Schueller et al.: Using a projector and microscope to read scales from Brevoortia tyrannus and B. patronus 29 
Percent agreement values were all above 86% among age 
estimates for samples overall between the microscope and 
Eberbach projector and were all above 83% between first 
and second readings within an instrument. Our results 
are similar to those from McBride (2015), who reported PA 
of around 80% with no bias and low variability. Values of 
PA were high for reference collections of Atlantic and Gulf 
menhaden (95.8% and 100.0%, respectively). A workshop 
on aging Atlantic menhaden that used the same reference 
collection calculated PA values of 58.2-86.7% for compar- 
isons of microfiche with an Eberbach projector between 
10 state laboratories and of 91.8% for comparisons on the 
Eberbach projector between reader 2 and another reader 
(ASMFC®). These values are similar to the PA of 91.7% 
between reader 1 and reader 2 in this study. An exchange 
of 100 fishery-independent scales between reader 1 
and an aging technician at the University of Southern 
Mississippi, aged on the same Eberbach projector, yielded 
a PA of 79.0% (Leaf’). The PA values from our study were 
also similar to or higher than those from studies for other 
species, with PA values of 39.8—100.0% when precision of 
estimates for multiple aging structures were compared 
(Khan and Khan, 2009; Khan et al., 2015; Kumbar and 
Lad, 2016) and of 68.1-92.4% when estimated precision 
was compared between readers (Goldman and Musick, 
2006; Khan et al., 2019). 
Overall, comparisons of age estimates made with the 
microscope and the Eberbach projector resulted in APEs 
under 4.4%, which is similar to or less than published 
APEs of about 8.0% (Campana, 2001; McBride, 2015), 
including the mean APEs for estimates from a simula- 
tion by McBride (2015) that indicated no bias and that 
decreased with increasing precision. Average percent 
errors from other studies have ranged from 0.7% to 16.1% 
when age estimates were compared among multiple aging 
structures or between readers (Giirsoy et al., 2005; Khan 
and Khan, 2009; Khan et al., 2015; Khan et al., 2019). 
Average percent errors in age readings were best for the 
reference collections, which had better quality, more leg- 
ible scales than the scales taken from fish sampled for 
this study. An APE of 8.2% was calculated at a workshop 
during which aging was done for a reference collection 
of Gulf menhaden (n=30) and during which microscope- 
based age estimates from a group of 6 state laboratories 
were compared with ages estimated with the Eberbach 
projector provided by reader 1 (VanderKooy®). Age esti- 
mates for scale subsamples from Gulf menhaden in our 
study had APEs of 9.7% when compared across devices 
and of 8.7% when compared among readings by a reader 
on the Eberbach projector, values that are higher than the 
benchmark identified in Campana (2001). 
5 ASMFC (Atlantic States Marine Fisheries Commission). 2015. 
2015 Atlantic menhaden ageing workshop report, 47 p. Atl. 
States Mar. Fish. Comm., Washington, D.C. [Available from 
website.] 
v Leaf, R. 2018. Personal commun. Univ. South. Miss. 703 East 
Beach Dr., Ocean Springs, MS 39564. 
: VanderKooy, S. 2019. Personal comm. Gulf States Mar. Fish. 
Comm. 2404 Government St., Ocean Springs, MS 39564. 
High APEs may indicate difficulty in maintaining con- 
sistency in the ages determined for Gulf menhaden with 
the Eberbach projector because of sample readability and 
quality. One third of the subsamples from Gulf menhaden 
were from 2005, a year in which scales had reduced quality 
due to improper storage. Although the use of both pieces of 
equipment yielded similar numbers of unreadable scales 
for reader 1, 55% of samples aged on the Eberbach projec- 
tor, versus 35% of samples aged on the microscope, were 
noted to be fuzzy, dirty, or unreadable. Comparisons of 
subsamples of scales from Atlantic menhaden may have 
been affected by samples from the bait fishery (20% of the 
samples were from the bait fishery); samples from the bait 
fishery can be harder to read because of differences in fish- 
ing and sampling practices. Specifically, handling time on 
the vessel and in storage is greater for samples from the 
bait fishery than for samples from the reduction fishery, 
and fish are caught at different times of the year. Also, 
staff members that mount scales from fish sampled from 
the bait fishery typically have less experience. 
The increase in APE values when comparing between 
the first and second readings for reader 1 could be due 
to greater experience. In 2017, when the first reading 
was done, reader 1 had been estimating ages of menha- 
den for 2 years, long enough to become proficient but still 
early enough that another year’s experience would likely 
result in some level of improvement at the time of the 
second reading in 2018. However, although the APEs for 
the first readings were higher, they are on par with val- 
ues of around 10% that have been observed for harder to 
read samples and for new readers in other studies (Potts’; 
McBride, 2015). 
Values of ACV in this study were comparable to or less 
than values in simulation analyses (McBride, 2015), a 
workshop for aging Atlantic menhaden (ASMFC°), and 
studies on other species. The highest ACV value in this 
study was for the samples of Gulf menhaden taken in 
2005 (6.2%), and this value is within the range of values 
reported in McBride (2015) on the basis of simulating age 
data with no bias across multiple precision levels but is 
higher than the value of 5% found to be accepted across 
data from several laboratories reviewed by Campana 
(2001). Analysis during a workshop on aging Atlantic 
menhaden resulted in ACVs that are generally above 
5% among data from several laboratories but that do not 
exceed 15% (ASMFC°). The workshop participants com- 
puted an ACV of 2.1% (ASMFC%), a value that is similar 
to the ACV of 2.2% computed in this study for the same 
reference collection. Values of ACV for age data for both 
Atlantic and Gulf menhaden across all comparisons are 
similar to the values of 1.0-22.8% from other studies 
across aging structures and between readers (Khan and 
Khan, 2009; Herbst and Marsden, 2011; Khan et al., 2015; 
Khan et al., 2019). Values of ACV from this study indicate 
that the use of the microscope and the use of the Eberbach 
° Potts, J. 2014. Personal commun. Natl. Mar. Fish. Serv. 101 
Pivers Isl. Rd., Beaufort, NC 28516. 
