The spatial pattern of Fusarium-infected kernels and their mycotoxin contamination was studied in four wheat fields in Germany using geo-referenced sampling grids (12–15 × 20–30 m, 28–30 samples per field) at harvest. For each sample, frequency of Fusarium-infected kernels and spectrum of species were assessed microbiologically; mycotoxin contents were determined by HPLC-MS/MS analysis. Spatial variability of pathogens and mycotoxins was analysed using various parameters including Spatial Analysis by Distance IndicEs (sadie®). Microdochium majus, the most frequent head blight pathogen in 1998, was less frequent in 1999 and could not be detected in kernels from two fields in 2004. Fusarium avenaceum, F. graminearum and F. poae were the most frequent Fusarium species, with 7–8 species per field. The frequency of Fusarium-infected kernels was 3–15% and the incidence of species showed considerable within-field variability. Spatial patterns varied among Fusarium species as well as from field to field. Although pathogens and mycotoxin were often distributed randomly in the field, F. avenaceum, F. graminearum, F. poae, F. sporotrichioides, F. tricinctum and the mycotoxin moniliformin had an aggregated pattern in at least one field. Patterns are discussed in relation to spread of Fusarium species depending on inoculum sources, spore type, kind of dispersal, availability of susceptible host tissue and micro-climate. Sampling of wheat fields for representative assessment of mycotoxins is complicated by random patterns of Fusarium-infected kernels, especially where the frequency of infection is small.
Decision-making in crop protection requires accurate and reliable assessment of the disease; in addition to the identification of the causal agent and the level of disease, the spatial pattern of disease in the field has recently gained more attention as it is crucial for representative sampling (Schmale et al., 2005; Montarry et al., 2009). Spatial pattern analyses have been used to describe patterns of primary inoculum, mechanisms of pathogen dispersal and effects of environmental factors (Xiao et al., 1997).
The situation may be very complex for diseases like fusarium head blight (FHB), which can be caused by several pathogens (Xu et al., 2005). Fungal species of the genera Fusarium and Microdochium infect and colonize developing kernels of cereals causing FHB symptoms. The disease is often associated with the production of mycotoxins and Fusarium species are reported to differ greatly in their ability to produce these toxic secondary metabolites (Desjardins, 2006). The occurrence of mycotoxins in wheat is a significant threat to its use for food and feed.
Incidence and severity of FHB and the composition of Fusarium species involved are reported to vary among geographical regions (Doohan et al., 1998; Xu et al., 2005) and years (large-scale variability) due to variation in climatic and weather conditions and cropping practices (Schaafsma & Hooker, 2007; Klix et al., 2008; Xu et al., 2008). Additionally, FHB is often characterized by the occurrence of ears with typical symptoms in close proximity to apparently non-affected ears (small scale variability). Within-field variability of FHB-causing fungi and associated mycotoxins however, is largely unknown.
Many diseases are distributed in the field in patches of different sizes which result from the survival of primary inoculum in some hot spots or the heterogeneity of environmental conditions – temperature and relative humidity (RH) – favouring the invasion of plant material by pathogens (Waggoner & Aylor, 2000). Patchiness has been detected particularly in the early stages in disease spread, whilst later stages are often characterized by rather homogenous disease incidence. Patchiness can manifest itself over a range of hierarchical scales, from the individual host plant to regional or global scales (Ristaino & Gumpertz, 2000). Spatial pattern has been described as the arrangement of disease entities relative to each other and to the architecture of the host crop (Gilligan, 1982). It is an ecological property of a species and reflects environmental and genetic heterogeneity and population growth influencing reproduction, dispersal and mortality (Madden, 1989). At growth stage (GS) 31/33 and GS 59 most leaf diseases of wheat exhibited a near random pattern on scales between 31 cm and 31 m; exceptions were early powdery mildew and yellow rust at GS 59/61 (Parker et al., 1997).
Characterization of the spatial disease pattern may result in information on the epidemiology of the disease, on the scale to study disease spread, as well as on the optimization of the sampling scheme (Madden & Hughes, 1999; Gosme, 2008). It may also contribute to the understanding of the underlying ecological processes resulting in the pattern, which may also influence control strategies (Campbell & Madden, 1990).
FHB is considered to be a monocyclic disease resulting from infections of wheat ears during anthesis, with mid anthesis being the most susceptible developmental stage (De Wolf et al., 2003). Other stages are less susceptible (Bai & Shaner, 1994; Meier, 2003) and as phenological stages may vary within fields, FHB incidence and severity as well as mycotoxin contamination of kernels are also very likely to vary on the field level. The incidence of FHB symptoms in New York State, predominantly caused by F. graminearum, has been reported to be random in all wheat fields sampled except one, indicating that the initial inocula were distributed randomly; clustering of diseased ears was ascribed to inoculum from corn debris in the soil (Shah & Bergstrom, 2001). Xu et al. (2008) conducted a 2-year study with 14 sites in four European countries and demonstrated considerable variability of Fusarium mycotoxins within single fields. They sampled ears from 16 squares randomly selected from a W-shaped walk across the sampling area; however, samples/data were not geo-referenced.
Disease incidence (proportion of diseased units) and severity (proportion of plant surface infected) are the most common measures of disease (Ridout et al., 2007). For severe epidemics of F. graminearum the percentage of blighted ears may be assessed easily in the field and may be a reliable measure of FHB incidence. In Europe, with a complex of Fusarium species differing in symptom expression often involved in FHB, this parameter is less reliable. Instead the frequency of infected kernels per ear/sample area may be used for the quantification of FHB severity, as infection by some Fusarium species may occur without symptoms.
Sampling also plays a crucial role in the determination of mycotoxin levels due to the sometimes very heterogeneous distribution of toxins in agricultural commodities (Krska et al., 2008). The European Commission and other authorities have therefore developed regulatory sampling criteria for commodities (EC, 2006; FDA, 2009). However, for non-harvested wheat crops there are no guidelines because of the lack of knowledge on the spatial pattern of Fusarium-infected kernels in the field. This information would be highly welcome for forecasting of the average mycotoxin contamination to be expected from a field and the potential separation of seed lots with no or low mycotoxin contamination from those from highly contaminated sites within a field at harvest.
This paper reports work in four field experiments in Germany to study the spatial pattern of Fusarium-infected kernels and their mycotoxin contamination in wheat fields using geo-referenced sampling grids close to harvest time.
Materials and methods
Four winter wheat fields were investigated 1–5 days before harvest for the spatial distribution of Fusarium-infected kernels; two at Kerpen-Buir, one in 1998 and one in 1999, and two at Klein-Altendorf and Lichtenhagen in 2004. At Kerpen-Buir (N 50°51′/E 06°35′; 110 m above sea level (asl); in both years previous crop sugar beet; wheat cv. Ritmo) the sampling areas were in the centre of 20 and 22 ha fields, respectively; at Klein-Altendorf (N 50°37′/E 06°59′; 80 m asl; previous crop sugar beet; wheat cv. Drifter) the sampling area was 45 m from the northern margin of a 9 ha field, and at Lichtenhagen (N 51°57′/E 09°23′, 310 m asl; previous crop oilseed rape; wheat cv. Ritmo) 12 m from the western margin of a 1·8 ha field adjoining a street. Wheat was grown in commercial fields according to local practice including control of leaf pathogens by fungicides; crops were not treated for FHB control during anthesis.
In 1998 and 1999, wheat ears were sampled at 28 (4 × 7) sampling sites using a sampling grid of 12 × 30 m representing a total area of 1 ha. Distances between sampling sites along the tracks (working width 12 m) were measured with a measuring tape and sampling sites were geo-referenced later on the computer. At each sampling site all wheat ears in an area of 0·5 × 0·5 m (0·25 m²) were harvested and threshed manually. In 2004, 30 (5 × 6) samples per field were taken. The size of the geo-referenced (Garmin GPSII) sampling grid was adjusted to the working width of machinery and was 15 × 20 m at Klein-Altendorf and 12 × 20 m at Lichtenhagen, representing total areas of 0·9 and 0·75 ha, respectively. The area harvested per sample was reduced to a circle of radius 0·19 m (0·11 m²) in order to emphasize spot-type sampling of wheat kernels.
Quantification and identification of Fusarium species
Two hundred kernels in each sample were surface sterilized with NaOCl (1·3% for 3 min), rinsed twice with sterile water, dried on filter paper, transferred onto CZID agar (Abildgren et al., 1987) and incubated for 7 days at 21 ± 2°C and 16 h NUV light to record the percentage of kernels infected by Fusarium spp. Fungal cultures were transferred to potato dextrose agar to obtain pure cultures for the morphological identification of species according to Nelson et al. (1983).
Analysis of mycotoxins
The spectrum and amount of mycotoxins in grain samples were analyzed for the samples from both 2004 sampling sites, Klein-Altendorf (n =28) and Lichtenhagen (n =30). A random sub-sample of 20 g kernels was ground to a fine powder using an ultra-centrifugal mill (Retsch ZM1). Mycotoxin analysis was performed as described in detail by Sulyok et al., 2007. In brief, 0·25 g of sample were extracted for 90 min with 1 mL of a mixture of acetonitrile:water:acetic acid (79:20:1, v:v:v) on a rotary shaker. After centrifugation, the raw extract was diluted 1:1 using acetonitrile:water:acetic acid (20:79:1, v:v:v) and directly injected into the liquid cromatograph mass spectrometer. Sample extracts exhibiting large enniatin concentrations were re-analyzed after being diluted 50-fold. Chromatographic separation was performed using an 1100 Series HPLC System (Agilent) equipped with a Gemini® C18-column, 150 × 4·6 mm i.d., 5 μm particle size, and a C18 4 × 3 mm i.d. security guard cartridge (Phenomenex) in gradient elution mode. Detection and quantification of the mycotoxins was performed with a QTrap 4000 LC-MS/MS System (Applied Biosystems) equipped with a Turbo V Ion source for electrospray ionization).
Quantification was performed using external calibration. Limits of detection were estimated from calibration standards at low concentration levels and were based on the signal-to-noise ratio (S:N = 3:1). Matrix effects were investigated by comparing the slopes of the linear, 1/x weighted calibration functions deriving from the analysis of standards prepared in solvent and in extracts of a blank control sample, respectively. Recoveries of the extraction step were determined by analyzing the same control sample after spiking it in triplicate on one concentration level prior to extraction. Mycotoxin concentrations in the contaminated samples were corrected accordingly; enniatin A (signal suppression by 25%), beauvericin (signal suppression by 11%), deoxynivalenol-3-glucoside (recovery of the extraction step: 82%) and moniliformin (recovery of the extraction step: 70%). For all the other toxins, no correction was applied.
Indices of species diversity – species richness S, Shannon’s Index H’, Hill’s N1, Hill’s N2, Simpson’s Index, and species evenness – Pielou’s J′ and modified Hill’s ratio, were calculated according to Ludwig & Reynolds (1988).
Spatial distribution of Fusarium infections of wheat kernels was analyzed with the Spatial Analysis by Distance IndicEs (sadie®) software (Perry et al., 1996; Madden et al., 2007) using the geo-referenced absolute numbers of Fusarium isolates from 200 kernels per sample. The analysis was done for each species separately. The degree of aggregation was calculated by comparing the observed pattern of count data with rearrangements of randomly distributed counts; the maximum of 5967 randomizations of data was compared to the actual distribution. The index of aggregation (Ia), i.e. the ratio of observed and expected distances to regularity, is <1, = 1, or >1 for regular, random, and aggregated patterns, respectively. sadie® calculates the significance of deviations from randomness (Pa) using the percentages of distances from the randomized rearrangements, and aggregated patterns and regular patterns were accepted with Pa< 0·1 and Pa> 0·9, respectively.
Potential spatial associations between fungal species, Fusarium mycotoxins, and mycotoxins and Fusarium species, i.e. the co-occurrence of fungal species and mycotoxins, respectively, in the same sample, were also calculated using sadie®. Two populations may be spatially dissociated (due to competition for the same limited resource) or positively associated (because of requirements for similar microclimates), or they may occur at random with respect to one another (Korie et al., 2000). Large values of local association are indicated by the coincidence of a patch cluster (= neighbourhoods of units with counts > sample mean m) for one set with a patch cluster for the other, or by the coincidence of two gaps (= several units nearby to one another with counts < m); dissociation is indicated by a patch coinciding with a gap (Perry, 1998). By employing randomizations in which the observed counts are permuted amongst the sample units, an index of clustering Ic is ascribed to each sample unit. Because it is based on cluster indices and not abundance, the method intrinsically allows for the spatial pattern in each component population, after Dutilleul adjustment (= detrending each data set) as required (Perry & Dixon, 2002).
As spatial structure in data should be described by more than one statistic, patchiness of pathogen distribution was also calculated using Lloyd’s index of patchiness (LIP; Lloyd, 1967; Madden et al., 2007) as well as the variance-to-mean ratio (VMR, s²/x−, also called index of dispersion; Yang, 1995; Madden et al., 2007). Statistical significance of deviations from randomness was tested with χ²-test, (N – 1) VMR was compared with the tabulated χ² distribution at N – 1 df (Campbell & Madden, 1990).
For the calculation of VMR and the sadie parameter Ia for the geo-referenced mycotoxin data, concentrations below the limit of detection (LOD) were set to 0.
Geo-referenced maps of Fusarium-infected kernels and mycotoxin concentrations were produced using ArcGIS software (V. 9.2, ESRI Geoinformatik GmbH).
Incidence of fusarium head blight pathogens in wheat kernels
The frequency of wheat kernels infected by FHB pathogens was low in 1999 and moderate in 1998 and 2004 (Table 1). In 1998, Microdochium majus was by far the most frequent species colonizing wheat kernels. In 1999, the frequency of this species was considerably lower and in 2004 it could not be detected at either site. The frequency of kernels infected by Fusarium species was low in 1998 and 1999 and reached 13–15% in 2004; incidence decreased in the order F. avenaceum, F. graminearum, F. poae, F. culmorum, F. cerealis, F. tricinctum, F. sporotrichioides and F. equiseti.
Table 1. Mean frequency of wheat kernels infected by Fusarium species and Microdochium majus at four field sites in Germany, 1998, 1999 and 2004. The total number of kernels investigated was 5600 in 1998 and 1999, and 6000 in 2004
Frequency of infected kernels [%]
Fusarium species, total
The number of fungal species isolated from wheat kernels varied from 7 to 9 in the fields, confirming FHB as a disease initiated by a complex of Fusarium species and Microdochium majus (Table 2). The mean number of species per sample site, S, was lowest in 1998 when M. majus dominated the species spectrum (see below) and 3 or higher in the other years. In 1999 and Lichtenhagen 2004, a minimum of two species occurred at each sampling site and a maximum of 6 and 5, respectively. Shannon’s Index, the most popular parameter for species diversity, varied considerably among sample sites within all fields (range 0–1·58); lower mean values in 1998 and Klein-Altendorf (H′ = 0·68 ± 0·08 and 0·81 ± 0·07, respectively, compared to 1·08 ± 0·06 and 1·00 ± 0·05 in 1999 and Lichtenhagen) reflected the dominance of M. majus and F. avenaceum and F. graminearum, respectively. Simpson’s Index, i.e. the probability that two individuals selected at random belong to the same species, reached 1 in a sample in 1998 when M. majus predominated, and in a sample at Klein-Altendorf, but was considerably lower in most cases. The number of abundant species and very abundant species – Hill’s N1 and N2– were highest in 1999 and at Lichtenhagen.
Table 2. Overview on the ecological diversity of fungal species causing fusarium head blight of wheat in four field experiments: Kerpen-Buir 1998, 1999, Klein-Altendorf 2004, Lichtenhagen 2004
(Smax = 9)
(Smax = 9)
(Smax = 7)
(Smax = 8)
Shannon’s index H ′
0·29 ± 0·33
0·38 ± 0·30
0·43 ± 0·29
0·46 ± 0·36
Pielou’s J ′
Modified Hill’s ratio
Spatial distribution of fusarium head blight pathogens in wheat kernels
In 1998, the frequency of pathogen species detected in wheat samples from Kerpen-Buir ranged from 4% for F. tricinctum to 64% and 100% for F. avenaceum and M. majus, respectively (Table 3a). Variance-to-mean ratios (VMR) often considerably exceeded 1 and indicated high within-field variability in the number of infected kernels for several FHB pathogens. Lloyd’s index of patchiness (LIP) and sadie parameters indicated an aggregated pattern of spatial distribution for F. graminearum (Ia= 1·414, P =0·10), but not for F. culmorum which had the highest number of infected kernels of all Fusarium species at sampling site A5 (sample 5, row A; Fig. 1a). Dominated by the almost regularly distributed species M. majus (Ia = 0·582, Pa = 0·986), spatial distribution of total FHB infections (Fusarium spp. and M. majus) was close to regular (Ia = 0·660) in this year.
Table 3. Parameters characterizing the incidence and spatial distribution of fungal species causing fusarium head blight of wheat at Kerpen-Buir, 1998 and 1999 (a), and Klein-Altendorf and Lichtenhagen, 2004 (b)
Fusarium spp., total
aFrequency of samples infected (n =28 in 1998 and 1999; n =30 in 2004).
In 1999, all sampling sites had Fusarium-infected kernels. However, the number of infected kernels varied from 1 to 17 out of 200 (Fig. 1b). Despite a very similar level of total Fusarium infection, VMR values were lower than in 1998 (Table 3a). LIP indicated patchiness in the distribution of F. cerealis and F. graminearum, whereas sadie statistics rated spatial distribution of the DON-producing species F. culmorum and F. graminearum in the middle between regular (Ia = 0·5) and random (Ia = 1·0). Distribution of F. sporotrichioides was highly aggregated (Ia = 1·794, Pa = 0·10). Spatial distribution of the total number of infected kernels per sampling site was random (Ia = 1·031).
In 2004, the overall percentage of Fusarium-infected kernels was 14·9% and 13·1% for Klein-Altendorf and Lichtenhagen, respectively (Table 1). In Klein-Altendorf, F. avenaceum and F. graminearum were detected at more than 80% of sampling sites, infecting 5% and 5·9% of all kernels sampled, respectively. Spatial distribution of F. avenaceum was random as indicated by Ia and LIP whereas VMR indicated large variability in the number of kernels infected among samples (Table 3b, Fig. 2). The parameter Ia indicated highest levels of aggregation, although non-significant at the 0·1 level, for F. graminearum and F. culmorum supported also by high VMRs. As the frequency of samples infected with F. graminearum was considerably higher, LIP was lower than for F. culmorum. Fusarium poae and F. tricinctum were present in about one quarter of samples; F. poae was distributed randomly in the field and showed lower variability in the percentage of kernels infected than F. cerealis (0–2·8% compared to 0–11·3%). The latter species had a high index of crowding but showed no aggregated pattern of spatial distribution.
At Lichtenhagen, F. avenaceum, F. poae and F. graminearum were detected in more than 80% of samples (Table 3b). Fusarium graminearum had by far the highest VMR of all species, and LIP indicated patchiness in the distribution of this species as well as F. culmorum, one of the species detected at low frequency (Fig. 3a). sadie parameter Ia indicated aggregation of F. avenaceum in the northern and southern part of the site and a gradient for F. poae with a decrease in the number of infected kernels from West to East (Fig. 3b). Although having an overall low infection rate (0·73% of kernels), two clusters of infection by F. tricinctum were separated by an area without infection as indicated by the high Ia (1·503, P =0·03) of this species.
The spatial association of Fusarium species colonizing wheat kernels exhibited large variability among sites and species. Table 4 summarizes the main results of the 121 tests. Species associations (P <0·05) proved to be significant for 3, 2, 1 and 3 pairs of Fusarium species in 1998, 1999, Klein-Altendorf and Lichtenhagen, respectively. Dissociation of species (P >0·95) was significant for 1, 6, 1 and 2 pairs of species, respectively. Frequencies of F. graminearum and F. equiseti were significantly associated (P =0·02) in 1999 and significantly dissociated (P =0·97) at Lichtenhagen. Unambiguous interrelationships could be identified only for less frequent species, which exhibited dissociation for the combinations FCER × FCUL, FEQU × FAVE, FEQU × FCER and FCER × FTRI, and positive association only for the combination FPOA × FSPO. Preference effects could not be established for all other species combinations.
Table 4. Statistical analysis on the association of fungal species causing fusarium head blight of wheat. Average, minimum and maximum values for coefficients of correlation and P according to Dutilleul (1993) for data from four locations in Germany in 1998, 1999 and 2004
cFAVE: Fusarium avenaceum, FCER: F. cerealis, FCUL: F. culmorum, FEQU: F. equiseti, FGR: F. graminearum, FPOA: F. poae, FSPO: F. sporotrichioides, FTRI: F. tricinctum, MMAJ: Microdochium majus.
dBold italic figures indicate P according to Dutilleul <0·05 (association) or >0·95 (dissociation).
−0·070 (−0·29 to 0·31)
0·090 (−0·29 to 0·30)
−0·338d (−0·56 to −0·15)
−0·090 (−0·15 to 0·04)
−0·057 (−0·31 to 0·29)
−0·194 (−0·42 to −0·03)
0·169 (−0·09 to 0·42)
0·62 (0·08 to 0·91)
−0·222 (−0·36 to −0·08)
−0·247 (−0·47 to −0·16)
0·095 (−0·12 to 0·46)
−0·118 (−0·26 to 0·06)
−0·044 (−0·24 to 0·40)
−0·265 (−0·59 to 0·02)
0·35 (0·09 to 0·92)
0·82 (0·62 to 0·98)
0·159 (−0·11 to 0·51)
−0·062 (−0·06 to −0·02)
0·056 (−0·16 to 0·21)
0·057 (−0·26 to 0·48)
0·084 (−0·34 to 0·33)
0·89 (0·72 to 0·99)
0·80 (0·68 to 0·98)
0·35 (0·01 to 0·65)
0·076 (−0·36 to 0·39)
0·158 (−0·11 to 0·30)
−0·015 (−0·76 to 0·41)
−0·057 (−0·30 to 0·19)
0·65 (0·43 to 0·77)
0·41 (0·02 to 0·72)
0·61 (0·54 to 0·76)
0·39 (0·02 to 0·97)
−0·051 (−0·28 to 0·15)
−0·052 (−0·33 to 0·28)
−0·028 (−0·22 to 0·21)
0·59 (0·13 to 0·90)
0·67 (0·38 to 0·88)
0·40 (0·14 to 0·79)
0·30 (0·07 to 0·72)
0·53 (0·24 to 0·89)
0·377 (0·04 to 0·75)
0·028 (−0·16 to 0·22)
0·56 (0·16 to 0·78)
0·59 (0·02 to 0·81)
0·48 (0·02 to 0·91)
0·40 (0·07 to 0·99)
0·58 (0·12 to 0·96)
0·19 (0·00 to 0·44)
0·076 (−0·24 to 0·37)
0·30 (0·01 to 0·65)
0·80 (0·48 to 0·98)
0·36 (0·06 to 0·95)
0·57 (0·17 to 0·94)
0·55 (0·13 to 0·90)
0·45 (0·14 to 0·77)
0·39 (0·02 to 0·86)
Spatial distribution of mycotoxins produced by Fusarium species
As the frequency of Fusarium-infected kernels was low in 1998 and 1999, no mycotoxin analysis was conducted for these years. In 2004, mycotoxin contamination of wheat kernels was considerably higher at Lichtenhagen than at Klein-Altendorf, although the highest DON content (2000 μg kg−1) was measured in a sample from Klein-Altendorf. Therefore, VMR for DON reached 1022 (Table 5). The random distribution of mycotoxin concentrations depicted in Figs 2c,d and 3c,d are also reflected by the Ia values. At both sites, all samples had enniatin levels above the level of detection (LOD). In contrast, DON could be detected in 50% and 90%, respectively. The frequency of sample sites with NIV was considerably lower. At low concentrations moniliformin showed an aggregated pattern at Klein-Altendorf, whereas spatial distribution at Lichtenhagen was more random. However, at this site, only NIV showed a crowded pattern of distribution at low concentrations. At both sites, all samples were additionally contaminated with tentoxin produced by Alternaria species with concentrations up to 13 ng g−1 (median at Klein-Altendorf and Lichtenhagen 2 and 5 ng g−1, respectively).
Table 5. Mycotoxin contamination of wheat kernels infected by fusarium head blight: frequency, intensity and spatial distribution, Klein-Altendorf and Lichtenhagen, 2004
Toxin content [ng g−1]
aFrequency of samples with mycotoxin concentration >LOD.
At both sites, the DON concentration of samples was positively correlated to the frequency of F. graminearum-infected kernels (P <0·001) as well as to the total number of Fusarium infected kernels (Table 6). For other Fusarium species-mycotoxin combinations, Pearson’s correlation was lower or non-existent, except for the highly significant correlation between the frequency of kernels infected by F. avenaceum and their contamination with moniliformin and enniatins at Lichtenhagen.
Table 6. Statistical analysis of correlations between the frequency of Fusarium-infected kernels, by species, and the amount of mycotoxins in wheat kernels from two locations, Klein-Altendorf and Lichtenhagen, in 2004
*, **, ***; significant at P <0·05, P <0·01, P <0·001, respectively.
Fusarium spp., total
Fusarium head blight in northwest Germany is caused by a complex of various Fusarium species, not only varying in species composition and disease severity from field to field, but also exhibiting considerable within-field variability. A relatively high proportion of M. majus-infected kernels contributed to moderate FHB incidence in 1998, whereas this non-mycotoxin producing species was not detected at all in 2004. The extensive use of strobilurin-containing fungicides since 1996 in Germany may have contributed to a considerable decline in the incidence of M. majus, which has been shown to be more sensitive to strobilurins than Fusarium species (Muellenborn et al., 2008). Despite the reduced maximum species diversity in 2004, species richness per site was not significantly affected. With generally low frequencies of infected kernels (4–15%) the level of infection had no effect on species richness in the fields investigated.
Indices commonly used in ecological studies were highly suitable to assess and quantify the diversity of FHB pathogens infecting wheat kernels in the field. Shannon’s index H′ was low when M. majus was the predominant species in 1998 and when high levels of infection by F. avenaceum and F. graminearum occurred (Klein-Altendorf). Consequently, the number of abundant (Hill’s N1) and highly abundant (Hill’s N2) species was lowest for these fields. With overall low infection levels the evenness of FHB-causing species at sampling sites was highly variable within fields and among fields.
Parameters such as VMR and LIP indicated considerable variability in the frequency of Fusarium-infected kernels; only five out of 35 VMR values were not significantly different from 1. However, only the sadie® parameter Ia was useful for describing whether the spatial variability resulted in patterns of aggregation or not. The VMR increases with the degree of aggregation, but variance values may vary greatly even when the degree of aggregation is constant (Yang, 1995). In contrast, similar VMR values can be associated with significantly different levels of aggregation (e.g. F. graminearum and M. majus in 1998) indicating that variability may have a regular or heterogeneous (= aggregated) pattern. sadie® takes into account the effect of distance and allows the differentiation of random distribution with large variability among sites (high levels of infection next to sites with no fungal infection) from patchiness where zones of high infection levels at various sites are separated by zones with low disease level. Similarly, high values of Lloyd’s index of patchiness were sometimes associated with no aggregation; however, low LIP values were consistent with a regular Fusarium pattern.
Rare Fusarium species, like F. equiseti and F. sporotrichioides, showed high spatial variability within and among fields. The extreme case, a single site in the field with infected kernels is categorized by sadie® as random; only with neighbouring samples also infected by the same pathogen does this sampling site becomes part of a cluster (= aggregation). Similarly, Fusarium species with higher incidence were not distributed regularly in the field; however, clear spatial aggregation zones of species were rare. Neither a consistent trend for a specific Fusarium species nor an obvious effect of the location could be detected. Only at Lichtenhagen, where a road was close to the western edge of the sampling grid, the frequency of F. poae-infected kernels showed a downwind gradient. This species exclusively produces microconidia, which may serve as air-borne inoculum and result in a disease gradient when inoculum is introduced from outside the field, or cause a random distribution of infected ears when the inoculum sources are within the field.
Fusarium avenaceum and F. tricinctum were also aggregated in the Lichtenhagen wheat crop. Fusarium avenaceum produces large macroconidia and is pathogenic to cereals as well as to oilseed rape, sugar beet and potato (Calman et al., 1986; Satyaprasad et al., 1997; Peters et al., 2008), which are often part of wheat crop rotations in Germany. Inhomogeneous distribution of debris from these host plants may result in a site-specific production of F. avenaceum inoculum in wheat crops. The spread of macroconidia from organic matter on the soil surface to the ears largely depends on splash dispersal via the wheat leaves (Jenkinson & Parry, 1994; Bateman, 2005). This type of dispersal reduces the probability of large-scale spread and, depending on the environment, may result in aggregation or randomness of FHB from F. avenaceum on the field scale. Successful infection requires the coincidence of infective inoculum, susceptible host tissue, and favourable environmental conditions in space and time. Moniliformin had an index of aggregation and a spatial pattern very similar to F. avenaceum. Higher heterogeneity in grain yield (Ia = 1·235) emphasized the heterogeneity of this wheat crop as compared to the rather homogenous crops in the other fields (Ia = 0·899). The crop itself may contribute considerably to spatial variability of FHB due to differences in plant density and ontogenetic development, and micro-climate resulting from within-field soil variability, spatial exposition, and the occurrence of weeds and leaf diseases.
Fusarium graminearum is the only Fusarium species for which ascospores have been reported to contribute significantly to epidemic spread (Suty & Mauler-Machnik, 1996). On a small scale, dispersal of ascospores of the teleomorph, Gibberella zeae, from a concentrated inoculum source may result in aggregation and gradients of infection (Schmale et al., 2005). Bentley et al. (2009) demonstrated clustering of wheat crown rot caused by G. coronicola (anamorph F. pseudograminearum) on a 1-m scale, resulting from plant debris as the primary inoculum source. In the present study, the distribution of infected kernels proved to be random in three fields and was aggregated on the field scale only in 1988. In wheat fields without maize in the crop rotation inoculum levels of F. graminearum are rather low. The random pattern of very high frequencies of infected kernels may be due to the fact that this species is a highly competitive and aggressive colonizer of wheat ears.
Dissociation of rather rare Fusarium species seems to be reasonably consistent with the low probability of having both species at the same site, especially for low overall infection frequencies. However, even in fields with high levels of kernel infection the preference of a few species is very likely as Fusarium species differ considerably in aggressiveness to the host and competitiveness among each other (Doohan et al., 2003). The positive association between F. poae and F. sporotrichioides may be due to the preferred production of microconidia in both species (Leslie & Summerell, 2006) and their relative low competitiveness and aggressiveness (Fernandez & Chen, 2005). The low overall level of spatial association/dissociation between Fusarium species underlines the problems in predicting the species composition of the FHB complex. Highly significant associations between species valid for some sites or years highlight differences in the ecological optima of the species. The variability of associations is likely due to the fact that various environmental factors – survival on the soil, sources of inoculum, temperature, rainfall, relative humidity, host susceptibility, etc. – and their spatial heterogeneity affect the incidence and epidemic development of pathogens.
Regularity of Fusarium-infected ears and mycotoxin contamination of kernels is likely to increase with overall FHB levels of wheat crops. However, the results demonstrate that subsamples even from fields with low to moderate disease levels may have high mycotoxin concentrations. With average DON levels considerably lower than the maximal acceptable concentrations in the EU (EC, 2006), the maximum concentrations per sampling site were 2·3 and 1·8 mg kg−1 at Klein-Altendorf and Lichtenhagen, respectively. Data on nivalenol and zearalenone also demonstrate that peak mycotoxin values are not correlated either to the field average or to the frequency of positive samples.
Other Fusarium mycotoxins which have recently attracted more attention were detected in high frequency – reaching 100% for enniatins – as well as in high amounts. Like fungi from some other genera, Fusarium species infecting wheat have been reported to produce beauvericin and enniatins; F. culmorum is able to produce enniatins (Jestoi, 2008). Low beauvericin concentrations in both wheat fields in 2004 are in agreement with reports from Norway and Finland (Uhlig et al., 2006). Enniatins were detected in all samples in 2004 and reached concentrations of 1 mg kg−1 at some sampling sites, stressing the risk of higher contaminations. Moniliformin has been reported to have immuno-suppressive activity (Jestoi, 2008); MON concentration at Klein-Altendorf was <100 μg kg−1 at all sites, but exceeded this level in 40% of samples at Lichtenhagen.
The predominant random spatial distribution of Fusarium species complicates representative sampling, and a high number of sampling points per field is necessary. More information with a higher number of sampling points/larger areas for the assessment of large-scale variability within fields, as well as studies on the ear-to-ear variability, would be required in order to be able to fully understand the spatial distribution of Fusarium species causing mycotoxin contamination of cereals.
The authors gratefully acknowledge technical support of field experiments at Kerpen-Buir by LWK Rhineland and technical assistance of Kerstin Lange, as well as financial support of A.M. by the German Research Foundation (DFG).