On the use of the gonosomatic index

Abstract

1. 1. Using data collected from two teleostean fish species the assumptions of the gonosomatic index (GSI) were examined primarily by regression analysis and analysis of covariance. 2. 2. Standard length of the fish and standard length cubed can have a higher degree of correlation with ovarian weight as compared to body weight. Therefore, an ovarian weight-body weight ratio is not always the best way of expressing a gonadal index. 3. 3. When ovarian weight is regressed on some expression of body size, intercepts of the regression equations have a non-zero intercept. Furthermore, intercepts of regression equations from fish in different ovarian development stages are not always statistically homogenous. These findings are inconsistent with the GSI assumption of a zero intercept. 4. 4. The relationship of ovarian weight to body size changes with the stage of oocyte development. Thus, the GSI is an inaccurate means of comparing gonadal activity between experimental groups. 5. 5. When independent and dependent variables are subjected to log transformations, the relationship of ovarian weight to body size is less likely to vary with stage of oocyte development. Thus, expressing ovarian weight as an exponential function of some measure of body size may provide a more appropriate gonadal index. 6. 6. The analyses presented disclose that the GSI is not an accurate indicator of gonadal activity.