Genetic combining ability of coriander genotypes for agronomic and phytochemical traits in response to contrasting irrigation regimes

Knowledge of genetic combining ability and gene action would help breeders to choose suitable parents and devise an appropriate breeding strategy for coriander. In the present study, six diverse genotypes of coriander, their 15 F1s and 15 F2s were evaluated through randomized complete block design with three replications to study genetic combining ability for agronomic and phytochemical traits in coriander. Plants were subjected to well-watered (WW), mild water-deficit stress (MWDS) and severe water-deficit stress (SWDS) irrigation regimes. The results indicate that water-deficit stress decreased all of the measured traits in both the F1 and F2 generations. General combining ability and specific combining ability effects were highly significant for all of the traits in both the F1 and F2 generations. Additive gene action was predominant for phonology and fruit yield component traits in all irrigation regimes in both the F1 and F2 generations. For fatty acid content and total lipid yield, non-additive gene action was predominant in the F1 generation while additive gene action was predominant in the F2 generation under MWDS and SWDS conditions. The P4 parent had the highest general combining ability for fruit yield components in both the F1 and F2 generations. The P6 parent had the highest general combining ability for phenological and phytochemical traits. The P4 and P6 parents are promising material to develop early flowering and early maturing genotypes coupled with high total lipids in advanced generations of segregation.


Introduction
Coriander (Coriandrum sativum L.) is an annual herb that belongs to the umbelliferous plant family, the Apiaceae. The rapid life cycle of some coriander genotypes allow them to be cultivated in the wide range of geographical areas throughout the world [1]. Fresh and dried leaves and seeds are commonly used as a seasoning and a general food ingredient [2]. Coriander is a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 corresponds to additive genetic effects, while the SCA is defined as the yield of a hybrid that deviates from what would be expected if traits were controlled by additive effects alone, i.e. it represents non-additive genetic effects [34].
There is insufficient knowledge of the genetic control of phenological and yield component characteristics in coriander. The objectives of this study were (1) to quantify heritability and the nature of gene action controlling phonological traits, yield components and total lipid yield traits, and (2) to estimate genetic combining ability of parents and hybrids.

Plant material and growth conditions
The coriander genotypes used to make diallel crosses had been evaluated in a preliminary experiment for drought tolerance by Khodadadi et al. [35]. Parents included the commercial genotype (P 1 ), TN-59-353 (P 2 ; relatively drought tolerant), TN-59-80 (P 3 ; drought susceptible), TN-59-160 (P 4 ; drought tolerant and relatively high yielding), TN-59-158 (P 5 ; highly drought susceptible) and TN-59-230 (P 6 ; highly drought tolerant but low yielding). All six parents were used in half diallel mating design, without reciprocals, to produce 15 F 1 hybrids in 2014. Seeds of these F 1 hybrids were used to produce 15 F 2 generations through self-pollination in isolated conditions. All of the six parents, the F 1 hybrids and F 2 populations were evaluated in different irrigation regimes in experiments with a randomized complete block design with three replications in each experiment during growing season of 2016. Tests were carried out at the research field of Tarbiat Modares University in Iran (51˚09 0 E longitude and 35˚44 0 N latitude, at an elevation of 1265 m above sea level). In treatment 1, genotypes were kept well-watered overall (WW). In treatment 2, genotypes were well-watered until the commencement of stem elongation when watering was withdrawn until the end of the flowering stage at which time only one recovery watering was applied (mild water-deficit stress; MWDS). In treatment 3, watering was normal until the commencement of the flowering stage, after which watering was cut off completely (severe water-deficit stress; SWDS). The soil's physical and chemical characteristics in the experimental field are presented in S1 Table.

Trait measurements
The traits which were measured included days to flowering (DTF), days to the end of flowering (DTEOF), days to ripening (DTR), umbel number per plant (UNPP), fertile umbel number per plant (FUNPP), fruit number per plant (FNPP), thousand fruit weight (TFW), fatty acid content (FAC) and total lipid yield (TLY). The timing of phenological traits were noted at the time at which 50% of plants in each plot had reached the target phonological stage. Sample size to measure yield components varied with genetic material; FAC and TLY traits were measured in ten plants in each plot for parental genotypes and F 1 hybrids and in 30 plants in each plot for F 2 populations.
To measure fatty acid content, two grams of powdered fruit samples of coriander were subjected extraction with a Soxhlet apparatus with 250 ml of petroleum ether for 6 h. Fatty acid content was measured after filtration and solvent evaporation under reduced temperature and pressure [18]. Finally, total lipid yield was estimated by multiplying fatty acid content with fruit yield per plant (g) for each plot. method 2, model 1 using a SAS program proposed by Zhang et al. [38]. Mean values of the traits in different irrigation regimes were compared using least significant difference (LSD) method. Estimates of s 2 g (general combining ability variance) and s 2 s (specific combining ability variance) were computed based on the random-effects model of Griffing's [37] method. These estimates were used to calculate σ A 2 (additive variance), σ D 2 (dominance variance), h 2 (heritability), and the GCA/SCA ratio [38]. The relative importance of variances due to GCA and SCA were computed for the traits using the method proposed by Baker [39] (Eq 1).
The GCA/SCA ratio reflects the degree of trait transmission from parent to the progeny. When the GCA/SCA ratio is closer to one, it shows that additive gene action is largely involved in the inheritance of the trait and it will be well transmitted from the parents to the progenies. Whereas, a GCA/SCA ratio closer to zero shows that non-additive gene action is predominant in the inheritance of the trait. Narrow-sense heritability (h N 2 ) was computed according to Eq 2 [40].
where σ E 2 and r are the error variance and number of replications, respectively. The genotypic correlation coefficients between traits were calculated according to the formula proposed by Holland [41]. The statistical analysis was carried out using SAS [42] software.

Combined analysis of variance of traits
Combined analysis of variance demonstrated that there were significant effects of different irrigation regimes on all of the traits in both the F 1 hybrids and the F 2 populations (S2 Table). Genetic differences between F 1 hybrids and between F 2 populations were highly significant for all of the studied traits. These results indicate that parents for diallel crosses had been properly selected. Also, genotype × irrigation regime interaction effects were significant for all traits in both F 1 hybrids and F 2 populations (S2 Table). The GCA and SCA effects were significant for all traits. Also, GCA × irrigation regime interaction effect was significant for all the traits in both the F 1 and F 2 generations. The SCA × irrigation regime interaction effect was significant for all traits in both the F 1 and F 2 generations except for TFW trait.

Nature of gene action
The GCA and SCA variances were highly significant for all traits in both the F 1 and F 2 generations (Tables 1 and 2). The GCA/SCA ratio values were high for phenological traits and relatively high for yield components in all irrigation regimes (Tables 1 and 2). These results indicate that additive gene action was predominant in controlling these traits. For FAC nonadditive gene action was predominant in the F 1 generation, while additive gene effects were important in the F 2 generation under all irrigation regimes (Tables 1 and 2). In addition, in WW condition, non-additive gene action was predominant for TLY in both the F 1 and F 2 generations (Tables 1 and 2). Under MWDS and SWDS irrigation regimes, non-additive gene action was predominant for TLY in the F 1 generation, while additive gene effects were important in the F 2 generation (Tables 1 and 2).

Narrow-sense heritability
Heritability estimates for all traits are presented in Tables 1 and 2

Genetic combining ability analysis
GCA values of parents in both the F 1 and F 2 generations showed that the P 6 parent was the best general combiner for phenological traits that enable plants to reach early ripening in all irrigation regimes; it had the largest negative GCA value for days to flowering (Table 3). In the case of UNPP, FUNPP and FNPP traits, the P 4 parent was the best general combiner in both the F 1 and F 2 generations in all irrigation regimes (Table 3). Also, the P 6 appeared as the best general combiner for TFW and FAC in all irrigation regimes in both the F 1 and F 2 generations. In the case of TLY, the P 4 parent was the best general combiner in WW conditions in both the F 1 and F 2 generations, while in MWDS and SWDS irrigation regimes, the P 6 parent had the largest positive GCA value for TLY in both the F 1 and F 2 generations ( Table 3).
Results of SCA analysis for DTF, DTEOF and DTR indicated that the progenies of P 6 (H 1 × 6 , H 2 × 6 , H 3 × 6 , H 4 × 6 and H 5 × 6 ) displayed negative significant SCA-effects in both the F 1 and F 2 generations in all irrigation regimes (Tables 4 and 5). In WW conditions, the crosses of H 1 × 6 and H 4 × 6 had the largest positive significant SCA values for UNPP in both the F 1 and F 2  generations. Under MWDS conditions, the SCA value for UNPP was not significant. Under SWDS conditions, the cross of H 1 × 6 had the largest positive significant SCA value for UNPP in the F 1 generation, whereas, in the F 2 generation, the population of H 3 × 6 had the largest positive significant SCA value for this trait (Tables 4 and 5).    (Tables 4 and 5).
In case of FNPP, the cross of H 4 × 5 had the largest positive significant SCA value in WW and SWDS irrigation regimes in both the F 1 and F 2 generations. In MWDS conditions, the crosses of H 2 × 4 and H 4 × 5 had the positive significant SCA values for FNPP in both the F 1 and F 2 generations. For TFW, the progenies of the P 6 parent (H 1 × 6 , H 2 × 6 , H 3 × 6 , H 4 × 6 and H 5 × 6 ) had positive significant SCA values in all irrigation regimes in both the F 1 and F 2 generations (Tables 4 and 5).
Under WW conditions, the crosses of H 1 × 4 , H 1 × 6 , H 3 × 6 , H 4 × 5 and H 4 × 6 had positive significant SCA values for FAC in the F 1 generation whereas, in the F 2 generation, the populations of H 1 × 4 and H 4 × 6 had the positive significant SCA values. In MWDS conditions, the crosses of H 1 × 6 , H 3 × 6 and H 4 × 6 had the positive significant SCA values for FAC in both the F 1 and F 2 generations. Also, in SWDS conditions, the crosses of H 1 × 6 and H 4 × 5 had a positive significant SCA value for FAC in the F 1 generation and the population of H 4 × 6 had a positive significant SCA value in the F 2 generation (Tables 4 and 5).
Under WW conditions, the crosses of H 1 × 4 and H 2 × 4 had the largest positive significant SCA values for TLY in both the F 1 and F 2 generations. Under MWDS conditions, the crosses of H 1 × 6 , H 3 × 6 and H 4 × 6 had the largest positive significant SCA values for TLY in both the F 1 and F 2 generations. Also, under SWDS conditions, the crosses of H 1 × 6 and H 4 × 5 had the largest positive significant SCA values for TLY in the F 1 generation and the population of H 4 × 6 had the largest positive significant SCA value in the F 2 generation (Tables 4 and 5).

Genetic correlation of total lipid yield with phenological and morphological traits
Genetic correlation analysis under WW conditions showed that there were positive correlations between total lipid yield and all of the phenological and yield components traits (S4 Table). In MWDS and SWDS conditions, total lipid yield was significantly and positively correlated with yield components while total lipid yield had the significant negative correlation with phenological traits (S4 Table).

Discussion
Iran has special a geographical location with high genetic diversity for coriander, as well as many other crops. It is a promising place to find and gather new genetic resources for coriander anda high genetic diversity was previously reported for drought stress tolerance for Iranian coriander genotypes [35]. Using drought-tolerant and high-yielding coriander genotypes as the parents in crossing programs can significantly increase the efficiency of coriander breeding schemes for developing high-yielding coriander genotypes for arid and semi-arid areas. In this study, a large genetic variation was observed for phonological and yield components, total lipid yield and fatty acid content among the parental genotypes, F 1 hybrids and F 2 populations. This indicates the existence of an excellent potential to study coriander genetics relative to crop improvement.
Results showed that flowering and maturity times were decreased in MWDS and SWDS irrigation regimes. In agreement with our results, Gales and Wilson [43] with studies in winter wheat, Bannayan et al. [20] in isabgol and black cumin and Alinian and Razmjoo [23] in cumin (Cuminum Cyminum L.) reported that water-deficit stress induced a reduction in the time to maturity. This effect is influenced by various factors including the level and duration of the stress, the genotype and the maturity time under non-stress conditions. Reduced time to maturity is known as a water-deficit stress avoidance mechanism in plants [43].
Yield components' measurements were higher under WW condition than with MWDS and SWDS irrigation regimes. Reduction in yield components may be due to lower availability of nutrients along with reduced photosynthesis and reduced translocation of photosynthesis products from source to sink area under drought stress [44]. The preferential allocation of biomass to the root growth has been associated with yield reduction in coriander under water stress conditions [26]. Similarly, Alinian and Razmjoo [23] reported that number of umbels per plant, number of seeds per umbel and 1000 seeds weight were reduced under water-deficit stress in cumin accessions.
The highest fatty acid content and total lipid yield values observed under WW conditions, conversely the lowest fatty acid content and total lipid yield values were obtained in SWDS conditions in both the F 1 hybrids and the F 2 populations. Similar results were reported by Singh and Ramesh [45] in rosemary, Zehtab-Salmasi et al. [19] in dill (Anethum graveolens L.), Hamrouni et al. [13] in safflower, Bettaieb et al. [11] in Salvia officinalis L. and Bettaieb et al. [14] in cumin (Cuminum cyminum L.). Those studies observed that fatty acid content and total lipid yield were significantly decreased by water-deficit stress. Reduction in total lipid yield under water-deficit stress could also be due to the reduction in days to flowering and maturity and some of the yield components for plants.
Narrow-sense heritability and GCA/SCA ratio values suggest that additive genetic effects were predominant in controlling phenological and yield components traits in all irrigation regimes in both the F 1 and F 2 generations. Therefore, breeding methods based on selection can be effective in the F 2 generations for improvement of these traits. Similar to our findings, Amiri-Oghan et al. [25] observed a high heritability for days to flowering and days to maturity in oilseed rape (Brasica napus L.). FAC and TLY In coriander were predominantly affected by non-additive gene action in the F 1 generation while additive gene action predominated in the F 2 generation under MWDS and SWDS conditions. Therefore, breeding methods based on selection in the F 2 and later generations will likely be effective to improve FAC and TLY in MWDS and SWDS conditions. The results of narrow-sense heritability and the GCA/SCA ratio for TLY in WW indicate that non-additive type of gene action were predominant in both the F 1 and F 2 generations. Therefore, for improvement of TLY under WW conditions, selection should be deferred to the later generations of segregation in which non-additive genetic effects have been reduced or fixed.
Blum [46] reported that indirect selection for yield components and other traits that have high heritability and which are strongly correlated with economical yield could be more efficient than direct selection for yield. In this study phenological and TFW traits had higher heritability estimates than total lipid yield. These traits also had a significant genetic correlation with total lipid yield. Thus, selection for DTF, DTEOF, DTR and TFW traits may be effective criteria for improvement of total lipid yield, especially under water stress conditions. The significant and negative genetic correlations between total lipid yield and phenological traits in MWDS and SWDS conditions suggest that simultaneous improvement of earliness to cope with drought stress and total lipid yield can be achieved in coriander. Flowering is the most critical stage that influences the yield of coriander. The time of flower initiation can have a strong influence on the number of flowers, umbel number per plant, fertile umbel number per plant and fruit number per plant. The development of early ripening coriander genotypes is important to avoid abiotic stresses, particularly drought and high temperatures at the end of the growing season in arid and semi-arid environments. Such conditions have been observed in some areas of Iran and coriander production has been restricted by the adverse effects of the terminal heat and drought stress which cause a reduction in the number of successfully pollinated flowers. Overall, the importance of phenological traits and the use of early genotypes as donor parents should be considered in coriander breeding programs to improve total lipid yield.
Achievement in breeding programs depends on the careful choice of parents. The selection of parents for hybridization programs should be based on their genetic value. High GCA values of the parents are mainly due to the additive genetic effects [37] that are heritable in the segregating generations. Therefore, the selection of parents for hybridization should be based on their GCA-effects which reflect on their potential to produce superior segregates in the F 2 and later generations. The GCA-effects of the six parents on the traits measured in both the F 1 and F 2 generations showed that the P 4 appeared as the best general combiner for UNPP, FUNPP and FNPP. P 6 was the best general combiner for DTF, DTEOF, DTR, TFW, FAC and TLY. These parents could be used to develop early flowering and early maturing types coupled with high total lipid yield genotypes in advanced segregating generations. The offspring of the P 4 and P 6 parents had high SCA values for FAC and TLY. They also exhibited significant SCA values for other phenological and yield components traits in both F 1 and F 2 generations. Results of the GCA and SCA analysis indicate that many of crosses which showed significant SCA-effects also had high GCA values for all traits.

Conclusion
Large genetic variability for phenological, yield components, total lipid yield and fatty acid content indicate a high potential of the studied germplasm for genetic improvement in coriander. The results indicated that water-deficit stress decreased DTF, DTEOF, DTR, UNPP, FUNPP, FNPP, TFW, FAC and TLY in both F 1 hybrids and F 2 generations. The high narrowsense heritability and GCA/SCA ratio for phenological traits indicates that these traits are mainly governed by additive genetic effects and suggest that these traits can be used as reliable and heritable selection criteria under drought stress. These traits were also correlated with total lipid yield and could be used as suitable surrogate selection criteria to enhance total lipid yield and to identify superior genotypes for drought stress conditions. Based on their general combining ability, the P 4 and P 6 parents can be used as promising parents for hybridization and selection of genotypes with high total lipid yield coupled with early ripening in advanced generations of segregation.