Deciphering Genotype-by-Environment Interaction to Identify Target Test Environments and Stable Stripe Rust-Resistant Wheat Genotypes
MohammadWarisHaider1✉Emailmohammad.waris99@gmail.com
JaspalKaur2✉Emailjassu75@pau.edu
RituBala2
AchlaSharma2
DineshKumarSaini3
JyotiKumari4
1
A
A
Department of Plant PathologyPAULudhianaIndia2Department of Plant Breeding & GeneticsPAULudhianaIndia
3Department of Plant and Soil ScienceTexas Tech UniversityTexasUSA
4National Bureau of Plant Genetics and Natural ResourcesNew DelhiIndia
Mohammad Waris Haider 1*, Jaspal Kaur 2*, Ritu Bala 2, Achla Sharma 2, Dinesh Kumar Saini 3 & Jyoti Kumari 4
1. Department of Plant Pathology, PAU, Ludhiana, India
2. Department of Plant Breeding & Genetics, PAU, Ludhiana, India
3. Department of Plant and Soil Science, Texas Tech University, Texas, USA;
4. National Bureau of Plant Genetics and Natural Resources, New Delhi, India
*Corresponding author(s). E-mail(s): jassu75@pau.edu; mohammad.waris99@gmail.com
Abstract
Climate change necessitates innovative approaches to account for environmental effects in future wheat breeding programs aimed at managing stripe rust. The most effective and environmentally friendly strategy for managing stripe rust is the incorporation of resistance genes into the background of elite cultivars. In this multi-environment evaluation, the GGE biplot approach was used to identify genotypes with stable resistance against stripe rust. A total of 441 germplasm entries were initially tested for their seedling responses against the most prevalent pathotypes of Puccinia striiformis f. sp. tritici (Pst) (viz., 238S119, 110S119, and 46S119). Based on infection-type data, 30 lines were highly resistant to all three pathotypes, while 27 were highly susceptible. Field responses were evaluated based on final rust severity (FRS) and the area under the disease progress curve (AUDPC) observed over a period of three years at Ludhiana and Gurdaspur. The lines were categorized as highly resistant, resistant, moderately resistant, moderately susceptible, or susceptible. GGE biplot analysis revealed inconsistency in genotype performance across environments. The first two principal components explained 93.2% of the total variation (87.87% and 5.33% by PC1 and PC2, respectively). IC111939 (G1) was identified as the “ideal” genotype, with a small projection from the average environment coordination (AEC) abscissa and a moderate resistance response to stripe rust. Ludhiana (2019-20) and Gurdaspur (2018-19) had small angles with the AEC abscissa, indicating a positive association, with Ludhiana showing the highest “desirability index” and serving as an “ideal” test site for screening rust resistance. Environmental effects were analysed using step-wise regression, revealing that maximum temperature, relative humidity, location, and the years 2019 and 2020 had positive effects, while 2021 had a negative effect on disease severity.
Key words:
stripe rust
wheat
AUDPC
resistance
susceptible
A
1. Introduction
A
Wheat (Triticum aestivum L.) is one of the most widely cultivated and essential staple food crops consumed globally (Igrejas and Branlard 2020). In India, wheat is the second most important crop after rice, grown on 27.505 million hectares with an annual production of 113.292 million tonnes during 2023–2024 (Ministry of Agriculture & Farmers’ Welfare 2024). Often referred to as the world’s breadbasket, wheat is constantly threatened by numerous diseases, including three types of rust caused by Puccinia spp. Among these, stripe rust (Puccinia striiformis f. sp. Tritici; Pst) is particularly devastating. It thrives in cooler temperatures and can infect crops from the early stages of growth to maturity (Chen 2005; Hovmøller et al. 2010). Stripe rust poses a serious threat to global wheat production, affecting 80% of cultivars and causing an estimated annual loss of 5 million tonnes, equivalent to USD 1 billion (Beddow et al. 2015). Between 1939 and 2016, 51 major epidemics of stripe rust were recorded globally, with yield losses ranging from 2% to complete crop failure (Chen et al. 2020). In India, severe outbreaks in Jammu and Kashmir, Himachal Pradesh, Uttar Pradesh, and the sub-mountainous regions of Punjab have led to significant yield losses (Saharan et al. 2015). Punjab alone reported a loss of INR 236 crore due to fungicide applications for managing the disease (Jindal et al. 2012). The current situation is even more concerning due to the emergence and rapid spread of more aggressive and genetically diverse populations of Pst that are capable of adapting to warmer temperatures (Milus et al. 2009; Hubbard et al. 2015; Hovmøller et al. 2016).Ongoing efforts to identify and integrate new sources of resistance against stripe rust in wheat breeding programs are essential to counter the evolving threat posed by Pst. Due to the fungus’s high genetic adaptability, several cultivars that were initially resistant to stripe rust have now lost their effectiveness. Therefore, it is imperative to continually incorporate novel genetic resistance sources into breeding objectives to develop improved and more resilient varieties (Milus et al. 2009; Kumar et al. 2023; Pal et al. 2022). Nevertheless, some cultivars have retained a certain level of resistance even after widespread cultivation, and such resistance is considered durable. Numerous studies have explored the quantitative nature of host resistance in various cereal-rust pathosystems by assessing disease severity, area under the disease progress curve (AUDPC), infection rate (r), and coefficient of infection (CI), particularly for slow rusting or adult plant resistance (Pathan and Park 2006; Shah et al. 2014; Singh et al. 2015). These parameters have proven effective in evaluating the degree of resistance in crops against rust pathogens. The epidemiology of stripe rust is influenced by both the presence of resistance genes in wheat cultivars and prevailing climatic conditions (Wellings et al. 2011). Factors such as temperature, humidity, rainfall, sunshine duration, and wind play a significant role in the occurrence, spread, and development of the disease (Sandhu et al., 2018). Understanding the interaction between weather and disease dynamics provides valuable insights into the onset, persistence, dissemination, and severity of stripe rust (Subba Rao et al., 1990).
An effective resistance breeding program requires a thorough understanding of environmental influences and genotype-by-environment interaction (GEI), which reflects the stability of both the pathosystem and host genotype across diverse locations. Identifying disease “hot spots” with consistent “repeatability” is critical for evaluating genotypes and determining their effectiveness in combating specific diseases. However, the influence of environmental conditions on host-pathogen interactions often complicates the identification and recommendation of genotypes with durable resistance (Forbes et al. 2005; Tekalign et al. 2017). Various stability analysis methods derived from multi-location trials across different crops have been widely utilized to assess GEI effects in the context of disease resistance (Robinson and Jalli 1999; Mukherjee et al. 2013). Regular monitoring of pathogen virulence is also essential for effective disease management and for guiding the incorporation of resistance genes (Wan et al. 2016). One strategy for breeding durable resistance involves identifying individual resistance components and integrating them as selection traits in new cultivars (Stuthman et al. 2007).
Considering the challenges posed by rapidly evolving Pst populations and the complexity introduced by GEI interactions, there is a pressing need to evaluate resistance stability across diverse environments using robust analytical tools. In this context, we hypothesize that multi-environment evaluation combined with GGE biplot analysis can identify stable resistance sources and ideal test locations. Accordingly, this study aimed to: (i) evaluate 441 wheat germplasm lines against three virulent Pst pathotypes at the seedling stage, (ii) assess adult plant resistance across locations and years, (iii) analyze genotype performance and stability using GGE biplots, and (iv) identify ideal genotypes and environments to guide resistance breeding efforts.
2. Materials and Methods
2.1 Plant material
A
A total of 441 wheat germplasm lines were obtained from the National Bureau of Plant Genetic Resources (NBPGR), New Delhi, India, along with three susceptible cultivars, including Agra Local, HD2967, and PBW343, which served as controls (Table S1). Seedling-stage evaluations were conducted by screening the germplasm lines against three prevalent stripe rust pathogen pathotypes: 46S119, 238S119, and 110S119, sourced from the IIWBR Regional Research Station, Flowerdale, Shimla. In addition, adult plant resistance (APR) assessments were carried out under field conditions at the Department of Plant Breeding and Genetics, Punjab Agricultural University (PAU), Ludhiana, India (30.9020° N, 75.8083° E) and its Regional Research Station in Gurdaspur, Punjab, India (32.0506° N, 75.4229° E) during the 2018-19, 2019-20, and 2020-21 crop seasons.2.2 Seedling stage evaluation
Wheat seedlings and susceptible checks were grown in plastic germination trays filled with a soil mixture consisting of sandy loam, cocopeat, farmyard manure, and vermicompost. The germplasm was sown in three sets, each designated for evaluation against one of the three stripe rust pathogen pathotypes: 46S119, 238S119, and 110S119. Regular irrigation was provided throughout the process. Ten-day-old seedling leaves were inoculated with urediniospores of the respective pathotypes and placed in separate polychambers. To promote successful infection, the trays were kept in a dew chamber under dark conditions for 48 hours before being transferred to a greenhouse with a photoperiod of 16 hours light and 8 hours darkness. Humidity levels were maintained through adequate irrigation. Seedling infection types were recorded 14 days post-inoculation using a 0–4 scale as described by Nayar et al. (1997) (refer Fig. 1).
Fig. 1
Infection types (0–4) for stripe rust at the seedling stage. The leaves show a range of infection responses, from immune to fully susceptible, progressing from left to right.
Adapted from McIntosh et al. (1995) under the Creative Commons license CC BY 4.0.
2.3 Multi-environment evaluation
A
Wheat germplasm lines were sown at both test locations during the second week of November in one-meter-long paired rows, with a row-to-row spacing of 20–22 cm. To ensure uniform distribution of rust inoculum, susceptible varieties (viz., Agra Local, HD2967, and PBW343) were interspersed after every 20 rows of test entries. Additional infector or spreader rows were planted around the periphery of the experimental plots. Under field conditions, a stripe rust epidemic was induced through repeated spray inoculations using Pst urediniospores (mixture of pathotypes). The urediniospores were extracted from infected leaves of susceptible varieties (viz., PBW343, Agra Local, A-9-30-1, and HD2967) immersed in water. The inoculum was prepared by suspending the urediniospores in 10 liters of water with a few drops of Tween-20 as a surfactant. Spray inoculations were conducted in the evening using an ultra-low volume sprayer on alternate days, beginning in mid-December and continuing until stripe rust symptoms appeared on the susceptible checks. Stripe rust development was monitored at regular intervals throughout the 2018–2021 cropping seasons, starting from the second week of January through the first week of March. Disease severity at the adult plant stage was recorded for each wheat germplasm line using the modified Cobb’s scale (refer Fig. 2). The area under the disease progress curve (AUDPC) was calculated using the following formulas-
Figure 2 Modified Cobb’s scale: (A) shows the actual percentage of leaf area covered by uredinia, while (B) illustrates the corresponding rust severity ratings according to the modified Cobb’s scale as described by Peterson et al. (1948). Adapted from Peterson et al. (1948) under the Creative Commons license CC BY 4.0.
2.4 GGE biplot analysis
GGE biplot analysis was performed using R software (R Core Team, 2013). This graphical method illustrates the interaction between genotype (G) and GEI effects, allowing for the assessment of the extent and nature of GE interaction in multi-location datasets. The biplot is generated by plotting the principal component scores of genotypes and environments, specifically, PC1 against PC2, derived from the singular value decomposition (SVD) of environment-centered data. Each element of the resulting matrix is estimated using the following formula (Yan et al. 2000; Yan and Kang 2003):
Where,
Yij = mean response of ith genotype (i = 1,...,I) in the jth environment (j = 1,..,J).
µ = grand mean.
ej = environment deviations from the grand mean.
λn = the eigen value of PC analysis axis.
γin and δjn = genotype and environment principal components scores for axis n
N = number of principal components retained in the model and
εij = Residual efect’’ ~ N (0, σ2)
The “average environment coordination” (AEC) view of the GGE biplot was used to assess genotype performance and stability. This view allows for the comparison of genotypes based on their mean disease scores and stability across environments within a defined “mega-environment” (Yan 2001; Yan 2002). A performance line passing through the origin of the biplot represents the mean genotype performance for rust severity, with the direction of the arrow indicating decreasing stability or increasing susceptibility (Yan and Falk 2002). Similarly, the “discriminating power vs. representativeness” view of the GGE biplot is utilized to evaluate test environments, where an “ideal” environment is both highly discriminating among genotypes and representative of the mega-environment (Yan et al. 2007). The “repeatability” of a test site was determined by calculating the mean genetic correlations across years to ensure consistent genotypic performance (Yan et al. 2011).
Additionally, the AEC view was used to assign each test site a “desirability score”, which accounts for the distance between test environments and the ideal genotype, as well as genotypic stability and adaptability (Yan and Holland 2010). Angles between environmental vectors were analyzed to assess the relationships among test locations and the correlations between environments (Yan and Kang 2003). To evaluate genotype performance across test environments and identify distinct “mega-environments”, a “which-won-where” view of the GGE biplot was also generated (Yan and Rajcan 2002). Furthermore, stepwise regression was utilized to evaluate the influence of environmental and other factors on disease severity. This statistical method iteratively adds or removes variables based on their significance, using criteria such as p-values, the Akaike Information Criterion (AIC), or the Bayesian Information Criterion (BIC). Let Y be the dependent variable and X the set of potential predictor variables. Starting from an empty model, variables are introduced or eliminated one at a time based on their contribution to model fit, continuing until a stopping criterion is met. Stepwise regression combines both forward selection and backward elimination to identify the most influential predictors while minimizing the risk of overfitting, thereby providing a systematic approach to model selection.
3. Results
3.1 Seedling evaluation
A
A
A
To conduct race-specific phenotyping, wheat germplasm entries (refer Table S1) were tested against three of the most virulent and prevalent Pst pathotypes: 110S119, 46S119, and 238S119. The results, based on infection type, revealed diverse resistance combinations (refer Fig. 3). Among the 441 wheat entries tested, 37 entries were highly resistant to pathotype 110S119, 108 were resistant, 97 showed resistant to moderately resistant reactions, 109 were moderately susceptible to susceptible, and 90 were highly susceptible (refer Table S2). Against pathotype 238S119, 42 lines were highly resistant, 85 were resistant, 83 were resistant to moderately resistant, 121 were moderately susceptible to susceptible, and 110 were highly susceptible (refer Table S3). For pathotype 46S119, 36 lines were highly resistant, 131 were resistant, 86 showed resistant to moderately resistant responses, 103 were moderately susceptible to susceptible, and 85 were highly susceptible (Table S4).Fig. 3
Seedling responses of wheat germplasm lines to three prevalent Pst pathotypes (viz., 46S119, 110S119, and 238S119). The histogram displays the number of lines falling into five infection type (IT) categories: 0; (highly resistant, HR), 1 (resistant, R), 2 (resistant to moderately resistant, R-MR), 3 (moderately susceptible to susceptible, MS-S), and 4 (highly susceptible, HS), based on their reaction to each pathotype.
The genetic relationships among wheat germplasm, as determined by their seedling responses to three stripe rust pathotypes, were analyzed. The resistance profiles were then compared, and the findings were visually represented in a Venn diagram (refer Fig. 4). This diagram is a powerful tool, clearly showing that some lines were resistant to only one race, some to two, and some to all three pathotypes, thereby providing a comprehensive view of the resistance profiles. Pathotype 46S119 exhibited 245 resistant lines, including 16 shared with 238S119, 34 shared with 110S119, and 47 unique to this race. Similarly, 238S119 had 210 resistant lines, of which 16 overlapped with 46S119, 19 with 110S119, and 27 were unique to this pathotype. Pathotype 110S119 showed 241 resistant lines, including 19 shared with 238S119, 34 with 46S119, and 41 unique to this race. Notably, 148 lines were resistant to all three pathotypes. These findings are critical for understanding the genetic basis of resistance to Pst and can aid the wheat resistant program with enhanced and broad-spectrum resistance.
Fig. 4
Venn diagram illustrating the distribution of resistant wheat germplasm lines across three Pst pathotypes: 46S119, 238S119, and 110S119.
A
A
A
A
A
A
Wheat germplasm lines were evaluated for adult plant response to stripe rust across two contrasting locations, including Ludhiana and Gurdaspur, over three consecutive seasons. Substantial variation in disease severity was observed among genotypes across locations and years (refer Fig. 5). Based on disease scores, the germplasm was grouped into four categories: highly resistant, resistant to moderately resistant, moderately susceptible, and highly susceptible. Classification was based on combined AUDPC and FRS values, which reflected the degree of resistance or susceptibility to stripe rust under field conditions. The consolidated data on FRS and AUDPC from Ludhiana for the 2018-19, 2019-20, and 2020-21 cropping seasons are presented in Fig. 5. The number of highly resistant germplasm lines recorded in these years were 48, 49, and 57, respectively. These were followed by 246, 245, and 274 lines classified as resistant to moderately resistant; 82, 72, and 64 lines as moderately susceptible to susceptible; and 65, 75, and 46 lines as highly susceptible (refer Tables S5, S6, and S7). At the Gurdaspur location, the corresponding numbers of highly resistant lines recorded during the same years were 70, 80, and 100. The resistant to moderately resistant categories included 256, 257, and 253 lines; the moderately susceptible to susceptible categories had 54, 42, and 54 lines; and the highly susceptible categories included 61, 62, and 34 lines, respectively (refer Tables S8, S9, and S10). A pooled analysis of seedling-stage data across the three tested Pst pathotypes and field performance across both locations over three years identified 25 lines as highly resistant, indicating the presence of all-stage resistance to stripe rust (refer Table 1).Fig. 5
Classification of wheat germplasm based on the AUDPC for stripe rust, recorded from 2018 to 2021 at two locations: Ludhiana (LDH) and Gurdaspur (GSP). Germplasm lines were grouped into four resistance categories: 0 (highly resistant, HR), 1-200 (resistant to moderately resistant, R-MR), 201–400 (moderately susceptible to susceptible, MS-S), and > 400 (highly susceptible, HS). Data are presented for each year and location individually, as well as for combined multi-year assessments.
Disease reaction | FRS | Infection type | Germplasm | Total number of lines |
|---|---|---|---|---|
Highly resistant | 0 | 0 | IC415825, IC553065, IC534662, IC534974, EC180031, IC529399, EC664198, EC693621, EC664239, EC635701, EC693322, EC635577, IC0078981-B, EC0635709, EC0610955, EC0612505, EC0612508, IC0078959-A, EC0610947, IC0599921, EC0633780, IC47034, IC469717, IC542117, and IC445516 | 25 |
Highly susceptible | > 40S | 3+,4 | IC543372, EC577481, IC531950, IC416043, IC335768, EC534382, IC532680, EC660948, EC635777, EC635684, IC0594610, EC0635708, IC0598726, PBW343 (SC), HD2967 (SC), and Agra local (SC) | 13 |
| *SC = Susceptible check | ||||
3.2 GGE biplot analysis
3.2.1 Performance of genotypes across multiple environments
A
To evaluate genotype performance across different environments, the data were analyzed using GGE biplot analysis. The AUDPC values for stripe rust severity revealed significant G, E, and GEI effects among the tested wheat germplasm. The susceptible checks (PBW343 and HD2967) exhibited high and variable disease severity (ranging from 60 to 80S) across locations and years, indicating adequate disease pressure for evaluation. Among the environments, Ludhiana during 2019-20 recorded the highest mean AUDPC (205.65), while Gurdaspur during 2020-21 recorded the lowest (129.74). The variation in stripe rust severity over time and locations, as captured by AUDPC values (refer Fig. 5), reflected inconsistent genotype performance. This inconsistency may be attributed to differences in pathotype aggressiveness, varying environmental conditions, and genetic diversity among genotypes. The analysis also highlighted the role of temperature in influencing disease development, with higher temperatures suppressing rust progression and promoting early teliospore formation. A key feature of the GGE biplot is its ability to reveal relationships among test environments and identify ideal testing locations. Stripe rust responses of wheat genotypes varied across sites, and the first two PCs of the biplot explained 93.21% of the total variation, including 87.81% by PC1 and 5.4% by PC2 (refer Fig. 6a).Figure 6 GGE biplot analysis of stripe rust severity in 441 wheat genotypes evaluated across six environments. (A) Presents the vector view of the GGE biplot, illustrating the interrelationships among test environments. Longer vectors indicate greater discriminative ability of the environment, and smaller angles between vectors denote higher correlations. (B) Shows the “mean vs. stability” view of the GGE biplot, identifying genotypes with high mean performance and stability across environments. Genotypes closer to the average environment axis and with smaller projections are considered more stable.
An AEC view of the GGE biplot was used to evaluate the mean performance and stability of genotypes across test locations (Fig. 6b). According to the analysis, genotypes IC111939 (G1), IC445516 (G19), IC416120 (G22), IC47034 (G111), IC535470 (G145), IC415825 (G402), and EC660684 (G403) showed lower stripe rust severity. Genotypic stability is assessed by the length of a genotype’s projection from the AEC abscissa, with shorter projections indicating greater stability. Genotypes with low rust severity and minimal projection from the AEC line are considered “desirable” due to their proximity to the “ideal” genotype, which reflects both resistance and consistency. In this context, IC111939 (G1) and IC415825 (G22) were identified as the most “ideal” genotypes, showing R to MR responses and stable performance. EC660684 (G145) and IC416120 (G111) also demonstrated desirable resistance and consistent performance across environments. Conversely, genotypes such as IC416143 (G112), EC660681 (G141), and EC0105950 (G312) were classified as unstable, exhibiting susceptible responses in Ludhiana over three years, despite showing R to MR responses at Gurdaspur. Similarly, IC445333 (G49), IC529410 (G76), and IC469420 (G428) were unstable due to their consistent susceptibility at Gurdaspur, although they showed MR to MS responses at Ludhiana.
3.2.2 Evaluation of the environments
In multi-environment trials, testing locations should be evaluated based on their ability to discriminate among genotypes (“discriminatory power”) and their similarity to the target production environment (“representativeness”). The GGE biplot view of “discriminativeness vs. representativeness” revealed that Ludhiana (2019-20) and Gurdaspur (2018-19) had the longest vector lengths for stripe rust, indicating the highest discriminatory power and ability to capture genetic variability among the germplasm. In contrast, both locations in 2020-21 exhibited the shortest vector lengths, suggesting low discriminatory ability. The angle between a test environment's vector and the AEC abscissa in the GGE biplot reflects its representativeness. Ludhiana (2019-20) and Gurdaspur (2018-19) showed small angles with the AEC abscissa, indicating strong representativeness and positive correlation with the overall genotype performance. Conversely, both locations in 2020-21 showed small vectors with large angles, indicating weak representativeness and limited usefulness for genotype evaluation. The “desirability index” of a testing site, which integrates both discriminatory power and representativeness, was highest for Ludhiana (2019-20), making it the most suitable or “ideal” location for screening wheat genotypes for stripe rust resistance (refer Fig. 7a).
Fig. 7
GGE biplot analysis illustrating genotype performance and environmental interactions for stripe rust severity across six environments in 441 wheat genotypes. (A) shows the "which-won-where" view, identifying the best-performing genotypes within specific environments and delineating mega-environment groupings. (B) presents the "discriminativeness vs. representativeness" view, where the length of environmental vectors indicates discriminative ability, and their proximity to the average environment axis reflects representativeness. This analysis aids in identifying both ideal test environments and superior, stable.
Notably, disease severity for stripe rust was most pronounced in Ludhiana during 2020, followed by Gurdaspur in 2019, reflecting higher disease pressure and favorable conditions for stripe rust development in these environments. In contrast, both locations in 2021 exhibited the lowest disease severity, likely due to environmental factors that limited pathogen spread or expression, further explaining their reduced discriminatory ability and representativeness in the GGE biplot analysis (see Fig. 8).
Fig. 8
GGE biplot showing the relationship among six test environments (Ludhiana and Gurdaspur across 2019–2021) based on disease severity data for stripe rust. The length and angle of the vectors represent the discriminative ability and correlation between environments, respectively.
3.2.3 Identification of mega-environments
The “which-won-where” view of the GGE biplot is a two-dimensional polygon that helps identify genotypes with superior performance in specific test environments. To divide the biplot into distinct sectors, perpendicular lines are drawn from the origin to each side of the polygon, with the “winning” genotype for each sector located at the polygon’s vertex. In this study, genotypes IC445384 (G370) and IC416143 (G112), positioned far from the origin and exhibiting the lowest rust severity, showed inconsistent performance, displaying resistance at Gurdaspur but moderately susceptible to susceptible reactions at Ludhiana (Fig. 7b). The local check PBW343 (CHK) was located in sector 1, near genotypes IC531950 (G17), IC529410 (G76), and EC635664 (G167), all of which showed high susceptibility to stripe rust. Genotypes IC0598225 (G279) in sector 4 and EC0612509 (G320) in sector 6, located along the AEC abscissa with minimal projection onto the AEC ordinate, demonstrated resistant to moderately resistant (R to MR) responses with the most stable performance. Across the three years of evaluation, the biplot was divided into seven sectors. Sector 1 can be considered the “mega-environment”, as all six environments clustered within this sector. This indicates that most genotypes showed similar, generally susceptible, responses across locations and years. In contrast, genotypes located in sectors 2 through 6 exhibited lower rust severity and greater variation in performance across environments.
3.3 Association of disease severity with exogenous variables
The year-wise association between disease severity and environmental parameters, including, maximum temperature, minimum temperature, relative humidity, and rainfall, was assessed using Pearson correlation analysis. The results indicate that both maximum (Tmax) and minimum temperatures (Tmin) were significantly correlated with disease severity across all years and in the overall dataset (see Fig. 9). Relative humidity (RH) exhibited a positive correlation with disease severity in 2019-20, but a negative correlation in 2020-21. To further examine the effects of environmental factors, including RH, Tmax, Tmin, and rainfall, along with year (2019, 2020, and 2021) and location (Ludhiana and Gurdaspur), stepwise regression analysis was performed. The selected variables, along with their estimated effects and statistical significance, are presented in Table 2. The analysis revealed that year, location, RH, and Tmax significantly influenced disease severity. Specifically, Tmax, RH, the years 2019 and 2020, and location had positive effects, while the year 2021 showed a negative impact on disease severity. Interestingly, the year 2021 had an opposite effect compared to 2020, which may be attributed to notable environmental shifts in RH and Tmax. To investigate this further, two separate regression models were developed for RH and Tmax, incorporating a dummy variable to assess the year-specific impact. The dummy variable was constructed using the following formula:
Coefficients | Estimate | Std. Error | p-value |
|---|---|---|---|
Intercept | -189.814 | 23.477 | < 0.001 |
Tmax | 6.566 | 0.423 | < 0.001 |
Year: 2020 | 11.620 | 2.938 | < 0.001 |
Year: 2021 | -13.483 | 2.786 | < 0.001 |
RH | 1.160 | 0.238 | < 0.001 |
Location: LDH | 6.886 | 2.307 | 0.004 |
R-square: 76.085% | Adj. R-square: 74.66% | ||
Fig. 9
Year-wise correlation analysis showing the association between stripe rust severity and key environmental parameters, including Tmax, Tmin, RH, and rainfall, during the 2019–2021 cropping seasons. The plot illustrates the strength and direction of associations for each year, aiding in the understanding of environmental influence on disease development.
A
A
The results of the two regression models are presented in Tables S11 and S12. These tables show that the dummy variable was not significant in the RH model but was significant in the Tmax model. This indicates that the change in disease dynamics observed in 2021 was primarily driven by variations in maximum temperature. Notably, during data collection, two heat shocks were recorded in February of the 2020-21 crop season, which likely disrupted stripe rust development.4. Discussion
4.1 Effective screening requires accounting for pathogen complexity and environmental variation
Stripe rust is one of the most devastating diseases affecting wheat crops worldwide, due to its capacity for long-distance migration (Brown and Hovmøller 2002), rapid shifts from avirulence to virulence through high mutation rates (Hovmøller and Justesen 2007), the emergence of new variants via sexual reproduction (Jin et al. 2010), and the existence of highly recombinant and genetically diverse populations (Thach et al. 2016). Furthermore, its ability to adapt to diverse climatic conditions (Milus et al. 2009) complicates control measures even further. Multi-environment screening is essential to identify stable resistance sources, given the variability in environmental conditions and the interactions between genotype and environment, which can obscure genetic advancements. Multi-location testing, although resource-intensive, is crucial for capturing the impact of weather-related factors on stripe rust severity and spread. To optimize screening in resource-constrained settings, identifying “hot spots” with consistently high disease pressure and defining “mega-environments” using multi-year data are effective strategies (Sharma et al. 2016; Parihar et al. 2018). In light of these considerations, the present study was conducted over three years at two distinct locations to evaluate 441 wheat germplasm lines for stripe rust resistance using GGE biplot analysis. The results facilitated the identification of resistant and stable genotypes and also revealed optimal environments for future screening and cultivar development in areas severely affected by stripe rust.
4.2 Differential resistance patterns reflect race-specific and field-level pathogen diversity
The variability in stripe rust resistance observed across seedling and adult plant stages highlights the complex interplay between host genetics and pathogen diversity. Among the tested Pst pathotypes, 238S119 exhibited the highest virulence, causing susceptible reactions in a greater proportion of genotypes, while 46S119 was relatively less virulent, with more genotypes showing resistance. The distinct infection patterns caused by different Pst pathotypes reflect their unique virulence profiles, consistent with previous findings on pathotype aggressiveness (Singh et al. 2020; Prashar et al. 2015; Gangwar et al. 2016; Kaur et al. 2018). These patterns suggest that race-specific resistance is critical at the seedling stage, where certain genotypes demonstrate strong resistance to specific pathotypes, while others succumb to infection. In field conditions, Gurdaspur generally supported a higher number of resistant genotypes compared to Ludhiana, likely due to differences in environmental conditions and pathogen pressure, which aligns with observations of location-specific resistance patterns, the combined effect of multiple races and environmental factors leads to a broader range of responses, from resistant to highly susceptible. This aligns with observations by Kokhmetova et al. (2018), who noted variable resistance across locations, and Sobia et al. (2017), who reported differential responses to Pst isolates. Genotypes with moderate susceptibility but low AUDPC values may possess durable resistance, as they allow initial infection but restrict disease progression through mechanisms like necrotic or chlorotic lesions with minimal spore production (Singh et al. 2008; Brown et al. 2001). These results underscore the need to assess both seedling-stage resistance and adult-plant resistance to fully capture the range of resistance mechanisms, particularly in environments with diverse pathogen populations.
4.3 Stable and broadly resistant genotypes offer promising resources for durable resistance
Durable resistance, characterized by consistent performance across environments and years, is a key target for wheat breeding programs. The identification of 126 germplasms with resistant to moderately resistant responses and low AUDPC values in this study aligns with the findings of Kaur and Bariana (2010), Ali et al. (2008), and Tsilo et al. (2010), who emphasize the value of slow-rusting traits in conferring durable resistance to stripe rust. These wheat germplasms play a vital role in confining disease advancement and repressing the emergence of new virulent Pst pathotypes, making them vital genetic resources for breeding programs aimed at developing wheat cultivars with durable resistance. Twenty-five genotypes exhibited high resistance at both seedling and adult plant stages across all locations and years, indicating the presence of major resistance genes effective against multiple pathotypes (Ali et al. 2007). However, the rapid evolution of Pst races can render race-specific resistance obsolete post-release (Wan and Chen 2012). Genotypes with moderate to high FRS but low AUDPC may still contribute to durable resistance and serve as valuable breeding parents (Singh et al. 2004). The GGE biplot analysis provided a robust framework for assessing genotype stability and adaptation, identifying lines with consistent performance across diverse environments (Yan and Kang 2003). For instance, the genotype IC111939 (G1) was identified as an ideal genotype due to its consistent performance and stability, small projection from the AEC abscissa, and moderate resistance. These results align with earlier GGE biplot findings (Yan 1999; Yan et al. 2007; Das et al. 2019).
4.4 Environmental factors influence disease dynamics and guide location-specific breeding strategies
A
Environmental factors, particularly temperature, play a critical role in stripe rust epidemiology, influencing spore germination, inoculum build-up, and disease spread (Kushwaha et al. 2007). The identification of Ludhiana as a suitable screening location, due to its high desirability index, underscores the importance of selecting test environments with strong discriminatory power and representativeness. As noted by Yan et al. (2007), representativeness, assuming sufficient discriminatory ability, is essential for selecting effective test locations in genotype evaluation. The observed variation in disease severity across years, with higher severity in environments conducive to stripe rust and less severity under unfavourable conditions, highlights the impact of climatic factors like maximum temperature on disease dynamics. Specifically, heat shocks, such as those observed in February 2021, likely inhibited disease progression by triggering early teliospore formation, reducing epidemic potential. These findings align with reports of crossover interactions (COI) in disease responses, where genotype performance varies due to environmental fluctuations, host genetics, and pathogen populations (Baker 1990; Singh et al. 1999; Xu et al. 2014). Genotypic responses varied across time and environments due to weather fluctuations, host genetic variability, and shifts in pathogen populations, patterns that have also been observed in other crops (Sharma et al. 2015, 2016; Parihar et al. 2018). By incorporating multi-environment data through GGE biplot analysis, breeders can optimize their selection strategies by focusing on genotypes that exhibit low FRS, AUDPC, and CI in hotspot environments, thereby strengthening their resistance breeding efforts.5. Conclusions
This study emphasises the significance of GGE biplot analysis in elucidating genotype-by-environment interactions to enhance breeding for stripe rust resistance in wheat. Notably, out of 441 germplasm, 30 showed high resistance to three virulent Pst pathotypes at the seedling stage, and 25 maintained moderate resistance across both growth stages and environments. The germplasm line IC111939 showed consistent performance across various locations, marking it a top candidate for breeding programs. Multi-environment trials conducted at Ludhiana and Gurdaspur highlighted significant environmental influences, with Ludhiana (2019-20) proving to be the most suitable testing site. Stepwise regression and correlation analyses confirmed the role of maximum temperature and relative humidity in disease modulation, particularly the suppressive effect of heat shocks in 2021. The combined use of phenotypic screening, GGE biplot modelling, and environmental analysis provides a robust framework for breeding climate-resilient, stripe rust-resistant wheat cultivars.
A
Acknowledgement
The authors gratefully acknowledge the support of the USAID-funded MSU GRAIN Project. They also thank the National Bureau of Plant Genetic Resources, New Delhi, for providing plant material; the IIWBR Regional Research Station, Flowerdale, Shimla, for supplying the disease inoculum; and the Department of Meteorology, Punjab Agricultural University, Ludhiana, India, for providing meteorological data.
A
Funding
Declarations
The USAID-funded MSU's GRAIN Project supported this work.
Ethics Declarations
Ethics approval and consent to participate
Not Applicable
Consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Electronic Supplementary Material
Below is the link to the electronic supplementary material
A
Author Contribution
J.K. and M.W.H. conceptualized the study. J.K., R.B., A.S., and J.Ku. provided guidance and contributed to the interpretation of results. M.W.H. conducted the investigation and analyzed the data. M.W.H. and J.K. developed the methodology. M.W.H. and D.K.S. drafted the original manuscript. All authors reviewed and critically revised the manuscript.
A
Data Availability
All data generated or analysed during this study are included in this article and its supplementary information files.
References
Ali, S., Shah, S. J. & Ibrahim, M. Assessment of wheat breeding lines for slow yellow rusting (Puccinia striiformis West. tritici). Pakistan J. Biol. Sciences: PJBS. 10 (19), 3440–3444 (2007).
Ali, S., Shah, S. J. A. & Maqbool, K. Field-based assessment of partial resistance to yellow rust in wheat germplasm. J. Agric. Rural Dev. 6 (1), 99–106 (2008).
A
Baker, R. J. & Kosmolak, F. G. Effects of genotype-environment interaction on bread wheat quality in western Canada. Can. J. Plant. Sci. 57, 185–191 (1977).Beddow, J. M. et al. Research investment implications of shifts in the global geography of wheat stripe rust. Nat. Plants. 1 (10), 1–5 (2015).
Brown, W. M. Jr, Hill, J. P. & Velasco, V. R. Barley yellow rust in North America. Annu. Rev. Phytopathol. 39 (1), 367–384 (2001).
Brown, J. K. & Hovmøller, M. S. Aerial dispersal of pathogens on the global and continental scales and its impact on plant disease. Science 297 (5581), 537–541 (2002).
Chen, X. Pathogens which threaten food security: Puccinia striiformis, the wheat stripe rust pathogen. Food Secur. 12 (2), 239–251 (2020).
Das, A. et al. Deciphering genotype-by-environment interaction for targeting test environments and rust resistant genotypes in field pea (Pisum sativum L). Front. Plant. Sci. 10, 825 (2019).
Gangwar, O. P. et al. Virulence pattern and emergence of new pathotypes in Puccinia striiformis f. sp. tritici during 2011-15 in India. Indian Phytopathol. 69 (4s), 178–185 (2016).
Hovmøller, M. S. & Justesen, A. F. Rates of evolution of avirulence phenotypes and DNA markers in a northwest European population of Puccinia striiformis f. sp. tritici. Mol. Ecol. 16 (21), 4637–4647 (2007).
Hovmøller, M. S. et al. Replacement of the European wheat yellow rust population by new races from the centre of diversity in the near-Himalayan region. Plant. Pathol. 65 (3), 402–411 (2016).
Hubbard, A. et al. Field pathogenomics reveals the emergence of a diverse wheat yellow rust population. Genome Biol. 16 (1), 1–15 (2015).
Igrejas, G. & Branlard, G. The importance of wheat. Wheat quality for improving processing and human health: 1–7. (2020).
Jin, Y., Szabo, L. J. & Carson, M. Century-old mystery of Puccinia striiformis life history solved with the identification of Berberis as an alternate host. Phytopathology 100 (5), 432–435 (2010).
Kashyap, S., Pannu, P. P. S., Kaur, G., Sandhu, S. K. & Singh, P. Effect of weather parameters on occurrence and development of stripe rust of wheat in central Punjab. Plant. Dis. Res. 33 (1), 76–81 (2018).
Kaur, J. & Bariana, H. S. Inheritance of adult plant stripe rust resistance in wheat cultivars Kukri and Sunco. J. Plant. Pathol. : 391–394. (2010).
Kaur, J. et al. Current status of wheat diseases in Punjab. Agric Res J, 55 (2). (2018).
Kokhmetova, A. et al. Evaluation of Central Asian wheat germplasm for stripe rust resistance. Plant. Genet. Resour. 16 (2), 178–184 (2018).
Kumar, S. et al. Comprehensive meta-QTL analysis for dissecting the genetic architecture of stripe rust resistance in bread wheat. BMC Genom. 24 (1), 259 (2023).
Kushwaha, C., Srivastava, C. P., Chand, R. & Singh, B. D. Identification and evaluation of a critical time for assessment of slow rusting in pea against Uromyces fabae. Field Crops Res. 103 (1), 1–4 (2007).
A
Lee, T. S. & Shaner, G. Transgressive segregation of length of latent period in crosses between slow leaf-rusting wheat cultivars. Phytopathology 75 (6), 643–647 (1985).McIntosh, R. A., Park, R. F. & Wellings, C. R. Wheat rusts: An atlas of resistance genes (CSIRO, 1995).
Milus, E. A., Kristensen, K. & Hovmøller, M. S. Evidence for increased aggressiveness in a recent widespread strain of Puccinia striiformis f. sp. tritici causing stripe rust of wheat. Phytopathology 99 (1), 89–94 (2009).
Ministry of Agriculture & Farmers’ Welfare, Press Information Bureau. Department of Agriculture and Farmers’ Welfare releases final estimates of major agricultural crops for 2023–24 (Release ID: 2058534). Government of India. (2024). https://www.pib.gov.in/PressReleasePage.aspx?PRID=2058534
Nayar, S. K., Prashar, M. & Bhardwaj, S. C. Mannual of current techniques in wheat rusts. Research Bulletin No.2, 32P. Regional Research Station, Directorate of Wheat Research, Flowerdale, Shimla 171002. (1997).
Pal, N. et al. Meta-QTLs for multiple disease resistance involving three rusts in common wheat (Triticum aestivum L). Theor. Appl. Genet. 135 (7), 2385–2405 (2022).
Parihar, A. K. et al. Targeting test environments and rust-resistant genotypes in lentils (Lens culinaris) by using heritability-adjusted biplot analysis. Crop Pasture Sci. 69 (11), 1113–1125 (2018).
Pathan, A. K. & Park, R. F. Evaluation of seedling and adult plant resistance to leaf rust in European wheat cultivars. Euphytica 149 (3), 327–342 (2006).
A
Prashar, M., Bhardwaj, S. C., Jain, S. K. & Datta, D. Pathotypic evolution in Puccinia striiformis in India during 1995–2004. Aust J. Agric. Res. 58 (6), 602–604 (2007).R Core Team. R: A language and environment for statistical computing. (2013).
Rao, S., Berggren, G. T. & Snow, J. P. Characterization of wheat leaf rust epidemics in Louisiana. Phytopathology 80 (4), 402–410 (1990).
Saharan, M. S., Kumar, S., Selvakumar, R. & Sharma, I. Wheat crop health report of Feb-March 2015201–7 (Wheat Crop Health Newsletter, Directorate of Wheat Research, 2015).
Sandhu, S. K., Dhaliwal, L. K. & Pannu, P. P. S. Effect of weather parameters on incidence and severity of stripe rust in wheat under natural and artificial conditions. J. Agrometeorol. 19, 272–277 (2017).
Shah, S. J. A., Hussain, S., Ahmad, M. & Ibrahim, M. Characterization of slow rusting resistance against Puccinia striiformis f. sp. tritici in candidate and released bread wheat cultivars of Pakistan. J. Plant. Pathol. Microbiol. 5 (2), 1 (2014).
Sharma, I., Tyagi, B. S., Singh, G., Venkatesh, K. & Gupta, O. P. Enhancing wheat production – A global perspective. Indian J. Agric. Sci. 85 (1), 3–13 (2015).
Sharma, R. C., Nazari, K., Amanov, A., Ziyaev, Z. & Jalilov, A. U. Reduction of winter wheat yield losses caused by stripe rust through fungicide management. J. Phytopathol. 164 (9), 671–677 (2016).
A
Sharp, E. L. & Fuchs, E. Additive genes in wheat for resistance to stripe (yellow) rust (Puccinia striiformis Westend). Crop Prot. 1 (2), 181–189 (1982).Singh, H., Kaur, J., Bala, R., Srivastava, P. & Bains, N. S. Virulence and genetic diversity of Puccinia striiformis f. sp. tritici isolates in sub-mountainous area of Punjab, India. Phytoparasitica 48 (3), 383–395 (2020).
Singh, M., Ceccarelli, S. & Grando, S. Genotype x environment interaction of crossover type: detecting its presence and estimating the crossover point. Theor. Appl. Genet. 99 (6), 988–995 (1999).
Singh, R. P. et al. Will stem rust destroy the world's wheat crop? Adv. Agron. 98, 271–309 (2008).
Singh, R. P., William, H. M., Huerta-Espino, J. & Rosewarne, G. Wheat rust in Asia: meeting the challenges with old and new technologies. Proc 4th Int Crop Sci Congr, 26. Brisbane, Australia: Published in CD-ROM. (2004).
Singh, V. K. et al. Characterization of yellow rust resistance genes by using gene postulation and assessment of adult plant resistance in some Indian wheat genotypes. Res. Crops, 16 (4). (2015).
Sobia, T., Muhammad, A. & Chen, X. Evaluation of Pakistan wheat germplasms for stripe rust resistance using molecular markers. Sci. China Life Sci. 53 (9), 1123–1134 (2010).
Stubbs, R. W., Prescott, J. M., Saari, E. E. & Dubin, H. J. Cereal disease methodology manualVol. 46 (CIMMYT, 1986).
Stuthman, D. D., Leonard, K. J. & Miller-Garvin, J. Breeding crops for durable resistance to disease. Adv. Agron. 95, 319–367 (2007).
Thach, T., Ali, S., de Vallavieille-Pope, C., Justesen, A. F. & Hovmøller, M. S. Worldwide population structure of the wheat rust fungus Puccinia striiformis in the past. Fungal Genet. Biol. 87, 1–8 (2016).
Tsilo, T. J., Jin, Y. & Anderson, J. A. Identification of flanking markers for the stem rust resistance gene Sr6 in wheat. Crop Sci. 50 (5), 1967–1970 (2010).
Vaibhav, K., Singh, G. P., Singh, P. K., Harikrishna, R. & Gogoi, R. Assessment of slow rusting resistance components to stripe rust pathogen in some exotic wheat germplasm. Indian Phytopathol. 70 (1), 52–57 (2017).
Wan, A. & Chen, X. Virulence, frequency, and distribution of races of Puccinia striiformis f. sp. tritici and P. striiformis f. sp. hordei identified in the United States in 2008 and 2009. Plant. Dis. 96 (1), 67–74 (2012).
Wan, A., Chen, X. & Yuen, J. Races of Puccinia striiformis f. sp. tritici in the United States in 2011 and 2012 and comparison with races in 2010. Plant. Dis. 100 (5), 966–975 (2016).
Wellings, C. R. Global status of stripe rust: a review of historical and current threats. Euphytica 179 (1), 129–141 (2011).
Xu, N. Y., Fok, M., Zhang, G. W., Jian, L. I. & Zhou, Z. G. The application of GGE biplot analysis for evaluating test locations and mega-environment investigation of cotton regional trials. J. Integr. Agric. 13 (9), 1921–1933 (2014).
Yan, W. & Falk, D. E. Biplot analysis of host-by-pathogen data. Plant. Dis. 86 (12), 1396–1401 (2002).
Yan, W. & Holland, J. B. A heritability-adjusted GGE biplot for test environment evaluation. Euphytica 171 (3), 355–369 (2010).
A
Yan, W. & Kang, M. S. GGE biplot analysis: A graphical tool for breeders, geneticists, and agronomists (CRC, 2002).Yan, W. & Rajcan, I. Biplot analysis of test sites and trait relations of soybean in Ontario. Crop Sci. 42 (1), 11–20 (2002).
Yan, W. A study on the methodology of cultivar evaluation based on yield trial data with special reference to winter wheat in Ontario. Doctoral dissertation, University of Guelph. (1999).
Yan, W. GGEbiplot—A Windows application for graphical analysis of multienvironment trial data and other types of two-way data. Agron. J. 93 (5), 1111–1118 (2001).
Yan, W., Hunt, L. A., Sheng, Q. & Szlavnics, Z. Cultivar evaluation and mega-environment investigation based on the GGE biplot. Crop Sci. 40 (3), 597–605 (2000).
Yan, W., Kang, M. S., Ma, B., Woods, S. & Cornelius, P. L. GGE biplot vs. AMMI analysis of genotype-by-environment data. Crop Sci. 47 (2), 643–653 (2007).
Yan, W., Pageau, D., Frégeau-Reid, J. & Durand, J. Assessing the representativeness and repeatability of test locations for genotype evaluation. Crop Sci. 51 (4), 1603–1610 (2011).