Kolmos et al.: Temporal changes in the life history of Hyporthodus niveatus off North and South Carolina 
311 
place in the 24 h after capture. For specimens with vitel¬ 
logenic oocytes or with hydrated oocytes, lunar day was 
equal to their date of capture. For specimens with post¬ 
ovulatory complexes, lunar day was determined to be the 
capture date -Id because spawning would have occurred 
in the 24-48 h prior to capture. The effect of lunar day 
on spawning probability (no. of spawners to no. of active 
females, i.e., presence of vitellogenic oocytes) was analyzed 
by using a generalized additive model (GAM), with bino¬ 
mial error distribution and logit link function. No other 
covariates were included in the model. 
Analyses 
Analyses were completed with statistical software R, 
vers. 3.0.2 (R Core Team, 2013), Microsoft Excel 2013 
(Microsoft Corp., Redmond, WA), and SAS (vers. 9.4; SAS 
Institute Inc., Cary, NC), and results were deemed sig¬ 
nificant ifP<0.05. Mean values of length, age, and length 
at age were compared between periods by using analysis 
of variance. All lengths in the text refer to maximum TL. 
Gear selectivity was assessed with density analysis of 
TL to determine how data from various gear types could 
be pooled for temporal comparisons related to age and 
growth; comparisons focused on 2 or 3 of the following 
periods: 1982-1985, 1993-1994, and 2008-2012 (hereaf¬ 
ter referred to as the 1980s, 1990s, and 2000s). Density 
analyses and the Kolmogorov-Smirnov test were used 
to compare length and age distributions between peri¬ 
ods. Nonlinear regression analysis was used to fit the 
von Bertalanffy growth function (von Bertalanffy, 1938) 
to length at age data by period. Following Wyanski et al. 
(2000), TLs were weighted by the inverse of the number 
of fish at each age, by period and overall, to moderate the 
effect of large and small sample sizes on the estimates 
of growth parameters from the von Bertalanffy growth 
function. 
All available reproductive samples from all gears were 
pooled for reproductive analyses, primarily to increase 
sample size. Pooling data from the 1980s and 1990s was 
necessary to increase sample size for the temporal aspect 
of analyses of sexual maturity and sex transition. Addi¬ 
tional samples (n=85) from the period 1979-1981 were 
used in maturity and spawning proportion analyses. 
For maturity analyses, pooling data over decades also 
ensured that more specimens, especially those within the 
range of size and age at which maturity was reached, were 
included in the temporal comparison (1980s and 1990s 
versus 2000s); these smaller specimens are caught more 
frequently with chevron traps, which were not used in the 
1980s. To estimate length at 50% maturity (L 50 mat ) and 
age at 50% maturity (A 50 mat ), a generalized linear model 
with a logit, probit, c-log, or Cauchy link was fit to matu¬ 
rity data in RStudio, vers. 3.5.2 (RStudio, 2018); length 
and increment data were not binned. The same analyses 
were run to estimate the length (L 50 male ) and age (A 50 male ) 
at 50% sex transition to male. The selected model had the 
lowest Akaike information criterion (Akaike, 1973). Sig¬ 
nificance was evaluated at the 5% error level by using the 
likelihood ratio chi-square (/ 2 ) test. In spawning analyses, 
data for sizes of fish from the spawning season (April- 
September) were grouped into 100-mm-TL bins starting 
from 501-600 mm TL and ending with a terminal group of 
>901 mm TL; analyses were performed overall and within 
months. The same spawning analyses were performed on 
ages of fish from the spawning season and grouped into 
2-year bins starting from ages 3-4 and ending with a ter¬ 
minal group of ages >11 years. Lunar effects on spawning 
were investigated with a GAM. 
Results 
Age and growth 
All specimens (n=5314) were collected between 29.66°N 
and 34.73°N and at depths of 18-302 m, but only 7 spec¬ 
imens were collected south of 31°N. We aged 2192 spec¬ 
imens during 1996-2012, making our total number of 
aged specimens 4342 when adding historical data from 
Wyanski et al. (2000). Ages in the combined data set 
ranged from 1 to 35 years, and lengths ranged from 130 
to 1137 mm TL. Males ranged in age from 6 to 32 years 
(mean: 13.0 years [standard error (SE) 0.5]; median: 
12 years) and in size from 620 to 1090 mm TL (mean: 
899 mm TL [SE 6.5]; median: 901 mm TL), whereas 
females ranged in age from 1 to 35 years (mean: 6.7 
years [SE 0.1]; median: 6 years) and in size from 208 
to 1113 mm TL (mean: 611 mm TL [SE 2.9]; median: 
619 mm TL). Transitional specimens ranged in age from 
5 to 12 years (mean: 8.4 years [SE 0.6]; median: 8 years) 
and in size from 627 to 895 mm TL (mean: 803 mm TL 
[SE 18.3]; median: 826 mm TL). 
To determine whether to combine length and age data 
for the 4 prevalent hook-and-line gear types, a density 
analysis was performed on TL by gear type for all periods 
combined (Fig. 1). Results indicate that the overall size 
range of snowy grouper was similar among gear types, 
but snapper reels caught a higher proportion of individ¬ 
uals <450 mm TL than the other gear types because this 
gear is fished in shallower depths and catches smaller 
fish (Wyanski et al., 2000). Given the notable differ¬ 
ence in the size selectivity of snapper reels, data from 
the remaining 3 gear types (long bottom longline, short 
bottom longline, and Kali pole) were combined, because 
of their relatively similar size selectivity, and used in 
further analyses of age and growth (Table 1). Another 
reason for combining the data is the similarity of the 
fishing hook sizes (#5-7) used in each longline gear and 
the Kali pole gear (Russell et al., 1988; Harris et al., 
2004; Smart et al. 4 ). The inclusion of data from only one 
period for short bottom longlines (2000s) and Kali poles 
(1980s) in the pooled data did not influence the tempo¬ 
ral comparisons because a Kolmogorov-Smirnov test of 
data from long bottom longlines between each period 
was significant (P<0.001) for size and age, indicating 
that the differences between periods are explained by 
time, not gear type. 
