This study examines direct and indirect interactions that influence the distribution of larvae of the bullfrog (Rana catesbeiana) and green frog (R. clamitans) along the environmental gradient of permanent to temporary ponds. The bullfrog is found in permanent ponds that typically contain fish, whereas the green frog is widely distributed along the gradient and is most successful in ponds where the bullfrog is absent. In a set of experimental ponds, bullfrogs were most abundant in ponds containing fish (bluegill, Lepomis macrochirus) and were rare in permanent ponds that lacked fish. In contrast, green frogs were most abundant in ponds that had been drained the previous fall, second most abundant in those that lacked fish, and sparse in ponds containing bluegill. When presence and absence of bluegill were experimentally manipulated in divided ponds, bullfrogs survived well in the presence of bluegill, whereas no individuals survived in the absence of bluegill. Green frogs survived in low numbers under both treatments with a tendency to better survivorship in the absence of bluegill. A series of laboratory choice experiments was conducted with three major predator types: the dragonfly larva Anax junius, the salamander Ambystoma tigrinum, and L. macrochirus. Bullfrog larvae were more vulnerable to Anax and Ambystoma than were green frogs (because of higher activity levels), and green frogs were more vulnerable to bluegill than were bullfrogs (because tadpoles were less noxious). We argue that the high densities of bullfrogs found in the presence of bluegill reflect an indirect facilitation through two pathways: bluegill had strong negative effects on invertebrate and salamander predators of bullfrogs and on the green frog, which is a bullfrog competitor. The direct and indirect effects of these suites of predators appear to explain the differences in species abundances along the environmental gradient. Finally, we discuss the trade-offs at the individual level contributing to these differences in species performance along the gradient.
