Publisher: NPG; Journal: Nature: Nature; Article Type: Physics article
DOI: 10.1038/nature12003
Page 1 of 31
Patterns and mechanisms of early Pliocene warmth
A. V. Fedorov
1
*, C. M. Brierley
1,2
*, K. T. Lawrence
3
*, Z. Liu
4
, P. S. Dekens
5
& A. C. Ravelo
6
1
Department of Geology and Geophysics, Yale University, New Haven, Connecticut 06511, USA.
2
Department of Geography, University College London, London, WC1E 6BT, UK.
3
Department of Geology and Environmental Geosciences, Lafayette College, Easton, Pennsylvania 18042,
USA.
4
Department of Earth Sciences, The University of Hong Kong, Hong Kong.
5
Department of Geosciences, San Francisco State University, San Francisco, California 94132.
6
Department of Ocean Sciences, University of California, Santa Cruz, California 95064, USA.
*These authors contributed equally to this work.
About five to four million years ago, in the early Pliocene epoch, Earth had a warm,
temperate climate. The gradual cooling that followed led to the establishment of
modern temperature patterns, possibly in response to a decrease in atmospheric CO
2
concentration, on the order of 100 parts per million, towards preindustrial values.
Here, we synthesize the available geochemical proxy records of sea surface temperature
and show that, compared with that of today, the early Pliocene climate had
substantially lower meridional and zonal temperature gradients, but quite similar
maximum ocean temperatures Using an Earth system model, we show that none of the
mechanisms currently proposed to explain Pliocene warmth can simultaneously
reproduce all three crucial features. We suggest that a combination of several
dynamical feedbacks, such as those related to ocean mixing and cloud albedo which are
underestimated in the models at present, , was responsible for these climate conditions.
Earth’s climate evolution over the last five million years (Myr), since the early Pliocene
epoch, has been meticulously studied by scientists (see, for example, refs 1–3). For years, the
focus of attention was on the origin
4–6
of glacial cycles, that is, the coming and going of
continental ice sheets evident in the δ
18
O record reflecting global ice volume and deep-ocean
temperature (Fig. 1a). However, a wealth of new data now show that the gradual onset
(around 3 Myr ago) and further amplification of Northern Hemisphere glaciation (Fig. 1b)
Publisher: NPG; Journal: Nature: Nature; Article Type: Physics article
DOI: 10.1038/nature12003
Page 2 of 31
was only one facet of this climate change and was a consequence of the global cooling rather
than its initial cause
7,8
.
The early Pliocene itself, the warm interval that preceded the glaciation, has attracted
a lot of attention as a possible analogue for future climate conditions. Despite relatively small
differences in climate control factors, including CO
2
concentrations (Fig. 1c), between the
early Pliocene and the present, the former was markedly different
9
. Palaeorecords indicate
vast changes in climate patterns since 5–4 Myr ago, including a contraction of the tropical
belt and oceanic warm pool
10
, emergence of strong temperature gradients along the
Equator
11
, cooling of coastal upwelling zones in the subtropics
12
, the shoaling of the ocean
thermocline
11,13
and cooling of the high-latitude and deep ocean
14,15
. Together these
observations imply a large structural change in climate, with major global and regional
implications.
By a structural change, we mean a transition from a climate with almost no zonal sea
surface temperature (SST) gradients (~1 °C or less) and a weak meridional SST gradient to a
modern climate with much more pronounced spatial temperature contrasts. This structural
climate change was paralleled by relatively small changes in atmospheric CO
2
concentration
(Fig. 1c), on the order of ~100 p.p.m. Thus, a key step towards understanding early Pliocene
climate, and possibly other warm climates throughout Earth history, is to explain what could
cause such changes.
The objectives of this article are twofold. First, we review recent geochemically
derived palaeorecords emphasizing the three criteria that any hypothesis must satisfy to
account for the climate of the early Pliocene (reduced meridional temperature gradient, weak
zonal gradients and SST stability). Then we evaluate various explanations for early Pliocene
climate with respect to these criteria, using a single modelling framework.
Observations
Substantial effort has been put into deriving palaeoclimate records for the Pliocene because it
is recognized as the most recent example of prolonged global warmth in the geological
past
2,16
. As part of the PRISM project, this effort produced a series of global temperature
reconstructions
17,18
for a ‘time slab’ of the mid Pliocene (3–3.3 Myr ago). These data sets and
Publisher: NPG; Journal: Nature: Nature; Article Type: Physics article
DOI: 10.1038/nature12003
Page 3 of 31
associated modelling studies enable a coordinated comparison between the Pliocene and the
present, and dramatically increased our understanding of Pliocene climate.
In the present study, we compile available SST proxy records, most of them
continuous, to describe the long-term climate trends from the warm early Pliocene to the
present (Fig. 2 and Methods). As several of these records indicate, the period from 4.4 to
4 Myr ago was probably the warmest interval within this timeframe—the Pliocene climatic
optimum (Fig. 2b, c). The subsequent cooling became evident roughly at the same time in
both hemispheres (Fig. 2c, d) and involved regions ranging from low-latitude upwelling
zones to mid and high latitudes. The expansion of the Northern Hemisphere ice sheets around
2.7 Myr ago, evidenced by an increase in the magnetic susceptibility of sediments affected
by ice-rafted debris (Fig. 1b), introduced a strong interhemispheric asymmetry in the climate
evolution over land
19
but to a lesser degree over the ocean (as shown by the similarity of
trends in Fig. 3c, d).
Within the global cooling pattern, local trends featured significant spatial and
temporal variations not necessarily tracking the growth of continental ice sheets inferred
from δ
18
O data (Fig. 1a). Regionally, temperature changes were as high as 11 °C over the
ocean
20
(Fig. 3c) and 19 °C over the land
19
. Below, we focus on the critical features that
characterize the evolution of climate structure since the early Pliocene. As is traditionally
done by the palaeoclimatology community, here we compare the mean trends incorporating
both glacial and interglacial intervals of the time series, even though some of our statements
contrast the late Quaternary interglacials and Pliocene interglacials. A comparison between
the mean and ‘interglacial’ trends supports our main conclusions (Methods and
Supplementary Figs 1–3).
Warm pool temperatures and CO
2
The first notable aspect of the Pliocene–Pleistocene climate evolution is the stability of
warmest SSTs in the tropical warm pool over the past 5 Myr. In all three tropical oceans,
these temperatures stayed fairly constant and the long-term average remained ~29 °C (Figs
2a and 3a), similar to the present. This is especially surprising given temperature changes
elsewhere. It has been suggested that Mg/Ca-based measurements, providing many warm
pool records, should be corrected for secular changes in seawater composition
21
. However,
Publisher: NPG; Journal: Nature: Nature; Article Type: Physics article
DOI: 10.1038/nature12003
Page 4 of 31
general agreement between Mg/Ca and alkenone data in locations where both proxies are
available (ref. 12 and Methods) indicates that this correction should be small, comparable to
calibration errors
12
.
Although a number of thermostat mechanisms capable of maintaining the stability of
warm pool SST in high-CO
2
climates have been proposed
22
, we find them unnecessary to
explain the Pliocene climate. A pervasive increase in atmospheric CO
2
is the driver for
present anthropogenic climate change, and is expected to cause even larger changes in the
future
16
. Similarly, CO
2
remains a suspected cause for early Pliocene warmth, and several
biogeochemical methods have been devised to estimate its concentration (Fig. 1c). Despite
large uncertainties, together these proxy data suggest that Pliocene CO
2
concentrations were
only ~100 p.p.m. higher than preindustrial values (~280 p.p.m). Climate models produce a
~1 °C temperature rise in the warm pool when the CO
2
concentration increases by
100 p.p.m.
16
, and a change by ~2 °C in doubling-of-CO
2
experiments
23
. Thus, within data
uncertainty (Methods), the stability of the warm pool temperatures is not inconsistent with
the relatively small CO
2
change.
Increasing meridional temperature gradients
High-latitude warmth and a reduced Equator-to-pole temperature gradient are other dominant
features of early Pliocene climate
2,24
. Marine data, available between latitudes 43° S and
58° N (Fig. 2e), suggest subsequent ocean cooling of 4–7 °C in the mid and high latitudes of
the Atlantic and Pacific (Figs 2d and 3d). The temperature of Atlantic deep waters (ODP
site 607; 3,400-m depth) basically follows SST evolution in the North Atlantic, implying a
similar cooling over the ocean convection regions at higher latitudes (~70° N). In terrestrial
regions of the Arctic, temperature fell by nearly 19 °C (ref. 19).
The meridional SST gradient as measured from the Equator to the subtropics was also
significantly smaller in the early Pliocene than at present
10
(Fig. 4a–c), and the meridional
temperature distribution within the Tropics was more uniform
10
. Consequently, despite little
difference in the warmest SSTs (Figs 2a and 3a), the meridional extent of the warm pool was
much broader in the early Pliocene. The subsequent cooling led to a gradual contraction of
the warm pool towards the Equator, as evidenced by the increase in meridional temperature
gradient. This contraction is also apparent from temperature records at the edge of the warm
Publisher: NPG; Journal: Nature: Nature; Article Type: Physics article
DOI: 10.1038/nature12003
Page 5 of 31
pool in the South China Sea
25,26
(Supplementary Fig. 9). Whereas data from inside the warm
pool show fairly constant temperatures, these particular records show a clear cooling trend.
The reduced meridional SST gradient seems to be important for understanding the
high global mean temperature and other characteristics of the early Pliocene
8,10
. For instance,
from the atmospheric perspective, as the temperature contrast from the Equator to the
subtropics increased, the surface high-pressure zones in low latitudes (subtropical highs)
strengthened, whereas the atmospheric meridional circulation (the Hadley cells) intensified
and contracted slightly towards the Equator. The strengthening of atmospheric circulation led
to a stronger subsidence, resulting in aridification of parts of Africa, Australia and North
America
8,27
.
From the tropical oceanic perspective, as the meridional SST gradient increased, so
did upper ocean stratification, because oceanic vertical thermal structure is directly related to
the surface temperature distribution
28,29
. This increase in stratification should have
contributed to the basin-wide shoaling of the tropical thermocline
9,30,31
, which has been under
way since 5 Myr ago
13,31
.
Strengthening of cold upwelling
As the meridional SST gradient increased and the tropical thermocline shoaled, cold waters
became present at low latitudes, with dramatic consequences for the tropics. Initially the cold
waters appeared in the coastal subtropical upwelling sites in the Atlantic and Pacific oceans,
in both hemispheres (Fig. 2c and 3c). The strengthening of anticyclonic winds within the
subtropical highs may have contributed to the stronger upwelling
8,10
.
The shoaling of the thermocline culminated in the appearance of colder water along
the Equator (Figs 2b and 3b) and the formation of a salient feature of the present-day SST
pattern, namely the equatorial ‘cold tongues’ on the eastern sides of the Pacific and Atlantic
basins. The corresponding development of the zonal temperature gradients along the Equator
(Fig. 4d–f) and the ensuing Bjerknes feedback (a stronger SST gradient leads to stronger
winds and, thence, a stronger gradient) led to the intensification of the zonal atmospheric
circulation—the Walker cell
3,10
.
In the present climate, an intermittent weakening of both the winds and the east–west
equatorial temperature gradient occurs during El Niño, but the mean climate is still
