Phase-change-mediated transport and agglomeration of fungal spores on wheat awns

Abstract

Wheat and other staple crops are devastated by fungal diseases. Many fungal plant pathogens are spread via active or passive discharge of microscopic spores. Here, we described the unique transport of spores of the fungal pathogen Epicoccum tritici, causal agent of black sooty mould, on wheat awns. The unique multi-scale architecture of wheat awns, coupled with condensation and evaporation of dew droplets, facilitated the transport and agglomeration of spores of the fungus. First, dew droplets spontaneously transported spores from the tips of awn hairs to the neighbouring stomatal ridges, driven by gradients in Laplace pressure and surface wettability. Subsequently, spores agglomerated into dry clusters due to the Cheerios effect and evaporation, increasing the likelihood of passive spore removal via wind shear and/or rainsplash. Future plant breeding approaches should consider the development of modified spike structures, such as those without awns or awn hairs, to reduce the potential for spread of fungal plant pathogens.

1. Introduction

Cereal crops occupy nearly half of all farmlands worldwide and their yield is projected to double by 2050 to meet increasing demand [1]. Wheat is the most widely grown cereal crop in the world, making it integral to global food and feed security [2]. An annual 15–20% yield loss due to fungal diseases challenges this continual drive to increase the yield and quality of wheat production [3]. Common examples of fungal diseases include leaf rusts [4], Fusarium head blight (FHB) [5] and black sooty head mould [6].

Most fungal plant diseases are spread in three stages: liberation of the pathogen-carrying spores from the diseased plant, drift, and deposition on target sites [7]. The liberation and dispersal mechanisms can be active or passive [8–10], with wind and rain being examples of the latter [11,12]. Recently, jumping-droplet condensation [13] was revealed to be a passive mechanism for liberating rust spores from superhydrophobic wheat leaves [14], with the subsequent dispersal getting an additional boost from a gentle breeze [15]. The effectiveness of these passive spore removal techniques depends on how strongly the spores are adhered to the plant surface [16]. The adhesion, and consequently, the force of removal, is primarily dependent on the size of a dispersal unit, i.e. whether the spores are removed singularly or in clusters [16]. Spores are generally aggregated due to physiological processes such as being embedded in a mucilaginous layer [17] or due to external stimuli such as humidity and electrostatic forces [18,19]. However, little is known about whether condensation or evaporation play an important role in the transport and aggregation of fungal spores on wheat awns, which are the thread-like appendages attached to the florets or seeds (figure 1a) [20,21].

Figure 1.

Awn structure and transport of fungal spores. (a) Photograph of a wheat spike with awns. (b) Close-up view of an awn, showing the conical hairs on the outside and a central groove. The black spherical bodies are the spores of the fungus Epicoccum tritici, causal agent of sooty mould. (c) Close-up view of black mould spores on the awn. These are found both singularly and as clusters. (d) Scanning electron micrograph of a wheat awn, showing the conical awn hairs and a central stomatal ridge. (e) Schematic of how condensation and the awn structures work synergistically to carry singular spores from the tip of the awn hair to the stomatal ridge to form clusters.

Using black sooty mould as a model system, we show that a natural dew cycle results in the agglomeration of pathogenic spores, resulting in spore clusters readily removed by wind or rain. We used optical microscopy to reveal that the formation of spore clusters from singular spores is a multi-step process that involves both condensation and the subsequent evaporation of the dew. First, spores ratchet to the base of the awn’s conical hairs by virtue of the asymmetric Laplace pressure of dew droplets. Second, droplets near the base of the hairs preferentially spread into hydrophilic stomatal ridges, bringing adhered spores with them. Third, spores within the wetted stomatal ridges self-assemble via the surface tension of their menisci (the Cheerios effect). Fourth, the subsequent evaporation of the dew leads to the formation of dry and closely packed spore clusters. Finally, we performed an analysis indicating that these spore clusters are more readily liberated by wind and/or rain compared to isolated spores. Our findings suggest a secondary role of the wheat awns beyond their important and established role as seed dispersal units [21]. Future plant breeding approaches should consider the development of modified spike structures, such as those without awns or awn hairs, to reduce the potential for the spread of plant pathogens.

2. Material and methods

2.1. Plant material

A series of healthy and diseased wheat spikes (heads) were collected from wheat research plots at Virginia Tech’s Kentland Farm, in Blacksburg, VA in 2016. The diseased wheat spikes were weathered, dark and discoloured. These symptoms were consistent with a disease complex known as black sooty head mould, caused by a number of saprophytic fungi [22].

2.2. Identification of fungus present on infected wheat heads

Wheat spikelets from the diseased wheat spikes were observed with a stereomicroscope. Numerous black spherical structures were observed on each of the spikes (figure 1b,c), signs of a fungus associated with black sooty mould [22,23]. Black sooty mould has been associated with a number of different species of fungi [22,24]. Consequently, additional work was warranted to carefully identify the fungus that produced the aforementioned symptoms and signs (see electronic supplementary material for details). BLAST queries of the resulting ITS sequence supported the identification of the fungus as Epicoccum tritici. The samples obtained from the field had already experienced natural dew cycles analogous to those we perform in the laboratory. Therefore, the awn surfaces already contained some phase-change-mediated agglomerations of spore clusters, in addition to having individual spores (approx. 10 mm⁻²). Most spore clusters were found near the stomatal ridge region. Singular spores were found throughout the awn, including the hair tips.

2.3. Phase-change experiments

For the condensation experiments, sections of diseased wheat awns were thermally bonded to the Peltier stage. The mean width of the awns was 0.45 ± 0.01 mm while the length varied. The Peltier stage temperature was set between $T_s=-5^{\circ }\mathrm{C}$ and 5°C, about 10°C lower than the dew point temperature which ranged from T_DP ≈ 5–15°C depending on day-to-day variations in the air’s relative humidity (RH = 50–70%) and temperature (T_∞ = 15–22°C). This accounts for the temperature drop across the awn (which is a poor conductor of heat) to ensure that the awn’s surface is just beneath the dew point temperature. In some cases, humid air from an ultrasonic humidifier (PureGuardian H940) was directed toward the thermally bonded awn using plastic tubing to aid in the condensation process. Technically, in this case, dew droplets would be growing simultaneously from condensation (i.e. diffusion) and from the impact of smaller fog droplets. To observe the evaporation-induced agglomeration process, 10 μl drops of distilled water were deposited on a diseased awn section with the Peltier stage set to T_s ≈ 40°C.

2.4. Droplet impact experiments

To simulate the effect of rain impact on spore dissemination from an awn’s surface, dyed water droplets were generated from a pipette fixed at a certain height. The awn sample was fixed by a clamp directly below the pipette. A white paper was placed under the sample at the start of each trial. After each droplet impact, the paper was collected and analysed under a microscope for any splashed droplet(s) containing spore(s). The set-up is shown in electronic supplementary material, figure S3.

2.5. Imaging

The experimental videos were obtained using a high-speed camera (Vision Research, Phantom v711). For top-down imaging, the camera was mounted on an optical microscope (Nikon, Eclipse LV150). A 20 × objective lens with a long working distance (Mitutoyo, 378-810-3) was used for the magnified view of the spores and the condensing droplets. Side-view imaging was done by attaching either a macro lens (Canon, MP-E 65 mm, 1–5× mag) or a long-distance microscopic lens (Infinity Infiniprobe TS-160, 1–16× mag) to the high-speed camera.

3. Results

The cross-section of a healthy wheat awn exhibits two stomatal ridges, one on either side (figure 1d), indicative of an awn’s photosynthetic capacity [25]. The intermediate periphery of the awn is covered by needle-like silicified hairs (figure 1d), which control the directional movement of the dispersal unit (i.e. awn and associated seed) and also act as hooks for passive seed dispersal [21]. Throughout this report, we will refer to the black spheroidal bodies (figure 1b,c) on the sooty mould-infected wheat awns as spores. When these diseased awns were subjected to a complete dew cycle (condensation followed by evaporation), it was observed that spores migrated from the awn hair tip or base to the stomatal ridge (width 740 μm ± 120 μm) and also formed clusters of spores (figure 1e). This multi-step clustering process will now be explained in detail.

3.1. Ratcheting water droplets

Condensation forming on awn hairs tend to nucleate at the tips of the hairs (figure 2a). The hair’s sharp spike facilitates condensation by increasing the local diffusive flux [26]. For a subset of experiments where an ultrasonic humidifier was used, these spikes are also effective at capturing small fog droplets (see electronic supplementary material, video S1). As shown in figure 2b, condensates forming on the tip of the conical hairs move towards the wider base of the hairs. This can happen due to a Laplace pressure gradient originated by the widening slope of the conical hair [27,28]:

$$\frac{\mathrm{d}P}{\mathrm{d}z}\approx \frac{\gamma }{{(r+R)}^2}\alpha ,$$ 3.1

where dP/dz is the Laplace pressure gradient along the hair length, γ is water surface tension, r and α are the mean radius and the half-apex angle of the hair, respectively, and R is the dew droplet radius. Comparing the capillary pressure gradient of equation (3.1) and the gradient of gravitational pressure, ρg, the Laplace pressure gradient is three orders of magnitude higher for r = 30 μm, R ∼ 10 μm, $2\alpha \approx {20}^{\circ }$, water density ρ ∼ 1000 kg m⁻³, and gravitational constant g ∼ 10 m² s⁻¹. This confirms that a dew droplet forming at the hair tip can move from the tip to the base regardless of the hair’s gravitational orientation (see electronic supplementary material, video S1). From the videos, the mean ratcheting speed of a droplet without any spores was measured to be about 60 μm s⁻¹, with a minimum of about 30 μm s⁻¹ (electronic supplementary material, figure S2b) and maximum of about 90 μm s⁻¹ (electronic supplementary material, figure S2a). Even though the size (or volume) of the droplets are similar in both sets of images, large differences in the mean speed indicates that local surface properties may play an important role. This can also be found from the expression of capillary driving force for such droplet motion which is given by $F_c={\int }_{\mathrm{base}}^{\mathrm{tip}}\gamma (\cos {\theta }_a-\cos {\theta }_r)\hspace{0.17em}\mathrm{d}l$, where θ_a and θ_r are the advancing and receding angles of the water droplet on awn hair [28,29]. Local changes in roughness and surface energy changes the advancing and receding angles and consequently the driving force F_c and the droplet velocity. Although a theoretical prediction of droplet velocity is beyond the scope of this study, we have calculated the mean velocity of a droplet sliding on conical cactus spines from the experimental images in [28]. We have chosen this study as the geometry of the cactus needles or spines is very similar to the awn hairs in our study. While the needles are much longer than the awn hairs, the length does not dictate the force on a sliding droplet due to the Laplace pressure difference. The apex angle of the cactus spines is about 12° while the width near the spine tip is about 26 μm. These values are very close to the apex angle of about 20° and width of about 29 μm in the case of the awn hairs. From analysis of the screenshots of droplet sliding in [28], we found the droplet speed to be about 24 μm s⁻¹, the same order of magnitude as in our experiments.

Figure 2.

Formation of spore clusters from singular spores. (a) Ratcheting motion of a droplet along an awn hair, carrying a spore from the tip to the base. (b) Schematic explaining the ratcheting motion due to a Laplace pressure gradient along the length of the conical hair. (c) Droplets at the base of the hairs coalesce and preferentially move towards the more hydrophilic stomatal ridge (light coloured stripe in the middle of t = 0 frame). (d) Schematic of the droplet movement from the hair to the stomatal ridge. (e) Coalescence-induced agglomeration of spores. Here, a singular spore is brought closer to a spore cluster as they are both connected by one droplet. (f) The spore-agglomeration is achieved through the deformation of the droplet air–water interface which brings neighbouring spores close to each other to form a cluster. (g) Evaporation-induced secondary agglomeration of spores. The temporal snapshots of the experimental video show how the spores within a cluster undergo secondary three-dimensional packing as the encapsulating thin film of water evaporates. (h) Schematic explains the evaporation-induced secondary spore aggregation mechanism. (i) A typical diseased wheat awn with singular and clustered spores.

This ratcheting motion of dew droplets on awn hairs serves two purposes: (1) freeing up space for more condensation on hair tips by continuously draining droplets, and (2) transporting adhered spores from the tip of the hair to the stomatal region (figure 2a). For a ratcheting dew droplet to pick up and carry a spore, the attractive force between the spore and droplet must overcome the adhesive force between the spore and the awn hair. For micrometric spores, the adhesion forces are mainly arising from either the electrostatic forces or van der Waals forces. In a humid environment, the electrostatic forces become negligible compared to the van der Waals forces [30,31]. The van der Waals interaction force between a spherical particle and a flat plate separated by a distance z scales as F_vdW ≈ (A/6z²)R, where A is the Hamaker constant and R is the particle radius [31]. Typical Hamaker constant values for dry organic materials are of the order of 10⁻²⁰ J [32]. Here, we use a representative value of A ≈ 5 × 10⁻²⁰ J for the dry spores and spore clusters on the awn surface [33]. As the micrometric spores can be safely assumed to be larger than the nanometric awn surface asperities, this would result in an improper contact between the surface and the spores increasing the separation distance z [33]. Typically, van der Waals forces are dominant at less than ∼10⁻⁸ m distances. Thus, we choose z ∼ 10⁻⁹ m which falls in between a molecular level separation of 10 ⁻¹⁰ m and the maximum effective van der Waals distance of 10⁻⁸ m. With these values, the van der Waals adhesion force is calculated to be F_vdW ∼ 10⁻⁹ N for a R ∼ 10 μm spore. In comparison, the capillary force exerted by a dew droplet on the spore scales as F_c ∼ γR, which for a spore of 10 μm radius is about an order of magnitude higher than the van der Waals force, F_vdW. This analysis shows that the capillary force exerted by the dew droplets is effective for dislodging and carrying small spores along the hair. Of course, the above calculation is based on an assumption of surface roughness due to a lack of experimental measurements of the awn surface roughness. If the spores are small enough to fit within the troughs and ridges of the awn surface, the separation distance will be lower than 1 nm. In that case the van der Waals attraction would also increase significantly and at some critical distance the capillary attraction due to the water droplet will not be able to detach the spores from its position (see electronic supplementary material, figure S4).

3.2. Spore agglomeration

Upon completion of the early stage of condensation, the droplets grow in size primarily by coalescence [34]. As shown in figure 2c and in the electronic supplementary material, video S2, the coalesced droplets were always seen to be moving towards the stomatal region and accumulating there as a film. It is well known that when two asymmetric droplets coalesce, it results in a directed motion of the merged droplet due to a net movement of its centre of mass [35,36]. The movement of the droplet is dictated by the imbalance of interfacial tensions on the surface which can be due to chemical or temperature gradients or a combination of both. In our case, as the awn is firmly attached to the peltier with thermal paste, any temperature gradient on the awn surface can be neglected. So there must be a chemical gradient between the region just adjacent to the hairs and the stomatal region. The hairs and their base region are comprised of epidermal cells which have a higher silicon content [25], rendering them more hydrophobic in comparison to the stomatal region. Furthermore, the stomatal band is overlaid on the chlorenchyma tissues, whose cell walls are made of hydrophilic polymers including pectin and cellulose [37,38]. Thus the chemical gradient arises from a wettability gradient between the hair base and the stomatal ridge and a coalesced droplet move towards the more hydrophilic region (figure 2d). We have calculated the average velocity of droplets outside the ridge during coalescence with droplets within the ridge. This average velocity 692 ± 372 μm s⁻¹ is higher than the average coalescence-induced velocity further away from the ridge (V ≈ 86 ± 39 μm s⁻¹) or within the ridge (V ≈ 502 ± 330 μm s⁻¹). Due to insufficient information regarding the spreading coefficient for the droplets on the solid surface as well as the chemical gradient, we could not provide a theoretical estimation of the velocity as derived extensively by Brochard [36].

Once enough droplets coalesce near or along the stomatal ridge, individual spores get pulled together by the water drop/film that spans between them (figure 2e). The self-assembly of small particles at a fluid interface is known as the ‘Cheerios effect,’ because of a similar aggregation phenomena seen for cereals in a bowl of milk [39]. The Cheerios effect is also seen in nature when small duckweed plants or mosquito eggs aggregate at the free surface of a pond [40]. The mechanism for the Cheerios effect is the meniscus that extends from each particle [41], where overlapping menisci zip together to minimize their interfacial energy [39,42] (figure 2f). As the coalesced droplets preferentially move towards the stomatal region, a large number of spore clusters are generally found in that region.

The newly formed spore clusters become more closely packed as the water subsequently evaporates (see electronic supplementary material, video S3). To observe this phenomenon, a water droplet was deposited on two spore clusters as shown in figure 2g. The humidity of the environmental chamber was decreased to promote a subsaturated ambient, resulting in the evaporation of the droplet. As the droplet continually decreased in volume, individual spores within each cluster were brought closer together. This led to the formation of multiple densely packed clusters as can be seen in the last frame of figure 2g. Comparing the projected area occupied by the two spore clusters in the first frame of figure 2g and the six smaller spore clusters in the last frame, a reduction of about 34% in area was measured due to evaporation-induced agglomeration. Although we were not able to measure the thickness of these spore clusters, we can safely state that each of these packed clusters had at least two layers or a minimum thickness of about 20 μm. Similar to the aggregation of the floating spores to form clusters, this phenomenon is due to the deformation of air–water interface which induces lateral capillary forces on the spores. But for this later case of evaporation-induced agglomeration, the contact line position and contact angle are more important than the weight (or buoyant force) of the spores [39,43]. Similar capillary aggregation is observed in the formation of two-dimensional ordered protein arrays [43], in the motion of micrometric particles in an industrial coating [44], or grouping of nematodes in a thin liquid layer [45]. The schematic in figure 2h explains this capillary aggregation process. During condensation over spore clusters, the individual spores within a cluster are joined by a layer of liquid. As the droplet evaporates into a thin film and the thickness of the film drops below the spore diameter, the spores are attracted to each other. The net attracting force is originated from the deformed interface at the three-phase contact line.

A typical diseased wheat awn, with both isolated spores and spore clusters, is shown in figure 2i. Most of the spore clusters were found on or near the stomatal ridge, which is not surprising considering the frequency of the coalescence/evaporation-induced spore agglomeration that occurs in that region. In the next section, we will discuss how the formation of such spore clusters are conducive to the spread of disease by virtue of their ease of removal compared to isolated spores.

3.3. Mechanisms of spore removal

The necessary first step for any fungal disease transport in plants is the efficient removal (i.e. liberation) of the disease-bearing spores from the host plant [12]. The removal is dependent on the shape and size of the dispersal units (isolated spore versus cluster), the adhesion strength of the unit with the plant surface [16], and the strength of the dislodging forces to overcome the adhesion forces [46]. While wind is the most abundant spore removal mechanism available in nature [16], it is not easy to detach spores from the surface because of the boundary layer extending from the plant surface [47]. The hydrodynamic force exerted by wind with a mean speed of U, on a spherical particle of radius R_p, can be defined as $F_{\mathrm{shear}}\approx 1.7\times 6\pi {\mu }_a(U/\delta ){R_p}^2$, where μ_a is the viscosity of air and U/δ is the shear rate or the velocity gradient within a boundary layer of thickness δ [48]. This shear force has to overcome the adhesion force exerted by the surface on the particle. For a completely dry solid surface with uncharged particles, this adhesion force is mainly due to the van der Waals force, F_vdW ≈ (A/6z²)R_p, where we have again used A ≈ 5 × 10⁻²⁰ and z ∼ 10⁻⁹. Of course, for spore clusters, due to even less contact between the surface and the spores, this value will increase further, decreasing the van der Waals attraction [32].

In figure 3a, the hydrodynamic shear force (green line) and adhesive van der Waals force (red line) are graphed as a function of the particle size. Note that an isolated spore would correspond to R_p ≈ 10 μm, with larger values of R_p corresponding to increasingly larger multi-spore clusters. For the average wind speed, we have used U = 10 m s⁻¹, which approximately corresponds to the global average [46]. As evident from the plot, an average wind speed of 10 m s⁻¹ cannot overcome the van der Waals attraction for an isolated spore. Although the wind shear overcomes the surface adhesion above a critical cluster size of R_p,c ≈ 60 μm, it only becomes an order of magnitude higher for a cluster size of about 200 μm. The effective size of the spore clusters observed in figure 1c and figure 2g are above this critical size and therefore fall within the green zone in figure 3a. Increasing the wind speed will further reduce the critical particle size, facilitating the removal of smaller spore clusters.

Figure 3.

Removal of spore clusters. (a) Phase map showing the criteria for successful removal of spore/spore clusters. The adhesion force or van der Waals attraction between the awn and the spores is shown by the red line. The removal forces are due to rain impact (blue line), wind shear (green line), and oscillatory forces due to wind or rain (grey continuous curve). The dashed grey line demarcates a typical single spore size of R_p ≈ 10 μm from larger sizes representing spore clusters. The dynamic force exerted during average rainfall (R = 2 mm, u = 6 m s⁻¹) is larger than the van der Waals force irrespective of spore size. But for an average wind speed of U = 10 m s⁻¹, the wind shear force can only overcome the surface adhesion if the spore size is more than about 60 μm. This critical size decreases in value as the wind speed increases (shown by the arrow). (b) Results of the water droplet impact on the wheat awn experiments. The number of splashes under the diseased awns are counted on the y-axis. The number of spore(s) in each of these splashes are then plotted on the x-axis. Experiments were performed for four different awn orientations. Most of the splashes contained only single spores.

Rain is also an important abiotic factor, which can remove spores via the splash of the impacted droplet (wet dispersal) [49,50] or by momentum transfer (dry dispersal) [51]. In wet dispersal, the advancing liquid–air interface of the impacted raindrop captures some spores which are ejected from the surface once fragmentation occurs at the end of spreading. Dry dispersal can occur in at least two different ways: elastic collisions with an impacting spore-laden droplet, or via vortices generated from drop impact [51]. The dynamic force exerted by a moving liquid–water interface after drop impacting on a solid surface is given by $F_{\mathrm{splash}}\approx C_D\rho {U_c}^2{R_{\hspace{0.17em}p}}^2$ [51], where C_D is the drag coefficient for the spores which depends on its shape and U_c is the average speed of the moving contact line. For spherical particles, C_D ≈ 0.5. The contact line speed after the drop impact on the spore-laden surface not only depends on the impact speed but also on the interaction between the droplet and the particles in front of the advancing rim. One recent study has experimentally found out that the velocity of a moving front just after a water droplet impacts a solid surface is about 7 m s⁻¹ for a drop radius of 1.8 mm and impact velocity of 3 m s⁻¹ [52]. The advancing velocity decreases exponentially as the spreading radius increases. Considering an average impacting rain droplet radius to be R = 2 mm and the impact velocity of u = 6 m s⁻¹, we can safely assume a value of U_c ≈ 14 m s⁻¹, which is twice the maximum advancing velocity of the spreading droplet in [52]. We then plot the variation of dynamic force due to the splash with the particle size R_p in figure 3a (blue line). The plot shows that for the average raindrop size and impact velocity, the impact force is enough to remove spores from the surface irrespective of the particle size.

Wind and rain can also detach spores and clusters from the awn surface indirectly, by triggering oscillation of the surface. Any singular spore or spore cluster adhered to the awn surface would experience an inertial force due to a wind or rain-induced vibration of the awn. For an oscillation frequency f and amplitude A_osc, this inertial force F_osc can be computed as $F_{\mathrm{osc}}\approx {\rho }_s{R_{\hspace{0.17em}p}}^2{(2\pi f)}^2{A}_{\mathrm{osc}}$, where ρ_s is the density of a spore. For a gentle breeze over the canopy, we can use an oscillation frequency of 1 Hz and an amplitude of 1 mm for the awns. We have plotted the variation of F_osc with the particle size R_p in figure 3a. For less than 100 μm size particles, this oscillatory force is much less than the van der Waals adhesion force. This agrees well with the findings of Nath et al. [14] where they also found rust spores on wheat leaves are difficult to remove from the surface via oscillation.

We also experimentally demonstrated the effect of rain impact on spore removal from a diseased awn. Figure 3b shows the experimental results where the total number of spore-containing splashes are plotted against the number of spores within each splash. Different colours of the bars represent the different angles at which the diseased awn was placed with respect to the rain impact direction. The majority of the splashed satellite droplets contained only one spore (irrespective of the angle), although in a few cases the maximum number of spores exceeded 10. This can be explained by comparing the force of impact with the cohesion force between the individual spores within a cluster. The force exerted by a droplet impacting on an unyielding surface is F_i ∼ ρu²R² where u and R are the velocity and radius of the raindrop, respectively [53]. For our experiments, the size of the impacting droplet was R ≈ 2 mm and u ≈ 3 m s⁻¹ which translates to an impact force of 0.04 N. But only a fraction of this total impact force would be utilized for breaking apart two spores in a cluster. This factor can be estimated by the ratio of the projected area of a spore (∼(10⁻⁵)² m²) to that of the impacting droplet (∼(10⁻³)² m²). With this assumption, the impacting force responsible for breaking apart two spores comes out to be 4 μN. Typical experimental measurements estimated an inter-particle force of about 1 nN for wheat leaf rust spores [51], still three orders of magnitude lower than the area-adjusted impact force. Therefore, a direct impact on a spore cluster might result in the breakage of the cluster and expulsion of only singular spores from the surface instead of a whole cluster. Of course, the actual impact force on the awns would be lower than the current calculation as the awns are not entirely unyielding [49].

The results of figure 3b can also be rationalized by the following argument. In our laboratory setting, the natural rain impact was simulated by generating single droplets from a point source above the awn surface. Slight changes in the trajectory of the falling droplets either due to small airflow in the laboratory or due to shedding vortices [49,54] ensures that not all raindrops will hit an individual spore or a spore cluster directly. In those cases, the spores or clusters can be dislodged by the spreading front of the impacted droplet as suggested by our calculation in the previous paragraph.

Once removed from the awn surface by either wind or rain, spore clusters will have a different aerodynamic behaviour than singular spores due to their difference in shape, size and density [19]. As the clusters are heavier than singular spores, they have higher settling velocities compared to individual spores. Thus once removed from the host, they are more likely to stay within a canopy to create multiple closely spaced disease focal points. By contrast, individual spores can be carried further in wind for long-distance disease dispersal [12].

4. Discussion

Though the function of the cereal awn in grain production [25], yield [55] and dispersal [21] has received considerable attention, little is known about the role of the awn in pathogen dispersal. Here, we showed through experiments how the multiscale hierarchical structure of wheat awns facilitates the formation of compact spore clusters in the presence of fog or dew and its subsequent evaporation. Cluster formation not only reduces the critical spore removal force but also changes the subsequent dispersal and deposition dynamics. Our experimental and theoretical analysis shows that rainfall impact is particularly effective in disintegrating the spore clusters and dispersing individual spores into the air stream. This helps explain the observations reported in other studies on the sudden increase in spore concentration in the air and also on neighbouring wheat heads after a heavy rainfall [56]. Moreover, our research also shows that apart from rain, spore removal via wind likely becomes easier with the increase in spore cluster size. Due to the widespread availability of wheat awns showing symptoms of black sooty mould, the present study uses spores of E. tritici as the model for studying the dynamics. However, we suggest that our characterization of phase-change mediated spore transport and liberation on wheat awns should be directly applicable to other fungal diseases on awns, such as FHB. Future plant breeding approaches should consider the development of modified spike structures, such as those without awns or awn hairs. Such experiments could help elucidate specific role of the awn and awn hairs in spore dispersal. The capillarity-induced phenomena described in this report were all done in a controlled laboratory environment with a narrow ambient temperature and relative humidity range. This may not be commensurate with the vastly different ambient temperatures across the wheat-growing regions in the United States as well as the world. Changes in the surface and air temperature changes the condensing droplet temperature [57] and their contact angle with the plant surface [58]. Future studies should look into the effect of different ambient conditions on the phenomena described in the manuscript.

Data accessibility

The data supporting the findings of this study are available within the article and the associated electronic supplementary material. Any other data are available from the corresponding author upon request.

Authors' contributions

G.J.I.: formal analysis, investigation, writing—review and editing; R.M.: formal analysis, writing—original draft, writing—review and editing; H.A.G.: resources, writing—review and editing; D.G.S.: conceptualization, methodology, writing—review and editing; S.J.: conceptualization, methodology, writing—review and editing; J.B.B.: conceptualization, methodology, supervision, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This work was supported by the USDA National Institute of Food and Agriculture, award no. 2018-67013-28063.