**3. Results**

## *3.1. Visual Data, Disease Index*

Table 3 presents the data across experiments. The four isolates had very distinct aggressiveness across experiments (Table 3A). The general tendency is that the most aggressive isolate, Fg 19.42, had the highest value alone, all combinations produced less aggressiveness. On the other hand, the least aggressive isolate showed higher aggressiveness in mixture in all cases than when it was used alone. The two least aggressive isolates gave results closer to the more weakly aggressive Fc 52.10. All eleven mixtures showed lower aggressiveness than the most aggressive component. The difference between inoculum means is highly significant, the maximum is 50.7, the minimum is 5.08, the distance

is 45.7, and the limit of significance LSD 5% for the 15 inocula is 2.06 (64 replicated behind each mean). As the mixing was made on *v*/*v* rate, it was anticipated that without specific mixing effects of the components, the arithmetical mean of the participating components will show the postulated performance. This was not the case. The measured aggressiveness across genotypes was seven cases higher than the mean of the components, and in four cases, the mean was lower than the arithmetical mean. The real data were between 74% and 173%. This means that the resulting aggressiveness is very variable and its aggressiveness cannot be forecasted. At present, we do not know the reason; further research should solve the problem. We can state, however, that the mixing will reduce the aggressiveness of the most aggressive component in each mixed inoculum, but balances much from the lower aggressive components in positive direction. On average, the mean shows a 28% increase in aggressiveness compared to the hypothetic arithmetical mean model. At the same time, the correlation coefficients, all are significant at *p* = 0.001 or higher at r-values between 0.976 and 0.985, indicating a similar response (Table 3B, *n* = 4)). This was not unexpected as we found it many times working with different isolates independently.

As the LSD 5% value 4.12% is valid for any difference among the data in Table 3A, the problem of significance can be identified without problem. When this is smaller than 4.12, no significant difference can be shown, when larger, it is proved. In many cases the behavior of genotypes (lines) or inocula (columns) is not so, and strongly varies. At low aggressiveness, no significant difference in resistance occurs for Isolates Fc52.10 and Isolate 4 Fg13.38 or the mixture 2 + 4. In other cases, such as Fc12375 3 + 4, the differences between genotypes are significant, with nearly 50% difference between them.

The variety reactions were compared for every inoculum (*n* = 15) so that the cultivar data were expressed at each cultivar to the mean of the four cultivars.

When the correlations are counted between the responses of the cultivars to different isolates, from the 105 correlations, only 13 were significant at *p* = 5% (Table 3C). Seven of the 11 mixed inocula contained Isolate 3. For the others, we have (altogether) four cases. The very variable correlations clearly show that the ranking of the genotypes at the different inocula (isolates and their mixtures) present a high diversity. In three inocula, the difference between genotypes is not significant (LSD 5% is smaller than 4.12%. Six of the genotypes have three cases without significant difference. Five cases were with no difference between two genotypes. Only one inoculum presented significant difference between all genotypes. This was the case also for the means of the four cultivars. For us the real problem is here—which inoculum is optimum to present differences in variety resistance? From the disease index, it seems that the mixing did not give to better differentiation of the genotypes. In this respect, the mixing is not the approach that would bring us closer to a more powerful methodology.

The four experiments (Table 4) had the same means for isolates and their combinations, but the means of the experiment differences are much larger, i.e., 44%, 8%, 26% and 28.5% (2013, 2014, 2015a, 2015b). The data proved that the differentiation between genotypes at low infection pressure is rather poor and not reliable compared to the data of the other years (Table 4B). All correlations (*n* = 4) where 2014 is a partner, gave correlations of r = 0.50, r = 0.66 and r = 0.69. For the rest, the correlations are between r = 0.85 and r = 0.99. Looking at the genotype correlations for the four years (*n* = 15) (Table 4C) from the 105 correlations, 52 were significant. Isolates 3 and 4 showed the least significant correlations with other inocula. It seems that aggressiveness level has a much higher importance in experimentation then mixing has. This is partly ecology-dependent, but is a result, also, of interaction between the aggressiveness level and differentiation of genotypes.


*Microorganisms* **2020**, *8*, 1036


**Table 4.** Response during the experiments to the different isolates and their mixtures. Disease index data (%) across cultivars, Szeged, 2013–2015.
