Potential sources of bias in the climate sensitivities of fish otolith biochronologies

Abstract

Analysis of growth increments in the hard parts of animals (e.g., fish otoliths) can be used to assess how organisms respond to variability in environmental conditions. In this study, mixed-effects models were applied to otolith data simulated for two hypothetical fish populations with assumed biological parameters and known growth response to environmental variability. Our objective was to assess the sensitivity of environment-growth relationships derived from otolith biochronologies when challenged with a range of realistic ageing errors and sampling regimes. We found that the development of a robust biochronology and the precision of environmental effect estimates can be seriously hampered by insufficient sample size. Moreover, the introduction of even moderate ageing error into the data can cause substantial underestimation of environmental sources of growth variation. This underestimation diminished our capacity to correctly quantify the known environment-growth relationship and more generally will lead to overly conservative conclusions concerning the growth response to environmental change. Careful study design, reduction of ageing errors, and large sample sizes are critical prerequisites if robust inferences are to be made from biochronological data.