concentrations (i.e., upwelling) of these nutrient-like tracers, and d¹³C probably responds to insolation variations, although this relationship is not constant for all coral d¹³C records from the Galapagos (McConnaughey, 1989).

Figure 4 compares the coral record of d¹⁸O anomaly from Punta Pitt with SST anomalies from Puerto Chicama and with an index of eastern-central Pacific SST anomalies (Wright, 1989). The d¹⁸O record captures most measured SST anomalies in these curves, both positive and negative. In certain cases where the d¹⁸O and Puerto Chicama records disagree (e.g., in 1980), the Wright SST index appears to corroborate the d¹⁸O record, suggesting that variability in the coastal SST record does not always represent conditions in nearby open ocean regions. Deser and Wallace (1987) have also noted that SSTs along the Peru coast occasionally become decoupled from large-scale variations in Pacific climate.

Cross-spectral analysis of the Punta Pitt coral records with records of Puerto Chicama SST indicate that correlations at the frequency bands associated with ENSO (about 2 and 3.8 years, in these records) reach values as high as 80 to 90 percent. Shen et al. (1992a) note that the observed degree of correlation, in both time and frequency domains, is truly remarkable, considering that these single proxy records derive from a site several hundred kilometers away from the coastal SST monitoring station at Puerto Chicama.

In the western Pacific, ENSO variability is characterized by dramatic shifts in rainfall patterns associated with the migration of the Indonesian Low from northern Australia/ Indonesia to the region of the equator and the date line.

Biweekly monitoring of surface ocean d¹⁸O at Tarawa Atoll demonstrates that at this site, rainfall causes a significant decrease in the d¹⁸O of the surface water in which corals grow (Cole, 1992). Coral d¹⁸O records from this site closely track these rainfall changes (Cole and Fairbanks, 1990; Cole et al., 1993). Our coral d¹⁸O record extends back to 1894 at monthly resolution, double the length of the instrumental record of Tarawa rainfall that began in 1946. Between 1976 and 1989, the Tarawa coral record suggests a period of increased rainfall that is unprecedented within this record. This apparent baseline shift is consistent with instrumental records from the Pacific (Trenberth, 1990; Kerr, 1992).

This d¹⁸O record correlates with large-scale indices of ENSO (Figure 5) at levels comparable to monthly instrumental records from individual Pacific stations, such as Tarawa rainfall and Darwin SLP. Monthly, seasonal, and annual d¹⁸O data linearly explain fractions of ENSO variance that are among the highest for any proxy record of ENSO (Cole, 1992; Cole et al., 1993). Despite the strong correlation between the coral and instrumental records of ENSO, the Tarawa d¹⁸O record does not capture the very strong ENSO warm anomaly of 1982-1983. Outgoing long-wave radiation measurements indicate that during that year the Indonesian Low moved much farther east than usual (Rasmusson and Wallace, 1983). Tarawa rainfall was not unusually intense at this time, and the coral d¹⁸O data reflect these conditions accurately. This anomaly that bypassed Tarawa is the only major ENSO extreme of the past century that the coral d¹⁸O record does not monitor, attesting both to the fidelity of the coral recorder and to the unusual nature of the 1982-1983 ENSO extreme. To reconstruct the full range of ENSO variability will require developing records from sites throughout the Pacific.

The coral record generally indicates warm and wet ENSO extremes during periods identified by Quinn et al. (1987) as "El Niño events" along the coast of South America. However, Figure 5 suggests exceptions to this correlation. In 1907, 1917, 1932, and 1943, Quinn et al. identify moderate or strong El Niño events, but neither the coral nor the instrumental indices of tropical Pacific climate indicate ENSO warm anomalies. In other years (e.g., 1946, 1963) the opposite pattern occurs; coral and instrumental indices demonstrate ENSO warm extremes that do not appear as moderate or strong in Quinn's most recent summaries (Quinn et al., 1987; Quinn and Neal, 1992), although these years were noted as weak anomalies in an earlier compilation (Quinn et al., 1978). Rasmusson et al. (1995, in this volume) have also noted, on the basis of instrumental climate records, that recurrence statistics derived from Quinn et al.'s "strong/very strong" events do not reflect basin-scale ENSO variability. Quinn et al. (1978, 1987, 1992) explicitly state that their index may not be entirely consistent with climatic records from across the tropical Pacific, such as the Southern Oscillation Index or central Pacific rainfall

FRONT MATTER (R1-R14)

EXECUTIVE SUMMARY (1-4)

INTRODUCTION (5-8)

THE ATMOSPHERE (9-232)

THE OCEAN (233-412)

COUPLED SYSTEMS (413-488)

PROXY INDICATORS OF CLIMATE (489-598)

CONCLUSIONS (599-604)

RECOMMENDATIONS FOR CLIMATE VARIABILITY RESEARCH PRIORITIES (605-608)

LIST OF PARTICIPANTS (609-614)

INDEX (615-630)