African Agricultural Practice Evidence from 2,916 Studies

The dataset synthesizes evidence from 2,916 agricultural studies conducted across Africa, containing 112,859 observations on the performance of 363 practice combinations across 87 outcome indicators. It was created by Todd Stuart Rosenstock and spans studies published from 1934 to 2024. Observations are geolocated and can be linked to environmental, economic, and social datasets.

Use Cases

• Analyze the performance of specific agronomic practice combinations across different environmental context variables and geographic coordinates.
• Model the relationship between experimental design variables and outcome indicators to identify effective climate adaptation strategies.
• Link geolocated observations to external datasets to study how local context shapes treatment comparisons for commodities like soil fertility or intercropping.
• Conduct meta-analysis using bibliographic metadata and outcome hierarchies to assess evidence trends in sustainable intensification over the 1934-2024 time period.

Strengths

• Large evidence base from 2,916 agricultural studies, providing 112,859 observations.
• Covers 363 distinct agronomic practice combinations and 87 outcome indicators, enabling detailed analysis.
• Modernized methodology using AI-powered tools (OpenAlex, OpenAI APIs) for improved consistency and reproducibility in evidence synthesis.
• Geolocated observations enable spatial analysis and linkage with open-source environmental, economic, and social datasets.

Limitations

• The temporal coverage includes studies from as early as 1934, where methodologies and reporting standards may differ significantly from modern practices.
• The dataset's focus is exclusively on Africa, limiting generalizability to other geographic regions.
• Reliance on automated extraction tools, while increasing throughput, may introduce specific errors or biases in data extraction that require validation.

Provenance

Source

The Alliance of Bioversity International and CIAT Dataverse.

Collection Method

Evidence synthesized from published agricultural studies using AI-augmented literature screening (OpenAlex) and semi-automated data extraction workflows (OpenAI APIs).

Time Range

1934 to 2024.

Freshness

Last updated in February 2026, with the v2 update extending coverage to include studies published through 2024.

Geography

Africa.