Introduction
Drought stress has been one of the most effective factors reducing agricultural productions in recent years. Oil seeds are considered as one of the main resources of providing the required energy for vital procedures in the body. Sunflower is one of the most important oil seeds in the world. Although sunflower is drought stress tolerant to some extent, its production and yield is greatly affected by drought stress (Chimenti et al., Reference Chimenti, Pearson and Hall2002; Gupta, Reference Gupta2015). Drought-stress results in the closing of apertures and thus losing the turgor pressure. Thus, the plant biomass growth and the cellular development would be diminished (Qadir and Ahmad, Reference Qadir and Ahmad2005; Jaleel et al., Reference Jaleel, Manivannan, Wahid, Farooq, Al-Juburi, Somasundaram and Panneerselvam2009; Ghaffari et al., Reference Ghaffari, Toorchi, Valizadeh and Shakiba2012). Among different phenological stages, drought stress from the blooming stage until the end of flowering stage would impose the highest negative effect on the production of sunflower hybrids (Stone et al., Reference Stone, Goodrum, Jaafar and Khan2001; Karam et al., Reference Karam, Lahoud, Masaad, Kabalan, Breidi, Chalita and Rouphael2007). In their reports, Farahvash et al. (Reference Farahvash, Mirshekari and Seyahjani2011) stated that moisture-stress results in the early ageing of the leaves, reduced leaf number and leaf area, reduced leaf diameter, reduced 1000-grain weight and ultimately reduced sunflower grain yield. Drought-stress tolerance is a complicated trait including several various traits. Thus, identifying, selecting and improving high-yielding and drought stress-tolerant genotypes require evaluation of the traits related to drought tolerance. Inter-simple sequence repeat (ISSR) markers have a benefit over the other markers by being reproducible and more reliable (Reddy et al., Reference Reddy, Sarla and Siddiq2002). An ISSR marker, in addition to its appropriable to genetic variation study, is highly polymorphic, cost effective and requires no prior information of the sequence (Bornet et al., Reference Bornet, Muller and Branchard2002). Traits are divided into two categories including quantitative and qualitative. Understanding the genetic structure of quantitative traits is a long-term challenge for quantitative geneticists and plant breeders, who wish to design efficient breeding programmes. With recent advances in molecular genotyping and high-throughput technology, unrevealing the genetic architecture of complex traits has become possible via quantitative trait loci (QTL) analysis (De Vienne, Reference De Vienne2003; Collard et al., Reference Collard, Jahufer, Brouwer and Pang2005). Molecular markers linked to QTLs/major genes of traits are being routinely recognized in many crops by utilizing genetic linkage map developed (Collard et al., Reference Collard, Jahufer, Brouwer and Pang2005). Many genetic variations have been revealed by ISSR (Joshi et al., Reference Joshi, Gupta, Aggarwal, Ranjekar and Brar2000). Polygenes usually control most of the traits that are in association with any type of stress. Therefore, quantitative genetic approaches can be used to explore genes controlling the selection traits (Neale, Reference Neale2007). Limitations of pedigree-based QTL can be offset by applying linkage disequilibrium (LD) mapping (Awais Khan and Korban, Reference Awais Khan and Korban2012). In association mapping (AM), genotypic and phenotypic correlations would be investigated in unrelated individuals; both LD and historical recombination within the gene pool would be employed. This approach has been applied for species with available genomic resources (Awais Khan and Korban, Reference Awais Khan and Korban2012).
Therefore, the objective of the current study was identifying some morphological and phenological traits related to drought tolerance of sunflower genotypes and the relationship between these traits and seed yield.
Materials and methods
The current study was conducted in the Dryland Agricultural Research Institute (DARI), in Sararood (Kermanshah) with the northern and eastern latitude and longitude of 47° and 34°, respectively, the altitude of 1351 m above sea level, the average annual rainfall of 416.8 mm and average temperature of 13.8°C. For this purpose, a lattice square-design experiment was conducted with 64 sunflower genotypes (Table 1) with an 8 × 8 pattern with two replications under non-stress and moisture-stress conditions (irrigation holding at the flowering stage). Before conducting the farm preparation and planting operations, the farm soil sampling was conducted at a depth of 50 cm; pH, the organic matter percentage, EC, macro elements (N, P and K) and soil texture type were determined (online Supplementary Table S1). The soil texture of the experimental field was silty clay loam. According to the fertilizer recommendation of the soil laboratory, the amount of potassium elements of the soil was balanced and there was no need for potassium fertilizers. However, 100 kg/ha of urea fertilizer and 40 kg/ha of phosphate fertilizer were used. After conducting the farm preparation operations (including plowing, disc harrowing, levelling and creating furrows), planting was conducted on the 9th of March 2017 in the middle of the furrow. The within row distance was 25 cm and the between rows distance was 50 cm. In each plot, five rows were considered; the side rows were not used due to marginal effects. The length of the plots was 4 m. In the distances between the plots, one row was regarded as uncultivated. In treating the moisture stress, irrigation was conducted until the flowering stage, and it was then holding. No rainfall occurred from this time until the end of the growing season (online Supplementary Table S2). However, under non-stress conditions, the irrigation was conducted twice including flowering and seed-filling stages. Each time, the irrigated water was controlled by a volume counter. During the plant growing season, the phenological and morphological traits were recorded. At the end of the growth season, as many as five plants were harvested from each plot. After moving these plants to the laboratory, it was attempted to measure head diameter, number of seeds per head and 1000-seed weight. In the lower half of each plot, dedicated to yield evaluation, the plants were harvested after removing the margins and counting the number of heads per unit area. After separating the seeds, the seed yield was measured separately. The oil content of seeds was measured by using NMR. The measurement of leaf relative water content (RWC) was conducted by using 15 leaf discs selected by a punch from the plant leaves at 10 a.m. After weighing the disks of plant leaves, the leaf discs were saturated in distilled water for 12 h, and they were weighed again. Then the weighed samples were placed in an oven with the temperature of 85°C for 24 h to determine the dry weight. The RWC was measured by using the following equation:
$${\rm RWC} = \left[{\displaystyle{{{\rm FW}-{\rm DW}} \over {{\rm TW}-{\rm DW}}}} \right]\times 100$$where FW is the fresh weight, DW is the dry weight and TW is the turgid weight.
Table 1. List of sunflower genotypes under study
Utilizing the protocol of Doyle and Doyle (Reference Doyle and Doyle1987) for DNA extraction, about 100 mg fresh leaf tissue of each sunflower genotype was used. DNA quality and quantity were tested by 0.8% agarose gel spectrophotometry and electrophoresis, respectively.
PCR reactions
To conduct polymerase chain reactions (PCRs), in 20 μl volume, 18 high reproducible polymorph ISSR (UBC) primers (online Supplementary Table S3) have been used. The volume included: 1.5 μl DNA, 20 μl PCR buffer 10×, 1.8 μl MgCl2 (20 mM), 0.4 μm dNTP (1 mM), 1.2 μl primer (10 pM), 0.3 μl Taq DNA polymerase (5 units). PCR cycles were carried out using a Bio-Rad c1000 thermal cycler. The PCR programme was: 95°C as the primary denaturation: 35 normal PCR cycles: denaturation (95°C for 60 s): annealing (primer specific T m (°C) for 60 s): extension (72°C for 2 min). Final extension time was 2°C for 10 min.
Population structure, kinship and association analysis
The genetic structure for the evaluated genotypes was analysed using frequency pattern of amplified alleles by STRUCTURE software (Pritchard et al., Reference Pritchard, Stephens and Donnelly2000). The length of the burn-in period was 100,000 followed by 10,000 Monte Carlo Markov chain replicates. The admixture model and matrix of correlated allele frequencies were considered in association analysis. The number of hypothetical subgroups (K) was set from 1 to 10 and three independent runs were made for each K. Optimal number of K determined by the log likelihood of the data [ln P(D)] in the STRUCTURE output and ΔK based on the rate of the change in ln P(D) between successive K values (Pritchard et al., Reference Pritchard, Stephens and Donnelly2000). The pairwise kinship coefficients among the genotypes estimated by the TASSEL program (Bradbury et al., Reference Bradbury, Zhang, Kroon, Casstevens and Ramdoss2007).
Results
Yield and yield components
The analysis of variance indicated that the effect of sunflower genotypes on seed yield was significant under both non-stress and moisture-stress conditions (Table 2). So, under non-stress conditions, Sil-96 genotype had the highest seed yield with more than 56 g/m2. This has been attributed to the higher number of seeds per head in the Sil-96 genotype. This genotype accounted for the highest number of seeds per head by producing 848 seeds more per head (Table 3). On the other hand, as for the other four genotypes i.e. Sil-48, Sil-82, Sil-74 and Sil-94, the seed yield was less than 22 g g/m2. This low yield is possibly resulted from the lower number of seeds per head in these genotypes (online Supplementary Table S4). Under moisture-stress conditions, Sil-96 showed the best seed yield performance (43.08 g/m2) through having a high number of seeds per head. There were four other genotypes whose seed yield was more than 36 g/m2 (online Supplementary Table S4). However, these four genotypes were different from the four genotypes having a high yield under non-stress conditions. Furthermore, under moisture-stress conditions, the seed yield of 18 genotypes did not even reach 20 g g/m2 (online Supplementary Table S4). Since the number of seeds per head (as one of the main components of yield) showed the strongest correlation with seed yield (online Supplementary Table S5), it seems that the changes of this component of yield can have a significant effect on sunflower seed yield. The effect of genotype on the number of heads per metre under both drought-stress and non-stress conditions was not significant (Table 2). In other words, under both drought-stress and non-stress conditions, no significant difference was observed between sunflower genotypes in terms of the number of heads per metre. The effect of genotype on the number of seeds per head was significant under both drought-stress and non-stress conditions (Table 2). Under non-stress conditions, Sil-96 genotype showed the highest number of seeds per head with the production of 842 seeds. However, the other 12 genotypes were classified in the same statistical group as this genotype; the number of seeds per head was more than 590 seeds per head (online Supplementary Table S4). In Sil-48 and Sil-82 genotypes, the number of seeds per head did not even reach 300 seeds. As for nine genotypes, the number of seeds per head was less than 400 (online Supplementary Table S4). Under drought-stress conditions, the number of seeds per head was more than 630 in 11 genotypes. However, the number of seeds per head was less than 400 in 16 genotypes (online Supplementary Table S4). The effect of genotype on the 1000 seed weight was not significant under drought-stress and non-stress conditions (Table 2). There was no significant difference between the studied genotypes in terms of 1000 seed weight. The effect of genotype on the oil yield of sunflower seed was significant under both stress and non-stress conditions (Table 2) that indicated the diversity of genotypes in terms of oil yield. Under non-stress conditions, Sil-95 and Sil-96 and Armaviresky had the highest amount of oil yield. Sil-82 had the lowest level of oil yield. Moreover, under drought-stress conditions, Sil-96 had the highest oil yield (online Supplementary Table S7). Sil-82 had the lowest amount of seed oil content. In comparison with the reduced seed yield, oil content was less affected by the drought stress.
Table 2. Analysis of variance for seed yield, two yield components, oil content and RWC of sunflower genotypes under stress and non-stress conditions
**Significant at 1% level; *significant at 5% level; ns, non-significant.
Table 3. Analysis of variance phenological and morphological traits of sunflower genotypes under two stress and non-stress conditions
**Significant at 1% level; *significant at 5% level; ns, non-significant.
Phenological trait
The analysis of variance indicated that under both drought-stress and non-stress conditions, no significant difference was observed between the genotypes in terms of the vegetative growth period (days to flowering) (Table 3). As the irrigation was conducted in the same way under stress and non-stress conditions until the flowering stage, the days to the flowering trait (the vegetative growth period) was not affected by drought stress. However, according to the analysis of variance, the effect of the genotypes on days to maturity trait was significant under drought-stress conditions (Table 3).
Morpho-physiological
The effect of genotype on the head diameter was not significant under drought-stress and non-stress conditions (Table 3). Among the genotypes, the head diameter was more than 11.4 cm in 27 genotypes. However, in 16 genotypes, the head diameter did not even reach 10 cm. Although, the head diameters of Sil-2, Sil-13, Sil-20 and Zarya genotypes were more than 14 cm, the difference between the genotypes was not significant (online Supplementary Table S7). In the current study, the effect of genotypes on the stem diameter was significant under moisture-stress and non-stress conditions (Table 3). Having a sufficient stem diameter is required for the sunflower, so that it can tolerate heavy heads. The effect of genotype on the number of leaves and plant height were non-significant under both stress and non-stress conditions (Table 3).
The effect of genotypes on the RWC of sunflower was not significant under non-stress conditions (Table 3). However, the effect of genotypes on the RWC of sunflower was significant under drought-stress conditions (Table 3). The investigation of the correlation between the traits indicated that there was a positive non-significant correlation between seed yield and the RWC of the plant (online Supplementary Table S6).
Descriptive statistics on phenotypic traits showed that phenotypic variation is very high for seed yield and its components, morpho-physiological and phenological traits of sunflower. This range of variation provides an appropriate diversity for association analysis of molecular markers with traits (Table 4).
Table 4. A summary of descriptive statistics of the traits measured in 64 sunflower genotypes under both conditions
The most important molecular diversity criteria calculated based on 100 polymorph fragments amplified at 18 ISSR marker loci across all 64 genotypes. Hierarchical clustering was carried out to determine more exact number of sub-populations in the studied regions. The presence of three separate groups was identified by using the UPGMA algorithm on Nei's pairwise distance matrix (Doyle and Doyle, Reference Doyle and Doyle1987) of genotypes (Fig. 1). The obtained dendrogram represents these three groups. Genetic structure of these findings was illustrated by the frequency of alleles across all genotypes. A set of sampled sunflowers was used to investigate the genetic structure pattern. The maximum value of ΔK genotypes as the number of distinct subpopulations in sunflower genotypes was three (Fig. 2). Q-plot (online Supplementary Fig. S1) shows that subpopulations that were obtained of pattern of allele frequency distributed across the genotypes.
Fig. 1. Grouping of genotypes by a UPGMA dendrogram constructed based on Nei's distance matrix.
Fig. 2. The posterior probabilities ln P(D) calculated for different Ks and determining the optimum K by calculating ΔK from dividing mean ln P(D) by mean of SD[ln P(D)].
Association mapping for yield and yield components
In AM with TASSEL software by using the general linear model (GLM) for seed yield, number of plants per unit area, number of seeds per head, 1000-seed weight and oil yield traits 14, 4, 8, 7 and 12 alleles, was respectively identified. Also, by using the mixed linear model (MLM) it was recognized that 6, 4, 4, 4, 4 and 4 alleles were related to seed yield, number of plants per unit area, number of heads per head, 1000-seed weight and oil yield traits, respectively (online Supplementary Table S5). The R 2 value for seed yield ranged from 0.1657 to 0.0621% and for plant number per square metre ranged from 0.1096 to 0.0687% and for seed number per head ranged from 0.1806 to 0.0631% and for weight of 1000-seeds ranged from 0.1638 to 0.0651% and for oil yield ranged from 0.1838 to 0.087% (online Supplementary Table S5). In fact, these values represented the percentage of justification of the trait changes by each of these genomic fragments. A number of common markers for different traits were identified by GLM and MLM. For oil yield and seed yield, 1000-seed weight and number of seeds per head and number of plants per square metre traits gene locus (ubc808 (10)), for two, oil yield and seed yield traits, gene loci (ubc835 (3)) and (ubc835 (1)) and (ubc835 (4)) and (ubc824 (4)) and (ubc835 (2)) and (ubc835 (6)) and (ubc886 (3)) and (ubc891 (5)) and (ubc888 (4)) and (ubc853 (3)) and (ubc808 (10)) and (ubc884 (4)) and for three, oil yield, seed yield and 1000-seed weight traits, gene loci (ubc888 (4)) and (ubc808 (10)) for two, 1000-weight seeds and number of seeds per head traits, gene locus (ubc808 (6)) and for three, oil yield, seed yield and number of seeds per head traits, gene loci (ubc808 (2)) and (ubc886 (3)) and (ubc853 (3)) and (ubc808 (10)) with GLM and MLM has been identified (online Supplementary Table S5). The presence of common markers in some investigated agronomic traits may arise from polytropic efficiency or linkage of the genome regions could engage to control this trait. Identifying common markers for simultaneous selection several agronomic traits through marker-assisted selection are important in many breeding plans.
Based on the results of AM between ISSR markers and oil yield for the GLM statistical model (online Supplementary Table S5) with 12 bands produced for primers, alleles 3, 1, 4, 2 and 6 of primer ubc835, allele 4 of primer ubc824, allele 3 of primer Ubc886, allele 5 of ubc891 primer, allele 4 of ubc888 primer, allele 3 of ubc853 primer, allele 10 of ubc808 primer and allele 4 of ubc884 primer and for the MLM statistical model (online Supplementary Table S5) with five bands generated sequences 3, 1 and 4 of ubc835 primer, allele 4 of ubc824 primer, allele 4 of ubc888 primer had higher relationship with oil yield traits than other alleles and entered into regression models. For seed yield trait for the GLM statistical model (online Supplementary Table S5) with 14 bands generated for primers, alleles 1 and 3, 4, 6, 2 and 8 of ubc835 primer, allele 4 of ubc824 primer, allele 3 of ubc886 primer, allele 5 of ubc891 primers, allele 4 of ubc888 primers, allele 3 of ubc853 primers, allele 10 of ubc808 primers and alleles 4 and 5 of ubc884 primers and for the MLM statistical model (online Supplementary Table S5) with six bands generated for primers, primers 4 and 5 alleles. The ubc835 allele 4 of ubc824 primer, allele 4 of ubc888 primer and allele 3 of ubc886 primer had higher relationship than other alleles with seed yield and entered into regression models. Also 1000-seed weight for the GLM statistical model (online Supplementary Table S5) with seven bands produced for primers, alleles 6 and 10 of ubc808 primer, allele 3 of ubc884 primer, allele 4 of ubc888 primer, allele 2 of Ubc840 primer, alleles 3 and 2 of ubc812 primer and for the MLM statistical model (online Supplementary Table S5) with four bands produced for primers, alleles 6 and 10 of ubc808 primer, allele 3 of ubc884 primer and allele 4 Ubc888 primers had higher relationship than other alleles with 1000-grain weight and entered into regression models. For the number of seeds per head trait for the GLM statistical model (online Supplementary Table S5) with eight bands produced for primers, alleles 10, 6, 2, 7 and 8 of ubc808 primer, allele 3 of ubc836 primer, allele 3 of ubc853 primer and allele 3 of ubc886 primer and for the MLM statistical model (online Supplementary Table S5) with four bands produced for primers, alleles 10, 6 and 2 of ubc808 primer and allele 3 of ubc836 primer had higher relationship than other alleles with the number of seeds per head and entered into regression models. For number of plants per square metre trait with two MLM and GLM statistical models (online Supplementary Table S5) among the four bands totally produced for primers. Allele 10 of ubc808 primer, allele 1 of primer ubc888, allele 1 of Ubc844 primer, allele 3 of ubc840 primer had higher relationship than other alleles with the number of plants per square metre and entered the regression models. In general, the result of association analysis study showed that 45 of ISSR markers in the GLM are associated with yield and yield components and 23 of them were verified using the MLM association approach.
Association mapping for phenological traits
Phenological traits, number of days to flowering and number of days to reach physiology associated with the GLM with 7 and 12 alleles. Also, using the MLM model, four and six fragments (alleles) associated with, days to flowering and number of days to reach physiology traits, was respectively identified. R 2 values for the number of days to flowering ranged from 0.02068 to 0.031% and for days to physiology ranged from 0.03 to 0.0602% (online Supplementary Table S5).
Phenological traits, number of days to flowering for the GLM statistical model (online Supplementary Table S5) with seven bands produced for primers, allele 4 of UBC807 primer, allele 1 of UBC824 primer, allele 6 of UBC840 primer, allele 1 of UBC836 primer, allele 5 of UBC891 primer, allele 1 of UBC844 primer and allele 3 of UBC810 primer and for the MLM statistical model with four bands produced for primers, allele 4 of UBC807 primer, allele 1 of UBC824 primer, allele 6 of UBC840 primer and allele 1 of UBC836 primer had higher relationship than other alleles with the number of days to flowering and entered into regression models (online Supplementary Table S5). To characterize the number of days to physiological maturation for the GLM statistical model (online Supplementary Table S5) with 12 bands produced for primers, allele 4 of UBC807 primer, allele 5 of UBC808 primer, allele 1 of UBC844 primer, alleles 1, 7 and 5 of ubc810 primer, alleles 1 and 3 of ubc812 primer, allele 4 of UBC823 primer, allele 3 of ubc888 primer, allele 1 of ubc834 primer and allele 5 of ubc811 primer and for the MLM statistical model (online Supplementary Table S4) allele 4 of UBC807 primer, allele 5 of UBC808 primer, alleles 1 and 3 of ubc812 primer and allele 5 of ubc811 primer had higher association than other alleles with the number of days to physiological maturation and entered into the regression model. Also, 19 informative markers were identified for phenological traits and 10 of them were verified by the MLM.
Association mapping for morpho-physiological traits
Based on the GLM model for the morpho-physiological traits, head diameter, stem diameter and number of leaves and plant height and RWC of leaf 6, 8, 4, 3 and 11 alleles were identified, respectively. Also, using the MLM 6, 8, 1, 3 and 6 (fragments) alleles related to head diameter traits, stem diameter, number of leaves and plant height and leaf RWC was determined, respectively. R 2 value for head diameter ranged from 0.1238 to 0.0672%, for stem diameter ranged from 0.1378 to 0.081%, for leaf number ranged from 0.2284 to 0.074%, for plant height ranged from 0.11594 to 0.0634% and for leaf RWC ranged from 0.1594 to 0.0634% (online Supplementary Table S5).
Based on the results of analysis of the association between ISSR markers and the head diameter with the two MLM and GLM statistical models (online Supplementary Table S5) with six bands produced for primers, allele 4 of ubc888 primer, allele 3 of primer ubc844, allele 4 of primer ubc823, allele 1 of ubc812 primer, allele 4 of ubc836 primer and allele 7 of ubc808 primer had higher association than other alleles with head diameter and entered into the regression models. For leaf number trait with two statistical models, for GLM (online Supplementary Table S5) with four bands produced for primers, alleles 5 and 7 of ubc811 primer, allele 3 of ubc834 primer and allele 2 of ubc890 primer. For the MLM (online Supplementary Table S5) with one band generated for the primer, allele 5 of ubc811 primer had higher association than other alleles with the number of leaf and entered into the regression models. For stem diameter trait with two MLM and GLM statistical models (online Supplementary Table S5) with eight bands produced for primers, alleles 5 and 9 of ubc811 primer, allele 4 of ubc888 primer, allele 4 of ubc823 primer, alleles 3 of ubc812 primer, allele 4 of ubc807 primer and allele 6 of ubc891 primer had higher association than other alleles with stem diameter and entered into the regression models. For the plant height trait with MLM and GLM statistical models (online Supplementary Table S5) with three bands produced for primers (online Supplementary Table S5), allele 1 of ubc834 primer, allele 1 of ubc886 primer and allele 8 of ubc808 primer had higher relationship than other alleles with plant height trait and entered into regression models. Based on the results of association analysis between ISSR markers and leaf RWC for the GLM statistical model (online Supplementary Table S5) with 11 bands produced for primers, alleles 5, 4 and 3 of ubc808 primer, allele 3 of ubc888 primer, allele 7 of ubc811 primer, allele 4 of ubc835 primer, alleles 5, 3 and 2 of ubc891 primer, allele 1 of ubc884 primer and allele 1 of ubc823 primer and for the MLM statistical model (online Supplementary Table S5) with six bands produced for primers and allele 5 of ubc808 primer, allele 3 of ubc888 primer, allele 7 of ubc811 primer, allele 5 of ubc891 primer and allele 1 of ubc884 primer had a higher relationship than other alleles with leaf RWC and entered into regression models. Finally, 32 markers were informative for morpho-physiological traits and 24 of them verified using the MLM. The impact of recognized genomic fragments (QTLs) on the studied traits is shown through R 2 values (online Supplementary Table S5) along with all details of this association analysis.
All association results with at least 95% confidence (P-value < 0.05) have been reported for 96 significant loci identified to be associated with 12 traits (online Supplementary Table S5). In the K model, the obtained results were reconsidered by the MLM procedure. All association results with at least 95% confidence (P-value < 0.05) have been reported for 57 ISSR loci identified to be association with 12 agronomical traits. The number of significant markers was the second largest after the naive model. Similarly, a total of 57 markers were detected as significantly associated with 12 traits in the Q + K method. The P-value for associations between ISSR markers and morphological traits in the Q + K model are shown in online Supplementary Table S5. The calculation of K-matrix was the pre-request of running MLM association analysis. Fusari et al. (Reference Fusari, Rienzo, Troglia, Nishinakamasu, Moreno, Maringolo, Quiroz, lvarez, Escande, Hopp, Heinz, Lia and Paniego2012) in a linkage analysis study, identified gene loci that control sclerotinia resistance in sunflower. They used the MLM to identify a candidate gene that accounts for 20% of the phenotypic data variation. Vanitha et al. (Reference Vanitha, Manivannan and Chandirakala2014) by using SSR markers identified 29 markers in sunflower that showed a significant relationship with different agro-morphological traits. Saeed et al. (Reference Saeed, Wangzhen and Tianzhen2014) in AM cotton germplasm found that the MLM helps to reduce false positive results (false marker-trait communication) and less-biased results.
Discussion
It seems that implementing moisture stress in flowering and seed-filling stages can significantly reduce sunflower seed yield. Göksoy et al. (Reference Göksoy, Demir, Turan and Dağüstü2004) reported that for achieving high yields in sunflower plants, the plant is required to be drought stress free in three stages i.e. head formation, flowering stage and milk stage. The significant role of hormones is included as one of the main reasons stated in this regard. In stress treatments, the amount of ABA hormone increases; this hormone not only closes the apertures and reduces assimilate production, but also prevents the activities of IAA and CK that are responsible for dividing cells and increasing their lengths (Soleimanzadeh et al., Reference Soleimanzadeh, Habibi, Ardakani, Paknejad and Rejali2010). Moreover, aborting fertile florets under drought-stress conditions (due to the reduced access to assimilates and aborted pollens) has resulted in reduced seed yield (Göksoy et al., Reference Göksoy, Demir, Turan and Dağüstü2004; Benlloch-González et al., Reference Benlloch-González, Quintero, García-Mateo, Fournier and Benlloch2015; Elsheikh et al., Reference Elsheikh, Schultz, Adam and Mehari Haile2015). Reduced yield of sunflower under drought-stress conditions has been reported by some other researchers as well (Flagella et al., Reference Flagella, Rotunno, Tarantino, Di Caterina and De Caro2002; Göksoy et al., Reference Göksoy, Demir, Turan and Dağüstü2004; Babaeian et al., Reference Babaeian, Tavassoli, Ghanbari, Esmaeilian and Fahimifard2011; García-López et al., Reference García-López, Lorite, García-Ruiz and Domínguez2014; Hussain et al., Reference Hussain, Saleem, Iqbal, Ibrahim, Ahmad, Nadeem, Ali and Atta2015).
According to the reports released, sunflower yield has three important yield components: the first of which is the number of heads per unit area. Since most of the sunflower cultivars have one head, the number of heads per unit area is determined only through their plant density (Sadras et al., Reference Sadras, Connor and Whitfield1993). In other words, the only factor that can affect the number of sunflower heads per unit area is plant density, and factors such as genotype or moisture stress do not have a significant effect on the number of heads per unit area. Ali et al. (Reference Ali, Ahmad, Khaliq, Ali and Ahmad2013) conducted study to investigate the effect of microelements on the yield and yield components of sunflower at different levels of plant density. They concluded that increasing plant density resulted in increased sunflower seed yield through increasing the number of heads per square metre. However, micronutrients did not affect the number of heads per unit area. The increased number of seeds per head resulted in the increased sunflower seed yield.
The number of seeds per head is one of the yield components that are affected by the potential of genotypes and environmental factors (Ion et al., Reference Ion, Dicu, Basa, Dumbrava, Temocico, Epure and State2015). In the flowering stage, drought-stress can affect the reduced number of seeds per head due to the drying of pollen seeds and stigma and reduced activities of the insects (Andrade and Ferreiro, Reference Andrade and Ferreiro1996). Given the positive and significant relationship between this yield component and the seed yield (online Supplementary Table S6), the genotypes having a higher potential for forming the number of seeds per head seem to have a higher level of production. Mobasser and Tavassoli (Reference Mobasser and Tavassoli2013) have reported that stress at different stages of growth can result in the reduced number of seeds per head. However, the reduction percentage is more intense in reproductive stages. Other researchers have referred to the formation of more unfilled seeds due to drought stress (Lyakh and Totsky, Reference Lyakh and Totsky2014; Totsky and Lyakh, Reference Totsky and Lyakh2015).
It seems that drought stress and reduced accessible moisture at flowering and seed-filling stages resulted in reduced production of photosynthetic materials and disordered remobilization and allocation of photosynthetic materials to the seeds; this brought about the reduction of 1000-seed weight. Nezamia et al. (Reference Nezamia, Boroumand Rezazadehb and Hosseini2008) attributed an 83% reduction of sunflower seed yield under drought-stress to the reduced 1000-seed weight as well as reduced number of seeds per head. Hussain et al. (Reference Hussain, Malik, Farooq, Khan, Akram and Saleem2009, Reference Hussain, Saleem, Ashraf, Cheema and Haq2010) have reported that drought-stress at the flowering stage results in the abortion of seeds; this reduces the number of seeds per head. Thus, lighter seeds (having a lower weight) are formed and the seed yield is significantly reduced (Hussain et al., Reference Hussain, Malik, Farooq, Ashraf and Cheema2008).
Dagdelen et al. (Reference Dagdelen, Yilmaz, Sezgin and Gurbuz2006) have reported that oil percentage is more tolerant to drought stress than the seed yield. According to Dagdelen et al. (Reference Dagdelen, Yilmaz, Sezgin and Gurbuz2006) this is possibly owing to the fact that seed oil content is less affected by the environmental conditions (Dagdelen et al., Reference Dagdelen, Yilmaz, Sezgin and Gurbuz2006). Drought stress at both vegetative and reproductive stages results in a significant reduction of oil content and oil quality of the seeds (Flagella et al., Reference Flagella, Rotunno, Tarantino, Di Caterina and De Caro2002; Hussain et al., Reference Hussain, Malik, Farooq, Ashraf and Cheema2008; Ali et al., Reference Ali, Ashraf and Anwar2009). The main reason behind changes in the quality of oil is the change in the gene expression of non-saturated enzymes relating to oleic acid and linoleic acid under drought-stress conditions (Anastasi et al., Reference Anastasi, Santonoceto, Giuffrè, Sortino, Gresta and Abbate2010). It has been reported that implementing drought stress from the flowering stage until the physiological maturity increases oleic acid and reduces linoleic acid (Flagella et al., Reference Flagella, Rotunno, Tarantino, Di Caterina and De Caro2002).
Since diameter determines the number of grain seeds per head, it is included as one of the traits that is highly effective (Vega et al., Reference Vega, Andrade, Sadras, Uhart and Valentinuz2001). A positive and significant correlation was observed between the number of seeds per head (online Supplementary Table S6). Thus, one of the main goals of sunflower breeders is increasing the head diameters as to produce genotypes having a larger head diameter (Darvishzadeh et al., Reference Darvishzadeh, Maleki and Sarrafi2011). Other researchers have referred to the reduced head diameter due to drought stress (Hammad et al., Reference Hammad, Tahir, Imran and Hussain2002), it is likely that the main reasons of the conflicting findings are different intensities and times of drought stress implemented as well as different potential of the investigated genotypes in these studies.
The shortening of the time required for the reproductive growth of plants experiencing stress at the flowering stage can be attributed to reduced photosynthesis, non-production of sufficient assimilates for plant growth and the accelerated ageing process due to the drought stress (Dickinson and Dodd, Reference Dickinson and Dodd1976). The reduced flowering period, seed filling and maturity (due to drought stress) have been reported for other plants as well (Muchow, Reference Muchow1985; Link et al., Reference Link, Gee and Downs1990; Lopez et al., Reference Lopez, Johansen and Chauhan1994). Moreover, as it was mentioned in the previous section, drought stress resulted in the reduction of 1000-seed weight. It has been reported that reduced irrigation through shortening the seed-filling period (reproductive growth period) leads to the reduction of 1000-seed weight (Lemon, Reference Lemon2007; Nezamia et al., Reference Nezamia, Boroumand Rezazadehb and Hosseini2008).
Stem diameter has a significant role in saving assimilates during the vegetative period and the possibility of moving these materials during the seed-filling stage; the larger the stem diameter, the higher the desired production potential of the plant (Blum et al., Reference Blum, Golan, Mayer and Sinmena1997). Stem diameter is closely related to the growth status, transmission of photosynthetic materials to this plant organ during vegetative and reproductive stages (Zaffaroni and Schneiter, Reference Zaffaroni and Schneiter1991). Having a sufficient stem diameter is required for the sunflower, so that it can tolerate heavy head. Sadras et al. (Reference Sadras, Wilson and Lally1998) reported that the main reason for reduced stem diameter under drought-stress conditions is reduced vegetative growth and cellular division. However, another important reason is that drought stress (at the flowering stage) triggers the remobilization of stored assimilates from the stem to the seed; this is likely to result in reduced biomass and stem diameter (Rawson and Evans, Reference Rawson and Evans1971). Other researchers have referred to reduced stem diameter due to drought stress (Buriro et al., Reference Buriro, Sanjrani, Chachar, Chachar, Chachar, Buriro, Gandahi and Mangan2015).
Since the drought stress was implemented at the end of the vegetative stage, the number of leaves was not affected by drought stress. At different vegetative stages, even minor stress levels can reduce the leaf growth rate. However, in some studies, it has been reported that implementing drought stress at the reproductive stage results in leaf wilting, leaf yellowing or leaf falling (Farooq et al., Reference Farooq, Hussain, Wahid and Siddique2012; Lisar et al., Reference Lisar, Motafakkerazad, Hossain and Rahman2012) and thus reduced photosynthesis (Tezara et al., Reference Tezara, Mitchell, Driscoll and Lawlor1999; Farooq et al., Reference Farooq, Hussain, Wahid and Siddique2012; Lisar et al., Reference Lisar, Motafakkerazad, Hossain and Rahman2012). The yield is ultimately reduced by shrinking the canopy leaf area and diminishing the effectiveness of light usage and photosynthesis (Ghobadi et al., Reference Ghobadi, Taherabadi, Ghobadi, Mohammadi and Jalali-Honarmand2013). The number of leaves had a positive and significant correlation with the number of days to maturity (online Supplementary Table S6). Thus, it is likely that genotypes with more leaves have a longer growth period with the continuity of photosynthesis; under non-stress conditions, this is likely to increase the oil yield.
The main reasons behind the reduced plant height are cell growth reduction (due to drought stress) and the allocation of plant energy for increasing the root length (Turhan and Baser, Reference Turhan and Baser2004).
The RWC of the plant refers to the plant capability in maintaining water values under stress conditions (Bajji et al., Reference Bajji, Lutts and Kinet2001). Reduced RWC and water potential have been reported in different plants (Nayyar and Gupta, Reference Nayyar and Gupta2006). A reduced RWC level in plants depends on several factors including stress severity and duration as well as the plant species (Yang and Miao, Reference Yang and Miao2010). Hussain et al. (Reference Hussain, Saleem, Iqbal, Ibrahim, Ahmad, Nadeem, Ali and Atta2015) have reported that drought significantly affects the water relationships of sunflower hybrids having different degrees of tolerance to drought stress. The main reason for the superiority of same genotypes is their high capability in absorbing water from deep soil layers, reduced water loss through apertures and maintaining turgor pressure (Siddique et al., Reference Siddique, Hamid and Islam2000; Terzi and Kadioglu, Reference Terzi and Kadioglu2006; Bayoumi et al., Reference Bayoumi, Eid and Metwali2008). Naeem et al. (Reference Naeem, Ahmad, Kamran, Shah and Iqbal2015) reported the high correlation between grain yield and RWC by evaluating physiological responses of wheat under drought-stress conditions. They concluded that if the plant is capable of maintaining the RWC at a high level and prevents the water reduction of the plant under stress conditions, it will have a higher chance of success in confronting the stress.
The Q-matrix as a cofactor was used in latter association analyses. A large number of markers suggesting associations the between genotypes and phenotypes and using genotyping and phenotyping data sets along with Q-matrix was detected by the association analysis by using the model without population structure and kinship GLM. The model was used to detect linked markers to yield and yield components, morpho-physiological and phenological trait of sunflower.
Conclusions
Due to the importance of sunflower oil in the production of community food and water shortages, it is necessary to improve oil yield and seed yield of sunflower. The findings of this study indicate that there is a large diversity in the drought tolerance traits. Through adopting improving methods, it is possible to create drought-tolerant lines having a higher yield under drought-stress conditions. In the current study, only the number of seeds per head had a higher correlation with the seed yield than the other traits. The technique (ISSR) is just confirmed genetic control of these traits. Association results with at least 95% confidence (P-value < 0.05) have been reported that 96 significant loci were identified to be associated with 12 traits.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1479262120000040.
