Recent sea surface temperature and salinity estimates in the eastern equatorial Pacific (EEP) and western Equatorial Pacific (Koutavas et al., 2002; Stott et al., 2002), and coupled ocean-atmosphere GCM results of Peltier and Solheim (2004) conclude that El Niños were more intense under glacial relative to Holocene conditions. This conclusion is diametrically opposed to that reached in previous model (Clement et al., 1999) and Papua New Guinea-coral 18O/16O-based paleoclimate (Tudhope et al., 2001) studies, which indicate intensified and/or more frequent El Niños under interglacial conditions. Data and modeling studies focusing on Holocene variability of ENSO provide a similarly contradictory set of conclusions. Colinvaux (1972) produced the first continuous record of Holocene ENSO variability by inferring water level variations in El Junco Lake on the Galapagos island of San Cristobal from the sedimentary abundance of Azolla microphylla remains, a water fern found only in shallow water. High concentrations of Azolla 12.1-8.8 and 7.1-3.2 ka indicated a shallow lake starved of El Niño rains, while a dearth of Azolla 8.8-7.1 and 3.2-0 ka signaled a high lake level, consistent with frequent and/or intense El Niño events (Fig. 1A). These results, in particular the wet period ~9-7 ka, the dry period ~7-3 ka, and the lack of any perceptible decrease in El Niño-related precipitation in the last 1.5 ky are difficult to reconcile with the Holocene ENSO reconstruction by Moy et al. (2002) from flood deposits in Laguna Pallacocha (Fig. 1B), a high Ecuadorian Andes lake, that indicate infrequent (severe) El Niños prior to ~7 ka, a trend toward increasing frequency of events from ~7-1.5 ka, and a sharp decrease to early Holocene frequencies in the last 1.5 ky. A third independent evaluation of ENSO variability in the Holocene by Koutavas et al. (2004), from SST reconstructions in the Warm Pool of the western equatorial Pacific (WEP) and the Cold Tongue of the eastern equatorial Pacific (EEP), provide a distinctly different picture than either of the other two ENSO reconstructions (Fig. 1C). A large zonal SST gradient ~10-4 ka is consistent with less frequent and/or weaker El Niño events through the early and mid-Holocene, while from ~12-10 and ~4-1.5 ka a diminished zonal gradient is consistent with more frequent or stronger El Niño events (Koutavas & Sachs, in prep; Koutavas et al., 2004).

On the modeling front, Clement et al. (2000) used a coupled ocean-atmosphere GCM (the Cane-Zebiak model), constructed specifically for ENSO-related studies, forced with Holocene changes in insolation to conclude that the frequency and intensity of El NiÒo events increased throughout the Holocene, until about 1.5-2 ka, after which both strength and frequency of events declined to the present, attaining values first surpassed 4-5 ka (Clement et al., 2000). The modeled ENSO variability during the Holocene is more-or-less consistent with the alluvial sedimentation reconstruction from Laguna Pallcacocha by Moy et al. (2002) and with Tudhope et al. (2001) who investigated the inter-annual range of 18O/16O ratios in corals from Papua New Guinea. But these studies are not easily reconciled with either the Galapagos lake level reconstruction of Colinvaux (1972), the SST reconstructions by Koutavas et al. (2004), or geoarcheological evidence from Peru which conclude that a near-permanent El NiÒo state characterized the early mid-Holocene (~8-5 ka) (Sandweiss et al., 1996). All of the aforementioned paleoclimate studies, except the SST reconstructions by Koutavas et al. (2004 & in prep.) are hampered by their reliance on a single proxy and/or a single site.