30 
Fishery Bulletin 119(1) 
projector are comparable in providing ages for Atlantic 
and Gulf menhaden. 
Our study used 3 bias tests to determine consistency in 
age readings, which were not consistent across all tests 
for Atlantic and Gulf menhaden. Bias in age estimates 
in our study is higher for some years than for others and 
higher for older samples than for more recent samples. For 
Atlantic menhaden, the only significant bias test was the 
Bowker’s test, which can be sensitive to one pair of age 
estimates with a large difference (McBride, 2015). Bias 
in the age estimates for scale subsamples indicates that 
use of the Eberbach projector results in less consistency 
in age readings, and the lack of bias in the readings of 
subsamples done with the microscope indicates that the 
use of the microscope lends well to consistency. During a 
workshop on aging Atlantic menhaden conducted in 2015, 
comparisons of ages between reader 2 and 10 laboratories 
yielded inter-laboratory and intra-laboratory P-values of 
0-1 depending upon the test (ASMFC°). 
Tests of symmetry for age data are often variable, with 
not all tests providing the same indication of bias or lack 
thereof. Results of some studies indicate no bias across 
tests (Goldman and Musick, 2006; Harry et al., 2011), 
but other studies have variable results pertaining to bias 
depending upon the test, reader, or age structure used 
(Harry et al., 2013; Baudouin et al., 2016; Stewart and 
Ogle, 2016). McBride (2015) used 100 samples across 20 
ages with 5 samples per age and found that bias was sub- 
ject to varying rates of type I and type II errors depending 
upon the underlying simulations. Given the sample size 
used by McBride (2015), the question becomes how sample 
size might affect those results. Campana et al. (1995) sug- 
gested that bias could be ignored when bias was low. Given 
the low numbers of age-4 fish and the expectation that the 
bias may have primarily crept in through ages estimated 
with the Eberbach projector, results of this study indicate 
that using the microscope for age estimation of menha- 
den species has a lower chance of introducing bias to the 
age estimates than using the Eberbach projector. Also, the 
main use of the age data from the long-term sampling pro- 
gram is to provide age compositions to stock assessments, 
and the simultaneous multinomial confidence intervals 
indicate that the composition data will not be affected by a 
switch to aging on the microscope. 
Scale distance measurements changed a small degree 
as the menhaden program of the Beaufort Laboratory 
switched from blue cards to the sonic digitizer pen to the 
microscope; many of the differences were not significant. 
With each update of the method, measurements may 
have become more precise. For example, for the method 
that includes the use of blue cards, the cards are lined 
up with the scale’s focus on a slanted screen below waist 
height, and annuli are ticked off with a pencil. Consis- 
tently placing the blue card ruler slightly below the focus 
that would have been selected with the sonic digitizer pen 
could lead to consistent, slightly larger measurements. 
Updating to the sonic digitizer pen decades ago allowed 
pen-point precision for measurement selection. Further 
updating of the method to the use of a microscope has 
allowed a clearer and full view of the scale at eye level, 
use of a straight guideline for annulus selection, more 
precise selection of measurement points, and use of imag- 
ing software to view the measurements. The scale dis- 
tance measurements indicate that similar structures are 
being measured to indicate true annuli. These data are 
not used in stock assessments. If these data are used for 
future research projects, the decreases in annulus mea- 
surements with changes in method should be taken into 
account. 
In general, age estimates for Gulf menhaden had lower 
agreement, and the scale samples of this species were 
more difficult to read than those of Atlantic menhaden. 
Several factors, including scale size, consistency in sam- 
pler over time, and environmental conditions, could have 
contributed to these differences. First, individual Atlantic 
menhaden and their scales tend to be larger in size than 
Gulf menhaden and their scales. Larger sizes could make 
it easier for the port sampler to take samples from the 
proper location on the fish, where scales are more regu- 
lar in shape and, therefore, more readable. Larger scales 
could result in more contrast and distance between per- 
ceived annuli, making them easier to read. Second, the 
same full-time employee has been the sampler of Atlantic 
menhaden since 1992. In the Gulf of Mexico, budgetary 
constraints have resulted in regular turnover of con- 
tract personnel that are selected, hired, and trained on 
an annual basis. Personnel differences would result in a 
consistently higher level of expertise in sampling at the 
port for the fishery in the Atlantic Ocean than that for 
the fishery in the Gulf of Mexico. Third, Atlantic menha- 
den caught in the fishery are exposed to environmental 
conditions that are more seasonally variable than those 
to which Gulf menhaden are exposed, given the differ- 
ences in latitude included in their habitat range. Such 
differences in the habitats of other species have been 
reported to result in more contrast between rapid- and 
slow-growth periods (Hoxmeier et al., 2001; Green et al., 
2009). The environmental differences could result in rel- 
atively higher-contrast scale circuli in Atlantic menha- 
den than the circuli of Gulf menhaden, therefore making 
them easier to age. 
Overall, on the basis of the results of this study, we 
have determined that the use of scales on a microscope 
is an acceptable method for aging menhaden. Age esti- 
mates from use of the Eberbach projector were repeat- 
able on the microscope, allowing continuity in the age 
data that have been maintained by the Beaufort Labora- 
tory for decades. The microscope has many advantages. 
First, stage, light, and magnification options offer mul- 
tiple ways to view samples more clearly while allowing 
for ergonomic considerations. Second, image analysis 
software used with a microscope allows more accurate 
straight-line measurements in units of the International 
System of Units. Third, because most laboratories have 
microscopes, samples can be viewed in a more uniform 
manner across facilities, making collaborations and 
comparisons easier. Fourth, data are more accessible 
and analysis of data is quicker because the data are no 
