A0500
Title: Shape analysis of European beech leaves
Authors: Alejandro Pereira - Georg-August-Universität Göttingen (Chile) [presenting]
Thomas Kneib - University of Goettingen (Germany)
Abstract: Statistical analysis of nonstandard data has become increasingly important in ecology and forestry, given the multimodal nature of the problems arising in these fields. One area of particular interest in Botany is the use of shapes of leaves, petals, crowns, whole plants, and more to use as a tool in their studies. A deeper understanding of the available tools of shape analysis for practitioners is then very important. A comparison study is performed on the statistical analysis of shape and variability via elliptical Fourier analysis, Kendall's landmarks shape space, and the elastic metric approach. The aim is to compare the performance of the selected methods and their uncertainty quantification (UQ) properties. The methods are illustrated by analyzing the shape of juvenile European beech (Fagus sylvatica) leaves obtained from a pot experiment conducted as a part of the research training group "Enrichment of European beech forests with conifers: impacts of functional traits on ecosystem functioning (EnriCo)". In the experiment, juvenile trees were grown under different mixture conditions as well as water and nitrogen level treatments. Further, the treatment classification performance of the methods is studied.
