Baum et al.: Bycatch of Hippocampus erectus in a Gulf of Mexico shirimp trawl fisfiery 



723 



United Stales 

 of America 



GULF OF 

 MEXICO 



ATLANTIC 



Florida OCEAN 



OI& Ofto C2003 - Km 



82^50' 



82^40' 



Figure 1 



Map of the eastern Gulf of Mexico, showing the study site of Hernando Beach on the western 

 coast of Florida. The shore is shaded; additional solid lines indicate depth contours in feet. 

 Letter and number combinations (e.g. N3) represent fishing areas along the coast. 



number of adult females because males that matured 

 after that size would have been included in our analysis 

 as females. We defined males as reproductively active if 

 they were pregnant or had recently released young (as 

 indicated by a loose pouch). Female reproductive state 

 was not included in this analysis because it is difficult to 

 determine reliably. We defined mortality to include sea- 

 horses already dead when the net was hauled and those 

 that died onboard. 



We evaluated temporal and spatial patterns in catch per 

 unit of effort (CPUE), standard length, population struc- 

 ture, and reproductive status of the seahorse bycatch. Spe- 

 cifically, we tested 1) temporal effects of year, lunar and 

 tidal phase, and 2) spatial effects of area and depth. In 

 our estimates of seahorse CPUE, we used standard length 

 instead of biomass because female weight changes with 



egg hydration and male weight increases greatly when 

 carrying embryos. Calculating CPUE per length (m) of 

 roller beam trawl controlled for variation in gear size. The 

 sampling unit was thus defined as the total or cumulative 

 standard length of seahorses caught during each tow (per 

 hour), per meter of trawl (per meter). Lunar phases were 

 defined as continuous variables by converting lunar day to 

 its angle, 6, based on a cycle of 29.5 days (=360°), with new 

 moon defined as 6 = 0°. These angles were then converted 

 to their cosine and sine functions for inclusion in linear 

 regression (deBruyn and Meeuwig, 2001). Tides were 

 semidiurnal in the Gulf of Mexico, ranging 1.3 m in tidal 

 level in the study area. High and low tides were defined as 

 each lasting two hours, with the remaining time classified 

 as ebb or flood accordingly. Spatial effects were analyzed 

 by dividing the total fishing ground into eight subareas ac- 



