USING ANGIOSPERM LEAVES TO BUILD A PALEOTEMPERATURE CURVE FOR THE LATE CRETACEOUS OF THE ROCKY MOUNTAIN REGION OF NORTH AMERICA

The Western Interior of North America contains thick sequences of interbedded marine and terrestrial rocks that can be correlated using ammonite biostratigraphy, radiometric ages from volcanic ash beds, and magnetostratigraphy. The physiognomy of fossil angiosperm leaves provides proxy data for mean annual temperature and mean annual precipitation. The one existing temperature record for the terrestrial sequence (Wolfe and Upchurch, 1987) is largely based on re-analysis of data from outdated monographs and has relatively coarse temporal resolution. Recognizing this problem, the Denver Museum of Nature & Science has begun to assemble a suite of newly collected Cretaceous leaf sites with the intention of building a reliable temperature curve for the Late Cretaceous. To date, 204 sites have been located, quarried, and analyzed. Coverage is strongly weighed towards the late Maastrichtian (149 sites from the Hell Creek, Laramie, Lance, and Denver fms.); followed by the early Maastrichtian (30 sites; Almond and Meeteetse fms.); Campanian (14 sites; Dinosaur Park, Mesaverde, Neslen, Rock Springs, and Two Medicine fms.); Cenomanian (10 sites; Dakota and Frontier fms.); and the Coniacian (1 site; Henefer Fm.). Two main problems hamper this effort: 1) Many of these floras are preserved in volcanic ash and represent herbaceous and shrubby vegetation which is not suitable for paleoclimate analysis; and 2) Many of the floras are not adequately sampled to produce (or the floras do not contain) the 20 species of dicotyledonous angiosperms needed to provide a reliable temperature estimate. In the late Maastrichtian, where sample coverage is dense, angiosperm species are abundant, and magnetic reversals are closely spaced, there is a tight correlation between terrestrial mean annual temperature and marine paleoclimate proxies. Lower in the section, angiosperms do not always dominate assemblages, ash bed floras are more common, and the Cretaceous long normal (C34n) precludes high resolution magnetostratigraphic correlation.