Agreement or Chance: How Exact are tree markings in forest management?

Abstract: Tree marking activities are usually assumed to follow textbooks and management plan guidelines. However, initial experiments starting in the 1990s have shown that there is much room for personal interpretation by field and operational staff. This study presents an analysis of tree selection variability among people selecting trees in marteloscopes as part of silvicultural training activities. The analysis has been done for two different thinning types, i.e. low and crown thinning, as well as for the selection of frame trees in twelve different forests in the United Kingdom. Two different methods have been used for assessing the agreement among participants, i.e. Fleiss’ kappa and a method based on the test statistic of the χ2 goodness-of-fit test. The Pearson correlation coefficient was calculated to identify possible relationships between the aforementioned two agreement characteristics and the structural parameters of the forests. In general the agreement between test persons was low for all types of tree selection. Overall, the agreement was higher for low thinning exercises followed by the selection of frame trees and the marked frame-tree competitors in crown thinnings. Both agreement characteristics indicated that test persons tend to agree more when selecting trees for low thinnings and when selecting frame trees. There were no consistent patterns that suggest that the structure of the forest influences the level of agreement between test persons.