%50=,9:0;@6-,)9(:2(05*635%50=,9:0;@6-,)9(:2(05*635
0.0;(364465:%50=,9:0;@6-,)9(:2(05*6350.0;(364465:%50=,9:0;@6-,)9(:2(05*635
!(7,9:05;/,(9;/(5+;46:7/,90*#*0,5*,:
(9;/(5+;46:7/,90*#*0,5*,:,7(9;4,5;
6-
)308<0;@7(*,+!306*,5,',:;5;(9*;0*0*,:/,,;6:*033(;065: )308<0;@7(*,+!306*,5,',:;5;(9*;0*0*,:/,,;6:*033(;065:
$(0:/
&0*;690(%50=,9:0;@6-',3305.;65
"!6>,33
69;/,95330560:%50=,9:0;@
96::.,6350<,+<
",=@
%50=,9:0;@6-,)9(:2(05*635
'03:65
%50=,9:0;@6- ;(.6
"#*/,9,9
69;/,95330560:%50=,9:0;@
#,,5,?;7(.,-69(++0;065(3(<;/69:
6336>;/0:(5+(++0;065(3>692:(;/;;7:+0.0;(3*64465:<53,+<.,6:*0,5*,-(*7<)
!(9;6-;/,(9;/#*0,5*,:64465:
(0:/$!6>,33",=@"'03:65#*/,9,9"$(3(90*690::,20,::,5!6470306
'03:65$(9;,9,65;6"<@),9:!*(@"!633(9+"6::'05;,9(99,;;!
96>5,6+@"6>(59(47;65<5)(9<5)(936905+6,)/(9+;9(/(4
(55(/(5:(9(1(9>66+(=0+,3305.,59@:#0556=</5@3,!(B<-,9
(C630!(.,5:(5+,95(*2*5;6:/'033(56905" /5,0:,9!(<3:,5$
!,9:0*6"(05,",,+"0,::,34(5#(.56;;0#*/40;;#1<55,:26.#;965.!$(=0(50
&6.,3#'03*/$(5+'0330(4:$ )308<0;@7(*,+!306*,5,',:;5;(9*;0*0*,:/,,;6:*033(;065:
!(7,9:05;/,(9;/(5+;46:7/,90*#*0,5*,:
/;;7:+0.0;(3*64465:<53,+<.,6:*0,5*,-(*7<)
$/0:9;0*3,0:)96<./;;6@6<-69-9,,(5+67,5(**,::)@;/,(9;/(5+;46:7/,90*#*0,5*,:,7(9;4,5;6-(;
0.0;(364465:%50=,9:0;@6-,)9(:2(05*635;/(:),,5(**,7;,+-6905*3<:06505!(7,9:05;/,(9;/(5+
;46:7/,90*#*0,5*,:)@(5(<;/690A,+(+4050:;9(;696-0.0;(364465:%50=,9:0;@6-,)9(:2(05*635
<;/69:<;/69:
$(0:/"!6>,33",=@'03:65"#*/,9,9$(3(90*690::,20,::,5!6470306$
'03:65(9;,9",65;6!<@),9:"*(@!633(9+"6::'05;,9!(99,;;96>5,"
6+@6>(59(47;65<5)(9<5)(936905+6,)/(9+;9(/(4(55(/
(5:(9(1(=0+(9>66+,3305.#,59@:0556=</5!@3,(B<-,9!(C630
(.,5:(5+,95(*2'*5;6:/033(5"6905 /5,0:,9$!(<3:,5!,9:0*6"(05,
",,+"0,::,34(5#(.56;;0#*/40;;#1<55,:26.!#;965.$(=0(50#&6.,3$'03*/(5+$
'0330(4:
$/0:(9;0*3,0:(=(03()3,(;0.0;(364465:%50=,9:0;@6-,)9(:2(05*635/;;7:+0.0;(3*64465:<53,+<
.,6:*0,5*,-(*7<)
LETTERS
Obliquity-paced Pliocene West Antarctic ice sheet
oscillations
T. Naish
1,2
, R. Powell
3
, R. Levy
4
{, G. Wilson
5
, R. Scherer
3
, F. Talarico
6
, L. Krissek
7
, F. Niessen
8
, M. Pompilio
9
,
T. Wilson
7
, L. Carter
1
, R. DeConto
10
, P. Huybers
11
, R. McKay
1
, D. Pollard
12
, J. Ross
13
, D. Winter
4
, P. Barrett
1
,
G. Browne
2
, R. Cody
1,2
, E. Cowan
14
, J. Crampton
2
, G. Dunbar
1
, N. Dunbar
13
, F. Florindo
15
, C. Gebhardt
8
, I. Graham
2
,
M. Hannah
1
, D. Hansaraj
1,2
, D. Harwood
4
, D. Helling
8
, S. Henrys
2
, L. Hinnov
16
, G. Kuhn
8
, P. Kyle
13
,A.La
¨
ufer
17
,
P. Maffioli
18
, D. Magens
8
, K. Mandernack
19
, W. McIntosh
13
, C. Millan
7
, R. Morin
20
, C. Ohneiser
5
, T. Paulsen
21
,
D. Persico
22
, I. Raine
2
, J. Reed
23,4
, C. Riesselman
24
, L. Sagnotti
15
, D. Schmitt
25
, C. Sjunneskog
26
, P. Strong
2
,
M. Taviani
27
, S. Vogel
3
, T. Wilch
28
& T. Williams
29
Thirty years after oxygen isotope records fr om microfossils deposited
in ocean sediments confirmed the hypothesis that variations in the
Earth’s orbital geometry control the ice ages
1
, fundamental questions
remain over the response of the Antarctic ice sheets to orbital cycles
2
.
Furthermore, an understanding of the behaviour of the marine-
based West Antarctic ice sheet (WAIS) during the ‘warmer-
than-present’ early-Pliocene epoch (
5–3 Myr ago) is needed to
better constrain the possible r ange of ice-sheet behaviour in the
context of future global warming
3
.Herewepresentamarineglacial
record from the upper 600 m of the AND-1B sediment core recovered
from beneath the northwest part of the Ross ice shelf by the
ANDRILL programme and demonstrate well-dated,
40-kyr cyclic
variations in ice-sheet extent linked to cycles in insolation influenced
by changes in the Earth’s axial tilt (obliquity) during the Pliocene.
Our data provide direct evidence for orbitally induced oscillations in
the WAIS, which periodically collapsed, resulting in a switch from
grounded ice, or ice shelves, to open waters in the Ross embayment
when planetary temperatures were up to
3 6C warmer than today
4
and atmospheric CO
2
concentration was as high as 400 p.p.m.v.
(refs 5, 6). The evidence is consistent with a new ice-sheet/ice-shelf
model
7
that simulates fluctuations in Antarctic ice volume of up to
17 m in equivalent sea level associated with the loss of the WAIS and
up to 13 m in equivalent sea level from the East Antarctic ice sheet, in
response to ocean-induced melting paced by obliquity. During inter-
glacial times, diatomaceous sediments indicate high surface-water
productivity, minimal summer sea ice and air temperatures above
freezing, suggesting an additional influence of surface melt
8
under
conditions of elevated CO
2
.
The Earth’s climate system during the Pliocene and early-
Pleistocene epochs was regulated by a ,40-kyr periodicity. The geo-
logical evidence for this is widespread and expressed in polar to
equatorial depositional environments including (1) ice volume from
oxygen isotope (d
18
O) records, that co-vary with the pattern of ice-
rafted debris in deep-sea sediments
9
; (2) ocean circulation
10
and
temperature also inferred from deep-sea sediment proxies
11,12
; (3)
atmospheric circulation from continental dust deposits
13
; and (4)
global sea-level fluctuations recorded in the shallow-marine continental
margins
14
.
The 40-kyr cycle is almost certainly linked to variations in the
Earth’s orbital obliquity. However, the specific nature of this forcing
and its influence on Antarctic glaciation remain unresolved owing to a
lack of well-dated climate records that directly sample past oscillations
of the ice sheet. The new AND-1B core provides such a record (Fig. 1).
In this Letter, we focus on the early Pliocene (,5–3 Myr ago) part of
the record, because for this period the response of Antarctic ice sheets
to orbital forcing can be studied without the complicating influence of
large Northern Hemisphere ice sheets on sea-level and deep-sea d
18
O
records
15
. Furthermore, polar ice-sheet boundary conditions were
similar to today, but the climate was warmer
3,5,6
. With anthropogenic
warming projected to rise an averageof ,3 uC in mean temperature by
the end of the twenty-first century, more significance is being placed
on the early Pliocene as an analogue for understanding the future
behaviour of the WAIS
3
and its contribution to global sea level
16
.
Far-field geological evidence for palaeoshorelines up to 25 m above
present
17,18
are consistent with ice-volume estimates from deep-ocean
d
18
O data
19
, and imply deglaciation of the Greenland ice sheet, the
1
Antarctic Research Centre, Victoria Universi ty of Wel lington, Kelburn Parade, PO Box 600, Wellington 6012, New Zealand.
2
GNS Science, 1 Fairway Drive, PO Box 30-368, Lower Hutt
5040, New Zealand.
3
Department of Geology & Environmental Geosciences, Northern Illinois University, DeKalb, Illinois 60115, USA.
4
ANDRILL Science Management Office,
Department of Geosciences, University of Nebraska-Lincoln, Lincoln, Nebraska 68588-0340, USA.
5
University of Otago, Department of Geology, PO Box 56, Leith Street, Dunedin,
Otago 9001, New Zealand.
6
Universita
`
di Siena, Dipartimento di Scienze delle Terra, Via Laterina 8, I-53100 Siena, Italy.
7
Ohio State University, Department of Geological Sciences,
275 Mendenhall Lab, 125 South Oval Mall, Columbus, Ohio 43210, USA.
8
Alfred Wegener Institute, Department of Geosciences, Postfach 12 01 6, Am Alten Hafen 26, D-27515
Bremerhaven, Germany.
9
Istituto Nazionale di Geofisica e Vulcanologia, Via della Faggiola 32, I-56126 Pisa, Italy.
10
Department of Geosciences, 233 Morrell Science Centre, University
of Massachusetts, Amherst, Massachusetts 01003-9297, USA.
11
Department of Earth and Planetary Sciences, Harvard University, Massachusetts 02138, USA.
12
Earth and
Environmental Systems Institute, Pennsylvania State University, University Park, Pennsylvania 16802, USA.
13
New Mexico Institute of Mining & Technology, Earth & Environmental
Sciences, Socorro, New Mexico 87801, USA.
14
Department of Geology, Appalachian State University, ASU Box 32067, Boone, North Carolina 28608-2067, USA.
15
Istituto Nazionale
di Geofisica e Vulcanologia, Via di Vigna Murata 605, I-00143 Rome, Italy.
16
Department of Earth and Planetary Sciences, Johns Hopkins University, Baltimore, Maryland 21218, USA.
17
Federal Institute of Geosciences & Natural Resources, BGR, Stilleweg 2, D-30655 Hannover, Germany.
18
Universita
`
Milano-Bicocca, Dipartimento di Scienze Geologiche e
Geotecnologie, Piazza della Scienza 4, I-20126 Milano, Italy.
19
Colorado School of Mines, Department of Chemistry & Geochemistry, 1500 Illinois Street, Golden, Colorado 80401,
USA.
20
US Geological Survey, Mail Stop 403, Denver Federal Center, Denver, Colorado 80225, USA.
21
University of Wisconsin-Oshkosh, Department of Geology, 800 Algoma
Boulevard, Oshkosh, Wisconsin 54901, USA.
22
Dipartimento di Scienze della Terra, Universita
`
degli Studi di Parma, Via Usberti 157/A, I-43100 Parma, Italy.
23
CHRONOS, Iowa State
University, Department of Geological & Atmospheric Sciences, 275 Science I, Ames, Iowa 50011-3212, USA.
24
Department of Geological and Environmental Sciences, School of Earth
Sciences, Stanford University, Stanford, California 94305, USA.
25
Department of Physics, Mailstop #615, University of Alberta, Edmonton, Alberta T6G 2G7, Canada.
26
Department of
Geology and Geophysics, Louisiana State University, Baton Rouge, Louisiana 70803, USA.
27
CNR, ISMAR
–
Bologna, Via Gobetti 101, I-40129 Bologna, Italy.
28
Albion College,
Department of Geology, Albion, Michigan 49224, USA.
29
Columbia University, Lamont-Doherty Earth Observatory, Palisades, New York 10964, USA. {Present address: GNS Science,
1 Fairway Drive, PO Box 30368, Lower Hutt 5040, New Zealand.
Vol 458
|
19 March 2009
|
doi:10.1038/nature07867
322
Macmillan Publishers Limited. All rights reserved
©2009
This article is a U.S. government work, and is not subject to copyright in the United States.
WAIS and the marine margins of the East Antarctic ice sheet (EAIS)
during the warmest early Pliocene interglacials.
Thirty-eight glacimarine cycles, each bounded by glacial surfaces
of erosion (GSEs), occur in the upper 600 m of the core, and record
oscillations in the extent of an ice sheet in Ross embayment during
the past 5 Myr (Fig. 2)
20
. The drilled strata accumulated in the rift axis
of the Victoria Land basin, ,100 km seaward of the coast.
Accommodation space for the preservation of the sediments, and
their protection from subsequent glacial erosion, was provided by
high rates of tectonic subsidence due to a unique combination of
regional rift extension and flexure of the crust by local volcanic
islands (Supplementary Information).
Figures 3 and 4 summarize our interpretations of individual sedi-
mentary cycles in terms of the vertical occurrences of lithofacies, that
is, sediments representing specific environments of deposition. These
range from marine-diatom-rich deposits and mudstones deposited
during interglacials to ice-proximal diamictites, conglomerates and
sandstones representing glacial periods. During glacial periods, the ice
sheet had a laterally extensive marine terminus extending well beyond
the drill site, out into the Ross Sea. During interglacials, the drill site
was either covered by an ice shelf or, when the ice sheet retreated, lay in
open water. The sedimentary characteristics of the cycles and the
approach used for their interpretation are given in more detail
elsewhere
21
(Methods). We note here that changes in lithofacies
through time primarily reflect the proximity of the ice-sheet ground-
ing zone and the thermal characteristics of the depositing ice (Fig. 2
and Supplementary Information). Such inferences are consistent with
depositional models from a variety of different glacimarine
regimes
22,23
, and permit the identification of 38 oscillations in the
extent of the ice sheet’s grounding line.
The composition of till (ice-contact diamictites), which overlie
GSEs and represent sediments transported at the base of a grounded
ice sheet, show that the ice originated from large outlet glaciers in the
Transantarctic Mountains (TAM), especially the Mulock and Skelton
glaciers south of Minna Bluff
21
. A new continental Antarctic ice-sheet
model
7
run for the past 5 Myr supports geological interpretations
that the provenance of grounded ice at the AND-1B site is always
from nearby southern-TAM outlet glaciers during glacial advances
(Fig. 2).
Notably, the model finds that local ice variations at the AND-1B
site are indicative of the overall West Antarctic glacial state, because
both are controlled by variations in ocean-induced melt. When
open-water marine sediments occur in the AND-1B core, the model
shows not only deglaciation in the western Ross embayment, but also
the collapse of the entire WAIS (Fig. 2). Pliocene–Pleistocene varia-
tions in ice volume are dominated by large WAIS advances and
retreats, while the high-altitude regions of the EAIS remain relatively
stable. This is because air temperatures never become warm enough
to cause significant surface melting on the EAIS
24
, whereas variations
in ocean-induced melt and sea level affect the marine-based WAIS
much more than the EAIS. Thus, the sedimentary cycles in the AND-
1B core both track local variations of the coastal margin of the EAIS
(for example TAM outlet glaciers) and provide physical evidence for
major changes in the mass balance of the WAIS (Supplementar y
Information).
Figure 2 illustrates the stratigraphic position of 26 chronological
datums that are used to constrain the age and duration of the 38 sedi-
mentary cycles and identify the time missing at cycle-bounding erosion
surfaces, that is, unconformities (Supplementary Table 1). The chro-
nology is developed from
40
Ar/
39
Ar ages of volcanic deposits and a
quantitative diatom biostratigraphy used to constrain the correlation
between the magnetic polarity stratigraphy and the geomagnetic
polarity timescale
21
. The approach used, and associated uncertainties,
are outlined in Methods. About 36% of the last 5 Myr is represented as
rock in the AND-1B core; the rest is lost at unconformities resulting
from erosion through long-term tectonic influences and shorter-te rm
volcanic and glacial processes. Chronostratigraphic constraints enable
identification of two types of unconformities: (1) those where the time
missing is longer than a Milankovitch cycle, interprete d as major
erosion due to tectonic influences and/or a major phase of glacial
advance; and (2) those of suborbital duration reflecting lesser glacial
erosion associated with a single glacial advance truncating only part of
the previous cycle. The chronology also allows the duration of relatively
continuous stratigraphic packages comprising more than one cycle to
be estimated with sufficient precision for the recognition of orbital
periods
25
(Methods).
One such interval is illustrated in detail in Fig. 3 and comprises six
early-Pliocene glacial cycles (cycles 38–33), spanning 265 kyr. In this
case, the identification of three palaeomagnetic reversal boundaries
allows one-to-one matching of the WAIS grounding-line oscillations
recorded in the AND-1B core to individual 40-kyr ice-volume cycles
in the deep-sea benthic d
18
O stack
26
and modelled ice-volume cycles
between 4.896 and 4.631 Myr ago. Figure 4 illustrates an interval in
which the chronology constrains 16 successive early- to mid-Pliocene
(3.60–2.87 Myr ago) cycles (32a–d to 20), spanning 700 kyr. In this
case, the cyclostratigraphic interpretation and the distribution of
time within the AND-1B core are not so straightforward, primarily
because a ,60-m-thick interval of marine diatomite, spanning
,200 kyr, occurs between 438 and 376 m.b.s.f. We use the distri-
bution of .2-mm-diameter iceberg-rafted debris (IBRD) through
the continuous, thick diatomite unit as an index for ice-rafting
intensity and glacial variability. The IBRD record identifies four or
five main orbital cycles (32a–d) reflecting glacial fluctuations during
an extended period of biopelagic deposition in the Ross embayment,
when the ice sheet remained landward of the drill site (Fig. 2). The
composition of the IBRD reflects oscillations of local outlet glaciers,
which remained near the coast with no significant expansion into the
Ross Sea
21
.
If the four or five IBRD cycles in the diatomite and the eleven
overlying unconformity-bounded cycles are distributed evenly over
this time interval, the resulting duration is ,40 kyr per cycle (Fig. 4).
Weddell
Sea
Vostok
Dome C
90º
90º
W 180º E
W 0º E
EAIS
Ross Sea
Prydz
Bay
Antarctic Peninsula
1,000
1,000
DSDP/ODP Plio-Pleistocene cores
Transantarctic Mountains
Boundary between East and
West Antarctic ice
Ice ow direction
WAIS
1096
1095
1097
1165
1100
1102/3
693
696
697
694
274
1101
IODP Wilkes Land scheduled
739–743
273
272–271
AND-1B
98% recovery
Leg 119/188
<40% recovery
Leg 28
<40% recovery
Leg 178
<15% recovery
1167
Leg 113
Atlantic Ocean
Indian Ocean
Pacic Ocean
Wilkes Land
Figure 1
|
Location of the ANDRILL McMurdo Ice Shelf Project AND-1B drill
site in the northwestern corner of the Ross ice shelf.
Also shown are the
locations of previous Deep Sea Drilling Project (DSDP), Integrated Ocean
Drilling Program (IODP) and Ocean Drilling Program (ODP) cores, Plio-
Pleistocene cores, percentage recovery (from shelf sites) and geographic
features including ice-sheet configuration and flow lines. (Supplementary
Figs 1 and 2 show more detail of the glaciologic and geologic setting). The
dashed contour indicates a depth of 1,000 m below sea-level.
NATURE
|
Vol 458
|
19 March 2009 LETTERS
323
Macmillan Publishers Limited. All rights reserved
©2009
The short-duration normal-polarity interval between the Mammoth
and Kaena subchrons is not represented in the AND-1B core and could
be missing at any of the GSEs between the bases of cycles 28–22.
However, if cycles 28–22 are distributed evenly across the amalgamated
reversed-polarity subchrons between 3.30 and 3.03 Myr ago, then a
one-to-one match with ,40-kyr d
18
O cycles results and suggests that
an intervening normal-polarity interval is most likely missing at the
basal GSE of cycle 26.
The 60-m-thick diatomite unit lacks a sea-ice-associated diatom
flora
27
, and sedimentological evidence implies that warmer-than-present
Fully glaciated Antarctica + Northern
Hemisphere (LGM eqv.)
West Antarctica and Greenland melted
Cycles of known duration
Unconformities of known duration
Cycles and unconformities of uncertain duration
Fully glaciated Antarctica (LGM eqv.)
Present-day ice volume
Diatomite
Volcanic
Mudstone/sandstone
Glacial surface of erosion (<1 cycle missing)
Diamictite
Siltstone
≥1 40-kyr cycle eroded at unconformity
Mammoth
Jaramillo
Benthic δ
18
O
stack (‰)
100
200
300
400
500
Age
Depth (m.b.s.f.)
Cycle type
Polarity
zonation
Mud Sand Grav.
Lithologic log
AND-1B
Glacial
proximity
Cycle
no.
Duration
(no.
cycles)
Chronostratigraphic
datum
From site survey
gravity cores
PleistocenePliocene
R5
R4
N6
N5
N4
R3
N3
R2
N2
N1
32.33
80.03
84.97
91.13
191.75
346.13
438.61
452.86
459.19
519.40
596.35
248
A/0.781
N/3.03
Q/3.596
R/<3.56
S/>4.29
Z/4.896 Myr ago
X/4.779 Myr ago
Y/<4.86 Myr ago
H/>2.01
I/>2.21
J/2.581
K/<2.79
L/<2.87
B/0.988
C/1.014±0.004
G/>1.945
F/1.67±0.03
0
I
PIDM
Cycle motif 2b (subpolar)
Cycle motif 1 (polar)
Global sea level
(metres below present)
600
N8
N7
R7
R6
630.12
637.99
654.48
29
59
60
58
56
K1
Km5
M1
M2
Mg2
Mg1
Mg3
Mg5
Mg6
Mg8
Mg10
Mg7
Mg11
Gi1
Gi5
Gi7
Gi4
Gi6
Gi2
Ns2
Ns4
Ns6
Ns3
Ns1
Ns5
Si1
Si2
Si4
Si5
Si3
89
95
97
99
101
100
198
96
92
90
88
86
G3
G5
G7
G11
G17
51
57
25
31
33
32
30
5e
9
11
13
15
17
7c
7a
19
18
16
14
12
10
8
6
2
?
1
2
3
4
5
6
7
8
9
11
12
13
14
15
16
17
18
19
22
20
23
21
25
24
26
27
28
29
30
31
32
33
34
35
36
37
38
800 kyr?
(8 cycles?)
80 kyr
(2 cycles)
80 kyr
(2 cycles)
40 kyr?
(1 cycle?)
400 kyr
(9/10 cycles)
265 kyr
(6 cycles)
Clast
abundance
32a
32b
32c
32d
W/4.800±0.076
D/1.072
10
?
?
?
E/1.65±0.03
T/4.493
V/4.631
U/<4.78
P/3.330
O
}
?
?
?
Modelled Antarctic
ice volume (m)
Obliquity
(º)
Mean annual
insolation, 75º S
(W m
–2
)
Ice volume (10
6
km
3
)
Kaena
Olduvai
Brunhes
0.5
1.0
0
1.5
2.0
2.5
3.5
3.0
4.5
4.0
5.0
ATS ,
2004
(Myr)
BrunhesGaussGauss MatuyamaMatuyamaGilbertGilbert Gilb.
Gilb.
Gilb.
Cochiti
Nunivak
Sidufjall
Thvera Mat.
G19
G20
G6
G2
G10
G15
G1
103
85
87
91
27
Summer
insolation,
75º S (W m
–2
)
0.08
0.06
0.04
0.02
0.1
0.2
0.2
0.4
0.6
0.8
0.2
0.05
0.1
0.15
0.2
0.4
0.6
0.8
0.3
0.4
0 0.02 0.04 0.06
0 0.02 0.04 0.06
1/f (kyr) 1/f (kyr) 1/f (kyr) 1/f (kyr) 1/f (kyr)
0 0.02 0.04 0.06 0 0.02 0.04 0.06
0 0.02 0.04 0.06
Power density
EETTPPET P ET P ET P
M/>2.87
Cycle motif 2a (subpolar to polar)
220 kyr
(6 cycles)
No. per 10 cm
Km3
5.0 4.5 4.0 3.5 3.0 –24 –16 –8 0 8 38 39 40 41 42 43
178
182
186
190
320
340
360
Summer sea-ice diatom
assemblage
Ice-free oceanic diatom
assemblage
Diatoms absent
or rare
a
b
cd e
Gi3
104
20
52
54
53
–120 –80 –40 0 40 50 40 35 30 25 20
f
ghij
060
LETTERS NATURE
|
Vol 458
|
19 March 2009
324
Macmillan Publishers Limited. All rights reserved
©2009
