Accurate classification of juvenile weakfish Cynoscion regalis to estuarine nursery areas based on chemical signatures in otoliths

We investigated the ability of trace element and isotopic signatures in otoliths to record the nursery areas of juvenile (young-of-the-year) weakfish Cynoscion regalis from the east coast of the USA. Juvenile C. regalis were captured with otter trawls at multiple sites in Doboy Sound (Georgia), Pamlico Sound (North Carolina), Chesapeake Bay (Virginia), Delaware Bay (Delaware) and Peconic Bay (New York), from July to September 1996. One sagittal otolith from each specimen was assayed for Mg/Ca, Mn/Ca, Sr/Ca and Ba/Ca ratios using inductively coupled plasma mass spectrometry (ICP-MS), while delta(13)C and delta(18)O Values from the other sagittal otolith in the pair were determined using isotope ratio mass spectrometry (IR-MS). A multivariate analysis of variance determined that there were significant differences in trace element signatures among locations. Bootstrapped 95% confidence ellipses on canonical variates indicated that all 5 locations were significantly isolated in discriminant space. On the basis of these differences, linear discriminant function analysis (LDFA) and artificial neural network (ANN) models were used to classify individual fish to their natal estuary with an overall error rate of 37 % for LDFA and 29.6% for ANN. Addition of delta(13)C and delta(18)O values to the LDFA and ANN models derived from the trace element data resulted in overall error around 10%. We will, therefore, be able to use chemical signatures from the juvenile portion of adult C. regalis otoliths to accurately classify these fish to their natal estuary.