Brown-Peterson et al.: Reproductive biology of Brevoortia patronus in the Gulf of Mexico 
287 
mercial fishery from 1964 to 2014 and archived at the 
National Marine Fisheries Service (NMFS) Beaufort 
Laboratory. At each commercial reduction plant oper- 
ating during these years, a sample of fish was taken 
from the top of the fish hold. Ten (1971-2014) or 20 
(1964-1970) fish were haphazardly selected from that 
sample. Each fish was weighed (in grams), measured 
for FL (Schueller et al., 2012), and a scale patch was 
removed for aging. For fish collected from 1964 to 
1970, sex was determined, and gonads were weighed 
(in grams) and macroscopically classified according to 
NMFS criteria (Huntsman and Chapoton, 1973). For 
these data, females were considered reproductively ac- 
tive with a GSI>1.0 and males were considered repro- 
ductively active with a GSI>0.5. 
Data analyses 
Reproductive biology Precision of macroscopic in con- 
trast to histological assessment of gender, as well as 
reproductive phase assessment for both males and 
females, was compared by calculating percent agree- 
ment (Klibansky and Scharf, 2015) and testing with a 
chi-square test, where differences were considered sig- 
nificant at P<0.05. When only 2 phases were compared, 
the Yates corrected chi-square test was used. The sex 
ratio was calculated for each month and overall and 
was analyzed with a chi-square test. The size-at-50%- 
maturity for males and females was determined with 
a logistic regression in R software (vers. 3.1.3; R Core 
Team, 2015). 
Distribution of males and females in each reproduc- 
tive phase for each age were compared among months 
when fish were reproductively active by using the 
Kolmogorov-Smirnov 2-sample test with a Bonferroni 
adjustment to determine whether a significant differ- 
ence existed in reproductive potential among months. 
Relative batch fecundity (RBF), expressed as number of 
eggs/g of fish weight, was calculated as BF/ovary-free 
fish weight. The relationship of BF and RBF to fish 
size (FL, W, age) and K was determined by using lin- 
ear and nonlinear regression analyses. Data were log 10 - 
transformed where necessary to meet assumptions and 
provide a better linear fit. Analyses were performed 
with IBM SPSS Statistics software, vers. 18 (IBM, Ar- 
monk, NY) and differences were considered significant 
if P<0.05. 
Comparisons of historical data Annual length-weight 
relationships were determined (1964-2014). Both 
length and weight were log-transformed and standard- 
ized for each year: 
S SD „ 
(3) 
and h = — — — , 
SD V 
(4) 
where x = the log-transformed length; 
x - the mean log-transformed length; and 
SD X = the standard deviation of x, y was the log- 
transformed weight; 
y = the mean log-transformed weight; and 
SD y = the standard deviation ofy (Kruschke, 2014). 
Standardizing the data was necessary given the cor- 
relation between the parameters in the length-weight 
relationship. 
Length-weight relationships were fitted by using 
Bayesian methods for linear regression in R and the 
program JAGS (Plummer, 2003): 
h = p 0 + Pig, (5) 
where length, g, was related to weight, h, through a 
slope, Pi, and intercept, p 0 . 
Priors for the slope and intercept parameters were 
normally distributed N(0, 0.1). A total of 30,000 chains 
were run for each year and a 10,000-chain burn-in 
period. 
Comparisons of FL, GSI, and K of reproductively 
active males and females collected in September and 
October during 1964-1970 and 2014-2015 were made 
with £-tests; equality of variance was assessed with 
Levene’s test. These analyses were performed with 
IBM SPSS Statistics software (vers. 18) and considered 
significant at P<0.05. 
Results 
Age determination 
Age estimates were determined from scales of 539 Gulf 
menhaden (269 females, 213 males, 57 fish of unknown 
sex) collected from 2014 through 2015 and used in our 
analyses of reproductive biology. Age estimates ranged 
from 1 through 4 years, with the majority of males and 
females in the age-2 class. Most fish of unidentified sex 
were age-1, corresponding to immature or young-of- 
year individuals. There was a large overlap in length- 
at-age, likely owing to the extended spawning season 
and variability in length-at-age. 
Reproductive biology 
A total of 697 Gulf menhaden were collected from 2014 
through 2016 for reproductive analysis, composed of 
337 females, 294 males, and 66 fish of unknown sex, 
of which 22 were <130 mm FL and sexually immature. 
Macroscopic differentiation of gender was particularly 
difficult for fish that were reproductively inactive or 
were frozen before analysis. Overall, there was 80% 
agreement between macroscopic and histological gen- 
der assignment for females and 84% agreement for 
males. Chi-square analysis showed a significant mis- 
classification of gender based on macroscopic observa- 
tions for males and females (x 2 59 3=239.1, P<0.001), sug- 
gesting histological analysis is necessary to accurately 
determine gender of Gulf menhaden, particularly with 
frozen specimens. 
