Plant-derived terpenoids as paleovegetation proxies: evaluation of the proxy with Paleocene and Eocene megafloras and plant biomarkers in the Bighorn Basin, USA

Plant terpenoids (defense compounds synthesized from the 5-carbon building block isoprene) have a long history of use as geochemical plant biomarkers, and potentially can be used to reconstruct changes in the abundances of major land plant groups in rocks and sediments that do not preserve plant megafossils or pollen. Pentacyclic triterpenoids are synthesized almost exclusively by angiosperms whereas conifers produce the tricyclic diterpenoids. Many previous studies have focused on the use of di- to triterpenoid ratios to reconstruct floral changes in the geologic past, however few studies have compared terpenoid-based paleoflora proxies to pollen or megafossils. Prior reconstructions also did not take into account differences in biomarker production between plant functional types, such as deciduous and evergreen plants, which can be quite large. To investigate the use of terpenoids as paleoflora proxies, we examined sediments from the Bighorn Basin (Wyoming, USA) where ancient megafloras have been studied in detail. We analyzed di- and triterpenoid abundances as well as plant leaf waxes (n-alkanes) and other biomarkers in a total of 75 samples from 15 stratigraphic horizons from the late Paleocene (62 Ma) to early Eocene (52.5 Ma). By comparing terpenoid ratios with abundances estimated from plant megafossils, we can evaluate the utility of terpenoids as paleovegetation proxies. In nearly all samples, angiosperm triterpenoids are significantly lower in abundance than conifer diterpenoids. This contrasts with leaf fossil data that indicate paleofloras were dominated by angiosperms in both abundance and diversity. Traditional use of terpenoid paleovegetation proxies would therefore significantly overestimate the abundance of conifers, even when accounting for plant production differences. To determine if this overestimate is related to the loss of angiosperm triterpenoids (rather than enhanced production of diterpenoids in the geologic past), we compared angiosperm triterpenoids to the n-alkane leaf waxes, which are produced primarily by angiosperms. After accounting for concentration differences between these two biomarker groups in modern plants, it is apparent that triterpenoid amounts are still significantly lower than expected in nearly all of our samples. We suggest a loss of triterpenoids might be related to enhanced diagenesis of triterpenoids in oxidizing terrestrial sediments. In order to reconstruct paleovegetation we propose a new method that employs a ratio of diterpenoids (conifers) to n-alkanes (angiosperms) and accounts for biomarker biomass abundance differences. Using this approach, we estimate paleovegetation communities similar to those predicted from megafossils. Although this new biomarker-based paleovegetation proxy works for sites within the Bighorn Basin, we stress this approach will need to be evaluated for other depositional environments using pollen or megafossil data.