last 200 years (Urban et al., 2000). Discussion continues on the statistical significance and long-term persistence of these switches and on whether they should be considered evidence of normal oscillations, of short-lived abrupt shifts, or of long-lived abrupt climate change (e.g., Rajagopalan et al., 1999; Trenberth and Hurrell, 1999a,b). Further back in the Holocene, the ENSO system might have been dramatically different from today, with much reduced variability and fewer strong events (Overpeck and Webb, 2000; Diaz and Markgraf, 2000; Cole, 2001; Sandweiss et al., 2001; Tudhope et al., 2001). Although the time at which modern ENSO variability became established is not known, there have been several model-based efforts to explain the changes, all tied to the response of the coupled atmosphere-ocean system to small orbitally induced insolation changes (Bush, 1999; Otto-Bliesner, 1999; Clement et al., 2000, 2001). The shift to more-modern ENSO variability also might have been coincident with other earth-system changes 4,000-5,000 years ago. Sandweiss et al. (2001) suggested that ENSO events were absent or substantially different from more recently between 8,800-5,800 years ago, present but reduced between 5,800-3,200 years ago, and increased to modern levels between 3,200-2,800 years ago, that would be consistent with other data that they summarize. Rodbell et al. (1999) placed the Holocene onset of El Niños at 7,000 years ago, with the beginning of modern levels reached 5,000 years ago.
Although there are other hints of important abrupt climate changes in the Holocene record, most of them have not been studied to the degree needed to place them in a coherent context (for example, examined at multiple sites). One important observation is that the landfall frequency of catastrophic hurricanes has changed rapidly during the Holocene. For example, the period about 1,000-3,500 years ago was active on the Gulf Coast compared with the last 1,000 years and changes in North Atlantic climate could be the primary cause (Liu and Fearn, 2000; Donnelly et al., 2001a,b). The period near 1,000 years ago was also possibly marked by a substantial change in hydrologic regimes in Central and North America (Hodell et al., 1995, 2001; Forman et al., 2001).
Climate variations within the last millennium are, in general, better resolved temporally and spatially than are variations earlier in the Holocene. This is due largely to the greater availability of annually dated records from historical documents, trees, corals, ice cores and sediments, but this availability is also due to greater emphasis on the last millenium by large paleoenvironmental science programs, such as PAst Global changES (PAGES) of the International Geosphere-Biosphere Programme (IGBP). Per-