INVITED: A NEW APPROACH FOR RECONSTRUCTING PALEOCLIMATIC AND PALEOECOLOGICAL VARIABLES FROM THE SIZES AND SHAPES OF FOSSIL PLANTS

The most successful proxy for terrestrial, pre-Pleistocene temperatures is based on the strong correlation in extant forests between the proportion of plant species that have untoothed leaf margins and mean annual temperature. However, this proxy, called leaf-margin analysis, relies on a single, binary variable. If the sizes and shapes (physiognomy) of leaves could be more fully described, improvements to leaf-margin analysis should be possible. Further, proxies for important ecological variables such as leaf mass per area and foliar nitrogen content have not been developed because of a general lack of appropriate variables that can be reliably measured on fossils. Here we present a new approach for using the physiognomy of fossil leaves to reconstruct paleoclimatic and paleoecological information. This approach, termed digital leaf physiognomy, is based on digital leaf images and largely automated computer routines that generate a suite of continuous size and shape variables related to tooth size, tooth abundance, and degree of leaf dissection. Our analysis of 17 extant floras along the east coast of the U.S. and the Republic of Panama shows the power of this new method. First, multivariate models based on these new variables can estimate mean annual temperature with significantly greater accuracy than leaf-margin analysis. The most accurate model that can be applied to fragmentary leaf fossils is composed of three variables: the ratio of tooth area to blade area, perimeter ratio (leaf perimeter / perimeter after all teeth are removed), and the margin percentage character used in leaf-margin analysis. Second, tooth area correlates significantly with both leaf mass per area and foliar nitrogen content. The leaf-climate correlations are not seriously degraded even when based on only one leaf image per species; moreover, the model most applicable for fossils shows no more signal degradation from leaf fragmentation than leaf-margin analysis. This new approach offers the potential, for the first time, to track in ancient ecosystems the relationships between climate and ‘leaf economic' variables such as leaf mass per area and foliar nitrogen content. A preliminary example from the fossil record will be presented.