Long-term shifts in otolith age interpretations

Multi-decadal time series using existing age data are increasingly being used to test important issues in ecology. Over the long term however, shifts in age determination protocol are almost unavoidable. This study uses an 86-yr time series of Icelandic cod (Gadus morhua) otolith samples (n = 906) to rigorously test the assumption of inter-year comparability of age interpretation across decadal changes in otolith preparation, viewing protocol and age reader. Comparison of the original age estimates with re-ageing of the same otoliths using modern preparation and interpretation protocols and a single age reader revealed a consistent bias, with the original readings over-estimating ages relative to modern. The extent of the bias was minimal at age 8 (0.18 yr), increasing to an average divergence of about 2 yr after age 14. The ageing bias was linked most strongly to the individual ager, whereas the transition from unsectioned (cracked or whole) to sectioned otoliths markedly improved precision. The implications of this study are encouraging, in that they suggest that old data sets incorporating historic age determinations based on specific methods may well be adequate for many purposes.