Diallel analysis of tropical and temperate sweet and supersweet corn inbred lines 1

The number of sweet corn cultivars adapted to the climatic conditions and with agronomic traits recommended to the Brazilian market is still limited. Thus, the aim was to investigate the general and specific combining ability (GCA and SCA, respectively) in relation to the grain yield (GY) and total soluble solid (TSS) contents of seven lines of sweet corn (su1) and eight lines of supersweet corn (sh2) by diallel crosses. For TSS, the inbred lines SC6 and SC7 of sweet corn, and SSC7 and SSC8 of supersweet corn showed higher GCA effect. For GY, the inbred lines SC1 and SC2 of sweet corn and SSC4 and SSC6 of supersweet corn showed higher GCA effect. For TSS in sweet corn, the hybrids ‘4 × 6’ and ‘1 × 7’ showed the superior Ŝij estimates across all evaluated environments. For GY, the hybrid ‘5 × 7’ was the most relevant because it showed higher Ŝ ij estimates in most of the evaluated environments. Additionally, the hybrids ‘1 × 4’, ‘3 × 4’, and ‘1 × 7’ showed positive Ŝ ij values across all environments. In supersweet corn, the hybrids ‘2 × 8’, ‘3 × 6’, and ‘5 × 6’ should be selected as the most promising for both traits (TSS and GY), because they presented positive and high Ŝij estimates across the six environments.


INTRODUCTION
Sweet corn (Zea mays L. saccharata) is considered one of the most popular vegetables in the US and Canada (TRACY, 2001) because of its improved shelf life and taste (DODSON-SWENSON;TRACY, 2015).In Brazil, the cultivation area of sweet corn is 36,000 ha (SANTOS et al., 2014), which represents only 14.4% of the area occupied by US sweet corn (UNITED STATES DEPARTMENT OF AGRICULTURE, 2014).The destination of sweet corn, mainly used as green maize, is exclusively for human consumption, due to the fact that it has sweet taste, fine pericarp and endosperm, and delicate texture (SA et al., 2010;TEIXEIRA et al., 2001;TEIXEIRA et al., 2014;TRACY, 2001).
The sweetness of sweet corn is a recessive trait controlled by the mutant genes sugary 1 (su 1 ), shrunken 2 (sh 2 ), brittle 1 (bt 1 ), sugary enhancer (se) and brittle 2 (bt 2 ).These genes may act in a single, double or triple combination (SINGH;LANGYAN;YADAVA, 2014).All of them promote changes in carbohydrate composition in the endosperm and differ in the starch and sugar proportions in the kernel and in relation to chromosome position (TRACY, 2001).Supersweet corn is a variety of corn in which the homozygous hybrid has the shrunken-2 (sh 2 ) gene (TRACY, 2001;YOUSEF;JUVIK, 2002) with about a 9 to 14% sugar content, and present about 15 to 25% of accumulated sugar in the endosperm.This higher accumulation results in a longer harvesting period, which implies a slower conversion of sugar to starch (MARSHALL;TRACY, 2003).
Sweet corn breeding programs target quality characteristics such as flavor, sweetness, creamy texture, appearance, aroma and low starch content.Specifically, high productivity stands out as the main objective of sweet corn breeding programs (LERTRAT; PULAM, 2007).In Brazil, some government and private companies are working to develop commercial cultivars adapted to tropical conditions.However, the number of cultivars recommended for the Brazilian market is still small in relation to the growing demand.Currently, 58 cultivars are registered in Brazil, although there is still the possibility for research and development of new cultivars (BRASIL, 2016).In order to be successful in breeding programs, it is essential to know a priori the available lines of behavior, their performance per se and their hybrid combinations.In this context, the diallel crosses system is an efficient genetic design to obtain information on the genetic control of quantitative traits (JINKS; HAYMAN, 1953).The most common methodologies of diallel analysis used are those proposed by Griffing (1956), which estimates the general and specific combining ability and provides information on the predominance of genetic action; the method proposed by Gardner and Eberhart (1966), which evaluates the effects of varieties and varietal heterosis; and the method proposed by Hayman (1954), which provides information on the basic inheritance mechanisms, genetic values and selection limits.
Thus, the aim was to investigate the general and specific combining abilities based on diallel crosses of sweet and supersweet corn and their F1's in different environments and evaluate the quality and yield traits.

MATERIAL AND METHODS
Six inbred lines of sweet (su 1 ) and eight of supersweet corn (sh 2 ) in S6 were obtained from successive self-crosses.The descriptions of the inbred lines are shown in Table 1.
The joint analyses of variance for each trait were performed according to the randomized complete block design with three replications.For the sweet group (su 1 ), 21 F 1 hybrids were evaluated in five environments: E1-Maringá, E2-Iguatemi, E3-Cidade Gaúcha, E4-Sabáudia, and E5-Umuarama.The variety Doce Cristal was used as check, and for the supersweet group (sh 2 ), 28 F 1 hybrids were evaluated in six environments: E1-Maringá, E2-Iguatemi, E3-Cidade Gaúcha, E4-Sabáudia, E5-Umuarama and E6-Cascavel (Table 2).The hybrid Tropical Plus ® (Syngenta) was used as check, because it is the most planted sweet corn cultivar in the country, and is considered a standard for the crop by breeders.For sweet and supersweet corn, the hybrids were arranged in randomized complete block design with three replications and were evaluated in the 2013-2014 crop season at each of the five and six environments, respectively.The plot size consisted of double rows with 5 m long, and 0.9 m spacing between rows, totaling a useful area of 9 m 2 .Fertilizer and field management practices recommended for optimum maize production were used in each environment.
In each plot, ears had the husks removed and those more than 15 cm in length, greater than 3 cm in diameter and free of pests and diseases were weighed in order to determine the marketable grain yield (GY) without the husk.Grain yield (GY) adjusted to 70% moisture was computed from grain weight in kg ha -1 .Total soluble solid (TSS, o Brix) contents were determined using a digital refractometer (PAL -1, Atago Co., Ltd., Tokyo, Japan).
Bartlett's test with χ 2 distribution (p<0.05) was applied to verify the homogeneity of residual variances and thus performed the joint analysis of environments.
Diallel analysis of tropical and temperate sweet and supersweet corn inbred lines  The model of the joint analysis studied was as follows: where: Y ijk is the observation of the k th block, evaluated in the i th genotype and in the j th environment; µ is the overall mean; B (j)k is the effect of the block k within environment j; G i is the effect of the genotype i; E j effect of the environment j; GE ij is the effect of interaction between genotype i and the environment j; ε ijk is the experimental error associated to the ijk observation.
The joint diallel analysis for Griffing's method 4 (GRIFFING, 1956) is illustrated in Table 3 and represented by the following model: where: Y ijk is the overall mean observed between the hybrid combination of the i th i nb r ed l in e a nd t he j th inbred line (i ≠ j) into the k th environment; µ is the overall mean; e is the environment effect; g i is the general combining ability effect associated to the i th inbred line; g j is the general combining ability effect associated to the j th inbred line; S ij is the specific combining ability effect between the inbred lines i and j; (eg) ik e (eg) jk is the general combining ability effect of the inbred lines with environments; (es) ijk is the specific combining ability effect between the i th and j th inbred lines with environments, and Ē ijk is the mean experimental error.The degrees of freedom of the genotype × environment interaction (G × E) were partitioned into the effects of GCA × E and SCA × E, according to Vencovsky and Barriga (1992)  Phenotypic, genotypic and environmental correlation between total soluble solids and grain yield for sweet and supersweet corn was implemented into the software GENES (CRUZ, 2013).

RESULTS AND DISCUSSION
In sweet and supersweet corn, for TSS and GY, Bartlett's test showed homogeneity of residual mean squares among the individual analyses (data not shown).Thereby, joint analysis of variance was conducted for sweet and supersweet corn across the five environments and six environments, respectively.
In sweet and supersweet corn, the average environmental coefficient of variation showed values of 8.00 and 8.10% for TSS, and 8.70 and 7.10% for GY, respectively (Table 3).These mean values were considered low when compared with Assunção et al. (2010) and Kwiatkowski, Clemente and Scapim (2011), suggesting the reliability of our results.
For GY and TSS, statistical differences (p<0.05) were observed among the sweet and supersweet inbreds and hybrids.Furthermore, for TSS and GY, significant (p<0.05)effects were observed for the G × E mean squares, except for the GY of supersweet corn (Table 3).The significant differences detected for both sweet and supersweet corn among the inbreds and hybrids for TSS and GY indicated that there was adequate genetic variability among the inbreds and hybrids to allow satisfactory progress from selection for the improvement of traits.These results were consistent with the findings of Allam et al. (2016), Djemel et al. (2015), Ordás et al. (2010), and Revilla et al. (2010), and represent a relevant report, since there is a need for lines to be adapted to different environments to obtain superior hybrids.
In sweet corn, both GCA and SCA effects for TSS and GY were estimated by environment because their interaction was significant (p<0.05)(Tables 4 and  5).The lines SC6 and SC7 showed the higher and more positive estimates of GCA (ĝ i ) across the five evaluated environments for TSS, suggesting that both inbred lines could generate promising hybrid combinations.For GY, the lines SC1 and SC2 showed higher and more positive estimates of GCA (ĝ i ) across the five evaluated environments.Hence, these inbred lines could contribute positively to their crossing, suggesting that higher and positive estimates of GCA could represent favorable allele frequencies when compared to contrasting lines in the diallel result.
In supersweet corn, for GY, GCA and SCA effects were estimated based on environment general mean, on the other hand, for TSS, GCA and SCA effects were estimated by environment, since GCA and SCA by environments were significant (p<0.05)(Tables 6  and 7).For GY, the lines SSC4 and SSC6 showed the higher and positive estimates of GCA (ĝ i ) across the mean of the six evaluated environments.For TSS, the lines SSC7 and SSC8 showed the best performance at the six evaluated environments revealing positive estimates of ĝ i .Consequently, these inbreds contribute with additive genetic effect and could increase the  contents of TSS.In supersweet corn, the inbred line SSC4 showed positive estimates of ĝ i for TSS in five of the six evaluated environments and positive average ĝ i across environments for GY.However, the GY estimate showed low values compared to the other inbred lines (Table 6).In sweet and supersweet corn breeding programs, the identification of inbred lines with high favorable genes for GY and grain quality traits such as TSS is highly desirable.However, in the present study, no inbred lines showed this condition.Similar results were showed by Ha (1999) and Kumari, Gadag and Jha (2007) who mentioned the difficulty in finding hybrids showing high GY and TSS.
In sweet corn, as mentioned, the GCA and SCA effects were estimated by environment.Consequently, our results indicated that hybrid vigor or positive heterosis is a major factor in the superior performance of F 1 s in terms of TSS.This confirms similar findings in sweet corn (ASSUNÇÃO et al., 2010;KHANDURI et al., 2010;SOLOMON;MARTIN;ZEPPA, 2012).Thus, for TSS, the hybrids '4 × 6' and '1 × 7' showed the superior Ŝ ij estimates across all the evaluated environments.For GY, the hybrid '5 × 7' was the most relevant because it showed the best Ŝ ij estimates (across environment means).Additionally, the hybrids '1 × 4', '3 × 4', and '1 × 7' showed positive Ŝ ij values across all environments.In fact, the simultaneous selection of hybrids based on both traits indicated that the hybrid '1 × 7' was the most promising.
In supersweet corn, the hybrids '2 × 8', '3 × 6', and '5 × 6' should be selected as the most promising for both traits, because they presented positive and high Ŝ ij estimates across the six environments.These hybrids have at least one of the parents with high GCA (inbred line SSC8 for TSS and SSC6 for GY).
In sweet corn, the hybrids '1 × 7' and '4 × 6' showed the highest percentages of TSS, surpassing the Doce Cristal check.Specifically, the hybrid '4 × 6' demonstrated the best performance for TSS across all evaluated environments, with values ranging from 27 (Environment E4) to 32 °Brix (Environment E5).These TSS values were higher than those reported by Pinho et al. (2010).For GY, the hybrids did not show a similar pattern of behavior.Consequently, none of the sweet corn hybrids could be considered as superior for TSS and GY, simultaneously.For supersweet corn, the hybrids '2 × 4', '2 × 8', and '3 × 8' showed the best performances for TSS across all the evaluated environments, with similar or superior behavior than that of the Tropical Plus ® check.
The genotypic, phenotypic and environmental correlation for TSS and GY, in both sweet and supersweet corn, was not significant in most of the evaluated environments.Only, in supersweet corn in environments E1, E2, E3 and E4, positive and significant phenotypic correlation was observed for TSS and GY (Table 8).On the other hand, in supersweet corn, non-significant genotypic and environmental correlations were observed between these traits in the same environments.There are very few studies correlating GY with TSS in sweet and supersweet corn, but our results may indicate that there is no possibility for indirect selection in sweet and supersweet corn breeding programs, because genetic gain in TSS through indirect selection of GY will not be possible.

CONCLUSIONS
1.For TSS, the inbred lines SC6 and SC7 of sweet corn, and SSC7 and SSC8 of supersweet corn, showed superior GCA effects.For GY, the inbred lines SC1 and SC2 of sweet corn, and SSC4 and SSC6 of supersweet corn, showed higher GCA effects; 2. For TSS in sweet corn, the hybrids '4 × 6' and '1 × 7' showed the most favorable Ŝ ij estimates across all the evaluated environments, and for GY, the hybrid '5 × 7' was the most relevant because it showed the most favorable Ŝ ij estimates.In supersweet corn, the hybrids '2 × 8', '3 × 6', and '5 × 6' should be selected as the most promising for both traits (TSS and GY), because they showed positive and high Ŝ ij estimates across the six environments.

Table 2 -
Geographic location, altitude and climate classification of the environments

Table 1 -
Description of the seven inbred lines of sweet corn (su 1 ) and eight inbred lines of supersweet corn (sh 2 ) used at the diallel crosses (Table3).

Table 3 -
Summary of analysis of variance for total soluble solids (TSS) and grain yield (GY), conducted at different environments in 2013/2014 crop season for sweet and supersweet corn * and ns = significant and non-significant at 5% probability according to F-test; DF = degrees of freedom; CV (%) = mean coefficient of environmental variation

Table 8 -
Genotypic, phenotypic and environmental correlation for total soluble solids and grain yield in sweet and supersweet corn by environments * = significant at 5% of probability by t-test; ns = not significant by t-test