Filling the gaps in the floral record : implications for paleobotanical paleoclimate models.

Plants are sensitive indicators of climate and environment. In particular, the size and shape (physiognomy) of woody dicotyledonous angiosperm leaves are correlated to climate, and these relationships are the basis for paleobotanical proxies for mean annual temperature (MAT) and mean annual precipitation (MAP). To improve these proxies, I tested two fundamental assumptions: (1) leaf phenotypic plasticity can be differentiated from developmental change and (2) modern leaf physiognomic-climate relationships hold through time and across all plant groups and geographic areas. Most leaf physiognomic climate proxies were developed using primarily temperate Northern Hemisphere modern floras with relatively few sites from the tropics, especially from Africa. However, limited research on African floras suggests that they have a different relationship between climate and leaf physiognomy than Northern Hemisphere floras, which calls into question the second assumption. This dissertation tests these assumptions in modern leaves and applies leaf physiognomic proxies to two fossil leaf sites from Kenya. The first assumption was tested by measuring the physiognomy of four species of Vitis from the USDA Germplasm Repository (Geneva, NY) from two growing seasons with different environmental conditions. This work demonstrated that the first assumption of leaf physiognomic proxies is reasonable. To test the second assumption, I used the Digital Leaf Physiognomy (DiLP) protocol to measure the physiognomy of woody dicot angiosperms leaves from modern sites across tropical Africa and estimated climate using DiLP and the univariate physiognomic models Leaf Area Analysis (LAA), Leaf Margin Analysis (LMA). These results indicate that assumption 2 is not a valid assumption and that existing physiognomic models should be applied with caution to African fossil floras. As a case study, I applied leaf physiognomic models to two Early Miocene African fossil floras on Rusinga Island (Lake Victoria, Kenya): Kiahera Hill and R3. I used LMA and LAA to reconstruct MAT and MAP for Kiahera Hill and R3, which I compared to previous work on the R5 flora. This work, coupled with previous research, suggests that the Hiwegi Formation on Rusinga Island samples multiple environments in the Early Miocene, which in turn likely influenced the evolution and habitat preferences of early hominoids