1. Data collection

To investigate the impact of plantation effects on disease prevalence and severity, a citizen science approach was used, following the methodology of previous studies (Beza et al., 2018). This widely used research approach has the advantage of allowing the collection of large amounts of data from both professional and non-professional volunteers and has been used successfully to explore general correlations in several topics (Silvertown, 2009). Furthermore, the method involved volunteers providing data on grapevine grey mould disease and the associated growing environment in different parts of the country, thus making it suitable for answering research questions at a large spatial scale (Gupta et al., 2022).

We collected geo-referenced data on the grey mould (Botrytis cinerea, BC), downy mildew (Plasmopara viticola, PV), and powdery mildew (Erysiphe necator, EN) diseases severity in a traditional wine- and grape-producing country, Hungary (area: 93,030 km², 9.7 million inhabitants). Citizen data were surveyed using an online, semi-quantitative questionnaire between July 2020 and March 2023. Around 5000 participants were recruited via e-mail, social media platforms and personally at professional events in Hungary interested in grape production (non-professional or professional). The questionnaire was completed by 181 volunteers, 112 of whom were qualified in plant protection (61.87 %) and gained independent observations (N = 239 vineyards) (Figure 1).

Figure 1. The map shows the geo-referenced vineyards from Hungary and nearby (red dot, N = 239 independent observations).

Save View full size Expand inline Collapse inline

A point with a darker shade reflects more vineyards from the same or close locations. The x and y axes represent the longitude and latitude coordinate scales, respectively.

2. Questionnaire survey

Volunteers provided information on the variables of the cultivated area, the average cultivation and general average plant health conditions over the last five years. As a result, overall multi-year crop protection effects and trends could be quantified, and the long-term relationships between factors influencing pathogen infestation could be explored. The surveyed variables belonged to three main sections: infection evaluation, vineyard features, and crop management, as the surveyed variables are described briefly in the following.

3. Statistical analysis

Before running the statistical analyses, we applied a couple of transformations regarding the two infection variables (such as infection ratio and infection severity) in each pathogen separately. First, we calculated a single general infection occurrence index based on the average of the two infection variables. Second, we applied a min-max normalisation procedure on these pathogen-specific infection occurrence indexes in each year (when the given data were provided) separately. For this procedure, the used transformation formula was x'=(x-xmin)/(xmax-xmin), where the x represented the general infection occurrence index, and the x_min and x_max were the yearly minimum and maximum infection values, respectively. As a result, all infection data were brought to a standard range between 0 and 1. Finally, due to the non-normal distribution of infection occurrences, these normalised infection variables were binarised based on the median of the corresponding year (0: below the median, 1: equal or higher median). These pathogen-specific, binarised infection occurrence (hereafter infection occurrence) variables were used in further statistical analyses.

In the present study, we tested the relationship between the infestation occurrence and the most relevant plantation characteristics by a Generalized Linear Mixed-Effect Model with binomial error structure (GLMER-b; using the ‘lme4' package, Bates et al., 2015). Models were run separately for each grape disease, such as BC, PV and EN. In the models, the response variable was the infection occurrence, and the predictor variables were the following covariates: inclination, adjacent plantation, age, distance, treatments and the survey year. In addition, the orientation, the inter-row management, leaf removal and the region ID were represented as factors, while the grape sort and the farmer ID were entered as random factors. In each statistical model, we applied a combined statistical weight considering the inequality data points given by the responses of qualified or unqualified grape producers (qualification yes: 1, no: 0.5) or the newly established plantations (if age ≤ 4 years: age × 0.2; or 5 ≤: 1).

In separate statistical analyses, we revealed the possible role of the characteristics of grape varieties regarding the occurrence of infestation. First, we calculated the mean of the normalised infection occurrence by grape varieties in each tested pathogen. Then, we applied general linear models in which the variety-specific infection occurrence mean was entered as the dependent variable. At the same time, the pathogen sensitivity, ripening period, berry skin thickness and bunch compactness were entered as covariates, while berry colour and utilisation were factors. The observations were weighted by their relative sample size (also considering the qualification and plantation age) in the statistical models. Grape varieties with three or more observations were included in further analyses. Thus, the final models were based on 22 grape varieties.

BC infection followed by specific climatic conditions may cause ‘noble-rot' as a part of some specific traditional wine production (dehydrated berries with increased sugar concentration) in some conventional grape varieties (i.e., Furmint, Sárgamuskotály, Hárslevelű) selected for this technology (Fournier et al., 2013). Therefore, we run the same model structure after excluding these varieties from the databases by eliminating the possible bias of these specific grape varieties in the relationship between BC infection and variety features.

In all statistical analyses, a likelihood ratio test (LRT) was applied to test the effects of each predictor in the extended models. In addition, the numeric predictors were centred and scaled before running the analyses, while the model diagnostics were always checked in the final statistical models. Finally, the derived variables were calculated, and all statistical analyses were run in the R environment (version: 4.1.0, R Development Core Team, 2019).

1. Roles of the plantation features and crop management

We found that the BC occurrence was significantly associated with the orientation (Figure 2A). Primarily, the NE–SW orientations increased the presence of the BC and the SE–NW row orientation less, while plantations with the N–S and the E–W directions resulted in moderate infection occurrences. The ratio of adjacent plantations suggested an apparent increase in the presence of BC infections (Figure 2B). Similarly, the occurrence of BC infections increased with the number of chemical treatments (Table 1) and with the elevation of the canopy wall from the ground (Figure 3). We found significant spatial variation among the growing regions (Table 1, Table S2, Figure S1). A marginal association was revealed with the observation year. We failed to detect any effect regarding the other plantation and crop management features, such as plantation age, slopes, inter-row management and leaf removal (Table 1, Table S2).

Figure 2. The figures show the relationships between the infection occurrence of grape pathogens, such as Botrytis cinerea (BC, blue) and Plasmopara viticola (PV, green), and two primary vineyard features, such as row orientation (A and C) and surrounding vineyards (B and D).

Save View full size Expand inline Collapse inline

The tested row orientations, such as SouthEast–NorthWest, North–South, NorthEast–SouthWest and East-West, were abbreviated by SE–NW, N–S, NE–SW and E–W, respectively. The ratio of the surrounding vineyards around the focal vineyard within a 1 km radius was also classified into the categories such as none (0–5 %), scattered (6–50 %), many (51–80 %) and high (81 % or higher) classes, respectively. The light (absent) and dark (present) shades illustrate the occurrence of infection. The mosaic size represents the ratio of the observation numbers within each category (width) according to the infection occurrence (height).

Figure 3. The figure shows the effects of the canopy wall height on infection occurrence in Botrytis cinerea (BC).

Save View full size Expand inline Collapse inline

The regression line (red) represents the probability of BC infection occurrence for the cordon heights with the confidence intervals (grey). Points were slightly jittered vertically for better visualisation.

We found that the infection of PV increased along with the magnitude of the inclination of the plantation (Table 1, Table S3). Similarly to the BC, different row orientations highly varied with the PV infection occurrence, such as an increased occurrence was revealed at the NE-SW and SE-NW row orientations and moderate ones at the N–S and the E–W directions (Figure 2C). In addition, the increased adjacent plantations around the focal plantations elevated the infection occurrence (Figure 2D). Furthermore, the observation year showed a significantly decreasing trend during the study. Similarly, growing regions reflected significant spatial variation in PV infection. Finally, a marginal association with the inter-row management was detected, suggesting that the covered inter-row caused more PV infection than the uncovered plantations. We could not detect any effect on plantation age, the elevation of the canopy wall, the number of chemical treatments, and leaf removal (Table 1, Table S3).

None of the tested variables affected the occurrence of EN. Besides these general results, some marginally insignificant effects were detected, suggesting patterns regarding row orientation, the height of cordon arms, leaf removal treatment and regional differences (Table 1, Table S4).

2. Roles of the grape variety

In BC, the observed infection occurrence means detectable increased with the level of literature-based susceptibility (Figure 4A). Infection increased with the delay of the ripening periods, so late-ripening varieties were more likely to be infected than early- or middle-ripening varieties (Figure 4B). Infection means could also vary among the berry colours, as blanc varieties suffered more from BC infections, while noir varieties enjoyed a relatively higher defence. Interestingly, berry skin thickness and bunch compactness also positively increased the long-term chance of BC infection. Finally, we could not detect any effect of the utilisation (Table 2, Table S5).

After excluding grape varieties selected for BC noble rot in an independent statistical analysis, we revealed similar significant results for the ripening period, berry colour and skin thickness. However, the susceptibility became marginally insignificant, while the bunch compactness and utilisation showed no detectable effect on the BC infection means (Table 2, Table S5).

We could not detect any significant effects among the tested variables regarding the grape varieties in PV and EN (Table 2, Table S5).

Figure 4. The figures show the relationship between the observed BC infection occurrence (variety means) and the literature-based susceptibility (A) and ripening (B).

Save View full size Expand inline Collapse inline

Figure 4A: A black point represents a variety-specific literature-based susceptibility with the observed infection occurrences. The point size reflects the sample size of the given variety (a larger point represents more observations). The regression line (grey) shows the significant relationship between the tested variables. The labelling of the points was based on the varieties' names according to the VIVC database (see details, Table S6). Figure 4B: The box size corresponds to the interquartile, while the whisker shows the nonoutlier range. The bold horizontal line indicates the median of the corresponding ripening category.

1. Vineyard features

Before establishing a plantation, a farmer has to decide on several plantation characteristics that cannot be modified afterwards; however, these environmental and biological conditions could affect the production system and the cultivation practices in the long term. Therefore, these characteristics have a long-term impact on the growth conditions and the crop's success from crop management, including pest control. For example, appropriately selected planting structures could contribute to reducing infections by plant pathogens by ensuring the vines are in the proper condition and by inhibiting the spread of pathogen propagules (Ellis, 2008; Stapleton et al., 1995).

Sloping and hilly areas are often preferred for vineyards due to favourable climatic conditions, such as higher solar radiation exposure and better air ventilation, especially in traditional grape-growing regions (van Leeuwen et al., 2004). Previously, we expected that row orientation and slopeness generate a considerably high spatial variance in environmental conditions among vineyards, which could be associated with a higher variance in the occurrence of specific plant pathogens. Supporting this prediction (see Figure 2, Table 1, Table S2, Table S3), we found that specific row orientations increased the occurrence of BC (i.e., NE-SW) and PV (i.e., NE–SW, SE–NW) infections, while the higher slopes generated a similar increasing pattern in PV. The result regarding orientation could be linked to the prevailing wind direction (i.e., N and NW) in Hungary, which changes to NE during summer (Mezősi, 2017), thereby promoting the spatial dispersion of air-borne spores. Furthermore, previous studies showed that row orientation and slopeness could indirectly influence pathogen infections via the environmental conditions affecting plant health or growth (van Leeuwen et al., 2018). The choice of row orientation also determines both the amount and the duration of irradiation reaching the vineyard, which also influences the ripening time of grape varieties through soil warming. Therefore, earlier ripening was expected in areas where the row orientations promote the rapid increase of the soil temperature better compared to those plantations with unfavoured row orientations (van Leeuwen et al., 2018). Although the BC infection can develop at any phenological stage, berries are markedly more susceptible from the ripening stage onwards (Kretschmer et al., 2007) since the susceptibility of berries to infection increases as the sugar content of the berries increases towards ripening (Deytieux-Belleau et al., 2009). Contrary to the result of the present study, it was previously found that plain vineyards suffered a higher risk of infection with more severe symptoms than those with slight or steep downhills because the slopes receive higher solar radiation, favouring the growing and ripening conditions and causing healthier vines; however, the susceptible phenological stage was shifted earlier (Seguin, 1986; van Leeuwen et al., 2018).

It may undoubtedly seem that increasing the distance between the lowest leaves and the ground makes work easier and isolates the foliage from the soil (Seguin, 1986). It also has plant protection benefits, as the distance of the lower leaves from the ground determines the chance of pathogens infecting the host plant from the ground, in case the pathogens overwinter on the soil surface or fallen plant residues, such as BC or PV. This is because leaves closer to the soil surface are more exposed to infection pressure from pathogens overwintering on fallen plant parts because the number and size of raindrops decrease as they move away from the soil surface (Rossi and Caffi, 2012). Nevertheless, our results indicate that canopy height increases BC infection incidence (Figure 3, Table S2). The potential role of the canopy height on pathogen infection has already been examined in grapes. For example, higher canopies result in larger shades for themselves, which affects the cluster environment and significantly affects the exposure of the fruit to solar radiation, leading to an increase in the incidence of BC (Smart et al., 2017). Also, from the point of view of PV, a shaded canopy, where drying is more gradual, is favourable for the infection process. The survival of sporangia and the release and movement of zoospores are highly dependent on the relative humidity and the presence of water (Kast and Stark-Urnau, 1999). Thus, consistent with previous studies, our result suggests that canopy height could be a critical factor influencing plant health and the incidence of pathogens such as BC.

Cultivated crops of the same species grown over a large area favour large-scale multiplication of plant pathogens, as the pathogen will always find a suitable susceptible host plant in large populations (Wolfe, 2000). In addition, the presence of sufficient amounts of infective propagules in the plantation during the susceptible phenological phase of the host plant is a prerequisite for the pathogen's survival and infection (Coertze et al., 2001). Supporting the expectations, our study revealed that a vineyard surrounded by a higher density of grape plantations suffered more from the infection of BC and PV but not EN (Figure 2, Table 1, Table S2, Table S3, Table S4). The combined results of several studies suggest that, although pathogen sporulation varies during vegetation (Mundy et al., 2012), there are sufficient quantities of infective conidia available at the phenological stages of grapes susceptible to BC infection (Warren et al., 1999).

Traditional grape and wine production is based on regional yield and quality specifications and standards, highlighting its characteristics as a brand mark of a specific terroir. Our large-scale study found similar regional variations in the prevalence of primary grape pathogens. The prevalence of BC and PV infections showed a high spatial heterogeneity, while the effect of EN infections remained hidden (Table 1, Table S2, Table S3, Table S4). Systematic spatial differences in grape infections may suggest a possible role of large-scale background mechanisms and their variations related to the macroclimatic, topographic and soil conditions. However, revealing these effects was out of the present study's focus; we noted that testing the relevance of grape production to these effects could be a fruitful research objective in the future.

In many host-pathogen interactions, individuals' susceptibility may vary with time, as specific organs or whole plants may become more or less susceptible or have variable responses with age (Burdon et al., 1989). Age-related susceptibility of a particular plant part is well-known in epidemic diseases; however, the link between infection susceptibility and the actual plant age was underrepresented in the literature. In the present study, we assumed that the infection rate would be higher and symptoms more severe in older plantations than in newly established ones. This is due to the reduced resistance of an older plant because the accumulation of endophytic parasites (i.e., viruses, trunk diseases) can be increased with years (Kovács et al., 2017), generating higher biotic stress and resulting in susceptibility to epidemic diseases. Contrary to our expectations, the results showed that the plantation age did not affect the infection incidence in each tested pathogen (Table 1, Table S2, Table S3, Table S4). This can be explained by the fact that most of the plantations included in the study applied intensive crop production management. In a production environment, avoiding the physiological and condition deterioration of vines is one of the most crucial issues for keeping the quantity and quality standards of the cultivated crop. Therefore, most farmers carefully and regularly monitor physiological deterioration symptoms and continuously prevent or mitigate it with specific cultivation technology interventions. Thus, an older culture growing in an excellent cultivation environment could be more resistant to various plant pathogen infections, while these effects could be much more pronounced in the case of degraded growth environments. Moreover, the majority of the studied plantations were far below the age of senescence (75 % of the surveyed plantations were 30 years old or younger) based on the biological characteristics of the species, as a 40-year-old vine was considered to be old (Riffle et al., 2022). Furthermore, according to grapevine production approaches, when the applied interventions were unsupported to prevent vineyard deterioration, farmers often completely eradicated the entire plantation. Thus, in cultivated grapes, the role of growing conditions on production is more substantial than age.

Grape studies draw special attention to the importance of the vintage effect (Khan et al., 2020). However, our study approach was to explore long-term temporal patterns by asking volunteers for 5-year average infection rates. The yearly infection variance was smoothed out in the present study to long-term differences. It reflected a general average effect, which most describes the long-term cultivation success against pathogens. Since the collection of infection data also takes several years (3 consecutive years), the data for the 5-year average in the different years only partially overlapped in particular sampling areas. Due to the applied methodological sampling procedure, despite the prior annual correction of the infection data, we detected systemic changes in pathogen occurrences in PV (Table 1, Table S3), marginally insignificant in BC (Table 1, Table S2). At the same time, this did not occur at all in the case of EN (Table 1, Table S4).

2. Crop management

Pest control treatments and other production management tools promote reducing the production risk in terms of crop quantity and quality, but the application frequency could be under consideration due to their long-term severe economic and agroecological consequences (Liere et al., 2017). Therefore, we hypothesised that increased chemical crop protection could effectively reduce infection rates and symptoms. However, we also expected that applying more chemical treatments has no effect beyond a certain point, and therefore, some of the chemical treatments may be unnecessary. In the present study, we found a systematic change between the frequency of pesticide application and the infection occurrence only for BC, in which the BC infection increased with the level of chemical treatments (Table 1, Table S2).

The pathogen-specific association between pesticide applications and infection occurrence could be explained by two, not mutually exclusive, explanations. First, BC has the shortest incubation period of all the tested pathogens (Corio-Costet et al., 2010; Gadoury et al., 2012; Nair and Allen, 1993). Additionally, unexpected injuries caused by environmental (e.g., hail) or biotic (e.g., insects) factors can facilitate the rapid initiation of BC infection on berries, while PV and EN could infect intact berries. In case of these unwanted events, a farmer often prefers to insert an additional chemical treatment into the regular fungicide spray schedule to reduce the risk of BC infection (Romanazzi et al., 2016). Due to the short incubation period of BC, these sprayings bore low efficiency or remained ineffective. In this way, the number of chemical treatments increased with the incidence of BC. Second, chemical fungicide resistance could determine the current infection rate (Hahn, 2014). The desired rotation of fungicides is challenging in the EU, as there are fewer and fewer authorised active substances available for plant protection in BC than in PV or EN, which is the main source of resistance. Although increasing the number of fungicide treatments may have the short-term effect of reducing pathogens, it may be counterproductive as this crop protection approach promotes the adaption of pathogens, leading to resistance and loss of fungicide efficacy in a long-term effect (Hahn, 2014).

In contrast with BC, we did not find detectable associations in PV and EN pathogens, suggesting that the frequency of the chemical treatments may represent a lower relevance compared to the quality (or effectiveness) of the applied pesticides and/or other external environmental factors on a long-term basis (Table 1, Table S3, Table S4). However, the applied active fungicide agents and their potential effectiveness (i.e., asked about an average of 5 years of application practice) would be an important influencing factor; we could not survey them in a separate question in the present questionnaire study (space limitations). Besides, it seems the fungicide spray schedules protect the vines sufficiently against diseases caused by PV and EN. This information is rather lacking in the literature, which might be a relevant and focused topic of a systematic large-scale study in the future.

Our findings could not support the effect of inter-row management on the disease occurrence in BC, PV or EN (Table 1, Table S2, Table S3, Table S4). Therefore, our study suggests that inter-row effects have a minor impact on disease patterns in complex systems, compared to other factors, such as vineyard characteristics, crop production, and grape variety, when considering the long-term consequences. Contrarily, specific and short-term experimental studies demonstrated detectable direct or indirect effects of inter-row crops on disease prevalence and severity, which often interfered with other environmental impacts. Generally, inter-row crops could shift the microclimatic conditions and interfere with the soil properties (Guerra and Steenwerth, 2012). The inter-row management could enhance the biological activity of the soil by the application of cover crops and other organic mulch types, which could accelerate the decomposition rate of the infected plant debris (Jacometti et al., 2007a; Jacometti et al., 2007b). So, plant pathogens could be effectively inhibited indirectly via the beneficial shift of soil properties (e.g. increased nitrogen mineralisation rates, organic matter, microorganism populations) by appropriate inter-row management (Guerra and Steenwerth, 2012). For example, the decomposition of overwintering plant debris along with overwintering structures reduced the infestation rate of grapes due to the low survival rate of BC sclerotia as the organ of conidia production (Jacometti et al., 2007a). Comparing inter-row management types, plantations with inter-row coverage cut down the infection level of the flowers and clusters than vineyards without using that (Jacometti et al., 2007a).

Our study failed to detect the relationship between the leaf removal application and the long-term consequences of disease occurrence in BC, PV and EN (Table 1, Table S2, Table S3, Table S4). Thus, our results indicated that leaf removal might be less important than previously expected compared to a complex model. However, short-term studies revealed that the increased canopy mass was associated with a more humid microclimate in the cluster zone, which promoted the development of grape grey mould disease (Mundy and Beresford, 2007). So, removing the front leaves covering the clusters is a traditional and widely applied procedure to reduce the chance of BC infection by reducing the relative humidity (Romanazzi and Feliziani, 2014). In addition, the absence of foliage contributed significantly to the increased exposure of the berries to solar UV radiation, which resulted in increased berry resistance against the pathogen by a relatively thicker skin (Mundy et al., 2012). Furthermore, opening the foliage at the cluster zone facilitates faster wetness evaporation and ensures optimal pesticide coverage (Evers et al., 2010). The inconsistent results could be explained by the fact that the pathogen occurrence and climatic factors often show a high yearly fluctuation, so the effect of the leaf removal could enhance the pathogen suppression only in specific years; however, the positive effects of this practice might be equalised on a longer temporal scale.

3. Grape variety features

Exhausted breeding efforts always attempt to improve production effectiveness, such as yield and quality, including tolerance and resistance against specific plant pathogens. Therefore, carefully selecting the most appropriate varieties is always crucial because the varietal-linked characteristics should fulfil the requirements regarding the production system, environmental regimes and consumer needs. This study found pathogen-specific infection associated with some variety-specific features only in BC but not in PV and EN, affecting the current infection rate (Figure 4, Table 2, Table S5). Thus, a promising variety can provide a complex and general response to the current production challenges. However, our results highlighted that relying exclusively on the beneficial features of a variety could not give an ultimate solution against all types of fungal attacks.

Grapevine varieties show high diversity in their phenotypic appearance due to their different genotypic backgrounds. Thus, the high variation of morphological traits and chemical properties (i.e. chemical defence) could be the primary source of the high heterogeneity in pathogen susceptibility. Supporting our expectations, the literature-based susceptibility to pathogen infections increased with BC's observed infection ratio but not PV or EN (Figure 4, Table 2, Table S5). These results suggested that BC infection could be primarily based on variety-specific features. In contrast, environmental and crop management factors could influence the infection in the other two tested pathogens. Moreover, some varieties were selected for botrytisation (i.e., facilitating noble-rot by the increased susceptibility to BC infection), which could bias the general pattern regarding BC infection. Thus, the susceptibility to a specific pathogen was linked strongly to the varieties as a complex representation of their properties as a whole package. Based on the reduced database, the additional analysis found that the susceptibility properties reduced the effect on the observed disease incidence (Table S5). Such a shift between the models supported the previous expectations that varieties represented a shared package regarding morphological and chemical features which could influence the pathogen attack. In the following, we provided a detailed description regarding the possible role of the specific features.

The general appearance of a cluster structure (i.e., compactness, berry number and size) is an inherited, variety-specific phenotypic feature determined by the genetic background; however, some environmental and production management factors could slightly shape it (Dai et al., 2011). Studies linked the bunch structure and the susceptibility to plant pathogens. Generally, a more dense cluster appearance explained the increased susceptibility to pathogens due to reduced isolation distance among berries (Gabler et al., 2003). Moreover, grape pests (e.g., vine moth larvae) could also contribute to the attack of the wound-induced pathogens by their hidden feeding (Fermaud, 1998), against which the effectiveness of the pesticide treatments is limited for a more dense cluster. Our survey revealed that the likelihood of BC infection could not be varied among the tested varieties with different cluster structures (Table 2, Table S5). However, the noble-rot varieties played an important role in shaping this general pattern, making the relationship detectable after their inclusion into the analyses. These patterns suggested that the cluster compactness might not always confidently determine the BC attack on a long-term basis. Additionally, the specially selected noble-rot varieties showed a relatively higher infection ratio compared to varieties with similar characteristics; however, noble-rot varieties show a heterogeneous morphological profile (i.e., compactness) (Table S5).

The berry skin represents a primary host barrier, so its thickness could determine the level of the constitutive defence of a grape variety (Commenil et al., 1997). Thus, varieties with thicker skins and more epidermis cell layers showed a greater defence capacity against pathogens (Gabler et al., 2003). A similar pattern was found in the leaf penetration tested in different grape varieties by the infection of EN and PV (Eftimová and Bacigálová, 2012). Contrary to our expectation, varieties with thicker epidermis increased with a higher ratio of BC infection, even when varieties for botrytisation were excluded from the analysis (Table 2, Table S5). Similarly, previous results suggested that the relationship between the epidermis thickness and the degree of infection ratio could not always be trivial. A study (Kretschmer et al., 2007) found that Riesling berries with thicker epidermis were more easily infected than the thin-skinned Pinot Noir ones, which pattern was explained by the fact that the Riesling had mechanically softer tissues and became more likely to rupture after infection than Pinot noir. Their further analysis revealed that the chemical defence could compensate for the greater vulnerability to physical injuries, reducing the negative consequences of the infection, which level of chemical defence shifted with ripening.

The chemical defence system is the second barrier against a berry infection manifested in the berry tissue (Gabler et al., 2003). It is well known that the production of the plant secondary metabolisms involved in chemical defence strategies varies among varieties (Tříska et al., 2017), and the more susceptible varieties tend to have lower concentrations of pathogen inhibitors (e.g., resveratrol) (Adrian et al., 1997). In this study, we compared the infection ratios among grape varieties depending on the berry colour (used as an indicator), which reflect highly different chemical composition. We found that white or pale-coloured varieties suffered more from BC disease than those with higher pigment content (i.e., rouge, noir). However, BC infection may be desirable as noble-rot in some varieties (all blanc), but excluding these noble-rot varieties could not modify the revealed pattern (Table 2, Table S5). Therefore, the berry colour could be a suitable representation of the chemical defence determining the berry infection. Other studies highlighted that some specific secondary metabolisms shifted during the ripening, which could define and represent an effective and inducible barrier against fungal attack (Kretschmer et al., 2007). The effectiveness of the chemical defence could also vary among different tissues within a berry due to the pH variation (Manteau et al., 2003).

Previous studies illustrated that the infection risk could change consistently with time, closely linked to fruit ripening. Due to morphological, structural and chemical properties changing over time, the ripening time is a variety-linked characteristic that could vary considerably from year to year according to the ongoing climatic conditions. Consequently, the known infection windows for grapevine pathogens also vary since the phenological development of plants (fruit ripening processes) is also linked to specific climatic conditions, and therefore, the risk of infection varies with time. Our results showed an apparent and detectable increase in BC infection rates with ripening time (Figure 4, Table 2, Table S5). The obtained pattern was consistent with the climatic requirements of BC (lower daily average temperature, more frequent precipitation) and the coincidence of the phenological stages of grape varieties suitable for infection (i.e., from pea-sized berries to harvest) (Ciliberti et al., 2015). Moreover, the fungus can also cause latent infection in young berries, and the berries remain asymptomatic until ripening, then the fungus reactivates, causing severe symptoms (Keller et al., 2003). Furthermore, the temporal variation in morphological and chemical processes previously described is well reflected in the differences in ripening processes observed between varieties. Since environmental, climatic and cultivation practices strongly influence these processes, deviations from optimal growing conditions can significantly increase the chances of BC infection.