Should reference conditions be drawn from a single 10 ha plot? Assessing representativeness in a 10,000 ha old-growth European beech forest

The management targets of modern forestry are often dictated by a desire to restore natural conditions, largely considered to be those found in contemporary reference sites. Beech reference sites are usually subjectively placed plots located in old‐growth reserves. Given the inherent variability in old‐growth, the validity of using a single such plot to guide restoration efforts is questionable. We therefore applied 3 methods to assess the representativeness of a 10 ha research plot in an old‐growth European beech (Fagus sylvatica L.) forest in Ukraine compared to an inventory of the entire 10,282 ha forest reserve. We compared the research plot to the 500‐m2 inventory plots using (1) permutation tests of structure metrics, (2) synthetic multivariate structural condition, and (3) functional condition via the proportion of area assigned to 8 forest development phases. Despite up to 82% distributional overlap for some metrics, both the averages and distributions of individual structural metrics (e.g. basal area, tree diameters) differed significantly between the RP and the inventory, as did the synthetic structural condition and the proportion of late optimal and decay phases. Extrapolations from this subjectively placed plot to the surrounding old‐growth matrix would overestimate several stereotypical “old‐growth” structures. These results support the need to draw on multiple reference sites and metrics and to select restoration target conditions that account for the variability associated with naturally dynamic ecosystems. A lack of absolute representativeness does not, however, necessarily preclude the generalizability of process‐based dynamics from old‐growth remnants.