PHYLOGENETIC SIGNAL IN LEAF TRAITS USED FOR PALEOCLIMATE ESTIMATIONS

Leaf physiognomy is well known as useful in paleoclimate estimation, because they display phenotypic plasticity in response to the environment and because species are thought to occur where they are best adapted to the environment. For example, there are significant, taxon-free positive correlations between the proportional richness of untoothed taxa and mean annual temperature, and between leaf area and precipitation. Several additional leaf shape variables are used to predict paleoclimate in the recently developed method known as digital leaf physiognomy. However, leaf characters are influenced not only by the environment in which species occur, but also by the phylogenetic history of trait and plasticity evolution. Phylogenetic signal, a tendency for close relatives to resemble one another, has been widely demonstrated across many species. Thus we hypothesize that there is phylogenetic signal in calibration floras which may affect climate inference models. Although, it is commonly assumed by paleobotanists that leaf physiognomy is largely independent of phylogeny, this has never been tested quantitatively. We used Phylocom and Phylomatic software assembly a phylogenetic supertree and test for phlyogenetic signal in a large leaf physiognomy dataset. Results show that there is low phylogenetic signal overall across the leaf character dataset, which is good news for conventional paleobotanical wisdom. However, there is non-trivial amounts of phylogenetic signal in several traits relating to leaf teeth, especially tooth area and number of teeth. The influence this phylogenetic signal on climate estimation, and potential methods to account for it will be discussed.