Upper Quaternary Sea Levels, Coral Terraces, Oxygen Isotopes and Deep-sea Temperatures. John CHAPPELL *

Transcription

1 Journal of Geography 103 (7) Upper Quaternary Sea Levels, Coral Terraces, Oxygen Isotopes and Deep-sea Temperatures John CHAPPELL * Abstract There are two widely used sea level curves for the last glacial cycle, one based on coral terraces at Huon Peninsula (HP) in Papua New Guinea, the other on oxygen isotope data from deep sea cores. Previous sea level estimates from HP are m higher than isotopic sea levels for the interval ka but new sea level data from HP support the isotopic sea levels. Computer simulations of coral terraces are more similar to observed HP terraces when the computer model is driven by the isotopic sea level curve than by the previous HP sea level curve. HP sea levels and benthic isotope records were used previously to calculate deep sea temperatures through the last glacial cycle the new HP results lead to a sharper indication that deep sea temperatures were up to 1.8 Ž cooler during the last glacial cycle than in the interglacials. The temperature estimates depend on the isotopic composition of Pleistocene ice sheets. However, if deep sea temperatures did not change, the HP sea levels require unrealistically low oxygen isotope values for the Pleistocene ice sheets. The model of cooler deep sea temperatures during glacial cycles is adopted here and was used to derive isotopic sea level changes using the V19-30 record over the last 330 ka. The isotopic sea levels before the last interglacial are consistent with the limited data available but need to be tested by high quality data from older coral terrace sequences. I. Introduction There are two widely-used sources of information about Late Quaternary glacio-eustatic sea level changes. One source is oxygen isotope data from deep sea cores (Shackleton, 1987) and the other comes from raised coral reef terraces that have been dated by the 230Th/234U method, principally at the Huon Peninsula (HP) in Papua New Guinea (Bloom et al., 1974 ; Chappell and Shackleton, 1986). Sea level estimates from coral terraces so far are restricted to the last few hundred thousand years but deep sea isotope records extend through the Quaternary into the Tertiary. However, there are differences between the two types of sea level curve that need to be resolved (Fig. 1), before oxygen isotopes are used to give sea levels for periods beyond the range of dated coral terraces. Sea level estimates by both methods rely on certain assumptions. In the case of coral terraces, the principal assumption is that tectonic uplift rate has been constant at a given terrace transect (Bloom et al., 1974 ; Chappell and Shackleton, 1986). For oxygen isotopes, the School of Pacific and Asian Studies, Australian National University. Canberra, ACT, 0200 Australia. 828

2 Fig. 1 Isotopic sea levels derived from deep sea core oxygen isotope data, from Shackleton (1987) (continuous line through open boxes), and sea levels derived from coral terraces at Huon Peninsula (HP) Papua New Guinea (small solid circles) Huon Peninsula points are from the HP2 data set tabulated by Chappell and Shackleton (1986). principal assumption is that the variation of mean oceanic d18o with ice volume is linear. Additio aally, a method for identifying and subtracting the effect of oceanic temperature changes must be devised (Shackleton, 1987). In this paper, the differences between the isotopic and coral terrace sea levels that are shown in Fig. 1 are examined by two methods. First, recent surveys and new 230Th/234U data from the HP terraces are reviewed. Second, a computer model of reef growth with changing sea level and tectonic uplift is used to simulate the formation of coral terraces. terraces are then compared with surveyed Simulated terrace sequences at HP. It is shown by both methods that the isotopic sea level curve of Shackleton (1987) is more accurate than earlier curves :rpm HP by Bloom et al. (1974) and by Chappell and Shackleton (1986). The relationships between ice volumes, deep sea temperatures and isotopes are then re-examined and the conclusion, reached by Chappell Shackle ton (1986), that and deep sea temperatures were colder during the last glacial phase than during the Holocene or the last interglacial phases is upheld. Using this result, sea levels for the last 3 glacial cycles (the last 320 ka= the last 320,000 years) are reconstructed from deep sea isotopic data. Computer simulations of reef terraees driven by this new 320 ka sea level curve are found to compare closely with surveyed terrace flights at HP and Timor. II. New Sea Level Data from Huon Peninsula Late Quaternary sea levels published by Chappell and Shackleton (1986) were based on field and230th/234u data from HP by Veeh and Chappell (1970), Bloom et al. (1974), Chappell and Veeh (1978 a) and some additional field data from Chappell (1983). New age measurements based on samples collected in 1988 from HP terraces in the age-range from 30 to 75 ka, done by electron spin resonance and 230Th/234U methods, showed some differ- 829

3 ences from the earlier results (Grun et al., 1992). Stein et al. (1992) published TIMS 230Th/234U results for samples that define the age range of the upper part of the last interglacial reef complex VII. In reviewing all previous alpha-spectrometric age measurements from HP, Omura et al. (1994) showed that several results failed strict criteria for acceptability and that several others age-errors. Previous discussion of HP sea levels has had substantial relied on generalised field descriptions of Chappell (1974) and height surveys done by Chappell in 1973 that were used by Bloom et al. (1974). Recognising that earlier sea level estimates may be affected by several sources of error, a new expedition to Huon Peninsula was led by Y. Ota and J. Chappell in 1992, with further precise dating, and careful resurvey of the terraces and their stratigraphy objectives (Ota, 1994). amongst the Coral terrace sea levels usually are calculated using the formula Si= (Hi+Di)-U.ti, with U=(H*-S*)/t* (1) where Si is sea level at time ti, Hi is the elevation of a reef feature with the measured age ti and Di is the estimated depth of water in which this reef feature formed, and U is the uplift rate, which is assumed to be uniform at a given field section. (1) gives negative values for sea levels lower than the present level. This convention is used throughout the paper. U is estimated from a reference terrace of elevation H*,age t* and palaeo sea level S*. The reference terrace that was used previously is the crest of the last interglacial terrace (terrace VII at HP) ; last interglacial sea level (S*) was taken as about +5 m (Bloom et al., 1974 ; Chappell and Veeh, ; Chappell and Shackleton, 1986). Uplift rate varies along the HP coast from <1 m/ka in the northwest to> 3 m/ka in the southeast and Chappell (1974) concluded that U may also have varied at each locality. Bloom et al. (1974) and Chappell and Shackleton (1986) assumed that U has been constant since the last interglacial at each locality. This assumption is supported by Ota et al. (1993), who found that the average Holocene uplift rate and the rate since the last interglacial were the same at all HP localities that they examined. The depth-correction Di was neglected in previous studies (Bloom et al., 1974 ; Chappell and Shackleton, 1986), where it was assumed that the age of a given reef terrace was the same the age of dated corals from the reef underlying the terrace. Sea levels Si thus were based on heights Hi of reef terraces. Several criteria must be met for sea level estimates to be judged as reliable. A field section must extend at least to the reference terrace (VII) ; it must not be interrupted by faults, and stratigraphic relationships between terraces must be identified and palaeo depths (Di) must be identifiable for each dated specimen. Finally, sea levels calculated by (1) should agree between uplift rates. different sections with different Previous HP sea levels are up to 40 m higher than isotopic sea levels of Shackleton (1987) in the age-range from 30 to 75 ka (Fig. 1). The HP coral terraces in this age range are associated with reef complexes II, III and IV. Each complex has more than one terrace, and complex III has at least two or perhaps three separate offlapping reefs (llla, Mb and IIIc) (Chappell, 1974, 1983). Two sections where reef complexes from II to VII are present and all the above field criteria are met (Kanzarua= KANZ, and Bobongara-= BOBO), and the Sialum-Kwambu lies beneath section (KW AM) where reef II Holocene reef I, were closely reexamined during the 1992 expedition. Topog- 830

4 Fig. 2 New sea level data from HP compared with the isotopic sea level curve of Figure 1 Data between 30 and 75 ka are from studies based on 1992 expedition to HP led by Y. Ota (Chappell et al., 1994 ; Omura et al., 1994, and in press). Data between 6 and 12 ka are from Edwards et al. (1993), and results of Stein et al. (1992) are added at 120 ka. Other HP points are same as Figure 1. raphy of these sections was surveyed in 1988 and in 1992 ; the closely (Ota, 1994). Stratigraphy two sets of surveys agree of the offlapping reef sequences from I to VII was carefully logged at BOBO and KANZ in 1992 by Dr J. Pandolfi (Pandolfi and Chappell, 1994 ; and in prep.). Samples for dating, collected in 1992 from the BO- BO, KANZ and KWAM sections, are tabulated by Omura et al. (1994) who also summarise diagene tic status of each sample. Not all samples proved to be suitable for dating. New 230Th/2U age measurements of suitable sam - ples were done by alpha spectrometry at Kanazawa University (Omura et al., 1994), and several were dated by the TIMS method at Australian National University by T. Esat and M. McCulloch. All new dates from reefs II to IV, together with relevant values of Hi, Di and U, are listed by Chappell et al. (1994) and Omura et al. (subm.), who also give values and standard errors of Si for each sample, calculated using (1). repeated here. The primary data are not Revised sea levels from HP for the last 140 ka are plotted against Shackleton's (1987) isotopic sea level curve in Fig. 2. HP data in the range ka are from the new results for reefs II to IV (Chappell et al., 1994 ; Omura et al., subm.), except for the low point at 65 ka which is based on the undated base of reef complex Ma. Low sea levels at 18 and 140 ka, and all points from 80 to 125 ka are from Chappell and Shackleton (1986) ; no new data have been established for these points. The timing of the last interglacial high sea level is based on the results for reef VII of Stein et al. (1992). Comparison of Figs. 1 and 2 shows that the new results from HP in the critical time range from 30 to 75 ka agree much more closely with Shackleton's sea levels than did the previous estimates. isotopic 831

5 Fig. 3 Observed versus computer simulated terraces at Kanzarua transect, HP Simulations ISO and HP, respectively, are driven by isotopic and previous HP sea level changes that are shown in Figure 1. Simulation shows surfaces at 1 ka intervals. Note that simulation ISO more closely resembles the observed profile (OBS) than does the HP simulatjon, especially for lower reefs III B, DIA and IV. III. Computer Simulation of Coral Terraces A computer program was written to simulate the development of coral terraces on a tectonically rising coast, with changing sea level. Comparison of simulated terraces with surveyed field examples provides a means for testing sea level curves. The program, developed from that described by Chappell (1980), uses the following principles. A primary bedrock surface is uplifted relative to a datum level at a given tectonic rate (U), and sea level is allowed to vary relative to the same datum. Simulated reef grows on the prior surface up to but not exceeding instantaneous sea level. The vertical rate of reef growth is assumed to be constant (R) down to a critical instantaneous depth (Z,), below which it declines linearly and becomes zero where the instantaneous depth equals the maximum depth of reef growth, Z,n. The reef surface is computed at successive time steps with the new initial surface being the surface at the end of the previous time step. Different from the program described by Chappell (1980), no allowance is made for reduced coral growth in back reef situations. start of a simulation. Values of U and R are input at the Model runs described below used R values in the range of 5-8 m/ka, which is typical of results from radiocarbon dated drillholes in the Great Barrier Reef (Hopley, 1982). Zc was set at 15 m but results with Zc= 20 m were very similar. Zin was set at 50 m. Two simulated terrace flights (ISO and HP) are compared with the surveyed KANZ transect in Fig. 3. The KANZ uplift rate of U= 2.8 m/ka (Chappell and Shackleton, 1986) was used for both simulations and shallow water reef growth rate (R) was set at 7 m/ka. The simulations show successive reef surface profiles at 1 ka intervals from 140 ka to the present. A simple rectilinear slope was used to simulate the prior bedrock surface, which field observations show to be a reasonable approximation for HP. Simulations ISO and HP were driven by the ISO and HP sea level curves 832

6 Table 1 Comparisons between observed and simulated terrace sequences (HP terraces I to VII) Height and width percentage errors were derived using equation (2). Results are for OBS versus ISOdriven and HP-driven simulations, respectively. of Fig. 1, respectively. Visually, the surveyed KANZ section is more similar to the ISO simulation than the HP simulation. The simulated growth structure of the reef sequence, with transgressive or vertical growth within each reef and offlapping relationships is generally between reefs, consistent with the observed stratigraphy described by Chappell (1974) and, in more detail, by Pandolfi and Chappell (1994). The observed and simulated sequences can be compared in terms of terrace elevations and widths by the following formulae : where Hi, hi and Wi, wi are simulated and (2) observed terrace elevations and widths, respectively, for each of the i terraces. Mh is the mean height interval between successive major terraces in the observed sequence from reef I to reef V JI, and My, is the mean observed terrace width. Summations are from i 1 to n, where n is the number of terraces in the sequence. Absolute magnitudes (ABS) are used to ensure that all terms are positive. The smaller the difference scores (H' and W') the better the match between simulations and observations. Values of H' and W' for ISO and HP simulations versus observed terrace profiles for the BOBO, KANZ and KW A transects are listed in Table 1. Results are substantially smaller for the ISO results than for the HP results, showing that a better match was obtained in each case with ISO than with the HP simulation. The results of the simulation experiments, together with the new sea level data reviewed in the previous section, support the isotopic sea level curve for the last 140 ka derived by Shackleton (1987). New sea level data from HP in the age range ka overlap the isotopic sea levels, within errors. Simulated terrace sequences generated by the isotopic sea level curve match the observed terraces closely. It is concluded that sea level estimates by Chappell and Shackleton (1986) in the age range from 30 to 75 ka were too high and that the isotopic curve is better. Sea levels curve could be tuned to produce best-fitting simulations of the terrace sequences but this is not attempted, owing to the approximations growth model. used in the reef IV. Oxygen Isotopes and Sea Level before the Last Interglacial Sea levels based on coral terraces are critically dependent on age measurements. The terrace sequences at HP extend to perhaps 400 ka but accurate last interglacial 230Th/234U dating beyond the is difficult owing to coral dingenesis, so that sea levels based on the older terraces are much less accurate than those derived from the young terraces. However, 833

7 sea levels before the last interglacial can be derived from deep sea isotopic records by a method similar to that used by Shackleton (1987) for the last 140 ka. This is now examined. Shackleton (1987) derived the isotopic sea level curve in Fig. 1 from 6180PDB measurements of the benthic foraminifer Uvigerina in east equatorial Pacific core V The 6180 values reflect a combination of deep sea temperature, global ice volume and the mean isotopic composition of the ice. The effect of deep sea temperature changes at V were eliminated by assuming that surface water temperature variations at equatorial core site RC were negligible during the past 140 ka. differences between the benthic V and planktic RC records were taken, smoothed and normalised so that the smoothed interglacial difference is zero, and subtracted from the V benthic record (Shackleton, 1987). Sea level was derived by using where ƒâ8oi is the value at point i in the reduced V benthic record, ƒâ18oo is the average value for the last few thousand years, and K approximately= 100. The average, smoothed 6180 difference between V and RC (about ) should represent the isotopic temperature factor in core V (Shackleton, 1987). The value of 0.4 ñ implies that deep water temperature at the V site was about 1.5 C cooler (3) change is (4) where T is temperature, and subscripts o and i refer to the mean late Holocene value and the value at point i, respectively. Deep water temperature changes for the last glacial cycle estimated by Chappell and Shackleton (1986), using (4) and the HP 2 sea level data (Fig. 1), are shown in Fig. 4 a. Estimates derived from the new HP sea level data (Fig. 2) are shown in Fig. 4 b. The earlier results suggest glacial temperatures cooler than interglacials by Ž with substantial scatter and a trend towards coolest temperatures at about 40 ka. The new results show less variation and indicate that deep water temperature through the last glacial period was uniformly about 1.8 C cooler than in the interglacials. This is very similar to Shackleton's (1987) estimate. Isotopic sea levels can be derived for earlier glacial cycles by assuming that the deep water temperature model of the last cycle, shown in Fig. 4 b, was repeated in earlier glacial cycles. The model holds temperature constant through each glacial cycle, 1.8 Ž below the interglacial temperature, with ramps joining the glacial and interglacial segments. Data in Fig. 4 are not sufficiently sharp to specify the duration of each ramp but a duration of 10 ka is assumed. This model was applied to the full V isotopic record, which extends to 340 ka. Resulting isotopic sea levels are plotted in Fig. 5. during the last glacial cycle than in interglacial isotope stages 1 and 5 e (the temperature fractionation coefficient for calcium carbonate is-0.23 cho. Ž-1). This is similar to the The raw V ƒâ80 data at approximately 0.5 ka intervals were provided by Professor N. J. Shackleton of Cambridge University. The data ware smoothed with a centre-weighted result obtained by Chappell and Shackleton 5-point filter (weights ) before 0.4 % (1986), who inferred temperature changes from the V record by subtracting an isotopic sea level factor based on HP sea levels and formula (3). For point i, the temperature (equivalent to 1.8 Ž) was subtracted from data points representing glacial periods. A different treatment was applied to isotope stage stage 7, however, for reasons explained below. Time- 834

8 a) b) Fig. 4 Equatorial east Pacific deep water temperatures at site V 19-30, calculated from oxygen isotope and HP sea level data (a) results of Chappell and Shackleton (1986) (b) results using the revised HP sea levels shown in Figure 2. proportionate fractions of 0.4 % were subtracted through the transitional ramp segments (stages 1-2, 5 d-5 e and 8-9). Revised HP sea levels (Fig. 2) are comparable with the extended isotopic sea level curve for the last glacial cycle (Fig. 5), as expected, but there are few cross-check data beyond 130 ka. High "interglacial" sea levels in isotope stages 7 and 9 are represented by older reef complexes at HP, Barbados and Timor, an d by raised coastal terraces dated by tephras in Japan and New Zealand (Chappell, 1974, 1983 ; Chappell and Veeh, 1978 b ; Radtke et al., 1988 ; Ota and Machida, 1987 ; Pillans, 1987) but tectonic uplift and substantial age-error terms preclude exact estimation of sea levels. Coastal beach-barrier formations at present sea level in New South Wales on the stable Australian continent have thermoluminescence ages comparable to stages 1, 5 e and 7 (Roy et al., 1992), indicating level in stage 7 was similar that sea to that of interglacial stages 1, 5 e and 9. However, if the glacial-interglacial temperature cycle of Fig. 835

9 Fig. 5 Extended isotopic sea level curve from 0 to 330 ka, derived from V isotopic data and simple model of deep water temperature changes (see text) Note that raw V19-30 data were smoothed, so that maxima and minima are less "peaky" than in Figures 1 and 2. Solid symbols represent revised HP sea level data from Figure 2. 4 b is applied to the smoothed isotope data for stage 7, the isotopic sea level peaks lie 30 m below present level. Stage 7 sea levels in Fig. 5 were obtained by subtracting 0.3 ñ from the isotopic data. If this procedure is appropriate, it follows that full interglacial conditions were not established in the deep water during isotope stage 7. Further research is needed to examine this. The 330 ka isotopic sea level curve was tested by comparing computer-simulated coral terraces with observed sequences. Terraces at Atauro Island, Timor, described by Chappell and Veeh (1978 b) are particularly suitable for this purpose. Any height differences between simulated and observed terraces will be large relative to terrace height spacing owing to the low uplift rate at Atauro (0.5 m.ka-1 Stratigraphic r elationships between terraces, raised reefs and the underlying stratovolcanic /234U but older reefs could not be dated owing to advanced diagenesis. The terrace flight was accurately surveyed. A computer simulation driven by the 330 ka isotopic sea level curve, with uplift (U)=0.5 m. ka-1 and the same coral growth as before, proved to be closely comparable with the Atauro sequence in terms of terrace heights, although less so in terms of terrace widths (Fig. 6). Stratigraphically the observed transgressive nature of each reef is simulated and the composite nature of Atauro reef 3 is replicated by reef components in the simulation that were generated by the sea level rises to isotope stages 7 a and 7 c. In summary, the extended isotopic sea level curve is supported by the Atauro (Timor) data and simulation. V. Ice Sheets, Isotopes and Deep Ocean Temperatures, slope are clearly exposed, as is the internal stratigraphy of each raised reef. Reefs of isotope stages 5 c and 5 e were well dated by 230Th Isotopic sea levels in Fig. 5 assume that deep water temperatures were about 1.8 Ž colder during glacial cycles. Mix and Pisias (1988) 836

10 Fig. 6 Observed and simulated coral terraces up to 330 ka at Atauro Island, Timor, which has a relatively low uplift rate (0.5m.ka-1) Simulated surfaces at 1 ka intervals as for Figure 3. Observed terraces Alb and A2 are U-series dated (104 and ka, respectively) A3 and A4 are inferred to represent isotope stages 7 and 9. Numbers on simulated terraces= ages in ka from isotopic sea level curve of Figure 5. Observed regressive (R) deposits within reef A3 match R-surfaces in simulated 213 and 237 ka reefs. Fig. 7 Mean oxygen isotope values for northern ice sheets during last glacial period, calculted by equation (5) assuming that deep sea temperature was the same in glacial and interglacial phases(solid symbols), or that glacial temperature was 1.8 C cooler than for interglacials (crosses) argued that this is unlikely because it could imply unrealistically rapid ocean turnover. Before Figs. 4 and 5 are accepted, the question of ice sheet isotopic values should be examined. To a first approximation, the shift DƒÂ18O of the average oceanic isotope composition, corresponding to a sea level change S, is (5) where H is mean ocean depth (3,790 m) and is the mean isotopic value of the ice that forms to produce S. Equation (5) assumes that the isotopic composition of the glacial-age sheets was constant for all values of S and that ocean volume is proportional ice to ocean depth. It follows from (5) that relationship (3) implies that the Pleistocene ice sheets had a mean 837

11 isotopic value of-37 ñ. Two alternatives are now examined. Either the deep water temperature was did not change and the V isotopic variations were caused solely by ice volume changes, or the deep water temperatures varied according to Fig. 4 b. Results for these alternatives calculated by (5) are shown in Fig. 7. The "constant temperature" assumption gives scattered, highly negative 6180i values (-55 %o to ), while the "cooler glacial temperatures" model gives value of-37 %o (as expected from eq. 3), with relatively little scatter. Results suggest that the constant temperature assumption is unrealistic because most excess ice during the last glacial cycle was on northern continents at similar latitudes to the Greenland ice sheet, which shows mean values around 35 to through the last glacial -period (Dansgaad et al., 1971). Hence, the "colder deep water" model is favoured. VI. Conclusions shown in Fig. 2, is adopted. Calculations using the new HP sea levels and isotopic data from V indicate that deep sea temperatures were about 1.8 cooler during the last glacial cycle than in the interglacials. Temperature estimates based on the new HP sea levels show less scatter than earlier results of Chappell and Shackleton (1986) and are consistent with Shackleton's (1987) deep water temperature estimates, which were based on benthic-planktic comparisons. However, deep sea temperatures based on sea levels and oxygen isotopes depend on an assumed relationship between sea level and the isotopic composition of sea level, which in turn assumes that oxygen isotopes in the Pleistocene ice sheets had a constant average value around-37 %o. The possibility that deep sea temperatures were constant was tested by calculating ice sheet PO values using the new HP sea levels. Results give highly scattered and unrealistically negative values of 61801, and the model of cooler deep sea temperatures during glacial cycles is New sea level data from HP, in the time range from 30 to 75 ka, agree within standard errors with isotopic sea levels estimated by Shackleton (1987) from the V deep sea benthic ƒâ180 record. Earlier estimates from HP of sea levels in this time range appear to be m too high. HP sea levels for the period 80 to 120 ka sea levels are consistent with Shackleton's isotopic sea levels and have not been revised. Computer simulations of coral reef growth on tectonically rising terrain produce offlapping reef terraces when the model is driven by fluctuating sea level. Simulations driven by the isotopic sea level curve match actual terrace flights more closely than simulations driven by the HP sea level curve of Chappell and Shackleton (1986). Hence, the new HP sea level data for the last glacial cycle, adopted. Isotopic sea levels were extended to 330 ka, using the V record. The deep-water temperature pattern deduced for the last glacial cycle was assumed to have also occurred in earlier glacial cycles, and the raw 00 data were adjusted on this basis. High sea levels before the last interglacial, on the extended isotopic sea level curve, are supported by observations from Australia, New Guinea and elsewhere although well dated data are sparse. The 330 ka isotopic sea level curve is supported by computer simulations, driven by the curve, which match observed flights of coral terraces. However, isotopic sea levels before the last interglacial have yet to be tested by exact 230Th/284U dating of older reef reef sequences. An interesting result, which warrants 838

12 closer study, is that observed sea levels in isotope stage 7 appear to require cooler deep water temperatures than the other interglacial stages. Acknowledgements The author particularly wishes to thank Professor Y. Ota for facilitating the 1992 Huon Peninsula ex. pedition, and Professor N. J. Shackleton of Cambridge University for providing the raw data file for core V The support of Monbusho International for the 1992 HP expedition is gratefully acknowledged. The Papua New Guinean people of the Huon Peninsula are thanked for their long-standing forebearance and assistance to field operations in their country. References Bloom, A. L., Broecker, W. S., Chappell, J., Matthews, R. K. and Mesolella, K. J. (1974) : Quaternary sea level fluctuations on a tectonic coast : New 233Th/234U dates from the Huon Peninsula, New Guinea. Quat. Res., 4, Chappell, J. (1974) : Geology of coral terraces, Huon Peninsula, New Guinea : A study of Quaternary tectonic movements and sea level changes. Geol. Soc. Am. Bull., 85, Chappell, J. (1980) : Coral morphology, diversity, and reef growth. Nature, 286, Chappell, J. (1983) : A revised sea level record for the last 300,000 years from Papua New Guinea. Search, 4, Chappell, J. and Shackleton, N. J. (1986) : Oxygen isotopes and sea level. Nature, 324, Chappell, J. and Veeh, H. H. (1978 a): 2"Th/23'U support for an interstadial sea level of-40 m at 30,000 years B. P. Nature, 276, Chappell, J. and Veeh, H. H. (1978 b) Late Quaternary tectonic movements and sea level changes at Timor and Ataura Island. Geol. Soc. Am. Bull., 89, Chappell, J., Omura, A., McCulloch, M., Esat, T., Ota, Y. and Pandolfi, J. (1994) : Revised late Quaternary sea levels between 70 and 30 ka from coral terraces at Huon Peninsula. In Ota, Y. ed. : Study on coral reef terraces at Huon Peninsula, Papua New Guinea. A preliminary Report on Project , Monbusho International Research Program, Japan, Dansgaard, W., Johnsen, S. J., Clausen, H. B. and Lancway, C. C., Jr. (1971): Climatic record revealed by the Camp Century ice core. In Turekian, K. K. ed. The Late Cenozoic glacial ages. Yale University Press, New Haven, Edwards, R. L., Beck, J. W., Burr, G. S., Donahue, D. J., Chappell, J., 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, Grun, R., Radtke, U. and Omura, A. (1992): ESR and U-series analysis on corals from Huon Peninsula, New Guinea. Quat. Sci. Rev., 11, Hopley, D. (1982) The geomorphology of the Great Barrier Reef. Wiley Interscience, New York, 453 p. Mix, A. C. and Pisias, N. (1988): Oxygen isotope analyses and deep sea temperature changes : Implications for rates of ocean mixing. Nature, 331, Omura, A., Chappell, J., Bloom, A. L., Pillans, B., MaCulloch, M., Esat, T., Sasaki, K. and Kawada, Y. (1994) Alpha-spectrometric 230Th/234U dating of Pleistocene corals. In Ota, Y. ed. : Study on coral reef terraces at Huon Peninsula, Papua New Guinea. A preliminary Report on Project , Monbusho International Research Program, Japan, Ota, Y. (1994): Study on coral reef terraces at Huon Peninsula, Papua New Guinea. A preliminary Report on Project , Monbusho International Research Program, Japan, 171 p. +Appendices. Ota, Y., Chappell, J., Kelley, R., Yonekura, N., Matsomoto, E., Nishimura, T. and Head, J. (1993): Holocene coral reef terraces and coseismic uplift of Huon Peninsula, Papua New Guinea. Quat. Res., 40, Ota, Y. and Machida, H. (1987): Quaternary sea level changes in Japan. In Tooley, M. J. and Shennan, I, eds. : Sea-level changes. Basil Blackwell, Oxford, Pandolfi, J. and Chappell, J. (1994): Stratigraphy and relative sea level changes at the Kanzarua and Bobongara sections, Huon Peninsula, Papua New Guinea. In Ota, Y. ed. : Study on coral reef terraces at Huon Peninsula, Papua New Guinea. A preliminary Report on Project , Monbusho International Research Program, Japan, Pillans, B. J. (1987): Quaternary sea level changes : Southern hemisphere data. In Devoy, R. J. N. ed. : Sea surface studies. Croom-Helm, London, Radtke, U., Grun., R. and Schwartz, H. (1988) : Electron spin resonance dating of the Pleistocene 839

13 reef tracts of Barbados. Quat. Res., 29, Roy, P. S, Zhuang, W-Y, Birch, G. F. and Cowell, P. J. (1992) : Quaternary geology and placer mineral potential of the Forster-Tuncurry shelf, southeastern Australia. Geological Survey Report : GS 1992: 201, Department of Mineral Resources, NSW, 164 p. Shackleton, N. J. (1987) : Oxygen isotopes, ice volume and sea level. Quat. Sci. Rev., 6, Stein, M., Wasserburg, G. F., Aharon, P., Chen, J. H., Zhu, Z. R., Bloom, A. L. and Chappell, J. (1992) : TIMS U-series dating and stable isotopes from the last interglacial event in Papua New Guinea. Geochim Cosmochim Acta, 57, Veeh, H. H. and Chappell, J. (1970) : Astronomical theory of climate change : Support from New Guinea. Science, 167, (Accepted 18 November, 1994) 840