Introduction

A decade has passed since the review of Lyons et al.1, describing the evolution of Earth’s atmosphere and oceans. Lyons et al.1 produced a seminal figure (that has more recently been updated2), colloquially termed the ‘Lyons curve’, which has been reproduced en masse, and acts as a reference frame, especially for understanding the broad trajectory of environmental conditions during the rise of complex life. The past 10 years have witnessed a rapid increase in the resolution (both spatial and temporal) of geochemical and palaeontological data, refinements and innovations in geochemical methodologies and chronostratigraphic calibration, and nuanced understanding of various geochemical systems and microbial ecology. However, a broad array of publications on the topic reveals that researchers are not always in agreement about the timing of oxygenation and its impact on the biosphere. Scientific consensus, or at least some degree of common and measured perspective of the scientific community on specific aspects of rapidly developing fields, has been identified to benefit both the researchers within those and adjacent fields, as well as science communication3,4.

Atmospheric oxygen concentration is generally understood to have undergone three broad transitions. The Great Oxidation Event - or Episode - (GOE; ~2.4–2.2 billion years ago, Ga5,6), marks the first appreciable rise in atmospheric oxygen. This transition is relatively well defined and constrained by the loss of mass independent fractionation of sulfur isotopes (MIF-S7,8). The loss of MIF-S signals an increase in atmospheric oxygen above a threshold concentration of 10−6 of the present atmospheric level (PAL)9,10, although this value is often quoted as 10−5 PAL after an earlier model11. The GOE is widely considered to result from an imbalance between the primary source of O2, (e.g. oxygenic photosynthesis) and a reducing sink of O2, (e.g. reduced gases9) the evidence for which process is otherwise poorly constrained by the fossil/rock record12,13. Following a period of apparent stability through the late Paleo-Mesoproterozoic, the Neoproterozoic Oxygenation Event (NOE; ~800–540 Ma14) is characterised by a dynamic oxygen trajectory and broadly coincided with the diversification of eukaryotic life in the late Tonian Period and rise of animal life in the Ediacaran Period15,16. The variability recorded by geochemical proxy records across this interval has led to modelling estimates of atmospheric oxygen between 1 and 50% PAL, and dynamic oxygenation and deoxygenation events that argue against the NOE being a simple unidirectional O2 increase17. Similar to the GOE, the NOE is also characterised by dramatic swings in global climate with multi-million year Snowball Earth events when ice sheets extended to low latitudes18. Uncertainty in the extent of oxygenation following the NOE is still highly debated, with notable spatial and temporal data variability throughout the early Phanerozoic19. The third stepwise event, the Paleozoic Oxygenation Event (POE; ~450–350 Ma20,21), is associated with the rise of the terrestrial biosphere during the mid-Paleozoic22,23, and culminated in the oxygenation of the deep ocean and an oxygenated world more reminiscent of the modern day.

Intertwined with the history of oxygenation is the rise and diversification of complex life. Broadly, molecular clocks suggest eukaryotes emerged after the GOE24 and diversified across the NOE interval25,26. However, it remains unclear to what extent early eukaryotes and early animals relied on O2 availability or stability27,28,29, as well as the cause and effect relationship between Earth’s oxygenation and the trajectory of eukaryotic evolution30,31,32. Even within this brief summary, we note several instances where uncertainties result in wide error envelopes when describing the history of oxygen and life on Earth. Here, we present the results of a questionnaire in an attempt to illuminate some of the most important remaining uncertainties in the Earth Science community associated with the coevolution of life and Earth’s surface oxygenation. We investigate responses to this questionnaire with the aim of semi-quantitatively assessing avenues for future collaborative research that may further our understanding.

The use of expert elicitation in helping refine scientific uncertainty is not a new phenomenon33. For example, several studies have aimed to constrain uncertainty in parameters associated with climate change as a means to better direct efforts34,35,36. We acknowledge that the use of this method does not provide a definitive answer to any questions that we, as a community, are attempting to ask37,38, and we are not attempting to resolve a consensus for the trajectory life and environment have taken. Instead, we use this study to convey a unique type of uncertainty quantification that can help contextualise the coevolution of life and oxygen, in ways that will support a community-based research agenda.

Methods

To investigate the state of the perception of the interplay between life and environmental evolution, we disseminated a survey to academics that were considered experts in the field.

An expert in this sense is defined by having a first author peer-reviewed publication, as identified by Google Scholar within the 5 years prior to questionnaire release (June 2017–2022) or someone who was found to be a co-author on three or more related peer-reviewed manuscripts. Search terms in Google Scholar included oxygen, Proterozoic, life, and evolution, leading to a total of 231 independent researchers after thematic review to ensure the publications were relevant. To ensure as large a sample size as possible, reminders were sent after 3 months and then on a monthly basis. The questionnaire received 133 individual responses (58% response success) between June 2022 and November 2022 indicating a wide range of specialisms, career stage and global representation (Fig. S1). The majority (37%) of respondents identified as postdoctoral or early-career with US (42%), UK (18%), Canada (7%) and China (7%) garnering the largest contributing countries (Fig. S2). We acknowledge that the 133 respondents do not make up the entire community view and only represents a subset. We do however believe the wide variety of respondents based on their academic demographics (Fig. S2) provides an array of views.

Initial questions within the survey allowed respondents to only answer questions they deemed themselves to be experts in, by asking what intervals of Earth history they were better acquainted within as well as their general method of research, i.e. geochemistry (32%), field-based (27%), palaeobiology (18%), modelling (17%) and other (6%), with the possibility to overlap responses.

The survey was approved by the Yale University Institutional Review Board (Yale IRB) and included a risk statement as follows: “You will not directly benefit from participation in this research study. Your responses will not be linked to your name. Your participation in this study is completely voluntary and the limited amount of personal information collected will be dissociated with your answers prior to data analysis. We are required to inform you that the only risk the Yale IRB review has identified is a possible loss of confidentiality of your answers.” We also note that no questions were compulsory, leading to variable response numbers.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Results and discussion

Defining an oxygenated world

In order to constrain Earth System changes associated with environmental oxygen availability, a variety of proxies have been developed. These proxies are calibrated through careful laboratory-based investigations that determine specific redox sensitive chemical behaviours, alongside threshold calibrations derived from large datasets of modern and ancient depositional environments, where redox/palaeoredox state is either known, or can be estimated using independent data (e.g., presence/absence of fossil organisms inferred to have had high metabolic demands). After careful calibration, these proxies are then applied to suitable lithologies throughout the sedimentary rock record, and oxygen dynamics are reconstructed based on collations and collective interpretations of geochemical proxy data.

Herein, experts were asked what proxies they preferred to use to define changes to Earth’s surface oxygen concentrations through time (n = 49). Despite only 49 responses to this question, there are a total of 41 self-identified geochemists in the survey, meaning that this is likely a fair representation of experts in the field (Fig. 1). To produce Fig. 1, to ensure each respondent with one or more chosen proxies had a singular vote, weightings were applied where for each additional proxy their vote was partitioned in the figure; for example, if someone chose two proxies they had half a vote for each. Respondents who self-identified as having multiple disciplines also had their vote partitioned into the separate disciplines in Fig. 1 but this does not alter the overall total votes for the proxies. The raw data is available at https://doi.org/10.6084/m9.figshare.29591216.v1.

Fig. 1: Current and future proxy development.
figure 1

A Geochemical proxies for constraining oxygenation through time, with preference allocated by expert respondents. B Proxies that may require further development, based on votes by expert respondents. Numbered proxies are listed in full in the Supplementary Information.

Full size image

The top four “favourite” palaeoredox proxies were: MIF-S (>14% of responses), redox sensitive element enrichments (RSEs) (>12% of responses), cerium anomalies (>11% of responses), and iron (Fe) speciation (>7% of responses).

As noted above, the loss of MIF-S has been used as a metric for defining the timing of the GOE for over two decades7 and constrains the point at which atmospheric oxygen rose across the 10−6 PAL threshold10,11. Based on the clear mechanistic interpretation for the loss of MIF-S, and consistent support for this interpretation based on independent multi-proxy records, it is considered as one of the defining characteristics, not only of the GOE, but as a benchmark for Earth’s overall oxygen trajectory. Redox sensitive elements were also cited as a preferred palaeoredox proxy, and incorporate a suite of elements including molybdenum (Mo), vanadium (V), uranium (U), and rhenium (Re), whose absolute and relative enrichments (most commonly calculated as enrichment factors, or ‘EFs’, in fine grained siliciclastic rocks, e.g., shale) can be used to distinguish the dominant redox conditions of overlying waters that existed during deposition39,40,41. Cerium anomalies (Ce/Ce*), which also constitute one of the top four favourite proxies, are distinguished by enrichments or depletions in Ce relative to a specific rare Earth element (REE) profile20,42. Ce is readily adsorbed onto manganese oxides, and oxygen concentrations sufficient to oxidise reduced Mn can therefore lead to more efficient Ce removal from seawater, equating to a negative Ce/Ce* anomaly (relative to the REE profile), most commonly recorded in early marine carbonate cements or carbonate sediments20,42. The fourth favourite palaeoredox proxy was Fe speciation. The Fe speciation protocol for modern sediments and ancient sedimentary rocks (commonly shales) has remained largely unchanged for almost 20 years43, and is contingent upon the quantification of the proportion of total Fe (FeT) considered highly reactive (FeHR) to biological or abiological reduction under anoxic conditions. FeHR constitutes the sum of Fe in carbonates (Fecarb), oxides (Feox) and magnetite (Femag), which are operationally defined via a three step sequential extraction procedure, in addition to pyrite (Fepy), which is quantified via a separate extraction43,44. All proxy datasets require thorough screening to determine the degree to which post-depositional conditions may have skewed geochemical data, and numerous articles have outlined best practices in the production and interpretation of Fe speciation data45.

Experts were also asked what proxies they would like to see better developed in the future (n = 44). Responses to this question can be interpreted in several ways; either the expert considers their chosen proxy to require a more thorough calibration prior to widespread use (amounting to concern with regards to overinterpreting premature datasets or methodological/analytical uncertainties), or that the expert believes that an underappreciated proxy is sufficiently novel to warrant more concerted development and application, or both. The top proxies that experts wish to see better developed are: redox sensitive elements (RSEs) (>8% of responses), carbonate-bound iodine (>13%), mass independent fractionation of oxygen isotopes (>7%), trace metal isotopes (>6%), and oxygen requirements of biology (>4%).

Only RSEs are found on both the favourite and further development lists, which may be due to the wide range of elements that can be grouped into this term, and ongoing research associated with the redox sensitivities and behaviours of specific elements. Carbonate-bound iodine has recently shown promise as a means to constrain shallow water oxygen concentrations in deep time, given that oxidised iodate is the only species of iodine that can be readily incorporated into the carbonate lattice46,47. Mass independent fractionation of oxygen isotopes attempt to capture a transition in atmospheric pO2 driven by ozone formation48,49,50. Interpretations of O-MIF data have also been taken further, in attempts to constrain productivity dependence on pO2 and pCO2 estimates51. The umbrella term ‘trace metal isotopes’ encapsulates a variety of elemental systems that, to describe in full, would be beyond the scope of this manuscript. As an example, Cr isotopes have been used over the past decade to constrain atmospheric O2 concentrations, given the sensitivity of Cr isotope fractionation to oxidative weathering and manganese cycling52,53. Meanwhile, Mo isotopes can be used to constrain the extent of oceanic sulfidic conditions54 and U isotopes can be used as a means to quantify the extent of oxic and anoxic global seafloor55, due to the long residence time of these elements in the modern ocean. The variety of proxies that sit within the term ‘trace metal isotopes’ demonstrates that they must be extensively calibrated independently, but will provide more nuanced interpretations of depositional redox conditions when employed efficiently, and in concert. Additionally, research on the O2 requirements of animals may offer a more direct means by which to constrain oxygen concentrations in the water column, given that specific oxygen concentrations are required to sustain specific metabolisms. However, determining the precise oxygen requirements of these metabolisms demands both careful study under laboratory conditions28, as well as assumptions regarding the affinities and life habits of complex multicellular organisms preserved in the fossil record42.

Understanding the evidence across disciplines: is there a holistic view?

Pieces of information are often misinterpreted during interdisciplinary communication. The Precambrian fossil record is rife in enigmatic fossils, many of which suffer an uncertain phylogenetic placement or even questioned biogenicity56. A specific example of a recent palaeontological finding that has been publicly refuted is the reassessment of a putative fossil from India that superficially resembled the Ediacaran organism Dickinsonia57 as the decaying impression of a modern beehive58. However, published rebuttals are relatively uncommon, and in the majority of cases, uncertain or disputed fossils themselves tend to be largely ignored in subsequent scientific literature, without published critique. From the perspective of an exterior discipline (e.g., geochemistry), such lack of published critique may be interpreted as acceptance of the outdated perspective, especially if open cross-discipline communication is lacking. In sum, this highlights the necessity for active cross-discipline communication.

Expert respondents were asked to self-categorise their primary discipline, and then to suggest which pieces of respective geochemical and palaeontological evidence are most often misinterpreted or misused. Experts who identified as palaeobiologists repeatedly outlined several instances of misinterpretation of the fossil record with regards to understanding the coevolution of life and oxygen (n = 34). Occurrences of stromatolites are often misinterpreted as direct evidence for the presence of cyanobacteria and thus onset of oxygenic photosynthesis (highest proportion of responses, c. 20%). This confusion likely derives from concerns with affinities of stromatolites in deep time. Recent microbialites, built largely by prokaryotic and eukaryotic photosynthesizers, are not always suitable analogues to Archean and Proterozoic ones59 and the metabolisms of ancient microbialite- and stromatolite-builders are not limited to oxygenic photosynthesis60,61. A lack of, or poor, age constraints and/or established affinity of Precambrian fossils are the second most often misinterpreted palaeontological evidence (>17%), and may lead to confusion about the timing of the appearance of major lineages. This not only impacts evolutionary studies, but also our understanding of the coevolution of life and the changing environment. This is associated with the third most often misinterpreted palaeontological evidence; the interpretation of fossil biogenicity, especially of some Archean microfossils (>14%). This response is likely driven by recent experimental studies that show inorganic growth of sulfur biomorphs with similar morphologies to early fossil cells62. Beyond the Archean, the biogenicity of the Paleoproterozoic ‘Francevillian biota’63,64 are also specifically referred to (>14% of responses). As an indicator of common scientific perspective, these expert responses identify areas of contention in deep time palaeontological research that might not always be obvious to external disciplines.

Numerous geochemical data are also often misinterpreted, leading to an extensive list of proxies named by self-categorised experts in geochemistry (n = 47) including; carbon and oxygen isotopes (>13% of responses), molecular biomarkers (>11% of responses), iron speciation (>9% of responses), trace metal concentrations (>9% of responses), chromium, sulfur and metal isotopes (>23% of responses in sum) and the S-MIF record (>3% of responses). In addition to these, specific reference was made to the misinterpretation of local/regional proxies as being informative of global environmental change (>5% of responses). The diversity of proxies being reported either suggests poor intra-disciplinary communication or a common agreement that the majority of proxies require ongoing detailed study and continued calibration (e.g., incorporating novel laboratory or field-based insights). It is important to note, however, that a large proportion of oxygenation/deoxygenation events are supported by a combination of proxies rather than relying on a single one.

A further, broader question posed to respondents, asked what they deemed to be the most pressing questions in the field (total of 106 unique responses from 45 respondents). For example, one question that was commonly raised by respondent experts is how to better distinguish between local and global signatures, or ‘how spatially representative is my geochemical dataset?’ (>16%). Specific responses highlighted the need to clarify local vs. global environmental drivers of oxygenation, as well as the extent of oxygen oases, throughout the Proterozoic. Several respondents also acknowledged that limited geological material exists with which to clarify a global understanding. Another pressing question that was repeatedly raised in the responses is how best to improve our proxy calibrations (>16% of responses). Geochemical proxies used to constrain past environmental change rely on calibrations that rely on using observations in modern depositional environments or under controlled laboratory conditions, which attempt to recreate representative environments that may have existed in the geologic past. Given the likelihood of non-uniformitarian environmental conditions throughout much of Earth’s history, it is necessary to continue investigating how these proxies behave in as many different chemical environments or depositional settings as possible. Furthermore, given that some ocean chemistries in the rock record are not available to study in nature today, theoretical or experimental studies are also required to supplement interpretations based on proxy data. The final, and potentially most pressing question, as defined by numerous experts, concerns whether or not O2 availability actually limits eukaryotic evolution, or indeed the timing, and pace, of early animal evolution (>20% of responses). Current estimates of atmospheric O2 requirements for early eukaryotes are between 0.001% and 0.4% PAL28,29,65. While the oldest eukaryotic fossils are found in ~1.63 Ga66 sedimentary rocks, some propose O2 concentrations of this magnitude are thought to have been reached on Earth much later, sometime during the Neoproterozoic Era, further complicating potential linkages between environmental oxygenation and early evolution1,2 (Fig. 2). Additional pressing questions arose repeatedly, including the timing of oxygenation events (>10%), which require both better chronostratigraphic constraints as well as a more integrated use of geochemical proxies, and a more nuanced understanding of geochemical proxy interpretations. Importantly, each of these thematic questions still only received a maximum of 20% of responses, highlighting that the broad research community is approaching this prospect with a wide variety of methods.

Fig. 2: Summary of current understanding and intervals of time for future research.
figure 2

A Reported intervals of high priority according to experts (n = 37). Highest reported interval of priority occurs during the mid-late Tonian 800–720 Ma (timescale overlap of 14 individual responses). B Atmospheric oxygen throughout Earth history2. C Global carbon isotope record based on a preliminary compilation72.

Full size image

Future directions

This questionnaire offers a direct method of investigating and communicating a subsample of the current state of research and cross-disciplinary understanding in the Earth sciences. It also provides an opportunity for open communication and discussion regarding areas of future research that may advance, or challenge, the current paradigm. With this in mind, experts were asked what field-based, experimental and/or modelling efforts they believed would be most interesting or useful going forward, to better understand the evolution of environmental oxygenation and its importance for the biosphere (total of 80 unique responses from 38 respondents). The most frequently requested items are listed in the Supplementary Information, but we expand on the most noted three here.

The first recurring item calls for more isotope fractionation experiments, in order to better understand how isotope systems inform specific environmental changes (>8%). As noted previously, this reiterates the widespread view that ongoing geochemical proxy calibration is essential.

The second item calls for more accurate temporal calibration of the rock record, resulting in a higher precision temporal calibration of geochemical and fossil data (>13%). Specifically, it is widely recognised that more data are required constrain age models for the Paleoproterozoic to early Neoproterozoic. The resulting dataset would ideally permit the calibration of geochemical and palaeontological information from mixed lithologies within a unified chronostratigraphic framework, or series of possible frameworks, each of which would be anchored in time by absolute ages derived from high-precision radiometric methods. Exploring alternative correlation frameworks may also help to discriminate between the relative likelihoods of alternative palaeogeographic reconstructions in deep time. In particular, more high-precision radiometric constraints are needed for time intervals of high research priority as identified by the survey responses (Fig. 2).

The third item calls for greater efforts to pursue global syntheses that demand interdisciplinary collaboration (>6%). Numerous international initiatives that aim to collate sedimentary, geochemical and palaeontological data throughout Earth history are enhancing our ability to constrain aspects of the rock record with greater confidence over both longer and shorter timescales. These databases include the Sedimentary Geochemistry and Palaeoenvironments Project67, the North American MacroStrat database68, and the growing Deeptime Digital Earth initiative (DDE, formerly known as the Geobiodiversity Database69), each of which not only collate data, but also facilitate large dataset interpretations by providing novel user interfaces and statistical methods for database interrogation.

Experts were also asked what they believe to be the greatest remaining uncertainties in attempts to constrain the influence of evolving atmospheric and oceanic oxygen concentrations on life, throughout Earth history. The following items dominated responses, and summarise three key questions that captivate the research community, as a whole:

(1) What environmental O2 concentration is required to facilitate the evolution of animal life?

(2) What were the major mechanisms driving long-term changes in environmental O2?

(3) What were the upper and lower limits of environmental O2 through time?

Attempting to answer these questions demands interdisciplinary studies that interrogate large swathes of Earth history. However, the patchiness of the rock record in deep time often limits research during certain periods. With this in mind, experts were asked what intervals of the geologic record they believe to be priority targets for future sampling and investigation (Fig. 2). This revealed a notable interval of interest during the mid-Paleoproterozoic (c. 2.2–1.8 Ga), in the immediate aftermath of the GOE, encompassing the Lomagundi-Jatuli Event and the estimated origin of eukaryotes (per Betts et al.24). Other target intervals include the late Mesoproterozoic (c. 1.3–1.0 Ga), coinciding with the radiation of eukaryotic lineages and the origin of multicellularity70,71, and the mid-late Tonian Period (c. 0.8–0.72 Ga) during eukaryotic diversification and the inception of climatic instability associated with the onset of the Sturtian Snowball Earth26.

Overall, we hope this study promotes future interdisciplinary research and conversations that attempt to further our understanding of how the Earth became habitable for complex life. We believe that future conferences and workshops should continue to promote interdisciplinary research and conversations in order to address several of the topics highlighted herein. More specifically, we hope that the suggestions here concerning specific sedimentary intervals, geochemical analyses, and palaeobiological assessments, which together represent a community-wide perspective of themes that warrant further investigation, provide direction and support for continued research.