Impacts of leaf age on the spectral and physiochemical traits of trees in Amazonian forest canopies

This doctoral research presents the first comprehensive analysis of the morphological, biochemical and spectral leaf traits of canopy and emergent tropical trees during natural (in situ) leaf ageing. It adopts an interdisciplinary approach and combines multiple scales of analysis to generate insights into the effects of natural leaf ageing on our current understanding of tropical leaf trait variation, chemometric models used to spectrally predict leaf traits, and together with other leaf phenological processes, on remotely-sensed vegetation indices (VIs) commonly used to monitor canopy dynamics in tropical evergreen forests. The first research paper of this thesis (Chapter 4) examines the effects of leaf age on morphological and biochemical leaf traits and demonstrates that leaf age is a significant driver of leaf trait variation in Amazonian canopy trees and that leaf age differences could potentially account for a significant fraction of what we currently understand as intra- and interspecific leaf trait variation. It also highlights that age-related trait variation within and between individual trees could play a significant role in shaping community composition and structure of tropical canopies. The leaf traits examined in Chapter 4, among others, have been shown to directly influence the spectral reflectance behaviour of leaves. Therefore, Chapter 5 investigates the effects of leaf age on leaf spectral properties within and across a tropical canopy tree community. This study reveals that trees with diverse leaf properties age in a similar manner in terms of spectral properties. This is one of the most important findings of this thesis and lead to the development of a novel chemometric partial least square regression (PLSR) model to predict leaf age from hyperspectral data. This model extends the utility of current spectroscopic methods and introduces a simple and efficient approach for predicting and monitoring leaf age in lowland tropical forests with important implications for remote sensing. Additionally, this study is the first to provide evidence of age-related reflectance changes in leaves that have significant impacts on vegetation indices commonly used to monitor productivity and canopy dynamics in tropical evergreen forests. Considering the findings of the previous two research chapters, Chapter 6 investigates if chemometric PLSR models used to spectrally predict some of the important leaf traits for plant physiology and economy (leaf mass per area, LMA; water content, LWC; phosphorous, P; nitrogen, N; and carbon, C content) investigated in the previous two research chapters could be significantly biased by variation in leaf age. This is particularly relevant as the current standard protocol is to use only "fully expanded mature leaves" to calibrate these models. This study demonstrates that PLSR models developed using the current standard protocol display age/temporal sensitivity, which has important implications for forest canopy communities with both synchronised and unsynchronised leaf phenology. The final research chapter of this thesis, Chapter 7, demonstrates that the phenological age-related changes in leaf spectral properties reported in Chapter 5 are also expressed at the canopy scale but influenced by both canopy leaf area (CLA) and the leaf phenological behaviour of individual trees. This study also reveals that the seasonality of greenness VIs such as NDVI and EVI2 are more strongly correlated to phenological changes in CLA then changes in leaf reflectance and proposes that NDWI (water content VI) which was found to be strongly correlated to age-related changes in leaf reflectance should complement greenness VIs in phenological studies. Furthermore, by combing leaf, canopy and community scale phenological observations, this study shows that complex and diverse leaf phenological behaviours exhibited by tropical canopy trees, at both the individual and community scale, challenge our current ability to remotely sense tropical canopy dynamics. Finally, this chapter highlights the need for more widespread phenological studies that examine the interaction, covariation, asynchrony and unique behaviours of tropical phenological processes at different scales. Such studies would enable the development of a significant mechanistic understanding of what creates and drives different phenological mosaics identified by remote sensing studies across tropical forests and in modelling their effects on water and carbon fluxes in tropical forest ecosystems.