Cold spells in the Nordic Seas during the early Eocene Greenhouse

The Eocene (56.0–33.9 Ma) represents the last epoch of pervasive greenhouse conditions. Despite extensive evidence for globally warm climates during the early Eocene, reports of glacial sediments and ikaite-derived glendonites at mid to high latitudes have suggested episodic cooler intervals. Glendonites, calcite pseudomorphs after hydrated calcium carbonate (ikaite), are traditionally regarded as cold-climate indicators because natural marine ikaite typically forms at ≤4 °C under conditions that inhibit anhydrous CaCO3 precipitation. Yet laboratory studies have demonstrated ikaite can nucleate at higher temperatures, casting doubt on the climatic significance of glendonites. The early Eocene Fur Formation (Denmark) contains abundant, unusually large glendonites within a depositional setting otherwise indicative of warm, subtropical conditions, creating a paradox. This study tests whether these glendonites indeed record near-freezing bottom waters by using clumped isotope thermometry, coupled with geochemical and biomarker analyses, to quantify temperatures and assess the environmental processes that enabled ikaite formation during a greenhouse world.

Previous studies document glendonites and possible ice-rafted debris from Eocene high to mid-latitudes (e.g., North America, Arctic Svalbard, Denmark), implying intermittent cool phases amid overall warmth. However, poor preservation and lack of quantitative temperature estimates have limited interpretations, and successful laboratory stabilization of ikaite at ~35 °C raised questions about using glendonites as paleotemperature indicators. Existing regional TEX86-derived SST records in the North Sea–Nordic Seas indicate cooling following the PETM, but with limited resolution. Global compilations of early Eocene SSTs and bottom-water temperatures generally show warm conditions, with Arctic SSTs ~17–25 °C and Antarctic SSTs ~24–34 °C, inconsistent with annual freezing. Hypotheses for regional cooling include aerosol-induced short-lived events associated with NAIP volcanism. There is also debate on the diagenetic stability of clumped isotope signals and potential ikaite-to-calcite transformation effects, necessitating careful petrographic, geochemical, and methodological controls.

• Geological context and sampling: The ~60 m-thick Fur Formation (early Eocene, northern Denmark) comprises laminated and structureless diatomites interbedded with 179 volcanic ash layers deposited during NAIP emplacement. Glendonites were sampled from ash horizons +15 and +60–62 (between PETM and ETM2). Large concretions (up to ~0.5 m) containing glendonites and isolated glendonites were halved; blades were sectioned for petrography.
• Petrography and mineral chemistry: Polished thin sections were examined by light microscopy, cathodoluminescence (CL), and SEM/EDS to distinguish calcite phases (Type I: primary ikaite-derived; Type II: early secondary calcite; Type III: late spar) and map minor elements (Mg, Mn, P, S, Sr, Fe) indicative of growth conditions.
• Stable isotopes and trace elements: δ13C and δ18O of calcite phases were measured (Isoprime triple collector IRMS), with reproducibility ±0.04‰ (δ13C) and ±0.08‰ (δ18O). Minor/trace element ratios (X/Ca) were measured by ICP-OES.
• Clumped isotope thermometry: Type I (brown, ikaite-derived) and Type III (late spar) calcites were hand-separated; concretion calcite was microdrilled. Powders were analyzed at ETH Zurich on a MAT253 with Kiel IV (LIDI mode), using ETH-1/2/3 standards and Brand parameter corrections in Easotope. Outliers were screened with Peirce criterion; Δ47 was reported on CDES with temperatures calculated via the Kele et al. calibration updated with ETH standard values. Quality metrics (mass 49 parameter, As offset) indicated negligible contamination. A modern calibration test on co-occurring recent glendonite/sedimentary carbonate and mollusk aragonite from the Kola Peninsula yielded consistent Δ47 temperatures (2–5 °C), supporting method applicability to glendonite calcite.
• Biomarker analyses: Total lipid extracts from glendonites, concretions, and sediments were obtained by Soxhlet extraction and analyzed by GC–MS (for alkenones, diols, sterols) and UHPLC–MS (for GDGTs). SSTs were reconstructed using TEX86 (BAYSPAR). Quality indices (BIT, ARI) screened for terrigenous GDGT influence and anomalous distributions. LDI and UK′37 were evaluated but deemed unreliable for this coastal/brackish setting. Additional indices (C37/C38 alkenone ratio, %C32 1,15-diol, Diol Index 2) assessed freshwater/riverine input and upwelling.
• Burial and diagenetic assessment: Low burial temperatures (~40–45 °C) were inferred from opal-A predominance over opal-CT, excellent microfossil preservation, and low thermal alteration indices (TAI 1–1+). These constraints, together with petrography, argue against diagenetic resetting of Δ47 to spuriously low values.

• Clumped isotope temperatures: Δ47 values for Fur Formation carbonates range ~0.71–0.77‰. Type I (ikaite-derived) calcite yielded average temperatures of 0.9 ± 4.7 °C at +15 and 9.1 ± 3.7 °C at +60–62. Type III (late spar) calcite yielded 8.5 ± 5.1 °C (+15) and 13.6 ± 4.0 °C (+62). Concretionary calcite averaged 4.5 ± 3.9 °C (+15) and 10.3 ± 8.3 °C (+60–62). Overall, glendonite and concretion calcites indicate mean crystallization temperatures near 5 ± 4 °C, representing near-freezing bottom waters.
• Fluid isotopes: Reconstructed δ18Ow values for all calcites are < −1‰ (ice-free seawater baseline), consistent with freshwater influence and/or early diagenetic alteration of volcanic detritus.
• Biomarkers and SSTs: Quality-screened TEX86 (BAYSPAR) SSTs from concretions/glendonites indicate ~13–15 °C at +15 and ~8–12 °C at +60–62, lower than coarser-resolution regional records. Sediments had BIT > 0.4 and/or ARI > 0.3 and were excluded. High C37/C38 alkenone ratios (≥1.8) and elevated %C32 1,15-diol (>10%) indicate strong riverine/coastal influence. LDI and UK′37 were unreliable due to non-open-ocean setting.
• Ikaite formation conditions: Elevated P and Mg, high Fe, and very low δ13C of Type I calcite imply conditions favoring ikaite nucleation in sediments driven predominantly by bacterial sulfate reduction, with little evidence for anaerobic oxidation of methane (low Methane Index ≤ 0.2; low archaeol/diether abundances).
• Diagenesis: Minimal burial heating and indistinguishable Δ47 temperatures between glendonites (Type I) and early concretion calcite (Type II) argue against diagenetic bias or a clumped isotope artifact during the ikaite-to-calcite transition.
• Water depth: Paleoenvironmental evidence supports a neritic shelf setting with average depth fluctuating around ~200 m (likely <300 m).

Quantitative clumped isotope temperatures from ikaite-derived calcite and coeval concretions demonstrate near-freezing bottom waters in the Danish Basin during specific intervals between the PETM and ETM2. These conditions are at odds with global early Eocene warmth and imply regionalized cold spells. The geochemical signatures (high P, Mg; low δ13C; biomarker evidence; low MI) indicate sulfate-reduction-driven ikaite formation in brackish-influenced sediments rather than methane seep settings. Reconstructed δ18Ow values and biomarkers point to freshwater input and early diagenetic influences consistent with a semi-enclosed basin receiving northern inflow. The cold events likely reflect regional forcing: multiple explosive NAIP eruptions could have injected sulfuric aerosols into the troposphere, causing short-lived regional cooling. The basin’s paleogeography would have favored winter surface cooling and density-driven cascading events that transported cold, oxygen-poor waters to the seafloor, enabling ikaite precipitation. The time-slice differences between proxies (Δ47 recording rapid, shallow subsurface conditions during ikaite growth; biomarkers integrating longer intervals) explain coexisting cold bottom-water snapshots with cool SSTs. Together, these findings reconcile glendonite occurrence with greenhouse climates by invoking seasonally or episodically cold regional bottom waters rather than global cooling.

This study provides the first quantitative near-freezing bottom-water temperature estimates from the early Eocene, using clumped isotope thermometry on ikaite-derived glendonite calcite from the Fur Formation. The results show bottom waters near ~5 °C at shallow shelf depths (<300 m), with coeval cool SSTs (~8–15 °C), demonstrating regional cold spells within a globally warm climate. Geochemical and biomarker evidence indicates ikaite formation under sulfate-reducing, brackish-influenced conditions, not methane seepage. The observed cooling is best explained by regional effects of NAIP volcanism and wintertime dense-water cascading in a semi-enclosed basin. Future work should target high-resolution sampling across glendonite- and ash-bearing horizons and apply fine-grid climate models to quantify the magnitude, seasonality, and spatial extent of these cold events.

• Proxy context: TEX86-based SSTs from sediments were excluded due to high BIT and ARI values; LDI and UK′37 are unreliable in coastal/brackish settings, constraining SST reconstructions to concretions/glendonites and limiting broader SST coverage.
• Spatial and temporal resolution: Δ47 temperatures represent short-lived early diagenetic conditions tied to ikaite growth/transformations and may not capture seasonal extremes or longer-term means; biomarkers integrate longer but still localized intervals.
• Modeling constraints: Current GCMs with coarse grids (~3°) are inadequate to resolve short-duration, small-scale cascading events, limiting direct model-data comparison.
• Ikaite-calcite transformation: Although data argue against Δ47 artifacts during transformation, the precise behavior of clumped isotopes during ikaite-to-calcite conversion is not fully constrained experimentally.
• Basin hydrography and δ18Ow: Freshwater and early diagenetic influences complicate precise deconvolution of original seawater δ18O and salinity.