266 
Fishery Bulletin 119(4) 
Figure 3 
(A) Position of the transverse dorsoventral section (0.5 mm thick) taken 
through the core (black dot) of a whole sagittal otolith from a black sea 
bass (Centropristis striata) collected from Massachusetts in May 2015. 
(B) Measurements for marginal increment analysis on a sectioned otolith 
taken from the core (white dot) to the distal edge of each annulus (black bars); 
the black sea bass from which this otolith was removed was captured from 
Massachusetts in May 2014. The measurements are R,, which depicts the 
otolith radius (core to edge); R,_,, the distance to the final annulus (core to 
the last opaque band); and R,_5, the distance to the penultimate annulus 
(core to the penultimate opaque band). The equation used to determine the 
marginal increment ratio (MIR) is MIR=(R,-R,_,)/(R,_,—-R,_2)- 
in MIA to confirm proper identification of the samples 
collected for this study. Length—frequency plots of the 
smallest fish (first 2 length modes) captured during the 
fall resource assessment survey (September 2016-2017) 
and the summer ventless trap survey (July—August 
2016-2017) were evaluated to confirm identification of 
the fall age-O and summer age-1 samples as YOY. Differ- 
ences in first annulus measurements between otoliths 
from fish collected in the regions north and south of the 
Hudson Canyon were also compared with Welch’s 
2-sample t-test. 
Assumptions for all statistical tests were evaluated by 
using visual diagnostic plots and were found to conform 
to assumptions of normality and homogeneity of variance. 
Type III sums of squares were used for both ANOVAs 
because of the unbalanced data. Post hoc multiple compar- 
ison analyses were conducted by using estimated marginal 
means, because sample sizes were not balanced among fac- 
tor levels (Lenth, 2019), and Tukey’s honestly significant 
difference test. A significance level of 0.05 was used in all 
statistical tests in this study. Model selection, ANOVAs, 
post hoc analyses, and visualizations were done by using 
R and the following packages in R: car, vers. 3.0-3 (Fox and 
Weisberg, 2019), emmeans, vers. 1.4.1 
(Lenth, 2019), multcomp, vers. 1.4-10 
(Hothorn et al., 2008), and ggplot2, vers. 
3.2.1 (Wickham, 2016). 
Results 
Marginal increment analysis 
Initial independent age determinations 
agreed for 1222 otoliths (89%); a consen- 
sus reading was required for the remain- 
ing samples. Precision was high and bias 
was low between readers in this study 
(CV=2.2%) (Fig. 4). Additionally, the pre- 
cision of each reader was high (CVs <2%) 
for both individuals from the reference 
collection, before and after samples col- 
lected for this study were examined. For 
samples that had been aged previously, 
final age estimates from this study were 
compared with previous estimates from 
collaborators, and age determinations 
differed for 107 fish (11%) and had a CV 
of 2.5%. There was no bias. 
The interactive model with the pre- 
dictors of Month Bin and Age Bin had 
the lowest AIC value and was selected 
for further analysis. An interaction 
between Month Bin and Age Bin for MIR 
(F=13.795, df=10, P<0.0001) revealed 
the lowest mean MIR occurred for 
AB1, followed by AB2 and AB3 (P<0.01) 
for the month bins January—February, 
March-April, and May—June; however, 
the remaining month bins had slightly different patterns. 
The MIRs for AB1 and AB2 were similar in July-August 
(P=0.3143) and November—December (P=0.3178) but were 
smaller than those for AB3 (P<0.001). Also, MIRs for AB2 
and AB3 were similar in September—October (P=0.8206) 
but were larger than those for AB1 (P<0.001). 
Campana (2001) noted that a minimum in the MIR should 
occur once per year and be significantly different from the 
MIR in other times of the year. Figure 5, A-C, shows that 
the minimum MIR in each age bin (indicated with the letter 
a above boxes for month bins) occurred once per year and 
was different from that of other month bins (P<0.0001). The 
only exception was for AB3, which appeared to have a mini- 
mum that extended from July—August through September— 
October (P=0.9849); whereas the minimums for AB1 and 
AB2 were both in July—August only. Plots of raw data show 
the monthly MIR for AB3 declined in July prior to reaching 
a minimum in August (Fig. 6C). Similarly, a depression in 
MIR occurred in May—June for AB1 (Figs. 5A and 6A). For 
all age bins, MIR gradually increased throughout the year 
after the minimum occurred. 
The model with 2-way interactions between Age Bin, 
Season, and Region had the lowest AIC value and was 
