Fig 1.
Study area (left) and a detailed view of the Koda stand with a year of the last coppice harvest marked (right part).
Fig 2.
Moderate (blue) or major releases (red columns) detected using the GA method displayed in 5-year intervals.
Years of coppicing recorded in archives are marked by black arrows. The green line indicates sample depth.
Table 1.
Summary of cored trees and release events (GA method).
Table 2.
Agreement between historical forest archives and releases detected by GA method expressed by sensitivity.
Percentages of trees showing release ± 5 years from the year of coppicing are shown.
Table 3.
Accuracy of coppice detection from tree release (GA method) compared to historical coppice records (± 5 years).
Positive predictive power (PPP) express probability of correct detection and is the inverse of the commission error accounting for false detections, whereas sensitivity is the inverse to the omission error accounting for undetected coppice events.
Fig 3.
Species composition within a 10 m radius of the cored standards, divided into single- and multi-stemmed (a), and by study site (b).
Sites were dominated by hornbeam and oak except the sandstone-based Mramor III site, where hornbeam was missing and pine formed 3% of the neighbours.
Fig 4.
Competition from neighbouring trees within a 10 m distance from cored standard trees expressed by distance-weighted basal area (a), and number of neighbouring stems (b), mean growth rate (c), and mean growth change identified by GA method.
KO = Koda, KB = Kobyla, M1–3 = Mramor I–III. Significance tests are summarized in S4 Table.
Table 4.
Models testing effect of growth parameters and neighboring competition on the tree response to the release (number of releases and mean growth change), and the accuracy of coppice detection from tree-ring analysis (expressed by sensitivity and PPP).
Significance was tested in generalized linear models using software R version 3.1.3. Best predictor variables were selected by stepwise model selection by Akaike’s Information Criterion using MASS package in R.