A global atmospheric methane record from a tropical ice core

Atmospheric CH₄ concentrations varied between approximately 350 and 800 parts per billion (ppb) over the past 800,000 years but since 1750 CE have risen approximately 160%, reaching 1,935 ppb in 2025⁹. Estimations of past fluxes and emissions still contain uncertainty, highlighting the need for improved constraints on the role of the tropics in CH₄ production^{5,10,11,12,13,14,15}. To better understand past changes in global levels of CH₄, the inter-polar difference (IPD) was reconstructed from the Greenland Ice Sheet Project 2 (GISP2; Greenland) and West Antarctic Ice Sheet (WAIS; Antarctica) cores⁷. From 800 to 1750 CE, the IPD is on average 44 ± 7 ppb, with GISP2 above the WAIS record⁷. This is interpreted as a reflection of the long-term dominance of Northern Hemisphere emissions. However, these polar-derived mixing ratios only indirectly reflect concentrations in the tropics, creating uncertainty as to the role of tropical sources in historical CH₄ variability^5,6.

Consistent, long-term CH₄ records from low-latitude ice cores would greatly improve our understanding of the atmospheric CH₄ history, but the few existing records are affected by notable complications. For example, a core from the Sajama ice cap in Bolivia (18° S) records large (100–900 ppb) spikes in CH₄ concentrations that are probably linked to elevated dust levels¹⁶, which is an issue also documented at smaller magnitudes¹⁷ in Greenland cores. Methane records from the Himalayan Dasuopu glacier (28° N) show an average concentration roughly 120 ppb higher than average Greenland levels; however, frequent melt layers are thought to impose unrealistically large variability as well as high uncertainty (±37 ppb) on the CH₄ record¹⁸. The Mount Everest East Rongbuk Glacier ice core (28° N)¹⁹ shows late PI Holocene CH₄ that are about 36 ± 17 ppb higher than in Greenland cores, but only 15 of the original 112 values are considered valid after outlier elimination.

In 2019, the Byrd Polar and Climate Research Center (BPCRC) ice core research team conducted an ice core drilling programme on Nevado Huascarán (−9.122° S, −77.605° W), the world’s highest tropical mountain located in the Cordillera Blanca in the Peruvian Andes (Extended Data Fig. 1). Two cores were drilled to bedrock on the summit of the South Peak (6,768 m asl). Methane measurements were made on discrete samples from SCA, yielding the first such CH₄ concentration record from the tropics.

The most recent 2,000 years of the SCA CH₄ data are compared with CH₄ data from the GISP2 and WAIS Divide cores⁷ to evaluate the latitudinal methane variability (Fig. 1a,b). During the PI period (0 CE to 1850 CE), the GISP2 CH₄ record shows an average concentration that is 48 ppb higher than the WAIS Divide record, indicating dominance of Northern Hemisphere CH₄ emissions⁷. Other polar records show comparable PI values, for example, the North Greenland Eemian Ice Drilling (NEEM) core²⁰ records a PI average only 2 ppb higher than that from the GISP2 core, whereas the average from the Law Dome Antarctic record²¹ is 16 ppb higher than that from the WAIS core. The increase in CH₄ after the PI period is the most prominent feature of the CH₄ records over the past 2,000 years (refs. ^9,20).

The higher methane concentrations in SCA compared with those in the polar records are expected due to the equatorial location of Huascarán and its proximity to dominant tropical sources^2,3. The average SCA CH₄ concentration during the PI is 46 ppb higher than the average GISP2 concentration, 48 ppb higher than NEEM, 78 higher than Law Dome and 94 ppb higher than WAIS Divide (Fig. 1a). The CH₄ concentrations in SCA and the polar records increase concomitantly after around 1850 CE; however, the SCA values remain above those in the Law Dome core but shift approximately 30 ppb below the NEEM record (Fig. 1b). Although timescales and sampling resolutions vary, the trends and magnitudes of the SCA CH₄ record over the past 2,000 years are broadly consistent with the polar records. One common feature in both the SCA and polar CH₄ records is an approximately 30 ppb increase in the late fifteenth century, followed by a decrease into the seventeenth century^7,20,21,22 (Fig. 1a). These types of CH₄ variation may align with periods of historical transition, including the global increase in agricultural emissions^23,24, the rise of the Inca farming civilization in the mid-fifteenth century and the subsequent Spanish conquest in the mid-sixteenth century²⁵. Furthermore, polar records show notable decreases in CH₄, CO₂ and N₂O concentrations after 1550 CE, with the minima occurring around 1600 CE (refs. ^26,27). The SCA record shows a decrease between 1530 and 1630 CE, with CH₄ levels declining from 817 to 730 ppb. This decrease aligns with the colloquially termed ‘Little Ice Age’²⁸. However, these historical events are not necessarily the direct cause of the observed CH₄ fluctuations but are instead helpful markers to assess existing records. Notably, the CH₄ concentrations in the SCA, WAIS⁷ and GISP2⁷ records over the past 2,000 years exhibit high covariance of the first principal component (91%) as well as equivalent loading coefficients (Extended Data Fig. 2), suggesting that they are all influenced by the same factors that cause global CH₄ fluctuations. A comparison between the SCA record and CH₄ concentrations (1985–2012) from the Mauna Loa, Hawaii station²⁹ indicates a ±4 ppb difference during their overlapping time periods, a further indication that the SCA record reflects global fluctuations.

Measurements of δ¹³C-CH₄ (Fig. 1c) provide information about the primary sources of CH₄, as these sources have distinct isotopic fingerprints; biogenic (for example, microbial): −50 to −70‰; pyrogenic (for example, biomass burning): −23‰; and thermogenic (for example, geologic and fossil fuels): −38 to −44‰. The combined source isotopic signature and a sink-driven kinetic isotope fractionation of about 6‰ (ref. ³⁰) result in a PI global atmospheric δ¹³C-CH₄ value of around −48‰ (ref. ²⁸). The SCA δ¹³C-CH₄ record broadly mirrors existing polar records, which show gradual depletion through the 1800s followed by a sharp enrichment after 1850 CE that is commonly linked to increasing fossil fuel emissions⁸. This trend supports the premise that SCA captures a global signal. However, the offset in the SCA isotopic values varies through time and differs between polar core records. Between approximately 1650 and 1820 CE, SCA δ¹³C-CH₄ values are more isotopically depleted than those from both Greenland and Antarctic records (approximately 0.24‰ and 0.52‰, respectively). Around 1900 CE, SCA values are 0.73‰ lower than Antarctic values and lower than both Antarctic and Greenland values around 1530 CE (roughly 1.51‰)^{22,23,31,32,33}. After accounting for the potential influence of firn fractionation³⁴ (Methods), we interpret the SCA δ¹³C-CH₄ values to be indicators of potentially stronger proportions of highly depleted microbial sources during the PI in the tropical regions. However, the small number of δ¹³C-CH₄ samples creates some limitations to this interpretation. Sampling for this analysis was restricted because of the high demand of SCA ice for other analyses, along with the large amount of ice needed for the isotope measurements owing to low air content. Should ice be available after the conclusion of all analyses on the Huascarán summit cores, further δ¹³C-CH₄ sampling would probably improve and solidify this interpretation.

In Greenland ice cores, excess CH₄ is associated with dust and melt features that affect rates of pore close off that indicate melting and refreezing in the ice column^17,35. A likely mechanism for excess CH₄ in Greenland ice is the abiotic decomposition of organics during wet extraction analysis³⁶. However, there is little evidence that any of these complications affect the SCA CH₄ values. There is no significant correlation between SCA CH₄ and Ca²⁺ and, although there is a moderate but negative Pearson correlation coefficient (−0.31; P value < 0.05) between SCA CH₄ and dust concentration, these three records show no visual similarities to each other (Extended Data Fig. 3). The dust and Ca²⁺ concentrations in SCA are also orders of magnitude lower than in other ice cores in which dust is probably responsible for excess CH₄ production (for example, Sajama in Bolivia¹⁶ and the North Greenland Ice Core Project (NGRIP) Greenland core¹⁷); however, the presence of dust does not have an effect on CH₄ in Antarctic cores^27,37. There is no evidence that temperature and air content in the Huascarán summit ice affects the SCA CH₄ concentrations and there are no indications of visible melt layers. Borehole temperatures recorded during drilling were between −9 and −14 °C and the ice was kept frozen during transport from the drill site to its final destination at the BPCRC freezers (Extended Data Fig. 4). The air content of the SCA ice is as expected on the basis of the elevation of the drilling site and that of cores from both polar and lower latitudes^{38,39,40,41,42} (Extended Data Fig. 5). The correspondence between SCA and polar ice core CH₄ trends over the past 2,000 years and the lack of evidence of CH₄ alteration establishes SCA as the first CH₄ low-latitude ice core to capture a consistent global CH₄ record, whereas its slightly elevated average over polar core gas concentrations points towards the inclusion of a low-latitude tropical source.

To better constrain the PI CH₄ latitudinal source strength distribution, we applied a four-box atmospheric model with the intention of partitioning sources across each of the four latitude bands. In previous implementations, the dominance of tropical source strengths remained implicitly inferred through hemispheric transport assumptions, whereas knowing from where the CH₄ originated remained unclear^2,5,8. With the input of only polar cores into the four-box model (Fig. 2c), the average interpolated source strength from 0 to 1800 CE for each of the tropical boxes (0–30° N and 0–30° S) are approximately 82 and 81 Tg year⁻¹, respectively, and the source strength for the northern box (30–90° N) is about 36 Tg year⁻¹. The addition of SCA CH₄ into the tropical south (0–30° S) box results in a 24% increase in combined tropics source strength, from approximately 163 Tg year⁻¹ to 213 Tg year⁻¹ (Fig. 2b). The overall average of the tropical north box increases from 82 to 88 Tg year⁻¹. We note that, in the southern box (30–90° S), the model source is fixed at a constant low level, as very few sources are noted in these latitudes^2,5. Although the model run ends at 1800 CE, we predict that anthropogenic production of CH₄ would create increases in the northern and tropical boxes.

Uncertainties in observations and parameters lead to uncertainties in source strength estimates from observational noise in the ice core CH₄ concentrations, assumed atmospheric lifetimes and latitudinal exchange rates, and general structural limitations owing to the reduced complexity of the box model. We have quantified these uncertainties using a Monte Carlo approach that propagates observational (CH₄ concentration) uncertainty, with a series of sensitivity tests that assess the influence of each lifetime and transport exchange rate (Extended Data Fig. 6; detailed in Methods). Source strength estimates are slightly sensitive to the assumed oxidation and transport parameters but even a 30% perturbation in CH₄ lifetime and exchange rates drawn from previous literature^43,44 produces smaller changes than in the relative latitude partitioning of source strengths from the inclusion of the SCA data. To investigate the sensitivity of the tropical north box interpolation, two more model runs were conducted using the limited data from the East Rongbuk Glacier and the tropical north data generated from the polar-only run as tropical north box inputs. These extra runs generated very slight changes (maximum 3 Tg year⁻¹) in all boxes, indicating that the model is not overly sensitive to the initial interpolated tropical north box (Extended Data Table 1). We note that this model is a simplified framework of atmospheric transport and chemistry and, as such, these derived source strengths will be subject to some degree of structural uncertainty.

Previous studies have only estimated historical tropical contributions from polar records^5,7. The addition of the SCA data confirms that fluctuations in the global CH₄ record reflect variations in tropical sources¹⁵, probably wetlands. These tropical CH₄ emissions are often considered the drivers of long-term atmospheric CH₄ variability, with notable large-scale fluctuations during the transitions into and out of glacial and interglacial stages^5,43. Although the surface area of low-latitude CH₄-producing wetlands has not substantially changed in the past 300 years (ref. ⁴⁵), climatic parameters (temperature, precipitation) have driven CH₄ fluctuations⁴⁶. Considering present CH₄ production drivers^1,2, past wetland reconstructions⁴² and previous estimates⁵, the box model results indicate a greater contribution from equatorial CH₄ sources, highlighting low-latitude data as an important constraint for improving previous estimations^42,47,48. The 2,000-year Huascarán SCA record fills a critical gap in historical CH₄ records and improves our understanding of the importance of tropical sources to the global CH₄ distribution.

Ice core measurements

Fifty-one discrete CH₄ samples between 40–60 g were cut below the firn to ice transition (about 41 m) at depths ranging from 44.01 to 67.40 m (out of the total 69.33 m) in Huascarán’s SCA. The samples were shipped frozen to Oregon State University, where they were analysed for CH₄ concentration using the standard and well-documented wet extraction procedure^5,49. The samples were evacuated for 90 min in a vacuum flask at −60 °C in a refrigerated bath. Once evacuated, the samples were submerged in a warm water bath (50 °C) to melt the ice and release the gas into the headspace above the water. After the samples were completely melted, they were refrozen in the refrigerated bath (−60 °C). The CH₄ concentration in the headspace was measured with a gas chromatograph^7,50. Calculations were performed using methane peak area and headspace pressure, compared with daily measurements of calibrated air standards. The air from each sample was measured four times and the mean standard deviation of these measurements across all samples was 3.7 ppb. All of the CH₄ concentration measurements were conducted in the same laboratory at Oregon State University, removing any potential issues for scale offsets. Not enough ice was available for replicate measurements.

Five more samples (from 47.24, 57.53, 60.76, 63.22 and 64.11 m depth) were analysed for their stable carbon isotopic composition of methane (δ¹³C-CH₄) at Oregon State University following the procedure detailed in previous studies^50,51. Ice samples for δ¹³C-CH₄ analysis were sealed in a glass extraction chamber held at −60 °C while the residual air was evacuated. The sample was melted in an ultrapure helium atmosphere and the air released was transferred to an extraction line using a helium carrier⁵⁰. A series of gas traps separated methane and other hydrocarbons from bulk air before a gas chromatographic column isolated CH₄ from any residual gases (for example, CO and CO₂). The CH₄ was then oxidized to CO₂ using a Thermo GC IsoLink interface and analysed for δ¹³C-CO₂ using a Thermo Delta V isotope ratio mass spectrometer in continuous-flow mode. δ¹³C-CO₂ was translated to δ¹³C-CH₄ by bracketing each sample with reference air standards of known isotopic composition and presented relative to the Vienna Pee Dee Belemnite scale. Besides the ice-melting steps, air standard measurements were treated identically to those of ice samples. Data were not corrected for gravitational settling⁵² or diffusive isotopic fractionation³⁴ during gas transport in the firn column, as the required data to do so did not yet exist. However, we expect these corrections to be small relative to the measured difference in δ¹³C-CH₄ between SCA and other polar records. We note an expected diffusive fractionation in which CH₄ growth rates are highest (for example, the oldest and youngest samples), data may be depleted (between 0.29 and 0.43‰ from diffusive fractionation)³⁴. Sample depths were selected to correspond to PI and post-industrial revolution dates (that is, before and after 1850 CE, respectively) and sample masses used (between 130 to 300 g) were chosen on the basis of the total air content and CH₄ concentration of the two closest samples measured during the initial CH₄ concentration analysis. A careful determination of the sample size was important to limit linearity effects that are caused by differences in the amount of CH₄ within each sample relative to that in each standard air measurement (7 cc of 1,900 ppb CH₄ in air). Linearity corrections were generally lower than 0.02% and much smaller than the longer-term reproducibility of the system (±0.14‰).

All of the sample depths noted above were analysed to catalogue the past 2,000 years of CH₄ concentration and more recent δ¹³C-CH₄ fluctuations. Sampling was limited to the past 2,000 years owing to: (1) the high rate of thinning (Extended Data Fig. 7) below 55 m, which particularly limited sampling below 67.40 m, and (2) limited availability of sufficient ice below 67.40 m owing to ice required for other analyses. Ice availability was further limited for the δ¹³C-CH₄ samples as a result of the low air content in SCA owing to high elevation. The sample size below 64.11 m required for the isotope analysis was approximately 380 g and the ice that remained after other analyses was not sufficient to collect a reliable sample.

Ice core timescale and ice conditions

The timescale for the top 55 m of SCA (2019 to 1910 CE) was established by annual layer counting using seasonal variations in δ¹⁸O and concentrations of dust and nitrate (NO₃⁻) to identify the annual cycles (Extended Data Fig. 8). The requirement for a year to be assigned is the alignment of at least two of the three parameters. The uncertainty in the assigned years at 55 m is ±2 years. Owing to compression, stratigraphy below 55 m thins rapidly (Extended Data Fig. 7), such that annual layer counting becomes difficult. To construct a timescale for this lower portion of SCA, the δ¹⁸O record was matched to the δ¹⁸O record from the Quelccaya summit core (Quelccaya Ice Cap, QIC) from southern Peru⁵³. The QIC is located 910 km southeast of Huascarán in the Cordillera Vilcanota just above the Amazon basin. Both Quelccaya and Huascarán receive most of their precipitation from the tropical Atlantic, carried by easterlies over the Amazon, and it is reasonable to conclude that they have broadly similar δ¹⁸O profiles. Using the AnalySeries software⁵⁴, δ¹⁸O records below the 1910 CE horizon (55.5 m in SCA, 75 m in QIC) in the two cores were matched by depth, allowing the QIC timescale to be transferred to SCA (Extended Data Fig. 9).

The linear correlation coefficient of the δ¹⁸O match between the two cores is +0.63 (P < 0.001). The difficulty of the matching increases with depth, increasing the potential for increasing timescale uncertainties with depth. The much higher stratigraphic thinning rate in SCA compared with QIC means that there is increasingly more time encompassed in each δ¹⁸O sample with depth relative to QIC. Also, the QIC δ¹⁸O data were smoothed with a 41-sample running mean, whereas the shorter SCA record was smoothed using a 9-sample running mean. The resulting timescale has the expected CH₄ rise in the early 1500s, occurring between 63.9 and 64.3 m in SCA, which corresponds to approximately 1480 to 1520 CE ice age (Δage: approximately 19 years; approximately 1500 to 1540 CE gas age), which lies within the time span of the same fluctuation in polar records⁷.

The ice samples were assessed for any potential alteration that may have occurred before or after coring that could have affected the CH₄ concentrations. After inspection, it was concluded that the SCA contains no indication of visible melt layers and all borehole temperatures recorded during drilling were between −9 and −15 °C between 0 and 10 m and around −9 °C from 10 m and below (Extended Data Fig. 4a). Following drilling and during transport, continuous temperature measurements were recorded with thermistors placed inside two randomly selected core section containment tubes. The thermistors indicate adequately low temperatures (consistent and below freezing) after drilling and throughout transport from the field to the freezer at BPCRC (Extended Data Fig. 4b,c). Furthermore, air content of SCA is consistent with what is expected on the basis of the site elevation and air content collected from other cores in both mid-latitude and polar regions (Extended Data Fig. 5). These assessments indicate that modern atmosphere has not influenced the sample concentrations^{38,39,40,41,42}.

Box model

Our model is based on a previously constructed four-box model used to constrain CH₄ source distribution⁵ but adapted to use the SCA CH₄ record as an extra constraint on the signal for the ‘tropical south’ box (0–30° S) instead of relying only on inputs for the northern and southern boxes from polar ice cores. The model consists of four boxes: northern (NH), 30–90° N; tropical north (TN), 0–30° N; tropical south (TS), 0–30° S; and southern (SH), 30–90° S. A Monte Carlo simulation (1,000 iterations) is used to vary CH₄ inputs. True tropical boundaries are at 23.5° N and 23.5° S but 30° N to 30° S are used so that each latitudinal box has an equal air mass.

Initially, the model was run with two inputs, only polar CH₄ data for the NH and SH boxes while solving for the TN and TS boxes using assumed constant transport fluxes. We defined a fixed lifetime for inter-hemispheric exchange and CH₄ lifetime based on previous literature with lifetimes of 15.6 years (NH), 6 years (TN and TS) and 24 years (SH) and transport exchange rates of 0.22 years (NH–TN), 0.45 years (TN–TS) and 0.45 years (TS–SH)^43,44. CH₄ observations were compiled from previously published ice core records GISP2¹⁵ and NEEM²⁰ (NH), WAIS¹⁵ (SH) and SCA (TS). A second model run with three inputs used the SCA record to prescribe CH₄ in the TS box while still using the polar data for NH and SH and solving for TN. Two runs with TN inputs were conducted using the limited data from the East Rongbuk Glacier¹⁹ (800–1800 CE) and the synthetically generated TN data from the polar-only run for the TN box. For all model versions, we assume a total tropospheric mass of 1.78 × 10²⁰ moles of dry air and use a CH₄ molar mass of 16 g mol⁻¹. To facilitate comparison, each record was interpolated onto a common 5-year age spacing based on the WAIS core using linear interpolation. Gaussian noise with a standard deviation of 5 ppb was added independently to each CH₄ value at every time step within each iteration, representing the measurement uncertainty of the system. This perturbation was included to represent the combined analytical and sampling uncertainties in ice core CH₄ measurements. Within the Monte Carlo framework, the observational inputs (CH₄ concentrations) are varied across iterations, the atmospheric lifetimes and transport exchange rates are held, with their influence assessed through the sensitivity analysis described below.

The resulting source strengths (Tg CH₄ year⁻¹) in this model were calculated using the methane concentration inputs, transfer fluxes between each box and loss rate within each box. Each box is connected to adjacent ones by transport fluxes represented as first-order exchange between neighbouring boxes and all boxes are subject to loss (representing all CH₄ sinks), with specific atmospheric lifetimes for each box. The following equation was then used to calculate output source strengths:

to solve for E_i as the calculated source strength (Tg year⁻¹), in which λ_i is the loss rate from oxidation (year⁻¹), C_i and C_j are the burdens in the box and neighbouring box, respectively (Tg), T_ij is the transport time between these boxes (years) and Σ_j is the net transport exchange across all of the boxes (Tg year⁻¹). The annual source strength is determined at each time step based on the above equation. SH source strength was fixed at 10 Tg CH₄ year⁻¹, as most studies suggest that the SH box source strengths are probably very low, have not varied and the SH box only receives transport exchange from higher latitudes⁵. All modelled results were smoothed to a 20-year average to eliminate short-term variability from any observational noise. The TN and TS outputs are summed for a ‘combined tropics’ category for discussion and visual analysis for low-latitude CH₄ source strength. Temporal variability was calculated as the standard deviation across Monte Carlo iterations for each box at each five-year time step. Within the Monte Carlo framework, observational noise is independently resampled for each CH₄ value for every iteration and time step, for which atmospheric lifetimes and transport exchanges are held at their set values.

Uncertainty of each source strength was quantified using the resulting ensemble spread from the Monte Carlo iterations. For each latitudinal box, the uncertainty was calculated as the standard deviation of the time-average source strength across all 1,000 simulations and is expressed as a percentage relative to the mean source strengths for that box. The average uncertainties for the three-input run are as follows: the NH box ±3.18%, the TN box ±0.50%, the TS box ±2.59% and the TS box 0%, as the source strength was hard-capped at 10 Tg year⁻¹. The uncertainty is reflected in the inferred source strength magnitude from both parameter assumption and observational noise. All four run uncertainties are available in Extended Data Table 1.

Sensitivity to the lifetime and exchange rates used were conducted by varying each parameter by ±30%. Although the resulting methane source strength is notably most sensitive to lifetime uncertainty, the lifetimes used are consistent with those used in previous studies^5,43,44. Past atmospheric chemistry model intercomparisons estimate that the CH₄ lifetime differs by approximately 15–34% between PI and present-day conditions, as well as by a similar magnitude across models within a fixed time span^55,56. This places our ±30% range within the uncertainty found in full chemical transport models. Transport rates between boxes are grounded in previous literature^43,44. Model source strengths record modest sensitivity to these terms, with a maximum ±1.6 Tg year⁻¹ between the NH to TN box, ±5.4 Tg year⁻¹ between the TN to TS box and ±17.5 Tg year⁻¹ between the TS to SH box (Extended Data Fig. 6). It should be noted that these source strengths are subject to some degree of structural uncertainty that will not be captured by the Monte Carlo approach, as this four-box model is a framework that represents a simplified explanation of atmospheric transport and the factors that affect methane lifetime. The extra runs with TN inputs derived from the East Rongbuk data from 800 to 1800 CE and the synthetically generated TN box from the polar-only run in the TN box from 0 to 1800 CE served to demonstrate that the initial interpolation of TN (the model run without TN input data) was not overly sensitive in the run with the SCA data. These four input runs resulted in minimal changes to the TN box source strength, with a maximum of 3 Tg year⁻¹. The resulting source strengths are detailed in Extended Data Table 1. Overall, the changes between the polar-only model configuration and the inclusion of the SCA data are more robust in their constraints of CH₄ distribution than they are sensitive to the interpolation of the TN data.

1. Saunois, M. et al. The global methane budget 2000–2017. Earth Syst. Sci. Data 12, 1561–1623 (2020).

   Article  ADS  Google Scholar 

2. Ma, S. et al. Satellite constraints on the latitudinal distribution and temperature sensitivity of wetland methane emissions. AGU Adv. 2, e2021AV000408 (2021).

   Article  ADS  Google Scholar 

3. Kirschke, S. et al. Three decades of global methane sources and sinks. Nat. Geosci. 6, 813–823 (2013).

   Article  ADS  CAS  Google Scholar 

4. Saunois, M. et al. The global methane budget 2000–2012. Earth Syst. Sci. Data 8, 697–751 (2016).

   Article  ADS  Google Scholar 

5. Riddell-Young, B. et al. Atmospheric methane variability through the Last Glacial Maximum and deglaciation mainly controlled by tropical sources. Nat. Geosci. 16, 1174–1180 (2023).

   Article  ADS  CAS  Google Scholar 

6. Chappellaz, J. et al. Changes in the atmospheric CH₄ gradient between Greenland and Antarctica during the Holocene. J. Geophys. Res. Atmos. 102, 15987–15997 (1997).

   Article  ADS  CAS  Google Scholar 

7. Mitchell, L., Brook, E., Lee, J. E., Buizert, C. & Sowers, T. Constraints on the late Holocene anthropogenic contribution to the atmospheric methane budget. Science 342, 964–966 (2013).

   Article  ADS  CAS  PubMed  Google Scholar 

8. Lan, X. et al. Improved constraints on global methane emissions and sinks using δ¹³C-CH₄. Global Biogeochem. Cycles 35, e2021GB007000 (2021).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

9. Thoning, K. W., Lan, X., Crotwell, A. M. & Mund, J. W. Atmospheric methane from quasi-continuous measurements at Barrow, Alaska and Mauna Loa, Hawaii, 1986-present. NOAA Global Monitoring Laboratory (GML), Boulder, Colorado, USA. Version 2025-04-26 (2025).

10. Craig, H., Chou, C. C., Welhan, J. A., Stevens, C. M. & Engelkemeir, A. The isotopic composition of methane in polar ice cores. Science 242, 1535–1539 (1988).

    Article  ADS  CAS  PubMed  Google Scholar 

11. Thompson, A. M. et al. The atmospheric CH₄ increase since the Last Glacial Maximum. Tellus B 45, 242–257 (1993).

    Article  ADS  Google Scholar 

12. Spahni, R. et al. Atmospheric methane and nitrous oxide of the Late Pleistocene from Antarctic ice cores. Science 310, 1317–1321 (2005).

    Article  ADS  CAS  PubMed  Google Scholar 

13. Rhodes, R. H. et al. Atmospheric methane variability: centennial-scale signals in the Last Glacial Period. Global Biogeochem. Cycles 31, 575–590 (2017).

    Article  ADS  CAS  Google Scholar 

14. Heimann, I. et al. Methane emissions in a chemistry-climate model: feedbacks and climate response. J. Adv. Model Earth Syst. 12, e2019MS002019 (2020).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

15. Loulergue, L. et al. Orbital and millennial-scale features of atmospheric CH₄ over the past 800,000 years. Nature 453, 383–386 (2008).

    Article  ADS  CAS  PubMed  Google Scholar 

16. Thompson, L. G. et al. A 25,000-year tropical climate history from Bolivian ice cores. Science 282, 1858–1864 (1998).

    Article  ADS  CAS  PubMed  Google Scholar 

17. Lee, J. E. et al. Excess methane in Greenland ice cores associated with high dust concentrations. Geochim. Cosmochim. Acta 270, 409–430 (2020).

    Article  ADS  CAS  Google Scholar 

18. Yao, T. et al. Temperature and methane changes over the past 1000 years recorded in Dasuopu glacier (central Himalaya) ice core. Ann. Glaciol. 35, 379–383 (2002).

    Article  CAS  Google Scholar 

19. Hou, S. et al. A new Himalayan ice core CH₄ record: possible hints at the preindustrial latitudinal gradient. Clim. Past 9, 2549–2554 (2013).

    Article  Google Scholar 

20. Rhodes, R. H. et al. Continuous methane measurements from a late Holocene Greenland ice core: atmospheric and in-situ signals. Earth Planet. Sci. Lett. 368, 9–19 (2013).

    Article  ADS  CAS  Google Scholar 

21. Rubino, M. et al. Revised records of atmospheric trace gases CO₂, CH₄, N₂O, and δ¹³C-CO₂ over the last 2000 years from Law Dome, Antarctica. Earth Syst. Sci. Data 11, 473–492 (2019).

    Article  ADS  Google Scholar 

22. Mischler, J. A. et al. Carbon and hydrogen isotopic composition of methane over the last 1000 years. Global Biogeochem. Cycles 23, GB4024 (2009).

    Article  ADS  Google Scholar 

23. Ferretti, D. F. et al. Unexpected changes to the global methane budget over the past 2000 years. Science 309, 1714–1717 (2005).

    Article  ADS  CAS  PubMed  Google Scholar 

24. Houweling, S., Van Der Werf, G. R., Klein Goldewijk, K., Röckmann, T. & Aben, I. Early anthropogenic CH₄ emissions and the variation of CH₄ and ¹³CH₄ over the last millennium. Global Biogeochem. Cycles 22, GB1002 (2008).

    Article  ADS  Google Scholar 

25. Frogley, M. R., Chepstow-Lusty, A., Thiele, G. & Chutas, C. A. Trees, terraces and llamas: resilient watershed management and sustainable agriculture the Inca way. Ambio 54, 7930–807 (2025).

    Article  Google Scholar 

26. Etheridge, D. M., Steele, L. P., Francey, R. J. & Langenfelds, R. L. Atmospheric methane between 1000 AD and present: evidence of anthropogenic emissions and climatic variability. J. Geophys. Res. Atmos. 103, 15979–15993 (1998).

    Article  ADS  CAS  Google Scholar 

27. Macfarling Meure, C. et al. Law Dome CO₂, CH₄ and N₂O ice core records extended to 2000 years BP. Geophys. Res. Lett. 33, L14810 (2006).

    Article  ADS  Google Scholar 

28. Jones, P. D. & Mann, M. E. Climate over past millennia. Rev. Geophys. 42, 2003RG000143 (2004).

    Article  Google Scholar 

29. Lan, X., Thoning, K. W. & Dlugokencky, E. J. Trends in globally-averaged CH₄, N₂O, and SF₆ determined from NOAA Global Monitoring Laboratory measurements. Version 2026-02 (2026).

30. Cantrell, C. A. et al. Carbon kinetic isotope effect in the oxidation of methane by the hydroxyl radical. J. Geophys. Res. Atmos. 95, 22455–22462 (1990).

    Article  ADS  CAS  Google Scholar 

31. Beck, J. et al. Bipolar carbon and hydrogen isotope constraints on the Holocene methane budget. Biogeosciences 15, 7155–7175 (2018).

    Article  ADS  CAS  Google Scholar 

32. Sapart, C. J. et al. Natural and anthropogenic variations in methane sources during the past two millennia. Nature 490, 85–88 (2012).

    Article  ADS  CAS  PubMed  Google Scholar 

33. Sowers, T. Atmospheric methane isotope records covering the Holocene period. Quat. Sci. Rev. 29, 213–221 (2010).

    Article  ADS  Google Scholar 

34. Buizert, C., Sowers, T. & Blunier, T. Assessment of diffusive isotopic fractionation in polar firn, and application to ice core trace gas records. Earth Planet. Sci. Lett. 361, 110–119 (2013).

    Article  ADS  CAS  Google Scholar 

35. Rhodes, R. H. et al. Local artifacts in ice core methane records caused by layered bubble trapping and in situ production: a multi-site investigation. Clim. Past 12, 1061–1077 (2016).

    Article  Google Scholar 

36. Mühl, M. et al. Methane, ethane, and propane production in Greenland ice core samples and a first isotopic characterization of excess methane. Clim. Past 19, 999–1025 (2023).

    Article  Google Scholar 

37. Fourteau, K. et al. Estimation of gas record alteration in very low-accumulation ice cores. Clim. Past 16, 503–522 (2020).

    Article  Google Scholar 

38. Martinerie, P., Raynaud, D., Etheridge, D. M., Barnola, J.-M. & Mazaudier, D. Physical and climatic parameters which influence the air content in polar ice. Earth Planet. Sci. Lett. 112, 1–13 (1992).

    Article  ADS  Google Scholar 

39. Jiule, L., Baiqing, X. & Chappellaz, J. Variations of air content in Dasuopu ice core from AD 1570–1927 and implications for climate change. Quat. Int. 236, 91–95 (2011).

    Article  Google Scholar 

40. Oyabu, I. et al. New technique for high-precision, simultaneous measurements of CH₄, N₂O and CO₂ concentrations; isotopic and elemental ratios of N₂, O₂ and Ar; and total air content in ice cores by wet extraction. Atmos. Meas. Tech. 13, 6703–6731 (2020).

    Article  CAS  Google Scholar 

41. Raynaud, D., Chappellaz, J., Ritz, C. & Martinerie, P. Air content along the Greenland Ice Core Project core: a record of surface climatic parameters and elevation in central Greenland. J. Geophys. Res. Oceans 102, 26607–26613 (1997).

    Article  ADS  Google Scholar 

42. Eicher, O. et al. Climatic and insolation control on the high-resolution total air content in the NGRIP ice core. Clim. Past 12, 1979–1993 (2016).

    Article  Google Scholar 

43. Brook, E. J. et al. On the origin and timing of rapid changes in atmospheric methane during the last glacial period. Global Biogeochem. Cycles 14, 559–572 (2000).

    Article  ADS  CAS  Google Scholar 

44. Rasmussen, R. A. & Khalil, M. A. K. Atmospheric methane in the recent and ancient atmospheres: concentrations, trends, and interhemispheric gradient. J. Geophys. Res. Atmos. 89, 11599–11605 (1984).

    Article  ADS  CAS  Google Scholar 

45. Fluet-Chouinard, E. et al. Extensive global wetland loss over the past three centuries. Nature 614, 281–286 (2023).

    Article  ADS  CAS  PubMed  Google Scholar 

46. Basso, L. S. et al. Amazon methane budget derived from multi-year airborne observations highlights regional variations in emissions. Commun. Earth Environ. 2, 246 (2021).

    Article  ADS  Google Scholar 

47. Severinghaus, J. P. & Brook, E. J. Abrupt climate change at the end of the Last Glacial Period inferred from trapped air in polar ice. Science 286, 930–934 (1999).

    Article  CAS  PubMed  Google Scholar 

48. Kaplan, J. O., Folberth, G. & Hauglustaine, D. A. Role of methane and biogenic volatile organic compound sources in late glacial and Holocene fluctuations of atmospheric methane concentrations. Global Biogeochem. Cycles 20, GB2016 (2006).

    Article  ADS  Google Scholar 

49. Miller, J. B. Development of analytical methods and measurements of ¹³C/¹²C in atmospheric CH₄ from the NOAA Climate Monitoring and Diagnostics Laboratory Global Air Sampling Network. J. Geophys. Res. Atmos. 107, 4178 (2002).

    Article  ADS  Google Scholar 

50. Riddell-Young, B. et al. Abrupt changes in biomass burning during the last glacial period. Nature 637, 91–96 (2025).

    Article  ADS  CAS  PubMed  Google Scholar 

51. Dlugokencky, E. J. et al. Conversion of NOAA atmospheric dry air CH₄ mole fractions to a gravimetrically prepared standard scale. J. Geophys. Res. Atmos. 110, 2005JD006035 (2005).

    Article  Google Scholar 

52. Craig, H., Horibe, Y. & Sowers, T. Gravitational separation of gases and isotopes in polar ice caps. Science 242, 1675–1678 (1988).

    Article  ADS  CAS  PubMed  Google Scholar 

53. Thompson, L. G. et al. Annually resolved ice core records of tropical climate variability over the past ~1800 years. Science 340, 945–950 (2013).

    Article  ADS  CAS  PubMed  Google Scholar 

54. Paillard, D., Labeyrie, L. D. & Yiou, P. AnalySeries 1.0: a Macintosh software for the analysis of geophysical time-series. Eos 77, 379 (1996).

    Article  ADS  Google Scholar 

55. Naik, V. et al. Preindustrial to present-day changes in tropospheric hydroxyl radical and methane lifetime from the Atmospheric Chemistry and Climate Model Intercomparison Project (ACCMIP). Atmos. Chem. Phys. 13, 5277–5298 (2013).

    Article  ADS  Google Scholar 

56. Osborn, T. J. & Wigley, T. M. L. A simple model for estimating methane concentration and lifetime variations. Clim. Dyn. 9, 181–193 (1994).

    Article  Google Scholar 

57. Lamantia, K. A global atmospheric methane record from a tropical ice core (v2). Zenodo https://doi.org/10.5281/zenodo.18657346 (2026).

Download references

Authors and Affiliations

1. Byrd Polar and Climate Research Center, Ohio State University, Columbus, OH, USA

   Kara A. Lamantia, Lonnie G. Thompson, Mary E. Davis, Emilie Beaudon & Ellen Mosley-Thompson

2. School of Earth Sciences, Ohio State University, Columbus, OH, USA

   Kara A. Lamantia, Lonnie G. Thompson & Emilie Beaudon

3. School of Geographical Sciences, University of Bristol, Bristol, UK

   Kara A. Lamantia

4. Cooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado Boulder, Boulder, CO, USA

   Ben Riddell-Young

5. College of Earth, Ocean, and Atmospheric Sciences (CEOAS), Oregon State University, Corvallis, OR, USA

   Ivo Strawson & Edward J. Brook

6. Doerr School of Sustainability, Stanford University, Stanford, CA, USA

   Newton Nguyen

Contributions

The CH₄ data analysis was conducted by K.A.L., L.G.T., M.E.D., E.B., E.M.-T. and B.R.-Y. The chronology was developed by M.E.D. The CH₄ isotopic data were measured and processed by I.S. The four-box model was developed by K.A.L., B.R.-Y., N.N. and E.J.B. The initial paper was prepared by K.A.L. and all authors contributed to the final paper.

Corresponding author

Correspondence to Kara A. Lamantia.

Competing interests

The authors declare no competing interests.

Peer review information

Nature thanks Alison Criscitiello and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

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 licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence 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 licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions