Seed predation is a pervasive, and often extensive, species interaction in tropical regions. The outcome of this interaction, however, can change across different spatial scales. Therefore, multi-scale approaches are needed to elucidate whether different species interactions, such as seed predation, are controlled by factors operating at different spatial scales and, ultimately, determine the scale at which patterns are most prominent. Here, we determined if geographical variation in seed predation of the palm Attalea phalerata by bruchid beetles of the genus Pachymerus (Family: Bruchidae) generates differences in survival. Moreover, we identified the scale at which seed predation is more variable (endocarp, tree, or locality). If seed predation operates at the local scale, then variation in survival should be greater among- than within localities. If, however, seed predation operates at the tree scale, then variation in survival rates should be greater within- than among localities, and thus, no geographical structure should be apparent, suggesting that the outcome of ecological interactions is evenly organized across space. We used a hierarchical design in which endocarp type (given by the number of seeds) was nested within each tree and trees were nested within each locality. For two years we quantified seed predation under 20 trees in each of 4 populations in the Bolivian lowlands.

Results/Conclusions

In general, levels of seed predation varied from 40 to 80%. The greatest variation was at the population level, providing evidence for a geographical structure in survival of individuals of both beetle larvae and palm seeds. Seed survival, as well as larval survival, showed the same patterns of variation across space. Thus, patterns of seed and larval survival are coupled and operate at the same scale. Moreover, changes in patterns of palm-beetle interactions are only evident at large scales. Collectively, results suggest that the outcome of seed predation varies among localities and, that for palm-beetle interactions differences in selective forces operating through seed predation may be more important at the population level. The implication of our results is that for phenotypic matching to occur, and thus coevolution, the reciprocal selective pressure most occur at the same scale, and must differentiate across space.