Diallel analysis in cotton ( Gossypium hirsutum L . ) for water stress tolerance

Information about combining ability is important in determining breeding strategies. The objective of this study was to estimate the general (GCA) and specific (SCA) combining abilities among twenty cotton hybrid combinations submitted to water stress, based on agronomic traits. The trails were carried out in semiarid environment (Barbalha, CE), during two years, in dry season under irrigation, and submitted to 23 days of water suppression. Wide variability was verified among parents and their hybrids, with additive effects to seed yield, lint percentage, blooming and boll weight. Dominance effect was found only to plant height. The combinations BRS 286 x CNPA 5M, BRS RUBI x CNPA 5M, FM 966 x CNPA 5M and BRS 286 x BRS Seridó showed the best SCAs and averages for the majority of the traits. In addition, at least one of their parents had high CGA. These combinations must be advanced in cotton breeding program.


INTRODUCTION
Cotton (Gossypium hirsutum L.) is one of the most important textile crop worldwide.It is cultivated on about 35 million hectares, generating a global production of 24 million tons of fiber (Abrapa 2016).China, India, United States, Pakistan and Brazil are the main fiber producers (FAO 2016).The crop management involves more than 350 million people, in Brazilian cotton belt, located at Cerrado region, with production exceeded 1.5 million tons (Abrapa 2016, Conab 2016).
Cotton plant has a phenological cycle ranging from 120 to 200 days, depending on the cultivar, with different responses to environment conditions in order to ensure the fiber yield.Reproductive phase are quite dependent on water and fertilization supplements.According to Batista et al. (2010), water suppression in cotton plants for 23 days causes reduction in plant height, fiber growth, bud and boll numbers, although root system length is increased as a response to dry condition.
In Brazil, the main commercial cultivars have limited adaptation to environments with water irregularities.Considering the current climate changes faced on last decades, especially those related to water scarcity, the identification of cultivars with ability to adapt to environments with water restriction is a valuable strategy adopted in several crop breeding programs.Therefore, the generation of populations with broad genetic base is necessary in order to increase the chances to identify water-tolerant materials in further selection procedures.Diallelic design is often adopted to study genetic inheritance of traits, supporting in breeding strategy, as selection of individuals to create a breeding population.The scheme consists in crossings among the parents, resulting in hybrids with all possible combinations and providing combining ability of the parents.
The estimative of general combining ability (GCA) refers to capability of parental to generate favorable combinations in order to promote genes predominantly additive.The higher values estimated, positive or negative, the parent is considered higher or lower than others included in a diallel.By other side, the specific combining ability (SCA) refers to dominance effects and also to non-additive interactions of genes, resulting from gene complementation between parents.Then, the hybrid combination that shows high SCA (s � ij ) and at least one parent with high GCA will be the most favorable (Cruz et al. 2012).
Several studies have been available in literature reporting on diallel analysis in cotton.Raza et al. (2013) evaluated five cultivars and detected additive effects to plant height, lint yield, lint percentage and boll weight.Khan and Hassan (2011) estimated the GCA and SCA in a diallel scheme involving 12 cotton cultivars and found differences in all hybrid combinations to lint yield, boll weight and boll numbers, with more genetic variances to SCA than to GCA indicating predominance of non-additive gene.
The genetic study of cotton under water restriction can support the cotton breeding program in semiarid environments.Thus, the objective of this research was to evaluate the hybrid combination, with the diallel design under water stress condition in order to identify the best combinations to water stress tolerance, based on GCA and SCA estimates.

MATERIAL AND METHODS
Twenty F 1 hybrid combinations generated from partial diallel crossing were used for this study using two parents groups: GI -formed by five cultivars with high fiber yield (Cia et al. 2014, Carvalho et al. 2015, Coutinho et al. 2015, Zonta et al. 2016) and GII -composed of by four drought tolerance cultivars (Freire et al. 1999, Vidal Neto andFreire 2013).These cultivars are inbred lines developed by Brazilian Company of Agricultural Research (Embrapa), Bayer and Mato Grosso Foundation (FMT) (Table 1).
The trials were carried out in Embrapa Cotton Experimental Station, in Barbalha (lat 7° 18′ 18″ S, long 39° 18′ 7″ W, alt 414 m asl), CE, in two years, during the dry season (July to December, 2014 and2015).The mean of temperature recorded during the trial was 24.1 °C.The soil is classified as Fluvic Neosoil, clay loam texture (Embrapa 1999).Based on fertilization soil analysis, a previous fertilization was performed by using 250 kg ha -1 of monoammonium phosphate (MAP) at sowing, and further 125 kg ha -1 of urea-N fertilization divided into two applications (35 and 50 DAE).
Plants were submitted to two water treatments: Control (C): normal watering, by sprinkler irrigation system during all cycle, and Stress (S): water suppression for 23 days, starting at 42 days after emergence (DAE), corresponding to blooming.The sprinkler irrigation was maintained with a flow rate of 1.49 m³ h -1 , according assays previously carried out by Bezerra et al. (2010).The plot consisted of two 5 m-rows, spaced in 0.90 m.Plants were spaced in 0.14 m, maintaining a population of 70 plants plot -1 .A randomized block design was adopted with 20 treatments and 3 repetitions.Plants were maintained free of pests and weeds by using pesticides.At harvest (128 DAE), the following traits were registered: plant height (PH, cm), cotton seed yield (CSY, kg ha -1 ), lint percentage (LP, %) and boll weight (BW, g).
The statistical-genetic analyzes were performed using the GENES program version 6.1 2014 (Cruz 2016).F and Scott and Knott (1974) tests were adopted to variance analysis (ANOVA) and mean comparisons, respectively.The joint analyses were performed considering genotypes as fixed effect and years and water treatments were random effects (Cruz et al. 2012).A partial diallel analysis model was adopted to genetic analysis, according to Griffing (1956)'s method 4 and adapted by Geraldi and Miranda Filho (1988) in order to estimate the effects of GCA of each parent and the effects of SCA among their combinations.

RESULTS AND DISCUSSION
Combinations resulting from a partial diallel crosses with upland and Moco cultivars were grown in Barbalha (CE), under irrigation and water suppression (for 23 days) treatments.The physiological symptoms caused by water stress in the plants were observed soon, after the 2 nd week of water suppression.In this occasion, plants showed slow growth and reduction in leaf turgidity.At the end of water suppression, irrigation was restored and the symptoms resulting from water stress were disappearing over plant growth.
The results of ANOVA (Table 2) showed that significant statistic differences (p<0.05) were verified in Genotype (G), Year (Y) and Water Treatment (WT) to all traits, excepting to boll weight (BW) in year (Y) and water treatment (WT).For these source of variation, the F test uses the interaction Y x WT.As it was very high and significant for this trait, it caused no-significance effect for them (Y and WT).The same results were seen for G x WT, that uses G x Y x WT interaction to test it, indicating different behavior of cultivars submitted to water treatments.On the other hand, the interaction G x Y did not demonstrate significant effect indicating that the genotypes had similar behaviors in both years (2014 and 2015).The CV values ranged from 2.14 for LP to 10.48% for CSY, and they are in accordance with the results found by Carvalho et al. (2015), Carvalho et al. (2016) and Zonta et al. (2016) for this crop.The means of traits (Table 3), obtained from different hybrid combinations submitted to two water treatments, demonstrate that the water suppression influenced the growth and production of plants at different levels.For all traits, some hybrid averages showed good potential to develop new genotypes with better performance in water stress condition.The number of groups formed by Scott-Knott test ranged from two groups (PH) to seven groups (LP).These results are expected due to the differences among the genetic variability and the influence of the environment in each trait.The reduction in height ranged from 12.88 to 24.71%.This difference promoted a direct impact in yield, with losses ranging from 3.08 to 44.78%.These results explain the effect seen in G x WT, in which some combinations were more negatively impacted by water stress, such as FMT 705 x BRS Seridó (1 x 9) with yield reduced of 44.78%, revealing a highly sensitive genotype.In contrast, BRS 286 x CNPA 5M (4 x 7) showed an excellent performance, demonstrated by high capacity to produce fibers, even in a situation of water stress, as imposed in this work.
Others five combinations are also productive in water stress conditions, clustered in "a" group, classified by Scott and Knott test: FM 966 x CNPA 5M (2 x 7), BRS RUBI x CNPA 5M (3 x 7) and BRS 286 x BRS Seridó (4 x 9), all of them with reduced losses in yield.By the way, BRS RUBI x CNPA 5M (3 x 7) showed a differential behavior, increasing the yield in 21.88% in water stress treatment.It means an interesting ability to physiological restoration, which is often associated with tolerance to drought.
Taking in account the others traits, we verified that FM 966 x CNPA 5M (2 x 7) and BRS 286 x BRS Seridó (4 x 9) also maintained slight gains for LP and BW, in stress treatment.As to blooming, we found that water suppression did not influence hardly in this trait due to it occurred in the period close to flowering (42 DAE).However, we found that the most combinations had a slight earliness to starting the first flower (Table 3).These results agree with Batista et al. (2010) that submitted cotton plants to different methods of irrigation and found that in water stress condition an earlier blooming took place in genotypes.As to Gowda and Hedge (1986), this earliness is often associated to reduction of plant cycle, and it is a desirable feature, especially in environments prone to "indian summer" (veranicos) or water rainfall irregularities.Consequently, the amount of reproductive structures is reduced (Baldo et al. 2009).
The joint diallel analysis of the F 1 hybrids for cotton traits (Table 4) was performed only under stress treatments in order to select the parents and their hybrid combinations for this condition.A significant effect of GCA was seen in both groups I and II to all traits (p<0.05),excepting to B, in GCA-II, indicating additive effects in the control of PH, CSY, LP and BW traits.For SCA, no significant effect was found only to BW, indicating non-aditive (dominance or epistasis) to others traits.These results agree with those found by Aguiar et  Crop Breeding and Applied Biotechnology -18: 24-30, 2018 UAA Vasconcelos et al. al. (2007), Bertini et al. (2005) and Menezes et al. (2013) and mean that, at least, one hybrid combination differs from others, suggesting complementation of favorable alleles between parents, in loci with dominance (Cruz et al. 2012).
These results demonstrate the importance of the GCA and SCA effects in the genetic variance of genotypes, especially GCA in all traits, accounting to whole variance of 24.27% (PH), 72.13% (CSY), 93.74% (LP), 87.64% (BW) and 64.12% (B), whose can be exploited in selection procedures of hybrid combinations in order to identify top lines and achieve expressive genetic gains.The effects of interaction G x Y, GCA-I x Y, GCA-II x Y and SCA x Y were all not significant (p>0.05),indicating similar behavior of genotypes during two years in this study.The only exception was found to BW, which revealed significant effect to all interactions, suggesting variations in behavior of parents (GCA-I and II) and hybrids (SCA).
The estimates of GCA (g � ) for genotypes of the groups I and II (Table 5) detected a significant and distinct contributions from the parents in order to improve the traits of hybrids demonstrating the contribution of diallel crossing to crop improvement, focusing on yield and adaptation to semiarid environment.Highlights were found to BRS 286 (GI) and CNPA 5M (GII) that contributed to increase the yield of hybrids in 515.06 kg ha -1 and 905.24 kg ha -1 , respectively.BRS 286 (GI) also contributed to increase the height of plants in 1.16 cm.Of course that this trait must be carefully maintained in selection procedures, in order to avoid further problems to mechanical harvesters.According to Chanselme and Ribas (2010), cotton plants should get 90 cm, at most, in the harvest period.

Table 1 .
Groups of cotton parents used in the partial diallelic crossing scheme UAAVasconcelos et al.