably, similar 18O budgets are also constructed for atmospheric O2 and which also involve uncertainties associated with the biological fractionations [generally termed the Dole effect (9, 10, 12,13)]. For 18O budgets of both CO2and O2 (Eq. 1 or its O2 equivalent), a key element is the contribution of the distinct plants term, due to18O enrichment of leaf water relative to soil (and ocean) water, and the required consistency in this term for both isotopic budgets.

We conducted a survey of CA activity and CO2 exchange rates in 52 species. This range included all major plant groups (trees, shrubs, herbs, and grasses) and both main photosynthetic types (C3 and C4 pathways). Trees and shrubs were collected from the Jerusalem Botanic Gardens, which houses a collection of species from all continents and most ecosystems. Herbaceous species were collected at the Weizmann campus. Leaf samples (n= 3 per species) were assayed for maximal CA activity, according to the method in (14). In vivo CA hydration rates,CA leaf, are lower, owing to low CO2 concentrations at the site of CO2-H2O exchange in leaves (c cs) (15). We used the maximal CA activity and c cs values obtained from leaf-scale gas-exchange measurements of the same plant species to estimateCA leaf (corrected also for leaf temperature).

The most striking feature of the survey [Fig. 1A and supplementary Web information (16)] was the low CA hydration rates (CA leaf) observed in C4 grass species and in particular in C4 grasses, consistent with observations in preliminary studies on CA activity in plant leaves (17, 18). Mean CA leaf in C4 species (80 μmol CO2 m−2s−1) was four times lower than in the next highest group. Among C3 groups, mean CA leaf for trees and shrubs was by far the highest (1350 μmol CO2m−2 s−1 compared with an average of 400 to 700 μmol CO2 m−2 s−1 for the other groups). Intra- and interspecific differences in CA activity may vary with climatic and environmental conditions (19). However, on the basis of the consistency of these data with preliminary data elsewhere on C3 versus C4 differences, and with emerging physiological explanations of high and low CA activity, respectively (17, 19), these results probably reflect the broad differences in CA activity among the major plant groups (Fig. 1).

Figure 1

(A) The activity of CA within leaves (CA leaf, open bars) and rates of gross CO2 flux into leaves (F al, shaded bars) and (B) extent of isotopic equilibrium, θeq, between CO2 and water in plant leaves. Bars represent the mean of the data, grouped according to major taxonomic or physiological category (the number of species sampled for each group is given in parentheses). Mean values of θeqwere determined for the entire group, with or without outlier species where θeq was >1 SD above or below the group mean (hence the two values shown in the figure), to allow for the potential effect of nonrepresentative species. Note that although θeq≈ 1 for C4 herbs, this group contributes negligibly to global GPP (32) and is ignored elsewhere. To calculate θeq according to Eq. 2, we derived in vivo CA activity for each species from assays on leaf CA extracts at 17.5 mM CO2. This was corrected to a CO2 concentration within aqueous leaf media (from applying Henry's law toc cs at measured leaf temperatures), assuming K m = 2.5 mM for C4 plants (17) and 5 mM for C3species (41). These were corrected to leaf temperature by adopting Q 10 = 2 (17). The gross CO2 influx, F al, was derived from gas-exchange measurements as F al =F la + A, whereF la = A·ɛ and ɛ =c cs/(c ac cs). Fick's law was applied to calculatec cs by A =g w(c ic cs), assuming an internal CO2conductance of 0.5 and 1.2 mol m−2 s−1 for woody and herb species, respectively (14).

The definition of isotopic equilibrium, θeq, developed from work on 18O exchange in the CO2/HCO3 reaction (20), was described (14) asEmbedded Image(2)where kτ (the number of hydration and dehydration reactions per molecule of CO2 dissolved in the leaf medium) is the product of the CA rate constant, k, and the residence time of CO2 inside the leaf, τ. This relationship has been confirmed by comparison of θeqdetermined both from CA activity and leaf-scale measurements of18O discrimination (21). In physiologically measurable terms, kτ =CA leaf/F al, whereCA leaf was obtained as above andF al, the unidirectional CO2 flux from atmosphere to leaf, was calculated from leaf-scale gas-exchange measurements under saturating light levels (22). We observed opposing trends for in vivo CA hydration rates and atmosphere-leaf CO2 flux (Fig. 1A) that amplified plant group differences in θeq, such that C4 grasses (lowCA leaf, high F al) had by far the lowest mean extent of isotopic equilibrium (Fig. 1B). This is in contrast to the θeq values close to 1 found in C3 plant groups (θeq > 0.95 in 26 of the 39 C3 species) (16). These measurements of θeq, spanning the entire range between 0 and 1, are in marked contrast to the common assumption that exchange between leaf CO2 and leaf water is nearly complete (5–8, 23–28).

A global view of the geographical distribution of variations in CA activity and the extent of isotopic equilibrium in leaves (Fig. 2) is obtained by applying our θeq values to corresponding plant groups described in 1° by 1° vegetation maps of 15 plant groups (29). This reveals a predominance of low θeq in the northern and southern subtropics (10° to 30°), where C4-based ecosystems, such as savanna and grassland habitats, dominate (Fig. 2). To obtain a first approximation of the extent of global disequilibrium, the above map of θeq was made compatible with that used in the SiB2 land biosphere model (8). We define three “mega-groups” with distinct θeq values for the terrestrial vegetation on the basis of the SiB2 vegetation groups (30) (see Table 1). Then, the value of θeq at each grid point is weighted byF la, the backflux of CO2 from plants to atmosphere, where F la =GPP·ɛ, ɛ =c cs/(c ac cs), and GPP (gross primary productivity) and ɛ are obtained from the SiB2 global simulation. Using this approach we estimate global mean θeq between CO2 and plant leaf water to be 0.78 (Table 1). Put simply, this indicates that, in contrast to current assumptions, only ∼80% of the diffusional CO2 backflux from plants to the atmosphere reflects the leaf water signal relevant to atmospheric C18OO budgets. Ignoring such incomplete equilibrium could result in about 20% underestimation of the gross CO2exchange with plants derived from atmospheric measurements of δ18O values (5–8). The relative contribution from each vegetation type to the global θeqsignal is calculated as [GPP·ɛ·(1 − θeq)]vegtype/[GPP·ɛ·(1 − θeq)]globe, where the numerator uses parameters for each vegetation type, and the denominator uses the global value. Consistent with Fig. 2, this analysis indicates that most of the 18O disequilibrium effect is due to C4grassland (∼54%) and C3 grassland (∼29%). The global perspectives discussed above (vegetation map and weighting by gross CO2 fluxes) show that C4-dominated vegetation has an overwhelming effect in reducing 18O exchange between CO2 and leaf water.

Figure 2

Estimated geographical distribution of the extent of isotopic equilibrium, θeq, for global vegetation types (29). These were inferred from mean θeq of the seven plant groups in Fig. 1. Values were directly applied where vegetation types were equivalent to the groupings in Fig. 1 (i.e., coniferous forest, broadleaf forest, C3 grassland, and C4 grassland) or merged from two appropriate groups [mixed forest, θeq = 0.91 (0.98); forest and C3 grassland, θeq = 0.92 (0.84); forest and C4 grassland, θeq = 0.69 (0.55)]. The value of θeq for cultivated land [0.69 (0.55)] was based on a 40:40:20 weighting for C4:C3grasses:C3 dicots; shrubs/bare ground and tundra were assumed to comprise C3 herbs and shrubs. The above numbers in parentheses refer to isotopic equilibrium on the basis of mean θeq, excluding outliers (i.e., the higher values in Fig. 1 with the exception of C4 grasses, where the lower value is that without outliers). Excluding outliers increases the potential difference between vegetation group θeq and is plotted in Fig. 2 to demonstrate the largest potential range of θeq, although in the quantitative analyses, mean values of all data are applied.

Table 1

Mean values of isotopic equilibrium (θeq) for grouped vegetation types weighted by the respective CO2 fluxes (from SiB2 land-biosphere simulation). CO2 concentrations inside chloroplasts were taken from the SiB2 model and adjusted by +18 parts per million to account for CO2 concentration gradient between the site of CO2 fixation and CO2 hydration (14). Numbers in parentheses refer to the equilibrium values derived from mean θeq values, excluding outliers. Allowing for the effect of outliers makes no difference to the overall global θeq.

View this table:

Considering the atmospheric 18O budget in the context of Eq. 1, the 18O signal of terrestrial plants is represented by an apparent discrimination against 18O during photosynthesis, ΔA (6). This was recently adapted to include θeq (14,21) for which a simplified version isEmbedded Image(3)where ā is the mean diffusional fractionation of C18OO from air to leaf, ɛ = c cs/(c ac cs) as above, and δe refers to the δ18O of CO2 in equilibrium with chloroplast water. Recent global budgets constructed with Eq. 1estimate ΔA to be 13.7‰ (6) and 9.0‰ (8). When our global mean estimate of θeq = 0.78 is used, instead of assuming θeq = 1 (as was commonly done in the original analyses), equivalent values for ΔA can be derived only by modifying ɛ (i.e., leaf internal CO2 concentrations), δe (i.e., leaf water δ18O), or both, assuming that δa is precisely measured andā is theoretically well constrained. These values are not directly measured and involve marked uncertainties. Focusing, for example, on the analysis of (6), we show that modifying ɛ values alone would require physiologically unrealistic values (Fig. 3A) compared with what is known from leaf-scale studies. Figure 3A indicates that including θeq in the calculations makes ΔAconsiderably less sensitive to uncertainties in ɛ. Estimated values of δe, however, could be increased by ∼3‰ (Fig. 3B) and remain within current bounds on the predicted global mean value of18O in leaf water. This is exactly what is required to reconcile the long-standing differences between δeestimates derived from CO2 studies (3.3 to 4.4‰) (6, 8) and from O2 studies (about 3 to 8.8‰) (Dole effect) (9, 10, 12,13). Notably, compensating for θeq= 0.78 by revising δe estimates in (6) brings δe into quantitative agreement with estimates of ∼8‰ obtained from most recent models of the Dole effect (31).

Figure 3

Global discrimination against C18OO, ΔA, as a function of (A) ɛ values and (B) leaf-water δ18O values. Discrimination is calculated from Eq. 3 with parameters given in (6) [δe = 4.8‰ in (A);a̅ = 7.4‰, ɛ = 1.32 in (B); +0.4‰ is included in δe to allow for 18O fractionation between leaf water and leaf CO2]. The sloping lines indicate the relationships for global vegetation with (light solid line) or without (heavy solid line) the disequilibrium effect used in the global 18O budget of (6). Also shown for comparison is the high sensitivity of global mean ΔAto the contribution of C4 plant productivity (dashed line, ɛ = 0.64). The horizontal line indicates global discrimination solved from the global mass balance of C18OO from (6). Vertical lines show how a shift in θeqfrom 1.0 to 0.78 can be compensated for by leaf water (realistically) or ɛ (unrealistically), while constrained to a constant ΔA.

The 18O disequilibrium effect is likely to have even wider implications for biogeochemical and climate-change research. For example, using Eq. 3 and typical values for C3 plants (c cs ∼200, θeq ∼0.93) and C4 plants (c cs ∼140, θeq ∼0.38), a distinction of about 8‰ in ΔA values is obtained (ΔA ∼15 or ∼7‰ for C3 and C4, respectively). This makes18O of atmospheric CO2 a sensitive indicator of changes in C4 contributions to GPP, at a time when two large-scale processes are expected to influence it. First, the rapid increase in atmospheric CO2 is expected to greatly disadvantage C4 plants (32, 33). Recent modeling indicates that the global land area favoring C4 plants after of a doubling of atmospheric CO2 could be completely eliminated (33). Assuming current estimates of C4 productivity [∼25% of GPP on land; (34)] and no net change in total productivity, a rough estimate of the upper limits of the associated18O effect indicates a potential forcing of 30 Pg C × 8‰ = 240 Pg ‰. Assuming that this elimination of C4plants occurs after a doubling of the present atmospheric CO2 pool (2 × 750 Pg C), the corresponding change in atmospheric C18OO could approach ∼0.2‰, a signal 10 times that of current analytical precision (35).

In contrast, large-scale land-use changes normally result in the conversion of C3 forests to crops and grasslands (with a large C4 component). It is estimated (36) that ∼4.5 Pg C year−1 is released as a result of deforestation, which is offset by ∼3 Pg C year−1 that is reabsorbed as a result of abandonment and regrowth (with a net release of ∼1.5 Pg C year−1). Abandonment and regrowth are likely to have large grassland and C4 components. This is because, first, forest recovery can take 50 years or longer (36, 37), which is preceded by grassland/C4 productivity; second, the increasing rate of turnover in land use does not allow full forest recovery in any case (36, 38). Roughly estimating an upper limit for the 18O signal involved, we assume that reabsorption of about 2 Pg C year−1 associated with abandonment, regrowth, and newly introduced cropland (the latter involves ∼107 hectare year−1 and consequently ∼0.2 Pg C year−1) carries a C4-like signal. This yields an atmospheric forcing of 2 × 8 = 16 Pg ‰ year−1, or a possible trend of ∼0.02‰ year−1 in atmospheric CO2 (under current CO2 concentrations). Although close to the detection limit, such a signal may be observed better on regional or local scales. Because the lifetime of 18O in atmospheric CO2 is several years, this signal could accumulate and produce a secular (decreasing) trend in the 18O of atmospheric CO2. This could help explain the trend of about −0.08‰ per year that was observed during most of the 1990s (39), and for which an explanation has yet to be offered.

  • * To whom correspondence should be addressed. E-mail: dan.yakir{at}


View Abstract

Stay Connected to Science

Navigate This Article