Fire in a Central Amazon forest: Lingering top canopy loss and initial understory regrowth revealed by repeated LiDAR

Understory fires in wet tropical forests impact trees across the size spectrum. However, the size specificity of these impacts—and whether they cascade through canopy structural changes to alter regeneration dynamics—remains poorly understood. We hypothesized that the demographic effects of understory fire, typically assessed through field surveys, produce distinct signals in three-dimensional canopy structure detectable from LiDAR point clouds, with significant consequences for forest regeneration. We analyzed leaf area vertical distributions and top-of-canopy heights across 980.6 ha of forests in the Central Amazon, surveyed by airborne LiDAR 2.5 and 3.5 years after fire. Overall, compared to unburned forests in both years, burned forests displayed lower leaf area density in the understory, lower-, and mid-canopy; with lower and more heterogeneous top-of-canopy heights. Burned forest top-of-canopy height profiles were bimodal, while unburned forest showed unimodal distributions. During the one-year interval between LiDAR surveys, total leaf area in fire-affected forests remained stable in magnitude but was reorganized, showing significant temporal shifts: increased leaf area density in the lower canopy (<13 m), reduced mid-canopy density (15–25 m), decreased upper canopy height, and loss of carbon density. Canopy damage was further characterized by expanding canopy gaps and new clusters of height loss associated with branch and whole-tree mortality. Together, our results reveal initial incipient recovery occurring simultaneously with delayed large-tree mortality—patterns undetected in prior field studies due to sample scale limitations—highlighting pervasive fire impacts that may increase rainforest sensitivity to climate change. • Burned forests had double the mortality of large trees in nearby unburned forests. • Marked dynamics were detected in 1-year interval driven by tree and branch mortality. • Including vertical displacement, LAWH outperformed LAI in tracking leaf area changes. • Bimodal top-of-canopy height profile was only observed in burned forests. • Multitemporal LiDAR complements field plots for linked canopy and demographic dynamics.