/AnMtgsAbsts2009.55426 The Use of Percent Checks and Its Impact On Genetic Parameters and the Prediction of Future Performance.

Wednesday, November 4, 2009: 10:15 AM
Convention Center, Room 337-338, Third Floor

Kelly Robbins, Jan Erik Backlund, Daniel Z Caraviello, Fabiano V C Pita and Jennifer C Mingus, Dow AgroSciences, LLC, Indianapolis, IN
Abstract:
The use of percent checks (%C) as an indicator of performance is common in plant breeding applications because it gives a standardized (adjusted based on the magnitude of the trait) measurement of the performance of hybrids relative to proven commercial lines; however, %C represents a mathematical transformation of trait of interest, and as such, can introduce unwanted variation into the selection process. Using derivations of the expected heritabilities of %C and the untransformed trait (UT) values, it is shown that the heritability of UT should always be greater than or equal to that of %C. To determine the extent of this difference, the heritabilities of UT and %C were calculated for five datasets using variance components estimated via REML. Results showed that estimates of heritability for UT were substantially higher when compared to the heritability of %C, resulting in more reliable estimates of genetic merit for UT. To evaluate the impact of using UT for selection decisions, when compared to %C, spearman correlations of line rankings were calculated for both general and specific combining abilities. In general, these correlations showed good agreement between methods, but examination of the top performing lines indicated the use of UT versus %C could impact decisions on line advancement.