*3.2. Fusarium Damaged Kernels (FDK)*

The values of the FDK data (Table 5) are much higher than the DI data; the mean was for DI 26.7% and for FDK 44.9. In the controls, no visual infection was recorded, so all infections originated from the artificial inoculation (Table 5A). The reduction of the aggressiveness through mixing is significant, but in extent, less than that of the FHB values. Here, the difference between the measured FDK and the counted is larger, 36% mean increase could be registered. The combinations having Isolate 3 (Fg 19.42) have a mean higher than 50% and in one case, higher than 60%. The mixing produced data compared to the arithmetical mean of the aggressiveness of the components between 91 and 184%. Actually, every mixture variant has more or less differing aggressiveness levels. The different compositions mean characteristic aggressiveness differences, which also influence the expression of resistance. The correlations between genotype means across years (Table 5B, *n* = 15) highly significant correlation above r = 0.90, indicating the similarity of the response of cultivars to different isolates. However, when we compare the aggressiveness of the mixture and the mean of participating isolates, we receive large deviations. From 91% to 184%, every possibility can occur. In Table 5C (*n* = 4) where the isolate reactions were compared for genotypes (*n* = 4), the variability in the correlations grow significantly, indicating different responses of the genotypes to the individual inocula.


**Table 5.** Response of the wheat cultivars to the different isolates and their mixtures. FDK data (%) across experiments, Szeged, 2013–2015.

However, the cultivar responses often differ in the different isolates and mixture. The two more resistant genotypes are the same we found for FHB, GK <sup>F</sup>ény and GK Csillag. In eight cases, the response of the two cultivars does not significantly differ from each other. In five cases, GK Csillag has lower value, and only in one case GK <sup>F</sup>ény. GK Garaboly and GK Futár do not show significant difference in six cases; in all other cases, GK Futár has lower values. The larger differences are more stable, GK Csillag has better resistance in each inocula compared to GK Garaboly, but compared to GK Futár, in seven cases, no significant difference was found. From Table 5A, the correlation between the genotype reactions were also computed (Table 5C). Of the 105 possible correlations, 29 were significant. The non-significant correlations varied strongly. This was 29, more than double than was found in DI (13, Table 3C). Of the 29, twenty-five were found between inocula containing Isolate 3 in one or more partner inocula for the correlation test. This shows that FDK provides a closer correlation matrix. This would mean that a mixture automatically does not solve the problem and does not secure a stable level. The problem is as it was for the DI—that in a regular case we have only one test result, and not 15 as in this test. It is sure that a mixture does not provide the increased security of testing we hope from it. Except for the several low aggressive versions we know better to avoid, even the same aggressiveness does not always guarantee the same variety response. Comparing 2 + 3 + 4 and 1 + 2 + 3 + 4 at 59.98 and 58.75 mean aggressiveness in this trait, in the first case, Futár is significantly more susceptible, and in the other case, they have the same number. This means that we have to look for another solution.

The response of the experiment means (Table 6) strongly di ffers; 2013 gave the highest FDK values, 2014 was five times less and 2015a and 2015b showed similar results to 2013. The response of the di fferent mixtures was very variable. At high infection pressure, there were rather small di fferences between FDK values at di fferent isolates and mixtures, except isolate 2 and the combination of isolates 2 + 4 that gave significantly less FDK than the others. There is another feature that needs attention. In 2013, isolate 3 gave 84% FDK, in 2014 39.88%. However, at the same performance in 2013 the 2 + 3 gave only 6%, 3 + 4 gave 9.94% and 1 + 3 + 4 only 6.38%. Another example is 1 + 2 + 4 where 72.40% and 0.84% are the two corresponding data. However, in a less epidemic year, independent of the causing agents, the forecasting of the numbers for a heavy epidemic situation is hardly possible.


**Table 6.** Response of wheat to the different isolates and their mixtures. FDK data (%) of the experiments across wheat cultivars, Szeged, 2013–2015.

It seems that the aggressiveness of the mixtures cannot be predicted based on the individual aggressiveness of the four basic isolates (Table 6B, *n* = 15). It depends also on the ecology, but the individual isolates behave di fferently in most of the mixtures. The correlations data support this, although the 2014 data do so only moderately. Significantly, the data of the higher epidemic situations correlate much better, above r = 0.90, when 2013 and the two 2015 experiments are compared. This agrees well with what we found for FHB data. The correlations between experiment reactions for every inocula were presented (Table 6C, *n* = 4). From the 105 possible correlations, 57 were significant compared to DI in Table 4 with 52 significant correlations. We have non-significant correlations where aggressiveness was very low, such as isolate 2 (Fc 52.10), where no other inocula gave significant correlation with this set.
