Introduction

Globally, wheat (Triticum spp.) production exceeds that of all other cereal crops. Germany is the eighth largest producer of wheat in the world, averaging an annual production of 19,203 TMT (USDA 2007). Plant breeders have so far been unable to locate complete resistance to pests in any of the wheat cultivars, some of which otherwise have shown varying degree of immunity against many insects (Berzonsky et al. 2002; McKenzie et al. 2002). Wheat midges and thrips are the major insect pests of wheat ears.

The wheat midge, Sitodiplosis mosellana (Géhin) (Diptera: Cecidomyiidae), is an important pest of winter wheat, T. aestivum L., in Germany (Volkmar et al. 2008; Gaafar et al. 2009) and UK (Oakley et al. 1998) and Canada (Lamb et al. 2003). It coexists with the lemon wheat blossom midge, Contarinia tritici (Kirby) another gall midge that attacks the wheat head, (Kurppa 1989; Harris et al. 2003). Management strategies that include host plant resistance would help to minimize the economic and environmental impact of S. mosellana (Olfert et al. 2009).

Resistance to the wheat midge is partially dominant due to expression of the Sm1 resistance as single gene (McKenzie et al. 2002), that mediates an induced hypersensitive response in the surface of developing seeds where wheat midge larvae begin feeding, resulting in larval death (Ding et al. 2000; Lamb et al. 2000a). In the wheat midge, the resistance act as an antibiotic reaction that includes elevation of phenol compound levels and results in the death of the young larvae shortly after they begin feeding (Ding et al. 2000; Harris et al. 2003). The adaptations that have been studied so far in the wheat midges have been due to alleles, usually completely or incompletely recessive, at single gene in the insects (Harris et al. 2003). Smith et al. (2007) stated that the density of wheat midge larvae developing on resistant wheat was lower compared to that of larvae developing on susceptible wheat cultivars and also mentioned that small numbers of larvae of S. mosellana matured in wheat cultivar were carrying the Sm1 gene. It was also reported that wheat midge infestation was associated with a reduced proportion of well-formed wheat seeds and yield losses (Lamb et al. 2000b; Doane and Olfert 2008).

Thrips fauna can also cause serious damage to winter wheat (Volkmar et al. 2009), and current methods of control are not sufficient to prevent crop damage. The widespread thrips species; Limothrips denticornis (Hal.), L. cerealium (Hal.), Haplothrips tritici (Kurd.), H. aculeatus (Fab.), Frankliniella tenuicornis (Uzel) and Thrips angusticeps (Uzel) were recorded on different wheat cultivars (Andjus 1996; Moritz 2006). Thrips feeding on the ovaries of immature wheat heads results in kernel distortion and abortion. This has considerable consequences on yield as well as on the baking quality of flour (Kucharzyk 1998).

This observation led us to assess the potential for wheat midge to adapt to resistant wheat carrying Sm1 by evaluating and comparing population of ear insects (wheat thrips and midges) on different cultivars. In order to explore the apparent differences in varietal characteristics leading to differential susceptibility to midge’s infestation, a field experiment was conducted at Silstedt using twelve cultivars with different characteristics in terms of susceptibility and resistance. The objective was to categorize wheat varieties (susceptible and resistant) based on wheat thrips and midge populations.

Materials and Methods

Winter wheat field

Twelve winter wheat cultivars were sown in sandy loam soil in RAGT 2n, Silstedt in central Germany (N 51.85°; E 10.85°) in October of 2008 and 2009. Five of these twelve cultivars (Glasgow, Welford, Robigus and Brompton) have proven resistance to S. mosellana (Ellis et al. 2009) as well as Skalmeje (Schliephake 2009, Personal communication). The experimental area was divided into plots. Two plots (replicates) were designated for each cultivar in a Completely Randomized Design.

Ear insect’s evaluation

Numbers of thrips species L. denticornis, L. cerealium, and T. angusticeps, and wheat midges larvae (S. mosellana and C. tritici) were assessed by randomly collecting 10 ears per plot per cultivar, during June from two growth stages (GS 65 and 73) (Tottman 1987) in 2008 and 2009. These samples were frozen at −20°C. The thrips and midges were counted in the laboratory after finishing field work. The ears were dissected under a low power microscope. Each wheat ear was divided into 3 parts (low, middle and upper). The numbers of spikelets and of thrips and midge’s larvae present on each as well as number of infested grains per ear were separately recorded for each part.

Kernel damage (shriveled, cracked, deformed kernels) in ears was also recorded. The relationship between numbers of thrips and midge larvae per ear was correlated with infested kernels in both years and among cultivars using Correlation Coefficient Statistics 9 (Thomas and Maurice 2008).

Statistical analyses

We assumed there was not a normal distribution of the pest count data within wheat ears. Figure 1 shows the distribution of the observed data for larval counts. In the present investigation we used a negative binomial distribution. The decision for this distribution compared with the frequently used Poisson distribution for counting data is based on a model selection with the use of the analytic criteria, AICC (Hurvich and Tsai 1989).

Fig. 1
figure 1

Wheat midge larvae (observed data 2008, 1st and 2nd growth stages)

Full size image

Accordingly we used a generalized linear model. Thus, the observed number of the i-th cultivar at date j in the k-th ear part of the l-th ear \({{\rm{y}}_{\rm{ijkl}}}\) is assumed to be the realization of a random variable \( {{\underline {\rm {y}}}_{\rm {ijkl}}} \) for those, the following is valid:

$$ \begin{array}{l} {\rm{P}}({\underline {\rm{y}}_ {\rm{ijkl}}} = {{\rm{y}}_{\rm{ijkl}}}) = {{\rm{f}}_{\rm{Negbin}}}({{\rm{y}}_{\rm{ijkl}}}|{{\mu}_{\rm{ijkl}}},\alpha ) \\({\rm i} = 1,...{\rm{a}};\;\!{\rm{j}} = 1,...,{\rm{b}};\;\!{\rm{k}} = 1,...{\rm{c}};\;\!{\rm{l}} = 1,...{\rm{n}}) \\\end{array} $$

We have multiple measurements within the same wheat ear; i.e. counts within different ear parts (low, middle and upper). Thus, the random effect of the ear is included in the model and we have to deal with generalized linear mixed models (GLMM). Between \({{\mu}_{\rm{ijkl}}}\) and\({{\eta}_{\rm{ijkl}}}\) we use the link function \(\log\!({{\mu}_{\rm{ijkl}}}) = {{\eta}_{\rm{ijkl}}},\) where \({{\eta}_{\rm{ijkl}}}\) is the so named “linear predictor”.

Furthermore, in the present case we assume for the linear predictor \({{\eta}_{\rm{ijkl}}}\) the following model:

$$ \begin{aligned} {\underline \eta_{\rm{ijkl}}} &= \mu+ {{\alpha}_{\rm{i}}} + {{\beta}_{\rm{j}}} + {{\gamma}_{\rm{k}}} + {(\alpha \beta)_{\rm{ij}}} + {(\alpha \gamma)_{\rm{ik}}}\\ &\quad+ {(\beta \gamma)_{\rm{jk}}} + {(\alpha \beta \gamma)_{\rm{ijk}}} + {{\underline {\rm{z}}}_{\rm{ijl}}} \\ &= \mu+ {{\delta}_{\rm{ijk}}} + {{\underline {\rm{z}}}_{\rm{ijl}}} \end{aligned} $$

The model contains all main effects of cultivar, date and ear part as well as their two-way and three-way interactions. All fixed effects are summarized in the effect \({{\delta}_{\rm{ijk}}}\), which describes the effect of the i-th cultivar and time j and ear part k. In addition, the random ear effect \( {{\underline {\rm{z}}}_{\rm{ijl}}} \) is included. For this effect we assume normal distribution \( {({\underline {\rm{z}}}_{\rm{ijl}}} \sim {\rm{N}(0,\sigma _{\rm{z}(\rm{j})}^2))} \). In the course of the model selection we found a better model fit in case of time dependent ear variance estimation \( {{\hat \sigma }}_{{\rm{z(j)}}}^{{2}} \). Therefore we used heterogeneous variances.

The parameters of the linear predictor \( {\underline \eta_{\rm{ijkl}}} \) and the variance components \( \sigma _{\rm{z(j)}}^{{2}} \) as well as their standard errors were estimated with the use of a conditional model and the maximum likelihood method with an adaptive Gauss-Hermite quadrature. The results presented in the next paragraph are on the original scale. Using the estimates of the linear predictor and the variance components we calculate

$$ {\hat \mu _{\rm{ijk}}} = \exp\!\left(\!\hat \mu + {\hat \delta}_{\rm{ijk}} + \frac{{\hat \sigma _{\rm{z}(\rm{j})}^2}}{2}\!\right) = \exp\!\left(\!{\hat \eta}_{\rm{ijk}} + \frac{{{\hat \sigma}_{\rm{z}(\rm{j})}^2}}{2}\!\right) $$

Derived from \({\hat \mu _{\rm{ijk}}}\) we are able to make any accumulation of effects. For the estimate of the i-th cultivar it follows \( {{\hat \mu}_{\rm{i}}} = \frac{1}{{\rm{b} \cdot \rm{c}}}\sum\nolimits_{{\rm{j}},{\rm{k}}} {{{\hat \mu }_{{\rm{ijk}}}}} . \) The standard error of the estimates, for example \({{\hat \mu}_{\rm{i}}}\), \( {{\hat \mu}_{{\rm{i}}^\prime}} \) as well as their differences, was calculated by delta-method (Greene 2003, p. 913 ff.). Statistical analysis of the differences are based on the t-value = difference/standard error (difference) and thus on the t-distribution. The calculations described above were done using the SAS-Software version 9.2 (Proc GLIMMIX, Proc IML) (SAS Institute 2009).

Results

Three parts of wheat ear were examined for the presence of thrips and midges within two growth stages during two years. Detailed results of the differences between the parts are not reported here. Therefore, the following results are presented as mean value of ear part.

Ear insect evaluation

1. 1. 2008

A. Thrips [larvae and adults]

Significant differences were found (P = 0.011) in the number of total thrips per ear part among cultivars. Esket and Türkis cultivars had the highest numbers of thrips 5.7 and 6.2/ear part, respectively, while moderate values (2.5 & 2.6/ear part) were recorded in cultivars of Welford and Carenius. The lowest number of total thrips was found in Glasgow and Robigus cultivars (2.1 & 2.2/ear part), respectively (Fig. 2a).

Fig. 2
figure 2

Population of thrips larvae & adults (a) and wheat midge larvae (b) in different winter wheat (resistant & susceptible) cultivars in Silstedt 2008. Different letters indicate significant differences

Full size image

There was a significant difference between Türkis and all cultivars except Esket (P = 0.91). There were significant differences between Esket and either Boomer (P = 0.038), Potenzial (P = 0.007), Tommi (P = 0.005), Global (P = 0.007), Carenius (P = 0.0008), Welford (P = 0.001), Skalmeje (P = 0.001), Brompton (P = 0.0003), Robigus (P = 0.001) or Glasgow (P = 0.001). Also, significant differences were found between Boomer and either Robigus (P = 0.040) or Glasgow (P = 0.044). Moreover significant differences were also recorded among Potenzial, Tommi, Global, Carenius, Welford & Skalmeje and Robigus or Glasgow as well (Fig. 2a).

B. Wheat midge’s larvae

Significant differences (P = 0.001) were also found between resistant and susceptible cultivars (Fig. 2b). There were significant differences (P = 0.014) among susceptible cultivars, whereas the highest midge’s larval population (10.2 larvae/ear part) has been found in Türkis as compared to other cultivars, while the moderate cultivars were reported in Carenius and Esket with value of 3.65/ear part. The least numbers of larvae were recorded in resistant cultivars namely Brompton, Skalmeje, Robigus, Welford and Glasgow (0.0, 0.1, 0.3, 0.6 and 1.2/ear part), respectively, there was no significant difference among resistant cultivars (Fig. 2b). There were four groups that have significantly different numbers of larvae. Most susceptible was Türkis. Tommi, Potenzial, Global and Boomer were less susceptible as Türkis, but with significantly higher larvae than Esket and Carenius, as well resistant cultivars, Glasgow, Welford, Robigus, Skalmeje and Brompton.

There was a significant difference between Türkis and all other cultivars without exception. Significant differences were recorded among Tommi and all cultivars except Boomer (P = 0.051), Global (P = 0.570), and Potenzial (P = 0.777). There were significant differences among Potenzial and other cultivars except Boomer (P = 0.115) and Global (P = 0.810). Significant differences were obtained among Global and other cultivars except Boomer (P = 0.153). Significant differences were recorded among Boomer and all cultivars except Global, Potenzial and Tommi. There were significant differences among Esket and all cultivars except Carenius (P = 0.871). There were significant differences among Glasgow, Welford, Robigus, Skalmeje & Brompton and susceptible cultivars; while there was no significant difference within these resistant cultivars (Fig. 2b).

1. 2. 2009

A. Thrips [larvae and adults]

Significant differences were found (P = 0.029) in the number of total thrips per ear part among cultivars. Global cultivar showed the highest numbers of thrips (5.6/ear part), while moderate numbers (2.4 and 2.7/ear part) were recorded in cultivars of Esket and Boomer. The lowest number of total thrips was found in Carenius and Brompton cultivars (1.3 and 1.6/ear part), respectively (Fig. 3a).

Fig. 3
figure 3

Population of thrips larvae & adults (a) and wheat midge larvae (b) in different winter wheat (resistant & susceptible) cultivars in Silstedt in 2009. Different letters indicate significant differences

Full size image

There were significant differences between Global and all cultivars as well as among Skalmeje and either Esket (P = 0.001), Tommi (P = 0.008), Potenzial (P = 0.001), Robigus (P = 0.001), Brompton (P = 0.003) or Carenius (P = 0.001) (Fig. 3a). Also, significant differences have been found among Glasgow, Esket (P = 0.49), Tommi (P = 0.048), Potenzial (P = 0.022), Robigus (P = 0.016), Brompton (P = 0.006) and Carenius (P = 0.001). Significant differences were also obtained among Türkis, Welford & Boomer and the following cultivars (Brompton and Carenius) and also between Esket and Carenius (P = 0.003) (Fig. 3a).

B. Wheat midge’s larvae

In 2009, the resistant cultivars (with exception of Skalmeje) were significantly separated from the other cultivars. Significant differences (P = 0.009) were found between resistant and susceptible cultivars (Fig. 3b). There were significant differences (P = 0.034) among susceptible cultivars, whereas Potenzial showed the highest larval population (0.7 larvae/ear part) compared to other cultivars, while the moderate cultivars were Boomer and Global with values of 0.2 and 0.3/ear part, respectively. The least WBM values were recorded in resistant cultivars (Glasgow, Robigus, Brompton, Welford, and Skalmeje) ranged from 0.01 to 0.2/ear part (Fig. 3b).

Significant differences were found between: (i) Potenzial and other cultivars [except Carenius (P = 0.758) and Tommi (P = 0.769)]; (ii) Tommi or Carenius and all other cultivars [except Türkis and Esket]; (iii) Türkis and rest of the cultivars [except Carenius, Tommi and Esket], Esket and all other cultivars [except Global, Boomer and Skalmeje]; (iv) Global and Boomer and other cultivars except [Esket and Skalmeje] and Skalmeje, Welford, Brompton, Robigus and Glasgow and susceptible cultivars were found significantly different (Fig. 3b).

1. 3. Pooled data of two years 2008 & 2009

A. Thrips [larvae and adults]

Generally, infestation of thrips was higher in 2008 compared to those in 2009. This may be due to the environmental conditions between the two years. There were significant differences (P = 0.001) between the two years in the following cultivars (Carenius, Esket, Potenzial, Global and Türkis), while there was no significant difference (P = 0.06) in cultivars of Boomer, Tommi, Brompton, Glasgow, Robigus, Skalmeje and Welford between both seasons. Significant differences were found (P = 0.022) in the number of total thrips per ear part among cultivars. Türkis, Global and Esket cultivars had the highest numbers of thrips 4.5, 4.3 and 4.1/ear part, respectively, while moderate ones (2.6 and 2.8/ear part) were recorded in cultivars of Welford and Glasgow. The lowest number of total thrips was found in Carenius and Robigus cultivars (1.9 and 2.0/ear part), respectively (Fig. 4a).

Fig. 4
figure 4

Population of thrips larvae and adults (a) and wheat midge larvae (b) in different winter wheat (resistant & susceptible) cultivars in Silstedt in 2008 and 2009. Different colours indicate different infestation grades. Different letters indicate significant differences

Full size image

There were significant differences between Türkis and all cultivars [except Global (P = 0.999)]. Significant difference was obtained among Global and Glasgow (P = 0.008), Welford (P = 0.005), Tommi (P = 0.001), Potenzial (P = 0.005), Carenius (P = 0.001), Robigus (P = 0.001) and Brompton (P = 0.001). Also, there were significant differences between Esket and each of Glasgow, Welford, Tommi, Potenzial, Carenius, Robigus and Brompton. Significant differences were recorded between Skalmeje & Boomer against Carenius (P = 0.001), Robigus (P = 0.012) and Brompton (P = 0.002) and also among Glasgow, Welford, Tommi & Potenzial against Carenius, Robigus and Brompton (Fig. 4a).

B. Wheat midge’s larvae

The proportion of ears infested with wheat midges also differed significantly (P = 0.0025) between the two years. In general, wheat midge’s infestation was lower in 2009 compared to those in 2008. There were significant differences (P = 0.001) in both years among all cultivars except in the cultivars of Brompton, Skalmeje and Welford. Significant differences (P = 0.006) were found among resistant and susceptible cultivars (Fig. 4b). There were significant differences (P = 0.019) among susceptible cultivars, whereas Türkis had the highest midge’s larvae population (5.3 larvae/ear part) compared to other cultivars, while the moderate cultivars were Esket and Carenius with value of 2.1/ear part. The least WBM values were recorded in resistant cultivars namely Brompton, Skalmeje, Robigus, Welford and Glasgow (0.1, 0.2, 0.2, 0.3 and 0.6/ear part), respectively, there was no significant difference among resistant cultivars (Fig. 4b).

There was a significant difference among Türkis and all cultivars. Significant differences were recorded among Tommi, Potenzial & Global and all cultivars except Boomer (P = 0.051). Significant differences were evaluated between Boomer and all cultivars [except Carenius and Esket], also between Carenius and all cultivars [except Esket (P = 0.862)]. Significant differences were found among resistant cultivars (Glasgow, Welford Robigus, Skalmeje and Brompton) and susceptible cultivars; while differences between resistant cultivars were not significant (Fig. 4b).

The number of midge larvae per ear was significantly positively correlated (r = +0.99) with the percentage of infested ears. There was no significant correlation between total thrips and infested kernels (r = +0.085) (Table 1).

Table 1 Correlation coefficient between ear insects (thrips & wheat midges) and infested kernels in 2008 and 2009, significant differences are at 0.05 level
Full size table

Ear insect’s Infestation Grades (Wheat Cultivars Groups)

The results indicated that winter wheat cultivars could be grouped into three categories based on thrips and wheat midges as shown in Fig. 4 in different colors among these groups in Table 2. The results indicated that the highest number of ear insects (thrips and midges) was found in cultivars of Türkis and Global, while the lowest values were recorded in Brompton and Robigus cultivars.

Table 2 Infestation grades of wheat cultivars based on population of thrips and wheat midge larvae in 2008 and 2009. Different letters indicate significant differences
Full size table

Discussion

Independent of the different levels of thrips and midge’s larvae attack in both experimental years, Türkis, Tommi and Potenzial proved the most susceptible cultivars to WBM, while Brompton, Skalmeje, Robigus, Welford and Glasgow showed a clear resistance reaction in the two years. These results are similar with Volkmar et al. (2008, 2009); Gaafar et al. (2009); Gaafar et al. (2010) who studied some wheat cultivars in Germany for their susceptibility to thrips and WBM infestations.

Wheat midge populations were significantly positively correlated with the number of infested ears among cultivars in both years. These findings coincide with that concluded by Olfert et al. (1985) and Smith and Lamb (2001) who mentioned that such a strong correlation was expected because midges prefer to oviposit the eggs in wheat ears in the flowering stage, and when hatched, quickly move into the ear and damage kernels in the milky stage.

Variety trials showed consistently low wheat midge larval infestations on the resistant cultivars such as Brompton, Skalmeje, Robigus, Welford and Glasgow. In contrast, infestation levels on other cultivars varied between seasons such Boomer, Potenzial, Global and Tommi. Although cultivar such as Potenzial has wm1 marker, but it is not resistant to orange wheat midges and its infestation vary from year to year. On the other hand, Skalmeje cultivar has also this wm1 marker, and it is a resistant cultivar, therefore, midge’s infestation was very low as confirmed by Schliephake (2009, personal communication). The low densities of wheat midge on resistant wheat in this study are consistent with those found in previous studies conducted with different objectives (Lamb et al. 2000a; Smith et al. 2004). Similar results were obtained by (Smith et al. 2007), pointing that wheat midge developing on resistant wheat was always very low compared with that of larvae developing on susceptible wheat cultivars and also they mentioned that small numbers of S. mosellana matured larvae in each wheat cultivar carrying the Sm1 gene for antibiosis resistance against this insect. Synchronicity between the susceptible ear emergence stage of the crop and the peak of WBM flight activity was another key factor in determining larval infestation levels. The resistant cultivars had the lowest levels of WBM larval infestation as expected. This result confirms the importance of monitoring pest numbers in order to make the decision for insecticide treatment.

This may be due to that the ancestors in these cultivars evolved a defense which reduces or prevent eggs laying on spikes, based on some explanations: The most resistant wheat cultivars had a higher constitutive level and a more rapid induction of ferulic acid than susceptible cultivars, which increased the mortality of newly hatched larvae. Analysis of phenolic acids in grain samples showed that levels of ferulic acid were higher in infested grains of some cultivars compared to uninfested grains. Levels of p-coumaric acid were greater in the infested than in the uninfested samples of all the tested cultivars indicating that WBM damage is inducing production of this acid in the seed as reported by Ellis et al. (2009). This suggests that there might be another mechanism of WBM resistance (Smith et al. 2007).

Understanding the biochemical basis of resistance: although it is clear that the Sm1 gene is responsible for resistance, as in Canada (Smith et al. 2007), the mechanism of resistance is still not understood. Canadian research suggested a correlation between increased levels of ferulic acid and resistance, but work with UK varieties does not support this. Further investigation is required to help future breeding programs (Ellis et al. 2009). Another demonstration by Ding et al. (2000) who mentioned that few wheat cultivars have a high level of antibiotic resistance to the larvae, which suppresses their growth and development; nearly all larvae develop successfully on susceptible wheat cultivars (Thomas et al. 2005).

There were less thrips or midges in the infested kernels of some cultivars than others in two years. Some wheat cultivars also have evolved a defense mechanism that deters oviposition by the wheat midge as mentioned by Berzonsky et al. (2002). These discrepancies might have been a result of speed ripening time as reported by Elliott et al. (2000). The wheat midge has evolved preferences for ovipositing at particular developmental stages of its host. This may have been sufficient to make some cultivars less favourable for oviposition (Lamb et al. 2001). Such cultivars are recommended to be cultivated in the next year.

In conclusion, to minimize the economic and ecological impact of S. mosellana, C. tritici and thrips, wheat producers in Germany should monitor ear insects, assess weather conditions and consider using resistant cultivars. Resistant wheat carrying the Sm1 gene for antibiosis to wheat midge is highly effective in preventing the development of larvae. However, a small proportion of larvae are capable of maturation on resistant wheat and surviving to adult emergence, although at a lower rate than larvae did on susceptible wheat. Another possibility is that occasional larvae survive because of variability in the expression of the Sm1 gene in wheat plants resulting from environmental influences as reported by Lamb et al. (2001). If a lower degree of infestation is predicted, producers may stick to their plans to grow wheat, but may choose a less susceptible wheat cultivar and early planting to avoid high midge populations during heading.