Auxiliary Material for
ENSO Reconstructions over the past 60ka using giant clams (Tridacna sp.) from Papua New
Guinea
Robin Driscoll1, Mary Elliot2, Tom Russon1, Kevin Welsh3, Yusuke Yokoyama4, Alexander
Tudhope1
1 - School of GeoSciences, University of Edinburgh, West Mains Road, Edinburgh EH9
3JW, UK
2 – Laboratoire de Planetologie et Geodynamique, Université de Nantes, France
3 – School of Earth Sciences, University of Queensland, Australia
4 – Atmosphere and Ocean Research Institute, Department of Earth and Planetary Sciences,
University of Tokyo, Japan
Correspondence should be addressed to Robin Driscoll (e-mail: r.e.driscoll@ed.ac.uk)
Introduction
This file contains two tables with δ18O details, calibrated age ranges or age estimations based
on reef age, and variance compared to the modern Tridacna gigas shell, a figure of the
location of the Huon Peninsula and the raw δ18O data from modern and fossil Tridacna sp.
shells.
Dating and Preservation
Fossil Tridacna sp. were collected from uplifted reef terraces along the Huon Peninsula,
Papua New Guinea (147.5°E, 6.5°S) in 2004. These terraces have been uplifted by the
collision of West Pacific and Australian plates, and have been well documented and
extensively dated [Chappell et al., 1996; Yokoyama et al., 2001; Yokoyama et al., 2000].
Fossil samples from terraces known to be younger than 45 ka were dated using 14C AMS
analysis of well-preserved subsamples of Tridacna sp. Shells collected from terraces known
to be older than 45ka were inferred from their position in terraces where previous studies
have collected reliable U/Th dates of in situ corals [Chappell et al., 1996; Yokoyama et al.,
2001; Yokoyama et al., 2000]. Samples from the inner layers of Tridacna sp. were used for
radiocarbon dating, which were analysed in two different laboratories, the NERC
Radiocarbon Laboratory in East Kilbride and the Department of Nuclear Engineering and
Management at Tokyo University. Holocene age shells were corrected by 407±52 years for
the local marine radiocarbon reservoir effect [Edwards et al., 1993]. Changes to circulation
in MIS3 mean that the global marine reservoir age (ΔR [Stuiver and Braziunas, 1993]) cannot
be reliably calculated, therefore a reservoir correction of 400 years was applied, with the
caveat that ages may be incorrect by several hundred years. Corrected radiocarbon dates
were calibrated using the Calib Rev 6.1.0 program [Stuiver and Reimer, 1993] with Marine
09.14 calibration, and ages are reported before present (BP). Mean ages were found to be
consistent with previously published coral radiocarbon data.
Fossil Tridacna sp. samples were screened for their mineralogical composition and
microstructure preservation before geochemical analysis, using X-ray diffraction (XRD).
Carbonate powder was collected using a handheld drill across the inner section of samples,
and a Bruker-AXS D8 Advance XRD using Cu K-alpha radiation (40kV) as the source and a
Sol-X energy dispersive detector was used to determine % calcite in carbonate powders. The
modern sample, MT7, had a 100% aragonite composition and all fossil shells used here had
less than 1% calcite. Additionally, SEM imaging of fossil Tridacna sp. show excellent
preservation of aragonite fibre and daily growth bands in all regions analysed.
Oxygen Isotopes
Profiles of δ18O were obtained by sampling polished 1mm thin sections the inner layer or
hinge of Tridacna sp. shells. A high precision MicroMill (New Wave) was used to sample
the shell for oxygen isotope analysis. Samples were milled with a resolution between 0.2 and
0.25 mm. Powdered aragonite samples were analysed in the Wolfson Stable Isotope
Laboratory at University of Edinburgh using a Thermo Electron Delta+ Advantage stable
isotope mass spectrometer with Keil Carbonate III preparation device. Age models were
constructed using day/night band counts and δ18O records, with δ18O maxima being assigned
an August date, as is commonly done for seasonally resolved stable isotopic time series
[Tudhope et al., 2001; Welsh et al., 2011]. Sampling resulted in between 6 and 20 samples
per year depending on the ontogenetic stage of shell growth, so all δ18O profiles were
interpolated to 6 samples per year, assuming linear growth between August δ18O maxima [de
Brauwere et al., 2009].
Variance of inter-annual anomaly time series
Any potential ontological trends were first removed from the interpolated δ18O time series by
subtracting the linear trend from each profile. Other methods of detrending, including a 2nd
order polynomial fit and a moving average were calculated and all gave similar results within
the same shell. The seasonal cycle was then removed from each detrended δ18O record by
calculating the mean δ18O value for each two-month period within that shell and subtracting
this from the δ18O record to give an anomaly time series (δ18OA). The variance of each
anomaly time series was calculated and compared to the modern Tridacna gigas.
Other methods for ENSO reconstruction and inter-comparison of results
We compare results from the inter-annual anomaly approach to those from both a bandpass
filter and a Hanning filter. For the bandpass filter, once each time series was interpolated to 6
samples per year, it was passed through a bandpass filter using Analyseries [Paillard et al.,
1996]. This removes any frequency outside the ENSO band of 2.5 to 7 years. The variance
of the filtered δ18O time series is then used as the ENSO metric. We also applied a 25 month
Hanning Filter [Correge et al., 2000] to each fossil shell to compare variance from this
method. The F values produced from the Hanning Filter, the 2.5-7 year bandpass filter and
δ18OA time series were similar (maximum difference in p-value of 0.23). Both filter methods
contain edge effect uncertainties at the ends of the time series, due to the finite width of the
filtering windows. To account for this, a period equivalent to at least the lower edge of the
filter passband should be excluded from the variance calculations, which effectively reduces
the durations of many of the records considered here by a large relative proportion. Another
method was to count events exceeding a specified threshold. Following the example of
McGregor and Gagan [2004], we calculated the δ18O anomaly for El Niño events occurring
during the lifetime of the modern shell as 0.22±0.07‰ (1σ), and for La Niña as -0.2±0.06‰
(1σ). If we therefore count positive or negative excursions above a threshold of ±0.15‰ for
more than 6 months, the number of events seen in fossil shells, normalised to the number of
events in MT7, is similar to that of the variance normalised to MT7 obtained from the δ18OA
time series (see Figure 1). This gives weight to our theory that the anomaly time series is
reflecting ENSO activity at this location.
Statistical Tests
We used the well-known F test to assess the significance of changes in variance. This is a
parametric test of the null-hypothesis that the two sample variances are drawn from
populations with the same population variance. This approach requires standard sampling
assumptions, namely that the data in the two samples are independent and identically
normally distributed.Whilst some ENSO-related metrics are known to be significantly
skewed (Eastern Pacific SST El Niño events tend to be larger than La Niña ones, for
example), the extent of non-normality is relatively small in West Pacific SST and in the
proxy data considered here. The effective numbers of degrees of freedom present in the subannually resolved data is estimated based on the assumption of an annual decorrelation time
for the anomaly data and a decorrelation time equivalent to the lower edge of the bandpass
filter passband for the bandpass filtered data. Were a lower decorrelation time of two years
used for the sub-annually resolved data, two fossil shells would remain statistically
significant out of the five original shells which are significant with an annual decorrelation
time.
File Explainer
1.ts01.txt Details of Holocene shells used in this study, including age range (ka BP), mean
d18O, d18OA variance and comparison to the modern Tridacna gigas
1.1 Column "Shell", Sample identification
1.2 Column "Species", Species of Tridacna sp.
1.3 Colummn "Reef", Reef terrace where sample was collected
1.4 Column "Age", cal ka BP, calibrated 2s.d. age range
1.5 Column "Length", years, length of stable oxygen isotope record
1.6 Column "Variance", variance of the anomaly oxygen isotope record
1.7 Column "% of modern", %, variance of each shell compared to MT7
1.8 Column "F test p Value", result of the statistical F-test
2.ts02.txt Details of MIS3 shells used in this study, including age range (ka BP), mean d18O,
d18OA variance and comparison to the modern Tridacna gigas
2.1 Column "Shell", Sample identification
2.2 Column "Species", Species of Tridacna sp.
2.3 Colummn "Reef", Reef terrace where sample was collected
2.4 Column "Age", cal ka BP, calibrated 2s.d. age range
2.5 Column "Length", years, length of stable oxygen isotope record
2.6 Column "Variance", variance of the anomaly oxygen isotope record
2.7 Column "% of modern", %, variance of each shell compared to MT7
2.8 Column "F test p Value", result of the statistical F-test
Auxiliary Figure 1 Raw δ18O profiles of modern and Holocene T. gigas shells.
Auxiliary Figure 2 Raw δ18O profiles of MIS3 Tridacna sp. shells
References
Chappell, J., A. Omura, T. Esat, M. McCulloch, J. Pandolfi, Y. Ota, and B. Pillans (1996),
Reconciliation of late Quaternary sea levels derived from coral terraces at Huon Peninsula
with deep sea oxygen isotope records, Earth and Planetary Science Letters, 141(1-4), 227236.
Correge, T., T. Delcroix, J. Recy, W. Beck, G. Cabioch, and F. Le Cornec (2000), Evidence
for stronger El Nino-Southern Oscillation (ENSO) events in a mid-Holocene massive coral,
Paleoceanography, 15(4), 465-470.
de Brauwere, A., F. De Ridder, R. Pintelon, J. Schoukens, and F. Dehairs (2009), A
comparative study of methods to reconstruct a periodic time series from an environmental
proxy record, Earth-Science Reviews, 95(3-4).
Edwards, R. L., J. W. Beck, G. S. Burr, D. J. Donahue, J. M. A. Chappell, A. L. Bloom, E. R.
M. Druffel, and F. W. Taylor (1993), A large drop in atmospheric C-14/C-12 and reduced
melting in the Younger Dryas, documented with Th-230 ages of corals, Science, 260(5110),
962-968.
McGregor, H. V., and M. K. Gagan (2004), Western Pacific coral δ18O records of anomalous
Holocene variability in the El Niño-Southern Oscillation, Geophysical Research Letters,
31(11).
Paillard, D., L. Labeyrie, and P. Yiou (1996), Macintosh Program performs time-series
analysis, Eos, 77(39), 379.
Stuiver, M., and T. F. Braziunas (1993), Modeling Atmospheric 14C Influences and 14C Ages
of Marine Samples to 10,000 BC, Radiocarbon, 35(1), 137-189.
Stuiver, M., and P. J. Reimer (1993), Extended 14C Database and Revised CALIB 3.0 14C
Age Calibration Program, Radiocarbon, 35(1), 215-230.
Tudhope, A. W., C. P. Chilcott, M. T. McCulloch, E. R. Cook, J. Chappell, R. M. Ellam, D.
W. Lea, J. M. Lough, and G. B. Shimmield (2001), Variability in the El Niño - Southern
oscillation through a glacial-interglacial cycle, Science, 291(5508), 1511-1517.
Welsh, K., M. Elliot, A. Tudhope, B. Ayling, and J. Chappell (2011), Giant bivalves
(Tridacna gigas) as recorders of ENSO variability, Earth and Planetary Science Letters,
307(3-4), 266-270.
Yokoyama, Y., T. M. Esat, and K. Lambeck (2001), Coupled climate and sea-level changes
deduced from Huon Peninsula coral terraces of the last ice age, Earth and Planetary Science
Letters, 193(3-4), 579-587.
Yokoyama, Y., T. M. Esat, K. Lambeck, and L. K. Fifield (2000), Last ice age millennial
scale climate changes recorded in Huon Peninsula corals, Radiocarbon, 42(3), 383-401.