## Article Figures & Data

### Figures

### Tables

## Supporting Information for Ould Estaghvirou, Ogutu, and Piepho, 2014

**Files in this Data Supplement:**- Supporting Information - Tables S1-S5 and Figures S1-S8 (PDF, 1 MB)
- Table S1 - The statistics, parameters and the symbols used to denote them in the text. (PDF, 112 KB)
- Table S2 - The variance components for the AgReliant real maize data set estimated by RR-BLUP models assuming genotypes are correlated according to the linear variance model. (PDF, 111 KB)
- Table S3 - The variance components for the KWS-Synbreed real maize data set estimated by RR-BLUP models assuming genotypes are correlated according to the linear variance model. (PDF, 111 KB)
- Table S4 -
Descriptive statistics for the difference between the heritability estimated using datasets with (
*r̂*^{2}_{g,ĝ,o}) and without (*r̂*^{2}_{g,ĝ}) outliers, taken as the benchmark, for the five methods (M1 to M5) in each of the 10 scenarios. (PDF, 114 KB) - Table S5 -
Descriptive statistics for the difference between the estimated predictive accuracy with (
*r̂*) and without (_{g,ĝ,o}*r̂*) outliers, taken as the benchmark, for the seven methods in each of the 10 scenarios. (PDF, 108 KB)_{g,ĝ} - Figure S1 -
Studentized residuals for yield for the small data set (
*n*=177 genotypes) contaminated with an outlier equal to five times the standard deviation of the residual error used to simulate the small datasets. (PDF, 162 KB) - Figure S2 -
Studentized residuals for yield for the large data set (
*n*=698 genotypes) contaminated with an outlier equal to five times the standard deviation of the residual error used to simulate the large datasets. (PDF, 167 KB) - Figure S3 -
Studentized residuals for yield for the small data set (
*n*=177 genotypes) contaminated with an outlier equal to 10 times the standard deviation of the residual error used to simulate the small datasets. (PDF, 154 KB) - Figure S4 -
Studentized residuals for yield for the large data set (
*n*=698 genotypes) contaminated with an outlier equal to 10 times the standard deviation of the residual error used to simulate the large datasets. (PDF, 155 KB) - Figure S5 -
Studentized residuals for yield for the small data set (
*n*=177 genotypes) contaminated with an outlier equal to eight times the standard deviation of the residual error used to simulate the small datasets. (PDF, 157 KB) - Figure S6 -
Frequency histograms of the deviations in the simulated true predictive accuracy
*r*from the estimated predictive accuracy for the datasets with_{g,ĝ}*r̂*(empty box and whiskers capped with brackets) and without_{g,ĝ,o}*r̂*outliers, regarded as the benchmark (green boxes), for each of the seven methods in Scenarios 1 to 3. (PDF, 417 KB)_{g,ĝ} - Figure S7 -
Frequency histograms of the deviations in the simulated true predictive accuracy
*r*from the estimated predictive accuracy for the datasets with_{g,ĝ}*r̂*(empty box and whiskers capped with brackets) and without_{g,ĝ,o}*r̂*outliers, regarded as the benchmark (green boxes), for each of the seven methods in Scenarios 4 to 6. (PDF, 200 KB)_{g,ĝ} - Figure S8 -
Frequency histograms of the deviations in the simulated true predictive accuracy
*r*from the estimated predictive accuracy for the datasets with_{g,ĝ}*r̂*(empty box and whiskers capped with brackets) and without_{g,ĝ,o}*r̂*outliers regarded as the benchmark (green boxes) for each of the seven methods in Scenarios 7 to 10. (PDF, 209 KB)_{g,ĝ}