Level and spatial pattern of overstory retention impose trade-offs for regenerating and retained trees

Variable retention (VR) has been adopted globally as an alternative to more intensive forms of regeneration harvest. By retaining live trees within harvest units, VR seeks balance among the commodity, ecological, and aesthetic values of managed forests. Achieving these multiple, often competing objectives requires an understanding of how level and spatial pattern of retention shape the abundance, growth, and mortality of regenerating and retained trees. Using long-term (18-19 yr) data from a regional-scale VR experiment, we explore the individual and interactive effects of retention level (15% vs. 40% of initial basal area) and pattern (dispersed vs. aggregated) on the post-harvest dynamics of forests of differing structure and seral composition. Level and pattern of retention imposed trade-offs for the density and growth of regenerating trees (≥0.1 m tall, <5 cm dbh) and ingrowth (trees attaining 5 cm during the study). Greater retention led to greater density of late-seral regeneration, but lower density of early-seral ingrowth, and slower growth of late-seral ingrowth. Dispersed retention enhanced the density of early- and late-seral regeneration (compared to aggregated treatments), but reduced the growth of early-seral ingrowth. We also observed trade-offs for retained trees. Lower retention enhanced the growth of smaller trees (<25 cm dbh), particularly in dispersed settings, but reduced the survival of larger trees, which were more susceptible to windthrow. Greater retention reduced the growth, but enhanced the survival of smaller trees. Pattern imposed similar trade-offs, with dispersed retention enhancing growth, but reducing survival of small trees. Finally, level and pattern resulted in trade-offs for productivity of regenerating vs. retained-tree cohorts. Ingrowth productivity was greater at lower retention and in aggregated treatments; retained-tree productivity was greater at higher retention and in dispersed treatments. Our results provide a unique, long-term perspective on the sensitivity of tree regeneration, growth, and mortality to key structural elements of VR systems. Strong responses to level and pattern of retention produce trade-offs for different ecological or resource objectives. Balancing these objectives may require the combined use of aggregates, dispersed retention, and clearings, to mimic the spatial heterogeneity of habitats, physical structures, and resource conditions that are produced by natural disturbances.