PS 7-62
A robust method to measure tree-rings in high noise stem cross-sections
We use dendroecology to document tropical forest growth by measuring tree-rings to reconstruct its historical growth and relate growth patterns to environmental influences. We have found that measuring the width of annual rings with multiple transects provides poor estimates of growth in non-circular or ring wedging species. Differing ring number across radii for individual stem cross sections reduce the ability to construct average growth estimates. In addition, growth for highly lobed stem cross-sections, cannot be accurately measured with four transects. Thus we develop and test new methods to obtain more quantitative rigor.
Tracing earlywood–latewood boundaries with Bézier curves using vector graphics software improves past season growth estimates for trees with asymmetric stem growth. The higher sampling detail involved with tracing, yields a two dimensional (area) instead of one dimensional (length) metric. Annual ring area and its distribution around the stem circumference may yield valuable information regarding climatic variables such as temperature and precipitation and their influence on tree physiology over its life history.
We measured Hawaiian sandalwood (Santalum paniculatum) samples using both the transect and the Bézier curve methods. We then applied correlation statistics to the results for individual tree stems in order to compare and contrast the two techniques.
Results/Conclusions
We found a significantly higher correlation (avg. spearman rho > 0.4) between time series among samples measured with boundary traces than with transects (avg. spearman rho < 0.1). The higher correlation between annual tree rings with the Bézier curve method may represent a growth signal with significantly less noise.
The lobed sandalwood stems contain incomplete rings that appear to lack annual growth on some sides of the stem. We tentatively attribute this asymmetry to uneven water stress as the Hawaiian sandalwood could express more stem growth on the sides closest to water conductive roots. The asymmetric morphology increases with age in our sandalwood samples, which could be caused by an expanding root matrix becoming less localized. Low growth years display a radial distribution of growth across less circumference, in comparison with high growth years.
Our Python software program, which employs Green’s Theorem to obtain high fidelity areal estimates of unique growth shapes, can be extended to many other tree species. Using Bézier curves to trace boundaries, reduces error and noise, especially in noncircular stemmed species. This method can be used to provide high resolution measures of forest secondary growth for natural resource managers.