I have a lot of plant-breeding data (tens of thousands of results) that are composed of phenotypes of parents and offspring. The plants however went trough years of selection and it is not balanced. Moreover, I have a lot of inbreeding data. What is the best way to estimate heritability from this type of data, having the exact pedigree relations and the phenotypes (But no accurate environmental data). Moreover, how would you estimate heritability from nominal traits (like color or resistance)?

  • 3
    $\begingroup$ The (biometric) heritability of a trait is the slope of the regression line of the mean phenotype of offspring on their mid-parent. When you have multiple traits, you need to use a general linear model rather than a simple linear regression. The GLM accounts for underlying genetic correlations among traits. Nominal data poses no problem, you simply assign numeric values to each category. For example, red = 1, blue = 0. For the relevant theory, I would consult Falconer & Mackay (1996) "Introduction to quantitative genetics". Also look up the 'multivariate breeder's equation'. $\endgroup$ May 15, 2016 at 23:31
  • 2
    $\begingroup$ d.umn.edu/~jetterso/Ecological%20Genetics/documents/… $\endgroup$
    – rg255
    May 16, 2016 at 7:00
  • $\begingroup$ Thank you very much, it was very helpful. I went on trying with the REML/MCMC methods, and will check how it goes. $\endgroup$ May 17, 2016 at 8:45


You must log in to answer this question.

Browse other questions tagged .