Lytton et al.: Age validation of Polyprion americanus based on bomb radiocarbon ( 14 C) 
81 
Reader 1 counts 
Age estimate (yr) 
Figure 2 
Bias plots of increment counts made by readers 1 and 2: (A) for identification of appro- 
priate specimens for bomb radiocarbon analysis before identification of strict increment 
identification protocols, and (C) for growth analyses. Coefficients of variation (CV) in age 
estimates made by (B) reader 1 and (D) reader 2. Error bars in panels A and C represent 
95% confidence intervals. Increment count-specific CVs increased (y=0.53x+9.86; coeffi- 
cient of determination [r 2 ]=0.12, P=0.005) with increment number before identification of 
strict aging protocols. For specimens analyzed for growth analysis, CVs decreased with 
increment count fy=-0.33x+25.47; r 2 =0.39, P<0.001). 
allowed for better resolution between tightly compact- 
ed increments that may have otherwise been grouped 
together. In many of the otoliths, readability decreased 
at various points along the chosen reading axis, forcing 
readers to shift to a new reading axis by following a 
growth increment along a lateral plane. 
Age validation 
Although increment counts for otoliths of wreckfish 
were relatively difficult to determine because incre- 
ments could be difficult to discern, no samples were 
identified as unreadable and all specimens (n=323) 
were included for possible selection for bomb radiocar- 
bon analysis. Lengths of the specimens ranged from 
880 to 1070 mm TL. Initial estimated increment counts 
ranged from 6 to 66 increments for reader 1 and from 
6 to 60 increments for reader 2. Count disagreements 
between readers ranged from 0 to 35 increments (aver- 
age disagreement of 9 increments). Readers produced 
identical counts only 4.3% of the time, were within 1 
increment 13.3% of the time, and were within 5 in- 
crements 45.7% of the time. Bias plots revealed that 
reader 2 counted fewer increments than reader 1 when 
reader 1 counted greater than 35 increments, indicat- 
ing a bias between readers in fish of older ages (Fig. 
2A). At increment counts of less than 35 for reader 
1, there was no bias pattern between readers. Coeffi- 
cients of variation (CV) increased significantly with age 
(P=0.005; Fig. 2B). The average CV between readers 
was 27.5% — a result that is likely due to the lack of 
validation-based aging criteria at this stage. 
The wreckfish used in bomb radiocarbon analysis 
(n=20) had increment counts ranging from 11 to 43, 
