Variability in leaf optical properties of Mesoamerican trees and the potential for species classification

• Published in: American journal of botany 2006-04, Vol.93 (4), p.517-530
• Main Authors: Castro-Esau, Karen L, Sánchez-Azofeifa, G. Arturo, Rivard, Benoit, Wright, S. Joseph, Quesada, Mauricio
• Format: Article
• Language: English
• Subjects: Botany; chemical constituents of plants; chlorophyll; Chlorophylls; Classification; Ecology; Forest trees; hyperspectral; Infrared reflection; interspecific; intraspecific; leaf level; Leaves; Mesoamerica; mesophyll; Optical reflection; plant physiology; reflectance; Remote sensing; seasonal variation; spatial variation; Species; spectral analysis; Spectral reflectance; taxonomy; thickness; tree crown; Tree crowns; Trees; tropical forests; tropical trees
• ISSN: 0002-9122; 1537-2197
• DOI: 10.3732/ajb.93.4.517

Leaf traits and physiological performance govern the amount of light reflected from leaves at visible and infrared wavebands. Information on leaf optical properties of tropical trees is scarce. Here, we examine leaf reflectance of Mesoamerican trees for three applications: (1) to compare the magnitude of within- and between-species variability in leaf reflectance, (2) to determine the potential for species identification based on leaf reflectance, and (3) to test the strength of relationships between leaf traits (chlorophyll content, mesophyll attributes, thickness) and leaf spectral reflectance. Within species, shape and amplitude differences between spectra were compared within single leaves, between leaves of a single tree, and between trees. We also investigated the variation in a species' leaf reflectance across sites and seasons. Using forward feature selection and pattern recognition tools, species classification within a single site and season was successful, while classification between sites or seasons was not. The implications of variability in leaf spectral reflectance were considered in light of potential tree crown classifications from remote airborne or satellite-borne sensors. Species classification is an emerging field with broad applications to tropical biologists and ecologists, including tree demographic studies and habitat diversity assessments.