Effect of non-normality and inadmissible estimates on estimation of heritability

Abstract

Heritability is one of the most important genetic parameter widely used in plant and animal breeding genetic improvement studies. In literature, several methodologies are available for estimation of heritability for different experimental situations. Unfortunately none of these provide always a valid estimate of heritability and the estimate is so inadmissible that no conclusions can be drawn for the inheritance of the trait under consideration. Further in particularly, there is no unique methodology, which is suitable for estimation of heritability in unbalanced situations. Keeping this in view , need has been felt that with the help of computer, the sensitivity and robustness of the very widely used genetic parameter, heritability might be studied at length. The sensitivity here refers to that how the estimate of heritability is dependent upon the aberrant or outliers. The paper contains some of the results as obtained by Bhatia et al (2003).