The use of digital image analysis and real-time PCR fine-tunes bioassays for quantification of Cercospora leaf spot disease in sugar beet breeding

Authors


E-mail: Bruno.Cammue@biw.kuleuven.be

Abstract

Cercospora leaf spot, caused by the fungus Cercospora beticola, is a major fungal sugar beet disease worldwide and the cause of significant yield losses. The disease is most successfully countered by the introduction of genetic tolerance into elite sugar beet hybrids. To this end, breeding programmes require high quality biological assays allowing discrimination of minor differences between plants within a segregating population. This study describes the successful implementation of image analysis software in the bioassays for quantification of necrotic lesions at different stages of C. beticola infection, allowing selection on minor phenotypic differences during the sugar beet breeding process for C. beticola resistance. In addition, a real-time PCR assay was developed for the quantification of C. beticola pathogen biomass in infected beet canopy. The use of both techniques, even in an early stage of infection, fine-tunes current bioassays, allowing more accurate and efficient selection of resistant breeding material.

Introduction

Cercospora leaf spot (CLS), caused by the fungal pathogen Cercospora beticola, is considered to be one of the most destructive foliar pathogens of sugar beet (Beta vulgaris) worldwide (Weiland & Koch, 2004). CLS is managed by planting partially resistant cultivars, cultural practices that reduce primary inoculum and/or multiple applications of fungicides (Miller et al., 1994). Without such measures, the leaf canopy of sugar beet fields can be destroyed by cyclic outbreaks of C. beticola, resulting in yield losses of up to 50% (Shane & Teng, 1992). Available sources of resistance to C. beticola are polygenic and quantitative, and thought to be controlled by up to five sugar beet genes (Smith & Gaskill, 1970). Combining high levels of CLS resistance with high yield in sugar beet is hampered because there is a genetic linkage between reduced yield and CLS tolerance (Smith & Campbell, 1996). As a result, commercial varieties generally have only moderate levels of resistance and still require fungicide applications to obtain adequate levels of protection against CLS (Miller et al., 1994; reviewed by Ioannidis & Karaoglanidis, 2010). Due to the high cost of fungicidal control, the appearance of fungicide resistant C. beticola isolates (reviewed by Karaoglanidis & Ioannidis, 2010) and the need to increase sustainability of sugar beet cultivation, novel, strong and durable genetic resistances are urgently needed. In that respect, the disease resistance potential of wild Beta species is considered to be immense (Francis & Luterbach, 2003). Useful resistance within Beta species to several diseases including CLS has been reported (Francis & Luterbach, 2003; Weiland & Koch, 2004).

A major aspect of the successful introduction of novel disease resistance traits in breeding programmes is the availability of appropriate evaluation methods. Screening programmes rely on quantifying the reaction of individual plants in the different breeding steps when challenged by pathogens in either artificial or natural environments. Traditionally, this has been achieved by examining plants visually for the presence of symptoms, and assessing the percentage of infected leaf area on a single-leaf basis as a measure for disease severity (e.g. Wolf & Verreet, 2005). However, visual assessments have been criticized for a lack of precision which is related to their qualitative and subjective interpretation of the symptoms (Sherwood et al., 1983; Nilsson, 1995; Steddom et al., 2005a). This criticism is even more pronounced when discrimination of minor differences in resistance level is required, such as in breeding programmes producing segregating populations with only minor differences in resistance level between individual plants due to quantitative genetic interactions (Setiawan et al., 2000; Weiland & Koch, 2004).

Recently, implementation of radiometric measuring of CLS has been shown to be more precise and accurate than traditional visual assessments, guarding against potential bias in disease evaluations (Steddom et al., 2005a). An alternative, more amenable evaluation method for assessment of foliar diseases appears to be image analysis using a digital camera and easy-to-handle image analysis software (Nutter et al., 1993; Lamari, 2002) as already used for disease evaluation in plants such as wheat (Strelkov et al., 2002; Steddom et al., 2005b), grapefruit (Bock et al., 2008) and Nicotiana benthamiana (Wijekoon et al., 2008). Assess 2·0 software, the digital image analysis tool used in this study, is based on selecting pixels that match certain criteria, also known as segmentation (Russ, 1998; Lamari, 2002). The ready availability of powerful personal computers and low cost digital cameras as well as the low cost of the software package make this a method of choice as an alternative or supporting tool for visual assessment of necrosis generating diseases such as CLS. Image analysis using Assess 2·0 was recently shown to be very robust for foliar disease quantification (Steddom et al., 2005b; Bock et al., 2008; Grünwald et al., 2008; reviewed in Bock et al., 2010). However, since development of disease symptoms does not always correlate with actual pathogen colonization (Bent et al., 1992; Hoffman et al., 1999; Thomma et al., 1999), it is recommended that the development of quantitative symptom evaluation methods is supported by techniques for quantification of pathogen proliferation. The latter includes quantitative or real-time PCR to quantify pathogen biomass and has been successfully implemented for several crop–disease systems (Böhm et al., 1999; Weller et al., 2000; Lees et al., 2002; Winton et al., 2002; Brouwer et al., 2003; McCartney et al., 2003; Yan et al., 2008).

To improve the accuracy of disease quantification, defined as the degree of closeness of measured values to the real value of disease (Bock et al., 2010), and guard against potential bias in analysis, this study set out to evaluate the use of digital image analysis software in CLS of sugar beet, as well as the development of a real-time PCR quantification method for C. beticola proliferation in sugar beet leaves. The study shows that both techniques can be used in CLS bioassays. In addition, both methods can be implemented at an early stage of infection.

Materials and methods

Pathogen growth

The inoculum source of C. beticola used in this study consisted of a mixture of three isolates obtained from naturally infected sugar beet plants in Belgium and France. A mixed pre-culture was prepared by grinding two agar pieces (1 × 2 cm) of fresh fully grown mother plates in 1200 μL water and inoculating five Petri dishes of solid V8 medium (200 mL L−1 V8 Campbell juice, 3 g CaCO3, 20 g L−1 agar) of pH 7·5 with a cotton plug. The isolates were allowed to grow for 14 days at 21°C in the dark. Spores and mycelium were harvested by adding 5 mL of sterile water per plate and gently scraping with a glass spatula. The plates were additionally rinsed with 4 mL of sterile water and the collected inoculum was divided with a cotton plug over 150 freshly prepared Petri dishes with solid V8 medium (200 mL L−1 V8 Campbell juice, 20 g L−1 agar) at pH 5·0, to enhance spore production. These plates were incubated for 3 days at 21°C in the dark. Subsequently, spores were harvested by adding 10 mL of sterile water per plate and gently scraping with a glass spatula. The spore concentration was determined using a cell counting chamber and adjusted to a final concentration of 2 × 106 spores mL−1 in 25% glycerol and stored at −80°C.

Plant growth conditions and inoculation procedure

Plants were grown for 8 weeks in a phytotron (22°C, a photoperiod of 16 h at 150 μmol m−2 s−1 light intensity, 60–70% relative humidity) and were treated in a randomized block design. Plants were spray-inoculated until runoff with a C. beticola conidial suspension at a concentration of 3 × 104 spores mL−1 (diluted in sterile water). Immediately after inoculation (=day 0), plants were kept at near-saturated air humidity (90–100% RH) for 24 h, without any light. From day 1 until the end of the experiment plants were kept at 60–70% RH with a 16 h photoperiod as described above.

Visual assessments

Disease progress of each plant was recorded according to the modified scale from Rossi (Rossi et al., 1991, 1999; Cercospora Development of Epidemics (CercoDep) model). In this scale, symptoms were estimated by a disease scale ranging from 0 (0% necrotic leaf area) to 100 (complete leaf covered with necrotic spots), as illustrated in Table 1. A single value was assigned for an entire plant based on the score of the most severely infected leaf at the moment of scoring.

Table 1.   Visual assessment scoring scale for Cercospora leaf spot on sugar beet leaves
Disease classDescription
0No necrotic spots
5Isolated spots occupying 1–5% of the leaf surface
10Confluent necrotic spots occupying about 5–10% of the leaf surface
20Blighted area occupies about 10–20% of the leaf surface
40Blighted area occupies about 20–40% of the leaf surface
60Blighted area occupies about 40–60% of the leaf surface
80Blighted area occupies about 60–80% of the leaf surface
100Leaf completely blighted

Disease quantification using image analysis software

Pictures of leaves were taken with a Canon EOS 350D in a closed box (50 × 50 × 100 cm). The focal distance was about 60 cm, representing about 600 cm2 area to lay out the leaves. Images were stored in high resolution-JPEG format and handled as described by Steddom et al. (2005b). All images were analysed using the Assess 2·0 image analysis software for plant disease quantification (Assess 2·0; L. Lamari, American Phytopathological Society, St. Paul, MN, USA) as described in the user manual for the software. In short, Assess 2·0 software ignores the background of a digital image and, once the correct leaf/lesion colour adjustments are set, is able to consecutively measure leaf and lesion area and calculate the percentage of infected leaf surface. Thresholds for leaf colour were established in the HSI (hue, saturation and intensity) colour space using saturation values between 80 and 126, separating the leaf area from the background. Then thresholds for lesions were set in the HSI colour space with hue values between 80 and 98. Images were segmented using these threshold values. Percentage area was then calculated as (lesion pixels/leaf pixels) × 100. In this way, a percentage leaf necrosis value was determined for all images under evaluation, as background and lighting were identical for all images to avoid potential bias.

The light quality of the numerical picture is a crucial parameter for a precise determination of the percentage necrotic area. Optimal light conditions were obtained by photographing the leaves under cool white light inside the closed box, using a blue coloured background. The conditions for success depend on the uniform photo quality, the relative position of leaves on the picture and smoothness of the leaf surface (highly susceptible leaves tend to curl). Depending on the homogeneity of the green colour versus the non-green colour and taking into account all mentioned precautions, the area of brown lesions could be determined with high precision.

DNA isolation

Depending on the experiment, either 100–500 mg of leaf tissue (experiments 3 and 4) or leaf disks of equal leaf area (experiment 5) were lyophilized, ground and used to extract DNA. For leaf disks with an equal area, sample material was collected by cutting out three pieces of leaf material twice, at either side of the main vein with a hole punch (15 mm diameter) (Fig. 1). Two DNA isolations were done per leaf, each from three leaf disks.

Figure 1.

 Sampling method for experiment 5. Sample material was collected by cutting out six leaf disks with a hole punch (15 mm diameter). Material was divided into two samples as represented by white and black disks in the figure.

Samples were collected in a deep well plate (National Scientific) and 500 μL of extraction buffer (50 mm Tris pH 8·0, 10 mm EDTA pH 8·0, 750 mm NaCl, 1% w/v CTAB, 0·1%β-mercaptoethanol) was added to each well. After incubation for 30 min at 65°C, 500 μL of chloroform was added and the samples were centrifuged (10 min at 2500 g). Two hundred and fifty microlitres of the upper layer was transferred to a new deep well plate and 500 μL of buffer (50 mm Tris pH 8·0, 1% w/v CTAB) was added. Samples were left for 15 min at room temperature. After centrifugation (10 min at 2500 g), the supernatant was discarded and 500 μL of TE-buffer (10 mm Tris pH 8·0, 1 mm EDTA pH 8·0, 1 m NaCl) was added to dissolve the pellet. After incubation for 10 min at 65°C, 1 mL of ethanol was added and samples were centrifuged (30 min, 2500 g). Supernatant was discarded and the DNA pellet was air-dried and resuspended in 200 μL DNase/RNase free water.

Real-Time PCR

Real-time PCR was performed using the TaqMan probe method. Primers and a probe were designed to specifically detect C. beticola calmodulin (FJ473441; CAL; Table 2). In this study the relative amount of C. beticola was calculated using two different approaches. In the comparative method, a ΔCt value was calculated by subtracting the Ct value of an endogenous sugar beet control from the Ct value of C. beticola calmodulin. A low ΔCt value corresponds to high C. beticola proliferation. Primers and probe for the endogenous sugar beet control (EC) were designed from a sequence that was verified to be highly monomorphic throughout sugar beet lines (unpublished data, Table 2). Both primer pairs were used in a multiplex PCR after optimization of the probe and primer concentration (data not shown). The ΔCt value did not vary in a dilution series up to 25-fold (slope = 0·01) (data not shown) meaning that this comparative method could be used. In a second method, an equal area of plant material was collected with a hole punch and a Ct value was measured. This method was introduced because normalizing pathogen amount to the amount of plant DNA can lead to an overestimation of pathogen biomass because necrosis of plant material can lead to DNA degradation of the host cells. For each PCR reaction, samples contained 2 μL DNA extract, 0·05 μL of CAL primers (75 μm), 0·5 μL CAL probe (10 μm), 0·05 μL EC primers (50 μm), 0·5 μL EC probe (5 μm), 6 μL of Taqman Universal Master Mix (Applied Biosystems) and 2·8 μL water. Forty cycles of amplification (95°C 15 s, 60°C 1 min) were carried out in a 7500 Fast Real-time PCR System instrument (Applied Biosystems).

Table 2.   Primers and probes used during this study
Primer/probe nameSequence (5’→3’)
CercoCal1-FCAAGGACGGCGATGGTATG
CercoCal1-RTTCGGTCGAGGTTAGTTCAGTACA
CercoCal1-probeTGTGCGCCCACCCTCTGCG
SbEc1-FACTTGCCTGGCTTTTGTTTCTAGT
SbEc1-RGCCAGGTGCTGACTTGATTATTT
SbEc1-probeCCCARAGCATTTTTCCAGTGCTCATATTGA

Experimental design

The different genotypes (inbred lines) used in this study were selected from SESVanderHave breeding programmes. Their discriminative levels of CLS symptom expression (from resistant to highly susceptible) were acknowledged from observations in field and greenhouse conditions (natural and artificial inoculation; data not shown).

Experiment 1

Comparison of visual assessment versus Assess-based disease quantification of CLS disease on sugar beet leaves. Leaves with different levels of symptom development (corresponding to disease classes, see Table 1) were photographed and the amount of symptoms was measured using Assess 2·0. Three to six plants per disease class were selected 21 days post-inoculation (dpi) with C. beticola.

Experiment 2

Optimization of visual scoring based on digital image analysis. New detailed disease sheets were made by taking, for each disease class, multiple pictures of leaves with similar Assess 2·0 scores but different symptom patterns (e.g. multiple necrotic spots versus larger necrotic zones). The main objective of the sheets is a pre-training of the eye to optimize visual symptom assessment.

Experiment 3

Comparison between CLS symptoms and C. beticola proliferation. During this experiment a total of 34 plants from resistant, susceptible and highly susceptible genotypes (selected from SESVanderHave breeding programmes) were used. From each plant, the leaf with the highest degree of disease symptoms (leaf with the most symptoms, LMS) after 21 dpi with C. beticola was collected for both digital image analysis (Assess 2·0) and C. beticola proliferation assays. The amount of C. beticola DNA was determined using multiplex real-time PCR and calculated according to the comparative ΔCt method.

Experiment 4

Determination of C. beticola proliferation at an early stage of infection. At four time points (4, 7, 10 and 13 dpi with C. beticola) three leaves (the three last appearing leaf pairs) from 10 plants of a highly susceptible and a resistant genotype (selected from SESVanderHave breeding programmes) were collected and C. beticola DNA amount was determined using multiplex real-time PCR and calculated according to the comparative ΔCt method.

Experiment 5

Comparison of the three methods: real-time PCR, visual assessment and digital image analysis in a CLS disease quantification assay. In order to test all methods in one experiment during current sugar beet breeding programmes at SESVanderHave, the leaf with the most symptoms from 20 plants of two resistant, two susceptible and two highly susceptible sugar beet genotypes was analysed 13 dpi with C. beticola. In a preliminary bioassay, resistance levels of the different genotypes were checked by measuring the percentage necrosis of the LMS by using Assess 2·0 after 21 dpi (Table 3). Leaves were first scored visually, then photographed and analysed with Assess 2·0. Cercospora beticola proliferation was determined on leaf disks (see DNA isolation) by real-time PCR using both the comparative ΔCt method and the Ct method normalized against equal leaf area.

Table 3.   Sugar beet lines, selected from a sugar beet breeding programme, with different degrees of resistance determined by Assess-based disease quantification
Sugar beet lines% necrosis (Assess 2·0)
Resistant A (RA)7·1 ± 0·7
Resistant B (RB)12 ± 2·6
Susceptible A (SA)56·5 ± 2·6
Susceptible B (SB)57·7 ± 5·4
Highly susceptible A (HS A)72·6 ± 2·2
Highly susceptible A (HS B)82·5 ± 2·2

Statistical analysis

Statistical significance was determined at a probability level of 95% by Student’s t-test (experiment 1, 2 and 4) or analysis of variance (anova) followed by Student–Newman–Keuls multiple comparison test (experiment 5). Statistical analysis was performed using GenStat Release 12·1.

Results

Correlation between visual assessment and digital image analysis in CLS quantification assays (experiments 1 and 2)

In experiment 1, classical visual assessment, routinely applied in breeding programmes, was compared with the use of digital image analysis software (Assess 2·0; Lamari, 2002). The data (Fig. 2a) revealed a clear correlation between the visual assessment data and the digital image analysis data (from herein called Assess-based) (R2 = 0·973). However, a slight discrepancy between both disease quantification methods was noticed and is based on an overestimation of the infected leaf surface by visual assessment (Fig. 2a). As such, the latter appears less accurate compared to the Assess-based analysis, especially for leaves of moderate disease class (for leaves of disease classes 20, 40 and 60, P-values equal 0·015, 0·022 and 0·048, respectively). This overestimation is potentially caused by differences in disease stages of the CLS on the scored leaves. The infected area of two leaves showing different stages of infections was quantified with both visual assessment and digital image analysis (Fig. 3). Although both leaves were classified in the 40–60% necrosis class as established with visual assessment, Assess-based disease quantification showed that the leaf with the isolated necrotic spots was assigned to the wrong disease class, since only 36·1% (20–40% class) of the leaf consisted of necrotic spots.

Figure 2.

 Comparison of visual assessment versus Assess-based disease quantification of Cercospora leaf spot on sugar beet leaves. (a) Leaves with different levels of infection (corresponding to disease classes, see Table 1) based on a visual assessment were quantified using digital image analysis (‘Assess’). x-axis represents the different disease classes. Bars represent the mean (±SE) for = 3–6. Bars with an * are significantly different from visual data by Student’s t-test at < 0·05. (b) Identical procedure as in (a), except that an Assess-based ‘disease sheet’ (with pictures of a range of leaves with a relevant degree of necrosis) was used to visually assess the percentage necrotic leaf area. x-axis represents the different disease classes. Bars represent the mean (±SE) for = 6–11.

Figure 3.

 Illustration of the effect of the Cercospora leaf spot disease pattern on disease quantification using visual assessment versus Assess-based analysis.

Because the data indicated a potential for overestimation of disease levels by visual scoring (Figs 2a & 3), corrected and more detailed disease sheets were developed for use in field experiments (experiment 2). Applying these optimized disease sheets gave a more fine-tuned disease assessment, and the discrepancy between this Assess-assisted visual disease assessment and the Assess-based quantification of leaf spot disease was removed (Fig. 2b). Moreover, more disease classes could be introduced resulting in more refined analysis.

Correlation between CLS symptoms and C. beticola proliferation (experiments 3 and 4)

In order to be able to easily detect and quantify C. beticola pathogen biomass, as well as overcome a potential discrepancy between CLS symptoms and pathogen biomass present as reported earlier for other pathogens (Bent et al., 1992; Hoffman et al., 1999; Thomma et al., 1999), a C. beticola-specific real-time PCR was developed and evaluated for its use in resistance breeding. Primers and a probe for amplification of the C. beticola calmodulin gene (Groenewald et al., 2005) were designed in order to specifically quantify the amount of C. beticola DNA in infected leaves of sugar beet. The related species Mycosphaerella fijiensis (synonym C. fijiensis) and C. apii could not be detected with the calmodulin specific primers (data not shown). In order to verify if a correlation exists between CLS symptoms and proliferation of C. beticola, real-time PCR data were analysed by the comparative ΔCt method and revealed that CLS symptoms and the relative C. beticola DNA amount correlated in the samples (R2 = 0·836; < 0·05) (Fig. 4).

Figure 4.

 Correlation between Cercospora leaf spot (CLS) symptoms and C. beticola proliferation. CLS symptoms were scored using digital image analysis (Assess 2·0) on the leaf showing most symptoms of 8-week-old sugar beet plants (= 34), 21 days post-inoculation (dpi) with C. beticola. Cercospora beticola proliferation was determined using multiplex real-time PCR and calculated according to the comparative ΔCt method where the Ct value of an endogenous sugar beet control is subtracted from the Ct value of C. beticola calmodulin. A low ΔCt value corresponds to high C. beticola proliferation.

Subsequently, the aim was to test if the real-time PCR method could be used to distinguish between susceptible and resistant plants in an early stage of disease development, at which time symptom development has not progressed enough to allow visual analysis. In addition, less necrotized tissue will not suffer from aberrant PCR kinetics (Bhat & Browne, 2010). To test this, leaves of C. beticola resistant and highly susceptible sugar beet plants were collected at different time points. At 10 dpi the first symptoms started to appear but only in the susceptible line (data not shown). However, at 4 dpi a statistically significant difference in ΔCt could already be detected between susceptible and resistant plants which increased drastically upon appearance of the first symptoms (Fig. 5).

Figure 5.

Cercospora beticola proliferation on sugar beet leaves, from a susceptible and a resistant genotype, at different time points after inoculation. Cercospora beticola proliferation was determined using multiplex real-time PCR and calculated according to the comparative ΔCt method where the Ct value of an endogenous sugar beet control is subtracted from the Ct value of C. beticola calmodulin. A low ΔCt value corresponds to high C. beticola proliferation. Analysis was done 4, 7, 10 and 13 days post- inoculation (dpi). At each time point the three last developed leaf pairs were collected and combined. Data points represent the mean (±SE) for = 10. Data points with an * are significantly different by Student’s t-test at < 0·05.

Combining real-time PCR, visual assessment and digital image analysis for CLS quantification in sugar beet breeding programmes (experiment 5)

To prove that both digital image analysis and real-time PCR can be used to select sugar beet lines with a different degree of resistance at an early stage of disease development, an experiment was set up with six different sugar beet lines with a different degree of susceptibility towards C. beticola. At 13 dpi, analysis of both symptom development (at all tested levels of resistance) and pathogen proliferation (using real-time PCR) was possible. The percentage necrotic leaf area was significantly different between the various categories of genotypes used (resistant, susceptible and highly susceptible, Fig. 6a). This means that differences in resistance can be seen at an early stage of infection. A clear correlation (R2 = 0·877; < 0·05) was found between the average percentage of necrosis determined using Assess 2·0 and visual methods for the different lines (Fig. 6a). However, overestimation still occurred when scoring visually. In addition, also based on the real-time PCR data (Fig. 6b), a distinction could be made between resistant, susceptible and highly susceptible genotypes. This resulted in strong correlations between ΔCt values and visual (R2 = 0·9658; < 0·05) or Assess 2·0 (R2 = 0·9945; < 0·05) scored data (Fig. 6c). In these real-time PCR experiments, proliferation of the pathogen was measured as total fungal DNA in samples of equal amounts of plant tissue, based on equal amounts of plant DNA. However, since tissue necrosis can lead to degradation of plant DNA, total fungal DNA in plant samples was also determined. Figure 6d indicates a clear correlation between Ct values and ΔCt values (R2 = 0·9598; < 0·05), at least at these infection levels.

Figure 6.

 Comparison of the three methods: real-time PCR, visual assessment and digital image analysis in a Cercospora leaf spot quantification assay with sugar beet genotypes of different sensitivity to C. beticola. Six sugar beet genotypes (= 20) with different susceptibility levels (R, resistant; S, susceptible; HS, highly susceptible; Table 3) were used in this assay. At 13 days post-inoculation (dpi) with C. beticola the leaves showing most symptoms were first scored visually, then photographed and analysed with Assess 2·0. Sample material for real-time PCR was collected as in Figure 1 such that real-time PCR could be performed twice on the same leaf. (a) Comparison of visual assessment versus Assess-based disease quantification. Bars with the same letter (upper case for visual, lower case for Assess) are not significantly different by anova followed by Student–Newman–Keuls multiple comparison test at < 0·05. (b) Cercospora beticola proliferation was determined using multiplex real-time PCR and calculated according to the comparative ΔCt method (see legend Fig. 5). Bars with the same letter are not significantly different by anova followed by Student–Newman–Keuls multiple comparison test at < 0·05. (c) Correlation between C. beticola proliferation, visual assessment and Assess-based disease quantification. The percentage of necrosis as determined via visual and Assess 2·0 scoring was expressed in log10-values since most of the leaves still had a very low percentage of necrosis. (d) Correlation between ΔCt and Ct values. Ct values represent the amount of C. beticola DNA without taking into account the amount of plant DNA.

Discussion

Disease evaluations are required for a number of agronomic studies. Visual assessments tend to be the most common as they are fast and easy to perform. However, several studies have shown that visual evaluations suffer from poor precision, accuracy and reproducibility (Sherwood et al., 1983; Shokes et al., 1987; Nutter et al., 1993). Some of these factors are linked to subjective scoring and represent an error factor for the trial. This study demonstrates the use of digital image analysis software and real-time PCR for quick and reliable assessment of C. beticola-induced disease quantification and pathogen colonization, respectively.

The implementation of Assess 2·0 in the CLS analyses described in this study ensures high accuracy and reliability of large-scale visual disease assessment (Figs 2 & 6). A clear correlation was detected between visual assessment and digital image analysis. However, visual scoring was shown to overestimate the necrotic leaf area in some cases, which was also noticed for other plant-pathogen studies (reviewed in Bock et al., 2010). This overestimation is probably due to the different disease stages of CLS (Fig. 3). The fungus typically enters the leaf through stomata followed by initial colonization of the tissue, causing local tissue collapse and several small necrotic lesions. These lesions can expand and coalesce to cover large areas of the leaf and eventually cause senescence of the complete leaf (Feindt et al., 1981; Weiland & Koch, 2004).

Additionally, accurate visual disease estimation is dependent on the rater’s experience, lesion size relative to infected area, time taken to assess disease etc. (reviewed in Bock et al., 2010). In contrast, the digital image analysis is more objective, and currently used on a regular basis in breeding programmes (e.g. at SESVanderHave for screening of CLS resistant lines) and to detect QTL markers specific for C. beticola resistance in greenhouse bioassays. However, since digital image analysis remains relatively labour-intensive it does not yet replace the fast and easy classical visual assessments in the field, especially not on large numbers of field plots. Therefore, disease sheets were developed with pictures of leaves where the symptoms were scored with Assess 2·0 and the use of these sheets was shown to improve accuracy of CLS disease assessments considerably. These pictures can be used by raters both as a preliminary training as well as during actual disease scoring in the field (Bock et al., 2010).

Furthermore, this study describes the development of a real-time PCR application to accurately and specifically assess C. beticola pathogen colonization. Several studies demonstrated the successful use of quantitative or real-time PCR for detection of pathogen biomass on different plant species (Böhm et al., 1999; Weller et al., 2000; Lees et al., 2002; Winton et al., 2002; Brouwer et al., 2003; McCartney et al., 2003; Yan et al., 2008). Previously, primers based on the sequences of the internal transcribed spacers (ITS) in the ribosomal DNA or actin were used to detect C. beticola in infected plant tissues or field soils (Weiland & Sundsbak, 2000; Lartey et al., 2003, 2010). However, Groenewald et al. (2005) proved that neither sequences could distinguish between C. beticola and the morphologically identical species C. apii. Moreover, primers of the ITS region were also able to detect other related species such as M. fijiensis (synonym C. fijiensis; data not shown). By using primers and a probe for C. beticola calmodulin (Groenewald et al., 2005), C. beticola could be specifically quantified and the data correlated well with both visual assessment and digital image analysis (Figs 4 & 6). As deduced from Figure 5, real-time PCR can be used at a very early stage of infection, even before pronounced symptoms appear, to discriminate lines with different resistance levels. Real-time PCR at an early stage of infection will not suffer from aberrant PCR kinetics because no extracts from very heavily necrotized tissues were used (Bhat & Browne, 2010). At this early stage a nice correlation between ΔCt and Ct values exists but one can expect that this correlation will decrease in heavily infected leaf tissue with more plant DNA degradation.

Real-time PCR may be especially useful in breeding programmes to reduce the potential bias in selection procedures for CLS tolerance. In contrast to real-time PCR, the accuracy of symptom assessments can suffer from non-disease related symptoms caused by senescence and environmental conditions. Overall, the use of both Assess-based image analysis and C. beticola-specific real-time PCR has the potential to increase accuracy and sensitivity of assessments of CLS of sugar beet, while reducing bias in the evaluations. This will be of particular importance in sugar beet breeding programmes, since C. beticola tolerance is recognized as multigenic (Setiawan et al., 2000; Weiland & Koch, 2004), requiring extreme sensitivity of disease assessment.

Acknowledgements

This research was partially financed by the Institute for the Promotion of Innovation by Science and Technology in Flanders (IWT) (project IWT-030437). BDC acknowledges the receipt of a postdoctoral grant from this institute.

Ancillary