LAUSR

Abstract The common stand-level diameter distributions of old-growth forests (i.e., reverse-J or rotated sigmoid) resemble the managerial targets of selection forestry, seemingly making it a potential tool in continuous cover/close-to-nature forestry for the restoration of natural forest structures. Yet while the structures associated with selection forestry are well known, the most prevalent diameter distributions exhibited in natural tree neighborhoods in primeval forests are less well documented. If the structures actually observed in sub-stand patches of natural forest in various forest development phases resemble those targeted by selection forestry, then it may indeed provide useful silvicultural guidance for restoration in future old-growth. To assess this for European beech, the spatial tessellation based floating neighborhood sampling approach was used to delineate natural tree neighbors at the second-order scale (averaging 21 trees over 759 m2 patches) in a primeval European beech forest in the Carpathian Mountains, Ukraine. These neighborhoods were assigned to one of seven Dominant Biomass Strata development phases based on the size class contributing the greatest relative live-tree volume (early Understory, Lower Midstory, and Upper Midstory phases, middle Lower Overstory and Overstory phases, and late Upper Overstory and Emergent phases). Cluster analysis then identified the predominant diameter distribution types (DDTs) in the neighborhoods assigned to each phase. As is typical of primeval beech forest, few neighborhoods were assigned to the early phases (2.1%), proportionally more to the middle phases (25%) and most to the late Upper Overstory phase (50%), with the remainder assigned to the final Emergent (23%) phase. The earliest phases had DDTs with negative exponential (U, LM) or slightly bimodal (UM) forms, but subsequent phases added bimodal (LO), rotated sigmoid (LO-UO), and multimodal (O-E) DDTs. All phases had a very high (50–97%) proportion of trees in the smallest size class, such that no sub-stand patch DDT exhibited a unimodal form, nor did the stand-level distribution resemble a balanced uneven-age distribution. Active management to restore these natural structures may require modifications to selection forestry and include crop-tree (crown) thinning to promote a high (>50/ha) density of large-diameter trees to accumulate high (>500 m3 ha−1 of volume) stocking levels, as well as the creation of mini-gaps (~0.05 ha) to maintain but not necessarily immediately release Understory trees and some small gaps to establish Midstory trees to promote natural multimodal diameter distributions.