Components of partial resistance to Plasmopara viticola enable complete phenotypic characterization of grapevine varieties

Six components of partial resistance (RCs) were studied in 15 grapevine varieties with partial resistance to Plasmopara viticola: (i) infection frequency (IFR, proportion of inoculation sites showing sporulation), (ii) latent period (LP50, degree-days between inoculation and appearance of 50% of the final number of sporulating lesions), (iii) lesion size (LS, area of single lesions in mm2), (iv) production of sporangia (SPOR, number of sporangia produced per lesion, and SPOR’, number of sporangia produced per mm2 of lesion), (v) infectious period (IP, number of sporulation events on a lesion), and (vi) infectivity of sporangia (INF, infection efficiency of sporangia produced on resistant varieties). Artificial inoculation monocycle experiments were conducted for a 3-year period on leaves collected at leaf development, flowering, and fruit development. Compared to the susceptible variety ‘Merlot’, the partially resistant varieties showed reduced IFR, longer LP, smaller LS, fewer SPOR and SPOR’, shorter IP, and lower INF. At leaf development, IFR, SPOR, and INF were higher and LP was shorter than at flowering and fruit development. RCs analysis through monocyclic experiments provides reliable assessments of the resistance response of grapevine accessions. The workload required for routine assessment in breeding programs could be reduced by measuring IFR and SPOR, while producing robust results.

sporulation 25 . These processes are called "components of partial resistance" and regulate the epidemic that can result from a chain of infection cycles during the host-growing season. In partial resistance, the phenotype is the result of different phases of the infection process in which resistance responses occur, and each phase is considered a component of resistance that contributes to the overall response 26 . The most studied components of resistance include reduced infection efficiency, a longer latent period, a shorter infectious period, reduced sporulation, and a smaller lesion size [27][28][29][30] . Resistance components have been analyzed 31 for a number of pathosystems, including Cercospora leaf spot of sugar beet 29,32 , yellow rust (Puccinia striiformis f. sp. hordei) of barley 33 , rice blast (Pyricularia oryzae) 34 , rice sheath blight (Rhizoctonia solani) 35 , leaf rust (Puccinia triticina) of wheat 36 , Fusarium head blight of wheat 37 , and Phoma black stem of sunflower 38 . However, such analysis has not been conducted on the DM-grapevine pathosystem.
From an epidemiological point of view, components of partial resistance all help to reduce the apparent infection rate (r) of the epidemic, thus slowing disease progress 25,[39][40][41][42] . The effects of partial resistance components (RCs) on the epidemic development have been investigated for groundnut rust 43 , potato late blight 44 , wheat stripe rust 45 , and sugar beet Cercospora leaf spot 32 .
The objective of this work was to measure the RCs to P. viticola in 15 Vitis accessions, most of them carry one or more Rpvs (www.vivc.de), or showing moderate susceptibility to downy mildew. For the sake of simplicity, these 15 accessions are referred to as "partially resistant varieties" in this paper. The following phenotypic traits were assessed in monocycle experiments conducted under environmentally controlled conditions: (i) degree of resistance, based on the OIV 452-1 descriptor 46 ; (ii) infection frequency; (iii) duration of the latent period; (iv) lesion size; (v) number of sporangia produced per lesion and per lesion surface unit, (vi) duration of the infectious period; and (vii) infectivity of the sporangia produced on DM lesions.

Results
Degree of resistance in the tested varieties. A factorial analysis of variance (ANOVA) for the degree of resistance to P. viticola according to the OIV 452-1 descriptor 46 showed significant effects of year, growth stage, variety, and their interactions (P < 0.001). Year and growth stage accounted for 3.7% and 3.5% of total variance, respectively, while variety accounted for 63% (Supplementary Table S1). The susceptible control 'Merlot' scored 1.4 in resistance to DM (very low resistance). Among the tested varieties, the lowest OIV 452-1 value was 1.6 (for 'Rkatsitelii'), and the highest values were 6.9 (for 'Bronner') and 7.9 (for 'Johanniter') ( Fig. 1).

Resistance components in 'Merlot' and in the partially resistant varieties.
ANOVAs showed a significant effect (P < 0.001) of variety for all RCs (Table 1). Variety and its interactions accounted for 67% to 89% of the total variance, depending on the RC (Supplementary Table S1). Plant growth stage at the time of RC measurement also had a significant effect (P ≤ 0.01) on RCs; the interaction variety × growth stage was significant for some RCs but not for others, and accounted for only a small percentage of the total variance (Supplementary  Table S1). Our analysis now focuses on the main effects of variety (Table 1) and growth stage ( Table 2).
The overall average values for the infection frequency (IFR) for 'Rkatsitelii' , 'Calandro' , 'Cabernet Volos' , 'Villaris' , and 'Regent' were not significantly different from that of the susceptible control 'Merlot' , which had an average IFR value of 0.95 (Table 1). IFR values were lower for 'Reberger' , 'Felicia' , 'Fleurtai' , 'Palava' , 'Merlot Khorus' , 'Solaris' , 'Calardis Blanc' , 'Merlot Kanthus' , 'Bronner' , and 'Johanniter' than for 'Merlot' , with a reduction relative to 'Merlot' ranging from 6% for 'Reberger' to 43% for 'Bronner' and 'Johanniter' ( Table 1). The average IFR Figure 1. Resistance to Plasmopara viticola assessed on leaf discs of 16 grapevine varieties (X axis) after artificial inoculation. The degree of resistance is expressed according to the descriptor OIV 452-1 46 : 1 = very low degree of resistance; 3 = low degree of resistance; 5 = medium degree of resistance; 7 = high degree of resistance; and 9 = very high degree of resistance. Values are means of data collected in a 3-year period (from 2014 to 2016) and at three growth stage, with 75 replicate leaf discs for each variety in each experiment + SE. Values with different letters are significantly different according to the LSD test (P = 0.05).
Area Under Infection Progress Curve (AUIPC) for 'Cabernet Volos' and 'Rkatsitelii' were significantly lower and higher, respectively, than the value for the susceptible control 'Merlot' (Table 1). AUIPC values for all of the other varieties were lower than the AUIPC for 'Merlot' , with reductions relative to 'Merlot' ranging from 4% to 50%. AUIPC values were lowest for 'Bronner' and 'Johanniter' . As was the case for IFR, the average AUIPC was significantly higher at stage 18 than at stages 65 or 79 ( Table 2).
The duration of the latent period (LP50) did not significantly differ between 'Villaris' and the susceptible control 'Merlot' (Table 1). Relative to 'Merlot' and 'Villaris' , the LP50 was shorter for 'Rkatsitelii' but was longer for all other varieties. The LP50 was longest for 'Johanniter' and 'Bronner' , with an increase of 20% and 17%, respectively, relative to 'Merlot' . The LP50 was significantly shorter at growth stage 18 than at the other two growth stages, which were not significantly different each other (Table 2).  Table 1. Values of the components of partial resistance (RCs) to Plasmopara viticola measured for the grapevine varieties described in Table 5.   Table 2. Values of the components of partial resistance (RCs) to Plasmopara viticola measured at three growth stages of grapevines, i.e. leaf development, flowering, and fruit development, corresponding to growth stages 18, 65, and 79, respectively 73  The lesion size (LS) was smaller for all the partially resistant varieties than for 'Merlot' (Table 1), with reductions ranging from 23% to >90% (93% for 'Calandro' and 96% for 'Johanniter').
The number of sporangia produced per lesion (SPOR) on 'Reberger' was not significantly different from the number on 'Merlot' (Table 1). In the partially resistant varieties, the SPOR was reduced by 44% to 99% compared to 'Merlot' . SPOR was lowest on 'Bronner' , 'Merlot Kanthus' , and 'Fleurtai' . Like SPOR' values, SPOR values were higher at stage 18 than at the other two stages ( Table 2).
For most varieties, the infectivity of sporangia produced on the DM lesions (INF) did not significantly differ from that of the sporangia produced on 'Merlot' (Table 1). INF, however, was significantly higher for 'Calandro' (by 22%) and significantly lower for 'Bronner' (by 19%). INF was lower at stage 65 than at the other two stages (Table 2).
Overall, the variability (as indicated by the coefficient of variability, CV) among grapevine varieties was high for SPOR' , LS, and SPOR but was low for LP50 (Table 1). (Table 3), there was a significant (P < 0.05) relationship between IFR (and AUIPC as a consequence), LP50, IP, and INF, indicating that as IFR decreased, LP50 increased and IP and INF decreased. LS was significantly (P = 0.032) correlated with LP50, so that LS at 11 dpi increased as the time for the lesions to appear shortened. SPOR' , in contrast, was not correlated with any of the other RCs, with the exception for SPOR.

Relationships among resistance components. As indicated by the correlation matrix
Cluster analysis. The hierarchical cluster analysis grouped the varieties at different rescaled distances ( Fig. 2); at the intermediate distance, four clusters were identified. Based on the OIV degree of resistance (standard OIV 452-1), these clusters represent "high resistance" (CLU1, with an average degree of resistance OIV = 7.0), "medium resistance" (CLU2, OIV = 5.4), "low resistance" (CLU3, OIV = 3.3), and "very low resistance" (CLU4, OIV = 1.5); no varieties showed a "very high" degree of resistance (i.e., with OIV = 9). The average values for each RC of each cluster are shown in Table 4. The among-cluster ANOVA revealed significant differences (P < 0.002) for IFR, LP50, SPOR, and INF, but not for IP (P = 0.07). CLU1 included 'Bronner' , 'Johanniter' , and 'Merlot Kanthus'; CLU1 had the lowest IFR value, the longest LP50, the lowest SPOR, the shortest IP, and the smallest INF. In contrast, CLU4, which included the susceptible control 'Merlot' and 'Rkatsitelii' , had the highest IFR, the  Table 3. Correlation coefficients (r values) among the components of partial resistance (RCs) to Plasmopara viticola measured for the grapevine varieties described in Table 5 CLU4 contained two varieties without Rpvs, while CLU3 contained varieties without Rpvs or with Rpv3. Rpv3 was also present, alone or pyramided with other Rpvs, in two varieties of CLU2 and two of CLU1. Similarly, Rpv10 was present in two varieties belonging to CLU1 and to CLU2, respectively. The three varieties with Rpv12 were all in CLU2. According to ANOVAs, the RCs IFR, LP50, IP, and INF were not significantly affected by the presence of a particular Rpv in the genome (P > 0.05, Supplementary Table S2). On the contrary, SPOR was significantly higher in the susceptible control 'Merlot' than in the varieties with no Rpvs and in the partially resistant varieties carrying Rpv3, Rpv10, or Rpv12; SPOR did not significantly differ among partially resistant varieties carrying Rpv3, Rpv10, or Rpv12 (Fig. 3).

Figure 2.
Dendrogram resulting from a hierarchical cluster analysis of the five components of partial resistance (RCs) to Plasmopara viticola measured in the grapevine varieties described in Table 5. The RCs are infection frequency (IFR), latency period (LP50), sporulation (SPOR), infectious period (IP), and the infectivity of the sporangia produced on the variety and inoculated on Merlot (INF). Four clusters were identified in the dendrogram (CLU1 to CLU4).  Table 4. Values of five components of partial resistance (RCs) to Plasmopara viticola for the four clusters in OIV is the OIV descriptor 452-1 for the degree of resistance, which ranges from 1 to 9, corresponding to very low to very high resistance, respectively. P-values indicate the significance of the effect of cluster in the ANOVAs. Means followed by the same letter in the same column are not significantly different according to Fisher-protected LSD test (P = 0.05).

Discussion
The purpose of the present research was to characterize the partial resistance to downy mildew (DM) in 15 grapevine varieties, most of which carried one Rpv from V. rupestris or V. amurensis 14,[47][48][49] . Resistance was assessed relative to a susceptible control, 'Merlot' . The study was conducted by inoculating leaf discs with P. viticola, which is a widely used bioassay in the study of genetic resistance to DM in grapevines 7,9,19,[50][51][52] . Inoculations were performed by using a suspension of P. viticola sporangia derived from sporangia collected in multiple vineyards and then maintained on the susceptible variety 'Merlot' for the duration of the study. A population of P. viticola was used instead of single isolates in order to minimize the effect of host-strain genotype interaction on the phenotypic evaluation of the resistance components [53][54][55] .
The study assessed six main resistance components (RCs) that cover the entire disease monocycle and that do not overlap in terms of processes 26,31 . RCs were expressed according to their proper dimensions (e.g., frequency for IFR, thermal time for LP50, or numbers of sporangia for SPOR).
IFR was measured as the proportion of inoculated sites that produced DM lesions in the leaf discs assay. IFR is a proxy of infection efficiency (IE), which has been measured as the number of DM lesions generated by each zoospore deposited on the leaf surface 56 . IE was not used in the current study because its measurement is too complex 56 given the number of experiments that were performed. In addition, IE is biased because the number of zoospores produced per sporangium is variable 56 , multiple germ tubes (from multiple zoospores) can penetrate the same stoma 57 , and a single lesion may result from multiple penetrations 23 . Reduction of infection efficiency (measured as IFR in this work) reflects the ability of the plant to reduce the number of pathogen propagules that successfully go through the different steps of the infection process, e.g., the ability of the plant to prevent germ tube penetration of stomata, the formation of substomatal vesicles, the growth of primary and secondary hyphae in the mesophyll, or the growth and functioning of haustoria. IFR can involve both pre-and post-infection mechanisms, such as callose deposition in stomata [58][59][60] , the presence of inner cuticular rims 22 , and a HR 14,60,61 .
The duration of the latent period indicates the time required for the pathogen to produce sporangia on the lesion surface. A longer latent period may result from reduced hyphal growth in the mesophyll of the host plant due to the thickness of the spongy mesophyll and cell walls 4,22,62 . A longer latent period may also result from a reduced number or functioning of haustoria resulting from HR 14 , such that the flow of nutrients from host cells to the intercellular mycelium is reduced. AUIPC was also calculated to account for the observation that DM lesions appeared later in partially resistant varieties than in the susceptible control. AUIPC is therefore related to the duration of the incubation period, which is the time required for lesions to appear. The incubation period is not a resistance component sensu Zadoks 31 or Parlevliet 40 , because it does not affect the epidemic development and it is part of the latent period.
The other RCs (i.e., the number of the sporangia produced per DM lesion and per area of DM lesion, the duration of the infectious period, and the infectivity of the sporangia produced on DM lesions) may also be related to a reduced flow of nutrients from the host to the pathogen. For instance, reduced sporangiophore density and reduced numbers of sporangia produced per unit of leaf area were observed in 'Bianca' as a consequence of an HR 14 . The size of DM lesions also reflects the ability of the host to limit the growth of the pathogen in the mesophyll. The synthesis of resveratrol and flavonoids, the increased peroxidase activity, and the formation of lignin in the host tissue all restrict the development of P. viticola and result in smaller lesions 63 .
INF was not previously considered in analyses of resistance components 31 , but reduced INF may slow the progress of DM epidemics in partially resistant varieties. The infectivity of P. viticola sporangia produced on DM lesions on leaves of partially resistant varieties has been poorly studied. However, finding similar to those for DM in the current study were described for leaf rust (Puccinia triticinia) on wheat 64 . Delmotte et al. 55 observed that cross-inoculation with the sporangia produced on the partially resistant variety 'Regent' on the susceptible  Table 5. Varieties were assigned to the following five groups: susceptible control 'Merlot' (test), resistant varieties with no Rpv (none), or with Rpv3, Rpv10 or Rpv12. Values are means of data collected in a 3-year period and at three growth stage of grapevines, with 75 replicate leaf discs for each variety in each experiment + SE. Values with different letters are significantly different according to the LSD test (P = 0.05). (2020) 10:585 | https://doi.org/10.1038/s41598-020-57482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ cultivar 'Cabernet Sauvignon' resulted in higher disease severity, higher sporangia production, and smaller sporangia than inoculation with sporangia produced on the susceptible variety. This seems in contrast with the findings of the current study, in which the sporangia produced on 'Regent' had lower infectivity on the susceptible 'Merlot' than the sporangia produced on 'Merlot' . These contrasting results might be explained by differences in methods or in the genotypes of both the host plant and the pathogen.
RCs were expressed to different degrees among the 15 varieties, with significant differences relative to the susceptible 'Merlot' . High among-variety variability was found for SPOR' , LS, and SPOR (i.e., the combination of LS and SPOR'), and low among-variety variability was found for IP. As expected, lower infection frequency, longer latent period, reduced sporulation, shorter infectious period, and lower infectivity of sporangia were found in the partially resistant varieties than in the susceptible 'Merlot' . Changes in some of these RCs were significantly correlated, so that varieties with low IFR values also had long LP50 values and low IP and INF values (Table 3). Similar results were found for yellow rust (Puccinia striiformis f. sp. hordei) on barley 33 , leaf rust (Puccinia triticina) on wheat 36 , and Phoma black stem of sunflower 38 , indicating that the mechanisms of resistance act on different stages of the infection process, as previously discussed. Surprisingly, sporulation was not correlated with the other RCs, and this requires further investigation.
Expression of resistance components was generally highest in the varieties carrying Rpv3, Rpv10 or Rpv12, which all grouped in the clusters of high and medium resistance (based on the OIV standards), i.e. CLU1 and CLU2, respectively; the exceptions were 'Felicia' , 'Calandro' , and 'Regent' , which were in the cluster with low resistance (i.e. CLU3), even though they carry Rpv3 (Fig. 2; Table 4). Unfortunately, no association was found between RCs and the Rpvs considered in this work, which would been helpful in breeding programs. On the one hand, the highest level of resistance was found in 'Bronner' , 'Merlot Kanthus' and 'Johanniter' (belonging to the same CLU1). 'Bronner' (carrying Rpv10) and 'Johanniter' (Rpv3) are known to be resistant to DM 10,65,66 , and both show an HR to P. viticola that leads to cell necrosis and reduced production of sporangia 14,67 . 'Merlot Kanthus' also carries Rpv3. On the other hand, 'Calandro' , 'Regent' , and 'Felicia' also carry Rpv3, but the resistance of these varieties was low in this work, as already observed in other studies 3,10,65 . The resistance expressed by 'Solaris' (CLU2, medium resistance level) involves both callose synthesis and an HR 60 . The varieties 'Merlot Khorus' , 'Cabernet Volos' , and 'Fleurtai' carry the Rpv12 (grouped in CLU2), inherited from V. amurensis 49 . A typical resistance response of V. amurensis consists in the activation of physical mechanisms involving callose deposition 15,58 that leads to the degeneration of portions of mycelium and the alteration of sporangiophore shape 7 . 'Rkatsitelii' , a Georgian grapevine variety, showed a very low level of resistance to DM, confirming the results of Bitsadze et al. 68 .
Overall, our findings indicate that the routine measurement of RCs in breeding programs can be simplified by the measurement of only IFR and SPOR. Table 4 shows that the assessment of IFR made it possible to distinguish CLU2 from CLU1 (medium and high resistance level, respectively), and from CLU4 (very low resistance level), but not CLU3 (low resistance level) from CLU4. At the same time, SPOR made it possible to distinguish CLU4 from CLU3. Therefore, IFR and SPOR, together, were able to distinguish all the four clusters. In addition, since IFR and SPOR are the major drivers of DM epidemic progress in vineyards 69 , which is caused by the concatenation of infection cycles, the phenotyping of partial resistant varieties by using these two RCs in monocyclic experiments would also account for their effect on epidemic development. This would reduce the workload of phenotyping while still producing correct assessment of resistance.
The current research documented differences in resistance expression among growth stages, with lowest resistance at stage 18, i.e., at leaf development. To our knowledge, no previous studies have reported this growth stage-related variability, and such variability requires further investigation. It is well known from Van der Plank's 25 studies of resistance to potato late blight (Phytophthora infestans) that resistance responses are inducible, which suggests that the expression of resistance in response to attack might be costly 70 . The metabolic cost of defence may be a form of allocation cost, in which resources devoted to self-protection are not available for other activities such as growth or reproduction 71 . Because fungicides must be used to protect even partially resistant grapevine varieties against DM 51,72 , a better understanding of the temporal dynamics of resistance expression may be useful in scheduling these fungicide applications.
In each vineyard and year, five plants of each variety were randomly selected at three growth stages: (i) shoot growing, (ii) flowering, and (iii) fruit development. These corresponded to the growth stages 18, 65, and 79, respectively, of Lorenz et al. 73 . At each growth stage, the fourth leaf from the apex of an actively growing shoot of each plant was sampled 74 . The detached leaves were then pooled, yielding 15 leaves per variety per stage.
The leaves were placed in a cooler (about 5 °C) and immediately transported to the laboratory where they were washed under tap water, disinfested with sodium hypochlorite (1%) for 1 minute, triple-rinsed using sterile demineralized water, and finally arranged under a sterile laminar flow until their surfaces were completely dry. (2020) 10:585 | https://doi.org/10.1038/s41598-020-57482-0 www.nature.com/scientificreports www.nature.com/scientificreports/ Following Staudt & Kassemeyer 75 and Rumbolz et al. 76 , 5 leaf discs, 21 mm in diameter, were cut from each leaf using a cork borer, so that 75 leaf discs were obtained for each variety (15 leaves, 5 leaf discs per leaf). The leaf discs were placed abaxial side up in Petri dishes (9 cm diameter) containing two layers of filter paper moistened with 3 ml of sterile demineralized water; the leaf discs were placed on nylon mesh to prevent direct contact with the moistened paper.

Inoculum preparation.
For the preparation of the inoculum to be used for artificial inoculations, field-produced sporangia were randomly collected and pooled with the aim of obtaining a diverse natural population of P. viticola sporangia and thereby minimizing the effect of a possible host × strain interaction on the phenotypic evaluation of the resistance components 54 . To obtain these sporangia, field-grown grape leaves showing typical DM lesions with fresh and abundant sporulation were collected in the spring of 2014 from several vineyards that had not been sprayed with fungicides; these leaves were collected from different V. vinifera varieties and from various locations in Northern Italy. The pooled leaves were brought to the laboratory, where sporangia were collected from lesions (with the help of a needle) and suspended in sterile double-distilled water. Droplets (10 μl) of this bulk suspension of sporangia were placed on the abaxial side of young, fresh leaves that had been detached from potted plants of 'Merlot' grown under isolation in a greenhouse at the University of Piacenza. The inoculated leaves were incubated for 7 days in a growth chamber at 20 °C and with high humidity and a 12-h photoperiod. The fresh sporangia produced on the DM lesions were suspended in sterile water, and the suspension was adjusted to 5 × 10 5 sporangia ml −1 . Immediately after preparation, these sporangial suspensions were used for the artificial inoculations described in the next section. The same suspensions were used to maintain the inoculum through repeated inoculations of Merlot leaves, using the same previously described method.
Inoculations. For inoculation of the leaf discs in Petri dishes, four 10-µl droplets of sporangial suspension were applied to each leaf disc using a micropipette. Petri dishes containing the inoculated leaf discs were then sealed with Parafilm (to maintain a saturated atmosphere) and incubated for 24 h in a growth chamber at 20 °C and with a 12-h photoperiod. The inoculum drops were then dried using blotting paper but without touching the leaf surface 46 . Petri dishes were then sealed again and kept in the growth chamber until the components of partial resistance were measured. Assessment of resistance. The degree of resistance was visually assessed 11 days post inoculation (dpi) using a modified OIV descriptor 452-1 46 . Leaf discs were rated from the most susceptible to the most resistant as follows: 1, dense sporulation at all four inoculation sites (i.e., very low degree of resistance); 3, dense sporulation at two or three inoculation sites (low resistance); 5, sparse sporulation at two or three inoculation sites (medium resistance); 7, sparse sporulation at one inoculation site (high resistance); and 9, absence of sporulation (very high resistance).
Measurement of the components of partial resistance. Leaf discs were examined daily with a stereomicroscope at 10-fold magnification for the assessment of the RCs as described in the following paragraphs. The 10-fold magnification allowed the screening of a high number of leaf discs with a resolution adequate to observe single sporangiophores of P. viticola 50 . No new DM signs appeared before 4 dpi or after 11 dpi.
Infection frequency. IFR was assessed at 11 dpi as the proportion of inoculation sites (i.e., sites where the inoculum drops were placed) showing typical DM lesions with sporulation 77 .  www.nature.com/scientificreports www.nature.com/scientificreports/ To account for the delay in lesion appearance following inoculation, the AUIPC was calculated by using the daily assessments of IFR between 4 and 11 dpi as follows: is the time interval between two consecutive assessments. The calculation follows the trapezoidal method for the estimation of the AUDPC (Area Under the Disease Progress Curve), which enables the discretization of the time variable and the calculation of the average disease intensity between each pair of adjacent time points 78 .
Duration of the latent period. The duration of the latent period (i.e., the time elapsed between inoculation and the start of sporulation on DM lesions) was measured as the thermal time 79 , i.e., as degree-days (DDs, base 0 °C) accumulated between the inoculation time and the time when 50% of the inoculation sites resulted in DM lesions at the end of the observation period (i.e., at 11 dpi); this is termed the LP50 80 . For instance, if IFR = 0.25 at 6 dpi (DD = 120, i.e., 6 days at 20 °C) and IFR = 0.75 at 7 dpi (DD = 140) on a leaf disc, then IFR = 0.5 is assumed to be at 6.5 dpi, corresponding to 130 DD; therefore, LP50 = 130.
Lesion size. At 11 dpi, a random sample of 50 leaf discs per variety, which have been inoculated in different growth stages and years, was photographed (13 megapixel f/1.9 resolution), and the area (in mm 2 ) of each lesion (referred as LS) was determined using Assess 2.0 (Image analysis software for plant disease quantification, by Lakhdar Lamari, APS PRESS, Saint Paul, Minnesota). Example of calculation of the lesion size of downy mildew lesions on leaf discs is provided in Supplementary Fig. S1.
Production of sporangia. Production of sporangia was determined as the number of sporangia produced per DM lesion (SPOR) and per mm 2 of DM lesion (SPOR') at 11 dpi. Sporangia were carefully collected from each leaf disc by using a needle; the collected sporangia were suspended in 100 µl of sterile-demineralized water; the suspension was shaken for 10 s, and the sporangia were counted using a hemocytometer. The total number of sporangia was then divided by the number of DM lesions assessed at 11 dpi on each leaf disc (for SPOR), and by the area of each lesion (for SPOR').
Duration of the infectious period. After 11 dpi, sporulating structures (sporangiophores and sporangia) were gently removed from leaf discs, which have been inoculated at fruit development in the second year, using a sterile cotton swab. Leaf discs were then incubated as previously described and observed every 2 days with a stereomicroscope (10 × magnification). When sporangiophores bearing sporangia re-appeared on lesions, sporulating structures were removed as previously described and the leaf discs were incubated again. This procedure was repeated until the lesions stopped producing new sporangia. The duration of the infectious period (IP, i.e., the period during which a lesion is fertile and continues to produce sporangia) was then expressed as the number of times the lesions produced sporangia 81,82 .
Infectivity of the sporangia produced on DM lesions. INF was assessed as the ability of the sporangia produced on DM lesions to cause infection on leaves of the susceptible variety 'Merlot' . At 11 dpi, sporangia produced on the DM lesions obtained in the second year were suspended in water as previously described for inoculum preparation, and the suspension was adjusted to 10 3 sporangia/ml. These sporangial suspensions were used to inoculate 15 leaf discs excised from 'Merlot' leaves that were collected and maintained as previously described. At 11 dpi, INF was assessed as the proportion of inoculation sites (i.e., sites where the inoculum drops have been placed) showing typical DM lesions with sporulation. Data analysis. The ANOVA was performed to determine whether RCs were significantly affected by grapevine variety, growth stage of vines, year (all considered as fixed factors), and their interactions. Before ANOVAs were conducted, SPOR and SPOR' data were transformed using the natural logarithm to stabilize variances and were then back-transformed using the inverse exponential function. When F values were significant, means were compared using the Fisher's protected least significant difference (LSD) post hoc test at P = 0.05.
To study the relationships among RCs, Pearson correlation coefficients were calculated between IFR versus AUIPC, LP50, LS, SPOR' , SPOR, IP, and INF.
To study the relationships between RCs and Rpvs, and among Rpvs, an ANOVA was performed by assigning the varieties into the following groups: susceptible control; varieties with no Rpvs; with Rpv3 (whether Rpv3 1 and Rpv3 2 ); with Rpv10; or with Rpv12.
A hierarchical cluster analysis was conducted for grouping grapevine varieties based on the within-group similarities and between-group differences in the average values of five RCs: IFR, LP50, SPOR, IP, and INF. AUIPC was not included in this analysis because it was closely correlated to IFR (see Table 3); LS and SPOR' were also not included because SPOR incorporates the information concerning the lesion size and the sporulation per surface unit of lesions. The Ward's method was used for clustering, in which the distance between two clusters is indicated by the increase in the sum of squares caused by merging of the clusters, with the Euclidean square distance for measuring similarities. Because the RC values were measured on different scales, some being much larger than others, the data were standardized by using the z-scores as follows: − (x x)/sd i , where x i is any value of a variable, x is the average for the variable, and sd is the standard deviation. An ANOVA was then performed to determine whether the groups of varieties formed by the cluster analysis were significantly different for each of the five RCs.
All analyses were performed using IBM SPSS software version 24 (SPSS, Chicago, IL).