Abstract Aim Human settlement establishment and reproduction of the baobab tree (Adansonia digitata) appear spatially and temporally dependent because baobabs are abundant in many settlement sites in Africa. This paper tests the spatiotemporal relationship between baobab and settlement distribution. Location South-western Mali. Methods In an area of 183 km2, 1240 baobabs were located and mapped, their diameters measured, and habitat characteristics recorded for each individual. All occupied (n = 9) and abandoned (n = 84) settlements were located and mapped, and occupation dates were determined through interviews. Chi-squared analysis indicated baobab habitat preferences, and bivariate point-pattern analysis tested baobab–settlement spatiotemporal independence. Results Baobabs and human settlements are positively spatially associated at most distances and for all baobab size-class–settlement age-class pairs. However, positive spatial association is significant only at distances < 500 m, and young settlements and large baobabs are not significantly associated. Positive association between small and large baobabs is marginally significant at <300 m, but observed significance is less than that for baobab–settlement positive association. Baobab abundance is not evenly distributed across the range of habitats it occupies; recruitment is strongest in settlements and fields, and on cliffs, while mortality is highest on cliffs. Ethnographic observations suggest that human settlement practices and fruit use are the main human factors contributing to baobab–settlement positive spatial association. Main conclusions There are three main conclusions: (1) Human settlement and baobab recruitment are spatially dependant because settlement leads directly and indirectly to the development of baobab groves at settlement sites. (2) The lower than expected abundance of mature individuals in natural habitats, and the habitat preferences of the observed population, suggest that baobabs were introduced to south-western Mali, probably centuries ago. (3) Human mobility over decadal time-scales is necessary to maintain baobab population structure in landscapes dominated by shifting land use, where baobabs are not purposefully planted. Baobab population processes in such landscapes occur at the scale of human settlement.