194 



Fishery Bulletin 97(1), 1999 



from each side of the vessel. These consisted 

 of a net attached to a stainless steel frame 

 with a slotted roller along the entire lower 

 portion of the frame. Stainless steel finger 

 bars, 0.8 m long, were fastened vertically, 5 

 cm apart, along the front top of the frame 

 to exclude seagrass and other debris (Ber- 

 keley et al., 1985). Each roller beam trawl 

 weighed -75 kg. Nets were constructed of 

 2.56-cm stretched mesh and had a 1.90-cm 

 stretched mesh tail bag. 



Turtlegrass 



Nine trawl areas were selected in turtle- 

 grass beds near Tarpon Key (Fig. 1). Each 

 consisted of a marked 40-m x 3.38-m plot of 

 continuous turtlegrass cover. Within each 

 40-m plot, a central 20-m x 3.38-m impact 

 area was marked. During August, we 

 sampled turtlegrass by removing cores of it 

 within the impact area of each plot prior to 

 trawling and after 1, 3, and 9 trawls. In 

 November, new plots were marked and 

 samples were collected prior to trawling and 

 again after 9 and 18 trawls. Trawl levels 

 were increased in November because pre- 

 liminary analysis of the August data did not 

 reveal a significant effect on turtlegrass at 

 the maximum trawl level. 



Five 15-cm diameter x 20-cm deep sub- 

 strate cores were taken randomly along a 

 lengthwise midline transect within each impact area 

 at each specified trawl level. For each core the shoot 

 density, longest blade length, total blade length per 

 shoot, and number of blades per shoot (from three ran- 

 domly selected shoots) were recorded. The turtlegrass 

 standing crop for each core was measured separately 

 for above- and below-ground dry -weight biomass ( dried 

 at 60°C for at least 48 hours). 



The sequential seagrass measurements within each 

 replicate plot were used in regression analysis to de- 

 tect rates of decrease for each of the different plant 

 parameters as a function of trawl level. The regi-ession 

 slope (rate of decrease) for each plot was calculated for 

 each plant parameter. Mean rate of decrease for each 

 plant parameter was calculated for each month (all 

 plots combined, n=9 for each month). T-tests were used 

 to compare differences (P<0.05 ) in mean rate of decrease 

 for each plant parameter for each month. 



Mortality of bycatch fish 



The bycatch collection site was southwest of Tierra 

 Verde, near the mouth of Tampa Bay (Fig. 1). Each 



bycatch 

 trawl site 



Figure 1 



Location of bycatch and turtlegrass trawl sites within Tampa Bay, Florida, 

 27 40' North latitude, 82 27' West longitude. 



month 30 trawls of 5-min duration were made at this 

 site over three consecutive nights. Trawl and culling 

 times selected were based on observations of com- 

 mercial live-bait shrimp fishing vessels and short- 

 est times observed used to obtain conservative esti- 

 mates of bycatch fish mortality. In August, the catch 

 from one net of a trawl pair was allowed to sit on the 

 culling table for 2 min prior to placement in the on- 

 deck live well. The catch from the other trawl net 

 was discarded. While the catch was sitting on the 

 culling table, algae and detritus were removed to 

 reduce subsequent fish entanglement. To reduce the 

 effect of predation on mortality estimates, hardhead 

 catfish, Arius felis, were removed from the catch. 

 Bycatch for each trawl was placed in its own sepa- 

 rate holding pen, which was held on deck inside an 

 aerated 300-L live well until transferred to a final 

 unvegetated holding site. Holding pens along with 

 bycatch contents were transferred from the on-deck 

 live well to the holding site in a 76-L seawater tank. 

 On-deck holding times were 15 min, transport 5 min. 

 Once at the holding site, pens were placed in the water 

 at a low tide depth > 1 m (the minimum) and anchored. 



