Introduction

Several Allium species are cultivated in South Africa (SA). The foremost crops include A. cepa L. var. cepa (bulb onion), A. fistulosum L. (Japanese bunching [Welsh, spring] onion), A. porrum L. (leek) and A. sativum L. (garlic). Bulb onions are the third most popular vegetable in the country and are widely cultivated, particularly in the Western Cape, Northern Cape, North West and Limpopo provinces with an annual production of approximately 700 000 tonnes (Anonymous 2020). In 2018 and 2019 the annual garlic production slightly exceeded 3000 tonnes (Anonymous 2020) and local demand continues to rely on imports (https://www.freshplaza.com/ accessed 24 November 2022). The leek and bunching onion industries are less structured and accurate data for hectarage and production were difficult to obtain.

The occurrence of rust on Allium species in SA has been known for several decades. Doidge (1948) reported the destructive occurrence of Puccinia allii on A. sativum and Crous et al. (2000), based on the host records of Doidge et al. (1953), listed P. allii (with P. porri as syn.) on A. cepa in the Western Cape and on A. porrum in the Western Cape and Eastern Cape. According to the rust database hosted at https://collections.daf.qld.gov.au/web/key/africarust/Media/Html/ (accessed 24 November 2022), P. porri has also been described on leek in Gauteng, SA. However, there is no published record on the response of Allium varieties to rust in SA. International reports on the susceptibility of Allium species to rust either lack clearly defined infection type (IT) scales, variety names or diversity in isolates (Jennings et al. 1990; Wako et al. 2015; Kwon et al. 2021).

Before the acceptance of molecular approaches, collections of rust fungi on species of Allium have often been identified as P. allii that consequently represented a species complex with unresolved taxonomic clarity (McTaggart et al. 2016b). Recent reports of rust outbreaks, incursion events and biosecurity threats emphasised the economic importance and necessity of the correct identification of rust fungi occurring on Allium species (Koike et al. 2001; Anikster et al. 2004; Furuya et al. 2009; Worku and Dejene 2012; Martínez-de la Parte et al. 2015; Geering 2017; Kwon et al. 2021).

A severe rust infection of bunching onion at Elim in the Western Cape, regular observations of rust-infected bunching onion and leek locally sold as fresh produce, and the occasional occurrence of rusted garlic crops prompted this investigation. Our objectives were to: (i) identify five rust isolates collected from bunching onions from three different provinces of South Africa; (ii) determine the response of Allium species and varieties to these isolates; and (iii) to study the rust infection process for selected varieties.

Materials and methods

Isolates

Allium fistulosum (bunching onion) plants heavily infected with rust were observed near Elim, Western Cape during August 2021 (Fig. 1). Infected plant material was accessioned as PREM63322 in the National Collection of Fungi, Pretoria, SA. From visits to supermarkets during March 2022, bunching onions infected with rust and sold as fresh produce were collected. These samples were traced to producers near Bloemfontein, Free State province (sample UFSPr2), the Cape Town metropolitan area in the Western Cape province (UFSPr3) and to the Gauteng province (UFSPr4). An additional rust collection (UFSPr1) from bunching onion sold as fresh produce in a supermarket in Bloemfontein in 2017 was also included, bringing the total number of isolates tested to five. Spores from a single uredinium from each of the individual samples were collected, stored at -80 °C and subsequently used to establish uredinial isolates on rust-free greenhouse-grown plants of the bunching onion variety White Welsh. Standard rust handling protocols were followed (Boshoff et al. 2020), which included a heat shock treatment of spores for 6 min at 46 °C after retrieval from the freezer. Three plants per isolate, 3–5 leaf stage, were spray-inoculated using a pressure pump (Vacuubrand® pump-model MZ2) at 25 kPa pressure setting connected to an inoculation device (Pretorius et al. 2019) with a suspension of urediniospores in 300 μl Soltrol® 130 isoparaffinic oil. To prevent contamination between isolates, inoculations were conducted in a closed room within an enclosed booth that was flushed for 60 s between sprays with warm water (± 65 °C).

Fig. 1
figure 1

Rust infection on naturally infected Japanese bunching onion near Elim, Western Cape in 2021 (top), in a commercial garlic production field (middle) and on the beaked spathe (uredinia) and scape (uredinia and telia) of garlic umbels (bottom) in the central Free State province in 2022

Full size image

Inoculated plants were dried off in a growth cabinet (200 μE/m2/s light; 25 °C) for 30 min and incubated in the dark in a dew simulation chamber overnight at 18 ± 1 °C. Upon removal from the dew chamber, plants were placed in separate compartments in a greenhouse. After 15 days, fresh spores from each isolate were sampled into size 00 gelatin capsules by connecting an air vacuum (Vacuubrand® pumpmodel MZ2) to cyclone spore collection devices (Pretorius et al. 2019) and used for inoculation of Allium varieties.

Identification of rust isolates

Infected leaf material containing two distinct pustules of each of the five rust isolates was used to extract total genomic DNA with a modified cetyltrimethylammonium bromide (CTAB) protocol (Visser et al. 2009). After diluting the DNA to 5 ng/µl, the 5.8S rRNA-ITS2-28S rRNA locus was PCR amplified using the Rust2Inv (5’-GATGAAGAACACAGTGAAA-3’; Aime 2006) and LR6 (5’-CGCCAGTTCTGCTTACC-3’; Vilgalys and Hester 1990) primer set. Each PCR reaction contained 10 ng genomic DNA, 0.25 µM of each primer, a 1 × concentration of the KAPA Plant PCR buffer and 0.5 U KAPA3G Plant DNA polymerase (KAPABiosystems, Wilmington, Massachusetts). The PCR regime consisted of 94 °C for 3 min, followed by 40 cycles of 94 °C for 30 s, 54 °C for 30 s and 72 °C for 1 min, with a final 5 min incubation at 72 °C.

The purified PCR amplicons were cloned into the pGem-T® Easy plasmid vector (Promega Corporation, Madison, Wisconsin). After transformation into Escherichia coli JM109 competent cells (Promega Corporation, Madison, Wisconsin), the cloned inserts of at least two recombinant plasmids per isolate were sequenced using the BigDye™ Terminator v 3.01 sequencing kit (ThermoFisher Scientific, Waltham, Massachusetts). Primers used for sequencing included Rust2Inv, LR6 and the internal LR0R (Moncalvo et al. 1995) and LR3 (Vilgalys and Hester 1990), respectively. Sequenced products were separated on a 3130 × 1 Genetic Analyzer (Applied Biosystems, California) using the StdSeq50_POP7 sequencing run module.

After resolving ambiguous nucleotides, the online CAP3 Sequence Assembly Program (Huang and Madan 1999) was used to construct a single contiguous sequence for each recombinant insert. The final consensus sequence for each rust isolate was deposited in GenBank (Table 1).

Table 1 Taxa used for the molecular identification of rust isolates collected from Allium fistulosum plants in South Africa
Full size table

Alignment of the five fungal sequences with selected reference sequences (Table 1) was done with the online MAFFT web interface (Katoh et al. 2019). Austropuccinia psidii was chosen as outgroup as indicated by McTaggart et al. (2016b). The 1341 bp 5.8S rRNA-ITS2-28S rRNA regions obtained after trimming both the 5’ and 3’ extending sequences, were used to determine the appropriate nucleotide substitution model with the Akaike Information Criterion (AIC) within jModelTest v 2.1.1 (Darriba et al. 2012; Guindon and Gascuel 2003). The GTR + G + I (n = 5) model was used for Bayesian inference (BI) analysis in MrBayes v 3.1.2 (Huelsenbeck and Ronquist 2001; Ronquist and Huelsenbeck 2003). BI analysis was started from a random tree using four Markov Chain Monte Carlo (MCMC) chains. The search was limited to 5 000 000 searches where every 500th generation was sampled. The average standard deviation of split frequencies was examined and the analysis stopped at a value below 0.01. The first 800 trees were discarded as burnin before analysis.

PAUP* v. 4.0b10 (Swofford 2003) was used to perform maximum parsimony (MP) analysis where all characters were weighted equally. The heuristic search was done with 1000 addition-sequence replicates with tree bisection and reconnection (TBR) branch swopping. Ten trees per replicate were saved. Bootstrap support for proposed branches was evaluated with 1000 replicates with 100 random addition-sequence replicates and TBR branch swopping.

Measurements of urediniospores (n = 30) and teliospores (n = 25) of isolate PREM63322 were recorded with a Leica DM500 microscope. Urediniospore cell wall thickness (n = 15) and teliospore apex thickness (n = 27) were also measured.

Allium infection studies

To assess the response of Allium species to the rust isolates, seed or bulbs of 65 varieties were purchased from retail outlets or by requesting certified sources directly from the licence holders (Table 2). Seeds, ± 50 per variety, were sown in 6-cm diameter plastic pots filled with Mikskaar Professional Potting Soil 70 (Hygrotech, Pretoria, SA) and watered daily with reverse osmosis purified water. Seedlings at the two leaf stage were transplanted into tapered plastic cones (4 × 4 × 10-cm, 95 cm3). Bulbs from the A. sativum varieties were split open and the cloves were directly planted into the plastic cones. Seedlings were fertilized weekly with a water-soluble fertilizer [Effekto Multifeed® Classic, NPK analysis 19:8:16 (43), 2.5 g/l]. Freshly collected spores from each rust isolate were used for inoculation as described above when the plants reached the 4–5 leaf stage. Inoculations were carried out in two independent trials with a minimum of three plants of each variety per rust isolate per trial. Twelve days after inoculation, rust infection types (ITs) were observed based on the commonly applied 0–4 cereal rust scale with 0 = no visible symptoms,; = small chlorotic or necrotic flecks, 1 = small pustules often with necrosis, 2 = small to medium size pustules with chlorosis but without necrosis, 3 = medium to large size pustules with chlorosis, 4 = large size pustules without chlorosis (McIntosh et al. 1995). Mesothetic IT responses were recorded from lowest to highest and when segregation between plants of a variety was observed, the most common IT response was recorded first.

Table 2 Allium varieties and infection types (ITs) to isolates of Puccinia porri
Full size table

Microscopy of infection process

The onion varieties White Lisbon, Hojem, Red Creole and bunching onion variety White Welsh were selected for microscopy. Sowing, transplanting, and plant maintenance were as described above. Inoculated leaf segments were sampled 24 and 48 h (hpi), and 5- and 14-days post inoculation (dpi).

For fluorescence microscopy, samples were cut into 10 mm segments and for scanning electron microscopy (SEM) into 5 mm segments. Fluorescence samples were stained using the modified method of Rohringer et al. (1977) as described by Moldenhauer et al. (2006). Leaf segments were stored in 50% (v/v) glycerol with a trace of lactophenol before viewing with an Olympus AX70 microscope (Rohringer et al. 1977). The blue wavelength epifluorescence cube with an excitation filter of 330–385 nm and a barrier filter of 420 nm was used to view stained fungal tissue, whereas an excitation filter of 450–480 nm and barrier filter of 515 nm were used to view autofluorescence associated with penetration points or colonies. The microscope was fitted with a CC12 digital camera for image capturing with Analysis LS Research version 2.2 software (Olympus Soft Imaging System). Abortive penetration (AP) was classified by Parlevliet and Kievit (1986) as infection sites not developing further than an appressorium or substomatal vesicle. For this study all nonpenetrating appressoria (NPA) and aborted substomatal vesicles (ASSV) were considered aborted. Fungal structures, except haustoria, fluoresced bright blue and host cells fluorescing orange-yellow were considered necrotic (Rohringer et al. 1977). The number of haustorium mother cells (HMC) were counted in small colonies (less than 30 HMC). When the infection sites exhibited more than 30 HMCs, the area was measured in square micrometers (μm2) using a closed polygon. The necrotic area associated with a colony was similarly measured and a hypersensitivity index (HI) was calculated by dividing the necrotic leaf area (μm2) with the corresponding colony size (μm2). Coalescing colonies were excluded from measurements.

For SEM observations, leaf segments were fixed according to the protocol of Glauert (1974). Dried segments sampled 24 and 48 hpi were directly mounted on 12.2 mm diameter metal stubs (Cambridge pin type). For observations on the outer surface of the leaves double-sided carbon tape was used, whereas double-sided sellotape was used for epidermal stripping to view fungal development inside the leaf tissue (Hughes and Rijkenberg 1985). The dried leaves were rolled open and mounted with the upper epidermis down on the stub before the lower epidermis was removed by stripping. The mounted leaf samples were sputter coated with gold (± 60 nm thickness) in a Bio-Rad sputter coater (United Kingdom) and examined with a JSM-IT200 InTouchScope™ SEM (New England, USA).

Results

Identification of South African Allium fistulosum rust isolates

When used in a BLASTn analysis, the five bunching onion rust 5.8S rRNA-ITS2-28S rRNA sequences shared the best similarity to that of eight P. porri isolates (E-value = 0.0; percentage identity > 99.9%). The phylogenetic analysis using both BI and MP analyses confirmed the isolates as P. porri with especially good PP support (Fig. 2). While all five isolates grouped with P. porri isolates originating from both different host plants and geographic origins, they grouped separately from P. mixta, as well as the P. alli sensu lato, sensu Koike and sensu Gäumann genetic lineages. All P. porri isolates were also significantly different from the two Puccinia spp. collected from garlic chives in the Philippines (KM249856) and Thailand (KU296910).

Fig. 2
figure 2

Identification of five rust isolates collected from Allium fistulosum, as Puccinia porri using a 1341 bp fragment of the 5.8S rRNA-ITS2-28S rRNA locus. Phylogenetic analysis was done using both Bayesian inference (BI) and Maximum parsimony (MP) with MrBayes and PAUP* respectively, with both the posterior probability (PP; > 0.8) and bootstrap (BS; > 80%) support values indicated above the branches. The five South African isolates are highlighted in bold

Full size image

Spore dimensions

Urediniospores [22-(26.3)-33 × 18-(20.3)-23 µm] were echinulate and subglobose to ellipsoidal. The cell wall width ranged from 1.4 to 2.3 µm (mean 1.8 µm). Teliospores were two-celled, clavate with a round to truncated apex and measured 45-(50.4)-56 × 19-(22.6)-26 µm. The apex thickness varied between 4.1 and 8.7 µm (mean 5.76 µm) (Fig. 3).

Fig. 3
figure 3

Cross sections of uredinia (a, b), urediniospores (c) and teliospores (d) of Puccinia porri

Full size image

Allium species and variety response

Infection responses of garlic, leek, bunching onion and bulb onion are shown in Fig. 4. The IT responses observed for the 65 varieties were almost identical for each isolate tested (Table 2). The 42 A. cepa varieties produced mostly low ITs (resistant ITs = 0; to;1) with the only exceptions being Lucinda (intermediate ITs = 12 +) and Cristalina (moderately susceptible ITs = 3). The varieties Australian Brown, Long Red Firenze and Texas Grano produced one susceptible (IT = 3 +) plant each. The bunching onion (A. fistulosum) variety Japanese Single Stem segregated (ITs = ;1 to 3 +) in its response with the varieties All Year Round Spring Onion, Slender Star, and White Welsh producing only susceptible (ITs = 3 + to 4) responses. The leek (A. porrum) varieties Carentan (ITs = ; or 3 +), Copenhagen Leek (ITs = ;1 or 3 +) and Elephant Leek (ITs = ;12 or 3 +) produced mixed plant responses while Bulgarian Giant Leek was susceptible (ITs = 3 +). Except for the garlic (A. sativum) varieties Kostyn's Russian Red (IT = 3C) and White Elephant Toe’s (IT = 3), that showed moderately susceptible responses (ITs = 3C), varieties from this species were consistently susceptible (ITs = 3 + C to 4). Varieties of chives (A. schoenprasum), garlic chives (A. tuberosum), and shallot (A. cepa var. aggregatum and A. oschaninii) were resistant (ITs = 0 to;1).

Fig. 4
figure 4

Infection types produced by Puccinia porri isolate PREM63322 16 days post-inoculation on White Welsh spring onion (a), Japanese Single Stem spring onion (b), Bulgarian Giant leek (c), (bottom, left to right): Egyptian Pink garlic (d), Pink Elephant Toes garlic (e), White Lisbon onion (f), and the four varieties selected for microscopy studies (left to right, White Lisbon, Hojem, Red Creole and White Welsh) (g)

Full size image

Microscopy of infection process

Infection types for the Allium varieties selected for microscopy are shown in Fig. 4g. White Lisbon and Hojem plants were highly resistant exhibiting a fleck (ITs = ;) whereas Red Creole showed small sporulating pustules (IT = 1) surrounded by chlorosis. White Welsh bunching onion was susceptible (IT = 4). The early infection process examined by SEM showed that after germination of the urediniospore, the germ tube protrudes from the germ pore and extends perpendicular to the parallel arrangement of leaf topography (Fig. 5a). When a stoma is encountered, an appressorium is formed above the stomatal opening (Fig. 5b). At 24 hpi almost all the appressoria had collapsed on top of the stomata and those on White Lisbon appeared smaller, covering about half of the stomatal opening (Fig. 5c). Inside the leaf the substomatal vesicle (SSV) appears as a spherical structure below the stomatal slit that elongates into the leaf to develop primary and secondary infection hyphae or a HMC (Fig. 5d and e). A non-penetrating appressorium (A) on the leaf surface of Hojem onion is shown in Fig. 5f.

Fig. 5
figure 5

Scanning electron micrographs of (a) a germinating urediniospore (US) on White Lisbon, (b) germ tube (GT) and collapsed appressorium (A) on top of a stoma on White Welsh, (c) a smaller appressorium on Spring White Lisbon, (d) a substomatal vesicle (SSV) inside a Red Creole leaf, (e) a SSV and infection hypha (IH) inside White Welsh, and (f) a non-penetrating appressorium on the leaf surface of Hojem. All micrographs reflect the development of Puccinia porri isolate PREM63322 24 h post inoculation

Full size image

Fluorescence microscopy revealed that the percentage of infection sites where P. porri failed to successfully penetrate and establish a colony (abortive penetration (AP) = combined NPA and ASSV) was the highest in Red Creole (44%), followed by Hojem (28%) and the lowest AP percentage was observed in White Lisbon (17%). All varieties, resistant and susceptible, supported a statistically (P < 0.05) similar number of HMC’s at 24 hpi (Fig. 6). Fewer HMC’s were observed 48 hpi (Fig. 6) for Hojem and White Lisbon, with Red Creole and White Welsh producing significantly (P < 0.05) more HMC’s. Small colonies were visible in the three resistant varieties at 5 dpi with the number of HMC’s varying between 6 and 13. For the susceptible variety White Welsh a value of 30 was allocated (Fig. 6) as accurate counting was not possible due to high HMC numbers (> 30) and different focus levels inside leaves.

Fig. 6
figure 6

The number of haustorium mother cells (HMC) of Puccinia porri isolate PREM63322 displayed by Hojem, Red Creole, White Lisbon and White Welsh for the respective sampling times, 24-, 48 h and 5 days post inoculation. Letters above bars represent the statistical differences (P < 0.05) between observations and error bars represent the standard error

Full size image

Measurements of colony size (Fig. 7) were significantly (P < 0.05) different between 5 dpi (only White Welsh measured) and 14 dpi. In general, smaller colonies were observed in Red Creole and White Lisbon compared to Hojem 14 dpi. Due to coalescing colonies, it was not possible to measure colony size in White Welsh. Host cell necrosis (HCN) (Fig. 8) was frequently observed at infection sites 24 hpi for plants of the three resistant varieties and increased exponentially as the infection progressed. Assessment of the necrotic area (μm2) indicated significant differences in the hypersensitivity index between the resistant varieties with HCN being most conspicuous in relation to visible colony size in Red Creole (Fig. 9).

Fig. 7
figure 7

Colony size (μm2) of Puccinia porri isolate PREM63322 measured at 5 days post inoculation (dpi) for White Welsh and at 14 dpi for Hojem, Red Creole, and White Lisbon. Letters above bars represent the statistical differences between observations (P < 0.05) and error bars represent the standard error

Full size image
Fig. 8
figure 8

Fluorescence micrographs indicating the infection sites of Puccinia porri isolate PREM63322, 5 days post inoculation (dpi) for White Lisbon (a fluorescence, c autofluorescence) and 14 dpi for Hojem (b fluorescence, d autofluorescence), displaying haustorium mother cells (HMC), intercellular hyphae (H) and the necrotic cells (NC)

Full size image
Fig. 9
figure 9

A hypersensitivity index (HI) determined 14 days post inoculation (dpi) for Hojem, Red Creole, and White Lisbon inoculated with Puccinia porri isolate PREM63322. Letters above bars represent the statistical differences (P < 0.05) between observations and error bars represent the standard error

Full size image

Discussion

Recent observations of rust on commercial plantings of bunching onion, leek and garlic indicated that the disease is widely distributed in SA and most likely underestimated in its importance. During March 2022, a high incidence of rust-infected bunching onion and leek was observed in plantings near Bloemfontein. These fields were established under drip irrigation and consisted of consecutive plantings of both crops at six-week intervals to uninterruptedly supply fresh produce to the market. Continuous cropping of rust susceptible Allium hosts in adjacent fields, favourable environmental conditions, including mild night temperatures and regular rainfall, as well as the ease of spore dispersal among plantings contributed to disease development. Under these conditions and despite the application of fungicides at regular intervals, rust infection remains a problem. Further to this, a severe rust outbreak occurred in a garlic seed production field established under overhead irrigation near Petrusburg in the Free State province during November 2022. Continuous moist conditions, susceptibility of the varieties planted, and the lack of timely chemical control resulted in early defoliation of plants and a consequent crop failure. Allium varieties Egyptian Pink, White Welsh and Carentan planted in these fields were confirmed either as susceptible or segregating in their response to the P. porri isolates used in the current study.

No reports of rust outbreaks on bulb onions in SA could be found. This is supported by the low ITs recorded for 41 of the 42 varieties assessed in this study. Similarly, previous studies have shown high levels of resistance in bulb onion varieties to isolates of P. allii (Jennings et al. 1990; Wako et al. 2015). These findings are further supported by the lack of records of rust on bulb onions based on the assessment of a wide collection of rust infected specimens from Allium species (Geering 2017). However, more recently a rust outbreak in an onion field caused by P. allii was reported from Korea (Kwon et al. 2021), which confirms the risk of bulb onion crops to rust should isolates with wider virulence evolve and spread to the major production areas. Varieties of chives, garlic chives and shallot expressed low ITs to the P. porri isolates in further support of host specificity.

In applying a morphological and molecular approach to identify the species of rust on a wide collection of Allium specimens, McTaggart et al. (2016b) reported P. porri on mainly leek and P. mixta on chives from Europe. Specimens infected with P. allii-like isolates were grouped into three genetic lineages including P. allii sensu Koike on garlic and chives from the USA; P. allii sensu lato on bunching onion and garlic from Australia and China; and P. allii sensu Gäumann representing isolates collected from bunching onion, garlic and shallot from Australia. Phylogenetic analysis in the current study indicated a close genetic relationship between all five rust isolates and isolates of P. porri collected from leek in Montenegro (KU296909), Albania (KU296901) and Greece (KU286908), wild leek in Germany (KU296898), bunching onion in England (KU296899) and Germany (KU296900) and chives in Serbia (KY492336) and New Zealand (ON495341). In addition, the dimensions of spore morphology determined in our study closely resembled those given for P. porri by McTaggart et al. (2016b).

Without testing of a much wider array of local isolates it is not possible to comment on whether the P. allii and P. mixta lineages, as described by McTaggart et al. (2016b), occur in SA. Interestingly, Doidge (1948) mentioned rust on Allium dregeanum (wild onion or Cape onion) in SA, but unfortunately the pathogen was not identified neither was a specimen deposited in a herbarium. Allium dregeanum occurs in the Overberg region of the Western Cape (Curtis-Scott et al. 2020) and should be surveyed for rust occurrence. The report of P. allii in the mid twentieth century in SA (Crous et al. 2000) should be interpreted within the context of morphology-based identification at the time. Future molecular work should include Allium rust specimens from the National Collection of Fungi in Pretoria and foreign isolates to resolve the taxonomic position of South African isolates. The resistant rust response observed for A. cepa in Australia (Geering 2017) as in our study may indicate similarities in the respective rust populations.

Microscopical investigation of the infection process and host response showed that the Allium-Puccinia porri interaction resembled that of many rust pathosystems (Heath 1974; Niks 1982, 1983; Maree et al. 2020; Bender et al. 2021; Boshoff et al. 2022). On the surface, the germ tubes grew perpendicular to the topography of the leaf, located a stoma, formed an appressorium and penetrated the stomatal opening. At 24 hpi no differences were observed among the varieties tested in infection structure development, including HMC formation. The number of HMC’s increased in the susceptible bunching onion White Welsh and in the moderately resistant Red Creole bulb onion at 48 hpi. All microscopic components supported susceptibility in White Welsh at 5 and 14 dpi. However, the responses of the resistant onion varieties ranked differently as would have been expected from their macroscopic ITs. For example, the HI of Red Creole, which showed small uredinia, was higher than the visually more resistant Hojem and White Lisbon. This implies a strong necrotic response in Red Creole which restricted mycelial growth.

Prehaustorial resistance, hindering the development of colonies, has been reported in both A. sativum and A. ampeloprasum by Fernández-Aparicio et al. (2011). In the present study the mechanisms of resistance acting before haustorium formation, calculated as abortive penetration, were operative in all resistant varieties with a high proportion of colonies in Red Creole failing to form any haustorium mother cells. Heath (1982) described prehaustorial resistance as typical of nonhost resistance to rust fungi, but according to Niks and Rubiales (2002) it plays a major role in partial resistance, that might be more durable than resistance controlled by hypersensitive resistance genes. Host cell death was already visible in the resistant varieties two days after inoculation with P. allii suggesting early acting hypersensitive response (HR) and at 5 dpi all infection sites were associated with HCN microscopically. The HI suggested that HR plays an important role in the defence/resistance mechanisms to impede fungal development at different stages.

In SA, azoxystrobin (Amistar®) and tebuconazole + trifloxystrobin (Nativo®) have been registered to control rust on bulb vegetables (Agri-intel; https://www.agri-intel.com, accessed April 2022). Both the Amistar® and Nativo® labels stipulate preventative applications with follow-up sprays at specified time intervals if necessary. An overlap in growing season between February and December for garlic, spring onion and leek may add to inoculum build-up in SA, especially in favourable environments assisted by irrigation. Through careful evaluation of varieties, farmers could make risk-based decisions should rust-susceptible varieties be grown. However, it may not be possible to grow resistant varieties, e.g. the popular garlic variety Egyptian Pink, and all other garlic entries tested here, are susceptible to rust. As in other crops, breeding for resistance remains an option but according to Fernández-Aparicio et al. (2011) the genetics of resistance in Allium sources may not have been extensively studied.

Conclusions

This study confirms P. porri as the causal organism of rust on Allium species in SA. Garlic and bunching onion varieties were mostly susceptible to P. porri, whereas leek varieties were either susceptible or segregating in their response. From visits to production fields, it was evident that overhead irrigation and continuous cropping contributes to higher incidence and severity of the disease. In the absence of resistant varieties producers will have to rely on integrated control measures when favourable conditions for disease development prevail. In accordance with feedback received from local breeders and seed growers, as well as with international reports, onion varieties were largely resistant to the pathogen. Microscopy observations indicated prehaustorial and early hypersensitivity as resistance mechanisms in bulb onions.