Quantifying site quality in a heterogeneous landscape: recruitment of a reef fish.

Heterogeneity in site quality can play an important role in patterns of abundance and population dynamics. Yet, estimating site quality in natural systems can be problematic because site quality can (1) vary through ontogeny for a focal organism, leading to shifts in site quality with age, (2) be confounded with (or masked by) variation in traits of individuals populating the sites, and (3) be correlated with local density. For example, if high-quality sites attract more individuals but vital rates are density dependent, then observed vital rates will be relatively homogeneous in space despite strong heterogeneity in site quality. Here, we operationally define site quality for a reef fish as the mean survival time of juveniles transplanted to sites at a common density and size structure, with random assignment of individuals to sites to remove potential confounding effects of local variation in individual quality and density. Our assays using juvenile age classes of the six-bar wrasse (Thalassoma hardwicke) showed that site quality varied in space (i.e., among patch reefs) but was constant through time. Site quality increased with availability of the branching coral Pocillopora (which is used as a refuge), but decreased with density of a predator, the arc-eye hawkfish, Paracirrhites arcatus (which also uses Pocillopora). We experimentally added colonies of Pocillopora to reefs and (1) increased site quality, (2) enhanced natural settlement rates of six-bar wrasse, but (3) attracted more hawkfish predators, and (4) did not increase survival of juvenile fish under ambient densities. Our results suggest that Pocillopora increases site quality, but attracts greater densities of settlers and predators, resulting in increased density dependence and predation, which mask the underlying effects of Pocillopora on site quality (supporting the hypothesis of "cryptic density dependence"). Variation in site quality and the possible confounding effects of density and individual traits warrant more experimental study.