Skip to main content
Rice
Fig. 2 | Rice

Fig. 2

From: Optimization of Multi-Generation Multi-location Genomic Prediction Models for Recurrent Genomic Selection in an Upland Rice Population

Fig. 2

Predictive ability (LSmeans with error-bars representing the standard error) within the Uni1 and Multi1 scenarios with three different models. We used the multi-site model without genotype by environment interaction (MM) and the multi-site model including the genotype-by-environment interaction with similar variances between environments (MDs) or with different variances between environments (MDe). Calibration and validation were performed within the PCT27B population phenotyped at the S0:4 generation in Santa Rosa (SRO) for the four traits of interest: flowering date (FL), plant height (PH), grain yield per plot (YLD) and grain zinc concentration (ZN). TS included 100% of the records in Palmira (PAL) and 70% of the records in SRO for all the models except SM where the TS included only 70% of the phenotypes recorded in SRO. For all models VS was 30% of phenotypes in SRO. Within a trait, the letters represent significant differences between models (p-value < 0.05)

Back to article page