Accurate canopy height information is essential for quantifying forest carbon, monitoring restoration and degradation, and assessing habitat structure, yet high-fidelity measurements from airborne laser scanning (ALS) remain unevenly available globally. Here we present Canopy Height Map version 2 (CHMv2), a global, meter-resolution canopy height map derived from high-resolution optical satellite imagery using a depth estimation model built on DINOv3 and trained against ALS canopy height models. Compared to existing products, CHMv2 substantially improves accuracy, reduces bias in tall forests, and better preserves fine-scale structure such as canopy edges and gaps. These gains are enabled by a large expansion of geographically diverse training data, automated data curation and registration, and a loss formulation and data sampling strategy tailored to canopy height distributions. We validate CHMv2 against independent ALS test sets and against tens of millions of GEDI and ICESat-2 observations, demonstrating consistent performance across major forest biomes.

Operational forest monitoring is essential for the effective implementation of national and international initiatives to reduce deforestation and forest degradation. Such monitoring systems are especially important within Brazilian humid tropical primary forests, where they support enforcement of national policies to prevent deforestation. Several satellite-based forest disturbance monitoring systems are operating in Brazil, including MapBiomas Alerta (MBA), Tree Cover Loss supported by Global Forest Watch (TCL), and Tropical Moist Forest developed by the European Commission’s Joint Research Centre (TMF). These systems differ in their forest disturbance definitions, input satellite data, and change detection methodologies. As a result, their annual estimates of primary forest disturbance are not fully consistent, which complicates the implementation of forest conservation policy and introduces uncertainties and potential bias in greenhouse gas emissions accounting. In this study, we followed good practice recommendations to evaluate the performance of these monitoring products to detect humid tropical primary forest disturbances in 2023 and 2024 using a probability reference sample. We compared the map-based and sample-based disturbance areas for each product and estimated their sensitivity and precision for primary forest disturbance detection. Our analysis showed that the MBA product mapped 64%–65% of the sample-based reference deforested area while maintaining the highest mapping precision among the three systems. The TCL product detected the highest percentage of the reference deforested area (82%–94%), while maintaining high precision for mapping high-severity, stand-replacement disturbances. TMF outperformed other products in capturing low-severity disturbances. The detailed comparative evaluation of the three operational monitoring systems highlights their respective strengths and limitations and explains the differences in forest disturbance reporting. Our results provide guidance for researchers, policymakers, and practitioners in selecting the most appropriate products for specific applications.

Quantifying the drivers of tree cover loss globally provides a synoptic understanding of pressures on the world's forests. Existing information about tree cover loss drivers relies on maps of coarse spatial and thematic resolution. In this study, we quantified the global extent of tree cover loss in 2018 at the scale of individual disturbances and provided a comprehensive accounting of land use outcomes using a global probability sample of 600 5 × 5-kilometer blocks mapped with high-resolution (3- to 10-meter) satellite data. Out of 277 thousand square kilometers of estimated global tree cover loss, nearly a third (29.0%) was due to long-term conversion of tree cover to other land uses, including conversion of natural tree cover to pasture (15.0%), cropland (6.4%), and nontimber tree plantations (3.8%).

Ontario’s forests are changing rapidly under the pressure of industrial logging, coupled with the effects of climate change, which is facilitating increased wildfires. The reduction of mature and old-growth forests negatively affects ecosystem functions, including wildlife habitat suitability and long-term carbon storage. Here, we present an operational mapping tool to identify and monitor the extent of high-carbon primary forests (HCPF) to support their protection and track progress toward the Kunming-Montréal Global Biodiversity Framework’s Target 3. We define HCPF as mature and old-growth, naturally regenerated forests that have not experienced tree canopy disturbance over the past 50 years and are located within unfragmented landscape patches with a minimum area of 1,000 ha. By integrating existing maps, satellite-based products, and detailed manual image interpretation, we mapped 28.4 Mha of HCPF in Ontario for 2020. HCPF comprised 80.7% of all forests outside industrial timber concessions, whereas within private and industrial forests, their percentage was much lower (29.7%). They represented less than 2% of all forests within the Mixedwood Plains ecozone of southern Ontario. The HCPF vegetation carbon stock density is higher than that of fragmented and recently disturbed forests by 7% within the entire province, and by 18% within the region of industrial forestry. Ontario’s unfragmented landscapes contain more than 72% of the total soil organic carbon stock of the province, illustrating their importance for long-term carbon storage. Unfragmented landscapes are essential for the survival of Ontario’s woodland caribou populations; however, their percentage within most caribou population ranges was below the management threshold of 65% and continued to decline. From 2021 to 2025, Ontario’s HCPF area decreased by 3.5% (0.2 Mha/yr) due to wildfires and logging. Within the region of industrial forestry, the proportion of 2021–2025 logging within HCPF exceeded that in non-HCPF forests, and the annual logging area within HCPF increased by 55%. This acceleration of logging and associated road construction illustrates that HCPF, as opposed to secondary forests, continue to be the major source of timber in Ontario. The HCPF method provides the tools and data required to accurately assess and monitor mature and old-growth primary forests, and to develop effective conservation strategies and policies at the provincial and national levels. As demonstrated for Ontario, the HCPF method can be applied to all provinces to facilitate the protection of some of the last remaining primary forests in Canada.

Abstract Remote forest landscapes provide critical references for understanding ecosystem functions (EFs) under low anthropogenic pressure, yet their capacity to sustain multiple EFs simultaneously remains poorly understood. We assessed landscape multifunctionality in western Patagonia by integrating satellite indicators, field data, and spatial modeling. Six EFs (carbon storage, nutrient availability, water regulation, erosion control, habitat quality, and ecological connectivity) were mapped, and their spatial relationships and hotspot distribution within and outside protected areas (PAs) were analyzed. Old-growth and secondary forests showed the highest functional performance. Strong synergies (ρ ≥ 0.6) between carbon storage and nutrient availability covered >50% of the landscape, whereas strong trade-offs (ρ ≤ –0.6) were spatially limited ( < 6%). Notably, 78% of multifunctionality hotspots occurred outside PAs, indicating that high-functional-value areas extend beyond formal conservation boundaries. These findings reveal spatial mismatches between multifunctionality and protection status and provide a replicable framework for integrating multifunctionality into conservation planning under global change.

High-resolution crop maps over large spatial extents are fundamental to many agricultural applications; however, generating high-quality crop maps consistently across space and time remains a challenge. In this study, we improved a workflow for crop mapping and developed an openly available, annual, 10 m spatial resolution maize and soybean map product over the Contiguous United States (CONUS) from 2019 to 2022 (available at https://glad.umd.edu/dataset/mapping-crops-10-m-resolution-united-states, last access: 26 December 2025). We obtained all available Sentinel-2 surface reflectance data between May and October for every year, applied quality assurance, corrected the bidirectional reflectance distribution function (BRDF) effects, and generated 10 d analysis ready data (ARD) composites. We then derived multi-temporal metrics from the 10 d ARD as training features for the national-scale wall-to-wall mapping. We implemented a stratified, two-stage cluster sampling, and then conducted annual field surveys and collected ground data. Utilizing the training data with Sentinel-2 multi-temporal metrics and topographic factors, we trained random forest models generalized for annual maize and soybean classification separately. Validated using field data from the two-stage cluster sample, our annual maps achieved consistent overall accuracies (OA) greater than 95 % with standard errors of less than 1 %. User's accuracies (UAs) and producer's accuracies (PAs) for maize were higher than 91 % and 84 % across the years, and UAs and PAs for soybean were greater than 88 % and 82 %, respectively. To illustrate the substantial improvement of the 10 m map over existing datasets, e.g., the 30 m Cropland Data Layer (CDL), we aggregated the 10 m maps to 30 m spatial resolution and quantified the number of mixed pixels that can be reduced by improving the mapping from 30 to 10 m. The counties with the most maize and soybean production in Iowa, Illinois and Nebraska had the lowest reduction in mixed pixels, ranging from 1 % to 7 %, whereas southern counties had a higher reduction in mixed pixels. Overall, the median percentages of mixed maize and soybean pixels reduction across all counties were 8 % and 9 %, respectively. With more Sentinel-2-like data available from continuous observations and incoming satellite missions, we anticipate that 10 m crop maps will greatly benefit long-term monitoring for agricultural practices from the field to global scales. The dataset is also available at 10.6084/m9.figshare.28934993.v2 (Li et al., 2025).

Irrigation plays a critical role in global food production and climate adaptation and exercises profound influence over humanity's water use. Yet despite its critical importance, there is a persistent lack of understanding of fine-scale irrigation patterns across the planet, knowledge which is essential for informing global food security and sustainability targets. Utilizing either statistical downscaling or remote sensing approaches, existing global irrigation datasets are constrained by coarse spatial resolutions, a lack of timeliness, or varying robustness and reliability. To address this gap, here we integrate multi-source Earth observation and environmental datasets and use machine learning to develop a medium-resolution (30 m) global irrigated area dataset for the 2023/24 growing season. Within existing cropland extent, we leverage a newly compiled set of georeferenced irrigated (N=230,683) and non-irrigated (N=153,194) ground-truth points and integrate seasonal vegetation metrics derived from Landsat 8/9 imagery with agroecological-zone information and hydroclimatic and topographic variables. We subsequently develop and evaluate two machine-learning frameworks, a continental Agro-Ecological Zone (AEZ) tile-based framework and a continental-scale framework, and apply the best-performing approach for each continent. Evaluation using held-out test samples yielded a global accuracy of 80.5 ± 2.1%. The resulting maps were also validated against independent global and national irrigation datasets and statistics, demonstrating broad agreement in the spatial distribution of irrigated areas. This approach is robust and reliable because it is built on a harmonized global ground-truth database, incorporates multiple predictors, and is rigorously validated using independent datasets. All code, ground-truth, and data products are freely and publicly available and can serve as a robust, scale-neutral, and fully reproducible framework for fine-resolution irrigation mapping. These advances provide the critical and long-needed foundation for near-real-time monitoring and early warning systems, and fine-scale land and water resource management.

Abstract Agricultural expansion continues to reshape landscapes across Africa, yet how patterns of cropland change vary across farming systems of different scales remains poorly understood. Here, we combined annual high-resolution field boundary mapping from Planet imagery with an empirically derived field-to-farm size relationship to examine cropland dynamics, farm-scale patterns, and their environmental and socioeconomic correlates across Zambia from 2018 to 2024. Cropland expansion was highly spatially concentrated, with local expansion rates reaching up to 170 ha yr -1 within 0.05◦ (≈5.5 km) grid cells, and occurred primarily along the margins of established agricultural regions. Areas dominated by medium-scale farms exhibited substantially faster cropland growth than areas dominated by smallholder farms. Similarly, when field size was considered as a continuous variable, cells characterized by larger median field sizes showed markedly higher expansion rates, exceeding 35 ha yr -1 in some areas, whereas expansion rates in areas dominated by smaller fields were generally below 5 ha yr -1 . Meanwhile, stable croplands showed little evidence of widespread field consolidation, with approximately 84% of grid cells exhibiting field-size trends within ±0.05 ha yr -1 , suggesting limited evidence for farm-scale expansion through widespread field consolidation within existing croplands. Environ mental and socioeconomic analyses further indicated that land tenure, represented by the proportion of land under state tenure, was the factor most consistently associated with the prevalence of medium scale farms, whereas climate stability was more strongly associated with total cropland area than with farm-size composition. Together, these findings suggest that recent agricultural growth in Zambia has been closely associated with the increasing prevalence of medium-scale farming, while widespread field consolidation within existing croplands appears limited. More broadly, the study demonstrates how high-resolution Earth observation can provide spatially explicit evidence of agricultural transformation and emerging farm structures across rapidly changing African farming systems.

Food security worldwide is increasingly threatened by population growth, shifting diets, geopolitical conflicts, and climate change impacts. Annual operational cropland monitoring is required to support the United Nations Zero Hunger Sustainable Development Goal. Landsat satellite data provide a foundation for such global, independent, high-cadence monitoring at 30-m spatial resolution suitable for agricultural policy and management interventions, policy responses, and market adjustments. Here, we used Landsat Analysis Ready Data developed by the Global Land Analysis and Discovery Lab (GLAD-ARD) and machine learning to map global cropland extent annually from 2015 to 2024. We showed that the global cropland area increased by more than 6% over the past decade. By combining sample-based cropland area estimates from our research and the earlier analysis (2003-2019), we estimate that the global cropland area has expanded by nearly 14% since 2003. Between 2015 and 2024, Africa accounted for the largest regional increase (+24.5 Mha). At the national scale, Brazil experienced the largest gain (+16.5 Mha) and Morocco the largest loss (-0.38 Mha). A third (33.3%) of all new cropland was established through natural vegetation clearing or irrigation expansion within natural drylands. The overall accuracies of the 2015 and 2024 cropland maps were 97.8% (Standard Error 0.3%) and 97.3% (SE 0.4%), respectively. Despite cropland expansion, population growth has outpaced cropland gains; between 2015 and 2024, per-capita cropland area declined from 0.166 to 0.161 ha per person. Our data illustrate the combined effects of changes in land use priorities, climate, water supply, international trade, and armed conflicts on global cropland extent dynamics during the last decade.

Distinguishing forest types---primary, naturally regenerating, planted, and plantation forests---from agricultural tree crops and other land uses is essential for carbon accounting, biodiversity assessment, conservation planning, and supply-chain regulation. However, no existing global dataset resolves this typology at high spatial resolution. We present the Forest Typology (ForTy) v1 dataset, a global 10-meter resolution map for 2020 that classifies all land into six categories aligned with FAO and EU Deforestation Regulation (EUDR) definitions: Primary Forest, Naturally Regenerating Forest, Planted Forest, Plantation Forest, Tree Crops and Agroforestry, and Other Land. A cascaded deep learning pipeline, trained on 1.7 million globally distributed samples, generates per-class probability maps from geospatial satellite embeddings by combining weakly supervised learning with active learning. Independent validation against 8,190 stratified random sites, each labeled by two experts, yields an overall accuracy of 90.2% for the six-class scheme, 94.8% for natural forest classification, and 95.5% for forest/non-forest classification.

Arable land has been expanding since the advent of agriculture, and now sustains over eight billion people by supplying food, fiber, and fuel. However, the rapid expansion of built-up areas is increasingly displacing arable land across diverse regions, raising concerns about the long-term sustainability of global food systems. Addressing these concerns requires a fundamental understanding of global arable-land dynamics, which remains largely unknown. Using our newly developed high-quality arable-land product—integrated with global settlement layers, Köppen–Geiger climate classifications, and historical land-use reconstructions—we show that competition between arable land and urban systems for favorable climate fundamentally shapes global arable-land dynamics. Globally, urbanization in the Northern Hemisphere is displacing arable land from climatically favorable regions toward more marginal tropical and arid environments. In North America, Europe, and Asia, early co-development of arable land and urbanization has transitioned into intensified urban encroachment, pushing cultivation into adjacent arid zones. By contrast, low-latitude tropical regions continue to expand both systems at the expense of natural ecosystems, driven in part by agricultural investment from Northern Hemisphere countries. This framework not only offers a new perspective for synthesizing existing regional arable-land dynamics but is also closely linked to ongoing dryland greening, tropical deforestation, and groundwater depletion in drylands. Reconciling agricultural production with rapid urban growth is therefore essential to safeguarding sustainable land resources, forest health, water security, and global food security under accelerating demographic pressures.

Global forests provide key ecosystem services, from climate regulation to biodiversity habitat, but are under increasing pressure from the combined impacts of climate and land use change. Here, we show that forest disturbance due to fire is growing globally, with the most dramatic increases in intact forest landscapes, highlighting an existential threat to remaining high biomass, high biodiversity forests. The global annual area of forest disturbance due to fire for 2023 and 2024 was highest since the beginning of monitoring in 2001. Compared to 2002-2022 average annual forest disturbance due to fire, the 2023-2024 average was 2.2 times higher globally and 3 times higher in the Tropics. More than ¼ of all 2024 forest disturbance from fire occurred in tropical forests. We found a statistically significant increasing trend of forest disturbance due to fire from 2002 to 2024 in all climate domains except Subtropical. High forest, low deforestation tropical countries were not exempt, with Guyana and the Republic of the Congo experiencing record forest disturbance due to fire. Our results agree with recently estimated increases in global forest fire emissions and active fire detections. The unprecedented scale of fires in the world's most remote forests is a potential harbinger of ecosystem tipping points. Protecting these remaining unfragmented high conservation value forests from this threat poses a daunting and as yet undeveloped policy and capacity challenge.

Recent advancements in data storage and computing, particularly cloud-based processing, enable mapping global land cover and change relatively quickly and easily. Multiple versions of a map could be produced within a matter of days with various adjustments of selected parameters of machine learning models. Sample-based validation is then required to establish correspondence between these map prototypes and the real world, thus turning them from algorithm data outputs into sources of information with quantified errors. Implementing global probability sampling of geographic data for the purposes of area estimation and map accuracy assessment presents multiple challenges, primarily linked to the way these geographic data are stored (coordinate systems and projections) and the objectives of the specific project. Here we summarize various approaches to global sampling aimed at assessing accuracy of global land cover and change maps and producing unbiased estimators of area along with the standard errors associated with these estimates for the target land cover classes. We provide a unified set of estimators that accommodate a variety of sampling designs by explicitly accounting for the area of each sample unit, as well as code and technical details necessary to implement the presented methods. While we do not compare relative precision of the presented sampling design options, our aim is to help practitioners select an appropriate sampling design and estimators for their specific data format and project objectives, and to facilitate the correct implementation and increased reproducibility of global sampling methods within the land cover mapping community.

Accurate mapping of rubber plantations is essential to understanding where deforestation due to rubber production occurs. Wang et al.1 produced 10-meter resolution maps of mature rubber with a reported high overall classification accuracy (OA = 0.95 ± 0.02), indicating that more than 4 million hectares of forests have been converted to rubber plantations in Southeast Asia and parts of China since 1993. However, we have serious concerns about the accuracy of these maps, as our analysis indicates that the area estimates by Wang et al.1 are significantly inflated. This observation underscores the need for caution in using these high-resolution maps for specific applications, such as assessing deforestation linked to rubber consumption under the EU regulation on deforestation-free products.

Accurate and high-resolution land cover (LC) information is vital for addressing contemporary environmental challenges. With the advancement of satellite data acquisition, cloud-based processing, and deep learning technology, high-resolution Global Land Cover (GLC) map production has become increasingly feasible. With a growing number of available GLC maps, a comprehensive evaluation and comparison is necessary to assess their accuracy and suitability for diverse uses. This particularly applies to maps lacking statistically robust accuracy assessment or sufficient reported detail on the validation procedures. This study conducts a comparative independent validation of recent 10 m GLC maps, namely ESRI Land Use/Land Cover (LULC), ESA WorldCover, and Google and World Resources Institute (WRI)'s Dynamic World, examining their spatial detail representation and thematic accuracy at global, continental, and national (for 47 larger countries) levels. Since high-resolution map validation is impacted by reference data uncertainty owing to geolocation and labelling errors, five validation approaches dealing with reference data uncertainty were evaluated. Of the considered approaches, validation using the sample label supplemented by majority label within the neighborhood is found to produce more reasonable accuracy estimates compared to the overly optimistic approach of using any label within the neighborhood and the overly pessimistic approach of direct comparison between the map and reference labels. Overall global accuracies of the maps range between 73.4% +/- 0.7% (95% confidence interval) to 83.8% +/- 0.4% with WorldCover having the highest accuracy followed by Dynamic World and ESRI LULC. The quality of the maps varies across different LC classes, continents, and countries. The maps' spatial detail representation was assessed at various homogeneity levels within a 3 x 3 kernel. Although considered as high-resolution maps, this study reveals that ESRI LULC and Dynamic World have less spatial detail than WorldCover. All maps have lower accuracies in heterogenous landscapes and in some countries such as Mozambique, Tanzania, Nigeria, and Spain. To select the most suitable product, users should consider both the map's accuracy over the area of interest and the spatial detail appropriate for their application. For future high-resolution GLC mapping, producers are encouraged to adopt standardized LC class definitions to ensure comparability across maps. Additionally, the spatial detail and accuracy of GLC maps in heterogeneous landscapes and over some countries are the key features that should be improved in future versions of the maps. Independent validation efforts at regional and national levels, as well as for LC changes, should be strengthened to enhance the utility of GLC maps at these scales and for long-term monitoring.

Aboveground biomass density (AGBD) estimates from Earth Observation (EO) can be presented with the consistency standards mandated by United Nations Framework Convention on Climate Change (UNFCCC). This article delivers AGBD estimates, in the format of Intergovernmental Panel on Climate Change (IPCC) Tier 1 values for natural forests, sourced from National Aeronautics and Space Administration's (NASA's) Global Ecosystem Dynamics Investigation (GEDI) and Ice, Cloud and land Elevation Satellite (ICESat-2), and European Space Agency's (ESA's) Climate Change Initiative (CCI). It also provides the underlying classification used by the IPCC as geospatial layers, delineating global forests by ecozones, continents and status (primary, young (≤20 years) and old secondary (>20 years)). The approaches leverage complementary strengths of various EO-derived datasets that are compiled in an open-science framework through the Multi-mission Algorithm and Analysis Platform (MAAP). This transparency and flexibility enables the adoption of any new incoming datasets in the framework in the future. The EO-based AGBD estimates are expected to be an independent contribution to the IPCC Emission Factors Database in support of UNFCCC processes, and the forest classification expected to support the generation of other policy-relevant datasets while reflecting ongoing shifts in global forests with climate change.

The national-level land cover database is essential to sustainable landscape management, environmental protection, and food security. In Afghanistan, the existing national-level land cover data from 1972, 1993, and 2010 relied on satellite data from diverse sensors adopted three different land cover classification systems. This inconsistent land cover map across the various years leads to the challenge of assessing landscape changes that are crucial for management efforts. To address this challenge, a 19-year national-level land cover dataset from 2000 to 2018 was developed for the first time to aid policy development, settlement planning, and the monitoring of forests and agriculture across time. In the development of the 19 year span of land cover data products, a state-of-the-art remote sensing approach, employing a harmonized classification scheme was implemented through the utilization of Google Earth Engine (GEE). Publicly accessible Landsat imagery and additional geospatial covariates were integrated to produce an annual land cover database for Afghanistan. The generated dataset bridges historical data gaps and facilitates robust land cover change information. The annual land cover database is now accessible through https://rds.icimod.org/. This repository ensures that the annual land cover data is readily available to all users interested in comprehending the dynamic land cover changes happening in Afghanistan.

The changing climate directly affects spatial and temporal patterns of snow and ice cover globally and in the Hindu-Kush Himalayan (HKH) region. In the HKH, around 3.3 billion people across 11 countries depend on water originating from mountain glaciers and snowfields, and melting snow cover has a direct impact on their livelihood and well-being. Various studies have shown that the snow and ice cover in the HKH is declining at an alarming rate but have been limited in geographic, spatial and temporal scales. Here, we employed the Global Land Analysis and Discovery analysis ready Landsat time-series data (GLAD ARD) to map changes in perennial snow and ice between 2001 and 2021 at five-year epochs using decision tree ensemble models. These maps were used to create a stratified sampling design for reference data collection to estimate area and map accuracy. All five-year epoch maps have user's accuracies above 90% and producer's accuracies above 91%. Our sample data analysis showed that one-eighth of the extent of perennial snow and ice in the HKH, totalling 15,770 km2 (CI +/- 3195 km2), disappeared over the last two decades with 105,935 km2 (CI +/- 3396 km2) remaining in 2021. From map-based estimates, the largest decline of perennial snow and ice cover among the mountain ranges was found in the Himalayas with an estimated reduction of 5741 km2. Among HKH countries, China had the largest area of perennial snow and ice reduction of 10,654 km2, Nepal had the highest net perennial snow and ice reduction rate (31%). Among the river basins, the maximum net loss of perennial snow and ice cover between 2001 and 2021 was in the Indus (24.8%), followed by Brahmaputra (18.3%), and Tarim (15.7%). Our results confirm wide-spread loss of perennial snow and ice across the HKH region and provide both regionally consistent maps with derived area estimates at landform, national and basin-levels and methodology to continue monitoring snow and ice melt.

Indonesia has experienced rapid primary forest loss, second only to Brazil in modern history. We examined the fates of Indonesian deforested areas, immediately after clearing and over time, to quantify deforestation drivers in Indonesia. Using time-series satellite data, we tracked degradation and clearing events in intact and degraded natural forests from 1991 to 2020, as well as land use trajectories after forest loss. While an estimated 7.8 Mha (SE = 0.4) of forest cleared during this period had been planted with oil palms by 2020, another 8.8 Mha (SE = 0.4) remained unused. Of the 28.4 Mha (SE = 0.7) deforested, over half were either initially left idle or experienced crop failure before a land use could be detected, and 44% remained unused for 5 y or more. A majority (54%) of these areas were cleared mechanically (not by escaped fires), and in cases where idle lands were eventually converted to productive uses, oil palm plantations were by far the most common outcome. The apparent deliberate creation of idle deforested land in Indonesia and subsequent conversion of idle areas to oil palm plantations indicates that speculation and land banking for palm oil substantially contribute to forest loss, although failed plantations could also contribute to this dynamic. We also found that in Sumatra, few lowland forests remained, suggesting that a lack of remaining forest appropriate for palm oil production, together with an extensive area of banked deforested land, may partially explain slowing forest loss in Indonesia in recent years.

The Southeast Asian rubber boom beginning in the early 20000s shaped a myriad of socioeconomic and environmental consequences, including deforestation, ecosystem impacts, shifts in community livelihoods, and altered local access to land and resources. Although there has been significant research assessing rubber production in this region, there has been less focus on economic inequality and polarization outcomes in rubber producing areas. This analysis explores the extent to which rubber production growth was associated with changes in rural economic inequality and polarization from 2007/08 to 2012/13, using Lao PDR as a case study. We also investigate the implications of these changes for voluntary sustainability programs focused on rubber production. We achieve this through a synthesis of land use change and economic data. First, we estimate rubber plantation extent based on Landsat time series data and supervised classification. We combine this with household expenditures data from the Laos Expenditure and Consumption Survey from 2007/08 to 2012/13, conducting Gini decomposition and Duclos Esteban Ray Index calculations to explore economic inequality and polarization in rubber and non-rubber producing areas. Our results indicate that rubber areas experience greater inequality and polarization compared to non-rubber areas. The Northern, Central, and Southern regions experience different economic inequality and polarization outcomes - inequality-enhancing effects appear to be greatest in the South, where large-scale concessions dominate rubber production. We assess the implications of our findings for voluntary rubber sustainability programs, arguing that these programs should address systemic drivers of inequality and polarization, including dispossession from land and forest resources, insufficient worker protections, livelihood vulnerability, and barriers for smallholders. Overall, our results underscore the importance of strong regulation, multi-stakeholder action, and environmental and social performance criteria in rubber production.& COPY; 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).