Reflected-light image analysis of conifer tree rings for dendrochronological research.

Abstract

The primary objective of this dissertation research is to use reflected-light image analysis to measure brightness of standard samples of conifer rings and then use brightness in dendrochronological research as a substitute for density. I developed an imaging system that ensures identical configuration of all components and measuring steps for all rings of a sample so that subsequent comparison of brightness between rings would be valid. From a mesic New England tree-ring site, I measured ring brightness of cores that had been previously measured using X-ray densitometry. Latewood brightness and density both correlate with April-May temperature such that they reconstruct that climate variable equally well. From a semiarid Southwest tree-ring site, I measured ring brightness of cores with severe extraneous color--mostly due to heartwood-sapwood color differences. Bleaching and organic extraction of cores did not overcome the problem of extraneous color, but autoregressively modeling brightness index series did. Various brightness and width variables combined to model July-October precipitation, a climate variable not usually reconstructed by Southwest tree-ring sites. From a stand of trees affected by a past earthquake, I measured ring brightness of one tree that responded to surface deformation with an apparent change in latewood density. Absolute latewood brightness did not change per se after the earthquake, but the amount of latewood relative to the total ring increased dramatically. Although technical and paleoenvironmental issues remain for future research, this study indicates that reflected-light image analysis is an excellent tool in dendrochronological research for increasing our understanding paleoenvironmental processes of the latest Holocene. The secondary objective of this dissertation research is to demonstrate a method for identifying low-frequency variation of tree-ring chronologies and/or past climate as reconstructed using tree-rings. This method provides confidence intervals with which to judge the significance or importance of low-frequency departures in tree-ring data as well as a visual basis for determining whether or not low-frequency variation is robustly estimated. This method is a re-ordering of the individual steps commonly used in constructing tree-ring chronologies or reconstructions.