Estimating carbon assimilation rates from fossil leaves and application to the mid-Miocene Clarkia forest

The rate of carbon assimilation in leaves (A) is a key trait central to a plant's economic strategy that has downstream impacts on the regional and global cycling of carbon and other nutrients. Most previous paleoecological studies estimate A from nearest living relatives or leaf vein density.

We present a method for reconstructing A using gas-exchange modeling that requires both measured (stomatal size and density, leaf δ¹³C) and inferred (e.g., atmospheric CO₂ concentration) inputs. We apply this method to ten extant taxa and nine fossil taxa representing common angiosperms of the exquisitely preserved mid-Miocene (~15.9 Ma) flora at Clarkia in northern Idaho, USA.

Application to extant taxa produces estimates of A that are near measured values on the same leaves (R² = 0.89 across all taxa). Median reconstructed A for fossil taxa range from 9.5-21.7 µmol m^-2 s^-1 with 95% confidence intervals ~+51%/-38% indicating that most species are statistically indistinguishable. Sensitivity tests show that our method is most reliable when CO₂ is well-constrained, but when that is impractical, taxa within single sampling horizons (with a presumed fixed CO₂ concentration) can be organized by A into a relative rank order with tighter confidence intervals (~+16%/-14%).

Following this relative approach at Clarkia, we reconstruct high A for taxa whose modern relatives are characterized by rapid growth and/or riparian habitats (Castanea and Platanus) and corroborate previous interpretations on the ecology of taxa whose modern relatives are less known (Quercus simulata).