On the rise of oxygen

Two-step rise of atmospheric oxygen linked to the growth of continents

Cin-Ty A. Lee, Laurence Y. Yeung, N. Ryan McKenzie, Yusuke Yokoyama, Kazumi Ozaki & Adrian Lenardic

Nature Geoscience 9,417–424 (2016); doi:10.1038/ngeo2707

Our paper on the rise of atmospheric oxygen has come out in Nature Geoscience.  You can access it from the NG website here or you can get it from my webpage at http://www.cintylee.org/#/publications/

This paper is an indirect byproduct of our work on the long-term whole-Earth carbon cycle (arc2climate.org).  Several years ago, we found ourselves working on skarns, which are the decarbonated end-products of the interactions between magmas and limestones. We hypothesized that these skarn-forming processes might be common during continental arc volcanism, where subduction zone magmas penetrate overlying continental lithosphere, interacting with ancient limestones stored on the continents. Such skarn-forming interactions could enhance the production of carbon dioxide out of volcanoes, so we hypothesized that global flare-ups in continental arcs might be correlated with higher volcanic degassing and possibly greenhouse conditions. Since then, a number of us at Rice, along with collaborators at the University of Texas and Pomona College have been trying to quantify how much carbon comes out of continental arcs.  We have also been working with other collaborators, like Ryan McKenzie, to tease out signals of continental arc magmatism from detrital zircons, under the premise that zircons preferentially sample felsic magmas, which preferentially sample continental arcs.  The detrital zircon work suggests that continental arc magmatism may wax and wane on a global scale and that during continental arc intervals, greenhouse climates operate, and during lulls, icehouse climates operate.  We are currently developing an independent test by compiling geologic maps so that we can map out continental arc distribution more directly through time.

But back to our oxygen paper.  All of this work on carbon started us thinking first about the possibility that Earth's long-term carbon cycle has never been at steady state. Surely, the amount of carbon stored in the continents has grown since the Archean.  Carbon dioxide leaks out of the Earth's interior through volcanoes and into the atmosphere. The carbon dioxide comes back out of the atmosphere by the precipitation of carbonates (through an intermediary step of chemical weathering) and organic carbon, the latter mediated by life of course.  But most of this precipitated carbon was deposited in shallow seas, on the margins of continents.  And because continents tend not to subduct and get recycled back into the deep Earth, what this means is that there is a nearly one way ticket out of the Earth's interior for carbon.  The crustal carbon available for metamorphic reactions and interaction with magmas has likely increased with time.

And this is where it gets interesting.  Production of molecular oxygen occurs primarily by oxygenic photosynthesis, where carbon dioxide reacts with water, driven by the sun's energy, to generate hydrocarbons and molecular oxygen as a by-product.  So the production of oxygen is intimately tied to the cycling of carbon.  The more metamorphic and volcanic degassing of carbon dioxide, the more carbon available to make oxygen, and the higher the production of oxygen.  All other things being equal, such as the efficiency of oxygen sinks (oxidative weathering or oxidation of volcanic gases), oxygen levels would rise.

This is how we came to think about oxygen.  Today's atmosphere has approximately 20% molecular oxygen, the rest being mostly molecular nitrogen.  We are the only planet in our solar system that has free molecular oxygen in its atmosphere.  On all planets, oxygen, being so reactive (it is highly electronegative), will bond with whatever is round.  On Earth, that means oxygen is bounded to metals to form oxides, essentially rocks.  Fifty percent of the Earth's mantle is in fact oxygen!  On more gaseous planets, oxygen might be bound up with hydrogen, nitrogen or carbon.

As it turns out, the oxygenated atmosphere we take for granted, you and me, actually arrived late in Earth's history.  Based on a variety of indirect tracers of atmospheric oxygen levels, it is widely agreed that Earth started off with no oxygen in the atmosphere. Then around 2.3 billion years ago, oxygen rose rapidly, but levels for the next billion years were still too low to sustain the complex animal life that requires respiration to survive.  The second and last major rise of oxygen appears to have occurred somewhere around 600 million years ago although the exact timing is still debated.

In our paper, we show that a growing crustal carbonate reservoir naturally leads to a gradual but very late rise in atmospheric oxygen. Since continents have been around for a very long time, the growth of a crustal carbonate reservoir is almost given, but it takes time to build to the critical mass beyond which production of oxygen reaches allows for runaway rise to a new steady state, that characterizing the last 500 million years or today, for that matter.

When we realized this link to continents, through the whole Earth carbon cycle, we were intrigued.  A close look at the detrital zircon chronology shows that most zircon populations occur after 2.7 billion years ago.  In fact, the first pulse of zircons occurs around 2.7 billion years ago.  And as mentioned above, zircons preferentially sample felsic rocks. To generate large amounts of felsic rocks requires significant amounts of water in the parental magmas, and the only way to do this is to recycle surface water back down into the mantle, presumably by subduction, and if so, this would imply that plate tectonics initiated at that point. There may be other ways to get water-bearing parental magmas without plate tectonics, but what is clear is that the first arrival of felsic crust, with characteristics like the continental crust we know today, appeared then for the first time.  From this point on, we started to accumulate carbon on the continents.

The appearance of felsic crust, however, had another effect. It is widely known that the total Fe and S contents of felsic rocks is substantially reduced compared to primitive basaltic magmas.  This is due to fractionation of Fe-rich phases into cumulates, along with the rapid decrease in sulfide solubility.  Ferrous iron and sulfide (S2-) are the primary sinks for oxygen.  While oxygen was being produced by photosynthesis, these reduced components react with the oxygen and prevent oxygen from building up.  For a given production, it is the efficiency of the oxygen sink that dictates atmospheric oxygen levels.  Prior to 2.7 billion years ago, the surface of the Earth was largely basaltic, with copious amounts of ferrous iron and sulfide, so the oxygen sink was very efficient.  After 2.7 billion years ago, with the conversion from mafic to felsic juvenile crusts, this sink was substantially reduced, allowing for oxygen levels to rise.

The way I think about this is through a leaky bathtub analogy.  The water level in a leaky bathtub is controlled by the magnitude of the input (how much water is flowing through the faucet) and the efficiency of the sink, that is, the size of the leaking sink.  The atmosphere is the bathtub and atmospheric oxygen is analogous to the water in the bathtub. The first rise in atmospheric oxygen occurred because we closed down the sink, allowing the atmospheric levels of oxygen to rise to another steady state.  The second rise of atmospheric oxygen occurred by increasing the production, that is, increasing the faucet of carbon dioxide and by implication oxygen. There's a little more to this than this simple analogy, but for now, it is probably not worth getting into all the complicated details.

So what next?  The astute reader will have seen that we make some clear and falsifiable predictions. We are not saying that our hypothesis must be correct, but that if we think there is life, plate tectonics and continents, we expect atmospheric oxygen to rise, in two steps.  There are many other hypotheses for the rise of oxygen. Our model does not negate any of these, but requires that all these other models must operate on top of this long term secular trend in carbon and oxygen cycling.

Where are we going from here?  I'm not sure, but one of the most exciting problems lying ahead is to quantify how much of the continents were above sea level and how this has changed with time.  It's also important to quantify where most of chemical weathering is occurring, in the lowlands or in the active mountain areas of mountain building.  These are all relevant to our understanding of the long term carbon and oxygen cycles.