Linking species composition shifts from satellite time series to disturbance regimes and Lidar-derived structural and mortality indicators in boreal mixedwoods

Understanding species composition shifts in boreal mixedwoods forests is essential for anticipating forest succession pathways under changing disturbance regimes. Species composition transitions in boreal forests reflect complex successional processes influenced by interactions between disturbance regimes, structural dynamics, and species traits. In this study, we integrated satellite-derived annual species composition data with airborne laser scanning (ALS) structural metrics, spatially explicit mortality estimates and disturbance history to investigate composition transitions across ∼288,000 ha of the Romeo Mallette Forest, Ontario. We focused on mid to late successional stages, identifying 27 species composition transitions and modeling their likelihood using extreme gradient boosting (XGBoost). From 2005 to 2018, 5% of the analyzed stands (∼42,000 ha) predominantly transitioned from hardwood to coniferous or mixed compositions. Transition probabilities were strongly associated with ALS-derived gap metrics, mortality rates, and cumulative years of spruce budworm and Forest Tent Caterpillar defoliation, while traditional site factors had limited predictive value. Notably, the number of years affected by spruce budworm defoliation significantly increased the likelihood of transition in stands dominated by more susceptible species. The results advance our understanding of mid-late succession pathways and support the integration of remote sensing time series into forest monitoring frameworks, improving inventory accuracy, and guiding adaptive management under evolving disturbance regimes.