Introduction
Bread wheat (Triticum aestivum) is one of the most important food crops in the world. Cultivation of wheat has a history of thousands of years and different aspects of its nutrition have made it a main crop for cultivation around the world. The root system and related traits have an important effect on adaptation to drought. Repeated studies indicate that root traits such as depth, volume and weight have a high diversity, so they can be used for selection (Manschadi et al., Reference Manschadi, Christopher, Devoil and Hammer2006; Cattivelli et al., Reference Cattivelli, Rizza, Badeck, Mazzucotelli, Mastrangelo, Francia, Mare, Tondelli and Santaca2008; Akbarabadi et al., Reference Akbarabadi, Kahrizi, Rezaizad, Ahmadi, Ghobadi and Molsaghi2015; Liakat-Ali et al., Reference Liakat-Ali, Luetchens, Singh, Shaver, Kruger and Lorenz2016; Ye et al., Reference Ye, Roorkiwal, Valliyodan, Zhou, Chen, Varshney and Nguyen2018; Verma et al., Reference Verma, Borah and Sarma2019; Chen et al., Reference Chen, Palta, Vara Prasad and Siddique2020).
Despite the necessity of the role of the root system in the plant's growth and adaptation, it has always been neglected (Herder et al., Reference Herder, Isterdael, Beakman and Smet2010). The limited research efforts about roots may be because of the difficulty in observing, measuring and manipulating (Shen et al., Reference Shen, Courtois, McNally, Robin and Li2001). Root traits are believed to be complex and controlled by many genes, each with a small genetic effect (Sharma et al., Reference Sharma, Xo, Ehdaie, Hoops, Close, Lukaszewski and Waines2011; Comas et al., Reference Comas, Becker, Cruz, Byrne and Dierig2013).
Most of the physiological traits are also quantitative characters, which are controlled by multiple genes and easily affected by environmental conditions. One of the important applications of molecular markers is to improve and enhance the efficiency of conventional plant breeding methods by the indirect selection of the molecular markers associated with traits. Since the molecular markers are not affected by the environment, they can be used at all stages of plant growth. Although quantitative trait locus analysis of physiological traits has made a good progress, these results are difficult to be used for the genetic improvement of wheat (Sharma et al., Reference Sharma, Xo, Ehdaie, Hoops, Close, Lukaszewski and Waines2011; Rostami-Ahmadvandi et al., Reference Rostami-Ahmadvandi, Cheghamirza, Kahrizi and Bahraminejad2013).
Association mapping is a method for understanding the effects of genes based on linkage mapping. This method recently has been used to identify the genotype–phenotype relationships in plants (Ni et al., Reference Ni, Li, Zhao, Peng, Hu, Xin and Sun2018). Unlike the conventional linkage analysis method used by a population designed to predict genotype–phenotype relationships, this method does not require the creation of a pure genetic population and uses the natural population. It can be used in marker-assisted selection (MAS) (Zhu et al., Reference Zhu, Gore, Buckler and Yu2008; Xu et al., Reference Xu, Li, Li, Ma, Shi, Xu, Ma and An2017; Tshikunde et al., Reference Tshikunde, Mashilo, Shimeli and Odindo2019). This approach has been successfully employed to identify the molecular markers associated with various phenotypic traits in different plant species, including grain traits in bread wheat (Mirdarikvand et al., Reference Mirdarikvand, Najafian, Bihamta and Ebrahimi2018), agronomic traits in hexaploid wheat (Brbaklic et al., Reference Brbaklic, Trkulja, Kondic-spika, Treskic and Kobiljski2013; Khaled et al., Reference Khaled, Motawea and Said2015) and root traits and drought resistance in bread wheat (Isras et al., Reference Isras, Ali, Ahmad and Inamullah2017).
Considering the important role of root and its related traits in maintaining and enhancing plant yield and ability of the plant to adapt to the climate changes as well as the effect of plant water use efficiency (WUE) especially on wheat, this study aimed to identify simple sequence repeat (SSR) and inter-simple sequence repeat (ISSR) markers associated with these traits in different wheat genotypes to map the association and genomic regions of these traits. The results can be useful for MAS.
Materials and methods
Plant material and phenotyping
A set of 102 bread wheat genotypes provided from the seed bank of Ilam University was evaluated (online Supplementary Table S1). Experiments were conducted in 2016/2017 in a greenhouse at Ilam University. The experiments were arranged in three steps. First, the root traits were measured at the embryonic stage (10 days after germination). At this stage, fresh leaves were sampled in immediately submerged in −105°C nitrogen and these were used for further studies. In the second step, root traits were measured at the seedling stage. Finally, in the third step, agromorphologic traits were measured at the whole plant stage.
At the two first stages, after washing and separating the roots from soil, some root and shoot traits, including seminal root length (SemRL) (cm), seminal root number (SemRN), seminal shoot length (SemShL) (cm), root fresh weight (RFW) (g), root dry weight (RDW) (g), root volume (RV) (cm3), root length (RL) (cm), shoot length (ShL) (cm), shoot weight (ShW) (g), leaf number (LN), tiller number (TN) and some root indexes, including root fineness (RF = RL/RFW (cm/g)), root diameter (RDM = [(4 × RFW)/(RL × 3.14)]) (Mandal et al., Reference Mandal, Hati, Misra, Ghosh and Bandyopadhyay2003), root water content (RWC = (RFW − RDW)/RFW), specific root length (SRL = RL/RDW (cm/g)) (Paula and Pausas, Reference Padilla and Pugnaire2011), root specific mass (RSM = RDW/SV (g/cm3)) (Mandal et al., Reference Mandal, Hati, Misra, Ghosh and Bandyopadhyay2003), root length density (RLD = RL × SV (cm/cm3), root surface area (RSA = RL × RDM × 3.14), root texture density (RTD = RDW/RV (g/cm3) (Paula and Pausas, Reference Padilla and Pugnaire2011) and root mass density (RMD = RFW/SV (g/cm3)) (Mandal et al., Reference Mandal, Hati, Misra, Ghosh and Bandyopadhyay2003) were estimated.
The determined traits at the whole plant stage were: plant height (PH) (cm), awn length (AL) (cm), peduncle length (PL) (cm), spike number (SN), fertile spike (FS), infertile spike (IS), number of spikelts per spike (NSPS), straw weight (SW) (g), spike length (SL) (cm), grain number for all spikes (GN/allS), grain number per spike (GN/perS), all grains weight (allGW) (g), thousand grains weight (TGW) (g), water use efficiency (WUE) (calculated as grain yield divided by water used), days to shooting (DtSh), days to heading (DtH), days to anthesis (DtA) and days to ripening (DtR).
Genotyping
For molecular analysis, leaf samples were harvested and freeze-dried. Genomic DNA was extracted using the CTAB method (Doyle and Doyle, Reference Doyle and Doyle1987). A set of 22 SSR and 10 ISSR markers were used (Tables 1 and 2). Polymerase chain reaction (PCR) was performed in 20 μl volume by a program as shown in Table 3 using a Bio-Rad C-1000 thermocycler machine (online Supplementary Table S2). The results of PCR products were scored as 0 and 1 and used in further analysis.
Table 1. Nucleotide sequence, annealing temperature and amplification results generated in the 102 wheat genotypes using ISSR markers
H (A, T, C); B (G, T, C).
Table 2. Nucleotide sequence, annealing temperature and amplification results generated in the 102 wheat genotypes using SSR markers
Table 3. Percentage of genotypes, differentiation index and genetic distance of the extracted subpopulations
Genetic analysis
The number of alleles produced by each one of the 31 markers was counted, and the polymorphism information content (PIC) value of each marker was computed across the whole set of genotypes. Significant P values of LD were estimated using Tassel v. 2.1 software. The squared correlation of allele frequencies (r 2) among loci was assessed. The population structure of 102 genotypes was performed using genetic data of 31 markers in STRUCTURE v. 2.3.4. The program was run for 1–10 subgroups (K value) with three independent runs for each K value. The iteration number for the Markov chain Monte Carlo (MCMC) algorithm was set as 100,000 following a burn-in period of 10,000 iterations. Possible sub-populations within the germplasm were evaluated using STRUCTURE software. The K-matrices and Q-matrix describing the assignment of each genotype to specific clusters were used in a mixed linear model. The genome-wide association analysis was performed to identify the markers associated with the evaluated traits, based on the mixed linear model using TASSEL 2.1 software, and markers related to each of the traits were identified and arranged.
Results and discussion
Phenotypic evaluation
Most of the measured traits exhibited a highly significant coefficient of variation (CV) among different genotypes (online Supplementary Table S3). The highest CV was for the SRL (79.83%) and the lowest was for the ShL (13.87%). Among the traits, the RFW, shoot fresh weight, RDW, RV, TN, RF, RSM, RL, RMD and RTD had a high CV (above 30%). Traits with a CV between 15 and 30%, containing the SemRL, SemRN, SemShL, LN, SRL, RLD, plant height, awn length, peduncle length, spike number, fertile spikes, infertile spikes, number of spikelts per spike, straw weight, spike length, grain number for all spikes, grain number per spike, all Grains weight, thousand grains weight, WUE, days to shooting, days to heading, days to anthesis and days to ripening considered as the traits with a moderate CV. Only two traits of the ShL and RWC had a low CV (<15%).
Such differences make exterior strategies for survival in dry conditions. However, the relationship between root growth and survival under drought conditions is still unclear. For example, Padilla and Pugnaire (Reference Padilla and Pugnaire2007) found a negative relationship between root growth and survival under drought conditions. There was a positive correlation between the RDW to shoot ratio and plant survival under drought conditions, while a positive relationship was found between the root depth and plant survival in other studies (Fensham and Fairfax, Reference Fensham and Fairfax2007).
Marker polymorphism
A total of 84 alleles were scored across 31 markers. In total, 43 and 39 alleles were amplified among 21 SSR and 10 ISSR markers, respectively. For ISSR markers, the number of alleles per marker ranged from 2 to 8, with an average of 3.9. The PIC value was from 0.06 to 0.32, with an average of 0.192 (Table 1).
Among SSR markers the number of alleles per locus was from 1 to 5, with an average of 2.04 alleles per SSR. The value of PIC varied from 0 (Xgwm132, Xgwm2, Xgwm608 and Xgwm469) to 0.35 (Xgwm369) with an average of 0.143 (Table 2).
Population structure
In genetic studies, population structure explains the relationships of individuals within and between populations and provides insights into the evolutionary relationships of individuals within a population. Determining the population structure using molecular information is a major prerequisite for communication analysis (Brbaklic et al., Reference Akbarabadi, Kahrizi, Rezaizad, Ahmadi, Ghobadi and Molsaghi2013).
Using the MCMC algorithm, the 102 genotypes were divided into two subpopulations (Fig. 1).
Fig. 1. Two-dimensional diagram based on K and ΔK statistics. The peak value of ΔK was at K = 2, suggesting two subpopulations in studied genotypes.
According to the results, K and ΔK statistics were extracted, and the two-dimensional diagram was drawn according to these criteria. The maximum value in the diagram was at K = 2. Therefore, based on the results, two possible subpopulations were identified in the germplasm (Fig. 2).
Fig. 2. Bar graph derived from the genetic analysis of 102 wheat genotypes using STRUCTURE software showing the membership based on SSR and ISSR markers (K = 2).
The second sub-population with 48 genotypes had the highest genetic differentiation index (FST = 0.3). For the first population with 54 genotypes, the amount of the genetic differentiation index was 0.25 (Table 3). The placement of the most genotypes in a subpopulation can be the result of similar selection pressure during breeding programmes.
Significant correlations and association mapping
Significant levels of correlations of the relevant markers and traits are shown in Table 4.
Table 4. Squared correlation of allele frequencies (r 2), significance level and position of the markers related to the evaluated traits in the 102 wheat genotypes using the genome-wide association analysis method
Morphologic traits
In the case of traits related to the yield, correlation analysis showed that marker xgwm369 was correlated with the all grains weight, grain number for all spikes, grain number per spike and thousand grains weight at two genetic loci on chromosome 3A and accounted for 7, 9, 6 and 12% of the phenotypic variation in these traits, respectively. This marker also was correlated with the number of fertile and infertile spikes with 0.035 and 0.005 probabilities, controlling 6 and 4% of the phenotypic variation in these traits, respectively. The reason that this marker is common in all of these traits can be attributed to the high correlation between these traits and their impact on the final yield.
For other morphological traits, the straw dry weight was associated with marker xgwm33 on chromosome 1A, at the probability level of 0.03, and justified 5% of the phenotypic variation. The spike dry weight was correlated with marker xgwm369 on chromosome 3A and it justified 8% of the phenotypic variation at a probability level <0.001.
Recent studies of the peduncle length have attracted much attention because of its importance in the prevention of the cluster diseases (Brbaklic et al., Reference Brbaklic, Trkulja, Kondic-spika, Hristov, Dencic, Micic, Tomicic and Kobiljski2015). In previous studies, this trait was significantly correlated with SSR markers on chromosomes 2A and 6A (Neumann et al., Reference Neumann, Kobiljski, Dencic, Varsheny and Borner2011), and in this study, it was associated with marker xgwm369 on chromosome 3A, and 4% of the phenotypic variation of this trait was justified at the probability of 0.005. Marker xgwm369 also showed a significant relationship with the LN and shoot fresh weight (P < 0.003) and justified 4 and 10% of the phenotypic variation in these traits, respectively. However, the ShL was correlated with marker ISSR14 at the probability level of 0.049, which explained 8% of the phenotypic variation.
In phenological traits, marker xgwm369 had a significant relationship with the days to anthesis and the days to heading on chromosome 3A. This marker accounted for 6 and 4% of the phenotypic variation of these traits, respectively. Marker ISSR14 with a probability level of 0.03 was correlated with the days to shooting and explained 14% of the phenotypic variation of this trait.
Root traits
Among the traits and indexes related to the root, at the embryonic stage, the RL was correlated with marker Xgwm33 on chromosome 1A and the root number was related to the ISSR14 and Xgwm369 markers on chromosome 3A. These markers accounted for 8, 4 and 8% of the phenotypic variation, respectively. The SemRL was significantly correlated with marker ISSR01 at the probability level of 0.01, which accounted for 6% of the phenotypic variation. Locations of B2 and B3 chromosomes in bread wheat have been reported for controlling the diversity in seminal roots under the water stress conditions. Genetic control of the RL traits has been reported in a wide range of wheat cultivars with different growth types (Narayanan et al., Reference Narayanan, Mohan, Gill and Prasad2014).
Regarding the traits evaluated at the seedling stage, the RV assessments showed a high coefficient of the phenotypic variation (45.07%) and finally was correlated with marker Xgwm33 on chromosome 1A and marker Xgwm369 on chromosome 3A, at the probability level of 0.02 and these markers accounted for 5 and 10% of the phenotypic variation in this trait, respectively. Iannucci et al. (Reference Iannucci, Marone, Anna Russo, De Vita, Miullo, Ferragonio, Blanco, Gadaleta and Mastrangelo2017) reported markers on chromosome 2A, associated with this trait.
The RDW also showed a very high CV (64.53%) in the phenotypic evaluation. Finally, marker ISSR10 at the probability level of 0.045 was associated with this trait and justified 5% of the phenotypic variation. For RFW, the CV was 36.88%, and marker ISSR12 at 0.048 probability level, and marker Xgwm369 at the probability level of 0.042, explained 10 and 6% of the phenotypic variation, respectively. The root surface density index had 20.20% of the CV in phenotypic assessments and was correlated with marker Xgwm369 at 0.006 probability level, which explained 7% of the phenotypic variation on chromosome 3A. The CV for this trait indicates that density and surface of the root need to be evaluated more accurately. Marker Xgwm33 on chromosome 1A was correlated with the RDM, RMD and root surface density indices at 0.042, 0.035 and 0.028 probability levels and explained 6, 10 and 4% of the phenotypic variation, respectively. The RMD was also correlated with marker Xgwm369 at the probability level of 0.035. This marker also showed a significant correlation with the SRL index which had a very high CV (79.83%) at 0.001 level and justified 11% of the phenotypic variation. The RSM index also showed a high phenotypic variation (57.57%). Marker Xgwm388 on chromosome 2B justified 5% of changes at the probability level of 0.032. The RWC, with 6% of the phenotypic variation was justified by UBC886 at 0.001 probability level.
The average WUE was found to be 0.8 g/l of the consumed water. Variation for this trait was very high, and the CV was 77.76%. Marker Xgwm369 was associated with this trait at two genetic locations. This marker on chromosome 3A at 0.001 and 0.018 probability levels justified 21% of the phenotypic variation of this trait.
In this study, 102 wheat genotypes were evaluated for some traits related to the root system structure at the embryonic and seedling stages and also their relationship with some agromorphological traits. The results showed a high level of genetic diversity in the studied genotypes. Descriptive statistics also revealed that most of these traits had a high level of the phenotypic variation coefficient. Genomic selection using markers that have a major role in controlling the phenotypic variation of traits can help early screening of the segregating populations in wheat breeding programmes. Genetic structure analysis revealed two distinct subpopulations. Significant correlations were found between genomic markers and evaluated traits in genome-wide association analysis. A total of nine markers were related to the studied traits, including four SSR markers in 1A, 3A, 2B and 5A chromosomal regions, and five different ISSR markers.
A total of 22 traits showed significant relationships with these markers. Among these, 19 traits were related to the up ground traits (such as the peduncle length, spike dry weight, awn length, etc.), and others (13 traits) were related to the root traits (such as the number and length of the seminal roots, RWC, RFW, RV, etc.).
The phenotypic variation justified by these alleles was 4–15%. The number of days to shooting and after that, the number of spikes, SRL, WUE and RV, had the highest allele frequencies (r 2). Among the related markers, marker Xwgm369 on chromosome 3A had the highest correlation with the desired traits. This marker was associated with 14 up ground traits and seven root traits. This may indicate a genetic association and a correlation between roots, and shoot growth.
The relationship between plant height and root system development is a controversial subject that has not been clearly defined, yet. There are various indications of previous studies that root traits were evaluated in different growth conditions and stages. Most recent studies have reported different genetic complexes to control stem and root growth. In some cases, there are pieces of evidence of a positive correlation between these traits (Genc et al., Reference Genc, Huang and Langridge2007). Obtained genetic information can be targeted for further validation and genetic analysis in the relevant populations or other breeding sets. In addition, the newly identified alleles indicate that association mapping has a higher resolution than other mapping methods and can be valuable for using in MAS studies.
Based on the results, some markers were involved in controlling several related traits. In such cases, where only a small number of markers control most of the variation in more than one trait, ideal genotypes can be reached by genomic selection in breeding populations at the least time.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1479262120000465
