Reconciling ice core CO₂ and land-use change following New World-Old World contact

Ice cores contain trapped ancient air that allows direct reconstruction of past atmospheric greenhouse gases. For the Common Era, the Law Dome and WAIS Divide ice cores provide the highest-quality CO₂ records, but they differ in both absolute values and variability, complicating interpretation of multidecadal features. A prominent example is the sharp CO₂ minimum centered around 1610 CE in the Law Dome record (the Orbis Spike), previously hypothesized to reflect pandemic-driven depopulation and land abandonment in the Americas after European contact. However, the WAIS Divide record lacks this sharp minimum, instead showing a gradual decline. This study aims to reconcile these records by providing a new high-resolution CO₂ dataset from the Skytrain Ice Rise core (1454–1688 CE), assessing the role of firn smoothing, and quantifying implied land carbon fluxes to evaluate whether a sharp or gradual CO₂ decline best represents the true atmospheric history.

Previous work established Law Dome as a high-accumulation site with minimal firn smoothing and early pioneering CO₂ measurements, though sampling resolution varied and datasets have undergone revisions. WAIS Divide offers higher analytical precision and sampling density but greater firn-induced smoothing. Differences between these records raise questions: some rapid CO₂ swings in Law Dome are challenging to reconcile with current carbon cycle understanding and Earth System Models, and absolute offsets of up to ~6 ppm exist between records for 750–1800 CE. The 1610 CE minimum in Law Dome has been linked to New World-Old World contact and land abandonment (Orbis hypothesis) and proposed as an Anthropocene marker. Land-use reconstructions differ substantially between low-end HYDE-based scenarios and high-end KK10-based scenarios, affecting expected carbon uptake. Methane records provide an independent constraint on firn smoothing because CH₄ exhibits multidecadal variability that should be similarly smoothed by firn processes.

• Core and interval: The Skytrain Ice Rise core (West Antarctica) was drilled to 651 m (2018/2019). This study analyzed gas ages 1454–1688 CE from depths 83.2–104.0 m. Accumulation ~13.5 cm w.e. yr⁻¹; 10 m temperature −25.9 °C. Chronology from ST22-WD, supported by prior discrete CH₄ measurements (~20-year resolution).
• CO₂ measurements: 31 depth intervals measured with the Oregon State University needle crusher and GC. Each sample had 2–3 replicates (true or proximal), pooled SD 1.0 ppm. Calibration to NOAA WMO X2019; average full-system blank correction 0.8 ppm; gravitational correction based on δ15N (avg ~0.7 ppm) applied. Data and errors provided in Supplementary Data.
• Continuous CH₄: 16.5 m of ice (1426–1617 CE) measured via Continuous Flow Analysis (BAS) with SARA spectrometer at 4 Hz. Calibrated to WMO X2004A, with dissolution correction (avg 11.7%) and binning at 30 s (pooled RSD 0.9%, ~6.3 ppb). Gaps due to prior sampling and breaks noted.
• Smoothing splines: Bootstrapped (n=10,000) cubic smoothing splines (MATLAB csaps), tuned per record for a 50-year cut-off (100-year for combined compilation). Means of iterations used as splines; uncertainties from iteration SDs.
• Firn filter modeling: OSU Firn Air Model used to generate site-specific age distributions (firn filters) for WAIS Divide and Skytrain, including processes of diffusion, advection, mixing, and bubble closure; CO₂ diffusivity 0.10 cm² s⁻¹. Validation by comparison with an independent WAIS Divide filter (CIC model) showed close agreement. Log-logistic functions were also tuned to represent and artificially enhance smoothing for sensitivity tests.
• Convolution tests: The Law Dome CO₂ series was convolved with WAIS Divide and Skytrain firn filters (and with enhanced log-logistic filters) to test whether a true atmospheric 1610 CE minimum would survive smoothing. Analogous tests applied to CH₄ to evaluate plausibility of enhanced smoothing. All convolutions performed at 0.2-year resolution with normalized filters.
• Deconvolution of land fluxes: The OSU Carbon Cycle Box Model (0.25-year timestep; 5000-year preindustrial spin-up) performed single deconvolution experiments to infer net land-atmosphere carbon fluxes needed to reproduce CO₂ histories from splines (50-year cut-off for individual cores; 100-year for combined). Ocean-atmosphere flux evolved freely. Results compared against bottom-up land-use scenarios (low-end HYDE 3.1/3.2 vs high-end KK10) implemented with LPX DGVM from literature. Offsets among cores were removed before combined analysis (+1.4 ppm to Law Dome excluding three low points near 1610; −1.3 ppm to WAIS Divide).

• New Skytrain CO₂ record (1454–1688 CE) shows a gradual 17th-century decline, not a sharp dip. Spline-based metrics: 8.0 ppm decrease over 157 years (max 285.0±1.2 ppm at 1515 CE to min 277.0±1.7 ppm at 1672 CE), average decrease rate ~0.5 ppm per decade.
• Comparison with existing records: Law Dome indicates a larger, faster decline (9.9 ppm over 84 years; ~1.2 ppm/decade; distinct minimum ~1610 CE), while WAIS Divide shows a smaller, gradual decline (6.7 ppm over 117 years; ~0.6 ppm/decade). Skytrain aligns with WAIS Divide in rate and timing.
• Firn smoothing tests: Convolution of the Law Dome CO₂ record with WAIS Divide and Skytrain firn filters preserves a pronounced 1610 CE minimum, contrary to actual WAIS Divide and Skytrain measurements. Eliminating the minimum requires artificially enhanced smoothing far beyond modeled firn age distributions; when applied to CH₄, such smoothing excessively damps observed CH₄ variability (~50% reduction), demonstrating incompatibility. Hence, the Law Dome 1610 CE CO₂ minimum is unlikely to be an atmospheric feature and is plausibly due to a small number of anomalously low data points.
• Inferred land carbon sink: Deconvolution indicates for 1570–1620 CE a mean net land sink of ~0.24±0.10 PgC/yr (WAIS Divide), ~0.28±0.13 PgC/yr (Skytrain), and ~0.26±0.05 PgC/yr in a combined compilation (equivalent to ~2.6 PgC per decade). The Law Dome-based deconvolution requires a much sharper and larger sink (~0.47±0.19 PgC/yr with peaks up to 1 PgC/yr at 1590–1600 CE), inconsistent with other cores.
• Model–data consistency: High-end land-use change scenarios (KK10-based) are consistent with the gradual sink inferred from WAIS Divide, Skytrain, and combined records, whereas low-end HYDE-based scenarios are inconsistent. A rapid, deep 1610 dip is incompatible with even extreme land-use change scenarios.
• Climate context: Decadal Northern Hemisphere coolings around 1600 CE could have contributed to gradual land uptake, but similar or colder decades elsewhere in the millennium did not produce comparable sharp CO₂ dips, arguing against a unique, rapid climate-driven CO₂ minimum at 1610 CE.

The Skytrain record supports a reconciled view in which atmospheric CO₂ declined gradually through the 16th–17th centuries rather than exhibiting an abrupt 1610 CE minimum. This resolves discrepancies between Law Dome and WAIS Divide by demonstrating that firn smoothing cannot explain the absence of a sharp minimum in the latter records and that the pronounced Law Dome minimum likely reflects non-atmospheric artefacts. The inferred land sink of ~0.26 PgC/yr (1570–1620 CE) aligns with high-end KK10 land-use change scenarios following New World-Old World contact, supporting the hypothesis of substantial land abandonment and regrowth in the Americas contributing to CO₂ drawdown. Conversely, the magnitude and abruptness implied by the Law Dome minimum are incompatible with both firn physics and realistic land-use change dynamics. While natural climate variability may have contributed to a gradual terrestrial uptake, the lack of comparable sharp CO₂ dips during other notable cool intervals suggests no exceptional, rapid climate-carbon feedback is necessary around 1610 CE. The results refine the atmospheric CO₂ history for climate-carbon modeling and IPCC-relevant applications and underscore the need to reassess the use of the 1610 CE feature as an Anthropocene marker.

This study presents a new high-resolution CO₂ record from the Skytrain Ice Rise that, together with WAIS Divide, indicates a gradual 6–8 ppm decline in atmospheric CO₂ from the early 16th to late 17th century. Convolution and CH₄-constraint tests show that the sharp 1610 CE minimum in the Law Dome record is unlikely to be atmospheric and is probably due to a small number of anomalous data points. Deconvolution of the combined CO₂ history implies a sustained terrestrial carbon sink of about 0.26±0.05 PgC/yr (1570–1620 CE), consistent with high-end land-use change scenarios after New World-Old World contact. Future work should focus on remeasurement or acquisition of new, very high-resolution Antarctic cores, improved high-resolution CH₄ and CO₂ isotope records (especially from Law Dome), and refined modeling of population dynamics and land-use change at higher temporal resolution to further constrain the drivers of the preindustrial CO₂ decline.

• Evidence against the 1610 CE minimum is indirect; the specific cause of the low Law Dome values remains unknown (potential post-coring storage or other artefacts).
• Firn models, while validated and intercompared, simplify complex layering and bubble closure processes; uncertainties in diffusivity and age distributions persist.
• High-resolution continuous CH₄ data from Law Dome are lacking, limiting independent constraints on firn smoothing; Skytrain CH₄ measurements do not extend through all predicted features (e.g., post-1600 leveling).
• Offsets between absolute CO₂ levels among cores (Law Dome vs WAIS Divide vs Skytrain) remain not fully explained.
• Bottom-up land-use scenarios depend on coarse (century-scale) population reconstructions and assumptions about per-capita land use, adding uncertainty to comparisons with top-down flux estimates.