Limits for Combustion in Low O2 Redefine Paleoatmospheric Predictions for the Mesozoic

See allHide authors and affiliations

Science  29 Aug 2008:
Vol. 321, Issue 5893, pp. 1197-1200
DOI: 10.1126/science.1160978

This article has a correction. Please see:


Several studies have attempted to determine the lower limit of atmospheric oxygen under which combustion can occur; however, none have been conducted within a fully controlled and realistic atmospheric environment. We performed experimental burns (using pine wood, moss, matches, paper, and a candle) at 20°C in O2 concentrations ranging from 9 to 21% and at ambient and high CO2 (2000 parts per million) in a controlled environment room, which was equipped with a thermal imaging system and full atmospheric, temperature, and humidity control. Our data reveal that the lower O2 limit for combustion should be increased from 12 to 15%. These results, coupled with a record of Mesozoic paleowildfires, are incompatible with the prediction of prolonged intervals of low atmospheric O2 levels (10 to 12%) in the Mesozoic.

Atmospheric O2 has played a key role in the development of life on Earth, with the rise of O2 in the Precambrian being closely linked to biological evolution (1). Variations in the concentration of atmospheric O2 throughout the Phanerozoic are predicted from models based on geochemical cycling of carbon and sulfur (2, 3). Such models predict atmospheric O2 concentrations as low as 10% (3) or 12% (2) (present atmospheric O2 is 20.9%) in the Mesozoic [251 to 65 million years ago (Ma)], and low (<12%) O2 atmospheres are hypothesized to be the primary driver of at least two of the “big five” mass-extinction events in the Phanerozoic (1, 4). Few proxies have been developed for testing models of past atmospheric O2 concentrations, particularly the low O2 levels predicted for the Permo-Triassic boundary (3) and throughout the Triassic (3) and Jurassic (2). The presence of charcoal in the geological record provides one means to test predicted paleoatmospheric O2 levels (5). Several studies have sought to test the limits of combustion under varying concentrations of O2 (68); early work observed that the minimum O2 level required for the combustion of CO and CH4 was 13% (6) and that paper would burn at 15% O2 (7). More recent work found that dry (0 to 2% moisture) pine dowels would burn at 12% O2 (8). No previous study has assessed the limits of combustion on natural (that is, not unnaturally dry) material, nor has any study been conducted in a realistic atmospheric environment.

We tested the hypothesis that prolonged low-O2 intervals occurred in the Mesozoic by using a series of combustion experiments and a compilation of geologic evidence for wildfires. These experimental burns have been undertaken in a >8 m3 walk-in plant-growth room equipped with a thermal imaging system and full atmospheric, temperature, and humidity control (9). The occurrence of paleowildfires (based on the occurrence of charcoals, inertinites, and geochemical evidence) in the Mesozoic is documented and compared with the predicted atmospheric O2 levels for this time, which are in turn reassessed on the basis of results from the combustion experiments.

Combustion is defined as a sequence of exothermic chemical reactions, between fuel and oxidant, accompanied by the production of heat or both heat and light in the form of either a glow or flames. These criteria were used to judge when combustion was occurring in a suite of rigorously controlled experimental atmospheres. Charring is not a measure of combustion, because charcoal can be made without the presence of O2 by using a heat source greater than 200°C (no heat is given off; the material is simply “cooked”). Combustion in different atmospheric O2 concentrations was measured either as production and duration of flaming and smoldering of matches, paper, candles, and Pinus caribaea (Ocote pine) wood, or as temperature change captured using continuous thermal imaging of sphagnum moss burns on a hot plate set at ∼450°C.

Atmospheric CO2 concentration had little effect on ignition or combustibility of the material (Fig. 1, A to D). A struck match produced a brief flame at 14% O2 (Fig. 1A), which was long enough to produce a flame and short-lived smoldering in paper (Fig. 1C). A candle and P. caribaea sticks sustained a flame briefly at 16% O2 (Fig. 1, B and D) but not at lower O2. A candle continually burned at 17% O2, whereas a match and paper burned to ash at 18% O2. P. caribaea wood burned with a sustained flame at 18% O2. The moss charred at all ranges of O2 concentration; however, no temperature change or flames were generated by the moss below 15% O2 (Fig. 1, E and G). The mean maximum temperature reached by the moss was found to increase with O2 concentration (Fig. 1E) [an r2 (coefficient of determination) value of 0.832 reveals a reliable correlation]. The moss did not show a temperature change at O2 concentrations below 15% (with no smoldering or flaming observed), indicating that combustion cannot occur at such low concentrations of atmospheric O2. The moss exceeded the temperature of the hot plate at 15% O2 by an average of 41.3°C [reaching maximum temperatures of 546.2°C (Fig. 1F)] with a brief flame observed, providing evidence that combustion first occurs at this concentration of atmospheric O2. In ambient O2, the moss exceeded the temperature of the hot plate on average by 332°C (reaching a maximum value of 882.5°C) and flamed readily.

Fig. 1.

Effects of O2 on combustion. (A to D) Ignition and flame duration for various O2 concentrations and in ambient and high CO2 (2000 parts per million) for (A) matches, (B) a candle, (C) paper, and (D) P. caribaea sticks (ignition source for all was matches). (E) Temperature changes observed in sphagnum moss heated on a hot plate set at 450°C for various O2 concentrations. All data are based on 15 repeated burns for each setting. A temperature change is first seen in the moss at 15% O2 (shown by a positive deviation in temperature of the moss from the temperature of the hot plate). (F) Thermal image of sphagnum moss heated in 15% O2, showing that the temperature of the moss is hotter than that of the hotplate at this concentration of atmospheric O2 and above, indicating that 15% O2 is the lower limit needed for combustion of plant material (that is, maximum temperature of some areas of moss are greater than that of the hot plate). (G) Thermal image of sphagnum moss heated in 14% O2, showing that no temperature change is observed (that is, maximum temperature of the moss is less than that of the hot plate).

Figure 1E shows that combustion does not occur below 15% O2, that there is limited combustion between 15 and 17% O2, and that combustion occurs readily at 17% O2 and above, in agreement with Watson (7), who showed that dry paper cannot be ignited by a spark below 17% O2. Thus, our results strongly suggest that the minimum O2 limits for combustion should be increased from 12% to 15% O2.

Our results contradict recently published evidence that pine wood dowels burn in 12% atmospheric O2 (8). These experiments were performed under conditions in which the ignition source was external to the low-O2 atmospheric environment; hence, material was ignited through a small door at the side of the apparatus. This probably affected the ignitability of the material, because the immediate area may have been flushed with present atmospheric levels (PAL) of O2. In addition, the moisture content of the wood dowels was between 0 and 2%, which we consider unrealistic for a natural system. The fuel-moisture content above which a fire will not spread ranges between 15 and 30% for temperate ecosystems (10) and has been suggested to be similar for tropical ecosystems (11). Foliar moisture contents of 1-year-old conifers (Douglas fir, grand fir, and ponderosa pine) during summer (North America), when the atmosphere has a relative humidity of 18 to 35% (dry), are 128 to 147% (12). In addition, pine needles in the same combustion experiments of the same moisture content did not burn until 16% O2, and pine dowels of 12% moisture failed to ignite at 12% O2. The average moisture contents of the P. caribaea wood, the matches, and the moss in our combustion experiments were 7.31%, 24.8%, and 48.3%, respectively, revealing that even at relatively low moisture contents, natural material cannot be ignited in a fully sealed atmospheric O2 environment below 15% O2. Furthermore, although the major cause of prehuman wildfires was lightning (13), the majority of lightning strikes do not initiate burning because they tend to be accompanied or followed closely by rain (14). During such circumstances, fuel moisture may be expected to rise beyond the fiber-saturation point (30%), making ignition, even under PAL of O2, virtually impossible (14). On the basis of our data, together with litter and vegetation moisture contents in a natural system (11), we argue that 17% O2 is likely to be a more realistic lower limit for the occurrence of natural wildfires, confirming the suggestions of Lenton and Watson (15); however, we have set the lower limit for combustion at 15%, which allows for wildfires to occur under exceptional circumstances.

Published records of wildfire-derived charcoal, inertinite, and geochemical proxies (polycyclic aromatic hydrocarbons) were collated from the literature on the Mesozoic to test whether atmospheric O2 could have dropped below 15% for prolonged intervals. We only include records in which the geological age of the stratigraphic units containing the wildfire evidence is known (for example, the Aptian stage) and in which the remains were considered not to have been substantially reworked (table S1). The occurrences of wildfire events throughout the Mesozoic were plotted against the predicted atmospheric O2 curves of Berner (2), Falkowski et al. (3), and Bergman et al. (16) (Fig. 2 and table S1). This analysis revealed that wildfires were prevalent throughout the Mesozoic and, coupled with data from our combustion experiments, did not support model-based predictions of low O2 (<15% O2) for the Triassic (3) or Jurassic (2) (Fig. 2). Limited occurrence of fire proxy data for the intervals 250 to 238 Ma and 236 to 222 Ma could potentially indicate a period of <15% O2. However, a detailed analysis of rock abundance and the availability of suitable strata to preserve charcoal (fig. S1, A and B) indicates that this apparent wildfire gap probably represents a preservational artifact rather than an indication of extremely low O2 levels.

Fig. 2.

Modeled O2 curves for the Mesozoic compared with the record of paleowildfires. Vertical black lines represent known occurrences of fires in the fossil record (table S1). O2 curves are replotted using data from the models of Berner (GEOCARBSULF) (2), Falkowski et al. (3), and Bergman et al. (COPSE) (16).

The various atmospheric O2 models use different input parameters that result in differing O2 records for the Mesozoic. The GEOCARBSULF (2) atmospheric O2 model is calculated using the difference in O2 input from organic carbon and pyrite sulfur burial and O2 removal by the weathering of organic carbon and pyrite sulfur in rocks, plus the oxidation of reduced carbon and sulfur gases emitted by volcanism and metamorphism. The Falkowski et al. (3) model is based on isotopic records for organic carbon, carbonates, and sulfates and on an earlier version of GEOCARBSULF [GEOCARB III (17)]. Both the Falkowski et al. (3) and GEOCARBSULF (2) models use δ13C data as a strong input forcing. The COPSE model for O2 (16) includes C, O, P, and S cycles, partial cycles for N and Ca, and also biological and geological processes (tectonic uplift; weathering; volcanic and metamorphic degassing; terrestrial and marine productivity; burial of marine and terrestrial organic matter; and marine inorganic burial). On the basis of our newly proposed low limit for combustion, both the Falkowski et al. (3) and GEOCARBSULF (2) models are currently incompatible with the record of fires in the Mesozoic, because they predict extensive periods of low (10 to 12%) atmospheric O2 throughout the Triassic (3) and Jurassic (2). Our experimental burn data (Fig. 1), coupled with the newly compiled Mesozoic wildfire record (Fig. 2) are consistent with the Mesozoic O2 record predicted by the COPSE model (16). This model, unlike GEOCARBSULF (2) and the model by Falkowski et al. (3), includes fire feedbacks and nutrient cycling and accounts for their controls on terrestrial productivity. These are known to affect weathering and organic carbon burial rates (18), which in turn play a role in regulating O2. Future models of paleoatmospheric O2 should strongly consider the effects of changing terrestrial and marine productivity, including fire feedbacks.

We have shown that extensive periods of low O2 (<15%) cannot have occurred in the Mesozoic, according to our revised lower limit for combustion coupled with the record of paleowildfires. This also suggests that the predicted low O2 (<13%) levels for the Frasnian (385 to 374 Ma) in the Palaeozoic (19, 20) needs reevaluation. The paleowildfire record provides a key means for testing low-O2 events in the geological record and highlights the need for high-resolution studies of paleowildfire across major mass-extinction events in order to test current hypotheses that advocate a primary role of short-term low-atmospheric-O2 events in catastrophic faunal diversity loss in the Permian-Triassic (4) and Triassic-Jurassic (21) mass-extinction events.

Supporting Online Material

Materials and Methods

Fig. S1

Table S1


References and Notes

View Abstract

Navigate This Article