4 O'Clock Forum: Using Prior information from Crop Variety Trials - A Bayesian Analysis of Barley Trials

Thursday, October 31, 2019 16:00
Raymond Building R3-048, 21111 Lakeshore Road, St Anne de Bellevue, QC, H9X 3V9, CA

4 O'Clock Forum continues this year's seminar series with a talk by Dr. Murari Singh, Assistant Professor, Concordia University. Everyone is welcome to attend.

In search of the high yielding and stable genotypes, crop improvement programs, including BIGM at ICARDA regularly evaluate crop variety trials in block designs, and generate a lot of information on genotypic and environmental variability. The frequentist approach of analysis followed in common practice uses only the current data but ignores the prior information available from the ongoing program. Bayesian approach makes use of such prior information by integrating it with the likelihood of current data. In doing so, it not only widens the statistical inference base and but also strengthens the estimates of parameters of interest such as genetic gain due to selection, heritability and predicted means, with more realistic estimates of precision. The yield data were generated from a winter barley trial conducted in resolvable incomplete block designs for 30 genotypes during 2011-12 season at Tel Hadya, Aleppo, Syria.  Bayesian analysis was carried out by developing various computing segments using Genstat,WinBUGS and R software codes. Although not guaranteed by the nature of the Bayesian approach itself, its application on the barley yield data, using priors determined from 20 similar trials, led to higher precision of predicted means of genotypes fromthe current trial, higher heritability and genetic gain due to selection. This implies that through Bayesian approach an enhanced breeding progress, or a relatively shorter breeding cycle, can be achieved by incorporating prior information which are routinely collected.


Dr. Murari Singh is an Assistant Professor on a Limited-Term Appointment at the Department of Mathematics and Statistics at Concordia University. Dr. Singh has generated his consultancy, research and training interest in plant biometrics primarily at IASRI, New Delhi; ICRISAT, Patancheru, India, and ICARDA, Aleppo, Syria and Amman, Jordan. More specific areas of his activities include design and analysis of experiments/surveys: G×E interaction, heritability, genotypic correlation, scaling tests, diallel crosses, diversity indices, molecular marker data analysis (optimum sample size); long-term trials, intercropping; spatial and temporal analysis of field trials; applied Bayesian analysis; training in biometrics, statistics and associated computing. This talk is based on a published study on barley trials conducted at ICARDA, Amman, Jordan, where he served as a Senior Biometrician before his retirement.

Back to top