Isotope Fractionation and Atmospheric Oxygen: Implications for Phanerozoic O2 Evolution

See allHide authors and affiliations

Science  03 Mar 2000:
Vol. 287, Issue 5458, pp. 1630-1633
DOI: 10.1126/science.287.5458.1630


Models describing the evolution of the partial pressure of atmospheric oxygen over Phanerozoic time are constrained by the mass balances required between the inputs and outputs of carbon and sulfur to the oceans. This constraint has limited the applicability of proposed negative feedback mechanisms for maintaining levels of atmospheric O2 at biologically permissable levels. Here we describe a modeling approach that incorporates O2-dependent carbon and sulfur isotope fractionation using data obtained from laboratory experiments on carbon-13 discrimination by vascular land plants and marine plankton. The model allows us to calculate a Phanerozoic O2 history that agrees with independent models and with biological and physical constraints and supports the hypothesis of a high atmospheric O2 content during the Carboniferous (300 million years ago), a time when insect gigantism was widespread.

A dominant feature of Earth's atmosphere is the presence of abundant free oxygen (O2), which signifies an active aerobic biosphere. Much attention has been directed toward understanding the rise of O2 during the Precambrian (1–5). Of equal interest, however, is the relative lack of variability in the partial pressure of atmospheric oxygen (pO2) during the Phanerozoic (the past 600 million years). Physical and biological constraints based on the fossil record of charcoal and forest fires limit extreme atmospheric composition variations to roughly 10 to 40% O2 at constant N2 mass (6, 7). This range is remarkably constant considering the potential variation in atmospheric O2 production and consumption rates that may have occurred over geologic time. On hundred- to thousand-year scales, O2 is controlled by global rates of photosynthesis and respiration. On longer time scales, the burial of organic matter (OM) and pyrite (FeS2) in sediments and the oxidative weathering of these materials on the continents (plus oxidation of reduced C- and S-containing gases liberated from them at depth) become the dominant controls on pO2 (8, 9). To maintain approximate constancy of pO2 through time, strong negative feedbacks must exist within global biogeochemical cycles. Proposed feedbacks include links betweenpO2 and forest fires (7, 10) and between the concentration of dissolved oxygen [O2]aq and nutrient availability in the oceans (11–13); however, none of these feedbacks have been formulated into a model that is consistent with constraints imposed by chemical and isotopic mass balances for carbon and sulfur.

The calculation of changes in pO2on geologic time scales can be achieved if reasonably accurate estimates of burial rates and weathering rates of OM and pyrite can be developed (14). Models based on the abundance of organic carbon and pyrite sulfur in sedimentary rocks through time (15) provide one approach to the modeling of PhanerozoicpO2 variations. Another approach is the use of mathematical models driven by carbon and sulfur isotope variations (9). In brief, the abundance of 13C in seawater-dissolved inorganic carbon reflects the partitioning of carbon between total global masses of carbonate and OM in sedimentary rocks; likewise, 34S abundance in seawater SO4 reflects partitioning of sulfur between global sedimentary sulfide and sulfate. These features arise because during photosynthetic fixation of CO2 there is a strong discrimination in favor of 12C; photosynthetic biomass is significantly depleted in 13C relative to ambient CO2 (17, 18). Analogously, bacterial sulfate reduction exhibits a strong discrimination in favor of 32S; biogenic sulfide minerals are significantly depleted in 34S relative to ambient dissolved sulfate (19). Several studies (16, 20–22) have attempted to use these features of the carbon and sulfur isotope systems to resolve pO2variations based on the isotopic records of marine carbonates and sulfates, which serve as proxies for oceanic values of13C/12C and 34S/32S. None has generated a Phanerozoic O2 history that is consistent with physical and biological constraints or the rock abundance model. Indeed, the use of geologically and biologically reasonable feedbacks in isotope-driven O2 models results in unavoidable positive feedbacks and catastrophic modeled O2histories (16, 23).

The possible dependence on pO2 of net carbon isotope discrimination during photosynthesis and biogenic sulfide production are two potentially important factors currently omitted from isotope-driven O2 modeling studies, although theoretical considerations indicate that both are likely. Increased rates of plant photorespiration under high ambient O2/CO2 are predicted because of the dual carboxylase-oxygenase function of Rubisco (24), potentially leading to a greater fraction of biomass being derived from13C-depleted respired CO2 (25). Sulfur cycling in marine sediments is related to O2availability, and more recycling results in a greater net biogenic sulfide 34S depletion (19). Therefore, to address the sensitivity of these current uncertainties on Phanerozoic O2 change, we report results from experiments determining the influence of elevated pO2 on the net13C discrimination of a herbaceous angiosperm (Ranunculus repens), a cycad (Macrozamia communis), and a marine diatom (Phaeodactylum tricornutum), all grown in the laboratory under a variety of CO2 and O2 concentrations. These data have been used, together with a best-guess estimate of sulfur isotope fractionation with O2 content, to derive a new mass balance model of O2 for calculating a revised Phanerozoic O2 history.

Each species of vascular land plant was grown under similar environmental conditions (i.e., light intensity, temperature, humidity,pCO2, and nutrient supply) but contrasting atmospheric O2/CO2 ratios [21 and 35% O2 with 330 parts per million (ppm) CO2] in controlled environment chambers described previously (26,27). The upper pO2 value was taken to represent a previous upper estimate of the past 300 million years (15). Table 1 shows the C isotopic composition, expressed as per mil (‰) deviation from the Pee Dee Belemnite standard, of CO2 in growth chamber air (δgas) and of the plant samples (δplant) grown at ambient and elevated O2 concentrations and the net carbon isotope discrimination ΔC[(δgas − δplant)/(1 + δgas/1000)]. These results show, in line with theoretical expectations, that plants grown at elevatedpO2 (35%) exhibit 13C-depleted biomass relative to plants grown under ambientpO2 (21%). The depletion, expressed as Δ(Δ13C) values in Table 1, averaged 1.4‰ for R. repens and 3.2‰ for M. communis.

Table 1

Fractionation of carbon isotopes during experimental vascular plant growth. All values are expressed as ‰. ΔC = 1000(1 − αC) = (δgas − δplant)/(1 + δgas/1000), where αC is the fractionation factor, numbers after ± indicate 1σ uncertainties, and values in parentheses indicate the number of individual gas or plant samples.

View this table:

The marine diatom P. tricornutum was grown under constant light, temperature, and nutrient supply (and thus had a constant growth rate) but variable [CO2]aqand [O2]aq in a continuous culture (chemostat) system described previously (28). Oxygen saturation of the solution relative to contemporary atmosphericpO2 ranged from 106 to 155% (equivalent to equilibrium with an atmospheric pO2 of 22 to 32.5%), whereas [CO2]aq varied from 3.7 to 20.1 μmol kg−1. Previous work has demonstrated that carbon isotopic fractionation in marine microalgae varies as a function of both [CO2]aq and algal growth rate (28, 29). The isotopic fractionation for P. tricornutum grown under similar ranges of growth rate (μ) and [CO2]aq, but in equilibrium with 21% O2 (30), were compared with the results of the present study in order to determine the effect of varying O2/CO2 on isotope fractionation corrected for fractionation due to varying CO2 (Table 2) (31).

Table 2

Carbon isotopic fractionation for the marine diatomP. tricornutum grown under different levels of [O2]aq and [CO2]aqbut at a constant growth rate μ of 0.60 per day. The predicted (Pred.) values for ΔC were calculated from the data and model of Laws et al. (30) for P. tricornutum for the same values of μ/[CO2]aq but for constant present atmospheric pO2 (21%). Obs., observed. Values of Obs.-Pred. represent changes in fractionation due to varying [O2]aq/[CO2]aqcorrected for isotopic fractionation due to varying [CO2]aq. For the method of Δ/ΔC determination, see (31).

View this table:

The results of the experiments can be applied directly to an isotope-driven O2 mass balance model by calculating the relation between ΔC and O2 asEmbedded Image Embedded Image(1)where ΔC is isotope fractionation (discrimination) for vascular land plant or algal carbon (in ‰); (ΔC)O = 25‰ = ΔC value for the present level of O2 but for an average higher-than-present CO2 level during the Phanerozoic (32, 33); O2 is the mass of oxygen in the atmosphere at any past time; 38 is the mass of oxygen in the present atmosphere (in 1018 mol); and J is an empirical coefficient used for curve fitting. Equation 1 closely mirrors that expected from theoretical modeling of photosynthesis and carbon isotope fractionation to the same increase in atmospheric O2 content (34).

Because of the complex effects of oxidation of sulfide and multiple fractionation during diagenetic recycling (19), no straightforward experiments on the effects ofpO2 on sulfur isotope discrimination during bacterial sulfate discrimination are possible. As an initial approximation, we assume a simple linear proportionalityEmbedded Image(2)where ΔS is fractionation for sulfur (in ‰) and (ΔS)O is the average fractionation for the present level of O2 (35‰).

The utility of a model that involves C and S isotopic fractionation depends on how O2 variations are calculated from carbonate and sulfate isotopic records. Changes with time toward13C-enriched carbonates indicate an increase in the total mass of OM in sediments and sedimentary rocks, either through increased OM burial or decreased OM weathering. Increased OM burial (or decreased weathering) rates correspond to elevated O2 production and increasing atmospheric pO2. However, at constant carbon and sulfur isotope discrimination during biomass production, the variations in atmospheric O2 calculated from observed variations in the δ13C of carbonate and δ34S of sulfate are too large to be realistic, resulting even in highly negative values for pO2. By allowing photosynthetic carbon and sulfur isotope discrimination to vary with atmospheric O2, changes in carbonate13C and/or sulfate 34S content result in damped O2 variations as compared with O2 variations derived without the O2 functionality.

The isotope model used in the present study is that of Garrels and Lerman (9), but with O2-dependent isotope discrimination and the use of both the carbon and sulfur isotopic record to calculate rates of weathering and burial of organic carbon and pyrite sulfur. Also included is rapid recycling analogous to what was done in modeling rock abundance data (15). Rapid recycling is a means of incorporating into a model the observation that younger rocks are more likely to be exposed and weathered on the continents than are older rocks. The change in atmospheric oxygen mass (O2) with time was calculated from the expressionEmbedded Image(3)where F refers to fluxes in mass per unit of time, the subscripts b and w refer to burial and weathering (including oxidation of reduced gases derived from deep processes), and the subscripts g and p refer to organic carbon and pyrite sulfur, respectively.

For carbon isotope fractionation, the values of J inEq. 1 were varied to obtain an O2 history bounded between 10 and 40% O2, which at the same time fitted our experimental plant and plankton growth results (Fig. 1). A curve for J = 2.5 is fit to the experimental data, although a somewhat higher value ofJ is more likely because cycads [showing the highest Δ(ΔC)] represent a primitive life form that is more representative of the burial of ancient terrestrial OM, as emphasized in the present study. Extensive sensitivity analysis showed that the variation of J from 0 to 5 results in a lowering of the peak value of O2 at 280 million years before the present (My B.P.) (Fig. 2) by about 25% and that values of O2 for 450 to 350 My B.P. are affected more by the inclusion of O2 dependence in sulfur isotope fractionation than by its inclusion in carbon isotope fractionation.

Figure 1

Plot of carbon isotopic fractionation as a function of % O2. Δ(ΔC) = ΔC − (ΔC)O = change in fractionation from that for 21% O2. Curves derived fromEq. 1 are shown for different values of J, with the experimental data fitted by J = 2.5. Solid squares represent the vascular plant results of Table 1; open circles are calculated from the data presented in Table 2 for P. tricornutum growth in seawater at 25°C equilibrated with 330 ppm CO2 (31).

Figure 2

Plot of O2 versus time for the Phanerozoic calculated by isotope mass balance modeling (the present study) and by the abundance of organic carbon and pyrite sulfur in sedimentary rocks (15). The upper and lower lines represent the range of estimated errors for the results of the rock abundance modeling.

The modeling results for the Phanerozoic history of atmospheric O2 based on Eqs. 1 through 3 and the isotopic composition of sedimentary carbonates (35, 36) and sulfates (16, 37, 38) as proxies for seawater isotopic composition are shown in Fig. 2 and are compared with the results of the sediment abundance model of Berner and Canfield (15). There is surprisingly good agreement between the results of these two totally independent approaches.

The most obvious feature in Fig. 2 is the largepO2 maximum centered around 300 million years ago, thought to be a response to the evolution of large vascular plants on the continents (8, 15, 16). Production of new sources of biomass (such as lignin) that were resistant to the then available pathways of OM degradation likely led to enhanced OM burial in swamps and in marine sediments after transport to the oceans by rivers. This explanation is consistent both with the abundance of coal deposits and OM in general that were preserved at this time (15, 16) and with the most severe enrichment in oceanic 13C content measured during the Phanerozoic (16, 36).

It has been suggested that variations in globally averaged ΔC values between sedimentary carbonate and OM correspond largely to variations in atmospheric CO2 content (33). This relation certainly holds for specific classes of organic compounds that show a strong relation between 13C depletion and ambient pCO2 values (39–41). However, ΔC values for the Permo-Carboniferous are larger than values immediately before and after this time span, a feature difficult to explain in terms of greatly increased pCO2. Indeed, Permo-Carboniferous CO2 concentration is likely to have been decidedly lower than during the preceding and succeeding periods (32). The paradox might be resolved if the pO2 of the Permo-Carboniferous reached 35%, a value indicated by modeling, because a greater measured ΔC at this time could be explained without relying entirely on the burial of anomalously13C-depleted OM (33). Plant growth experiments demonstrate that plants can grow in atmospheres of up to 40% O2 (42) with the balance between vegetative and reproductive growth remaining unaffected, indicating the possibility of maintaining viable populations under such conditions. Furthermore, a value of 35% O2 (with pCO2 = 300 ppm) is compatible with continued biogeochemical cycling of carbon by terrestrial ecosystems (43). Other physiological studies suggest that elevated pO2 during the Permo-Carboniferous may help explain patterns in evolution. Flight metabolism in arthropods is enhanced at elevated O2concentrations (44), and the sudden rise and subsequent fall of insect gigantism documented from the fossil record by, for example, giant dragonflies with wingspans of 70 cm is coincident with the Permo-Carboniferous maximum inpO2 (45, 46). Other patterns have been linked to elevated O2, such as changes in organisms with diffusion-mediated respiration and the invasion of the land by vertebrates (45, 46).


View Abstract

Stay Connected to Science

Navigate This Article