A PALEOBOTANICAL METHOD FOR RESTORING FAR–TRAVELLED TERRANES

The proportion P of species of “smooth–margined” woody dicotyledonous angiosperm leaves within a flora is strongly correlated to mean annual temperature (MAT=28.9P + 1.66 (R²=0.88). Modern MAT measurements from 1,944 stations in North America, corrected to sea level using a spatially averaged environmental lapse rate of 2.21°C/km, are well correlated with latitude (R²=0.89), showing that MAT and P are a strong function of latitude and useable for paleolatitude estimation. Numerous studies suggest that many Mesozoic and Cenozoic lithotectonic terranes on the western margin of North America are latitudinally offset. We use estimates of P from fossil floras in two of these terranes to restore their original latitude by comparing their P to contemporaneous floras from stable North America. First, we estimate the mid–Cretaceous latitudinal position of the Baja BC block from the indigenous flora of the Winthrop Formation in northern Washington State. This flora has a P value of 0.76 ± 0.18, which translates to a MAT of 23.7°C ± 2.5°C indicating a paratropical to tropical climate (confidence intervals are cited at 95%). In comparison, nearby mid–Cretaceous floras in the Northern Rocky Mountains show a P » 0.5 or a MAT » 16°C indicating a warm temperate climate. An offset of 2900 ± 2000 km is estimated relative to a paleolatitudinal reference of P for stable North America based on 12 mid–Cretaceous floras. Similarly, the early Tertiary Yakutat terrane of southeast Alaska is hypothesized, on plate tectonic and other evidence, to have originated ~2,000 km south. Our paleofloral analysis yielded P=0.59 or MAT=19°C. In comparison to stable North American floras of the same age, this P indicates minimum of 2,000 km of northward offset. This paleobotanical method for restoring far–traveled terranes requires only a fossil flora on an offset terrane and a cratonal (stable) reference set of floras over the paleolatitudinal range of the possible offset. P estimates must be corrected for paleoelevation using either lapse rates or geological arguments. Uncertainties depend on sample and reference frame size and are on the order of 1,000–1,500 km, which is slightly larger than those for paleomagnetic studies.