Middle Miocene Southern Ocean Cooling and Antarctic Cryosphere Expansion

See allHide authors and affiliations

Science  17 Sep 2004:
Vol. 305, Issue 5691, pp. 1766-1770
DOI: 10.1126/science.1100061


Magnesium/calcium data from Southern Ocean planktonic foraminifera demonstrate that high-latitude (∼55°S) southwest Pacific sea surface temperatures (SSTs) cooled 6° to 7°C during the middle Miocene climate transition (14.2 to 13.8 million years ago). Stepwise surface cooling is paced by eccentricity forcing and precedes Antarctic cryosphere expansion by ∼60 thousand years, suggesting the involvement of additional feedbacks during this interval of inferred low-atmospheric partial pressure of CO2 (pCO2). Comparing SSTs and global carbon cycling proxies challenges the notion that episodic pCO2 drawdown drove this major Cenozoic climate transition. SST, salinity, and ice-volume trends suggest instead that orbitally paced ocean circulation changes altered meridional heat/vapor transport, triggering ice growth and global cooling.

The middle Miocene climate transition (MMCT), 14.2 to 13.8 million years ago (Ma), is one of the three major steps in Earth's Cenozoic climate evolution (13). The ∼1‰ increase in the oxygen-isotopic composition (δ18O) of benthic foraminifera describes a combination of Antarctic ice growth and global cooling at ∼14 Ma, as is also indicated by Southern Ocean ice-rafted detritus, eustatic change, and the fossil record (16). However, because δ18O records both temperature and global ice volume, fundamental questions and uncertainties exist concerning the magnitude and phasing of middle Miocene ice growth and cooling. The development of Mg/Ca, an independent paleotemperature proxy measured on the same foraminiferal calcite (CaCO3) as δ18O, has facilitated isolation of the ice-volume component of δ18O records (712). The Mg/Ca content of foraminifera increases exponentially with temperature (∼9 ± 1% per 1°C) and is relatively insensitive to salinity and ice-volume fluctuations (7, 8). Low-resolution paired benthic foraminifer Mg/Ca and δ18O studies designed to constrain the timing and magnitude of pre-Quaternary ice-volume fluctuations suggest substantial Antarctic ice growth (∼0.85‰) and a concomitant deep ocean cooling (2°C to 3°C) during the MMCT (11, 12). The magnitude of Antarctic ice growth and rapidity of this climate transition [<0.5 million years (My)] suggests that Earth's climate system was highly sensitive to oceanic, atmospheric, and cryospheric feedbacks.

Ocean circulation and atmospheric pCO2 variations are often cited as potential catalysts of the MMCT (1317). Large-scale reorganizations of ocean circulation driven by atmospheric circulation changes and/or tectonic reorganizations of gateway regions may have altered poleward heat and moisture transport, resulting in Antarctic ice growth and global cooling (1315). Ocean circulation hypotheses are supported by δ13C proxy evidence (14, 15, 18, 19) and the timing of tectonic events in the eastern Tethys/Indonesia (4, 20) and the North Atlantic (13). Alternatively, atmospheric pCO2 drawdown, through organic carbon sequestration on the mid-latitude continental margins (16) and/or enhanced silicate weathering rates (17), may have driven Antarctic ice-sheet expansion and cooling at ∼14 Ma. Support for this “Monterey Hypothesis” comes from thick, organic carbon-rich Miocene sedimentary sequences around the Pacific Rim (4, 16) and a corresponding ∼ 1‰ increase in global deep sea δ13C (4, 16, 21, 22). A potential complication of the hypothesis is revealed by paleo-pCO2 estimates (2325), which indicate that atmospheric pCO2 levels declined >3 My before the MMCT and provide little support for either elevated atmospheric pCO2 during the warm Miocene climatic optimum (MCO) (17 to 14 Ma) or a semipermanent atmospheric pCO2 decrease at the MMCT. These estimates indicate that factors other than those related to global carbon cycling may contribute to this major Cenozoic climate transition. To evaluate the processes and feedbacks involved in the MMCT, detailed information is needed regarding the phasing of carbon cycling, Antarctic ice growth, and high-latitude oceanic/atmospheric cooling. Acquiring this information has thus far proven difficult because of the limited availability of CaCO3-rich Southern Ocean sediments and the lack of an unambiguous paleotemperature proxy.

Here, we present an independent record of middle Miocene high-latitude Southern Ocean sea surface temperature (SST). To establish the thermal and hydrographic response of Southern Ocean surface waters and the phasing of high-latitude SST change, Antarctic cryosphere expansion, and global carbon cycling between ∼17 and 13 Ma, we generated paired Mg/Ca and δ18O records from surface dwelling planktonic foraminifer Globigerina bulloides in conjunction with benthic foraminifer (Cibicidoides mundulus) δ18O and δ13C records at Ocean Drilling Program (ODP) Holes 1170A (47°90'S, 146°02.98'E; 2704 m) and 1171C (48°30'S, 149°06.69'E; 2150 m) on the South Tasman Rise (STR) (26) (Fig. 1). Plate tectonic reconstructions indicate that Sites 1170 and 1171 were situated at ∼55°S in the middle Miocene (27), with calculated paleodepths of 2100 m and 1600 m, respectively (28). We used G. bulloides for several reasons. First, a well-defined modern subantarctic G. bulloides Mg-temperature calibration exists (7). Second, previous studies demonstrate the utility of G. bulloides Mg/Ca in reconstructing Quaternary subantarctic SSTs (7, 9). Finally, G. bulloides is continuously present through both STR middle Miocene sequences (26). Hole 1171C Miocene G. bulloides Mg/Ca ranges between ∼1.7 and 4.3 mmol/mol (Fig. 2 and table S1); minimum Mg/Ca values are similar to those from subantarctic core tops (7, 9). The main feature of the G. bulloides Mg/Ca record is the 1.8 mmol/mol point-to-point decrease centered at 166.4 m below sea floor (mbsf) that marks the transition from relatively high to relatively low Mg/Ca. Three step-like features are superimposed on this transition (Fig. 2). Hole 1171C G. bulloides δ18O ranges between ∼0.0‰ and ∼1.8‰ (29) (Fig. 2 and table S1). The main feature of the G. bulloides δ18O record is the 0.9‰ point-to-point increase centered at 165.6 mbsf that marks the transition to more positive δ18O values. The step-like structure of the Mg/Ca transition does not appear in the G. bulloides δ18O record, which reflects ice volume and salinity influences in addition to temperature.

Fig. 1.

Annual average SSTs in the subantarctic southwest Pacific (39) with Ocean Drilling Program (ODP) South Tasman Rise (STR) sites indicated. ODP Site 1171 is located at the intersection of the southward-flowing East Australian Current (EAC) and the eastward-flowing Antarctic Circumpolar Current (ACC), one of three locations where heat is introduced to the Southern Ocean (26). ODP Site 1170 is located 50 km northwest of Site 1171, beyond the influence of the EAC (26). In the middle Miocene, the STR was situated at ∼55°S, 5° to 7° south of its present location (27). Backtracked middle Miocene paleodepths of Sites 1170 and 1171 are 2100 m and 1600 m, respectively (28). Currently, STR hydrography is controlled by the Subtropical Convergence (STC) to the north and the Subantarctic Front (SAF) to the south. Seasonal SST variation is ∼4°C (26, 39). G. bulloides calcifies in the mixed layer during the austral spring when regional surface temperatures range between 8°C and 12°C (9, 39).

Fig. 2.

Climate proxy data from South Tasman Rise (STR) ODP Hole 1171C (48°30'S, 149°06.69'E; 2150 m). The Mg/Ca record was generated on planktonic foraminifer species G. bulloides at 20-cm intervals from 125 to 154mbsf and from 154to 207 mbsf and at 10-cm intervals between 154and 172 mbsf. Each point represents an average of one to four analyses; pooled SD of all replicates (df = 64) is ±0.21 mmol/mol (±7.8%). Measured modern southwest Pacific core top Mg/Ca is ∼1.6 mmol/mol (9). Oxygen isotope (δ18O) data were generated from G. bulloides at 10-cm intervals between 125 and 207 mbsf. The black arrow denotes the midpoint of the major 1.8 mmol/mol Mg/Ca transition, and the gray arrow denotes the midpoint of the 0.9‰ G. bulloides δ18O transition. Gaps in the data are due to incomplete core recovery (26).

Diagenesis and dissolution may potentially alter the primary Mg/Ca signal encoded in planktonic foraminiferal calcite and, thus, the inferred SST record (811) (SOM Text). Several lines of evidence suggest that these processes have not biased our STR middle Miocene Mg/Ca values: (i) G. bulloides ultrastructures are visible, and shells do not exhibit extensive pitting, fragmentation, infilling, or overgrowth; (ii) the average CaCO3 content of the middle Miocene ooze is stable at ∼94% by weight (26) as a result of the intermediate STR paleodepths (28); and (iii) benthic and planktonic foraminifers exhibit interspecific δ18O and δ13C offsets (30).

We converted STR G. bulloides Mg/Ca values to temperatures using the calibration of Mashiotta et al. (7): SST = ln(Mg/Ca/0.474)/0.107 (SE, ±0.8°C). This calibration successfully estimates subantarctic Pacific (45°S to 56°S) and Indian Ocean (43°S) austral spring SSTs in both the modern and Quaternary (7, 9). As ours is the first preQuaternary study employing this calibration, we assume that the environmental preferences of middle Miocene subantarctic G. bulloides and physical processes governing Mg uptake are similar to those in modern subantarctic G. bulloides. These processes and seawater Mg/Ca (31) may have changed through time and could contribute an additional uncertainty of up to 3°C in the conversion of Mg/Ca to absolute temperature (SOM Text). However, this uncertainty would not affect the magnitude of inferred temperature shifts on time scales <1 My. Age models derived from magnetostratigraphy, biostratigraphy, and stable isotope datums (32) provide the chronologic framework for the STR records (SOM Text).

Mg/Ca evidence from Hole 1171C indicates that regional SSTs were ∼2°C cooler after ∼14 Ma (14.7 ± 1.1°C; 13.9 to 12 Ma) than during the preceding MCO (17.0 ± 1.2°C; 17 to 14 Ma) (Fig. 3) (3, 4). Paleobotanical and faunal paleotemperature estimates from southern Australia and New Zealand support this trend and are similar to, or slightly cooler than, G. bulloides Mg/Ca-derived SST estimates (4, 33). The major feature of the STR SST record is a ∼7°C (point-to-point) transition (14.2 to 13.9 Ma) from the warmest MCO to cooler post-MMCT SSTs (Fig. 3). Inferred orbital-scale [∼400 and 100 thousand years (ky)] SST variability is particularly pronounced during the MMCT. Cooling occurred in three distinct steps (midpoints, 14.2, 14.0, and 13.9 Ma) and was followed by an interval of relatively cold SSTs (13.7 ± 1.1°C; 13.9 to 13.8 Ma) (Figs. 3 and 4). At ∼13.8 Ma, regional SSTs warm by ∼2°C to 3°C and remain relatively stable until ∼12 Ma. We confirmed the Hole 1171C SST estimates by measuring G. bulloides Mg/Ca at a second STR location, Hole 1170A [∼54°S (27); 2100 m (28)] (Figs. 1 and 3, table S2), with a different depositional history (26). Similarities in both magnitude (±0.5°C) and structure of the two records (Fig. 3) further support our interpretation of STR G. bulloides Mg/Ca as a primary climate signal. We attribute several offsets between the records to age model and sampling resolution differences (30) (SOM Text).

Fig. 3.

Hole 1171C middle Miocene (12 to 17 Ma) G. bulloides δ18O, SST, and benthic foraminifer (C. mundulus) δ18O (32) records versus age. Mg/Ca–derived SSTs from Hole 1170A (open circles) are plotted with the Site 1170 age model. Age models at Sites 1170 and 1171 were constructed using magnetostratigraphy, biostratigraphy, and stable isotope datums (26, 32). The G. bulloides Mg/Ca values were converted to SST by using the equation of Mashiotta et al. (7). Both the Mg/Ca and SST scales are given. The C. mundulus δ18O record serves as a general proxy for Antarctic ice volume (11, 30) (SOM Text).

Fig. 4.

Expanded view of the 13- to 15-Ma interval. Site 1171 SSTs are compared with C. mundulus δ18O and δ13C records (32). Orbital solutions (36) are plotted; eccentricity (black) and obliquity (gray). Paleo-pCO2 estimates derived from southwest Pacific alkenone δ13C (23); maximum (solid line) and minimum (dashed line) estimates are plotted. All records are plotted on the Site 1171 age scale (32) (SOM Text). Midpoints of the SST and C. mundulus δ18O transitions are 14.07 Ma (black arrow) and 14.01 Ma (gray arrow), respectively. The final δ13C increase of the “Monterey” δ13C excursion (CM6) begins at 13.8 Ma and terminates at 13.6 Ma.

Large-amplitude SST fluctuations over the STR suggest substantial reorganization of Southern Ocean surface waters during the MMCT. One possible explanation is an orbitally paced intensification of the Antarctic Circumpolar Current (ACC) system, perhaps related to increased westerly wind strength. This idea is supported by a paleobotanic record of regional wind strength (4, 34), substantial evidence for increased meridional thermal gradients (4), the timing of major Southern Ocean unconformities (4, 5), and STR surface water δ18O (δ18Ow). We used Hole 1171C G. bulloides Mg/Ca SST and δ18O (table S1) to calculate δ18Ow (7, 8) (SOM Text) (fig. S1), which should reflect some combination of ice volume and local salinity effects. Large SST fluctuations during the MMCT obscure the G. bulloides δ18O increase, which indicates that STR δ18Ow likely reflects regional salinity variations. Strict interpretation of the δ18Ow results suggests a gradual freshening of regional surface waters between 14.2 and 13.8 Ma, consistent with enhanced ACC strength and the resultant increased influence of subantarctic surface waters over the STR. A notably similar pattern of oceanographic change is recognized in the southwest Pacific during Quaternary glaciations (9, 35).

Orbital-scale cyclicity (400 and 100 ky) is inferred in the STR records of G. bulloides Mg/Ca and C. mundulus δ18O and δ13C. This cyclicity indicates a potential role for Milankovitch forcing during the MMCT, which suggests that Earth's climate system was particularly sensitive to forcing at eccentricity frequencies between ∼15 and 13.5 Ma. Comparison of orbital curves (36) to untuned STR SST and C. mundulus δ18O, an ice-volume proxy (32) (SOM Text), reveals that the MMCT occurred within an interval of high obliquity and moderate eccentricity variance (Fig. 4) (36). This relationship suggests that, at 14.2 Ma, Antarctic ice sheets were large enough to survive warm summer orbital configurations but small enough to respond dynamically to orbital forcing (37). Surface cooling generally corresponds with long-period (∼400 ky) eccentricity minima between 14.2 and 13.8 Ma (which are similar in amplitude to previous and subsequent minima) and precedes Antarctic ice growth by ∼60 ky (as judged by the midpoints of the major Mg/Ca and C. mundulus δ18O transitions). Our data suggest that orbital forcing paced both Southern Ocean SSTs and Antarctic ice growth during the MMCT. There is, however, no obvious orbital anomaly akin to that recognized at the Oligocene/Miocene boundary (38), and, assuming a synchronous atmospheric and SST response, the substantial time lag (∼60 ky) between orbital forcing and ice growth supports the notion that additional feedbacks (e.g., carbon cycling, ocean circulation, and/or ice/albedo feedbacks) may be required for substantial and rapid Antarctic ice growth. Heightened sensitivity to eccentricity forcing between 14.2 and 13.8 Ma and the phasing of SST and ice volume (δ18O) are clues to identifying the feedbacks involved in this climate threshold.

Model results stress a fundamental role for atmospheric pCO2 in Cenozoic climate change and indicate that the Antarctic cryosphere may react sensitively to climate feedbacks only when atmospheric pCO2 is relatively low and within some narrowly defined range (37). Sensitivity to eccentricity forcing is apparent in many middle Miocene δ13C records between 17 and 13.5 Ma (4, 14, 15, 18, 21, 22), which suggests at least some role for the carbon cycle in the MMCT. However, the >3 My discrepancy between the initial atmospheric pCO2 decline (23, 24) and the MMCT has shifted focus to the role of episodic pCO2 drawdown, associated with more positive δ13C carbon maxima (CM) events, in triggering this major climate step (21). Moderate resolution records suggest that CM events, which are attributed to climate-induced variations in marine organic carbon sequestration, co-vary with analogous δ18O events at the 400-ky frequency (14, 15, 18). Comparison of STR G. bulloides Mg/Ca to C. mundulus δ18O (ice volume) and δ13C, a general proxy for carbon cycling and atmospheric pCO2, reveals that Southern Ocean surface-water cooling and Antarctic ice growth occurred between CM5 and CM6 during an interval of increasing pCO2 (Fig. 4) (23). This sequence of change challenges the notion that episodic pCO2 drawdown was the primary forcing that triggered the MMCT, again implying that other feedbacks (e.g., ice/albedo) played a more significant role in this climate transition. Our records also indicate that CM6, one of the largest CM events, corresponds with maximum Antarctic ice volume and a slight warming of the high southern latitudes. This relationship raises the possibility that feedbacks related to Antarctic cryosphere expansion may have exerted control over the global carbon cycle through enhanced ventilation of intermediate and deep ocean waters and falling sea levels (4, 6, 23). The long-term trend observed in the global δ13C record after the rapid expansion of Antarctic ice volume supports this interpretation (3).

Our results demonstrate that STR SSTs cooled 6°C to 7°C between 14.2 and 13.8 Ma, revealing that the Southern Ocean was a dynamic component of the MMCT. Eccentricity-paced Southern Ocean surface cooling and freshening suggest that atmospheric/oceanic circumpolar circulation intensified in response to orbital forcing, increasingly thermally isolating Antarctica during the MMCT (14.2 to13.8 Ma). Middle Miocene intensification of the ACC may have played a major role in Cenozoic climate evolution, both directly through changes in meridional heat transport and indirectly through changes in vapor transport. We speculate that sensitivity to eccentricity forcing increased at 14.2 Ma, immediately following peak warmth of the MCO (2325) as a result both of low atmospheric pCO2 (37) and of a fundamental reorganization of the climate system, specifically a tectonically mediated reduction in meridional heat/vapor transport related to the constriction of the Eastern Tethys at ∼15 Ma (4, 14, 20). The presence of 100-ky variability (14.2 to 13.8 Ma) and a rare shift in eccentricity cadence between 15 and 14 Ma (36) are intriguing, and future efforts should focus on understanding the evolution of the climate spectrum on orbital time scales during the MMCT.

Supporting Online Material

Materials and Methods

SOM Text

Fig. S1

Tables S1 and S2


References and Notes

Stay Connected to Science

Navigate This Article