498 



Fishery Bulletin 97(3), 1999 



to be biologically significant when the value exceeds 

 0.60(Keast, 1978; MacPherson, 1981; Wallace, 1981; 

 Langton, 1982). Bias-corrected bootstrap 95% confi- 

 dence intervals, based on 1000 simulations, were 

 used to estimate the reliability of these indices (Efron 

 and Tibshirani, 1986; Hall et al., 1990). Differences 

 in diet composition and stomach fullness by fish size 

 and month were tested by x^ on combined data for 

 both years because diet did not vary significantly 

 between the two years. The logit model used to in- 

 vestigate monthly variations in the proportion of full 

 and empty stomachs, as a function offish size was 



log ipjp^i - 1 ) = 



',/ + ^d^ + ^ 



where p^ = the probability that a stomach is 

 nonempty in month d; 

 L = the fish length; and 

 e = the error term. 



One-way AN OVAs were used to compare the mean 

 number, the mean weight, and the mean size of prey 

 items among the size classes and the a posteriori 

 Tukey's test was employed to locate the source of any 

 differences. Numbers and weights of prey items were 

 log-transformed to remove the dependency of the 

 variance on the mean (Zar, 1984). Mean number and 

 mean weight of prey per fish in each size class was 

 based only on specimens with food items in their 

 stomach. All statistical inferences were based on the 

 0.05 significance level. 



Results 



Distribution patterns 



Red porgy were found to occur to 250 m depth, be- 

 tween temperatures of 14.5 and 24.2"C and salini- 

 ties of 38.1 and 40.07m (Table 1). Abundance in the 

 trawlable fishing grounds was negatively related to 

 depth during all seasons and positively related to 

 temperature during summer and winter (Table 2). 

 No significant correlations between abundance and 

 temperature were observed during spring. 



The distribution of red porgy by depth intervals 

 indicated that although fish generally occur in all 

 zones, their abundance significantly increased in 

 shallow waters (zone I), with highest values during 

 summer (Fig. 2). Relative abundance was signifi- 

 cantly higher in the shallow zone, with a maximum 

 between 20 and 50 m depth for all seasons. Further- 

 more, the relative abundance increased progressively 

 from winter to a maximum of 2985.6 individuals per 

 nmi^ in summer. These differences of the relative 



abundance in the shallow zone were mainly due to 

 recruitment during summer. 



The relationships between mean FL and depth 

 were significantly positive in summer and winter 

 (Table 2). Maximum FL increases with depth only in 

 summer, whereas the minimum and the mean FL 

 increase during summer and winter. However, no 

 significant correlation between FL and temperature 

 was found (Table 2). Depth distribution by size re- 

 vealed that fish larger than 186 mm do not occur in 

 shallow waters, which were dominated by young fish 

 (Fig. 3). Moreover, individuals larger than 147 mm 

 were poorly represented at depths below 80 m. The 

 abundance of the size classes 0-2 was greatest in 

 warm waters during all seasons (Fig. 4). 



Feeding intensity 



Of the 634 stomachs of red porgy examined, 226 were 

 found to be empty (35.7%). The proportion of empty 

 stomachs varied significantly among the size classes 

 offish examined (x~=ll-78, P<0.001), with a maxi- 

 mum of 39.6% for the size class (Table 3). The logit 

 model revealed that the interaction between month 

 and fish size on the probability that a stomach is 

 nonempty was significant (F=2.07, P<0.01 ). The slope 

 was always >0, indicating a positive allometry on full 

 stomachs, whereas no differences on the estimated 

 probabilities were detected between the two years 

 (^test=-0.50, P>0.05). 



Composition of the diet 



There were at least 58 different prey species belong- 

 ing to four major groups (decapods, small crusta- 



