Poultry farm distribution models developed along a gradient of intensification

• Intensive farms were found clustered in Nigeria, Thailand, Argentina and Belgium. The gradient of clustering did not reflect the intensification. We developed farm distribution models (FDM) to predict location sizes. Improvements should consider covariates explaining better locations Efficient planning measures limiting epidemic spread requires information on sizes (number animals per farm). However, such data are rarely available. intensification process which is operating most low- middle-income countries (LMICs), comes together with a spatial farms, characteristic epidemiological sensitive to. predicting both number farm, while accounting for data-poor countries, using poultry production as an example. selected four Belgium, along expressed by capita Gross Domestic Product (GDP). First, we investigated chicken spectrum Second, built based censuses commercial each point pattern random forest models. As external validation, predicted Bangladesh. increased gradually line GDP following order: Interestingly, find clustering. Our modelling procedure could only partly reproduce observed datasets sample internal validation. be reproduced predictors poorly explained Further improvements methodology explore other intensity sizes, well methodology. Structural transformation, economic development environmental conditions essential characteristics extrapolation our FDM procedure, generalisation appeared challenging. believe ultimately used predictive tool countries.