Weedy plants with intermediate (domesticated x wild) phenotypes occur in most pearl millet fields in West Africa, even in the absence of wild populations. They are usually found, in high numbers, both inside and outside of drills. Questions pertaining to the evolutionary dynamics of diversity within the pearl millet complex (domesticated-weedy-wild forms) were addressed in this study. The diversity of the different components of this complex sampled in two pearl millet fields in two villages of southwestern Niger was assessed at both molecular (AFLP) and morphological levels. Results show that, in both fields, weedy plants found outside of drills are morphologically distinct from weedy plants found inside drills, despite their close similarity at AFLP markers. The data suggest some introgression from the wild to the weedy population but nevertheless that the gene flow between the parapatric wild and domesticated populations is very low. This challenges the traditional view that regular hybridization between domesticated and wild pearl millets explains the abundance of these weedy plants despite farmers' seed selection. The level of genetic differentiation between fields from the two villages was low when considering domesticated and weedy plants. This could be explained by high gene flow resulting from substantial seed exchanges between farmers. The fact that it is very difficult for farmers to keep their own selected seeds, and the consequent substantial seed exchanges between them, is probably the main factor accounting for the maintenance and dispersal of weedy pearl millets in the region, even in areas where no wild forms have been observed.

In the Sahel of Africa, farmers often modify their cultivation practices to adapt to environmental changes. How these changes shape the agro-biodiversity is a question of primary interest for the conservation of plant genetic resources. We addressed this question in a case study on pearl millet in south western Niger where farmers used to cultivate landraces with different cycle length in order to cope with rain uncertainty. Early and late landraces were previously grown on distant fields. Nowadays, mostly because of human population pressure and soil impoverishment, it happens that the two types of landraces are grown on adjacent fields, opening the question whether gene flow between them may occur. This question was tackled through a comparative study among contrasting situations pertaining to the spatial distribution of early and late landraces. Observations of flowering periods showed that pollen flow between the two landraces is possible and has a preferential direction from early to late populations.

No abstract