We reconstruct the subtropical North Atlantic water column structure during the Miocene Climate Optimum warming (MCO; 17–14.8 Ma) and the Middle Miocene Climate Transition cooling (MMCT; 14.8–12.8 Ma) by analyzing δ18O and δ13C in four species of foraminifera (surface dwellers Dentoglobigerina altispira and Trilobatus quadrilobatus, thermocline dweller Dentoglobigerina venezuelana, and benthic Planulina wuellerstorfi) from Site 558 (37.8°N). At the end of the MCO, δ18O of surface and thermocline dwellers increased by >1‰, suggesting at least 2°C cooling in the upper ocean as ice growth increased global δ18Osw by ∼0.5‰. The difference in δ18O values between thermocline and surface-dwelling species increased during the MMCT, coinciding with the development of a largely permanent East Antarctic Ice Sheet, and persisted into the Late Miocene. We interpret this increase in vertical δ18O gradient as a strengthening of the thermocline due to intensification of subtropical gyre circulation in response to the MMCT cooling.
The trend of cooling over the past 50 Myr was punctuated during the Miocene by a period of global warming known as the Miocene Climate Optimum (MCO; Holbourn et al., 2013a; Steinthorsdottir et al., 2020). The MCO is generally recognized in benthic foraminiferal δ18O records by minimum Pacific Ocean values from 17 to 14.8 Ma (Holbourn et al., 2013a) and from a sea-level perspective as a period of reduced ice volume from 17 to 13.8 Ma (Miller et al., 2020). Global mean sea surface temperatures (SSTs) were ∼3–4°C warmer than modern (You et al., 2009; Pound et al., 2011) and terrestrial temperatures similarly warmer (Goldner et al., 2014) during the MCO. The MCO provides one of the best past analogs for future warming (e.g., Steinthorsdottir et al., 2020). Although atmospheric CO2 concentration estimates are still poorly constrained (Pagani et al., 1999; Royer et al., 2001; Demicco et al., 2003; Kurschner et al., 2008; Foster et al., 2012), best estimates range from 400–600 ppm (Steinthorsdottir et al., 2020). Antarctic ice sheets are thought to have been significantly reduced (Holbourn et al., 2015) or possibly eliminated (Levy et al., 2016; Miller et al., 2020) during the MCO. Causes and mechanisms of MCO warming are widely debated; hypotheses include high pCO2 levels (Goldner et al., 2014) and large-scale changes in ocean circulation (Shevenell et al., 2004).
The MCO was followed by a period of sharp, steady global cooling referred to as the Middle Miocene Climate Transition (MMCT) which consists of three stepwise benthic foraminiferal δ18O increases at 14.8 Ma (Mi2a, ∼30 m sea level fall, ∼0.7°C bottom water cooling); 13.8 Ma (Mi3, ∼50 m fall, ∼1.2°C bottom water cooling); and 12.8 Ma (Mi4, 20–30 m sea-level fall, ∼1.0°C bottom water cooling; Miller et al., 2020). Physical evidence indicates that the East Antarctic Ice Sheet (EAIS) grew to near its present state and remained permanent after 12.8 Ma (Lewis et al., 2008) and the Antarctic ice sheets became too large to be strongly influenced by precession and eccentricity that dominated its pacing prior to this (Miller et al., 2020). Estimates for the timing and magnitude of surface water cooling during the MMCT vary regionally (Super et al., 2020). Furthermore, there are discrepancies in SST estimates among the few studies of North Atlantic SSTs during the Miocene (Goldner et al., 2014; Herbert et al., 2016; Super et al., 2020), and little work has been done in studying North Atlantic thermocline variability and surface-water circulation change during the Miocene. There is no consensus on what caused the MMCT cooling, though decreasing atmospheric CO2 likely played a role (Goldner et al., 2014; Steinthorsdottir et al., 2020).
Another notable feature of the Miocene is a positive carbon isotope (δ13C) excursion between 17.5 to 13.5 Ma (Vincent & Berger, 1985) that is well-expressed in globally distributed planktonic and benthic foraminiferal records as well as basin-wide, organic-rich deposits along continental margins. This global pronounced shift in δ13C, referred to as the Monterey Excursion, is hypothesized to have occurred due to rapid burial of organic carbon and drawdown of atmospheric CO2. Multiple hypotheses exist regarding the causes of the Monterey Excursion and its association with MMCT cooling. The most established of these, the Monterey Hypothesis (Vincent & Berger, 1985), proposes that an increased latitudinal gradient caused by long-term global cooling led to more upwelling, leading to increased organic carbon burial and subsequent lowering of atmospheric CO2, further amplifying the cooling.
Deep water formation and circulation plays an important role in the distribution of ocean heat and salt and is tied to surface and thermocline water changes as part of Atlantic Meridional Overturning (AMOC; Broecker, 1991; Ferreira et al., 2018). The Miocene saw a transition in deep water formation with reduction in warm salty Tethyan Indian Saline Water (Woodruff & Savin, 1989) and strengthening of North Atlantic Deep Water/Northern Component Water (NADW/NCW). Interbasinal benthic foraminiferal δ13C gradients among the Atlantic, Pacific, and Southern Oceans indicate little NADW/NCW in the earliest Miocene, high production from 18–17 Ma, a shutdown from 17–12.5 Ma, and high production with a geochemically modern distribution from 12.5 Ma (e.g., Wright et al., 1992). These Miocene changes in NADW/NCW likely played a role in global heat distribution and have been attributed to changes in the Greenland-Scotland Ridge sill depths due to variations in the Iceland hot spot (Wright & Miller, 1996). Benthic foraminiferal stable isotopes record from Deep Sea Drilling Project (DSDP) North Atlantic sites have played an important role in these reconstructions of Atlantic deep-water circulation (Wright et al., 1992), but planktonic foraminiferal stable isotopes from these sites have only been cursorily examined (Miller & Fairbanks, 1985).
In the modern North Atlantic Ocean, surface waters from the subtropical gyre are a major reservoir of heat and salt, with winter temperatures greater than 18°C and salinities greater than 37 psu (Fig. 1). These warm, salty surface to thermocline waters are transported by the Gulf Stream/North Atlantic Current to higher northern latitudes (Fig. 1). There, they cool and become increasingly dense, convecting in the winter in the Norwegian-Greenland Sea (e.g., Broecker, 1991). A cold, dense water mass known as North Atlantic Deep Water (NADW) forms, sinks to depths of 2 to 4 kilometers and is transported southward combining with other water masses to form the Deep Western Boundary Current (e.g., Broecker, 1991). NADW is integral to AMOC, a conceptual model of large-scale global circulation in which water, along with salt, heat, carbon, and nutrients, circulates through global oceans. Therefore, AMOC and the North Atlantic play crucial roles in modulating the ocean's response to climate change.
Many of these modern North Atlantic circulation patterns developed during the Miocene. During the Oligocene to Early Miocene, an important component of modern ocean circulation, a strong Antarctic Circumpolar Current (ACC), developed due in part to the continued opening of the Drake Passage (e.g., Livermore et al., 2007). Once the Drake Passage had opened, the westerly wind-driven Antarctic Circumpolar Current in the Southern Ocean was established. Through the Miocene, the widening of the Drake Passage and the northward movement of the Australian continent enabled the ACC to strengthen. As the Antarctic Circumpolar Current intensified through the Miocene, Antarctica became increasingly thermally isolated from the warm surface waters of the surrounding oceans (Beu et al., 1997). During the MMCT, a permanent EAIS began to form (Pierce et al., 2017) due in part to the strong ACC cutting off the flow of warm surface water towards Antarctica. As the ACC continued to strengthen, the EAIS continued to expand, causing Antarctic Bottom Water (AABW) to become increasingly colder (Huang et al., 2017). Furthermore, opening of the Fram Strait, the North Atlantic-Arctic Gateway, around 17.5 Ma allowed water and ice exchange between the Arctic and North Atlantic and established a steeper meridional thermal gradient, coinciding with the onset of the MCO (Jakobsson et al., 2007). Though the closing of the Isthmus of Panama was largely a Pliocene event, shoaling of the Isthmus coupled with intensified meridional thermal gradients in the Middle Miocene resulted in an intensification of the Caribbean Loop-Gulf Stream/North Atlantic Current system (Mullins et al., 1987).
Materials and Methods
Samples and Preservation
Recovery in cores 7R to 11R (215–262 mbsf) was moderate (Bougault et al., 1984) and thus the upper Middle Miocene section is not as well-characterized as the lower Middle Miocene and older section (Fig. 4). Recovery in the top (cores 1R–6R) and bottom (cores 12R–17R) of Hole 558 was excellent despite being only rotary cored with one hole (Fig. 4). Thus, the interval representing the MCO (Fig. 5) and the Upper Miocene are well-represented. The sediments are nannofossil ooze in cores 1R–12R (158–272 mbsf) and marly nannofossil chalk in cores 16R–17R (300.5–319.5 mbsf) with transitional chalky ooze lithologies in cores 13R–15R (272–300.5 mbsf; Bougault et al., 1984). Preservation of foraminifera is very good to moderate, with occasional poorly-preserved specimens (Fig. 9), using the calcareous nannofossil preservation criteria identified in IODP Expedition 356 Methods (Gallagher et al., 2017). We selected good to very good, with occasional moderate, specimens for isotopic analyses (Fig. 9).
Samples were provided by the International Ocean Discovery Program Bremen Core Repository. The sampling interval averaged one per section or 1.5 m. Sediment samples were dried, washed with sodium metaphosphate, wet-sieved using a 63-μm sieve, dried overnight at 50°C in an oven, and subsequently weighed to determine the percent coarse fraction (>63 μm).
To reconstruct the vertical thermal evolution of the North Atlantic subsurface, thermocline, and bottom waters, we analyzed the stable isotope composition of the following species of foraminifera: surface-dwelling Trilobatus quadrilobatus/trilobus (d'Orbigny, 1846; Reuss, 1850) and Dentoglobigerina altispira (Cushman & Jarvis, 1936), thermocline-dwelling Dentoglobigerina venezuelana (Hedberg, 1937), and benthic epifaunal Planulina wuellerstorfi (Schwager, 1866). Most of the Trilobatus species analyzed were T. quadrilobatus, so we hereafter refer to those samples as T. quadrilobatus. Identification of planktonic foraminifera was based on Kennett & Srinivasan (1983) and benthic foraminifera on Holbourn (2013b). More detailed foraminiferal taxonomy and description are provided under Systematics.
Planktonic foraminiferal specimens of D. venezuelana, T. quadrilobatus, and D. altispira (Plate 1) were picked from restricted size fractions (250–300 μm, 212–250 μm, and 212–250 μm, respectively). Benthic foraminiferal specimens of P. wuellerstorfi (Plate 1) were picked from size fractions of >300 μm above Sample 558-9R-1-90 cm (234.9 mbsf; 12.54 Ma); analyses were combined with published isotopic data of Cibicidoides (primarily C. mundulus) from Miller & Fairbanks (1985; Appendix). Though we accept the generic assignment of P. wuellerstorfi to Planulina (see also Miller & Katz, 1987) and not Cibicidoides as often reported in isotopic studies, we note that most (but not all) species of both genera yield similar stable isotopic results. Stratigraphic resolution is limited by a zone of poor calcareous preservation due to dissolution between ∼205 and 220 mbsf and poor core recovery between 215 and 262 mbsf. Four to five specimens of planktonic foraminifera and three specimens of benthic foraminifera were analyzed per sample. Each sample was ultrasonically cleaned for nine seconds to clean residual calcareous material and dried at 50°C for 24 hours prior to analysis.
A period of low δ18O values associated with MCO warming occurred between approximately 17.0 to 14.3 Ma at Site 558 (Fig. 5). During this interval, surface dwellers T. quadrilobatus and D. altispira recorded δ18O decreases of 1.1 and 1.0‰, respectively, and the thermocline dweller D. venezuelana recorded a δ18O decrease of 0.7‰, while δ18Osw decreased by only 0.2‰ (Fig. 5). The most pronounced δ18O increase associated with MMCT cooling occurred between 14.3 and 13.0 Ma and is recorded as an increase of 0.8‰ in benthic foraminifera P. wuellerstorfi, 1.0‰ in D. venezuelana, and 0.7–0.8‰ in T. quadrilobatus and D. altispira. Planktonic foraminifera show no general δ18O trend in between 13.0 and 7.5 Ma, punctuated by two intervals of higher δ18O values at ∼11–10.4 Ma and 8.8–8.5 Ma, suggesting two additional smaller-magnitude phases of cooling.
The initial positive δ13C excursion of ∼1.4‰ associated with onset of the Monterey Event occurred between 17.1 and 16.2 Ma (Figs. 4, 5). Peak δ13C values in surface and thermocline dwellers lasted from 16.2 to 14.2 Ma and sharply decreased through 13.5 Ma, marking the end of the Monterey Event. Planktonic foraminiferal δ13C values across all species gradually declined by 1.6–1.8‰ from 13.5 Ma to around 12 Ma, after which we observe no significant trend. While the benthic foraminiferal stable isotope data are low resolution in the interval encompassed by the Monterey Event, benthic foraminiferal δ13C values were relatively unchanged after 13.5 Ma.
ΔΔ18O between the Surface and Thermocline
To examine first-order changes between surface and the thermocline, we calculate δ18O difference between the surface and the thermocline (Δδ18Os-t) by subtracting δ18O values of T. quadrilobatus, the more abundant surface-dwelling species, from thermocline-dwelling D. venezuelana δ18O values at the corresponding sample (Fig. 7A). While surface and thermocline δ18O values were similar prior to the MCO (Δδ18Os-t = 0.03±0.02‰), they diverged slightly during the MCO (Δδ18Os-t = 0.037±0.01‰) and then diverged more during the MMCT (Δδ18Os-t = 0.54±0.01‰), with a peak difference of ∼1‰ at 13.2 Ma (Fig. 7A). A one-way analysis of variance test performed on the data results in a p-value of <10−4, indicating that this increase in Δδ18Os-t from the MCO to the MMCT is highly statistically significant at a 1% significance level.
Further one-way ANOVA tests were performed to determine statistically significant Δδ18Os-t changes between each of the four time intervals (Figs. 6, 7). Prior to the MCO (Fig. 6A), the mean Δδ18Os-t was 0.03±0.02‰ (p-value = 0.001), during the MCO (Fig. 6B) mean Δδ18Os-t was 0.37±0.01‰ (p-value = 0.001), and during the MMCT (Fig. 6C) mean Δδ18Os-t was 0.54±0.01‰ (p-value = 0.03). The mean Δδ18Os-t of 0.52±0.01‰ for the following interval from 10 to 7.8 Ma, shows that the increased oxygen isotopic gradient between the sea surface and the thermocline persisted into the Late Miocene (Fig. 6D).
To better observe trends in the record, δ18O data were interpolated to every 0.1 Myr and smoothed using a 7-point Gaussian convolution filter to remove periods shorter than 350 kyr (Fig. 7B). The divergence in δ18O values between the surface, particularly those recorded in T. quadrilobatus, and the thermocline is more easily observed in the smoothed records (Fig. 7B). Although sample resolution is lower in the interval prior to the MCO and data interpolated across the 1.1 Myr hiatus, Δδ18Os-t increased in the early MCO (Fig. 7C). The most pronounced Δδ18Os-t of ∼0.5‰ occurred across the MMCT and persisted at a smaller magnitude following 12.8 Ma (Fig. 7C).
We interpret an increase in Δδ18Os-t to reflect thermocline intensification during the MMCT cooling. There are two alternative explanations for the increased gradient between subsurface and thermocline dwellers: 1) D. venezuelana lived farther up in the water column prior to and during the MCO and migrated to a deeper habitat following the MCO (Matsui et al., 2016), recording colder temperature; 2) the divergence in δ18O values is due to a seasonality bias in planktonic foraminifera – if all thermocline foraminifera analyzed in each sample calcified during a cooler season than the planktonic species, the resulting δ18O value would be higher, indicating colder temperatures. However, development of a coherent cold season bias starting at the MMCT cooling would be counterintuitive and thermocline intensification is the most straightforward mechanism for interpreting the distinct increase in Δδ18Ot-s at the onset of the MMCT. Though changes in seasonality or depth migration could potentially explain the observed increase in Δδ18Os-t (e.g., Chapman, 2010), there are no extant relatives of some of the species analyzed in this work and investigating seasonal biases in the record would require single specimen analyses that are out of the scope of this study.
ΔΔ18O Between the Surface and Deep Ocean
Cross plots show that the δ18O gradient between the surface, thermocline, and benthic foraminifera dramatically increased from the Early to Late Miocene (Fig. 6). Prior to the MCO, there was no pronounced difference in the δ18O values between the surface, thermocline, and deep water (Fig. 6A). While benthic foraminiferal δ18O data are scarce over the next phase corresponding to the MCO (Fig. 6B), their δ18O values were clearly distinct from those surface and thermocline, suggesting relative cooling of the deep western North Atlantic Ocean. During the MMCT (Figs. 6C, 7), there were pronounced δ18O differences between surface/thermocline and benthic foraminifera, despite minor overlap between δ18O values of surface and thermocline species. By the Late Miocene (Figs. 6D, 7), there was a very strong δ18O gradient between the deep sea, thermocline, and sea surface, suggesting a strongly stratified ocean.
Using the smoothed records (Fig. 7B), we compute Δδ18Osurface-benthic and Δδ18Othermocline-benthic (Fig. 7C). Both δ18O gradients were weak prior to the MCO, then strengthened during the early MCO, and further strengthened with the MMCT. Limited sampling resolution of the δ18Obenthic record precludes comments about the timing and amplitude of the strengthening entering the MCO. Both Δδ18Osurface-benthic and Δδ18Othermocline-benthic remain relatively stable through the MCO and strengthen consistently by ∼1‰ from 13.5 to 7.8 Ma.
We do not observe any significant change in δ13C gradients between the surface ocean and thermocline in the subtropical gyre during the Early to Middle Miocene (Fig. 6). Surface to bottom δ13C gradients were relatively constant, though they were apparently reduced in the Late Miocene (Figs. 5, 6) likely due to increased NCW production (Wright & Miller, 1996).
The increase in benthic foraminiferal δ18O values during the MMCT was 0.5‰ greater than the amplitude of δ18Osw increase over this interval (Fig. 5), suggesting that bottom waters cooled by ∼2° between 13.5 and 12.8 Ma (Fig. 8). Bottom waters record a warming of approximately 2°C between 12–8 Ma, with a net change from 8°C in the MCO to ∼4°C after the MMCT.
Cause of the Thermocline Intensification
The strongest period of thermocline intensification in our records occurred between 14.3 and 12.8 Ma (Fig. 7), coeval with the expansion of a permanent EAIS (e.g., Lewis et al., 2008) and general global cooling associated with the MMCT. We suggest that the primary mechanism driving thermocline intensification was increased tropic-to-polar temperature gradients due to high latitude cooling, as might be expected from development of polar ice sheets. Additionally, thermocline intensification might have resulted from development of a stronger North Atlantic subtropical gyre circulation following the MCO. Stronger flow of wind-driven surface currents, namely the Gulf Stream/North Atlantic Current in the North Atlantic, would result in an intensification of subtropical gyre circulation (Fig. 1). While both seismic stratigraphic and deep-sea drilling data suggest evidence of Gulf Stream flow in the late Oligocene (Schlager et al., 1988) to as far back as the Cretaceous (Sheridan et al., 1981), such flow was likely sluggish due to the presence of the Central American Seaway. A shoaling of the Isthmus of Panama coupled with intensified meridional thermal gradients resulted in an intensification of the Caribbean Loop-Gulf Stream/North Atlantic Current system from approximately 15–12 Ma (Mullins et al., 1987; Sentman et al., 2018). In turn, an invigorated Gulf Stream/North Atlantic Current would spin up circulation of the North Atlantic subtropical gyre, resulting in the increased thermocline gradients as we observe at Site 558, and thermally isolating surface waters within the subtropical gyre.
We provide the first detailed surface and intermediate water stable isotope records from the Miocene North Atlantic subtropical gyre and focus on reconstructing surface to intermediate changes in the gyre. Our stable isotope records of planktonic and benthic foraminifera from Site 558 provide valuable insights into thermal evolution of the North Atlantic subtropical gyre during the Middle to Late Miocene. A ca. 17 Ma sharp decrease of ∼1.0‰ in surface δ18O and ∼0.7‰ in thermocline δ18O compared to only a 0.2‰ decrease in global δ18Osw suggests a 2–4°C warming of the surface ocean at the onset of the Miocene Climate Optimum (MCO; 17–14.8 Ma). Covarying positive excursions in oxygen isotopic data suggest three distinct surface cooling phases in the surface ocean following the MCO: a gradual cooling of at least 2°C across the Middle Miocene Climate Transition (MMCT; 14.8–12.8 Ma) and two more rapid but lower-magnitude phases between 11–10.3 Ma and 8.8–8.4 Ma. The low vertical δ18O gradients prior to the MCO imply a weaker thermal stratification of the water column in the subtropical gyre. The first significant increase in δ18O gradients occurred coincident with the onset of the MCO. The most pronounced divergence of ∼0.75‰ between sea surface and thermocline δ18O occurred during the MMCT and persisted through the Miocene, which we interpret to be a strengthening of the thermocline. We propose that invigorated Gulf Stream/North Atlantic Current transport caused by the partial closing of the Isthmus of Panama intensified subtropical gyre circulation in the North Atlantic, in turn strengthening the subtropical thermocline and thermally isolating subtropical surface waters from those at higher latitudes. This proposition can be tested by comparison with stable isotopic data particularly from higher latitude locations, and future studies should extend these studies to higher and lower latitudes.
Genus Trilobatus Spezzaferri et al., 2015
Type species: Globigerina trilobaReuss, 1850
Trilobatus quadrilobatus (d'Orbigny, 1846)
Diagnosis. Trilobatus quadrilobatus is distinguished by having a subtriangular outline. Globular chambers have depressed sutures. The umbilicus is open with a large primary aperture with distinct supplementary apertures. This species is distinguished from its descendant T. sacculifer by absence of the elongate final chamber/sac (Kennett & Srinivasan, 1983).
Genus Trilobatus Spezzaferri et al., 2015
Type species: Globigerina trilobaReuss, 1850
Trilobatus trilobus (Reuss, 1850) Figure 9A
Diagnosis. Trilobatus trilobus has three globose chambers, with a significantly larger final chamber. The umbilicus is narrower than T. quadrilobatus and features a slit-like broad primary aperture. Trilobatus trilobus is further distinguished from its descendant T. quadrilobatus by the presence of three main chambers as opposed to 3.5–4 (Poole & Wade, 2019).
Genus DentoglobigerinaBlow, 1979
Type species: Globigerina galavisiBermudez, 1961
Dentoglobigerina altispira (Cushman & Jarvis, 1936) Figure 9B
Diagnosis. Dentoglobigerina altispira is large with a high trochospire. The surface is cancellate with a wide and deep aperture containing a large central umbilical tooth. We only selected high trochospiral specimens with four chambers in the final whorl to avoid overlap with D. globosa, which has 5–6 chambers. The final chamber is flatter than that of D. globosa and D. globularis.
Dentoglobigerina venezuelana (Hedberg, 1937) Figure 9C
Diagnosis. Dentoglobigerina venezuelana is distinguished by having a large globular test that is circular in outline with 3.5–4 chambers in the final whorl. The final chamber is commonly reduced in size, appears flattened relative to the other three chambers, and can overhang the umbilicus. The aperture is narrow and rectangular in shape. The test has a cancellate wall structure, which in combination with its circularity and reduced final chamber makes it quite recognizable.
Family PlanulinidaeBermúdez, 1952
Genus Planulina Spezzaferri et al., 2015
Type species: Planulina ariminensesd'Orbigny, 1826
Planulina wuellerstorfi (Schwager, 1866)
Diagnosis. Planulina wuellerstorfi is referred to Planulina following van Morkhoven et al. (1986) and Holbourn et al. (2013b). Planulina wuellerstorfi is distinguished by ten narrow, rapidly increasing in size, and strongly curved chambers in the final whorl. The test is large, flattened, involute and gently convex dorsally with a blunt keel on the early part of the final whorl. This taxon appeared in the earliest Middle Miocene (ca. 16 Ma; Miller & Katz, 1987).
This study was conducted as part of the senior honors thesis of the senior author (M.G.) and was supported by NSF grant OCE 16-56960. We thank R.A. Mortlock for help with stable isotope analyses. We thank A. Lam and A. Shevenell for thoughtful reviews. Oxygen and carbon isotope data are available for download through NCEI (https://www.ncei.noaa.gov/access/paleo-search/study/37238). The Appendix can be found linked to the online version of this article.
Stable isotopic data.