In many of the world's drylands genetic and soil erosionare widespread. To enhance success of any large scale restorationeffort in these hazardous conditions, it is essential to set upsystematic seed production of key native plant species. The aim ofthe traditional outline is to gather local yet ample geneticvariation to avoid the risks of non-adaptation. But forStipa lagascae R. and Sch, aNorth-African perennial bunch grass, we carried out an outlineinspired from conventional forage breeding practice, with the aim toselect for vigour. Here, we assess its outcomes as for the risk ofselecting non-local and/or non-ample geneticvariation. We established a collection composed of 276 clonal linesout of individuals coming from 70 accessions, spread over 6 majorecogeographic origins within Presaharian Tunisia (100mm < average annual rainfall < 200mm), both source and target region. During 7successive growing seasons with stark contrasts for rainfall andfertility, we observed the collection for vigour and phenology.Principal Components Analysis has been carried out on a data matrixof 276 lines and 11 variables. Both for vigour and for phenology,individuals differentiated strongly, but irrespective of accession orecogeographic origin. Moreover, differences for vigour persistedthroughout contrasting growing seasons. Both facts combined(1) do not indicate any local adaptation within the targetregion, (2) suggest instead intense genetic structuring dueto isolation by distance, (3) make it easy to cull alimited number of lines to combine considerable genetic variationwith consistently superior vigour.
