168 
Fishery Bulletin 120(2) 
_A 
® All BL fish (n=741) 
a! 
400 
! 
200 
600 800 1000 
Stretch total length (mm) 
Frequency (%) 
= Female aged fish (n=392) 
= Male aged fish (n=369) 
suffer from deviations from uniformity, outliers 
(Suppl. Fig. 1) (online only), dispersion (P=0.92), or 
zero inflation (P=0.87). The results of the variance 
inflation factor analysis indicate a lack of multi- 
collinearity, given that all variance inflation fac- 
tors were less than 2. The variable for year was 
not significant (P=0.13), and there were no trends 
within the standardized index (Fig. 5), indicating 
that the declines in the nominal CPUE data for 
2007-2010 reflect increases in offshore sampling 
effort beginning in 2010 rather than changes in 
a relative abundance of red drum. 
Spatial analysis 
During 2006-2018, bottom longline sets were dis- 
tributed fairly evenly between state (46%) and 
federal (54%) waters. Nominal CPUE was high- 
est less than 3 nmi from shore (1.13 [SE 0.10], 
602 stations), followed by 3-6 nmi from shore 
(0.72 [SE 0.18], 103 stations), 6—9 nmi from shore 
(0.385 [SE 0.19], 58 stations), and greater than 
9 nmi from shore (0.08 [SE 0.03], 533 stations). 
The one-way analysis of variance found that 
distance from shore was significant (P<0.01). 
Peal = T I 
0 200 400 
T T T 
600 800 1000 
Total length (mm) 
Figure 3 
1200 The results from the Tukey’s multiple-pairwise- 
comparison test indicate that nominal CPUE was 
significantly higher less than 3 nmi from shore 
compared with 6—9 nmi from shore (P<0.01) and 
greater than 9 nmi from shore (P<0.01). Nomi- 
Length—frequency distributions for (A) red drum (Sciaenops ocella- 
tus) encountered during bottom longline (BL) surveys (sexes com- 
bined) and (B) red drum encountered during BL and gill-net surveys 
(by sex) in the north-central Gulf of Mexico from 2006 through 2018. 
The vertical dashed line represents size at 50% maturity, reported 
nal CPUE was also significantly higher 3-6 nmi 
from shore than greater than 9 nmi from shore 
(P<0.01). Both ages (D=0.41, P<0.01) and length 
distributions (D=0.42, P<0.01) were significantly 
by Bennetts et al. (2019). n=number of samples. 
have a higher L,, value compared with that of males. The 
VBGF versions for female (F) and male (M) red drum, 
respectively, are as follows: 
ag 3S VOEEE Sop eval (7) 
Lian = 932.71(1 = GAVEL) (Fig. 4B). (8) 
All VBGF parameter estimates from this study are listed 
in Table 1. Estimates of M were 0.12 for the Hoenigs..,.. 
method, 0.14 for the Hoenig,,, method, and 0.39 for the 
Pauly,,.7 method. 
Relative abundance 
The final version of the negative binomial generalized 
linear model included variables for year, depth, surface 
temperature, dissolved oxygen, and bottom salinity. 
The variables for latitude, longitude, bottom tempera- 
ture, surface salinity, and day length were also tested but 
were excluded from the final version of the model. Model 
fit was deemed appropriate because the model did not 
different for red drum caught in state versus fed- 
eral waters. Notably, fish were older and larger 
in state waters (average age of 18 years and 
average total length of 938 mm) than in federal 
waters (average age of 12 years and average total length 
of 887 mm). Further examination revealed a negative cor- 
relation between age and distance from shore (coefficient 
of correlation [r]=—0.239, P<0.01) and between size and 
distance from shore (r=—0.274, P<0.01). 
Model performance and interpretation 
Model performance was assessed for all red drum across the 
3 sampling seasons: spring, summer, and autumn. The AUC 
scores for training data were high across all seasons (0.90), 
indicating “very good” model performance according to cri- 
teria defined in Lane et al. (2009) (Table 2). Cross-validated 
AUC scores were 0.85—0.86 (SE 0.01), indicating that model 
overfitting was negligible (Hijmans and Elith, 2013). 
Habitat suitability 
Across all seasons, northward velocity of the surface cur- 
rent, surface temperature, and depth were the 3 most 
influential predictors of abundance of red drum (Table 2). 
