Consequences of inappropriate criteria for accepting age estimates from otoliths, with a case study for a long-lived tropical reef fish

Fish ages estimated from increments in otoliths are uncertain because of various sources of error, including increment interpretation. Interpretation error is often addressed by reading each otolith multiple times and accepting age estimates only if readings satisfy certain consistency criteria. Choice of an inappropriate acceptance criterion may significantly bias the accepted age estimates and derived parameters such as mortality. The frequencies and magnitudes of discrepancies from replicate readings of otoliths increased with age for the red bass, Lutjanus bohar. The trend was best described by a constant probability of misinterpreting each increment, indicating an age acceptance criterion that allowed for increasing discrepancy between readings with age. Simulations of three error processes in reading otoliths, two processes of error accumulation within readings, and six acceptance criteria illustrated the biases in age-based metrics that arise from choosing inappropriate acceptance criteria. Biases were largest for static constant, rather than proportional, acceptance criteria, leading to elevated exclusion of older otoliths, overestimation of mortality, and underestimation of mean age. von Bertalanffy growth parameters were generally estimated with little bias. We recommend formal analysis of alternative models of ageing error to choose appropriate acceptance criteria and minimise biases in age-based demographic metrics.