Error patterns in age estimation and tooth readability assignment of grey seals (Halichoerus grypus): Results from a transatlantic, image-based, blind-reading study using known-age animals

Abstract

We analysed error patterns in a first interlaboratory grey seal (Halichoerus grypus) age-reading experiment. The experiment involved ten readers, who estimated age using images of cementum growth layers from teeth of 68 known-age seals (0-22 years). The percentages of correct estimates ranged from 32.4% to 60.3% among readers, and 89.3% of all errors were by ±1-2 years. Six readers showed increasing underageing with increasing seal age. An elevated risk of underestimation by 1 year occurred in teeth collected 0-5 months after breeding and was attributed to more frequent absence of a distinct growth layer for the new year and lack of information on months between the last birthday and the date of sample collection (plusmonths). For plusmonths 6-11, positive bias was predominant, suggesting that overestimation is the more common error when plusmonth information is available. Readers assigned readability scores to the tooth sections, and 79.1% of all ageing errors occurred in sections of low or intermediate readability. Excluding these sections would, however, also exclude 43.0% of all correct estimates. Neither levels of age estimation error nor predictive values of readability assignments were associated with reader experience levels. Analyses of image markings identified common errors in delineations of annual increment layers. © 2012 International Council for the Exploration of the Sea.