Table 5. Anova on classification performance of realistically oriented density gradient growth layouts.
| Factor | Accuracy | Youden index J | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Sum Sq. | d.f. | Mean Sq. | F | Prob>F | Sum Sq. | d.f. | Mean Sq. | F | Prob>F | |
| species | 2.00 | 1 | 2.00 | 1847.0 | 0 | 4.40 | 1 | 4.40 | 1283.7 | 0 |
| origins | 1.24 | 2 | 0.62 | 572.8 | 0 | 5.52 | 2 | 2.76 | 804.0 | 0 |
| growth mode | 0.28 | 2 | 0.14 | 128.7 | 0 | 8.03 | 2 | 4.02 | 1170.7 | 0 |
| Error | 1.94 | 1794 | 0.00 | 6.15 | 1794 | 0.00 | ||||
| Total | 5.46 | 1799 | 24.11 | 1799 | ||||||
A three-way analysis of variance was performed for both classification accuracy (see Fig 8, Table 4) and Youden index J (see Fig 9, Table 4), testing for effects of the factors ‘species’, ‘number of origins’, and ‘growth mode’. Sum Sq., Sum of squares; d.f., degrees of freedom; Mean Sq., mean squares = Sum.Sq. / d.f.