MEPS 636:169-187 (2020)  -  DOI: https://doi.org/10.3354/meps13229

ABSTRACT: Otolith chemistry is frequently employed in the reconstruction of fish environmental histories. While some elements have been strongly correlated with environmental factors (e.g. salinity, temperature, water chemistry), others may not indicate exogenous factors and simply add endogenous variability to a data set. Several commonly assessed elements were previously identified as being only present in the proteinaceous fraction of endolymph from black bream Acanthopagrus butcheri, suggesting that the choice of elements in otolith multi-elemental fingerprinting could influence their utility as natural environmental markers. To test this hypothesis, we performed several cluster analyses based on different sets of trace element data extracted from both the core and larval region of otoliths of juvenile black bream. We clustered in 3 different ways: (1) all elements analysed; (2) elements identified as being primarily in the salt fraction of the endolymph (i.e. inorganic); and (3) elements identified as being associated with the protein fraction of the endolymph. We subsequently assessed the power of the resulting clusters to resolve cohort identity and/or collection location based on differences in otolith chemistry using multinomial logistic regression. Our results indicate that clustering based solely on salt-fraction elements is best for resolving spatio-temporal variability in spawning sources and larval nursery habitats of black bream.

KEY WORDS: Gippsland Lakes · Nominal logistic regression · Hierarchical clustering · Sparid · Natural environmental tag · Sagittae