No abstract

Scatterhoarding rodents often place caches in the open where pilferage rates are reduced, suggesting that they tradeoff higher risks of predation for more secure cache sites. We tested this hypothesis in two study systems by measuring predation risks inferred from measures of giving-up densities (GUDs) at known cache sites and other sites for comparison. Rodent GUDs were measured with small trays containing 3 L of fine sand mixed with sunflower seeds. In the first experiment, we relied on a 2-year seed dispersal study in a natural forest to identify caches of eastern gray squirrels (Sciurus carolinensis) and then measured GUDs at: (i) these caches; (ii) comparable points along logs and rocks where rodent activity was assumed highest; and (iii) a set of random points. We found that GUDs and, presumably, predation risks, were higher at both cache and random points than those with cover. At the second site, we measured GUDs of eastern gray squirrels in an open park system and found that GUDs were consistently lowest at the base of the tree compared to more open sites, where previous studies show caching by squirrels to be highest and pilferage rates by naïve competitors to be lowest. These results confirm that predation risks can influence scatterhoarding decisions but that they are also highly context dependent, and that the landscape of fear, now so well documented in the literature, could potentially shape the temporal and spatial patterns of seedling establishment and forest regeneration in systems where scatterhoarding is common.