Testing the accuracy of palaeo-CO2 estimates from the Franks leaf gas exchange model

In light of anthropogenic climate change, the use of palaeo-CO₂ proxies to reconstruct past atmospheric pCO₂ is essential for understanding the impact of CO₂ variations on the Earth's climate. The mechanistic leaf gas exchange model of Franks et al. (2014) utilises the relationship between atmospheric pCO₂ and chemical and morphological traits of leaf fossils to reconstruct pCO₂. While the model has shown promising results, its broader applicability remains unclear, particularly whether there are consistent off-sets in pCO₂ estimates among major plant clades. Here, we evaluate the performance of the model across different phylogenetic groups (five ferns, six gymnosperms, five angiosperms), with generic input values and recommended adjustments to isotopic values, assimilation rates and the scaling of stomatal conductance. Results show species-specific variations with generic values; a clade-level bias is therefore unlikely. As a consequence, the correction factors to remove a phylogenetic effect from isotopic values do not show improvements across plant groups. Adjustments to the assimilation rate based on phylogeny, habit and habitat result in less species-specific variation and in an improvement in pCO₂ estimates. While the minor change in the scaling value of stomatal conductance for woody angiosperms results in inconsiderable differences in accuracy, precision decreases due to higher error rates. Overall, the results of this study show that the Franks model has a strong potential for reconstructing past atmospheric pCO₂ when applied in a multitaxon framework, especially with recommended assimilation rate values.