Pedigree Relationships in the University of Florida Southern Highbush Blueberry Germplasm

The blueberry breeding program at the University of Florida (UF) currently uses a phenotypic recurrent selection strategy for trait improvement. This results in overall genetic improvement of the population, but gains are typically slow with the selection cycle ranging from 10 to 15 years. Choosing superior parents at an earlier stage could result in improved gain from selection, especially for low heritability traits. Best linear unbiased prediction (BLUP) analysis is being used with great success in the animal breeding industry for parental selection. BLUP analysis uses available pedigree information to estimate additive breeding values which can be used to more effectively select parents.  We used pedigree information dating back to 1909 to construct an A-matrix that described the genetic relationship of the UF blueberry breeding program germplasm.  However, the current A-matrix model assumes disomic inheritance and doesn’t account for double reduction, which is a common phenomenon in autotetraploid species. To determine the level of double reduction, several A-matrices were constructed assuming disomic and tetrasomic inheritance. Using these matrices, we identified the most frequent genotypes in pedigrees, estimated inbreeding coefficient, and calculated the proportion of double reduction, The mean value of the A-matrix diagonal was 1.052, when assuming a tetraploid model with no double reduction, which indicates the presence of inbreeding in the population. The three individuals most represented in the germplasm were Earliblue, Windsor, and E-30. This was the same across all matrices. The two models estimated different amounts of inbreeding, but the individuals with the highest level of inbreeding were the same in both models. We are collecting phenotypic data for a range of traits and populations to fit a BLUP model comparing various levels of double reduction. These analyses are necessary to enable BLUP analysis and the use of genome-wide prediction models for blueberry breeding.