Technical Comments

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.1186544


Sosdian and Rosenthal (Reports, 17 July 2009, p. 306) used magnesium/calcium ratios in benthic foraminifera from the North Atlantic to reconstruct past bottom-water temperatures. They suggested that both ice volume change and ice-sheet dynamics played important roles during the late Pliocene and mid-Pleistocene climate transitions. We present evidence that their record of deep ocean temperature is not reliable, thus raising doubts about their conclusions.

Sosdian and Rosenthal (1) used magnesium/calcium (Mg/Ca) ratios in the benthic foraminifera Cibicidoides wuellerstorfi (epifaunal) and Oridorsalis umbonatus (infaunal) to derive a 3.2-million-year record of bottom-water temperatures (BWTs) for the deep North Atlantic. Combined with benthic oxygen isotope ratios (δ18O), this allowed them to calculate deep seawater δ18O values, which provide an estimate of ice volume changes. Based on this, they concluded that the increase in ice volume during the late Pliocene transition was caused by global cooling, whereas an additional change in ice-sheet dynamics was likely to have induced the shift to 100,000-year cycles at the mid-Pleistocene transition. Here, we argue that their reconstructed down-core BWTs are not reliable, and therefore their conclusion is not substantiated and should be considered with caution.

The regional Mg/Ca-temperature calibration [Mg/Ca = 0.15 × BWT + 1.16 (Eq. 1)] used in (1) is defined by the regression of two data points: their core-top Mg/Ca and modern BWT versus Mg/Ca and an estimated BWT during the Last Glacial Maximum (LGM). We use published core-top and down-core data to demonstrate that this method is not valid. Because of incomplete information about past deep water conditions, a baseline is that a robust BWT reconstruction method should produce BWTs for core-top samples from extensive geographic regions closely matching modern values within the cited error. It is not acceptable to obtain seemingly reasonable results for only one core. Equation 1 does not fit core-top C. wuellerstorfi and O. umbonatus samples from global oceans (24) (Fig. 1A). Applying Eq. 1 to these two species shows large BWT deviations from expected temperatures (5) (Fig. 1B). The difference is not randomly distributed and is up to 5°C with an average estimation error of ±2.7°C, significantly larger than the ±1.1°C error quoted in (1). Application of Eq. 1 to a 2-km water depth core BOFS 11K, which is located at the same region as the cores used in (1), yields a ~2°C warmer BWT at the LGM than that during the Holocene, clearly at odds with benthic δ18O and colder LGM BWT estimated by porewater measurements (6) (Fig. 1D). Therefore, Eq. 1 used by (1) lacks support from both core-top and down-core data sets.

Fig. 1

Mg/Ca (A) and ΔT (B) versus BWT for global core-top C. wuellerstorfi (2) and O. umbonatus (3, 4). ΔT = BWT by Eq. 1 − BWT estimated from modern hydrographic data (5). The shaded bar in (B) indicates the ±1.1°C error quoted in (1). A constant interspecies offset in Mg/Ca of 0.16 mmol/mol (1) is applied before converting O. umbonatus Mg/Ca to BWTs using Eq. 1. (C) Mg/Ca in C. wuellerstorfi (2) and O. umbonatus (3, 4) versus deep water ΔCO32-. (D) C. wuellerstorfi Mg/Ca derived BWT using Eq. 1 for BOFS 11K (55.2°N, 20.4°W, 2004 m) compared with benthic δ18O (2). Also shown are BWTs at modern and LGM periods for a nearby core ODP 981 (55.48°N, 14.65°W, 2184 m) estimated by porewater measurements (6). Mg/Ca in O. umbonatus from the Norwegian Sea are corrected by 0.2 mmol/mol to account for cleaning effect (3). All other Mg/Ca values were measured using the same cleaning procedure (24).

Different from the planktonic foraminiferal and inorganic carbonates, C. wuellerstorfi and O. umbonatus from global oceans show no clear relationship between their Mg/Ca and growth BWTs in the range of –1°C to 4°C (24) (Fig. 1A). The elevated Mg/Ca in shells from the deep Norwegian Sea highlights an influence of the degree of carbonate ion saturation (ΔCO32−) on benthic foraminiferal Mg/Ca (2, 3) (Fig. 1C). The larger scatter of O. umbonatus (Fig. 1C) may indicate a complication from porewater chemistry associated with this species. The residual Mg/Ca for core-top C. wuellerstorfi after removing the ΔCO32− effect shows a negligible dependence on BWTs (2). Based on a comprehensive sensitivity test on a sequence of cores from 1- to 4-km water depth in the North Atlantic, it has been shown that it is not possible to use C. wuellerstorfi Mg/Ca for down-core BWT reconstructions even with the information about deep water ΔCO32− from an independent proxy (2). Deep North Atlantic waters (>~3 km water depth) are expected to have high ΔCO32− during the interglacial and low values during glacial periods (2). This ΔCO32− change could partially explain the lowered benthic Mg/Ca during glacial periods observed in cores used in (1).

Obtaining BWT and ice volume changes is important for understanding the mechanisms responsible for climate changes in the past. The precondition is to use a robust method and fully consider possible complicating factors. The fact that Eq. 1 proposed by Sosdian and Rosenthal (1) fails to yield reasonable BWTs for core-top and down-core samples casts doubt on the validity of their BWT and ice volume reconstructions involving Mg/Ca in benthic foraminifera.


View Abstract

Stay Connected to Science

Navigate This Article