Statistical analysis on otolith data of anadromous fishes

Abstract

Stable oxygen and carbon isotope (δ18O and δ13C) analyses of otoliths are becoming increasingly common in fisheries science and management. However, little is known about the statistical properties of isotopic data and few attempts have been made to explore appropriate statistical methods that could be used for otolith data analysis. In this paper, we present a pilot study on δ18O and δ13C data from otoliths of two anadromous fish species, Atlantic salmon (Salmo salar) and Pacific sockeye salmon (Oncorhynchus nerka). The results indicated that the salmon otolith data were not normally distributed, so that linear discriminant function analysis and commonly-used statistical tests such as ANOVA and the t-test may not be appropriate. Using non-parametric k-sample nearest neighbor discriminant analysis, we were able to discriminate with high accuracy among five hatcheries for Atlantic salmon and the origins of wild and hatchery sockeye salmon. Analyses also indicated that the sample sizes required to estimate δ18O and δ13C means based on the different sources of variability (between group or within group) and precision levels (≤ ±5.0 %) were not large. These results and conclusions not only address the statistical considerations of isotopic data from otoliths, but also have practical importance for fisheries management as well.