An early productive ocean unfit for aerobics.

G rasping the big picture of the early ocean used to be easy: no oxygen, and then oxygen. The crux of that popular model hinged on the almost universally accepted Great Oxidation Event, or GOE, during which appreciable free oxygen (O2) first accumulated in the atmosphere about 2.4 billion years ago. At the same time, as the old argument goes, the O2-free, iron-rich earlier ocean gave way to one with oxygen at all depths. About a decade ago, Canfield (1) offered a very different possibility—that ventilation of the deep ocean lagged behind the GOE by more than a billion years, resulting in a vast, deep reservoir of hydrogen sulfide, but long-held presumptions about photosynthetic life in the surface waters remained untouched. In the first comprehensive biogeochemical model of this ‘‘Canfield Ocean,’’ Johnston et al. (2) in a recent issue of PNAS present a stunningly different take on those early photosynthesizers—one in which the upper, light-containing layers indeed drove biological production but without the expected concomitant release of oxygen. And it is this feedback that may explain a troubling uncertainty about the Canfield Ocean and this time interval in general— exactly how oxygen in the biosphere remained at only a fraction of modern levels for so long after the GOE. Populations of photosynthesizing phytoplankton abound in the sunlit portions of the modern surface ocean, often to depths of 100 m or more. Using energy from the sun, they represent the dominant pathway by which carbon dioxide is fixed in the ocean and are the foundation (the primary producers) of the food chain. And O2 is their famous waste product. Much of the biomass formed in the surface ocean is degraded by aerobic bacteria on the spot, consuming the oxygen at a rate almost as fast as the gas is produced. The rest settles to the deep ocean, where the O2-consuming respiration continues. This shallow-to-deep flow of material is known as the biological pump. Canfield (1) speculated on how an analogous biological pump and associated oxygen loss in the deep ocean might have operated in the early world, after the GOE but well before the rise of animals about 700 million years ago—during a time interval known as the Proterozoic. There are two key pieces in this puzzle. First, because O2 levels in the atmosphere, in general, were still only a fraction of today’s, Canfield could confidently infer similarly lower concentrations in the ocean, given the exchange between the two. Add in the O2 consumption facilitated by a biological pump assumed to be robust, and anoxic conditions may have been pervasive throughout the deep global ocean. Although O2 remained low or absent in the deep ocean, post-GOE levels in the atmosphere were sufficiently high (it doesn’t take much) to oxidize the mineral pyrite (FeS2) during weathering on the continents, leading to sulfate (SO4 2 ) delivery to the ocean by rivers. This new flux of sulfate is key to the story because as it accumulated in the still anoxic deep ocean, sulfate-loving, O2loathing bacteria rereduced the sulfur to hydrogen sulfide (H2S), driving the solubility of dissolved iron in seawater to near null levels through pyrite formation rather than iron oxidation. This loss of iron dealt a deathblow to the vast banded iron formations (BIFs)—the smoking gun for the earlier ferruginous ocean. In other words, Canfield (1) argued that H2S began to build up throughout much, if not all, of the deep ocean beginning 1.8 billion years ago and persisted ubiquitously for a billion or more years. The Canfield Ocean is nothing short of paradigm-busting, and evidence for euxinic (anoxic and sulfidic) conditions in the Proterozoic ocean has emerged over the past decade (refs. 3–7, reviewed in ref. 8). In fact, confidence in this model has inspired a new generation of research exploring the potential biological implications of global euxinia, such as decreased availability of bioessential trace metals (9). Molybdenum and copper, for example, are less soluble in the presence of H2S and play key enzymatic roles in the nitrogen cycle, including N2 fixation by cyanobacteria (7, 9, 10). Enter Johnston et al. (2). Astutely, the authors recognized that a challenge to the Canfield model is maintaining oxygen deficiency over such a long period. The problem: a tiny but critically important percentage of sinking biomass escapes decay in the deep ocean and underlying sediments and is buried, typically yielding a net increase in atmospheric and seawater O2. And a strong negative feedback is in place because the anoxia induced by organic degradation actually increases the fraction of organic matter that escapes decay. As is often the case, old rocks make more sense when viewed through a lens focused on the modern ocean. After having such a look, Johnston et al. (2) asked a key question: what if the primary production in the surface ocean that settles and decays to facilitate O2 loss in the deep ocean was triggered by a photosynthetic pathway that does not produce oxygen? Today, we only have to look to the Black Sea and many stratified lakes to see that such anoxygenic photosynthesis does occur and can even dominate primary production, at least in lakes. There are two key ingredients to such systems. First, the subsurface waters in