Jeffers et al.: Habitat and bycatch effects on population parameters of Synodus foetens in the Gulf of Mexico 
419 
mesh. The sole deviation from SEAMAP protocol was 
the inclusion of a 0.7-cm mesh bag within the codend 
to retain small fish. Three replicate tows were made 
in each habitat type inside and outside the AR zone 
during quarterly sampling. Before trawl sampling, a 
hydrographic cast was made at each station with a Sea- 
bird 19plus CTD (Sea-Bird Electronics Inc., Bellevue, 
WA) to measure temperature, salinity, and dissolved 
oxygen. Then, trawling was conducted at 4.6 km/h for 
approximately 10 minutes. Up to ten fish were randomly 
sampled from each trawl sample and immediately fro- 
zen in ziploc bags and brought to the laboratory for 
processing. Individuals were thawed, weighed to the 0.1 
g, measured to the nearest mm total length (TL), and 
their sex was determined by macroscopic examination 
of gonads. Both sagittal otoliths were then extracted for 
age analysis as detailed below. 
Additional fish were opportunistically sampled with 
trawls approximately monthly from the north central 
GOM (Fig. 1). In the laboratory, sampled individuals 
were measured, weighed, and their sex determined as 
above. Their sagittae then were extracted and stored. 
The purpose of opportunistic collections was to provide 
monthly samples to identify the timing of otolith opaque 
zone formation by means of marginal condition analysis, 
but those samples were also included in age and growth 
analyses. 
Inshore lizardfish density was estimated as the num- 
ber of individuals caught divided by the area swept 
by the trawl. Trawl width under tow was estimated 
as 0.8x12.8 m, which was based on National Marine 
Fisheries Service (NMFS) unpublished data of tow- 
ing conditions for their standard trawl survey gear. 
Distance towed was estimated with GPS coordinates 
recorded while the trawl was on the seafloor. Analysis 
of variance (ANOVA; 5x4x2 factorial design) allowed 
us to test sampling quarter, habitat, and exposure to 
trawling effects, as well as their interactions, on in- 
shore lizardfish density and size. All statistical analy- 
ses were computed with SAS (SAS Institute Inc., Cary, 
NC). Density values were transformed by taking the 
natural log of (density +1) and size was log-transformed 
to meet parametric assumptions of normality and het- 
eroscadascity. Analysis of variance test statistics (F- 
tests) were computed for both density and fish size 
with type-III sums of squares despite the two missing 
cells (i.e., no high shell data outside the AR zone for 
either winter or spring 2005) in the factorial design 
noted above. However, additional ANOVAs were also 
computed after removing all high shell habitat data to 
confirm results from the unbalanced models computed 
with the full data set. 
One sagitta from each fish was used to determine age. 
Otoliths were set in epoxy and a 0.5-mm transverse sec- 
tion including the core was cut with a Buehler (Buehler 
Ltd., Lake Bluff, IL) Isomet saw fitted with a diamond 
blade. Each section was mounted on a glass slide with 
thermoplastic cement, sanded with 3200 grit wet-or-dry 
sand paper, and polished with an alumina powder slur- 
ry on a felt polishing cloth. Marginal condition analysis 
was performed to verify that opaque zones formed an- 
nually in adults (Beckman et al., 1990). Opaque zones 
were then counted to determine age. Otolith thin sec- 
tions were read with a compound microscope with both 
reflected and transmitted light. Two different readers 
counted opaque zones in each sample; readers counted 
opaque zones independently without knowledge of fish 
size or the other reader’s age estimate to ensure that no 
bias occurred during process. If counts for a given oto- 
lith differed, the otolith was read a second time by both 
readers. Average percent error (APE) was computed 
to assess precision between readers (Campana, 2001). 
Length-at-age was plotted and then least-squares fits of 
the nonlinear von Bertalanffy growth curve were com- 
puted to estimate inshore lizardfish growth for males, 
females, and both sexes jointly: 
L t = L x ( l- e -K«-*o», (1) 
where L t = estimated length at age t\ 
L^= asymptotic length; 
K = growth coefficient; 
/ = age in years; and 
t 0 = hypothetical age at zero length. 
A likelihood ratio test was computed to test whether 
sex-specific growth functions were significantly different 
from one another (Kimura, 1980). 
Three independent estimates of natural morality (M) 
were computed based on maximum observed longev- 
ity according to the methods of Royce (1972), Hoenig 
(1983), and Pauly (1980): 
M = 4.6 l(t max ) (Royce, 1972), (2) 
In (M) = 1.44 -0.982 In (t max ) (Hoenig, 1983), (3) 
ln(M) = -0.0152 -0.2791n(LJ + 0.65431n(JO 
+ 0.4631n(T) (Pauly, 1980), (4) 
where M = natural mortality/yr; 
tmax ~ the oldest aged fish in sample; and 
T - mean temperature °C. 
Total instantaneous annual mortality (Z) was estimated 
with catch curve analysis (Ricker, 1975). Bycatch mortal- 
ity (F b ) was computed by subtraction: 
F b - Z - M (5) 
where F h = instantaneous annual fishing (bycatch) 
mortality; 
Z = instantaneous annual total mortality; and 
M = instantaneous annual natural mortality. 
The difference in total mortality between habitats 
exposed to trawling versus nontrawled habitats was 
tested with analysis of covariance (ANCOVA) test for 
equal slopes. Exploitation ratio ( E ) was estimated as 
the ratio of F b /Z (Ricker, 1975). 
