Sensitivity of ocean circulation to warming during the Early Eocene greenhouse.

Multiple abrupt warming events ("hyperthermals") punctuated the Early Eocene and were associated with deep-sea temperature increases of 2 to 4 °C, seafloor carbonate dissolution, and negative carbon isotope (δ13C) excursions. Whether hyperthermals were associated with changes in the global ocean overturning circulation is important for understanding their driving mechanisms and feedbacks and for gaining insight into the circulation's sensitivity to climatic warming. Here, we present high-resolution benthic foraminiferal stable isotope records (δ13C and δ18O) throughout the Early Eocene Climate Optimum (~53.26 to 49.14 Ma) from the deep equatorial and North Atlantic. Combined with existing records from the South Atlantic and Pacific, these indicate consistently amplified δ13C excursion sizes during hyperthermals in the deep equatorial Atlantic. We compare these observations with results from an intermediate complexity Earth system model to demonstrate that this spatial pattern of δ13C excursion size is a predictable consequence of global warming-induced changes in ocean overturning circulation. In our model, transient warming drives the weakening of Southern Ocean-sourced overturning circulation, strengthens Atlantic meridional water mass aging gradients, and amplifies the magnitude of negative δ13C excursions in the equatorial to North Atlantic. Based on model-data consistency, we conclude that Eocene hyperthermals coincided with repeated weakening of the global overturning circulation. Not accounting for ocean circulation impacts on δ13C excursions will lead to incorrect estimates of the magnitude of carbon release driving hyperthermals. Our finding of weakening overturning in response to past transient climatic warming is consistent with predictions of declining Atlantic Ocean overturning strength in our warm future.

North Atlantic surface ocean warming and salinization in response to middle Eocene greenhouse warming.

Quantitative reconstructions of hydrological change during ancient greenhouse warming events provide valuable insight into warmer-than-modern hydrological cycles but are limited by paleoclimate proxy uncertainties. We present sea surface temperature (SST) records and seawater oxygen isotope (δ18Osw) estimates for the Middle Eocene Climatic Optimum (MECO), using coupled carbonate clumped isotope (Δ47) and oxygen isotope (δ18Oc) data of well-preserved planktonic foraminifera from the North Atlantic Newfoundland Drifts. These indicate a transient ~3°C warming across the MECO, with absolute temperatures generally in accordance with trace element (Mg/Ca)-based SSTs but lower than biomarker-based SSTs for the same interval. We find a transient ~0.5‰ shift toward higher δ18Osw, which implies increased salinity in the North Atlantic subtropical gyre and potentially a poleward expansion of its northern boundary in response to greenhouse warming. These observations provide constraints on dynamic ocean response to warming events, which are consistent with theory and model simulations predicting an enhanced hydrological cycle under global warming.

High-latitude biomes and rock weathering mediate climate-carbon cycle feedbacks on eccentricity timescales.

The International Ocean Discovery Programme (IODP) and its predecessors generated a treasure trove of Cenozoic climate and carbon cycle dynamics. Yet, it remains unclear how climate and carbon cycle interacted under changing geologic boundary conditions. Here, we present the carbon isotope (δ13C) megasplice, documenting deep-ocean δ13C evolution since 35 million years ago (Ma). We juxtapose the δ13C megasplice with its δ18O counterpart and determine their phase-difference on ~100-kyr eccentricity timescales. This analysis reveals that 2.4-Myr eccentricity cycles modulate the δ13C-δ18O phase relationship throughout the Oligo-Miocene (34-6 Ma), potentially through changes in continental weathering. At 6 Ma, a striking switch from in-phase to anti-phase behaviour occurs, signalling a reorganization of the climate-carbon cycle system. We hypothesize that this transition is consistent with Arctic cooling: Prior to 6 Ma, low-latitude continental carbon reservoirs expanded during astronomically-forced cool spells. After 6 Ma, however, continental carbon reservoirs contract rather than expand during cold periods due to competing effects between Arctic biomes (ice, tundra, taiga). We conclude that, on geologic timescales, System Earth experienced state-dependent modes of climate-carbon cycle interaction.