Table 3.
Statistical analysis of competitive success of vir9 isolates in the field competition experiments
| Source | denm. | DF | MD | F | Pr > F |
| Year | 1+3–5 | 1,1 | 1.13 | 0.09 | 0.8072 |
| Cultivar | 2+3–5 | 1,1 | 10.71 | 0.88 | 0.5203 |
| Pair | 1+2–5 | 3,4 | 47.80 | 15.57 | 0.0114 |
| Year*Pair (1) | 4 | 3,3 | 1.84 | 3.23 | 0.1807 |
| Cultivar*Pair (2) | 4 | 3,3 | 1.45 | 2.54 | 0.2321 |
| Year*Cultivar (3) | 4 | 1,3 | 11.36 | 19.93 | 0.0209 |
| Year*Cultivar*Pair (4) | 5 | 3,16 | 0.57 | 0.57 | 0.6438 |
| Residual (5) | 16 | 1.00 | - | - |
Based on a logistic regression, Analysis of Deviance tested for effects of pair (Avir/vir), host cultivar and year on final frequencies of vir9 isolates. Analogous to Analysis of Variance, mean deviances (MD = 2.log-likelihood ratio/DF) were used to calculate quasi-F values. The denominator (denm.) column indicates the error terms used for the F-tests; where necessary, linear combinations of different model factors were made [69]. The DF column indicates the numerator and denominator degrees of freedom. All factors were considered as random effects. Model fitting was analogous to SAS-type II model fitting [68].