*3.4. Interrelations between Traits*

Figure 3 shows the means of the three traits of the different tested inocula. The data correlate well, but we see that isolates 2 and 4 behave differently from the others. Isolate 4 causes relatively more FHB and FDK, but it is poor in DON production. Isolate 2 behaves oppositely, here the visual data are much lower, but the DON data are higher. Their mixture shows well the transitional result between components. The correlations are very close, they are between r = 0.95 and 0.97 (calculated without the control). The rates for DI/DON and FDK/DON were also calculated (Figure 4). For 1% FDK we had DON content between 1.53 to 0.19 mg/kg. Isolate 3 and its mixtures gave values between 1.25 and 1.53, the others between 0.19–1.12. This clearly means that the forecast of DON contamination via FDK is not possible. We draw also similar conclusions for the DON/FHB rate; however, the different behavior of the DI/DON rate is clear, following a decrease to 1 + 3 the rate increases again. It shows also that the FDK is more precise to describe the DON relation than DI is.

**Figure 3.** Mixtures and aggressiveness, means for the three traits, Szeged, 2013–2015. LSD 5% between any data for DI 2.06%, FDK 3.82% and DON 9.35 mg/kg.

**Figure 4.** Influence of mixtures on FHB test results. DON production for one percent FHB and FDK infection, 2013–2015.

The cultivar reactions as means across inocula and years illustrate the well-known fact that visual head data are less informative for DON than the FDK data are (Figure 5). Based on the FHB values, no DON forecast is possible. GK Csillag is important as this cultivar performed well after a fungicide treatment than any other cultivars tested. Based on data from the milling industry, all staples were bought from this cultivar in 2010 when we had the national FHB epidemic. The correlation between FHB and DON is r = 0.66, which is not highly significant, whereas between FDK and DON the correlation is r = 0.9968, and is significant at *p* = 0.001. It means that the FDK signalizes the DON content much better than the visual scores.

**Figure 5.** General performance of the cultivars across experiments and inocula for FHB (DI), FDK and DON contamination. LSD 5%: DI: 1.06, FDK: 1.97, DON: 4.68.

Correlations were counted between DI, FDK and DON for every cultivars tested to see whether there are different variety to the 15 inocula responses or not (Figure 6). The first conclusion is that the FDK/DON correlations are much closer than the DI/DON correlations, this agrees well with earlier experience. The DI/FDK data did not show any significant relations in the four cultivars, indicating that the same cultivar might have different responses at different inocula and mixtures of inocula.

The correlations between DI, FDK and DON were counted also for all isolates separately for the four cultivars (Table 10). In five inocula, all three correlations are highly significant and nearly no difference exist between them. Two of them were single isolates and three mixtures were found (1 + 3, 1 + 4 and 1 + 2 + 4). In eight cases, the FDK/DON correlations were closer than the DI/DON correlations. Thus, it seems that the isolates and their mixtures may have different profiles that can influence resistance expression. The only advantage of the mixture is that the very diverse behavior of 2 Fc 52.10 and 4 Fg 13.38 could be balanced by the mixing to some extent by increasing the aggressiveness. It is important that the DI/DON correlation across all inocula is much lower (r = 0.6668) than the FDK/DON relation (r = 0.9968). The DI/FDK is also low (r = 0.6170). This means that for forecasting DON the FDK is more precise means than visual scores are. This is an argumen<sup>t</sup> that FDK and DON contaminations should be treated more seriously than it normally happens.

**Figure 6.** Correlations between genotype reactions to different FHB traits (data from Figures 2–4, for each isolate the performance of the cultivars was related to the mean of genotypes). Limit for *p* = 0.05 = 0.51, *p* = 0.02 = 0.59, *p* = 0.001 = 0.77.



\* *p* = 0.05, a Means for all inocula. b averages of correlations (from inoculum 1 to mixture 1+2+3+4) across all inocula.
