Quantification of large uncertainties in fossil leaf paleoaltimetry

Estimates of paleoelevation potentially constrain geodynamic models of continental deformation and inform interpretations of landscape and climate evolution. One widely used, paleobotanical approach reconstructs paleoelevation from the difference in estimated atmospheric enthalpy between a known sea level and a targeted, coeval, elevated fossil floral site. Enthalpy is estimated using Climate-Leaf Analysis Multivariate Program (CLAMP) on 31 leaf size and shape variables that have been calibrated in living forests. Errors related to CLAMP are significantly greater than often reported, and there are many sources of large potential error related to this method that are either difficult to quantify or unquantifiable and are thus not documented. Here, we quantify one significant bias, toward underestimation of leaf area in the CLAMP data set (∼50%), that affects all CLAMP climate estimates, including enthalpy. Crucially, errors in paleoelevation when the leaf size bias is included are in the range of ±2 km or more, at least 2 times the previous estimates, and exceeding the plausible paleoelevations of many fossil sites. Previously published paleoelevations derived from this technique are unlikely to be accurate either in magnitude or in estimated error.