SEARCH

SEARCH BY CITATION

Keywords:

  • Allium cepa;
  • disease management;
  • epidemiology;
  • integrated disease management

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Botrytis leaf blight (BLB) caused by Botrytis squamosa is a major leaf disease of onion. Various forecasting systems have been developed to help growers manage the disease. To improve forecasting reliability, the influence of temperature and wetness duration on B. squamosa infection was quantified by inoculating onion leaves with a conidial suspension and incubating them under various combinations of temperature (10–30°C) and leaf wetness duration (0–84 h). Infection was measured as the number of lesions per cm2 of leaf and converted to the proportion of maximum infection (PMI). Regardless of leaf wetness duration, only a few lesions developed at 30°C and the number of lesions increased as the temperature rose from 10 to 20°C but decreased at 25°C. Between 10 and 25°C the number of lesions per cm2 of leaf area increased gradually with increasing leaf wetness duration from 12 to 72 h. Relative infection was modelled as a function of both temperature and wetness duration using a modified version of the Weibull equation, which provided a precise description of the response of B. squamosa (R2 = 0·88). To facilitate field validation, receiving operating characteristic curve analysis was performed to determine the accuracy of various sets of criteria for establishing the length of an infection event based on field weather data. The total number of leaf wetness and RH >90% hours over a 72 h period was the best criterion, regardless of the wetness interruption pattern (sensitivity = 90·91, specificity = 84·62, area under the receiving operating curve = 0·878). The model describing the relationship of PMI to temperature and leaf wetness duration, and field observations on airborne conidium concentration (ACC) were used to calculate the risk of infection (RIBLB) as RIBLB = PMI × ACC. In 2009 and 2010, this risk index was compared to the observed rate of BLB progress (RateBLB+5 days) during the following 5 days. There was a linear relationship between RIBLB and RateBLB+5 days indicating that this new risk indicator was reliable for predicting the risk of BLB development. These findings will help to improve the timing of fungicide applications for BLB management.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Allium crops, including bulb onions (Allium cepa), are important vegetables worldwide. In Canada, a total area of 6118 ha was seeded with onion in 2009 (FAOSTAT, 2010), with 1200 ha of this total being in the muck soil area southwest of Montreal. In this area botrytis leaf blight of onion (BLB), caused by Botrytis squamosa, is the most problematic field onion disease (Hancock & Lorbeer, 1963; Carisse et al., 2011). Several species of Botrytis are known to infect onion leaves and their prevalence varies with the region of production (Yohalem et al., 2003). However, in Canada B. squamosa is largely predominant (Carisse et al., 2011). Botrytis squamosa causes two types of symptoms on onion leaves. During the leaf spotting phase of the disease, B. squamosa causes small, discrete greyish to white lesions that are slightly depressed in the centre. During the leaf blight phase, B. squamosa causes leaf dieback that begins at the leaf tip and progresses downward. In eastern Canada, the disease is endemic; however, depending on disease severity and time of onset, BLB causes various levels of reduction in bulb size and reduced storability mainly associated with reduced absorption of sprout inhibitors applied to onion foliage near harvest (Sutton et al., 1986). Most cultivars that correspond to market demand in terms of bulb size and storability are susceptible to the disease; hence BLB is managed through frequent applications of fungicides (Carisse et al., 2011). In some years, up to 16 fungicide sprays may be required to keep the disease under the economic threshold.

In the 1970s and 1980s, in Ontario and Quebec, Canada as well as in New York, Michigan and Ohio in the United States, the disease posed a serious threat to commercial onion crops. Conventional fungicide spray programmes for BLB management at that time were based on the application of fungicides every 7–10 days from the 3–4 leaf stage until shortly after onion lodging (Shoemaker & Lorbeer, 1971; Lorbeer & Ellerbrock, 1976). Even though the disease is present every year, the severity of BLB varies significantly from year to year. Therefore, fixed-interval fungicide spray programmes have resulted in unnecessary fungicide applications during years with weather less favourable for disease development (Sutton et al., 1986). In view of the threat posed by BLB, the routine use of fungicides and the risk of development of fungicide resistance, research programmes on BLB epidemiology and management were initiated in Ontario, Michigan and New York (Shoemaker & Lorbeer, 1977a; Swanton, 1977; Sutton et al., 1978, 1983; Alderman & Lacy, 1983, 1984a,b; Lacy & Pontius, 1983; Alderman et al., 1985; Vincelli & Lorbeer, 1988a,b). These research programmes led to the development of several BLB forecasting systems and management tools (Lacy & Pontius, 1983; Sutton et al., 1986; Vincelli & Lorbeer, 1989). Carisse et al. (2011) recently reviewed the existing BLB management tools. In brief, two tactics have been proposed to manage BLB. First, monitoring- or weather-based indicators have been developed to determine the best time to initiate a fixed-interval fungicide spray programme (Shoemaker & Lorbeer, 1977b; Lorbeer & Jares, 1981; Sutton et al., 1986; Carisse et al., 2005). Secondly, weather-based indicators have been developed to determine the optimum interval between fungicide sprays (Lacy & Pontius, 1983; Vincelli & Lorbeer, 1989).

Recently, the accuracy of these BLB predictors was examined (Carisse et al., 2008a). In that study, BLB risk indicators based on field monitoring of lesion density or airborne conidial concentrations were found to be the most reliable. Among the weather-based indicators, those predicting the risk of BLB based on sporulation potential were more reliable than those based on infection potential (Carisse et al., 2008a). Of the seven BLB risk indicators evaluated in that study, two included an infection component, specifically the botcast and blight-alert forecasting systems (Sutton et al., 1986; Vincelli & Lorbeer, 1989). In botcast, hourly temperature, relative humidity, leaf wetness and rainfall data are used to calculate a daily infection value, which represents the favourability of weather conditions for infection. Daily infection values are expressed as categorical data, with 0, 1 and 2 representing weather that is unfavourable for infection, favourable for light infection or favourable for severe infection, respectively (Sutton et al., 1986). Similarly, in blight-alert, temperature and leaf wetness duration are used to categorize leaf wetness episodes as not conducive to infection, or as conducive to light, moderate or severe infection (Vincelli & Lorbeer, 1989). In their comparison of BLB risk indicators, Carisse et al. (2008a) reported that these two forecasting systems were less reliable for predicting the risk of BLB being above the damage threshold of one and five lesions on the youngest and oldest green leaves, respectively. This could be explained at least in part by the incomplete knowledge available at the time these tools were developed regarding the combined effect of a wide range of temperatures and leaf wetness durations on lesion development. Several authors have investigated the influence of temperature and/or leaf wetness on infection of onion leaves by B. squamosa (Shoemaker & Lorbeer, 1977a; Swanton, 1977; McDonald, 1981; Tanner & Sutton, 1981; Ramsey & Lorbeer, 1986). Those studies reported an optimum temperature of 18°C (McDonald, 1981; Tanner & Sutton, 1981) and 20°C (Shoemaker & Lorbeer, 1977a; Swanton, 1977) for infection. At temperatures between 9 and 25°C, the number of lesions increases with increasing temperature and leaf wetness duration; however, conflicting reports have been published on the optimum leaf wetness duration. Minimum leaf wetness requirements of 5, 6, 9 and 12 h at 18–20°C have been reported, while optimum leaf wetness durations of 12, 18 or 48 h have also been reported (Shoemaker & Lorbeer, 1977a; Swanton, 1977; McDonald, 1981; Tanner & Sutton, 1981). Alderman & Lacy (1983) reported that dry spores sprayed onto dry onion leaves survived for up to 2 days in the absence of leaf wetness and kept their potential to infect onion leaves that subsequently became wet. The same authors reported that the infection level was low at 15°C, optimal at 20°C, and significantly reduced at 25°C; infection was initiated after 6 h of leaf wetness, and the number of lesions increased sigmoidally over 32 h of leaf wetness. When the leaf wetness period was interrupted, the number of lesions was reduced in proportion to the length of the dry period, and as few as 0·3–1·7 h of dryness was sufficient to reduce the number of lesions produced (Alderman et al., 1985).

In the onion production area of the province of Quebec, scouting services are widely available and information on BLB lesion density, airborne inoculum concentration (ACC) and sporulation index values (SIV) (Lacy & Pontius, 1983) was collected weekly, three times weekly and daily, respectively. Sporulation-based BLB indicators (ACC and SIV) are used to time both the initiation of the fungicide spray programme and the interval between sprays (Van der Heyden et al., 2012). Lesion density is mainly used to ensure that the previous sprays were efficient at keeping the disease below the damage threshold and to determine which fungicide should be used. In other words, when BLB is below the lower threshold of one and five lesions on the youngest and oldest green leaves, respectively, the use of contact fungicides is recommended. When BLB is at or above the upper threshold of 5 and 10 lesions on the youngest and oldest green leaves, respectively, the use of translaminar fungicides is recommended. This approach has proven efficient for managing BLB (Carisse et al., 2008a; Van der Heyden et al., 2012). However, because BLB risks are mainly estimated from either measured (ACC) or forecast (SIV) inoculum, it is possible that sprays are advised when the weather is not favourable for infection and hence lesions do not develop. Considering the increasing cost of fungicides, their adverse impact on the environment and on farm worker health, the risk of development of fungicide resistance and consumer concern about pesticide residues, it is important to avoid recommending unnecessary fungicide applications. In addition, knowledge about the proportion of airborne conidia that can cause lesions under different conditions can assist in interpreting both AIC and SIV values. Therefore, the objectives of this study were threefold: (i) to model the relationship between leaf wetness duration, temperature and lesion density for conidial infection of onion leaves by B. squamosa; (ii) to evaluate the reliability of various criteria derived from information in the scientific literature for use in calculating the length of the infection period in order to detect infection events; and (iii) to formulate and validate a new BLB risk indicator based on the infection efficiency of airborne conidia of B. squamosa.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Modelling of the relationship between leaf wetness duration, temperature and lesion density

Inoculum production

A mixture of eight single conidial isolates of B. squamosa, obtained from naturally infected onion leaves collected in 2004 at the Agriculture and Agri-Food Canada experimental farm in Ste-Clothilde, Quebec, was used for inoculation. Single conidial isolates were prepared using a modified version of the method described by Tremblay et al. (2003). Conidia from single lesions were collected using a sterile cotton-tipped applicator and placed in a sterile 1·5 mL tube containing 300 μL of a sterile glycerol solution (5%) amended with novobiocin (100 μg mL−1; Sigma) and tetracycline (50 μg mL−1; Sigma) to prevent bacterial contamination. The spore suspension was agitated and poured onto a 90 mm diameter Petri dish containing water agar amended with novobiocin (100 μg mL−1) and tetracycline (50 μg mL−1), and incubated at room temperature for 6 h. Pieces of agar containing only one spore were cut and placed on a Petri dish containing potato dextrose agar (PDA; Difco Laboratories). The plates were incubated in the dark at 18·5°C until mature sclerotia were produced, generally within 4–6 weeks. Sclerotia were harvested, washed in a 1% sodium hypochlorite solution, and rinsed in sterile distilled water prior to incubation under UV lights (Sylvania F20T12/BLB, 310–420 nm) as previously described (Tremblay et al., 2003). Inoculum was prepared by soaking sporulating sclerotia in a 15 mL tube containing 10–12 mL of a 5% glycerol solution. The stock inoculum suspensions for each isolate were pooled together and agitated for 2 min, and then filtered through two layers of cheesecloth. The concentration of the inoculum suspensions was adjusted to 7·5 × 104 conidia mL−1 with a haemocytometer. Aliquots of at least 40 mL were prepared in 50 mL tubes with each conidial suspension and stored at −20°C for up to 18 months until required for the inoculation. Because a different batch of conidial suspensions was used for each inoculation, germination of the inoculum was estimated for each batch by plating three droplets of 20 μL each onto 15 PDA plates and incubating them at room temperature for 48 h. Germination on each plate was estimated by observing 50 randomly selected conidia and counting the number of conidia with germ tubes at least half their length.

Plant production

Onion seedlings (cv. Bastille) were produced in bedding plant trays (36 plugs). Each hole in the tray was filled with 40 g of the general purpose peat-based growing medium Promix BX (Premier Horticulture Inc.). Seeded bedding plant trays were incubated for 4–6 weeks at 20°C and 70% RH in a growth chamber (Conviron model E15; Controlled Environments Limited) with a 16 h photoperiod provided by fluorescent and incandescent light bulbs placed 90 cm above the seedlings, until the onions reached the three leaf stage or the cotyledon senescence stage. The plants were transplanted into 9 cm pots and maintained under the same conditions until inoculation (Carisse & Tremblay, 2007).

Inoculation procedures

For each inoculation, 45 onion seedlings were placed in a transparent plastic tower 67 cm high and 155 mm in diameter. The seedlings were inoculated by placing the nozzle of an airbrush atomizer (Phantom 100 Createx Colors) upright to a perforated plastic cover placed at the top of the inoculating tower. Each seedling was inoculated with the spore suspension at 172·4 kPa using an airbrush compressor (Phantom 100 Createx Colors). Immediately after inoculation, the onion plants were placed in a growth chamber (Conviron model E15) with a 10 h photoperiod and the perforated covers were replaced with non-perforated ones to maintain continuous leaf surface wetness. After a post-inoculation wetness period of 0, 8, 12, 24, 36, 48, 60, 72 or 84 h, five plants were transferred to a growth chamber maintained at 18°C and 60% RH. This procedure was repeated for temperatures in the growth chamber of 10, 15, 20, 25 and 30 ± 1°C. To minimize variation, the order in which plants were inoculated, the temperature selected, and the distribution of plants in the growth chamber were randomized for the eight leaf wetness durations. Typical lesions of B. squamosa were counted on the first (oldest) and second leaf. Height and diameter at the base of each leaf were measured using an electronic caliper. Leaf area was calculated using the formula to calculate the surface area of a cone. The experiment was set up as a split-plot experiment with temperature as the main plot and leaf wetness duration as a sub-plot, using five replicates (five plants). The entire experiment was conducted three times.

Data analysis

Within each experiment, the mean number of lesions per cm2 of leaf was converted to a proportion of maximum infection (PMI) by dividing the mean number of lesions per cm2 leaf area by the maximum infection level obtained from any of the temperature–leaf wetness duration combinations in one experimental run. Effects of temperature and leaf wetness duration on mean number of lesions per cm2 were analysed using a linear mixed model with the repetitions as random effects. All statistical analyses, including modelling procedures (proc nlin), were performed using the Statistical Analysis System (SAS) program (sas v. 9.2; SAS Institute Inc.).

Model development

To model the combined effects of temperature and leaf wetness duration on PMI, a modified version of the Weibull equation was fitted to the data according to the methodology proposed by Carisse et al. (2000) using Eqn 1.

  • image(1)

where PMI is the proportion of maximum infection, lw is the leaf wetness duration in hours, A is the upper asymptote, B is the intrinsic rate of increase in PMI, C is the length of delay in the response of PMI to increasing leaf wetness duration, and D is the portion of the period of leaf wetness in which PMI decelerates. First, the upper asymptote (A) was established by expressing observed maximum PMI as a function of temperature (T):

  • image(2)
  • image(3)

where A is the upper asymptote (maximum PMI) at a given temperature (T), E is the maximum response, F is a location parameter proportional to the optimum temperature, G is the intrinsic rate of decline from the maximum as the temperature departs from the optimum, and H is the degree of asymmetry of the curve. To estimate the value of parameters E, F, G and H, partial derivative equations were calculated and the Marquardt iterative method of the non-linear procedure in sas was used (proc nlin). Because PMI responded immediately to an increase in leaf wetness duration >8 h, the parameter C in Eqn 1 was set to 8. To estimate the value of parameters B and D, partial derivative equations were calculated and the Marquardt iterative method of the non-linear (nlin) regression procedure in sas was used. To predict PMI as a function of both temperature (T) and leaf wetness duration (lw), Eqn 2, which predicts the asymptote (A), estimated value of B (mean overall temperatures), the fixed value for C (C = 8) and estimated value of D (mean overall temperatures), was integrated into Eqn 1. The model was evaluated by performing a simple regression analysis between the predicted values and the corresponding observations as paired data (goodness of fit between predicted and observed values). Analyses of the linear regressions were done by testing whether the intercept and the slope were significantly different from 0 and 1, respectively.

Reliability of criteria used to calculate the length of infection periods in order to detect infection events

Data were collected from 2005 to 2008 at the Agriculture and Agri-Food Canada experimental farm located in Sherrington County, about 50 km southwest of Montreal. In the first week of May, a 15 × 15 m plot was established by sowing onion (cv. Tribute) at a density of 30 seeds per metre within rows, with row spacing of 0·36 m. A pre-emergence herbicide (Prowl at 7·5 L ha−1, Pendimethalin 400 g L−1; BASF Canada Inc.) was applied, and weeding was done four to five times by hand during the course of the experiments. The plots were treated with insecticide when necessary. The number of lesions per leaf was counted on 10 plants per plot selected at random three times weekly from mid-June to harvest, for a total of 36 sampling days per field (12 weeks × 3 sampling days per week). Weather data were obtained using a CR-21X automatic weather station (Campbell Scientific Inc.) placed in the onion fields. Air temperature, relative humidity, rainfall and leaf wetness data were monitored every 15 min and hourly averages or totals were used in the analysis.

Data analysis

In this study, as in most studies on the influence of leaf wetness duration on B. squamosa infection conducted under controlled conditions, optimum infection was observed after a long leaf wetness period, generally >48 h. However in the field, such a long leaf wetness period is rare. In practice, leaf wetness periods are interrupted by dry and/or high relative humidity periods. Various criteria to calculate the length of infection periods were derived from published results of experiments on the influence of interrupted leaf period, initial wet period, and relative humidity on infection of onion leaves by B. squamosa (Alderman & Lacy, 1983, 1984a,b; Alderman et al., 1985). The following sets of criteria were evaluated for their reliability in identifying infection events: SET 1: a minimum of 6 h of leaf wetness is required to trigger infection, leaf wetness interruptions of 2 h or less do not affect infection and are considered leaf wetness hours, provided that the interruption is followed by at least 6 h of leaf wetness; a dry period >2 h ends the infection period; SET 2: same as SET 1, except that RH >90% is considered equivalent to leaf wetness, and a dry period >2 h ends the infection period; SET 3: same as SET 2, except that interruptions of 6 h or less are considered not to affect infection and counted as leaf wetness hours, provided that the total leaf wetness or RH >90% period is at least 24 h; a dry period >6 h ends the infection period; SET 4: same as SET 3, except that interruptions of >6 but <12 h are considered to affect infection and are not counted as leaf wetness hours, provided that the total leaf wetness or RH >90% period is at least 24 h; a dry period >12 h ends the infection period. SET 5: daily cumulative number of wet or RH >90% hours during the next 72 h (regardless of interruption patterns). For each infection period, the duration and mean temperature during the period was calculated using the five sets of criteria and relative infection was estimated using Eqn 5 (see Results). An infection was considered to occur when the predicted PMI was >0·15. The 144 (4 years × 36 sampling days annually) samplings were divided into two groups based on observed BLB lesion density. ‘Cases’ were defined as sampling days for which increases in the number of lesions per plant were observed over a 5 day period and the ‘controls’ were defined as the sampling days with no increases in number of lesions per plant. Receiving operating curve (ROC) analysis was performed to evaluate the ability of each set of criteria to calculate the duration of the infection period in order to detect infection periods. For each set of criteria, the true positive proportion (TPP) was calculated by dividing the number of true positives (BLB increases observed) by the total number of cases; the true negative proportion (TNP, specificity) was calculated by dividing the number of true negatives (BLB increases not observed) by the total number of controls; the false positive proportion (FPP) was calculated as 1 – TNP and the false negative proportion (FNP) as 1 – TPP (Yuen & Hughes, 2002; Hughes & Madden, 2003; Carisse et al., 2008a). The area under the ROC curve (AUC) was used to evaluate the accuracy of each set of criteria. A set of criteria that provides no discrimination between disease increase or no disease increase will have an AUC equal to 0·5 (TPP = FPP). The non-parametric approach of DeLong et al. (1988) was used to compare the AUC for the different sets of criteria. The statistical package ROCtools Beta.1 (Institut Philippe-Pinel de Montréal) was used to produce the ROC curves and to calculate the AUC.

Correlation between risk of infection and rate of disease progress under field conditions

The experiment was conducted at the Agriculture and Agri-Food Canada experimental farm in Ste-Clotilde, located southwest of Montreal. The experiment was conducted in 2009 and 2010 with the onion cultivar Infinity, which is considered susceptible. Both years, on 6 May, a 10 × 10 m plot was established by sowing onions at a density of 30 seeds per metre within rows, with row spacing of 0·36 m. The plots were not treated with insecticides unless necessary and weeding was done by hand four times during the course of the experiments. The airborne concentration (ACC) of B. squamosa conidia was measured in the centre of each plot using rotating-arm impaction spore samplers placed 10–15 cm above the canopy. The samplers were set to run from 10·00 to 12·00 h, three times a week on Mondays, Wednesdays and Fridays. The number of B. squamosa conidia per rod was estimated using a TaqMan PCR assay (Carisse et al., 2009). Spore counts were converted to concentrations of conidia per cubic metre (number of spores per rod × 1000 L m−3)/(20·65 L min−1 rod × 60 min h−1 × 2 h). On each sampling day, the number of lesions per leaf was counted on 10 plants per plot, selected at random within the plot.

Weather data were obtained from the Environment Canada weather station located at the experimental farm and used to run the model predicting PMI (Eqn. 5). For each spore sampling day, PMI was calculated using weather data for the following 3 days. It was assumed that the viability of airborne conidia was 100, 50% and 25% on the day of sampling, and one and 2 days later, respectively (Alderman et al., 1985). The risk of infection (RIBLB) was calculated as ACC × PMI and compared with the rate of BLB development (RateBLB) calculated as (lesions t2 − lesions t1)/(t2 − t1), where t1 and t2 are the day of year of two consecutive BLB assessments and lesion t1 and lesion t2 are the corresponding mean number of lesions per plant. The relationship between RIBLB and RateBLB was established by performing a simple regression analysis between calculated RIBLB values and the corresponding observed RateBLBt+5 (value obtained 5 days later) as paired data (goodness of fit between predicted and observed values). An analysis of the linear regression was performed by testing whether the intercept and the slope were significantly different from 0 and 1, respectively.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Modelling of the relationship between leaf wetness duration, temperature and lesion density

Regardless of the temperature, no lesions developed at a leaf wetness duration shorter than 8 h (Fig. 1). At 30°C, the number of lesions per cm2 of leaf area remained near zero until leaf wetness reached 60 h, then increased slowly to reach a maximum of 0·15 lesions per cm2 of leaf (Fig. 1). The pattern of lesion density progress with increasing leaf wetness duration was similar at temperatures from 10–25°C. The number of lesions per cm2 of onion leaf area increased rapidly from 12–72 h of leaf wetness and slowly from 72–96 h of leaf wetness (Fig. 1). When the numbers of lesions per cm2 of leaf area were converted to proportion of maximum infection, the asymptote (maximum PMI, parameter A in eq. 1) was well described by Eqn 2 (Fig. 2). The asymptote was highest at 20°C with a value of 1·0 and lowest at 30°C with a mean value of 0·09 (Fig. 2). The response of the upper asymptote to temperature was well described by the following Eqn (Fig. 3):

  • image(4)
  • image(5)
image

Figure 1.  Number of lesions per cm2 of onion leaf area caused by Botrytis squamosa on leaves exposed to various temperatures and leaf wetness durations. The lines represent the mean values (average over three experimental repetitions).

Download figure to PowerPoint

image

Figure 2.  Response of the upper asymptote (A) to temperature in the model used to describe the proportion of maximum infection (PMI) as a function of temperature (T) and leaf wetness duration (lw) as: inline image The line represents the predicted value calculated using equations 4 and 5, and the symbols represent the observed values for each experimental repetition.

Download figure to PowerPoint

image

Figure 3.  Three-dimensional response of relative infection of onion leaves by Botrytis squamosa to the combined effects of temperature and leaf wetness duration. The predicted values of the proportion of maximum infection (PMI) were calculated using equations 4, 5 and 6.

Download figure to PowerPoint

Parameter estimates and their associated errors are presented in Table 1. The regression of the predicted values against the observed values was significant (< 0·0001, R2 = 0·96). The intercept was −0·0295, which was not significantly different from 0 (P = 0·3758), and the slope was 1·066, which was not significantly different from 1 (< 0·2248). The surface response of PMI to the combined effects of temperature and leaf wetness duration was obtained by integrating Eqn 4, describing the response of the asymptote (parameter A) to temperature, into Eqn 2 with = 8. Estimates of parameters B and D, along with their associated errors, are presented in Table 1. The three-dimensional representation of the response of Eqn 6 is given in Figure 3 and was calculated as:

  • image(6)
Table 1.   Parameter estimates and associated statistics for the Weibull equation used to model the response of the proportion of maximum infection (PMI) of onion leaves by Botrytis squamosa to combined effects of temperature (T) and leaf wetness duration (lw) as inline image, where inline image, and inline image
ParameteraEstimateAsymptotic standard errorAsymptotic 95% confidence interval
LowerUpper
  1. aParameters E, F, G and H were used in Eqns 4 and 5 and parameters B and D in Eqn 6 (see text).

E  1·0010·0235 0·9687 1·0334
F 21·0451·108218·605823·4841
G  0·49540·0421 0·4027 0·5881
H  2·15290·7866 1·4215 3·8843
B  0·0260·0012 0·0252 0·0270
D  1·9990·0874 1·9015 2·0984

where A was calculated from Eqns 4 and 5. The model provided a good prediction of relative infection for all leaf wetness durations (Fig. 3). The intercept of the regression of the predicted values against the observed PMI, equal to 0·017, was significantly different from 0 (= 0·1068) with a slope of 0·9421, which was not significantly different from 1 (= 0·0641), and the coefficient of determination was 0·88.

Reliability of criteria used to calculate the length of infection periods in order to detect infection events

Of the 144 sampling days analysed, 45·83% of days were classified as cases (disease increase observed). The sensitivity and specificity of the five sets of criteria varied from 19·70% to 90·91% and from 88·46% to 80·77%, respectively. The set of criteria with the lowest sensitivity was SET 1 with 19·70. As the criteria included less restrictive conditions, such as RH >90% equivalent to leaf wetness or a longer dry period during the wetness interruption period, the sensitivity increased from 31·82% to 81·82%. However, the most accurate set of criteria was SET 5, which considers the total number of hours of leaf wetness or RH >90% regardless of the interruption pattern, with a sensitivity of 90·91%. Similarly, the area under the ROC curves (AUC) varied between 0·541 and 0·878 and was different from the no discrimination line (AUC = 0·5) for all sets of criteria except SET 1 (AUC = 0·541) (Table 1). From the ROC analysis, it was concluded that SET 5, which considered the total number of hours of leaf wetness or RH >90% over a 72 h period, was the most accurate set of criteria for the detection of infection periods (sensitivity = 90·91, specificity = 84·62, AUC = 0·878).

Correlation between risk of infection and rate of disease progress under field conditions

For the 2 years of the experiment, BLB progress exhibited an initial phase from mid-May to the third week of July (133–200) with slow disease progress followed by a phase during which the disease progressed rapidly (Figs 4a and 5a). During the first phase, the rate of BLB (RateBLB) varied between 0 and 1·5 lesions per day, whereas during the phase of rapid disease progress, the rate varied between 5 and 300 lesions per day per plant (Figs 4a and 5a). Similarly, airborne conidium concentration (ACC) was low during the first phase with concentrations ranging from 0 to 191 conidia m−3 and high during the phase of rapid disease progress with concentrations ranging from 114 to 7416 conidia m−3 (data not shown). Overall the weather was less favourable for BLB development in 2010 than in 2009 mainly because of periods of dry weather unfavourable for infection of onion leaves by B. squamosa. This was reflected in the temporal pattern of PMI (Figs 4b and 5b). During the onion growing period from mid-May (emergence) to mid-August (onion lodging), the predictions were as follows for 2009 and 2010, respectively: 10 and 27 days without infection; 54 and 40 days with light infection (PMI < 0·15); 21 and 14 days with moderate infection (PMI 0·15–0·30); and 10 and 14 days with severe infection (PMI > 0·30) (Figs 4b and 5b). The difference in weather favourability for BLB is further illustrated by the relative risk of infection, which is the product of the predicted PMI and ACC based on an incubation period of 5 days, which was lower in 2010 than in 2009 (Figs 4a and 5a). The rate of botrytis leaf blight development was near 0 when the risk of infection was <1·5, and then increased linearly with increasing RI (Fig. 6). Hence, the linear regression of RateBLB+5 days against RIBLB was significant (< 0·0001) with an intercept of 1·7272, which was significantly different from 0 (= 0·5758), and a slope of 0·3119, which was significantly different from 1 (< 0·0001).

image

Figure 4.  Rate of botrytis leaf blight progress (RateBLB+5 days), observed number of lesions per plant and risk of botrytis leaf blight infection (RIBLB) (a). Predicted proportion of maximum infection (PPMI) calculated from the mean temperature during leaf wetness periods (calculated as the total number of leaf wetness, or RH >90%, hours over the following 72 h period) in 2009 (b).

Download figure to PowerPoint

image

Figure 5.  Rate of botrytis leaf blight progress (RateBLB+5 days), observed number of lesions per plant and risk of botrytis leaf blight infection (RIBLB) (a). Predicted proportion of maximum infection (PPMI) calculated from the mean temperature during leaf wetness periods (calculated as the total number of leaf wetness, or RH >90%, hours over the following 72 h period) in 2010 (b).

Download figure to PowerPoint

image

Figure 6.  Regression of the observed rate of botrytis leaf blight development. (RateBLB+5 days) against calculated risk of botrytis leaf blight infection (RIBLB).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Understanding the factors that trigger the development of botrytis leaf blight epidemics is essential in order to create and implement effective management strategies. This has motivated a large body of research addressing the effects of weather conditions on botrytis leaf blight development in several countries where the disease is prevalent. This research has given rise to monitoring- and weather-based control systems aimed at rationalizing onion protection (Shoemaker & Lorbeer, 1977b; Lorbeer & Jares, 1981; Lacy & Pontius, 1983; Sutton et al., 1986; Vincelli & Lorbeer, 1989; Carisse et al., 2005). Recent reports on evaluations of these management systems undertaken in the Netherlands, France and Canada indicate that the estimation of infection is the weakest part of these systems (De Visser, 1996; Huchette et al., 2005; Carisse et al., 2008a). Hence, in this study the objective was to refine the prediction of infection based on weather conditions. The first experiment was conducted under controlled conditions and consisted of studying the influence of a wide range of temperatures and continuous leaf wetness durations on lesion production. Regardless of the temperature, no lesions developed when the leaf wetness duration was 8 h or less. At 30°C the density of lesions remained low, regardless of the length of the leaf wetness period. At temperatures between 10 and 15°C, lesion density gradually increased with increasing leaf wetness duration (starting at 8 h) to reach a maximum level at leaf wetness of 72 h. These results are in agreement with some published results of studies examining the influence of weather conditions on lesion production (McDonald, 1981; Ramsey & Lorbeer, 1986). From the results of this first experiment, a model was created that describes the combined effect of temperature and leaf wetness duration on relative infection of onion leaves by B. squamosa conidia. A modified version of the Weibull model (Carisse et al., 2000) was used to fit a response surface model to observed relative infection data, mainly because of its flexibility in accounting for asymmetry in asymptote values with increasing temperature and because of the shape of the relative infection progress curve over time (here, leaf wetness duration). The model provided a good fit of the data with an R2 of 0·88. Although some infection indexes already exist (Sutton et al., 1986; Vincelli & Lorbeer, 1989), a new index was developed in this study because of the considerable disagreement found between scientific studies on the relationships between temperature, leaf wetness duration and BLB development (Shoemaker & Lorbeer, 1977a,b; Swanton, 1977; McDonald, 1981; Tanner & Sutton, 1981; Ramsey & Lorbeer, 1986). In addition, the previously published infection indexes are categorical, with the severity of infection being expressed as absent, light, moderate or severe. In this study, a continuous scale of 0–1 was used to allow for greater precision in predicting the favourability of weather conditions for infection. Nevertheless, it is challenging to take a predictive model developed from controlled conditions studies and apply it to field situations mainly because, unlike the case with controlled conditions, inoculum, temperature and wetting conditions fluctuate in the field, while leaf area increases over time (Alderman et al., 1987). Leaf wetness is probably the most critical of these factors. There are only a few studies on the influence of interrupted leaf wetness period on infection of onion leaves by B. squamosa in the scientific literature (Alderman et al., 1985). It is difficult to determine the ability of B. squamosa to withstand interruptions in wetness periods during infection, as it is virtually impossible to test all possible combinations of initial wet period, interrupted wetness (dry) period and final wet period. Nevertheless, Alderman et al. (1985) reported that an initial 6 h of onion leaf wetness was required to initiate infection and that a 2 h period of wetness interruption did not reduce lesion production; however, fewer lesions were produced as the duration of the interrupted wetness period increased from 2 to 24 h. In this study, an empirical approach was taken and the accuracy of several sets of criteria taken from the scientific literature was evaluated (McDonald, 1981; Alderman et al., 1985; Sutton et al., 1986) using receiving operating curve analysis. To select the best sets of criteria, sensitivity, specificity and area under the ROC curves were considered. Sensitivity represents the proportion of true positives, that is, the proportion of sampling days where the disease increases and for which an infection period is detected; specificity represents the opposite situation, that is, the proportion of sampling days where disease increase is not observed and infection is not detected. Finally, the area under the ROC curve represents the accuracy of the set of criteria for detecting both disease increase (cases) and no disease increase (controls). The best set of criteria was SET 5, which considered the total number of hours of leaf wetness or RH >90% during a 72 h period regardless of the wetness interruption pattern. Using these criteria, 90·91% of the sampling days characterized by an increase in disease intensity 5 days later were detected, while 84·62% of the sampling days characterized by no increase in disease intensity 5 days later were detected.

The use of an infection index alone to estimate risk of BLB may produce misleading results unless information about the abundance of inoculum is also taken into account. Therefore, the third part of this study consisted of correlating the risk of infection (RI) based on the amount of airborne inoculum (ACC) and the favourability of weather conditions for infection (PPMI) with the observed rate of BLB progress. A linear relationship was found between the rate of BLB progress and the risk of infection (R2 = 0·90). Results showed that even early in the season, when onion leaves are young and highly susceptible to B. squamosa infections and weather is favourable to infection, the scarcity of inoculum is responsible for the low increase in lesion density (Figs 4a and 5a). The BLB risk indicator developed in this study is based on a monitoring-based component and a weather-based component: airborne inoculum concentration and proportion of maximum infection, respectively. In the area southwest of Montreal, a scouting network was implemented more than 25 years ago. Growers who are members of the network are advised about the risk of BLB based on the critical disease level determined from scouting of individual fields. This approach has contributed to improved BLB management (Carisse et al., 2011); however, warnings based on critical disease levels come ‘after the fact’. In other words, lesion density is an indication of existing infection, which cannot be controlled with most fungicides registered for BLB management. Most of them are protectants and hence should be applied before inoculum deposition on onion leaves. To circumvent this problem, the risk indicators used by the scouting network was expanded to include B. squamosa airborne inoculum monitoring (Carisse et al., 2008a,b; Van der Heyden et al., 2012). Since 2008, information on airborne inoculum concentrations has been available to onion growers belonging to the scouting group. Nonetheless, it was felt that information on how favourable weather conditions are for this inoculum in terms of ‘causing lesions on onion leaves’ would help in interpreting the airborne inoculum information. In this study, field observations showed that even though airborne inoculum is present, if the weather conditions are not favourable, lesions will not be produced. Therefore, current airborne inoculum concentrations and forecast weather conditions can be used to estimate the imminent risk of BLB infection and the corresponding need for fungicide sprays. In areas where real-time information on airborne inoculum concentration is not available, the new infection risk index (PPMI) could be used in conjunction with sporulation index values (SIV), such as the index proposed by Lacy & Pontius (1983). Nevertheless, additional research is needed to validate the new BLB risk indicator developed in this study as a tool for timing fungicide sprays for management of the disease.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The authors are grateful to Daniel Rolland, Annie Lefebvre, Mathieu Tremblay, Jean-Francois Desteredjian and Nadia Surdek, as well as all the summer students and the scouts from the PRISME consortium for their technical assistance and spore quantification. This work was financially supported by Agriculture and Agri-Food Canada. The contribution of H. Van der Heyden to this work was partly supported by the Compagnie de Recherche Phytodata Inc.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Alderman SC, Lacy ML, 1983. Influence of dew period and temperature on infection of onion leaves by dry conidia of Botrytis squamosa. Phytopathology 73, 10203.
  • Alderman SC, Lacy ML, 1984a. Influence of leaf position and maturity on development of Botrytis squamosa in onion leaves. Phytopathology 74, 14613.
  • Alderman SC, Lacy ML, 1984b. Influence of temperature and moisture on growth and sporulation of Botrytis squamosa. Canadian Journal of Botany 62, 27937.
  • Alderman SC, Lacy ML, Everts KL, 1985. Influence of interruptions of dew period on numbers of lesions produced on onion by Botrytis squamosa. Phytopathology 75, 80811.
  • Alderman SC, Biernbaum JA, Hoffhines M et al. , 1987. A simulation model of the spread of Botrytis leaf blight of onion and its effect on onion yields. Agriculture Ecosystem and Environment 19, 5570.
  • Carisse O, Tremblay DM, 2007. Incidence and significance of iprodione-insensitive isolates of Botrytis squamosa. Plant Disease 91, 416.
  • Carisse O, Bourgeois G, Duthie JA, 2000. Influence of temperature and leaf wetness duration on infection of strawberry leaves by Mycosphaerella fragariae. Phytopathology 90, 11205.
  • Carisse O, McCartney HA, Gagnon JA, Brodeur L, 2005. Quantification of airborne inoculum as an aid in the management of leaf blight of onion caused by Botrytis squamosa. Plant Disease 89, 72633.
  • Carisse O, McRoberts N, Brodeur L, 2008a. Comparison of monitoring- and weather-based risk indicators of botrytis leaf blight of onion and determination of action thresholds. Canadian Journal of Plant Pathology 30, 44256.
  • Carisse O, Savary S, Willocquet L, 2008b. Spatiotemporal relationships between disease development and airborne inoculum in unmanaged and managed Botrytis leaf blight epidemics. Phytopathology 98, 3844.
  • Carisse O, Tremblay DM, Lévesque CA, Gindro K, Ward P, Houde A, 2009. Development of a TaqMan real-time PCR assay for quantification of airborne conidia of Botrytis squamosa and management of Botrytis leaf blight of onion. Phytopathology 99, 127380.
  • Carisse O, Tremblay DM, McDonald MR, Brodeur L, McRoberts N, 2011. Management of Botrytis leaf blight of onion: the Quebec experience of 20 years of continual improvement. Plant Disease 95, 50414.
  • De Visser CLM, 1996. Field evaluation of a supervised control system for Botrytis leaf blight in spring sown onions in the Netherlands. European Journal of Plant Pathology 102, 795805.
  • DeLong ER, DeLong DM, Clarke-Pearson DL, 1988. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 44, 83745.
  • FAOSTAT, 2010. FAO available at: [http://faostat.fao.org/site/339/default.aspx].
  • Hancock JG, Lorbeer JW, 1963. Pathogenesis of Botrytis cinerea, B. squamosa, and B. alli on onion leaves. Phytopathology 53, 66973.
  • Huchette O, Guerber-Cahuzac B, Brouant L, 2005. Validation of forecasting systems for downy mildew and Botrytis squamosa in spring sown onion in France. Acta Horticulturae 688, 2018.
  • Hughes G, Madden LV, 2003. Evaluating predictive models with application in regulatory policy for invasive weeds. Agricultural Systems 76, 75574.
  • Lacy ML, Pontius GA, 1983. Prediction of weather-mediated release of conidia of Botrytis squamosa from onion leaves in the fields. Phytopathology 73, 6706.
  • Lorbeer JW, Ellerbrock LA, 1976. Failure of ethylene bisdithiocarbamates to control Botrytis leaf blight of onion. Proceedings of the American Phytopathological Society 3, 7584.
  • Lorbeer JW, Jares TW, 1981. Field sampling patterns for determining the critical disease level for initiating fungicidal control of Botrytis leaf blight of onion. Phytopathology 71, 238. (Abstract).
  • McDonald MR, 1981. Effect of Environmental and Host Factors on Botrytis Leaf Blight of Onions. Guelph, ON, Canada: University of Guelph, MSc thesis.
  • Ramsey GR, Lorbeer JW, 1986. The role of temperature and moisture in onion flower blight. Phytopathology 76, 6126.
  • Shoemaker PB, Lorbeer JW, 1971. Spray volume, interval, and fungicide rate for control of Botrytis leaf blight of onion. Plant Disease Reporter 55, 5659.
  • Shoemaker PB, Lorbeer JW, 1977a. The role of dew and temperature in the epidemiology of Botrytis leaf blight of onions. Phytopathology 67, 126772.
  • Shoemaker PB, Lorbeer JW, 1977b. Timing initial fungicide application to control Botrytis squamosa leaf blight epidemics on onions. Phytopathology 67, 40914.
  • Sutton JC, Swanton CJ, Gillespie TJ, 1978. Relation of weather variables and host factors to incidence of airborne spores of Botrytis squamosa. Canadian Journal of Botany 56, 24609.
  • Sutton JC, James TDW, Rowell PM, 1983. Relation of weather and host factors to an epidemic of Botrytis leaf blight in onions. Canadian Journal of Plant Pathology 5, 25665.
  • Sutton JC, James TDW, Rowell PM, 1986. botcast: a forecasting system to time the initial fungicide spray for managing Botrytis leaf blight of onions. Agriculture Ecosystem and Environment 18, 12343.
  • Swanton CJ, 1977. Influence of Environmental Factors on Development and Control of Botrytis Leaf Blight of Onions. Guelph, ON, Canada: University of Guelph, MSc. thesis.
  • Tanner MR, Sutton JC, 1981. Effect of leaf wetness duration and temperature on infection of onion leaves by Botrytis squamosa. Phytopathology 71, 565. (Abstract).
  • Tremblay DM, Talbot BG, Carisse O, 2003. Sensitivity of Botrytis squamosa to different classes of fungicides. Plant Disease 87, 5738.
  • Van der Heyden H, Carisse O, Brodeur L, 2012. Comparison of monitoring based indicators for initiating fungicide spray programs to control Botrytis leaf blight of onion. Crop Protection 33, 218.
  • Vincelli PC, Lorbeer JW, 1988a. Forecasting spore episodes of Botrytis squamosa in commercial onion fields in New York. Phytopathology 78, 96670.
  • Vincelli PC, Lorbeer JW, 1988b. Relationship of precipitation probability to infection potential of Botrytis squamosa on onion. Phytopathology 78, 107882.
  • Vincelli PC, Lorbeer JW, 1989. blight-alert: a weather-based predictive system for timing fungicide applications on onion before infection periods of Botrytis squamosa. Phytopathology 79, 4938.
  • Yohalem DS, Nielsen K, Nicolaisen M, 2003. Taxonomic and nomenclatural clarification of the onion neck rotting Botrytis species. Mycotaxon 85, 17582.
  • Yuen J, Hughes G, 2002. Bayesian analysis of plant disease prediction. Plant Pathology 51, 40712.