World Agroforestry Centre (ICRAF) - TerraFund Verification Results

Project Snapshot

Drought, overgrazing and demand for fuelwood have heavily degraded Eastern Rwanda’s Nyagatare and Kayonza Districts. Restoring this ecosystem — the breadbasket of densely populated and growing Rwanda — is especially urgent. It's simple: As soil loses its fertility, farmers produce less food, and fewer people have enough to eat.

In 2022, TerraFund financed World Agroforestry Centre (ICRAF), a research organization, to grow trees on these degraded farms and rangelands. ICRAF has a long track record in improving access to the diverse, high-quality seeds. The healthier the seeds, the higher the chance that trees survive — and benefit people for decades to come.

With that mission in mind, ICRAF's goal was to help farmers plant a mix of 340,000 native, fruit, and pollinator species across 3,400 hectares of land.

In the project's first years, ICRAF strengthened local nurseries and built Rural Resource Centers, small plots for experimentation, demonstration, and training. After learning at the centers, farmers went out to plant trees to stabilize and improve soil fertility, grow woodlots to redirect demand for firewood from natural forests, and increase resilience to drought. By 2025, local community members had planted 443,551 trees across 3,371 hectares and employed 1,743 people, with women and youth each representing more than 40% of the total.

Verified Results

The Target

Metric	Target	Reported
Trees Planted	340,000	443,551
Survival Rate	70%	45.5%
Trees Grown	201,815

The Analysis

The Results

Metric	Verified Result
Trees in March 2022	471,317
Trees in August 2024	752,466
Verified New Trees	281,149

Due to its scale, diligent reporting practices, and community-centered model, WRI selected ICRAF's project to test a seemingly simple question: How many of those 443,551 trees survived two years after they were planted?

ICRAF’s self-reported data indicates that the project exceeded its target by 30%, reaching 443,551 trees planted. WRI then applied its AI-powered tree count methodology, powered by the DINOv2 computer vision model, to independently verify the reported data. WRI also validated these insights through a field visit to gather perspectives from the technical staff and the farmers themselves.

Because not all planted trees survive, WRI verifies the number of “trees grown” rather than trees planted by multiplying the project’s reported survival rate by the total number of trees planted. Based on ICRAF’s reported survival rate of 45.5%, a conservative estimate, WRI anticipated that 201,815 trees would have survived in the 2.5 years after planting began. That became the target.

To ensure that each partner farmer could get credit for their work, ICRAF and WRI had to collect 35 separate geospatial project boundaries, known as "polygons," that enclosed the 3,371 hectares of land where farmers had planted. WRI then examined high-resolution satellite imagery for 22% of that area, a statistically sufficient percentage to assess overall change in the number of trees growing across the project.

Comparing imagery before the project started to 2.5 years after planting finished, WRI's analysis detected a net increase of 281,149 trees, a significant improvement over ICRAF's own estimates.

These findings represent early insights into the project's long-term outcomes. Agroforestry systems develop gradually, and monitoring will continue for three more years until the trees are fully mature.

WRI's Assessment

What’s Working

ICRAF’s project demonstrates that farmers can successfully grow trees in large-scale agroforestry systems, even in dry landscapes where trees struggle to survive. The project's combination of scientific expertise with hands-on learning in the field helped farmers adopt restoration practices that fit local conditions. Through its rural resource centers, ICRAF helped farmers experiment with different tree species and management practices with little risk.

At the same time, well-managed nurseries full of drought-tolerant species ensured a reliable supply of seedlings for farmers and allowed technical staff to quickly replace any trees that died in this harsh ecosystem. By pairing the best of research with sustained community engagement, WRI expects that this model will lead to long-term adoption by farmers.

What's Challenging

Eastern Rwanda is a dry environment with heavily eroded soils. Drought and pests kill many trees before they grow to maturity. This justifies the project’s reported survival rate of 45.5%, which is significantly lower than 70% to 80% range reported by most TerraFund projects.

However, WRI's tree count analysis shows that the number of new trees was 39% higher than ICRAF's own estimation. ICRAF's conservative approach to calculating the survival rate corrects a commonly cited challenge from local restoration projects: Many organizations over-estimate their survival rates because of the pressure they feel from donors. Independent verification can encourage implementers to set more achievable targets and educate donors on the challenge of operating in drylands.

Pathway to Scale

The project offers several lessons for scaling agroforestry led by smallholder farmers. First, restoration efforts need to match local conditions, including drought risk, soil quality, and how farmers use the land. Projects that adapt their approach to these realities are more likely to succeed.

Second, restoration works best when farmers have the knowledge and support they need to manage trees on their farms over the long term. Once a tree is planted, the task shifts to protecting and conserving this new natural asset.

Finally, trees take several years to grow, but new datasets like WRI's AI tree count algorithm can help assess progress earlier and more accurately than before. Peer organizations have much to learn from the depth and rigor of ICRAF's work.

_{Cover image credit: World Agroforestry (ICRAF)/TerraFund}