Hyperspectral species maps and LiDAR‐based structured population models show future forest fire frequency may compromise forest resilience

Abstract Forest disturbances are accelerating biodiversity loss and altering tree productivity worldwide. Post‐disturbance recovery time, a component of resilience, is critical for identifying vulnerable areas and targeting conservation but varies with environmental conditions. Monitoring recovery at scale requires tracking tree dynamics, yet traditional ground‐based approaches are resource‐intensive. We present a pipeline to parameterize integral projection models (IPMs) using LiDAR data and hyperspectral‐based species maps to assess post‐fire recovery across large, forested areas. Focusing on the fire‐adapted Picea mariana , we model passage times to reproductive heights and life expectancy under different fire regimes as indicators of recovery time. To do this, we combined hyperspectral‐based species maps and LiDAR‐based crown heights to track individual tree survival and growth at the Caribou‐Poker Creek Research Watershed (BONA) from 2017 to 2023. We incorporated fire history, aspect, slope, elevation and surrounding canopy height into our models and found partial support for their expected effects on survival and growth. Once accounting for topography and competition, we estimated passage times to reproductive maturity (11–22 years). Life expectancy in the absence of fire is shortest on North‐facing slopes with recent fire (581 years). Sensitivity analyses highlight fire history and aspect as key modulators of population resilience, with elevation exerting strong influence on life expectancy across all conditions. Our results demonstrate that remotely sensed IPMs can effectively quantify forest recovery at scale, revealing that in some contexts, stands of P. mariana may not recover between fire disturbances. We discuss the implications of these findings for advancing modelling of resilience and highlight both the challenges and opportunities of using LiDAR and hyperspectral data to build demographic models for forecasting forest dynamics.