Massive corals that live near the sea surface produce annual density bands of aragonite (calcium carbonate) that can be sampled and used to reconstruct monthly climate records by examining their geochemical composition. In particular, the ratio of 18O to 16O (commonly referred to as δ18O)1 in coralline aragonite decreases with increasing seawater temperature and with decreasing δ18O in water at the time of formation. The δ18O of water is often strongly correlated with salinity. Hence, δ18O values from a coral that grew in an open-ocean location can be used to reconstruct a combined signal of sea surface temperature (SST) and salinity for that oceanic region.
Sites for coral sampling are selected on the basis of proximity to open ocean and generally well-flushed locations. The largest corals are usually found in leeward locations that are minimally influenced by storms and silt. The morphology of the coral head should show a rounded shape with minimal erosion at the base caused by boring organisms or physical damage. In most cases, more than one coral is sampled from each site to obtain the longest record. Ideally, coral-based reconstructions would be based on multiple cores from the same site, but in practice few sites have been studied in this detail (for an exception, see Hendy et al. 2002). More commonly, coral reconstructions are calibrated closely with instrumental data, and high correlation coefficients lend confidence to the reconstruction of past conditions (for reviews, see Gagan et al. 2000, Cole 2003). The primary limitation of using coral δ18O for SST reconstruction is the added variable of water δ18O, which is important in areas of high rainfall, evaporation, or river input, such as the western Pacific warm pool. Recently, it has been discovered that this limitation can be overcome by simultaneously measuring the elemental ratio of strontium to calcium (denoted Sr/Ca) in coral bands, because the Sr/Ca ratio changes mainly as a function of SST. Sr/Ca ratios have been shown to track SST quantitatively with a high degree of precision (Beck et al. 1992), yielding an uncertainty in SST reconstructions of less than 0.3°C. There are several long-term reconstructions of SSTs available using coral Sr/Ca ratios including the Great Barrier Reef (Hendy et al. 2002), Rarotonga and Fiji (Linsley et al. 2004), Madagascar (Zinke et al. 2004), and Hawaii (Druffel et al. 2001). A few studies have also used uranium/calcium (U/Ca) ratios to reconstruct SSTs (e.g., Hendy et al. 2002).
Continuous coral δ18O records for the last 100–400 years are available for regions in the tropical and subtropical Pacific, Indian, and Atlantic oceans (see summary papers: Cole 2003, Lough 2004, Gagan et al. 2000). Corals from most of these sites display an overall decrease in δ18O values toward the 20th century, which indicates surface seawater that is warmer, fresher (lower salinity), or both. Most of these records also show abrupt shifts toward warmer/fresher waters during the mid-