Global potential for natural regeneration in deforested tropical regions

The study addresses how to achieve large-scale, cost-effective forest restoration to meet climate and biodiversity targets. Tropical forests are pivotal due to their biodiversity, ecosystem services, rapid growth, and extensive degradation. Global platforms (for example, the Bonn Challenge and the Global Biodiversity Framework) set ambitious restoration targets by 2030. However, large-scale tree planting is expensive and may not always restore native biodiversity effectively, whereas natural regeneration can be cheaper and often more successful under suitable conditions. Planners lack tools to predict where natural regeneration is likely to occur and how quickly benefits will accrue. Prior restoration potential maps often used coarse data and assumptions rather than observed regrowth, potentially underestimating natural regeneration. This study develops a high-resolution, pantropical model to identify where natural regeneration is most likely, informing restoration planning and policy.

Previous approaches to mapping restoration potential relied on coarse-scale datasets, expert opinion, and assumptions about potential forest cover rather than empirically observed regrowth, leading to underestimation of natural regeneration potential. Tree planting, though common, is costly (US$105–25,830 per ha) compared with natural regeneration or assisted natural regeneration (US$12–3,880 per ha) in tropical/subtropical contexts and can yield lower biodiversity outcomes under similar conditions. Earlier global estimates suggested biophysical feasibility for natural forest regrowth across 349–678 Mha, larger than the pan-tropical scope here. A pantropical remote-sensing analysis previously identified 31.6 ± 11.9 Mha of natural regrowth (2000–2012 persisting to 2016) in 4.78 million patches. Field studies highlight the importance of proximity to intact forests for seed sources, microclimate moderation, and seed dispersers. Recent work mapped global carbon accumulation potential from natural forest regrowth, informing sequestration estimates used here.

The authors modeled the potential for natural regeneration across tropical forest biomes (to +25° latitude) at 30 m resolution using empirically observed natural regrowth (2000–2016) as the response variable. Building on a pantropical remote-sensing dataset that detected natural regrowth patches (>0.45 ha, vegetation >5 m tall), they sampled 5.4 Mha of regrowth to train machine-learning classifiers that distinguish where regeneration did or did not occur. Predictor variables encompassed local to landscape scales and included: distance to nearest tree cover; local forest density within 1 km2; land cover; 12 soil metrics for top 30 cm; 19 bioclimatic variables reduced to 5 principal components; slope; net primary productivity; mean monthly burned area (2001–2017); distance to water; population density; GDP; Human Development Index; road density; distance to urban areas; and protected area status. Spatial predictions were generated for circa 2015 and projected to 2030 under the assumption that 2000–2016 conditions persist. The model output is a continuous 0–1 potential for natural regeneration per pixel, interpreted as probability; an expected (weighted) area per pixel was computed by multiplying potential by pixel area, with a sensitivity analysis using a >50% threshold. Model selection compared versions with biophysical+socioeconomic predictors versus biophysical-only; due to similar accuracies and greater stability/resolution of biophysical data, final spatial predictions used biophysical variables only. Validation used 4.87 million independent, stratified random points, yielding 87.9% accuracy (out-of-bag accuracy 87.8%). Biome-level accuracies were ~87.8–87.9% (moist, dry broadleaf, and coniferous tropical/subtropical forests). Partial dependence indicated positive association with local forest density (within 1 km) and negative association with distance to existing forest. A random sample of 62,493 30×30 m cells showed that 98.1% of cells with potential >0.5 were within 300 m of a forest edge. Soil organic carbon content was a positive predictor, consistent with higher values in or near forests and lower values with intensive land use. Above-ground carbon sequestration potential from predicted regenerating areas was derived using recent spatially explicit accumulation rates for natural regrowth.

- Estimated potential area for tropical natural forest regeneration: 215 Mha (95% CI: 214.78–215.22 Mha) by 2030. Regional breakdown: Neotropics 98 Mha (CI 97.80–98.20), Indomalayan tropics 90 Mha (CI 89.82–90.18), Afrotropics 25.5 Mha (CI 25.47–25.53). - Five countries account for 52% of global potential: Brazil 20.3%, Indonesia 13.6%, China 7.2%, Mexico 5.6%, Colombia 5.2%. Twenty-nine additional countries each have >1 Mha potential. - Model performance: validation accuracy 87.9% (4.87 million points); out-of-bag 87.8%; lowest accuracies in parts of Southeast Asia; biome-level accuracies ~87.8–87.9%. - Key predictors: higher local forest density (1 km radius) increases potential; greater distance from existing forest decreases potential. In a random sample, 98.1% of cells with potential >0.5 lay within 300 m of a forest edge. Soil organic carbon positively associated with potential. - Above-ground carbon sequestration potential over 30 years if regeneration occurs across 215 Mha: 23.4 Gt C (range 21.1–25.7). Regional contributions: Neotropics 11.1 Gt (CI 10.0–12.2), Indomalayan 5.42 Gt (CI 4.87–5.96), Afrotropics 3.1 Gt (CI 2.83–3.37). Including belowground biomass (+22–28%) implies 28.6–30.0 Gt C total, up to ~1 Gt C yr−1, increasing current global tropical/subtropical forest carbon sequestration by ~14.3% per year. - Potential mitigation context: regeneration could mitigate pantropical forest carbon losses by ~90.5% per year or constitute ~26.9% of total potential carbon across global deforested areas, under assumptions noted.

The study closes a key information gap by providing a high-resolution, empirically informed map of where tropical forests can naturally regenerate, enabling cost-effective restoration planning. Findings confirm that proximity to existing forests and higher soil organic carbon strongly favor regeneration, emphasizing landscape context and the importance of conserving remaining forests as seed sources and for microclimate regulation. The sizable 215 Mha potential, concentrated in a subset of countries, highlights opportunities for targeted national strategies (for example, incentives such as payments for ecosystem services and improved compensation mechanisms). The estimated carbon gains underscore natural regeneration’s role in achieving climate mitigation targets, complementing protection of intact forests. The authors note positive feedbacks: as forests regrow, they expand seed sources and ameliorate local conditions, potentially increasing regeneration potential over time. The maps can be integrated into multiobjective planning for biodiversity, carbon, and cost, and help identify areas where additionality and permanence for offsets may be greatest. However, success depends on local contexts, tenure, policy, and market conditions; assisted natural regeneration may be needed to overcome establishment barriers, and protection is required to ensure persistence against reclearance risks. Certification and carbon market methodologies must evolve to capture additionality and permanence of naturally regenerating forests. Community-based approaches, livelihood diversification, and supportive national policies can help scale implementation while addressing equity considerations.

Natural forest regeneration can deliver large-scale, cost-effective restoration with substantial climate mitigation, biodiversity, water regulation, erosion control, and resilience co-benefits. The pantropical, 30 m resolution maps identify where natural regeneration is most likely to succeed now, helping to set ambitious yet feasible restoration targets and to prioritize areas where planting costs can be avoided or minimized. Because regenerating forests can catalyze further regeneration via positive feedbacks, there is a strong rationale to support these processes immediately. Recognizing and leveraging the significant regeneration capacity of tropical forests is essential to stabilize climate, reduce biodiversity loss, and advance regenerative land use alongside protection of intact forests and reduced deforestation.

- Potential underestimation: (1) positive feedbacks from new regrowth expanding future regeneration potential are not captured; (2) analysis focuses on areas supporting dense natural forests and excludes non- or sparsely forested ecosystems (for example, savannas). - Climate uncertainty: future drought, fire risk, precipitation changes, and CO2 fertilization could alter successional trajectories, impeding or facilitating regeneration. - Socioeconomic and feasibility not modeled: final maps use biophysical drivers for predictions; actual implementation depends on tenure, policies, market dynamics, and local priorities. - Persistence risks and leakage: young forests are vulnerable to reclearance; restoration can cause leakage if pressures shift elsewhere; realized carbon benefits may be lower than maximum potential. - Carbon accounting scope: above-ground biomass only; belowground biomass not mapped (would add ~22–28%). Initial non-forest biomass replacement not considered but likely small on cropland/pasture. - Data resolution: although outputs are 30 m, some predictors are coarser, influencing local predictions; accuracy varies spatially (lower in parts of Southeast Asia). - Methodological constraints: equal accuracy for models with and without socioeconomic variables led to biophysical-only mapping; local conditions and long-term land-use history can strongly affect outcomes and should guide site-level decisions.