Search

Recent studies investigating the relationship between crop genetic diversity and human cultural diversity patterns showed that seed exchanges are embedded in farmers' social organization. However, our understanding of the social processes involved remains limited. We investigated how farmers' membership in three major social groups interacts in shaping sorghum seed exchange networks in a cultural contact zone on Mount Kenya. Farmers are members of residence groups at the local scale and of dialect groups clustered within larger ethnolinguistic units at a wider scale. The Chuka and Tharaka, who are allied in the same ethnolinguistic unit, coexist with the Mbeere dialect group in the study area. We assessed farmers' homophily, propensity to exchange seeds with members of the same group, using exponential random graph models. We showed that homophily is significant within both residence and ethnolinguistic groups. At these two levels, homophily is driven by the kinship system, particularly by the combination of patrilocal residence and ethnolinguistic endogamy, because most seeds are exchanged among relatives. Indeed, residential homophily in seed exchanges results from local interactions between women and their in-law family, whereas at a higher level, ethnolinguistic homophily is driven by marriage endogamy. Seed exchanges and marriage ties are interrelated, and both are limited between the Mbeere and the other groups, although frequent between the Chuka and Tharaka. The impact of these social homophily processes on crop diversity is discussed.

Understanding the effects of actions undertaken by human societies on crop evolution processes is a major challenge for the conservation of genetic resources. This study investigated the mechanisms whereby social boundaries associated with patterns of ethnolinguistic diversity have influenced the on-farm distribution of sorghum diversity. Social boundaries limit the diffusion of planting material, practices and knowledge, thus shaping crop diversity in situ. To assess the effect of social boundaries, this study was conducted in the contact zone between the Chuka, Mbeere and Tharaka ethnolinguistic groups in eastern Kenya. Sorghum varieties were inventoried and samples collected in 130 households. In all, 297 individual plants derived from seeds collected under sixteen variety names were characterized using a set of 18 SSR molecular markers and 15 morphological descriptors. The genetic structure was investigated using both a Bayesian assignment method and distance-based clustering. Principal Coordinates Analysis was used to describe the structure of the morphological diversity of the panicles. The distribution of the varieties and the main genetic clusters across ethnolinguistic groups was described using a non-parametric MANOVA and pairwise Fisher tests. The spatial distribution of landrace names and the overall genetic spatial patterns were significantly correlated with ethnolinguistic partition. However, the genetic structure inferred from molecular makers did not discriminate the short-cycle landraces despite their morphological distinctness. The cases of two improved varieties highlighted possible fates of improved materials. The most recent one was often given the name of local landraces. The second one, that was introduced a dozen years ago, displays traces of admixture with local landraces with differential intensity among ethnic groups. The patterns of congruence or discordance between the nomenclature of farmers' varieties and the structure of both genetic and morphological diversity highlight the effects of the social organization of communities on the diffusion of seed, practices, and variety nomenclature.