Wheat (Triticum aestivum L.) is one of the most important sources of human nutrition and plays a critical role in global food security. The global wheat production reached 760.7 million tons during 2019, and 31% of it came from the two Asian countries, i.e., China and India, ranked first and second in global wheat production, respectively (FAO, 2021). Wheat production has been continuously threatened by several abiotic and biotic factors including diseases. Stem (black) rust caused by Puccinia graminis f. sp. tritici Eriks. and Henn. (Pgt) is a highly devastating disease of wheat and could result in wheat yield losses up to 100% on susceptible wheat cultivars (Anonymous, 1982; Worku et al., 2016). It is an obligate biotrophic, macrocyclic, and heteroecious fungus. The Pgt population comprises pathotypes with a highly diverse virulence nature or in some cases could be extremely uniform, which can disperse over long distances through wind and evolve new pathotypes that could overcome the stem rust resistance in wheat cultivars (Singh et al., 2015; Prasad et al., 2016).

Stem rust epidemics have occurred in the majority of wheat-growing regions of the world including East Africa, Eastern Europe, Australia, North America, Canada, the United States, and India (Dill-Macky et al., 1990; Eversmeyer and Kramer, 2000; Prasad et al., 2017b,2020). In India, stem rust is known to affect wheat crops in approximately 7 Mha areas covering the Central and Southern parts of the country. Historically, severe stem rust outbreak on Einkorn wheat was reported during 1907, followed by epidemics of 1946–1947 and 1948–1949 in Central and Southern India, respectively (Mehta, 1952; Prasad et al., 2018). With the help of strong disease surveillance, availability of suitable stem rust-resistant wheat cultivars, and high varietal replacement rate, the rusts have been kept under check; and no wheat rust epidemics were reported from any of the wheat-growing areas from India in the post green revolution period. Mehta (1952) reported a total loss of approximately 60 million Indian rupees annually in wheat and barley due to rust diseases in India. Stem rust resistance genes Sr2, Sr5, Sr6, Sr7a, Sr7b, Sr8a, Sr8b, Sr9b, Sr9e, Sr11, Sr12, Sr13, Sr17, Sr21, Sr24, Sr25, Sr30, and Sr31 are very common in Indian wheat cultivars. Among these, Sr2, Sr11, and Sr31 are more frequent in bread wheat, whereas Sr7b, Sr9e, Sr11, and Sr13 have been postulated in many durum cultivars (Jain et al., 2013). The emergence of TTKSK (also known as Ug99), one of the most virulent forms of Pgt, was first identified from Uganda in 1998 (Pretorius et al., 2000) and its variants (15 races), reported in 14 countries including Egypt, Yemen, and Iran (Singh et al., 2015; https://rusttracker.cimmyt.org/?page_id=22) and has rendered 90% of the world wheat varieties susceptible to stem rust (Singh et al., 2015; Worku et al., 2016; Soko et al., 2018). Another stem rust race, namely, Digalu (TKTTF) caused 100% yield loss in many Ethiopian regions during 2013–2014 and 2014–2015 (Olivera et al., 2015). Thus, stem rust has become a massive menace for world wheat production and has threatened global food security (Patpour et al., 2016; Prasad et al., 2016; Olivera et al., 2019).

The stem rust management in wheat primarily relies on the cultivation of disease-resistant varieties in combination with appropriate agronomic practices and the use of fungicides (Bhardwaj et al., 2016; Savadi et al., 2017). However, race-specific resistance is often overcome by the frequent introduction or evolution of new Pgt pathotypes through mutation, somatic recombination, or sexual recombination (Patpour et al., 2016). Therefore, regular monitoring of variations in the Pgt populations is essential for the identification of new resistance sources and anticipatory breeding for rust resistance to effectively manage the newly emerging Pgt isolates. The understanding of the virulence and genetic diversity, distribution, and population structure of the Pgt population allows effective monitoring of the emerging stem rust challenge, which is critical in formulating wheat stem rust management strategies (Abebe et al., 2012).

Virulence-based phenotyping of Pgt races in India started way back in 1932, and since then, 29 pathotypes of Pgt have been identified from the Indian subcontinent based on their pathogenicity on Indian stem rust differentials (Prasad et al., 2018). However, the genetic analysis of these pathotypes using molecular markers is not reported till date. Genetic analysis and genotypic variability of plant-pathogen populations have assisted researchers to reveal the genetic relationship among different pathotypes and track the pathways by which these pathogens evolve over time and space (Olivera et al., 2015; Prasad et al., 2017a). The simple sequence repeat (SSR) or microsatellite markers are most extensively used and established for exploring population dynamics and evolutionary patterns of Puccinia spp. due to the multi-allelic, reproducible, and highly polymorphic nature of these markers (Zhong et al., 2009; Savadi et al., 2020). This study was conducted to explore the virulence-based phenotype and SSR marker-based genotype patterns, population structure, and other diversity parameters in a repository of 29 Pgt pathotypes collected during the past ∼90 years from India and neighboring countries and maintained in vivo at ICAR-Indian Institute of Wheat and Barley Research, Regional Station (ICAR-IIWBR, RS), Shimla, India. Every year 100–200 isolates of Pgt collected from different wheat stem rust-prone areas in the Indian subcontinent are analyzed. Since the alternate host for stem rust is not functional, only 29 pathotypes have been identified in Pgt from this region. These pathotypes have been identified from different locations and years. The information on the distribution of these pathotypes during different years and locations is being published regularly, and the latest was by Prasad et al. (2018).

Puccinia graminis tritici (Pgt), causing wheat stem rust, is highly destructive and one of the most extensively studied plant pathogens. Stem rust is a serious threat to world wheat production and food security. The emergence and dispersal of Pgt races, such as TTKSK (Ug99), TTRTF, TKTTF (Digalu race), and their variants, have posed a serious risk to global wheat production (Pretorius et al., 2000; Olivera et al., 2015; Patpour et al., 2020). The occurrence of such virulence has not been reported from any wheat-growing regions in the Indian subcontinent; however, strategic monitoring to track the change in the virulence pattern of Pgt is in place for early detection of likely new virulence (s), which might be introduced from elsewhere or evolved locally.

This study reports virulence phenotype and SSR genotype-based diversity analysis of Pgt repository (29 pathotypes) from the Indian subcontinent. Virulence phenotype data revealed both inter- and intra-lineage diverse nature of all 29 pathotypes on stem rust differentials and other Sr genes (Table 1). The virulence frequencies of Pgt pathotypes on stem rust differentials ranged between 0% and 100%. It was in the range of 20%–60% for the majority of the differentials. Ten Sr genes (i.e., Sr26, Sr27, Sr31, Sr32, Sr33, Sr39, Sr40, Sr43, SrTmp, and SrTt3) were found effective to the Indian population of Pgt, whereas virulence for Sr24 and Sr25 was found only in two (34-1 and 40-1) and one pathotypes (40-2), respectively. Among the effective all stage resistance (ASR) genes in India, seven (Sr26, Sr27, Sr32, Sr33, Sr39, Sr40, and Sr43) confer resistance to Ug99 and other variants in its lineage and are possibly the most useful ASR genes (Singh et al., 2015). Some of these genes, Sr26 in particular, have performed well against stem rust pathotypes worldwide and, therefore, could be a valuable asset for utilization in breeding programs if used in combinations with other effective genes (Singh et al., 2015; Terefe et al., 2016). Moreover, the availability of promising slow-rusting adult plant resistance (APR) genes such as Sr2, Sr55, Sr56, Sr57, Sr58, and others, and their pyramiding with effective ASR genes in cultivars with superior agronomic traits could further improve the effectiveness and durability of stem rust resistance (Huerta-Espino et al., 2020).

Virulence phenotype-based cluster analysis grouped Pgt pathotypes into five distinct VGs (I–V). All the pathotypes in race 117 lineage were in VG I (Figure 3). The clustering of pathotypes 40A, 40-1, and 40-3 justifies their similar virulence phenotypes on majority of the differentials except for: (i) Sr24 (virulent: 40-1; avirulent: 40A and 40-3), (ii), Sr13, and (iii) Sr30 (virulent: 40-3; avirulent: 40A and 40-1; for both Sr13 and Sr30). The identical virulence phenotypes of 40A and 40-1, except for Sr24 (virulent: 40-1), suggested and have been established that pathotype 40-1 had evolved through a single-step gain of virulence for Sr24 in 40A (Bhardwaj et al., 1990). Other pathotypes in the same race lineage and belonging to one VG include 184 and 181-1 in VG II, 122 and 295 in VG IV, and 21, 21A-2, and 21-1 in subgroup IIIa. Likewise, pathotypes 11 and 11A were clustered in subgroups IIIa and IIIb (VG III), respectively.

The possible evolutionary lineages deduced based on the virulence-avirulence assays of 29 Pgt pathotypes classified them into seven lineages. In race 11 lineage, the first pathotype was 11 (virulence for Sr11 and Sr30), which was detected in 1962 from Maharashtra (India). Subsequently, another pathotype in its lineage with additional virulence on Sr37 was identified in 1974 as 11A (Sinha et al., 1976). Other difference between pathotypes 11 and 11A is that the latter has an avirulence phenotype on Sr11. In race 21 lineage, the first pathotypes to be identified was 21, which was simultaneously detected from Lyallpur, Pakistan, and Himachal Pradesh (India) in 1935. The one-step gain in virulence for Sr9b and Sr30 in pathotype 21 caused the emergence of 21A-2 in 1962 from Madhya Pradesh (India). Likewise, additional virulence for Sr8b and loss of virulence for Sr13 in pathotype 21 resulted in the emergence of 21-1 in 1985 from Uttar Pradesh (India) (Bahadur et al., 1985b). Pathotype 40 in race 40 lineage was detected in 1932 from Pune (India; Mehta, 1940), while a new independent pathotype 40A emerged in the Nilgiri Hills (Wellington, India) in 1974 (Sharma et al., 1975). Additional virulence for Sr24 in pathotype 40A resulted in the emergence of 40-1 (Bhardwaj et al., 1990), while simultaneous gain and loss of virulence for Sr25 and Sr11, respectively, in 40A resulted in the emergence of 40-2 (Jain et al., 2009). Other pathotypes 40-3 and 15-1 in race 40 lineage emerged from Dharwad (Karnataka) in 2008, resulting from the gain of virulence for Sr13 and Sr30 (for both 15-1 and 40-3) and loss of virulence for Sr11 (for 15-1) in 40A (Jain et al., 2013).

Race 117 lineage has been the most variable collection of Pgt pathotypes in the Indian subcontinent. Pathotype 117 in this lineage was first detected from Madhya Pradesh in 1945 (Prasada and Lele, 1952). Since then, eight additional variants of this pathotype have been detected. The loss of virulence for Sr7b and Sr13 in pathotype 117 resulted in the emergence of 117A in 1960 from Karnataka (Patil et al., 1963). Consequently, pathotypes 117-2 and 117-6 evolved as a result of loss of virulence to Sr11 (Bhardwaj et al., 1990) and one step gain of virulence on Charter (Bhardwaj et al., 1994), respectively, in pathotype 117. Additional virulence in pathotype 117A for Sr9b and Charter resulted in the emergence of 117A-1 (Sharma et al., 1979), while the gain in virulence for Sr37 in 117A contributed to the advent of 117-1 (Bhardwaj et al., 1989). Added virulence for Sr7b and NI5439 in pathotype 117-1 resulted in the emergence of 117-4 and 117-5, respectively, whereas pathotype 117-1 advanced to 117-3 after gaining additional virulence for Sr7b and Sr13 (Bhardwaj et al., 1994). The gain in virulence was the principal mode of evolution in the race 117 lineage except for the emergence of 117A (117-Sr13 and Sr7b) and 117-2 (117-Sr11), where the loss of virulence in pathotype 117 resulted in their evolution. The virulence-based hypothetical evolutionary hierarchy proposed in this study is to a great extent in agreement with previously discussed virulence phenotype data-based NJ cluster analysis. The evolutionary behavior of Pgt isolates reveals that the gain of virulence for Sr13 was most frequent. The resistance genes present in the most widely used cultivars always play a critical role in influencing the prevalence of pathogen races in a specific region or location. The Sr13, one of the most widely used Sr genes in durum wheat, is very common in Indian cultivars as revealed by gene postulation using the gene matching technique. Moreover, the areas covering Madhya Pradesh, Karnataka, Maharashtra, and Tamil Nadu states, where most of the new pathotypes were initially detected, have all three types of wheat (i.e., aestivum, durum, and dicoccum) in cultivation. Adequate host diversity among aestivum, durum, and dicoccum wheat cultivars might be facilitating easier survival for newly evolved rust isolates.

The SSR genotype pattern of each of 29 Pgt pathotypes was estimated using 37 Puccinia spp. specific SSR markers developed in previous studies (Karaoglu et al., 2013; Prasad et al., 2017a, 2018; Savadi et al., 2020). Four of these markers (i.e., PGTG 03066, PGTG 04483, PGTG 00856, and PGTG 07438) were gene-based functional SSRs, which are known to be most reliable and reproducible for genetic diversity analysis (Anderson and Lubberstedt, 2003; Zhong et al., 2009). The MAF, gene diversity, heterozygosity, and PIC values indicated that the SSR markers used in this study were highly informative and, therefore, yielded good quality genetic diversity analysis data. Thirty-two SSRs had PIC values more than 0.50, which is considered the most appropriate for conducting genetic diversity analysis (Botstein et al., 1980; Szabo, 2007).

The cluster analysis based on the SSR genotype data grouped 29 Pgt pathotypes into five MGs, of which MGs I and III were further divided into three and two subgroups, respectively (Figure 5). The results of the PCoA explaining 25.11% cumulative variance corroborated the NJ cluster analysis. The Pgt pathotype grouping as explained through NJ cluster analysis and PCoA was entirely independent of their geographical distribution. For instance, the pathotypes from different regions, i.e., 117/117A-1 (Karnataka), 42B (Maharashtra), and 14 (Madhya Pradesh) were grouped in cluster MG-V, which reflected high genetic similarity among them (Figure 5). Similarly, pathotypes 21, 21-1, and 21A-2 shared almost analogous virulence phenotype and SSR genotypes but detected in distal geographical regions, i.e., 21 and 21-1 in North India and 21A-2 in Central India. This relationship of the geographical distribution of Pgt pathotypes with their SSR genotypes could also be endorsed by the findings of AMOVA (Supplementary Figure 2 and Supplementary Table 4). An explanation to justify such distribution of Pgt pathotypes could be the existence of gene flow (the transfer of genetic information from one population to the other) over long distances. Similar findings for cereal rust pathogens are reported from India, South America, North America, and Europe (McCallum et al., 1999; Ordonez and Kolmer, 2007; Prasad et al., 2017a). Conversely, clonal lineages were described for Puccinia spp. evolution from Australia {Pgt (Zhang et al., 2017), P. coronata f. sp. avenae (Brake et al., 2001), and P. graminis f. sp. avenae (Keiper et al., 2006)} and South Africa {Pgt (Visser et al., 2009)}.

The correlation analysis of virulence phenotype and SSR genotype data of all Pgt pathotypes revealed a positive (R² = 0.1546) correlation between these two variables, which suggests a poor correlation between the virulence phenotypes and SSR genotypes. This is logical because the SSRs used in this study were unrelated to pathogenicity/virulence-specific genes. Similar findings were observed in our earlier studies (Prasad et al., 2017a,2018; Gangwar et al., 2021). Such correlation was also witnessed in the Puccinia striiformis population by Chen et al. (2010), who clarified that SSR genotypes were independent of virulence and that the whole genomic region of the pathogen evolved much faster than its pathogenicity genes. In contrast, a strong correlation between virulence phenotype and SSR genotype data was observed in a population of Puccinia triticina from China (Kolmer, 2015) and Canada (Wang et al., 2010) and of P. striiformis from Europe (Hovmoller et al., 2002) populations. Such a positive strong correlation could be the result of SSR genotypic data acquired by the markers designed from virulence/pathogenicity genes, which are distributed throughout the pathogen genome (Ordonez and Kolmer, 2009; Kolmer, 2015). Moderate correlation (0.43; P = 1.0) between virulence phenotypes and SSR genotypes was reported for the P. triticina population from North America (Ordonez and Kolmer, 2009).

The value of different genetic diversity parameters, namely, Na, Ne, I, He, uHe, and %P was maximum for the Pgt population from Maharashtra and minimum for North India. These results suggest that the Pgt population is highly diverse in Maharashtra followed by Tamil Nadu, while the North Indian population has poor genetic diversity. These findings support our previous results for P. triticina and Pgt populations in the Indian subcontinent, wherein we documented that wheat cultivation round the year and availability of green bridges in higher hills of Tamil Nadu (Nilgiris), which receives high UV intensity sunlight, might be helping these pathogens to mutate and survive without difficulty (Prasad et al., 2017a,2018). All these factors are known to be critical for the evolution and survival of new virulences in nature (Cheng et al., 2014; Singh et al., 2015; Zhao et al., 2020). The new Pgt isolates evolved in these hills migrate to the foothills of Tamil Nadu and Karnataka, and subsequently, to plains of peninsular India (Karnataka, Maharashtra) and Central India (Madhya Pradesh), where these are established on susceptible wheat cultivars (Mehta, 1952). There are speculations that the primary source of inoculum of Pgt is also present in parts of Karnataka other than Nilgiris (Bhardwaj et al., 1989). Furthermore, the cultivation of susceptible bread, durum, and diacoccum wheat species in these areas might be helping an array of Pgt pathotypes with a diverse avirulence/virulence formula to survive and evolve locally. In contrast, cultivation of genetically uniform wheat species (mostly aestivum) in a small stem rust-prone geographical area in North India together with a short favorable period for stem rust development followed by harsh summers prohibits Pgt isolates to evolve and survive until the next crop season, which is reflected by poor genetic diversity within Pgt population in this region (Bhardwaj et al., 2019). The SSR genotype-based population structure analysis using a model-based tool clustered 29 Pgt pathotypes into two major subpopulations (i.e., SP1 and SP2), containing eleven pathotypes in each, and an admixture comprising seven pathotypes (i.e., 184, 117, 184-1, 34-1, 14, 40-1, and 42). Similar differentiation comprising subpopulations and admixture, using population structure analysis, was reported for the Pgt population from South Africa (Terefe et al., 2016; Boshoff et al., 2018) and P. graminis f. sp. avenae population from Australia, Brazil, and Sweden (Gnocato et al., 2018). However, four subpopulations were explained by a combined analysis of Pgt isolates from South Africa and Australia (Visser et al., 2019). Like cluster analysis, PCoA, and AMOVA, the grouping of Pgt pathotypes explained by population structure analysis was not related to their geographical distribution. For example, pathotypes 117-2, 21, and 184 belonged to the North India region while population structure analysis classified them in SP1, SP2, and admixture, respectively. Our results comprehend that the Pgt population in the Indian subcontinent might have evolved from two clonal events, and subsequently, gene flow, long-distance migration, and somatic hybridizations could have contributed to the development of the admixture population, which is not exceptional in Puccinia spp. (Park et al., 1999; Park and Wellings, 2012). The low Nei’s genetic distance between different population combinations indicated that the Pgt population belonging to different geographical regions in the Indian subcontinent are closely related and thus explain poor genetic divergence. Similar findings are reported for two P. striiformis populations from Bajaur and Malakand in Pakistan, while substantial genetic divergence was observed between P. striiformis populations from Nepal, Pakistan, and Bhutan (Khan et al., 2019). Likewise, genetic divergence among P. triticina isolates collected from Aegilops speltoides, durum, and bread wheat; delimited by host specificity was reported from Israel (Liu et al., 2014). Genetic drift, gene flow, accumulated mutation, adaptation, and host specificity are suggested as the driving forces behind genetic divergence among Puccinia spp. (Kohn, 2005; Kolmer et al., 2020).

The repository of 29 Pgt pathotypes collected during the last ∼90 years exhibited significant variability for virulence phenotype and SSR genotypes. The SSR markers used in this study were highly informative in explaining the genetic diversity among Pgt pathotypes. As revealed by different statistical programs and hypothetical evolutionary hierarchy, the gene flow, long-distance dispersal, and single-step mutation for gain and/or rare loss of virulence could have contributed to cryptic genetic variation and evolution in the Pgt population from the Indian subcontinent. The findings of population structure and genetic diversity analysis of Pgt population, supplemented by deployment of effective Sr genes (Sr26, Sr27, Sr31, Sr32, Sr33, Sr35, Sr39, Sr40, Sr43, SrTmp, and SrTt3) in combination with other APR genes, could be utilized for strategic wheat stem rust management.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

PP, SS, and SB conceived the idea and designed the experiments. PP, RT, and SK performed the virulence phenotyping. PP and RT performed SSR genotyping. PP, SS, OG, CL, and SA performed the data analysis and prepared first draft of the manuscript. PP and SB reviewed and revised the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.