A Spurious Correlation in an Interpopulation Comparison of Atlantic Salmon Life Histories

We tested two hypotheses concerning geographical variation in Atlantic salmon (Salmo salar) life histories: (1) mean age at first reproduction is positively correlated with growth rate at sea and (2) within-population variation in age at first reproduction first increases and then decreases with latitude. Data on growth and age at first reproduction were compiled from 41 populations in eastern North America. Data reliability was checked by a redetermination of ages based on scale examination. The proportion of fish that were incorrectly aged was small (°0.7%); however, aging errors were primarily of one kind; salmon that had previously spawned were misclassified as virgin fish of an older age class. Growth rate at sea was found not to be positively correlated with age at maturation. Schaffer and Elson's (1975) positive correlation between growth and age at first reproduction can be attributed to a subtle statistical artifact caused by aging errors. We also found that within-population variation of age at maturation was not related to latitude. We conclude that tests of life history theories should not assume constancy in life history traits, such as mortality, among populations.