Role of predators in the early post-settlement demography of coral-reef fishes.

Populations with dispersive larvae are often demographically open such that local reproduction and subsequent larval settlement are not linked. Thus, understanding whether and how settlement patterns are established and subsequently modified is central to understanding local demography. Settlement is typically not measured directly, but rather it is estimated by recruitment, which is the observation of new individuals sometime after settlement. At Lizard Island on the Great Barrier Reef, I examined how patterns of recruitment of coral-reef fishes were modified across a range of natural recruit densities in the presence and absence of resident predators. Resident predators decreased recruitment and increased mortality for all species, but these effects varied considerably among species. The effects of predators on recruitment were at least partly due to mortality within 2 days after settlement. At their most extreme, predators caused recruitment failure of several species of butterflyfish. For one species of damselfish (Pomacentrus amboinensis), predators both induced weakly density-dependent mortality and obscured any relationship between recruitment and subsequent abundance, while for another damselfish (Neopomacentrus cyanomos), mortality was density-independent and subsequent abundance was a function of recruitment. These contrasting results may reflect differences in prey behavior. P. amboinensis tended to feed near or within the branches of coral inhabited by resident predators, while N. cyanomos tended to feed higher in the water column above the reefs, and thus farther away from resident predators. These results highlight the speed and extent to which patterns of settlement are modified, indicating that caution should be exercised when attributing patterns of recruitment to patterns of settlement. Tremendous between-species variation in how patterns of recruitment, and presumably settlement, were modified by predation indicates that generalizations or between-species extrapolations about the magnitude of these effects may be unwarranted.