Unexpected pattern of pearl millet genetic diversity among ethno-linguistic groups in the Lake Chad Basin

Abstract

Despite of a growing interest in considering the role of sociological factors in seed exchanges and their consequences on the evolutionary dynamics of agro-biodiversity, very few studies assessed the link between ethno-linguistic diversity and genetic diversity patterns in small-holder farming systems. This is key for optimal improvement and conservation of crop genetic resources. Here, we investigated genetic diversity at 17 SSR markers of pearl millet landraces (varieties named by farmers) in the Lake Chad Basin. 69 pearl millet populations, representing 27 landraces collected in eight ethno-linguistic farmer groups, were analyzed. We found that the farmers' local taxonomy was not a good proxy for population's genetic differentiation as previously shown at smaller scales. Our results show the existence of a genetic structure of pearl millet mainly associated with ethno-linguistic diversity in the western side of the lake Chad. It suggests there is a limit to gene flow between landraces grown by different ethno-linguistic groups. This result was rather unexpected, because of the highly outcrossing mating system of pearl millet, the high density of pearl millet fields all along the green belt of this Sahelian area and the fact that seed exchanges among ethno-linguistic groups are known to occur. In the eastern side of the Lake, the pattern of genetic diversity suggests a larger efficient circulation of pearl millet genes between ethno-linguistic groups that are less numerous, spatially intermixed and, for some of them, more prone to exogamy. Finally, other historical and environmental factors which may contribute to the observed diversity patterns are discussed.

Figure 1

Geographical location of sampled pearl millet populations. Each population is represented with a specific colour for the ethno-linguistic group to which the farmer who provided the seeds belongs. Circles and diamonds represent early and late-flowering landraces, respectively. Stars represent pairs or triplets of early and late-flowering landraces obtained from the same farmer.

Figure 2

Genetic distance-based neighbour-joining tree showing the genetic similarity among sampled populations labelled by the population number given in Table 1. Coloured branches correspond to landraces belonging to different ethno-linguistic group. All populations collected in eastern side of the Lake Chad have the same colour.

Figure 3

Genetic structure of pearl millet populations revealed by a clustering Bayesian analysis (K=6) in the Lake Chad Basin. Populations are arranged in the same order as in Table 1. (a) Bar plots of all solutions obtained from the Bayesian analysis. Each thin vertical line corresponds to an individual. Coloured segments represent the proportion of each individual’s genome assignment to each cluster. Numbers on the right of bar plots show how many times each solution was observed among 100 repetitions. (b) Illustration of the most probable solution (A) based on the population average proportion of genome assignment to each cluster, on the geographical map. Each pie chart represents one pearl millet population and each colour represents each inferred genetic cluster.

Figure 4

Geographical distribution of genetic clusters inferred from the clustering Bayesian analysis carried out on pearl millet populations sampled in the Zarma-Songhay and the Hausa socio-cultural areas. Hatched zones correspond to social mixing areas. Only the major solution (found for 9 runs out of 10) is shown. Each pie chart represents one pearl millet population and each colour represents each inferred genetic cluster. The delimitation of cultural areas is approximate.