See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/8554523

Regional climate shifts caused by gradual global cooling in the Pliocene Epoch

Article  in  Nature · June 2004

DOI: 10.1038/nature02567 · Source: PubMed

CITATIONS READS
683 1,820

5 authors, including:

Ana Christina Ravelo Mitchell Lyle

University of California, Santa Cruz Oregon State University
144 PUBLICATIONS   7,947 CITATIONS    1,838 PUBLICATIONS   12,362 CITATIONS   

SEE PROFILE SEE PROFILE

Michael W Wara
Stanford University
45 PUBLICATIONS   3,445 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

IODP Expedition 355 View project

Plio-Pleistocene tropical Pacific sea-surface temperatures View project

All content following this page was uploaded by Mitchell Lyle on 30 May 2014.

The user has requested enhancement of the downloaded file.

 articles

Regional climate shifts caused by

gradual global cooling in the
Pliocene epoch
Ana Christina Ravelo1, Dyke H. Andreasen2*, Mitchell Lyle3, Annette Olivarez Lyle3 & Michael W. Wara1
1
Ocean Sciences Department, and 2Earth Sciences Department, University of California, Santa Cruz, California 95064, USA
3
Center for Geophysical Investigation of the Shallow Subsurface, Boise State University, 1910 University Drive, Boise, Idaho 83725, USA
* Present address: Rutgers University, Institute of Marine and Coastal Sciences, 71 Dudley Road New Brunswick, New Jersey 08904, USA

...........................................................................................................................................................................................................................

The Earth’s climate has undergone a global transition over the past four million years, from warm conditions with global surface
temperatures about 3 8C warmer than today, smaller ice sheets and higher sea levels to the current cooler conditions. Tectonic
changes and their influence on ocean heat transport have been suggested as forcing factors for that transition, including the onset
of significant Northern Hemisphere glaciation ,2.75 million years ago, but the ultimate causes for the climatic changes are still
under debate. Here we compare climate records from high latitudes, subtropical regions and the tropics, indicating that the onset
of large glacial/interglacial cycles did not coincide with a specific climate reorganization event at lower latitudes. The regional
differences in the timing of cooling imply that global cooling was a gradual process, rather than the response to a single threshold
or episodic event as previously suggested. We also find that high-latitude climate sensitivity to variations in solar heating
increased gradually, culminating after cool tropical and subtropical upwelling conditions were established two million years ago.
Our results suggest that mean low-latitude climate conditions can significantly influence global climate feedbacks.

Global climate change of the past 4 million years (Myr) includes the intensification of NHG. Yet the global impact of these tectonic
end of the early Pliocene warm period (5–3 Myr ago) and significant events has not been adequately examined. Finally, bidirectional
intensification of Northern Hemisphere glaciation (NHG) high–low latitude interactions may explain important features of
,2.75 Myr ago (Fig. 1a). The amplitude of 104–106-year climate the transition. For example, intensification of NHG could have
oscillations increased as climate cooled. The past 4 Myr, unlike the resulted in cooler deepwaters (formed at high latitudes) and a
more recent past, can be studied to assess climate theories that: subsequent increase in deep ocean stratification. Increased stratifi-
involve climate components with relatively long timescales of cation may have caused the ventilated thermocline to shoal, allow-
response (for example, deep ocean, cryosphere), predict different ing cold water to upwell in tropical and subtropical regions, thereby
behaviour in warm versus cold conditions, and are best tested by altering global climate patterns8.
examining changes to average conditions that are large relative to To test hypotheses that explain the end of the warm period, we
the seasonal signal. In addition, well-understood changes in solar compare distant palaeoceanographic records to examine tropical–
heating (Milankovitch cycles) occur on long timescales, providing extratropical interactions. This analysis results in the fundamental
an excellent natural experiment to examine climate responses to conclusion that major long-term cooling steps in different regions
perturbations in the Earth’s radiative balance. (for example, intensification of NHG, reorganization of tropical
Relative to today, the Pliocene warm period was characterized by: circulation) did not all occur at the same time. Thus, regionally
,3 8C higher global surface temperatures, 10–20 m higher sea level, specific processes caused cooling phases at different times, and the
enhanced thermohaline circulation1,2, slightly reduced Antarctic ice end of the warm period was not forced by a single episodic event
sheets, emerging but small Northern Hemisphere ice coverage3, and whose effects propagated globally7,11. Rather, it must have been
slightly (30%) higher atmospheric carbon dioxide concentrations1,4. forced gradually. This conclusion is backed up by the analysis of
changes in the amplitude of high-latitude climate variability, which
A small decrease in carbon dioxide concentration could explain the
indicates that climate sensitivity also increased gradually.
cooling at the end of the warm period if coupled with positive
feedbacks, as suggested for the onset of significant Antarctic glacia-
tion5. However, whether these feedbacks primarily involved low- or High-latitude climate trends
high-latitude processes has been controversial. The oxygen isotopic (d18O) composition of benthic foraminifera
Although high-latitude feedbacks (for example, related to ocean reflects deepwater temperature and ice volume changes, both
heat transport or ice albedo6,7) may have accelerated cooling once indicative of high-latitude climate change. The d18O record
NHG began, the impact of glaciation on global-scale cooling still (Fig. 1a) indicates that during the Pliocene warm period, high-
needs to be explored. Alternatively, long-term reorganization of latitude climate was, on average, warmer than interglacial periods of
tropical conditions could have strongly influenced global climate, as the past 1 Myr. The onset of significant NHG, reflected by the first
occurs interannually with the El Niño–Southern Oscillation obliquity-related cycles with high d18O values, or the beginning of
phenomenon8,9. Even small changes in tropical temperature pat- the ‘41-kyr world’, occurred ,2.75 Myr ago (Fig. 1a). These intense
terns can profoundly affect extratropical conditions on geological glaciations evident in the d18O record caused pronounced ice-sheet
timescales10. Thus, low-latitude tectonic events (restriction of calving as documented by North Pacific13 (Fig. 2b) and North
Panamanian or Indonesian seaways) may have changed the distri- Atlantic14 ice-rafted debris records. Other evidence of major North-
bution of heat between basins7,11,12, causing reorganization of ern Hemisphere high-latitude climate reorganization includes the
climate patterns, the end of the warm period, and ultimately sudden increase in North Pacific surface-water stratification13
NATURE | VOL 429 | 20 MAY 2004 | www.nature.com/nature
©2004 Nature Publishing Group 263
articles
(Fig. 2c), and pronounced reductions in North Atlantic1,15 and in estimating sedimentation rate, because they are reproducible
Pacific16 thermohaline overturn rates. using independent dating methods (for example, high-resolution
correlations, biostratigraphy), and because the change 1.7 Myr ago is
Subtropical climate trends also evident in records (%CaCO3, %Corg, seasonality) (Fig. 2f, g, h)
Two major steps of climate reorganization are evident in many that do not depend on accurate estimates of sedimentation rate.
records of subtropical cooling at the end of the Pliocene warm Today, CaCO3-MAR on the California coast is greatest in offshore
period. Both cooling steps are well expressed in the best-studied
subtropical continental regions: those influenced by the Asian and
African (Fig. 2d) monsoon systems17–19. The first step (between 3.0
and 2.5 Myr ago) was coincident with the onset of significant NHG.
However, the second step (between 2.0 and 1.5 Myr ago) occurred
well after the onset of significant of NHG. Sea surface temperature
(SST) in the West African upwelling system also began to decrease
around 3.0 Myr ago (ref. 20) with the majority of the 10 8C-cooling
trend after the onset of significant NHG (Fig. 2e). In all, these
records indicate that at least the first step in subtropical climate
change may be related to NHG. However, the second step in
subtropical climate change occurred when high-latitude climate
was relatively stationary (Fig. 1a).
To characterize further the subtropical climate change in the
Pliocene, we generated new California margin records (Fig. 2f–i).
Ocean Drilling Program (ODP) Site 1014 (32850 0 N, 119859 0 W,
1165 m water depth) is located in a sensitive region between the
warm North Pacific subtropical gyre and the highly productive
cooler upwelling conditions of the California margin. Our records
of calcite mass accumulation rate (CaCO3-MAR) and a seasonality
proxy (see Methods) are used to monitor the evolution of the
upwelling system, and like records from other subtropical regions,
indicate that changes occurred in two steps over the last 4 Myr. The
first step, an increase in CaCO3-MAR just after 3.0 Myr ago, was not
accompanied by a change in seasonality. The second step, a
pronounced decrease in CaCO3-MAR and increase in seasonality,
occurred about 1.7 Myr ago. The first-order trends in CaCO3-MAR
are unlikely to be an artefact of changes in dissolution21 given the
shallow water depth and unchanging bottom water conditions of
this site16 (and preliminary measurements of alkenone-MAR, an
organic marker of coccolithophore productivity, are in agreement
with the CaCO3-MAR record). Nor are the trends related to errors

Figure 2 Comparison of climate evolution in different regions. a, Benthic foraminifera

d18O record from Fig. 1. b, Magnetic susceptibility record from Subarctic Pacific ODP Site
882 (508 N, 1688 E; water depth 3,244 m) reflects ice-rafted debris input to sediments13.
c, Mass accumulation rate (MAR) of biogenic opal from ODP Site 882 interpreted as
inversely related to surface stratification13. d, Terrigenous wt% at ODP Site 66418
(68 N, 238 W; water depth 3,806 m) indicates increased aridity at the onset of NHG and
further aridification just after ,2.0 Myr ago. Time interval when 41-kyr cycles dominate
variability is annotated. e, Alkenone-saturation-index-derived SST record from ODP Site
1084 (268 S, 138 E; water depth 1,992 m) reflects strength of West African margin
upwelling20. f, Carbonate (wt%) from ODP Site 1014. g, Organic carbon (wt%) from ODP
Site 1014. h, Carbon isotopic difference between O. universa and G. bulloides, two
Figure 1 Evolution of high-latitude climate and sensitivity (adapted from ref. 44). species of planktonic foraminifera that proliferate in different seasons from ODP Site
a, Benthic foraminifera d18O record of high-latitude climate change. b, Deconvolution of 1014, reflects surface water seasonality (red dots). i, MAR of biogenic carbonate from
the obliquity component of scaled oxygen isotope (orange curve) and of solar forcing ODP Site 1014 reflects the accumulation rate of carbonate in the sediments. j, Oxygen
(black curve) records (smoothed records are thick lines). The period of greatest high- isotope difference (measured on G. sacculifer) between eastern Pacific ODP Site 851 (ref.
latitude response to solar forcing is shaded in green. c, Gain record, a measure of climate 26) (38 N, 1118 W, water depth 3,761 m) and western Pacific ODP Site 806 (refs 49, 50)
sensitivity, is the ratio of two deconvoluted records. Note that the transition to the high- (08, 1598 E, water depth 2,521 m) reflects the development of a hydrographic gradient
amplitude d18O cycles of the ‘41-kyr world’ was not related to a change in sensitivity. across the tropical Pacific. k, Gain record, a measure of climate sensitivity (from Fig. 1).
264 ©2004 Nature Publishing Group NATURE | VOL 429 | 20 MAY 2004 | www.nature.com/nature
 articles
windstress-curl-driven upwelling regions, where well-stratified sur- development of strong Walker circulation was potentially related to
face waters22 and coccolithophore productivity develop seasonally. the second major transition in subtropical conditions (Fig. 2d, e, h, i).
Coccolithophore productivity is favoured in the subsurface chloro-
phyll maximum, where cool nutrient-rich waters from the under- Gradual global cooling
lying ventilated thermocline (the strong thermal gradient between In the following section, we integrate important aspects of palaeo-
warm surface waters and cool deep water) upwell into the photic climate observations from around the globe, and conclude that the
zone. Thus, the increase in CaCO3-MAR approximately coincident transition from the Pliocene warm period to the Pleistocene cold
with onset of significant NHG may reflect shoaling of the ventilated period cannot be explained by a single tectonic or threshold event,
thermocline or an increase in windstress curl at the edge of the but rather, was forced gradually. The onset of significant NHG
North Pacific gyre, without a notable change in surface seasonality. (,2.75 Myr ago) (including major changes in thermohaline circu-
These conditions allowed for the delivery of subsurface nutrient- lation and deep ocean density and nutrient stratification1,2,15,16),
rich waters into the photic zone to support coccolithophore occurred at the same time as subtropical regions began to cool,
production, while maintaining well-stratified conditions. 1.7 Myr confirming a strong linkage between high latitude and subtropical
ago, a further increase in wind-driven upwelling or shoaling of the regions17–20. However, the onset of significant NHG (Fig. 2b, c) was
thermocline, accompanied by decreased stratification at least season- not related to fundamental changes in tropical conditions. It
ally, as occurs today, would have caused total CaCO3 production to occurred when tropical conditions were stable and El Niño-like,
drop and seasonality to increase. This second major change on the about a million years after an initial, possibly tectonically driven
California margin and in other subtropical regions occurred when change due to a closing tropical seaway, and about a million years
long-term high-latitude climate was relatively stable (Fig. 2a–c). before the establishment of strong Walker circulation (Fig. 2j).
Thus, the rapid transition ,2.75 Myr ago that characterizes climate
Tropical climate trends evolution in high latitude and some subtropical regions, must have
The state of the tropics is intimately tied to subsurface thermocline strictly involved extratropical processes such as those involving ice
conditions. The source of thermocline waters is surface water albedo, thermohaline circulation, monsoon strength, or biogeo-
subducted at mid-latitudes23. Today, the thermocline is sufficiently chemical13 changes, and was potentially initiated by reaching a
shallow and cool at the eastern boundaries of the tropical and greenhouse gas35,36, ice-sheet substrate37, or tectonic38 threshold. A
subtropical Pacific and Atlantic Oceans that upwelling results in different threshold was reached just after 2.0 Myr ago that switched
cool SSTs in these regions, setting up west–east gradients in SST and the tropics and subtropics (Fig. 2i) into the modern mode of
pressure. In the tropical Pacific, the west–east gradients strengthen circulation with relatively strong Walker circulation and cool sub-
the easterly trade winds and reinforce cool upwelling in the east, tropical temperatures.
thereby further augmenting the temperature and pressure gradients. Although the asynchronicity of climate change steps attests to the
This positive air–sea feedback is required to maintain strong zonal, importance of regionally distinct processes to explain those steps,
or Walker, circulation. Small perturbations that weaken Walker they do not preclude the existence of important feedbacks between
circulation are amplified by these same air–sea feedbacks, causing distant regions. For example, the observed increase in wind-driven
the thermocline to deepen in the east and El Niño conditions to subtropical upwelling could be related to the increase in the
develop24. Teleconnnections then cause climate anomalies in the latitudinal temperature gradient39 that occurred first with high-
tropical regions to influence climate patterns on a global scale. latitude cooling 2.75 Myr ago, and then with the enhancement of the
Although limited data compilations indicate that the tropics were western tropical warm pool just after 2.0 Myr ago. However, recent
probably not, on average, significantly warmer than today25 during modelling results40 predict that high-latitude cooling should have
the early Pliocene, there is evidence that the pattern of tropical caused the subtropical winds to weaken. This apparent contradic-
conditions resembled a permanent El Niño: the thermocline was tion could indicate that subtropical upwelling records reflect
deep, and the west–east SST gradient was greatly reduced compared changes either in the position, rather than the strength, of the
with modern normal conditions26–28. Furthermore, long-term subtropical gyre or in thermocline conditions, rather than changes
extratropical climate patterns in the early Pliocene were similar to in wind strength. Thus, the ocean’s thermocline, rather than the
those manifested during a modern El Niño12. Mid-latitude surface atmosphere, may have served as a link between distant locations. A
waters, which are subducted into the thermocline thereby control- recent study emphasizes that global cooling of the past 3 Myr was
ling its character23,29, were several degrees warmer than today in the related to changes in the depth of the ventilated thermocline8. The
Pliocene warm period26. With the thermocline either deeper or study asserts that as high-latitude deepwater formation regions
warmer than today, wind-driven upwelling would not have cooled cooled, deepwater temperature decreased, ocean stratification
the eastern tropical Pacific sufficiently to maintain strong Walker increased, and the thermocline shoaled. This caused tropical and
circulation. Thus, mid-latitude warmth, and tropical–extratropical subtropical upwelling regions to cool, invigorating west–east tem-
coupling via the thermocline, could explain the weak Walker perature and pressure gradients and reinforcing the cooling trend.
circulation of the Pliocene warm period. The observed changes in Pacific subtropical and tropical upwelling
The possible effects of tropical processes on extratropical climate regions ,2.0 Myr ago suggests that distant locations are being
events (for example, NHG) can be assessed by considering the influenced by a shoaling thermocline and supports the hypothesis.
timing of tropical climate changes and the development of strong The hypothesis is also supported by evidence that during the warm
Walker circulation. Tropical records indicate that significant tropi- period relative to today, deep ocean temperature was high41 and
cal climate reorganization occurred twice: once between 4.5 and ocean stratification was reduced16 at the same time that the
4.0 Myr ago, well before significant NHG, and once between 2.0 and thermocline was deep26. However, further testing of the idea using
1.5 Myr ago, well after the onset of significant NHG. Between 4.5 the climate change of the past 4 Myr awaits the generation of more
and 4.0 Myr ago, a marked shift in surface-water hydrographic data on the detailed history and global nature of changes in deep
gradients between the Pacific and Caribbean30 and between the ocean stratification, the depth of the ventilated thermocline, and the
Caribbean and the western tropical Atlantic31, shoaling of the temperature of upwelling regions.
thermocline in the east Pacific28, and circulation changes in Although the changes in the ventilated thermocline8 and the
the Atlantic32–34, were possibly forced by tectonic event(s) (for temperature gradient39 each describe possible linkages between
example, restriction of the Panamanian30 or Indonesian11 seaways). distant locations and positive feedbacks to cooling, they provide
However, the SST gradient across the Pacific (Fig. 2i) did not no explanation of the ultimate cause of global cooling. Certainly, the
increase until between 2.0 and 1.5 Myr ago26,27, indicating that the fact that the global cooling transition occurred with regionally
NATURE | VOL 429 | 20 MAY 2004 | www.nature.com/nature
©2004 Nature Publishing Group 265
articles
distinct timing suggests that it was not caused by a single event. A the maintenance of Walker circulation and subtropical land–sea
single event could not have acted as a ‘trigger’ that initiated the pressure gradients, amplified small perturbations to the radiative
transition, as it is unlikely that subsequent interactions between forcing resulting from changes in Earth’s tilt. This idea is consistent
climate system components over a million years or so could with the low-amplitude tropical climate cycles, and their lack of
have by themselves ‘fortuitously’ led to global cooling. Thus, coherency to solar forcing, before the establishment of Walker
gradual changes in atmospheric greenhouse gas composition, circulation27. It is also consistent with the observation that obli-
basin geometry, or land-surface conditions, occurring at least over quity-related cycles dominated African monsoon variability after
the past ,4–1.5 Myr, most probably forced global cooling and need 1.8 Myr ago19 (Fig. 2d). To test our conclusions, a quantitative
further investigation. assessment of changes in low-latitude climate sensitivity to solar
forcing is needed once long high-resolution records from tropical
Climate sensitivity and subtropical regions are generated. Furthermore, since about
Glacial–interglacial cycles have been a striking feature of climate 35% of the variance in the d18O record before 1 Myr ago is not
change for the past 4 Myr (Fig. 1a). What controls the frequency of directly related to orbital solar forcing, a thorough evaluation of
these cycles? Variability of seasonal solar heating at all latitudes is ‘sensitivity’ should also focus on the source of higher-frequency
predominantly controlled by precession of the Earth’s position on variations.
its orbit during summer, with cycles of 19–23 kyr and amplitudes The observation that average lower-latitude conditions influ-
often over 100 Wm22. However, only 12% of the variance in the enced high-latitude climate sensitivity has implications on hypoth-
d18O record, before 1 Myr ago, is in the precession band, whereas eses that explain how obliquity solar cycles might drive climate
43% is in the obliquity band (cycles of ,41 kyr). Thus, variations in change. One hypothesis8, generally supported by our observations,
solar heating due to changes in the Earth’s angle of tilt, or obliquity, predicts that thermocline temperature was influenced by annual
must be an important driving force of glacial cycles before the Late average heating (controlled by obliquity) at mid-latitudes where
Pleistocene period8,42. In fact, although precession controls seasonal thermocline waters are subducted. Changes in thermocline tem-
heating, obliquity controls over 85% of the variance in the annual perature influenced SST in tropical upwelling regions, which
average solar heating at all latitudes and in the high-to-low latitu- influenced high-latitude climate through teleconnections. The
dinal gradient of solar heating. Clearly, the frequency of glacial climate response to solar forcing was weak in the Pliocene warm
cycles was in large part controlled by obliquity-related solar vari- period because the thermocline was too deep to influence SST in
ations; but why did the amplitude of glacial cycles change through upwelling regions, regardless of obliquity-forced perturbations.
time? Comparing the non-stationary d18O record to the long-term However, the sensitivity (Fig. 2k) increased as the long-term average
regional trends of the past 4 Myr can address this question. conditions of the thermocline shoaled or cooled gradually. A
Because the onset of significant NHG and increased amplitude of different theory42 suggests that low-to-high latitude gradient in
obliquity-related d18O variations (the beginning of the ‘41-kyr solar forcing, coupled with strong ice-albedo feedbacks, controlled
world’) both occurred ,2.75 Myr ago (Fig. 1a), it seems that obliquity-related climate change. However, our data indicate that if
there was a causal relationship between average high-latitude the gradient in solar forcing controlled glacial cycles in the late
conditions and the amplitude of glacial–interglacial cycles. How- Pliocene, it may have instead done so with the help of tropical or
ever, an evaluation of changes in the climate’s response to solar subtropical processes.
forcing must account for how the forcing itself changed. In fact, the
amplitude of obliquity solar cycles was not constant; it was modu- Implications for understanding climate change
lated by interactions with eccentricity cycles of the Earth’s orbit43. Several lessons can be drawn from the comparison of Plio-Pleisto-
The amplitude modulation of obliquity cycles in both the forcing cene climate change records from distant locations. First, although
(solar) and response (d18O) records was quantified using complex changes in forcing were gradual, strong regional nonlinear
demodulation performed on the filtered 41-kyr components of responses generated pronounced regional climate changes includ-
these records44 (Fig. 1b). The gain (the ratio of the two demodulated ing the onset of significant NHG. Second, the ventilated thermo-
records) (Fig. 1c) reflects the amplitude of the d18O response relative cline and/or latitudinal temperature gradient may have played an
to solar forcing, thereby providing a measure of high-latitude important role in linking subtropical conditions to change in other
climate sensitivity. The increase in the amplitude of d18O variability regions. Finally, tropical and subtropical conditions, specifically the
between 3.0 and 2.5 Myr ago (Fig. 1b), or the beginning of the ‘41- time-averaged strength of coldwater upwelling in the eastern Pacific,
kyr world’, was a direct response to the increasing amplitude of solar and of Walker circulation, had a strong influence on the climate
forcing, and therefore cannot be directly attributed to the onset of response to radiative changes. Thus, the last 4 Myr illustrates that as
significant NHG. Climate sensitivity increased gradually after globally average conditions change, so do the feedbacks or ‘rules’
4.0 Myr ago, culminating in a period of highest sensitivity after that determine climate sensitivity. This conclusion is relevant to
,2.0 Myr ago. A calculation of the sensitivity, or gain, using the studies of future global warming because it emphasizes the import-
precessional components of the d18O and solar-forcing records ance of ‘background’ or average tropical conditions in predicting
provides the same result. high-frequency climate change. Furthermore, understanding pro-
Comparison of the sensitivity record (Fig. 2k) to changes in the cesses responsible for recent climate change of the last hundreds or
mean state of climate in different regions leads us to several thousands of years, when average background conditions changed
conclusions. Sensitivity gradually increased and then reached its very little, is unlikely to be sufficient to predict climate variability for
highest level during the ‘41-kyr world’ (Figs 1, 2k), even while average periods with different globally averaged conditions. This highlights
high-latitude conditions remained relatively stable (Fig. 2a–c). the importance of developing theory to explain ocean and atmos-
Thus, the strength of feedbacks that increasingly amplified solar pheric change, and testing that theory using records from geologic
cycles after 4.0 Myr ago was probably independent of long-term time periods that represent a large dynamic range of climate
high-latitude conditions such as ice-sheet size or deepwater for- conditions. A
mation strength. In contrast, the approximate temporal correlation
between the enhancement of Walker circulation, seasonal subtro-
pical upwelling, and climate sensitivity, just after ,2.0 Myr ago, Methods
Measurements of d18O on benthic foraminifera (Cibicides mckennai) were made in the
suggests that important processes that amplify obliquity-related stable isotope facility at the University of California, Santa Cruz on samples from ODP Site
solar forcing may reside in tropical and/or subtropical regions. As 1014, with an approximate average resolution of ,3–4 kyr (data not shown). The age of
upwelling regions cooled, potent air–sea feedbacks associated with the sediment, from 0–3.0 Myr ago, was determined by correlating this new d18O record to

266 ©2004 Nature Publishing Group NATURE | VOL 429 | 20 MAY 2004 | www.nature.com/nature
 articles
the d18O record from ODP Site 846 (ref. 45). Because there were core gaps in the circulation and thermocline depth. Paleoceanography 12, 805–820 (1997).
sedimentary section older than 1.0 Myr ago, the age model was further refined by tuning 27. Wara, M. On the Interaction Between Tropical and High Latitude Climate Over the Past 5 Million Years.
magnetic susceptibility46 (measured with the down-hole Geologic High Resolution Thesis, Univ. California, Santa Cruz (2003).
Magnetic Tool on Hole 1014A after drilling) to obliquity43. Measurements of weight per 28. Chaisson, W. & Ravelo, A. C. Pliocene development of the East-West hydrographic gradient in the
cent of calcium carbonate (%CaCO3) (Fig. 2f) and organic carbon (%Corg) (Fig. 2g) of Equatorial Pacific. Paleoceanography 15, 497–505 (2000).
ODP Site 1014 samples were made at Boise State University using a UIC coulometer 29. Gu, D. & Philander, G. S. Interdecadal climate fluctuations that depend on exchanges between the
attached to a modified version of the CM-5120 furnace module, as described in a previous tropics and extratropics. Science 275, 805–807 (1997).
study47. The CaCO3 component at this site is made primarily of coccoliths46. The 30. Haug, G. H. et al. Role of Panama uplift on oceanic freshwater balance. Geology 29, 207–210 (2001).
‘seasonality’ (Fig. 2h) is the mathematical difference between carbon isotopic values 31. Billups, K. et al. A link between ocean heat transport, thermohaline circulation and Intertropical
(Dd13C) measured on two species of planktonic foraminifera, Orbulina universa and Convergence Zone in the early Pliocene Atlantic. Geology 287, 319–322 (1999).
Globigerina bulloides , that favour different seasonal conditions48. The record of 32. Chaisson, W. & Ravelo, A. C. Changes in upper water-column structure at Site 925, Late Miocene–
CaCO3-MAR (Fig. 2i) was calculated using %CaCO3, sedimentation rate, and ODP Site Pleistocene: Planktonic foraminifer assemblage and isotopic evidence. Proc. Ocean Drill. Prog. Sci. Res.
1014 shipboard dry bulk density measurements46. All ODP Site 1014 data will be archived 154, 255–268 (1997).
on the NDGC-NOAA palaeoclimate database website: www.ngdc.noaa.gov/paleo/ 33. Norris, R. D. Miocene–Pliocene surface-water hydrography of the eastern equatorial Atlantic. Proc.
paleo.html. Ocean Drill. Prog. Sci. Res. 159, 539–555 (1998).
34. Billups, K., Ravelo, A. C. & Zachos, J. C. Early Pliocene climate: A perspective from the western
Received 6 November 2003; accepted 7 April 2004; doi:10.1038/nature02567. equatorial Atlantic warm pool. Paleoceanography 13, 459–470 (1998).
1. Raymo, M. E., Grant, B., Horowitz, M. & Rau, G. H. Mid-Pliocene warmth: Stronger greenhouse and 35. Saltzman, B. & Verbitsky, M. Y. Multiple instabilities and modes of glacial rhythmicity in the Plio-
stronger conveyer. Mar. Micropaleo. 27, 313–326 (1996). Pleistocene: a general theory of late Cenozoic climatic change. Clim. Dyn. 9, 1–15 (1993).
2. Ravelo, A. C. & Andreasen, D. H. Enhanced circulation during a warm period. Geophys. Res. Lett. 27, 36. Raymo, M. E. Glacial puzzles. Science 281, 1467–1468 (1998).
1001–1004 (2000). 37. Clark, P. U., Alley, R. B. & Pollard, D. Northern hemisphere ice-sheet influences on global climate
3. Haywood, A. M., Valdes, P. J. & Sellwood, B. W. Global scale palaeoclimate reconstruction of the change. Science 286, 1104–1111 (1999).
middle Pliocene climate using the UKMO GCM: initial results. Glob. Planet. Change 25, 239–256 38. Ruddiman, W. F. & Raymo, M. E. Northern hemisphere climate régimes during the past 3 Ma: possible
(2000). tectonic connections. Phil. Trans. R. Soc. Lond. B 318, 411–430 (1988).
4. Van der Burgh, J., Visscher, H., Dilcher, D. L. & Kürschner, W. M. Paleoatmospheric signatures in 39. Rind, D. Relating paleoclimate data and past temperature gradients: Some suggestive rules. Quat. Sci.
Neogene fossil leaves. Science 260, 1788–1790 (1993). Rev. 19, 381–390 (2000).
5. DeConto, R. M. & Pollard, D. Rapid Cenozoic glaciation of Antarctica induced by declining 40. Haywood, A. M., Valdes, P. J. & Sellwood, B. W. Global scale palaeoclimate reconstruction of the
atmospheric CO2. Nature 421, 245–249 (2003). middle Pliocene climate using the UKMO GCM: initial results. Glob. Planet. Change 25, 239–256
6. Raymo, M. E., Rind, D. & Ruddiman, W. F. Climatic effects of reduced Arctic sea ice limits in the GISS (2000).
II general circulation model. Paleoceanography 5, 367–382 (1990). 41. Lear, C. H., Rosenthal, Y. & Wright, J. D. The closing of a seaway: ocean water masses and global
7. Driscoll, N. W. & Haug, G. H. A short circuit in thermohaline circulation; a cause for Northern climate change. Earth Planet. Sci. Lett. 210, 425–436 (2003).
Hemisphere glaciation? Science 282, 436–438 (1998). 42. Raymo, M. E. & Nisancioglu, K. The 41 kyr world: Milankovitch’s other unsolved mystery.
8. Cane, M. A. & Evans, M. Do the tropics rule? Science 290, 1107–1108 (2000). Paleoceanography 18, 1011 (2003).
9. Philander, S. G. & Fedorov, A. V. Role of tropics in changing the response to Milankovitch forcing 43. Laskar, J., Joutel, F. & Boudin, F. Orbital, precessional, and insolation quantities for the Earth from
some three million years ago. Paleoceanography 18, 23-1–23-11 (2003). 220 Myr to þ10 Myr. Astron. Astrophys. 270, 522–533 (1993).
10. Yin, J. H. & Battisti, D. S. The importance of tropical sea surface temperature patterns in simulations 44. Andreasen, D. H. Isolating Sources of Plio-Pleistocene Paleoclimate Variability. Thesis, Univ. California,
of last glacial maximum climate. J. Clim. 14, 565–581 (2001). Santa Cruz (2001).
11. Cane, M. A. & Molnar, P. Closing of the Indonesian Seaway as a precursor to East African aridification 45. Shackleton, N. J., Crowhurst, S., Hagelberg, T., Pisias, N. G. & Schneider, D. A. A new Late Neogene
around 3–4 million years ago. Nature 411, 157–162 (2001). time scale: application to Leg 138 Sites. Proc. Ocean Drill. Prog. Sci. Res. 138, 73–101 (1995).
12. Molnar, P. & Cane, M. A. El Niño’s tropical climate and teleconnections as a blueprint for pre-Ice Age 46. Lyle, M. et al. Site 1014. Proc. Ocean Drill. Prog. Init. Rep. 167, 175–222 (1997).
climates. Paleoceanography 17, 11–11 (2002). 47. Lyle, M. et al. Millennial-scale CaCO3 and Corg events along the northern and central California
13. Haug, G. H., Sigman, D. M., Tiedemann, R., Pedersen, T. F. & Sarnthein, M. Onset of permanent margin: Stratigraphy and origins. Proc. Ocean Drill. Prog. Sci. Res. 167, 163–182 (2000).
stratification in the subarctic Pacific Ocean. Nature 401, 779–782 (1999). 48. Sautter, L. R. & Thunell, R. C. Seasonal variability in the 18O and 13C of planktonic foraminifera from
14. Ruddiman, W. F. & Raymo, M. E. Northern Hemisphere climate regimes during the past 3 Ma: an upwelling environment: sediment trap results from the San Pedro Basin, Southern California
possible tectonic connections. Phil. Trans. R. Soc. Lond. B 318, 411–430 (1988). Bight. Paleoceanography 6, 307–334 (1991).
15. Raymo, M. E., Hodell, D. & Jansen, E. Response of deep ocean circulation to initiation of northern 49. Berger, W. H., Bickert, T., Schmidt, H. & Wefer, G. Quaternary oxygen isotope record of pelagic
hemisphere circulation (3–2 Ma). Paleoceanography 7, 645–672 (1992). foraminifers; Site 806, Ontong Java Plateau. Proc. Ocean Drill. Prog. Sci. Res. 130, 381–395 (1993).
16. Kwiek, P. & Ravelo, A. C. Pacific Ocean intermediate and deep water circulation during the Pliocene. 50. Jansen, E., Mayer, L. A., Backman, J., Leckie, R. M. & Takayama, T. Evolution of Pliocene climate
Palaeogeogr. Palaeoclimatol. Palaeoecol. 154, 191–217 (1999). cyclicity at Hole 806B (5–2 Ma); oxygen isotope record. Proc. Ocean Drill. Prog. Sci. Res. 130, 349–362
17. Clemens, S., Murray, D. W. & Prell, W. L. Nonstationary phase of the Plio-Pleistocene Asian monsoon. (1993).
Science 274, 943–948 (1996).
18. DeMenocal, P. B. Plio-Pleistocene african climate. Science 270, 53–59 (1995). Acknowledgements We thank L. Anderson, M. Cane, B. Chaisson, P. Dekens, G. Haug,
19. Tiedemann, R., Sarnthein, M. & Shackleton, N. J. Astronomic timescale for the Pliocene Atlantic d18O K. Lawrence, P. Molnar and G. Philander, for suggestions that influenced this work. M. Flower,
and dust flux records of Ocean Drilling Program site 659. Paleoceanography 9, 619–638 (1994). Z. Irons and K. Ziegler helped with the data generation. G. Haug and P. deMenocal supplied their
20. Marlow, J. R., Lange, C. B., Wefer, G. & Rosell-Melé, A. Upwelling intensification as part of the Plio- data for Fig. 2. The National Science Foundation funded this work. Samples were provided by the
Pleistocene climate transition. Science 290, 2288–2294 (2000). Ocean Drilling Program.
21. Ravelo, A. C. et al. Pliocene carbonate accumulation along the California Margin. Paleoceanography
12, 729–741 (1997). Authors’ contributions D.H.A. conceived of the ideas presented in Fig. 1. M.L. and A.O.L. were
22. Okada, H. & Honjo, S. The distribution of oceanic coccolithophorids in the Pacific. Deep-Sea Res. 20, responsible for carbonate and organic carbon analyses. All authors contributed scientific ideas.
355–374 (1973).
A.C.R. was primarily responsible for generating isotope data on ODP Site 1014 and constructing
23. Harper, S. Thermocline ventilation and pathways of tropical-subtropical water mass exchange. Tellus
age model, formulating the main conclusions, and writing the paper.
A 52, 330–345 (2000).
24. Bjerknes, J. Atmospheric teleconnections from the tropical Pacific. Mon. Weath. Rev. 97, 163–172
(1969). Competing interests statement The authors declare that they have no competing financial
25. Dowsett, H., Barron, J. & Poore, R. Z. Middle Pliocene sea surface temperatures: A global interests.
reconstruction. Mar. Micropaleo. 27, 13–25 (1996).
26. Cannariato, K. & Ravelo, A. C. Plio-Pleistocene evolution of eastern tropical Pacific surface water Correspondence and requests for materials should be addressed to A.C.R. ([email protected]).

NATURE | VOL 429 | 20 MAY 2004 | www.nature.com/nature

©2004 Nature Publishing Group 267
View publication stats