Fusarium graminearum is the most important species in the complex causing Fusarium head blight (FHB) of small grain cereals in many areas of the world (Osborne and Stein 2007). The fungus produces both ascospores and macroconidia, mainly on residues of the previous crop (Sutton 1982), and both the spore types can cause FHB (Mitter et al. 2006; Salgado et al. 2008; Stack 1989). Ascospores are produced in perithecia; mature ascospores are forcibly discharged from perithecia and become air-borne. Conidia are produced in sporodochia and are mainly splash-borne (Parry et al. 1995; Sutton 1982). Ascospores were reported to travel short distances in some reports (de Luna et al. 2002; Fernando et al. 1997, 2000; Keller et al. 2010; Markell and Francl 2003; Paul et al. 2004; Paulitz 1996, 1999) and long distances in others (Del Ponte et al. 2003; Francl et al. 1999; Maldonado-Ramirez et al. 2005; Prussin et al. 2014; Schmale et al. 2006, 2012). Because macroconidia can be splashed only a limited distance (≤60 cm) above the soil surface (Hörberg 2002; Jenkinson and Parry 1994; Rossi et al. 2002), only a very small proportion of conidia produced on the crop debris on the soil surface can directly reach the ears by rain splashes. Isolation of several Fusarium species, among which F. graminearum, from apparently healthy leaves and from leaves with necrotic spots, suggests that the fungus can survive on leaves throughout the season (Ali and Francl 2001; Osborne and Stein 2004). Researchers have therefore speculated that leaves may provide an important bridge between conidia on crop residue and ears (Xu 2003).

Although infection by F. graminearum is likely to depend greatly on inoculum dispersal, the distribution patterns of ascospores and conidia within the wheat canopy in the presence of an inoculum source on the soil surface have seldom been studied. Paul et al. (2004) determined the distribution of F. graminearum propagules (as colony forming units, CFUs) at 0, 30, and 100 cm above maize residue or infested maize kernels on the soil surface during separate rain episodes. Similarly, Osborne et al. (2002) and Osborne and Stein (2004) quantified the temporal dynamics of ascospores and CFUs on ears, flag leaves, and leaves at the second and fourth position from the top.

This report describes a 2-year study (2013 and 2014) on the dispersal patterns of F. graminearum spores within canopies of wheat crops located at the experimental stations of ARVALIS—Institut du végétal (Boigneville, France, common wheat cv. Pakito) and Horta (Ravenna, Italy, durum wheat cv. Saragolla). Wheat crops were managed according to standard practices at both locations; soils were not tilled and fungicides were not applied to control FHB. At the Boigneville location, maize residues from the previous maize crop were naturally present on the soil surface. At the Ravenna location, maize residues infested with F. graminearum (inoculated as indicated by Manstretta 2015) were distributed on the soil surface at the end of February. Spore deposition within the wheat canopies was assessed by using passive spore traps attached to spore samplers.

A spore sampler consisted of a wooden stake (4 × 4 cm in cross section) that extended vertically 90 cm from the surface of the soil. Each spore sampler had seven spore traps. Three metal clothes-pegs were attached to each stake at 10, 30, and 60 cm above the soil surface, and each held one horizontal microscope slide that served as a surrogate leaf; the horizontal slides are referred to as “leaf-like traps”. Each stake also had four vertical slides that served as surrogate wheat ears (and are referred to as “ear-like traps”) at 90 cm above the soil surface, with four vertical slides per stake side (Fig. 1). A piece (50 × 12 mm) of double-sided transparent adhesive tape (Tesa, Hamburg, Germany) was attached to one side of each slide. There were six spore samplers (stakes) per location arranged in two rows, with 2 m between rows and 2 m between samplers in the same row. The adhesive tape on the horizontal slides was placed upward on three random samplers and downward (facing the ground) on the remaining three samplers. The adhesive side on the four vertical slides always faced away from the stake and had different orientations: north, south, east, or west. Slides were replaced once each week between booting (stage 41 of the scale of Meier 2001) and grain maturity (stage 99); there were 11 samplings at Ravenna and 10 at Boigneville that covered 77- and 70-day periods, respectively. Slides were examined with a light microscope (250× magnification) to assess the presence of F. graminearum ascospores and macroconidia. Two lengthwise stripes (0.213 mm wide) were scanned on each slide so that the total surface examined per slide was 21.3 mm2. Spores were identified as ascospores or macroconidia based on morphology (Headrick et al. 1988; Leslie and Summerell 2006); the two spore types were counted separately, and the results were expressed as numbers of ascospores or conidia per cm2.

Fig. 1
figure 1

Spore trap scheme. Leaf-like traps a, b, c are made of horizontal slides placed at different height above ground, ear-like traps d, e, f, g are made of vertical slides at the height of wheat heads. Each slide had a side coated with double-side adhesive tape as trapping surface

Full size image

An analysis of variance (ANOVA) was performed on spore numbers, separately for ascospores and conidia, using SPSS (ver. 21, IBM SPSS Statistics, IBM Corp. USA). To make variances homogeneous, spore numbers (n) were transformed by the following natural logarithm function: ln(n + 1). For leaf-like traps, factors tested with ANOVA were: height above ground (fixed factor, 3 levels: 10, 30, and 60 cm), trapping side (fixed factor, 2 levels: upward and downward), and location (random factor, 2 levels: France and Italy) For ear-like traps, factors tested with ANOVA were: orientation (fixed factor, 4 levels: north, south, east, and west) and location (random factor, 2 levels: France and Italy). Number of replicates was six (i.e., the six spore samplers).

Among all F. graminearum spores trapped, 93 % were ascospores and 7 % were conidia. This difference in the type of spores trapped was consistent between the two locations (P = 0.436 and P = 0.435 for the effect of location on numbers of ascospores and conidia, respectively). Prevalence of ascospores is not surprising because maize residues are a favourable substrate for ascospore production (Pereyra and Dill-Macky 2008) and ascospores have been found to be prevalent over conidia in other studies (Fernando et al. 2000; Inch et al. 2005; Markell and Francl 2003; Mitter et al. 2006; Panisson et al. 2002). However, F. graminearum survives on a wide range of plant material (Parry et al. 1995) and there is limited information on whether the substrate affects the ascospore/conidium production ratio.

A total of 46.5 ascospores/cm2 of leaf-like spore trap were sampled between booting and harvest at 10 cm above the ground (average of the two locations), which represented 41.4 % of the total ascospores sampled. Fewer ascospores were trapped at 30 and 60 cm than at 10 cm above the ground (P = 0.021 for the effect of height above ground on ascospore numbers) (Fig. 2a). Numbers of ascospores sampled did not significantly differ between the upper and the lower side of the leaf-like traps (P = 0.389 and P = 0.071 for the effect of trap side and for interaction “height above ground × trap side”, respectively) (data not shown). In the same period (i.e., between booting and harvest), a total of 18.3 ascospores/cm2 of spike-like spore trap were sampled at the ear layer (90 cm above the ground) (Fig. 2a). The number of ascospores trapped was not significantly affected by the orientation (north, south, east, or west) of the ear-like traps (P = 0.446). Peaks of ascospore trapping occurred at Ravenna from 7 to 14 May 2014, with 15.9 ± 6.5 ascospores/cm2 on the leaf-like traps (10 to 60 cm above ground) (Fig. 3a), and from 30 April to 7 May 2014, with 19.9 ± 4.4 ascospores/cm2 on the ear-like traps (Fig. 3b).

Fig. 2
figure 2

Numbers of Fusarium graminearum ascospores (a) and conidia (b) trapped at different heights above the soil surface within a wheat canopy. Passive spore traps were used and replaced once each week between wheat booting and grain maturity. Values are the means of 2 years; bars represent standard errors

Full size image
Fig. 3
figure 3

Numbers of Fusarium graminearum ascospores (white bars) and conidia (black bars) trapped at the leaf level (a 10, 30, and 60 cm above the soil surface) and ear level (b 90 cm above the soil surface) within a wheat canopy at Ravenna (North Italy) in 2014. Passive spore traps were used and replaced once each week between wheat booting and grain maturity; bars represent standard errors. Weather data measured during the trapping period are shown in (c)

Full size image

A total of 9.1 conidia/cm2 of leaf-like spore trap were sampled between booting and harvest at 10 cm above the ground (average of the two locations), which represented 77 % of the conidia sampled. Of the conidia sampled at 10 cm above the ground, 69 and 8 % were sampled on the upper and lower side of the leaf-like traps, respectively (P = 0.034 and = 0.031 for “height above ground” and “trap side”, respectively; P = 0.435 for the interaction “height above ground × trap side”). Of the conidia trapped, 10 and 8 % were from samplers at 30 and 60 cm above ground (on the upper side of the sampler only), and 5 % were from the ear layer (90 cm above ground) (Fig. 2b). Peaks of conidia occurred at Ravenna from 7 to 14 May 2014, with 2.7 ± 2.2 conidia/cm2 recorded for leaf-like traps (10 to 60 cm above the ground) (Fig. 3a), and from 16 to 23 April 2014 and from 30 April to 7 May 2014, with 0.5 ± 0.3 conidia/cm2 for ear-like traps (Fig. 3b).

Paul et al. (2004) placed spore traps at 0, 30, and 100 cm (100 cm was the height of wheat ears) above the soil surface to collect rain splash in wheat fields with maize residue or infested maize kernels during separate rain episodes; they quantified the numbers of F. graminearum spores dispersed in rain splashes in terms of CFU/mL. Spores were recovered from samplers at all heights for all sampled rain events. Both conidia and ascospores were found based on microscopic examination of random samples of splashed rain. Spore density was higher at 0 and 30 cm than at 100 cm. However, spore density was only 30 % lower at 100 cm than at the other heights. In contrast to Paul et al. (2004), we collected spores continuously for more than 2 months between booting and grain maturity, during which time both rainy and dry periods occurred (Fig. 3c), and we counted ascospores and conidia separately. Our results confirm the finding of Paul et al. (2004) that spore density decreases with height of wheat canopy. In addition, our data show that changes in vertical distribution within the wheat canopy are generally similar for ascospores and conidia across dry and rainy periods throughout the season (Fig. 3a and b). The vertical distribution of spores in the canopy was investigated also by Osborne et al. (2002), who found both ascospores and conidia on wheat leaves throughout the canopy; and identified a bimodal distribution of ascospores on wheat leaves within the canopy (i.e., higher concentrations on the highest and lowest healthy leaves than on intermediate level healthy leaves), suggesting the importance of both soil surface-/lower canopy-derived propagules and airborne propagules (from distant or local sources) as ascospore sources. However, Osborne and Stein (2004) found the highest inoculum levels on lower leaves and did not find a bimodal distribution.

Conidia of F. graminearum are typically splash-borne (Sutton 1982). The vertical distribution pattern of conidia may be caused by splashing droplets (that can travel to heights of 20 to 60 cm) generated by rain drops that strike the rain-soaked, residue-covered soil surface and also by the splashing of water generated by rain drops that strike the upper leaves of the wheat canopy (at heights >30 cm) (Walklate et al. 1989). Unlike conidia, ascospores are forcibly discharged from perithecia (Trail 2007) and are then carried by wind (Keller et al. 2014). The vertical distribution pattern that we found for ascospores may be explained by the release of ascospores during daylight hours (Bergstrom and Schmale 2007; Francl et al. 1999) when turbulent air currents cause the ascospores to ascend above the ground (Isard and Gage 2001; Oke 1987; Schmale et al. 2005a); because night hours often lack turbulent air currents, ascospores often settle in the crop canopy at night (Schmale et al. 2006). However, Ingold (1933) reported that perithecia of Pyrenomycetes (presently known as Sordariomycetes, the Ascomycetes class to which F. graminearum belongs) can exude ascospores in a mucilaginous mass through the ostiole instead of being forcible discharged, so that they can also be splash dispersed (Fitt et al. 1989).

To further investigate the possible splash dispersal of ascospores, we studied the degree to which trapped ascospores were spatially aggregated on our leaf-like and ear-like traps. If the spores were randomly distributed, the data should fit the Poisson distribution (Madden and Hughes 1995). The variance-to-mean (VM) ratio is commonly used to measure aggregation; this is ultimately a variance-to-variance ratio because the variance equals the mean in a Poisson distribution (van Maanen and Xu 2003). Therefore, a random distribution of observed data should result in a VM of 1. The index of dispersion (D) is a generalization of the VM ratio; D is the ratio of the observed to theoretical variance, so that D = 1 for random data, D > 1 for over-dispersed data, and D < 1 for under-dispersed data; thus, values of D > 1 suggest aggregation. For binary sampling such as spore trapping (i.e., spore present or absent), D = sy 2 /(ȳ(1−ȳ)/n), where sy 2 is the observed variance and ȳ(1−ȳ)/n is the binomial variance for a sample size of n. Under the null hypothesis of a random distribution, (n − 1) × D follows a χ2 distribution with (n − 1) degrees of freedom. For each location, D was calculated for each sampler (Dsampler) but separately for the leaf-like traps and ear-like traps. D was also calculated for each location; Dlocation was based on the average data from each sampler at a location. Because only a small number of conidia were trapped, D values were calculated only for ascospores. A chi-square test was used to reject or accept the null hypothesis of random distribution (Pielou 1977).

At the ear level, 80 % of the Dsampler and all of the Dlocation values were ≤ 1 with P < 0.01 (Fig. 3a and b), indicating the prevalence of a random distribution of the trapped ascospores. Aggregation at the sampler level was observed in 20 % of cases (Fig. 4a), when spores were found on only one or two of the four ear-like traps, i.e., slides that were oriented north, south, each, and west (data not shown). In these cases, however, there was no prevalence of one orientation over the others. Similar to the data at the ear level, the data at the leaf level indicated a random spatial distribution of ascospores, i.e., both Dsampler and Dlocation values were predominantly ≤ 1 at the leaf level; only 16 to 22 % of these values were > 1 (Fig. 4c and d).

Fig. 4
figure 4

Frequency distribution of the index of dispersion (D) for Fusarium graminerum ascospores trapped within a wheat canopy. D was calculated for each sampler (Dsampler) and for each location (Dlocation) at the ear level (panels a and b, respectively) and at the leaf level (panels c and d, respectively). Percentage values in each panel indicate the frequencies for D ≤ 1 (indicating a random spatial distribution) and D > 1 (indicating an aggregated spatial distribution). Passive spore traps were used and replaced once each week between wheat booting and grain maturity. In panels c and d, black and white bars refer to ascospore trapped on the upward and downward sides, respectively, of the leaf-like spore traps

Full size image

The spatial pattern of spore deposition or plant infection has been used to evaluate the degree of association among sampling units and to develop biological and environmental hypotheses about pathogen dispersal patterns (Madden and Hughes 1995; Madden et al. 2007). An aggregated pattern has been traditionally related to splash dispersal, in which droplets originating from raindrops falling on sporulating lesions may carry many spores (Madden 1997; Waggoner and Rich 1981). A predominantly random pattern, in contrast, has been related to deposition of spores from a diffuse, well-mixed aerial population (Del Ponte et al. 2003; Schmale et al. 2005b).

In previous works (Del Ponte et al. 2003; Schmale et al. 2005b; Shah and Bergstrom 2001; Shah et al. 2000), FHB incidence was spatially aggregated only in fields with large concentrations of maize residues; on the contrary, FHB incidence was randomly distributed in fields with no or low amount of maize residues, indicating that the inoculum for infecting ears was distributed randomly and was thus likely to be dominated by airborne ascospores originating from external sources. Our findings showed a prevalent random distribution in ascospore deposition at the ear level also with a relevant inoculum source within the field. In our wheat plots, the random distribution of the ascospores within the wheat canopy and at the ear level was accompanied by a clear vertical distribution pattern based on the numbers of ascospores trapped across all samplers and sampling periods (Fig. 2a). The latter data indicate an upward movement of the ascospores from the maize residues on the soil surface.