SUMMARY OF PROJECT RESULTS TO DATE

The timing of the last interglacial period (LIG) can be determined most precisely by U-series dating of corals from emergent reefs and terraces on coasts and islands. Based on U-series dating of corals from Hawaii, California, Florida, and the Bahamas, sea levels were higher than present from ~130 ka to ~112 ka (oxygen isotope substage 5e and possibly part of 5d), much longer than is indicated by the deep-sea record of SPECMAP (see Fig. 4A: U-series ages from these localities). Shallow marine paleotemperatures can be inferred from marine terrace faunas that have reliable U-series ages or can be correlated to dated localities using amino acid geochronology. Faunal assemblages (mostly mollusks and ostracodes) from most ~125 ka-old marine terrace deposits on the Pacific, Atlantic, and Arctic coasts of North America contain extralimital southern species indicating warmer-than-present waters during the peak of the LIG (see Fig. 4B).

Figure 4A

Figure 4B

LIG pollen and macrofossil records from North America are rare compared to Europe. However, a number of localities have been found on the Pacific coast, midcontinent, southeastern Canada, Alaska, and Arctic Canada (see Fig. 5: pollen and macrofossil localities for North America). Almost all localities indicate peak LIG conditions as warm or warmer than present. British Columbia and Washington had cedar-hemlock-Douglas fir forest similar to present, California was dominated by oak woodland, Illinois had deciduous forest/savanna, Alaska and Yukon had boreal forest over a greater extent than present, and in Arctic Canada, shrub tundra migrated north to areas now occupied only by herb tundra (see Fig. 6: last interglacial treelines). During the LIG, greater summer warmth is indicated by hickory, spruce and shrub birch treelines that were all north of their present limits. At localities where particularly detailed records of the LIG exist, there is evidence for considerable complexity in vegetation changes. In central Alaska, for example, there were at least two periods of boreal forest separated by an interval of tundra vegetation, indicating a complex summer temperature history; in addition, the earlier period of boreal forest growth may have had higher precipitation than present. These shifts in vegetation within the LIG sensu lato may may correspond in part to the complexity of the marine oxygen isotope record represented by substages 5e through 5a.

Figure 5

Figure 6

Where pollen or macrofossil records are absent, the climate of the LIG in the North American midcontinent may be inferred from properties of LIG paleosols, because soil geography is to a great extent a function of prevailing climate and vegetation. Based on TL dating of bracketing loesses, the LIG soil formed between ~130 ka and ~45 ka (all of oxygen isotope stages 5 and 4 and part of stage 3) in the midcontinent of North America. Modern soils in a transect from Ohio to Colorado (a precipitation gradient of ~1000 mm/yr to ~400 mm/yr) show distinct morphological changes, particularly over the forest-prairie boundary. LIG soils in this same transect have morphological properties and carbon isotopic data that indicate that the forest-prairie border may have been west of its present boundary, indicating higher precipitation in the midcontinent during at least parts of the LIG.

Figure 7

A synthesis of all of the data compiled thus far allows reconstruction of a synoptic-scale paleoclimate picture for North America during the last interglacial (see Fig. 7: synoptic map of North America). Comparisons to atmospheric general circulation model (AGCM) simulations of the LIG show both agreement and disagreement with the geologic record. Both AGCM results and geologic data indicate greater warmth in the midcontinent and high latitudes of North America during the LIG. However, the greater precipitation in interior Alaska and the midcontinent shown in the geologic record contrasts strongly with AGCM results indicating lower precipitation in these regions during the LIG. The comparisons suggest that AGCMs are better at simulating temperature fields than they are at simulating precipitation fields, an important difference to consider when using AGCMs for future climate simulations.