Land tenure drives Brazil's deforestation rates across socio-environmental contexts

Tropical deforestation, largely driven by conversion of forestlands to agriculture and other human-dominated uses, degrades biodiversity and carbon stocks. Land-tenure rights determine who can use forestlands and how, making them central to sustainability challenges and deforestation dynamics. Tenure rights are contested, leading to shifts such as protection of public lands, recognition of indigenous and local community claims, and privatization through liberalization. The study defines a land-tenure regime as the combination of governance-related tenure factors over a parcel that is stable over time, including bundles of rights, implications for tenure security, and predisposition to policies or regulations. Different stakeholders claim that various tenure interventions—from privatization to recognition of communal rights—improve forest conservation, yet scientific evidence is mixed: theoretical predictions conflict, and empirical syntheses have been limited to meta-studies with heterogeneous cases or large-n single-scale analyses focused on few regimes. Systematic, large-n assessments comparing alternative tenure regimes across multiple spatial and temporal contexts are lacking. This study addresses that gap by testing and synthesizing the effects of tenure regimes on deforestation across Brazil's forestlands over three decades, informing policy on tenure interventions consistent with sustainable development goals.

The paper juxtaposes competing theoretical hypotheses on how tenure regimes and changes affect deforestation. For undesignated/untitled public lands, some theories suggest such status may inhibit deforestation by limiting access to credit and creating regulatory uncertainty, while others argue these areas become de facto open access, prone to exploitation and state incentives to clear. Privatization could either reduce deforestation via enhanced tenure security and enforceability of environmental obligations (e.g., Brazil’s Forest Code) or increase it by facilitating investment, credit access, and market transfers that favor agricultural expansion. Recognizing Indigenous and local community (IPLC) rights may reduce deforestation through collective rules and social enforcement, but may also fail due to collective action problems. Privatizing public lands could improve governance efficiency or, conversely, undermine conservation where private incentives do not internalize societal environmental values, indicating a need for state-led governance. Prior empirical work is fragmented: meta-studies across incomparable cases and single-scale analyses limit generalization. The paper builds on these debates and evidence gaps by providing systematic, cross-scale, quasi-experimental estimates.

Data and study scope: The authors analyze forest-to-agriculture conversion across Brazil from 1985 to 2018 using MapBiomas 30 m annual land-use/cover data. Land-tenure categories come from Imaflora’s Atlas (v.1812), covering 83.4% of Brazil and integrating 18 sources with expert conflict resolution. Tenure categories: (1) Undesignated/untitled public lands (publicly owned with poorly defined rights; includes undesignated lands and rural settlements, excluding Terra Legal program parcels), (2) Private lands (individual ownership from CAR and SIGEF, excluding Terra Legal properties), (3) Strict-protection protected areas, (4) Sustainable-use protected areas, (5) Indigenous lands, (6) Quilombola lands (privately owned by Afro-Brazilian descendant communities). Communal lands were analyzed separately due to heterogeneity and low sample sizes and excluded from main results. Covariates for matching and modeling include market accessibility (travel time to nearest city), agricultural suitability (slope, elevation), human population density, and parcel area, chosen to reflect deforestation drivers, policy relevance, and siting biases. Study design: The authors estimate longer-term differences in forest-to-agriculture conversion associated with tenure regimes, not short-term impacts of specific titling events. They conduct analyses across 49 spatiotemporal contexts: Brazil-wide and for each biome (Amazônia, Caatinga, Cerrado, Mata Atlântica, Pampa, Pantanal), over the full period (1985–2018) and subperiods (1985–1990, 1991–1995, 1996–1999, 2000–2004, 2005–2012, 2013–2018). Counterfactuals: For each tenure regime, effects are compared to two counterfactuals where relevant: (i) undesignated/untitled public lands and (ii) private lands. Quasi-experimental matching: Coarsened exact matching (CEM) is applied separately for each regime comparison and spatiotemporal extent using the R ‘cem’ package. Variables are coarsened into bins: travel time (0–2, 2–6, 6–12, >12–24, >24 h), parcel area (0–2, 2–5, 5–15, 15–50, >50–100, >100–500, >500–1,000, >1,000–5,000, >5,000–10,000, >10,000–50,000, 50,000–100,000, >100,000–500,000, 500,000–1,000,000 ha), with automated coarsening for elevation, slope, and population change. Unmatched observations are dropped. Post-matching balance is assessed via L1 imbalance. Generalizability: Recognizing matched subsets may not represent the full population, the authors assess generalizability using Tipton’s T-index comparing matched samples to the population of parcels at each scale. Absolute standardized mean differences (ASMDs) diagnose covariate differences. To improve population representativeness, inverse-odds weights are generated using lasso-estimated probabilities of being matched; higher weights are assigned to observations more representative of the population. Effect estimation: Weighted generalized linear models (GLMs) with binomial errors and logit link estimate average treatment effects, including uncoarsened covariates, federal-state fixed effects, and municipality-clustered standard errors to account for spatial autocorrelation. Effects are synthesized across scales, with robustness checks including balance weighting and filtering of protected areas established after period starts or with unknown dates. Additional steps manage potential tenure timing issues by excluding certain programs (e.g., Terra Legal) and performing robustness tests in regions/periods with known tenure changes.

• Extent and loci of deforestation: 17.4% of Brazil’s originally forested 30 m pixels were converted to agriculture between 1985 and 2018. Most deforestation occurred on private lands (78%) and undesignated/untitled public lands (19%). Undesignated/untitled public lands cover ~963,357 km² in Brazil.
• Effect of poorly defined tenure: Across Brazil (1985–2018), undesignated/untitled tenure increased deforestation relative to other regimes, with estimated effects ranging approximately from −12.4% to +23.2% depending on the comparator regime. Across 196 regime-by-context tests (48 context combinations plus Brazil-wide), undesignated/untitled tenure led to higher deforestation in 141 cases, lower in 6, and non-significant in 49; effects were generally highly generalizable (high T-index values).
• Private tenure vs undesignated/untitled: Privatizing undesignated/untitled lands generally reduced deforestation, particularly where private lands faced stringent environmental obligations (e.g., the Amazon Forest Code, soy/beef moratoria). However, private tenure was the least reliable and among the least effective alternatives compared with other well-defined regimes: it had the highest risk among alternatives of increasing deforestation over the undesignated/untitled counterfactual (8.8% of scales; 8.4% balance-weighted) and was least likely to achieve large reductions (2.9%; 2.2% balance-weighted). It underperformed relative to all alternatives except quilombola regimes.
• Conservation regimes: Strict-protection and sustainable-use protected areas reliably reduced deforestation. Significant reductions occurred in 88.2% (strict) and 76.5% (sustainable use) of narrower context tests. Sustainable-use regimes were about five times more likely to outperform than to underperform other regimes (e.g., largest/smallest reductions in 41.2%/8.8% of cases; similar when balance-weighted and time-filtered), while strict-protection’s relative performance was more mixed but still strongly beneficial overall.
• IPLC regimes: Brazil-wide, indigenous and quilombola regimes reduced deforestation relative to both counterfactuals, but results across contexts were inconsistent: significant reductions appeared in only ~58.3–59.8% of comparisons. Indigenous tenure reduced deforestation vis-à-vis undesignated/untitled in 76.5–82.8% of cases, but vs private in only 59.4–70.4% and with low generalizability (T-index ≥0.5 in only 17% of those cases, largely in the Cerrado), reflecting siting biases (e.g., distance from cities, elevation).
• Amazonia exception: In Amazonia (hosting 90.5% of remaining undesignated/untitled forest), private tenure shifted from deforestation-increasing (1985–1990) to one of the strongest deforestation-decreasing regimes from the early 2000s onward, likely due to stricter private-land environmental obligations and moratoria, while public reserve regimes’ advantages over undesignated/untitled were weaker—possibly due to limited on-the-ground enforcement capacity.
• Public vs private ownership more broadly: Replacing public regimes other than undesignated/untitled (i.e., protected or indigenous lands) with private tenure would likely increase deforestation in most contexts: 66.7% of country-wide, 77.8% of biome-specific long-term, and 75% of biome-specific short-term tests (mean increases 1.6–28.2%). In Amazonia, privatizing protected or indigenous lands would likely increase deforestation in most periods, especially after 2000.
• Policy implication: Extending stringent private-actor-focused environmental policies (akin to the Forest Code’s high legal reserve in Amazonia) to other remote biomes dominated by private tenure (e.g., Cerrado: 80.4% private; Pantanal: 92.8%) could substantially reduce deforestation.

The study addresses how alternative land-tenure regimes influence deforestation and whether effects generalize across socio-environmental contexts. Consistent evidence indicates that poorly defined tenure on public lands (undesignated/untitled) drives higher agriculture-driven deforestation, supporting interventions that replace ambiguity with clearer governance. While privatization of such lands can reduce deforestation—particularly when strong, enforceable environmental obligations apply—private tenure generally performs worse than conservation-oriented public regimes across diverse contexts. Strict-protection and sustainable-use protected areas most reliably lower deforestation, with sustainable-use often achieving the largest reductions, aligning with theories that well-enforced, conservation-designed bundles of rights restrict forest conversion. Effects of IPLC regimes (indigenous and quilombola) are beneficial on average but context-dependent and less generalizable, reflecting heterogeneity in local governance capacity, siting, and enforcement contexts. Amazonia presents a notable divergence: stringent private-land environmental requirements and commodity moratoria can render private tenure effective in reducing deforestation where state enforcement on public reserves is challenged by remoteness and capacity limits. However, wholesale privatization of protected or indigenous public lands would likely increase deforestation, indicating that privatization addresses specific mechanisms relevant to undesignated/untitled lands but not those managed for conservation or indigenous stewardship. Overall, aligning tenure interventions with governance capacity and coupling private rights with robust, enforceable environmental obligations emerges as key to achieving deforestation reductions.

This study provides a systematic, cross-scale, quasi-experimental assessment of how tenure regimes shape deforestation across Brazil. The most consistent, transferable finding is that converting undesignated/untitled public lands with poorly defined rights to any other tenure regime substantially reduces deforestation. Conservation-focused regimes (strict protection, sustainable use) most reliably reduce forest loss, with sustainable-use protected areas often delivering the largest reductions. Private tenure is generally the least reliable option for reducing deforestation relative to other well-defined regimes, but can be effective under strong environmental obligations, as seen in Amazonia. Privatizing protected areas or indigenous lands would likely increase deforestation in most contexts. For regimes with context-dependent effects, notably IPLC tenure, policy design should be informed by detailed local conditions. Future research should extend these cross-scale analyses to other tropical countries, improve temporal resolution of tenure-change data, and enhance transparency and availability of parcel-level tenure information—especially for private and IPLC lands—to enable stronger causal inference and transferability assessments.

• Tenure timing and stability: Many tenure datasets lack precise formalization dates; analyses assume basic tenancy type remained stable over study periods. The authors mitigate this by excluding programs initiated mid-period (e.g., Terra Legal), conducting robustness checks, and comparing results across regions/periods with known changes.
• Matching and siting bias: Despite CEM improving balance, residual imbalance and siting biases remain (e.g., matched samples tended to be at lower elevations, farther from cities, and larger parcels). Generalizability was evaluated with Tipton’s T-index, and inverse-odds weighting was used, but not all comparisons achieved high generalizability (notably indigenous vs private outside Cerrado).
• Omitted variables and enforcement heterogeneity: Unobserved factors (e.g., local enforcement intensity, commodity market shocks, infrastructure changes) may influence results despite controls for key covariates and state fixed effects.
• Protected-area establishment dates: Some protected areas’ establishment dates are unknown or post-date period starts; sensitivity analyses filtered later-established areas with qualitatively robust results.
• External validity beyond Brazil: While multiple Brazilian contexts were analyzed, transferability to other tropical countries depends on institutional and policy similarities and data availability.