Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Archaeal lipid biomarker constraints on the Paleocene-Eocene carbon isotope excursion

Abstract

A negative carbon isotope excursion recorded in terrestrial and marine archives reflects massive carbon emissions into the exogenic carbon reservoir during the Paleocene-Eocene Thermal Maximum. Yet, discrepancies in carbon isotope excursion estimates from different sample types lead to substantial uncertainties in the source, scale, and timing of carbon emissions. Here we show that membrane lipids of marine planktonic archaea reliably record both the carbon isotope excursion and surface ocean warming during the Paleocene-Eocene Thermal Maximum. Novel records of the isotopic composition of crenarchaeol constrain the global carbon isotope excursion magnitude to −4.0 ± 0.4‰, consistent with emission of >3000 Pg C from methane hydrate dissociation or >4400 Pg C for scenarios involving emissions from geothermal heating or oxidation of sedimentary organic matter. A pre-onset excursion in the isotopic composition of crenarchaeol and ocean temperature highlights the susceptibility of the late Paleocene carbon cycle to perturbations and suggests that climate instability preceded the Paleocene-Eocene Thermal Maximum.

Introduction

The early Paleogene witnessed frequent intervals of transient warming, so-called hyperthermals, superimposed on a greenhouse background climate1. The most intense hyperthermal, the Paleocene-Eocene Thermal Maximum (PETM), was a ~200 ka global climate perturbation2 associated with increased atmospheric CO2 levels3,4, ocean warming5, and ocean acidification6. Widespread observations of a negative carbon isotopic excursion (CIE) in various geological substrates indicate substantial emissions of 13C-depleted carbon into the biosphere-ocean-atmosphere (exogenic) reservoir during the PETM7,8,9. Due to the fossil nature of the carbon and the potentially rapid release, the PETM represents a close albeit imperfect analog for modern anthropogenic climate change10,11,12. Accurately reconstructing the amount, origin, and timing of carbon emissions during the PETM13,14 is a critical step towards understanding the associated climatic forcing and feedback mechanisms, and thus the impacts of carbon cycle perturbations on the past and future Earth system15.

The duration and magnitude of the CIE are important constraints on the amount and source of the emitted carbon8,16. The magnitude of the CIE can in principle be reconstructed through δ13C measurements of carbon-bearing substrates in geologic records, and is thus commonly used for mass balance calculations of PETM emission8,9,16 (compiled in Supplementary Data 1) and as a tuning target in simulated emission scenarios11,13,17,18. Despite more than 160 published CIE estimates from various sample substrates16, the true magnitude of the CIE of the exogenic carbon reservoir remains elusive. In particular, it remains unclear whether the CIE reconstructed from a given sample substrate is representative of the CIE of the entire exogenic carbon reservoir, since equilibration between the exogenic sub-reservoirs may depend on emission rates and size, carbon isotopic composition and location of the source, and oceanographic changes19. The CIEs recorded in different sample substrates vary from −0.6 to −5‰ in marine carbonates to −1 to −8‰ in bulk organic carbon from both marine and terrestrial records16. The wide range of CIE values has led to a diversity of PETM carbon emission scenarios ranging from 2000–20,000 Pg C (refs. 16,17 Supplementary Data 1). These scenarios invoke a wide array of possible carbon sources, including enhanced methane emissions from hydrate dissociation8, geothermal heating of organic carbon20, and oxidation of permafrost or other generic organic carbon sources9,21. Carbon cycle modeling based on observed changes in ocean carbonate saturation and ocean pH yield CIE-independent, but diverging estimates of ~3000–7000 Pg C (refs. 13,14) and > 10,000 Pg C (ref. 17), respectively. However, reconstructions of the isotopic composition of the carbon source from these models still depend on assumptions about the CIE magnitude18.

The discrepancy in the magnitude of the CIE between sample substrates and across sampling locations likely originates from changes in geochemical and biological factors across the PETM that influence δ13C values of the substrate. These factors include diagenetic overprints, source heterogeneity, and vital effects of the source organisms16,19,22. For instance, the carbon isotopic composition of soil bulk organic matter is affected by variations in higher plant assemblages23 as well as by variable fractionation imparted by changes in atmospheric CO2 concentrations and associated growth rates3. These factors similarly affect the carbon isotopic composition of marine phytoplankton biomass24 and thus bias reconstructions from bulk sediment organic matter (often towards larger CIEs), in addition to offsets resulting from varying admixtures of terrigenous organic matter into marine sediments19. By contrast, δ13C records of bulk carbonate and foraminifera may underestimate the magnitude of the CIE due to biases imposed by carbonate dissolution25, diagenesis26, and vital effects27. Collectively, these uncertainties confound not only the quantification of carbon emissions during the PETM16, but also our understanding of temporal release patterns11,17,28, climate sensitivity15, and ecosystem recovery10.

A proxy for the CIE that is insensitive to mineral dissolution and has negligible or readily quantifiable vital effects would help constrain carbon emissions during hyperthermals. The archaeal lipid biomarker crenarchaeol may satisfy these criteria for the marine environment of the Paleogene. This organic compound originates from Thaumarchaeota29, a group of marine planktonic archaea that fix dissolved inorganic carbon (DIC)30. δ13Ccren should thus primarily reflect δ13CDIC at the habitat depth of Thaumarcheaota31,32, which predominantly reside at the base of the photic and upper sub-photic zones33,34. Analysis of thaumarchaeal lipids additionally provides coupled ocean temperature estimates through the TEX86 proxy35, such that both δ13CDIC and ocean temperature can be obtained on the same substrate, circumventing stratigraphic uncertainties inherent to CIE–temperature correlations based on different substrates.

Here we apply spooling-wire microcombustion–isotope ratio mass spectrometry of δ13Ccren (described in detail in the Methods section and in ref. 32) to reconstruct the PETM CIE in thaumarchaeal lipids from three globally distributed sites. Values of δ13Ccren are then compared to the CIE recorded in newly generated and previously published δ13C records of total organic carbon (TOC)36, benthic foraminifera37, dinocysts19,38, and n-alkanes39. We argue that under PETM boundary conditions (pH ~7.5, ref. 6; [DIC] ≥ 2 mM) δ13Ccren is offset by a predictable fractionation factor from δ13CDIC (refs. 40,41) and that the change in δ13Ccren during the PETM closely approximates the CIE in the global exogenic carbon reservoir. We further couple δ13Ccren to TEX86 analyses which allows simultaneous reconstruction of CIE and temperature change in low-carbonate sediments, thereby circumventing confounding effects of carbonate dissolution prevalent during the PETM and other climate warming events. Our independent CIE estimate derived from δ13Ccren represents an important step towards constraining PETM carbon emissions.

Results

Study area

We sampled the Paleocene-Eocene boundary in three marine sediment cores (Fig. 1) retrieved during expeditions ODP 174AX (Ancora site, Hole A/B, New Jersey shelf), ODP 189 (Hole 1172D, core 15R section 4–5, Tasman Sea), and IODP 302 (Hole M0004A, cores 28X-32X, Arctic Ocean). During the Paleocene and Eocene, the New Jersey shelf and Tasman Sea sites were located on continental margins of the north-west Atlantic and south-west Pacific Oceans, respectively36,42. The Arctic Ocean site was located on the submerged northern flank of the Lomonosov Ridge43. Records of δ13Ccren were generated for all three sites and were compared against δ13C changes in other substrates at these sites. Newly generated (New Jersey) and existing (Arctic Ocean43, Tasman Sea36) TEX86 records were used to compare changes in ocean temperature and δ13Ccren across the PETM. To assess the potential influence of terrigenous input of crenarchaeol on δ13Ccren, records of BIT (branched-over-isoprenoid tetraether index), a semi-quantitative tracer for input of soil organic matter44, were generated for the New Jersey core, while literature data were used for the Tasman Sea36 and Arctic Ocean43 cores. δ13CTOC records were generated for the Ancora and Arctic Ocean cores and literature data were used for the Tasman Sea site36.

Fig. 1
figure1

The Paleocene-Eocene carbon isotope excursion recorded in crenarchaeol and other marine substrates. Carbon isotopic ratios of crenarchaeol (δ13Ccren), total organic carbon (δ13CTOC; Arctic Ocean, New Jersey: this study; Tasman Sea: ref. 36) and the foraminifera Anomalinoides acutus and Cibicidoides spp. (δ13CForaminifera; ref. 37) at three globally distributed sites (a, Arctic Ocean; b, New Jersey shelf; c, Tasman Sea). Foraminifera are preserved only in the New Jersey shelf record. Closed and open symbols for δ13Ccren represent samples with high and low signal-to-background ratios (F2/F1; see Methods), respectively. Error bars of δ13Ccren represent one standard deviation of replicate analyses (n = 5). Error bars of δ13Ccren are not displayed where they are smaller than the symbol. Boundaries of the PETM and pre-onset event are shown in blue (based on refs. 36,42,43) and gray shading, respectively. Error bars in the stratigraphic column of the Arctic Ocean record indicate uncertainty in sample position due to poor core recovery. Inset map (d) shows sampling locations and plate configuration (reconstructed using the Ocean Drilling Stratigraphic Network webservice, http://www.odsn.de/ and the model of ref. 77) during the late Paleocene (ca. 55 Ma)

Crenarchaeol stable carbon isotope records

Our new δ13Ccren data allow an assessment of the magnitude of the crenarchaeol CIE (CIEcren) at each location. For the Ancora and Tasman Sea datasets, we estimate CIEcren as the difference between mean pre-PETM and mean peak-PETM values (with the CIEcren error representing the propagated uncertainties, i.e., 1σ standard deviations of the means)16. The intervals of the pre- and peak-PETM were determined through ramp function fitting45 (rampfit; Supplementary Fig. 1). For the Arctic Ocean record, owing to the coring gaps, we report the CIE as the difference between the pre-PETM δ13Ccren mean and the lowest δ13Ccren value during the PETM. We find a CIEcren of −3.12 ± 0.20‰ for the Ancora site and −3.28 ± 0.21‰ for the Tasman Sea (Fig. 1). The CIEcren estimate for the Arctic Ocean is −3.31 ± 0.23‰. These estimates are all consistent within 1σ uncertainties, with an average CIEcren of −3.24 ± 0.37‰.

Following our rampfit approach, the CIE in total organic carbon (TOC) is −4.31 ± 0.39‰ at Ancora, −3.06 ± 0.37‰ in the Tasman Sea (data from ref. 36), and −6.67 ± 0.22‰ in the Arctic Ocean. These values are significantly more heterogeneous than our δ13Ccren-based estimates, possibly due to variable proportions of terrigenous and marine input influencing the δ13CTOC signal, which leads to an overestimation of the CIE at least for the Arctic Ocean record19. At the New Jersey shelf, the only site studied here with sufficiently preserved carbonates, benthic foraminifera37,46 indicate a CIE of −3.54 ± 0.34‰ at Ancora and −3.76 ± 0.44‰ at Bass River (rampfit approach; Supplementary Fig. 1).

Temperature records and provenance of archaeal lipids

Low BIT indices of ~0.3 indicate a marine provenance of glycerol dibiphytanyl glycerol tetraethers (GDGTs) throughout the Ancora record, which is similar to previous records from Bass River38 and the Tasman Sea43 (Supplementary Fig. 2). In contrast, BIT values for the Arctic Ocean core43 indicate significant terrigenous input before the CIE, with a drop in BIT during the CIE indicating decreased relative contributions from terrigenous GDGTs during the PETM.

At the depth of the CIE, TEX86 values increase by ~0.2 in the Ancora record, indicating ocean warming, and return to pre-CIE levels after the PETM (Fig. 2). A similar pattern is observed in the Bass River38 and Tasman Sea records36 (Fig. 2). By contrast, TEX86 values do not increase significantly over the course of the CIE in the Arctic Ocean record (Supplementary Fig. 3; ref. 43), which has previously been attributed to unusually high abundances of tricyclic GDGT at this site relative to modern core-top compilations43. However, the modified TEX86’ index, which omits tricyclic GDGT but is still well-correlated to SST in modern core-top sediments43, shows an increase of 0.07 in parallel with the CIE, and returns to pre-CIE levels after the PETM (Fig. 2; data from ref. 46). For comparison, TEX86’ values calculated for the Ancora and Tasman Sea records also increase in parallel with TEX86 (Supplementary Fig. 3). There is no consistent depth offset between changes in the TEX86 (or TEX86’) and δ13Ccren signals at the PETM onset, potentially indicating synchronous change in temperature and CIE. However, incomplete recovery in the Arctic Ocean record39,43 and potential condensation in the Tasman Sea record47 could obscure potential lead-lag relationships.

Fig. 2
figure2

Coupled ocean temperature proxy and crenarchaeol stable carbon isotope data across the Paleocene-Eocene Thermal Maximum. TEX86 and TEX86’ index values (plotted on an inverted axis; Arctic Ocean: ref. 43, New Jersey shelf: this study, Tasman Sea: ref. 36) and carbon isotopic ratios of crenarchaeol (δ13Ccren) during the late Paleocene and early Eocene at three globally distributed sites (a, Arctic Ocean; b, New Jersey shelf; c, Tasman Sea; as in Fig. 1). Error bars for δ13Ccren represent one standard deviation of replicate analyses (n = 5). Error bars of δ13Ccren are not displayed where they are smaller than the symbol

Discussion

The distinctive biology of Thaumarchaeota yields important constraints on the carbon isotope systematics of their lipids. Thaumarchaeota fix carbon using the 3-hydroxypropionate/4-hydroxybutyrate pathway exclusively from bicarbonate (HCO3) rather than from CO2 (ref. 30). Flux balance modeling41 and environmental data40 suggest that carbon isotopic fractionation in marine Thaumarchaeota originates from passive CO2 transport into the cell combined with slow conversion to HCO3-, resulting in a dependence of the expressed fractionation factor on the dissolved CO2 concentration. Importantly, the effect is inverse to the influence of pCO2 on carbon isotopic fractionation in phytoplankton, resulting in a smaller fractionation factor (εDIC-Cren) at high dissolved CO2 concentrations40,41. Due to the pCO2 increase during the PETM εDIC-Cren may have been smaller by 0.75 ± 0.15‰ (range 0.6–0.9‰) during the peak-PETM compared to the pre-PETM (assuming Paleocene-Eocene atmospheric pCO2 of ~850–2200 ppm, refs. 13,17 Supplementary Fig. 4). Accounting for this effect leads to a δ13Ccren-based CIE in DIC (CIEDIC) of −3.87 ± 0.25‰ at Ancora, −4.03 ± 0.27‰ in the Tasman Sea, −4.06 ± 0.27‰ in the Arctic Ocean, yielding an average of −4.0 ± 0.4‰. Tighter constraints on Paleocene-Eocene pCO2 will allow reducing the uncertainty in εDIC-Cren estimates. Regardless, we expect the pCO2-dependent sensitivity of εDIC-Cren to be relatively constant across archaeal species, as cell morphology and physiology are highly similar among marine planktonic Thaumarchaeota48,49,50 and the 3-hydroxypropionate/4-hydroxybutyrate pathway is phylogenetically conserved30. Therefore, minimal effects are expected from changes in the diversity or distribution of thaumarchaeal clades between the modern ocean and past ecosystems. Studies with additional thaumarchaeal cultures will be needed to fully assess potential impacts of changes in community composition on εDIC-Cren and δ13Ccren records.

Due to their ecological niche as chemolithoautotrophic ammonia oxidizers, Thaumarchaeota are most abundant in a narrow interval at the base of the photic zone33,34. Consequently, thaumarchaeal lipids appear to be predominantly sourced from this zone51, despite the presence of Thaumarchaeota throughout the ocean water column52. Contributions from meso- and bathypelagic Thaumarchaeota should therefore not be a major confounding factor for sedimentary δ13Ccren records, and the signal should principally reflect basal photic zone δ13CDIC. A point-source origin of δ13Ccren from the basal photic zone is further supported by δ13Ccren records of Pleistocene Mediterranean sapropel events S3 and S5 (ref. 53), where significant changes in water column hydrography and changing contributions from deep-water Thaumarchaeota did not lead to significant changes in sediment δ13Ccren. The consistency of δ13Ccren data through the sapropel events (including pre-event comparison values) imply that strong changes in ecosystem states, such as eutrophication, habitat depth changes, and anoxic/euxinic conditions do not significantly impact δ13Ccren values53. This is explained by recognizing that the depth habitat of Thaumarchaeota is linked to oceanographic parameters such as the deep chlorophyll maximum and the depth of maximum remineralization. The thaumarchaeal population should thus maintain a similar relative position along the δ13C gradient, rather than being fixed to a specific water depth. Although it is possible that warming-induced changes in remineralization rates during the PETM could have impacted δ13Ccren values by changing δ13CDIC gradients through the thaumarchaeal depth habitat, the extent of these changes likely was minimal given the warm greenhouse background climate of the Paleocene-Eocene54. Overall a primary advantage of estimating the CIEDIC from δ13Ccren is the apparent niche stability of marine Thaumarchaeota.

Putative mixotrophy (i.e., autotrophy supplemented by heterotrophic organic carbon assimilation) by marine archaeal consortia could affect values of δ13Ccren, assuming autotrophic δ13Ccren is > −20‰ and δ13C of the organic substrate would be < −20‰50,55,56. However, close examination of earlier studies demonstrates that there is currently no direct evidence for mixotrophy among epipelagic Thaumarchaeota57. Previous estimates indicating 17 ± 11% heterotrophic carbon assimilation were based on suspended particulate matter samples from 700 m water depth56 and thus were not representative of the major habitat and export depth of Thaumarchaeota, which is centered around ~150 m water depth58. Second, previous reports of amino acid utilization55 solely demonstrated uptake of exogenous amino acids into thaumarchaeal cells but not contribution to biomass carbon. Finally, recent 13C-label experiments demonstrated that all cultivated Thaumarchaeota are strictly autotrophic57. In summary, present knowledge supports the idea that values of δ13Ccren reflect autotrophic growth and reliably record δ13CDIC.

Systematic offsets between δ13Ccren values at our three study sites additionally corroborate an autotrophic origin of crenarchaeol. Preceding the CIE, absolute values of δ13Ccren differ by up to 3‰ between sites, with the most positive δ13Ccren values found in the Tasman Sea and the most negative values observed in the Arctic Ocean, with an intermediate gradient of ~1 to 1.5‰ in δ13Ccren values between the Tasman Sea and New Jersey shelf; the latter gradient is similar in magnitude but opposite in direction to the modern gradient between the North Pacific and North Atlantic59. Although lower δ13Ccren values in the northern Atlantic and Arctic compared to the Pacific Ocean support previous arguments for distinct, and potentially reversed, ocean circulation patterns during the Paleocene-Eocene60,61,62,63, similar CIE magnitudes at all sites suggest that the effect of potential deep water circulation reversals on the CIE was minor in comparison to whole-ocean δ13CDIC changes. Calculating pre-PETM δ13CDIC from an expected εDIC-Cren of −19.7 ± 0.1‰ (Supplementary Fig. 4) yields reasonable δ13CDIC values for the base of the photic zone of ~0.8 ± 0.2‰ (Tasman Sea), −0.5 ± 0.2‰ (New Jersey shelf), and −2.2 ± 0.3‰ (Arctic Ocean). These δ13CDIC reconstructions may prove useful as paleoclimate model targets.

The close agreement in CIE estimates at all sites suggests no significant influence of changes in local environmental conditions on CIE expression. Still, comparatively low δ13Ccren values in the Arctic Ocean, when compared to the Tasman Sea and New Jersey shelf, indicate distinct environmental conditions for the Arctic basin. Lower absolute δ13Ccren values may reflect accumulation of 13C-depleted DIC in the chemocline and anoxic zone of the Arctic Ocean due to reduced mixing under enclosed, euxinic conditions similar to the modern Black Sea or Cariaco Basin64. Further, post-PETM δ13Ccren values in the Arctic Ocean are equal to or higher than pre-PETM δ13Ccren values, while they are lower than pre-PETM values in the Tasman Sea and New Jersey records (Fig. 1). This suggests that during the recovery phase the Arctic Ocean entered a new ecosystem state with a distinct local δ13CDIC signal. Bulk sediment δ15N records65 support the notion that Arctic Ocean ecosystem changes persisted past the peak PETM. Specifically, lower δ15N values in the PETM recovery phase than during the pre-PETM suggest continued suboxic conditions that were more pronounced than prior to the PETM. However, it remains unconstrained how these ecosystem changes, potentially resulting from variability in ocean stratification, affected local δ13CDIC values. Because δ13CDIC dynamics may differ between shelf environments (New Jersey, Tasman Sea), enclosed basins (Arctic Ocean), and deep open ocean environments, additional records from deep ocean environments could help constrain the influence of local effects on δ13CDIC and δ13Ccren. Terrigenous input of crenarchaeol into the proximal Arctic Ocean also could yield depleted δ13Ccren values and a biased CIEcren, given that terrigenous crenarchaeol would likely be more depleted in 13C than marine-derived crenarchaeol32. The BIT record at our deep Arctic Ocean site indicates significant soil input only during the pre-PETM interval (Supplementary Fig. 2). However, it seems unlikely that terrigenous input could significantly impact marine δ13Ccren values as crenarchaeol is a minor or trace component in soils compared to other GDGTs66. This view is supported by the consistency of the CIEcren between sites throughout the analyzed time interval regardless of the magnitude of the local BIT value or the absolute δ13Ccren values (Supplementary Fig. 2).

The lack of biases from source heterogeneity and the agreement between all three marine sediment cores yields confidence in our δ13Ccren-based CIEDIC estimate of 4.0 ± 0.4‰ (Figs. 1, 3). Our δ13Ccren records confirm an earlier CIEcren estimate of −3.6 ± 0.3‰ (CIEDIC= −4.3 ± 0.3‰) using crenarchaeol degradation products from a North Sea record31, although this previous record is too sparsely resolved to allow rampfit estimates.

Fig. 3
figure3

Estimates of the carbon isotope excursion and corresponding emission scenarios during the Paleocene-Eocene Thermal Maximum. a Carbon isotope excursion (CIE) magnitude for the Paleocene-Eocene Thermal Maximum (PETM) reconstructed from crenarchaeol at our three study sites and one additional site from the North Sea, alongside estimates from other substrates of marine (blue) or terrigenous origin (green) compiled previously16 (TOC: total organic carbon). Whiskers indicate maximum/minimum values, boxes represent the 25 and 75% quartiles, and the horizontal line identifies the median. b Compilation of published PETM emission scenarios with reported CIE value and estimate of total carbon emissions (numbers refer to information listed in Supplementary Data 1; note that studies with multiple scenarios are tabulated and shown in this figure as separate entries), which are superimposed on a parameter space of the δ13C of the emission carbon source (color shading) defined by the CIE magnitude and the mass of carbon emitted; assuming a fixed mass (40,000 Pg C)71 and isotopic composition (δ13C = −2.5‰)78 of the exogenic carbon reservoir. Symbols denote the sample substrate on which the CIE was measured, and color coding indicates the assumed source of carbon in each emission scenario. Vertical bars represent median CIEs recorded in crenarchaeol (this study and ref. 31), benthic foraminifera16, and a plant biomass consensus estimate23

Comparison with established proxies suggests that δ13Ccren may circumvent problems that confound other types of δ13C records. Among the most commonly used proxies, bulk marine, and soil organic matter as well as plant and algal biomarkers tend to overestimate the CIE due to increased isotopic fractionation at higher CO2 levels67. Bulk measurements are additionally prone to biases resulting from shifting plant/algal community composition or from varying admixtures of terrigenous and marine organic carbon to the bulk marine TOC pool19. These effects are reflected in variable CIE magnitudes of −3.1‰ to −6.7‰ in δ13CTOC at our three study sites (Fig. 1). Similarly, CIEs for all compiled records of marine and terrigenous organic matter range from −1‰ to −8‰ (Fig. 3; ref. 16). CIEs reconstructed from more specific algal and plant lipid biomarkers likely suffer from similar biases, yielding diverging results of ~ −6.0 ± 0.2‰ for plant-derived C27 n-alkanes but no discernable CIE for putatively algal C17 n-alkanes at our Arctic Ocean study site39 (Supplementary Figs. 1, 5). Likewise, CIEs reconstructed from species-specific dinoflagellate cyst δ13C data from Bass River vary from −1.8 ± 0.4‰ to −4.0 ± 0.5‰ (Supplementary Fig. 5), possibly due to divergent ecological preferences68. A plant biomass consensus estimate, which accounts for changes in fractionation induced by precipitation, vegetation, temperature, and altitude yields a continental (or atmospheric) CIE of −4.6‰ (ref. 23). However, this estimate does not account for changes in pCO2-dependent fractionation and thus may overestimate the true CIE.

In contrast to bulk organic matter, bulk marine carbonate and foraminiferal carbonate may underestimate the CIE due to carbonate dissolution in the global deep ocean during the PETM (ref. 6; e.g., below ~1500 m water depth in the South Atlantic25), and/or post-depositional carbonate recrystallization26. In addition, vital effects in foraminifera may result in δ13C carbonate-seawater disequilibria27,69,70. These confounding effects are reflected in a large spread of foraminifera-based CIE estimates between −0.6‰ and −5.1‰ for benthic species (median −2.6‰)16 and −0.7‰ to −4.4‰ for planktonic species (median −2.6‰)16. At the New Jersey shelf, the only site studied here with well-preserved (glassy) foraminifera, benthic foraminifera indicate a CIE of −3.5 ± 0.3‰ (Ancora37) and −3.8 ± 0.4‰ (Bass River46; Supplementary Fig. 1), while the CIE derived from planktonic foraminifera (Bass River) is somewhat more variable (−3.6 to −4.3‰)46. Along with other CIE estimates from well-preserved foraminifera from the South Atlantic (−3.4 ± 0.2‰, paleo-water depth of ~1500 m; Supplementary Fig. 1)25 and the North Atlantic (−3.3 ± 0.2‰; Supplementary Fig. 1)37, these estimates are consistent with our δ13Ccren-based CIEDIC estimate, emphasizing that the global median CIE of foraminiferal records of −2.6‰ (ref. 16) likely is biased by generally poor carbonate preservation. Additional biases from vital effects69,70 may have amplified (e.g., decrease in δ13C of the food source) or dampened (e.g., decrease in carbonate ion concentration) the CIE in foraminifera. However, it is currently not possible to quantify the net bias resulting from these effects.

The observation of similar CIE magnitudes in both benthic foraminifera and planktonic archaea, which broadly represent the bathypelagic and epipelagic DIC pools, respectively, suggests complete propagation of the CIE signal from the surface into the deep ocean within the temporal resolution of our records (> 3–4 ka). Equilibration of the surface and deep ocean carbon pools within < 4 ka is supported by model simulations across a wide range of carbon emission rates, magnitudes, and isotopic compositions18. Accordingly, paleorecords with higher temporal resolution could yield larger initial CIEs that were not equilibrated across the entire exogenic carbon reservoir18. Indeed, very negative δ13C values at the PETM onset were observed in single foraminifera26 and putatively annually-resolved bulk carbonate records28. Taken together, the δ13Ccren records narrow down the fully-mixed marine CIE to −4.0 ± 0.4‰. As most exogenic carbon is stored as DIC in the ocean and is mixed with the other sub-reservoirs on millennial timescales, we argue that this estimate may approximate the average total CIE for the exogenic carbon reservoir.

The magnitude of the CIE reconstructed from δ13Ccren can be used to refine plausible carbon emission scenarios for the PETM. To this end, we compiled previous estimates of CIE magnitude as well as the mass and isotopic composition of emitted carbon (Fig. 3b, Supplementary Data 1). Based on assumed or measured CIEs of −2.5 to −5.5‰, these previous estimates of carbon emissions vary widely from 1100 Pg to 20,000 Pg C. Our CIE estimate of −4.0 ± 0.4‰ is significantly larger than predicted from most marine records and is therefore also larger than CIEs used in most mass balance calculations and carbon cycle models8,13,16,17. Mass balance calculations using our CIE estimate (Fig. 3b) constrain emissions to 4400–15,200 Pg C if the carbon source is thermogenic methane and CO213C = −17‰ to −42‰)17,20, to 3000–3700 Pg C if the source is only methane (e.g., hydrates; δ13C = −50‰ to −60‰)8,13,22,71, and to 8200–10,300 Pg C if the source is organic matter (δ13C = −22‰ to −26‰)9,14,21. These ranges of δ13Ccren-based emissions estimates are not only narrower than previous mass balance estimates but also more precise, with an uncertainty of ~10% compared to the much larger uncertainties resulting from the wide scatter of other substrate-specific CIEs16 (Fig. 3a).

Although a mass-balance approach assumes a single emission pulse, our δ13Ccren-based CIEDIC estimate also allows for re-evaluation of previous model-derived estimates that incorporate more complex features such as sustained emissions and carbon burial. It is notable that many of the more complex modeling approaches to date have been based on smaller magnitudes for the CIE of −2 to −3‰13,17. For instance, coupled carbonate dissolution data-model simulations suggest emission of <3000 Pg C (ref. 13), with a CIE of −3‰ in foraminifera used to independently constrain the δ13C of the carbon source to −50‰. The -4.0 ± 0.4‰ CIE in crenarchaeol instead implies that the carbon isotopic composition of the emitted carbon would be required to be more negative (< −60‰) for this scenario. Similarly, ocean pH data-model comparisons17 suggesting larger emissions (5700–20,000 Pg C) used a CIE of −2.6‰ derived from planktonic foraminifera to constrain the δ13C of the carbon source to −11‰ (ref. 17), implying emission of mostly mantle-derived CO2. Our larger δ13Ccren-based CIE again would imply a more 13C-depleted carbon source; specifically, it would scale the mean emission δ13C value to −17‰ (using a CIE value of −4‰ but otherwise following the approach of Gutjahr et al.17). A revised emission δ13C value of −17‰ suggests larger contributions from organic matter, such as from thermogenic methane generated by North Atlantic sill emplacement17,20,72, sedimentary organic matter oxidation9,21,73, or methane hydrate dissociation8. In summary, the tighter constraints on the CIE from δ13Ccren suggest that previous estimates based on smaller CIEs may have pointed to carbon source δ13C values that are too enriched in 13C and/or underestimated the magnitude of carbon emissions. Carbon emissions must have been larger than 3000 Pg C for a release from methane hydrates and larger than 4400 Pg C for any scenario involving contributions from sedimentary organic matter oxidation and/or geothermal heating. Building on our refinement of the CIE magnitude and uncertainty, emerging highly resolved records of pCO2 or carbonate system parameters (e.g., [DIC]) will allow closer evaluation of the rates and mechanisms of PETM carbon emissions.

Analysis of archaeal lipids further provides coupled temperature data via the TEX86-paleothermometer, which can be used to evaluate the role of pre-CIE warming as a trigger for carbon emissions. This approach avoids biases from diagenetic carbonate overgrowth and dissolution affecting foraminiferal δ13C values26 and stratigraphic ambiguities that arise when separate substrates are used for the CIE and warming. Although incompleteness of the PETM onset in the Arctic Ocean record, and potentially the Tasman core, as well as a lack of well-constrained age models limit the determination of lead-lag relationships, the parallel changes in δ13Ccren and TEX86 (Fig. 2) suggest that any warming preceding the CIE must have occurred no more than 3–4 ka prior, consistent with previous estimates from the New Jersey shelf38,74, the equatorial Atlantic74, and the Southern Ocean74. The coupled δ13Ccren-TEX86 approach described here could be instrumental for determining high-resolution lead-lag relationships of temperature and CIE during the PETM and other climate events.

The coupled δ13Ccren-TEX86 records further yield important insights into the functioning of the carbon cycle immediately prior to the PETM. Our highest resolution record from the Tasman Sea captures a decrease of ~ −1‰ in δ13Ccren and a coeval increase in TEX86 (equivalent to 1–2 °C warming, depending on calibration; Supplementary Fig. 6) preceding the PETM (Fig. 2). This event may be correlated to a 3‰ decrease in δ13C of pedogenic carbonates, the pre-onset event (POE), recorded in a terrestrial section from the Bighorn Basin (Wyoming, USA)12. If these events were indeed coeval, the POE must have been a global event. The POE was an ephemeral event that lasted < 2 ka before recovery to pre-POE conditions12, as recorded in the decrease of both TEX86 and δ13Ccren values in the Tasman Sea immediately prior to the PETM onset (Fig. 2). Similarly, δ13C values decrease to pre-POE values before the PETM onset in the Bighorn Basin12. In contrast to the PETM, which must have been sustained by continuous emissions to explain its protracted recovery75, the POE could have originated from a single carbon release, such as from gas hydrates. The rapid recovery suggests that critical thresholds for positive feedback mechanisms were not reached.

Due to their contrasting behavior, high-resolution δ13Ccren and TEX86 studies of the PETM and other Paleogene hyperthermals, including the POE, may yield constraints on the relative timing of warming, carbon emissions, and recovery. The diagenetic stability and potential for high-resolution records of archaeal lipids suggest that δ13Ccren is a promising proxy, particularly for reconstructing transient CIEs that exhibit poor carbonate preservation, such as during Mesozoic Oceanic Anoxic Events and early Cenozoic hyperthermals. Further ground-truthing using a wider range of thaumarchaeal cultures and globally representative environmental datasets will help further constrain uncertainties of the δ13Ccren proxy and will strengthen its application to the paleoenvironment.

Methods

Lipid extraction

Sediment samples (10–20 cm³) were freeze dried, homogenized, and subsequently extracted in teflon vessels in three steps (50:50 dichloromethane:methanol, 90:10 dichloromethane:methanol, 100% dichloromethane) using a MARS5 microwave-assisted extraction system (CEM Corporation, Matthews, NC, USA). The microwave program consisted of 30 min heating to 100 °C and 20 min hold time, after which the solvent was decanted and fresh solvent was added. The extracts from the successive extraction steps were combined into a total lipid extract (TLE), dried under N2, and stored at −20 °C until measurement. The TLEs were split into two aliquots for (i) purification and carbon stable isotopic analysis of crenarchaeol and (ii) analysis of lipid abundances for determination of TEX86 and BIT.

Purification and carbon isotopic analysis of crenarchaeol

TLE aliquots were separated over pre-combusted SiO2 (130–270 mesh) to yield a GDGT-containing fraction32. The GDGT-containing fraction was further purified by orthogonal semi-preparative high performance liquid chromatography (HPLC) on an Agilent 1200 series HPLC equipped with a fraction collector, following previously described protocols32,53.

The carbon stable isotopic composition of crenarchaeol was analyzed using a spooling wire microcombustion–isotope ratio mass spectrometer (SWiM–IRMS)32. The fractions were dissolved in ethyl acetate and 30–100 ng were manually injected onto the wire (3 replicate injections for F1, 5–6 replicates for F2). The raw data were corrected for blanks and absolute offsets using dilution series of the C46-GTGT standard. The 1σ precision averaged over all samples was 0.15‰. Purity of the crenarchaeol fraction was assessed using the ratio of SWiM–IRMS F2/F1 peak areas (Supplementary Data 2). Analysis of F2/F1 versus δ13Ccren shows that F2/F1 ratios smaller than 2 yield negatively biased δ13Ccren values, due to generally more negative δ13C values of the background organic matter relative to crenarchaeol. Therefore, only data with F2/F1 ratios >2 were used for calculating the CIE magnitude. All isotopic ratios are reported relative to the Vienna Pee Dee Belemnite scale.

Analysis of lipid abundances

Relative abundances of GDGTs (Supplementary Data 3) from site ODP 174AX were determined from aliquots of the same TLE used for crenarchaeol isotopic analyses, using the instruments and conditions described in ref. 53.

TEX86 and BIT values were calculated from relative abundances of GDGTs for the Ancora record. Published TEX86 and BIT data from adjacent samples were used for the Arctic Ocean43 and Tasman Sea records36. TEX86 values were corrected for differences in peak separation between the traditional single-column HPLC-MS method and the newer protocol76 employed here by using a cross-calibration of 27 Arabian Sea and Atlantic Ocean sediments. Due to the ambiguity associated with choosing appropriate deep-time TEX86 sea surface temperature calibrations, all data are presented as TEX86 or TEX86’ ratios rather than putative sea surface temperatures, with the understanding that higher ratios correspond to relatively warmer temperatures.

Carbon isotopic analysis of bulk sediment

Content and carbon isotopic composition of TOC at sites ODP 174AX and IODP 302 (Supplementary Data 4) were analyzed on acidified bulk sediment aliquots using a ThermoScientific Flash EA Delta V Plus IRMS. Carbon isotopic compositions were peak-size-corrected and offset-corrected using laboratory and authentic reference standards (USGS40, USGS41a). The standard deviation of δ13CTOC as determined by replicate analysis of reference standards was 0.04‰.

Calculation of carbon isotope fractionation and carbon isotope excursion

εDIC-Cren was calculated for pre-PETM and peak-PETM boundary conditions (Supplementary Fig. 4) using the flux balance model by Pearson et al.41. Boundary conditions13,17 were set as: 25 °C seawater temperature, salinity 35, [DIC] 2000–3000 µmol kg-1, pH 7.75–7.45. εDIC-Cren depends on assumed pCO2 change across the PETM, following results from Gutjahr et al.17, εDIC-Cren changed by 0.9‰, from of 19.7 ± 0.1‰ for the pre-PETM to 18.8 ± 0.05‰ for the peak-PETM. However, using the smaller pCO2 change suggested by Zeebe et al.13, εDIC-cren changed by only 0.6‰. Uncertainty in pCO2 change across the PETM thus leads to an uncertainty in εDIC-cren of ± 0.15‰ from an average of 0.75‰. CIEcren was calculated from pre-PETM and peak-PETM δ13Ccren data using the rampfit software45. To derive CIEDIC, pre-PETM and peak-PETM εDIC-cren were applied to CIEcren to derive CIEDIC and uncertainty was determined by propagating the measurement uncertainty, the rampfit estimated error of CIEcren, and the uncertainty in εDIC-cren.

Data availability

All data underlying main text and supplementary figures are available as a Source Data file and have been deposited in the Pangaea repository (https://doi.org/10.1594/PANGAEA.900498) and the Harvard Dataverse (https://doi.org/10.7910/DVN/OJJGYN).

References

  1. 1.

    Zachos, J. C., McCarren, H., Murphy, B., Röhl, U. & Westerhold, T. Tempo and scale of late Paleocene and early Eocene carbon isotope cycles: implications for the origin of hyperthermals. Earth Planet. Sci. Lett. 299, 242–249 (2010).

    CAS  ADS  Article  Google Scholar 

  2. 2.

    Westerhold, T. et al. Synchronizing early Eocene deep-sea and continental records—cyclostratigraphic age models for the Bighorn Basin Coring Project drill cores. Climate 14, 303–319 (2018).

    Google Scholar 

  3. 3.

    Schubert, B. A. & Jahren, A. H. Reconciliation of marine and terrestrial carbon isotope excursions based on changing atmospheric CO2 levels. Nat. Commun. 4, 1653 (2013).

    ADS  Article  Google Scholar 

  4. 4.

    Cui, Y. & Schubert, B. A. Atmospheric pCO2 reconstructed across five early Eocene global warming events. Earth Planet. Sci. Lett. 478, 225–233 (2017).

    CAS  ADS  Article  Google Scholar 

  5. 5.

    Dunkley Jones, T. et al. Climate model and proxy data constraints on ocean warming across the Paleocene–Eocene Thermal Maximum. Earth-Sci. Rev. 125, 123–145 (2013).

    CAS  ADS  Article  Google Scholar 

  6. 6.

    Penman, D. E., Hönisch, B., Zeebe, R. E., Thomas, E. & Zachos, J. C. Rapid and sustained surface ocean acidification during the Paleocene-Eocene Thermal Maximum. Paleoceanography 29, 357–369 (2014).

    ADS  Article  Google Scholar 

  7. 7.

    Kennett, J. P. & Stott, L. D. Abrupt deep-sea warming, palaeoceanographic changes and benthic extinctions at the end of the Palaeocene. Nature 353, 225–229 (1991).

    ADS  Article  Google Scholar 

  8. 8.

    Dickens, G. R., O’Neil, J. R., Rea, D. K. & Owen, R. M. Dissociation of oceanic methane hydrate as a cause of the carbon isotope excursion at the end of the Paleocene. Paleoceanography 10, 965–971 (1995).

    ADS  Article  Google Scholar 

  9. 9.

    Higgins, J. A. & Schrag, D. P. Beyond methane: towards a theory for the Paleocene-Eocene Thermal Maximum. Earth Planet. Sci. Lett. 245, 523–537 (2006).

    CAS  ADS  Article  Google Scholar 

  10. 10.

    Zeebe, R. E. & Zachos, J. C. Long-term legacy of massive carbon input to the Earth system: Anthropocene versus Eocenetransactions. Philos. T. R. Soc. A 371, 20120006 (2013).

    ADS  Article  Google Scholar 

  11. 11.

    Zeebe, R. E., Ridgwell, A. & Zachos, J. C. Anthropogenic carbon release rate unprecedented during the past 66 million years. Nat. Geosci. 9, 325–329 (2016).

    CAS  ADS  Article  Google Scholar 

  12. 12.

    Bowen, G. J. et al. Two massive, rapid releases of carbon during the onset of the Palaeocene–Eocene thermal maximum. Nat. Geosci. 8, 44–47 (2015).

    CAS  ADS  MathSciNet  Article  Google Scholar 

  13. 13.

    Zeebe, R. E., Zachos, J. C. & Dickens, G. R. Carbon dioxide forcing alone insufficient to explain Palaeocene–Eocene thermal maximum warming. Nat. Geosci. 2, 576–580 (2009).

    CAS  ADS  Article  Google Scholar 

  14. 14.

    Panchuk, K., Ridgwell, A. & Kump, L. R. Sedimentary response to Paleocene-Eocene thermal maximum carbon release: a model-data comparison. Geology 36, 315–318 (2008).

    CAS  ADS  Article  Google Scholar 

  15. 15.

    Pagani, M., Caldeira, K., Archer, D. & Zachos, J. C. An ancient carbon mystery. Science 314, 1556–1557 (2006).

    CAS  Article  Google Scholar 

  16. 16.

    McInerney, F. A. & Wing, S. L. The Paleocene-Eocene Thermal Maximum: a perturbation of carbon cycle, climate, and biosphere with implications for the future. Annu. Rev. Earth Planet. Sci. 39, 489–516 (2011).

    CAS  ADS  Article  Google Scholar 

  17. 17.

    Gutjahr, M. et al. Very large release of mostly volcanic carbon during the Palaeocene–Eocene Thermal Maximum. Nature 548, 573–577 (2017).

    CAS  ADS  Article  Google Scholar 

  18. 18.

    Kirtland Turner, S. & Ridgwell, A. Development of a novel empirical framework for interpreting geological carbon isotope excursions, with implications for the rate of carbon injection across the PETM. Earth Planet. Sci. Lett. 435, 1–13 (2016).

    CAS  ADS  Article  Google Scholar 

  19. 19.

    Sluijs, A. & Dickens, G. R. Assessing offsets between the δ13C of sedimentary components and the global exogenic carbon pool across early Paleogene carbon cycle perturbations. Glob. Biogeochemical Cycles 26, GB4005 (2012).

    ADS  Google Scholar 

  20. 20.

    Svensen, H. et al. Release of methane from a volcanic basin as a mechanism for initial Eocene global warming. Nature 429, 542–545 (2004).

    CAS  ADS  Article  Google Scholar 

  21. 21.

    DeConto, R. M. et al. Past extreme warming events linked to massive carbon release from thawing permafrost. Nature 484, 87–91 (2012).

    CAS  ADS  Article  Google Scholar 

  22. 22.

    Dickens, G. R. Down the Rabbit Hole: Toward appropriate discussion of methane release from gas hydrate systems during the Paleocene-Eocene thermal maximum and other past hyperthermal events. Climate 7, 831–846 (2011).

    Google Scholar 

  23. 23.

    Diefendorf, A. F., Mueller, K. E., Wing, S. L., Koch, P. L. & Freeman, K. H. Global patterns in leaf 13C discrimination and implications for studies of past and future climate. Proc. Natl Acad. Sci. 107, 5738–5743 (2010).

    CAS  ADS  Article  Google Scholar 

  24. 24.

    Laws, E. A., Popp, B. N., Bidigare, J. R. R., Kennicutt, M. C. & Macko, S. A. Dependence of phytoplankton carbon isotopic composition on growth rate and [CO2]aq: Theoretical considerations and experimental results. Geochimica et. Cosmochimica Acta 59, 1131–1138 (1995).

    CAS  ADS  Article  Google Scholar 

  25. 25.

    McCarren, H., Thomas, E., Hasegawa, T., Röhl, U. & Zachos, J. C. Depth dependency of the Paleocene-Eocene carbon isotope excursion: Paired benthic and terrestrial biomarker records (Ocean Drilling Program Leg 208, Walvis Ridge). Geochem., Geophysics, Geosystems 9, Q10008 (2008).

    ADS  Article  Google Scholar 

  26. 26.

    Kozdon, R., Kelly, D. C. & Valley, J. W. Diagenetic attenuation of carbon isotope excursion recorded by planktic foraminifers during the Paleocene-Eocene Thermal Maximum. Paleoceanogr. Paleoclimatology 33, 367–380 (2018).

    Article  Google Scholar 

  27. 27.

    Si, W. & Aubry, M.-P. Vital effects and ecologic adaptation of photosymbiont-bearing planktonic foraminifera during the Paleocene-Eocene Thermal Maximum, implications for Paleoclimate. Paleoceanogr. Paleoclimatology 33, 112–125 (2018).

    Article  Google Scholar 

  28. 28.

    Wright, J. D. & Schaller, M. F. Evidence for a rapid release of carbon at the Paleocene-Eocene thermal maximum. Proc. Natl Acad. Sci. USA 110, 15908–15913 (2013).

    CAS  ADS  Article  Google Scholar 

  29. 29.

    Elling, F. J. et al. Chemotaxonomic characterisation of the thaumarchaeal lipidome. Environ. Microbiol. 19, 2681–2700 (2017).

    CAS  Article  Google Scholar 

  30. 30.

    Könneke, M. et al. Ammonia-oxidizing archaea use the most energy-efficient aerobic pathway for CO2 fixation. Proc. Natl Acad. Sci. USA 111, 8239–8244 (2014).

    ADS  Article  Google Scholar 

  31. 31.

    Schoon, P. L. et al. Recognition of Early Eocene global carbon isotope excursions using lipids of marine Thaumarchaeota. Earth Planet. Sci. Lett. 373, 160–168 (2013).

    CAS  ADS  Article  Google Scholar 

  32. 32.

    Pearson, A. et al. Stable carbon isotope ratios of intact GDGTs indicate heterogeneous sources to marine sediments. Geochimica et. Cosmochimica Acta 181, 18–35 (2016).

    CAS  ADS  Article  Google Scholar 

  33. 33.

    Santoro, A. E., Casciotti, K. L. & Francis, C. A. Activity, abundance and diversity of nitrifying archaea and bacteria in the central California Current. Environ. Microbiol. 12, 1989–2006 (2010).

    CAS  Article  Google Scholar 

  34. 34.

    Church, M. J., Wai, B., Karl, D. M. & DeLong, E. F. Abundances of crenarchaeal amoA genes and transcripts in the Pacific Ocean. Environ. Microbiol. 12, 679–688 (2010).

    CAS  Article  Google Scholar 

  35. 35.

    Schouten, S., Hopmans, E. C., Schefuß, E. & Sinninghe Damsté, J. S. Distributional variations in marine crenarchaeotal membrane lipids: a new tool for reconstructing ancient sea water temperatures? Earth Planet. Sci. Lett. 204, 265–274 (2002).

    CAS  ADS  Article  Google Scholar 

  36. 36.

    Sluijs, A. et al. Southern ocean warming, sea level and hydrological change during the Paleocene-Eocene thermal maximum. Climate 7, 47–61 (2011).

    Google Scholar 

  37. 37.

    Cramer, B. S. & Kent, D. V. Bolide summer: The Paleocene/Eocene thermal maximum as a response to an extraterrestrial trigger. Palaeogeogr., Palaeoclimatol., Palaeoecol. 224, 144–166 (2005).

    Article  Google Scholar 

  38. 38.

    Sluijs, A. et al. Environmental precursors to rapid light carbon injection at the Palaeocene/Eocene boundary. Nature 450, 1218–1221 (2007).

    CAS  ADS  Article  Google Scholar 

  39. 39.

    Pagani, M. et al. Arctic hydrology during global warming at the Palaeocene/Eocene thermal maximum. Nature 442, 671–675 (2006).

    CAS  ADS  Article  Google Scholar 

  40. 40.

    Hurley, S. J. et al. CO2-dependent carbon isotope fractionation in Archaea, Part II: the marine water column. Geochimica et. Cosmochimica Acta 261, 383–395 (2019).

    CAS  ADS  Article  Google Scholar 

  41. 41.

    Pearson, A., Hurley, S. J., Elling, F. J. & Wilkes, E. B. CO2-dependent carbon isotope fractionation in Archaea, Part I: Modeling the 3HP/4HB pathway. Geochimica et. Cosmochimica Acta 261, 368–382 (2019).

    CAS  ADS  Article  Google Scholar 

  42. 42.

    Stassen, P., Thomas, E. & Speijer, R. P. Integrated stratigraphy of the Paleocene-Eocene thermal maximum in the New Jersey Coastal Plain: toward understanding the effects of global warming in a shelf environment. Paleoceanography 27, PA4210 (2012).

    ADS  Article  Google Scholar 

  43. 43.

    Sluijs, A. et al. Subtropical Arctic Ocean temperatures during the Palaeocene/Eocene thermal maximum. Nature 441, 610–613 (2006).

    CAS  ADS  Article  Google Scholar 

  44. 44.

    Hopmans, E. C. et al. A novel proxy for terrestrial organic matter in sediments based on branched and isoprenoid tetraether lipids. Earth Planet. Sci. Lett. 224, 107–116 (2004).

    CAS  ADS  Article  Google Scholar 

  45. 45.

    Mudelsee, M. Ramp function regression: a tool for quantifying climate transitions. Computers Geosci. 26, 293–307 (2000).

    ADS  Article  Google Scholar 

  46. 46.

    John, C. M. et al. North American continental margin records of the Paleocene-Eocene thermal maximum: Implications for global carbon and hydrological cycling. Paleoceanography 23, PA2217 (2008).

    ADS  Article  Google Scholar 

  47. 47.

    Sluijs, A. et al. Southern ocean warming, sea level and hydrological change during the Paleocene-Eocene thermal maximum. Climate 7, 47–61 (2011).

    Google Scholar 

  48. 48.

    Könneke, M. et al. Isolation of an autotrophic ammonia-oxidizing marine archaeon. Nature 437, 543–546 (2005).

    ADS  Article  Google Scholar 

  49. 49.

    Santoro, A. E. & Casciotti, K. L. Enrichment and characterization of ammonia-oxidizing archaea from the open ocean: phylogeny, physiology and stable isotope fractionation. ISME J. 5, 1796–1808 (2011).

    CAS  Article  Google Scholar 

  50. 50.

    Qin, W. et al. Marine ammonia-oxidizing archaeal isolates display obligate mixotrophy and wide ecotypic variation. Proc. Natl Acad. Sci. USA 111, 12504–12509 (2014).

    CAS  ADS  Article  Google Scholar 

  51. 51.

    Hurley, S. J., Lipp, J. S., Close, H. G., Hinrichs, K.-U. & Pearson, A. Distribution and export of isoprenoid tetraether lipids in suspended particulate matter from the water column of the Western Atlantic Ocean. Org. Geochem. 116, 90–102 (2018).

    CAS  Article  Google Scholar 

  52. 52.

    Karner, M. B., DeLong, E. F. & Karl, D. M. Archaeal dominance in the mesopelagic zone of the Pacific Ocean. Nature 409, 507–510 (2001).

    CAS  ADS  Article  Google Scholar 

  53. 53.

    Polik, C. A., Elling, F. J. & Pearson, A. Impacts of Paleoecology on the TEX86 sea surface temperature proxy in the Pliocene-Pleistocene Mediterranean Sea. Paleoceanogr. Paleoclimatology 33, 1472–1489 (2018).

    Article  Google Scholar 

  54. 54.

    John, E. H., Wilson, J. D., Pearson, P. N. & Ridgwell, A. Temperature-dependent remineralization and carbon cycling in the warm Eocene oceans. Palaeogeogr., Palaeoclimatol., Palaeoecol. 413, 158–166 (2014).

    Article  Google Scholar 

  55. 55.

    Ouverney, C. C. & Fuhrman, J. A. Marine planktonic archaea take up amino acids. Appl. Environ. Microbiol. 66, 4829–4833 (2000).

    CAS  Article  Google Scholar 

  56. 56.

    Ingalls, A. E. et al. Quantifying archaeal community autotrophy in the mesopelagic ocean using natural radiocarbon. Proc. Natl Acad. Sci. USA 103, 6442–6447 (2006).

    CAS  ADS  Article  Google Scholar 

  57. 57.

    Kim, J.-G. et al. Hydrogen peroxide detoxification is a key mechanism for growth of ammonia-oxidizing archaea. Proc. Natl Acad. Sci. USA 113, 7888–7893 (2016).

    CAS  Article  Google Scholar 

  58. 58.

    Lincoln, S. A. et al. Planktonic Euryarchaeota are a significant source of archaeal tetraether lipids in the ocean. Proc. Natl Acad. Sci. USA 111, 9858–9863 (2014).

    CAS  ADS  Article  Google Scholar 

  59. 59.

    Kroopnick, P. M. The distribution of 13C of ΣCO2 in the world oceans. Deep Sea Res. Part A. Oceanographic Res. Pap. 32, 57–84 (1985).

    CAS  Article  Google Scholar 

  60. 60.

    Bice, K. L. & Marotzke, J. Could changing ocean circulation have destabilized methane hydrate at the Paleocene/Eocene boundary? Paleoceanography 17, 1018 (2002).

    ADS  Google Scholar 

  61. 61.

    Nunes, F. & Norris, R. D. Abrupt reversal in ocean overturning during the Palaeocene/Eocene warm period. Nature 439, 60–63 (2006).

    CAS  ADS  Article  Google Scholar 

  62. 62.

    Zeebe, R. E. & Zachos, J. C. Reversed deep-sea carbonate ion basin gradient during Paleocene-Eocene thermal maximum. Paleoceanography 22, PA3201 (2007).

    ADS  Article  Google Scholar 

  63. 63.

    Luo, Y., Boudreau, B. P., Dickens, G. R., Sluijs, A. & Middelburg, J. J. An alternative model for CaCO3 over-shooting during the PETM: Biological carbonate compensation. Earth Planet. Sci. Lett. 453, 223–233 (2016).

    CAS  ADS  Article  Google Scholar 

  64. 64.

    Fry, B. et al. Stable isotope studies of the carbon, nitrogen and sulfur cycles in the Black Sea and the Cariaco Trench. Deep Sea Res. Part A. Oceanographic Res. Pap. 38, S1003–S1019 (1991).

    Article  Google Scholar 

  65. 65.

    Knies, J., Mann, U., Popp, B. N., Stein, R. & Brumsack, H.-J. Surface water productivity and paleoceanographic implications in the Cenozoic Arctic Ocean. Paleoceanography 23, PA1S16 (2008).

    Article  Google Scholar 

  66. 66.

    Weijers, J. W. H., Wiesenberg, G. L. B., Bol, R., Hopmans, E. C. & Pancost, R. D. Carbon isotopic composition of branched tetraether membrane lipids in soils suggest a rapid turnover and a heterotrophic life style of their source organism(s). Biogeosciences 7, 2959–2973 (2010).

    CAS  ADS  Article  Google Scholar 

  67. 67.

    Schubert, B. A. & Jahren, A. H. The effect of atmospheric CO2 concentration on carbon isotope fractionation in C3 land plants. Geochimica et. Cosmochimica Acta 96, 29–43 (2012).

    CAS  ADS  Article  Google Scholar 

  68. 68.

    Sluijs, A., van Roij, L., Frieling, J., Laks, J. & Reichart, G.-J. Single-species dinoflagellate cyst carbon isotope ecology across the Paleocene-Eocene Thermal Maximum. Geology 46, 79–82 (2018).

    CAS  ADS  Article  Google Scholar 

  69. 69.

    Hesse, T. et al. Modelling δ13C in benthic foraminifera: Insights from model sensitivity experiments. Mar. Micropaleontol. 112, 50–61 (2014).

    ADS  Article  Google Scholar 

  70. 70.

    Uchikawa, J. & Zeebe, R. E. Examining possible effects of seawater pH decline on foraminiferal stable isotopes during the Paleocene-Eocene Thermal Maximum. Paleoceanography 25, PA2216 (2010).

    ADS  Article  Google Scholar 

  71. 71.

    Dickens, G. R. Modeling the Global Carbon Cycle with a Gas Hydrate Capacitor: Significance for the Latest Paleocene Thermal Maximum. in Geophysical Monograph Series (eds. Paull, C. K. & Dillon, W. P.) 19–38 (American Geophysical Union, 2001). https://doi.org/10.1029/GM124p0019.

  72. 72.

    Frieling, J. et al. Thermogenic methane release as a cause for the long duration of the PETM. Proc. Natl Acad. Sci. USA 113, 12059–12064 (2016).

    CAS  ADS  Article  Google Scholar 

  73. 73.

    Lyons, S. L. et al. Palaeocene–Eocene Thermal Maximum prolonged by fossil carbon oxidation. Nat. Geosci. 12, 54–60 (2019).

    CAS  ADS  Article  Google Scholar 

  74. 74.

    Frieling, J. et al. Widespread warming before and elevated barium burial during the Paleocene-Eocene Thermal Maximum: evidence for methane hydrate release? Paleoceanog. Paleocl. 34, 546–566 (2019).

  75. 75.

    Zeebe, R. E. What caused the long duration of the Paleocene-Eocene Thermal Maximum. Paleoceanography 28, 440–452 (2013).

    ADS  Article  Google Scholar 

  76. 76.

    Becker, K. W., Lipp, J. S., Versteegh, G. J. M., Wörmer, L. & Hinrichs, K.-U. Rapid and simultaneous analysis of three molecular sea surface temperature proxies and application to sediments from the Sea of Marmara. Org. Geochem. 85, 42–53 (2015).

    CAS  Article  Google Scholar 

  77. 77.

    Hay, W. W. et al. Alternative global Cretaceous paleogeography. in Special Paper 332: Evolution of the Cretaceous Ocean-Climate System 332, 1–47 (Geological Society of America, 1999).

  78. 78.

    Killops, S. & Killops, V. Introduction to Organic Geochemistry (Blackwell Science, 2005).

Download references

Acknowledgements

This research used samples provided by the Ocean Drilling Program (ODP) and the International Ocean Discovery Program (IODP) and was supported by the National Science Foundation (Grants 1702262 and 1843285, to A.P.), the Gordon and Betty Moore Foundation (A.P.), the Deutscher Akademischer Austauschdienst (RISE Worldwide Fellowship #US-ES-2997 to K.D.D.) and the Deutsche Forschungsgemeinschaft (KU2842/1-1, to S.K.; and GO2294/2-1, to J.G.). Laura Kattein is thanked for assistance with sample preparation.

Author information

Affiliations

Authors

Contributions

F.J.E., A.P., and S.K. designed the study. F.J.E., K.D.D., A.P., S.K., S.J.H., and J.G. performed the analyses. F.J.E., A.P., and J.G. wrote the paper with contributions from all authors.

Corresponding author

Correspondence to Felix J. Elling.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature Communications thanks Gerald Dickens and Appy Sluijs for their contribution to the peer review of this work. Peer reviewer reports are available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Elling, F.J., Gottschalk, J., Doeana, K.D. et al. Archaeal lipid biomarker constraints on the Paleocene-Eocene carbon isotope excursion. Nat Commun 10, 4519 (2019). https://doi.org/10.1038/s41467-019-12553-3

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing