Threat Assessment for Protected Areas in Bolivia

Analysis of Forest Cover Loss and Burned Areas (2000-2024)

Bolivia
Protected Areas
Forest Loss
Burned Areas
Threat Assessment
Author
Affiliation

Johannes Schielein

MAPME Initiative

Published

January 12, 2026

Modified

February 4, 2026

Abstract

This report analyzes threat levels in protected areas (PAs) of the Bolivian Amazon Basin, examining forest cover loss and burned area dynamics from 2000 to 2024. The analysis utilizes Global Forest Watch data and MODIS burned area products to identify high-priority areas for conservation investment and support project preparation activities.

Introduction

This analysis supports peer-discussion on the threat level of protected areas (PAs) in the Bolivian Amazon Basin. The goal is to assist KfW and its partners in the project preparation phase and to derive information about the current portfolio’s relevance for protecting the most threatened areas.

The analysis examines a set of predefined PAs, ranks them according to their threat level, and compares them to past KfW support. We utilize publicly available data from:

  • World Database of Protected Areas (WDPA/IUCN) - For protected area boundaries and metadata
  • Global Forest Watch (Hansen et al., 2013) - For forest cover loss detection
  • MODIS MCD64A1 - For burned area mapping
NoteAnalysis Period Update

This report has been updated to cover the period 2000-2024, extending the original analysis by 4 years. Additionally, fire analysis now uses burned area in hectares instead of active fire counts, providing a more accurate measure of fire impact.

Analyzed Protected Areas

This analysis examines a set of protected areas within the Bolivian Amazon and Gran Chaco regions. Of these areas, five have received financial support from KfW through current or past conservation projects. The remaining areas represent additional protected zones of conservation interest that could benefit from future investment.

Forest Cover Loss (2000-2024)

Forest cover loss is quantified using Global Forest Watch data (Hansen et al., 2013). Forest cover loss is defined as “a stand-replacement disturbance, or a change from a forest to non-forest state.” Loss can result from human activities or natural factors such as droughts, forest fires, or hurricanes.

Key Indicators

We focus on two primary outcome indicators:

  1. Total forest cover loss: Measures the cumulative loss area in hectares. This identifies PAs with the highest primary forest cover loss, useful for targeting areas where we might achieve the largest impact in reducing emissions from deforestation and forest degradation.

  2. Relative forest cover loss: Measures the percentage of primary forest cover lost compared to the total primary forest area in 2000. PAs with high relative losses may have lost large parts of their functional forest habitat, making them priorities for protecting floral biodiversity, fauna, and dependent human communities.

Interactive Map

The map below depicts relative and absolute forest cover loss in the selected areas. Circle size indicates total loss (larger = more loss), and color indicates relative loss (red = higher percentage lost). From a threat perspective, large red circles indicate especially relevant areas for conservation.

Absolute Forest Cover Loss

The lollipop plot below ranks protected areas by total forest cover loss (2000-2024). Areas supported by KfW projects are highlighted.

Burned Area Analysis (2015-2023)

Comparing Burned Area and Forest Cover Loss Data

This analysis uses burned area in hectares from the MODIS MCD64A1 product (Giglio et al., 2018). It is important to understand the methodological differences between burned area and forest cover loss data:

  • Burned Area (MCD64A1): Cumulative sum of all fire-affected areas. If the same area burns multiple times over the analysis period, each burn event is counted separately. This means burned area totals can exceed the total land area of a protected area.

  • Forest Cover Loss (Global Forest Watch): Each forest pixel is only counted once when it transitions from forest to non-forest. Even if an area burns multiple times or experiences repeated disturbances, the loss is recorded only at the initial deforestation event.

Implication for interpretation: Burned area values are typically higher than forest cover loss values because they represent cumulative fire exposure, not permanent land cover change. High burned area with low forest loss may indicate recurrent fires in grasslands or savannas, while matching high values in both metrics suggest fire-driven deforestation.

GFW Detection Limitations in Dry Forest Ecosystems

A critical limitation of the Global Forest Watch data concerns its reduced detection accuracy in dry forest ecosystems, such as those found in Kaa-Iya del Gran Chaco National Park and other Chaco-type environments. This limitation stems from several technical factors:

  1. Spectral Confusion: The “brown” signature of deciduous trees during the dry season is often indistinguishable from bare soil in global satellite baselines, leading to underdetection of forest cover.

  2. Deciduous Phenology: Dry forests naturally lose their leaves seasonally, which can be misinterpreted as permanent forest loss or, conversely, cause actual losses to be missed when trees are leafless.

  3. Algorithm Bias: Research indicates that GFW algorithms are fundamentally optimized for humid tropical biomes and frequently misclassify the sparse, clumped canopies characteristic of the Gran Chaco as shrubland or non-forest.

  4. Masking Practices: GFW often masks out non-humid regions from its primary forest datasets entirely, resulting in systematic underrepresentation of dry forest dynamics.

Interpretation Guidance: Low or zero forest cover loss values for dry forest protected areas (such as Kaa-Iya del Gran Chaco) should be interpreted as a technical artifact rather than evidence of low threat. Nevertheless, the relatively low burned area values observed in Kaa-Iya del Gran Chaco suggest that this area may indeed have been spared from the major fire events that affected other regions in the last decade.

Fire is a significant threat to protected areas, whether from natural causes (e.g., El Niño-related droughts) or human activities (land clearing, agricultural burning). The MODIS MCD64A1 burned area product provides monthly burned area detection at 500m resolution, allowing us to quantify the total hectares affected by fire within each protected area over the analysis period (2015-2023).

Interactive Map

The map shows burned area totals for analyzed protected areas. Larger circles indicate greater cumulative burned area.

Total Burned Area by Protected Area

Interpretation & Discussion

Most Threatened Protected Areas

Based on the combined analysis of forest cover loss and burned area data, the following protected areas emerge as the most threatened and should be prioritized for conservation intervention.

NoteThreat Index Methodology

The Composite Threat Index used in this analysis follows established approaches in conservation prioritization (Margules and Pressey 2000; Wilson et al. 2005). The index combines three normalized indicators:

  1. Absolute Forest Loss (ha): Total hectares of forest lost since 2001, capturing the magnitude of habitat destruction.
  2. Relative Forest Loss (%): Percentage of original forest cover lost, indicating the intensity of pressure relative to the area’s forest resources.
  3. Total Burned Area (ha): Cumulative fire exposure since 2015, reflecting fire-related threats.

Calculation Method: Each indicator is normalized to a 0-1 scale using min-max normalization: ( = ). The final threat score is the arithmetic mean of all three normalized indicators, giving equal weight to each threat dimension.

Interpretation:

  • Scores close to 1.0 indicate areas with the highest combined threats across all dimensions
  • Scores close to 0.0 indicate areas with relatively lower threat levels
  • The index is relative to the analyzed portfolio and should not be compared across different study areas
WarningCritical Findings: Most Threatened Areas

Based on the composite threat index and supporting evidence from documented fire events:

1. Otuquis National Park - Ranks among the highest-threat areas due to catastrophic fire events during the 2019-2020 Bolivian fire crisis. According to reports from the Fundación Amigos de la Naturaleza (Fundación Amigos de la Naturaleza (FAN) 2019), this Pantanal-Chaco transition zone experienced unprecedented fire activity that burned over 5 million hectares across eastern Bolivia in 2019 alone.

2. San Matías Integrated Management Natural Area - High burned area and forest loss values. This area, located near the Brazilian border in the Chiquitano dry forest region, was severely affected by the 2019 fires, which were linked to agricultural expansion and land clearing practices.

3. Areas in the Chiquitano Dry Forest Region - The Chiquitano forest, the world’s largest and best-preserved tropical dry forest, experienced devastating fires in 2019 that destroyed approximately 1.4 million hectares according to satellite analysis. Several protected areas in this region show elevated threat scores.

4. Kaa-Iya del Gran Chaco National Park - While showing high absolute values due to its large size (3.4 million ha), this area appears to have been relatively spared from the worst fire events. Note that GFW detection limitations in dry forest ecosystems (see methodology note above) may underestimate actual forest dynamics in this area.

Note on Data Interpretation: The rankings are based on available satellite data and should be interpreted alongside ground-truth information and local ecological knowledge. Some areas may show low threat scores due to detection limitations rather than actual low threat levels.

Does Forest Cover Loss Coincide with High Burned Areas?

The analysis reveals a moderate positive correlation (r = 0.67) between absolute forest cover loss and burned area. This relationship suggests that fire and deforestation are interconnected threats in many protected areas. Key observations:

  • High fire, high loss areas (upper right quadrant): These areas experience fire-driven deforestation, where fire is likely a direct cause or facilitator of forest loss. They should be prioritized for integrated fire management and deforestation control.
  • High fire, low loss areas (upper left quadrant): These may represent ecosystems where fire occurs but does not lead to permanent forest loss—either because fires occur in non-forest vegetation (savannas, grasslands) or because forests regenerate after fire events.
  • Low fire, high loss areas (lower right quadrant): Deforestation in these areas is driven by factors other than fire, such as agricultural conversion or logging. Different intervention strategies focused on land-use planning and enforcement may be more effective.
  • Areas near the origin: Protected areas with low values for both metrics may either be well-protected or, in the case of dry forest ecosystems, may be affected by the detection limitations discussed above.

Is KfW Already Present in the Most Threatened Areas?

NoteKfW Portfolio Coverage Analysis

Current KfW presence in threatened areas:

  • Of the top 10 most threatened protected areas, 1 are currently supported by KfW projects.
  • This indicates that KfW is underrepresented in the most critical areas requiring conservation investment.

Coverage by threat level:

Temporal Patterns and Emerging Threats

The heatmaps reveal important temporal patterns that align with documented events in the region:

  1. 2019-2020 Fire Crisis: Multiple protected areas show dramatic spikes in both burned area and forest loss during 2019-2020. This corresponds to the well-documented Bolivian fire crisis of 2019, during which fires—many linked to agricultural clearing—burned an estimated 5.3 million hectares across Bolivia (Fundación Amigos de la Naturaleza (FAN) 2019). The Chiquitano dry forest, the world’s largest and best-preserved tropical dry forest (Pennington, Prado, and Pendry 2000), was particularly affected, with approximately 1.4 million hectares burned. International media and conservation organizations extensively documented this event, which triggered national emergency declarations.

  2. Post-Crisis Patterns (2020-2023): Following the 2019 peak, some areas show reduced fire activity. This may reflect:

    • Reduced fuel availability after extensive burning
    • Increased monitoring and enforcement following international attention
    • Natural variation in fire occurrence
    • However, 2020 also saw significant fire activity in the Pantanal region, affecting areas like Otuquis National Park
  3. Persistent Deforestation Patterns: Areas showing continuous, steady loss patterns (rather than fire-related spikes) suggest ongoing land conversion pressures, likely related to agricultural expansion—a pattern consistent with Bolivia’s agricultural frontier dynamics in the lowlands.

Recommendations for Conservation Investment

TipImportant Caveat

This analysis examines conservation priorities exclusively through the lens of environmental threats (forest loss and fire). A comprehensive project prioritization would also need to consider:

  • Biodiversity values: Species richness, endemism, and ecological uniqueness
  • Socioeconomic factors: Local livelihoods, economic development needs, and poverty contexts
  • Social and cultural dimensions: Indigenous territories, traditional land use, and community perspectives
  • Institutional capacity: Existing management structures, governance quality, and implementation feasibility

The recommendations below should therefore be understood as one input among many for investment decisions, highlighting areas where threat mitigation interventions may be most urgent from a forest conservation perspective.

Based on the findings of this analysis, we suggest the following priorities:

Priority 1 — Target High-Threat Areas The protected areas with the highest composite threat scores should be prioritized for conservation intervention. This includes areas experiencing both high absolute forest loss and significant fire exposure. Among these, areas not currently receiving KfW support represent potential opportunities for new investments to expand coverage of the most threatened sites.

Priority 2 — Invest in Fire Management Capacity The 2019-2020 fire crisis demonstrated the vulnerability of protected areas in the Chiquitano and Pantanal-Chaco transition zones to catastrophic fire events. Investments in fire prevention (fuel management, firebreak maintenance), early detection systems, and rapid response capacity could significantly reduce future losses in these ecosystems.

Priority 3 — Strengthen Monitoring for Verification Given the detection limitations of satellite-based monitoring in dry forest ecosystems (discussed in the Methods section), ground-based monitoring and the integration of alternative data sources (such as MapBiomas Chaco) should be considered to improve threat assessment accuracy in areas like Kaa-Iya del Gran Chaco.

Discussion of Methods and Data

Forest Cover Loss Data (Global Forest Watch)

The GFW methodology defines forest cover loss as “a stand-replacement disturbance, or a change from a forest to non-forest state” (Hansen et al., 2013). Key interpretation considerations:

  • Detection scope: GFW detects loss of tree cover ≥5m in height with ≥30% canopy cover, which may miss sparse or degraded forests
  • Cause attribution: GFW does not differentiate between natural and human-caused loss; sudden spikes may indicate wildfires or storms, while continuous patterns suggest agricultural conversion
  • Temporal resolution: Annual data allows trend analysis but may miss within-year dynamics
  • Dry forest limitations: As noted above, GFW algorithms are optimized for humid tropical forests and systematically underdetect changes in dry and deciduous forest ecosystems

Burned Area Data (MODIS MCD64A1)

The MODIS MCD64A1 burned area product (Giglio et al., 2018) provides monthly burned area detection at 500m resolution. Key considerations:

  • Spatial accuracy: The 500m resolution captures the actual spatial extent of burned areas, not just fire detections
  • Ecosystem context: Fire impacts vary by ecosystem—savannas may tolerate and even depend on regular fire, while moist rainforests are highly fire-sensitive
  • Fire types: The product does not distinguish between natural fires, controlled burns, and fire used for land clearing
  • Cumulative counting: Unlike forest loss (counted once per pixel), burned area accumulates each time an area burns, so totals can exceed actual land area if recurrent fires occur

Composite Threat Index Limitations

The threat index developed for this analysis provides a standardized comparison across protected areas but has inherent limitations:

  • Equal weighting: The three indicators are weighted equally, though in some contexts one threat dimension may be more important than others
  • Relative scoring: The index is relative to the analyzed portfolio; an area with a “low” score is only low relative to other areas in Bolivia
  • Detection bias: Areas with systematically underdetected threats (e.g., dry forests) may appear artificially low-risk

Recommendations for Future Analysis

  1. Alternative data sources: Integrate MapBiomas Chaco data to accurately capture dry forest dynamics and loss, addressing GFW’s detection limitations in the Gran Chaco region
  2. Land cover analysis: Overlay loss areas with land cover data to distinguish permanent agricultural conversion from natural disturbance and regeneration
  3. Buffer zone analysis: Examine threats in PA buffer zones and corridors, as edge pressures often drive internal degradation
  4. Spatial hotspot mapping: Create fine-resolution heatmaps within large PAs to identify specific threat frontiers
  5. Temporal baseline comparison: Compare recent fire patterns to historical norms (pre-2015) to contextualize current threat levels
  6. Socioeconomic drivers: Integrate data on agricultural commodity prices, land tenure, and road infrastructure to understand and predict deforestation drivers

Data Sources & References

Fundación Amigos de la Naturaleza (FAN). 2019. “Reporte de Incendios Forestales En Bolivia 2019.” https://www.fan-bo.org.
Margules, Christopher R, and Robert L Pressey. 2000. “Systematic Conservation Planning.” Nature 405 (6783): 243–53.
Pennington, R Toby, Darién E Prado, and Colin A Pendry. 2000. “Neotropical Seasonally Dry Forests and Quaternary Vegetation Changes.” Journal of Biogeography 27 (2): 261–73.
Wilson, Kerrie A, Michael I Westphal, Hugh P Possingham, and Jane Elith. 2005. “Measuring and Incorporating Vulnerability into Conservation Planning.” Environmental Management 35 (5): 527–43.
  • WDPA: UNEP-WCMC and IUCN (2024), Protected Planet: The World Database on Protected Areas (WDPA)
  • GFW: Hansen, M. C., et al. (2013). High-Resolution Global Maps of 21st-Century Forest Cover Change. Science, 342(15 November): 850–53. Data version: GFC-2024-v1.12.
  • MODIS MCD64A1: Giglio, L., Boschetti, L., Roy, D. P., Humber, M. L., & Justice, C. O. (2018). The Collection 6 MODIS burned area mapping algorithm and product. Remote Sensing of Environment, 217, 72-85. Available at: https://modis-fire.umd.edu/ba.html

Report generated: 2026-02-04