Oat crown rust, caused by the fungus Puccinia coronata f. sp. avenae has long been considered as one of the most devastating diseases on oats worldwide (Nazareno et al. 2018; Simons 1985). Like other cereal rusts, the life cycle of P. coronata includes two hosts and five spore stages (Simons 1970; Leonard and Szabo 2005; Jin et al. 2010), and an aecial host is a prerequisite for the fungi to complete the sexual part of their life cycle. The disease is named after the crown-like structures on the dark teliospores of P. coronate; however, it is the clonal, uredinial stage causing orange pustules on the oat leaves that causes the major yield losses.

The aecial hosts of P. coronata are Rhamnus spp. and Frangula spp. (Anikster and Wahl 1979; Hemmami et al. 2006; Dietz 1926; Gäumann 1959; Nazareno et al. 2018). In Sweden Rhamnus cathartica and Frangula alnus are prevalent in the agricultural landscape (Mossberg and Stenberg 2003) and both species harbor aecia annually. In spring, the teliospores on the remaining grass leaves germinate into basidiospores that can infect young leaves of the aecial hosts, where haploid pycnia are formed and fertilized by pycniospores from another pycnium. Fertilized pycnia form aecia that produce dikaryotic aeciospores. The aeciospores are dispersed to oats and other grass hosts in early summer (June in Sweden) where uredinia containing urediniospores are formed. The first disease symptoms of crown rust in oats are usually observed three to 4 weeks later.

Early studies by Eriksson and Henning (1896) showed that urediniospores from one grass host were unable to infect other grass hosts, and consequently divided the pathogen into different formae speciales. The type infecting oats was named P. coronata f. sp. avenae. Gäumann (1959) reported that the formae speciales infecting oats and some wild grasses (f. sp. avenae, f. sp. agropyri, f. sp. arrhenatheri, f. sp. bromi, f. sp. festucae, f. sp. lolii) infected the alternate host R. cathartica while other formae speciales (f. sp. agrostis, f. sp. calamagrostis, f. sp. holci and f. sp. phalaridis) infected the alternate host Frangula alnus. Phylogenetic studies of other cereal rusts have shown clear genetic differentiation between formae speciales (Abbasi et al. 2005; Liu and Hambleton 2010; Liu and Hambleton 2013). Leonard (2007) reported high phenotypic diversity in terms of numbers of virulence phenotypes in both aecial and uredinial populations of P. coronata in North America. Furthermore, the diversity in pathogenicity of a populations of P. coronata f.sp. avenae sampled in Minnesota, in proximity to the alternate host R. cathartica, was found higher than that of a population sampled in Texas, far away from the alternate host (Simons et al. 1979). In addition, studies of the genetic diversity of the rust fungi on their aecial host showed high genetic diversity within each aecial cluster, with each aecium (aecial cup) potentially producing spores of distinct genotypes (Anikster et al. 1999; Berlin et al. 2017; Rodriguez-Algaba et al. 2017), illustrating the diversity within cereal rusts in the presence of their alternate hosts.

The objective of this study was to investigate the role of the aecial hosts of P. coronata in the epidemiology of oat crown rust with the aim to improve the understanding of this pathosystem. We hypothesized that genotypes of P. coronata belonging to the same population can be found on both R. cathartica and oats, and that the inoculum initiating crown rust epidemics is dispersed from aecial hosts in close proximity to infested oat fields.

Material and method

Samples from the aecial hosts were collected at two locations (59.2011 N, 15.5988 E and 59.2922 N, 15.0651 E) in July 2015. To infer the population structure and diversity of P. coronata infecting oats, samples from the uredinal host were collected in three oat fields in close proximity to the aecial host locations, Field L (59.2009 N,15.5983 E), Field F (59.2837 N, 15.0642 E) and Field N (59.3017 N, 15.0566 E). At location L, R. cathartica was present 350 m from the sampled section of the field. Field F was located 950 m south of the F. alnus sampling site, and Field N 1.2 km North West of the F. alnus site. The distance between the two sampling locations (field F and field N) was 28 km. 30 oat leaves clearly showing uredinial infections were collected in each field in a regular grid design. Sampling was done along three rows, 10 m apart, and 10 samples were taken in each row at a distance of 10 m. All samples were kept in separate paper bags and were air-dried before further processing.

For DNA extractions, large aecial clusters were cut into three sections, while small aecial cluster were kept intact. This was done to reduce the risk of having more than one genotype within a sample (Berlin et al. 2017). The samples were put into 2-ml plastic tubes together with 30 glass beads and diatomaceous earth and extracted using the OmniPrep kit (GenoTech. St. Louis) according to manufacturer’s instructions for fungal tissues with minor modifications as described by Berlin et al. (2013). For samples collected from oats, single uredinia were carefully picked from the infected leaves. The DNA extractions were performed according to the same procedure as for the aecial samples. All samples were analyzed using 12 microsatellite markers specifically developed for Puccinia coronata f. sp. avenae (Dambroski and Carson 2008) (Table 1). Amplification of the microsatellites were performed in a PCR reaction including each of the following to the final concentration of 2 ng μl−1 DNA template, 0.02 mM dNTP, 0.2 m μM of each primer, 2.75 mM MgCl2, 0.05 U μl−1 DreamTaq DNA polymerase and DreamTaq buffer according to the manufacturer’s recommendation. The PCR program included an initial denaturation at 94 °C for 5 min, followed by 35 cycles of denaturation at 94 °C, annealing at 57 °C, and extension at 72 °C, each for 30 s, and a final extension step for 7 min at 72 °C. The PCR products were analysed by capillary electrophoresis (ABI 3730 XL DNA Analyzer). The length of the fragments was determined using GeneMarker (Softgenetics) and compiled to identify the multi locus genotype (MLG) of each sample.

Table 1 Microsatellite markers developed for Puccinia coronata f. sp. avenae by Dambroski and Carson (2008) used in this study, the alleles detected on the different hosts and their calculated expected heterozygosity and evenness
Full size table

The markers’ suitability to infer the population structure in this particular study was evaluated by calculating the expected heterozygosity (Hexp), evenness of the allelic distribution and possible deviation from Hardy-Weinberg equilibrium (Kamvar et al. 2014). Genotypic accumulation curves were calculated to confirm that adequate numbers of loci were included in this study (Kamvar et al. 2014). For the population biology analysis, samples were grouped based on host and location of the sampled oat fields. The genetic and genotypic diversity for each strata and strata combination were calculated in the R package Poppr (Kamvar et al. 2014), and the genotypic diversity is reported as the number of unique MLGs and expected heterozygosity (Hexp). The linkage disequilibrium was evaluated by calculating the \( \overline{r} \)D values (Agapow and Burt 2001) on clone corrected data to accommodate for unequal sample sizes (Lott et al. 2010). To determine the population differentiation between the different hosts, an analysis of molecular variance (AMOVA) between the different strata was performed using Poppr (Kamvar et al. 2014). The calculations of pairwise FST was performed using GenAlEx 6.51 (Peakall and Smouse 2012) and based on the heterozygosity between individuals within each population. The significance of each values was calculated by 9999 permutations. In addition, the Bayesian clustering software Structure version 2.3.4 (Pritchard et al. 2000) was used to infer the common ancestry among the different samples. The assumptions for this analysis were set to mixed populations with independent alleles. The range of presumed number of genotypic clusters (K) was 1–12 and the program was set to a burn-in period of 300,000 iterations followed by 300,000 iterations. The software calculates the Log likelihood for each K, and the highest likelihood was chosen as the most likely number of genotypic clusters. Five independent simulations were performed to test the consistency of the results. The most likely number of genetic groups was determined based on the Evanno method (Evanno et al. 2005) using STRUCTURE HARVESTER (Earl and von Holdt 2012).

Results

All 12 microsatellite loci used in this study were informative (Table 1) and found to be at Hardy-Weinberg equilibrium. The genotypic accumulation curve for the full data set was saturated at nine loci (data not shown). A total of 188 alleles were detected, ranging from 8 to 26 alleles by locus (Table 1).

In total, 143 P. coronata samples were successfully genotyped and 142 unique MLGs detected (Table 2). Samples were omitted if there was a suspicion of mixture of genotypes or if the amplifications were unsuccessful. The samples collected from F. alnus clearly differed from the samples collected from R. cathartica and A. sativa. Two loci (PcaSSRA04 and PcaSSRA66) completely failed to amplify samples collected from F. alnus. For the loci PcaSSRB25, PcaSSRC26, PcaSSRC76 and PcaSSRC52, samples from one of the F. alnus sites failed to amplify the loci using these primers (Table 1). Result from the Bayesian cluster analysis showed that number of inferred clusters, K = 3 had the strongest support according to the Evanno method (Evanno et al. 2005), clearly dividing the samples into different populations based on the host from where it was collected. In Fig. 1, K 2 to 4 is presented, highlighting the robust population differentiation. The variation within and between the different hosts was examined by AMOVA. The largest variation was found within samples (71%), followed by the variation between hosts (25%) (data not shown). The pairwise FST between populations based on hots ranged from 0.296 to 0.401 and were all significant (Table 4). The large values reflects the differentiation between the populations visualized in the Bayesian cluster analysis (Fig. 1).

Table 2 Population statistics of Puccinia coronata collected from the telial hosts Avena sativa and from the two aecial hosts Rhamnus cathartica and Frangula alnus. Statistical analyses are based on clone corrected data
Full size table
Fig. 1
figure 1

Bayesian cluster analysis for K 2–4. When K = 2, the populations are divided between oats and the alternate hosts. At K = 3 and onwards, the differentiation between the alternate hosts R. cathartica and F. alnus becomes evident

Full size image

The population genetic data for the oat fields clearly shows a large genotypic diversity, as all MLGs were unique except for one MLG detected twice (Table 3). In addition, no linkage disequilibrium was observed in any of the three oat field populations, as the \( \overline{r} \)D values were not significant (Table 3). The variation within and between the three oat fields was examined by AMOVA. The largest variation was found within each MLG (99%), and no differentiation was found between fields (data not shown). The Bayesian clustering analysis confirms that all samples collected from the three different oat fields belonged to the same population when compared to samples from the other hosts (Fig. 1).

Table 3 Population statistics of the P. coronata f. sp. avenae populations collected in three fields
Full size table

Discussion

In this study, P. coronata sampled from oat fields and from the two alternate hosts R. cathartica and F. alnus was shown to differentiate into genetically discrete groups (Tables 1 and 4, Fig. 1). A clear effect of host origin was observed as only the samples from F. alnus were split into two genetic clusters (Fig. 1), possibly caused by a large number of missing values due to the markers’ inability to amplify the targeted regions within that particular group of samples. Several formae speciales may simultaneously infect the same alternate hosts (Dracatos et al. 2010). The 12 microsatellite markers selected in our study were developed for P. coronata f. sp. avenae (Dambroski and Carson 2008), and all markers amplified samples collected on oats and most samples collected from R. cathartica, but performed less well on samples from F. alnus. The majority of alleles were shared between P. coronata infecting R. cathartica and oats (Table 1). Since the allele sizes detected in the samples from A. sativa and R. cathartica did not overlap with the ones from F. alnus, it is evident that these genetic groups are genetically different (Tables 1 and 4 ; Fig. 1). A reason for this could be that formae speciales infecting F. alnus are genetically more distant to P. coronata f. sp. avenae than the formae speciales infecting the same alternate host, R. cathartica. The established concept that P. coronata f. sp. avenae does not complete its sexual cycle on F. alnus (Gäumann 1959), thus excluding F. alnus as an alternate host in the oat crown rust epidemiology, is hereby supported. Concurrently, the absence of linkage disequilibrium suggests that P. coronata f. sp. avenae undergoes sexual reproduction in Sweden (Table 3).

Table 4 Pariwise FST between samples collected from the three hosts, Avena sativa, Rhamnus cathartica and Frangula alnus are presented below the diagonal and the level of significance above the diagonal (9999 permutations)
Full size table

Earlier studies show that common buckthorn, R. cathartica is the aecial host of P. coronata f. sp. avenae (Gäumann 1959; Anikster and Wahl 1979), and it was hypothesized that at least a proportion of the genotypes collected from R. cathartica and oats respectively would belong to the same population. This genetic link was not detected in this study, indicating that the importance of R. cathartica in the epidemiology of oat crown rust is unclear. The absence of a genetic link between the samples from oats and R. cathartica underlines that R. cathartica is host also for other formae speciales of P. coronata (Dracatos et al. 2010; Gäumann 1959). Both R. cathartica and F. alnus are prevalent within the studied agricultural area (Mossberg and Stenberg 2003), contributing to the possibility for the fungus to perform its sexual cycle between seasons. Increased prevalence of oat crown rust in recent years is likely linked to changes in agricultural practices such as increased practice of no tillage and an extensive use of oats as cover crop when establishing grass and legumes for forage ley, leaving infected crop residues in the field. In spring, the teliospores germinate into short-lived basidiospores that are released under wet conditions such as dew or rain showers, and infect the alternate hosts in the close surroundings (Zhao et al. 2016). Possibly, the sampled bushes of aecial hosts were not surrounded by oats the year prior to sampling, limiting the infections from P. coronata f. sp. avenae basidiospores on the alternate hosts. The missing genetic link between samples collected on R. cathartica and P. coronata found in the field in close proximity to the alternate host suggests that the inoculum originated from a more distant source. Thus, only a fraction of the large diversity within P. coronata aeciospores released from alternate hosts will constitute the initial source of inoculum causing oat crown rust. In a previous study of the stem rust pathogen P. graminis, Berlin et al. (2012) reported that samples of P. graminis collected from the aecial host barberry, harboured a range of different populations, and only few samples grouped in the same populations as samples collected from oats and rye (Secale cerealis). To determine the role of the aecial host for P. coronata f. sp. avenae, an extensive number of samples from potential aecial hosts need to be genotyped, in order to detect genotypes shared between populations collected from aecial hosts on one side, and populations of P. coronata f. sp. avenae collected from its uredinial host on the other side.

Most formae speciales of P. coronata have R. cathartica as aecial host (Gäumann 1959), and the aecial genotypes found in this study on R. cathartica and on F. alnus probably belonged to formae speciales other than P. coronata f. sp. avenae. From our results, it is also evident that the aecial hosts adjacent to an infested oat field are not the only source of inoculum for a particular field. Other aecial hosts within the area, or even within a region, will contribute to the pathogen diversity within a field. Basidiospores cannot disperse far because they are colorless, sensitive to UV light, and to dry conditions (Zhao et al. 2016). Aeciospores have a limited capacity of dispersal and primarily infect grass hosts in close proximity to the aecial host (Roelfs 1985). The urediniospores may spread for hundreds of kilometers (Nagarajan and Singh 1990), as they are colored and adapted to long distance dispersal (Simons 1970), hence, at the uredinial stage, genotypes may further spread between distant oat fields. Since the samples collected from different oat fields in this study belonged to the same P. coronata f. sp. avenae population, we suggest that the dispersal of inoculum initiating crown rust epidemics is not limited to fields in close proximity to the aecial hosts.