956 



Fishery Bulletin 97(4), 1999 



Effects of food supply on recruitment and growth 

 of croaker in varying habitats 



Field experiments were conducted in East Lagoon, 

 located at the eastern most end of Galveston Island, 

 TX (29°20'N, 94°44'W). East Lagoon is 1.6 km long, 

 0.48 km wide, and has a maximum depth of 4.6 m. 

 Water is exchanged tidally by means of seven 0.92 m- 

 diameter cement culverts with the Galveston Ship 

 Channel, which runs from the Gulf of Mexico into 

 Galveston Bay. A detailed description of this site can 

 be found in Levin et al. ( 1997). Seagrass, once wide- 

 spread in Galveston Bay, including East Lagoon, has 

 decreased by 90% from peak levels (Pulich and White, 

 1991); no natural seagrass habitats presently are 

 found in East Lagoon. The absence of natural 

 seagrass beds allowed us to establish artificial 

 seagrass beds with desired characteristics, without 

 the confounding effects of a natural seagrass bed. 

 Experiments were located >8 m from the Sparfina 

 alterniflora dominated marsh edge and placed at an 

 average low tide depth of 42 cm. 



To test the null hypothesis that food supply does 

 not limit abundance or growth rates of croaker re- 

 cruits in different habitats, we conducted an experi- 

 ment in which food supply was manipulated in sand 

 and seagrass habitats. On 20 February 1996 we cre- 

 ated five blocks each consisting of four 1-m- experi- 

 mental plots. Within each block, food supply and 

 habitat type were manipulated orthogonally. To con- 

 trol for differences in seagi'ass structure or seagrass- 

 associated food resources, we used artificial seagrass 

 habitats. Artificial seagrass habitats were con- 

 structed from a 1-m- polyvinylchoride (PVC) ( 1.3 cm 

 diameter) frame, strung with monofilament to form 

 a grid consisting of 576 points. At each of these points 

 a 16 cm X .5 cm strand of green ribbon was woven in, 

 such that the frame consisted of 576 shoots of 

 seagrass, each shoot having two leaves. No exces- 

 sive fouling was observed on the frame or ribbon for 

 the duration of the experiment. We performed a pre- 

 liminary experiment to determine if the structure of 

 the PVC-frame would attract more fish recruits than 

 bare sand, and no difference was found between the 

 bare sand plot and the PVC-frame (F, 2,3=0.512, 

 0!=0.61, 1-/3=0.76). Consequently we performed sub- 

 sequent experiments without a PVC-frame control. 



Food supply was experimentally manipulated with 

 feeding tubes in each experimental plot. Feeding 

 tubes were constructed of a 7.5-cm diameter length 

 of PVC pipe attached to a l.S-cm diameter PVC pipe 

 stake, with the bottom of the tube about 15 cm from 

 the substratum, and the top always above the water 

 line. We provided supplemental food daily for seven 

 days, from 23 February to 3 March 1996, to half of 



the sand and seagrass replicates (i.e. five sand and 

 five seagrass plots received food). Food consisted of 

 200 g of fish fiesh and 300 mL of water blended to 

 produce plankton-size particles (Forrester, 1990; 

 Levin et al., 1997). The fish puree was placed in ice 

 cube trays and frozen. Each frozen cube yielded 

 11.8 g offish flesh. One cube of frozen food was placed 

 in the feeding tubes of appropriate replicates, 

 whereas control plots received one ice cube and no 

 food was added. As the ice cube containing food 

 melted, it delivered a continuous stream of particles 

 to the habitat for 5-15 min. We observed fish readily 

 consuming the supplemented food in both the field 

 and laboratory. 



On 4 March 1996, the experiment was terminated 

 by sampling each plot. Recruit density was quanti- 

 fied by using 1 m'' (1x1x1 m) drop samplers 

 (Zimmerman et al., 1984; Fonseca et al., 1990). Drop 

 samplers were constructed of 9.5-mm diameter rebar 

 covered on four sides with taut 2-mm nylon mesh. A 

 dip net (90 x 100 cm, 2-mm nylon mesh) was used to 

 retrieve fish from the samplers, and replicates were 

 considered adequately sampled when five consecu- 

 tive passes of the dip net yielded no fish (Fonseca et 

 al., 1990). A blocked two-factor analysis of variance 

 was used to test the hypotheses that the abundance 

 of newly recruited croaker did not vary among habi- 

 tat or food supplemented treatments. 



Five fish from each replicate were haphazardly 

 selected for further analysis. We measured the 80 

 selected fish to the nearest 0.1 mm (SL), removed 

 their otoliths, and stored them in immersion oil for 

 one week. Fish age was then determined by enumer- 

 ating the daily growth rings on the lapillar otolith 

 by using an image analysis system. The existence of 

 daily rings on croaker otoliths has been validated 

 previously ( Nixon and Jones, 1997 ). Each otolith was 

 examined independently three times. If two of the 

 three counts did not agree, the fish was discarded 

 and another selected. When two of the three counts 

 were the same, that count was used as a datum in 

 the analysis. 



Differences in growth rates were examined by us- 

 ing otolith microstructure. Because otolith diameter 

 was correlated to fish length (r=0.73, n=75), we used 

 otolith measures as a proxy for growth rate. Mea- 

 surements were taken inward from the edge of the 

 otolith to the seventh ring. This distance corre- 

 sponded to growth during our seven days of food 

 supplementation. Otolith distances (/i) were then 

 converted into daily growth rates (mm SL/day), by 

 using the following equation generated from a re- 

 gression of otolith diameter on fish length; 



Growth ^{iotnlith (//.stance + 0.002) / 0.014J/7. 



