Coincidence of maximum severity of powdery mildew on grape leaves and the carbohydrate sink-to-source transition

Authors


E-mail: Katherine.Evans@utas.edu.au

Abstract

Grapevine leaves infected with powdery mildew are a source of inoculum for fruit infection. Leaves emerging on a single primary shoot of Vitis vinifera cv. Cabernet Sauvignon were exposed to average glasshouse temperatures of 18°C (0·23 leaves emerging/day) or 25°C (0·54 leaves emerging/day). All leaves on 8–10 shoots with approximately 20 leaves each were inoculated with Erysiphe necator conidia to assess disease severity after 14 days in the 25°C glasshouse. Two photosynthetic ‘source’ leaves per shoot on the remaining 8–10 shoots were treated with 14CO2 to identify, by autoradiography, the leaf position completing the carbohydrate sink-to-source transition. There was a clear association between the mean modal leaf position for maximum severity of powdery mildew (position 3·7 for 18°C; position 4·4 for 25°C) and the mean position of the leaf completing the sink-to-source transition (position 3·8 for 18°C; position 4·7 for 25°C). The mean modal leaf position for the maximum percentage of conidia germinating to form secondary hyphae was 4·2 for additional plants grown in the 25°C glasshouse. A higher rate of leaf emergence resulted in a greater proportion of diseased leaves per shoot. A Bayesian model, consisting of component models for disease severity and leaf ontogenic resistance, had parameters representing the rate and magnitude of pathogen colonization that differed for shoots developing in different preinoculation environments. The results support the hypothesis that the population of leaves in a vineyard capable of supporting substantial pathogen colonization will vary according to conditions for shoot development.

Introduction

Erysiphe necator causes powdery mildew in species of Vitaceae, including Vitis vinifera, which is widely cultivated for table and wine grapes. Cultivars of V. vinifera are generally highly susceptible to powdery mildew (Gadoury et al., 2012). Infected leaves are a source of inoculum for fruit infection, with powdery mildew incidence on leaves at flowering associated positively with disease severity on bunches (Calonnec et al., 2006). Pathologists propagating E. necator know that for abundant inoculum (conidia) production, it is necessary to select about the fourth leaf below the shoot apex that has a shinier cuticle than older leaves and is not quite fully expanded. Models of leaf growth for V. vinifera cv. Cabernet Sauvignon support these observations: leaves predicted to be 85–98% of their final lamina length at the time of inoculation with E. necator developed severe powdery mildew (Lee et al., 2012). Inoculated older leaves develop less powdery mildew (Doster & Schnathorst, 1985; Singh & Munshi, 1993).

Disease severity of many biotrophic pathogens infecting woody perennial plants varies with leaf age or position following inoculation under controlled-environment conditions (e.g. Sharma et al., 1980). Disease signs and symptoms may be absent altogether if a leaf is younger than the latent period of the pathogen. The term ontogenic or age-related resistance is applied when it can be demonstrated that the host has overcome, completely or to some degree, the effect of the pathogen as the leaf ages. Doster & Schnathorst (1985) reported that the percentage of germinated E. necator conidia that developed hyphae 48 h after inoculation of leaves of V. vinifera cultivars declined as leaves matured beyond a midvein length of 5 cm. Ontogenic resistance was clearly expressed as a corresponding decline in hyphal length from germinated conidia with appressoria. Reduced infection, growth and sporulation of E. necator on V. vinifera cultivars with increasing leaf maturity was also observed by Singh & Munshi (1993).

Little is known about the cellular and molecular genetic basis of ontogenic resistance in leaves. Studies of mutants have shown connections between plant development and defence, but age-related factors that confer competency to respond with a defence reaction have not been identified (Whalen, 2005). Moreover, leaf position relative to the shoot apex can indicate the age of a leaf relative to another, but it does not reveal the capacity of a leaf at a particular position to support pathogen growth and reproduction. For example, the relationship between the maturity of leaves (days) on shoots of Populus species and the number of uredinia mm−2 of Melampsora larici-populina varied with shoot age (Sharma et al., 1980). Rather than identify mechanisms of ontogenic resistance, which are inextricably linked to plant organ development, this current study sought to understand the metabolic state of leaves capable of supporting substantial pathogen growth and reproduction.

As a leaf develops, it converts from being a net importer (sink) to a net exporter (source) of carbohydrate (Turgeon & Webb, 1973), hereafter referred to as (carbon) assimilates. The timing of the transition from leaf sink to source is correlated with attainment of a positive carbon balance: import stops and export begins when the supply of assimilates exceeds the growth and respiratory needs of the leaf (Pate & Atkins, 1983). Leaves of Vvinifera cv. Muscat of Alexandria start exporting assimilates when they are about one half their final area (Hale & Weaver, 1962). There is a period of time when a leaf is importing and exporting concurrently, although not necessarily from the same part of the leaf (Turgeon, 1989). Biotrophic pathogens act as a sink for assimilates, with increases in invertase activity in infected leaves resulting in an accumulation of soluble carbohydrates (e.g. Gamm et al., 2011). Chou et al. (2000) suggested that an increase in apoplastic invertase activity in diseased leaves may facilitate phloem unloading of sucrose into cells adjacent to fungal mycelium, thus converting regions of a source leaf into a (fungal) sink for assimilates. This source-to-sink transition was documented clearly for Plasmopara viticola infecting V. vinifera (Gamm et al., 2011). Such a strategy would aid efficient nutrient acquisition by the pathogen, which presumably influences the extent of host tissue colonization. Host–pathogen studies like these typically use the most ontogenically susceptible leaf. Less is known about the extent of host tissue colonization in relation to the capacity of a leaf, or leaf section, to export assimilates to the pathogen.

Coleman (1986) postulated that leaves pass through a stage of maximum susceptibility to attack by biotrophic pathogens and phytophagous insects that coincides with the sink-to-source transition. The primary grapevine shoot is the one that emerges from the primary bud in the dormant compound bud. The present study tests the hypothesis that the leaf on a primary grapevine shoot that expresses maximum severity of powdery mildew is the one that was infected just when the leaf ceased importing assimilates. A Bayesian model (Berger, 1993) was constructed to quantify a nonlinear change in powdery mildew severity as a function of increasing leaf position on primary shoots of cv. Cabernet Sauvignon. This mechanistic model is used to aid interpretation of the effect of the preinoculation environment on the proportion of diseased leaves per shoot and disease severity per leaf position. A secondary objective of the present study was to quantify leaf ontogenic resistance under near optimum conditions for host and pathogen growth by describing the relationship between leaf position and percentage of E. necator conidia germinating to form secondary hyphae. The significance of the results are discussed in terms of the need for more precise identification of the population of leaves in a crop that are susceptible over time in order to understand the interventions needed to manage powdery mildew.

Materials and methods

Plant material and leaf measurements

Dormant rooted cuttings of Vitis vinifera cv. Cabernet Sauvignon, clone Q390-05, pruned to one bud, were grown in moist, well-drained commercial potting medium (Horticultural and Landscape Supplies) containing controlled-release Osmocote® fertilizer (Total All Purpose, Scotts; N:P:K 19·4:1·6:5 plus trace elements) in 15-cm-diameter pots (one cutting/pot). Plants were grown in a glasshouse with a 16-h day length supplemented by 400-W mercury halide lights. Water was reapplied as soon as the surface of the potting mixture began to dry. Plants were fertilized every 14 days with water-soluble Thrive All Purpose Plant Food (Yates Australia; N:P:K 27:5·5:9 plus trace elements). In two separate experiments, 20 plants were grown in a glasshouse with the temperature set at 25°C (experiment one) and 16 plants were grown in a glasshouse set at 18°C (experiment two). Temperatures measured during shoot growth confirmed means equal to the set temperatures, but temperatures were variable by ±5 and ±8°C, respectively. These two environments were intended to result in relatively fast (near-optimum) or slow (suboptimum) rates of leaf emergence (Winkler, 1970), respectively.

In experiments one and two, leaves were numbered according to position from the shoot apex to the base, starting with leaf position 1 for the first leaf with a lamina length ≥30 mm, then 2, 3, etc. for older leaves. Once shoots had developed four or five leaves, lamina lengths on leaves of each shoot were measured every 3–4 days, until shoots developed approximately 20 leaves. The rate of leaf emergence for each shoot was calculated from the slope of the linear regression of plastochron index (Erickson & Michelini, 1957), with a reference lamina length of 30 mm, against number of days since the first measurement of lamina length. The mean rate of leaf emergence for all shoots from each experiment was then calculated.

Leaves emerging on shoots were maintained free of powdery mildew using vapours of penconazole (Topas, Syngenta Crop Protection Pty Ltd) (Szkolnik, 1983), which was discontinued 7 days prior to inoculation with E. necator. Few of these shoots developed inflorescences, and where these did occur they were removed as soon as they were observed. Half the plants with approximately 20 leaves from each glasshouse environment were used for inoculations and the other half were used for treatment with 14CO2 to identify the sink-to-source transition. Lamina length on each leaf was measured immediately prior to inoculation or treatment with 14CO2. This leaf variable is easy to measure and can be used to estimate leaf area (Merry, 2011).

Inoculation, assessment of disease severity and conidial germination

A bulk isolate of E. necator was collected from a commercial vineyard in southern Tasmania and cultured on detached leaves of cv. Cabernet Sauvignon as described by Evans et al. (1996). Conidia were shaken off leaves of 12-day-old cultures into distilled water containing 0·05% Tween 20. Ten plants from the 25°C glasshouse and eight plants from 18°C glasshouse were selected for inoculations. For both experiments, a suspension of 105 E. necator conidia mL−1 water was applied to the adaxial side of all leaves using a handheld atomizer. Plants were then incubated in the 25°C glasshouse (±5°C) for 14 days before visual estimation of disease severity on the adaxial surface of each leaf. Disease severity was assessed with the aid of a standard area diagram (R. W. Emmett, Primary Industries Victoria, Australia, unpublished data) that displays the percentage of leaf area covered by epiphytic hyphae, conidiophores and conidia of E. necator.

A third experiment (experiment three) was conducted, in which 10 plants were grown in the 25°C glasshouse (±5°C), as described previously, until approximately 20 leaves developed. All leaves were then inoculated by transfer of dry conidia from 12-day-old cultures with an artists' paintbrush. Plants were incubated in the glasshouse at 25°C (±5°C) as before. At 72 h after inoculation, a 2- × 4-cm section adjacent to the basal end of the midvein of each leaf was removed. Sections were cleared in 3:1 ethanol:glacial acetic acid for 48 h, softened in lactoglycerol for 24 h and stained with lactoglycerol with 0·1% trypan blue for 48 h. Light microscopy with ×400 magnification was used to determine the germination status of the first 40 conidia observed on each section. The presence of one or more secondary hyphae per germinated conidium was taken to indicate that infection had proceeded to penetration of the leaf cuticle (Ficke et al., 2003).

Identification of the sink-to-source transition

Ten plants from the 25°C glasshouse (experiment one) and eight plants from 18°C glasshouse (experiment two) were selected for treatment with 14CO2. Two source leaves for assimilates on plants from both experiments, which were the two youngest leaves per shoot that had reached maximum lamina length (data not presented), were each enclosed in a polythene bag sealed around the petiole at 08·00 h when conditions outside the glasshouse were sunny. A total of 17·8 kBq 14CO2 was released inside the polythene bag by addition of 0·5 mL 20% (v/v) lactic acid to a 1·2 mL solution of NaH14CO3. Photosynthesis was allowed to continue for 2 h before removal of the polythene bag. After 24 h, leaves exposed to 14CO2 and all leaves distal to the exposed leaves were cut from the shoot and dried for 7 days in a plant press at room temperature. Dried leaves of each shoot were then exposed to general-purpose X-ray film in a 35- × 43-cm medical X-ray cassette for 3 weeks prior to film development. After inspection of the autoradiograph, the leaf position for the cessation of the sink-to-source transition was designated as the first (proximal) leaf showing no visual evidence of 14C accumulation.

Bayesian model construction

A nonlinear modelling framework with parameters that can be interpreted biologically was used to describe the relationship between leaf position and disease severity using data from experiments one and two. Bayesian inference is the process of fitting a probability model to a set of data and summarizing the results via probability distributions on the parameters of the model (Mila & Carriquiry, 2004). In a Bayesian model, the parameters are considered to be uncertain and are therefore represented by probability distributions, known as prior distributions, or priors. These are combined with the model likelihood to obtain posterior distributions that provide their most probable values given the data and modelling assumptions (Berger, 1993). The posterior distributions are summarized using summary statistics. Models developed using classical or frequentist statistics generally require that experiments be repeated to demonstrate reproducibility. A Bayesian approach was selected to allow a unified analysis of both experiments and use of priors common to both, thus providing comparisons of parameters between them and allowing insights that would otherwise not have been available.

The model consists of two linked components, one termed the disease severity model and the other the leaf resistance model. In the first, it was assumed that the pathogen would eventually colonize the entire leaf area. While this may not actually occur, the aim was to describe the situation where the host does not suppress the pathogen. In the second, the host's suppression of the pathogen was described separately. This structure enabled a more complicated problem to be approached in simpler steps, each of which may correspond to processes in nature that are not observed directly. In both models, j denoted individual plants, p denoted the leaf position, and i denoted the preinoculation environment (25 or 18°C on average).

Disease severity model

To describe the change in the proportion of leaf area colonized, a logistic function was adopted, given by Eqn (1) in which aj,p was the proportional area colonized by the pathogen on leaf position = 1, 2,….Pj on plant j, and βj and γj were constants to be estimated. The constant γj may be interpreted as an indicator of the rate of colonization, while βj corresponded to the magnitude of initial colonization.

display math(1)

Leaf resistance model

The second model is given in Eqn (2), in which sj,p is the resistance, and δj and εj are parameters to be estimated. The parameter δj can be interpreted as the rapidity of the disease resistance response with increasing leaf position, while εj controls the lag in response by the host.

display math(2)

Finally, it was assumed that the observed disease severity was Gaussian-distributed with mean given by the product of the outcomes of the disease severity and leaf resistance models: mj,p ∼ N(aj,p × sj,pτ) , where τ is the precision (reciprocal variance).

To allow for variation between plants within each preinoculation environment the plant-level parameters shared a common prior distribution (Table 1). For example, βj was assigned a Gaussian distribution, inline image , in which the mean inline image was common to all the plants that were in the preinoculation environment i. Inference was obtained in the form of posterior means and variances obtained via Markov chain Monte Carlo (Brooks, 1998) (MCMC) simulation. The probability that a parameter in one preinoculation environment was larger than in the other was calculated by summing the number of MCMC iterations in which this occurred and dividing by the total number of iterations. Each parameter was updated in turn using either a Metropolis–Hastings (Chib & Greenberg, 1995) or Gibbs update (Brooks, 1998), depending upon the form of the associated posterior conditional distributions. The model was run for 500 000 iterations and a 50% burn-in was used. Sensitivity studies and standard diagnostic techniques were used (Brooks & Roberts, 1998) to assess model validity. Comparison of the posterior parameter means for both the disease severity and leaf resistance models was done by calculating the probabilities that one would exceed the other.

Table 1. Common ‘prior’ distributions for each preinoculation environment in Bayesian models of the relationship between grapevine leaf position and powdery mildew disease severity. The Gaussian distribution is denoted by N and the Gamma distribution by Ga
ModelPriors  
Disease severity
display math
 
display math
display math
display math
display math
display math
Leaf resistance
display math
display math
display math
display math
display math
display math
Observational precision 
display math
 

Other data analyses

For each shoot inoculated with E. necator, the modal leaf position at maximum disease severity or maximum percentage of conidia with secondary hyphae was identified by a bootstrap approach (Efron & Tibshirani, 1993). The mean and range for leaf position of the leaf expressing maximum powdery mildew severity for each preinoculation environment were then compared with the corresponding mean and range of leaf position in the sink-to-source transition, as determined by cessation of assimilate import.

For experiment one, lamina lengths immediately prior to inoculation at leaf positions with maximum powdery mildew severity were compared with those at leaf positions for the sink-to-source transition using a two-sided t-test. The same comparison was done using data from experiment two. Linear regressions for estimating rates of leaf emergence and t-tests were calculated using excel 97–2003 (Microsoft). Linear regressions for germinating conidia against leaf position were calculated using GenStat v. 12.1 (VSN International Ltd).

Results

On average, 0·54 leaves emerged per day for shoots in the 25°C glasshouse and 0·23 leaves emerged per day for shoots in the 18°C glasshouse. A nonlinear change in powdery mildew severity with increasing leaf position was evident from the results for experiments one and two (Fig. 1). The higher rate of leaf emergence resulted in a greater proportion of diseased leaves per shoot (Fig. 1): no powdery mildew was observed at leaf positions ≥17 on shoots with the higher rate of leaf emergence, nor at leaf positions ≥11 for shoots with the lower rate of leaf emergence.

Figure 1.

Effect of leaf position (values increase as leaves age) on shoots of glasshouse-grown Vitis vinifera cv. Cabernet Sauvignon vines on mean severity of powdery mildew (Erysiphe necator) 14 days after determination of leaf position and inoculation. Plants were grown at an average temperature of 25°C (a) or 18°C (b) prior to inoculation. Error bars represent standard error of the mean.

There was a clear association between the leaf position for maximum severity of powdery mildew and the position of the leaf completing the sink-to-source transition for shoots exposed to the same preinoculation or pretreatment environment (Table 2; Fig. 2). The leaf position for the leaf completing the sink-to-source transition occurred on average at leaf positions 3·8 and 4·7 for the 18 and 25°C pretreatment environments, respectively. The modal leaf position at maximum disease severity occurred on average at leaf positions 3·7 and 4·4 for the corresponding glasshouse environments. In both experiments one and two, there was no significant difference in mean lamina length prior to inoculation or treatment between leaves completing the sink-to-source transition and those with maximum disease severity (Table 2).

Figure 2.

Autoradiograph (a) illustrating ‘source’ leaves from a primary shoot of Vitis vinifera cv. Cabernet Sauvignon treated with 14CO2 and the outline of the corresponding leaves detached from this shoot and placed in an X-ray cassette (b). The youngest leaf that had ceased importing assimilates (indicated with an arrow) lacks a dark image on the autoradiograph.

Table 2. Mean leaf position or lamina length of the leaf expressing maximum powdery mildew severity or of the leaf in the sink-to-source transition, as determined by cessation of assimilate import, for grapevine cv. Cabernet Sauvignon shoots grown in different environments prior to inoculation or treatment with 14CO2. Lamina lengths were measured immediately prior to inoculation with Erysiphe necator or treatment with 14CO2
Average temperature (°C) preinoculation or pre-14CO2 treatmentLeaf positionLamina length (mm)Significance (P) of t-test for mean lamina lengtha
Maximum disease severitySink to source leafPrior to inoculationPrior to 14CO2 treatment
MeanRangeMeanRange
  1. a

    For means within rows.

  2. b

    Mean of the modes.

  3. c

    Standard deviation of the mean in parentheses.

183·7b2·1–4·83·83·0–5·092 (8·5)c90 (7·0)0·718
254·4b3·0–6·64·73·0–7·083 (12)87 (6·1)0·433

The maximum percentage of conidia germinating among sections from different leaf positions per shoot was in the range 45–95% (data not presented). The mean modal leaf position of 4·2 (range 3·0–6·6) for the maximum percentage of conidia germinating to form secondary hyphae was not significantly different from that for maximum disease severity (4·4 for the 25°C preinoculation environment; = 0·939 for two-sided t-test). The penetration of epidermal cells by germinating E. necator conidia declined as leaves matured beyond leaf position 4 (Fig. 3a). In contrast, there appeared to be a slight decline in the percentage of conidia with primary germ tubes but no secondary hyphae as leaves aged (Fig. 3b). Linear regressions of conidial germination against leaf position were significant (< 0·001), with adjusted R2 values of 0·4 and 0·15 for germinating conidia with or without secondary hyphae, respectively.

Figure 3.

Effect of leaf position (values increase as leaves age) on shoots of glasshouse-grown Vitis vinifera cv. Cabernet Sauvignon vines on (a) the percentage of germinated Erysiphe necator conidia with secondary hyphae, or (b) the percentage of germinated E. necator conidia with only primary germ tubes. Each circle represents the percentage for 40 conidia observed per leaf section 72 h after inoculation. A natural spline curve was fitted to the data (Green & Silverman, 1994). Plants were grown at an average temperature of 25°C before and after inoculation.

Visually, the Bayesian model gave a good fit for all shoots (Fig. 4). The magnitude of initial colonization and the rate of colonization, indicated by βj and γj, respectively in the disease severity model, were both greater on vines with a faster rate of leaf emergence before inoculation (Table 3). The probability that the posterior inline image was less than inline image was 1·0, and the probability that the posterior inline image was less than inline image was 0·98. For the leaf resistance model parameters there was little difference between the two preinoculation environments. The probabilities that the posterior inline image was less than inline image was 0·117, and the probability that the posterior inline image was less than inline image was 0·287.

Figure 4.

Observed data and fitted values from the Bayesian model for data from primary shoots of Vitis vinifera cv. Cabernet Sauvignon grown at an average temperature of (a) 18°C or (b) 25°C prior to inoculation of all leaves with Erysiphe necator conidia. Values of leaf position increase as leaves age. The curves fitted to observed data (circles) are the overall model (–), the leaf resistance model (- - -) and the disease severity model (….).

Table 3. Posterior means (with standard deviations in parentheses) for preinoculation environment hyper-prior parameters inline image , inline image , inline image , inline image in Bayesian models of the relationship between grapevine leaf position and powdery mildew disease severity
Average preinoculation temperature (oC)inline image (magnitude of initial pathogen colonization)inline image (rate of pathogen colonization)inline image (rapidity of host resistance response)inline image (lag in response by host)
183·29 (0·30)0·96 (0·24)4·30 (2·08)0·61 (0·09)
255·00 (0·41)1·65 (0·22)3·52 (0·87)0·49 (0·07)

Discussion

Leaves of Cabernet Sauvignon vines were most prone to development of severe powdery mildew when infected immediately after leaves had ceased importing assimilates. This association was maintained regardless of preinoculation or pretreatment environment. An ideal ecological niche for infection by E. necator might be created immediately after completion of the sink-to-source transition as a result of high amounts of sugar available for nutrition of germinating conidia that had relatively high penetration rates (Fig. 3a; Sutton et al., 1999). In healthy leaves, apoplastic invertase gene expression declines suddenly as the leaves mature, coinciding with the transition of the leaf from a sink to a source organ (Sturm et al., 1995; Godt & Roitsch, 1997). Even though a biotrophic pathogen can act as a localized sink for assimilates, any increase in invertase activity of host origin (e.g. Chou et al., 2000) might be restricted in older leaves. The results of this study raise the possibility that fungal nutrition is a factor in the level of disease severity observed; however, this level will be influenced to some degree by unknown mechanisms of ontogenic resistance preventing fungal penetration of the leaf cuticle and/or cessation of fungal growth after formation of haustoria.

Low disease severity on newly unfolded leaves that were sinks for assimilates might have been the result of rapid expansion of leaf area between inoculation and disease assessment; however, the fitted curve for the percentage of germinating conidia with secondary hyphae suggested these leaves were less susceptible to infection by E. necator (Fig. 3a). Given that there are two sink organs, namely the fungus and the leaf, then one hypothesis is that the host cells, with greater biomass relative to the fungus, have a greater competitive ability or ‘sink strength’ for importing and metabolizing assimilates to support rapid organ growth (Ho, 1988). Specific biochemical mechanisms for resistance might also be operating, given that there are significant changes in foliar chemistry during leaf expansion (Cole, 1966; Coleman, 1986). Again, the cause of any change in susceptibility to infection by E. necator conidia remains unknown.

The current study used primary grapevine shoots without inflorescences or fruit: grape bunches with unripe fruit are sinks for assimilates, with sink strength generally increasing as berry development progresses (Hunter & Visser, 1988). It is not known how the presence of reproductive tissues or secondary shoots would affect assimilate distribution dynamics in relation to powdery mildew development on a particular leaf; strong sinks such as grape berries are usually supplied with assimilates from a nearby source, for example, basal leaves (Hunter & Visser, 1988). Doster & Schnathorst (1985) observed that ‘…late in the season U. necator develops profusely again on presenescent leaves of certain cultivars, thus exhibiting a distinct gradient in development’. Distribution of assimilates to vegetative sinks is resumed once grape berries are ripe and the sink capacity of the grape clusters decreases (Hunter & Visser, 1988). There is potential to further explore disease severity as a function of the ability of a plant organ to provide the parasite with assimilates, depending on assimilate distribution dynamics at the time of infection. Spatial and temporal resolution of the photosynthetic sink-to-source transition within individual leaves might also aid studies identifying connections between leaf development and ontogenic resistance.

The expression of powdery mildew on grape leaves was clearly different in plants exposed to different preinoculation environments. Previous studies (Doster & Schnathorst, 1985; Singh & Munshi, 1993) have not modelled the relationship between disease severity and leaf position. This study appears to be a new application of the Bayesian approach in plant pathology; most recent applications appear to be for molecular genetic analyses and, occasionally, for plant disease prediction. The Bayesian analysis provided insight to the host–pathogen relationship that would not have been possible using classical or frequentist statistics; that is, the parameters of the leaf resistance model were similar for plants grown in the two different environments, yet there was a difference in these parameters for the disease severity model. Mechanistic modelling enabled these effects to be separated; i.e. the interpretation was that the warmer preinoculation environment produced leaf tissue capable of supporting greater initial colonization by E. necator and a higher rate of colonization.

The results of this study suggest that either the preinoculation environment had a direct effect on the nutritional quality of the plant tissue to be colonized by E. necator and/or there was a direct effect of the environment, perhaps temperature, on the expression of preformed and/or induced mechanisms of resistance. Few pathogenicity studies have investigated the effects of temperatures prior to infection which permit normal host growth but are below the accepted optimum for the species; in one example, Rubio-Covarrubias et al. (2006) found that the penetration frequency of Phytophthora infestans zoospores in potato tubers was lower when whole plants were preconditioned at 16°C rather than at 24°C.

The development of secondary hyphae of E. necator was severely inhibited as leaves of Cabernet Sauvignon vines aged, thus confirming leaf ontogenic resistance. This finding corresponds with full expression of ontogenic resistance in older berries, where successful penetration, formation of haustoria and development of secondary hyphae can decline to nil or trace levels (Ficke et al., 2003). Germination and appressorium formation were not affected by berry age (Ficke et al., 2003). The lack of germination of E. necator conidia on the oldest leaves in the current study suggests further research is needed to elucidate whether or not pathogen development is inhibited during the prepenetration phase. Conidia of Venturia inaequalis germinate on apple leaves up to 4 months old; leaf ontogenic resistance in this system is manifest as slow mycelial growth (Li & Xu, 2002). Singh & Munshi (1993) found no change in germination percentage of E. necator conidia or appressorium formation between the first fully expanded leaf and the eighth leaf of the susceptible V. vinifera cv. Thompson Seedless, a result that is consistent with the data presented in Figure 2b.

Models simulating spatiotemporal development of powdery mildew within and between grapevines in the field can benefit from a better understanding of variation in host tissue susceptibility. In developing a simulation model for development of powdery mildew on a single vine, Calonnec et al. (2008) accounted for leaf ontogenic resistance in the sense that a constant decay rate for leaf susceptibility was used for pathogen infection and colony growth. The current study suggests that the ‘decay rate’ will depend on environmental conditions that influence the rate of leaf emergence, which for V. vinifera cvs Chardonnay and Pinot Noir vines varies according to growing season, shoot position on a single vine and/or crop stage (Merry, 2011). A higher shoot leaf number early in the season has been associated with a greater incidence of leaves with powdery mildew and more diseased berries (Valdés-Gómez et al., 2011). If the interaction between vine genotype and environment also leads to a dense vine canopy, pathogen development may also be promoted through increased humidity and reduced UV light exposure (Austin et al., 2011).

From a disease management perspective, this study provides quantitative evidence that the conditions under which shoots develop will affect the rate of leaf emergence and the incidence and severity of powdery mildew on a grapevine shoot. More precise identification of the population of leaves (number and location) in a crop that are susceptible over time, especially during the period when grape bunches are highly susceptible to E. necator infection, will aid understanding of the interventions needed to influence the development of powdery mildew.

Acknowledgements

This work and the PhD scholarship of AMM were supported by the University of Tasmania and Australian grape growers and winemakers through their investment body the Grape and Wine Research and Development Corporation (Project UT0401). Support from the Australian wine sector was matched by the Australian Government. The authors thank Professor Philip Brown for advice on radiolabelling.

Ancillary