Plant material receptivity

Disease resistance in plants is genetically defined but, in some cases, is organ-specific (Hermanns et al., 2003; Strugala et al., 2015). For grapevine, cases with large differences in the level of resistance between organs (leaves and bunches) are quite rare and seem to apply to only a few accessions (Kennelly et al., 2005; Buonassisi et al., 2018). Leaves are the most rated organ for disease resistance, probably because they are numerous, available throughout the vegetative season, bidimensional, and can be scored for different variables. Resistance to DM and PM may also be assessed on berries, rachises, and canes in the vineyard (Riaz et al., 2011; Pap et al., 2016; Karn et al., 2021).

Environmental conditions may cause host stress-induced resistance and reduced DM and PM infections. For instance, temperatures below 8°C (Moyer et al., 2010), nutrient deficiency (Keller et al., 2003), and drought (Heyman et al., 2021) reduce the development and susceptibility of grapevine organs to diseases, delaying epidemics (Moyer et al., 2016). In addition, cultural strategies (soil tillage and fertilization) and stimulated grapevine growth increase both leaf and berry susceptibility to pathogens (Fernandes de Oliveira et al., 2021).

For grapevines growing under optimal conditions, resistance phenotypes are affected by the development stage of the organ, its physiology, and source-sink relationships. Young tissue shows higher susceptibility to DM and PM with the exception of very early leaves to P. viticola because of low stomata receptivity (Kennelly et al., 2007). Old grapevine tissue shows ontogenic resistance (Delp, 1954; Kennelly et al., 2005; Develey-Rivière and Galiana, 2007). For instance, leaves are less susceptible to E. necator after their physiological transition from sink to source status (Doster and Schnathorst, 1985; Calonnec et al., 2018). They are also characterized by an overexpression of pathogenesis-related (PR) protein genes, and ethylene and phenylpropanoid pathways (Calonnec et al., 2021). The ontogenic resistance process is promoted by stress and appears to be cultivar-dependent (Doster and Schnathorst, 1985; Barba et al., 2015; Calonnec et al., 2018, 2021). In particular, the gain in resistance is considered higher in partially resistant cultivars (Peros et al., 2006). Grape berries and rachises are receptive to DM until stomata convert to lenticels, 4–6 weeks post-bloom (Kennelly et al., 2005) and are receptive to PM until they reach 8% sugar content (Delp, 1954; Gadoury et al., 2003).

Evolution of pathogen populations and isolates in natural environment

Pathogen populations in the vineyard are characterized by some genetic diversity and undergo evolutionary processes with varied selection pressures. Plasmopara viticola and Erysiphe necator originated in North America where the genetic variability is high (Brewer and Milgroom, 2010; Rouxel et al., 2013). For instance, cryptic species of P. viticola are described on different Vitis spp. (Rouxel et al., 2013; Taylor et al., 2019) and specialized isolates of E. necator are detected on V. rotundifolia (Frenkel et al., 2010). In Europe, Australia, and other countries, pathogen diversity is lower due to the bottleneck at the time of their introduction (Brewer and Milgroom, 2010; Fontaine et al., 2013, 2021; Taylor et al., 2019). In Europe, two dominant E. necator isolates termed EU-A and B have been characterized (Délye et al., 1997; Brewer and Milgroom, 2010; Csikós et al., 2020). Group A populations are clonal while group B populations are sexual, but these two groups differ little in aggressiveness (Peros et al., 2006).

Distinct populations of P. viticola and E. necator may be involved in divergent resistance phenotyping outputs and mapping studies (Cadle-Davidson, 2008; Cadle-Davidson et al., 2016). Selective pressures, such as fungicide treatments (Frenkel et al., 2015; Delmas et al., 2017; Santos et al., 2020; Massi et al., 2021) and host resistance, are responsible for changes in population biology over time. In Europe, extensive asexual reproduction and mutation rates of P. viticola, despite little genetic differentiation, yield virulent isolates on Rpv3.1 (Peressotti et al., 2010; Venuti et al., 2013; Delmotte et al., 2014), Rpv10 (Heyman et al., 2021), and Rpv12 (Wingerter et al., 2021) grapevine genotypes. Similarly, phenotyping data in North America suggest the selection of natural virulent E. necator isolates on Ren2 (Cadle-Davidson et al., 2016) and Run1 (Feechan et al., 2015) grapevine genotypes.

Additionally, DM and PM inoculum pressure and distribution vary with geographic location, vineyard, and season in any given year (Montarry et al., 2008; Boso et al., 2019). Extremely low humidity and low or high temperatures are the main factors that negatively affect P. viticola and E. necator growth and sporulation (Caffi et al., 2013; Choudhury et al., 2014). Assessments for resistance are conducted one to three times annually. However, the end of the vegetative season, when the damage of DM and PM is usually more intense, is the preferred period for scoring resistance with dates changing depending on the year (Ramming et al., 2011; Riaz et al., 2011; Pap et al., 2016; Zyprian et al., 2016; Teh et al., 2017; Zendler et al., 2017; Karn et al., 2021).

Resistance traits can be measured by several variables and scales

Resistance phenotypic variables are scored by visually monitoring disease symptoms that are determined by the plant response to inoculation (necrosis) and/or pathogen growth. In the vineyard, P. viticola early infections cause yellow lesions on leaves that appear as translucent oil spots. On older leaves in late season, infections are limited and form yellow to reddish-brown spots in a mosaic-like patchwork. The sporulation of P. viticola is white, downy, and cottony, and develops on the abaxial leaf surface. The infection of E. necator displays on green tissues as whitish and powdery colonies that get thicker and larger with pathogen mycelium growth and sporulation. Numerous brown to black necrosis can be present during both DM and PM infections. Young berries infected by DM produce sporulation or shrivel completely in later infections, while in the case of PM infections, can break off during growth.

DM and PM symptoms in the vineyard are evaluated on one to three plants per genotype by determining the number of infected leaves (disease incidence) or by different visual scales (disease severity) (Table 3). The use of standardized descriptors determined by the Organization Internationale de la Vigne et du Vin (OIV; OIV, 2009) or the International Plant Genetic Resources Institute (IPGRI) represents a solid foundation for cross-comparison of results (Cadle-Davidson et al., 2019). The OIV codes 452 and 455 for leaf resistance, and 453 and 456 for berry resistance to P. viticola and E. necator, respectively, refer to a semi-quantitative scoring method. The scales comprise five classes, ranging from 1 for very susceptible phenotypes (with extended and dense pathogen sporulation) to 9 for totally resistant phenotypes (with only small necrosis or no symptoms) (OIV, 2009) (Table 3). Inverted scales from IPGRI positively correlated with the rate of infection are frequently considered to reduce the risk of possible confusion between the rate of disease severity and the degree of plant resistance (Pap et al., 2016; Teh et al., 2017; Zendler et al., 2017). This scale is also applied to various organs (Riaz et al., 2011; Pap et al., 2016; Karn et al., 2021) (Table 3).

Control of phenotyping sources of variation in the greenhouse

Phenotyping in the greenhouse usually relies on two to four replicates of each genotype obtained from hard-wood or soft-wood cuttings. Plants are maintained for 1–3 years in pots and are fertilized and pruned to maintain young growing shoots during the evaluation period. Plants in the greenhouse may be affected by only minor non-homogeneous conditions such as light exposure (Cadle-Davidson, 2008). As applied in the vineyard, grapevines are maintained in the greenhouse with or without a minimal fungicide protection.

Conditions for P. viticola infections in the greenhouse can be natural (Bhattarai et al., 2021) or induced (Merdinoglu et al., 2003). Artificial P. viticola inoculations are usually preferred and performed using a natural pathogen population from the vineyard (Moreira et al., 2011), or using a defined isolate maintained on controlled potted plants (Merdinoglu et al., 2003; Bellin et al., 2009; Vezzulli et al., 2019). Sporangia are sprayed by a water solution on receptive leaves of potted plants (Merdinoglu et al., 2003; Bellin et al., 2009) or on whole plants (Vezzulli et al., 2019; Bhattarai et al., 2021). DM is evaluated once annually between 7 (Merdinoglu et al., 2003) and 15 days post-inoculation (dpi) (Moreira et al., 2011) (Table 5).

In the greenhouse, E. necator encounters optimal growth conditions and may develop natural infections (De Nardi et al., 2019). However, to encourage PM and its homogeneous distribution, naturally or artificially infected V. vinifera plants are often placed among the population of grapevines to be phenotyped (Hoffmann et al., 2008; Ramming et al., 2011; Mahanil et al., 2012; van Heerden et al., 2014; Karn et al., 2021). Then, the epidemic develops over several months and plants are evaluated two or three times (Hoffmann et al., 2008; Ramming et al., 2011; van Heerden et al., 2014; Karn et al., 2021). Alternatively, the pathogen may be inoculated by spraying and resistance is evaluated after a few weeks (Pauquet et al., 2001; Pap et al., 2016).

In the greenhouse, resistance is scored according to the OIV descriptors as done in the vineyard (Merdinoglu et al., 2003; Hoffmann et al., 2008; van Heerden et al., 2014; Vezzulli et al., 2019) or by using other similar semi-quantitative scales (Moreira et al., 2011; Ramming et al., 2011; Bhattarai et al., 2021) (Table 5).

Plant material management

For the bioassays performed in the laboratory, young leaves are collected from plants growing in optimal conditions either in the greenhouse (Blanc et al., 2012; Vezzulli et al., 2019; Possamai et al., 2021) or vineyard (Bellin et al., 2009; Ochssner et al., 2016; Divilov et al., 2017; Sapkota et al., 2019). Leaves are usually collected according to their position along the shoot to avoid ontogenic resistance (Merry et al., 2013; Calonnec et al., 2018, 2021) (Table 7). The maximal susceptibility to PM infection is conserved up to the fourth to fifth unfolded leaf on cuttings (Merry et al., 2013) and up to the second to third leaf on shoots in the vineyard (Barba et al., 2015; Calonnec et al., 2018). Some authors suggested that shine is also an important factor to identify susceptible leaves (Sapkota et al., 2019), particularly for PM resistance evaluations (Feechan et al., 2011; Cadle-Davidson et al., 2016; Possamai et al., 2021).

At the beginning of phenotyping experiments, particularly for E. necator resistance evaluations, a well-established practice is leaf disinfection and rinsing (Ramming et al., 2011; van Heerden et al., 2014; Pap et al., 2016; Divilov et al., 2017; Vezzulli et al., 2018; Sapkota et al., 2019; Bhattarai et al., 2021; Karn et al., 2021; Possamai et al., 2021; Zendler et al., 2021a). This avoids contamination when plant material originates in the vineyard and for long-lasting experiments.

Between 8 and 16 leaf disks are usually excised for DM resistance evaluations (Bellin et al., 2009; Divilov et al., 2017; Vezzulli et al., 2019), while for PM resistance evaluations, up to eight whole leaves (Ramming et al., 2011; Blanc et al., 2012; Pap et al., 2016) or leaf disks are assessed (Pauquet et al., 2001; Karn et al., 2021; Possamai et al., 2021; Zendler et al., 2021a).

Inoculum and infection management

P. viticola and E. necator inocula can either be collected as a natural population in the vineyard (Ramming et al., 2011; Schwander et al., 2012; Lin et al., 2019; Fu et al., 2020), or produced under controlled conditions on potted plants, for P. viticola (Bellin et al., 2009; Vezzulli et al., 2019), or on in vitro leaves for both pathogens (Blanc et al., 2012; Pap et al., 2016; Divilov et al., 2017; Sapkota et al., 2019; Possamai et al., 2021). Controlled inocula may originate as a mixture of isolates (Hoffmann et al., 2008; Bellin et al., 2009; Zyprian et al., 2016), or monosporangia/monoconidial cultures (Bellin et al., 2009; Teh et al., 2017; Possamai et al., 2021) (Table 7). In mapping studies, isolates are usually characterized by the pathogen origin (Bellin et al., 2009; Barba et al., 2015; Sargolzaei et al., 2020) and a few defined isolates (Bellin et al., 2009; Venuti et al., 2013). Leaves with P. viticola sporulation, or sporulated lesions, can be conserved by freezing (Delmotte et al., 2014), but the aggressiveness of the pathogen inoculum requires an infection cycle on leaves before resistance phenotyping. In contrast, E. necator isolates need to be maintained on green tissue (Miazzi et al., 1997; Gao et al., 2016).

For DM leaf disc bioassays, the pathogen inoculum is prepared by soaking leaves showing fresh P. viticola sporulation in water (Bellin et al., 2009; Divilov et al., 2017). Then, inoculations are performed by spraying the solution on the tissues (Bellin et al., 2009; Sapkota et al., 2019; Vezzulli et al., 2019; Sargolzaei et al., 2020; Bhattarai et al., 2021) or by the application of solution drops (10–50 μl) on the abaxial surface of the leaves (Schwander et al., 2012; Divilov et al., 2017; Lin et al., 2019; Fu et al., 2020). Experiments used a concentration of the spore solution ranging from 2 × 10⁴ to 5 × 10⁵ sporangia per ml (Table 7). Plasmopara viticola droplets remain on leaves for up to 24 h and need to be removed to avoid tissue rotting (van Heerden et al., 2014; Zyprian et al., 2016).

Erysiphe necator is inoculated on in vitro leaves by spraying conidia suspensions (Ramming et al., 2011; Cadle-Davidson et al., 2016; Karn et al., 2021; Zendler et al., 2021a) or using settling towers (Pauquet et al., 2001; Blanc et al., 2012; Pap et al., 2016; Possamai et al., 2021). Experiments used PM pressure ranging from 5 × 10⁴ to 2 × 10⁵ conidia per ml for spraying, and from 2 to 8 conidia per mm² for settling towers (Table 7).

Petri dishes containing leaves or leaf disks are usually inoculated at the sampling date and incubated according to P. viticola and E. necator optimal growth and sporulation conditions between 20 and 25°C and with high relative humidity. Photoperiods vary between 12 and 19-h light (Schwander et al., 2012; Bhattarai et al., 2021) (Table 7). P. viticola infection is usually evaluated between 5 and 7 dpi (Bellin et al., 2009; Schwander et al., 2012; Lin et al., 2019; Fu et al., 2020; Bhattarai et al., 2021), while E. necator infection is assessed between 3 and 14 dpi (Blanc et al., 2012; Pap et al., 2016; Possamai et al., 2021).

Assessment of resistance variables from in vitro bioassays

In the laboratory, resistance rating methods are very diverse (Tables 6, ​,7).7). As commonly practiced for the other environments, DM and PM sporulation is scored for incidence (Pauquet et al., 2001; Bellin et al., 2009; Blasi et al., 2011; Vezzulli et al., 2019) and severity using different visual semi-quantitative scales (Divilov et al., 2017; Sapkota et al., 2019; Vezzulli et al., 2019), such as the OIV 452-1 and 455-1 descriptors (OIV, 2009). In vitro bioassays facilitate the use of machine vision analyses (Divilov et al., 2017; Fu et al., 2020). Phenotyping in the laboratory is often enriched by the collection of additional variables obtained by stereomicroscopes and microscopes. Some examples are: DM density of sporangiophores (Bellin et al., 2009; Zyprian et al., 2016; Fu et al., 2020), DM mesophyll colonization (Bellin et al., 2009), PM mycelium growth (Blanc et al., 2012; Possamai et al., 2021), dimension of DM sporangia (Zyprian et al., 2016), count of PM hyphal transects (number of hyphae crossed by a line traversing the leaf disc; Cadle-Davidson et al., 2016; Karn et al., 2021), and spore production (Zyprian et al., 2016; Possamai et al., 2021) (Tables 6, ​,7).7). In vitro phenotyping is also often enriched by the calculation of infection indexes (Townsend and Heuberger, 1943; Lin et al., 2019; Sargolzaei et al., 2020).

Plant necrosis is assessed either qualitatively (Bellin et al., 2009; Zyprian et al., 2016; Possamai et al., 2021) or quantitatively for its frequency and extension (Blasi et al., 2011; Zyprian et al., 2016; Divilov et al., 2017; Zendler et al., 2021a), in particular for P. viticola infections (Tables 6, ​,77).

Time series ratings are frequently recorded in the laboratory (Bellin et al., 2009; Venuti et al., 2013; Divilov et al., 2017; Sapkota et al., 2019; Vezzulli et al., 2019; Possamai et al., 2021) (Table 6), allowing several comparisons among phenotyping data and providing the opportunity to obtain new variables that consider the complete infection process, such as the Area-Under-Disease-Pressure-Curve (AUDPC; Jeger and Viljanen-Rollinson, 2001).

Environments, plant material, and pathogen inocula

Phenotyping in the vineyard yield resistance data at the whole plant level during natural polycyclic infections and under unstable epidemic pressure. Some studies suggested that the leaves may be a proxy of whole plant resistance (Welter et al., 2007; Ramming et al., 2011; Pap et al., 2016; Zyprian et al., 2016; Karn et al., 2021). However, other studies suggested limited correlation between resistance of grapevine leaves and bunches (Calonnec et al., 2013; Buonassisi et al., 2018; Vezzulli et al., 2018). This may be due to either the presence of factors that influence the susceptibility of the organs or a possible low reliability of some resistance data. For these reasons, resistance should be observed on different organs, at least leaves and bunches. Furthermore, to standardize phenotyping results, the time of plant organ scoring should be defined according to the grapevine phenological phases (Lorenz et al., 1995; OIV, 2009). For disease dynamic, epidemic pressures could be estimated by assessing control plants (Riaz et al., 2011; Ochssner et al., 2016; Zyprian et al., 2016), or by capturing and quantifying pathogen airborne spores (Mahaffee and Stoll, 2016; Thiessen et al., 2018; Brischetto et al., 2020). Finally, pedoclimatic data should be recorded to better understand the resistance phenotyping results and QTL analyses (Zyprian et al., 2016; Zendler et al., 2017, 2021a).

Phenotyping under natural conditions of infection in the greenhouse has similar characteristics to phenotyping in the vineyard but with more consistent environmental conditions. Plants produced in the greenhouse for subsequent greenhouse or in vitro experiments allow a better follow-up of plant growth and juvenility of the leaves (Welter et al., 2007; Zyprian et al., 2016). The growing degree-day and/or the leaf position on the shoots of the leaf collected can be helpful to record the stage of the plant material used in phenotyping. Pathogen inoculations could be also standardized using artificial inoculation of isolates maintained under control conditions. However, the management of plant growth and infection in the greenhouse may result in expensive and complex practices in terms of space and time, limiting bioassay replicates. Probably for these reasons, only a limited number of experiments are carried out in the greenhouse.

In vitro bioassays do not affect the original plants and can be performed at different periods and in multiple replicates with several experimental units during the vegetative season. To date, pathogen isolates used in mapping studies have been poorly described. Therefore, it would be advantageous to standardize pathogen inocula in vitro using a set of isolates characterized for their virulence and aggressiveness. In vitro evaluations are effective in describing the biological and genetic resistance components during one pathogen cycle, but the bioassays do not consider possible derivative effects of the polycyclic behavior of the disease (Bove et al., 2019). Furthermore, plants are tested in extremely favorable conditions for the diseases, leading to a possible underestimation of the resistance in comparison with the vineyard (Calonnec et al., 2013). Therefore, the extreme diverse experimental conditions may explain the limited correlation between resistance sometimes observed on detached leaves or leaf disks and bunches in the vineyard (Calonnec et al., 2013; Zyprian et al., 2016; Buonassisi et al., 2018; Vezzulli et al., 2018; Bove et al., 2019). For all these reasons, resistant plants with new identified QTL should be assessed also under vineyard conditions.

Regardless of the phenotyping environment, the use of a range of control genotypes (with known resistance loci correlated to various resistance levels) growing and rated in the same environment and experimental design as the candidate plants to be assessed, would validate the phenotyping practices (plant material growth and sampling, pathogen inoculum preparation, and distribution and infection environmental conditions). Finally, due to the possible pathogen genetic variability, the identified resistance QTL should be checked in multi-year and multi-time evaluations, and in different vineyard areas before their introduction into breeding programs (Bellin et al., 2009; Teh et al., 2017; Karn et al., 2021; Possamai et al., 2021).

Resistance variables

The different variables used to score P. viticola and E. necator symptoms show significant correlations between them and usually locate the main QTL at the same genomic positions (Bellin et al., 2009; Cadle-Davidson et al., 2016; Pap et al., 2016; Zyprian et al., 2016; Karn et al., 2021; Possamai et al., 2021). Sporulating P. viticola and E. necator is primarily utilized to characterize grapevine resistance because it is macroscopic and appraisable in all phenotyping environments (Bellin et al., 2009; Venuti et al., 2013; Pap et al., 2016; Zyprian et al., 2016; Teh et al., 2017; Vezzulli et al., 2018; Fu et al., 2020; Sargolzaei et al., 2020; Özer et al., 2021). However, different studies suggest that necrosis or cell death (Pitsili et al., 2020) should be also assessed because the plant response may be differentially expressed according to plant–pathogen interactions (Zyprian et al., 2016; Gómez-Zeledón et al., 2017; Divilov et al., 2018; Possamai et al., 2020).

The official visual semi-quantitative scales, such as those of OIV or IPGRI, are considered subjective and error-prone (Poland and Nelson, 2011). However, annual training of observers in visual estimation of disease symptoms could reduce possible biases. To further reduce the possibility and effects of human errors and to improve the biological description of resistance and QTL identification, it is advisable that resistance assessments should be performed by several people, on multiple biological replicates and in multiple bioassays. The disease severity rating method may be an alternative to the previous scales, but the assessment of the percentage of infected surface may also be subjective. The disease incidence is probably the easiest method to assess resistance but fails to record complete information about quantitative phenotypes.

Tools and strategies for improving grapevine resistance phenotyping objectivity and automation

In resistance mapping studies, despite the numerous experiments and large-scale samples, phenotyping is generally conducted by traditional visual rating of disease symptoms with some exceptions (Pap et al., 2016; Divilov et al., 2017, 2018; Fu et al., 2020). Other studies showed that metabolomes such as stilbenoids in grapevines (Malacarne et al., 2011; Chitarrini et al., 2017) and lipids in pathogens (Negrel et al., 2018) can be studied to characterize DM and PM infections. Alternatively, grapevine pathogens can also be detected and quantified by molecular techniques such as qPCR and transcriptional analyses (Valsesia et al., 2005; Dufour et al., 2011; Malacarne et al., 2011; Hariharan and Prasannath, 2021). Different metabolic responses to P. viticola have already been associated with resistance loci (Eisenmann et al., 2019; Chitarrini et al., 2020; Ciubotaru et al., 2021) and could provide guidelines to define new resistance variables. However, these methods do not allow monitoring of disease progression.

Assessment of grapevine tissue, using spectrometers and spectral imaging sensors, relies on non-destructive measurements capable of early detection of pathogen infection, tracking of infection progression, and differentiation of responses mediated by resistance loci (Cséfalvay et al., 2009; Peressotti et al., 2011; Oerke et al., 2016; Štambuk et al., 2021). Nowadays, machine vision approaches can be highly automated in image capture, and their analysis can use advanced recognition algorithms (Hamuda et al., 2017) and convolutional neural networks (CNNs) (Rawat and Wang, 2017). These elements increase the machine vision throughput and performance that rival human observers in grapevine disease detection and classification (Bierman et al., 2019; Liu et al., 2020; Zendler et al., 2021b). Disease severity assessed by machine vision appears to be time and cost-effective (Divilov et al., 2017), but presents specific challenges, such as droplet dispersion (Fu et al., 2020), leaf morphology (Divilov et al., 2017; Fu et al., 2020), and low pathogen structure (E. necator hyphae) resolution (Bierman et al., 2019). Machine vision probably represents one of the most interesting opportunities to produce objective and quantitative measures of grapevine resistance.

The authors thank Daniele Migliaro (CREA-VE), Komlan Avia (INRAE Grand Est-Colmar), and Marc Fuchs (Cornell University) for their helpful suggestions and discussions in the final review of the manuscript.