North Pacific warmth synchronous with the Miocene Climatic Optimum Open Access

The middle Miocene is distinguished as the major warm anomaly in the Cenozoic transition from global greenhouse conditions to the modern icehouse world (Westerhold et al., 2020). During the Miocene Climatic Optimum (MCO, ca. 16.95–13.95 Ma), ice sheets retreated, leaving Antarctica at least partially deglaciated and occupied with temperate and tundra ecosystems (Lewis et al., 2008; Warny et al., 2009). Cooling and ice sheet growth intensified during the Middle Miocene Climate Transition (MMCT, ca. 13.95–12.8 Ma), culminating in the reestablishment and expansion of the East Antarctic Ice Sheet during the Mi3 glaciation (ca. 13.8–13.9 Ma; Miller et al., 1991). Outside of Antarctica, global ecosystems shifted from equable MCO conditions toward stronger meridional gradients and terrestrial aridity (Pound et al., 2012).

While qualitative warmth during the MCO and cooling during the MMCT are well supported by geologic evidence, the quantitative evolution of global surface temperatures during these periods remains unclear. Benthic foraminiferal oxygen isotope records (benthic δ¹⁸O) have provided high-resolution information on global climate throughout the early and middle Miocene (Holbourn et al., 2014, 2015), but surface temperature proxy records at similar resolution are lacking. As a result of this gap, estimates of middle Miocene global warmth range anywhere from ~5 °C (Westerhold et al., 2020) to more than 10 °C above modern (Burls et al., 2021; Herbert et al., 2022).

Existing Northern Hemisphere sea surface temperature (SST) records are limited to sites in the eastern North Atlantic and North Sea (Herbert et al., 2016, 2020; Super et al., 2020; Sangiorgi et al., 2021; Fig. 1). These sites are influenced by regional dynamics of the North Atlantic Current, so SSTs at these sites may not be representative of broader Miocene Northern Hemisphere temperatures. In addition, the temporal resolution of these records does not resolve orbital-scale SST variability. Higher-resolution records from other regions are needed to constrain the Miocene evolution of Northern Hemisphere SSTs.

Here, we present high-resolution SST proxy data from the North Pacific spanning the entire MCO and MMCT. We measure the unsaturation ratios of alkenones (U^k′₃₇ [see Supplemental Material¹]) from Ocean Drilling Program (ODP) Site 884 in the northwest Pacific (Rea et al., 1993; see the Supplemental Material¹ for site description). U^k′₃₇ SST reconstructions are largely robust to species effects (Villanueva et al., 2002), assumptions of past seawater chemistry, and methods of calibration to SST (Müller et al., 1998; Tierney and Tingley, 2018), and this proxy has been successfully applied to quantifying global mean annual SSTs in the late Miocene (Herbert et al., 2016). High core recovery (>90%) and magnetic reversal stratigraphy (Table S1 in the Supplemental Material) provide a continuous and well-dated section for high-resolution records of the MCO and MMCT at Site 884 (see the Supplemental Material for age model). Our record extends continuous records of SSTs in the North Pacific to span the past 18 million years.

Our measurements of U^k′₃₇ at Site 884 provide a continuous high-resolution (6 k.y. average sampling interval) record of SSTs in the subpolar North Pacific from 18.2 to 12.6 Ma, with lower resolution (54 k.y.) data to 8.8 Ma. After column chromatography, alkenones were analyzed using gas chromatographic or high-performance liquid chromatographic methods (see the Supplemental Material for analytical methods). U^k′₃₇ varied between 0.393 and 0.882, providing reconstructed SSTs between 10.6 and 25.4 °C. We apply the calibration of Müller et al. (1998), though other choices of calibration do not significantly affect the reconstructed SSTs for the range of U^k′₃₇ obtained (Tierney and Tingley, 2018).

Previous work suggests seasonal biases in the modern application of the U^k’₃₇ proxy in the subpolar North Pacific, although the implications for Miocene interpretations are uncertain. The majority of alkenone production in the region occurs between July and November (Harada et al., 2006), which average 2.9 °C warmer than mean annual SSTs at Site 884 (Huang et al., 2021). Regional core-top U^k′₃₇ correlates most strongly with June through August monthly SSTs (Tierney and Tingley, 2018), which are 2.3 °C warmer than mean annual SSTs at Site 884 (Huang et al., 2021). While winter light limitation was similar, differences in the seasonal variation of SSTs and nutrient availability between the Miocene and modern are unknown, complicating any assumption of similar biases in U^k′₃₇ as well.

In addition, alkenone concentrations were strongly anti-correlated with U^k′₃₇, with warmer proxy values corresponding to exponentially lower alkenone concentrations (Fig. S1 in the Supplemental Material). Our Miocene anti-correlation implies higher production of alkenones associated with cooler temperatures at Site 884, potentially biasing reconstructed SSTs toward colder values. Because of conflicting directions for potential biases, we interpret U^k′₃₇ at Site 884 as representative of mean annual SSTs during the Miocene, though we acknowledge the potential for some bias (<3 °C) toward warmer seasonal temperatures.

We find a pronounced SST maximum in the North Pacific synchronous within age uncertainties to the MCO as defined by benthic δ¹⁸O records (Holbourn et al., 2014, 2015). Long-term mean temperatures during the MCO peak above 22 °C, more than 16 °C warmer than the modern mean annual SST at Site 884. Though changes in SSTs at high latitudes were likely amplified compared to the global mean (Liu et al., 2022), these temperatures are more consistent with high-end estimates of MCO global warmth of at least 10 °C warmer than modern (Herbert et al., 2022: ~12 °C; Burls et al., 2021: ~11.5 °C) than more modest estimates (Westerhold et al., 2020: ~5 °C).

The coincident timing of Miocene warmth and subsequent cooling in the North Pacific with analogous changes in benthic δ¹⁸O establishes a globally synchronous evolution of Miocene ocean temperatures. The benthic δ¹⁸O signal originates from the southern high latitudes, which were the source of both bottom water formation and terrestrial ice volume during the early–middle Miocene (Shevenell et al., 2008), controlling both deep sea temperatures and seawater δ¹⁸O. Geologic records from Antarctica and the polar Southern Ocean (Lewis et al., 2008; Warny et al., 2009; Pierce et al., 2017) corroborate reduced MCO ice volume and ice sheet expansion during the MMCT interpreted from benthic δ¹⁸O (Holbourn et al., 2014, 2015). While previous temperature records have found high-latitude cooling events during the MMCT, either preceding or coincident with the Mi3 glaciation between 13.8 and 14.2 Ma (Shevenell et al., 2004; Super et al., 2020; Sangiorgi et al., 2021), ours captures the significant warming at the onset of the MCO as well as MMCT cooling in a continuous record at a single site.

In light of ice-distal SST records from both the North Atlantic and North Pacific, the Miocene decay and subsequent growth of ice sheets in Antarctica cannot be attributed primarily to regional dynamics (Shevenell et al., 2008; Hou et al., 2023) but rather as part of global temperature changes. Direct ice sheet feedbacks (ice albedo, positions of Southern Hemisphere frontal zones, etc.) from the Antarctic would have been insufficient to explain high-latitude warmth in both the North Pacific and North Atlantic. Changes in greenhouse gas concentrations likely provided this global forcing, though direct evidence for synchronous changes in atmospheric pCO₂ proxy records have yielded widely varying results (Cenozoic CO₂ Proxy Integration Project [CENCO₂PIP] Consortium, 2023, and references therein; Fig. 2A). While pCO₂ proxies show a local maximum coinciding with the MCO, the increase in atmospheric pCO₂ appears to lag the increase in SSTs at Site 884 and in benthic δ¹⁸O (Fig. 2). However, we caution this may be an artifact of low pCO₂ proxy data density covering the Early Miocene before the MCO. The eruption of the Columbia River Flood Basalts (northwestern United States) is commonly considered the source of this inorganic carbon to the ocean-atmosphere system (Holbourn et al., 2015; Sosdian et al., 2020), but this mechanism is challenged by the later timing of eruptions with respect to warming (Kasbohm and Schoene, 2018) and questions regarding the ability for these emissions to cause such long-sustained (~3 m.y.) warmth. While flood volcanism may have contributed to peak warmth early during the MCO, elevated tectonic degassing (Herbert et al., 2022) provides a more sustained mechanism of greenhouse gas emissions on million-year time scales. The beginning of long-term cooling after 15 Ma aligns with declining seafloor spreading rates (Fig. S2). Slowing mantle degassing could explain the initiation of cooling, culminating in passing a glaciation threshold (DeConto et al., 2008) during the Mi3 glaciation. However, our temperature records alone cannot distinguish between the importance of reductions in CO₂ sources to the atmosphere and increasing weathering sinks, such as the uplift of highly weatherable silicates in the tropics (Park et al., 2020). Changes in carbon cycling within the ocean-atmosphere system (Holbourn et al., 2015) were associated with climate variability on shorter time scales (tens to hundreds of thousands of years) imposed on these longer-term (millions of years) tectonic forcings.

Our high-resolution data reveal rapid warming at the beginning of the MCO as well as cooling synchronous with East Antarctic Ice Sheet expansion during the MMCT. SSTs at Site 884 warm more than 7°C within 50 k.y. after 17 Ma (Fig. 3), a time scale shorter than that over which tectonic processes could have been the forcing mechanism. Although our reported timing of this warming is younger than that observed in the benthic δ¹⁸O records (Holbourn et al., 2015), we interpret these as synchronous within error of our age model (Supplemental Material). In addition, SSTs in the North Pacific cool 4 °C after 14 Ma, though a gap in sediment recovery between cores prevents precise determination of the duration of cooling. This cooling event aligns with the abrupt expansion of the East Antarctic Ice Sheet observed in Antarctic geologic records (Lewis et al., 2008; Pierce et al., 2017) and in benthic δ¹⁸O records (Shevenell et al., 2004; Holbourn et al., 2014) as well as decreasing atmospheric pCO₂ observed in some proxy records (Badger et al., 2013; Raitzsch et al., 2021; Witkowski et al., 2024).

North Pacific SSTs during the MCO were highly variable, with glacial-interglacial cycles of as much as 6 °C in amplitude (Fig. 3) and a longer cooler interval surrounding the Mi2 glaciation (Miller et al., 1991) after 16 Ma (Fig. 2). Significant periodicities correspond to orbital frequencies near eccentricity (109 k.y., 99% confidence level) and obliquity (43 k.y., 90% confidence level) across the high-resolution record (Fig. S3). The large amplitude of orbital-scale temperature variability, particularly during the MCO, is consistent with large variability found in contemporaneous δ¹⁸O (Holbourn et al., 2014, 2015), atmospheric pCO₂ proxies (Greenop et al., 2014), and Antarctic ice sheet extent (Levy et al., 2016) on orbital time scales. Our record confirms that temperature variability in the high latitudes distant from Southern Hemisphere ice sheets accompanied these changes as well.

To assess the context of North Pacific temperatures within the northern subpolar regions, we compile Miocene U^k′₃₇ records from sites between 50°N and 60°N (Herbert et al., 2016, 2020; Super et al., 2020; Sangiorgi et al., 2021; Fig. 4; Fig S4). SST anomalies in the North Atlantic relative to modern mean annual SST were persistently higher than in the North Pacific throughout the Miocene, resulting in a larger than modern gradient between the basins. While the gradient appears to reduce during the MCO, we caution that this may be an artifact of proxy saturation (U^k′₃₇ approaches 1) at North Atlantic sites. The proxy response to further SST variability at these high temperatures is likely truncated, whereas North Pacific U^k′₃₇ is much further from saturation. Because reconstructed SSTs in the North Atlantic depend more on the choice of linear (Müller et al., 1998) or nonlinear (Tierney and Tingley, 2018) calibrations, our reconstructed inter-basin gradient has higher uncertainty during the MCO.

The gradient between North Pacific and North Atlantic SST anomalies (Fig. 4) suggests persistent differences in poleward oceanic heat transport between the two basins. While the modern North Atlantic is warmer than the North Pacific due to the greater poleward influence of the North Atlantic Current (NAC) compared to the North Pacific Current (Fig. 1), these differences were enhanced throughout the Miocene (Fig. S5). Super et al. (2020) hypothesized an invigorated NAC as responsible for extreme North Atlantic warmth, extending into the mid-latitudes as well, and our North Pacific record confirms this warmth was exaggerated in the Atlantic basin. These proxy records contradict predictions from model simulations (Burls et al., 2021) of a reduced or even reversed Miocene temperature gradient between the northern ocean basins (Fig. S6). A stronger inter-basin gradient during the Miocene, when the Central American Seaway was open, also challenges the idea of a weaker NAC prior to seaway closure during the Pliocene (Haug and Tiedemann, 1998). While these paleotemperature records alone cannot directly rule out effects of other possible mechanisms, such as cloud radiative feedbacks and atmospheric heat transport, greater northward oceanic heat transport in the Atlantic basin provides a compelling potential mechanism for enhanced Miocene warmth observed in the North Atlantic.

Our high-resolution SST proxy record spanning the entire MCO and MMCT at a single site extends our understanding of continuous temperature evolution in the North Pacific over the past 18 million years. Elevated temperatures in the northern high-latitudes coincide with minimal Antarctic ice sheet volumes and higher bottom water temperatures, showing that the MCO was a globally synchronous climate interval in the high latitudes. The bipolar nature of warmth between ca. 17 and 14 Ma suggests a global forcing mechanism, likely elevated greenhouse gas concentrations, though proxy constraints on these concentrations vary widely (Cenozoic CO₂ Proxy Integration Project [CENCO₂PIP] Consortium, 2023). The coincidence of high SST variability on orbital time scales in the North Pacific with similarly large variations in benthic δ¹⁸O implies either large changes in atmospheric pCO₂ on these time scales or high sensitivity of Miocene climate to smaller changes in greenhouse gas concentrations or orbital forcing. While high-resolution temperature records from more regions are needed to assess the global pattern and magnitude of temperature variability, the synchronous timing of early–middle Miocene warmth in both the North Pacific and Antarctic provides strong circumstantial evidence for a common forcing.

This study used samples and data provided by the International Ocean Discovery Program (IODP) and its predecessors. Funding was provided by National Science Foundation grants 1635127 and 2202760 to Herbert. We thank Jonah Bernstein-Schalet, Emily Clements, and Sofía Gilroy for their assistance in sample preparation. Thanks also to Junsheng Nie and two anonymous reviewers for their constructive reviews. Alkenone data is also available at https://doi.pangaea.de/10.1594/PANGAEA.965068.