- We explicitly tested the following predictions using Gerbillus allenbyi and G. pyramidum foraging on artificial arrays of natural seedbank on a semi-stabilized sand dune: 1. that the gerbils will forage using a quitting harvest rate rule, 2. that larger seeds are preferred due to higher encounter rates, and 3. that there are community level consequences for the annual plants as a direct result of foraging by the desert rodents. Natural seedbank, separated into two size classes, was placed in seed trays in the field at four different densities (1/16 x 1/4 ×, 1 ×, and 2 x normal). Following exposure to granivory, the remaining seeds were weighed and germinated to test for community level effects. Only the 1 x and 2 x normal density plots were heavily foraged, and at both seed sizes, which suggests that the gerbils employed a quitting harvest rate rule. In support of our second prediction, the two species of gerbils tended to consume more of the larger seeds, particularly at higher densities. The mean number and total number of plant species that germinated in plots exposed to granivores was not significantly different from unexposed samples. At the community level however, there was no net association of germinated plant species in the four treatment groups exposed to granivores, but a significant net positive association in the unexposed control seedbank. Gerbil foraging on annual plant seedbank may thus subtly change the entire community structure (from positive to neutral), although not necessarily shift the species distributions significantly. Our results corroborate other studies involving artificial food types such as millet and suggest that foraging decisions may affect the plant community.