Diagnosing Inconsistency in Compartment-Level Growing Stock Volume Estimates Across Inventory Methodological Transitions: Evidence from a Greek Uneven-Aged Forest

Sustainable forest management depends on reliable compartment-level growing stock volume (GSV) estimates, which inform the prescribed cut. At the University Forest of Pertouli (Greece), four management-plan inventories (1989–2028) span a methodological transition from full-callipering in 1988 and 1997 to a systematic-sample-plot inventory with empirical compartment-level upscaling in 2008 and 2018. We distinguish methodological discontinuity (a change in method) from methodological inconsistency (variation in empirical sample-plot weighting between sample-based inventories) and develop a four-hypothesis diagnostic framework. Hypothesis 1 tests temporal change via a linear mixed model on 149 compartments. Hypotheses 2–4 use non-parametric bootstrap on 231 common sample plots under three nested resampling schemes, which bound a documented plot-center relocation. The 2008 compartment-level estimator underestimated the sample-plot benchmark by 19% (deviation B2008 = +54.9 m3 ha−1, 95% confidence interval (CI) +39.6 to +70.8); in 2018 the two converged (B2018 = −2.2 m3 ha−1). The implied decadal changes differ by difference-in-differences DID = +57.0 m3 ha−1 (robust across three schemes), with ~96% attributable to baseline-year underestimation; the 2008 underestimation was confirmed in all nine orographic units. The framework provides a retrospective consistency assessment for inventory time series crossing methodological transitions. The 19% deviation documented here is invisible without it; its identification restores estimation reliability to compartment-level estimates.