Seed Predation by Rodents on Three Ant‐Dispersed Plants

Seed predation by small mammals has been proposed as one selective force in the evolution of ant—dispersal (myrmecochory) in many herbaceous plants of mesic deciduous forests. Three myrmecochorous species, Asarum canadense, Jeffersonia diphylla, and Sanguinaria canadensis, were studied in the Monongahela National Forest, West Virginia, USA, to (a) assess the impact of seed predators on seed mortality, (b) compare seed removal rates when ants were present vs. absent, and (c) evaluate the effects of elaiosome removal, seed burial, and doubling seed number per depot on rates of seed predation. Seed predators, especially rodents, cause more mortality among these woodland herbs than previously thought. Predispersal seed predation by rodents destroyed 24% and 47% of the seed crops of A. canadense and J. diphylla, respectively. The proportions of A. canadense seeds lost to predators varied among sites. Sanguinaria canadensis also suffered seed predation before capsules dehisced. Peromyscus leucopus was the most common seed predator. In exclosure experiments, seed predators removed 70% of the A. canadense seeds when dispersal by ants was prevented, but only 24—39% of the seeds suffered predation when ants were allowed access. Seed predators removed 84% of the S. canadensis seeds when ants were excluded, but 13—43% were lost when also exposed to ants. These estimated minimal proportions of seeds lost to predators when ants were allowed access differed between two study sites, but myrmecochory clearly reduced seed predation at parent plants. In laboratory experiments, P. leucopus located buried A. canadense and S. canadensis seeds less frequently than seeds on the surface (67.5% vs. 100%). Consumption of buried seeds by rodents was reduced when elaisomes were removed, as would be done by ants in natural conditions. Increasing the detectability of seeds through adding olfactory cues or doubling seed number per depot did not increase the frequency of predation, because mice were able to locate depots without these additional cues.