PALEO-CO2 RECONSTRUCTIONS USING FOSSIL PLANTS FROM THE MIDDLE MIOCENE CLARKIA LAKE DEPOSIT

To better understand the current climate change and its sensitivities caused by anthropogenic CO₂ emissions, precise reconstructions of ancient atmospheric CO₂ levels are necessary for building a more reliable relationship between global temperatures and CO₂ levels. Among various paleo-CO₂ estimate methods, those based on plant fossils have been focused on morphological and/or physiological features of higher plants that are sensitive to CO₂ levels through photosynthesis.

The ancient Clarkia Lake was created due to the Columbia River Basalt damming the proto–St. Maries River in northern Idaho during the middle Miocene (15.78 Ma). The lacustrine deposit provides a thousand years of high-resolution record with finely laminated depositional layers yielding abundant, diversified, and well-preserved fossils for the reconstruction of paleoclimate, paleoenvironments, and paleoecosystem during the Miocene warming period. In the past decade, our lab has used the empirical stomatal frequency—SF (including stomatal density—SD, stomatal index—SI, and stomatal number per leaf needle length) methods, the C₃ plant carbon isotope fractionation (Δ¹³C) method, and the mechanistic gas exchange modeling method such as the Franks Model to both gymnosperm and angiosperm fossils (mainly Metasequoia, Taxodium, and Lithocarpus) from this world-renowned ancient lake deposit.

We show that two deciduous conifers (Metasequoia and Taxodium) that are most commonly found from Clarkia fossil sites provide more consistent and reliable paleo-CO₂ reconstruction than most angiosperm fossils. Recent results confirmed that the SF methods generally underestimate paleo-CO₂. For example, the mean values of Metasequoia and Taxodium using SF methods are 282ppm and 371ppm respectively. The C₃ plant Δ¹³C method yields wide range of estimates, from 255ppm to 787ppm, averaged 388ppm which is also lower than expected. Compared with these two methods, the Franks Model’s results using gymnosperms are more consistent, around 500ppm, with most of the previously reported middle Miocene CO₂ levels. We also tested the reliability of these methods by using their nearest living relatives (NLRs) grown under known CO₂ levels and compared our results with CO₂ reconstructions for the middle Miocene warming using a variety of methods.