Technical Comments

Response to Comment on “Deep-Sea Temperature and Ice Volume Changes Across the Pliocene-Pleistocene Climate Transitions”

See allHide authors and affiliations

Science  18 Jun 2010:
Vol. 328, Issue 5985, pp. 1480
DOI: 10.1126/science.1186768


Yu and Broecker argue that the paleoceanographic interpretation of our 3.2-million-year record of North Atlantic deep-sea temperature hinges on the determination of whether temperature or carbonate saturation is the primary driver of benthic foraminiferal magnesium/calcium ratios from the North Atlantic. Here, we present evidence supporting our argument that bottom-water temperature variability is the primary control on benthic foraminiferal Mg/Ca at our site.

Over the past 10 years, magnesium/calcium (Mg/Ca) ratios in benthic foraminifera have provided new insights into the Cenozoic history of deep ocean temperature and continental ice volume on various time scales. Yu and Broecker (1) question the reliability of our 3.2.-million-year record (2) of ocean temperature in the deep North Atlantic based on such ratios because of the confounding influence of carbonate ion saturation (ΔCO3).

Current calibrations suggest a dominant temperature control on benthic foraminiferal Mg/Ca above 5°C with a sensitivity of ~0.12 mmol mol−1 per °C (3). At temperatures below 4°C, a decrease in deep water ΔCO3 below 20 μmol kg−1 leads to an increasingly higher offset of Mg/Ca data from the expected warm-temperature calibration line (4, 5). Deep Sea Drilling Program (DSDP) site 607 (41°N, 32°W; water depth 3427 m), which we studied in (2), is characterized by modern bottom-water temperature (BWT) = 2.6°C and ΔCO3 = ~33 μmol kg−1 and therefore is currently minimally affected by the [CO3] ion effect (4, 5). This was likely also the case for other interglacials, and hence the uncertainty in reconstructing interglacial BWTs is relatively small. The largest uncertainty in our reconstruction (2) is, therefore, associated with reconstructing BWTs during the late Pleistocene glacial intervals when the presence of nutrient-rich, low [CO3] Antarctic Bottom Water at this site could have biased Mg/Ca estimates of BWT. Based on an estimated decrease of ~20 μmol kg−1 during the Last Glacial Maximum (LGM) from a nearby site (5) and Mg/Ca-ΔCO3 sensitivity of 0.0085 mmol mol−1 per μmol kg−1, we calculated ~0.17 mmol mol−1 LGM-to-Holocene (HL) change in Mg/Ca due to lower saturation state. Considering that 0.17 out of the measured 0.54 mmol mol−1 LGM-HL increase in Mg/Ca at site 607 was due to the increase in [CO3], we calculated an LGM-HL change of ~3.1°C, based on the temperature sensitivity of 0.12 mmol mol−1 per °C (3). This estimate is entirely consistent with the benthic foraminiferal oxygen isotope (δ18Ob) record showing an LGM-HL change of about –1.5 per mil (‰) in the companion piston core [Chain 82-24-23PC; 43°N, 31°W; water depth 3406 m (2)] assuming –0.8‰ change in LGM seawater composition in the North Atlantic (6). In contrast, if entirely due to the saturation effect, the site 607 Mg/Ca record would imply a >60 μmol kg−1 LGM-HL change in [CO3], which is larger than current estimates of 10 to 25 μmol kg−1 (79). In our report (2), we accounted for the [CO3] ion effect by using apparently greater temperature sensitivity of ~0.15 mmol mol−1 per °C, which yields similar LGM-HL change of 3.3 ± 1.1°C.

The above comparison demonstrates that ~70% of the LGM-HL Mg/Ca signal at site 607 is attributable to changes in BWT, in contrast to Yu and Broecker’s contention that carbonate ion changes primarily influence the Mg/Ca record at site 607 whereas BWT has only a weak effect. Nonetheless, we agree with Yu and Broecker that additional estimates of changes in [CO3] saturation could reduce the uncertainty in our BWT reconstruction (estimated at ±1.1°C). In the absence of these, we validate our 3.2-million-year reconstruction by comparing it to other climate records. Our temperature record suggests that climate cooling over the past 3.2 million years occurred primarily through two distinct events associated with the late-Pliocene and mid-Pleistocene shifts in the global δ18Ob record (2). BWT variations generally covary and are coherent with the δ18Ob record in frequency, average long-term trend, and glacial-interglacial (G-I) amplitude, and are consistent with the low-resolution ostracod temperature record from this site (10). In contrast, the lack of coherency and synchronicity between Mg/Ca, δ13C (as a nutrient/[CO3] analog) and deep Atlantic dissolution records across the Pleistocene suggests that saturation changes do not dominate the Mg/Ca variability at this site (2). Furthermore, our BWT reconstruction is consistent with other late-Pleistocene benthic foraminiferal Mg/Ca-derived BWT records from the Atlantic, Southern, and Pacific Oceans despite different carbonate histories in each basin (Fig. 1A) (11, 12) and with estimates based on δ18Ob in benthic foraminifera, suggesting G-I temperature variability of 2 to 4.5°C (13, 14). Of particular interest is the coherency and synchronicity (relative to δ18Ob) with the BWT record from Southern Ocean ODP (Ocean Drilling Program) site 1123. This record is derived from the benthic foraminifer Uvigerina sp., a species that is allegedly “free” of [CO3] ion effect (12), and thus supports the application of our site-specific calibration.

Fig. 1

Comparison of the Mg/Ca-derived BWT record from DSDP site 607 with (A) late Pleistocene BWT records derived from Uvigerina Mg/Ca from the southern Pacific Ocean (12), eastern equatorial Pacific [(11); recalculated using calibration from 12)], and porewater estimates for the North Atlantic (triangles) and South Pacific Ocean (circles) (6). Note the consistency among all the records, suggesting that during glacial maxima BWTs were similar and close to freezing in all sites, in agreement with conclusions derived from porewater data for the LGM. (B) Alkenone-derived SST record from the high-latitude North Atlantic ODP site 982 exhibits significant coherency in both the long-term trend and G-I variability (16).

The timing and magnitude of temperature change in our BWT record (2) show strong similarity with high-latitude sea surface temperature (SST) records across the Pliocene-Pleistocene climate transitions and on G-I time scales, indicating that our BWT record is indeed capturing variations in the temperature of high-latitude source waters (Fig. 1B) (15, 16). Notably, the agreement holds for both interglacial (when [CO3] effect is minimal) and glacial (when [CO3] effect is the largest) intervals. These lines of evidence and the consistency between our results and other late Pleistocene sea-level reconstructions (2) suggest that we are correctly accounting for changes in carbonate saturation and that therefore, within the quoted errors, our BWT reconstruction is valid.


View Abstract

Navigate This Article