Terrestrial records of two hyperthermal events in the Cretaceous-Paleogene boundary suggest different control mechanisms

A series of late Cretaceous–early Paleogene hyperthermals show transient warming and negative carbon isotope excursions (CIEs). Two events across the K–Pg boundary, the Late Maastrichtian Warming Event (LMWE) and the Early Danian Dan‑C2 event, are linked to major climatic and biotic changes. The Dan‑C2 event is documented in the Atlantic and Tethys and parts of Eurasia, but marine records show inconsistencies: CIEs are mainly in bulk and planktonic records, benthic signals are scarce, and bottom-water warming is generally lacking, raising questions about its global significance and whether it is a true hyperthermal. Two hypothesized drivers for the LMWE and Dan‑C2 CIEs are Deccan volcanism and carbon pools modulated by orbital forcing. Both CIEs align with maxima of the 405‑kyr eccentricity cycle, and Dan‑C2’s double CIEs align with two 100‑kyr eccentricity maxima, implicating orbital forcing. Overlap with Deccan volcanism suggests volcanic CO2 may have contributed, especially to the LMWE, but causal links remain debated, and CO2 release scenarios from Deccan differ among models. Most existing carbon isotope records are marine; terrestrial records are scarce. This study generates a high‑resolution terrestrial δ13Ccarb record from the Nanxiong Basin across the K–Pg boundary to: (1) document terrestrial expressions of the LMWE and Dan‑C2 and assess Dan‑C2’s global significance; and (2) evaluate relative contributions of Deccan volcanism and orbital forcing to carbon‑cycle perturbations during these events.

Prior work has identified LMWE and Dan‑C2 in marine sequences of the Atlantic and Tethys and in some Eurasian terrestrial settings. Marine Dan‑C2 records often show CIEs confined to bulk and planktonic foraminifera, with weak or absent benthic responses and limited evidence for deep‑water warming. Hyperthermals often coincide with maxima of the 405‑kyr eccentricity cycle, and Dan‑C2’s twin CIEs correspond to 100‑kyr eccentricity maxima, indicating orbital pacing. Deccan volcanism temporally overlaps both events, but its causal role is unresolved due to conflicting CO2 emission reconstructions (more than half, half, or one‑third pre‑K–Pg). PETM studies show larger terrestrial than marine CIEs, with marine signals attenuated by carbonate dissolution and sedimentation rate effects, and amplification in terrestrial systems via environmental and photosynthetic fractionation. Similar processes likely influenced Dan‑C2 marine CIE magnitudes and completeness. Post‑K–Pg ocean carbon cycling exhibited spatial heterogeneity (“Resilient”/“Heterogeneous Ocean” concepts), with variable primary/export productivity that could yield inconsistent regional δ13C responses. These contexts frame the need for terrestrial records to constrain mechanisms and global significance of LMWE and Dan‑C2.

Study area and sampling: The Nanxiong Basin (southeastern China) preserves continuous Upper Cretaceous–lower Paleocene red fluvial–lacustrine clastics. In the CGD–CGY (Datang) section dominated by muddy siltstone/silty mudstone with interbedded sandstone and conglomerate, 274 fresh bulk samples were collected at ~1 m spacing from the upper Zhenshui Formation to the lower Xiahui part of the Shanghu Formation. Chronology follows Ma et al., tied to the paleomagnetic framework of Clyde et al., with the K–Pg boundary as a control point.

Carbon isotope analyses: Bulk carbonate δ13C was measured at IGGCAS using a Thermo Fisher MAT 253 IRMS with GasBench II. Carbonate reacted with phosphoric acid at 72 ± 0.1 °C for 60 min in continuous flow; CO2 was carried by 99.999% He to the IRMS. Analytical precision (1σ) for δ13C was better than 0.15‰ based on replicate internal calcite standards. Results are reported relative to V‑PDB. Comparisons with previously measured pedogenic carbonate nodules indicate minimal diagenetic overprint; no significant δ13C–δ18O correlation supports this.

Time-series analysis: To assess orbital pacing, δ13C records from Nanxiong and ODP Site 1262 were analyzed with Acycle. Series were linearly interpolated at 5 kyr (Nanxiong) and 2 kyr (ODP 1262) and detrended using LOWESS. Fast Fourier Transform evolutionary power spectra were computed with sliding windows of 500 kyr for complete records and 150 kyr for discrete windows that exclude large, abrupt excursions. Significance of periodicities (405‑kyr long and 100‑kyr short eccentricity) was evaluated across time windows.

Comparative datasets and proxies: The terrestrial δ13Ccarb record was compared to marine δ13Cbulk at ODP Site 1262, Deccan eruption models based on U‑Pb chronologies (Schoene et al., Sprain et al.), and La2010b orbital solutions (filtered to 405‑kyr eccentricity). Published Hg/Al2O3 data from Nanxiong provided an independent volcanism proxy. Cross‑site comparisons of CIE magnitudes versus paleo‑depth, sediment thickness, and carbonate concentration were compiled for PETM and Dan‑C2 sites to assess effects of dissolution and sedimentation on recorded CIEs. Long‑term ocean‑atmosphere‑sediment carbon cycle (LOSCAR) simulations were used to evaluate plausible Deccan CO2 release scenarios before versus after the K–Pg boundary.

• The Nanxiong Basin terrestrial δ13Ccarb record clearly resolves both LMWE and Dan‑C2 across the K–Pg boundary. δ13Ccarb declines from 66.4 to ~66.2 Ma, with a muted negative CIE of ~1.5‰ at ~66.2 Ma (LMWE), then gradually recovers towards the boundary. Immediately post‑boundary, δ13Ccarb drops sharply and then shows double negative excursions at ~65.8 Ma (~3‰) and ~65.7 Ma (~2‰), characteristic of the Dan‑C2 event; an additional ~2‰ excursion at ~65.3 Ma corresponds to the Lower C29n event.
• Event durations differ: LMWE onset–peak–recovery spans ~200–300 kyr, significantly longer than each Dan‑C2 CIE (~100 kyr). Terrestrial CIE magnitudes (1.5–3‰) exceed those at ODP 1262 (marine bulk 0.5–1‰), consistent with generally larger terrestrial responses.
• Orbital pacing: Both terrestrial (Nanxiong) and marine (ODP 1262) records exhibit significant 405‑kyr eccentricity cycles from ~66.6 to 65 Ma. The 100‑kyr short‑eccentricity cycle is robust in terrestrial records both below and above the K–Pg boundary but weak to absent in the ODP 1262 marine record from ~66.2 to 66.0 Ma (LMWE interval), reappearing after the boundary.
• Marine Dan‑C2 inconsistencies are explained by oceanographic and preservational effects: Across sites, CIEbulk magnitudes during Dan‑C2 correlate negatively with paleo‑depth and positively with sediment thickness and carbonate concentration, indicating attenuation by carbonate dissolution at greater depths and incomplete recording at low sedimentation rates; analogous patterns are known for the PETM.
• Mechanistic interpretation: Multiple lines of evidence (Hg anomalies and isotopes, overlap with central Deccan activity, modeling) indicate Deccan volcanism and orbital forcing both contributed but with differing dominance. Early Deccan magmas were enriched in CO2; passive degassing of several thousand Gt C with δ13C ≈ −5‰ likely triggered a muted, prolonged LMWE CIE and disrupted marine short‑eccentricity expression, while still enhancing climate sensitivity to orbital forcing to yield ~4 °C warming. After the K–Pg, reduced volcanic CO2 output and greater influence of orbitally modulated, 13C‑depleted carbon pools (e.g., peat, methane hydrates) produced larger Dan‑C2 CIEs and strong 100‑kyr cyclicity, with limited deep‑ocean warming.
• LOSCAR simulations support greater volume and/or rate of Deccan CO2 release before the K–Pg boundary (e.g., intrusive:extrusive ≈ 5:1 or 50:50 with higher pre‑K–Pg rate), consistent with volcanism contributing more to LMWE than to Dan‑C2.

The new terrestrial δ13Ccarb record from Nanxiong directly addresses uncertainties about the expression and drivers of LMWE and Dan‑C2. Clear identification of both events on land and close correspondence with marine bulk δ13C establishes their broader geographic footprint and supports Dan‑C2 as a bona fide hyperthermal. The longer duration and muted magnitude of the LMWE CIE, combined with the disappearance of 100‑kyr eccentricity power in marine records during ~66.2–66.0 Ma, implicate a strong volcanic perturbation to the carbon cycle, likely via passive degassing of CO2 from early Deccan magmas. In contrast, the Dan‑C2 event’s double CIEs align with 100‑kyr eccentricity maxima, with robust short‑eccentricity pacing in both terrestrial and marine records, pointing to dominant orbital control of carbon pools with more negative δ13C, and reduced volcanic influence. Cross‑site analyses demonstrate that dissolution and sedimentation effects can suppress marine CIE magnitudes, explaining regional inconsistencies, while post‑extinction heterogeneity in primary/export productivity further contributed to variable marine δ13C responses. Together, these findings reconcile terrestrial–marine differences and differentiate the relative roles of Deccan volcanism and orbital forcing across the two hyperthermals.

This study provides a high‑resolution terrestrial δ13Ccarb record from the Nanxiong Basin that resolves the LMWE and Dan‑C2 across the K–Pg boundary and enables detailed comparison with marine records. The results expand the documented global distribution of both events and support Dan‑C2’s classification as a true hyperthermal. We propose that passive degassing of CO2 from early Deccan magmas muted and prolonged the LMWE CIE and enhanced climate sensitivity to orbital forcing, whereas the Dan‑C2 event was primarily paced by orbital forcing acting on 13C‑depleted carbon reservoirs under diminished volcanic influence. LOSCAR simulations corroborate greater Deccan CO2 emissions prior to the K–Pg boundary. Future work should target additional high‑resolution terrestrial and marine records to better quantify spatial heterogeneity in post‑K–Pg ocean carbon cycling, refine Deccan CO2 emission histories and mechanisms, and integrate multiproxy constraints on temperature, dissolution, and sedimentation effects on CIE recording.

• Marine records of Dan‑C2 remain geographically sparse and uneven, leading to unresolved spatial heterogeneity in δ13C responses and incomplete understanding of underlying mechanisms.
• Attenuation of marine CIEs by carbonate dissolution and variable sedimentation rates complicates cross‑site magnitude comparisons and may underrepresent true excursion amplitudes.
• The precise magnitude and temporal distribution of Deccan CO2 emissions are debated; passive degassing scenarios and their sufficiency to produce ~4 °C warming remain uncertain.
• The study relies on bulk carbonate δ13C; although diagenesis was assessed and found minimal, multi-phase carbonates or site-specific biases cannot be entirely excluded.
• Modeling (LOSCAR) scenarios are simplified and do not capture all complexities of coupled climate–carbon cycle feedbacks and post‑extinction ecosystem dynamics.