Quaternary Integrated Pest Management concept for powdery mildew in sugar beet. III. Economic damage threshold and loss prediction

Quaternäres Integriertes Pflanzenschutz-System für den Echten Mehltau der Zuckerrübe. III. Ökonomische Schadensschwelle und Verlustprognose

Despite the seemingly dependence on dry weather conditions, the chance of predicting the epidemic onset of powdery mildew precisely is quite poor (Wolf et al., 2006). Therefore, some other tools of Integrated Pest Management (IPM) are required to match the objectives of reducing fungicide treatments to an essential amount. These are acting thresholds as well as the damage threshold where the loss prediction is based on. The tools are linked with each other through our "Quaternary IPM-concept" (Wolf et al., 2004; Wolf and Verreet, 2003). The linking is compensating insufficiency of the single elements. As already mentioned, prediction of disease is not accurate enough in order to pinpoint fungicide sprays precisely. So far, only a so-called Negative-Prognosis is possible, which is determining the risk of first symptom occurrence. During the risk period observation of beet fields is advisable and as well to score the disease level in view of exceeding the acting threshold. The acting threshold is a quite early stage of the epidemic. It is determined by a plant infection of 50% or respectively a leaf infection prevalence of 5% (Wolf, 2002; Wolf et al., 2007). Mostly, it is coincident with the first symptom occurrence. Although the acting threshold is setting the efficiency of fungicide treatments to an optimum, this epidemic stage does not really damage the beet. Thus, if the exceeding of the acting threshold is indicated, a loss prediction is necessary. The loss prediction is focussed on to predict the likelihood of disease will exceed the damage threshold at harvest time. Only in this case, fungicide applications are justified according to the demands of IPM.

The principles of the Quaternary IPM-concept were already presented for the Cercospora beticola (Wolf et al., 2004; Wolf and Verreet, 2003; Wolf and Verreet, 2005a; Wolf and Verreet, 2005b). This paper depicts the specific implications of powdery mildew in sugar beets concerning the damage threshold and loss prediction in order to complete the Quaternary conception.

2 Material and methods

Each trial included at least a fungicide untreated control – below predicated as "Diseased" - and, besides different epidemic oriented IPM-applications, a three times treated variant to determine the disease free respectively site specific yield optimum – in the following predicated as "Healthy". Thus, every mark in the graphs (see results) is related to a single field study reflecting the difference between Diseased and Healthy.

Further descriptions, about field experiments and the scoring of disease incidence were already published (Wolf et al., 2006) as well as yield measurement had been remarked (Wolf et al., 1998; Wolf, 2002).

2.2.1 Influence of incidence on yield and quality factors. The damage potential was evaluated through one-dimensional regression analysis where the AUDPC (Area Under Disease Progress Curve)-value was the independent variable and yield and quality factors were the dependent variables. Due to the specific epidemiological behaviour of powdery mildew, i.e. the effect of decreasing visible incidence towards the end of the season, the AUDPC value (Kranz and Holz, 1993) – this value (for this evaluation divided by 100) is related to the entire disease course - was preferred instead of disease severity (DS).

In the charts, the changes of non sucrose (Potassium, Sodium, αAminoN) contents and recoverable sugar are displayed as absolute differences of Diseased and Healthy. In terms of sugar content, beet yield and recoverable sugar yield, the respective differences are calculated as percentage in relation to Healthy.

2.2.2 Loss prediction. The AUDPC, loss of beet and recoverable sugar yield are estimated through a non-linear, one-dimensional regression model. The influence variable is here the time of first symptom appearance.

3 Results

Fig. 1. Relationship^a between powdery mildew incidence level (AUDPC^b) and the relative difference of sugar content to “Healthy”. ^a Δ-sugar content (%) = 0,21*AUDPC; r²=0,27.^b AUDPC=Area Under Disease Progress Curve = DS (%) x days / 100; DS= Disease severity (% infected leaf area).

Higher non-sucrose contents of Potassium, Sodium and αAminoN may reduce the recoverable sugar. Nevertheless, powdery mildew has no significant effect on the contents of Potassium and Sodium in sugar beets (Fig. 2-A,B). The deviations vary in the range of ±4 mmol/kg beet and are assumed as randomly effects which are confirmed by low coefficients of regression.

Fig. 2. Relationship between the level of powdery mildew in­cidence (AUDPC^d) and non-sucrose components. In each case the absolute difference to “Healthy” is shown. A: Δ-Potassium^a, B: Δ-Sodium^b, C: Δ-αAminoNitrogen^c. ^aΔ-Potassium = -0,067*AUDPC; r²=0,02. ^bΔ-Sodium = 0,043*AUDPC; r²=0,02. ^cΔ-αAminoN = 1,46 + 0,12*AUDPC; r²=0,06. ^d Please see text figure 1.

In contrast, powdery mildew seems to increase the content of αAminoN (Fig. 2-C). The difference of Diseased and Healthy, however, is largely independent on the severity of disease. Apparently, there are already significant effects even in case of low incidence. This may be also interpreted as a synergistic effect of used fungicides. With other words, the fungicides are reducing the αAminoN contents in the variant Healthy, even if there is no incidence of powdery mildew.

This trend continues by the analysis of recoverable sugar (Fig. 3); not surprisingly, because this quality parameter comprises the effects of the non-sucrose components and powdery mildew had no impact on the contents of Potassium and Sodium. Thus, decrease of recoverable sugar is mainly dependent on αAminoN. Because the latter was correlated positive linear with the incidence, the regression to recoverable sugar must convert to negative-linear. Anyway, the reductions of 1 - 2% at the maximum are rather slight.

Fig. 3. Relationship^a between powdery mildew incidence level (AUDPC^c) and the difference of recoverable sugar^b to “Healthy”. ^a Δ-recoverable sugar = -0,06*AUDPC; r²=0,23. ^b Calculated after the Braunschweiger Formel (Glattkowski and Thielecke, 1995) ^c Please see text figure 1.

The beet yield is also negative-linear dependent on the height of incidence (Fig. 4). If the AUDPC increases by the value of 1, losses of about 0,7% will follow. In case of early epidemic onset, AUDPC-values may rise to 15 at the maximum, which is equal to a loss of 10%. In view of a proper definition of a damage threshold, the conclusion is that a mostly randomly affected variability of beet losses is evident till AUDPC 2.

Fig. 4. Relationship^a between powdery mildew incidence level (AUDPC^b ) and relative difference of beet yield to “Healthy”. ^a Δ-beet yield = -0,67*AUDPC; r²=0,52. ^b Please see text figure 1.

In terms of economical affairs, the recoverable sugar yield may be assumed as the most important factor. This parameter comprises quality factors – non sucrose components and sugar content - as well as beet yield. The impact of AUDPC on recoverable sugar yield is not basically different to them of beet yield, although the extent of losses is increased. By analysing the negative-linear relationship, the value of AUDPC 1 causes a reduction of 0,9% of recoverable sugar yield (Fig. 5). Variation of the correlated values is also high and rather randomly affected within the range of AUDPC 0-2; if incidence is further increasing, losses must be expected with high probability. Under conditions of central Europe, Erysiphe betae may cause 15% losses of recoverable sugar yield at the maximum.

Fig. 5. Relationship^a between powdery mildew incidence (AUDPC^b) and relative difference of recoverable sugar yield to “Healthy”. ^a Δ-recoverable sugar yield = -0,87*AUDPC; r²=0,64. ^b Please see text figure 1.

When overall summarising yield and quality factors, as a tolerance limit of incidence respectively a damage threshold the value of AUDPC=2 is defined. According to the regression equation of recoverable sugar yield, losses would account to 1,8%.

The severity of epidemic, corresponding with the disease severity curve over the entire season, is expressed by the AUDPC-value. The latter is dependent on the time of disease initiation and cultivar susceptibility. The respective estimation of AUDPC may be performed by a negative-sigmoidal function (Fig. 6). If the epidemic onset takes place in July and cultivar susceptibility is high, severe epidemics are most likely, which exceed the damage threshold significantly by AUDPC-values in the range of 5 - 17. The extent of epidemic is declining strongly, if first symptom appearance is delayed in August. In case of high susceptibility, the sigmoidal curve fit indicates a drop of the curve beneath the tolerance limit at the end of August with further alignment to zero during September. If cultivar susceptibility is low, loss risk caused by Erysiphe betae is generally low (Fig. 6). There is just a marginal risk when first symptoms appear before mid-August.

Fig. 6. Relationship of first symptom appearance and AUDPC^c. Prediction of AUDPC is indicated through curves, full line representing highly susceptible^a respectively dotted line low susceptible cultivars^b. The dotted horizontal line at AUDPC=2 indicates the economic damage threshold. ^a Highly susceptible: AUDPC = 18/(1+exp (-(date-218)/-10)); r²=0,64. ^b Low susceptible: AUDPC = 10/(1+exp (-(date-220)/-8)); r²=0,55. ^c Please see text figure 1.

The relationship of disease initiation and losses of beet yield respectively recoverable sugar yield is expressed by an exponential function of the type "y=-y_max*exp(-a*(x-t₀)", where "y_max" indicates the maximum loss and "x" the calendar day of disease initiation; the term "x-t₀=0" determines the begin of the estimation respectively the time, where yield loss is the maximum. Deviations of correlated values from the curve fit are once again high and, therefore, the coefficients of determination rather low (Fig. 7). Overall, this may be a consequence of the slight damage potential of the biotrophic parasite. On the other hand, field experiments are naturally affected with randomly deviations which may explain part of the total variation. Maximum beet yield loss of 8% occurs if disease is initiated in the period of mid to end of July. The more the epidemic onset is delayed the smaller the differences between Diseased and Healthy. A conditional risk is evident until the end of August, particularly if cultivar susceptibility is high. Therefore, the incidence of powdery mildew can be neglected under conditions as follows: If cultivar susceptibility is low and first symptoms appear after mid-August respectively valid for high susceptible cultivars, if disease is initiated first during September.

Fig. 7. Relationship of first symptom appearance and Δ-beet yield^b. Loss prediction of beet yield is indicated through curves, full line representing highly susceptible^a respectively dotted line low susceptible cultivars^b. ^a Highly susceptible: Δ-beet yield (%) = -15*exp (-0,05*(date-192)); r²=0,30. ^b Low susceptible: Δ-beet yield (%) = -8*exp (-0,065*(date-194)); r²=0,08. ^b Difference of "Healthy" and "Diseased", expressed as percentage in relation to "Healthy".

The principles of the above reported relationship are the same if the recoverable sugar yield is the dependent variable (Fig. 8). Only the extent of losses is different. Under conditions of first symptoms appearance in the period of mid-end of July and high cultivar susceptibility, losses may achieve 10-15% at the maximum. The trend of decreasing losses continues during August; the average of sugar loss accounts to ≈2% at the middle of the month. In particular, if high susceptible cultivars are grown, economic losses cannot be excluded if first symptoms are found till the end of August. In terms of low susceptible reaction, however, the average losses tend to zero already from mid-August.

Fig. 8. Relationship of first symptom appearance and Δ-recoverable sugar yield^c. Loss prediction of sugar yield is indicated through curves, full line representing highly susceptible^a respectively dotted line low susceptible cultivars^b. The scheme at the bottom indicates risk periods for economic losses dependent on first symptom appearance and cultivar susceptibility. ^a Highly susceptible: Δ-Recoverable sugar yield (%) = -17*exp (-0,048*(date-194)); r²=0,25. ^b Low susceptible: Δ-Recoverable sugar yield (%) = -10*exp (-0,047*(date-194)); r²=0,12. ^c Difference of "Healthy" and "Diseased", expressed as percentage in relation to "Healthy".

These findings may be summarised through a simple loss prediction scheme by distinguishing periods of risk and no risk (Fig. 8, bottom). The assessment of loss risk is dependent on the time of disease initiation and cultivar susceptibility. Hence, there is a loss risk if first symptoms appear till the end of August respectively mid-August in case of low susceptible cultivar. The periods are equal to action zones concerning fungicide applications.

Even though the damage potential is moderate, we have to conclude that there is a substantial loss risk worth to face with fungicide applications. Hence, the question is, to what situations it is opportune to spray. The strategy of handling fungicide treatments is in accordance to the principles of the Quaternary IPM-concept as already presented for Cercospora beticola (Wolf et al., 2004; Wolf and Verreet, 2002; Wolf and Verreet, 2005b). Nevertheless, some modifications had to be introduced to match the specific biology and epidemic behaviour of Erysiphe betae. The negative-prognosis may be summarised briefly, as in Germany disease initiation doesn't take place before mid-July, in low susceptible cultivars even later, from the beginning of August onwards (Wolf et al., 2006). A further important statement was that only fungicide sprays during the phase of disease initiation are most effective and provide best monetary profits (Wolf, 2002).

Therefore, in order to back decisions of fungicide treatments, this paper depicted the other two elements of the quaternary principle, the economic damage threshold and loss prediction. In view of a proper definition of the damage threshold, the AUDPC-value was preferred instead of the final disease severity. The object here was to prevent false correlations of yield factors with the incidence of powdery mildew. The epidemiology of powdery mildew involves a period of more or less steep increase followed by culmination, and, as a typical feature, sometimes disappearance of the mycel layer towards the end of the season (Wolf, 2002; Wolf et al., 2006). Hence, the calculation of AUDPC comprises all epidemic phases and may be assumed as suitable for disease loss relationships. In this way, if quality and yield parameters are summarised overall, a damage threshold of AUDPC 2 is derived. Ahrens (1985) stated economic yield losses under conditions of disease initiation before mid-August and incidence progressions exceeding 30%. This finding also is fairly according with our results. Exceptionally, the risk period should be extended till the end of August if a high susceptible cultivar is grown.