Unused forest roads fragment boreal landscapes but also offer opportunities for carbon recovery through reforestation. We develop a Mixed-Integer Linear Programming (MILP) model that plans long-term road restoration within a boreal forest context while enforcing realistic implementation limits (budget, workforce, logistics) and explicit spatial–temporal rules (cohesion, continuity, and priority spatial units). The formulation integrates road degradation, site productivity, transport distance, and treatment options, producing deterministic and reproducible schedules relevant for boreal forest management. Applied to a Forest Management Unit (FMU 02371) in Québec’s boreal forest, carbon-oriented scenarios attained the highest sequestration efficiency (14.1 MgCha−1) but dispersed efforts across the FMU. Cohesion-oriented scenarios concentrated restoration within a small subset of spatial units, reducing efficiency by about 40% (to 8.3 MgCha−1) while restoring a comparable proportion of area (∼350 ha). Temporal rules further shaped schedules: front-loaded plans accelerated early carbon recovery and formed more contiguous restored areas but required strong upfront capacity, whereas smoother allocations eased short-term demands at the cost of delayed gains. Once ecological filters (excluded segments, minimum reforestable area, and the degradation index) were fixed, sensitivity analyses showed that spatial and temporal rules governed most of the variation in model outputs—more than the budget envelope itself. To our knowledge, this is among the first spatially explicit, fully constrained MILP applications to forest-road restoration that jointly capture carbon returns, spatial cohesion, and operational feasibility at the FMU scale. The model is transferable and supports transparent policy design by transforming spatial rules into measurable levers with clear trade-offs.

Spruce budworm (Choristoneura fumiferana) outbreaks are a major disturbance in northeastern American forests. Monitoring, forecasting, and mitigating outbreak risks require information on how local population densities translate to defoliation. Given landscape-scale heterogeneity, forest structure, and local population dynamics, one expects the relationship between population densities and defoliation to vary spatially and temporally. We analyzed 17 years of larval density data from over 1000 locations in Québec, Canada, to investigate how larval densities and environmental context translate into observable defoliation. We found a positive latitudinal gradient and a positive effect of hardwood species on insect population growth rates. Further, we identified a 3-year cumulative effect of larval densities on defoliation, with a 2-year lag having the strongest influence. On average, a density of 15 larvae per branch corresponded to a 29% probability of defoliation. However, this probability varied widely (13%?52%) with the proportions of balsam fir and black spruce. Our study elucidates the relationship between SBW populations and defoliation, highlighting the importance of cumulative larval density effects and environmental context in defining defoliation risk.