Towards a geography of omnivory: Omnivores increase carnivory when sodium is limiting.

Towards understanding the geography of omnivory, we tested three hypotheses that predict the proportion of animal tissue consumed: the sodium limitation hypothesis predicts that omnivores increase animal consumption in Na-poor environments because Na bioaccumulates from plants to predators; thus, heterotrophs are Na-rich sources. The nitrogen limitation and habitat productivity hypotheses use the same logic to predict more animal consumption in N-poor and productive environments respectively. Omnivory is a common trophic strategy, but what determines the balance of plant and animal tissue omnivores consume is relatively unexplored. Most of what we know comes from single populations at local scales. Here we quantitatively test these three hypotheses at a large geographic scale and across 20 species of omnivorous ants. We tested each hypothesis using N stable isotopes (δ N) to quantify the degree of carnivory in ant populations in 20 forests that span 12° latitude from Georgia to Maine, USA. We used the difference in δ N between 20 ant conspecifics in 10 genera between two paired forests (10 pairs of 20 forests) that consisted of a coastal and inland forests on the same latitude to determine if the proportion of animal tissue consumed could be predicted based on Na, N or net primary productivity. Sodium gradients accounted for 18% of the variation in δ N, 45% if one outlier ant species was omitted. In contrast, the nitrogen limitation and habitat productivity hypotheses, which predict more animal consumption in N-poor and more productive environments respectively, failed to vary with δ N. Our results reveal a geography of omnivory driven in part by access to Na.