Date of Award
Doctor of Philosophy (PhD)
Cynthia M. Jones
Dayanand N. Naik
Identifying the natal sources of fish is an important step in understanding its population dynamics. Adult recruits are often sourced from multiple nursery areas, with good quality locations contributing disproportionately more fish to the adult stock. Because population persistence is strongly influenced by nursery habitat, methods that correctly identify the source of recruits are necessary for effective management. Within the last decade, otolith chemistry signatures have been increasingly used as a natural marker to delineate fish from a mixture of nursery sources. Despite the widespread use of otolith trace element and stable isotope ratios as habitat markers, the statistical approaches to handle these data have been slow to develop. Limited guidelines have been offered for constructing the discriminatory function in terms of the number of chemical variables used, the information conveyed by each variable, and the overall stability of important variables with time. Almost all studies argue that juvenile signatures must be collected anew each year at considerable expense. In this study, Rao's test for additional information was used to identify the most useful discriminatory variables for identifying the nursery seagrass habitats for spotted seatrout (Cynoscion nebulosus) in Chesapeake Bay. Additionally, Akaike information criterion (AIC) and Bayes information criterion (BIC) were used to select the discriminant function analysis (DFA) model that minimized the prediction error for determining provenance of adult fish. The AIC technique was also used to construct a short-term multi-year habitat tag for the Bay. Variable selection using Rao's test show that classification accuracy was heavily dependent on the type and number of variables used in the model. Barium was the most important variable and it was the most stable variable over time. From the AIC model selection, adult fish were correctly assigned to nursery area with over 94% accuracy within-year, while the AIC multi-year tag accurately identified the source of historical collections of adult fish with over 80% classification accuracy. These results show that by using correct statistical approaches to construct the discriminatory model, the probability of misclassification for subsequent survivors is minimized. Additionally multi-year models can be developed, directing research for other species.
Beharry, Stacy K..
"Evaluating Methods for Optimizing Classification Success From Otolith Tracers for Spotted Seatrout (Cynoscion nebulosus) in the Chesapeake Bay"
(2011). Doctor of Philosophy (PhD), dissertation, Ocean/Earth/Atmos Sciences, Old Dominion University, DOI: 10.25777/7kg5-8y61