The approximately 10,000-year-long Last Glacial Maximum, before the termination of the last ice age, was the coldest period in Earth's recent climate history. Relative to the Holocene epoch, atmospheric carbon dioxide was about 100 parts per million lower and tropical sea surface temperatures were about 3 to 5 degrees Celsius lower. The Last Glacial Maximum began when global mean sea level (GMSL) abruptly dropped by about 40 metres around 31,000 years ago and was followed by about 10,000 years of rapid deglaciation into the Holocene. The masses of the melting polar ice sheets and the change in ocean volume, and hence in GMSL, are primary constraints for climate models constructed to describe the transition between the Last Glacial Maximum and the Holocene, and future changes; but the rate, timing and magnitude of this transition remain uncertain. Here we show that sea level at the shelf edge of the Great Barrier Reef dropped by around 20 metres between 21,900 and 20,500 years ago, to -118 metres relative to the modern level. Our findings are based on recovered and radiometrically dated fossil corals and coralline algae assemblages, and represent relative sea level at the Great Barrier Reef, rather than GMSL. Subsequently, relative sea level rose at a rate of about 3.5 millimetres per year for around 4,000 years. The rise is consistent with the warming previously observed at 19,000 years ago, but we now show that it occurred just after the 20-metre drop in relative sea level and the related increase in global ice volumes. The detailed structure of our record is robust because the Great Barrier Reef is remote from former ice sheets and tectonic activity. Relative sea level can be influenced by Earth's response to regional changes in ice and water loadings and may differ greatly from GMSL. Consequently, we used glacio-isostatic models to derive GMSL, and find that the Last Glacial Maximum culminated 20,500 years ago in a GMSL low of about -125 to -130 metres.

[1] Paleotemperature estimates based on coral Sr/Ca have not been widely accepted because the reconstructed glacial-Holocene shift in tropical sea-surface temperature ($4-6°C) is larger than that indicated by foraminiferal Mg/Ca ($2-4°C). We show that corals over-estimate changes in sea-surface temperature (SST) because their records are attenuated during skeletogenesis within the living tissue layer. To quantify this process, we microprofiled skeletal mass accumulation within the tissue layer of Porites from Australasian coral reefs and laboratory culturing experiments. The results show that the sensitivity of the Sr/Ca and d 18 O thermometers in Porites will be suppressed, variable, and dependent on the relationship between skeletal growth rate and mass accumulation within the tissue layer. Our findings help explain why d 18 O-SST sensitivities for Porites range from À0.08‰/°C to À0.22‰/°C and are always less than the value of À0.23‰/°C established for biogenic aragonite. Based on this observation, we recalibrated the coral Sr/Ca thermometer to determine a revised sensitivity of À0.084 mmol/mol/°C. After rescaling, most of the published Sr/Ca-SST estimates for the Indo-Pacific region for the last $14,000 years (À7°C to +2°C relative to modern) fall within the 95% confidence envelope of the foraminiferal Mg/Ca-SST records. We conclude that two types of calibration scales are required for coral paleothermometry; an attenuated Porites-specific thermometer sensitivity for studies of seasonal to interannual change in SST and, importantly, the rescaled À0.084 mmol/mol/°C Sr/Ca sensitivity for studies of 20th-century trends and millennial-scale changes in mean SST. The calibration-scaling concept will apply to the development of transfer functions for all geochemical tracers in corals.

A new geochemical reference material, coral Porites sp. JCp‐1 has been prepared by the Geological Survey of Japan (GSJ). Provisional values for twenty one major, minor and trace elements are presented. The homogeneity tests showed that all elements studied are considered to be homogeneously distributed.