Abstract
The goals of this dissertation are to provide new data on white marlin life history through development of an age and growth model and to explore whether gains in assessment precision and confidence can be realized by applying more complex models using this information. Cross-sections of anal fin spines were collected from fisheries Venezuela, Brazil, and the USA and used to age white marlin (Tetrapturus albidus, also known as Kajikia albida). A novel technique based on k-means clustering was developed to correct for the loss of rings in the spine sections due to vascularization. These corrected ages, along with information on size, sex, and location of capture were used to fit a von Bertalanffy growth curve with nonlinear least squares regression. Likelihood ratios and the Akaike Information Criterion (AIC) were used to test the reductions in residual error between the full and reduced model to examine the effects of including sex or area as a covariate. Statistically significant differences were found between the growth curves of males and females, but not between fish from the northern and southern hemispheres, although the latter may be due to the low sample size from the southern hemisphere. A single growth curve was used to inform an age-structured surplus production model fit to lumped biomass estimates of catch and a multi-fleet composite CPUE. The population estimates agreed with the results of the 2002 assessment, indicating the stock was highly overfished and suffering overfishing at the end of the 20th century.