Antarctica is the least explored continent on our planet Earth, largely due to today's massive ice cover on the continent, reaching a thickness of 4500 m in places, leaving only 0.3% of the land area uncovered. This ice sheet, however, was not always in place, and its inception ∼34 m.y. ago at the Eocene−Oligocene boundary marked one of the most fundamental climate transitions in recent Earth history: the transition from the greenhouse world of the Cretaceous and early Cenozoic to the icehouse world we are currently living in (e.g., Zachos et al., 2008). The paper by Scher et al. (2011, p. 383 in this issue of Geology) provides a detailed record of pulses in Antarctic continental weathering through this glacial onset.

Global climate in the early Cenozoic seems to have been characterized by low latitudinal temperature gradients and subtropical temperatures at high latitudes (e.g., Bjil et al., 2009). There was no or only very little ice on the poles, and atmospheric CO2 levels were probably well in excess of 1000 ppm. State-of-the-art climate models and palaeoclimatic proxy data suggest that the main triggering mechanism for initial inception and development of the Antarctic ice sheet was the drop of atmospheric CO2 concentrations below a critical threshold (∼750ppm; DeConto et al., 2008). While it remains a topic of debate whether the tectonic configuration of Southern Ocean gateways influences the sensitivity of Antarctic temperatures to atmospheric CO2 concentrations (e.g., Sijp et al., 2009), changes in silicate weathering are arguably the most important mechanisms for long-term draw down of atmospheric CO2 (e.g., Kent and Muttoni, 2008).

The Eocene−Oligocene transition, however, reveals a very rapid response of the climate system to initial cooling and ice buildup in Antarctica. The marine geological record documents a two-step increase in deep-sea benthic foraminiferal oxygen isotopes in less than 300 k.y. (Coxall et al., 2005), marked surface and deep ocean cooling of 4–5 °C at high latitudes (Liu et al., 2009), deposition of the first ice-rafted debris layers and a switch from chemically to physically weathered clay minerals in Southern Ocean sediments (e.g., Barker et al., 2007), pronounced deepening of the carbonate compensation depth in the Pacific Ocean (Coxall et al., 2005), increased productivity in the Southern Ocean, and replacement of carbonate-rich facies on passive margins by siliciclastics (see the discussion in Merico et al., 2008).

Taken together, these observations are indicative of the close interrelationships between Earth's cryosphere, ocean chemistry, and the carbon cycle. Scher et al. take an innovative approach to investigate this interplay across the Eocene−Oligocene transition. First, they produced a seawater Nd isotope record, extracted from fossil fish teeth, to trace the flux of continental weathering−derived Nd to the deep waters of the Prydz Bay region of the Southern Ocean. They then combined this with a set of oxygen isotopes from deep-sea benthic foraminifera, which track both the volume of continental ice sheets and deep ocean cooling. Their results show a stunning two-stepped Nd isotope excursion that correlates very well with the global deep-sea oxygen isotope record, and slightly predates the arrival of the first ice-rafted debris, a direct proxy for continental-scale glaciation. Scher et al. suggest that the seawater Nd isotope record can be interpreted as two surges of weathering, generated by Antarctic ice growth—a novel idea that requires some further explanation.

Seawater chemistry (at any point back in time) principally depends on the flux of solutes from the continents, which in turn depends on rates of physical denudation and chemical weathering. Weathering of silicate rocks not only acts as a long-term sink for atmospheric CO2, but also strongly influences global biogeochemical cycles by determining continental runoff. While chemical weathering is vital to the flux of nutrients to the ocean (e.g., Raiswell et al., 2006), this flux is also strongly coupled to the availability of fresh mineral surfaces with high reactivity. Global field studies, encompassing a variety of climate zones and erosional regimes, show a tight coupling between the supply of fresh material and chemical weathering rates (e.g., Millot et al., 2002). High weathering rates are not necessarily linked to tropical areas, as often assumed. In contrast, temperate glaciers that have water available at their bed to facilitate basal sliding and physical erosion are ‘mineral surface factories,’ and studies on mountain glaciers reveal some of the highest mechanical and chemical denudation rates (for a summary, see Anderson, 2007).

Glacial grinding does not change the bulk mineralogy of source rocks, but exposes accessory phases and makes them accessible to chemical weathering. Among others, the radiogenic isotope systems of U/Th-Pb and Lu-Hf can monitor such changes in the style of weathering, as the parent/daughter ratios show significant variations between different mineral phases. As a consequence, solute continental runoff for these isotope systems can deviate significantly from the bulk rock signature (e.g., Harlavan et al., 1998). Seawater Pb and Hf isotope records have been used previously for studies of changes in the style of weathering in the Northern Hemisphere during the late Quaternary (e.g., van de Flierdt et al., 2002, Foster and Vance, 2006), and would also be ideally suited to provide insights into the dynamics of Antarctic continental weathering across the Eocene−Oligocene transition.

But what about Nd isotopes? While a number of studies indicate that rare earth element mobility during weathering could lead to small Sm/Nd fractionation (measurable Nd isotope effects in weathered glacial tills, boreal river water, and sediment leachates; von Blanckenburg and Nägler, 2001, and references therein), it seems unlikely that this effect is large enough to have an impact on seawater budgets. Overall Nd isotopes are probably not significantly fractionated during weathering, and seawater records still reflect the isotopic fingerprint of the continental source area that has been eroded. This notion is supported by similar proportions of Sm and Nd being incorporated in most common rock-forming minerals (see the compilation in Bayon et al., 2006), and is also in agreement with the observation that dissolved and suspended loads of rivers generally exhibit similar Nd isotopic compositions (e.g., Goldstein and Jacobsen, 1987). Hence, there are strong indications that dissolved Nd isotopes in seawater monitor the flux of weathered Nd from the continents, rather than the actual style of weathering.

Consequently, Scher et al. interpret the observed two-stepped seawater Nd isotope excursion across the Eocene−Oligocene boundary as a two-stepped change in Antarctic weathering flux, reflecting increased continental runoff created by the interplay of physical and chemical denudation. This interpretation adds to the evidence that the latest Eocene paleoenvironment in Antarctica was characterized by small isolated mountain glaciation, and a fluvial erosion pattern not too dissimilar to today's ice drainage pattern (Jamieson and Sugden, 2008). At the end of the Eocene, a permanent transition took place from an environment dominated by chemical weathering to one dominated by physical weathering featuring an ice-covered Antarctic continent (e.g., Barker et al., 2007). Stepwise advance of temperate glaciers could have provided the grinding and water needed for chemical reactions to create distinct pulses of weathering runoff. Furthermore, the expansion of ice onto areas of the continent that were not previously covered by riverine drainage may have facilitated erosion of Nd-rich iron(hydr)oxides, pre-formed in an Eocene (subtropical) weathering environment (Bayon et al., 2004). If such iron(hydr)oxides were formed in areas of older bedrock geology (e.g., Southern Prince Charles Mountains; see the appendix in Williams et al., 2010) they could have provided a large flux of Nd with a particularly low Nd isotopic composition.

The short-lived nature of the spectacular Nd isotope excursion observed by Scher et al., and its coincidence with the benthic deep-sea oxygen isotope record, intrinsically ties it to the major ice expansion in Antarctica. It seems plausible that a significant weathering flux would precede the arrival of ice on the continental margin, and hence ice-rafted debris production (Scher et al., this volume). At this point, one could be tempted to use the Nd isotope information in a more quantitative way to constrain silicate weathering rates and potential effects on atmospheric CO2 draw down during the Eocene−Oligocene transition. While the community should strive to achieve such a quantitative understanding in the future by means of robust modern process studies and modeling of geochemical budgets (e.g., Vance et al., 2009), we have to acknowledge that one record from the Southern Ocean is not sufficient to take this last step yet. It is, however, studies like the one presented by Scher et al. that stimulate the application of novel climate proxies to the field and foster new avenues of research.

The material for such research is already in sight: two recent Integrated Ocean Drilling Program (IODP) expeditions have recovered some of the sedimentary archives needed to further refine our understanding of the Eocene−Oligocene transition. Expeditions 320 and 321 sailed in 2009 and retrieved a Pacific Equatorial Age Transect (PEAT) of cores, containing Eocene−Oligocene sections (Pälike et al., 2010). These distal records will provide far-field constraints on Antarctic ice buildup, and complement the first ever proximal record drilled across the Eocene−Oligocene boundary at the Antarctic Wilkes Land margin (IODP Expedition 318; Expedition 318 Scientists, 2010). Exciting times lie ahead for advancing our understanding on the complex interaction of climate, tectonics, and ocean biogeochemical cycles.