Table 1.
Environmental means, Root Mean Squared Error (RMSE) and the Pearson correlation between predicted and observed days to heading (DTH) for 112 rice genotypes tested in 51 environments in Japan.
| CV00 | CV0 | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Mean | RMSE | Corr | Mean | RMSE | Corr | |||||
| Real | CB | GS | CB | GS | CB | GS | C | C | C | |
| Tsukubamirai 2004 Late | 79.4 | 76.7 | 85.9 | 7.0 | 12.6 | 0.86 | 0.70 | 76.3 | 4.6 | 0.99 |
| Tsukubamirai 2004 Early | 87.5 | 93.2 | 84.1 | 10.0 | 12.9 | 0.84 | 0.55 | 92.2 | 6.4 | 0.97 |
| Tsukubamirai 2005 Early | 98.5 | 102.1 | 87.1 | 6.9 | 14.0 | 0.87 | 0.75 | 101.0 | 3.5 | 0.98 |
| Tsukubamirai 2005 Late | 76.7 | 74.8 | 84.8 | 5.7 | 12.4 | 0.86 | 0.68 | 73.7 | 3.4 | 0.99 |
| Kasai 2006 | 87.4 | 83.8 | 88.3 | 7.9 | 8.8 | 0.85 | 0.75 | 82.8 | 6.8 | 0.93 |
| Fukuyama 2006 | 80.6 | 83.0 | 82.5 | 5.6 | 13.4 | 0.87 | 0.29 | 82.0 | 2.0 | 0.99 |
| Fukuyama 2007 | 84.1 | 83.4 | 86.6 | 5.0 | 6.1 | 0.87 | 0.84 | 82.5 | 2.9 | 0.98 |
| Tsukuba 2008 | 104.5 | 107.7 | 87.2 | 9.1 | 19.2 | 0.87 | 0.85 | 107.3 | 4.0 | 0.98 |
| Kasai1st 2008 | 102.2 | 99.9 | 88.7 | 9.0 | 17.8 | 0.89 | 0.71 | 99.2 | 4.3 | 0.99 |
| Fukuyama 2008 | 77.8 | 82.2 | 85.6 | 7.8 | 11.1 | 0.86 | 0.77 | 81.3 | 4.5 | 0.98 |
| Tsukuba 2009 | 103.9 | 107.7 | 87.1 | 8.6 | 19.0 | 0.89 | 0.81 | 108.0 | 5.2 | 0.98 |
| Kasai1st 2009 | 101.9 | 100.6 | 89.4 | 7.6 | 15.1 | 0.89 | 0.83 | 100.0 | 2.7 | 0.99 |
| Fukuyama 2009 | 78.7 | 78.7 | 89.6 | 5.2 | 13.2 | 0.86 | 0.83 | 77.8 | 1.5 | 0.99 |
| Tsukuba 2010 Late | 83.5 | 89.1 | 88.9 | 8.4 | 11.1 | 0.86 | 0.66 | 88.4 | 5.9 | 0.97 |
| Kasai1st 2010 | 100.6 | 100.7 | 89.5 | 7.1 | 14.3 | 0.88 | 0.77 | 100.1 | 1.9 | 0.99 |
| Fukuyama 2010 | 76.2 | 79.5 | 88.2 | 5.9 | 15.3 | 0.86 | 0.50 | 78.1 | 2.6 | 0.99 |
| Fukuyama 2010 Late | 59.0 | 65.4 | 87.7 | 9.2 | 29.7 | 0.84 | 0.82 | 62.5 | 4.2 | 0.96 |
| Tsukuba 2011 Early | 105.9 | 109.9 | 85.5 | 9.9 | 23.9 | 0.87 | 0.66 | 109.9 | 5.3 | 0.98 |
| Tsukuba 2011 Late | 82.8 | 80.8 | 90.0 | 5.9 | 11.9 | 0.86 | 0.78 | 80.4 | 4.1 | 0.95 |
| Tsukuba 2011 Middle | 94.7 | 95.7 | 90.9 | 6.5 | 11.1 | 0.88 | 0.67 | 95.4 | 3.0 | 0.98 |
| Kasai 2011 | 101.4 | 100.6 | 89.3 | 7.8 | 15.7 | 0.89 | 0.76 | 100.0 | 2.4 | 0.99 |
| Fukuyama 2011 | 75.7 | 79.4 | 91.1 | 6.4 | 16.9 | 0.87 | 0.83 | 78.6 | 3.1 | 0.99 |
| Fukuyama 2011 Late | 66.4 | 72.7 | 97.9 | 8.9 | 32.4 | 0.68 | 0.58 | 66.2 | 2.9 | 0.95 |
| Tsukuba 2012 Early | 97.1 | 96.0 | 86.9 | 5.7 | 12.4 | 0.81 | 0.74 | 101.1 | 5.8 | 0.97 |
| Tsukuba 2012 Late | 84.3 | 81.6 | 88.2 | 6.1 | 7.5 | 0.86 | 0.80 | 81.0 | 4.3 | 0.96 |
| Tsukuba 2012 Middle | 92.6 | 95.4 | 88.4 | 6.3 | 8.3 | 0.87 | 0.82 | 94.8 | 3.7 | 0.97 |
| Kasai 2012 | 103.8 | 102.7 | 86.4 | 7.9 | 21.3 | 0.89 | 0.60 | 102.1 | 2.7 | 0.99 |
| Fukuyama 2012 | 75.2 | 78.8 | 89.9 | 6.2 | 16.3 | 0.87 | 0.83 | 78.0 | 3.1 | 0.99 |
| Fukuyama 2012 Late | 71.5 | 77.0 | 103.1 | 7.5 | 32.4 | 0.70 | 0.66 | 68.6 | 3.4 | 0.98 |
| Fukuoka 2012 | 79.6 | 76.0 | 89.1 | 6.7 | 13.0 | 0.84 | 0.69 | 75.4 | 5.2 | 0.96 |
| Akita 2012 | 113.1 | 115.9 | 90.1 | 8.3 | 26.1 | 0.86 | 0.54 | 114.5 | 2.8 | 0.98 |
| Kasai 2013 | 104.8 | 104.1 | 87.3 | 8.0 | 20.1 | 0.90 | 0.80 | 103.5 | 3.2 | 0.98 |
| Fukuyama 2013 | 74.9 | 80.1 | 89.4 | 7.4 | 16.3 | 0.87 | 0.82 | 79.3 | 4.6 | 0.99 |
| Fukuyama 2013 Late | 62.1 | 64.8 | 85.7 | 6.3 | 27.1 | 0.83 | 0.36 | 64.1 | 3.6 | 0.93 |
| Fukuoka 2013 | 77.4 | 77.3 | 93.3 | 5.7 | 17.5 | 0.86 | 0.77 | 76.7 | 1.9 | 0.99 |
| Akita 2013 | 116.6 | 115.8 | 91.1 | 8.5 | 28.1 | 0.88 | 0.66 | 115.1 | 2.4 | 0.99 |
| Kasai 2014 | 102.4 | 105.1 | 88.9 | 8.4 | 18.0 | 0.90 | 0.66 | 104.4 | 2.8 | 0.99 |
| Fukuyama 2014 | 79.6 | 82.1 | 87.9 | 5.9 | 12.1 | 0.88 | 0.68 | 81.3 | 2.1 | 0.99 |
| Fukuyama 2014 Late | 67.6 | 64.5 | 88.1 | 6.7 | 21.4 | 0.83 | 0.79 | 62.9 | 5.1 | 0.98 |
| Fukuoka 2014 | 81.8 | 77.7 | 80.4 | 6.7 | 11.5 | 0.87 | 0.45 | 77.1 | 5.4 | 0.97 |
| Akita 2014 | 115.2 | 116.2 | 89.6 | 9.2 | 28.5 | 0.88 | 0.68 | 115.6 | 2.2 | 0.99 |
| Fukuyama 2015 | 76.3 | 72.6 | 88.1 | 6.5 | 14.4 | 0.86 | 0.63 | 71.9 | 4.6 | 0.99 |
| Akita 2015 | 118.1 | 119.8 | 89.4 | 7.7 | 29.9 | 0.81 | 0.76 | 117.6 | 2.5 | 0.99 |
| Kasai 2016 | 102.3 | 101.2 | 85.6 | 7.2 | 20.4 | 0.90 | 0.61 | 100.7 | 3.1 | 0.98 |
| Fukuyama 2016 | 79.1 | 81.1 | 85.2 | 6.0 | 8.9 | 0.86 | 0.82 | 80.8 | 2.4 | 0.99 |
| Fukuyama 2016 Late | 67.5 | 65.2 | 89.6 | 6.8 | 26.1 | 0.79 | 0.63 | 64.6 | 4.5 | 0.95 |
| Tsukubamirai 2016 Late | 73.6 | 77.9 | 84.6 | 7.0 | 13.6 | 0.90 | 0.80 | 77.2 | 4.9 | 0.98 |
| Kasai 2017 | 108.6 | 106.2 | 85.7 | 8.3 | 26.3 | 0.89 | 0.56 | 105.5 | 4.3 | 0.98 |
| Fukuyama 2017 | 77.4 | 80.2 | 87.7 | 6.3 | 16.3 | 0.88 | 0.42 | 79.6 | 2.7 | 0.99 |
| Fukuyama 2017 Late | 68.4 | 67.7 | 88.8 | 6.0 | 22.9 | 0.82 | 0.69 | 67.0 | 3.0 | 0.97 |
| Tsukubamirai 2017 Late | 69.5 | 73.1 | 86.9 | 6.8 | 19.8 | 0.87 | 0.65 | 72.2 | 4.3 | 0.98 |
| Mean | 7.2 | 17.5 | 0.86 | 0.69 | 3.7 | 0.98 | ||||
Two cross-validation schemes were implemented for mimicking realistic prediction scenarios: CV0 corresponds to the scenario of predicting tested genotypes in unobserved environments; CV00 considers the prediction of untested genotypes in unobserved environments. For CV0, the C method was used combining phenotypic and day length information (DL) of genotypes tested in other environments (one at a time). For CV00, two methods were considered: (i) the conventional GS implementation; and (ii) the CB method, which combines phenotypic and DL information from tested genotypes (training set) and genomic BLUP values from untested genotypes (testing set). In both cases, the prediction procedure was conducted by leaving one genotype out across environments and by deleting all phenotypic information from the target environment.