Author Correction: Broadly stable atmospheric CO₂ and CH₄ levels over the past 3 million years

Correction to: Nature https://doi.org/10.1038/s41586-025-10032-y Published online 18 March 2026

Following publication of this article, we are correcting Extended Data Fig. 8. This figure contained boron-based CO₂ from the compilation by Rae et al., 2021 (ref. 18 in the article), which had slightly different criteria for excluding studies than the CenCO₂PIP Consortium dataset that we used for Fig. 3. For continuity, we are replacing Extended Data Fig. 8b with a new version that contains boron-based CO₂ from the CenCO₂PIP database. The previous Extended Data Fig. 8b contained 176 points distinguished by core site (upside-down triangles for site 999, plus sign for site 926 and squares for site 668). The new Extended Data Fig. 8b contains 32 additional data points and distinguishes boron-based CO₂ reconstructions by foraminiferal species instead of core site (G. ruber (upside-down triangles) and T. trilobus (plus signs)). The added data produced from T. trilobus (plus signs) were originally from Bartoli et al.¹ and exhibit a closer agreement with our new ice core CO₂ record (this study). These data were always included in Fig. 3 and thus this correction does not change that figure. Additionally, the foraminiferal species were incorrectly assigned to the various studies and have now been corrected in the new figure and in the wording in the Methods. Four boron isotope proxy studies used G. ruber (Foster et al., 2008 (ref. 79), Chalk et al., 2017 (ref. 35), Martinez-Botí et al., 2015 (ref. 80) and de la Vega et al., 2020 (ref. 83)) while three used T. trilobus (Hönisch et al., 2009 (ref. 36), Dyez et al., 2018 (ref. 81) and Sosdian et al., 2018 (ref. 82)). Note that Sosdian et al. also include G. ruber data from the paper by Chalk et al.. We thank Bärbel Hönisch and Thomas Chalk for bringing these corrections to our attention.

Additionally, we have made a small correction to the code that produced Extended Data Figs. 5 and 6. The EDC layer thickness curve exaggerated by one order of magnitude (yellow) was missing an interglacial (Marine Isotope Stage 11) that was included in all other layer thickness curves in our experiment. The implication can be observed in the light blue data in Extended Data Fig. 6: including MIS11 shifts the light blue distribution of CH₄, CO₂ and δ¹⁸O_atm closer to the pink distributions. This correction has no bearing on the conclusions of the paper. We have edited the sentence in the Methods pertaining to this data.

Extended Data Figs. 5, 6, 8 and Methods text have been amended in the HTML and PDF versions of the article.