Geographic variations in the correspondence between diaspore phenotypes and disperser behavior are thought to determine the evolution of plant-animal dispersal mutualisms. Helleborus foetidus is a widely distributed plant in Western Europe, which seeds bear a lipid rich elaiosome attracting ant dispersers. Laboratory cross-tests were conducted to check the correspondence between diaspore phenotypes and ant preference in two localities of the Iberian Peninsula, Caurel and Cazorla, separated by 750 km. Diaspores from Caurel were systematically preferred to those from Cazorla by Formica lugubris (the major disperser at Caurel), and Aphaenogatser iberica and Camponotus cruentatus (both major dispersers at Cazorla). Further bioassays conducted on A. iberica only showed that differences in elaiosome traits were sufficient to explain ant preference. Separation of the lipid fractions composing the elaiosome revealed that triglycerides, diglycerides and free fatty acids were all dominated by oleic acid. The elaiosomes from Caurel contained relatively more free oleic acid but were less concentrated in linoleyl-containing triglycerides, free palmitic acid and free linoleic acid than those from Cazorla. The three lipid fractions were attractive to ants but dummies soaked with the free fatty acids extracted from Caurel were preferred to those from Cazorla. Taken together, these results reinforce the idea that oleic acid is a major releaser of seed collection by ants and suggest that geographic variations in free fatty acid composition affect the probability of diaspore removal by ants which in turn potentially determine plant demography.

The effect of local ant species on the dispersal success of a myrmecochorous plant, Helleborus foetidus, was analyzed in two populations of the Iberian Peninsula (Caurel and Cazorla, respectively). The contribution of the various local ant species to dispersal was very unequal. While 5 and 19 ant taxa visited the plants of Caurel and Cazorla, respectively, most removal activity (67 and 80%) was performed by two species only (Formica lugubris and Camponotus cruentatus, respectively). Visits by dispersers were also unequally distributed between neighboring plants. While some plants were always visited during the period of seed release, others were never visited. A regression model indicated that this pattern might be explained by two plant traits: ants preferred to visit plants that released more seeds and whose elaiosomes were richer in oleic acid. Although it has long been known that this compound triggers removal by ants, it is the first demonstration that quantitative variations in elaiosome traits contribute to variation in dispersal success. Finally, other variables being equal, morphological traits (seed size, elaiosome size, and elaiosome/seed size ratio) did not affect ant behavior. Although myrmecochory has long been considered a diffuse interaction, our results support the idea that, at local scale, a limited number of ant species may be decisive to its evolution.