For over a century, leaf shape in woody non-monocotyledonous species has been used to reconstruct aspects of climate related to temperature and precipitation. These methods are typically calibrated with complete leaves from living forests. An understudied issue is how appropriate these calibrations are to incomplete leaf compression fossils that dominate most fossil leaf assemblages. We sought to test the fidelity of the leaf-climate signal with analogous modern leaves, excavated from within the sediment layer at two temperate research forests, Harvard Forest and the Smithsonian Environmental Research Center. At both locations, we targeted leaves that underwent transport and deposition in environments that commonly contain fossil leaves in the geologic record: swamp and river margins. In this regard, these incomplete leaves represent a natural experiment of leaf taphonomy that is a good partial analog for leaf fossilization. We are measuring a suite of size and shape variables that are used in the multivariate leaf-climate method called digital leaf physiognomy. These variables are mostly related to tooth size, tooth abundance, and leaf dissection. We will then compare the distributions of these shape variables to complete fresh leaves of the same species sampled from the same sites. We anticipate that our results will provide a more robust framework for interpreting paleoclimate from leaf shape.