FISHERY BULLETIN: VOL. 85, NO. 1 



Table 3.— ANOVA results from predation experiments with pinfish 

 using the number of brown shrimp eaten per fish as the observa- 

 tion. Probability values are also listed from an ANOVA using the 

 weight of shrimp eaten as the observation. Turb/subs = 

 Turbidity/substratum. 



the results. Pinfish were voracious feeders eating 

 between 19.8% (turbid/sand) and 57.1% (clear/no 

 sand) of their body weight in shrimp over the 12-h 

 experimental period. Predation rates were probably 

 underestimated in clear-water treatments without 

 sand, since in three out of four of these compart- 

 ments the three pinfish ate all of the available 

 shrimp. 



The duration of the pinfish II experiment was 

 reduced to 6 h (0700-1300 h) to lower the overall 

 number of shrimp eaten by the predators. Similar 

 trends were apparent in the number of shrimp eaten 

 for each treatment combination (Fig. 2B), but the 

 interaction term (P = 0.082) in the ANOVA was not 



significant at the 0.05 level (Table 3). The size range 

 of the prey in the second experiment was slightly 

 larger than in pinfish I (Table 1), and variability in 

 the size of shrimp available or small differences in 

 size-selection may have affected our results. Using 

 the weight of shrimp eaten as the observation should 

 reduce this problem, and in this ANOVA (Table 3) 

 both turbidity and substratum were significant ef- 

 fects, but the F-test for interaction had a probabil- 

 ity value of 0.110. 



To increase the error degrees of freedom and 

 hence the power of the statistical test, the data from 

 both pinfish experiments were combined and ana- 

 lyzed. In one such ANOVA, day was considered to 

 be a blocking variable (no interaction with other fac- 

 tors), and the results on the number of shrimp eaten 

 were similar to those from the pinfish I experiment, 

 showing a significant interaction between turbidity 

 and substratum (P = 0.021). We also analyzed the 

 data in a completely randomized crossed design with 

 day as a main effect (Table 3). In this ANOVA the 

 turbidity/substratum interaction was highly signif- 

 icant, but the turbidity/day interaction was also 

 significant indicating that the effect of turbidity on 

 predation was less during the second experiment. 

 In addition to the shorter duration of pinfish II, 

 overall light levels were higher during this second 

 pinfish experiment (clear sunny day) compared with 

 the first experiment (overcast day) (Table 1). 



Atlantic Croaker 



Mean predation rates for Atlantic croaker were 

 highest in clear-water tanks without sand, and rates 

 in all turbid tanks were low (Table 4A). The ANOVAs 

 with both the number (Table 4B) and weight of 



Table 4.— Predation on brown shrimp by Atlantic croaker. A) Number of 

 shrimp eaten per fish over the 12-h experimental period for treatment com- 

 binations of turbidity and substratum. B) ANOVA results using the number 

 of shrimp eaten per fish as the observation. Probability values from an 

 ANCOVA using incident light as the covariate are also included. 



64 



