Tropical moist forests (TMFs) are crucial for global ecosystem services, including climate regulation, carbon sequestration, and biodiversity. Despite their importance, TMFs are being lost at an alarming rate, with degradation from selective logging, fires, and edge effects significantly impacting their functionality. These degradation processes are occurring at a rate comparable to, and sometimes exceeding, deforestation. Edge effects, defined as changes in forest structure and function at forest edges due to fragmentation, further compound the issue. Degraded forests are also more susceptible to additional disturbances, such as climate extremes, reducing resilience and threatening long-term survival. Reducing forest degradation is key to reducing carbon emissions and enhancing carbon sequestration, yet significant uncertainties remain in quantifying its contribution to global carbon fluxes. Accurate estimates are vital for supporting REDD+ activities under the UNFCCC. While remote sensing has advanced, a pantropical assessment of forest degradation's impact on forest structure is still lacking. The deployment of the GEDI instrument offers a unique opportunity to address this gap. This study aims to assess the impact of human-induced degradation on global TMF structure and their recovery capacity. Specifically, it quantifies the extent of forest degradation in 2022, the impact of different disturbances on forest structure and their persistence, recovery rates after various degradation types, and the vulnerability of degraded forests to subsequent deforestation.

Previous research has highlighted the significant contribution of tropical forest degradation to carbon and biodiversity loss (Pearson et al., 2017; Barlow et al., 2016). Studies have shown that tropical forests are a net carbon source (Baccini et al., 2017), and that degradation processes like selective logging and fire are substantial contributors to these losses (Lapola et al., 2023). The impact of edge effects on forest structure and function has also been recognized, though the extent of these effects remains uncertain (Chaplin-Kramer et al., 2015; Silva Junior et al., 2020; Ordway & Asner, 2020). Existing studies often rely on optical remote sensing, which has limitations in detecting small-scale or low-intensity disturbances (Gao et al., 2020; Dalagnol et al., 2023). Previous assessments of intact forest landscapes have also varied in their methodologies and resulting estimates (Potapov et al., 2017). The study builds upon advancements in remote sensing capabilities, particularly the use of LiDAR data from GEDI, to provide a more comprehensive assessment of forest degradation and its consequences.

This study combines wall-to-wall data on forest degradation, deforestation, and regrowth dynamics from Landsat imagery (30m resolution) with spatially discontinuous estimates of forest canopy structure from GEDI. The analysis uses two key canopy height metrics from GEDI: RH98 (top of the canopy) and RH50 (height of median energy). Aboveground biomass density (AGBD) is also considered. The study analyzes data from intact forests (areas with no human activity detected in the past three decades and located at least 3 km from forest/non-forest edges), degraded forests (affected by logging, fire, or natural disturbances), forest edges (within 120m of forest/non-forest boundaries), and forest regrowth areas. The research assesses the spatial patterns of forest canopy heights across different regions, comparing intact and disturbed forests. The scale and magnitude of edge effects are evaluated using two indicators: the distance at which RH98 reaches 95% of the value in intact forests, and the distance at which differences in RH98 between edge and intact forests are no longer statistically significant (using ANOVA). The persistence of edge effects over time is assessed by analyzing changes in RH98, RH50, and AGBD over the period 1990-2022. The study also investigates the recovery dynamics of forests after different disturbances (selective logging, fire, edge effects) and the vulnerability of degraded forests to subsequent deforestation using statistical models relating canopy height, AGBD, distance to edges, and deforestation probability.

The study reveals that the magnitude of degradation effects on canopy structure is greater than previously reported. Selective logging resulted in a 15% decrease in forest height, while fire caused a 50% decrease. Recovery was slow, even after 20 years. Edge effects extended much farther than previously assumed, up to 1.5 km into forest interiors. These edge effects impacted 18% (approximately 206 Mha) of remaining TMFs, over double previous estimates. Degraded forests with more than 50% canopy loss showed significantly higher vulnerability to subsequent deforestation. The impacts of edge effects were particularly pronounced in the Americas (up to 7 km), followed by Africa (1.7 km) and Asia (1.7 km). The study found that logging impacts were highest in Asia (20% decrease in RH98, 50% decrease in AGBD), while fire impacts were most severe in Asia (60% decrease in RH98, 80% decrease in AGBD). Recovery rates after logging varied across continents (25% for Americas, 15% for Africa, 27% for Asia for RH50), with slower recovery for AGBD. Forest regrowth after deforestation showed a plateau at 60% of intact forest RH98 after 10-15 years, with slower AGBD recovery. Degraded forests were significantly more likely to experience subsequent deforestation, especially in the Americas.

The findings demonstrate the significant underestimation of tropical forest degradation in previous studies. The extensive reach of edge effects and the persistent impacts of logging and fire on forest structure highlight the need for a reassessment of carbon emissions and biodiversity loss associated with degradation. The study's integration of LiDAR and optical data provides a more comprehensive picture of forest degradation than relying on optical data alone. The strong predictive power of canopy height and distance to forest edge for deforestation risk emphasizes the importance of considering forest structure in conservation efforts. These findings underscore the limitations of relying solely on deforestation rates to understand forest loss and have significant implications for REDD+ and other conservation initiatives.

This study demonstrates that the combined use of GEDI and JRC-TMF datasets offers a novel approach for assessing pantropical forest degradation and recovery. The magnitude of degradation effects, particularly edge effects, is significantly greater than previously thought, impacting a much larger area than previously estimated. The persistent nature of degradation and the vulnerability of degraded forests to deforestation highlight the urgent need for enhanced conservation efforts. Future research should focus on understanding the complex interactions between different degradation drivers and developing more effective strategies for mitigating the impacts of degradation.

The study's reliance on GEDI data, which has spatial limitations, may influence the accuracy of the findings, particularly in areas with sparse GEDI coverage. The 30-year time frame might not capture the full long-term recovery dynamics of forests after various disturbances. The study primarily focuses on the structural aspects of degradation; future studies should incorporate other ecological factors to provide a more holistic understanding.