Ca ratios of a porites coral from Sanya Bay, Hainan Island, South China Sea and their relationships to sea surface temperature

Ca ratios of a porites coral from Sanya Bay, Hainan Island, South China Sea and their relationships to sea surface temperature

Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74 www.elsevier.nl/locate/palaeo Mg/Ca, Sr/Ca and U/Ca ratios of a porites coral from...

430KB Sizes 0 Downloads 16 Views

Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74 www.elsevier.nl/locate/palaeo

Mg/Ca, Sr/Ca and U/Ca ratios of a porites coral from Sanya Bay, Hainan Island, South China Sea and their relationships to sea surface temperature Gangjian Wei a,b, *, Min Sun b, Xianhua Li a, Baofu Nie c a Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, PO Box 1131, Wushan, Guangzhou, People’s Republic of China b Department of Earth Sciences, The University of Hong Kong, Pokfulam Road, Hong Kong, Hong Kong c South China Sea Institute of Oceanology, CAS, Guangzhou, People’s Republic of China Received 25 March 1999; accepted for publication 20 March 2000

Abstract High-resolution Mg/Ca, Sr/Ca and U/Ca ratios are reported in the skeleton of Porites sp. collected from Sanya Bay, on the south coast of the Hainan Island, China using inductively coupled plasma mass spectrometry combined with isotope dilution technique (ICP–MS–ID). The coralline Mg/Ca ratios covary with sea-surface temperatures (SST ) for the 10 year interval, giving the relationship between Mg/Ca and SST as T (°C )=(−14.13±0.31)+ (8.846±0.068)×103Mg/Ca with the mean square weighted deviation (MSWD) of 9.7. Using this equation, we are able to reconstruct SST record with a precision better than ±0.5°C. Several Sr/Ca and U/Ca ratios show poor linear relationships with the measured SST. This study indicates that the growth rates of the coral do not significantly affect coralline Mg/Ca, Sr/Ca and U/Ca ratios. However, the high meteoric precipitation in this area, which may cause a heavy run-off from the Island, may change seawater Sr/Ca and U/Ca ratios and ruin the relationships between coralline Sr/Ca or U/Ca ratios and SST. This implies that in coastal areas like Hainan, coralline Mg/Ca ratios are valid for reconstructing the SST records, but Sr/Ca and U/Ca ratios may be suspected. © 2000 Elsevier Science B.V. All rights reserved. Keywords: coral; Mg/Ca, Sr/Ca, and U/Ca ratios; sea-surface temperature; South China Sea

1. Introduction The concentrations of Sr, Mg and U in coral skeletons are considered to be potential proxies of sea surface temperature (SST ) ( Weber, 1973; Smith et al., 1979; Swart, 1981; Swart and Hubbard, 1982; Shen and Sanford, 1990; Beck et al., 1992; Shen * Corresponding author. Present address: Gangjian WEI, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, PO Box 1131, Wushan, Guangzhou, People’s Republic of China. E-mail address: [email protected] (G. Wei)

et al., 1992, 1996; de Villiers et al., 1994; McCulloch et al., 1994, 1996; Min et al., 1995; Shen and Dunbar, 1995; Mitsuguchi et al., 1996; Alibert and McCulloch, 1997). Beck et al. (1992) reported several high-quality coralline Sr/Ca data by the thermal ionization mass spectrometry–isotope dilution method (TIMS–ID). Their data covary with the SST records, and consequently, a coral Sr/Ca thermometer was constructed. More recent studies confirmed that coralline Sr/Ca ratios are dominantly controlled by SST and are suitable for reconstruction of paleo-SSTs (de Villiers et al., 1994, 1995; McCulloch et al., 1994, 1996; Alibert

0031-0182/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S0 0 3 1 -0 1 8 2 ( 0 0 ) 0 0 10 5 - X

60

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

and McCulloch, 1997). Several recent studies showed that the U/Ca and Mg/Ca ratios in coral skeleton are also closely related with SST (Min et al., 1995; Shen and Dunbar, 1995; Mitsuguchi et al., 1996). These two ratios have steeper relationships with respect to temperature and therefore should be potentially easier to measure than Sr/Ca. As mentioned in the literature, biological factors and environmental factors other than SST can affect trace element intake of corals (Swart, 1981; de Villiers et al., 1995). It is necessary to consider such factors when using Mg/Ca, Sr/Ca and U/Ca ratios in a coral skeleton to reconstruct paleoSSTs. For instance, growth rate, a comprehensive indicator of biological factors, and species differences may significantly affect the coralline Sr/Ca ratios (de Villiers et al., 1994, 1995). To minimize such interference, recent studies have tended to take samples along the optimal growth axis within coral colonies. Unfortunately, agreements among the coral Sr/Ca thermometers from different areas are still poor, even if they are constructed by the same species in similar growth rates (de Villiers et al., 1995; Shen et al., 1996; Sinclair et al., 1998). Similar problems also exist for U/Ca and Mg/Ca thermometers. Min et al. (1995) reported two U/Ca thermometers constructed on Porites from Tahiti and New Caledonia, respectively, with the temperature disparity up to 1–2°C. Wei et al. (1998) derived a Mg/Ca thermometer for Porites collected from the north offshore of the South China Sea. Compared with that of Mitsuguchi et al. (1996), this Mg/Ca thermometer gives a temperature disparity of about 1°C in SST <25°C, but less disparity in SST >25°C. It is likely that the relationships between SST and coralline Mg/Ca, Sr/Ca as well as U/Ca ratios may not always be the same in different marine environments. Therefore, the variation in seawater Mg/Ca, Sr/Ca and U/Ca ratios should be considered. All these thermometers can only be valid for corals growing in seawater with constant Sr/Ca, U/Ca and Mg/Ca ratios. Because the residence time of Ca, Sr, Mg and U in open marine is quite long (Goldberg and Arrhenius, 1973; Cochran, 1992), this prerequisite is generally satisfied. However, in some marginal seas, such as the South China Sea, Sr/Ca ratios of seawater in reef areas are much

lower and vary within a greater range (Guo et al., 1992) than those of other reef areas (de Villiers et al., 1994). This may significantly affect the relationships between coralline trace element ratios and SST in this area. The South China Sea, which partly involves the West Pacific Warm Pool, plays an important role in controlling the East Asian monsoon climate ( Wang, 1999). The large number of coral reefs in SCS can be used as valuable archives for paleoclimate reconstruction (Nie et al., 1996; Wei et al., 1999). Consider that the SCC is the largest margina sea of the west Pacific with a semi-close environment, coral reefs in SCS are potential in examining whether coralline Mg/Ca, Sr/Ca and U/Ca ratios are valid for paleo-SST reconstruction in marginal sea environments. In this study, using inductively coupled plasma–mass spectrometry and the isotope dilution technique (ICP–MS–ID), Mg/Ca, Sr/Ca and U/Ca ratios were determined for the skeleton of a Porites sp. collected from the south offshore of the Hainan Island, China. Our data cover the 10 years from 1977 to 1986, with a sampling density of more than 12 samples per year. The relationships between these ratios and the instrument-measured SSTs at a nearby hydrometric station were examined. Further evaluation of these results with climatic records, such as meteoric precipitation at Hainan Island, was made to better understand environmental control on these ratios in the coral skeleton.

2. Sampling and analytical methods 2.1. Sampling A Porites lutea specimen, 40 cm in diameter, living 2 m below the low tidal level, was collected on 26 October 1996 from Sanya Bay, south offshore of the Hainan Island (18°12∞N, 109°29∞E ) ( Fig. 1). To minimize any growth-rate problems, the whole colonies were carried out of water and cut with a stainless-steel saw. A slab about 1 cm in thickness and 5 cm in width, containing the optimal growth axis, was firstly sliced from the center and further cut to obtain a thin working

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

61

Fig. 1. Simplified map of Hainan Island an Sanya Bay. The Porites coral was collected at Sanya Bay, south offshore of the Hainan Island. The SST records were measured at the Yinggehai hydrometric station.

slab 2 mm in thickness and 3 cm in width along the optimal growth axis. The working slab was immersed in fresh water for 24 h, then rinsed with fresh water and placed in a clean plastic box, ultrasonically washed with deionized water, 10% NaOH, 1% H O , 0.1 N 2 2 HNO and deionized water sequentially to remove 3 organic materials and adherent contaminants. Because high-density bands on the X-ray photograph correspond to cold seasons, and the lowest annual temperatures usually appear in January in this area (Nie et al., 1996), the high-density bands represent annual boundaries. An X-ray photograph of this slice was taken to reveal its density

bands. For the purpose of duplication, two adjacent parallel sets of cubes of 1 mm3 were hand-cut with a thin stainless-steel blade, about 0.1 mm thick, along the edge of the working slab. 2.2. Analytical methods In order to measure the coral Sr/Ca ratio, the TIMS–ID method is used. However, the technique of ICP–MS–ID is commonly used as a method for measuring trace elements in marine carbonate materials, such as coral and foraminifera (Lea et al., 1989; Linn et al., 1990; Shen et al., 1992; Lea and Boyle, 1993; Lea and Martin, 1996).

62

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

Compared with the TIMS–ID method, the ICP– MS–ID technique is less time-consuming and more efficient. However, the data precision using the traditional ICP–MS–ID method is usually insufficient for a thermometer construction (Lecornec and Correge, 1997). Lecornec and Correge (1997) developed an internal standard ICP–MS–ID technology, in which Sc and Y were added as an internal standard and spike for Ca and Sr measurement. This method can significantly improve the precision and makes them suitable as a Sr/Ca thermometer. This study employed Lecornec and Correge’s internal standard ICP–MS–ID technique to conduct the Ca and Sr measurements. Usually, coralline Mg can be precisely measured by ICPAES (Mitsuguchi et al., 1996) or using the external standard calibration method of ICP-MS ( Wei et al., 1999). In order to minimize any deviations between different methods and acquire a higher efficiency, we measured Mg, Ca and Sr at the same time. Similar to the Ca measurement, Sc was used as an internal standard and spike for Mg measurement. Our results indicate that this method can provide reasonably steady and reproducible Sr/Ca and Mg/Ca ratios with a precision of about 0.2 and 0.5% (1s), respectively. The small coral cubes, weighing approximately 1 mg, were precisely weighed and dissolved using 100 ml of 6 N HNO . The solutions were diluted 3 to 6 ml, and then an aliquot of about 1 ml was further diluted to 10 ml for Mg, Sr and Ca measurement. The Mg/Ca and Sr/Ca ratios were measured on

a VG Plasma Quad 3 ICP-MS in the Department of Earth Sciences, University of Hong Kong. About 5 ml of the diluted solution were mixed with an equal amount of Sc+Y solution (20 ppb) to ensure that the analyte solution contained 10 ppb Sc and 10 ppb Y. The instrument was optimized with a 10 ppb multi-element solution by automatically adjusting machine parameters, such as the XYZ position of the plasma, r.f. power, lens voltages and gas flows to attain a maximum signal of 59Co. Some of the machine parameters for the measurement are listed in Table 1. Five-multi-element solutions (FME ) with Mg contents ranging from 10 to 75 ppb, Sr from 30 to 200 ppb and Ca from 1 to 10 ppm were used as external standards, which also contained 10 ppb Sc and 10 ppb Y. Regression equations of 24Mg/45Sc to Mg, 48Ca/45Sc to Ca and 86Sr/89Y to Sr were calculated from the five external standard solutions (Lecornec and Correge, 1997). The correlation coefficients for these equations were all better than 0.9995. A modern Acropora sp. collected from Sanya Bay was used as a monitoring standard to check the reproducibility of this method. The monitoring standard, one blank solution and FME solutions, were analyzed before and after 15 Porites coral sub-samples. The final results of monitoring standard gave averages of Mg=(1133±6.5) mg/g, Ca=(38.14±0.16) wt%, Sr=(8122±21) mg/g and Mg/Ca (atomic ratios)=(4.958±0.027) mmol/ mol, Sr/Ca=(9.716±0.039) mmol/mol, after 10 runs (errors are 1s). The U/Ca ratios were determined on the parallel

Table 1 Main machine parameters for ID–ICP–MS measurement

Machine Mode Mass ICP r.f. power Nebulizer gas flow Auto lens Dwell time per amu Integration time per mass Acquisition time per replicate Replicates

Mg, Ca and Sr measurement

U measurement

VG PQ3 Peak hopping 24Mg, 26Mg, 45Sc, 43Ca, 48Ca, 86Sr, 88Sr, 89Y 1320 W Auto On 10 ms – 1.5 min 6

PE Elan 6000 Peak hopping 235U, 238U 1000 W 0.79 l/min off ( lens voltage 11.3 kV ) 50 ms 5500 ms – 9

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

set of Porites sub-samples. U and Ca analyses were conducted on a PE Elan 6000 ICP-MS in Guangzhou Institute of Geochemistry, Chinese Academy of Sciences. Ca content was measured by the external standard calibration method ( Wei et al., 1999), and the precision was between 0.5 and 1.0% (1s). U contents were measured using the isotope dilution technique for which the 235U isotope spike solution was used. The instrument was optimized with 1 ppb of U solution before measurement to attain a maximum signal of 238U. The main machine parameters for U measurement are also listed in Table 1. The NBS standard U 500 solution was measured during the U analysis. Ten replicated runs of the U 500 gave an average 238U/235U ratio of 1.0354±0.0006. A correction factor for mass bias of the ICP-MS was derived from the ratio of this value to the certified value of 1.0003 (Chen et al., 1986), and used for our samples. A normal U standard solution, routinely measured in the same lab by TIMS, was used for checking the reproducibility. This solution gave an average of 22.43±0.09 (1s) ppb of U content after 10 measurements, consistent with the TIMS result of 22.38±0.07 (1s).

3. Results and discussions 3.1. Sequential variation of Mg/Ca, Sr/Ca and U/Ca ratios One hundred and forty-seven sub-samples cut from Porites living from 1977 to 1986 were measured in this study. The Sr/Ca, Mg/Ca and U/Ca ratios are presented in Table A1 in Appendix A. The estimated growth rate of the sample, according to the density bands, is 11–16 mm/year with an average of 14.5 mm/year. The coralline Mg/Ca, Sr/Ca and U/Ca ratios all vary seasonally, suggesting that temperature is the key factor controlling the concentrations of these elements in the coral skeleton. Among these three ratios, the Mg/Ca ratios show the greatest seasonal amplitude (35%) from 5.138 to 3.778 mmol/mol. This is consistent with data reported by Mitsuguchi et al. (1996) for a Porites from Ryukyu Islands, west Pacific. The seasonal

63

Mg/Ca amplitude in this study is larger than previous data of the same coral from 1981 and to 1983, which were measured using the external standard calibration method on a PE Elan 6000 ICP-MS ( Wei et al., 1999). The maximum Mg/Ca ratio of the previous study is about 1% higher than that of this study, while the minimum Mg/Ca of this study is about 7% lower. This deviation is mainly caused by the sampling resolution. Compared with the monthly resolution of the former study, the sampling resolution of this study is much higher, up to 16–18 samples in most of the year. A higher sampling resolution can provide much more detail of the winter records, but not of the summer records due to the slow winter growth rate (Nie et al., 1996). However, the disparity of these two data sets may be caused by their different analytical methods to some extent. With the external standard calibration method, the former data set has a precision up to 1%. For this study, if samples were measured by the internal standard isotope dilution method, the precision can be improved to less than 0.5%. Moreover, the reproducibility and accuracy were also better than the former. It is still worth noting that the subsamples are parallel samples, not the same. There would be some inherent slight difference between the two sets of sub-samples. The U/Ca ratios vary from 1.042 to 1.338 mmol/mol. The variation amplitudes are similar to the corals from Tahiti and New Caledonia (Min et al., 1995). The Sr/Ca ratios also vary within the range of previous studies (Beck et al., 1992; de Villiers et al., 1994; Alibert and McCulloch, 1997). 3.2. Sea-surface temperature records of the south coast of Hainan Island A continuous 30 year instrument-measured SST record at Yinggehai Hydrometric Station is available. The station is located on the southwestern coast of Hainan Island, about 90 km west of Sanya Bay. The SST difference between Yinggehai and Sanya is supposed to be very small. According to the temperature records of the Meteorological Reports of Guangdong Province ( Zhong, 1997), the monthly average temperatures of these two

64

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

localities differ by less than 0.2°C and vary synchronously with a correlation coefficient of 0.98. Therefore, the SST records from Yinggehai can be used to calibrate thermometer for the coral from Sanya. The weekly average SST records between 1976 and 1988 reveal that the maximum SSTs of each year are not very different, about 30–31°C, and the SSTs generally reach a maximum in July or August of each year. SSTs usually reach a minimum in January, but the values significantly vary from year to year. The warmest winters were in January of 1979 and 1987 with minimum SSTs higher than 23°C. The coldest winters occur in January of 1977 and 1983 with the minimum SST lower than 20°C. These extremely cold winters may be caused by the enhancement of the northeast monsoon in the second year during 1976–1977 and during the 1982–1983 El Nin˜o. 3.3. Relationships of Mg/Ca, Sr/Ca and U/Ca ratios with SST The annual growth interval of the coral is manifested as density bands indicated on the X-ray photograph. The Mg/Ca ratios can be correlated to SST year by year. First, the weekly SST records were smoothed to the resolution as the sampling density, and then, the maximum SST was matched to maximum Mg/Ca and minimum SST to minimum Mg/Ca. The rest of the Mg/Ca ratios, between the maximum and minimum, can be correlated to a new SST curve, which is built by re-averaging the weekly SST record between the ultimate values. Because the Sr/Ca and Mg/Ca ratios are measured on the same sub-sample at the same time, the SSTs matched to Mg/Ca ratios can be directly correlated to Sr/Ca ratios. The right process was also applied to U/Ca ratios later (Fig. 2a–c). The variation of Mg/Ca ratios is well concordant with the instrument SST records (Fig. 2a). The agreement between Mg/Ca ratios and SSTs is better in winter than in summer. Two Mg/Ca maximum (1978 and 1979) results in the calculated SST show temperatures 1°C higher than the warmest SST records of the same years. However, even in the abnormally cold winter of 1976 and 1982, the Mg/Ca ratios still closely match

the SSTs very well. In contrast, the consistency of the Sr/Ca and U/Ca maximum and minimum with the SST minimum and maximum of each year is not as good as that of the Mg/Ca ratios ( Fig. 2b and c). Linear regression for the above data gives the following calibrated equations (Fig. 3), and all the measured data were used in the calculation. In order to evaluate uncertainties for the calculated SST based on regression, we calculated the thermometer equations according to Ludwig (1992) as follows with 1s errors: T(°C )=(−14.13±0.31)+(8.846±0.068)× 103Mg/Ca (MSWD=9.7 and r=0.97) T(°C )=(210.2±2.4)−(19.83±0.26)×103Sr/Ca (MSWD=24 and r=0.75) T(°C )=(81.15±0.57)−(48.14±0.48)×106U/Ca (MSWD=31 and r=0.85). The linear relationship between Mg/Ca ratios and SST is excellent but relatively poor for the relationship between Sr/Ca and SST and between U/Ca and SST. Analytical uncertainties could be partly ascribed to the discrepancy in the Sr/Ca ratios with SST. As estimated from the above equations, the Sr/Ca analysis relative deviation of 0.2–0.5% can attribute to temperature errors of 0.3–0.9°C. However, such effects on the relationships between Mg/Ca or U/Ca ratios and SST are trivial due to the large seasonal variation of Mg/Ca (~30%) and U/Ca (~30%). Considering the large discrepancies, some factors other than temperature seem also play a role in affecting the coralline Sr/Ca and U/Ca ratios of the Porites coral from Sanya. 3.4. Control factors other than temperature 3.4.1. Growth rate Growth rate has been considered to significantly affect coralline Sr/Ca ratios (de Villiers et al., 1994, 1995). However, a recent study by Shen et al. (1996) indicated that the effect of growth rate might have been overestimated. These authors found that the Sr/Ca~SST relationship is the same for the two sections of different growth rate, i.e. 18 and 23 mm/year, respectively, along the same axis. More recently, Alibert and McCulloch (1997) suggested that sampling should be con-

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

65

(a)

(b)

(c) Fig. 2. Variations of Mg/Ca, Sr/Ca and U/Ca ratios of the coral with SST records. (a) Mg/Ca to SST; (b) Sr/Ca to SST; (c) U/Ca to SST. The time on the x-axis corresponds to the beginning of the year.

66

(a)

(b)

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

ducted along the optimal growth axis to minimize the effect of growth rate on the Sr/Ca ratios. For Mg/Ca and U/Ca ratios, the growth rate is not considered to significantly affect these ratios because of the fairly large variation of coralline Mg/Ca and U/Ca ratios (Shen and Dunbar, 1995; Mitsuguchi et al., 1996) Because all our sub-samples were taken along the optimal axis, the growth rate is not thought to be important in affecting the coralline element ratios. To confirm this argument, we used the approach of Shen et al. (1996). The sub-samples can be divided into a low-growth-rate section (LGR) and a high-growth-rate section (HGR). The LGR section contains those between 1978 and 1979 with a growth rate of about 11~12 mm/year, and the remaining part is the HGR section with a growth rate of about 14~16 mm/year, except those of 1983 (Section S-83), which will be discussed separately. The relationships between the Mg/Ca, Sr/Ca and U/Ca ratios and SST for each section are re-calculated, respectively (Fig. 4a–c). The difference in the Mg/Ca~SST relationships between LGR and HGR is small (Fig. 4a). During the winter, the discrepancies for temperatures calculated from the two equations are about 0.5–0.8°C, and during the summer, the discrepancies are less than 0.5°C, i.e. within the 2s errors of the temperature derived from analytical uncertainties. It seems that growth rate does not significantly affect our Mg/Ca ratios. Discrepancies between S-83 and HGR are less than 0.5°C within the seasonal SST oscillation range from 18 to 32°C of this area. However, the discrepancies between S-83 and LGR are more than 1.0°C in the winter and summer (Fig. 4a). Therefore, factors other than growth rate might affect the coralline Mg/Ca~SST relationship. In spite of the poor correlation coefficient, the Sr/Ca~SST relationships of LGR are almost the same as that of HGR (Fig. 6b). The discrepancies in the temperatures calculated using the two equations are less than 0.5°C, indicating that the growth rate does not seem to affect the Sr/Ca ratios of

Fig. 3. Linear regression of Mg/Ca, Sr/Ca and U/Ca against measured SSTs. (a) Mg/Ca to SST; (b) Sr/Ca to SST; (c) U/ Ca to SST. All data measured were used in regression.

(c)

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

67

our Porites. However, a different Sr/Ca-SST relationship appears in S-83. The slope and intercept of the equation of S-83 are almost twice that of other sections, even though its growth rate is the same as that of HGR. These discrepancies do not seem to be due to growth-rate differences. The differences in the U/Ca~SST relationships among these three sections are much greater than those of the Mg/Ca and Sr/Ca ratios (Fig. 4c). It is difficult to estimate the extent to which the growth rate affects the U/Ca ratios of our Porites. However, the discrepancies in both the slope and the intercept of the equations between S-83 and HGR, the two intervals with similar growth rates, are much greater than those between LGR and HGR, the two with different growth rates. This implies that the growth rate has little effect on the U/Ca ratios of our Porites. As a whole, we suggest that the growth rate is not an important factor in influencing the Mg/Ca, Sr/Ca and U/Ca ratios in the Porites coral from Sanya. Several unusual conditions in the year 1983 probably caused different responses of Sr/Ca and U/Ca ratios to SSTs in the South Coastal of Hainan Island. 3.4.2. Variation of the seawater composition A variation in trace-element contents in seawater is one of the important factors in controlling the intake of trace elements into coral skeletons (Swart and Hubbard, 1982; Shen and Sanford, 1990). The Sr/Ca ratios in global reef water seem to remain constant (de Villiers et al., 1994; Shen et al., 1996). However, several abnormal Sr/Ca values can also be seen in some coral reefs (Shen et al., 1996). The U contents of global seawater vary by about 3.8% (Chen et al., 1986). The coralline U/Ca ratios are obviously affected by the variation of U contents

Fig. 4. Linear regressions for each section. (a) Mg/Ca: LGR: T (°C )=7.578×103Mg/Ca−8.249, r2=0.97; S-83: T (°C )= 9.617×103Mg/Ca−17.39, r2=0.93; HGR: T (°C )=8.854× 103Mg/Ca−12.94, r2=0.98. (b) Sr/Ca: LGR: T (°C )=133.8– 11.60×103Sr/Ca, r2=0.75; S-83: T (°C )=227.7–21.67× 103Sr/Ca, r2=0.71; HGR: T (°C )=139.6−12.22×103Sr/Ca, r2=0.57. (c) U/Ca: LGR: T (°C )=64.42–30.89×106U/Ca, r2=0.61; S-83: T (°C )=96.80–58.05×106U/Ca, r2=0.95; HGR: T (°C )=69.77–36.30×106U/Ca, r2=0.80.

68

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

in seawater, which is subjected to a variation in salinity by heavy rainfall or run-off (Shen and Dunbar, 1995). Very few records of the variation of seawater Mg/Ca have been reported. Schifano (1982) suggested that the Mg contents of seawater vary with salinity, and so, the Mg/Ca ratios in seawater may change with salinity. Guo et al. (1992) reported several Mg, Ca and Sr contents of seawater from the Nansha Islands, the southern part of the South China Sea, which is the nearest available trace-element data set. According to their data, the maximum relative variability for Mg is 4%, 7% for Ca, and 15% for Sr. Because Hainan Island is located much closer to the continent than the Nansha Islands, the variations of these elements and element ratios in the seawater near the Hainan Island are supposed to be greater. At least, we can conclude that the Sr/Ca ratios in the seawater of coastal Hainan Island are not as constant as those reported in the open marine environment (de Villiers et al., 1994) The contents of trace elements in seawater are assumed to covary with salinity, and so, the temporal change of these elements can be estimated by the change of salinity. In Yinggehai, an area close to Sanya, the seawater is generally diluted 2 months after the precipitation maximum in Hainan Island, with a minimum salinity (in October) about 5% lower than the maximum (in April ) (Fig. 5). The salinity of Yinggehai strongly covaries with monthly precipitation (2 months forward), with a correlation coefficient of 0.92 with records from the Dongfang Meteorological Observatory, and 0.85 with the average of monthly precipitation of the Haikou, Qionghai and Dongfang Observatories (Fig. 5). Because there is no large river influx into this area, the time while peak flux of the run-off from the island reaches the area may be later than the precipitate time. The 2 month delay in salinity change with precipitation implies that not only local rainfall but also the related run-off from the island affects the salinity in South Coast of Hainan Island significantly. Therefore, the change in salinity in this area is not a simple dilution by fresh water but is mixed with elements from the continent, which probably causes a variation in Mg/Ca, Sr/Ca and U/Ca. As indicated from precipitation records, in Hainan Island, about 80–90% of the annual rain-

Fig. 5. Comparison of monthly average salinity in Yinggehai to the monthly precipitation data: (a) compared with the monthly precipitate of Dongfang Station; (b) compared with the average of Haikou, Qionghai and Dongfang. The precipitate records had been pulled back by 2 months. The month of January on the x-axis corresponds to November for precipitation, February to December, March to January and so forth. Salinity data are from Zhong (1997), and precipitation data are from the National Climatic Data Center of NOAA (http://www. ncdc.noaa.gov), Global Historical Climatology Network ( Vose et al., 1992).

fall occurs during the season from May to October. Rainfall in September and October accumulates to half of the total annual precipitation, though the annual amounts of precipitation differ from year to year. We used the average of the monthly precipitation records of Haikou, Qionghai and Dongfang to represent the precipitation of the Hainan Island and found a high precipitation in 1978, 1980 and 1982 (1785~1891 mm/year), and a low precipitation in 1977 (1018 mm/year) and

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

1983 (1154 mm/year). Due to the significant variation in annual precipitation, the salinity of the coastal seawater should have varied significantly from year to year, as did the Mg/Ca, Sr/Ca and U/Ca in the seawater. The above-mentioned change in these element ratios would make the calculated SST biased from the measured SST. Assuming that the variation in element ratios in the seawater is similar to that of salinity, we can estimate the temperature bias of these coral thermometers induced by the seawater compositional variation based on the thermometer

69

equations. If the variation of Mg/Ca, Sr/Ca and U/Ca ratios in seawater reaches the maximum variation of salinity in Yinggehai, 5% (Fig. 5), the Mg/Ca and U/Ca ratios result in a temperature bias of about 1.2~1.5°C. In contrast, the Sr/Ca ratios will give 6~7°C temperature errors. If we presume that the variation of these ratios in seawater reaches a 1s range of salinity change (2%), the temperature errors of Mg/Ca and U/Ca thermometers would be about 0.5°C, but 3°C for the Sr/Ca thermometer. Obviously, the Mg/Ca thermometer is much more accurate than the Sr/Ca thermometer.

Fig. 6. Comparison of U/Ca, Sr/Ca anomaly with precipitation anomaly. (a) U/Ca anomaly; (b) relative precipitation anomaly; (c) Sr/Ca anomaly. The time for precipitation had been pulled back by 2 months.

70

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

In addition, the large discrepancies between the U/Ca thermometer and the measured SST imply that the variation of the U/Ca in seawater should be more complicated. The precipitation records can account for some of the ‘abnormal’ U/Ca ratios. Fig. 6 exhibits the correlation of relative precipitation and U/Ca ratio anomaly. The relative precipitation anomaly of each month (RPA) and the U/Ca ratio anomaly (DU/Ca) are defined as: RPA=100×(monthly precipitation−mean of this month)/(mean of this month) and (DU/Ca)=measured U/Ca−expected U/Ca. The definition of DSr/Ca is similar to that of DU/Ca. We use a 30 year average to represent the mean precipitation of each month, and the expected U/Ca ratios are calculated from SST according to the U/Ca~SST relationship above. Considering the 2 month lap of salinity change in Yinggehai and rainfall in Hainan (Fig. 5), we pull the RPA back by 2 months in Fig. 6. Many of the U/Ca anomalies, such as 1979, 1981 and 1983– 1984, can therefore be correlated with precipitation anomalies. This indicates that high precipitation in the Hainan Island causes a higher U/Ca in the seawater of South Coast of Hainan, and vice versa (Fig. 6). This can be supported by a previous study that documented that the U/Ca ratios in river water are several times higher than those in seawater (Cochran, 1992). Therefore, the discrepancies between the measured SST and the calculated SST from U/Ca ratios can be attributed to the abnormal precipitation. As revealed by Fig. 6, some of the ‘abnormal’ U/Ca ratios, such as the higher U/Ca in 1977– 1978, are difficult to interpret by abnormal rainfalls. The coralline Sr/Ca ratios behave in the same way as the U/Ca ratios ( Fig. 6), but have a poorer correlation. This implies that the control factors for the U/Ca and Sr/Ca ratios in the Porites are quite complicated. It is worth noting again that the differences between measured Mg/Ca ratios and calculated ratios from SST are much smaller than those of U/Ca and Sr/Ca.

Porites and SST records. Our relative analytical uncertainties for Mg/Ca ratios, ranging from 0.5 to 1%, together with the regression errors of 0.8 and 1.5%, respectively, for the slope and the intercept will give 0.4–0.5°C errors for the calculated SST. The small errors allow us to build the Mg/Ca thermometer for SST reconstruction. On the contrary, due to the possible variation of their composition in the seawater from the south coast of Hainan Island, the Sr/Ca and U/Ca ratios of our Porites show a fairly poor linear relationship, even though the analytical precision is good enough for a thermometer construction. This indicates that coralline Sr/Ca and U/Ca ratios are not valid in reconstructing SST records, at least in this area. Our Mg/Ca thermometer is consistent with another Mg/Ca thermometer from a Porites lutea coral near the Ryukyu Islands, west Pacific, reported by Mitsuguchi et al. (1996) (Fig. 7). Since SST in both Ryukyu and the Hainan Island mostly ranges between 22 and 32°C, the difference in the calculated SST derived from these two different Mg/Ca thermometers for the same data set will be less than 0.5°C in the summer. In the winter, the calculated SSTs using our Mg/Ca thermometer are 1.1°C lower than those from Mitsuguchi et al. (1996) in the maximum (Fig. 7). This indicates that the regional difference for the coralline Mg/Ca

3.5. Coralline thermometers for the coast of Hainan Island As discussed above, there is a good linear relationship between the Mg/Ca ratios of our

Fig. 7. Comparison of the coralline Mg/Ca thermometer. The solid line represents our result, and the dashed line is that of Mitsuguchi.

71

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

thermometer is not significant, and hence the coralline Mg/Ca thermometer is a reliable and versatile tool for paleo-SST reconstruction.

Sciences Foundation Grant to G.J.W. (49803003), President Grant of Chinese Academy of Sciences ( KJ95-321) and Grant for Overseas Chinese Scholar of CAS (941101, 941111) to X.H.L.

4. Summary Appendix A A new analytical method, an internal standard together with ICP–MS–ID, was employed to measure the Sr/Ca and Mg/Ca ratios, and the ICP– MS–ID method was utilized for U/Ca measurement. The analytical precision and reproducibility are about 0.5~1.0, 0.2~0.5 and 0.5~1.0% for Mg/Ca, Sr/Ca and U/Ca, respectively. A Porites specimen from Sanya Bay, the southern offshore part of the Hainan Island, was analyzed with an interval of 10 years. Based on our data set, the Mg/Ca thermometer is not sensitive to environmental factors other than temperature, and seems versatile in different areas. Therefore, a high-precision (better than 0.5°C ) Mg/Ca thermometer is constructed for the SST study. Due to the high run-off and other possible inputs from the continent, Sr/Ca and U/Ca thermometers are not valid for the SST study in the study area.

Acknowledgements We thank Marine Department of the Hainan Province, Sanya Coral-Reef Nature Protection District, and Sanya Oceanic Station of the South China Sea Institute of Oceanology, Chinese Academy of Sciences for help in the sample collection. We are grateful to Dr M.F. Zhou of the University of Hong Kong, Ms Y. Liu, Mr H.C. Liu and Mr G.L. Wang of the Guangzhou Institute of Geochemistry, CAS, for help in the ICP-MS analysis, and Dr H.W. Zhou for splitting the coral slice. We are also grateful for valuable advice from Professor M.T. McCulloch of the Australia National University in Canberra and Dr Y.G. Chen of National Taiwan University in Taipei. The constructive comments from P. Swart, G. Mortimer and another anonymous reviewer are valuable for us to improve the manuscript. The research was supported by Hong Kong RGC Grant (HKU510/96P) to M.S., Chinese National

Table A1 Mg/Ca, Sr/Ca and U/Ca ratios of our poritesa Sample number

Sr/Ca (mmol/mol )

Mg/Ca (mmol/mol )

Sample number

U/Ca (mmol/mol )

SYa001 SYa002 SYa003 SYa004 SYa005 SYa006 SYa007 SYa008 SYa009 SYa010 SYa011 SYa012 SYa013 SYa014 SYa015 SYa016 SYa017 SYa018 SYa019 SYa020 SYa021 SYa022 SYa023 SYa024 SYa025 SYa026 SYa027 SYa028 SYa029 SYa030 SYa031 SYa032 SYa033 SYa034 SYa035 SYa036 SYa037 SYa038 SYa039 SYa040 SYa041 SYa042 SYa043 SYa044

9.270 9.198 9.266 9.318 9.243 9.244 9.155 9.034 9.034 9.047 9.177 9.084 9.052 9.151 9.102 9.269 9.293 9.413 9.533 9.389 9.175 9.248 9.017 9.081 9.136 9.007 9.026 9.140 9.109 9.188 9.359 9.322 9.147 9.050 9.043 8.967 9.068 9.271 9.327 9.430 9.630 9.654 9.428 9.166

4.439 4.370 4.214 3.896 3.944 4.208 4.292 4.445 4.792 5.100 4.829 5.036 4.880 4.949 4.940 4.653 4.396 4.320 4.006 4.225 4.383 4.645 4.663 4.938 5.028 5.013 4.829 4.991 5.090 4.528 4.283 4.338 4.527 4.890 4.931 4.977 5.138 4.911 4.601 4.539 4.152 3.979 4.277 4.550

SYb001 SYb002 SYb003 SYb004 SYb005 SYb006 SYb007 SYb008 SYb009 SYb010 SYb011 SYb012 SYb013 SYb014 SYb015 SYb016 SYb017 SYb018 SYb019 SYb020 SYb021 SYb022 SYb023 SYb024 SYb025 SYb026 SYb027 SYb028 SYb029 SYb030 SYb031 SYb032 SYb033 SYb034 SYb035 SYb036 SYb037 SYb038 SYb039 SYb040 SYb041 SYb042 SYb043 SYb044

1.119 1.158 1.179 1.185 1.214 1.222 1.246 1.295 1.284 1.287 1.319 1.321 1.213 1.190 1.112 1.130 1.134 1.140 1.142 1.134 1.183 1.195 1.230 1.269 1.225 1.191 1.142 1.169 1.233 1.181 1.169 1.166 1.220 1.213 1.225 1.268 1.322 1.336 1.308 1.225 1.226 1.229 1.093 1.104

72

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

Table A1 (continued ).

Table A1 (continued).

Sample number

Sr/Ca (mmol/mol )

Mg/Ca (mmol/mol )

Sample number

U/Ca (mmol/mol )

Sample number

Sr/Ca (mmol/mol )

Mg/Ca (mmol/mol )

Sample number

U/Ca (mmol/mol )

SYa045 SYa046 SYa047 SYa048 SYa049 SYa050 SYa051 SYa052 SYa053 SYa054 SYa055 SYa056 SYa057 SYa058 SYa059 SYa060 SYa061 SYa062 SYa063 SYa064 SYa065 SYa066 SYa067 SYa068 SYa069 SYa070 SYa071 SYa072 SYa073 SYa074 SYa075 SYa076 SYa077 SYa078 SYa079 SYa080 SYa081 SYa082 SYa083 SYa084 SYa085 SYa086 SYa087 SYa088 SYa089 SYa090 SYa091 SYa092 SYa093 SYa094 SYa095 SYa096 SYa097 SYa098 SYa099

9.089 8.946 9.067 9.087 9.071 9.072 9.118 9.118 9.162 9.250 9.197 9.308 9.572 9.370 9.310 9.322 9.199 9.172 9.053 9.120 9.106 9.038 8.996 8.994 9.065 9.120 9.438 9.504 9.681 9.643 9.425 9.175 9.259 9.206 9.199 9.097 9.107 9.000 9.066 9.121 9.109 9.112 9.539 9.687 9.762 9.571 9.465 9.400 9.469 9.425 9.337 9.246 9.151 9.196 9.140

4.763 4.854 4.968 4.871 5.015 4.992 4.961 4.657 4.538 4.530 4.334 4.034 4.192 4.463 4.538 4.539 4.731 4.919 4.912 4.967 4.965 4.981 4.945 4.869 4.718 4.638 4.601 4.503 4.116 4.221 4.193 4.636 4.590 4.793 4.985 4.922 4.992 5.029 5.096 4.920 5.052 4.822 4.384 4.261 4.020 3.778 4.202 4.401 4.284 4.459 4.528 4.834 4.874 4.963 4.928

SYb045 SYb046 SYb047 SYb048 SYb049 SYb050 SYb051 SYb052 SYb053 SYb054 SYb055 SYb056 SYb057 SYb058 SYb059 SYb060 SYb061 SYb062 SYb063 SYb064 SYb065 SYb066 SYb067 SYb068 SYb069 SYb070 SYb071 SYb072 SYb073 SYb074 SYb075 SYb076 SYb077 SYb078 SYb079 SYb080 SYb081 SYb082 SYb083 SYb084 SYb085 SYb086 SYb087 SYb088 SYb089 SYb090 SYb091 SYb092 SYb093 SYb094 SYb095 SYb096 SYb097 SYb098 SYb099

1.144 1.197 1.268 1.288 1.338 1.170 1.096 1.094 1.143 1.121 1.091 1.062 1.115 1.171 1.115 1.210 1.244 1.242 1.204 1.214 1.187 1.159 1.146 1.151 1.171 1.205 1.230 1.207 1.244 1.286 1.246 1.131 1.097 1.105 1.110 1.097 1.119 1.206 1.289 1.311 1.264 1.225 1.177 1.146 1.170 1.152 1.166 1.138 1.317 1.234 1.199 1.184 1.134 1.156 1.111

SYa100 SYa101 SYa102 SYa103 SYa104 SYa105 SYa106 SYa107 SYa108 SYa109 SYa110 SYa111 SYa112 SYa113 SYa114 SYa115 SYa116 SYa117 SYa118 SYa119 SYa120 SYa121 SYa122 SYa123 SYa124 SYa125 SYa126 SYa127 SYa128 SYa129 SYa130 SYa131 SYa132 SYa133 SYa134 SYa135 SYa136 SYa137 SYa138 SYa139 SYa140 SYa141 SYa142 SYa143 SYa144 SYa145 SYa146 SYa147

9.183 9.118 9.189 9.157 9.334 9.160 9.321 9.356 9.350 9.089 9.093 9.153 9.103 9.036 9.020 9.140 9.117 9.093 9.259 9.412 9.259 9.302 9.344 9.387 9.480 9.342 9.216 9.200 9.396 9.040 9.003 9.139 9.055 9.039 9.085 9.320 9.076 9.337 9.320 9.391 9.439 9.580 9.466 9.427 9.440 9.555 9.405 9.304

4.815 4.968 4.938 4.820 4.644 4.437 4.128 4.048 4.221 4.537 4.609 4.792 4.910 5.008 5.048 4.878 4.975 4.768 4.965 4.625 4.763 4.453 4.369 4.227 4.250 4.341 4.327 4.342 4.568 5.072 4.946 4.885 5.076 5.008 5.003 4.771 4.866 4.773 4.697 4.695 4.332 4.220 4.099 4.145 4.103 4.371 4.475 4.638

SYb100 SYb101 SYb102 SYb103 SYb104 SYb105 SYb106 SYb107 SYb108 SYb109 SYb110 SYb111 SYb112 SYb113 SYb114 SYb115 SYb116 SYb117 SYb118 SYb119 SYb120 SYb121 SYb122 SYb123 SYb124 SYb125 SYb126 SYb127 SYb128 SYb129

1.088 1.089 1.094 1.116 1.125 1.200 1.256 1.227 1.258 1.284 1.282 1.286 1.216 1.074 1.090 1.059 1.078 1.042 1.138 1.183 1.208 1.204 1.236 1.282 1.221 1.320 1.248 1.201 1.150 1.136

a Mg, Ca and Sr were measured by the internal standard isotope dilution ICP-MS method for Mg/Ca and Sr/Ca ratios. For U/Ca ratios, U was measured by the ordinary isotope dilution ICP-MS method, and Ca by the external standard calibration method. Relative standard deviations (1s) are 0.5~1.0%, 0.2~0.5% and 0.5~1.0% for Mg/Ca, Sr/Ca and U/Ca, respectively.

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

References Alibert, C., McCulloch, M.T., 1997. Strontium/calcium ratios in modern Porites corals from the Great Barriers Reef as a proxy for sea surface temperature: Calibration of the thermometer and monitoring ENSO. Paleoceanography 12, 345–363. Beck, J.W., Edwards, R.L., Ito, E., Taylor, F.W., Recy, J., Rougerie, F., Joannot, P., Henin, C., 1992. Sea-surface temperature from coral skeletal strontium/calcium ratios. Science 257, 644–647. Chen, J.H., Edwards, R.L., Wasserburg, G.J., 1986. 238U, 234U and 230Th in seawater. Earth Plant. Sci. Lett. 80, 241–251. Cochran, J.K., 1992. The oceanic chemistry of the uranium and thorium series nuclides. In: Ivanovich, M., Harmon, R.S. ( Eds.), Uranium-Series Disequilibrium: Applications to Earth, Marine and Environmental Sciences, second ed., 334–395. Clarendon Press, Oxford. de Villiers, S., Shen, G.T., Nelson, B.K., 1994. The Sr/Ca–temperature relationship in coralline aragonite: Influence of variability in (Sr/Ca) and skeletal growth parameters. seawater Geochim. Cosmochim. Acta 58, 192–208. de Villiers, S., Nelson, B.K., Chivas, A.R., 1995. Biological controls on coral Sr/Ca and d18O reconstructions of sea surface temperatures. Science 269, 1247–1249. Goldberg, E.D., Arrhenius, G.O.S., 1973. Chemistry of Pacific pelagic sediments. Geochim. Cosmochim. Acta 13, 153–212. Guo, L.F., Zhu, Y.Z., Zhao, H.T., 1992. Contents of some elements in the seawater of the reef zone and adjacent sea area in Nansha Islands, the South China Sea. In: Zhao, H.T., Song, C.J., Huang, C.L. ( Eds.), Collection of Research Papers on Geophysics and Coral Reef of the Nansha Islands and their Adjacent Regions, Section 2. Science Press, Beijing, pp. 95–107 (in Chinese). Lea, D.W., Shen, G.T., Boyle, E.A., 1989. Coralline barium records temporal variability in equatorial Pacific upwelling. Nature 340, 373–376. Lea, D.W., Boyle, E.A., 1993. Determination of carbonatebound barium in Foraminifera and corals by isotope-dilution plasma mass-spectrometry. Chem. Geol. 103, 73–84. Lea, D.W., Martin, P.A., 1996. A rapid mass-spectrometric method for the simultaneous analysis of barium, cadmium and strontium in Foraminifera shells. Geochim. Cosmochim. Acta 60, 3143–3149. Lecornec, F., Correge, T., 1997. Determination of uranium to calcium and strontium to calcium ratios in corals by inductively-coupled plasma-mass spectrometry. J. Anal. Atom. Spectrosc. 12, 969–973. Linn, L.J., Delaney, M.L., Druffel, E.R.M., 1990. Trace-metals in contemporary and 17th-century Galapagos coral— records of seasonal and annual variations. Geochim. Cosmochim. Acta 54, 387–394. Ludwig, K.R., 1992. In: ISOPLOT—A Plotting and Regression Program for Radiogenic–Isotopic Data, Version 2.57. U.S.G.S. Open-File Report 91-445, 40. McCulloch, M.T., Gagan, M.K, Mortimer, G.E., Chivas, A.R.,

73

Isdale, P.J., 1994. A high resolution Sr/Ca and d18O coral record from the Great Barrier Reef, Australia and the 1982–1983 El Nin˜o. Geochim. Cosmochim. Acta 58, 2747–2754. McCulloch, M.T., Mortimer, G., Esat, T., Li, X.H., Pillans, B., Chappell, J., 1996. High resolution windows into early Holocene climate: Sr/Ca coral records from the Huon Peninsula. Earth Planet. Sci. Lett. 138, 169–178. Min, G.R., Edwards, R.L., Taylor, F.W., Recy, J., Gallup, C.D., Beck, J.W., 1995. Annual cycles of U/Ca in coral skeletons and U/Ca thermometry. Geochim. Cosmochim. Acta 5, 2025–2042. Mitsuguchi, T., Matsumoto, E., Abe, O., Uchida, T., Isdale, P.J., 1996. Mg/Ca thermometry in coral skeletons. Science 274, 961–963. Nie, B.F., Chen, T.G., Liang, M.T., Wang, Y.Q., Zhong, J.L., Zhu, Y.Z., 1996. Relationship between sea surface temperature and coral growth rate in the north of South China Sea during the last 100 years. Sci. China, Series D 26, 59–66. in Chinese. Schifano, G., 1982. Temperature–magnesium relations in the shell carbonate of some modern marine gastropods. Chem. Geol. 35, 321–332. Shen, C.C., Lee, T., Chen, C.Y., Wang, C.H., Dai, C.F., 1996. The calibration of D[Sr/Ca] versus sea temperature relationship for Porites corals. Geochim. Cosmochim. Acta 60, 3849–3858. Shen, G.T., Sanford, C.L., 1990. Trace element indicators of climate variability in reef-building corals. In: Glynn, P.W. ( Ed.), Global Ecological Consequences of the 1982–83 El Nin˜o-Southern Oscillation. Elsevier, Amsterdam, pp. 255–283. Shen, G.T., Cole, J.E., Lea, S.W., Linn, L.J., McCounnaughey, T.A., Fairbanks, R.G., 1992. Surface ocean variability at Galapagos from 1936~1982: calibration of geochemical tracers in corals. Paleoceanography 5, 563–588. Shen, G.T., Dunbar, R.B., 1995. Environmental controls on uranium in reef corals. Geochim. Cosmochim. Acta 59, 2009–2024. Sinclair, D.J., Kinsley, L.P.J., McCulloch, M.T., 1998. High resolution analysis of trace elements in corals by laser ablation ICP-MS. Geochim. Cosmochim. Acta 62, 1889–1901. Smith, S.V., Buddemeier, R.W., Redalje, R.C., 1979. Strontium–calcium thermometry in coral skeletons. Science 204, 404–407. Swart, P.K., 1981. The strontium, magnesium and sodium composition of recent scleractinian coral skeletons as standards for palaeoenvironmantal analysis. Palaeogr., Palaeoclimatol., Palaeoecol. 34, 115–136. Swart, P.K., Hubbard, J.A.E.B., 1982. Uranium in scleractinian coral skeletons. Coral Reefs 1, 13–19. Vose, R.S., Schmoyer, R.L., Steurer, P.M., Peterson, T.C., Heim, R., Karl, T.R., Eischeid, J., 1992. The Global Historical Climatology Network: Long-term Monthly Temperature, Precipitation, Sea Level Pressure, and Station Pressure Data. ORNL/CDIAC-53, NDP-041. Carbon

74

G. Wei et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 162 (2000) 59–74

Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, TN. Wang, P.X., 1999. Response of Western Pacific marginal seas to glacial cycles: paleoceanographic and sedimentological feastures. Mar. Geol. 156, 5–39. Weber, J.N., 1973. Incorporation of strontium into reef coral skeletal carbonate. Geochim. Cosmochim. Acta 37, 2173–2190.

Wei, G.J., Li, X.H., Nie, B.F., Sun, M., Liu, H.C., 1999. High resolution Porites Mg/Ca thermometer for the north of the South China Sea. Chinese Sci. Bull. 44, 273–276. Zhong, J.L., 1997. Hydrometric and climatic characters of coral reef zone of the south coast of Hainan Island. In: The Relationship Between Reef Coral and Environmental Changes of Nansha Islands. Science Press, Beijing, pp. 21–22.