Inferring multilocal typologies of agrosystems and farmers’ practices: A methodological basis for the setting of participatory breeding designs

Abstract

Family farms play an essential role in agroecological transition in Sahelian countries and worldwide. They present diversified features in terms of socio-economic organization, agrobiodiversity management and cropping systems diversity. Decentralized participatory breeding approaches aim to sustain the diversity of varieties adapted for such smallholder farmers' contexts. However the lack of clear target population of environments limits the focus and the efficiency of these approaches given the large diversification of the local contexts. In this study, we surveyed variables linked to agrosystems, crop management options and farmers’ criteria of varietal evaluation from 254 family farms sampled along 13 locations spanning the target area of a decentralized participatory breeding program of cowpea crop (Vigna unguiculata L. Walp) in Niger. The objective of our study was to infer typologies of family farms in the study area based on relevant variables supporting the setting of target population of environments (TPEs) to be considered in the breeding program. We used factorial analysis of mixed data (FAMD) and the Discriminant Analysis of Principal Components (DAPC) to infer the clusters. Chi square test, analysis of variance and generalized linear model were used to identify key variables discriminating the clusters. These clusters were geographically mapped to analyze their multilocal distribution. So, we identified and characterized four clusters structuring the diversity of the local agrosystems (Typologie G), five clusters structuring the diversity of cowpea cropping management options (Typologie C) and five clusters structuring the diversity of criteria used by farmers to evaluate the performance of cowpea varieties in the local contexts (Typologie P). Typology G distinguished farms based on discriminating variables linked for instance to secondary activities, cultivated species, soil fertility management practices and farm resources including land and livestock. Typology C distinguished farms based on cowpea management pratices including the secondary crop intercropped with cowpea (sorghum, Guinea sorrel, sesame or groundnut) and the use of cowpea harvest products (seeds, haulms, hulls). Typology P was based on discriminating performance criteria including cycle length, insect resistance, striga resistance, drought resistance, haulm production and economic value of cowpea variety. This methodology provides a robust and replicable way for the definition of clusters capturing the diversity of farms and local contexts. We discussed the perspective of using these clusters to set target population of environments for decentralized breeding programs.