Abstract

Plantation-forest planning must reconcile biological growth, silvicultural timing, economic return and the spatial arrangement of harvests, yet these components usually reside in disconnected software, leaving the assumptions that link a published growth trajectory to a recommended schedule uninspectable. We present Treem & uuml;n v2.0.0, an open-source Python framework that encodes published Chilean dry-aboveground-biomass trajectories for Pinus radiata and Eucalyptus globulus as 34 stratum-specific continuous surrogate functions, propagates them through event-based simulation of thinning, final harvest and regeneration across multiple rotations, and passes the resulting stand-policy coefficients to solver-independent mixed-integer harvest scheduling and multi-objective trade-off analysis. Objectives span net present value, standing-carbon stock-time and green-up conflicts, optimized singly or through weighted-sum and epsilon-constraint formulations, under optional even-flow and ending-stock constraints, with GIS-ready export. We report numerical consistency at two levels rarely examined together. Curve-level agreement with 526 tabulated values yields R2 = 0.9982 and RMSE = 6.38 Mg ha-1. A decision-level error-propagation analysis, re-optimizing a georeferenced 105-stand, 834.27 ha landscape (780 stand-policy alternatives, 30-year horizon) with a table-driven implementation of the same source trajectories, shows that agreement between growth curves does not guarantee identity of the selected plan: 33 stands changed their selected policy, yet cross-evaluated economic regret stayed below 0.9% and the two annual harvest schedules correlated at 0.95. Because both engines derive from the same tables, this tests internal numerical consistency rather than biological accuracy. CPLEX, CBC and HiGHS returned identical plans. In the single landscape instance examined in this study, retaining 50% of initial stock cost 1.01% of NPV while raising carbon stock-time by 3.57%; the carbon anchor raised it by 35.14% for a 4.78% NPV loss; and a conflict-focused plan cut green-up conflicts from 200 to 57 for 3.16% of NPV, whereas zero-conflict and even-flow requirements proved infeasible. These magnitudes are properties of the instance, not of the framework. Treem & uuml;n demonstrates, for this instance, how biological approximation, accounting scope, carbon retention and spatial coordination propagate into management recommendations.