FISHERY BULLETIN: VOL. 85, NO. 1 



of shrimp eaten by a predator in a tank over the ex- 

 perimental period as the observation in a two-way 

 analysis of variance (ANOVA). A second ANOVA 

 was also performed on the weight of shrimp eaten 

 per fish. In experiments with pinfish and Atlantic 

 croaker, where both compartments within a tank 

 contained predators, observations from the two 

 compartments were considered to be within tank 

 replicates or subsamples. With southern flounder, 

 only one compartment was used in each tank. This 

 experiment was repeated on a second day, and day 

 was considered a blocking variable in the analysis. 

 Because differences in incident light among tanks 

 could potentially affect predation rates and increase 

 within treatment variability, an analysis of 

 covariance (ANCOVA) was also performed on the 

 data from all experiments using incident light as the 

 covariate. 



The size range of shrimp available to predators 

 was kept as narrow as possible (Table 1) to avoid 

 problems associated with size-selective predation. 

 In addition, we attempted to keep the distribution 

 of shrimp within this size range similar for all rep- 

 licates. The size-frequency distributions of shrimp 

 placed in the tanks and shrimp removed from the 

 tanks after each experiment were compared to 

 check for evidence of size-selective predation. 



Turbidity and Burrowing of 

 Brown Shrimp 



The effect of turbidity on burrowing by juvenile 

 brown shrimp was examined in eight rectangular 

 tanks each with a bottom area of 0.92 m^. Water 

 depth was maintained at 25 cm, and temperature 

 and salinity were adjusted to 25 °C and 25°/oo, 

 respectively. Light was provided through white 

 translucent skylights. Lengths of PVC pipe were in- 

 stalled along the walls on the bottom of each tank. 

 The tanks were filled with washed beach sand to a 

 depth of 5 cm, and the sand surface was approx- 

 imately 5 mm below the top of the PVC pipe. The 

 number of shrimp burrowed was determined using 

 a net composed of fiberglass screen mounted on a 

 wooden frame. The frame was the same width as 

 the tanks and was pushed over the PVC runners 

 along the bottom, passing just above the sand sur- 

 face. Shrimp caught in the net were assumed to be 

 in the water column or on the surface of the sub- 

 stratum. 



Ten brown shrimp (50-100 mm) were placed in 

 each tank on the day before an experiment. Before 

 sunrise on the day of the experiment, kaolinite was 

 added to four of the tanks through the settling 



bucket system. Airstones in all tanks provided 

 enough turbulence to keep the clay in suspension. 

 At 1100 h, turbidity and light levels were measured 

 in the center of the water column in each tank, and 

 nonburrowed shrimp were collected. The tanks were 

 then drained, and the burrowed shrimp were recov- 

 ered. The experiment was repeated with different 

 shrimp on a second day, and an ANOVA, with day 

 used as a blocking variable, was performed to test 

 for an effect of turbidity. The percentage of shrimp 

 burrowed in a tank was used as the observation after 

 an arcsin transformation. The accuracy of our col- 

 lecting technique was examined by comparing visual 

 observations of the number of shrimp burrowed in 

 the clear tanks with the catch in the net. All non- 

 burrowed shrimp were captured in six out of seven 

 trials, but one nonburrowed shrimp avoided capture. 

 In one trial, a burrowed shrimp was collected. 



RESULTS 



Predation Experiments 



Data from the two control compartments (one 

 turbid and one clear) used in each experiment in- 

 dicated that mortality of prey was low. Only 1.6% 

 of the 250 control shrimp were not recovered alive. 

 This mortality was considered negligible, and all 

 shrimp not recovered in predation experiments were 

 assumed eaten by predators. The use of a relative- 

 ly narrow size range of prey also appeared to 

 eliminate problems associated with size-selective 

 predation. Comparisons of size-frequency distribu- 

 tions of shrimp introduced into experimental com- 

 partments to those removed following the experi- 

 mental period showed no apparent size-selective 

 predation in any of the experiments. 



Southern Flounder 



Predation by southern flounder was highest in 

 tanks with turbid water and without sand substrata 

 (Table 2A). The interaction term in the ANOVA was 

 not significant, and both main effects of turbidity 

 and substratum were significant at the 0.05 level 

 (Table 2B). Predation rates of these fish increased 

 from a mean of 2.2 shrimp/fish in clear water to 4.4 

 shrimp/fish in turbid water. Predation rates were 

 reduced in the presence of sand from a mean of 4.8 

 shrimp/fish in tanks without sand to a mean of 1.9 

 shrimp/fish in tanks with sand. An ANCOVA with 

 incident light and an ANOVA using the weight of 

 shrimp eaten as the observation gave similar results. 

 The mean weight of shrimp eaten, expressed as a 



62 



