as a PDF

Earth and Planetary Science Letters 209 (2003) 309^322
www.elsevier.com/locate/epsl
Neodymium isotopic reconstruction of late Paleocene^early
Eocene thermohaline circulation
Deborah J. Thomas a; , Timothy J. Bralower a;1 , Charles E. Jones b
a
b
Department of Geological Sciences, University of North Carolina, Chapel Hill, NC 27599-3315, USA
Department of Geology and Planetary Science, University of Pittsburgh, Pittsburgh, PA 15260, USA
Received 13 November 2002; received in revised form 27 January 2003; accepted 8 February 2003
Abstract
High-resolution, fish tooth Nd isotopic records for eight Deep Sea Drilling Project and Ocean Drilling Program
sites were used to reconstruct the nature of late Paleocene^early Eocene deep-water circulation. The goal of this
reconstruction was to test the hypothesis that a change in thermohaline circulation patterns caused the abrupt 4^5‡C
warming of deep and bottom waters at the Paleocene/Eocene boundary ^ the Paleocene^Eocene thermal maximum
(PETM) event. The combined set of records indicates a deep-water mass common to the North and South Atlantic,
Southern and Indian oceans characterized by mean ONd values of V38.7, and different water masses found in the
central Pacific Ocean (ONd V34.3) and Caribbean Sea (ONd V1.2). The geographic pattern of Nd isotopic values
before and during the PETM suggests a Southern Ocean deep-water formation site for deep and bottom waters in the
Atlantic and Indian ocean basins. The Nd data do not contain evidence for a change in the composition of deep
waters prior to the onset of the PETM. This finding is consistent with the pattern of warming established by recently
published stable isotope records, suggesting that deep- and bottom-water warming during the PETM was gradual and
the consequence of surface-water warming in regions of downwelling.
; 2003 Elsevier Science B.V. All rights reserved.
Keywords: seawater Nd isotopes; PETM; Ocean Drilling Program; thermohaline circulation
1. Introduction
1.1. Overview of late Paleocene^early Eocene
climate and oceanography
Gradual global warming during the late Paleocene through early Eocene led to the warmest
climatic conditions of the last 90 million years.
Stable isotope records from deep-sea sedimentary
sections indicate that high-latitude surface waters
increased from V11 to 15‡C, while global deepwater temperatures warmed from V8 to 12‡C [1^
4]. Terrestrial records suggest a s 2‡C mean annual temperature increase during the latest Paleo-
* Corresponding author. E-mail address: dthomas1@email.unc.edu (D.J. Thomas).
1
Present address: Department of Geosciences, Pennsylvania State University, University Park, PA 16802, USA
0012-821X / 03 / $ ^ see front matter ; 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0012-821X(03)00096-7
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
310
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
cene (e.g. [5]), corroborating trends in the deepsea records.
One of the most abrupt and signi¢cant global
warming events in the geologic record occurred at
the Paleocene/Eocene boundary (P/E boundary)
in the middle of the gradual, long-term late Paleocene^early Eocene warming trend. This transient event (V210 kyr in duration), the Paleocene^
Eocene thermal maximum (PETM), was characterized by profound changes in the Earth’s climate, oceans, and biota. During the PETM,
southern high-latitude sea-surface temperatures
(SSTs) increased by V6‡C [6], subtropical SSTs
within the South Atlantic increased by as much as
4‡C [7], and deep-ocean temperatures increased by
V4^5‡C [2,6,8]. One explanation for the rapid
warming of the deep ocean invokes a change in
the location of deep-water formation from cool,
high-latitude regions to warmer, subtropical areas
(e.g. [6]). Warming of the deep ocean may have
triggered the thermal dissociation of methane hydrates which is the likely source of the excursion
[9].
The cause of the extended interval of extreme
warmth as well as the transient PETM event may
be intricately related to the evolution of thermohaline circulation patterns (e.g. [10^13]). Multiple
operating modes of the oceanic thermohaline conveyor are thought to be possible, and switches in
the operating mode may have dramatic e¡ects on
the equilibrium state of the climate system (e.g.
[14]). However, very little is known about the nature of thermohaline circulation during the warm
climate of the early Paleogene. The lack of a modern analog for the extreme warmth, as well as the
signi¢cantly di¡erent ocean basin con¢guration,
only permit us to speculate about where deepwater masses formed and how they circulated.
Thus a general understanding of the nature of
thermohaline circulation in the early Paleogene
is required before possible changes in the prevailing pattern can be investigated.
Early Paleogene oceans were characterized by
reduced equator-to-pole and surface-to-deep temperature gradients (e.g. [4] and references therein).
The occurrence of high-latitude SSTs signi¢cantly
higher than those found today implies enhanced
meridional heat transport. Warmer deep-water
masses, such as those that existed in the early
Paleogene, have a diminished capacity for heat
transport, thus the intensity of thermohaline thermal transport must also have been greatly enhanced. However, such a scenario is di⁄cult to
reconcile with a reduced equator-to-pole thermal
gradient [15].
Modern thermohaline circulation is dominated
by two primary sources of deep waters: the cold
and relatively saline waters produced in the Nordic Seas that over£ow into the North Atlantic
(North Atlantic deep water, NADW), and the
very cold and relatively fresh waters formed
around Antarctica (Antarctic bottom water,
AABW). In addition, a similar magnitude of
heat transport is accomplished by intermediate
water circulation (e.g. [16] and references therein).
Because rifting and sea£oor spreading within the
Norwegian and Greenland seas began during the
late Paleocene^early Eocene (e.g. [17] and references therein), the northern North Atlantic was not
a signi¢cant source of deep waters during the
early Paleogene. However, there is evidence for
possible deep-water formation in the Southern
Ocean during the late Paleocene and early Eocene. A late Paleocene erosional event in seismic
records from the western North Atlantic was
linked to bottom-water formation and circulation
from the south, analogous to modern AABW
[18]. Ramsey and others [19] interpreted the distribution of early Cenozoic hiatuses from the Indian Ocean as evidence of a Southern Ocean
deep-water source existing since at least the early
Eocene. In addition, indirect evidence for this circulation pattern comes from benthic foraminiferal
faunal distribution [20] and stable isotope values
[21,22].
Other potential sources of deep waters during
warmer climatic intervals are evaporative subtropical regions such as the eastern (and possibly
western) Tethys and the Gulf of Guinea [12,23].
While warm saline deep waters may have contributed to the water mass structure during past
greenhouse climatic intervals, they were most
likely not the dominant water mass. Consideration of salt mass balance, moisture £ux and runo¡ distribution patterns (e.g. [24,25]), as well as
the diminished ability of a warm saline water
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
311
Fig. 1. Cenozoic ONd records derived from Fe^Mn crusts establishing North Atlantic deep waters as the most non-radiogenic
(ONd V310 to 312), the North Paci¢c as the most radiogenic (ONd V35 to 36), and Indian Ocean deep waters as intermediate
in composition between those two end members (ONd V38). The range of variability exhibited by the new high-resolution ¢sh
tooth data from the Atlantic sites is depicted by the brackets on the ONd axis.
mass to transport heat to the poles [26] preclude
the protracted formation of warm saline deep
waters.
Here we attempt to determine the nature of late
Paleocene^early Eocene thermohaline circulation,
and assess the possibility that a change in thermohaline circulation caused the PETM.
1.2. Neodymium as a deep-water mass proxy
The short oceanic residence time of Nd (V1000
yr; e.g. [27]) with respect to the mixing time of the
oceans (V1500 yr; [28]) implies that the oceans
are not homogeneous in their Nd isotopic composition. Thus distinct interbasinal Nd isotopic differences exist, making Nd a useful tracer of deepwater mass transit (e.g. [29^31]), similar to the
temperature, salinity, and nutrient characteristics
of modern water masses. Continental weathering
and runo¡ is the principal source of Nd to the
oceans (e.g. [32^35]); thus the Nd isotopic composition of individual deep-water masses is derived from the composition of dissolved, and to
a lesser extent, suspended materials draining into
the source regions [32,36,37].
Most investigations of paleo-Nd isotopic composition employ analyses of the layers of Fe^Mn
crusts (e.g. [38]). Such records of deep-water Nd
isotopic composition demonstrate the general
long-term evolution of water mass composition
associated with the opening and closure of major
oceanic gateways (e.g. [38]). Analysis of Nd in
Fe^Mn crusts dredged from the sea£oor has established the Cenozoic evolution of deep-water
masses in the North Atlantic, Indian, and North
Paci¢c basins [38^40] (Fig. 1). However, these
records are unlikely to reveal a potential change
during the late Paleocene^early Eocene due to
their low resolution and the slow precipitation
rate of Fe^Mn crusts (Vmm/Myr). In order to
¢ll the gaps in the late Paleocene^early Eocene
Nd isotopic reconstruction, we have analyzed fossil ¢sh teeth at high stratigraphic resolution (2^10
cm across the P/E boundary corresponding to a
temporal resolution of V1 to several kyr).
Fossil ¢sh teeth are useful for paleo-Nd investigations because of their relatively high Nd concentrations (100^1000 ppm) (e.g. [41^43]), as well
as their resistance to dissolution in corrosive bottom waters. Fish teeth (as well as other biogenic
apatite) acquire their enhanced Nd concentrations
during an early diagenetic reaction at the sediment/water interface (e.g. [43]). Thus the fossil
material records the Nd isotopic composition of
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
312
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
2. Methods
from ¢ve to 15 teeth per sample, depending on
size and availability). In this study, only teeth
were employed. Samples were then cleaned using
an established reductive/oxidative cleaning protocol [46,47]. We analyzed samples as NdOþ using a
multi-collector Micromass Sector 54 at the radiogenic isotope facility at UNC-CH. Monitor peak
(144 Nd16 O) beams of V1 V were achieved by introducing pure oxygen into the source via a leak
valve. External analytical precision based upon
replicate analysis of the UNC Ames Nd standard
(as NdOþ ) was 0.512140 Q 0.000014 (2 S.D.). Replicate analyses of eight of nine samples yielded Nd
isotope values within error limits (table 2 in the
Background Data Set2 ). Nd isotope values are reported using the epsilon notation, ONd , which normalizes the analyzed 143 Nd/144 Nd ratio to the bulk
Earth value of CHUR (chondritic uniform reservoir) [48].
We analyzed several samples for Sm to determine the range of 147 Sm/144 Nd ratios recorded by
teeth at each of the sites. The maximum range of
147
Sm/144 Nd ratios is 0.11747^0.13971 for all of
the sites investigated, similar to other analyses
of ¢sh teeth [44]. We applied the mean 147 Sm/
144
Nd value of 0.1286 to all samples to calculate
ONd (t) values (table 2 in the Background Data
Set2 ).
2.1. Site selection
2.3. Age model
Fish tooth Nd isotope stratigraphies were generated at eight DSDP and ODP sites that contain
relatively complete sections of the upper Paleocene to lower Eocene transition including the
PETM (Fig. 2). These sites provide broad geographic coverage, and enable investigation of
both intermediate- and deep-water mass Nd composition (table 1 in the Background Data Set2 ).
Numerical ages for the early Paleogene are in a
constant state of revision, primarily due to the
paucity of radiometric tie-points, discovery of
stratigraphic hiatuses, lack of complete deep-sea
sedimentary sequences that preserve biostratigraphic markers and yield a reliable magnetostratigraphy, and the paucity of biostratigraphic datums to correlate from high to low latitudes (e.g.
[49,50]). Recent e¡orts to re¢ne the time scale
focus on development of an orbitally tuned chronology based on deep-sea proxy records from relatively expanded and complete sections [51,52].
However, correlation to other deep-sea sections
with poorer recovery or stratigraphic hiatuses is
di⁄cult. The di⁄culties in establishing a reliable
age model for the PETM are compounded at Sites
527, 549, and 1001 due to carbonate dissolution.
the overlying bottom water (e.g. [41,44]). The Nd
isotopic signal in ¢sh teeth has been shown indirectly to resist diagenetic exchange with pore
water during burial (e.g. [44,45]), providing a
means to reconstruct the temporal record of bottom-water composition.
The advantage of ¢sh teeth over Fe^Mn crusts
is that they are present, albeit rare, in most deepsea sedimentary sections. This permits construction of higher-resolution Nd isotopic records,
both spatially and temporally. By analyzing ¢sh
teeth, we can detect short-term changes in water
mass composition that would otherwise have been
averaged out during the slow precipitation of Fe^
Mn crusts (Fig. 1). Moreover, the age of deep-sea
sedimentary sections is more precisely determined
than in crusts.
To better determine the nature of early Cenozoic thermohaline circulation, we generated the
¢rst high-resolution, late Paleocene^early Eocene
(V50.6 through V56.5 Ma) seawater Nd isotopic
records. In addition, we use these records to assess the possibility that a change in the nature of
circulation caused the deep-water warming associated with the onset of the PETM.
2.2. Analytical methods
Fish teeth were handpicked from the s 63 Wm
size-fraction of washed samples, and multiple
teeth were used in each analysis (in general,
2
http://www.elsevier.com/locate/epsl
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
313
Fig. 2. Paleogeographic reconstruction of the late Paleocene showing locations of the DSDP and ODP sites investigated (from
the Ocean Drilling Stratigraphic Network).
Poor recovery and hiatuses a¡ect Sites 401, 865,
and 1001.
Despite these di⁄culties, we have determined a
reasonable age model based on the orbital stratigraphy [52] and the well-characterized carbon isotope stratigraphy [6,53] of the Site 690 PETM
interval. Site 690 contains the most expanded
and complete stratigraphic record of the PETM
interval [52,54] and the V4x planktonic foraminiferal N13 C excursion is assumed to re£ect the full
magnitude of the marine carbon isotope excursion
(CIE). Inter-site correlations are based on comparison of the Site 690 N13 C curve structure and
excursion magnitude with those recorded at the
other sites. For example, sites that contain a dissolution interval and record a planktonic foraminiferal CIE magnitude signi¢cantly less than 4x
do not contain the true peak of the event and
were visually correlated to the ‘recovery’ portion
of the 690 N13 C curve. Because the long-term Nd
records cover varying time spans, we supplemented P/E interval chemostratigraphic tie-points
with biostratigraphic datums above and below the
N13 C records (table 3 in the Background Data
Set2 ).
Numeric ages assigned to the biostratigraphic
datums [55] have not yet been revised to comply
with the recent proposal for the P/E boundary
global stratotype (the onset of the CIE now designated to be 55.0 Ma). To maintain age con-
sistency with available biostratigraphic datums
we employed the formerly accepted age of 55.5
Ma for the base of the CIE [8,56], and assigned
55.28 Ma to the asymptote of the carbon isotope curve recovery based on the 220-kyr duration of the entire event [52]. We assumed linear
sedimentation rates between datum ages, and
adopted the visual correlation of carbon isotope
stratigraphies to help compensate for artifacts of
rapid and brief lithologic changes which characterize many late Paleocene^early Eocene deep-sea
sections.
3. Results
The resolution of each Nd isotopic record
varies as a function of ¢sh teeth availability.
These records demonstrate high-frequency £uctuations of V0.5 up to V1.5 ONd units, especially
in the densely sampled Paleocene^Eocene transition interval where similar or higher-magnitude
changes occur on a scale of a few centimeters
(in the span of several thousand years). In all of
the records, intervals with a higher sample density
have an increased frequency of ONd £uctuation.
Lower-resolution portions of the records show
smoother trends. The data generated in this investigation are available as table 2 in the Background
Data Set2 .
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
314
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
3.1. DSDP Site 401 (Bay of Biscay, North
Atlantic)
Throughout the Paleocene^Eocene transition at
Site 401 (paleo-water depth 1900 m), ONd values
were V38.8 (Fig. 3A) (204.09^198.65 mbsf;
V55.6^55.1 Ma). The ONd values in the high-resolution portion of the record £uctuate between
V310.0 and 38.0, with a trend towards slightly
more radiogenic (less negative) values. Above the
high-resolution segment at Site 401 ONd values increase to 35.6 at 194.57 mbsf (54.12 Ma) and
then decrease, averaging 38.7 throughout the remainder of the section from 190.93 through
181.54 mbsf (53.23^50.78 Ma).
3.2. DSDP Site 549 (Goban Spur, North Atlantic)
DSDP Site 549 (V3900 m paleo-water depth)
ONd values (Fig. 3B) vary between 39.2 and 38
Fig. 3. Site 401, 549, 527, 690, 213, 865, and 1001 ONd records generated in this study (two analyses from Site 1051 not plotted),
labeled with paleo-water depth and paleo-latitude information. The ONd records (diamonds) are plotted with previously published
carbon isotope records (small squares) for stratigraphic context, particularly within the PETM interval (carbon isotope data sources listed in table 1 in the Background Data Set2 ). Panels A^D (Sites 401, 549, 527, and 690) consist of a long-term record (lefthand side) with the high-resolution portion of the record spanning the P/E boundary (between dashed lines) expanded on the
right-hand side. Site 213 data (panel E) only spans a portion of the early Eocene interval, and Sites 865 and 1001 (panels F and
G) only consist of a few analyses. Carbon isotope data from Sites 401, 527, 549, and 865 were derived from multi-specimen analyses of planktonic foraminifera, the shaded areas in the Site 690 panel represent the range of N13 C values from single-specimen
analyses of planktonic foraminifera, and the data from Sites 213 and 1001 came from bulk-sediment analyses.
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
from 355.50 to 337.14 mbsf (56.4^55.4 Ma), with
a mean of 38.5. ONd values increase upward,
peaking at 34.7 (335.28 mbsf; 54.36 Ma). The
subsequent early Eocene ONd trend (from 334.99
to 305.00 mbsf; 53.92^52.90 Ma) is a gradual decrease from 36.6 to 38.4.
3.3. DSDP Site 527 (Walvis Ridge, South
Atlantic)
Site 527 (V3400 m paleo-water depth) records
ONd values that oscillate between V39.9 and
315
V38.0 from 203.3 to 199.49 mbsf (55.58^55.34
Ma) (Fig. 3C), with a mean of 39.0. In this portion of the Paleocene^Eocene transition, a signi¢cant positive excursion in ONd from 39.4 to 37.2
occurs at 200.31 mbsf (55.41 Ma). This is followed by a decrease in ONd to 39.5 from 199.73
mbsf (55.36 Ma) to 189.6 mbsf (54.46 Ma), and
another positive excursion in ONd to 37.2 at the
top of the analyzed record.
3.4. ODP Site 690 (Maud Rise, Southern Ocean)
Mean ONd values at Site 690 (V1900 m paleowater depth) are V39.1 throughout the Paleocene^Eocene transition from 174.02 to 164.18
mbsf (Fig. 3D). In the higher-resolution portion
of the record spanning the P/E boundary from
172.00 to 170.74 (55.59^55.506 Ma), ONd values
£uctuate between 39.9 and 39.0. The magnitude
of the oscillations increases above this interval, as
ONd values vary from 310.2 to 38.0.
3.5. DSDP Site 213 (proto-Indian Ocean)
The base of the ONd record at Site 213 (V3000
m water depth) lies above the P/E boundary (Fig.
3E). Site 213 ONd values increase from V39.5 at
147.57 mbsf (55.43 Ma) to 35.8 at 146.30 mbsf
(55.406 Ma). Subsequently, ONd values decrease to
37.7 at 144.82 mbsf (55.324 Ma), and 38.6 at
142.96 mbsf (54.710 Ma). The mean value of
Site 213 values is 38.1.
3.6. ODP Site 865 (Allison Guyot, Paci¢c)
We were able to analyze only six samples from
Site 865 (V1500 m paleo-water depth) due to
insu⁄cient ¢sh teeth (and debris) in most of the
samples processed. The range of ONd values is
35.0 to 33.5 (Fig. 3F ; table 2 in the Background
Data Set2 ), with a mean value of 34.3.
3.7. ODP Site 1001 (Lower Nicaragua Rise,
Caribbean)
Fig. 3 (Continued).
Only ¢ve samples from Caribbean ODP Site
1001 (V2500 m paleo-water depth) contained suf¢cient ¢sh teeth for Nd isotopic analysis (Fig.
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
316
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
Fig. 4. All Nd isotopic data plotted together versus age. Site symbols are listed in the legend. Panel A contains the entire longterm data set for Sites 213, 401, 527, 549, 690, and 1051. The arrow represents ‘average’ early Paleogene Fe^Mn crust ONd values
for the northern Atlantic (light gray) [37]. The orange circle at V53 Ma represents the only Tethyan phosphate analysis available
for the late Paleocene^early Eocene interval [59]. The yellow box indicates the portion of the graph that is expanded in panel B.
We include the planktonic foraminiferal N13 C curve [6] for stratigraphic context with respect to the onset of the PETM. Panel C
contains the entire Nd isotopic data set for Sites 865 and 1001 (note the change in ONd scale to accommodate the more radiogenic values). The dark gray arrow indicates average North Paci¢c crust values for the early Paleogene [39]. The PETM portion of
panel C (yellow box) is expanded in panel D.
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
317
Fig. 5. General late Paleocene^early Eocene mean ONd values with inferred deep-water formation regions and circulation patterns.
3G), precluding construction of a high-resolution
ONd stratigraphy. Mean ONd values through the
Paleocene^Eocene transition (240.15^238.55 mbsf)
were 1.2, and decrease to 34.3 at 237.76 mbsf
(55.41 Ma). An additional sample at 208.79
mbsf yielded an ONd value of 31.0.
3.8. ODP Site 1051 (Blake Nose, North Atlantic)
Only two samples from Site 1051 (V2000 m
paleo-water depth) could be analyzed due to insu⁄cient ¢sh teeth. These samples yielded nonradiogenic ONd values (39.2 and 38.3) (Fig. 4).
4. Discussion
4.1. Early Paleogene thermohaline circulation
The high-resolution ¢sh tooth Nd isotopic
records corroborate trends in the low-resolution
Fe^Mn crust records, yet reveal considerably
more structure and variability in the composition
of the deep Atlantic, Southern, Indian, and Paci¢c
Oceans and Caribbean Sea during the late Paleocene^early Eocene (Fig. 4). One important aspect
of the new data is that analyses from Site 213
constrain the Nd isotopic composition of the Indian Ocean back to the early Eocene, beyond the
late Oligocene extent of the Indian Ocean crust
data (Fig. 1). In addition, Site 1001 data provide
the ¢rst Caribbean Sea values for the Cenozoic.
Three deep-water Nd isotopic provinces are evident from the data (Fig. 4). Sites in the Atlantic,
Indian, and Southern oceans (Sites 213, 401, 527,
549, 690, and 1051) are generally characterized by
relatively non-radiogenic Nd isotopic values
(V38.7 ONd ). Such values are similar to, although
slightly more negative (less radiogenic) than,
present-day Indian and Southern ocean values,
but are more positive (more radiogenic) than
modern North and South Atlantic ocean values
(e.g. [29^31]). The occurrence of more radiogenic
ONd values of V34.3 at intermediate depths in
the Paci¢c (V1500 m) suggests that this water
mass had a di¡erent source of Nd than the Atlantic, Indian and Southern oceans. This value is
similar to modern Paci¢c Intermediate Water,
which receives radiogenic Nd values of V34
from the drainage of relatively young volcanic
arc terranes [29]. One of the interesting aspects
of the long-term data set is that the most radio-
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
318
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
genic ONd values analyzed were found at Caribbean Sea Site 1001. Caribbean Sea ONd values averaged V1.2, most likely re£ecting the weathering
of volcanics from the contemporaneously erupting
proto-Antilles arc [56,57].
The similarity of ONd values in the North and
South Atlantic, Indian, and Southern oceans suggests either that distinct water masses with similar
Nd isotopic compositions bathed these sites or,
alternatively, that these sites were bathed by a
common water mass. The former scenario would
imply that the di¡erent water masses received Nd
from continental sources with similar isotopic
compositions. If this were the case, any changes
in the relative contributions of such di¡erent
water masses could not be resolved using Nd isotopes. We argue, however, that despite the rather
broad range in paleo-water depths (V1900 to
V3900 m), it is plausible that the six sites we
studied sampled the same water mass. For example, in the modern oceans, the £ow of NADW can
be traced from its source southward into the
South Atlantic sector of the Southern Ocean (to
V40‡S), encompassing up to 4 km of the water
column. Even toward the southern limit of its
circulation path, NADW has a 1^2 km depth extent. Although NADW did not exist in the early
Paleogene, a water mass of similar extent would
have encompassed all of the Atlantic, Indian, and
Southern ocean deep-sea sites, given the V2 km
spread in paleo-water depths. However, the most
compelling evidence for a common water mass is
the remarkable coincidence of ONd values and
trends at the two North Atlantic Sites (401 and
549), which span a depth range of V2 km (Fig.
4).
The most cohesive paleoceanographic interpretation of the general Nd isotopic records invokes
a major source of Atlantic, Indian, and Southern
ocean deep waters in the surface waters of the
Southern Ocean between the location of the future Drake Passage and Australia (Fig. 5). We
can rule out the possibility of a contribution of
northern Paci¢c Ocean intermediate waters (ONd
V34.3) to the deep Atlantic, Indian, and Southern ocean basins given the non-radiogenic ONd
values that dominate them (ONd V38.7). The
North Atlantic could also be discounted as a sig-
ni¢cant source of deep waters, as sea£oor spreading in the Labrador Sea probably only began between magnetic Chron 31 (Maastrichtian) and
Chron 27 (early Paleocene) [17], and the Norwegian^Greenland Sea began opening during the latter stages of Chron 24 (early Eocene) (e.g. [17]). It
is unlikely that the incipient basins were sites of
volumetrically signi¢cant deep-water mass formation in the late Paleocene^early Eocene. In addition, we can rule out the North Atlantic region
south of the incipient Arctic basins as a potential
site of downwelling based on temperature constraints. SST estimates based on N18 O values of
planktonic foraminifera indicate temperatures of
V23‡C for the northeastern North Atlantic (4),
while those in the Southern Ocean recorded at
Site 690 are V11‡C and those in the Indian sector
of the Southern Ocean as warm as V14‡C (4).
Thus the Indian and Atlantic sectors of the
Southern Ocean, which were characterized by
the coolest high-latitude sea-surface waters of
the late Paleocene^early Eocene interval and
were considerably more isolated than they are today, seem to be the only likely locations of highlatitude deep-water formation. Downwelling within the Indian or South Atlantic sectors of the
Southern Ocean could supply deep waters to the
Indian Ocean as well as the Atlantic Ocean, imparting the same Nd isotopic signature to those
water masses (Fig. 5).
The slight yet systematic increase in ONd values
from the Southern Ocean to the North Atlantic
Ocean (Fig. 5) may be additional evidence of
northward £ow of deep waters from a southerly
source. Small contributions of radiogenic Nd, either from Caribbean sources or weathering of
North Atlantic Igneous Province basalts, to the
northward £owing deep waters might account
for the geographic trend. For example, simple
mixing calculations suggest that only a small
(V4%) contribution of Caribbean waters could
explain the small northward increase in Atlantic
ONd values. Alternatively, Icelandic basalts that
erupted V55 Ma (during North Atlantic Igneous
Province emplacement) have ONd values of +V7
(e.g. [17]). Thus an even smaller contribution of
waters draining these provinces would have been
required to increase ONd values, given Nd concen-
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
trations of only V20 pmol/kg in seawater at
depths of 1900^3900 m (e.g. [58]).
4.2. Thermohaline circulation during the PETM
Nd isotopic records can be used to test if a
change in thermohaline circulation was responsible for a change in deep-water conservative properties (temperature and possibly salinity) during
the PETM. A proposed scenario for such a
change is that sea-surface warming in high-latitude regions caused downwelling to cease [6].
The consequent ‘density void’ was possibly ¢lled
by warmer and more saline waters from a subtropical, evaporative region, presumably Tethys
[6,12]. These warmer waters potentially led to
the thermal dissociation of methane hydrates
and extinction of benthic organisms [6,8,9,56].
There is no substantial evidence in any of the
Nd isotopic records for a change in deep-water
mass composition at the onset of the PETM
(Fig. 4B). However, the high-resolution portions
of the records from Sites 401, 527, 549, and 690
(Fig. 3A^D) are characterized by short-term £uctuations of V1^1.5 ONd units, a feature that merits discussion. One potential cause of the shortterm ONd £uctuations is alternating contributions
from two di¡erent deep-water sources. Two likely
sources of competing deep waters during the late
Paleocene^early Eocene interval were the Southern Ocean and the Tethys. One published analysis
of an uncharacterized, uncleaned phosphate from
the early Eocene (V53 Ma) Tethys yielded an ONd
value of 38.3 [59]. If this value can be considered
representative of late Paleocene^early Eocene
Tethyan waters, then Tethyan deep waters may
have been a more radiogenic end member and
the Southern Ocean, with average ONd values of
V39.1, a more non-radiogenic end member.
While it is tempting to speculate that the shortterm £uctuations may have resulted from alternating dominance of these two deep-water sources
(Fig. 3A^D), the available evidence argues against
such a scenario. All four of these sites contain ONd
values more radiogenic than 38.3 within the
PETM portion of the records (e.g. 55.3^55.6
Ma). In addition, we argue that even if the
short-term £uctuations did have paleoceano-
319
graphic signi¢cance (i.e. millennia-scale changes
in water-mass composition), such changes could
not have caused the PETM deep- and bottomwater warming because they occur before, during,
and after the onset of the CIE. An alternative
explanation for the short-term £uctuations may
involve changes in the Nd supplied to the oceans,
a possibility discussed by Scher and Martin [60].
Future work will help constrain the source of
these short-term variations. Given the available
evidence, it is unlikely that a fundamental change
in thermohaline circulation patterns caused the
global deep- and bottom-water warming during
the PETM.
When the new Nd isotopic data are considered
in the context of high-resolution PETM stable
isotopic data from Site 690 [54], it becomes apparent that a change in thermohaline circulation
patterns may not have been necessary to generate
deep- and bottom-water warming during the
PETM. The lack of any Nd isotopic change at
the onset of the PETM is consistent with the ¢nding that PETM warming originated in the atmosphere and propagated into oceanic surface
waters and then mixed downward to thermocline
and deeper waters [58]. The lag between surfaceand deep-water warming at Site 690 suggests that
surface-water warming in regions of subduction
and downwelling would have transmitted the
thermal anomaly to intermediate and deep/bottom waters [54].
Several intriguing features emerge from the ONd
data subsequent to the PETM warming. Two possible excursions in the data, one during the recovery phase of the CIE (V55.4 Ma) and a second at
V54.3 Ma, may indicate transient changes in
thermohaline circulation patterns. Further work
is needed to test this hypothesis.
5. Conclusions
High-resolution, deep-sea Nd isotopic records
using fossil ¢sh teeth from eight DSDP and
ODP sites are used to reconstruct the nature of
late Paleocene^early Eocene thermohaline circulation. Several signi¢cant features emerge from the
data. The data indicate a common deep-water
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
320
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
mass (ONd V38.7) in the North and South Atlantic, Southern, and proto-Indian oceans (south of
the eastern Tethys), with di¡erent water masses in
the central Paci¢c Ocean (ONd V34.3) and Caribbean Sea (ONd V1.2). The Nd isotopic patterns
suggest a Southern Ocean deep-water formation
site, and these waters ¢lled the deep basins both
northward into the Atlantic Ocean and eastward
into the proto-Indian Ocean.
The Nd data do not contain evidence for a
change in the composition of deep waters prior
to the onset of the PETM. Deep- and bottomwater warming during the PETM may have
been a consequence of warming of surface waters
in downwelling regions.
[6]
[7]
[8]
[9]
[10]
Acknowledgements
We gratefully acknowledge the thorough reviews of Ellen Martin and Adina Paytan that improved the manuscript. Many thanks to Brent
Miller and Drew Coleman for thoughtful discussions. Samples were provided by the Ocean Drilling Program, and we thank the curatorial sta¡ for
all of their help. This work was funded by a JOI/
USSAC Fellowship (D.J.T.) and NSF Grant
EAR-98-14604 (T.J.B.).[KF]
[11]
[12]
[13]
[14]
[15]
References
[1] K.G. Miller, T.R. Janacek, M.E. Katz, D.J. Keil, Abyssal
circulation and benthic foraminiferal changes near the Paleocene/Eocene boundary, Paleoceanography 2 (1987)
741^761.
[2] L.D. Stott, J.P. Kennett, N.J. Shackleton, R.M. Cor¢eld,
The evolution of Antarctic surface waters during the Paleogene: inferences from the stable isotopic compositions
of planktonic foraminifers, ODP Leg 113, Proc. ODP Sci.
Results 113 (1990) 849^863.
[3] J.C. Zachos, K.C. Lohmann, J.C.G. Walker, S.W. Wise,
Abrupt climate change and transient climates during the
Paleogene: a marine perspective, J. Geol. 101 (1993) 191^
213.
[4] J.C. Zachos, L.D. Stott, K.C. Lohmann, Evolution of
early Cenozoic marine temperatures, Paleoceanography
9 (1994) 353^387.
[5] S.L. Wing, Late Paleocene-Eocene £oral and climatic
change in the Bighorn Basin, Wyoming, in: M.-P. Aubry,
S.G. Lucas, W.A. Berggren (Eds.), Late Paleocene-Eocene
[16]
[17]
[18]
[19]
[20]
Climatic and Biotic Events in the Marine and Terrestrial
Records, Columbia Press, New York, 1998, pp. 380^400.
J.P. Kennett, L.D. Stott, Abrupt deep-sea warming, palaeoceanographic changes and benthic extinctions at the
end of the Palaeocene, Nature 353 (1991) 225^229.
D.J. Thomas, T.J. Bralower, J.C. Zachos, New evidence
for subtropical warming during the late Paleocene thermal
maximum: Stable isotopes from Deep Sea Drilling Project
Site 527, Walvis Ridge, Paleoceanography 14 (1999) 561^
570.
E. Thomas, N.J. Shackleton, The Paleocene-Eocene
benthic foraminiferal extinction and stable isotope
anomalies, in: R.O. Knox et al. (Eds.), Correlation of
the Late Paleocene - early Eocene in Northwest Europe,
Geol. Soc. London Spec. Publ. 101 (1996) 401^441.
G.R. Dickens, J.R. O’Neil, D.C. Rea, R.M. Owen, Dissociation of oceanic methane hydrate as a cause of the
carbon isotope excursion at the end of the Paleocene, Paleoceanography 10 (1995) 965^971.
L.A. Frakes, E.L. Kemp, Palaeogene continental positions and evolution of climate, in: D.H. Tarling, S.K.
Runcorn (Eds.), Implications of Continental Drift to the
Earth Sciences, Academic Press, San Diego, CA, 1973, pp.
539^559.
E.J. Barron, S.L. Thompson, S.H. Schneider, An ice-free
Cretaceous? Results from climate model simulations, Science 212 (1981) 501^508.
G.W. Brass, J.R. Southam, W.H. Peterson, Warm saline
bottom waters in the ancient ocean, Nature 296 (1982)
620^623.
T.J. Crowley, Past CO2 changes and tropical sea surface
temperatures, Paleoceanography 6 (1991) 387^394.
W.S. Broecker, Paleocean circulation during the last deglaciation: a bipolar seesaw?, Paleoceanography 13 (1998)
119^121.
L.C. Sloan, J.C.G. Walker, T.C. Moore, Possible role of
oceanic heat transport in early Eocene climate, Paleoceanography 10 (1995) 347^356.
L.D. Talley, Some aspects of ocean heat transport by the
shallow, intermediate and deep overturning circulations,
Geophys. Monogr. 112 (1999) 1^22.
A.D. Saunders, J.G. Fitton, A.C. Kerr, M.J. Norry, R.W.
Kent, The North Atlantic Igneous Province, in: Large
Igneous Provinces: Continental, Oceanic, and Planetary
Flood Volcanism, AGU Geophys. Monogr. 100, 1997,
pp. 45^93.
G.S. Mountain, K.G. Miller, Seismic and geologic evidence for late Paleocene - early Eocene deepwater circulation in the western North Atlantic, Paleoceanography 7
(1992) 423^439.
A.T.S. Ramsey, T.J.S. Sykes, R.B. Kidd, Waxing (and
waning) lyrical on hiatuses: Eocene-Quaternary Indian
Ocean hiatuses as proxy indicators of water mass production, Paleoceanography 9 (1994) 857^897.
E. Thomas, Late Cretaceous - early Eocene mass extinctions in the deep sea, in: V.L. Sharpton, P. Ward (Eds.),
Global Catastrophes in Earth History: an Interdiscipli-
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
nary Conference on Impacts, Volcanism, and Mass Mortality, GSA Spec. Publ. 247, 1990, pp. 481^496.
D.K. Pak, K.G. Miller, Paleocene to Eocene benthic foraminiferal isotopes and assemblages: Implications for
deepwater circulation, Paleoceanography 7 (1992) 405^
422.
J.C. Zachos, D.K. Rea, K. Seto, N. Niitsuma, N. Nomura, Paleogene and Early Neogene Deep Water History
of the Indian Ocean: Inferences from Stable Isotopic Records. in: R.A. Duncan, D.K. Rea, R.B. Kidd, U. von
Rad, J.K. Weissel (Eds.), The Indian Ocean: A Synthesis
of Results from the Ocean Drilling Program, AGU
Monogr. 70, 1992, pp. 351^386.
S.B. O’Connell, M.A. Chandler, R. Ruedy, Implications
for the creation of warm saline deep water: late Paleocene
reconstructions and global climate model simulations,
GSA Bull. 108 (1996) 270^284.
K.L. Bice, E.J. Barron, W.H. Peterson, Continental runo¡ and early Cenozoic bottom-water sources, Geology 25
(1997) 951^954.
K.L. Bice, J. Marotzke, Numerical evidence against reversed thermohaline circulation in the warm Paleocene/
Eocene ocean, J. Geophy. Res. 106 (2001) 11529^11543.
M. Lyle, Could early Cenozoic thermohaline circulation
have warmed the poles?, Paleoceanography 12 (1997)
161^167.
K. Tachikawa, C. Jeandel, M. Roy-Barman, A new approach to the Nd residence time in the ocean: the role of
atmospheric inputs, Earth Planet. Sci. Lett. 170 (1999)
433^446.
W.S. Broecker, R. Gerard, M. Ewing, B.C. Heezen, Natural radiocarbon in the Atlantic Ocean, J. Geophys. Res.
65 (1960) 2903^2931.
D.J. Piepgras, G.J. Wasserburg, Isotopic composition of
neodymium in waters from the Drake Passage, Science
217 (1982) 207^214.
C.J. Bertram, H. Elder¢eld, The geochemical balance of
the rare earth elements and neodymium isotopes in the
oceans, Geochim. Cosmochim. Acta 57 (1993) 1957^1986.
C. Jeandel, Concentration and isotopic composition of
Nd in the South Atlantic Ocean, Earth Planet. Sci. Lett.
117 (1993) 581^591.
S.L. Goldstein, S.B. Jacobsen, Nd and Sr isotope systematics of river water suspended material: Implications for
crustal evolution, Earth Planet. Sci. Lett. 87 (1988) 249^
265.
H. Elder¢eld, M.J. Greaves, The rare earth elements in
seawater, Nature 296 (1982) 214^219.
A.N. Halliday, J.P. Davidson, P. Holden, R.M. Owen,
A.M. Olivarez, Metalliferous sediments and the scavenging residence time of Nd near hydrothermal vents, Geophys. Res. Lett. 19 (1992) 761^764.
C.E. Jones, A.N. Halliday, D.K. Rea, R.M. Owen, Neodymium isotopic variations in the North Paci¢c modern
silicate sediment and the insigni¢cance of detrital REE
contributions to seawater, Earth Planet. Sci. Lett. 127
(1994) 55^66.
321
[36] H. Elder¢eld, R. Upstill-Goddard, E.R. Sholkolvitz, The
rare earth elements in rivers, estuaries, and coastal seas
and their signi¢cance to the composition of ocean waters,
Geochim. Cosmochim. Acta 54 (1990) 971^991.
[37] E.R. Sholkovitz, The geochemistry of rare earth elements
in the Amazon River estuary, Geochim. Cosmochim.
Acta 58 (1993) 2181^2190.
[38] K.W. Burton, H.-F. Ling, R.K. O’Nions, Closure of the
Central American Isthmus and its e¡ect on deep-water
formation in the North Atlantic, Nature 386 (1997)
382^385.
[39] R.K. O’Nions, M. Frank, F. von Blanckenburg, H.-F.
Ling, Secular variation of Nd and Pb isotopes in ferromanganese crusts from the Atlantic, Indian, and Paci¢c
Oceans, Earth Planet. Sci. Lett. 155 (1998) 15^28.
[40] H.-F. Ling, K.W. Burton, R.K. O’Nions, B.S. Kamber,
F. von Blanckenburg, A.J. Gibb, J.R. Hein, Evolution of
Nd and Pb isotopes in Central Paci¢c seawater from ferromanganese crusts, Earth Planet. Sci. Lett. 146 (1997)
1^12.
[41] J. Wright, R.S. Seymour, H. Shaw, REE and Nd isotopes
in conodont apatite: Variations with geological age and
depositional environment, GSA Spec. Paper 196 (1984)
325^340.
[42] H.F. Shaw, G.J. Wasserburg, Sm-Nd in marine carbonates and phosphates, Geochim. Cosmochim. Acta 49
(1985) 503^518.
[43] H. Staudigel, P. Doyle, A. Zindler, Sr and Nd isotope
systematics in ¢sh teeth, Earth Planet. Sci. Lett. 76
(1985) 45^56.
[44] E.E. Martin, B.A. Haley, Fossil ¢sh teeth as proxies for
seawater Sr and Nd, Geochim. Cosmochim. Acta 64
(2000) 835^847.
[45] D.J. Thomas, Ph.D. Dissertation (2002).
[46] E.A. Boyle, Cadmium, zinc, copper, and barium in foraminifera tests, Earth Planet. Sci. Lett. 53 (1981) 11^35.
[47] E.A. Boyle, L.D. Keigwin, Comparison of Atlantic and
Paci¢c paleochemical records for the last 250,000 years:
changes in deep ocean circulation and chemical inventories, Earth Planet. Sci. Lett. 76 (1985) 135^150.
[48] D.J. DePaolo, G.J. Wasserburg, Nd isotopic variations
and petrogenetic models, Geophys. Res. Lett. 3 (1976)
248^252.
[49] M.-P. Aubry, W.A. Berggren, L. Stott, A. Sinha, The
upper Paleocene-lower Eocene stratigraphic record and
the Paleocene-Eocene boundary carbon isotope excursion: implications for geochronology, in: R.W.O’B.
Knox, R.M. Cor¢eld, R.E. Dunay (Eds.), Correlation of
the late Paleocene - early Eocene in Northwest Europe,
Geol. Soc. Spec. Publ. 101 (1996) 353^380.
[50] W.A. Berggren, M.-P. Aubry, A late Paleocene-early Eocene NW European and North Sea magnetobiochronological correlation network, in: R.W.O’B. Knox, R.M.
Cor¢eld, R.E. Dunay (Eds.), Correlation of the late Paleocene - early Eocene in Northwest Europe, Geol. Soc.
Spec. Publ. 101 (1996) 309^352.
[51] R.N. Norris, U. Ro«hl, Carbon cycling and chronology of
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart
322
[52]
[53]
[54]
[55]
D.J. Thomas et al. / Earth and Planetary Science Letters 209 (2003) 309^322
climate warming during the Paleocene/Eocene transition,
Nature 358 (1999) 319^322.
U. Ro«hl, T.J. Bralower, R.N. Norris, G. Wefer, A new
chronology for the late Paleocene thermal maximum and
its environmental implications, Geology 28 (2000) 927^
930.
S. Bains, R.M. Cor¢eld, R.N. Norris, Mechanisms of
climate warming at the end of the Paleocene, Science
285 (1999) 724^727.
D.J. Thomas, J.C. Zachos, T.J. Bralower, E. Thomas, S.
Bohaty, Warming the fuel for the ¢re: Evidence for the
thermal dissociation of methane hydrate during the Paleocene-Eocene thermal maximum, Geology 30 (2002) 1067^
1070.
W.A. Berggren, D.V. Kent, C.C. Swisher III, M.-P.
Aubry, A revised Cenozoic geochronology and chronostratigraphy, in: W.A. Berggren, D.V. Kent, M.-P.
Aubry, J. Hardenbol (Eds.), Geochronology, Time Scales
and Global Stratigraphic Correlations: Framework for an
Historical Geology, SEPM Spec. Publ. 54 (1995) 129^212.
[56] T.J. Bralower, D.J. Thomas, J.C. Zachos, M.M. Hirschmann, U. Rohl, H. Sigurdsson, E. Thomas, D.L. Whitney, High-resolution records of late Paleocene thermal
maximum and circum-Caribbean volcanism: Is there a
causal link?, Geology 25 (1997) 963^966.
[57] H. Sigurdsson, R.M. Leckie, G. Acton, Proc. ODP Init.
Rep. 165 (1997).
[58] D.J. Piepgras, G.J. Wasserburg, Rare earth element transport in the western North Atlantic inferred from Nd isotopic observations, Geochim. Cosmochim. Acta 51 (1987)
1257^1271.
[59] P. Stille, M. Steinmann, S.R. Riggs, Nd isotope evidence
for the evolution of the paleocurrents in the Atlantic and
Tethys Oceans during the past 180 Ma, Earth Planet. Sci.
Lett. 144 (1996) 9^19.
[60] H. Scher, E.E. Martin, Eocene to Miocene Southern
Ocean deep water circulation revealed from fossil ¢sh
teeth Nd isotopes, EOS Trans. AGU 82, Fall Meet.
Suppl., Abstr. F 639 (2001).
EPSL 6588 2-4-03 Cyaan Magenta Geel Zwart