NEWS & VIEWS

294 nature geoscience | VOL 1 | MAY 2008 | www.nature.com/naturegeoscience

PALAEOCLIMATOLOGY

A tale of two climates

Katharina Billupsis at the College of Marine and Earth Studies, University of Delaware, 700 Pilottown Road, Lewes, Delaware 19958, USA.

e-mail: kbillups@udel.edu

S ubstantial Antarctic glaciation commenced rapidly 34 million years ago at the geologic interval known

as the Eocene–Oligocene boundary, which separates one of the warmest intervals of the past 65 million years from one of its coldest periods1,2 (Fig. 1). Th is important climatic step from a ‘greenhouse’ to an ‘icehouse’ has long interested researchers because insights may be gained into the factors that drive climate from one state to another. Assembling a picture of such a major climatic event may seem relatively easy because of the large footprint left in the geologic record. However, quantifying what is probably the most dramatic climatic transition of the Cenozoic era in terms of generated ice volume and temperature drop has been a challenge to palaeoceanographers, in part owing to the lack of deep-sea evidence for the suspected oceanic cooling. On page 329 of this issue, Katz and co-authors3 bring together geochemical and sedimentary evidence to describe in fi ne detail climatic change and sea-level history across the Eocene to Oligocene transition.

Large-scale climatic changes, such as the expansion of Antarctic ice during the Eocene–Oligocene climate transition, are an ideal example of the application of foraminiferal geochemistry to the study of palaeoclimate. Because the formation of polar ice sequesters water molecules with the lightest of the oxygen isotopes, 16O, the growth and decay of an ice sheet brings about changes in the oxygen isotopic composition of seawater (the ‘ice-volume’ eff ect). Th ese isotopic changes are preserved in the oxygen isotopic ratios (δ18O) of organisms that secrete calcium carbonate, such as the single-celled microscopic foraminifera (Fig. 1).

A large increase in ice volume at the Eocene–Oligocene boundary is thus

recorded as a rapid increase in the δ18O values of foraminifera that accumulated on the sea fl oor during this interval of time. However, in foraminifera from diff erent sediment cores, the measured increase in the δ18O values varies between 1 and 1.5‰2–5. Variability such as this arises from diff erences in regional seawater temperature, which aff ect the distribution of the 18O between the calcite and the sea water. As the temperature drops, the heavier isotope is preferentially incorporated in the foraminiferal calcite. Additionally, the δ18O values of surface water depend on the regional evaporation versus precipitation patterns (salinity). Th erefore, bottom- dwelling (benthic) foraminifera are more commonly used to reconstruct the global seawater signal because they are not as strongly infl uenced by regional water-mass salinity changes. Still, the δ18O signal of the benthic foraminifera is controlled by both temperature and ice volume.

One approach has been to combine Mg/Ca ratios, which provide an independent measure of the calcifi cation temperature, with the δ18O measurements.

Early Mg/Ca measurements showed no cooling associated with the Eocene–Oligocene transition, thus calling on ice-sheet build-up in the Northern Hemisphere to account for the entire oxygen isotope shift 2. However, large changes in the carbonate chemistry of the deep ocean also occurred at the Eocene– Oligocene transition, which would have aff ected the incorporation of Mg into the shell carbonate, potentially biasing the temperature record2.

Katz and co-authors have analysed both the Mg/Ca ratios and δ18O values of the benthic foraminifera from a shallow site on the continental shelf, thereby avoiding uncertainties related to regional water-mass salinity and deep-sea carbonate changes. Together, these measurements reveal the seawater cooling (2.5 °C) that occurred across this climate transition and constrain the δ18O seawater increase (1.2‰) refl ective of ice-volume growth.

However, the researchers point out that an increase in the δ18O value of sea water of 1.2‰ presents a conundrum: on the basis of late Pleistocene calibrations between

The generally warm and ice-free conditions of the Eocene epoch rapidly declined to the cold and glaciated state of the Oligocene epoch. Geochemical evidence from deep-sea sediments resolves in detail the climatic events surrounding this transition.

30 35 40 45

Oligocene Eocene

Antarctic glaciation

1

2

3

4

δ18O

(‰) δ18O seawater

Temperature

Ice volume(global)

Salinity(regional)

War

mer

Figure 1 The Eocene to Oligocene climate transition as recorded in foraminiferal oxygen-isotope records from all major ocean basins1. In this low resolution compilation of records1, the Eocene to Oligocene climate transition is outlined by a single major climatic step. Inset: The foraminiferal δ18O value is controlled by both the oxygen isotopic composition of seawater, a function of global ice volume and salinity, and the seawater temperature. Katz and co-authors3 use the Mg/Ca ratios of the foraminiferal calcite to constrain the temperature component to present a high-resolution record that reveals three climate steps. Image in inset reproduced with permission from ref. 9. Copyright (1994) Cambridge University Press.

NEWS & VIEWS

nature geoscience | VOL 1 | MAY 2008 | www.nature.com/naturegeoscience 295

SEISMOLOGY

Do faults shimmy before they shake?

Michael R. Brudzinski is in the Department of Geology, Miami University, 114 Shideler Hall, Oxford, Ohio 45056, USA.

e-mail: brudzimr@muohio.edu

J ust like us humans, the Earth has many ways to release its stress, ranging from the equivalent of a yell

in the form of a destructive earthquake to the gentle whisper of a tremor. Along plate boundary faults, two tectonic plates move in relation to each other in response to external stresses, but these faults sometimes get locked together because of friction. As a result they accumulate stress while the surrounding rock continues to deform. Th e release of this elastic strain, when the locked area fi nally gives way, generates a potentially devastating earthquake. Recent observations reveal that at depth, plate boundary faults can also release strain through slow slip, which can recur with remarkable regularity1. Th e conditions under which slow slip occurs and its spatial, temporal and potential causal relationships with earthquakes

were discussed at a recent workshop on “Aseismic Slip, Non-Volcanic Tremor, and Earthquakes” that was held in Sidney, British Columbia this February2.

Unlike normal earthquakes that happen instantaneously, slow slip episodes may last for hours to years and are usually seismically silent. However, they are often accompanied by a low-level rumbling called non-volcanic tremor (NVT)3, which generates detectable ground vibrations that are less impulsive than seismic waves from earthquakes (Fig. 1). NVT and slow slip have been observed in an increasing variety of tectonic environments4, including several subduction zones, the Californian San Andreas Fault, the detachment fault in the south flank of Kilauea of Hawaii and other natural systems (for example, landslides and glaciers).

Episodic tremor and slip (ETS) was fi rst discovered in the Cascadia subduction zone in 2003 and can be understood as slow slip on the interface between the overriding and subducting

plates at depths of 25 to 50 km. Across these depths, normal earthquake behaviour fades out and eventually gives way to aseismic creep on the plate interface1,3. Th e high-resolution seismic network of Japan revealed the not-quite-silent signature of this slow slip: the seismometers detected small-amplitude signals with a long duration at frequencies lower than those characteristic of earthquakes. Th e location of these signals and the style of faulting inferred from them suggests that these small-amplitude signals are indicative of relative motion between the plates (D. Shelly, Univ. California- Berkeley)5,6.

Although a recent analysis of NVT in Cascadia supports its origin at the plate boundary interface, (K. Creager, Univ. Washington), several other studies indicate that a signifi cant amount of NVT occurs in the overriding plate (H. Kao, Geological Survey of Canada)7, suggesting that NVT is not always associated with slow slip or that some portion of slow slip occurs within the overriding plate.

Not only do plate boundary faults generate earthquakes, they also produce slow slip and non-volcanic tremor. New observations on these phenomena provide fresh insights into the conditions that dictate earthquake behaviour.

changes in the isotopic composition of sea water and sea level6, a δ18O increase of this magnitude would translate into a sea-level drop of 120–135 m. Th is estimate is larger than the stratigraphic evidence for a 55–70 m eustatic sea-level fall3,7. However, the eustatic estimate incorporates the isostatic rebound of the ocean basin, and, aft er accounting for this eff ect, refl ects an 82–105 m sea level lowering solely due to the removal of water3. On the basis of these observations, Katz and co-authors propose a new calibration of Oligocene δ18O sea water to sea level (0.12‰ increase in δ18O per 10 metre sea level decrease).

Katz and co-authors also discuss that the δ18O value of the polar ice is an important uncertainty in quantitatively relating changes in foraminiferal δ18O values to sea level, as this value is not preserved in the geologic record. Under simple mass balance considerations, to produce a given change in foraminiferal δ18O, a smaller change in

ice volume is required if the ice is more depleted in 18O. Th us to bring in line the observed δ18O seawater change of 1.2‰ with the 105 m sea-level fall, they argue for early Oligocene ice that was as depleted in 18O as Antarctic ice today (–45‰).

Katz and co-authors conclude that the Eocene to Oligocene climate transition occurred in three steps, beginning 33.8 Myr ago with a deep-water cooling of 2.5 °C, followed by sequential ice-volume expansion and concomitant sea-level fall, and ending about 33.5 million years ago in an Antarctic ice sheet that was about one quarter larger than the one that exists in the present day.

Th is study is timely; the overall results agree well with work published earlier this year5, which showed an early 2.5 °C cooling in tropical sea surface waters concurrent with similar deep-sea temperature changes at the study site and ended with the expansion of ice. Even though there are discrepancies

between the two studies concerning the amount of ice that accumulated during the transition, the implications for advancing the climate sciences are the same. Th ese studies fi nally provide evidence for signifi cant deep-sea cooling at a time of major ice-sheet development. Importantly, these studies provide constraints for numerical climate models incorporating global climate forcing factors such as greenhouse-gas levels8.

References1. Zachos, J. Pagani, M., Sloan, L., Th omas, E. & Billups, K.

Science 292, 686–693 (2001).2. Coxall, H. K. & Pearson, P. N. Geol. Soc. London (2007).3. Katz, M. E. et al. Nature Geosci. 1, 329–334 (2008).4. Kennett, J. P. & Shackleton, N. J. Nature

260, 513–515 (1976).5. Lear, C. H, Bailey, R., Pearson, P. N., Coxall, H. K. &

Rosenthal, Y. Geology 36, 251–254 (2008).6. Fairbanks, R. Nature 342, 637–642 (1989).7. Miller, K. et al. Science 310, 1293–1298 (2005).8. DeConto, R. M. & Pollard, D. Nature 421, 245–249 (2003).9. Bolli, H. M., Beckmann, J. P. & Saunders, J. B. Benthic

Foraminiferal Biostratigraphy of the South Caribbean Region (Cambridge Univ. Press, 1994).