Fungicide efficacy and timing for the management of Stemphylium vesicarium on onion

Abstract

Stemphylium leaf blight (SLB), caused by Stemphylium vesicarium (Wallr.) E.G. Simmons, has become an important disease of onion (Allium cepa L.) in Ontario, Canada, and the northeastern USA in recent years. The disease presents as elongated lesions on the leaves and severe leaf dieback. The effect on yield is unclear, but the extensive leaf dieback limits uptake of sprout inhibitors that are applied to onion foliage prior to harvest. This can result in high losses in storage. There are no resistant commercial onion cultivars and growers apply foliar fungicides at 7–14-day intervals to manage the disease. Field trials to evaluate fungicide efficacy and disease-forecasting models were conducted at the Muck Crops Research Station, Holland Marsh, Ontario, from 2011 to 2019. Fungicide efficacy declined over the years. The disease-forecasting models reduced the number of fungicide spray applications, but none of the models, including calendar-based applications, reduced SLB severity. Seed treatments containing penflufen, combined with calendar-based fungicide applications, reduced SLB severity. Assessments of fungicide insensitivity in isolates collected locally in 2018 and 2019 demonstrated that 90% of isolates (n = 48) were insensitive to azoxystrobin (a QoI fungicide, FRAC 11) and 57% (n = 47) were insensitive to pyrimethanil (an AP fungicide, FRAC 9). Both fungicides are used extensively on onion in the Holland Marsh but are no longer effective against the population of S. vesicarium that is present in the region. Additional studies on the efficacy of seed treatments, fungicide insensitivity, and potential of biological fungicides are required.

Résumé

La brûlure stemphylienne (BS), causée par Stemphylium vesicarium (Wallr.) E. G. Simmons, est devenue, au cours des dernières années, une importante maladie de l’oignon (Allium cepa L.) en Ontario, au Canada, et dans le nord-est des États-Unis. La maladie se manifeste par des lésions allongées sur les feuilles et un grave dépérissement de ces dernières. L’effet sur le rendement n’est pas clair, mais le dépérissement général limite l’absorption des antigermes appliqués sur feuillage de l’oignon juste avant la récolte. Cela peut engendrer de lourdes pertes durant l’entreposage. Il n’existe pas de cultivars commerciaux d’oignon résistants et, afin de gérer la maladie, les producteurs appliquent des fongicides foliaires à intervalle de 7 à 14 jours. Des essais au champ ont été menés, de 2011 à 2019, à la Station de recherche sur la culture de terres noires d’Holland Marsh, en Ontario, pour évaluer l’efficacité des fongicides et des modèles de prédiction de la maladie. L’efficacité des fongicides a décru au fil des années. Les modèles de prédiction de la maladie ont permis de réduire le nombre d’applications de fongicide par pulvérisation, mais aucun des modèles, y compris celui des applications établies en fonction du calendrier, n’a pu réduire la gravité de la BS. Des traitements de semences à base de penflufen, combinés aux applications établies en fonction du calendrier, ont réduit la gravité de la BS. Des évaluations relatives à l’insensibilité aux fongicides chez des isolats collectés localement en 2018 et 2019 ont démontré que 90% des isolats (n = 48) étaient insensibles à l’azoxystrobine (un fongicide QoI, FRAC 11) et que 57% (n = 47) étaient insensibles au pyriméthanile (un fongicide AP, FRAC 9). Les deux fongicides sont utilisés extensivement sur l’oignon à Holland Marsh, mais ne sont plus efficaces contre les populations de S. vesicarium de la région. D’autres études sur l’efficacité des traitements de semences, l’insensibilité aux fongicides et de possibles fongicides biologiques sont nécessaires.

Keywords:

• fungicide sensitivity
• onion
• Pleospora
• Stemphylium leaf blight

Mots clés:

• brûlure stemphylienne
• oignon
• Pleospora
• sensibilité aux fongicides

Introduction

Stemphylium leaf blight (SLB) is caused by the fungal pathogen Stemphylium vesicarium (Wallr.) E.G. Simmons (teleomorph: Pleospora herbarum [Pers.] Rabenh., syn P. allii), which is an important foliar disease of onion (Allium cepa L.) and garlic (A. sativum L.) in many countries around the world (Gupta et al. 1994; Hassan et al. 2007; Mishra and Singh 2017). SLB can lead to premature defoliation of the crop, resulting in loss of yield. Field trials in New York demonstrated yield losses of 28–38% (Hoepting 2018a, 2018b) and up to 74% premature plant mortality when disease pressure is high (Hoepting 2017a). This pathogen also causes purple spot and Stemphylium leaf spot of asparagus and brown spot of pear (Lacy 1982; Hausbeck et al. 1999; Singh et al. 1999). SLB was first noted on onion in Ontario, Canada in 2008 (Paibomesai et al. 2012). None of the onion cultivars commercially grown in Ontario are resistant to S. vesicarium (Tayviah 2017).

Regular application of preventative or curative fungicides is an important tool for management of diseases caused by fungal pathogens (Llorente et al. 2012), especially where genetic resistance is not available. Onion growers often use a calendar-based method to apply fungicides, weekly or bi-weekly, to manage foliar diseases, in the absence of disease-forecasting information. Onion growers in the Holland Marsh, Ontario typically begin fungicide applications for SLB when the disease is reported in the local area (Tayviah 2017). This method does not depend on weather conditions or knowledge of the biology of the pathogen, and can result in more applications than necessary (Llorente et al. 2012). Applying more fungicide sprays than needed is not economical, and also increases the risk that fungicide insensitivity will develop in the pathogen population (Alberoni et al. 2010b). Fewer applications than necessary could result in unacceptable levels of disease and loss of yield. Forecasting models use environmental factors to identify conditions that are conducive to disease development and to recommend when growers should apply pesticides. A disease-forecasting model should indicate the optimum number and timing of fungicide applications to manage the disease without compromising crop yield or quality.

Effective forecasting models can provide the same level of disease suppression as calendar-based methods that call for a greater number of fungicide applications. For instance, FAST (Forecasting for Alternaria solani on Tomato), a forecast model developed for A. solani Sorauer on tomatoes (Solanum lycopersicum L.), achieved comparable disease management for brown spot of pear (BSP) as the 7-day calendar spray, with 28% fewer applications (Montesinos and Vilardell 1992). TOMcast, a modification of FAST, was developed for management of Septoria leaf spot and anthracnose on tomato caused by Septoria lycopersici Speg. and Colletotrichum coccodes (Wallr.) S. Hughes, respectively (Pitblado 1992). In the TOMcast model, disease severity values (DSVs) are calculated based on leaf wetness duration and the average temperature during the wet period, and fungicide sprays are recommended when cumulative DSVs meet or exceed the set threshold value (Madden et al. 1978). When TOMcast with a DSV threshold of 15 was used for management of purple spot of asparagus, fungicide application was reduced by 60% (Meyer et al. 2000). On some cultivars of asparagus, TOMcast improved the suppression of Stemphylium leaf spot without increasing the number of fungicide applications (Foster and McDonald 2018). The BOTcast (BOTrytis foreCASTer) model was developed for management of botrytis leaf blight of onion, caused by Botrytis squamosa Walker. BOTcast uses similar weather parameters as TOMcast but uses a different combination of temperature and leaf wetness to estimate disease risk based on favourable conditions for infection (Sutton et al. 1986). A combination of daily inoculum value (0–2) and daily infection value (0–2) is used to calculate cumulative disease severity values (CDSI). Fungicide applications are recommended at one of two thresholds; medium risk at 21–30 CDSI, and high risk of disease at 31–40 CDSI. In preliminary trials at the Muck Crop Research Station in Ontario, BOTcast showed potential to reduce the number of spray applications for managing SLB (McDonald and Vander Kooi 2013). A third model, BSPcast (Brown Spot of Pear foreCASTer), was developed specifically for use with S. vesicarium in pear (Pyrus spp. L.) orchards (Montesinos et al. 1995). As with the previous models, this model integrates the effect of daily leaf wetness duration and the temperature during the wetness period to calculate a fungicide application recommendation. BSPcast-guided schedules for application of thiram used 20–50% fewer sprays with the same degree of disease suppression as weekly application on pear trees (Llorente et al. 2010).

Evidence is mounting that the foliar fungicides currently used by producers do not provide economic levels of suppression of SLB on onion in North America (Pethybridge et al. 2016). High concentrations of mancozeb, azoxystrobin, propiconazole, and propineb in vitro inhibited the growth of S. vesicarium (Mishra and Gupta 2012) but had little or no impact on the incidence of SLB in the field (Hoepting and Pethybridge 2016). Fungicides with different modes of action are registered in Canada for management of SLB on onion but growers in the Holland Marsh are often frustrated by a perceived lack of efficacy.

Several studies have reported that populations of S. vesicarium are insensitive to specific fungicides (Alberoni et al. 2005, 2010b; Hay et al. 2019). This is not unexpected; S. vesicarium is at risk for developing fungicide insensitivity because it has a short asexual reproductive cycle, produces multiple generations in a season, and has the ability to produce many air-borne spores through sexual and asexual reproduction (Tayviah 2017). In addition, disease management often involves multiple fungicide applications during each growing season (Misawa and Yasuoka 2012). To date, most publications on fungicide insensitivity of S. vesicarium are focused on brown spot of pear, but the same risk factors are present for SLB on onion.

Several strobilurin fungicides (quinone outside inhibitors, QoI; FRAC group 11) are registered to manage SLB on onion in Canada. A strobilurin-resistant isolate of S. vesicarium was first detected in an Italian pear orchard (Collina et al. 2007) and the same mutation was detected recently in populations of S. vesicarium in New York State (Hay et al. 2019). Also, cross-resistance to two QoI fungicides, azoxystrobin and pyraclostrobin, has been documented (Hay et al. 2019). Similarly, several isolates of S. vesicarium collected in New York State were insensitive to the anilinopyrimidine fungicides (AP; FRAC group 9) cyprodinil and pyrimethanil (Hay et al. 2019). Most of the S. vesicarium isolates collected recently in New York were sensitive to the succinate dehydrogenase inhibitor (SDHI; FRAC group 7) fungicides fluopyram and fluxapyroxad, but over half were insensitive to another SDHI fungicide, boscalid (Hay et al. 2019). Also, the SDHI fungicide carboxin inhibits mycelial growth and conidia germination of a close relative, S. lycopersici (Enjoji) Yamamoto (Rajani et al. 1992). In contrast, isolates from New York were sensitive to the phenylpyrrole (PP; FRAC group 12) fungicide fludioxonil (Hay et al. 2019). Insensitivity to demethylation inhibitor fungicides (DMI; FRAC group 3) has not been found in S. vesicarium. Similar group 3 fungicides difenoconazole and propiconazole provided acceptable control of a closely related pathogen species, Stemphylium solani G.F. Weber on cotton, however this result was not consistent (Mehta and Oliveira 1998). Of the 46 S. vesicarium isolates recently collected in New York, 97.8% were sensitive to the DMI fungicide difenoconazole (Hay et al. 2019). Insensitivity to multi-site fungicides, such as groups M03 and M05, is rare. Recently, a mancozeb-resistant (FRAC group M03) strain of Alternaria alternata (Fr.) Keissl., a pathogen also in the Pleosporaceae family, was identified in southern Greece on tomatoes (Malandrakis et al. 2015). This appears to be the only report of insensitivity to this fungicide group, and the mechanism of insensitivity was not explained. No insensitivity to fungicides in FRAC group M05, such as chlorothalonil, has been reported (FRAG-UK 2016).

For disease-forecasting models to be effective, the fungicides applied must be effective against the target organism. Screening for insensitivity could assist growers in avoiding fungicides that will not be effective and may explain the lack of efficacy of some fungicides in the field. However, there have been no assessments of fungicide sensitivity of S. vesicarium in onion production in Canada.

A research programme was initiated to develop an integrated pest management approach for the management of SLB of onion. The objectives of this research were: (a) to assess fungicides for the management of SLB symptoms in the field, (b) to assess existing disease-forecasting models for management of SLB, (c) to investigate seed treatments as alternative or additional management methods for SLB and (d) to assess the in vitro sensitivity of S. vesicarium to commonly used fungicides. This research focused on yellow bulb onion grown in the Holland Marsh, Ontario, which is characterized by organic muck soil (50–75% organic matter, pH 5.5–7.2), where the majority of Ontario onions are produced.

Materials and methods

Fungicide efficacy trials

Commonly used cultivars of yellow cooking onions (Allium cepa L.) were direct seeded or transplanted in organic soil (organic matter ~65.5%, pH ~7.0) at or near the Muck Crops Research Station, Holland Marsh, Ontario over nine years (2011–2019). Depending on the weather, seeding/transplanting occurred between May 7th and June 1st for each year. The cultivars used were ‘Tahoe’ (Bejo Seeds, Oceano, CA; 2011), ‘Patterson’ (Bejo Seeds; 2012, 2013, 2014, 2015), ‘La Salle’ (Stokes Seeds, Thorold, ON; 2016, 2019), and ‘Ridgeline’ (Stokes Seeds; 2017, 2018). All trials were laid out as a randomized complete block design with four replicates. One plot consisted of one bed 5–6-m long with four rows of onions in 2011–2014, and two side-by-side beds totalling eight rows in 2015–2019. The beds were 42 cm apart, rows within beds were 40 cm apart and the onion transplants were 10 cm apart within the row. In 2011, 2012, 2017, and 2019 the crop was direct seeded ≈ 35 seeds m⁻¹ with the same bed size and spacing as the transplanted trials in early May using a Stanhay precision seeder.

In 2011–2014 the fungicides were applied using a CO₂ backpack sprayer equipped with four flat fan nozzles (XR8002VS, TeeJet, Springfield, PA, USA), 40 cm apart, calibrated to apply 400 L ha⁻¹ at 240 kPa (boom height). In 2015–2019 a tractor-mounted sprayer fitted with TeeJet spray nozzles AI9503-EVS at 415 kPa (2015) or TeeJet D-3 hollow cone nozzles at 620 kPa (2016–2019) and delivering 500 L ha⁻¹. For all fungicide applications, the boom height was maintained at approximately 30 cm above the canopy. The nozzles were changed year-to-year based on spray coverage observations and recommendations from the application technology specialist at the Ontario Ministry of Agriculture, Food and Rural Affairs. The switch to a tractor mounted sprayer was made to more closely represent grower practices. The fungicides tsted and rates used are presented in Table 1 and spray dates for each year in Table S1.

Table 1. Fungicides, active ingredients (a.i.) and rates assessed in field efficacy trials against stemphylium leaf blight of yellow onion at the Muck Crops Research Station, Holland Marsh, Ontario, 2011–2019

Active ingredient	Trade name	A. I. (%)	Rate (ha^−1)	Rate (a.i ha^−1)	FRAC group	Source	Years Tested
Mancozeb	Dithane™	75	3.3 kg	2475 g	M03	Dow Agro-Sciences	2013–2016
Mancozeb	Manzate® 750 F	37	3.3 kg	1221 g	M03	DuPont	2011, 2012
Chlorothalonil	Bravo® 500	40	4.8 kg	1920 g	M05	Syngenta	2011, 2012
Difenoconazole	Inspire®	23	0.51 L	117 mL	3	Syngenta	2011–2015
Penthiopyrad	Fontelis® 20SC	20	1.4 L	280 mL	7	DuPont	2011–2017
Fluxapyroxad	Sercadis®	27	0.33 L	89 mL	7	BASF	2016–2019
Pydiflumetofen	Miravis® Bold^1	18.3	0.38 L	70 mL	7	Syngenta	2016, 2017
Difenoconazole + Benzovindiflupyr	Aprovia® Top	11 + 8	767 mL	84 mL 61 mL	3 + 7	Syngenta	2019
Azoxystrobin + Difenoconazole	Quadris Top®	18 + 11	1.0 L	180 mL 110 mL	11 + 3	Syngenta	2012–2018
Fluopyram + Pyrimethanil	Luna Tranquillity®	11 + 34	1.4 L	476 mL	7 + 9	Bayer CropScience	2011–2019
Boscalid + Pyraclostrobin	Pristine®	25 + 13	1.3 kg	325 g 169 g	7 + 11	BASF	2011–2019
Fluxapyroxad + Pyraclostrobin	Merivon®	21 + 21	0.8 L	168 mL 168 mL	7 + 11	BASF	2017, 2018
Cyprodinil + Fludioxonil	Switch® 62.5WG	38 + 25	0.98 kg	372 g 245 g	9 + 12	Syngenta	2011–2015
Trichoderma atroviride	T-77	2 × 10^9 spores g^−1	0.25 kg	n/a	n/a	Andermatt Biocontrol	2018, 2019

¹Miravis® Bold was previously known as Syngenta AL9649B.

Disease ratings were made based on visual assessments of percent yellowing (chlorosis) of the length of the leaves. The disease rating was modified over time, to improve the rating system, and thus the data cannot be pooled or directly compared across years.

The disease assessment was conducted in mid-August just before the onions were expected to begin lodging. In 2011–2014, 2015, and 2017–2019 a total of 10, 6, and 20 plants from the inside of two rows of each plot, respectively, were harvested and rated for SLB severity. In 2016, onions from a 2-m section of one row were rated (16–38 plants). In 2011–2013, the selected plants were assigned a rating between 0% and 100% based on percentage of foliage visibly affected by SLB and the mean value for each plot was calculated. In 2014–2017, all leaves were removed from each plant and sorted into classes based on the percentage of the leaf length showing symptoms of SLB. The six classes were: 0 = no disease, 1 = 1–10%; 2 = 11–25%; 3 = 26–50%; 4 = 51–75% and 5 = >75% of the leaf with symptoms of SLB. These data were used to calculate a Disease Severity Index (DSI), which is a value that ranges from 0 to 100 (Chester 1950). In 2018 and 2019, disease was assessed on the three oldest leaves and was based on the percentage of leaf length with symptoms. The leaves were given a rating on a 0–4 scale where 0 = no SLB symptoms, 1 = 1–10% of leaf, 2 = 11–20%, 3 = 21–60% and 4 = 61–100% of leaf area with symptoms. The disease severity index in all years where the leaves were grouped into classes was calculated as:

$DSI=\frac{\sum \left[\left(classno.\right)\left(\begin{array}{c}no.ofleaves\\ ineachclass\end{array}\right)\right]\times 100}{\left(\begin{array}{c}totalno.leaves\\ assessed\end{array}\right)\left(no.classes-1\right)}$

Disease forecasting & fungicide timing trials

Trials were carried out over four years at two sites with a history of SLB. In 2015, the trial was conducted at the Jane Street research site, approximately 1 km from the Muck Crops Research Station (MCRS) on an organic soil (organic matter ≈ 62%, pH 7.2). In 2016, 2018, and 2019, the trial was conducted at the MCRS, also on organic soil (organic matter ≈ 70%, pH 5.9). For three years, onion cv. ‘LaSalle’ (Stokes Seeds) was used because it was shown to be susceptible to SLB in preliminary trials. In 2019, this cultivar was not available and cv. ‘Fortress’ (Stokes Seeds, Thorold, ON) was used. In 2015, onion transplants (3 seeds per plug) were transplanted in late May using a mechanical transplanter with four rows per bed. The beds were 42 cm apart, rows were 40 cm apart and the plugs were 10 cm apart within the row. In 2016, 2018, and 2019, the crop was direct seeded (≈ 35 seeds m⁻¹) in early May using a Stanhay precision seeder. There were four double rows per bed, with 40 cm between rows, and the plants were approximately 7.5 cm apart within the row. The rows were 6 m long. Each plot consisted of two adjacent beds.

The treatments included fungicide regimens that growers would typically use with weekly sprays, fungicides applied according to disease-forecasting models TOMcast, BOTcast, STEMcast, and a modified BSPcast, seed treatments containing the active ingredients (a.i.) azoxystrobin and penflufen, and an unsprayed control for comparison (Table 2). The fungicides were applied with a tractor-mounted sprayer, fitted with flat fan TeeJet AI9503 EVS nozzles at 620 kPa calibrated to deliver 500 L ha⁻¹. In 2015, the foliar fungicide used was fluopyram plus pyrimethanil (Luna Tranquillity® applied at 150 g ha⁻¹ fluopyram, 450 g ha⁻¹ pyrimethanil; Bayer Crop Science Inc., Calgary, AB) for every spray. In 2016, mancozeb (Dithane™ applied at 2.5 kg a.i. ha⁻¹; Dow Agro Sciences, Calgary, AB) was applied as an initial protective spray, followed by subsequent applications of fluopyram plus pyrimethanil. In 2018, a treatment was added where a food-grade mineral oil product (Civitas™ applied at 25.5 L ha⁻¹ mineral oil; Suncor Energy Inc., Calgary, AB) was applied as a drench at emergence, followed by weekly calendar sprays starting at the 2-leaf stage. Drench applications were applied to the base of the plants with a CO₂ backpack sprayer equipped with a single Syngenta 65–06 vegetable nozzle calibrated to deliver 1000 L ha⁻¹ along the row. Products containing mineral oil have been registered for use against several foliar diseases of turfgrasses, and this product has demonstrated induced systemic resistance against three diseases of Agrostis stolonifera L. caused by Ascomycete fungi. (Cortes-Barco et al. 2010). In 2018, the foliar sprays consisted of azoxystrobin plus difenoconazole (Quadris Top® applied at 200 g ha⁻¹ azoxystrobin, 125 g ha⁻¹ difenoconazole; Syngenta, Guelph, ON) alternated with fluopyram plus pyrimethanil and in 2019 benzovindiflupyr (Aprovia® applied at 75 g a.i. ha⁻¹; Syngenta, Guelph, ON) was alternated with chlorothalonil (Bravo® Zn applied at 1.8 kg a.i. ha⁻¹; Syngenta, Guelph, ON) in an effort to use fungicides that might be more effective to reduce SLB. The seed treatments used in 2018 and 2019 were EverGol Prime® (2.5 g penflufen kg⁻¹ seed; Bayer Crop Science, Guelph, ON) and Farmore F300 (0.025 g azoxystrobin kg⁻¹ seed, 0.075 g mefenoxam kg⁻¹ seed, 0.0275 g fludioxonil kg⁻¹ seed; Syngenta, Guelph, ON). The seed treatments were applied by Incotec (Salinas, California) using industry-standard procedures. The weekly calendar sprays were applied every 7–10 days (Table S2).

Table 2. Fungicide timing treatments for the field efficacy trials on stemphylium leaf blight (SLB) of yellow onion at the Muck Crops Research Station, Holland Marsh, Ontario, 2015–2019. Values within brackets indicate the number of foliar fungicide applications

	Fungicide timing treatments by year
Treatment	2015	2016	2018	2019
Control	Unsprayed (0)	Unsprayed (0)	Unsprayed (0)	Unsprayed (0)
Weekly, early	Weekly, start when conidia present, crop ≥ 3-leaf stage (10)		Weekly starting at 2-leaf stage (7)	Weekly starting at 2-leaf stage (7)
Weekly, late	Weekly, start after 1^st SLB in region (8)	Weekly, start when conidia present + 4-to 6-leaf stage (4)	Weekly, start at 4-leaf stage (5)
BOTcast	Spray at 21 CDSI, then weekly (8)	Spray at 21 CDSI, then weekly (3)
STEMcast (modified BOTcast)	Spray at 21 CDSI, then weekly (8)
TOMcast 15	TOMcast every 15 DSVs (6)	TOMcast every 15 DSVs (5)	TOMcast every 15 DSVs (5)	TOMcast every 15 DSVs (6)
TOMcast 15 R		TOMcast every 15 DSVs + rainfall predicted (3)
TOMcast 25		TOMcast start at 15 DSVs, spray every 25 DSVs (3)
BSPcast (modified)			Start at emergence (6)	Start at emergence (5)
Mineral Oil			Drench at emergence, then weekly start at 2-leaf stage (7)
Azoxystrobin seed treatment + spray			Seed treated, weekly sprays start at 2-leaf stage (7)	Seed treated, weekly sprays start at 2-leaf stage (7)
Penflufen seed treatment + spray			Seed treated, weekly sprays start at 2-leaf stage (7)	Seed treated, weekly sprays start at 2-leaf stage (7)
Azoxy seed treatment only				Seed treated (0)
Penflufen seed treatment only				Seed treated (0)

Disease severity assessments for the fungicide timing trials were conducted using different methods for 2015–2016 and 2018–2019. Note that these methods do not coincide precisely with the methods used for the fungicide efficacy trials. In 2015 and 2016, the amount of leaf dieback was measured using a clear ruler on the oldest two leaves of 16 plants per plot in 2015, and 20 plants per plot in 2016. The mean percentage of necrotic or chlorotic length of the leaf compared to the total leaf length was calculated for each plot. In 2018 and 2019, a scale of 0 to 4 was used to visually estimate the percent yellowing (chlorotic or necrotic) area of the three oldest leaves for 20 onions per plot: 0 = no yellowing, 1 = 1–10% yellowed, 2 = 11–25% yellowed, 3 = 26–50% yellowed, 4 > 51% yellowed area. The disease severity index was calculated as described above.

Yield assessment

At the end of each growing season, late September to early October, the onions in two 2.3 m sections of row were pulled from the middle rows of each plot for a yield sample. Onions were weighed and sorted into the following size classes based on widest diameter: jumbo (>76 mm), medium (45–76 mm), and small (<45 mm). Onions in the jumbo and medium classes are deemed marketable, and the percentage of marketable onions was calculated for each treatment.

In vitro fungicide sensitivity

Isolates of S. vesicarium were collected from onion and leek in 2018 and 2019 from the Grand Bend Marsh (beside Lake Huron), the Holland Marsh, and Keswick Marsh (south end of Lake Simcoe) areas of Ontario (Table S3). An unexposed baseline isolate was revived from a collection that had been isolated from oats in Saskatchewan in 1995. Germination or mycelial growth on fungicide-amended media was compared to fungicide-free controls to assess sensitivity to azoxystrobin (a strobilurin fungicide, FRAC group 11) and pyrimethanil (an anilinopyrimidine fungicide, FRAC group 9). Sensitivity to azoxystrobin was assessed based on germination of conidia (32 isolates). A discriminatory concentration of 5 µg a.i. mL⁻¹ was set based on previous testing (data not included) and a higher dose of 100 µg a.i. mL⁻¹ was used to capture any isolates that were extremely insensitive. The discriminatory concentration was 10× higher than assessments of S. vesicarium isolates collected in Italian pear orchards (Alberoni et al. 2010a) but is consistent with a recent survey of isolated collected from onion fields in New York State (Hay et al. 2019). Technical-grade azoxystrobin (Sigma-Aldrich, St. Louis, MS) was added to potato dextrose agar (PDA) at rates of 0, 5, or 100 µg a.i. mL⁻¹, and 100 µg mL⁻¹ of salicylhydroxamic acid (SHAM) was added to each rate to inhibit the alternative respiration pathway (Wood and Hollomon 2003). Conidia suspensions were pipetted onto the amended PDA and germination of ~100 conidia was assessed under a light microscope after 24 h. Conidia were considered germinated if the germ tube was greater than the length of the conidium. Sensitivity to pyrimethanil was evaluated using a mycelial growth assay (42 isolates). PDA was amended with technical-grade pyrimethanil (Sigma-Aldrich, St. Louis, MS) at rates of 0, 5, or 100 µg a.i. mL⁻¹ as per Secor and Rivera (2012). The diameter of each colony was marked after 48 h and 144 h. Three replications were used for each isolate.

Data analysis

Statistical analyses were conducted using SAS University Edition (SAS Institute Inc., Cary, NC). Mixed model analysis of variance was used to assess the disease and yield data (PROC GLIMMIX). The normality of each data set was assessed using PROC UNIVARIATE. The Disease Severity Index values and marketable yield values expressed as a percentage were transformed by dividing by 100 for analysis using a beta distribution. Variance was portioned into random (block or replicate) and fixed (treatment) effects. Years were not pooled due to differences in treatments and assessment methods. Pearson’s correlation (PROC CORR, P ≤ 0.05) was used to test for a linear relationship between yield and disease severity for each field season. Percent inhibition caused by in vitro exposure to fungicide active ingredients was calculated as conidia germination or mycelium growth in the control minus germination or growth in the treatment divided by the control × 100. A probit analysis (PROC PROBIT) analysis was used to calculate the EC₅₀ value (effective concentration to cause 50% inhibition) in preliminary experiments, which was used to determine a discriminatory concentration. Isolates were sorted into three classes based on the reaction to the discriminatory concentration of the each fungicide: sensitive (>50% inhibition at 5 µg a.i. mL⁻¹ compared to the control), insensitive (no inhibition at 5 µg a.i. mL⁻¹), or no effect (no inhibition at 100 µg a.i. mL⁻¹). Means were separated using Tukey’s Honest Significant Difference (P ≤ 0.05) to compare treatments.

Results

Fungicide efficacy evaluation

Several fungicides reduced the severity of SLB in the trials in 2011, 2012, and 2013, but none reduced severity compared to the untreated control in subsequent years. In 2011, all fungicides reduced disease and penthiopyrad was more effective than chlorothalonil, mancozeb, and cyprodinil plus fludioxonil (Table 3). In 2012, most fungicides reduced disease severity, with the exception of chlorothalonil and cyprodinil plus fludioxonil. Azoxystrobin plus difenoconazole (12 DSI), fluopyram plus pyrimethanil (13 DSI), and penthiopyrad (19 DSI) had the numerically, but not statistically, lowest disease severity (control = 32 DSI). In 2013, only mancozeb (16 DSI), fluopyram plus pyrimethanil (13 DSI) and azoxystrobin plus difenoconazole (12 DSI) reduced SLB severity (control = 30 DSI). In 2013–2019, none of the fungicides tested reduced severity compared to the untreated control (Table 3). The biocontrol agent T-77 containing the mycoparasitic fungus Trichoderma atroviride was tested in 2018 and 2019; however, the amount of disease severity (2018 = 59 DSI, 2019 = 27 DSI) did not differ relative to the unsprayed control in either year (control = 54 and 33 DSI, respectively) (Table 3).

Table 3. The effect of foliar fungicides on the mean disease severity index (DSI) of stemphylium leaf blight on onion in field trials at the Muck Crop Research Station over nine years

	Disease severity index by year
Treatment	2011		2012		2013		2014		2015		2016		2017			2018		2019
Unsprayed control	67	a^1	32	a	30	a	56	ns^2	96	ns	28	ns	46	ns	54		ns	33	ns
Chlorothalonil	38	b	23	ab	-		-		-		-		-		-			-
Mancozeb	37	b	20	bc	16	b	61		89		35		-		-			-
Cyprodinil + Fludioxonil	37	b	23	ab	20	ab	66		95		-		-		-			-
Difenoconazole	25	bc	17	bc	20	ab	55		88		-		-		-			-
Boscalid + Pyraclostrobin	34	bc	20	bc	19	ab	55		93		30		43		60			31
Fluopyram + Pyrimethanil	21	bc	13	c	13	b	47		88		31		28		52			35
Penthiopyrad	20	c	19	bc	19	ab	59		95		28		49		61			-
Azoxystrobin + Difenoconazole	-		12	c	12	b	50		96		27		37		56			-
Fluxapyroxad	-		-		-		-		-		36		47		59			27
Pydiflumetofen	-		-		-		-		-		36		47		-			-
Fluxapyroxad + Pyraclostrobin	-		-		-		-		-		-		46		56			-
T-77	-		-		-		-		-		-		-		59			27
Difenoconazole + Benzovindiflupyr	-		-		-		-		-		-		-		-			31

¹Means in a column followed by the same letter do not differ based on Tukey’s HDS (P < 0.05).

²ns – not significant.

Yield fluctuated by year, from 21 to 68 t ha⁻¹ but was generally not affected by fungicide treatment (Tables 5, S4 and S5). This variation was likely due to differences in seeding date, use of transplants versus direct-seeded crops, and weather during the growing season. The only exception was 2015, where the unsprayed treatment produced 94% marketable yield and the weekly early treatment had 99%, but total yield (40–47 t ha⁻¹) was not affected (Tables S4 and S5). For the fungicide efficacy trials, there was a significant negative relationship between disease severity and total yield (r = −0.43, 49–61 t ha⁻¹) in 2012, a year with moderate disease severity (unsprayed control = 32 DSI), and between disease severity and percent marketable onions in 2019 (r = −0.43, 98–99% marketable), a year that also had similar moderate disease severity (unsprayed control = 33 DSI). However, there was a significant positive relationship between disease and percent marketable yield in 2015 (r = 0.46, 39–47 t ha⁻¹), which was unexpected as this was a year with high-disease pressure with 96 DSI in the untreated check (Tables 3 and S5). For the disease-forecasting trials, there was a positive relationship between disease severity and total yield in 2019 (r = 0.41, 70–75 t ha⁻¹,) and a negative relationship with percent marketable yield in 2018 (r = −0.41, 87–95% marketable). The impact of disease on yield was inconsistent over the years, which may be due to confounding factors such as weather, the use of transplants or direct seeding, the different cultivars used, and different seeding dates. Weather likely had the greatest effect on the variation in yield from year-to-year. For instance, the stress of high temperatures could reduce yields but not directly influence disease severity. The correlations may also be influenced by the different methods used for disease assessment in different years. Overall, only a small proportion of the variation in yield was explained by variance in SLB severity and there was no relationship between yield and disease severity for most years (Table S5).

In 2011, 2015, 2017 and 2018, SLB severity was high (unsprayed control DSI = 67, 96, 46, 54, respectively), whereas in 2016 and 2019, severity was low (unsprayed control DSI = 28 and 33) (Table 3). The summers of 2016 and 2019 were uncharacteristically dry, with monthly rainfall below the previous 10-year average for most months (May, June, July, August, and September for 2016; June, July and August for 2019 and within average for other months). The years with high SLB severity had higher than average rainfall for at least one month (May and August for 2011; June for 2015, May and June for 2017; August for 2018) (McDonald et al. 2019). This suggests that higher rainfall may increase overall disease pressure.

Disease forecasting & fungicide timing trials

TOMcast with spray applications after every DSV of 15 was the only treatment assessed in all four years of these trials. Weekly sprays that started early (2-leaf stage) or late (4–6 leaf stage) were evaluated in three of the four years. BOTcast, BSPcast and seed treatments plus foliar sprays were assessed in two of the four years. Modifications of TOMcast, seed treatments alone, and the mineral oil drench were only evaluated in one year each (Table 2). In the 2015 trial, all spray regimes reduced SLB severity compared to the untreated check, however, in 2016 only the modified TOMcast systems reduced SLB severity. In 2018, only the seed treatments combined with foliar sprays of fluopyram plus pyrimethanil reduced disease severity and in 2019 only the penflufen seed treatment plus foliar sprays of benzovindiflupyr alternated with chlorothalonil reduced disease severity (Table 4). There were no differences in the weekly spray programmes that began at two true leaves or four true leaves.

Table 4. Effect of fungicide timing treatments on the severity of stemphylium leaf blight (assessed as % dieback or disease severity index) on yellow onion in field trials at the Muck Crop Research Station from 2015–2019

	Dieback (%)				Disease severity index
Treatment^1	2015		2016		2018		2019
Control	65	a^2	43	a	57	a	37	a
Weekly late	43	b	31	ab	46	ab	-
Weekly early	46	b	-		51	a	27	ab
Mineral oil	-		-		42	abc	-
Azoxystrobin seed trt + spray	-		-		32	bc	30	ab
Azoxystrobin seed trt only	-		-		-		34	a
Penflufen seed trt + spray	-		-		28	c	21	b
Penflufen seed trt only	-		-		-		29	ab
STEMcast	46	ab	-		-		-
BOTcast	43	b	27	ab	-		-
BSPcast	-		-		45	ab	28	ab
TOMcast (15)	42	b	30	ab	40	abc	29	ab
TOMcast (15 R)	-		24	b	-		-
TOMcast (25)	-		25	b	-		-

¹The fungicides applied were fluopyram plus pyrimethanil (2015), mancozeb followed by subsequent applications of fluopyram plus pyrimethanil (2016), azoxystrobin plus difenoconazole alternated with fluopyram plus pyrimethanil (2018), and benzovindiflupyr alternated with chlorothalonil (2019).

²Means within a column with the same letter do not differ based on Tukey’s HDS (P < 0.05).

Table 5. Effect of fungicide timing treatments on total yield (t ha⁻¹) and percent marketable yield (%M) in yellow onion in field trials at the Muck Crop Research Station from 2015–2019

	2015		2016		2018		2019
Treatment	Yield	%M	Yield	%M	Yield	%M	Yield	%M
Nonsprayed control	40 ns^2	94 b^1	20 ns	80 ns	40 ns	89 ns	74 ns	99 ns
BOTcast	47	96 ab	25	82	-	-	-	-
Weekly late	44	96 ab	25	87	33	87	-	-
Weekly early	46	99 a	-	-	35	90	73	99
STEMcast	44	96 ab	-	-	-	-	-	-
TOMcast15	42	97 ab	25	80	-	-	-	-
TOMcast 25	-	-	22	79	47	93	75	99
TOMcast 15 R	-	-	25	79	-	-	-	-
BSPcast	-	-	-	-	37	89	70	98
Penflufen seed treatment +spray	-	-	-	-	35	95	72	97
Penflufen seed trt	-	-	-	-	-	-	75	97
Azoxystrobin seed treatment +spray	-	-	-	-	37	91	74	99
Azoxystrobin seed trt	-	-	-	-	-	-	72	99
Mineral oil	-	-	-	-	44	90	-	-

¹Values within column with the same letter do not differ based on Tukey’s HDS (P < 0.05).

²ns – not significant: values do not differ within column.

In three seasons of field trials, the use of TOMcast with a DSV threshold of 15 decreased the number of recommended sprays by one, two, or four applications compared to the early weekly spray programme (2019 = 7, 2018 = 7, 2015 = 10) (Table 2). Most of the other models also either reduced the number of sprays compared to the weekly spray schedule or recommended the same number. The only exception was in 2016 where TOMcast 15 recommended five sprays and the weekly schedule (late, starting at the four to six true leaf stage) recommended four (Table 2). In 2015, TOMcast 15 reduced the number of sprays by four and BOTcast reduced the number by six, and reduced disease severity from 65 in the unsprayed control to 42–43. In 2016, TOMcast 15 R and TOMcast 25 reduced disease severity from 43 in the control to 23–24 and resulted in one fewer spray than the weekly applications. For all four years, SLB severity for the forecasting models did not differ from that in the weekly spray programme (Table 4). However, the disease severity in the weekly spray programs in 2016, 2018, and 2019 did not differ from the unsprayed control, and this lack of response may be linked to a lack of efficacy of the fungicides assessed, as discussed below.

In vitro fungicide sensitivity

The assessment of conidia germination demonstrated that 79% of isolates exhibited insensitivity to 5 µg mL⁻¹ azoxystrobin (FRAC group 11) and 10% were highly insensitive and exhibited no inhibition at 100 µg mL⁻¹ (n = 48). Furthermore, 57% of isolates (n = 47) were found to be insensitive to 5 µg mL⁻¹ pyrimethanil (FRAC group 9) (Table S3). Forty-two isolates were tested for insensitivity to both azoxystrobin and pyrimethanil and 48% were insensitive to both fungicides.

Discussion

Stemphylium leaf blight can result in severe defoliation of onion in Ontario. The management of SLB has traditionally focused on foliar fungicide applications, however, field trials during the past six years have not been able to identify a fungicide that can provide effective disease suppression. Fungicide insensitivity is developing in the field to at least two important fungicide active ingredients, azoxystrobin and pyrimethanil, and is likely associated with the poor performance of fungicides containing these active ingredients.

Disease-forecasting models may reduce the number of fungicide applications by recommending the best time to apply, based on weather parameters. When used correctly, forecasting models can provide the same or better disease suppression as regular calendar-based fungicide application, that is, when spraying starts at the same date or same growth stage each year and continues on a 7–10-day schedule. In this study, the disease severity for six forecasting models assessed over four years was either equal to or less than the weekly spray programme, with fewer fungicide applications for the most part. This demonstrates that disease-forecasting models for SLB can reduce the number of fungicide applications while providing the same or better level of disease suppression. Repeated fungicide applications increase the risk of fungicide insensitivity developing in the pathogen population. Therefore, reducing the number of applications by using disease-forecasting models should extend the useful life of a fungicide in the field.

The SLB severity in 2016 and 2019 was low to moderate, which was likely associated with lower than average seasonal rainfall. However, the disease-forecasting programs still recommended several fungicide applications. The TOMcast15 model recommended five or six applications in each year, and the TOMcast 15 R and TOMcast 25 models recommended only three in 2016. However, the DSI for the unsprayed control was 28 in 2016 and 33 in 2019. This level of severity did not merit the number of fungicide applications recommended, and perhaps no fungicides were needed in these years, especially as there was no reduction in yield compared to the untreated check. There is room for improvement in the disease-forecasting programs to indicate when disease risk is low, and no fungicides are required.

The results of this study differ considerably in both fungicide efficacy and fungicide timing, compared to trials conducted in New York State, even though the onion production systems are quite similar in the two locations. The 2016 and 2017 fungicide efficacy trials on muck soils in New York State included many of the same fungicides as this study, but all fungicides reduced the SLB score compared to the untreated check and fluopyram plus pyrimethanil (Luna Tranquillity) was very effective. Azoxystrobin plus difenoconazole (Quadris Top) was included in the 2016 and 2017 trials and also effectively reduced the SLB score (Hoepting 2017b, 2018b). These fungicide combinations were no longer effective during the same years in this study. Seed treatments provided some reduction in the severity of SLB when combined with foliar fungicide applications. This may indicate that S. vesicarium is a seed-borne pathogen, as suggested by Aveling et al. (1993), or that infection occurs very early in the season, as soon as onion leaf tissue is present. There is a possibility that S. vesicarium can infect the leaf tip of the cotyledon when the onion seed-coat remains attached as the seedling emerged from the soil, which occurs with neck rot of onion (caused by Botrytis allii Munn) (Maude and Presly 1977).

In this study, the seed treatment containing penflufen reduced SLB severity when used in conjunction with weekly foliar sprays. Penflufen is a new active ingredient in the SDHI group (FRAC group 7) that was registered for use on onion in Ontario in 2019. The in vitro sensitivity of S. vesicarium to penflufen remains to be assessed. The role of early infection of the onion cotyledon or first or second true leaf by airborne inoculum should also be investigated further.

The results from similar trials on fungicide timing in New York State again came to a different conclusion. The recommendations from the 2016 and 2017 studies were to begin weekly spray programmes near the end of July, or around the time a 2.5-cm bulb had formed (Hoepting 2017b, 2018b). This is a later start than indicated by this study, where sprays at the four true leaf stage occurred on June 24th for onion transplants and at June 29th and July 8th on seeded onions. These differences cannot be compared directly as there were differences in onion cultivar and seeding date between the two locations, and initial infections and rate of disease development could have been very different. The severity of SLB in New York State was moderate in the untreated checks (2.15–3.2, on a scale of 0–6) but the percentage of premature plant mortality was very high by mid to late September (Hoepting 2017a, 2018a). This is a symptom that is only occasionally seen in the Holland Marsh, so is not included in the ratings.

Fungicides have been grouped into FRAC groups that differ in their intrinsic risk of developing fungicide insensitivity. For example, multi-site inhibitors such as M03 and M05 have a low risk of developing insensitivity, whereas FRAC group 12 is listed as low-medium, 3 and 9 are medium risk, 7 is medium-high, and 11 is at high risk (FRAC 2017). Products containing active ingredients in FRAC group M03 were registered in Canada for diseases of onions in 1988, group 29 in 2003, group 9 in 2005, groups 11 and 40 in 2006, M05 in 2008, group 3 in 2011, groups 7 and 45 in 2012, group 4 in 2014, and group 49 in 2015 (Health Canada 2020). Penthiopyrad (group 7) and boscalid plus pyraclostrobin (groups 7 & 11) reduced SLB disease in 2011 and 2012 but had no effect in subsequent years. Fluxapyroxad plus pyraclostrobin (groups 7 & 11), fluxapyroxad (group 7), and pydiflumetofen (group 7) were tested in 2016 or later years and had no effect on SLB severity. Difenoconazole (group 3) suppressed disease symptoms in 2011 and 2012 but had no effect in 2013–2015. The fungicide containing difenoconazole plus benzovindiflupyr (groups 7 & 3) was newly registered in 2019 and also did not decrease SLB symptoms. This suggests that the S. vesicarium population may have become insensitive to fungicides in FRAC groups 3, 7, 9, and 11 as early as 2014. It may also explain the lack of efficacy of a product such as Aprovia® Top which was recently registered but contained a.i.’s from these groups (groups 3 & 7).

Active ingredients with a high risk of developing insensitivity are often formulated in fungicide combinations with lower-risk active ingredients. Azoxystrobin plus difenoconazole (groups 11 & 3) and fluopyram plus pyrimethanil (groups 7 & 9) exhibited promising results in early testing years, with significant suppression in two or three seasons, respectfully. However, in 2014–2019 these two products did not reduce SLB severity relative to the unsprayed control. It is also concerning that the fungicide containing cyprodinil plus fludioxonil (groups 9 & 12), both of which are considered low to medium risk, reduced severity in 2011 but not in 2012–2015. This indicates that insensitivity to this fungicide may be developing in the pathogen population. The G143A mutation conferring resistance to QoI fungicides has been confirmed in S. vesicarium isolates in New York that exhibited decreased sensitivity to azoxystrobin (Hay et al. 2019).

Fungicide insensitivity was confirmed in isolates of S. vesicarium from Ontario. Ninety percent were insensitive to azoxystrobin and 57% were insensitive to pyrimethanil. Of these isolates, 48% were insensitive to both azoxystrobin (group 11) and pyrimethanil (group 9) (n = 42). This is very similar to results from S. vesicarium from onions in New York State where 88% of isolates from high input fields were insensitive to azoxystrobin and 33% were insensitive to pyrimethanil (Hay et al. 2019). Fungicide products that are commonly used in Ontario include active ingredients from these two FRAC groups: Switch®, Pristine®, Luna Tranquillity, Quadris Top, and Merivon®. Luna Tranquillity (contains FRAC groups 7 & 9) and Quadris Top (contains FRAC groups 3 & 11) have been registered for use on onions in Canada since 2012 (Health Canada 2020). The failure of these products to perform in the field trials could be explained by insensitivity.

Products containing mancozeb (FRAC group M03), such as Manzate® 750 F and Dithane™, are under re-evaluation by the Pest Management Regulatory Agency and are at the risk of being deregistered for use on onion in Canada due to their human toxicity (PMRA 2018). Chlorothalonil reduced severity compared to the unsprayed control in 2011 but not in 2012, Manzate was effective in 2011 and 2012, and Dithane reduced severity in 2013 but had no effect in 2014, 2015, or 2016. Recently, the use of chlorothalonil (FRAC group M05) on onion has been limited to two applications per year (Pest Management Regulatory Agency 2018).

Developing and registering a new fungicide takes an enormous amount of time, effort, and funding so it is important to manage the products that are currently registered. Rotating between FRAC groups, fewer applications, or using biofungicides in place of synthetic fungicides are management strategies that can be used. If there is a fitness cost associated with the mutation conferring fungicide insensitivity, then discontinuing the use of that fungicide is a management tactic that could shift the pathogen population back to sensitivity (Suzuki et al. 2010).

Biocontrol agents (BCAs) have not been classified with a FRAC code. The efficacy of BCAs is often affected by abiotic factors, competition in the ecosystem, or poor formulation of the final product (Ruocco et al. 2011). Even though the BCA product T-77 in 2018 and 2019 did not suppress SLB symptoms, it may still have a role as part of the integrated pest management toolkit to manage this disease. Further testing of this and other BCAs is warranted.

Stemphylium leaf blight did not decrease yield in this study, however the small-plot trials conducted at the MCRS may be unable to capture small changes in yield, even though differences in yield were found in similar sized trials in New York State (Hoepting 2017b, 2018a). Defoliation caused by SLB may not necessarily affect yield directly, especially if it occurs later in the growing season, but it can reduce the efficacy of sprout inhibitors. The sprout inhibitor malic hydrazide should be applied to green leaves shortly before lodging (Ilić et al. 2011). Onions with moderate to severe SLB would not have enough green leaves near the end of the season to take up the sprout inhibitor. If sprout inhibitor is not applied, the crop will have a shorter storage life and shelf life. Foliar sprays in combination with seed treatments may be the best combination to manage SLB.

It has been suggested that a portion of SLB symptoms on onion and garlic may be due to a host-specific toxin produced by the pathogen (Basallote-Ureba et al. 1999). Previous studies have shown that secondary metabolites produced by S. vesicarium caused leaf dieback, reduced yield, and may increase susceptibility to other diseases (Tesfaendrias et al. 2014). A few lesions of S. vesicarium on a plant early in the growing season may result in severe defoliation due to the production of host-specific toxins. Subsequent fungicide sprays would not control the earlier infections and thus be less effective in reducing the damage.

Further research is needed to develop recommendations for the integrated pest management of SLB of onion. It should focus on (a) development of a forecasting model that is effective under high and low-disease pressure, (b) continued assessment of the efficacy of new fungicides and of biological controls in field trials, and (c) identifying the in vitro sensitivity of S. vesicarium to existing fungicides in the spring of each year to warn growers against using ineffective fungicides and d) screening for insensitivity to new chemistries such as penflufen.

Acknowledgements

The authors thank M. Tesfaendrias, L. Riches, K. Vander Kooi, and S. Janse for technical assistance in field trials, and P. Hildebrand for consultation.

Supplemental data

Supplemental data for this article can be accessed here.

Additional information

Funding

The authors thank the Ontario Agri-Food Innovation Alliance, the Fresh Vegetable Growers of Ontario and the Bradford Cooperative Storage Inc. for partial funding of this research.