Geographic variation in the adaptive landscape for seed size at dispersal in the myrmecochorous Helleborus foetidus

Abstract

Literature on seed dispersal mutualisms suggests that plant populations should hardly adapt to their current dispersers. We address the predictions that selection pressures exerted by ants on dispersal-related diaspore traits of the ant-dispersed Helleborus foetidus are highly variable in space, and that geographic (inter-population) variation in these traits is unrelated to selection by current dispersers. To test these predictions we use the concept of the quantitative adaptive landscape for seed size at dispersal. Such landscape depicts the relationship between the population’s mean trait value (mean seed size in the present study) and the population’s mean fitness (mean dispersal probability in the present study). Adaptive landscapes make it possible to assess whether the mean population’s phenotype agrees with one favored by selection. We first analyse, in 12 populations of H. foetidus from southern Spain, the extent of divergence among populations in seed and elaiosome size, and the abundance, composition, and behavior of the ant communities. Seeds from a fixed set of five of these populations were offered to ants in all the study sites to fit the adaptive landscape for seed size. In addition, seeds from the local population were also offered in each site. Our results show that seed size has undergone a larger divergence among populations than elaiosome size. Despite geographic variation in ant assemblages, the adaptive landscapes for seed size at dispersal were remarkably similar among sites: ants create disruptive selection on seed size in 10 out of 12 study sites. As predicted, the basic features of these adaptive landscapes (curvature and location of the minimum) varied geographically in accordance with variation in the size of seed dispersers. Also as predicted, in most populations, the observed mean seed size does not agree with that expected from the adaptive landscapes at dispersal. However, the relevance of dispersers for seed size evolution should not be neglected since the agreement between observed and optimum seed size was stronger where dispersers were more abundant. Thus, against the general view, our results evidence that, in H. foetidus, the observed geographic variation in dispersal-related plant traits is partly linked to selection exerted by current dispersers. Geographic variation in ant assemblages determines both the existence of a selection mosaic and the degree of adjustment of populations to the patterns of selection in the mosaic.