Oxygen

by Nick Lane

  Unfortunately, ice-core data can take us back only about 200 000 years, before which the ice runs out. We have gone just 0.0007 per cent of the distance.

  Oxygen and the Rise of the Giants • 87

  The situation seemed hopeless until the mid 1980s, when Gary Landis, a geochemist at the US Geological Survey in Denver, had a clever idea. Tiny bubbles trapped in amber might conceivably contain ancient air, which had once dissolved in the resin of trees and later formed pressured bubbles, as the resin hardened to form amber. As luck would have it, Landis had the right equipment for the job: a quadrupole mass spectrometer, recently designed by the US Geological Survey to analyse the amounts and chemical identities of gases in tiny samples. The instrument is sensitive enough to detect gases at concentrations as low as 8

  parts in a billion, and quick enough to analyse samples released in milliseconds from bubbles as small as a hundredth of a millimetre (10

  micrometres) in diameter.

  Amber jewellery, containing embalmed insects and spores, has been highly prized ever since Neolithic times. Trade in Baltic amber is known to stretch back at least 5000 years. The amber itself ranges in age from Carboniferous (300 million years old) to Pleistocene (the period of the last major ice age) so many trapped insects are in fact very ancient. The reputation of amber as a time capsule reached its apotheosis in the notion that dinosaur genes might survive intact in the abdomens of blood-sucking insects that were engulfed in resin soon after their meal, an idea made famous by the novelist Michael Crichton in Jurassic Park. Although controversial, the idea did have sufficient scientific merit to attract serious attention, and ancient DNA has indeed been isolated from insects in amber dating from the Cretaceous (140 million years ago). If flimsy DNA molecules could survive, why not air?

  Working with Robert Berner, Landis used his quadrupole mass spectrometer to detect the gases trapped in air bubbles by mechanically crush-ing the amber in a vacuum. To piece together a time line, Berner and Landis ground up amber samples that dated to successive geological periods, from the Cretaceous (140 million years ago) to modern times.

  There was, of course, a danger that the different types of amber would produce a range of values simply because they were different types of amber. In an attempt to exclude the possibility that the measurements reflected only local conditions, Berner and Landis used amber from a variety of sources, from the Baltic States to the Dominican Republic, and from open beaches to buried strata.

  Their results were published in the journal Science in March 1988 and caused an immediate stir. The data implied that oxygen levels had been higher than 30 per cent during the Cretaceous, falling to our modern level

  88 • THE BOLSOVER DRAGONFLY

  of 21 per cent around 65 million years ago, a time that corresponded a little too closely for coincidence to the mass extinction of the dinosaurs.

  Might it be the case, Berner and Landis wondered, that, like the dragonflies, the dinosaurs needed high oxygen levels to achieve their giant size, and could not survive in our thin modern atmospheres? By August of that year, the letters pages of Science were filled with detailed technical criticisms.

  There are few better forums for a display of erudition than letters pages, and on a good day the top scientific journals extend this tradition into a marvellous parade of eclectic learning. No academic exchange is more chastening to an erring author, or more entertaining to an interested onlooker. The amber results were subjected to scrutiny from just about every angle, from the diffusion constants and solubility ratios of matrix gases to the unusual chemistry of fractured polymers and the geometrical behaviour of pressurized bubbles. Curt Beck of the Amber Research Laboratory in New York, for example, drew attention to the methods by which the Romans restored translucence to milky (or bone) amber. Bone amber is rendered opaque by the presence of microscopic air bubbles, but can be made transparent, as well as dyed, by heating it in oil.

  The Romans apparently used the fat of suckling pigs, whereas the great nineteenth-century German authority, Dahms, recommended rapeseed oil. The method works because the air bubbles become filled with the oil, which has the same refractive index as the amber. This means that the oil can penetrate the amber matrix completely. That is, the bubbles are not in fact entirely isolated from the environment, and so the air inside them can exchange with air from outside, and thus would not accurately reflect the composition of the air at the time they were formed.

  The overall consensus, which has now persisted for more than a decade, was that the air trapped in amber could not have been ancient, and that the results of Berner and Landis must be artefacts of inappropriate methodology. While Landis claims to have refuted his critics in a new round of experiments, his compendious data are not yet published in any detail, and so have not been widely accepted by the geological community. Berner himself acknowledges that their preliminary data were flawed, and concedes, with a rather charming candour, that Landis has so far failed to persuade him that their critics were wrong.

  Oxygen and the Rise of the Giants • 89

  In rushing to denounce the amber results, one or two critics claimed that the old status quo had been restored, and that the “previously held views of palaeontology, geology and atmospheric science are still intact”. Berner and Landis retaliated that these ‘previously held views’ were not at all intact, but were being revised independently of the amber studies. This brisk exchange seems to me to stand at the heart of the debate, and reveals quite a lot about the nature of scientific enquiry, which is very rarely the kind of dispassionate induction process that philosophers like to call the ‘scientific method’. Most scientists instead cling to their favourite stories, or hypotheses, until they are either proved, or discredited to the point that even an obstinate old professor must concede their fallacy. The two sides here were entrenching themselves for battle, with a broad stretch of no-man’s land separating the advocates of an unchanging atmosphere from the campaigners for high-oxygen air — a situation reminiscent of the squabbles that once took place in physics between the exponents of a steady-state universe and the big bang theory.

  Clearly, Berner’s and Canfield’s atmospheric model would convince some geologists but not others, and only fresh data could point to a way out of the impasse. If amber does not contain ancient air, it is hard to think of anything else that might. Is there another way of corroborating the model’s results?

  Well, ye-ess, is the best answer: there is another way, but it has a few problems of its own. As we saw in Chapters 3 and 4, carbon-isotope ratios can be used to measure changes in oxygen levels. The idea is more or less the reverse side of the coin to measuring carbon burial directly, and the problem has always been to make the opposite sides of the coin correspond to each other.

  The isotope method hinges on the fact that life prefers the light carbon isotope, carbon-12. Organic matter therefore becomes enriched in carbon-12. When organic matter is buried, relatively more carbon-12

  is buried and more carbon-13 is left behind in the air as carbon dioxide.

  The carbon dioxide in the air exchanges freely with carbonates in the oceans, as well as those in swamps, lakes and rivers; the whole world is interconnected. In shallow seas, the dissolved carbonates may in turn precipitate as carbonate rocks — marine limestones. Because of the equilibrium between carbonates in the seas and swamps and carbon dioxide in the air, periods of high carbon burial on land leave a strong carbon-13

  signature in marine limestones. These signatures can be extrapolated backwards to calculate how much organic carbon must have been

  90 • THE BOLSOVER DRAGONFLY

  buried.5 The advantage of the isotope method is that it paints a picture of organic burial based on changes in the ocean pool. This gives an idea of the total global rate of burial, as, like salt, carbonate is dispersed by the tides and currents at a roughly equal concentration throughout the oceans.

  There can be little doubt from carbon-isotope measurements that the rate o
f burial of organic matter did vary substantially over geological time, with a large peak during the Carboniferous and early Permian. The problem now, as before, is how to constrain the predicted changes in oxygen levels within reasonable limits. The difficulty with estimates based on isotope evidence is the extreme sensitivity of the predicted atmospheric oxygen levels to small changes in the predicted rate of burial. We noted in Chapter 2 that the amount of buried organic matter dwarfs the organic content of the living world, by a factor of 26 000 according to Robert Berner. This means that a very small change in the calculated rate of burial (from isotope ratios), sustained over a period of millions of years, can result in extreme variations in calculated oxygen levels — shifts that are incompatible with life, and so cannot really have happened. Some kind of feedback mechanism must have constrained these variations —

  the trouble is, we don’t know what it is.

  After testing numerous possibilities in his atmospheric models, with what must have been a dispiriting lack of success, Berner finally had an ingenious idea that was open to experimental verification. What if life’s preference for carbon-12 itself varied with oxygen levels? In other words, what if the degree of carbon-12 enrichment in organic matter varied with the amount of oxygen in the air? When we measure carbon burial, we assume that a fixed proportion was buried as carbon-12. If we then detect more carbon-12 in buried carbon (or more carbon-13 in limestone), we take this to mean that there was an increase in the rate of burial. However, it could also mean that there was a rise in the proportion of carbon-12

  buried. If this happened, the total amount of buried carbon might remain 5 A similar argument applies to sulphate-reducing bacteria, which discriminate between the two sulphur isotopes, sulphur-32 and sulphur-34. Burial of iron pyrites resulting from the activities of sulphate-reducing bacteria raises oxygen levels because the organic matter used as a carbon source by the bacteria is not completely oxidized. Strong sulphur-34 signatures in sulphate evaporites, such as gypsum, correspond to a high burial of sulphur-32 in iron pyrites. The studies discussed in the main text do take sulphur isotopes into consideration but a detailed exposition here runs the risk of losing any remaining readers. Berner himself writes clearly and engagingly on the subject, so for those who want more detail, I recommend his primary papers, listed in Further Reading.

  Oxygen and the Rise of the Giants • 91

  constant, but the amount of carbon-12 measured in it could still increase.

  If we then misguidedly applied the law of fixed proportion, we would have an exaggerated impression of the total amount of carbon that had actually been buried. Any mechanism that makes plants use more carbon-12

  than normal could have this effect. Correcting for the distortion would decrease the amount of buried carbon estimated by the model and so blunt the predicted variations in oxygen. In other words, the impossibly extreme fluctuations in atmospheric oxygen predicted by the unmodified isotope method could be made more realistic by taking into consideration the selectivity of plants for carbon-12 at different oxygen levels. If plants became more selective for carbon-12 at high oxygen levels, and less selective at low oxygen levels, the predicted fluctuations in atmospheric oxygen would be constrained within more reasonable bounds.

  Convinced? Perhaps not, but in theory our old friend photorespiration could have exactly this effect. The carbon dioxide produced by photorespiration is released into the leaf spaces and may then either diffuse out of the leaf, or be taken up a second time by Rubisco and converted back into sugars, proteins and fats. Because the carbon dioxide released by photorespiration is formed from organic matter, it is already enriched in carbon-12. When subjected to a second round of discrimination, the newly made organic matter will be even more enriched in carbon-12. We noted a similar effect with sulphate bacteria in Chapter 4, analogous to re-breathing air in a plastic bag. The rate at which this super-enrichment occurs depends on the rate of photorespiration — and this, as we have seen, increases with rising oxygen levels. In theory, then, high oxygen levels should increase life’s preference for carbon-12 and distort our calculations of atmospheric oxygen. The theory seems to fit; but do the numbers add up?

  Berner teamed up with specialists from the University of Sheffield in England, including David Beerling, and the University of Hawaii. They chose photosynthesizers from diverse evolutionary groups including angiosperms, cycads and marine single-celled algae, and grew them in the laboratory under different oxygen concentrations. Their findings were published in Science in March 2000 and are an amazingly close fit to the wildest dreams of the theorists. All the species they tested responded to increasing oxygen levels by becoming more selective for carbon-12.

  Plants that had evolved during the Carboniferous or early Permian, however, showed the greatest selectivity. Cycads, for example, left behind an excess of 17.9 parts per thousand carbon-13 in the air at 21 per cent

  92 • THE BOLSOVER DRAGONFLY

  oxygen, but when oxygen levels were raised to 35 per cent, this rose to 21.1 parts per thousand. This is a relative increase in carbon-12 enrichment of 18 per cent. With fewer stomata in their leaves, the plastic-bag effect was enhanced. When this figure is taken into consideration in the isotope calculations, we find a very close match between the two sides of the coin: between the calculations based on carbon burial directly (mass balance), and those based on carbon isotopes. Both independent calculations indicate that oxygen levels reached 35 per cent during the Carboniferous. Berner had finally balanced his books.

  These results do not prove conclusively that the air in Carboniferous and early Permian times was rich in oxygen, but they do place the boot firmly on the other foot: it is now up to those who disbelieve that such changes could have taken place to produce strong empirical evidence supporting their case. In the meantime, an expanding party of researchers from other fields have taken the high-oxygen case at face value and started to ask pertinent questions about the likely implications. Many of these implications can be examined directly by a fresh look at the fossil record, or by measurements of physiological performance in high-oxygen atmospheres, such as dragonfly flight. But first, we still have a paradox on our hands: fire. Shouldn’t everything burst spontaneously into flames at such high oxygen levels? What about the runaway catastrophe of burning forests that we discussed earlier?

  One of the difficulties that scientists face today is the sheer breadth of knowledge. Even within a particular field, such as medical research, it is hard to keep abreast of new research, while spending most working hours at the bench or in the clinic. Researchers typically have a minutely detailed knowledge of their own immediate discipline, for example population genetics, while maintaining enough of a broad understanding to weigh up the importance of developments in related subjects, such as molecular biology. Once we get further afield, however, scientists, like everyone else, just have to accept a lot on trust. Fire gives an idea of just how deeply a concept can become embedded in scientific lore without anybody questioning its experimental basis.

  During the 1970s, Lovelock and Watson argued that “above 25 per cent oxygen very little of our present land vegetation could survive the raging conflagration which would destroy tropical rain forest and arctic

  Oxygen and the Rise of the Giants • 93

  tundra alike”, and that even wet vegetation would have “a significant probability of ignition… a fire could be kindled even during a rainstorm.”

  To back such strong statements, we may be forgiven for assuming that experiments must have been done, and the data filed away long ago as established fact; this, at least, was my assumption, until I became sufficiently troubled by the story to look up the original experiments. The work had indeed been done: Andrew Watson, then a graduate student of Lovelock’s, devoted his 1978 PhD thesis to a detailed examination of burning in different oxygen atmospheres. Unfortunately, somewhere along the way his conclusions were stripped of their context.

  To control the parameters of his exp
eriments, allowing him to compare like with like, Watson worked mostly with paper strips. These he moistened to varying degrees and ignited. He went on to stage hundreds of small burns under conditions of controlled oxygen and moisture content, and drew curves for the probability of ignition by electrical discharge, the rate of fire spread, and the amount of water needed to extinguish the flames. The results confirmed our intuitive judgement that high oxygen intensifies fire and counteracts the dampening effect of moisture.

  There is nothing wrong with these results. The problem is what they left unsaid, as Watson himself acknowledges. Critically, paper is a poor surrogate for the biosphere, as anyone who has attempted to light a wood fire using newspaper knows. As we noted in Chapter 4, paper manufacture involves the removal of most of the lignin, and this increases its flammability. Lignin does not burn well, but instead tends to smoulder gently.

  Trees with a high lignin content in their bark are relatively fire resistant.

  Nor can paper retain water by osmosis as living cells do. The moisture content of comparably thin plant tissues, such as leaves, is therefore much higher than paper. While Watson measured the flammability of paper up to a maximum of 80 per cent moisture saturation, some leaves approach an equivalent of 300 per cent saturation. Plants at high risk of fire may also accumulate high levels of fire retardants, such as silica. Some straws, for example, have an unusually high content of silica, which can make it hard to burn agricultural waste. Housewives apparently discovered this trick long ago: silicate paints were commonly applied to curtains during the Second World War to retard the spread of fire in the aftermath of bombing.