Ecological forestry rooted in natural disturbance-based silviculture (NDBS) aims to help maintain or facilitate recovery of ecosystem structure and function toward conditions found in unharvested stands. In the short term, leaving a portion of a stand’s merchantable trees intact at harvest should increase similarity in composition and function relative to stands following non-stand-replacing natural disturbances such as low-intensity fires, windthrow, or insect outbreaks, and thus maintain biodiversity and other ecosystem services.

Testing NDBS requires long-term data, while assessing the generality of responses to NDBS requires such data from multiple experiments across diverse stand and site conditions. To date, our understanding of forest responses to NDBS has been limited by a lack of opportunities to explore responses across time. In this study, we utilized data from three long-term experiments encompassing 21 combinations of harvest treatments by stand type plus unharvested reference stands. These experiments provide data to assess responses for 12 to 20 years post-treatment. We quantified the direction (increasing or decreasing), duration (persisting or transient), magnitude, and temporal dynamics of responses in stand-level tree growth, mortality, and regeneration abundances.

High-intensity treatments (≥40%) produced strong initial impacts, particularly increased tree mortality (although not statistically significant across all treatments), which in turn had prolonged effects on stand-level growth through reduced tree density. These treatments also promoted increased broadleaf regeneration, likely due to greater canopy opening and enhanced light availability. We observed inconsistent stand-level growth response to treatments, with some treatments showing initial decline while others eventually resulting in higher growth in harvested stands. In mixed-species stands, coniferous species showed growth benefits when coexisting with broadleaf species, possibly due to reduced intraspecific competition or improved access to resources mediated by broadleaf neighbors. This growth advantage was not observed for broadleaf species, indicating asymmetric benefits in mixed-species stands.

Response patterns and magnitude vary over time and are influenced by local factors such as stand type, overstory retention, species, and initial treatment impact. This variability highlights the potential for tailoring NDBS approaches locally to achieve desired outcomes and underscores the need to incorporate stand-specific conditions into ecological forestry practices.

This dataset contains tree growth, mortality, and regeneration data from four long-term forest experiments: EMEND, SAFE1, SAFE3, and Cochrane, all located within the boreal forest biome of Canada. For each experiment, three separate data files are uploaded, corresponding to growth, mortality, and regeneration.

Detailed information on the study sites, experimental design, data collection protocols, and analytical methods is described in the Methods section of Bose et al. (2026), Journal of Applied Ecology.