958 



Fishery Bulletin 97(4), 1999 



0.20 -1 



0.18 



to 



■D 



^ 0.16 



E 



5 

 2 

 CD 



0.14 



12 



1 



n Food Addition 

 Control 



Habitat P = 09 

 Food P = 61 



Grass 



Sand 



Figure 3 



Growth rate (mean +1 SEi determined from 

 analysis of otolith microstructure in croaker 

 from l-m- artificial seagrass (grass) and sand 

 habitats with (food addition) and without (con- 

 trol) food supplementation. P values are from 

 a blocked two-factor analysis of variance, and 

 n, the number offish sampled, is given at the 

 base of each bar. 



however, given the small difference in average 

 croaker density between food-supplementation and 

 control plots (1.1/m-) and the high within-treatment 

 variation, we would have needed 55 replicates to 

 achieve sufficient power (1-/^=0.95) to accept the null 

 hypothesis of no difference between treatment means. 

 Growth rates of newly recruited croaker were the 

 same in sand and grass habitats {F-^ ^0=2. 79, a=0.09, 

 /3<0.001 ), as well as with or without food supplemen- 

 tation (Fi^„=0.26, a=0.61, /j=0.004) (Fig. 3). The in- 

 teraction between habitat and food-supplementation 

 on the growth rate of newly recruited croaker was 

 not significant (Fj 7,1=0.49, a=0.49). Growth averaged 

 0.148 mm SL/day (SE=0.03) in grass and 0.158 mm 

 (SE=0.03)in sand. Average growth rates of 0.1 53 mm/ 

 day ( SE=0.03) were obsei-ved for both control and food 

 addition treatments (Fig. 3). These growth rates are 

 similar to growth rates reported elsewhere ( Warlen, 

 1981; Cowan, 1988; Nixon and Jones, 1997), suggest- 

 ing that our back calculation of growth rates from 

 otolith measures were not seriously biased. 



Effects of predation on recruitment of croaker to 

 varying habitats 



When we examined recruitment of croaker to experi- 

 mental plots with or without predator access, we were 



45 



40 - 



35 - 



30 - 



25 



20 



10 



5 - 



I I Predator Excluded 

 Control 



Habitat P = 34 

 Predator P = 92 

 n = 5 



Grass 



Sand 



Figure 4 



Atlantic croaker density (mean -t-l SE) in 1 m- 

 artificial seagrass (grass) and sand habitats 

 with predators excluded and allowing preda- 

 tory fish and decapod access (control). P values 

 from a blocked two-factor analysis of variance. 



unable to detect an effect of predation on croaker 

 recruitment (Fj i2=0.01, a=0.92) (Fig. 4). Croaker 

 density averaged 10.7/m- (SE=3.8) in the caged rep- 

 licates, and 15.8/m- ( SE=10. 1 ) in cage controls ( Fig. 4 ). 

 In contrast with the food supplementation experi- 

 ment, we did not detect a difference in recruitment 

 between grass and sand habitats (Fj j,;=0.96, a=0.34) 

 (Fig. 4). The interaction between habitat and preda- 

 tor access was also not significant (Fj j2=0.10, 

 a=0.76). This experiment also suffered from low 

 power ( 1-/^=0.06 ). Sufficient power to accept the null 

 hypothesis of no difference in croaker density be- 

 tween cage and cage-control treatments (l-/i=0.95), 

 would have required 550 replicates. 



Discussion 



Recruitment of fishes with open populations is af- 

 fected by variability in larval supply (Jenkins et al., 

 1996; Hamer and Jenkins, 1997), habitat selection 

 by settling larvae (Bell et al., 1987) and postsettle- 

 ment mortality (Orth et al., 1984), growth (Levin et 

 al., 1997), and migration (Sogard, 1989). Understand- 

 ing how these processes interact with each other to 

 determine population size has been a major focus of 

 researchers on tropical and temperate reefs (Doherty 

 and Williams, 1988; Caley et al., 1996) and recently 

 in seagrass meadows (Bell et al., 1987; Jenkins et 

 al., 1996; Hamer and Jenkins, 1997). There has also 



