26 
Fishery Bulletin 119(1) 
scales from the reference collections having the best agree- 
ment with 95.8% for Atlantic menhaden and 100.0% for 
Gulf menhaden. The readings for more recent samples, 
those taken in 2017, had better PA than the readings of 
the older samples from 2013 and 2005. Percent agreement 
in the age estimates for scale subsamples ranged from 
83.6% to 97.1% (Table 2). For age data on subsamples of 
Atlantic menhaden, PA was consistently >88%. For age 
data on subsamples of the Gulf menhaden, PA was more 
variable, with comparisons of the second readings done 
with the Eberbach projector and microscope and compari- 
sons of the first and second readings done with the micro- 
scope having PA above 90%. 
For all scale samples examined in our study, the propor- 
tion of scales for each of the determined ages reflected the 
proportion of samples for each observed age in the long-term 
sampling program for the Atlantic and 
Gulf menhaden fisheries (Fig. 2; for age 
compositions used in the respective stock 
assessments, see SEDAR, 2018, 2020 and 
Supplementary Figures 1-6 [online only]), 
and the age bias plots indicate that 95% 
confidence intervals for mean ages were 
small (Suppl. Figs. 7-14) (online only). Dif- 
ferences in aging were greatest at the older 
ages (>3 years) and at age 0; however, sam- 
ple sizes for these ages were smaller than 
those for the ages of 1 year and 2 years. 
Patterns in the number of samples for 
each of the ages determined for subsam- 
ples of Atlantic and Gulf menhaden on the 
same device are similar to patterns for the 
full sample, with small numbers of devia- 
tions in age that are both above and below 
the 1:1 line in age bias plots, except for the 
ages of 3 and 4 years estimated with the 
Eberbach projector for Atlantic menhaden 
(Suppl. Fig. 15) (online only). 
For samples from the reference col- 
lections and samples taken in 2017 
for both Atlantic and Gulf menhaden, 
no bias in age estimates was found 
in comparisons between the 2 pieces 
of equipment—across all of the tests of 
symmetry (Table 3). For the samples 
of Gulf menhaden from 2005, however, 
the results of all 3 tests of symmetry 
indicate significant bias. A significant 
bias in age estimates for samples of 
Atlantic menhaden from 2013 is indi- 
cated only in results from the Bowker’s 
test. For all samples of Atlantic men- 
haden in total, results of the Bowker’s 
test indicate some bias, likely influ- 
enced heavily by the data set for 
samples from 2013, and the results 
of the other tests do not indicate sig- 
nificant bias in age data between the 
Eberbach projector and microscope. For 
Microscope age 
o 
D 
© 
® 
Q 
fe} 
12) 
no 
fe} 
_ 
2 
= 
all samples of Gulf menhaden in total, results of all 3 
tests of symmetry indicate significant bias in age esti- 
mates, influenced heavily by the data set for samples 
from 2005. For the subsamples, results from all tests for 
both species indicate significant bias in ages estimated 
with the Eberbach projector; results of these tests for 
both species indicate that bias was not significant for 
ages determined with the microscope. In comparisons of 
ages for the subsamples between estimates made with 
the Eberbach projector and those made with the micro- 
scope, results of some tests indicate significant bias, but 
results from other tests do not, outcomes that are likely 
influenced by the bias in ages determined with the Eber- 
bach projector. 
Overall simultaneous multinomial confidence intervals 
were slightly different for the ages of 1 year and 2 years 
2 
Eberbach projector age 
2 
Eberbach projector age 
Figure 2 
Age bias plots showing the number of samples for each age estimated with 
an Eberbach projector and stereo microscope for all (A) Atlantic menhaden 
(Brevoortia tyrannus) and (B) Gulf menhaden (B. patronus) collected from 
landings along the Atlantic and Gulf coasts of the United States. Samples 
included Atlantic menhaden collected in 2013 and 2017, Gulf menhaden col- 
lected in 2005 and 2017, and reference collections for each species. The 1:1 line 
represents agreement between the age estimates from use of the 2 devices. 
