Seedling recruitment in Oryzopsis hymenoides: Are desert granivores mutualists or predators?

Granivorous animals that cache as well as consume seeds may actually serve as mutualists to their plant resources. Seeds of Indian ricegrass (Oryzopsi s hymenoides), a perennial bunchgrass in North American deserts, are consum ed by various desert granivores and dispersed by seed-caching heteromyid ro dents. We used a three-way factorial design at a western Nevada site to sel ectively exclude or allow access to experimental plots by granivorous roden ts and seed harvester ants, and to subsequently follow the fate of radiolab eled Indian ricegrass seeds introduced to the plots. In addition to the pre sence or absence of rodents or ants, the third experimental treatment facto r was to allow "initial caching" of the radiolabeled seeds by single Merria m's kangaroo rats (Dipodomys merriami), which were confined to certain plot s for one night. Both rodents and ants larder-hoarded seeds in their burrow s, but seedlings rarely established from larders. Only rodents also placed seeds in scatterhoards: shallowly buried surface caches distributed about a n animal's home range. Following initial caching by a kangaroo rat, the num ber of seedlings established from scatterhoards was significantly greater ( usually by more than an order of magnitude) than those from seeds unharvest ed by either type of granivore. With no initial caching, rodents reduced se edling recruitment from unharvested seeds but facilitated compensatory seed ling recruitment from scatterhoards. Seeds harvested by ants seldom establi shed seedlings. We used a seed fate model to estimate that, on average, rod ents and ants harvested 96% and 7%. respectively, of seeds to which they ha d exclusive access, and that the probability of seedling establishment for a seed harvested by a rodent was an order of magnitude greater than from a seed harvested by an ant and slightly greater than for an unharvested seed. The predicted rank order of seedling recruitment among nine experimental t reatments based on expected effects of rodent seed caching closely matched the observed ranking pattern, indicating that rodents determined seedling r ecruitment patterns of Indian ricegrass. Because rodents harvested such a l arge majority of seeds and their caches enhanced seedling establishment., t hey played a central role in the population dynamics of Indian ricegrass.