Validation of a classification protocol: meeting the prospect requirement and ensuring distinctiveness when assigning forest development phases

Aims: Evaluation and modification of a classification protocol for delineating development phases of the beech forest life cycle (Tabaku, Struktur von Buchen-Urwaldern in Albanien im Vergleich mit deutschen Buchen-Naturwaldreservaten und-Wirtschaftswaldern, 2000). Location: The Uholka-Shyrokyi Luh primeval beech forest, Carpathian Biosphere Reserve in the Ukrainian CarpathianMountains. Methods: The classification protocol was evaluated by examining the requirements for each type based on the literature and empirically assessing the suitability of the typology resulting from applying it to non-overlapping 156.25 m(2) grid cells superimposed on a 10-ha stem-mapped old-growth European beech stand. Modifications were proposed and the modified protocol applied to the 10-ha old-growth stand and evaluated using permutation tests to assess the distinctiveness of primary attributes among types in the resulting typology. Results: We found a consistent bias in the classification protocol resulting from a failure of the prospect requirement: the compound classification rule rendered it virtually impossible for grid cells to be assigned to the Late Optimum phase. The proposed modified protocol alleviates this problem and was found to result in distinct types with statistically significantly different mean values for many primary attributes. Conclusions: Interpretations of a purported 'signature texture' for old-growth beech forests dependent on the absence of the Late Optimum phase when using the original classification protocol require reconsideration. The proposed modified protocol would be less susceptible to this bias, but caution is still advised, given the inherent constraints of the classification protocol at small scales.