Sediment selection by juvenile sea scallops (Placopecten magellanicus (gmelin)), sea stars (Asterias vulgaris verrill) and rock crabs (Cancer irroratus say).

ABSTRACT We examined sediment selection by juvenile sea scallops (Placopecten magellanicus) and their predators, sea stars (Asterias vulgaris) and rock crabs (Cancer irroratus). In laboratory trials, groups of scallops (~30 mm shell height) were simultaneously offered four sediment types in the presence and absence of a sea star or a rock crab: (1) glass representing a homogeneous, hard bottom; (2) sand; (3) granule and (4) pebble. As well, individual predators were offered the four sediment types without scallops. The number of scallops and the proportion of time predators spent on each sediment type were monitored over time. When compared: with the expected distribution, scallops avoided glass and tended to select granule and pebble sediments when alone. In the presence of a rock crab, scallops also avoided glass. However, scallop distribution was similar to the expected distribution when a sea star was present, because sea star encounters are an important trigger of scallop swimming, leading to frequent redistribution of scallops. Sea stars spent less time than expected on glass, whereas crabs spent more time than expected on sand. For both predators, distribution did not change significantly in the presence or absence of scallops. In sum, scallop distribution appears more dependent on predator distribution than the reverse. Predation of scallops by sea stars and rock crabs was not influenced by sediment type. Our results have implications for the bottom culture of scallops. Because scallops select heterogeneous sediments over homogeneous ones, dispersal of scallops may be important on unsuitable sediments. Also, dispersal may he higher when sea stars are present at an aquaculture site.

KEY WORDS: bottom culture; distribution, rock crab, Cancer irroratus; sea scallop, Placopecten magellanicus, sea star, Asterias vulgaris, sediment type, selection, substrate

INTRODUCTION

The bottom culture of sea scallops (Placopecten magellanicus), where juvenile scallops are released (seeded) onto the sea bottom and allowed to grow to commercial size, has gained considerable interest in the last two decades on the coast of the northwest Atlantic (Couturier et al. 1995). However, the success of this type of scallop aquaculture has been limited because large portions of seeded scallops (40% to 99%) often do not survive (Barbeau et al. 1996, Hatcher et al. 1996, Nadeau & Cliche 2004). The two main processes influencing loss of seeded scallops are predator-related mortality (specifically by predatory sea stars Asterias spp. and rock crabs Cancer irroratus in coastal areas of Atlantic Canada), and dispersal of seeded scallops away from the culture site (Barbeau et al. 1996, Wong et al. 2005). Both of these processes can be influenced by the physical characteristics of the bottom culture site, such as sediment type, water temperature and current velocity. In this study, we focus on sediment type. Certain sediment types may reduce a predator's searching and handling ability of prey, or provide prey with physical refuges that reduce detection by predators (Sponaugle & Lawton 1990, Arsenault & Himmelman 1996). Additionally, predators and prey may disperse from habitats where sediments interfere with feeding behaviors, chemoreception, or general movement (Sponaugle & Lawton 1990). Previously, Wong and Barbeau (2003) examined the effects of sediment type on predator-prey interactions when sea stars and rock crabs preyed on juvenile sea scallops. This previous study, while directly investigating effects of sediment type on predation, did not examine choice of sediment by predators or prey. If offered a number of sediment types simultaneously, sea stars, rock crabs and juvenile scallops may select a particular sediment type to which they will disperse and spend the majority of time there.

Field observations of sea star, rock crab and scallop abundances in a variety of habitats suggest that these animals select specific sediments. Juvenile sea scallops (P. magellanicus) (<50 mm shell height, SH), although rarely observed in the field, have been observed on smaller gravel sediments (2.00-15.00 mm diameter; Thouzeau et al. 1991), gravel and shells embedded in a silt matrix (Barbeau et al. 1996, Wong et al. 2005) or attached to substrates that provide vertical relief, such as hydrozoan branches or amphipod tubes (Larsen & Lee 1978). Adult sea scallops ([greater than or equal to] 50 mm shell height) are found on a variety of sediment types, but they are more commonly found on gravel (2.00-64.00 mm diameter) or gravelly-sand (0.0625-16.00 mm diameter) sediments (Langton & Uzmann 1989, Langton & Robinson 1990, Stokesbury & Himmelman 1995, Stokesbury 2002, Wong et al. 2005). Sea stars (Asterias vulgaris) and rock crabs (C. irroratus) are often observed on the same sediment types as adult and juvenile sea scallops (Wong et al. 2005). Sea stars tend to be more abundant on coarser sediments (Langton & Uzmann 1989, Himmelman & Dutil 1991), but are found on most sediment types ranging from mud to boulders (0.06 [micro]m-1024 mm). Rock crabs are also common on all sediment types (Scarratt & Lowe 1972, Bigford 1979, Drummond-Davis et al., 1982); those found on sand (0.0625-2.00 mm diameter) are often wholly or partially buried (Scarratt & Lowe 1972).

Dispersal from one sediment type to another requires a well-developed locomotory ability. Juvenile sea scallops are able to actively disperse and potentially select specific sediment types by swimming. Rapid clapping of the valves causes enclosed water to be forced out near the shell hinge and the scallop is propelled forward by jet propulsion. Scallops 11-80 mm in shell height (SH) are efficient swimmers and capable of extended horizontal flight (~0.5-3 m) (Dadswell & Weihs 1990, Carsen et al. 1996). Smaller scallops (<11 mm SH) tend to swim vertically up into the water column with little horizontal displacement (Manuel & Dadswell 1991). Large adult scallops (>100 mm SH) are limited by heavy shells, and can only shuffle for short distances along the seabed (Dadswell & Weihs 1990). Sea stars and rock crabs are highly mobile and likely have little difficulty dispersing to certain sediment types. Sea stars use tube feet to travel at moderate velocities (~2 cm [min.sup.-1]), whereas rock crabs use four pairs of walking legs to travel at relatively high velocities (~66 cm [min.sup.-1]) (Barbeau et al. 1994).

Sediment selection by juvenile sea scallops, sea stars and rock crabs is likely to have important consequences for the survival of seeded scallops during bottom culture. For example, seeding scallops onto a sediment type that enhances filter feeding and is avoided by sea stars and crabs may increase survival of scallops. The objective of our study was to determine the sediment selection by juvenile sea scallops, sea stars and rock crabs when offered four different sediment types simultaneously. A laboratory experiment was conducted in which the distribution of scallops on the four sediment types when predators were present and absent was quantified over time. Additionally, the proportion of time sea stars and rock crabs spent on each sediment type in the presence and absence of scallops was quantified.

MATERIALS AND METHODS

Experimental Materials

Sediment selection by juvenile sea scallops (Placopecten magellanicus), sea stars (Asterias vulgaris) and rock crabs (Cancer irroratus) was investigated from 29 July to 13 August 2001, using a laboratory experiment at the Hunstman Marine Science Center, St. Andrews, New Brunswick, Canada. Fiberglass circular tanks (measuring 120 cm in diameter and 27 cm high) were used. Flow-through sea water was sand filtered to 20 [micro]m and regulated between 0.6-2 L [min.sup.-l], the light regimen was 14 h light: 10 h dark, and water temperature ranged between 12[degrees]C and 14[degrees]C.

Juvenile sea scallops were obtained from Sea Perfect Cultivated Products, Arichat, Nova Scotia. They were held in 260 L tanks for six days before the experiment began. Scallops were fed algae paste (Innovative Aquaculture Products Ltd) prior to and during the experiments at a concentration of 1.0 x [10.sup.4] cells [mL.sup.-1] of water (Hollett & Dabinett 1989). Scallops were 31.2 [+ or -] 3.9 mm (mean [+ or -] S.D., n = 30) in shell height (SH), measured as the distance between the middle of the dorsal hinge to the furthest ventral edge of the shell.

All predators originated from Passamaquoddy Bay in the Bay of Fundy, New Brunswick. Sea stars were collected by trawling, and were held in small tanks (measuring 60 cm long x 30 cm wide x 30 cm high) for 3 wk before the experiments began. Only intact sea stars with all 5 arms were used. Sea stars were 213.5 [+ or -] 35.3 mm (mean [+ or -] S.D., n = 16) in diameter, measured as the distance between the tip of an average sized arm to the middle of the oral region, multiplied by two. Rock crabs were obtained by trawling or from local crab fishers, and were held in separate small tanks for 3 wk before the experiment began. Only male crabs with both chelae and all walking legs were used. Crabs were 108.4 [+ or -] 7.0 mm (mean [+ or -] SD, n = 16) in carapace width (CW), measured as the distance between the two most distal marginal teeth. To standardize hunger level, both predator species were fed blue mussels until 6 d prior to the experiment. Animals were only used once during the experimental trials.

Experimental Design

Trials of 12-h duration were conducted in which groups of scallops and/or individual sea stars and rock crabs were placed in tanks with four different sediment types: glass, sand, granule and pebble. Glass consisted of a flat, 0.625-cm thick Lexan sheet, and represented a homogeneous, hard substrate. Sand consisted of 21% granule and 79% sand, and granule was 12% sand, 29% pebble and 59% granule. Pebble was uniform with no sand or granule present, and ranged from 30 x l0 mm to 75 x 35 mm in particle size (Wong & Barbeau 2003). These sediment classifications follow Wentworth scale of particle sizes, where sand is 0.0625-2.00 mm, granule 2.01-4.00 mm, and pebble 4.01-64 mm in diameter (Wentworth 1922, Folk 1974). Sediments were allowed to soak overnight in seawater before each trial to allow a biofilm to accumulate. For each trial, sediment types were placed into specific, randomly chosen pie-shaped sections that prevented mixing of the sediment types using a separation apparatus (Wright 2002). Sediments were ~15 cm in depth. Replicates of the following treatments were randomly allocated into the tanks: scallops alone (group of 40), scallops with one sea star, scallops with one rock crab, one sea star alone, and one rock crab alone. Eight replicates of each treatment were conducted, for a total of 40 trials. In trials with a sea star or crab, the predator was placed on a randomly chosen sediment 12 h prior to the beginning of the trial. In trials with scallops, 10 scallops were released onto each sediment near the center of the tank and allowed to acclimate for 5 min before the start of the trial.

Collection of Data

To determine sediment selection by scallops, the number of scallops on each sediment type was noted at hourly intervals during the 12-h trial. For sea stars and rock crabs, the time spent on each sediment type was quantified during five continuous l-h observation periods, randomly chosen during the 12-h trial. The proportion of time predators spent on each sediment type was calculated as the total time spent on one sediment type divided by the total observation time. Animals were considered on a certain sediment type when [greater than or equal to] 50% of the body surface was in contact with that sediment. Sea stars were observed on the walls of the tank, and so this was designated as a fifth substrate ("other") in trials involving sea stars. If a scallop was consumed, it was immediately replaced in the center of the sediment type on which consumption took place.

Statistical Analyses

Scallop and predator data were analyzed separately. Sediment selection by scallops was investigated using two methods. In the first scallop analysis, a one-sample Hotelling [T.sup.2] test was conducted to determine if scallop distribution at the end of the trial (720 min) differed from the expected distribution of 10 scallops on each sediment type. F values were calculated as:

F = n - p/p(n - 1) [T.sup.2],

where n = number of replicates, p = number of sediment types and [T.sup.2] = calculated Hotelling [T.sup.2] statistic (Johnson & Wichern 1998). A separate Hotelling [T.sup.2] test was conducted for scallops alone, scallops with sea stars and scallops with rock crabs. If scallop distribution differed significantly from the expected distribution, 95% confidence intervals were calculated for each sediment type by:

[[bar.x].sub.i] [+ or -] [[[(1/n)[s.sup.2.sub.i](n - 1)p/n - p [F.sub.[alpha];p,n-p].sup.1/2],

where [[bar.x].sub.i] = mean number of scallops on sediment type i, and [s.sub.i.sup.2] = sample variance for sediment type i (Johnson & Wichern 1998). A sediment type contributed to the significant result when the confidence interval did not include the expected value (Roa 1992).

In the second scallop analysis, scallop distribution over time in the presence and absence of predators was analyzed using a split-plot multivariate analysis of variance (MANOVA), with predator presence (3 levels: no predator present, sea star present, crab present) as a fixed factor, time (13 levels from 0-12 h) as the split-plot factor, and tank (8 replicate tanks) as the random plot factor. Significant results were identified using Pillai's trace. This is a robust statistic that is not influenced by violation of multivariate normality, which could have resulted from our relatively small sample size (Scheiner 2001). When significant results were obtained, we conducted canonical analyses for the significant factors. We observed the squared canonical correlation of each canonical variate generated to determine the percentage of variance explained by the model. We also observed the magnitude and sign of the coefficients of the first standardized canonical variate to aid interpretation of significant main effects. Similar signs of coefficients indicate that dependent variables are positively correlated across treatments; in other words, selection for sediment changes in a similar direction across treatments. Magnitude of values indicates the relative contribution of each sediment type to the significant result (Scheiner 2001).

Sediment selection by predators was also investigated using two methods. In the first predator analysis, one-sample Hotelling [T.sup.2] tests were used to determine if the proportion of time predators spent on each sediment type was different from the expected distribution (i.e., sea star: 0.126 on each of the four sediments and 0.496 on "other"; rock crab: 0.25 on each sediment; calculated by dividing the area covered by the substrate by the total area available to the predator). Separate tests were conducted for predators alone and in the presence of scallops. In the second predator analysis, two-sample Hotelling [T.sup.2] tests were used to compare the proportion of time predators spent on each sediment type when scallops were present and when scallops were absent. F values for two-sample tests were calculated as:

F = ([n.sub.1 + [n.sub.2] - p - 1)[T.sup.2]/([n.sub.1] + [n.sub.2] - 2)p.

Significant results in above-mentioned Hotelling [T.sup.2] tests were investigated by calculating 95% confidence intervals as described for the scallop analyses. A two-sample t-test was also used to compare the proportion of time sea stars spent on vertical surfaces (i.e., "other" substrate) in the presence and absence of scallops.

One-sample Hotelling [T.sup.2] tests were also used to determine if predation rate on scallops on each sediment type by predators was different from the expected predation rate. Expected predation rate on each sediment type took into account the distribution of scallops; therefore, expected predation rate for a particular sediment type was calculated by multiplying the sum of observed predation rates over all sediment types (averaged over the 8 replicates) by the proportion of scallops on that sediment type at 720 min (averaged over the 8 replicates).

In most Hotelling [T.sup.2] tests conducted, linear dependence between columns of the data matrix existed because the data were restricted to total to 40 scallops (when analyzing distribution of scallops) or to 1 (when analyzing proportion of time predators spent on each sediment type). This linear dependence leads to an undefined inverse covariance matrix and the Hotelling [T.sup.2] statistic cannot be calculated (see Johnson & Wichern 1998 for matrix algebra). To alleviate this problem, we removed data for one sediment type that was similar to another sediment type (see Table captions for the exact data removed). Data were not removed from analyses of scallop distribution when sea stars and crabs were present, because some scallops were consumed and data were not restricted to total to 40. The assumption that data came from a normal multivariate population was tested using probability plots for data on each sediment type (Roa 1992). This assumption was not violated for any case. The assumption of equal covariance matrices between groups for two-sample Hotelling [T.sup.2] tests were tested using Cochran test for data on each sediment. For the MANOVA, equal covariance was determined when the sign of the correlation between dependent variables at each level of the treatment variables were the same (Scheiner 2001). In all cases, the assumption of equal covariance between treatment levels was not violated.

RESULTS

Sediment Selection by Scallops

At the end of the experiment, the number of scallops on each sediment type in the absence of a predator was significantly different from the expected distribution (10 scallops per sediment type) (Table 1; Fig. 1). The number of scallops was significantly lower than expected on glass (Table 1). When a rock crab was present, the number of scallops was also significantly lower than expected on glass (Table 1; Fig. 1). Scallop distribution was not different than expected when a sea star was present (Table 1; Fig. 1).

[FIGURE 1 OMITTED]

Scallop distribution on the different sediment types over the duration of the experiment was influenced by predator presence and time (Table 2; Fig. 2). When alone, the number of scallops was highest on granule and pebble and lowest on glass (Fig. 2). Scallops selected all sediment types over glass when crabs were present (Fig. 2), but did not select a specific sediment type when sea stars were present. Scallop distribution changed from the initial distribution to selected sediments after 60 rain when alone and after 180 min when a crab was present, and remained fairly stable for the rest of the trial. In contrast, scallop distribution when a sea star was present did not stabilize over the duration of the trial (Fig. 2). MANOVA generated four eigenvectors (i.e., squared canonical correlation) that explained 65.6% and 56.5% of the total variation in the model, for the predator presence and time effect, respectively. Of these, the first eigenvector explained 54.4% and 48.1% of the variation for predator presence and time effect, respectively. We interpreted the coefficients of the standardized canonical variates associated with these first eigenvectors (Scheiner 2001) (Table 2). For the predator effect, the number of scallops on granule and pebble contributed most to the significant MANOVA result, followed by sand and glass (Table 2). All coefficients had the same sign, indicating that the various sediments contributed in a similar direction to the predator effect (Table 2). For the time effect, sand and granule contributed most to the significant result. All coefficients had the same sign, and so the various sediments contributed in a similar direction to time effect.

[FIGURE 2 OMITTED]

Sediment Selection by Sea Stars

Sea stars spent significantly less time than expected on glass when alone and when scallops were present (Table 1; Fig. 3). When sea stars alone and sea stars in the presence of scallops were directly compared, the distribution did not change significantly (Table 3). In both situations, sea stars spent a large proportion of time on "other" substrate. However, sea stars spent less time on vertical surfaces in the presence of scallops than when alone ([t.sub.14] = 2.83, P = 0.013). Sea star predation rate on scallops on the 4 different sediments did not differ from the expected rate (Table 4, Fig. 4).

[FIGURES 3-4 OMITTED]

Sediment Selection by Rock Crabs

Rock crabs spent significantly more time than expected on sand, and so significantly less time than expected on the other sediment types (Table 1; Fig. 3). When on sand, crabs spent large proportions of the total observation time (~80%) wholly or partially buried in the sand with little activity. When crabs alone and crabs in the presence of scallops were directly compared, the distribution did not change (Table 3). Rock crab predation rate on scallops on the 4 different sediments did not differ from the expected rate (Table 4; Fig. 4).

DISCUSSION

Sediment Selection by Animals

Our study showed that juvenile sea scallops (Placopecten magellanicus, ~30 mm SH) selected specific sediment types. In the absence of a predator, scallops selected granule and pebble bottom over the other sediment types. Similar results were observed by Barbeau (unpublished data) when smaller juvenile scallops (9.0-10.4 mm SH) were offered glass bottom, sand, granule, and a pebble/cobble mixture simultaneously. Also, using pair-wise choice experiments, Bourgeois (2004) found that juvenile scallops (~25 mm SH) selected granule and a coarse sand/shell sediment over homogeneous sand. In our current experiment, sediment selection of scallops changed in the presence of sea stars (Asterias vulgaris). Scallops no longer avoided glass bottom; instead, scallops did not select any specific sediment type over others, and their distribution did not differ from the expected distribution of 10 scallops per sediment type. This probably resulted because juvenile sea scallops usually swim when encountered by sea stars (Barbeau & Scheibling 1994, Wong & Barbeau 2003), and so redistributed relatively frequently. Contrary to these results, sediment selection by scallops when rock crabs (Cancer irroratus) were present did not change from when scallops were alone, because scallops still avoided glass. Interestingly, scallops did not avoid sand when a crab was present, even though crabs clearly selected sand over other sediment types. Although crabs spent the majority of their time on sand buried, crabs did consume some scallops on this sediment. Previous experiments showed that rock crabs spend only a small percent of their time foraging (<25%), but that when they do forage, they are very effective at encountering, capturing and consuming scallops (Barbeau & Scheibling 1994, Wong & Barbeau 2003, Wong & Barbeau 2005). Scallops may not have avoided sand sediment when crabs were present because juvenile scallops tend to tightly close their valves upon encounter with crabs (which provides some protection against predation), instead of the typical swimming escape response used to evade sea stars (Barbeau & Scheibling 1994). Thus, the distribution of scallops in the presence of predators seems to be dependent on the type of antipredator strategy used by the prey.

In general, the distribution of predators did not depend on the presence or absence of scallops, because distributions did not change significantly between the two situations. Sea stars avoided glass bottom and rock crabs spent more time than expected on sand regardless of whether scallops were present or not. Sea stars did, however, spend less time on vertical surfaces when scallops were present than when they were absent; clearly, the presence of scallops induced sea stars to move to the sediments and forage. Because the selection for specific sediment types by predators was generally not influenced by the presence or absence of scallops, sediment choice by scallops was more likely to be dependent on the location of predators than vice-versa.

Our results are consistent with field observations of scallops, sea stars and rock crabs on specific sediment types. In the field, juvenile scallops (<50 mm SH) as well as adult scallops tend to be found on gravelly- or mixed sediments (Thouzeau et al. 1991, Barbeau et al. 1996, Wong et al. 2005), which we observed in our laboratory experiment. Juvenile scallops often move to gravelly sediments after the initial postlarval phase (spat) settle on various materials such as hydrozoans, amphipod tubes (Larsen & Lee 1978), shells of live P. magellanicus (Naidu 1970), shell fragments (Caddy 1968), bryozoans (Baird 1953, Caddy 1972) and red algae (Naidu 1970). By choosing sediments that provide vertical relief, spat may attain protection from shifting bottom sediments or from predation (Larsen & Lee 1978). Scallop species other than P. magellanicus also select specific sediment types when juveniles. For example, juvenile bay scallops (Argopecten irradians) attach to sea grass blades to reduce the probability of detection by predators (Pohle et al. 1991).

Our results for sea stars are also mostly consistent with field observations. Although found on most sediment types in the field, sea stars are usually observed on coarse sediments such as granule or cobble (Langton & Uzmann 1989). However, Himmelman & Dutil (1991) also observed high densities of sea stars (A. vulgaris and Leptasterias polaris) on bedrock. The selection of the vertical surface of the tank, especially when scallops were absent, has been observed during previous laboratory experiments (M. Wong, pers. obs.), and may be an experimental artifact. The burial behavior of rock crabs in sandy sediments has been observed in the field (Scarratt & Lowe 1978), and is well documented for other crab species, such as the dungeness crab (Cancer magister) and red rock crab (Cancer productus) (Bellwood 2002, McGaw 2005).

Sediment Selection and Predator-prey Interactions

In our study, we observed sea stars and crabs consuming scallops on all sediment types (except granule for sea stars). These results are comparable to those for similar-sized juvenile scallops (24-28 mm SH) in Wong and Barbean (2003), where sediment type did not affect sea star and rock crab predation rate. However, in Wong and Barbeau (2003), predation rate of smaller scallops (11-15 mm SH) by sea stars decreased with increasing particle size of sediments. If we had used such smaller scallops in our current study, we may also have detected an effect of sediment particle size on predation rate.

The effect of sediment type on predation of sea scallops by sea star and crab predators has been examined in field experiments in Baie des Chaleurs, Gulf of St. Lawrence, Canada (Stokesbury & Himmelman 1995). In these experiments, tethered scallops were used to quantify predation of juvenile sea scallops (35-45 mm SH) at sites with sand, gravel (composed of pebble and small cobble), or bedrock. Predator-related mortality of scallops was significantly higher at a site with sand than a site with gravel or bedrock. Furthermore, scallop shell remains indicated that ~50% to 80% of scallop predation was by decapods (Cancer irroratus, Homarus americanus, Hyas ssp.), whereas the remaining predation was by sea stars (Asterias vulgaris, Crossaster paposus, Leptasterias polaris). Because density of rock crabs was positively correlated with predator-related mortality of tethered scallops, and the highest density of these crabs (0.56 [+ or -] 0.43 individuals per 10 [m.sup.2]; mean [+ or -] SD) was at the sandy site, these results may be directly related to predator density and not to differential foraging abilities of predators on the various sediment types.

The complexity of sediment types, and not simply the size of sediment particles, can also influence predator-prey interactions. Talman et al. (2004) found that mortality of tethered New Zealand scallops (Pecten novaezandiae), caused by predation, decreased as habitat complexity increased. In this case, habitat complexity was defined in terms of biological features (e.g., diversity of biota) and physical features (e.g., number of sediment types, density, or structure). In our experiments, granule and pebble could be considered heterogeneous sediments because they provide physical crevices, whereas glass and sand are homogeneous sediments. However, we did not observe decreased predation on more complex sediment types, as observed for the New Zealand scallops. This may have resulted from the absence of biotic heterogeneity in our sediment types.

The lack of effect of sediment type on predation in our current study may be directly associated with the type of antipredator behavior used by sea scallops. As discussed in Wong and Barbeau (2003), there are two types of behaviors that prey can use to reduce predation risk: avoidance behaviors and escape behaviors (Sih 1987). Prey that use avoidance behaviors reduce the probability of detection and thus the probability of being encountered by predators, so these prey are likely to make use of features of the substrate to evade predators. However, prey that use escape behaviors after an encounter with a predator reduce the probability of being attacked or captured, and would be less likely to use substrate features. Indeed, juvenile sea scallops (P. magellanicus, [greater than or equal to] 11 mm SH) use active escape mechanisms by swimming horizontally through the water column, and do not actively seek refuge by hiding in crevices between sediment particles (as do Iceland scallops, Chlamys islandica; Arsenault & Himmelman 1996) or by attaching to sea grass blades (as do juvenile bay scallops; Argopecten irradians, Pohle et al. 1991). The lack of effect of sediment type may also be related to predator foraging behaviors on the different sediments. Specific behaviors, such as searching for prey, handling prey and encounters between predators and prey would require quantification to further investigate this (as done in Wong & Barbeau 2003).

Implications for Future Experiments and Aquaculture

The experimental design used in our study allowed comprehensive examination of sediment selection by juvenile sea scallops, sea stars and rock crabs. Observed distributions of animals were compared with expected distributions. Further, distributions when animals were alone were compared with distributions when they were in the presence of a predator or prey. While we were able to identify selection of sediment types, our experimental design did not evaluate preference for sediment types. Preference is an active behavioral choice, and can only be identified if it is not confounded with "accessibility" of that choice (Underwood et al. 2004). Accessibility in our case refers to the ease with which a sediment type can be found or occupied (Olabarria et al. 2002). For example, scallops may be found mostly on a pebble sediment because they have difficulty swimming away from the cracks and crevices between sediment particles, and not because they actively chose this particular sediment type over others. The experimental design necessary to identify preference would include treatments that allow animals to choose between sediment types, as well as treatments that do not allow a choice of sediments (Olabarria et al. 2002, Barbeau et al. 2004). In our experiments, additional tanks where each divided segment is filled with the same sediment type would need to be included (i.e., replicated treatments would include only glass, sand, granule or pebble sediment). Preference would be evident if animals chose a particular sediment more often in the choice situation (i.e., in choice treatments) than expected by chance when no choice is available (i.e., in no-choice treatments). Thus, the observed association of animals on particular sediment types in our experiments indicates selection of that sediment, but not necessarily an active behavioral choice.

Our results have implications for the bottom culture of scallops, and suggest that aquaculturists should carefully consider the sea bottom characteristics of the site before seeding scallops. Juvenile scallops select more heterogeneous sediment types (e.g., granule or pebble) over more homogeneous substrates (sand or flat, hard bottoms). Therefore, dispersal of scallops away from a site will be higher on less suitable sediment types. This has been observed in early seeding trials off the Magdalen Islands, Quebec, Canada, where sea scallops were released on sand (Picard & Vigneau 1992). Dispersal from certain sediment types has been observed in other scallop species, such as bay scallops (Argopecten irradians), which swim from an unnatural sediment (sand) to their normal substrate (eelgrass blades) (Winter & Hamilton 1985), and great scallops (Pecten maximus), which have a higher probability of swimming from bedrock than from sediments suitable for recessing (Baird 1958, Baird 1966). Furthermore, in our system, sea scallop dispersal will be higher if sea stars are present, because sea star encounter seems to be a primary trigger for scallop swimming. Encounter with crab predators may also trigger swimming of sea scallops, but not to the extent of encounters with sea stars. Note, however, that crabs can be much more effective scallop predators than sea stars (Barbeau & Scheibling 1994). Based on research with another scallop species (Talman et al. 2004), the foraging efficiency of both predator types may be reduced in habitats with high structural complexity (e.g., high density of macrophytes). Additionally, the sediment type of an aquaculture site should be considered in terms of scallop growth. Brand (1991) reported faster growth of various scallop species (e.g., P. megallanicus, A. irradians, P. maximus) on sediments with little silt or mud. Thus, seeding sea scallops onto sediments with large particle sizes may lead to faster growth. Generally, sediment type remains an important factor that can influence the success of bottom culture operations by affecting predator-prey interactions, scallop movement, and scallop growth.

ACKNOWLEDGMENTS

The authors thank M. Dowd and an anonymous reviewer for useful comments. Huntsman Marine Science Centre provided laboratory facilities. M.C.W. was supported by the Network of Centres of Excellence for Mathematics of Information Technology and Complex Systems (MITACS) of Canada, and L.D.W. by the Summer Career Placement Program (Human Resources and Development, Canada). Research was funded by a Discovery grant to M. A. B. from the Natural Sciences and Engineering Research Council of Canada.

LITERATURE CITED

Arsenault, D. J. & J. H. Himmelman. 1996. Size-related changes in vulnerability to predators and spatial refuge use by juvenile Iceland scallops Chlamys islandica. Mar. Ecol. Prog. Ser. 140:115-122.

Baird, F. T. 1953. Observations on the early life history of the giant scallop (Pecten magellanicus). Maine Dept. Seashore Fish. Res. Bull. 14:2-7.

Baird, R. H. 1958. On the swimming behaviour of scallops (Pecten maximus L.). Proc. Malac. Soc. Lond. 33: 67-71.

Baird, R. H. 1966. Notes on an escallop (Pecten maximus) population in Holyhead Harbour. J. Mar. Biol. Assoc. U.K. 46:33-47.

Barbeau, M. A., K. Durelle & R. B. Aiken. 2004. A design for multifactorial choice experiments: an example using microhabitat selection by sea slugs Onchidoris bilamellata (L.). J. Exp. Mar. Biol. Ecol. 307:1-16.

Barbeau, M. A. & R. E. Scheibling. 1994. Behavioural mechanisms of prey size selection by sea stars (Asterias vulgaris Verrill) and crabs (Cancer irroratus Say) preying on juvenile sea scallops (Placopecten magellanicus Gmelin). J. Exp. Mar. Biol. Ecol. 180:103-136.

Barbean, M. A., B. G. Hatcher, R. E. Scheibling, A. W. Hennigar, L. H. Taylor & A.C. Risk. 1996. Dynamics of juvenile sea scallop (Placopecten magellanicus) and their predators in bottom seeding trials in Lunenburg Bay, Nova Scotia. Can. J. Fish. Aquat. Sci. 53:2494-2512.

Barbeau, M. A., R. E. Scheibling, B. G. Hatcher, L. H. Taylor & A. W. Hennigar. 1994. Survival analysis of tethered juvenile sea scallops Placopecten magellanicus in field experiments: effects of predators, scallop size and density, site and season. Mar. Ecol. Prog. Ser. 115: 243-256.

Bellwood, O. 2002. The occurrence, mechanics and significance of burying behaviour in crabs (Crustacea: Brachyura). J. Nat. Hist. 36:1223-1238.

Bigford, T. E. 1979. Synopsis of biological data on the rock crab, Cancer irroratus Say. NOAA Technical Report 426:1-26.

Bourgeois, M. 2004. Effet du substrat sur la survie, la croissance et la dispersion du petoncle geant juvenile (Placopecten magellanicus). M.Sc. thesis. Rimouski, Quebec, Canada: Universite du Quebec Rimouski. 150 pp.

Brand, A. R. 1991. Scallop ecology: distributions and behaviour. In: S. E. Shumway, editor. Scallops: biology, ecology and aquaculture. Amsterdam: Elsevier, pp. 517-584.

Caddy, J. F. 1968. Underwater observations on scallop (Placopecten magellanicus) behaviour and drag efficiency. J. Fish. Res. Bd Can. 25: 2123-2141.

Caddy, J. F. 1972. Progressive loss of byssus attachment with size in the sea scallop, Placopecten magellanicus (G.). J. Exp. Mar. Biol. Ecol. 9:179-190.

Carsen, A.E., B.G. Hatcher & R.E. Scheibling. 1996. Effect of flow velocity and body size on swimming trajectories of sea scallops, Placopecten magellanicus (Gmelin): a comparison of laboratory and field measurements. J. Exp. Mar. Biol. Ecol. 203:223-243.

Couturier, C., P. Dabinett & M. Lanteigne. 1995. Scallop culture in Atlantic Canada. In: A.D. Boghen, editor. Cold-Water Aquaculture in Atlantic Canada. Moncton, NB: Canadian Institute for Research on Regional Development, pp. 297-340.

Dadswell, M. J. & D. Weibs. 1990. Size-related hydrodynamic characteristics of the giant scallop, Placopecten magellanicus (Bivalvia: Pectinidae). Can. J. Zool. 68:778-785.

Drummond-Davis, N. C., K. H. Mann & R. A. Pottle. 1982. Some estimates of population density and feeding habits of the rock crab, Cancer irroratus, in a kelp bed in Nova Scotia. Can. J. Fish. Aquat. Sci. 39:636-639.

Folk, R. L. 1974. Petrology of Sedimentary Rocks. Austin, TX: Hemphill Publishing Company. 184 pp.

Hatcher, B. G., R.E. Scheibling, M. A. Barbeau, A. W. Hennigar, L. H. Taylor & A. Windust. 1996. Dispersion and mortality of a population of sea scallop (Placopecten magellanicus) seeded in a tidal channel. Can. J. Fish. Aquat. Sci. 53:38-54.

Himmelman, J. H. & C. Dutil. 1991. Distribution, population structure and feeding of subtidal seastars in the nothern Gulf of St. Lawrence. Mar. Ecol. Prog. Ser. 76:61-72.

Hollett, J. & P. E. Dabinett. 1989. Effect of ration on growth and growth efficiency of spat of the giant scallop, Placopecten magellanicus (Gmelin). Aqua. Assoc. Can. Bull. 89:71-73.

Johnson, R.A. & D. W. Wichern. 1998. Applied multivariate statistical analysis. Upper Saddle River, N.J.: Prentice-Hall. 816 pp.

Langton, R. W. & W. E. Robinson. 1990. Faunal associations on scallop grounds in the western Gulf of Maine. J. Exp. Mar. Biol. Ecol. 144: 157-171.

Langton, R.W. & J.R. Uzmann. 1989. A photographic survey of the megafauna of the central and eastern Gulf of Maine. Fish. Bull. (Wash. DC) 87:945-951.

Larsen, P. R. & R. M. Lee. 1978. Observations on the abundance, distribution and growth of postlarval sea scallops, Placopecten magellanicus, on Georges Bank. Nautilus 92:112-116.

Manuel, J. L. & M. J. Dadswell. 1991. Swimming behavior of juvenile giant scallop, Placopecten magellanicus, in relation to size and temperature. Can. J. Zool. 69:2250-2254.

McGaw, I.J. 2005. Burying behaviour of two sympatric crab species: Cancer magister and Cancer productus. Sci. Mar. 69:375-381.

Nadeau, M. & G. Cliche. 2004. Evaluation de la survie des petoncles geants (Placopecten magellanicus) juveniles lors d'ensemencements exprrimentaux menes aux Iles-de-la-Madeleine. Ministere de l'Agriculture, des Pecheries et de l'Alimentation. Cahier d'information 140:28 pp.

Naidu, K. S. 1970. Reproduction and breeding cycle of the giant scallop Placopecten magellanicus (G.) in Port au Port Bay, Newfoundland. Can. J. Zool. 48:1003-1012.

Olabarria, C., A.J. Underwood & M.G. Chapman. 2002. Appropriate experimental design to evaluate preferences for microhabitat: an example of preferences by species of microgastropods. Oecologia 132: 159-166.

Pohle, D.G., V.M. Bricelj & Z. Garcia-Esquivel. 1991. The eelgrass canopy: an above-bottom refuge from benthic predators for juvenile bay scallops Argopecten irradians. Mar. Ecol. Prog. Ser. 74:47-59.

Picard, P. & S. Vigneau. 1992. Essais d'ensemencement de pttoncle geant sur les fonds a potentiel pectinicole aux Iles-de-la-Madeleine. Rapport annuel, Roche Ltte Groupe-Conseil, Sainte-Foy, Quebec, Canada.

Roa, R. 1992. Design and analysis of multiple-choice feeding-preference experiments. Oecologia 89:509-515.

Scarratt, D. J. & R. Lowe. 1972. Biology of rock crab (Cancer irroratus) in the Northumberland Strait. J. Fish. Res. Board Can. 29:161-166.

Scheiner, S. M. 2001. MANOVA: multiple response variables and multi-species interactions. In: S. M. Scheiner & J. Gurevitch, editors. Design and analysis of ecological experiments. New York, NY: Oxford University Press. pp. 99-115.

Sih, A. 1987. Predators and prey lifestyles: an evolutionary and ecological overview. In: W. C. Kerfoot & A. Sih, editors. Predation: direct and indirect impacts on aquatic communities. London: University Press of New England London. pp. 203-224.

Sponaugle, S. & P. Lawton. 1990. Portunid crab predation on juvenile hard clams: effects of substrate type and prey density. Mar. Ecol. Prog. Ser. 67:43-53.

Stokesbury, K. D. E. & J. H. Himmelman. 1995. Biological and physical variables associated with aggregations of the giant scallop Placopecten magellanicus. Can. J. Fish. Aquat. Sci. 52:743-753.

Stokesbury, K. D. E. 2002. Estimation of sea scallop abundance in closed areas of Georges Bank, USA. Trans. Am. Fish. Soc. 131:1081-1092.

Talman, S.G., A. Norkko, S.F. Thrush & J.E. Hewitt. 2004. Habitat structure and the survival of juvenile scallops Pecten novaezelandiae: comparing predation in habitats with varying complexity. Mar. Ecol. Prog. Ser. 269:197-207.

Thouzeau, G., G. Robert & S. J. Smith. 1991. Spatial variability in distribution and growth of juvenile and adult sea scallops Placopecten magellanicus (Gmelin) on eastern Georges Bank (Northwest Atlantic). Mar. Ecol. Prog. Ser. 74:205-218.

Underwood, A. J., M. G. Chapman & T. P. Crowe. 2004. Identifying and understanding ecological preferences for habitat or prey. J. Exp. Mar. Biol. Ecol. 300:161-187.

Wentworth, C. K. 1922. A scale of grade and class terms for clastic sediments. J. Geol. 30:377-392.

Winter, M. A. & P. V. Hamilton. 1985. Factors influencing swimming in bay scallops, Argopecten irradians (Lamarch, 1819). J. Exp. Mar. Biol. Ecol. 88:227-242.

Wong, M. C. & M. A. Barbeau. 2003. Effects of substrate on interactions between juvenile sea scallops (Placopecten magellanicus (Gmelin)) and predatory sea stars (Asterias vulgaris Verrill) and rock crabs (Cancer irroratus Say). J. Exp. Mar. Biol. Ecol. 287:155-178.

Wong, M. C. & M. A. Barbeau. 2005. Prey selection and the functional response of sea stars (Asterias vulgaris Verrill) and rock crabs (Cancer irroratus Say) preying on juvenile sea scallops (Placopecten magellanicus (Gmelin)) and blue mussels (Mytilus edulis Linnaeus). J. Exp. Mar. Biol. Ecol. 327:1-12.

Wong, M. C., M. A. Barbeau, A. W. Hennigar & S. M. C. Robinson. 2005. Protective refuges for seeded juvenile scallops (Placopecten magellanicus) from sea star (Asterias spp.) and crab (Cancer irroratus and Carcinus maenas) predation. Can. J. Fish. Aquat. Sci. 62:1766-1781.

Wright, L. D. 2002. Substrate preferences of juvenile sea scallops (Placopecten magellanicus), sea stars (Asterias vulgaris), and rock crabs (Cancer irroratus). B.Sc. honours thesis. Fredericton, NB: University of New Brunswick. 41 pp.

MELISA C. WONG, * (1) LISA D. WRIGHT AND MYRIAM A. BARBEAU

Department of Biology, University of New Brunswick, Fredericton, New Brunswick, E3B 7B8 Canada

* Corresponding author. E-mail: [e-mail]

(1) Current address: Institute of Marine Sciences University of North Carolina--Chapel Hill, Morehead City, North Carolina 28557.

TABLE 1.
Results of one-sample Hotelling [T.sup.2] tests for the number of
scallops (Placopecten magellanicus) on each sediment type at the end
of the experiment (720 min) and the total proportion of time sea stars
(Asterias vulgaris) and rock crabs (Cancer irro ratus) spent on each
sediment type compared to the expected distributions. Expected
distribution for scallops = 10 scallops on each of glass, sand,
granule, pebble; sea stars = 0.126 on each of glass, sand, granule,
pebble, and 0.494 on "other"; rock crabs = 0.25 on each of glass,
sand, granule, pebble. For scallop distribution when alone, linear
dependency in the data matrix was corrected by removing data on
granule. Data on granule and on pebble were removed for the sea star
and rock crab analyses, respectively.

   Analysis      Treatment    [T.sup.2]   [F.sub.df1,df2]       p

Scallop          Alone        75.51       17.973              0.004#
  distribution
                 With sea     28.02        [4.003.sub.4.4]    0.104
                   star
                 With rock    103.1       [14.72.sub.4,4]     0.012#
                   crab

Sea star         Alone        144.7       [20.67.sub.4,4]     0.006#
  distribution
                 With         352.5       [50.34.sub.4,4]     0.001#
                   scallops

Rock crab        Alone        243.2       [57.90.sub.3,5]    <0.001#
  distribution
                 With         388.4       [92.48.sub.3,5]    <0.001#
                   scallops

   Analysis      Treatment    Mean [+ or -] 95 % CI

Scallop          Alone        Glass: -2.095 [less than or equal to]
  distribution                  2.875 [less than or equal to] 7.845 *
                              Sand: 6.263 [less than or equal to]
                                9.375 [less than or equal to] 12.49
                              Pebble: 9.819 [less than or equal to]
                                14.50 [less than or equal to] 19.18
                 With sea
                   star
                 With rock    Glass: -2.945 [less than or equal to]
                   crab         3.500 [less than or equal to] 9.945 *
                              Sand: 1.544 [less than or equal to]
                                12.25 [less than or equal to] 22.96
                              Granule: -2.469 [less than or equal to]
                                12.37 [less than or equal to] 27.22
                              Pebble: 1.678 [less than or equal to]
                                11.63 [less than or equal to] 21.57
Sea star         Alone        Glass: 0 [less than or equal to] 0.020
  distribution                  [less than or equal to] 0.091
                              Sand: 0 [less than or equal to] 0.047
                                [less than or equal to] 0.157
                              Pebble: 0 [less than or equal to] 0.062
                                [less than or equal to] 0.376
                              Other: 0.282 [less than or equal to]
                                0.804 [less than or equal to] 1.00
                 With         Glass: 0 [less than or equal to] 0.016
                   scallops     [less than or equal to] 0.075 *
                              Sand: 0 [less than or equal to] 0.114
                                [less than or equal to] 0.555
                              Pebble: 0 [less than or equal to] 0.361
                                [less than or equal to] 1.00
                              Other: 0 [less than or equal to] 0.381
                                [less than or equal to] 1.00
Rock crab        Alone        Glass: 0 [less than or equal to] 0.049
  distribution                  [less than or equal to] 0.117 *
                              Sand: 0.536 [less than or equal to]
                                0.826 [less than or equal to] 1.00 *
                              Granule: 0 [less than or equal to] 0.058
                                [less than or equal to] 0.166 *
                 With         Glass: 0 [less than or equal to] 0.043
                   scallops     [less than or equal to] 0.114 *
                              Sand: 0.503 [less than or equal to]
                                0.829 [less than or equal to] 1.00 *
                              Granule: 0 [less than or equal to] 0.086
                                [less than or equal to] 0.249 *

df1 = numerator df; df2 = denominator df; CI = confidence interval.

Significant results are indicated in bold, and sediment types that
contributed to the significant result (based on CI calculations) are
indicated with an asterisk.

Note: Significant results are indicated in bold indicated with #.

TABLE 2.
Split-plot MANOVA results for distribution of scallops (Placopecten
magellanicus) over time in the presence and absence of a predator
(Asterias vulgaris, Cancer irroratus).

Source of
Variation   Error Term    [F.sub.df1,df2]         p      Glass

P           M (P)        [2.32.sub.8,38]       0.039#   -0.962
T           T x M (P)    [3.54.sub.48,1008]   -0.001#    0.887
P x T       T x M (P)    [1.23.sub.96,1008]    0.075

Source of
Variation
             Sand    Granule   Pebble
P
T           -1.267   -1.458    -1.464
P x T        1.134    1.130     0.945

P = predator presence, T = time, M = tank. F = F-ratio for Pillai
trace. Significant results are indicated in bold. Coefficients of
the first standardized canonical variate are provided to aid
interpretation of significant main effects. Similar signs of
coefficients indicate a positive correlation between sediment
types, and magnitude of values indicates the relative contribution
of each sediment type to the significant result.

Note: Significant results are indicated in bold.

TABLE 3. Results for two-sample Hotelling [T.sup.2] tests for the
distribution of sea stars (Asterias vulgaris) and rock crabs (Cancer
irroratus) in the absence and presence of scallops (Placopecten
magellanicus). Data on granule and on pebble were removed for sea
star and rock crab analyses, respectively. df1 = numerator df,
df2 = denominator df.

Analysis    [T.sup.2]   [F.sub.df1,df2]        P

Sea star    12.28       [2.411.sub.4,11]   0.112
Rock Crab    0.016      [0.005.sub.3,12]   0.997

TABLE 4.
Results for one-sample Hotelling [T.sup.2] tests for predation rate
(number of scallops consumed over 12 h) of sea stars (Asterias
vulgaris) and rock crabs (Cancer irroratus) on each sediment type,
compared to the expected number of scallops consumed (sea stars:
glass = 0.230, sand = 0.420, pebble = 0.270; rock crabs: glass =
0.374, sand = 1.323, granule = 1.323, pebble = 1.243). For sea stars,
data on granule were removed to correct for the linear dependency in
the data. df1 = numerator df, df2 = denominator df.

Analysis    [T.sup.2]   [F.sub.df1,df2]     P

Sea star      3.000     [0.71.sub.3,5]    0.584

Rock crab     9.135     [1.31.sub.4,4]    0.401