Plant-derived terpenoids, long recognized as biomarkers, can help reveal the major taxonomic groups of land plants present in ancient environments, even if rocks and sediments do not preserve plant macro- or microfossils. Previous studies have used simple di- to triterpenoid ratios to reconstruct floral changes in the geologic past, but few have compared terpenoid ratios with estimates of floral composition from fossils. Further, reconstructions have not taken into account differences in biomarker production (i.e. concentration relative to leaf biomass) between different types of plants. Here we have examined terpenoids from early Cenozoic fluvial rocks from the Bighorn Basin (Wyoming, USA), where fossil flora has been studied in detail. We analyzed the distributions of diterpenoids, triterpenoids and n-alkanes from leaf wax in a total of 43 samples from 15 stratigraphic horizons of late Paleocene (63 Ma) to early Eocene (53 Ma) in age. In nearly all samples, triterpenoids, derived from angiosperms, were significantly lower in abundance than conifer-specific diterpenoids, a finding that contrasts with plant fossil evidence for the same rocks. This suggests that di- to triterpenoid ratios severely underestimate the abundance of angiosperms in paleovegetation. Angiosperms dominated n-alkane production among modern plants, and we propose a new paleovegetation proxy based on the ratio of diterpenoids (conifers) to n-alkanes (angiosperms), corrected for lipid production estimated from extant vegetation. Using diterpenoids and alkanes, we infer the composition of paleovegetation to be similar to that inferred from plant fossils. Although the approach works well for the Bighorn Basin, we stress the new paleovegetation proxy will need to be evaluated for other time periods, communities, paleogeography and depositional environments with pollen or megafossil data available.
