Résumé
In Benin, identifying agro-ecological zones and agricultural development poles is crucial for implementing effective agricultural policies. However, current large-scale zoning methods for agrarian systems rely on heterogeneous data sources and often involve subjective selection of socio-economic and environmental variables. These approaches face challenges in representativeness and reproducibility, limiting their utility for policy and planning. To overcome these limitations, we propose a novel approach grounded in the principle that landscapes - as a reflection of the interplay between biophysical and human factors - can serve as proxies for land use and agricultural practices in rural areas. This makes landscape zoning a viable tool for approximating agrarian system zoning. Recognizing that traditional landscape mapping relies on extensive, multi-scale data with varying degrees of accuracy, we introduce an innovative method called radiometric landscape mapping (Lemettais et al., 2024). This approach derives landscapes exclusively from remote sensing data, bypassing the need for measured variables (e.g., climate data) or interpreted products (e.g., land cover maps). It offers a statistically robust, scalable, and cost-effective solution that is applicable across different locations and scales. Data and Methods Radiometric landscapes were calculated using the first principal components of a series of MODIS NDVI (Normalized Difference Vegetation Index) images from 2018 to 2022. These analyses resulted in the identification of 36 homogeneous radiometric landscapes, which were subsequently classified into nine broader radiometric zones. For comparative analysis, we utilized data from the 2017 " Typologie des exploitant(e)s des sites de recherche et développement du Bénin " survey (Sossou et al., 2019), which covered 477 villages. This survey collected data on agricultural households, focusing on socio-economic characteristics (e.g., household composition, assets) and agricultural practices (e.g., crop types, mechanization, irrigation). The analysis identified three primary farming system types: irrigated systems, mechanized systems, and intensive systems relying on chemical inputs. Results and Implications The comparison of the agrarian system types distribution with radiometric zoning revealed strong alignment in terms of land cover composition and an intensification of the agriculture. Radiometric zones effectively discriminated between land cover types and provided a robust framework for analyzing agrarian systems. Conclusion These findings highlight the potential of radiometric zoning to redefine zoning frameworks for agricultural and land-use planning policies. By offering a replicable, data-driven, and scalable approach, radiometric landscapes present a promising tool for supporting sustainable agricultural development in Benin and beyond.