in greater detail, including specific techniques for sample cleaning and analysis. Studies of growth rate and species influences show very limited evidence of biological mediation of metal incorporation (de Villiers et al., 1993; G.T. Shen, unpublished results). However, SST may play a role in the incorporation of Ba (Lea et al., 1989), and the dependence of Sr incorporation on ambient SST (Smith et al., 1979) makes possible precise reconstruction of SST from coral Sr records (Beck et al., 1992).
The modern distribution of cadmium follows that of marine nutrients. Low levels in surface waters reflect biological removal, while higher levels at depth result from the regeneration of organic matter (Boyle et al., 1976; Martin et al., 1976). In coral records, the skeletal Cd content usually depends on the balance between Cd-rich upwelled deep water and Cd-poor oligotrophic surface waters. In Galapagos corals, Cd/Ca ratios directly reflect upwelling variations associated with both seasonal cycles (Linn et al., 1990; Shen and Sanford, 1990) and interannual ENSO variability (Shen et al., 1987, 1992a).
This trace metal exhibits nutrient-like behavior akin to cadmium's (Chan et al., 1977). Higher concentrations in both seawater and corals render coral Ba records less susceptible to contamination. Seasonal-resolution records of Ba/Ca from Galapagos corals reflect regional upwelling variability (Lea et al., 1989). Relative to Cd, Ba may exhibit greater sensitivity to periods of weak upwelling, possibly because the biological uptake of Cd occurs at rates comparable to the slow rate of supply during these times. Ba is also enriched in continental runoff waters, and Ba/Ca records from corals near continental margins may reflect this input (Shen and Sanford, 1990). However, a slight temperature effect may occur upon incorporation of Ba into the coral skeleton, which would complicate paleoclimatic reconstruction from coral Ba/Ca records (Lea et al., 1989).
Unlike Cd and Ba, Mn reaches high concentrations in surface waters. A mid-depth maximum coincides with the local O2-minimum zone, where particulate Mn oxides are reduced and solubilized; concentrations then diminish with increasing depth. Aeolian and fluvial input provide important Mn sources, as do reducing environments where particulate Mn is degraded, such as the O2-minimum zone, shelf sediments, and lagoons. The interpretation of coral Mn/Ca records is thus usually site-specific, reflecting transport and mixing of water masses with varying Mn levels. In the Galapagos, for example, seawater Mn levels reflect a combination of aerosol deposition on the surface waters and long-range advection of Mn-enriched surface water from the continental shelf (Shen et al., 1991). In localized reef settings, dissolved Mn in the water column can be augmented by local sediment fluxes, producing very high skeletal concentrations of Mn in corals from the Gulf of Panama and some Caribbean islands (Shen et al., 1991). Diagenetic Mn fluxes from lagoonal sediments offer useful indicators of climate variability at certain Pacific atoll sites far removed from continental sources of Mn (Shen et al., 1992b).
Work by Smith et al. (1979) demonstrated that the Sr/ Ca ratio of coral skeletons may provide a monitor of past temperature changes. However, the measurement technique used in that study was not sufficiently precise to offer detailed SST reconstructions at a useful resolution for the tropics. The recent application of thermal ionization mass spectrometry (TIMS) to these measurements has greatly improved the precision of Sr/Ca determinations (Beck et al., 1992); recent results indicate that monthly SST can be reconstructed with an apparent accuracy of better than 0.5°C (Beck et al., 1992; de Villiers et al., 1993). SST reconstructions based on skeletal Sr determinations are less susceptible to artifacts associated with hydrologic changes than d18O-based reconstructions. The paired application of d18O and Sr/Ca data may thus allow the separation of SST changes from seawater d18O variability.
Many studies have documented the paleoclimatic utility of isotopic and trace-metal measurements by developing short (2-20 yr) data sets that correlate with nearby instrumental records (e.g., Fairbanks and Dodge, 1979; Pätzold, 1984; Shen et al., 1987, 1992b; Carriquiry et al., 1988; Lea et al., 1989; Cole and Fairbanks, 1990; Winter et al., 1991; Cole et al., 1992; Beck et al., 1992). With the growing interest in long records of natural climate variability, many groups are applying these techniques to the development of proxy records that extend beyond the period of local instrumental coverage. Efforts to develop long, high-resolution reconstructions of climate are in progress using coral records from sites throughout the low-latitude oceans. Several studies have focused explicitly on reconstructing interannual-to-decadal variability in the ENSO system; we detail current results from the equatorial Pacific in the following section. Other sites from which coral records extending beyond the present century have been developed include Florida Bay (Kramer et al., 1991), Cebu Island, Philippines (Pätzold and Wefer, 1986), Espiritu Santo Island (Quinn et al., 1992), the southern Great Barrier Reef (Druffel and