Petrik et al,: Recruitment of Micropogonias undulatus 



957 



We used a blocked two-factor analysis of variance to 

 test the hypothesis that gi-owth rates did not vary 

 among habitat or food supplementation treatments. 



Effects of predation on recruitment of croaker to 

 varying habitats 



The null hypothesis that predation has no effect on 

 recruitment of croaker in sand or seagrass habitats 

 was tested by using cages to limit predator access to 

 experimental plots. On 18 March 1996, a fortuitous 

 seasonal low tide completely exposed the study site 

 to air. allowing us to erect cages and to ensure that 

 no recruits or predators occupied the cages at the 

 start of the experiment. Cages (2 x 2 x 1 m) were 

 constructed with 25-mm mesh on four sides, whereas 

 cage controls had mesh on two sides. Mesh was large 

 enough to be transparent to croaker recruits, but fine 

 enough to prevent predators from entering the cages. 

 A randomized block design was employed, with habi- 

 tat (sand or grass) and predator access (cage or cage 

 control) as fixed effects. Replicates were placed 8 m 

 apart, and blocks were separated by 10 m. The experi- 

 ment was terminated after 7 d. Drop samplers were 

 used to quantify recruits as described previously. 



Results 



Habitat preference and use by newly recruited 

 croaker 



The density of newly settled croaker (mean TL=14.48, 

 SE=0.15) did not differ among sand, seagrass, or 

 marsh edge habitats (F^ ^=0.86, «=0.44, /3=0.13) 

 (Fig. 1). The density of croaker recruits also did not 

 vary between sites (Fj jy=0.09, «=0.77) and the in- 

 teraction between habitat and site was not signifi- 

 cant (^2,12=1-07, «=0.37). 



Our observation on the behavior of croaker recruits 

 in mesocosms did not reveal a preference between 

 sand and seagrass habitats (one way /-test, ^=1.64, 

 df=ll.a=0.13). An average of36'::HSE=16> of the time 

 was spent in seagrass, and QA^'i (SE = 16) in sand. 



Effects of food supply on croaker recruitment and 

 growth in varying habitats 



The abundance of newly settled croaker differed be- 

 tween experimental habitats (Fi.ii=5.98, a=0.03) with 

 greater recruitment in sand ( X=31.6/m-, SE=6.5) 

 than in seagrass habitats (X=13.7/m-, SE=5.9) 

 (Fig. 2). Conversely, we did not detect a difference 

 (Fj ,,=0.13, a=0.73) in the number of croaker in food- 

 supplemented plots ( X =23.2/ m-, SE=8.3 ), compared 



12 n 



•S 4 



□ Edge 

 Grass 

  Sand 

 Habitat P = 44 

 Site P = 77 



Site 1 



Site 2 



Figure 1 



Atlantic croaker densities (mean +1 

 SE ) sampled from three estuarine habi- 

 tats; marsh edge (edge), seagrass bed 

 (gi-ass). and barren sand (sand). P val- 

 ues from two-way analysis of variance. 



40 



o 



10 



n Food Addition 

 Control 



Habitat P = 03 

 Food P = 73 

 n = 5 



Grass 



Sand 



Figure 2 



Atlantic croaker density (mean +1 

 SE) in 1-m- artificial seagrass (grass) 

 and sand habitats, with (food addi- 

 tion) and without ( control ) food supple- 

 mentation. P values from a blocked 

 two-factor analysis of variance. 



with control plots ( X =22. 1/m^ SE=4. 1 ) ( Fig. 2 ). The 

 interaction between habitat and food-supplementa- 

 tion was not significant (Fj jj=0.96, «=0.35 ). The sta- 

 tistical power of this experiment was low ( 1-/3=0.18); 



