How accurate are our current BAI estimation methods? An evaluation of how eccentricity and coring affect BAI estimation accuracy

Link To Share This Poster: https://cdmcd.co/nQ7mG6
Live Discussion Link: https://cdmcd.co/5mqGgD

Christina A. Rossi and Julie Messier, Biology, University of Waterloo, Waterloo, ON, Canada, Andrew Trant, School of Environment, Resources and Sustainability, University of Waterloo, Waterloo, ON, Canada

Background/Question/Methods
In forestry, it is important to understand tree growth in terms of forest productivity and health, especially to estimate lumber yield. To measure tree growth, basal area increment (BAI) is used and typically calculated from annual radii increments measured on tree cores. Current methods for estimating BAI from radii assume that trees are perfect circles with the pith at the center. This method leads to BAI estimation error because often trees deviate from a perfect circle (i.e. eccentricity). This can be due to pith offset from the center (pith eccentricity) or stem out-of-roundness (stem eccentricity). This study aims to address three research questions: (Q1) How much does tree eccentricity (both pith and stem eccentricity) affect BAI estimation accuracy, (Q2) Which method of BAI estimation best accounts for eccentricity (produces the least error), and (Q3) How does the number of cores used affect estimation accuracy in more or less eccentric trees? To address these questions we used one hundred tree sapling cross-sections from 25 different tree species from Mont Saint-Hilaire, Quebec, Canada. Four BAI estimation methods are explored: arithmetic, geometric, quadratic, and ellipse area, to determine which method produces the least estimation error. Additionally, we use one to four cores for each BAI estimation.
Results/Conclusions
The preliminary data analysis with a full model revealed that accounting for out-of-roundness, in addition to using only one core during BAI estimations, significantly impacts the absolute percent error but only accounts for 13 percent of the variance. Further, the full model also reveals that pith offset does not impact the absolute percent error. Additionally, the four estimation methods preformed similarly, with the exception of when using four cores to estimate BAI. With four cores, the geometric method leads to significantly higher absolute percent error than the other methods. Based on our current results, we suggest that when estimating BAI from tree cores it is best to use two to four cores with any estimation method (with the exception of the geometric method with four cores). If only one core is available and/or the stem is exhibiting eccentricity, then the estimation error can be accounted for (~13 percent). Our study looks to see how BAI estimation can be improved by considering the interplay among BAI estimation method, tree eccentricity, and number of cores. We hope to inform those in forestry on the best way to minimize error while calculating BAI to improve growth and lumber yield estimates.