Supplementary Material
Methods and Study Sites
Two coral reef sites have been examined in this study, Havannah and Pandora and Reefs (Figure 1).
At Havannah Reef a Porites core of 5.3 metres length that extends back to the late 1600’s, was collected by
Dr Isdale from AIMS in 1988. Additional shorter cores were collected by the authors in 1998 from both
Havannah and Pandora Reefs. Although these shorter cores only extend back to the late 1960’s, they
nevertheless enable crosschecks to be made on reproducibility of the coral records both within and between
reefs (Figure 2c). Coral cores were slabbed into ~6 mm slices, x-rayed and photographed under UV light.
The UV light illuminates luminescent bands caused by river flood plumes, with floods generally occurring
in the months of January and February. Luminescent flood bands therefore provide excellent chronological
markers for the inner GBR19,20,28. A luminescent chronology was established by counting of individual
flood bands, which was crosschecked using x-ray density bands. The translation from distance to time in
the coral was undertaken independently, using Sr/Ca and U/Ca ratios (determined with laser ablation ICPMS) and assuming that the winter minimum sea surface temperature (i.e. maximum Sr/Ca and U/Ca ratios)
occurred on the 20th July of each year. Time was interpolated linearly between winter minimums. The
chronology of the Ba/Ca records was thus determined independently from the luminescent flood band
chronology.
The coral slices were cut into 20 mm x 50 mm pieces for analysis in the laser ablation cell and
cleaned in distilled water using a high intensity ultra-sonic probe. Slices were then scanned beneath an ArF
excimer laser in a mixed He, Ar cell with ablated material being carried in an Ar stream to the ICP-MS
plasma. The laser operates at a wavelength of 193 nm and has the unique feature of having a large depth of
field, a consequence of the extended focal length optics. This is important for analyses of corals, as it
enables the laser beam to remain in focus despite the relatively irregular coral surface and hence minimise
elemental fractionation. Prior to analysis the coral slice was pre-ablated14 using a 100x800 micron slit with
the Ar gas stream disconnected from ICP-MS. Elemental analysis of the coral pieces were undertaken with
a 50x500 micron slit with backgrounds and standards being determined at the beginning and end of the
analysis of each piece. High abundance minor elements B, Mg, Sr and U were standardised against a
pressed powder coral standard15. The isotopes
11
B,
25
Mg,
31
P,
46
Ca,
55
Mn,
84
Sr,
138
Ba and
238
U were
measured with 46Ca being used as an index isotope to correct for signal fluctuations. Samples were scanned
at a rate of ~1 mm per minute, with data being collected in 1 second cycles. To improve counting statistics
data was averaged into 10 second blocks, representing a spatial resolution of ~0.15 mm equivalent to a
temporal resolution of 3-4 days. See Sinclair et al14 and Fallon et al15 for more details.
Determination of the Magnitude of pre-European Floods
For the Burdekin River, daily discharge data is available from 1921 to the present. Prior to that time
there are occasional records of the maximum height of flood peaks (measured near Home Hill) that extend
back to 1870, when European settlers first arrived. Two methods have been used to estimate the magnitude
of pre-European floods. The first approach is based on the method of Lough et al.,28 using visual
determination of flood band intensity under UV light. Based on this approach, flood bands were divided
into three categories, strong, average, and weak. The modern coral record (i.e.1921 to 1988) was used to
calibrate these categories against the measured Burdekin River flow discharge. This modern calibration
was then applied to the pre-European flood bands (see figure below).
To crosscheck this semi-quantitative approach, a limited number of flood-bands were also sampled
at high resolution (0.25 mm intervals) for Sr/Ca and δ 18O isotope ratio measurements. Following the
approach of McCulloch et al.,29 the salinity change at the coral reef due to freshwater input by the river
flood plumes can be determined by subtracting the temperature component from the δ 18O signal using
Sr/Ca ratios. This is because δ18O ratios in corals reflect changes in both temperature and salinity (i.e. the
δ18O composition of seawater), whereas Sr/Ca ratios are mainly dependent on sea surface temperature29. In
the Havannah coral, δ18O-Sr/Ca systematics were determined for the floods of 1958, 1968, 1970, 1972,
1974 and 1981 and then compared with the weekly averaged records of Burdekin River flow (see
supplementary figure). For the modern part of the coral record there is a reasonable correlation between
salinity changes derived from δ18O–Sr/Ca ratios in the coral and the Burdekin River discharge records of
flood events. This approach was then extended to the older part of the coral record for the floods of 1819,
1826, 1831, 1859, 1860 and 1864. The flood magnitudes determined using this approach are shown in the
supplementary figure and are consistent with those obtained using the intensity of luminescent bands.
Flood plume events were recorded throughout the entire coral record with their magnitude being dependent
on salinity changes inferred from δ18O-Sr/Ca ratios29 and the intensity of luminescent flood bands19,20,28.
Figure Caption
Coral Ba/Ca versus the Burdekin River flow showing the calibration procedures used to determine
the magnitude of pre-European floods. Modern luminescent bands are shown as strong (dark blue),
average (blue) and weak (light blue) calibrated for flood magnitude using recorded range of maximum
Burdekin River flows for the period from 1921 to 1998 (red line). Luminescent bands for pre-European
floods are shown as strong (dark grey), average (light grey) and weak (open), with flood magnitudes
determined using the modern calibration (green line).
Salinity changes at the reef site were also
independently determined using δ18O-Sr/Ca relationships29. These are shown as red circles for modern
floods and green for pre-European floods. Open circles connect measured Burdekin River flow (weekly
maximum) to the modern flood events used to calibrate δ 18O-Sr/Ca relationships. Both approaches show
that following European settlement, Burdekin River flood plumes are distinguished by having much higher
Ba/Ca ratios and hence higher (x5 to x10) sediment fluxes for the same magnitude flood event.