Estimating age of sea otters with cementum layers in the first premolar

We assessed sources of variation in the use of tooth
cementum layers to determine age by comparing counts in premolar tooth
sections to known ages of 20 sea otters (Enhydra lutris). Three
readers examined each sample 3 times, and the 3 readings of each
sample were averaged by reader to provide the mean estimated age. The
mean (SE) of known age sample was 5.2 years (1.0) and the 3 mean
estimated ages were 7.0 (1.0), 5.9 (1.1) and, 4.4 (0.8). The
proportion of estimates accurate to within +/- 1 year were 0.25, 0.55,
and 0.65 and to within +/- 2 years 0.65, 0.80, and 0.70, by reader.
The proportions of samples estimated with >3 years error were 0.20,
0.10, and 0.05. Errors as large as 7, 6, and 5 years were made among
readers. In few instances did all readers uniformly provide either
accurate (error <1 yr) or inaccurate (error >1 yr) counts. In most
cases (0.85), 1 or 2 of the readers provided accurate counts.
Coefficients of determination (R2) between known ages and mean
estimated ages were 0.81, 0.87, and 0.87, by reader. The results of
this study suggest that cementum layers within sea otter premolar
teeth likely are deposited annually and can be used for age
estimation. However, criteria used in interpreting layers apparently
varied by reader, occasionally resulting in large errors, which were
not consistent among readers. While large errors were evident for
some individual otters, there were no differences between the known
and estimated age-class distribution generated by each reader. Until
accuracy can be improved, application of this ageing technique should
be limited to sample sizes of at least 6-7 individuals within age
classes of >/=1 year.