Abstract
Understanding changes in abundance and survival in Atlantic salmon populations requires knowledge of growth rates and age. Salmon are typically aged through scale reading, but such estimates are rarely validated against age-verified fish from the wild. Here, we present a unique dataset of scales from 254 PIT-tagged Atlantic salmon with known sea-age. In addition, the freshwater age is known for 81 of these fish, through genetic parent-offspring identification. This dataset was used to estimate precision and bias in age readings and back-calculated length, as estimated by three independent experienced salmon scale readers. Overall, readers had an accuracy of 97.1% for sea-age and 71.7% for freshwater-age. For sea-age, scale reading was less accurate for salmon that had spent 2 or more years at sea than for salmon that had spent 1 year at sea. Freshwater age did however not affect scale reading accuracy. None of the scale readers erroneously misclassified freshwater- or sea-age with more than one year, and there was no significant pattern of misclassified ages to be under- or overestimate by the scale readers. Back-calculated smolt length was significantly different to length when measured as a smolt prior to seaward migration: it was shorter than the measured body-length for small smolts and longer for large smolts. This unique dataset, including the age-validated images of all scales, is now made openly available providing an important resource for training and testing salmon scale readers globally.