An Analysis of Morphometrics of Bluefin Tuna

Elkana Ngwenya

Faculty of Fisheries and Marine Environment, Australian Maritime College, P.O Box 21, Beaconsfield, Tasmania, Australia 7270
Email: e.ngwenya@fme.amc.edu.au

It is generally argued that the weight-length relation is best described using a power function. The parameters of the power function are often estimated using simple ordinary least squares (OLS) techniques. However, various studies estimating the power relation suggest that diagnostic tests of OLS assumptions are ignored. In this paper, the OLS estimation of weight-length relationship in morphometric analysis is conducted, and diagnostic tests are used to identify and remedy any statistically significant violations of OLS assumptions. Analyses are based on data from the study by Shingu, Warashima and Matsizaki (1974). Eight morphometric characters, recorded for 200 bluefin tuna (Thunnus thynnus) caught by longlining, are used. Associations between morphometric variables are identified, quantified, and assessed for statistical significance. Hypothesis testing of parameters of the weight-length relationship is conducted. Model significance testing and diagnostic testing of ordinary least squares (OLS) assumptions are also conducted. The results show (i) low levels of correlation among most morphometric measures, and (ii) statistically significant parameter estimates of the length-weight measurements. Hypotheses testing of the parameters of the weight-length relationship suggest that the parameters of the power function are sensitive to corrections to violations of OLS assumptions. This sensitivity has implications for the use of estimated weight measurements in subsequent biomass estimation.

Room 3 Thursday 2.30 pm