Geochimica et Cosmochimica Acta, Vol. 64, No. 14, pp. 2401–2416, 2000
Copyright © 2000 Elsevier Science Ltd
Printed in the USA. All rights reserved
0016-7037/00 $20.00 ⫹ .00
Pergamon
PII S0016-7037(99)00422-6
U-Th dating of deep-sea corals
HAI CHENG,1 JESS ADKINS,2,* R. LAWRENCE EDWARDS,1 and EDWARD A. BOYLE3
1
3
Minnesota Isotope Laboratory, Department of Geology and Geophysics, University of Minnesota, Minneapolis, MN 55455 USA
2
Geochemistry 62, Lamont-Doherty Earth Observatory, Rt. 9W, Palisades, NY 10964 USA
Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139-4307 USA
(Received February 24, 1999; accepted in revised form November 16, 1999)
Abstract—230Th, 232Th, 234U and 238U compositions of several deep-sea solitary corals, mainly the species
Desmophyllum cristagalli, were determined by thermal ionization mass spectrometry (TIMS). It is possible to
obtain high precision ages on modern pristine corals that have low [232Th] (5 to a few hundred ppt). However,
because older deep-sea corals tend to have higher [232Th] compared to surface corals, and the initial
230
Th/232Th ratio is uncertain, older deep-sea corals have larger age uncertainties (⫾several hundred years for
samples with a few thousand ppt 232Th). Therefore, the key hurdle for precise U-Th dating is to remove or
account for contaminants which contain elevated 232Th and associated 230Th not due to closed system decay
within the coral lattice. A modification of the trace metal cleaning methods used for foraminifera and surface
corals can significantly reduce this contamination. By counting the visible growth bands and measuring the
mean age of a single septum, the extension rate of D. cristagalli was estimated to be between 0.1 and 3.1
mm/year. In a mean sense, bands appear to be precipitated annually, but this estimate has a large uncertainty.
If appropriate tracer calibrations can be established, these corals are therefore suitable to record decadal or
sub-decadal oceanographic changes over the course of their lifetime. The ␦234U values of all modern samples
from different localities and different depths are similar (mean 145.5 ⫾ 2.3‰) and indistinguishable from the
data obtained from surface corals. At a precision of about ⫾2‰, we find no structure in the oceanic profile
of ␦234U ratios over the top 2000 m of the water column. Copyright © 2000 Elsevier Science Ltd
sands of samples. While reported depth habitats of deep-sea
coral range from 60 – 6000 meters (Cairns and Stanley, 1981),
most specimens are found between 500 –2000 meters. This
depth range makes deep-sea corals ideal for studies of intermediate and upper deep water masses. The great potential of
this archive stems from the fact that the density banded coral
skeletons are not subject to bioturbation and the fact that the
skeletons are potentially datable by uranium-series techniques
(Cheng et al., 1995; Goldstein et al., 1996; Smith et al., 1997).
Thus, deep-sea corals can provide absolute-dated records with
temporal resolutions not generally attainable in deep-sea sediments and are a new archive of deep circulation rates by
coupling U-series ages with 14C dates (Adkins and Boyle,
1998; Mangini et al., 1998). However, realization of this potential requires evaluating the U-Th dating systematics (Lomitschka and Mangini, 1999) and growth rates of deep-sea
corals. This study focuses on the ubiquitous pseudo-colonial
species, Desmophyllum cristagalli, using both a modern and a
fossil sample set.
Several previous studies have examined deep-sea coral
growth rate and its relationship to species specific density
banding. Duncan (1877) and Pratje (1924) reported extension
rates of 6.8 and 7.5 mm/year for specimens of the genus
Lophelia that were attached to underwater transatlantic cables.
Teichert (1958) also estimated growth rates of 7.5–15 mm/year
for Lophelia. Grigg (1974) determined a value of about 20
mm/year vertically and 3 mm/year horizontally for two gorgonains by tagging colonies of Muricea californica and Muricea
furticosa from a relatively shallow depth of 14 –20 m. He
suggested that the periodicity of growth ring formation was
annual based on matching the estimated ages to the number of
growth bands. Using the 210Pb dating method, Druffel et al.
1. INTRODUCTION
Uranium series dates from surface corals provide constraints
on several late Quaternary climate processes. Past sea level
estimates from ␣-counted 230Th dates on raised coral terraces
from Barbados supported the Milankovitch hypothesis on the
relation between glacier ice volume and insolation (Broecker et
al., 1968; Mesolella et al.,1969). Later, more precise and accurate thermal ionization mass spectrometry (TIMS) confirmed
this result in a variety of locations (Edwards et al., 1987a,b;
Gallup et al., 1994; Muhs et al., 1994; Stirling et al., 1995;
Szabo et al., 1994). Drill cores of submerged corals that have
been precisely dated by 230Th methods contain our most detailed and continuous record of sea-level since the last glacial
maximum (Bard et al., 1993; Bard et al., 1996; Edwards et al.,
1993). Coupled high precision radiocarbon and uranium series
dates from surface corals constrain the history of atmospheric
⌬14C beyond the tree ring calibration (Bard et al., 1993; Bard
et al., 1990; Edwards et al., 1993). In addition, TIMS dates
provide precise ages for coral tracer-based studies of past
oceanographic conditions (Beck et al., 1997; Beck et al., 1992;
Gagan et al., 1998; Guilderson et al., 1994; McCulloch et al.,
1996).
Analogous studies of deep-sea corals promise to constrain a
number of deep ocean processes (Adkins et al., 1998; Mangini
et al., 1998; Smith et al., 1997). Several cosmopolitan genera of
deep-sea scleractinia inhabit all major ocean basins. Coral has
been dredged from the ocean floor since at least the days of the
Challenger expedition, so the world’s collections contain thou*Author to whom correspondence should be addressed: MS 100-23
Caltech 1200 E. California Blvd., Pasadena, CA 91125-0001 (jess@
gps.caltech.edu).
2401
2402
Cheng et al.
Table 1. Deep sea coral samples used in this study.
Sample
set
Sample
number
Locality
Analysis
number
Collection
time
Depth
(m)
Lat
Long
DC-2 a, b
DC-4 a, b
DC-3 a, b
DC-5
DC-1
Jan. 21, 1964
Nov. 23, 1986
Feb. 25, 1990
Oct. 28, 1973
Apr. 16, 1987
421
806
990–1150
1420–1470
2110–2180
50°38⬘S
0°14⬘N
43°47S
48°40⬘N
38°45⬘N
167°38⬘E
91°36⬘W
150°29⬘E
10°54⬘W
72°39⬘W
1684–1829
1684–1829
1784
1954
1954
1954
42°N
42°N
38°N
38°N
38°N
38°N
29°W
29°W
60°W
62°W
62°W
62°W
Coral species
Solitary
or colonial
Modern
47413
84820
85080
48740
78459
Desmophyllum
Desmophyllum
Desmophyllum
Desmophyllum
Desmophyllum
cristagalli
cristagalli
cristagalli
cristagalli
cristagalli
Solitary
Solitary
Solitary
Solitary
Solitary
?
Solenosmillia sp.
Desmophyllum cristagalli
Desmophyllum cristagalli
Desmophyllum cristagalli
Desmophyllum cristagalli
?
Colonial
Solitary
Solitary
Solitary
Solitary
Paleo
JFA-2
JFA-17
JFA-24C
JFA-20A
JFA-20B
JFA-20C
2
17
24
20A
20B
20C
(1990) calculated a mean value of 0.11 ⫾ 0.02 mm/year in
width for the calcitic deep-sea gorgonian Corallium niobe,
which grew at 600 m depth. In that study, the periodicity of
growth band formation did not appear to be annual. Our approach is to use U-series disequilibria to constrain dating and
growth systematics in several deep-sea corals.
The behavior of 238U, 234U, 230Th, and 232Th in reef-building corals and the suitability of reef-building corals for 230Th
dating is the subject of an extensive literature starting with
Barnes et al. (1956) and continuing up to the present, see
Ivanovich and Harmon (1992) and references therein. The
behavior of uranium-series nuclides in surface-dwelling solitary corals has also received attention (Ku and Kern, 1974;
Muhs et al., 1994; Stein et al., 1991; Szabo, 1985). However,
similar studies on deep-sea corals are generally lacking (Adkins
et al., 1998; Cheng et al., 1995; Goldstein et al., 1996; Smith et
al., 1997; Lomitschka and Mangini, 1999). In many respects,
one would expect the U-Th dating systematics of deep-sea
corals to be similar to those of their surface-dwelling counterparts. However, deep-sea coral systematics should differ in at
least two key respects. First, deep ocean waters have much
higher 230Th concentrations than surface waters. Thus, deepsea corals may incorporate a significant amount of unsupported
230
Th during growth. If so, a correction for initial 230Th must
be made, and the precision and accuracy of the 230Th age will
depend on the precision and accuracy of the correction. Second,
the environments in which deep-sea corals age are quite different
from those in which surface corals age. Thus, diagenetic processes
may affect corals in the two settings in different ways.
In this study, we investigate the U-Th dating systematics of
several species of aragonitic deep-sea corals, focusing on Desmophyllum Cristagalli, which is the most abundant coral in our
collection of dredged specimens. We establish growth rates and
initial chemical and isotopic characteristics by analyzing 238U,
234
U, 230Th, and 232Th concentrations in modern samples. We
constrain diagenetic shifts in the pertinent nuclides by comparing the isotopic characteristics of the modern specimens to
those of fossil samples, and by analyzing coatings on the
specimens. By analyzing different sub-samples of the same
coral, we show that deep-sea corals generally have significant
amounts of initial and/or added thorium. We investigate
sources of this thorium, as well as methods to remove or correct
for initial 230Th. Our results are specifically applicable to the
study of Adkins et al. (1998), which established that deepocean circulation changed dramatically in less than several
decades at 15.4 ka. The time and the duration of the circulation
change is established with uranium-series data and interpretations presented here. Beyond this specific application, we envision that our results will be more generally applicable to
further studies in the emerging field of deep-sea coral paleoceanography.
2. SAMPLES AND METHODS
2.1. Samples
Deep-sea coral samples were provided by the Smithsonian Institution and the Woods Hole Oceanographic Institution dredge collection.
Five modern samples of D. cristagalli were collected by dredging
programs in the Pacific, the Atlantic, and the Southern Oceans over the
past 30 years (Table 1 and Fig. 1). Sample depths range from 420 to
2200 m. Samples were judged to be modern either because of intact
organic matter, noted upon recovery, or very fresh looking preservation
of the septa in the dredge collections. Prior to cleaning, the modern
samples often were covered with a yellowish organic coating. Six fossil
samples were dredged from the Atlantic between depths of 1700 to
2000 m (Table 1). All specimens are D. cristagalli except for one
Solenosmilia sp. specimen and one unidentified sample. As opposed to
the modern samples, the fossil samples are almost always covered with
a black crust. Energy-dispersive electron microprobe analysis indicates
that this crust is composed of iron and manganese oxides mixed with
detrital aluminosilicates (Fig. 2a). It is also possible to find discrete
metal sulfide deposits in the crust’s matrix (Figure 2b).
2.2. Sample Cutting
Samples were cut into small slabs parallel to the radially symmetric
septa, which are connected around the outer rim by a thick-walled theca
(Fig. 3). Septa in individual deep-sea corals are classified on the basis
of their relative cross-sectional areas in a radial plane, where the S1
septa have the largest cross-sectional areas (Fig. 3, refer to (Cairns,
1981) for classification of septa). Each slab generally consists of one
whole S1 septum associated with either one or two smaller septa on
both sides of the S1. S1 septa typically have tens of density bands
sub-parallel to their upper surfaces (Fig. 4). The thickness of individual
septa are about 0.1– 0.7 mm, with the S1 being the thickest. Each
septum’s thickness decreases vertically from a bulge near the top and
increases from the interior towards the theca. After cutting, the slabs
were set aside for one of two cleaning procedures described below.
Slabs of the five modern D. cristagalli samples were further subsampled into smaller pieces roughly parallel to their growth bands
using a small blade (Fig. 3). Final weights of these modern coral
sub-samples ranged from 28.9 to 143.3 mg.
U-Th dating of deep-sea corals
2403
Fig. 1. Sample locations used in this study. Sample depths, in parentheses, are in meters.
2.3. Ultrasonic Cleaning
All sub-samples of the five modern samples (Table 2) and some
sub-samples of the six fossil corals (labeled “N” in Table 3) were
cleaned with the following ultrasonic cleaning technique. Under a
binocular microscope, samples were checked for organic coatings or
iron-manganese oxide crusts, which were scrubbed with small dental
tools. Samples were put in a plastic bottle, ultrasonically cleaned in
deionized water for 10 min. and then completely rinsed. This process
was repeated several times until the coral looked clean (free of crusts)
under the microscope. The above procedure was then repeated three
times in a Teflon beaker with 5 min. of ultrasonic cleaning each time.
Finally, the sample was dried in an oven at about 70°C and set aside for
isotopic analysis. Some of the yellowish organic material from modern
samples 85080 and 78459 (labeled “D” in Table 2) and some of the
iron-manganese crust from fossil samples JFA 2 and JFA 24C (also
labeled “D” in Table 3) were also saved for isotopic analysis.
2.4. Chemical Cleaning
Fig. 2. Electron micro-probe analysis of black crusts from a fossil
sample of D. cristagalli. The bulk crust (a) is a mixture of detrital
minerals and authegenic iron and manganese oxides. Discrete sulfide
mineral phases (b) are also found in the crust matrix. This figure is
meant as a description of some of the contaminants found in coral
crusts, not as a quantitative measure of contamination.
Early in this project it became clear that one of the major limiting
factors in U-Th dating of deep-sea corals would be the ability to
accurately and precisely account for initial and added thorium. Removal of thorium on or near coral surfaces can help solve this problem.
We therefore modified chemical and physical cleaning techniques
developed some years ago at M.I.T. for trace metal analysis (Boyle,
1981; Boyle and Keigwin, 1985/6; Shen and Boyle, 1988) and applied
them to the remaining modern and fossil sub-samples. As has been
demonstrated previously for surface corals and foraminifera, these
techniques can remove exterior contaminants from fossil samples and
improve accuracy and reproducibility. We followed the procedure of
Shen and Boyle (1988) except that the pre-cleaning step was altered to
better remove the black crusts. After scrubbing with a brush and
deionized water, samples were placed in plastic tubes with clean
distilled H2O and ultrasonicated for 15 min. Corals were then submerged in a 50/50 mixture of 30% H2O2 and 1M NaOH for 15 min.
with ultrasonication. This step was repeated several times until there
was little crust left. Occasionally samples were scrubbed with a brush
between oxidative leaches to remove loose material. The last step was
a brief (30 sec. to 2 min.) rinse in a 50/50 mixture of 30% peroxide and
1% HClO4. This step effectively removes all organic stains left on the
coral but also removes about 5–10% of the skeleton. The cleaning
solution in this last step always had a pH ⬍ 2. Samples were then
thoroughly rinsed with clean distilled water and left to dry in a HEPA
filtered laminar flow bench. The visible iron-magnesium oxide crusts are
easily removed with these oxidizing steps, flaking off in small sheets. This
suggests that the crust is bound by an organic “glue”, possibly remnant
polyp organic matter. Subsequent cleaning followed the method of Shen
and Boyle (1988) and used clean reagents throughout.
2404
Cheng et al.
Fig. 3. (a) A typical side view of the deep-sea solitary coral Desmophyllum cristagalli. The drawing is adapted from
(Cairns, 1981). (b) The location of sub-samples from the five modern D. cristagalli. Each sub-sample includes a main
septum (S1) and one or two smaller septa. The fine dashed lines indicate the pattern of growth bands. The sample analysis
numbers are on the right side of each sub-sample and the heavy dashed lines are their boundaries. Where there are multiple
septa analyzed from the same coral (see Table 2), septum “a” is the one pictured. The middle piece of 85080 was lost during
sampling.
2.5. Thermal Ionization Mass Spectrometry
Procedures for chemical separation and instrumental analysis of
uranium and thorium are modifications of those previously described
for surface corals (Chen et al., 1986; Edwards et al., 1993; Edwards,
1988; Edwards et al., 1987b). Ultrasonically or chemically cleaned
samples were slowly dissolved in nitric acid. This was sufficient to
dissolve all samples completely, except for a residue in the ironmanganese oxide crust samples. The residue, presumably detrital aluminosilicate contained in the crust, dissolved readily in hydrofluoric
acid. The solutions were spiked with 233U-236U (233U/236U ⫽
1.010527) and 229Th solutions of known concentration. Following the
chemical separation of thorium and uranium, the thorium fraction was
loaded on a graphite-coated single Re filament, which had been previously checked for its thorium blank. The filament blank for 232Th is
generally 100 –150 counts per second at around 1700°C and increases
with increasing temperature. This is a significant fraction of the total
232
Th beam in typical sample runs, particularly for the runs on small
samples with relatively low 232Th concentrations. We corrected for this
blank assuming an uncertainty of ⫾75% in the value of the filament
blank. In order to minimize the filament blank, we measured 232Th at
temperatures under 1700°C; nevertheless, as has been shown previously (Edwards et al., 1987b), the uncertainty in the filament blank
correction is the main source of error in the 232Th measurements. The
uranium fraction was loaded on a Re filament without graphite and run
with the double-filament technique. Uranium and thorium were measured on the Minnesota Isotope Laboratory’s Finnigan-MAT 262-RPQ
mass spectrometer on the first stage electron multiplier prior to the
static quadrupole second stage. Tails were accounted for by measuring
count rates at half masses.
3. RESULTS AND DISCUSSION
3.1. Reproducibility
Uranium and thorium isotope compositions of modern and
fossil samples are presented in Tables 2 and 3 respectively. We
measured a number of samples in replicate. In one case we
analyzed separate aliquots of the same solution (Table 2: DC-1
T(I) and DC-1 T(II)). These analyses agreed within error for the
230
Th age and all uranium and thorium concentrations and
isotope ratios. In four cases (Table 3), we analyzed two separate septa from the same fossil coral that were cleaned using the
same method. Three pairs were treated using the ultrasonic
cleaning method; JFA 2 (N) I and II, JFA 24C (N) I and II, and
JFA 20C (N) I and II, and one pair was treated using the
chemical cleaning method; JFA 24C I and II. In all four cases,
the 238U and 232Th concentrations did not replicate within
analytical errors, indicating that concentrations of these nuclides differ in separate sub-samples of the same coral. In three
of the cases the 230Th ages, uncorrected for initial 230Th, agreed
within analytical errors. In the fourth case (the JFA 24C (N)
pair), the uncorrected 230Th ages differed significantly. In this
case, the 232Th concentrations differed by more than a factor of
two and the 230Th ages quite likely differed because of differing initial 230Th contents (see discussion below). In all four
cases the measured ␦234U values (defined as ((234U/238U)sample/
(234U/238U)standard⫺1) ⫻1000) replicated within errors. In sum,
U-Th dating of deep-sea corals
2405
Fig. 4. Transmitted light image of the side view of modern D. cristagalli sample number 47413. This single septum was
placed in a standard photographic enlarger to expose the density banding structure. X-rays were not required to see the
bands. Alternating light and dark density bands are visible sub-parallel to the top edge of the septum. The white area on the
bottom is the thickened portion where the septum connects to the rest of the skeleton around the outer edge of the coral.
White flecks are from dust on the sample. Bands can be seen with the naked eye but are more clear in transmitted light.
replicate analyses agreed within analytical error in all cases
where one would expect agreement.
3.2. Modern Samples
3.2.1. Uranium concentrations and isotopic ratios
Ku et al. (1977) and Chen et al. (1986) demonstrated that
within several percent uranium is conservative in seawater. Our
measured uranium concentrations in modern D. cristagalli, on
the other hand, vary from 2963 to 5531 ppb (Table 2). These
uranium values are higher than those of most hermatypic corals
(e.g., 2– 4 ppm, (Burnett and Veeh, 1992)) and are similar to
those of other ahermatypic corals (Stein et al., 1991; Thompson
and Livingston, 1970; Lomitschka and Mangini, 1999). In
surface corals, uranium variations are correlated to changes in
sea surface temperature (Min et al., 1995; Shen and Dunbar,
1995) . However, for a single septum of D. cristagalli, the
measured range of uranium concentration is larger than the
surface coral data, while the deep water temperature variations
are much smaller. Clearly there is a combination of other
environmental parameters, such as temperature, pH, carbonate
ion concentration, growth rate, and/or a biologically induced
“vital effect” affecting the D. cristagalli uranium concentration
(Gvirtzman et al., 1973). In general, the primary uranium
concentrations in deep-sea corals are higher than those of most
other biogenic deep-sea carbonates. For example, foraminifera
shells have primary uranium concentrations about 200 times
lower than deep-sea corals (Delaney and Boyle, 1983; Henderson and O’Nions, 1995; Ku, 1965; Russell et al., 1994) . These
high uranium contents make deep-sea corals ideal candidates
for uranium-series dating.
The ␦234U value for modern hermatypic corals measured in
the Minnesota lab is 145.8 ⫾ 1.9‰ (2␴) (Cheng et al., 1999;
Edwards et al., 1993). This ratio is about 3‰ lower than earlier
reports because we have used the new ␭234 value determined by
Cheng et al. (1999). Each modern initial ␦234U value for our
deep-sea corals is within error of the surface coral value (Fig.
5, Table 2). The mean and two sigma error of all 20 values is
145.3 ⫾ 2.3‰. As the deep-sea samples were collected from
nine different depths across the Pacific, Atlantic, Indian and
Southern Oceans, these data indicate that the sea water ␦234U
value is conservative in the upper 2000 meters of the world’s
oceans within about ⫾2‰. In addition, surface coral data
indicate that since the last glacial maximum (Bard et al., 1993;
Edwards et al., 1993) and for sea level high stands over the past
200,000 years (Gallup et al., 1994; Gallup and Edwards, 1997)
the ␦234U of surface seawater has been the same as the modern
value within several per mil (Henderson et al., 1993). Since
seawater ␦234U appears to be temporally and spatially constant,
the ␦234U values in fossil deep-sea corals may be used as a
check of diagenetic alteration.
2406
Cheng et al.
Table 2. U and Th isotopic composition and
Sample Analysis Weight
number number (mg)
47413
84820
85080
48740
78459
DC-2a
T
M1
M2
B
DC-2b
T
M
DC-4a
T
M
B
DC-4b
T
B
DC-3a
T
B
D
DC-3b
T
B
DC-5
T
B
DC-1
T(I)
T(II)
B
D
238
U
(ppb)
232
Th
(ppb)
230
Th ages of modern solitary deep sea corals.
230
Th/232Th
(atomic, ppm)
␦234U
(measured)
145.1 ⫾ 1.5
145.8 ⫾ 1.6
147.2 ⫾ 1.1
145.4 ⫾ 2.9
Th/238U
(activity)
Age
(yearb)
␦234U
(initial)
0.000406 ⫾ 0.000047
0.000461 ⫾ 0.000058
0.000507 ⫾ 0.000058
0.000629 ⫾ 0.000069
7.7 ⫾ 4.5
12.9 ⫾ 5.5
17.2 ⫾ 5.5
28.9 ⫾ 6.5
145.1 ⫾ 1.5
145.8 ⫾ 1.6
147.2 ⫾ 1.1
145.4 ⫾ 2.9
230
129.19
114.24
100.13
81.09
3305 ⫾ 2
2963 ⫾ 2
3088 ⫾ 2
3257 ⫾ 2
5 ⫾ 18
40 ⫾ 21
52 ⫾ 24
153 ⫾ 31
4296 ⫾ 15050
570 ⫾ 304
501 ⫾ 242
221 ⫾ 56
43.87
47.57
3003 ⫾ 2
3024 ⫾ 15
63 ⫾ 41
69 ⫾ 49
344 ⫾ 244
386 ⫾ 292
72.04
66.05
139.90
5232 ⫾ 2
5335 ⫾ 3
4898 ⫾ 2
139 ⫾ 35
107 ⫾ 37
76 ⫾ 27
1028 ⫾ 259
1138 ⫾ 397
1667 ⫾ 607
144.5 ⫾ 1.2 0.001655 ⫾ 0.000072 149.5 ⫾ 6.8
145.1 ⫾ 1.7 0.001380 ⫾ 0.000063 123.3 ⫾ 6.0
143.6 ⫾ 1.3 0.001558 ⫾ 0.000096 140.4 ⫾ 9.1
144.6 ⫾ 1.2
145.1 ⫾ 1.7
143.6 ⫾ 1.3
41.30
40.30
5531 ⫾ 2
4893 ⫾ 3
140 ⫾ 63
109 ⫾ 58
1134 ⫾ 512
1282 ⫾ 689
145.5 ⫾ 1.3 0.001743 ⫾ 0.000097 157.8 ⫾ 9.2
144.0 ⫾ 1.5 0.001724 ⫾ 0.000096 156.2 ⫾ 9.1
145.5 ⫾ 1.3
144.0 ⫾ 1.5
135.71 3729 ⫾ 2
58.60 3779 ⫾ 2
0.81 21104 ⫾ 36
66 ⫾ 17
373 ⫾ 103
2293 ⫾ 42
154 ⫾ 5
1.38 ⫻ 105 ⫾ 3 ⫻ 103 90.3 ⫾ 21.1
10.7 ⫾ 11.8 144.6 ⫾ 1.9
22.3 ⫾ 12.7 143.8 ⫾ 5.2
145.0 ⫾ 1.2 0.000399 ⫾ 0.000037 33.0 ⫾ 3.5 145.0 ⫾ 1.2
145.4 ⫾ 1.4 0.005672 ⫾ 0.000150 535.5 ⫾ 14.3 145.7 ⫾ 1.4
108.1 ⫾ 36 0.035705 ⫾ 0.008356
78.06
94.52
3520 ⫾ 2
3208 ⫾ 2
21 ⫾ 30
159 ⫾ 26
95.62
28.89
3674 ⫾ 2
3769 ⫾ 4
971 ⫾ 37
2196 ⫾ 102
91 ⫾ 4
69 ⫾ 5
146.4 ⫾ 1.5 0.001456 ⫾ 0.000093 117.1 ⫾ 8.8 146.4 ⫾ 1.5
142.8 ⫾ 3.2 0.002425 ⫾ 0.000183 210.0 ⫾ 17.0 142.9 ⫾ 3.2
1438 ⫾ 67
1374 ⫾ 24
1112 ⫾ 35
4 ⫻ 105 ⫾ 3 ⫻ 103
122 ⫾ 9
132 ⫾ 4
128 ⫾ 5
12 ⫾ 1
146.3 ⫾ 1.2 0.002871 ⫾ 0.000150 265.3 ⫾ 14.2 146.4 ⫾ 1.2
145.7 ⫾ 1.1 0.002967 ⫾ 0.000064 274.6 ⫾ 6.1 145.8 ⫾ 1.1
156.7 ⫾ 1.2 0.002445 ⫾ 0.000048 224.6 ⫾ 4.7 146.8 ⫾ 1.2
143 ⫾ 126 0.07775 ⫾ 0.00701
354.60
143.30
13.00
3711 ⫾ 2
3710 ⫾ 2
3516 ⫾ 2
3815 ⫾ 221
897 ⫾ 1303
316 ⫾ 63
144.6 ⫾ 1.9 0.000437 ⫾ 0.000124
143.8 ⫾ 5.2 0.000559 ⫾ 0.000138
146.0 ⫾ 2.2 0.000323 ⫾ 0.000068
146.6 ⫾ 1.4 0.000889 ⫾ 0.000114
25.8 ⫾ 6.5 146.0 ⫾ 2.2
79.5 ⫾ 10.8 146.6 ⫾ 1.4
Uranium series data of modern D. cristagalli used in this study. Samples labeled “a” and “b” are different septa from the same individual. T(I) and
T(II) are different aliquots of the same sample after it was dissolved. T, M and B refer to the Top, Middle and Bottom of septa respectively as shown
in Figure 3. (D)— organic material scrubbed off the surface of the coral. Half-lives of 234U and 230Th used in the calculations are 245,250 ⫾ 490 yr
and 75,690 ⫾ 230 yr respectively (Cheng et al., 2000). The decay constant for 238U is 1.551 ⫻ 10⫺10 yr⫺1 (Jaffey et al., 1971). Ages are calculated
according to equation 1 with no correction for initial 230Th. All errors are 2␴. b Age before collection.
3.2.2. Initial
230
Th
230
As opposed to uranium, thorium is not conservative in seawater. Since both 230Th concentrations and 230Th/232Th ratios
increase with depth (Anderson et al., 1983; Bacon and Anderson, 1982; Cochran, 1992; Cochran et al., 1987; Guo et al.,
1995; Huh and Beasley, 1987; Huh et al., 1989; Moran et al.,
1995; Nozaki and Horibe, 1983; Nozaki et al., 1987; RoyBarman et al., 1996), we expect much higher initial 230Th
concentrations in deep sea corals than in surface corals. Therefore, we need to estimate the initial 230Th/238U ratio for deep-sea
corals in order to calculate their 230Th ages. We can write the
230
Th age equation, including the term for initial 230Th/238U:
230
Th
⫽1⫹
238
U
⫹
冉冉 冊 冊
冉
冊
230
Th
238
U
␦ 234U共0兲
1000
0
⫺ 1 e ⫺␭230t
␭ 230
共1 ⫺ e 共␭234⫺␭230兲t兲
␭ 230 ⫺ ␭ 234
(1)
where all isotope ratios are activity ratios, the ␭’s are decay
constants, t is the 230Th age, (230Th/238U)° is the term for initial
230
Th/238U, and ␦234U(0) is as defined previously (Edwards et
al., 1987b). We can substitute (232Th/238U)(230Th/232Th)° for
(230Th/238U)°:
Th
⫽1⫹
238
U
⫹
冉冉 冊冉 冊 冊
冉
冊
232
Th
238
U
␦ 234U共0兲
1000
230
Th
232
Th
0
⫺ 1 e ⫺␭230t
␭ 230
共1 ⫺ e 共␭234⫺␭230兲t兲
␭ 230 ⫺ ␭ 234
(2)
As (232Th/238U) can be measured, our problem reduces to one
of estimating the initial 230Th/232Th ratio, an issue that we will
discuss at length below. Because (230Th/232Th)° cannot be
determined exactly, the error in its estimation introduces error
in t, the 230Th age. We will show that this uncertainty is the
main source of error in the 230Th age for most deep-sea corals.
In general one might expect two possible sources of initial
thorium: one from the water column, similar to the “hydrogenous” component of Lin et al. (1996), and one from continentially derived detrital particles. Both dissolved and particulate
sea water 230Th/232Th generally increase with depth and water
mass age. Surface waters are as low as about 5–10 ⫻ 10⫺6
(atomic ratio) (Cochran et al., 1987; Guo et al., 1995; Nozaki et
al., 1987), while maximum values in the deep Pacific are
⬎700 ⫻ 10⫺6 (Moore, 1981). On the other hand, the thorium
isotopic composition of detrital clays is expected to be close to
the bulk earth value of 4 ⫻ 10⫺6, which is based on a 232Th/
238
U value of 3.8 and the assumption of secular equilibrium
2
2 (N) (I)
2 (N) (II)
2 (N) (D)
17
24 (I)
24 (II)
24 (N) (I)
24 (N) (II)
24 (N) (D)
20 A
20 A (N)
20 B
20 B (N)
20 C
20 C (N) (I)
20 C (N) (II)
Analysis
number
Th
(ppt)
1723 ⫾ 26
4217 ⫾ 49
6749 ⫾ 43
757300 ⫾ 9350
877 ⫾ 22
2901 ⫾ 26
3255 ⫾ 36
7960 ⫾ 53
19231 ⫾ 176
4511000 ⫾ 37800
1623 ⫾ 24
8472 ⫾ 48
4809 ⫾ 43
6555 ⫾ 35
856 ⫾ 17
2108 ⫾ 36
3043 ⫾ 69
3752 ⫾ 2
4059 ⫾ 2
4196 ⫾ 2
4114 ⫾ 4
3851 ⫾ 1
3740 ⫾ 3
3858 ⫾ 1
3293 ⫾ 2
3521 ⫾ 5
2941 ⫾ 40
4524 ⫾ 2
4935 ⫾ 3
4507 ⫾ 1
4473 ⫾ 2
3114 ⫾ 2
3109 ⫾ 1
3319 ⫾ 25
232
238
U
(ppb)
4936 ⫾ 78
2195 ⫾ 28
1421 ⫾ 16
102 ⫾ 2
9372 ⫾ 232
3317 ⫾ 31
3020 ⫾ 35
1084 ⫾ 9
502 ⫾ 5
48 ⫾ 1
23470 ⫾ 343
4943 ⫾ 37
6834 ⫾ 64
5045 ⫾ 29
30920 ⫾ 612
12501 ⫾ 213
9086 ⫾ 212
230
Th/232Th
(atomic ⫻ 10⫺6)
141.3 ⫾ 1.5
142.5 ⫾ 1.3
141.5 ⫾ 1.4
143.2 ⫾ 3.0
143.6 ⫾ 1.3
144.5 ⫾ 2.8
141.3 ⫾ 1.3
144.8 ⫾ 2.1
145.2 ⫾ 4.1
206.3 ⫾ 79
116.6 ⫾ 1.2
119.7 ⫾ 1.3
119.7 ⫾ 1.2
120.9 ⫾ 1.3
128.8 ⫾ 1.0
142.7 ⫾ 1.4
140.3 ⫾ 8.7
␦234U
(measured)
0.13731 ⫾ 0.00068
0.13842 ⫾ 0.00076
0.13814 ⫾ 0.00124
1.1357 ⫾ 0.0172
0.12923 ⫾ 0.00057
0.15581 ⫾ 0.00058
0.15435 ⫾ 0.00059
0.15870 ⫾ 0.00064
0.16612 ⫾ 0.00092
4.4194 ⫾ 0.0687
0.50999 ⫾ 0.00133
0.51388 ⫾ 0.00248
0.44160 ⫾ 0.00151
0.44781 ⫾ 0.00135
0.51485 ⫾ 0.00166
0.51334 ⫾ 0.00144
0.50461 ⫾ 0.00462
Th/238U
(activity)
230
147.0 ⫾ 1.6
148.3 ⫾ 1.3
147.2 ⫾ 1.5
149.0 ⫾ 1.3
151.2 ⫾ 2.9
147.8 ⫾ 1.4
151.6 ⫾ 2.2
152.4 ⫾ 4.4
140.4 ⫾ 1.5
144.3 ⫾ 1.6
139.5 ⫾ 1.4
141.3 ⫾ 1.5
155.0 ⫾ 1.3
171.1 ⫾ 1.7
167.6 ⫾ 10.2
13063 ⫾ 63
15932 ⫾ 73
15820 ⫾ 68
16245 ⫾ 77
17058 ⫾ 121
65787 ⫾ 254
66200 ⫾ 445
54209 ⫾ 248
55109 ⫾ 229
65589 ⫾ 296
64180 ⫾ 266
62914 ⫾ 1025
␦234U
(initial)
13962 ⫾ 75
14065 ⫾ 1.3
14046 ⫾ 135
Age
(BP year†)
Th ages of old deep sea corals.
230
65610 ⫾ 239
65353 ⫾ 863
53682 ⫾ 532
54386 ⫾ 706
65455 ⫾ 246
63854 ⫾ 356
62471 ⫾ 708
12951 ⫾ 120
15552 ⫾ 363
15406 ⫾ 396
15058 ⫾ 1131
14359 ⫾ 2607
13736 ⫾ 223
13556 ⫾ 488
13258 ⫾ 760
Cor. age
(BP year⫹)
140.4 ⫾ 1.4
144.0 ⫾ 1.6
139.3 ⫾ 1.4
141.0 ⫾ 1.5
155.0 ⫾ 1.2
170.9 ⫾ 1.7
167.4 ⫾ 10.4
149.0 ⫾ 1.3
151.0 ⫾ 2.9
147.6 ⫾ 1.4
151.1 ⫾ 2.2
151.2 ⫾ 4.4
146.9 ⫾ 1.6
148.1 ⫾ 1.4
146.9 ⫾ 1.5
Cor. ␦234U
(initial)
Uranium series data of fossil corals analyzed in this study. Samples labeled with an (N) were ultrasonically cleaned. All other samples were chemically cleaned as described in the text. (I) and (II) are
separate pieces of the same sample. (D)— detrital material scrubbed off the surface of the coral. All errors are 2␴. † Age before 1950. ⫹ Corrected using an initial 230Th/232Th ratio of 85 ⫾ 80 ⫻ 10⫺6.
Errors are due only to the uncertainty in the initial ratio. See text.
JFA 20C
JFA 20B
JFA 20A
JFA 17
JFA 24C
JFA 2
Sample
number
Table 3. U and Th isotopic composition and
U-Th dating of deep-sea corals
2407
2408
Cheng et al.
through isochron methods, in which the sub-samples have
different parent-daughter ratios, but the same age and initial
daughter isotope composition. We use a variation of this approach because our assumptions do not always fit those used in
the isochron method and the stratigraphy of samples from the
same septum are an added constraint on the age. Instead, we
use development diagrams (Fig. 6, see Faure, 1986 or DePaolo,
1981 for analogous examples with other isotope systems), in
which we plot (230Th/232Th)° versus age for each sub-sample.
Assuming closed system behavior, each sub-sample is represented by a curve (in age vs. (230Th/232Th)° space) which is the
solution to Eqn. 2. For modern samples, this curve closely
approximates a straight line over the time scale of interest. This
can be demonstrated by taking the zero and first order terms of
a MacLaurin Expansion of Eqn. 2 (about t), assuming that
decay the of initial 230Th and the ingrowth of 234U are insignificant, and simplifying:
冉 冊 冉 冊冉 冊
230
Fig. 5. Plot of the mean ␦234U values of modern deep-sea corals
versus depth. Number of measurements per point are in parentheses.
Error bars show 2␴ analysis errors. The top dark gray rectangle
corresponds to the mean ␦234U value of 20 modern surface corals from
the Minnesota Isotope Lab (Cheng et al., 1999) . The long light gray
rectangle represents the mean and 2␴ error for the ␦234U values of all
five modern individuals (20 total measurements).
(Taylor and McLennan, 1985). Thus, for deep-sea corals, one
might expect (230Th/232Th)° to be a mixture between these two
end-members.
Our modern deep-sea corals have large variations in their
[232Th] and 230Th/232Th ratios (Table 2). The lowest 232Th
concentrations (47413, 84820, 85080) are similar to surface
corals (several tens or a few hundreds of ppt, (Edwards, 1988;
Gallup et al., 1994)). These samples also have the highest
230
Th/232Th ratios, suggesting that the aragonite skeletons
themselves have relatively low 232Th contents (generally in the
low 100’s of ppt). Supporting this idea is the fact that the four
modern sub-samples with concentrations significantly higher
than 1000 ppt all contained visible remnants of yellowish,
presumably organic material, on their exterior surfaces after
ultrasonic cleaning. Direct measurement of this yellowish material (Table 2, analyses DC-1-D and DC-3a-D) yielded 232Th
concentrations two to four orders of magnitude higher than the
bulk samples.
3.2.3. Development Diagrams
Determination of initial 230Th/232Th is not as straightforward
as making an accurate measurement on a “modern sample”. If
the coral was alive at the time of collection, we know the age
of the youngest portion of the skeleton. However, each subsample has a finite mass and therefore contains material that
grew over some interval of time potentially ranging from the
time of collection back to the initiation of growth. A priori, the
amount of radiogenic 230Th that formed during the sample
interval is not known. To calculate an initial 230Th/232Th ratio
from Eqn. 2 one needs to know the age (or more exactly, the
mean age weighted by mass) of the sample. So, for each
sub-sample, we have one equation (Eqn. 2) and two unknowns
(initial 230Th/232Th and age).
The classic approach to solving this type of problem is
Th
232
Th
238
⬇
m
U
232
Th
234
m
U
238
U
冉 冊
230
␭ 234t ⫹
m
Th
232
Th
0
(3)
Here all isotope ratios are atomic ratios, the subscript “m”
refers to the measured value, and the superscript “°” refers to an
initial value. Because the measured 230Th/232Th value is fixed,
older ages correspond to smaller initial 230Th/232Th ratios, and
vise-vera. Pairs of ages and initial ratios lie along a line for
which we know the slope and a single point. We can transform
the abscissa of this line such that the zero time reference point
is the measurement date rather than the initiation of growth. In
this case, we are extrapolating back from measured 230Th/232Th
ratios instead of extrapolating forward from initial 230Th/232Th
ratios. Here the variable t⬘, the new x-axis, is equal to ⫺(t ⫹
sample age). Eqn. 3 then becomes:
冉 冊 冉 冊冉 冊
230
Th
Th
232
0
238
⬇⫺
U
Th
234
232
U
U
m
冉 冊
230
␭ 234t⬘ ⫹
238
m
Th
Th
(4)
232
m
By assuming that the initial 230Th/232Th ratios in all subsamples of the same coral have the same value, we can use this
equation to constrain the initial 230Th/232Th value (Fig. 6). If
the sub-samples have the same stratigraphic position, the age
and initial 230Th/232Th value are represented by the intersection
of the lines of the sub-samples. If the stratigraphic sequence of
ages is known, the initial 230Th/232Th value must fall within a
range that gives a sequence of ages which agrees with the
known stratigraphic order. No values of initial 230Th/232Th that
give ages younger than the time of collection are possible.
The development diagram for septum 47413a is shown in
Figure 6. Data, including the 2␴ error bars, for the bottom
sub-sample of 47413a describes the dark gray wedge. Only
values of initial 230Th/232Th that lie within the wedge and to the
right of the time of collection are possible. Similarly, the light
gray area describes the range of acceptable pairs of ages and
initial 230Th/232Th ratios for the top of 47413a. The sides of the
light gray area are curved because of measurement error propagation through the various isotopic ratios in Eqn. 4. Due to the
stratigraphic order of the top and bottom pieces, the point
marked “A” is the first stratigraphically acceptable initial ratio.
Values higher than this intersection (younger ages) make the
dark gray wedge younger than the light gray area. However,
because point “A” is younger than the time of collection, the
U-Th dating of deep-sea corals
2409
Fig. 6. Development diagram for septum A of the modern D. cristagalli sample 47413. Top, middle and bottom pieces
of this septum are represented by wedges of light gray, black lines and dark gray respectively. Boundaries for the “wedges”
are calculated by propagating errors through Eqn. 4. Sample ages and initial 230Th/ 232Th ratios are constrained by the
intersection of the wedges coupled with stratigraphic constraints. See text for a full discussion. The y-intercept is the
measured thorium isotopic composition and the x-intercept is the age if the initial 230Th/232Th ratio were zero, the maximum
possible age. The slope is a function of the 238U/232Th ratio, the higher the value, the steeper the slope.
largest stratigraphically acceptable atomic (230Th/232Th)° value
is really 147 ⫻ 10⫺6. All values between 0 and 147 ⫻ 10⫺6
give ages for 47413a sub-samples that agree with stratigraphic
constraints and do not give negative ages for the septum. One
of the two middle pieces from this septum is shown for completeness (black lines), but does not further constrain the system. In this manner, each coral septum in Table 2 can be
analyzed for both the minimum and maximum allowable initial
230
Th/232Th ratios.
Due to lack of space, we do not show all data in graphical
form. Instead, we summarize the results of this analysis in
Table 4 and Figure 7. Outside of one coral, all samples have
initial isotopic compositions that fall between 0 and 160 ⫻
10⫺6 (the black dashed line in Fig. 7). Sample 84820 gives a
very large range of possible initial 230Th/232Th values and
correspondingly old ages. This sample was probably already
dead when dredged from the ocean bottom and the large
isotopic ratio is due to radiogenic 230Th rather than a high
initial ratio (see below).
3.3. Growth Rate and Band Periodicity from
Measurements on Modern Samples
In order to use individual D. cristagalli specimens as archives of oceanographic time series, we need to constrain their
growth rate and band periodicity. Given the constraints on
initial 230Th/232Th deduced in the previous section, we can use
the age equation and the mass weighted mean of our isotope
data to calculate the mean septal age for each sample (Table 4).
Ages for all sub-samples are less than 150 yr., and, for the most
part, in the range of tens of years or even less. Furthermore, the
age differences between stratigraphically older and younger
sub-samples of the same septa (Table 2) are generally less than
20 yr. For four of the five specimens, the youngest sub-sample
of each septum has an uncorrected age within error of 0 to 20
yr., and a whole septum mean age less than 80 yr. However, for
both septa analyzed from sample number 84820, the ages of the
youngest sub-samples are much greater than 20 yr. (150 ⫾ 7
and 158 ⫾ 9 yr., uncorrected for initial 230Th). Either this coral
grew more slowly than the others, or it had a very high
230
Th/232Th initial ratio, or it died prior to collection. Because
the difference in age between the top and bottom portions is
similar to the same difference in the other 4 specimens, we
conclude that coral 84820 died long before it was collected. In
this case we can not constrain the (230Th/232Th)° or the mean
age of the sample so we do not discuss the calculated growth
rate for 84820.
For the other samples, we calculate growth rates from mean
2410
Cheng et al.
Fig. 7. Plot of calculated initial 230Th/ 232Th ratio vs. sample number.
Upper and lower bounds (gray and black circles respectively) for each
septum were calculated using development diagrams (see Fig. 6 and
text). Nearby water column values (open squares) and measured detrital
values (open circles) are shown where data exist. Sample 85080 has a
measured detrital component that lies in the middle of the estimated
range of initial ratios. The range for the deep North Atlantic sample
78459 is more restricted because initial 230Th/232Th ratios below 50 ⫻
10⫺6 violate the stratigraphic order. Reported isotopic compositions of
thorium in North Atlantic filtered sea water range from 50 to 150 ⫻
10⫺6 (Moran et al., 1997 and Hoff et al., unpublished data). Sample
78459’s range of initial 230Th/232Th values is consistent with these
seawater data. Because the top of the coral must be younger than the
bottom of the sample, sample 48740’s initial 230Th/232Th ratio must be
less than 70 ⫻ 10⫺6. Water values from the area around the anomalous
sample 84820 are much lower than our estimated range, lending support to the argument that this sample died before collection. In general,
where data to make such comparisons exist, the initial thorium isotope
ratios of the aragonitic portion of the specimens are consistent with sea
water and detrital values. The overall range of possible initial 230Th/
232
Th ratios is between 0 and 160 ⫻ 10⫺6.
septal ages by assuming that the rate of mass accumulation is
constant with time and that the coral died when it was collected.
Given these assumptions, twice the mean age divided by the
septum’s length is the average growth rate. The mass-weighted
mean age for a septum is calculated by summing the individual
mass-adjusted 238U, 234U, 230Th, and 232Th concentrations of
the sub-samples and then recalculating the age from equation 4
(see Table 4 caption for which (230Th/232Th)° values were
used). Growth rates range from 0.1–3.1 mm/yr. The youngest
sample, 47413, has the largest error bars because its relative
age errors are large. For sample 85080a, only the top portion of
septum a was used in Table 4. In the bottom portion of this
septum, the 232Th concentration and the corresponding initial
230
Th are so large that the age is not very well constrained.
However, this bottom piece does place strong constraints on the
upper value of the initial 230Th/232Th ratio. To account for
using only the top piece to calculate the mean age, we adjusted
the total length and number of bands accordingly. On average
the growth rate for D. cristagalli seems to be about 1 mm/yr,
but there is evidence that separate specimens can have different
rates. The growth rates measured here indicate that deep-sea
corals may be used as recorders of deep-sea changes at annual
to centennial time scales (Smith et al., 1997). Adkins et al.
(1998) used this range of growth rates to constrain the life span
of individual D. Cristagalli corals that recorded a deep-ocean
circulation change at 15.4 ka to be under 160 yr. While the
fractional errors on our calculated growth rates are large, the
minimum rates, and therefore maximum ages of individual
corals, are well constrained. Measured 230Th concentrations are
small in our modern data set and can not lead to growth rates
slower than our minimum estimates where there is no initial
230
Th assumed.
Growth band frequency can also be calculated by dividing
the total number of bands in a septum by 2 times the mean age
(Table 4). Band counts were estimated using a transmited light
microscope. For the four corals that were alive when collected,
band frequency is between 0.3 and 3.0 bands/year. Taken at
face value, 2 of the 4 samples are not consistent with annual
banding. However, given uncertainties in both mass accumulation rate and especially band counting, we cannot rule out
annual banding. At this point, we can say that banding frequency is within a factor of a few of 1 pair/yr. Given the water
Table 4. Summary of evolution isochron data, growth rates and band periodicity.
Mean septal Agea
230/232
Sample
number
Analysis
number
min
max
Growth Rate
Error
Years
⫹
Periodicity
Error
⫺
Length
(mm)
mm/yr
⫹
Error
⫺
Bands
#
err
bands/yr
⫹
⫺
47413
DC-2a
0
147
7.5
6
5
41
2.7
4.8
1.3
40
8
2.7
4.7
1.3
DC-2b
0
994
6.6
39
6
41
3.1
84
2.7
40
8
3.0
82
2.7
85080
DC-3ab
0
155
2.5
10
10
11b
0.4
0.3
0.1
17b
10
0.7
0.6
0.4
DC-3b
0
155
41
7
6
33
0.4
0.1
0.1
50
10
0.6
0.2
0.2
48740
DC-5
0
67
72
9
9
19
0.1
0.02 0.01
—
—
—
—
—
78459
DC-1
50
130
53
10
10
28
0.3
0.06 0.04
35
8
0.3
0.1
0.1
-----------------------------------------------------------------------------------------------------------------------84820
DC-4a
183
1375
129
10
7
25
0.1
0.01 0.01
30
6
0.1
0.02 0.02
DC-4b
0
2060
146
19
9
25
0.1
0.01 0.01
30
6
0.1
0.02 0.02
Summary of initial 230Th/232Th ratio results from the development diagrams (see Fig. 6). Mean septal ages are calculated using a weighted average
of all U-series data from a single septum. Growth rates/periodicities are calculated by dividing the length/bands by twice the mean septal age. a Ages
use an initial atomic 230Th/232Th ratio of 80 ⫾ 80 ⫻ 10⫺6 except for 48740 and 78459 where the ratio is bettter constrained by the development
diagram data. Ages are also corrected for time since collection. b Includes data from only the top piece. Length and bands are adjusted accordingly.
U-Th dating of deep-sea corals
depths of coral growth, banding cannot be controlled by seasonal variations in sunlight, temperature or other climate variables. The banding pattern may be governed by endogenous
physiological rhythms (Emiliani et al., 1978) or by the raining
food supply from above (Deuser et al., 1981).
3.4. Fossil Samples
3.4.1. Sources of
232
Th and diagenetic Uranium
Ultimately we would like to use data from modern samples
to correct fossil coral ages for initial and added thorium. Fossil
samples are clearly enriched in 232Th relative to modern samples (compare Tables 2 and 3). Largely this enrichment is due
to the black crusts that coat older corals and are elevated in
232
Th by about 104 times over modern samples (labeled “D” in
Table 3). As these crusts also contain 230Th associated with the
232
Th, they represent a clear source of contamination to Useries ages. Our strategy has been to use chemical cleaning
techniques to remove a significant fraction of the added thorium. Then, using both Eqn. 2 and isochrons, we try and
account for residual added as well as initial thorium to calculate
an age. Chemically cleaned fossil corals have 232Th concentrations ranging from 856 to 4809 ppt, about 30 – 80% lower than
their untreated counterparts. As lower 232Th concentrations
imply smaller age errors in Equation 2, and a larger spread in
isotopic ratios for the isochrons (where dirtier samples are also
being measured), the chemical leaches are an important part of
obtaining precise and accurate ages on deep-sea corals. Without
removing the added thorium in black crusts, we could not place
any reasonable constraints on fossil coral ages (Lomitschka and
Mangini, 1999). In the sections that follow, we will show that
this cleaning does not significantly bias the calculated ages.
The elevated 232Th in fossil corals raises the issue of open
system behavior for uranium. Concentrations for the fossil
sub-samples are similar to the range in modern corals, indicating that none of the fossil corals have undergone large net
diagenetic gains or losses of uranium. In contrast to the situation with 232Th, the two Fe-Mn rich crusts have uranium
concentrations comparable to the coral values. Thus, any residual crust remaining after either the ultrasonic or chemical
cleaning procedure is not as likely to contribute a significant
fraction of uranium to the sample.
Uranium isotopic values for our fossil data set fall into two
categories. The eight sub-samples of coral that have ages between 10 and 20 kyr also have initial ␦234U values between 146
and 152‰, equal to or within several per mil of the modern
marine value (145.8 ⫾ 1.9). It is known from analysis of surface corals that the ␦234U of the surface ocean between 10 and
13 ky ago was indistinguishable from the modern value (Edwards et al., 1993). Thus, this slight elevation in ␦234U above
the modern value is quite likely due to small amounts of
diagenetic exchange of uranium. The seven sub-samples of
coral in the older range also have initial ␦234U values close to
the modern value, but have a larger spread: 139 to 171‰. In at
least one case (JFA 20C), we can resolve the ␦234U values of
two sub-samples of the same coral. While the value of marine
␦234U in this time interval is unknown, models suggest that it
is not likely to differ by large amounts from modern values
(Edwards, 1988; Gallup and Edwards, 1997; Richter and
2411
Turekian, 1993), nor is it likely to have shifted significantly
over time scales of several thousand years or less. Thus, the
range in ␦234U, the fact that most of the values are distinguishable from the modern value, and the differing ␦234U within
specimen JFA 20C all suggest diagenetic exchange of uranium.
The degree of exchange is more extensive in the older corals
than in the younger samples. Although the source of the diagenetic uranium is unknown, sea water and pore fluid uranium
could plausibly provide uranium of the appropriate isotopic
composition to cause these effects. In the following sections,
we investigate and compare two calculation methods that correct for these potential problems to U-series dating in fossil
deep-sea corals.
3.4.2. Correcting fossil coral ages: Equation 2
In section 3.2.2 we established that all samples in our modern data set have an initial atomic 230Th/232Th ratio lower than
160 ⫻ 10⫺6 (Table 4, Fig. 7). Using Eqn. 2 and a value of
80 ⫾ 80 ⫻ 10⫺6 for (230Th/232Th)°, we can calculate a conservative estimate for the age range of a fossil sample. However,
as discussed above, initial thorium is not the only component of
non-radiogenic thorium in our fossil corals. If we assume that
the added thorium has the same isotopic composition as the
initial thorium, within the broad bounds that we have set, and
that the thorium was added soon after coral growth, then Eqn. 2
is a solution to the age equation. The actual timing of thorium
addition, whether episodic or continuous, early or late is not
critical for corals significantly younger than the half-life of
230
Th (75 ky), but becomes an issue for samples older than a
good fraction of a 230Th half-life.
Given the measured modern initial Th isotopic range, we can
evaluate the sensitivity of calculated age errors to the amount of
measured 232Th. Figure 8 is a plot of error in age versus age for
a (230Th/232Th)° value equal to 80 ⫾ 80 ⫻ 10⫺6 and a uranium
concentration of 4 ppm. The thick black line represents typical
analytical errors as a function of age. The solid circles are the
age and 2␴ error in age for all sub-samples, uncorrected for
non-radiogenic thorium. The open circles represent the same
analyses corrected for non-radiogenic 230Th using Eqn. 2 and
the 80 ⫾ 80 value. For these, uncertainties are a combination of
analytical error and the assumed range in the (230Th/232Th)°
value. As the open circles plot above the solid circles, in some
cases by more than an order of magnitude, the dominant source
of error is clearly not analytical but the range in intial thorium
isotopic composition. Thin black lines are the calculated total
error in age as a function of age, contoured in 232Th concentration. Where these lines are horizontal, age errors are dominated by the (230Th/232Th)° ratio. The thick dashed line marks
the region where analytical errors begin to contribute to the age
error as the influence of the error on the (230Th/232Th)° ratio
falls away. These contours illustrate quantitatively the point
made above, that all other factors being equal, the lower the
232
Th concentration, the lower the error in age. For paired
230
Th-14C studies aimed at determining the ventilation age of
past water masses, one would like to determine 230Th age to
⫾102 years or better (Adkins et al., 1998; Mangini et al., 1998).
Figure 8 shows that to achieve this, one must clean deglacial
age corals to the level of about 1000 ppt. Table 3 indicates that
our cleaning procedure has achieved these levels in some cases,
2412
Cheng et al.
Fig. 8. Age plotted versus its 2␴ error. Solid points are the uncorrected deep-sea coral data and are close to the typical analytical
uncertainty (solid black line). After correction (open circles) using Eqn.
2, points are shifted to younger ages and larger errors. Each row in
Tables 2 and 3 represents one pair of points, one corrected (open) and
one uncorrected (solid). The relation between measured [232Th] and the
error in age for an assumed initial 230Th/232Th ratio of 80 ⫾ 80 ⫻ 10⫺6
is given by the thin lines. Higher 232Th results in a larger absolute age
uncertainty. For young ages, the error is dominated by uncertainty in
the initial Th correction. The older the age and the smaller the measured
[232Th] the lower the relative age error. In the region between the
dashed and thick black lines, age uncertainty is jointly determined by
analytical and initial Th errors. It is clear that further cleaning of 232Th
from older corals can significantly lower age uncertainties.
but it is clear that improvements to remove more exterior
thorium would further reduce the errors.
3.4.3. Correcting fossil coral ages: Isochrons
In addition to the method of correcting for initial 230Th using
Eqn. 2 and our estimate of initial 230Th/232Th, we have also
calculated ages by applying isochron methods to different
fragments of the same coral. Isochrons, in essence, allow one to
calculate a radiogenic end member isotopic composition from a
set of two or more materials that lie on a two-component
mixing line between radiogenic and non-radiogenic components. Given the isotopic composition of the radiogenic component, one can calculate an age. Inherent in the calculation is
the assumption that there are only two end members, an assumption that deep sea corals potentially violate. For example,
each coral is deposited over a finite interval of time so that the
radiogenic component of the oldest part of the coral may have
a different isotopic composition than the radiogenic component
of the youngest part of the coral. Because the lifetime of the
coral (typically tens to 200 yr based on our growth rate determinations) is similar to our analytical error in age (for corals
older than about 10 ka), this is not a significant problem for
corals older than about 10 ka. Also, non-radiogenic components are added at different times (as evidenced by fossil corals
with generally higher 232Th concentrations than modern corals). Earlier generations of non-radiogenic component may
have changed isotopic composition by radioactive decay and
ingrowth by the time subsequent generations of non-radiogenic
component are added, again violating the two-component assumption. Isochron methods do not require that the non-radiogenic component be present initially, but do require that the
non-radiogenic component is introduced at one instant in time.
For samples significantly younger than the half-life of 230Th,
the addition of a radiogenic component over a period of time
should not introduce significant inaccuracy. Therefore the timing of addition should not be an issue for our 10 to 20 ka old
samples. A third potential problem is the fact that there are at
least two possible non-radiogenic thorium components, the
hydrogenous and detrital components discussed above. A
fourth potential problem relates to the issue of the cleaning of
some of our samples by chemical (as opposed to physical)
techniques. This set of procedures could potentially shift U/Th
ratios either by preferential leaching from the solid or by
preferential adsorption onto the solid. To a certain degree, we
can test for all of these potential problems with isochrons. If
there are three components present in different proportions in
different sub-samples, the isotopic composition of the subsamples would likely deviate from a line in an isochron plot. If
cleaning techniques shift U/Th ratios, chemically treated samples should also deviate from a line defined by untreated
samples.
Results of isochron calculations (Ludwig, 1993; Ludwig and
Titterington, 1994) are shown in Table 5 and Figure 9. For all
three corals with enough data points, the MSWD statistic
exceeds one. As a measure of the linear fit of each isochron,
this result implies that one or more of the assumptions outlined
above is not valid. However, this statistic is sensitive to the
errors assigned to each measurement. Doubling the 232Th uncertainties reduces the MSWD to near or below one for each
sample. As 232Th errors are difficult to evaluate due to the
uncertainty in the filament blank, underestimating this value is
the likely reason for elevated MSWDs on our regressions.
Alternately, the slightly elevated MSWDs may reflect slight
deviations from linearity due to natural processes or chemical
treatment. If so, these deviations are on the order of analytical
error, suggesting that the isochron approach is valid at about the
level of our analytical precision. Additionally, chemically
treated and untreated sub-samples are co-linear at about the
level of our analytical precision, suggesting that chemical
cleaning has not shifted U/Th ratios significantly.
Corals in Table 3 fall into two age ranges, 10 –20 ka and
50 –70 ka. Assuming that the non-radiogenic component was
added early for the younger samples JFA-2 and JFA 24C, we
can calculate initial 230Th/232Th ratios for both corals from the
intercepts in Figure 9. For each sample these values (19 ⫻ 10⫺6
for JFA-2 and 37 ⫻ 10⫺6 for JFA-24) fall within the 80 ⫾ 80 ⫻
10⫺6 range that we established with modern corals. Thus, for
these younger samples, both the isochron and initial 230Th/
232
Th methods appear to be valid. In each case, it is important
that at least one sub-sample have low 232Th concentration to
get a reasonable age estimate. In practice, this requires the use
of some sort of chemical cleaning technique to remove surficial
contaminants.
For the three samples (JFA-20A, JFA-20B, and JFA-20C)
U-Th dating of deep-sea corals
2413
Table 5. Results of the isochron calculations for fossil deep sea corals.
Sample
number
Data
points
Corrected
(230Th/238U)act
Corrected
(234U/238U)act
Corrected
age (year)
95% Conf.
error (yr)a
Corrected
␦234U(T)
MSWD
value#
JFA-2
JFA-24C
JFA-20A
JFA-20B
JFA-20C
4
5
2
2
3
0.13322 ⫾ 0.00048
0.15241 ⫾ 0.00033
0.50896 ⫾ 0.00178
0.42499 ⫾ 0.00647
0.51718 ⫾ 0.00270
1.1418 ⫾ 0.00084
1.1428 ⫾ 0.00103
1.1158 ⫾ 0.00159
1.1163 ⫾ 0.00549
1.1195 ⫾ 0.00212
13510 ⫾ 50
15580 ⫾ 40
68680 ⫾ 340
51830 ⫾ 1100
66800 ⫾ 500
⫾280
⫾130
147.3 ⫾ 1.0
149.2 ⫾ 1.1
139.4 ⫾ 1.9
134.6 ⫾ 6.4
144.3 ⫾ 2.6
15.1
7.39
⫾2200
4.39
Results of the isochron calculations for fossil deep-sea corals. Calculations are based on “Rosholt Type-II diagrams” (Ludwig and Titterington,
1994) and were performed using the UISO program (Ludwig, 1993). ¶ Age before 1950. # Mean Square of Weighted Deviates of data from the
regression line.
between 50 and 70 ka old, we have also calculated ages using
both correction schemes (Tables 3 and 5). Corrected and uncorrected ages for each sample differ by no more than a few
thousand years, suggesting that the ages are accurate to about
this level. However, there are some consistencies and some
inconsistencies in the details. Sample JFA 20A gives consistent
corrected ages (Table 3). Sample JFA 20B yields an isochron
age that is slightly younger than the ages corrected using the
80 ⫾ 80 ⫻ 10⫺6 initial value, while sample JFA 20C gives an
isochron age that is older. However, the initial 230Th/232Th
ratio calculated for this sample, assuming that the non-radiogenic component was added early, is negative (Fig. 9). This
observation is consistent with a non-radiogenic component that
had a 230Th/238U ratio significantly less than that of the primary
coralline aragonite at the time of addition; and that was added
significantly after original precipitation of the primary coralline
aragonite. Thus, it appears that both uranium and thorium could
have been added to this sample. The uranium isotopic composition of the sub-samples supports this idea. The sub-samples
with the highest 232Th concentrations have initial ␦234U values
that are furthest from the marine value. The correction for
initial 230Th using the 80 ⫾ 80 ⫻ 10⫺6 value is not designed
to correct for uranium addition and therefore yields an inaccurate age in this case. The isochron method, on the other hand,
may well yield an accurate age. In addition to co-linearity of the
sub-sample points, the initial ␦234U value calculated using the
Fig. 9. Isochrons for three fossil deep-sea corals. These plots are half of a traditional Rosholt “Type II” isochron diagram.
Calculation of isochrons and error ellipses follow the method of (Ludwig and Titterington, 1994) using the program of
(Ludwig, 1993). Normally isochron plots use activity ratios, however we use atomic ratios to emphasize the implied initial
230
Th/ 232Th represented by the y-intercepts. The lowest points in JFA 2 and JFA 24C are from detrital pieces scraped from
the coral exterior. Gray scale points at the high end of each plot were chemically cleaned.
2414
Cheng et al.
isochron method is indistinguishable from the marine value. It
also appears that our chemical cleaning method removed most
of the added uranium as well as most of the added thorium. The
two sub-samples cleaned only by ultrasonic methods, have high
␦234U (and high 232Th), whereas the sub-sample that was also
chemically cleaned, has a ␦234U much closer to the modern
value (and low 232Th).
In sum, both isochron methods and methods that use modern
230
Th/232Th values to correct for initial 230Th give consistent
results for samples in the 10 to 20 ka age range. Samples in the
50 to 70 ka age range give consistent results within a few
thousand years, but in at least one case show clear evidence for
diagenetic addition of both uranium and thorium. In this case,
the isochron method appears to have corrected for this diagenetic component. For all fossil samples, chemical cleaning
methods remove added thorium and uranium. This cleaning
step is important in reducing errors in U-Th dating of fossil
corals. Further improvements in cleaning techniques would
further reduce dating errors.
4. CONCLUSIONS
Uranium rich deep-sea corals are suitable for precise dating
by the 238U-234U-230Th decay scheme. Measurements on a
suite of modern D. cristagalli constrain the growth rate of this
pseudo-colonial species to be between 0.1 and 3.1 mm/yr.
Because of uncertainties in the amount of initial 230Th, relative
errors on growth rates appear large. However, maximum ages,
and minimum growth rates, are well constrained because of the
relatively low amounts of measured 230Th. Combining age
determinations with counts of density bands provides an estimate of mean band periodicity of between 0.3 and 3 bands/yr.
␦234U data from modern deep-sea corals indicate that this
species is conservative in the upper 2000 meters of the water
column. Because this sample set comes from a variety of open
ocean settings, ␦234U initial measurements in fossil corals can
be an indicator of diagenetic alteration. Age errors on fossil
samples exceed the analytical uncertainty because of the range
of assumed initial 230Th/232Th ratios, 80 ⫾ 80 ⫻ 10⫺6. Reducing the amount of measured 232Th, by using more rigorous
cleaning techniques, is the most promising way to improve the
precision of fossil ages. Isochron ages for samples of deglacial
age agree with our 232Th corrected ages indicating that the
proposed method properly accounts for all initial 230Th. Older
specimens with more diagenetic alteration, may require better
cleaning and/or isochron measurements to obtain accurate and
precise ages.
Acknowledgments—D. Richards and J. Doral provided stimulating
discussions about many of the issues raised in this work. S. Cairns
helped identify the coral samples. T. Kleindinst of WHOI produced the
photograph in Figure 4. We would like to thank G. Henderson and D.
Muhs for helpful reviews of the manuscript. JFA thanks the UCAR
Post-Doctoral Fellowship Program and LDEO for support during the
writing of this paper.
REFERENCES
Adkins J. F. and Boyle E. A. (1997) Changing atmospheric ⌬14C and
the record of paleo-ventilation ages. Paleoceanography 12, 337–
344.
Adkins J. F., Cheng H., Boyle E. A., Druffel E. R. M., and Edwards
R. L. (1998) Deep-sea coral evidence for rapid change in ventilation
of the deep North Atlantic 15,400 years ago. Science 280, 725–728.
Anderson R. F., Bacon M. P., and Brewer P. G. (1983) Removal of
Th-230 and Pa-231 from the open ocean. Earth Planet. Sci. Lett. 62,
7–23.
Bacon M. P. and Anderson R. F. (1982) Distribution of Thorium
isotopes between dissolved and particulate forms in the deep sea. J.
Geophys. Res. 87, 2045–2056.
Bard E., Arnold M., Fairbanks R., and Hamelin B. (1993) 230Th-234U
and 14C ages obtained by mass spectrometry on corals. Radiocarbon
35, 191–199.
Bard E., Hamelin B., Arnold M., Montaggioni L., Cabioch G., Faure
G., and Rougerie F. (1996) Deglacial sea-level record from Tahiti
corals and the timing of global meltwater discharge. Nature 382,
241–244.
Bard E., Hamelin B., Fairbanks R. G., and Zindler A. (1990) Calibration of the 14C timescale over 30,000 years using mass spectrometric
U-Th ages from Barbados corals. Nature 345, 405– 410.
Barnes J. W., Lang E. J., and Potratz H. J. (1956) Ratio of Ionium to
Uranium in coral limestone. Science 124, 175–176.
Beck J. W., Recy J., Taylor F., Edwards R. L., and Cabioch G. (1997)
Abrupt changes in early Holocene tropical sea surface temperature
drived from coral records. Nature 385, 705–707.
Beck W. J., Edwards L. R., Ito E., Taylor F. W., Recy J., Rougerie F.,
Joannot P., and Henin C. (1992) Sea-Surface temperature from coral
skeletal strontium/calcium ratios. Science 257, 644 – 647.
Boyle E. A. (1981) Cadmium, zinc, copper, and barium in foraminifera
tests. Earth Planet. Sci. Lett. 53, 11–35.
Boyle E. A. and Keigwin L. D. (1985/6) Comparison of Atlantic and
Pacific paleochemical records for the last 250,000 years: Changes in
deep ocean circulation and chemical inventories. Earth Planet. Sci.
Lett. 76, 135–150.
Broecker W. S., Thurber D. L., Goddard J., Ku T.-L., Matthews R. K.,
and Mesolella K. J. (1968) Milankovitch hypothesis supported by
precise dating of coral reefs and deep-sea sediments. Science 159,
297–300.
Burnett W. C. and Veeh H. H. (1992) Uranium-series studies of marine
phosphates and carbonates. In Uranium-series disequilibria, applications to earth, marine and environmental sciences. (ed. M. Ivanovich and R. S. Harmon). Oxford Science Publications.
Cairns S. D. (1981) Marine flora and fauna of the Northern United
States. Scleractinia. NOAA Technical Report NMFS Circular 438,
1–15.
Cairns S. D. and Stanley G. D. Jr. (1981) Ahermatypic coral banks:
Living and fossil counterparts. Proc. of the Fourth Intnl. Coral Reef
Symp., Manila 1, 611– 618.
Chen J. H., Edwards R. L., and Wasserburg G. J. (1986) 238U, 234U and
232
Th in seawater. Earth Planet. Sci. Lett. 80, 241–251.
Cheng H., Adkins J. F., Edwards R. L., and Boyle E. A. (1995) 230Th
dating of deep-sea solitary corals. AGU, F292.
Cheng H., Edwards R. L., Hoff J., Gallup C. D., Richards D. A., and
Asmerom Y. (2000) The half-lives of Uranium-234 and Thorium230. Chem. Geol. in press.
Cochran J. K. (1992) The oceanic chemistry of the uranium- and
thorium-series nuclides. In Uranium-series disequilibrium, applications to earth, marine and environmental sciences (2nd edition)
(ed. M. Ivanovich and R. S. Harmon). Oxford Science Publications.
Cochran J. K., Livingston H. D., Hirschberg D. J., and Surprenant
L. D. (1987) Natural and anthropogenic radionuclide distributions
in the norhtwest Atlantic Ocean. Earth Planet. Sci. Lett. 84,
135–152.
Delaney M. L. and Boyle E. A. (1983) Uranium and thorium isotope
concentrations in foraminiferal calcite. Earth Planet. Sci. Lett. 62,
258 –262.
DePaolo D. J. (1981) Nd isotopic studies: Some new perspectives on
earth structure and evolution. EOS 62, 137–140.
Deuser W., Ross E. H., and Anderson R. F. (1981) Seasonality in the
supply of sediment to the deep Sargasso Sea and implications for the
rapid transfer of matter to the deep ocean. Deep-Sea Res. 28, 495–
505.
Druffel E. R. M., King L. L., Belastock R. A., and Buesseller K. O.
U-Th dating of deep-sea corals
(1990) Growth rate of a deep-sea coral using 210Pb and other isotopes. Geochim. Cosmochim. Acta 54, 1493–1500.
Duncan P. M. (1877) On the rapidity of growth and variability of some
Madreporaria on an Atlantic cable with remarks upon the rate of
accumulation of foraminiferal deposits. Ann. Mag. Nat. Hist. 20,
361–365.
Edwards L. R., Beck W. J., Burr G. S., Donahue D. J., Chappell
J. M. A., Bloom A. L., Druffel E. R. M., and Taylor F. W. (1993) A
large drop in atmospheric 14C/12C and reduced melting in the
Younger Dryas, documented with 230Th ages of corals. Science 260,
962–967.
Edwards R. L. (1988) High-Precision Thorium-230 ages of corals and
the timing of sea level fluctuation in the late Quaternary. Ph.D.
Thesis, California Institute of Technology.
Edwards R. L., Chen J. H., Ku T.-L., and Wasserburg G. J. (1987a)
Precise timing of the last interglacial period from mass spectrometric determination of Thorium-230 in corals. Science 236,
1547–1553.
Edwards R. L., Chen J. H., and Wasserburg G. J. (1987b) 238U-234U230
Th-232Th systematics and the precise measurement of time over
the past 500,000 years. Earth Planet. Sci. Lett. 81, 175–192.
Emiliani C., Hudson J. H., Shinn E., and George R. Y. (1978) Oxygen
and carbon isotopic growth record in a reef coral from the Florida
Keys and a deep-sea coral from Blake Plateau. Science 202, 627–
629.
Faure G. (1986) Principles of isotope geology (2nd edition). Wiley and
Sons.
Gagan M. K., Ayliffe L. K., Hopley D., Cali J. A., Mortimer G. E.,
Chappell J., McCulloch M. T., and Head M. J. (1998) Temperature
and surface-ocean water balance of the Mid-Holocene tropical Western Pacific. Science 279, 1014 –1018.
Gallup C. D., Edwards L. R., and Johnson R. G. (1994) The timing
of high sea levels over the past 200,000 years. Science 263,
796 – 800.
Gallup C. D. and Edwards R. L. (1997) Constraints from the uranium
isotopic composition of well-preserved corals on past changes in the
flux or isotopic composition of riverine uranium. AGU Fall Meeting,
San Francisco.
Goldstein S. J., Murrell M. T., Lea D., Chakraborty S., and Kashgarian
M. (1996) 231Pa and 230Th dating of deep-sea coral. EOS: Transactions, AGU, 1996 Spring Meeting, F291.
Grigg R. W. (1974) Growth rings: Annual periodicity in two Gorgaonian corals. Ecology 55, 876 – 881.
Guilderson T. P., Fairbanks R. G., and Rubenstone J. L. (1994) Tropical temperature variations since 20,000 years ago: Modulating interhemispheric climate change. Science 263, 663– 665.
Guo L., Santschi P., Baskaran M., and Zindler A. (1995) Distribution
of dissolved and particulate 230Th and 232Th in seawater from the
Gulf of Mexico and off Cape Hatteras as measured by SIMS. Earth
Planet. Sci. Lett. 133, 117–128.
Gvirtzman G., Friedman G. M., and Miller D. S. (1973) Control and
distribution of Uranium in coral reefs during diagenesis. J. Sediment.
Petrol. 43, 985–997.
Henderson G. M., Cohen A. S., and O’Nions R. K. (1993) 234U/238U
ratios and 230Th ages for Hateruma Atoll corals: Implications for
coral diagenesis and seawater 234U/238U ratios. Earth Planet. Sci.
Lett. 115, 65–73.
Henderson G. M. and O’Nions R. K. (1995) 234U/238U ratios in
Quaternary planktonic foraminifera. Geochim. Cosmochim. Acta 59,
4685– 4694.
Huh C.-A. and Beasley T. M. (1987) Profiles of dissolved and particulate thorium isotopes in the water column of coastal southern
California. Earth Planet. Sci. Lett. 85, 1–10.
Huh C.-A., Moore W. S., and Kadko D. C. (1989) Oceanic Th-232: A
reconnaissance and implication of global distribution from manganese nodules. Geochim. Cosmochim. Acta 53, 1357–1366.
Ivanovich M. and Harmon R. S. (1992) Uranium-series disequilibrium,
2nd Edition. Oxford University Press.
Jaffey A. H., Flynn K. F., Glendenin L. E., Bently W. C., and Essling
A. M. (1971) Precision measurement of half-lives and specific activities of 235U and 238U. Phys. Rev. C 4, 1889 –1906.
2415
Ku T.-L. (1965) An evaluation of the 234U/238U method as a tool for
dating pelagic sediments. J. Geophys. Res. 70, 3457–3474.
Ku T.-L. and Kern J. P. (1974) Uranium-series age of the upper
Pleistocene Nestor Terrace, San Diego, California. GSA Bulletin 85,
1713–1716.
Ku T. L., Knauss K. G., and Mathieu G. G. (1977) Uranium in the open
ocean: Concentration and isotopic composition. Deep-Sea Res. 24,
1005–1017.
Lin J. C., Broecker W. S., Anderson R. F., Hemming S., Rubenstone
J. L., and Bonani G. (1996) New 230Th/U and 14C ages from Lake
Lahontan carbonates, Nevada, USA, and a discussion of the
origin of initial thorium. Geochim. Cosmochim. Acta 60, 2817–
2832.
Lomitschka M. and Mangini A. (1999) Precise Th/U dating of small
and heavily coated samples of deep sea corals. Earth Planet. Sci.
Lett. 170, 391– 401.
Ludwig K. R. (1993) UISO-A program for calculation of 230Th-234U238
U isochrons. USGS Open File Report 93-531.
Ludwig K. R. and Titterington D. M. (1994) Calculation of 230Th/U
isochrons, ages and erors. Geochim. Cosmochim. Acta 58, 5031–
5042.
Mangini A., Lomitschka M., Eichstadter R., Frank N., Vogler S.,
Bonani G., Hajdas I., and Patzold J. (1998) Coral provides way to
age deep water. Nature 392, 347–348.
McCulloch M., Mortimer G., Esat T., Xianhua L., Pillans B., and
Chappell J. (1996) High resolution windows into early Holocen
climate: Sr/Ca coral records from the Huon Peninsula. Earth Planet.
Sci. Lett. 138, 169 –178.
Mesolella K. J., Matthews R. K., Broecker W. S., and Thurber D. L.
(1969) The astronomical theory of climatic change: Barbados data. J.
Geol. 77, 250 –274.
Min G. R., Edwards R. L., Taylor F. W., Recy J., Gallup C., and Beck
J. W. (1995) Annual cycles of U/Ca in coral skeletons and U/Ca
thermometry. Geochim. Cosmochim. Acta 59, 2025–2042.
Moore W. S. (1981) The thorium isotope content of ocean water. Earth
Planet. Sci. Lett. 53, 419 – 426.
Moran S. B., Charette M., Hoff J. A., and Edwards R. L. (1997)
Distribution of Th-230 in the Labrador Sea and its relation to
ventilation. Earth Planet. Sci. Lett. 150, 151–160.
Moran S. B., Hoff J. A., Buesseler K. O., and Edwards R. L. (1995)
High precision 230Th and 232Th in the Norwegian Sea and Denmark
by thermal ionization mass spectrometry. Geophys. Res. Lett. 22,
2589 –2592.
Muhs D. R., Kennedy G. L., and Rockwell T. K. (1994) Uranium-series
ages of marine terrace corals from the Pacific coast of North America
and implication for last-intergalcial sea level history. Quat. Res. 42,
72– 87.
Nozaki Y. and Horibe Y. (1983) Alpha-emitting thorium isotopes in
northwest Pacific deep waters. Earth Planet. Sci. Lett. 65, 39 –50.
Nozaki Y., Yang H.-S., and Yamada M. (1987) Scavenging of thorium
in the ocean. J. Geophys. Res. 92, 772–778.
Pratje O. (1924) Korallenbanke in tiefen und kuhlen Wasser. Cebtralbl
f Min. Geol., 410 – 415.
Richter F. M. and Turekian K. K. (1993) Simple models for the
geochemical response of the ocean to climatic and tectonic forcing.
Earth Planet. Sci. Lett. 119, 121–131.
Roy-Barman M., Chen J. H., and Wasserburg G. J. (1996) 230Th-232Th
systematics in the central Pacific Ocean: The sources and fates of
thorium. Earth Planet. Sci. Lett. 139, 351–363.
Russell A. D., Emerson S., Nelson B., Erez J., and Lea D. W. (1994)
Uranium in foraminiferal calcite as a recorder of seawater uranium
concentrations. Geochim. Cosmochim. Acta 58, 671– 681.
Shen G. T. and Boyle E. A. (1988) Determination of lead, cadmium
and other trace metals in anually-banded corals. Chem. Geol. 67,
47– 62.
Shen G. T. and Dunbar R. B. (1995) Environmental controls of
Uranium in reef corals. Geochim. Cosmochim. Acta 59, 2009 –
2024.
Smith J. E., Risk M. J., Schwarcz H. P., and McConnaughey T. A.
(1997) Rapid climate change in the North Atlantic during the
Younger Dryas recorded by deep-sea corals. Nature 386, 818 –
820.
2416
Cheng et al.
Stein M., Wasserberg G. J., Lajoie K. R., and Chen J. H. (1991)
U-series ages of solitary corals from the Californian coast by mass
spectroscopy. Geochim. Cosmochim. Acta 55, 3709 –3722.
Stirling C. H., Esat T. M., McCulloch M. T., and Lambeck K. (1995)
High precision U-series dating of corals from Western Australia and
implications for the timing and duration of the last interglacial. Earth
Planet. Sci. Lett. 135, 115–130.
Szabo B. J. (1985) Uranium-series dating of fossil corals from marine
sediments of the United States Atlantic coastal plain. GSA Bulletin
96, 398 – 406.
Szabo B. J., Ludwig K. R., Muhs D. R., and Simmons K. R. (1994)
Thorium-230 ages of corals and duration of the last interglacial
sea-level high stand on Oahu, Hawaii. Science 266, 93–96.
Taylor S. R. and McLennan S. M. (1985) The Continental Crust: Its
Composition and Evolution. Blackwell, Oxford.
Teichert C. (1958) Cold and deep-water coral banks. Bull. Am. Assoc.
Petrol. Geol. 42, 1064 –1082.
Thompson G. and Livingston H. D. (1970) Strontium and uranium
concentrations in aragonite precipitated by some modern corals.
Earth Planet. Sci. Lett. 8, 439 – 442.