Data Visualization Project · LSE · 2026

No One's Land

Power vacuums, armed groups and deforestation before and after Colombia's 2016 Peace Agreement

Image: Renaldo Matamoro in Unsplash
Context & Research Question

Colombia ranks among the world's most biodiverse nations, harboring approximately 10% of global species[1] across its Amazon rainforest, Andean ecosystems, and Pacific Chocó region. This environmental wealth, however, has coexisted with over 52 years of internal armed conflict, the longest in Latin American history.

The Revolutionary Armed Forces of Colombia (FARC), at their peak, maintained presence in roughly 71% of municipalities, predominantly in remote, forested territories where state institutions remained historically weak. In November 2016, the Colombian government and FARC signed a comprehensive peace agreement, ending five decades of hostilities.

The accord has been widely celebrated as a humanitarian success. Yet emerging evidence suggests an unintended consequence: a significant surge in deforestation following the agreement.

Research Question

How has the relationship between territorial control, armed actors, and deforestation reconfigured following Colombia's 2016 Peace Agreement, and what are the implications for conservation policy in post-conflict settings?

Data Sources & Methodology

  • Deforestation API

    Global Forest Watch (Hansen et al., University of Maryland)

    Tree cover loss 2001-2024, 30m resolution via Google Earth Engine — globalforestwatch.org

  • Conflict Events 11 APIs 241,118 records

    SIEVCAC (Sistema de Información de Eventos de Violencia del Conflicto Armado)

    Batch download via Socrata API loop: massacres, kidnappings, forced displacement, sexual violence, landmines, terrorist attacks, and more — datos.gov.co

  • Population SCRAPING 1,122 municipalities

    DANE Census 2018 via Wikipedia

    Web scraping for per-capita normalization of conflict data — dane.gov.co

  • Geographic Boundaries TopoJSON

    Colombia Municipalities (1,031 geometries)

    Name standardization achieved 94% match rate across datasets

View Colab Notebook →

The Paradox

The Peace Agreement achieved its primary humanitarian objective: conflict-related events—including homicides and kidnappings—dropped by over 90% between 2016 and 2024, representing a historic success.

However, the post-conflict period revealed an unexpected environmental consequence. Before 2016, deforestation fluctuated between 122,000-270,000 hectares annually but in 2017, Colombia registered a 46% increase in deforestation—the highest in two decades—coinciding precisely with FARC's territorial withdrawal.

This inverse correlation suggests that armed conflict had inadvertently functioned as a barrier to forest exploitation.

The Governance Vacuum

To test whether FARC presence functioned as a deforestation barrier, we compare municipalities by their pre-2016 FARC presence levels.

The pattern is striking: municipalities with high FARC presence experienced the largest deforestation spikes after 2016, while areas without FARC presence showed minimal change.

Spatial Distribution

Where were these high-FARC municipalities? The maps show the geographic concentration: the Amazon —Caquetá, Guaviare, Meta, Putumayo— and Pacific coast. These regions share high forest cover, weak state presence, and economies linked to coca and cattle.

Use the sliders to explore. FARC data ends in 2018; deforestation extends to 2024.

Territorial Reconfiguration

FARC's withdrawal triggered violent competition for control. Dissident FARC factions, ELN, and criminal organizations like Clan del Golfo moved into former strongholds, seeking coca routes, illegal mining, and extortion networks.

Unlike FARC's stable governance, these actors operate with shorter time horizons. The result: accelerated deforestation for cattle ranching and coca expansion.

Predictors of Deforestation

To quantify the relationship between armed group presence and forest loss, we estimated an OLS regression model predicting the change in deforestation (post-2016 minus pre-2016) for each municipality.

Municipalities with higher historical FARC presence (+180 ha) and active FARC dissident factions (+91 ha) experienced significantly greater increases in deforestation after the Peace Agreement.

Interestingly, ELN presence shows less deforestation change (-44 ha), likely because ELN operates in mountainous regions with lower forest cover.

Significant (p<0.05) Not significant
R² = 17.3% | n = 1,032 municipalities
Conclusion

This analysis reveals a paradox of peace: FARC's territorial control inadvertently protected Colombia's forests. The 2016 agreement, while a humanitarian triumph, created governance vacuums that accelerated deforestation—particularly in high-FARC municipalities (+123 ha/year).

These findings suggest that post-conflict conservation requires rapid state presence in former armed group territories. Peace agreements must anticipate environmental consequences and integrate forest governance into transition planning.

Analytical Challenges

Data integration posed significant challenges. Conflict data from 11 SIEVCAC APIs (241,118 records) required filtering to 67,797 cases with identified perpetrators. Municipality name standardization across datasets achieved 94% match rate. FARC presence was proxied through violent events—a limitation, as high violence may indicate contested zones rather than consolidated control. Population data was web-scraped from Wikipedia for per-capita normalization.

Word count: 774 (including modal content)

References

[1] Colombia.co (2024). "Colombia: Second greatest biodiversity in the world." https://colombia.co

[2] Hansen, M. C., Potapov, P. V., Moore, R., et al. (2013). "High-Resolution Global Maps of 21st-Century Forest Cover Change." Science, 342(6160), 850-853. Global Forest Watch

[3] Centro Nacional de Memoria Histórica (2024). Sistema de Información de Eventos de Violencia del Conflicto Armado Colombiano (SIEVCAC). datos.gov.co

[4] DANE - Departamento Administrativo Nacional de Estadística (2018). Censo Nacional de Población y Vivienda. dane.gov.co

[5] Wikipedia (2024). "Anexo: Municipios de Colombia por población." Web scraping for municipal population data. es.wikipedia.org

[6] Anthropic (2025). Claude AI Assistant. This project used generative AI to support Python code cleaning and to resolve technical issues in Vega-Lite, HTML, and CSS. The AI assisted with error debugging and code optimization. All design choices, written content, and analytical methods are the author's own work. claude.ai