Abstract
The objective of this work was to evaluate the efficiency of genomic selection in different population structure scenarios in popcorn, by estimating the effects of and using markers in the same reference population and in populations not related to each other, and to determine the effect of the effective size and kin relationships on the estimation population over the prediction accuracy. Populations were simulated with different linkage disequilibria (LD) and additive variances, taking into account different characters, marker densities, heritabilities, and generations, totaling 144 scenarios. Populations structured in progenies were also simulated. The prediction accuracy of additive breeding values was obtained by correlating true breeding values and breeding values predicted by RR-BLUP (ridge regression-best linear unbiased prediction). In populations with low LD and lower additive variance, the use of higher SNP density (10 SNP 0.1 cM-1) is recommended; besides, the selection candidate should be related to the estimation population in order to obtain satisfactory prediction accuracy. The use of a selection population in the same generation of the estimation population reduces accuracy in at least 8%. Populations structured in progenies with stronger relationships and smaller effective size increase the efficiency of genomic selection.
Index terms:
Zea mays; prediction accuracy; genomic selection efficiency; breeding value prediction; kin relationship; effective size
