Tree Mortality in Boreal Primary Forests is Sensitive to Climate and Stand Structure: High-Resolution Evidence Across a Gradient of Protected Landscapes

Climate-driven increases in tree mortality represent a major uncertainty in projections of boreal forest carbon balance and ecosystem resilience. Detecting climate sensitivity in mortality patterns is particularly challenging in managed landscapes, where harvesting and silvicultural legacies obscure underlying ecological signals. Here, we analyse a large dataset of individual standing dead trees collected in 2023–2024 across 11 protected primary boreal forest areas along a north–south gradient in Finland, using aerial image-based detection methods. This setting provides an opportunity to assess tree mortality drivers under near-natural conditions.

The analysis covered 7,495 forest stands spanning 69,791 ha, with a total of 304,458 standing dead trees detected in the upper canopy layer (typically DBH ≥20–25 cm), paired with stand-level data on forest structure, development stage, species dominance, and habitat characteristics. Tree mortality was modelled using a two-part hurdle framework that separates the occurrence of mortality from its intensity (dead trees per hectare). Occurrence was estimated using ridge-regularized logistic regression, while mortality intensity was modelled with negative binomial generalized linear models to account for strong overdispersion. Stand area was included via offsets, and model robustness was evaluated using leave-one-area-out sensitivity analyses.

Across all model formulations and spatial subsets, stand structural attributes—most notably total standing volume—emerged as the strongest and most stable predictors of mortality intensity. A one-standard-deviation increase in log-transformed volume was associated with a 55–85% increase in expected mortality, indicating that biomass-rich stands exhibit heightened vulnerability to mortality processes. Field-measured total deadwood volume, where available, further amplified mortality signals, consistent with cumulative effects of past disturbance or chronic stress.

Forest development stage showed systematic but secondary effects. Relative to old-growth stands, younger and mid-successional development classes consistently exhibited lower mortality intensity, while differences among mature and old stands were modest once structural variation was accounted for. This suggests that apparent age-related patterns are largely mediated through biomass accumulation and stand structure rather than chronological stand age alone. Dominant tree species had comparatively weak effects: spruce-dominated and mixed species stands tended to show slightly lower mortality than pine-dominated stands, while birch-dominated stands exhibited reduced mortality in some model formulations. Overall, species effects were less stable than structural predictors.

Categorical habitat descriptors, including vegetation type and Natura 2000 habitat class, exhibited limited explanatory power after accounting for stand structure, development stage, and species dominance. Together, these results indicate that the climate sensitivity of tree mortality in boreal primary forests is primarily mediated by structural factors rather than habitat type. High-biomass stands may amplify the impact of climate-related stressors—such as drought, thermal extremes, or biotic agents—by increasing competition and physiological demand via increased evaporative area and metabolic costs.

Our findings provide a quantitative baseline for detecting climate-induced changes in boreal forest mortality and highlight the importance of structurally explicit approaches for assessing ecosystem vulnerability under ongoing climate change.