CEES Extra seminar: Quantifying inbreeding depression: the tripple whammy of measurement error
By Lukas Keller, University of Zurich
Quantitative estimates of the magnitude of inbreeding depression in traits of interest is of central importance in evolutionary genetics, animal and plant breeding, and conservation biology. At the face of it, obtaining such estimates is labour intensive but seems straightforward.
Standardized methods using weighted linear regression have existed for nearly 60 years, and approaches using mixed models have become popular in the last two decades. However, like most statistical analyses in ecology, evolution, and behaviour, these methods currently ignore the measurement errors and uncertainties in covariates. Measurement error and uncertainties in covariates can lead to severely biased parameter estimates and reduced statistical power, known as the double whammy of measurement error. In this talk I will show that estimates of inbreeding depression are also affected by such biases because all measures of inbreeding, whether derived from pedigrees or molecular markers, contain uncertainties and measurement errors. I will introduce Bayesian methods that allow us to obtain correct parameter estimates in the presence of uncertainties, provided that good estimates of the uncertainties exist.
The latter is not always trivial. I will end with a discussion of the challenges that lay ahead when correcting for measurement error and uncertainties in statistical analyses in ecology, evolution, and behaviour.