The evolution of dispersal at range margins received much attention recently; especially in the context of dynamic range shifts, such as those following climate change. However, much less attention has been devoted to study variation in and selection on dispersal at non-expanding range margins, where populations are often small and isolated, and empirical test is dearly missing. To fill this gap, we tested whether dispersal of an ant-dispersed perennial plant (Sternbergia clusiana), is quantitatively and/or qualitatively reduced toward a non-expanding range margin. We evaluated plant investment in dispersal structures (elaiosome), seed removal rates, and the relative abundance, activity and behavior of low- and high-quality seed dispersing ants in six sites ranging from mesic Mediterranean site (>600 mm of annual rainfall) to arid site (<100 mm of annual rainfall), which marks the southern range margin of the species. In a set of cafeteria and baiting experiments, we found that overall seed removal rates, the contribution of high-quality dispersers, maximum dispersal distance and dispersal-conducive ant behavior decreased towards range margins. These findings agree with a lower investment in reward by range margin plant populations, as reflected by lower elaiosome/seed ratio, but not by variation in the reward chemistry. More than variation in traits controlled by the plants, the variation in ant-seed interactions could be attributed to reduced presence and activity of the more efficient seed-dispersing ants in the marginal populations.  Specifically, we found a mismatch between local distribution of potentially effective seed dispersers, and that of the plant, even though those dispersers were observed in the study site. Interestingly, although the observed variation in the outcome of ant-seed interactions supported the prediction of reduced dispersal at non-expanding range margins with small and isolated populations, the underlying mechanisms seems to be incidental difference in the seed-dispersing ant community rather than a plant-mediated response to selection

The data was collected in field experiments and lab analyses. It was analyzed using R and SPSS