David C. Marvin
Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, MI 48109, UNITED STATES OF AMERICA.
Humans are altering tropical ecosystems and their biotic and biogeochemical processes with unprecedented scale and severity. The increasing size and abundance of tropical lianas (woody climbing plants) relative to trees may be the result of global anthropogenic change, and may further alter forest function. Yet the mechanisms responsible for this reported phenomenon are unclear, and the scale at which it occurs has been unexamined. In this dissertation, I use a combination of empirical experimentation, ground-based forest censuses, and advanced airborne remote sensing imagery and analysis to investigate the question of why lianas are increasing and at what spatial scale. First, I tested the hypothesis that elevated CO2 gives species of lianas a growth advantage relative to trees, especially during periods of seasonal drought. In the first experiments to directly compare the relative response of tropical liana and tree species to elevated CO2, I found no significant differences between the two growth forms. Both lianas and trees responded equally well to elevated CO2, even when soil water was limited by seasonal drought. Second, I extended the tests of liana-tree response to CO2 to include the effects of soil nutrient availability. No interactions between elevated CO2 and either soil nitrogen or phosphorus availability were found for lianas. Instead, changes to soil nutrient availability or CO2 alone had strong and significant effects on lianas. Finally, I used data collected from my field censuses to train machine learning algorithms to detect severe liana coverage in tree canopies using high-resolution hyperspectral imagery. This method proved to be very accurate at distinguishing severe liana cover from liana-free cover in tree canopies, and quantified severe liana infestation as 11.9%-18.0% of the total canopy cover over a 600-ha tropical forest. The results of the experiments and the development of landscape-scale liana detection methods are key steps toward a full understanding of the mechanisms and scope of the liana increase.