394 
Fishery Bulletin 99(3) 
Hence, radiometric age was determined on the basis of 
measured activity ratios and no adjustment for exogenous 
210 Pb was necessary. When the radiometric and annulus 
reading (±12% CV) age ranges were considered graphical- 
ly, most of the data points still did not agree with the ex- 
pected ingrowth curve (2? 0 =0.0). 
A direct comparison of annulus-derived age with radio- 
metric age indicated that the ages were evenly distributed 
on either side of a line of agreement and that the slope of 
the regression (slope=0.915) was close to 1 (Fig. 2). Radio- 
metric age was not statistically different from the annu- 
lus-derived age estimates (paired two-tailed t-test, df=14, 
t=0.4181, P=0.6822). The coefficient of determination was 
low (adjusted /- 2 =0.55; Kvalseth, 1985) and the power of 
the test to detect significant differences was low because 
of the variability of the data and the small sample size. 
The analytical uncertainty associated with radiometric 
age determination or the reading range of the annulus- 
derived age encompassed the line of agreement for only 
four samples. The sample with the closest agreement was 
a female (1780 mm FL) whose annulus-derived age was 
36 years. The annulus-derived age estimates of the juve- 
nile age groups were high by 1.9 and 2.4 years. For older 
age groups and individual cores, annulus-derived age was 
higher than radiometric age by as much as 21.8 years and 
lower than radiometric age by as much as 23 years. Otolith 
sections that had radiometric and annulus-derived age es- 
timates that differed considerably were photographed to il- 
lustrate the difficulty in aging otoliths (Fig. 3). 
Von Bertalanffy growth functions fitted to the radiomet- 
ric ages of males and females (Figs. 4 and 5) were similar 
to annulus-derived growth functions determined by Crab- 
tree et al. (1995). However, the low number of radiometric 
data points resulted in large confidence intervals (±95%) 
for growth model parameters. Radiometric-age growth pa- 
rameters for males indicated an asymptotic length (LJ of 
1550 ±83 mm FL and a growth coefficient ( k ) of 0.19 ±0.12 
(Fig. 4). Radiometric-age growth parameters for females 
indicated an of 2030 ± 227 mm FL and a growth coef- 
ficient ( k ) of 0.08 ±0.04 (Fig. 5). 
Discussion 
In four cases, the radiometric age range encompassed the 
annulus-derived age estimate. The oldest female tarpon 
had an estimate (55 yr) that was lower than the radio- 
metric age by 23 yr. The radiometric age range, however, 
encompassed the annulus-derived age by 4 years. This 
may indicate that the annulus-derived age estimate was 
correct in this case. The radiometric age val- 
idation, by itself, confirms the longevity of 
female Atlantic tarpon to at least 50.6 yr, but 
it may indicate that age can meet or exceed 
78.0 yr (Table 3). Similarly, the longevity 
of male Atlantic tarpon was confirmed to 
at least 31.8 yr, but may exceed 41.0 yr. 
When the CV for annulus-derived age was 
included in the range of age estimates, some 
additional samples encompassed the radio- 
metric age estimates, but most deviant ages 
could not be explained by this variation. 
The wide dispersion of residuals and the 
low coefficient of determination indicated 
that there were differences between radio- 
metric age and annulus-derived age that 
could be explained in the interpretation of 
otolith growth zones (Fig. 2). According to 
Crabtree et al. (1995), aging otoliths of the 
Atlantic tarpon was confusing and many 
were not readable. Although Crabtree et 
al. (1995) validated the annual periodicity 
of growth zone formation for 12 out of 18 
young fish up to an age of 9 yr in captivity, 
the pattern of growth zone formation in 
older, wild fish may not be annual. Because 
the tarpon is a migratory fish that inhab- 
its inshore and estuarine waters of vary- 
ing salinity and temperature and spawns 
offshore (Zale and Merrifield, 1989; Crab- 
tree et al., 1992), there is a high potential 
for irregular growth zone formation from 
the stresses of extreme changes in habitat 
(Pannella, 1980; Campana, 1983). Suban- 
nual growth zones may explain annulus- 
100 
80 -- 
60 -- 
o 40 -- 
20 -- 
S 
_o 
•3 
c4 
Upper limit undefined 
Line of agreement 
Regression 
• Male 
o Female 
y = 0.9 15r -0.085 
Adj r 2 = 0.55 
20 40 60 
Annulus-derived age (yr) 
80 
100 
Figure 2 
Comparison of male and female Atlantic tarpon (Megalops atlanticus) 
annulus-derived ages and radiometric ages. A linear regression and a line 
of agreement are plotted for comparison. Vertical bars represent low and 
high radiometric age estimates based on the analytical uncertainty of 210 Pb 
and 226 Ra measurements. Horizontal bars represent the range of annulus- 
derived ages for age groups. The upper limit of the radiometric age esti- 
mate for the oldest female is undefined because the analytical uncertainty 
of 210 Pb: 226 Ra exceeds 1.0. 
