DIGITAL LEAF PHYSIOGNOMY: USING LEAF SIZE AND SHAPE TO RECONSTRUCT ANCIENT CLIMATES

For over 100 years paleobotanists have used the sizes and shapes (physiognomy) of fossil leaves to reconstruct paleoenvironments. The most commonly used methods are leaf margin analysis, which is based on the correlation between mean annual temperature (MAT) and the proportion of untoothed woody dicot angiosperm species, and leaf area analysis, which is based on the relationship between mean annual precipitation (MAP) and leaf area. Although these methods appear robust, they only utilize a single aspect of leaf physiognomy. Over the last few years, we and other colleagues have developed a multivariate technique for reconstructing climate called digital leaf physiognomy. This method uses digitally-measured, reproducible, continuous variables, such as number of teeth and perimeter:area relationships, to estimate MAT and MAP. Here we present revised models for estimating MAT and MAP that utilize digitally-measured variables from an expanded dataset of 95 sites from around the world. The sites have highly different phylogenetic histories and represent almost all extant biomes. Applications of our revised models to early and middle Paleocene fossil floras from the Fort Union Formation in the Williston Basin of North America yield MAP and MAT estimates that are more precise than the traditional univariate methods. The standard errors MAT estimates made using the digital leaf physiognomy models are ~ 1 °C more precise than leaf margin analysis, and for MAP are ~50 cm more precise than leaf area analysis. Importantly, the digital leaf physiognomy estimates are 3-4 °C warmer than estimates from leaf margin analysis. These new warmer estimates are more in line with expected temperature estimates for this time period, and are probably due to the incorporation of additional variables related to number of teeth and tooth size and to the shape of the leaf lamina. This suggests that the digital leaf physiognomy technique may be a more accurate and precise method for estimating paleoclimate in the past.