Limnol. Oceanogr., 44(5), 1999, 1341–1347

To test whether larval selectivity at settlement contributes to distributional patterns of benthic infauna, we conducted three reciprocal sediment transplant experiments at 15-m-deep coarse-sand and muddy-sand sites (;3 km apart) on the continental shelf near Tuckerton, New Jersey. During 3to 5-d deployments in 1994, larvae of the surfclam, Spisula solidissima, selected coarse sand over muddy sand, and capitellid polychaetes selected muddy sand over coarse sand, regardless of site. Thus, larvae of both taxa selected sediments typical of adult habitats, displaying selectivity consistent with previous flume experiments. Settlement intensity changed significantly over the ;6-week period during which experiments were conducted. Several other taxa exhibited selectivity consistent with field distributions, and several were nonselective. Significant site differences in settlement intensity were also observed on some dates for several taxa. Plankton pump samples taken at the sites during tray experiments suggested that significant differences in supply of surfclam larvae over the 3-km scale contributed to between-site differences in settlement. In the later experiments, surfclams in experimental trays were larger than would be expected for recent settlers, suggesting that postsettlement migration and selectivity occurred on small scales. These experiments demonstrate that differences in larval supply over scales of kilometers and time scales of weeks can affect settlement intensity but that habitat selection by settling larvae of some species of soft-sediment invertebrates may set initial distribution patterns. Many studies have demonstrated spatially and temporally persistent sediment-associated patterns in marine benthos (reviewed by Snelgrove and Butman 1994), but the relative importance of larval supply and settlement in creating these patterns has been a subject of some debate (Butman 1987; Ólafsson et al. 1994; Snelgrove and Butman 1994). The potential importance of larval habitat selection was realized long ago in small-scale laboratory still-water experiments demonstrating that larvae delay or avoid settling in the absence of a particular sedimentary cue (e.g., Mortensen 1921). More recent laboratory flume experiments (e.g., Butman et al. 1988) have demonstrated that some species have the capacity for selectivity in realistic bottom flows. There is also evidence that small-scale flows affect initial settlement patterns (e.g., Eckman 1979; Butman and Grassle 1992; Snelgrove et al. 1993), and settlement therefore appears to be influenced by both behavioral and hydrodynamic processes. Several field studies have documented a relationship between densities of settled juveniles and larval abundance in the water column (e.g., Muus 1973; Cameron and Rumrill 1982), but most studies have been done over time scales of weeks or longer, thus confounding larval supply and postsettlement processes. It is inarguable that processes such as predation (e.g., Ambrose 1984), competition (e.g., Peterson 1977), and postsettlement redistribution (e.g., Emerson and Grant 1991) play important roles in structuring soft-sediment communities, but the potential importance of larval supply and habitat selection remains poorly known and is likely very important. A key problem is that flume studies have focused on settlement, and field studies have focused on recruitment. Thus, the importance of habitat selection could not be inferred from field studies, and it is unknown whether laboratory studies can be extrapolated to the field (e.g., Schneider et al. 1997). In the present study, we performed short-term reciprocal sediment transplants between two sites with different sediment composition and faunas to evaluate how larval supply and habitat selection may contribute to observed patterns. The settlement of two taxa, for which we had previously conducted laboratory flume studies, allowed us to draw inferences on the applicability of flume studies to field situations. Experiments were designed to determine the role of larval supply and selective behavior in establishing faunal distributions at the LEO-15 site (von Alt and Grassle 1992) on Beach Haven Ridge, a ;15-m-deep, 5-km-long by 1.5-kmwide ridge that is one of many such shore-oblique ridges along the coast of New Jersey. On the southern landward side of the ridge (39827.699N, 74815.819W) is a sandy habitat (f ; 1.0, Craghan unpubl. data), in which the surfclam, S. solidissima, is an abundant and conspicuous component of the fauna. On the northern landward side of the ridge (39829.299N, 74814.489W) is a muddy-sand habitat (f 5 3.7, Craghan unpubl. data) where polychaetes are dominant, and S. solidissima is rare. These sites will be referred to as the ‘‘sandy’’ and ‘‘muddy’’ sites, respectively. Faunal composition at the two sites was determined from 7-cm-diameter cores (surface area 5 38 cm2) pushed ;10 cm into the sediment by SCUBA divers in September 1994. Randomization of nine replicate cores was achieved using the hierarchical sampling approach of Morrisey et al. (1992). The bottom at the sandy site was rippled, so divers collected paired crest/trough samples rather than single cores, which resulted in a total of 18 samples. Samples were processed over a 300-mm sieve, fixed in buffered 10% formalin, and transferred to 80% ethanol with Rose Bengal. During the early summer of 1994, bulk sediment was collected from the sandy and muddy sites with a 0.04-m2 Van Veen grab. The upper 2–3 cm of sediment was retained from multiple grabs, frozen, washed with fresh water, and pushed through a 1-mm sieve to remove shell fragments and large, dead invertebrates. These sediments were refrozen and later