Background/Question/Methods   Seed-caching granivores may act as seed predators and hurt plant recruitment or as seed dispersers and aid recruitment. This dual role makes it difficult to interpret seed removal studies. We developed a heuristic model of antagonistic/mutualistic relationships for quantifying the net outcome of plant-seed consumer interactions.

Results/Conclusions   According to our model, granivory is beneficial when the probability of caching removed seeds exceeds the ratio of establishment on the ground to establishment from caches. Our model also breaks down plant-consumer interactions into readily measurable field data (e.g. probability of germination when granivores are excluded; fraction of seeds removed when granivores are present; proportion of seeds removed that are cached rather than consumed) that can be obtained for specific habitat types and combined algebraically to estimate habitat-specific interactions. We used it to evaluate the outcome of plant-rodent interactions described in published studies, demonstrating cases where rodents most likely act as mutualists and cases where granivory is unlikely to benefit the plant.