Detection and quantification of airborne spores from six important wheat fungal pathogens in southern Alberta

Abstract

Wheat is affected by many fungal diseases that can cause severe yield and quality losses. Disease prediction models generally employ weather data to estimate potential for infection to determine timing for fungicide applications, but these models fail to account for the presence and quantity of pathogen inoculum. This study adapted highly specific qPCR primers to identify and quantify, in real-time, inoculum present in air for the six most important wheat pathogens in Canada. Fungal spores were collected using either Burkard spore collectors and quantified using qPCR or microscope slides covered with adhesive tape and identified and quantified using microscopy. Samples were collected from seven different sites in southern Alberta throughout the 2015–2017 growing seasons. The results demonstrated that qPCR can reliably identify and quantify spores from Puccinia striiformis f. sp. tritici, P. triticina, P. graminis f. sp. tritici, Blumeria graminis f. sp. tritici, Pyrenophora tritici-repentis, and Fusarium graminearum. The limits of detection of DNA for primer pairs in singleplex tests ranged from 0.0001 ng for P. graminis to 0.001 ng for P. tritici-repentis, which corresponded to approximately 3 spores for P. tritici-repentis and F. graminearum and 1 spore for the other pathogens. Conversely, microscopy permitted identification of rusts to the genus but not to the species level and was ineffective in quantification of the remainder of the wheat pathogens. This study will contribute to the development of a fast and reliable forecasting system that will enable identification and quantification of airborne pathogens in real-time before initial disease symptoms appear.

Résumé

Le blé est affecté par plusieurs maladies fongiques pouvant causer des pertes sévères au rendement et à la qualité des grains. Les modèles de prédiction des maladies utilisent généralement des informations météorologiques pour évaluer un potentiel d’infestation afin de déterminer l’utilité d’épandage de fongicides. Toutefois, ces modèles ne tiennent pas compte de la présence et de la quantité d’inoculum des pathogènes. Dans cette étude, nous avons adapté ou développé des amorces hautement spécifiques de qPCR afin d’identifier et de quantifier très rapidement la présence dans l’air d’inoculum de six des plus importants pathogènes du blé au Canada. Les spores de champignons ont été collectées soit par un collecteur de spores de type Burkard et quantifiées par qPCR ou soit en utilisant des lame microscopiques recouvertes de ruban adhésif et identifiées et quantifiées par microscopie optique. Les échantillons collectés proviennent de sept différents sites du sud de l’Alberta durant les saisons de 2015 à 2017. Les résultats obtenus démontrent que la qPCR permet d’identifier et de quantifier de façon très fiable les spores de Puccinia striiformis f. sp. tritici, P. triticina, P. graminis f. sp. tritici, Blumeria graminis f. sp. tritici, Pyrenophora tritici-repentis, et Fusarium graminearum. La limite de détection de l’ADN des champignons en utilisant une paire d’amorces par réaction de PCR se situe entre 0.0001 ng pour P. graminis et 0.001 ng pour P. tritici-repentis ce qui correspond approximativement à 1 spore pour P. striiformis, P. graminis, P. triticina et B. graminis et 3 spores pour P. tritici-repentis et F. graminearum. Vice-versa, la méthode basée sur l’utilisation de la microscopie optique n’a permis que l’identification au niveau du genre pour les trois rouilles du blé, mais pas au niveau de l’espèce et était ineffective pour quantifier les autres espèces de champignons pathogènes. Les résultats de cette études vont permettre le développement d’un système de prédiction fiable presqu’en temps réel pour identifier et quantifier les pathogènes de blé transmis par voie aérienne et cela avant même que des symptômes de maladie soient apparents.

Keywords:

• fungal pathogen detection
• fungal pathogen quantification
• qPCR analyses
• spore trapping
• wheat fungal diseases

Mots clés:

• analyses par qPCR
• détection de pathogènes fongiques
• maladies fongiques du blé
• quantification de pathogènes fongiques
• trappage de spores

Introduction

The disease triangle concept states that a susceptible host, a virulent pathogen and appropriate environmental conditions are required for disease development (Agrios 2005; Scholthof 2007). Employing genetic resistance is the most economically and environmentally efficient method to control plant diseases; however, in many crops, genetic resistance is lacking, and producers must rely on fungicides and cultural methods to control diseases (Agrios 2005; Bailey et al. 2009). The reliance of modern agriculture on fungicides for disease control may not be sustainable due to the negative environmental impact of fungicides and the on-going evolution of pathogen resistance to fungicides (Wilson and Tisdell 2001). Thus, management strategies involving an overall reduction in fungicide applications coupled with their judicious use must be considered if their efficacy is to be maintained (Hobbelen et al. 2014). It may be possible to eliminate routine application of costly fungicides if pathogens are not present or at very low numbers and environmental conditions are non-conducive for disease development. Accurately assessing the presence and quantity of virulent crop pathogens, in real-time, and integrating this information with weather data into crop-forecasting models that reliably predict risk of yield loss so that control methods such as fungicides can be applied if needed, is technologically feasible (Sankaran et al. 2010; Donatelli et al. 2017). Ultimately, the goal within the context of sustainable agriculture, should be timely application of fungicides to crops only when there is a known threat by pathogens to crop yield or quality.

Identifying and quantifying virulent pathogens in real-time in the air and within the crop canopy remains challenging particularly for rusts and other pathogens that can travel long distances in air currents and are deposited on crops in ‘spore showers’ (Agrios 2005; Chen 2005). Because of the latency period between infection and expression of symptoms and signs of the pathogen (Agrios 2005; Chen 2005), monitoring for airborne pathogen propagules prior to the development of damaging symptoms during the crop season is requisite to responding effectively to the threat to crops posed by individual pathogens. Current disease monitoring methods mostly rely on the presence of visual symptoms or pathogen signs on crops (Agrios 2005; Bailey et al. 2009). Methods to monitor pathogen propagule numbers in the field include passive and active sampling. Passive sampling frequently uses sticky adhesive materials such as gelatine or sticky tape mounted on microscope slides or Petri dishes that are placed in or near field areas to capture airborne propagules (Alcázar and Comtois 1999; Isard et al. 2011). Active sampling employs instruments that extract propagules from predetermined volumes of air over a specified time (Sterling et al. 1999; Núñez et al. 2017). Once trapped, propagules must be identified and quantified. Microscopic identification methods are time consuming and may be inaccurate depending on the propagule numbers and condition and distinctiveness of the pathogen propagules. Conversely, molecular techniques are rapid, sensitive, highly accurate, becoming more cost-effective and the current focus of strategies to forecast diseases and protect crops (Khan et al. 2005; Friesen et al. 2006; Tsui et al. 2011; Randhawa et al. 2013; Cao et al. 2016). These techniques frequently employ methods such as real-time PCR (qPCR) assays, DNA sequencing, and multiplex PCR assays (Cao et al. 2016; Núñez et al. 2017). Effective molecular approaches to monitoring wheat fungal pathogens require methods for extraction of nucleic acids through adequate cell disruption and efficient DNA extraction (Gao et al. 2004; Tsui et al. 2011) and design of highly specific primers and/or probes and their optimization (Reischer et al. 2004; Nicolaisen et al. 2009; Cao et al. 2016). Pathogen-specific primers for PCR-based assays often target pathogen ITS (Internal Transcribed Spacers) rDNA regions whose variability permits differentiation among sub-genera but generally not among closely related species. Therefore, the design of more specific assays targeting unique genes, such as pathogen virulence and/or toxin genes, has been suggested (Nicolaisen et al. 2009).

A variety of pathogens infect wheat (Triticum aestivum L. and T. turgidum L. ssp. durum (Desf.) Husn.) crops worldwide. In western Canada, where approximately 10 million ha are seeded annually to wheat, the most important diseases are stripe rust (Puccinia striiformis Westend. f. sp. tritici Erikss., (Pst)), leaf rust (Puccinia triticina Erikss., (Pt)), stem rust (Puccinia graminis Pers. f. sp. tritici Erikss. & Henning (Pgt)), powdery mildew (Blumeria graminis (DC.) Speer f. sp. tritici Em. Marchal (Bgt)), tan spot (Pyrenophora tritici-repentis (Died.) Drechs. (Ptr)), and Fusarium head blight (FHB) caused predominantly by Fusarium graminearum (Schwein) Petch. (Fg) (Bailey et al. 2009). Under optimum conditions, these pathogens are capable of causing severe losses up to 50–100% in wheat (Sykes and Bernier 1991; Waalwijk et al. 2004; Chen 2005; Bailey et al. 2009; McMullen and Adhikari 2009; FAO 2010; Fernandez et al. 2010, 2016; Huerta-Espino et al. 2011; Marone et al. 2013; Tittlemier et al. 2013; Abdullah et al. 2017).

The impact of stripe rust is greatest in the western Prairies, where higher elevation and cooler environmental conditions prevail (Chen 2005; Gaudet et al. 2015, 2016; Xi et al. 2015), whereas the threat of leaf rust and stem rust is greatest in the eastern Prairies where warmer and more humid conditions prevail during the growing season (Eversmeyer and Kramer 2000; McCallum et al. 2007; Aboukhaddour 2017). Rusts are wind dispersed primarily by asexual urediniospores that are capable of moving long distances within and even between continents (Wegulo and Byamukama 2012; Chen et al. 2014). On the North American continent, leaf rust and stem rust are primarily dispersed throughout the Great Plains via the ‘Puccinia Pathway’ (Eversmeyer and Kramer 2000; McCallum et al. 2007). Stripe rust tended to be restricted to the Rocky Mountain regions and infected wheat in the Pacific Northwest USA and southern Alberta (Chen 2005; Gaudet et al. 2015, 2016) until a high-temperature race emerged in 2000; this new race permitted the spread of stripe rust into the Great Plains (Chen 2005). The urediniospore morphologies between the three rusts are similar and are characterized by the presence of germ pores, echinulation, outside hyaline walls, with shapes varying from round to obovoid, and sizes varying from 26 µm to 30 µm × 18 µm to 32 µm (Hiratsuka and Sato 1982; Savile 1984; Liu and Hambleton 2010; Baka and Rabei 2013). Spore colour varies among the rust fungi, from yellow to orange for stripe rust, reddish to brown for leaf rust and brown to black for stem rust (Agrios 2005; Baka and Rabei 2013).

The remainder of the important wheat pathogens are stubble-borne, where they overwinter and produce conidia during the spring and summer. Infection and lesions produced on wheat leaves result in a decrease in photosynthetic capacity followed by yield and quality losses (Agrios 2005; Faris et al. 2013; Moffat et al. 2014). The powdery mildew pathogen is spread on wind currents predominantly via oblong-shaped hyaline conidia (25–40 µm × 8–10 µm in size) that are produced throughout the growing season (Conner et al. 2003; Agrios 2005; Parks et al. 2011; Wicker et al. 2013; Bayer 2018).

Tan spot (also called yellow spot) is one of the most destructive leaf spot diseases in Canada (Strelkov and Lamari 2003; Fernandez et al. 2010, 2016; Aboukhaddour et al. 2013). Ptr produces a combination of three necrotrophic effectors (Ptr ToxA, Ptr ToxB, and Ptr ToxC) which can induce necrosis or chlorosis on susceptible wheat genotypes (Lamari and Strelkov 2010). Ptr ToxA is the predominant necrotrophic effector in the Americas, and it was found in 98% of the tested isolates in Alberta (Aboukhaddour et al. 2013). The tan spot pathogen overwinters as pseudothecia on wheat stubble, and in early spring the sexual ascospores mature inside the pseudothecia from where they are released to initiate the primary infection. Several cycles of secondary infection may follow initiated by multi-celled asexual spores (conidia) that develop on conidiophores during the growing season (De Wolf et al. 1998). Conidia are multinucleate and cylindrical in shape with four to seven septa and vary in size from 117 µm to 217 µm × 15 µm to 18 µm. Ascospores are also multinucleate with three septa, oval to globose shape, brown in colour, and size varies from 48 µm to 58 µm × 18 µm to 24 µm (Benslimane et al. 2017). Whether the ascospores alone or both spore types are considered to be the primary inoculum of new infections is still unclear (Morrall and Howard 1975; Ciuffetti and Tuori 1999).

FHB can be caused by several different facultative saprotrophic Fusarium species, but in western Canada it is predominantly caused by Fg (Gräfenhan et al. 2013). In addition to yield losses, Fusarium diseases can cause quality losses due to contamination of grains with toxins (Waalwijk et al. 2004; Gautam and Dill-Macky 2012; Harris et al. 2015). Increased FHB severity and toxin accumulation is correlated with high temperatures and humidity during or just after the anthesis growth stage (Fernando et al. 2000; Kikot et al. 2011; Gautam and Dill-Macky 2012). Fg overwinters and produces ascospores (sexual stage) and conidia (asexual stage) on infected plant debris during the warm and humid weather of spring. Both types of spores may serve as the primary inoculum, and are spread by rain, insects, and wind (Goswami and Kistler 2004). Macroconidia produced by Fg are translucent canoe or banana-shaped with five or more septa and a distinctly foot-shaped basal cell and vary in size from 25 µm to 50 µm × 3 µm to 4 µm. Ascospores range from translucent to light brown in colour, are slightly curved with rounded ends, and vary in size from 19 µm to 24 µm × 3 µm to 4 µm (Schmale and Bergstrom 2003; Leslie and Summerell 2006; Keller et al. 2014). It is difficult to distinguish Fg conidia from several Fusarium spp. that commonly occur in cereal fields because of their morphological similarities (Leslie and Summerell 2006).

The objectives of this study were to: (a) design or adapt highly pathogen-specific PCR primers available in public databases for a qPCR assay and to assess the efficiency of this assay for detection and quantification of Pst, Pt, Pgt, Bgt, Ptr, and Fg; (b) compare active spore sampling using a Burkard air sampler coupled with qPCR identification and quantification of spores with passive spore trapping (e.g. sticky microscope slide) that employed microscopy for identifying and quantifying spores of Pst, Pt, Pgt, Bgt, Ptr, and Fg; and (c) determine the incidence of airborne Pst, Pt, Pgt, Bgt, Ptr, and Fg fungal spores from air samples taken in wheat fields in southern Alberta during the 2015–2017 crop seasons. These results will contribute to development of forecasting models for wheat diseases that provide the critical information to producers about potential disease outbreaks in western Canada.

Materials and methods

DNA extraction

Methods were optimized by using pure spore preparations from the wheat pathogens Pst and Bgt, obtained from wheat plants grown in growth chambers, Pt and Pgt (courtesy of Dr. H. Randhawa, Lethbridge, Alberta), Ptr (courtesy of Dr. S. Strelkov, Edmonton, Alberta), and Fg (courtesy of Dr. N. Foroud, Lethbridge, Alberta), with the following DNA extraction protocols: alkaline polyethylene glycol (PEG)-based (Chomczynski and Rymaszewski 2006); OmniPrep kit (Biosciences); DNeasy PowerSoil kit (QIAGEN); and NaOH solution (modified from Werner et al. 2002). For the modified NaOH method samples in 100 µL of 50 mM NaOH were heated at 95°C for 15 min and cooled on ice and neutralized with 10 µL 1 M Tris-HCl, pH 8.0). Pst (LSW3-2012 and SWS9-90), Pt, Pgt and Bgt (uncharacterized isolates) fungal spores were collected from wheat plants reared and inoculated under controlled conditions whereas Ptr (various isolates from races 1 and 2) and Fg (strain GZ3639) were grown on V8-PDA and PDA, respectively. Also, in order to test the ability of the different protocols to isolate DNA in dirty air samples, various simulations using soil, insects, and the six fungal spores were conducted. Following DNA extraction, samples were kept at −20°C until analyzed.

Optimization of pathogen-specific primers

Sets of pathogen-specific primers were designed based on sequences available on public databases and were evaluated using standard PCR with the QIAGEN HotStarTaq Master Mix kit and protocol (2.5 μL DNA template added). Cycle conditions were: 1 cycle at 94°C for 15 min, 35 cycles at 94°C for 1 min, 55 to 65°C for 1 min and 72°C for 1 min followed by a final extension period at 72°C for 10 min after the last cycle. After six PCR primer pairs were selected (one pair for each pathogen) (Table 1), they were adapted and/or optimized with DNA samples from the six pathogens. Primer pairs were further evaluated using different qPCR assay approaches for quantitative determination of the pathogens. A minimum of two replicates for each primer set including three technical reps for each sample was conducted.

Table 1. Sequence data and annotation for primer sets selected for standard PCR and qPCR assays for the six wheat pathogens

Primer pairs	Target^a	Sequences (5’-3’) Annotation	Size bp	Tm^c °C	Optimum^d Tm °C	GenBank accession	References
PSBTQ	Pst	F: CAATCACCGTCC	151	74	55-60	HM067995	Liu et al. 2015
		CAGAGTTGACATC
		R: ACGGACAGCAT
		GTTCTCTTCGACT
		Β-tubulin gene
Pg2BTQ	Pgt	F: AAGAGAACATG	146	61	53-55	HQ317591	Liu et al. 2015
		CTGTCCGTT
		R: TTTGGAACTACA
		AGGGTATCAAG
		Β-tubulin; mRNA
EF1-Pt2-032	Pt	F: TCGTCATTGGAC	504	72		JX533507.1	Liu et al. 2015
		ACGTCGATTCGT
EF1-Pt1-113		R: ATCCTCACAGCG
		GTAAATCCACCA
EF1-Pt1-113*		R: TGGTGGATTTAC	104	72	53-59	JX533507.1	This manuscript
		CGCAGTGAGGAT
		Elongation factor 1-α
Bgt-6	Bgt	F: CGAGGGTTGGA	157	60	55-57	HQ631368	Parks et al. 2011
		ACTGGATAA
		R: TTCTCTGTTTGA
		CGCTCTCG
		Bgt-6 microsatellite sequence
Tox A1	Ptr	F: GTCATGCGTTCT	294	59	53-57	HM234155.1	GenBank
		ATCCTC
Tox A2		R: CCTATAGCACCA
		GGTCGTCC
		ToxA-like gene
Fgram379	Fg	F: CCATTCCCTGG	58	55		KX702800.1	Nicolaisen et al. 2009
		GCGCT
Fgram411		R: CCTATTGACAGG
		TGGTTAGTGACTGG
Fgram411^b		R: CAGGTGGTTAG	50	56	53-57	KX702800.1	This manuscript
		TGACTGG
		Elongation factor 1-α

^aTarget species were Puccinia striiformis f. sp. tritici (Pst); Puccinia graminis f. sp. tritici (Pgt); Puccinia triticina (Pt); Blumeria graminis f. sp. tritici (Bgt); Pyrenophora tritici-repentis (Ptr); Fusarium graminearum (Fg).

^bModifications made on the original sequence.

^cOptimum temperature reported in literature.

^dOptimum temperature range observed under our experimental conditions.

Singleplex, duplex, and multiplex qPCR analyses and digital droplet PCR (ddPCR) were evaluated for their ability to detect and quantify the six wheat pathogens. PerfeCTa^® SYBR^® Green SuperMix, Low Rox (QuantaBio) for singleplex and duplex qPCR; and Prime Time^® Gene Expression Master Mix probe-based (Integrated DNA Technologies) (TaqMan) for multiplex qPCR were used. QX200™ ddPCR™ EvaGreen Supermix (Bio-Rad Laboratories, Inc.) was used for ddPCR. One probe was designed for each primer set (data not shown).

Samples collected by the spore sampling instruments in the field were analyzed for fungal identification and qPCR quantification using an Applied Biosystems™ QuantStudio™ 6 Flex Real-Time PCR System using the PerfeCTa^® SYBR^® Green kit and protocol using 2.5 μL of extracted DNA. Cycle conditions were: 1 cycle at 95°C for 3 min, 40 cycles at 94°C for 15 s, 55°C for 30 s, and 72°C for 30 s, with melting curve formation at the end of the run. Analyses of the results that included standard curves, Ct values, melting curves, and other parameters, were conducted using QuantStudio™ Software V1.3. Standard curves to determine the quantity of DNA amplified, using qPCR, employed known spore numbers for each pathogen. These results were used to evaluate the method consistency and to calculate how many spores are needed to obtain 1 ng of DNA for each pathogen. The qPCR results of the air samples (in ng of DNA) were then converted into spore numbers.

Spore trapping in southern Alberta

Of the different commercial spore trapping instruments available, the Burkard 7-days sampler has been shown to be a reliable instrument for sampling airborne spores for DNA-based analyses (Cao et al. 2016; Núñez et al. 2017). Therefore, the active spore sampling employed in this study used the Burkard Automatic Cyclone Air Sampler (Burkard Manufacturing Co. Ltd., Rickmansworth, Hertfordshire, UK), and air was sampled at a rate of 16.5 L min⁻¹. The Burkard Cyclone One-vial, which samples constantly into one tube until the tube is manually changed, and Burkard Cyclone Multi-vials which permitted programmed sampling into 8 individual tubes, were employed. The samples were collected in 1.5 mL microfuge tubes (Burkard Manufacturing Co. Ltd.). The instruments were placed in or near wheat fields at several locations in southern Alberta (Table 2). Instruments were set up on top of a pallet platform and powered by a battery and solar panel. The collection orifice of the spore collector sitting on a pallet was 90 to 100 cm above ground level. The collection orifice was always facing the prevailing wind direction due to the instrument’s wind vane. During the 2015, 2016, and 2017 growing seasons, samples were generally collected at weekly intervals but, in a few instances, samples were collected for more than seven days due to weather or other operational restrictions. Samples were dried using a Savant SpeedVac^® SC110 Concentrator and stored at −80°C prior to DNA extraction.

Table 2. Sampling sites in southern Alberta where the Burkard Cyclone instruments and microscope slides were installed during the wheat growing seasons 2015, 2016, and 2017

Growing season	Location	Coordinates
2015 2016 2017	Weather Station – Lethbridge Research and Development Centre (LeRDC), Lethbridge, AB	N 49° 41.685ʹ W 112° 46.080’
2015	Field Pathology Building LeRDC, Lethbridge, AB	N 49° 42.067ʹ W 112° 44.449’
2016	Crop Diversification Centre South, Brooks, AB	N 50° 32.612ʹ W 111° 50.609’
2016 2017	Stanford Farm, Magrath, AB	N 49° 23.556ʹ W 112° 56.510’
2016 2017	Stamp Seeds Farm, Enchant, AB	N 50° 10.047ʹ W 112° 26.030’
2017	Welsh Farm, Milk River, AB	N 49° 6.423ʹ W 111° 55.137’
2017	Fairfield – LeRDC, Lethbridge, AB	N 49° 42.493ʹ W 112° 41.738’

Passive spore trapping employed double-sided adhesive tape (3M Scotch^® Removable Poster Tape, 3/4” (19 mm) wide, clear) applied to a 19 mm × 50 mm area of a microscope slide (25 mm × 75 mm). Slides were attached to the cap of the Burkard Cyclone instrument, just below the collection orifice, and always facing the prevailing wind. The slides were stored in a microscope slide box at room temperature for up to a few days until analyzed using a light microscope.

Microscopic analyses

Slides with adhesive tape were observed, directly without any staining, using a Leica Microsystems DM6000 B brightfield microscope equipped with a DFC310 FX camera. Photos of spores were taken; fungi were identified based on spore shape, size, and colour and quantified using a grid slide underneath (with a defined area of 40 quadrants of 5 mm × 5 mm in size, covering a total area of 20 mm × 50 mm) to facilitate the orientation on the slide while counting the spores.

Temperature and precipitation

Records for temperature, precipitation and growing degree days (above 5°C) for different locations were obtained from the Alberta Agriculture and Forestry web site (https://agriculture.alberta.ca/acis/alberta-weather-data-viewer.jsp).

Statistical analyses

Statistical analyses were performed using SPSS Statistics (IBM Analytics, release 23.0.0.0). Descriptive statistics analyses were conducted on the data. Simple Linear Regression (SLR) was used to analyse the results of air samples. Pearson correlation test was used to evaluate the relationship between spores counted under the microscope and spore quantities detected by qPCR assays. A data clean-up removing zeros (no spores detected) and logarithm base 10 transformations (log10) of spore numbers were used to correct for skewed data (McDonald 2014).

Results

Environmental conditions

Total accumulated rainfall (TAR) and growing degree days (GDD) from May 1 to September 1 for the Weather Station location were 100 mm and 1450 GDD in 2015; 212 mm and 1400 GDD in 2016; 133 mm and 1600 GDD in 2017 whereas the long-term averages of TAR and GDD for the site were 208 mm and 1300 GDD. These results showed that the 2015 and 2017 growing seasons were warmer and dryer than the long-term averages whereas 2016 was warmer but similar to the long-term averages for precipitation. Similar trends were observed for Magrath and Enchant in 2016 and 2017 (data not shown).

Sampling and identification of fungal pathogens

Results of passive sampling employing microscopic identification and quantification of fungal spores demonstrated that rusts and other pathogens were more abundant in 2015 than in 2016 and 2017 at the Weather Station site (Fig. 1a-c). Rust urediniospores of Pst, Pgt and Pt are nearly morphologically identical (Agrios 2005; Bailey et al. 2009) and it was not possible to accurately distinguish them microscopically. Rust distribution was similar between the Field Pathology Building and Weather Station sites in 2015 (Fig. 1a, d). Similarly, Fg macroconidia were generally indistinguishable on slides from those of other Fusarium spp. so all canoe-shaped macroconidia were grouped together. Collectively, the rusts were the most abundant in all years followed by Bgt and Ptr. Similar trends were observed for the other site/years, whereas Fusarium spp. were infrequently observed in all of the site/years.

Fig. 1 Weekly numbers of fungal spores collected using passive spore collection and identified and enumerated using a compound microscope. Samples were collected at the Weather Station, Lethbridge, AB, during the wheat growing seasons 2015 (a), 2016 (b), 2017 (c), and at the Field Pathology Building, Lethbridge, in 2015 (d). The collection dates mark the beginning of the sampling week

Fig. 2 Relationship between log number of spores used to generate standard curves and Ct values in qPCR assays for identification and quantification of: (a) Puccinia striiformis f. sp. tritici (Pst), (b) Puccinia triticina (Pt), (c) Puccinia graminis f. sp. tritici (Pgt), (d) Blumeria graminis f. sp. tritici (Bgt), (e) Pyrenophora tritici-repentis (Ptr), and (f) Fusarium graminearum (Fg). Overlapping data points are represented with symbols with a darker outline (in some cases, more than two data points were overlapping)

Of the DNA extraction protocols evaluated, the most effective was the DNeasy PowerSoil kit (QIAGEN) because it yielded higher DNA quality and consistent results among technical replicates. Minor adaptations made to the original QIAGEN protocol consisted of the following: air sample tubes were washed (4×) with the beads buffer and C1 solution, and spores were ground at room temperature using a PRECELLYS^® 24 (Bertin Instruments) using program 5000 rpm with 2 × 30 s grinding and a 10 s pause between the two periods. Following the extraction process, DNA was eluted in 60 μL of C6 solution to increase DNA concentration.

The most consistent six of 29 primer pairs tested (Table 1), one for each of wheat pathogens, were selected for optimization. All primers were first tested by standard PCR using a temperature gradient (55 to 65°C) to identify the annealing temperature range of detection. They were all tested using DNA from spores of the six target pathogens. The optimum annealing temperatures for Pst, Pgt, Pt, Bgt, Ptr, and Fg represented by the brightest bands were 55 to 60°C, 53 to 55°C, 53 to 59°C, 55 to 57°C, 53 to 57°C, and 53 to 57°C for primer pairs, respectively. Several tests were conducted using standard PCR before advancing to different qPCR assay approaches and ddPCR (data not shown). One of the tests was to confirm specificity of the chosen primers; it consisted in running PCR assays for each primer set with DNA of all six target pathogens and, for a few primer sets, other pathogens’ DNA were also used. For example, each rust primer set was tested for all three rusts plus the other three target pathogens, and the Fg primer set was tested for Fg, seven other Fusarium spp., and the other five target pathogens (data not shown). In all cases, among the six pathogens only the target species’ DNA was amplified. Among all qPCR and ddPCR methods tested, qPCR with a single pair of primers (singleplex) presented the brightest bands and most consistent results. An annealing temperature of 55°C produced strong and consistent amplifications and was selected for a multiplex of all six pathogen primer pairs. All multiplex and most duplex primer combinations yielded inferior results when compared to simplex results as these amplifications yielded additional fragments detected by melting curve analysis (qPCR) or additional DNA fragments in electrophoresis gels of amplified products (PCR) or amplification failure of pure DNA samples. Each primer pair was singly tested by qPCR assays using 10-fold serial dilutions of DNA of known concentration. Standard curves testing the efficiency of the qPCR assays with known amount of DNA from spores of each pathogen subjected to regression analysis demonstrate a very high association between Ct values and the log10 values for spore number for all six species (Fig. 2a-f) (P < 0.001). The response was linear over four orders of magnitude for each fungal species evaluated. The limits of detection of DNA for the primer pairs ranged from 0.0001 ng for Pgt, 0.0002 ng for Pst, Pt, Bgt and Fg to 0.001 ng for Ptr which corresponded to approximately one spore for Pst, Pgt, Pt and Bgt and 3 spores for Ptr and Fg. Analyses of active air samples collected at the seven sites during the project showed that the majority of the samples results presented a significant linear relationship (P < 0.05) between log10 of spore numbers and Ct value for the six pathogens in 61 of the 66 pathogen/site/years (Table S1). Where non-significant values were encountered, pathogen spore numbers were very low (often zero) and of the same value thus preventing generation of equations (Table S1).

The major advantage of molecular detection using qPCR over microscope detection and identification was that spores could be accurately identified and quantified at the species level (Fig. 3). Overall, results using microscopy and qPCR were correlated for the rusts during 2015 (R = 0.68, P < 0.03) and 2017 (R = 0.61, P < 0.05) but not in 2016. For Bgt, Ptr, and F. spp., no association was identified as detection levels for these pathogens were so much higher using qPCR.

Fig. 3 Weekly number of fungal spores collected using active spore collection and identified and enumerated using qPCR at the Weather Station, Lethbridge, AB, in 2015 (a-b), in 2016 (c-d), and in 2017 (e-f), respectively. The collection dates mark the beginning of the sampling week. Note the difference in scale for spore abundances among the different panels

qPCR permitted detection of the six pathogens in all of the seven collection sites during the study in southern Alberta (Table 3). Fg spores generally predominated among the six pathogens followed by Ptr and Bgt at most sites (Fig. 3, Table 3). Among the rusts, Pst, spores were generally most abundant followed by Pt and finally Pgt (Table 3, Fig. 3a, c, e). Conversely, the highest average occurrence detected using Burkard/qPCR among all six pathogens was observed for Fg, averaging 866 spores/sample including sites with greater than 33 000 total spores observed for the 3 years of study (Table 3, Fig. 3b, d, f). The highest total spore numbers throughout the growing season detected using qPCR were also observed for Fg which frequently attained levels of 1000–2000 spores in individual samplings compared with slide counts that ranged from 0 to 6 (Table 3; Figs. 1a-d, 3a-f). For Bgt, spore counts ranged from 200 to 625 at the majority of sites using qPCR. The highest total numbers of spores observed for all species in the 3 years of the study using qPCR was for Ptr which attained approximately 5000 in a late August sample in 2016 at the Weather Station site (Fig. 3d). In general, the highest total spore numbers were observed in 2016 followed by 2017 and 2015. This ranking corresponds to the rankings for TAR for the 3 years.

Table 3. Comparison of the total spore numbers for the 3 years of the study over the growing season detected by active sampling or passive sampling at various locations in southern Alberta from 2015–2017

Site	Year	Pathogen	Active^b Sampling	Pathogen	Passive^c Pathogen Sampling
Weather Station (LeRDC^a)	2015	Pst	349	rusts	848
		Pt	33
		Pgt	5
		Bgt	1105		369
		Ptr	211		8
		Fg	1693	F. spp	3
Weather Station (LeRDC)	2016	Pst	256	rusts	537
		Pt	104
		Pgt	83
		Bgt	1695		238
		Ptr	8820		14
		Fg	16005	F. spp.	1
Weather Station (LeRDC)	2017	Pst	103	rusts	486
		Pt	17
		Pgt	59
		Bgt	335		173
		Ptr	900		10
		Fg	10730	F. spp.	0
Field Pathology Building (LeRDC)	2015	Pst	638	rusts	711
		Pt	265
		Pgt	14
		Bgt	2741		219
		Ptr	226		2
		Fg	2549	F. spp.	6
Fairfield Station (LeRDC)	2017	Pst	1095	rusts	1465
		Pt	630
		Pgt	82
		Bgt	306		387
		Ptr	102		1
		Fg	302	F. spp	1
Magrath	2016	Pst	99
		Pt	90
		Pgt	10
		Bgt	899
		Ptr	59
		Fg	603
Magrath	2017	Pst	36
		Pt	14
		Pgt	4
		Bgt	352
		Ptr	534
		Fg	1425
Enchant	2016	Pst	93
		Pt	64
		Pgt	13
		Bgt	2599
		Ptr	90
		Fg	93
Enchant	2017	Pst	5
		Pt	13
		Pgt	12
		Bgt	11
		Ptr	8
		Fg	279
Brooks	2016	Pst	267
		Pt	27
		Pgt	29
		Bgt	1027
		Ptr	1858
		Fg	28941
Milk River	2017	Pst	113
		Pt	25
		Pgt	1
		Bgt	224
		Ptr	831
		Fg	4275

^aLeRDC = Lethbridge Research and Development Centre.

^bActive sampling employed used the Burkard Automatic Cyclone Air Sampler at an air sampling rate of 16.5 L min and qPCR for spore identification and enumeration.

^cPassive spore trapping employed double-sided adhesive tape on a 19 mm x 50 mm microscope slide and spores were identified and enumerated using a compound microscope.

While spore counts through the growing season differed among years and locations (P < 0.001), some patterns of spore release were observed. Pst spore numbers tended to increase in early June, attain maximum levels in early- to mid-July and decrease towards the end of the growing season (Figs. 1, 3). Ptr spore numbers fluctuated markedly according to year and location. Very low spore numbers of Pgt were generally observed in mid-July but increased gradually until early September (Figs. 1, 3). For Bgt, Fg and Ptr, qPCR quantification determined that spores were generally widespread throughout the sampling periods but increased to maximum numbers later in the growing season.

Discussion

This study focused on detection and quantification of six important airborne fungal wheat pathogens using an active sampling method combined with molecular identification and quantification and contrasted with a passive sampling method combined with microscopic identification. Studies on identification and quantification of airborne wheat pathogens are a priority worldwide because of the importance of the crop as a food and feed source. Additionally, there is a demand for accurate real-time methods for forecasting models employed to sustainably mitigate damaging effects of disease to wheat production through the judicious use of pesticides for disease control (Fernando et al. 2000; West et al. 2008; Isard et al. 2011; Guarín et al. 2015; Cao et al. 2016).

For the qPCR-based analyses, the six best out of the 29 pathogen-specific PCR primer pairs evaluated were selected for their ability to reliably amplify target sequences from wheat pathogen genomes during the optimization period. The limits of detection of DNA for primer pairs in singleplex tests ranged from 100 fg for Pgt to 1 pg for Ptr which corresponded to approximately 1 spore for Pst, Pgt, Pt and Bgt, and 3 spores for Ptr and Fg. These values are consistent with sensitivities which vary between 200 fg and 1–100 pg or between 1–10 spores or conidia that have been developed for other pathogens in qPCR-based tests (Gao et al. 2004; Lievens et al. 2006; van Doorn et al. 2007; Carisse et al. 2009, 2014; Mirmajlessi et al. 2015; Canfora et al. 2016; Gu et al. 2019). The utilization of probes based on rDNA sequences has been associated with improved assay sensitivity (Gao et al. 2004; Lievens et al. 2006; van Doorn et al. 2007; Carisse et al. 2014; Gu et al. 2019) whereas in other studies, no improvement was observed (Carisse et al. 2009; Canfora et al. 2016). Studies have demonstrated the need for a linear range between 3 and 6 orders of magnitudes in order to produce meaningful results (Gao et al. 2004; Lievens et al. 2006; van Doorn et al. 2007; Carisse et al. 2009, 2014; Mirmajlessi et al. 2015; Canfora et al. 2016; Gu et al. 2019). The linear range in this study for detection of all six fungal pathogens was superior to four orders of magnitude. For the rusts, detection limits were within the range of one spore; therefore, housekeeping genes (β-tubulin and elongation factor 1-α) were abundant enough to permit the detection of these fungi. Because, rust spores are dikaryotic, this represented the ability to detect the sequences of both nuclei present in the spore as these housekeeping genes are present in both nuclei in stripe rust (Zheng et al. 2013). Similarly, macroconidia of Fg and Ptr are both multi-cellular, 6–8 per macroconidia, and multi-nucleate (Roper et al. 2011; Shahi et al. 2014). Consequently, the true sensitivity of the primers employed in our qPCR assays is proportionally increased. Key to the success of any detection/quantification assay is an efficient and effective extraction of the DNA from spores with the absence of PCR amplification inhibitors (Gao et al. 2004; van Doorn et al. 2007; Carisse et al. 2009, 2014; Mirmajlessi et al. 2015; Canfora et al. 2016).

PCR primers for Tox A1/2 in Ptr have the potential to amplify sequences among Parastagonospora (syn. Stagonospora; Phaeosphaeria, Septoria) nodorum (Berk.) Quaedvlieg, Verkley & Crous (Pn) strains that cause leaf blotch of wheat and also produce the Ptr ToxA protein (Phan et al. 2016). Tox A has also been identified in Bipolaris sorokiniana (Sacc.) Shoemaker (Bs) causal agent of spot blotch in wheat (Friesen and Holmes 2018; McDonald et al. 2018). Whereas Ptr is the predominant leaf spot pathogen in Alberta (Aboukhaddour et al. 2013), the Ptr primers may also be useful to detect P. nodorum which is also a member to the leaf spot complex. The prevalence of these main species can vary depending on the soil type, agricultural practice, region and other factors (Fernandez et al. 2016). For example, P. nodorum is the most prevalent pathogen in the black soil zones in Saskatchewan (Fernandez et al. 2016) further supporting a previous report of it being the main leaf spotting pathogen in eastern Canadian Prairies (Bailey et al. 2009). Thus, based on all these evidences and for the purpose of this study, all the spores detected by the Tox A primer set are being treated as Ptr due to the absence of Pn and Bs in southern Alberta. Regardless, spore numbers of these species should be included in future forecasting models.

Identification and quantification of all six pathogen spores in air samples collected at all seven locations in southern Alberta were possible using qPCR analyses. Correlations between log10 number of spores and Ct values detected for most of the air samples collected by the cyclone instruments demonstrated the consistency of this analytical method. Active sampling using the Burkard system and qPCR was markedly more sensitive than passive sampling coupled to microscopic identification and quantification of the six wheat pathogens thereby confirming results from previous studies (Alcázar and Comtois 1999; Alcázar et al. 2003; West and Kimber 2015; Cao et al. 2016). When active sampling/qPCR and passive sampling/microscopy were compared, the number of Puccinia spp. spores enumerated using the latter was generally higher but individual species could not be determined because of the morphological similarities among urediniospores belonging to the three species. Pst was the predominant rust species at all seven locations in southern Alberta whereas the least abundant pathogen was Pgt. Pst is endemic to the western Prairies and occurs in most years (Gaudet et al. 2015, 2016; Brar et al. 2017). The lower incidence of both Pgt and Pt is likely due to their natural occurrence along the Puccinia pathway in the central Great Plains of North America, resulting in late entry of these two species in southern Alberta in most years (Wegulo and Byamukama 2012). Correlative analyses between the two methods suggested that microscopic analyses, albeit less sensitive, could be used to quantify the wheat rusts collectively but not individually in our study although stem and leaf rust urediniospores have been distinguished by pathologists on the Prairies since the initial studies on rusts’ impact on wheat production (Green 1965). In Alberta and western Saskatchewan, leaf and stem rust usually cause only minor infections and yield losses whereas stripe rust will overwinter or arrive early in the spring and causes serious yield and quality losses (Gaudet et al. 2015, 2016; Brar and Kutcher 2016; Aboukhaddour and Amundsen 2018). Therefore, obtaining accurate stripe rust spore numbers in spore showers is critical to implementing control strategies in western Canadian Prairies. Elsewhere on the Canadian Prairies, identification of rust species is also critical to assessing whether seeded varieties are resistant to prevailing rusts because resistance to leaf and stem rust is common in current wheat varieties. However, stripe rust resistance is lacking in several wheat varieties (McCallum and DePauw 2008; Randhawa et al. 2012; Martens et al. 2014).

For Fg, Ptr and Bgt, spore numbers collected passively on sticky slides and identified and quantified using microscopy were insufficient to be of epidemiological value. Conversely, the highest average occurrence detected using Burkard/qPCR among all six pathogens was observed for Fg including samples with greater than 10 000 spores. These results are similar to those reported by Fernando et al. (2000) in eastern Canada. Bgt or Ptr were also abundant depending on the location, averaging 2217 and 1968 spores/season/location, respectively. These three pathogens overwinter in stubble and are widespread in western Canada (Aboukhaddour et al. 2013; Gräfenhan et al. 2013; Abdullah et al. 2017). Ranges of 50 to 500 000 spores/sample were observed for Bgt spores in wheat fields in Hebei Province (Cao et al. 2016); however, no reports of spore numbers for the rusts and tan spot were found in the literature. The greatest factor attributing to the discrepancy between the spore numbers quantified by the two methods was likely due to a tendency for slides to be washed by rainfall and/or irrigation, and general weathering of the slides fully exposed to the sun and wind during the one-week sampling periods. This is a major limitation for the adhesive tape method (Alcázar et al. 2003). Adding to the discrepancy for Fg, macroconidia are less common than ascospores at 1 m height where wheat heads occur resulting in their underrepresentation on microscope slides (Fernando et al. 2000). Additionally, ascospores can be difficult to characterize morphologically (Fernando et al. 2000). Conidia of Pyrenophora teres, the causal agent of net blotch of barley, is also restricted to a zone of 1 m above the ground (Martin et al. 1984) which might be similar for Ptr.

As these two sampling methods and their associated identification and quantification methods were so different, it was not possible to directly compare them. Both methods have their advantages; microscopic sampling and analyses require a relatively low level of technology, material cost, and setup compared to the Burkard/qPCR method, but microscopic counts are time-consuming, involve more costly labour, required a much higher expertise to distinguish them and provide limited to no information at the species level. Conversely, the Burkard Cyclone instrument as well as qPCR instruments and related technologies are currently expensive but future widespread adoption of these technologies will undoubtedly reduce costs. Field setup procedures with the Burkard Cyclone is more complicated than microscope slide placement and there is a risk of instrument failure due to power supply failure or mechanical issues. The collector orifice on the Burkard Cyclone instrument was set up at 90 to 100 cm from the ground surface which is appropriate for sampling pathogens that may attack the flag and penultimate leaves and the head; these three wheat organs contribute the greatest to yield and quality in wheat and therefore require the greatest protection against pathogens if losses are to be avoided (Patterson et al. 1990; Dimmock and Gooding 2002). In this study, microscopic analysis served as a complementary technique providing contrasting yet supporting evidence for identification and quantification of rust airborne fungal spores using Burkard sampling/qPCR. In situations where active samplers are in limited number or not available, an alternative is performing qPCR on adhesive tape samples as was demonstrated when wheat powdery mildew spores were effectively released from adhesive tape prior DNA extraction (Cao et al. 2016).

The upper sensitivity limit of urediniospore detection, which is similar to other biological systems (Gao et al. 2004; Lievens et al. 2006; van Doorn et al. 2007; Carisse et al. 2009, 2014; Mirmajlessi et al. 2015; Canfora et al. 2016; Gu et al. 2019), is much higher than what is needed for detection of pathogens. For example, a minimum threshold level of up to10⁴ to 10⁶ urediniospores is needed for development of stripe rust under controlled environment and field conditions (Araujo 2018). Whereas our results generally demonstrated a reduced sensitivity of qPCR when primers were used in a multiplex vs singleplex manner, the high sensitivity demonstrated in our results suggest that a multiplex containing primer pairs for several or all of the six wheat pathogens could be devised. Although the prospect for development of a multiplex of primers for all the important wheat pathogens remains to be determined, it is the goal in the field assessment of wheat pathogens for future forecasting models. Additionally, future breakthroughs in understanding the nature of pathogenicity and virulence genes will permit the identification of individual pathogen races and hence, increase the overall precision of disease forecasting systems.

Variance in spore numbers for the different pathogens in air samples was observed among the different locations during the three crop seasons from 2015 to 2017. In general, based on qPCR results, there were higher spore numbers from 2016 compared to 2015 and 2017. These changes closely followed the long-term average rainfall/moisture and lower GDD in 2016 compared to 2015 and 2017 that were warmer and drier. The prevailing wind directions in southern Alberta are from west, west southwest and southwest (Vickers et al. 2001; AAF 2020) which would likely create spore showers for stripe rust originating from the Pacific Northwestern USA (Chen 2005) and influence the secondary spread of rust spores. This demonstrates the need for establishing a representative array of sampling sites throughout southern Alberta in order to document the location and timing of stripe rust spore showers and its subsequent spread during the growing season.

In summary, this study demonstrated that it was possible to reliably identify and quantify spores from six fungal wheat pathogens in air samples from southern Alberta areas by selecting pathogen-specific primers and an optimized qPCR protocol. A qPCR-based technique coupled with Burkard Cyclone instruments for air sampling provided a means of generating efficient and reliable pathogen identification and monitoring in real-time. The selected pathogen-specific primers were highly sensitive and served to identify and quantify, in real-time, the incidence of airborne spores of wheat pathogens sampled at specific locations in southern Alberta. These methods will be important in developing reliable forecasting models based on pathogen presence and abundance that will be indispensable for estimating the risk of crop disease and for establishing the best practices for the use of fungicides that will help maintain fungicide efficacy, reduce production costs, and protect the environment.

Acknowledgements

We would like to thank Dr. Michael Harding, Crop Diversification Centre South, Alberta Agriculture and Forestry, Brooks, AB and the owners of the Stanford Farm, Magrath, Stamp Seeds Farm, Enchant, and Welsh Farm, Milk River for hosting and collecting samples from the spore collectors.

Supplementary material

Supplemental data for this article can be accessed online here: https://doi.org/10.1080/07060661.2020.1817795

Additional information

Funding

This work was funded by the Peer Review Program of Agriculture and Agri-Food Canada