Specialty Timbers New ZealandImplementation of genomic selection in provenance/progeny test of Douglas-fir
By Jaroslav Klápšte, Gancho Slavov, Toby Stovold, Heidi Dungey, June 2019.
Download SWP-T084 (pdf)
Executive summary
Two Douglas-fir progeny trials planted in Kaingaroa and Gowan Hill in 1996 were assessed in 2007 and re-assessed in February 2017. Since this material contains genetically broad material coming from populations across Oregon and California, these two progeny trials were also selected as training populations in a genomic selection project for Douglas-fir.
Since the business model connected to Douglas-fir marker array allows for genotyping of each individual for $75 USD, the strategy for genotyping was based on a partial genotyping effort equally distributed across all open-pollinated families (~2,200 individuals in total) and thus a single-step genetic evaluation approach combining phenotypic, pedigree and genomic data was implemented in this study.
Implementation of genomic resources developed by Oregon State University in the evaluation of New Zealand Douglas-fir breeding populations in this study resulted in substantially improved prediction accuracy and response to selection compared with pedigree-based analysis. An additional increase in the response to selection was found when only ancestry informative markers were used in the analysis of traits with strong population differentiation (e.g. DBH). Thus, the currently available SNP array appears to be a useful genotyping platform for the New Zealand Douglas-fir breeding program.
Implementation of the metafounders approach (i.e. inference of relatedness between pedigree founders of the provenance/progeny trial) resulted in increased prediction accuracy not only for genotyped but also for non-genotyped individuals. However, more complex modelling of population demography resulted in a reduction in model fit and lower prediction accuracy compared with a simple model with a single metafounder population. Therefore, reliable modelling of population structure in forest trees is challenging, even with the availability of abundant genetic marker data. The important finding from this study is that consideration of the distance of populations from native populations is important when building an implementation strategy for genomics.
No posts yet
Add a post