Diallel analysis of local sweet corn varieties for grain chemical quality

Local varieties of sweet corn from the far west of Santa Catarina, southern Brazil, show outstanding potential for several important agronomic traits aimed at sweet corn genetic breeding. However, there is no data in the literature on the general and specific combining ability of these varieties for the chemical quality of the grains and the interactions of the general and specific combining ability with the environment. The percentage of total soluble sugars, starch content, and the relationship between sugars and starch of the grains were evaluated in experiments designed in complete randomized blocks, with two replications, in two environments of this region. The splitting of the diallel analysis into general and specific combining ability indicated a predominance of non-additive effects for all evaluated traits and showed non-significant effects (p ≤ 0.05) of the interactions of the general and specific combining ability x environment. Varieties 2255A and 2276A presented a higher concentration of favorable alleles for increasing the chemical quality of grains. The intervarietal hybrids F1's 2255A x 319A, 2255A x 2029A, 2255A x 2276A, Cubano x 2276A, 2276A x 2029A and 2255A x 741B stood out for interpopulation improvement. The biparental compounds derived from the combinations 2255A x 319A and 2255 x 2276A, the triple compound 2255A x 2276A x 2029A, and the quadruple compounds 2255A x 741B x 2276A x 319A and 2255A x 2276A x 2029A x 319A are the most suitable for the formation of composite populations followed by a cyclic process of recurrent selection, aiming to increase the chemical quality of the grains. model The quantification of total s oluble was performed using the glucose standard curve (5; 10; 25; 50; 75; 100 mg/mL; R² = 0.99; y = 0.0081x). as milligrams of total soluble sugars per gram of dry mass. The quantification of starch content was performed using the residue from the total soluble sugar extraction, following the method by McCready, Guggolz and Owens (1950); 2 mL of 30% Perchloric Acid were added and the extract was centrifuged (4000 rpm for 10 minutes). The supernatant was collected, and the residue was again extracted with 2 mL of 30% Perchloric Acid and centrifuged (4000 rpm for 10 minutes). 80 μL of the supernatants were diluted in 3920 μL of MCW (adjusted according to the sample). 1 mL aliquots of the extract, plus 2 mL of 0.2% Anthrone solution, were vortexed and heated in a water bath at 100 °C for three minutes. With the samples cooled, 300 μL were transferred to a microplate for reading the absorbance (620 ηm), in a microplate reader, model Spectramax Paradigm. Starch quantification was performed using the standard starch curve (5; 10; 25; 50; 75; 100 mg/mL; R² = 0.98; y = 0.0065x). Results were expressed as milligrams of starch per gram of dry mass. The data were submitted to individual preliminary examinations for normality and homogeneity of variances. The diallel analysis was performed according to Model II by Griffing (1956), in which the parents and the F1's hybrids are included in the analysis, without considering the reciprocal hybrids. The following random statistical- mathematical model was Yijkl = μ + gi + gj + sij + lz + lgiz + lgjz + lsijz + bk(z) +  ijkz being: Yijkz = average value to hybrid combination (i ≠j) or parent (i=j); μ = overall average of treatments; gi e gj = effect of


Introduction
Sweet corn is a special type of corn intended exclusively for human consumption (Pereira Filho & Teixeira, 2016). In Brazil, the cultivated area is not very representative (40 thousand hectares) but expanding (Teixeira et al., 2013).
Sweet corn differs from other types of corn due to the presence of at least one of the mutant genes that affects the expression of enzymes involved in starch anabolism. Thus, they cause an increase in the concentration of sugars in the endosperm of the grains and wrinkling when they reach physiological maturity (Boyer & Shannon, 1984). Among the mutated genes best known for conferring this trait are: sugary1 (su1); dull (du) e amilose-extender (ae), shrunken-2 (sh2); brittle1 (bt1); sugary enhancer (se); brittle-2 (bt2) e waxy (wx) (Boyer & Shannon, 1984).
Although there are several mutant genes, the genetic basis of sweet corn is narrow. It is believed that in the world there are only 300 open-pollinated varieties of this type of corn. In Brazil, the genetic base is even smaller, as only 20 accessions are kept in the Maize Active Gene Bank (Maize BAG) of the Brazilian Agricultural Research Corporation (Embrapa), most of them imported or derived from breeding programs (Teixeira et al., 2019). There are few commercial sweet corn cultivars in the Brazilian seed market, which restricts the expansion of cultivation in Brazil .
However, an important genetic pool of sweet corn is conserved in the diversity microcenter of Zea mays L. of the far west region of the State of Santa Catarina (FWSC), in southern Brazil (Costa et al., 2016). Thirteen genotypes with wrinkled grains were identified in this region (Souza et al., 2020). Of these, seven have the sugary1 gene and one has the shrunken2 gene for the sweet phenotype (Souza et al., 2021a). The FWSC sugary1 local sweet corn varieties show genetic diversity between and within populations, excellent performance in several desirable agronomic traits and are promising to be used in genetic breeding programs (Souza et al., 2021b). The temperature amplitude due to the different altitudes in FWSC influences the agronomic performance of these varieties (Souza et al, 2021b). However, there is no information on the performance of these varieties regarding the chemical quality of the grains, especially for the percentage of soluble sugars and starch in the dry matter of the grains, when cultivated at different altitudes, nor on the interactions of the general and specific combining ability with environment for these characteristics.
The diallel analysis allows inferences about parameters useful in breeding (Cruz & Regazzi, 2001). The diallel analysis methodology proposed by Griffing (1956) allows obtaining estimates of the effects of general and specific combining ability and inferring the type of predominant genetic action in character control (Hallauer et al., 2010). The general and specific combining ability studies have been frequently used in the genetic breeding of sweet corn (Hossain & Lakhera, 2010;Khanduri et al., 2008;Sadaiah et al., 2013;Souza, 2019;Suzukawa et al., 2018;Teixeira et al., 2013). Souza (2019) used diallel analysis to study the genetic basis of phenological, morphological and agronomic characters of the FWSC sugary1 sweet corn local varieties and proved the existence of variability resulting from the action of additive effects for most of the characters evaluated. However, there is no information about the combining ability and the predominant genetic effects in the determination of the characters related to the chemical quality of the grains of these varieties. Therefore, the present work aimed at evaluating the combining ability of local varieties of sugary1 sweet corn for total soluble sugars, starch content, and relationship sugars/starch, aiming to identify the type of predominant gene action in the determination of these related characters and to define breeding strategies for sweet corn in the FWSC.

Methodology
Samples of grains in milky grain stage of 21 genotypes were obtained, being: six varieties of sweet corn sugary1 (five local varieties from FWSC and the access Cubano from Embrapa's Maize BAG) and 15 intervarietal hybrids F1's resulting from the crossing between these varieties, following the complete diallel scheme.
The local varieties evaluated in the present study are part of the total set of 13 sweet corn varieties conserved in situ-on farm at the FWSC, which were collected by members of the Nucleus of Studies in Agrobiodiversity (NEABio) of the Federal University of Santa Catarina (UFSC), in farms of the municipalities of Anchieta and Guaraciaba, between 2013 and 2016 (Vidal et al., 2020). These varieties were characterized for some morphological and agronomic characters and were stored in the UFSC's Maize BAG (Souza 2019;Souza et al., 2021a). The five local varieties selected to be part of this research represent the genetic diversity of sweet corn conserved on farm in the FWSC (with a 90% confidence level) for the following quantitative grain characteristics: ear weight; ear length; grain volume; and grain weight. This result was obtained using the formula: n = t(0,05;GL) 2 . S² ȳ−µ , being S²: population variance; μ: population average; e ȳ: sample average (Cochran, 1977). The result was adjusted for a finite population, using the formula: n ′ = n 1+ N , where N: finite population size (Bartlett, Kotrlik & Higuins, 2001).
The 21 genotypes were evaluated in two experiments carried out at FWSC, in the municipalities of Guaraciaba (624 m of altitude) and Anchieta (717 m of altitude). The cultivation sites represent, in terms of altitude, the conditions under which local sweet corn varieties are grown in that micro-region, given that the cultivation altitude of the varieties varies from 515 to 833 m (Souza, 2019).
The experimental design was complete randomized blocks, containing two replications and plots represented by two rows of four linear meters in length, at 1.0 m spacing between rows and under a plant density of 50.000 plants ha -1 after thinning.
The useful plot area of 2.0 m² was constituted by the central part of the two rows and from which the grain sample was obtained from five random ears of each genotype, manually pollinated, and harvested on the 21st day after manual pollination.
The grain samples were freeze-dried, fragmented, and submitted to the procedures of quantification of total soluble sugars and starch in triplicates.
The extraction of total soluble sugars was performed according to the methodology proposed by Shannon (1968). 2 mL of the MCW solution (methanol: chloroform: water) (12:5:3, v/v) were added to a sample of 50 mg of dry mass. The solution was centrifuged (4000 rpm for 10 minutes) and then the supernatant was collected. The residue was diluted again with 2 mL of the MCW solution, centrifuged (4000 rpm for 10 minutes) and the supernatant removed. 80 μL of the supernatants were diluted in 3920 μL of MCW (adjusted according to the sample). Subsequently, 1 mL of chloroform and 1.5 mL of water were added, and the extract was centrifuged again (4000 rpm for 5 minutes). After centrifugation, the extract formed two phases, the upper phase being collected and used for sugar quantification, according to the method of Umbreit and Burris (1964). 1 mL aliquots of the extract plus 2 mL of 0.2% Anthrone solution (200 mg of Anthrone in 100 mL of concentrated Sulfuric Acid) were vortexed and heated in a water bath at 100°C for three minutes. With the samples cooled, 300 μL were transferred to a microplate for reading the absorbance (620 ηm), in a microplate reader, model Spectramax Paradigm. The quantification of total soluble sugars was performed using the glucose standard curve (5; 10; 25; 50; 75; 100 mg/mL; R² = 0.99; y = 0.0081x). Results were expressed as milligrams of total soluble sugars per gram of dry mass.
The quantification of starch content was performed using the residue from the total soluble sugar extraction, following the method by McCready, Guggolz and Owens (1950); 2 mL of 30% Perchloric Acid were added and the extract was centrifuged (4000 rpm for 10 minutes). The supernatant was collected, and the residue was again extracted with 2 mL of 30% Perchloric Acid and centrifuged (4000 rpm for 10 minutes). 80 μL of the supernatants were diluted in 3920 μL of MCW (adjusted according to the sample). 1 mL aliquots of the extract, plus 2 mL of 0.2% Anthrone solution, were vortexed and heated in a water bath at 100 °C for three minutes. With the samples cooled, 300 μL were transferred to a microplate for reading the absorbance (620 ηm), in a microplate reader, model Spectramax Paradigm. Starch quantification was performed using the standard starch curve (5; 10; 25; 50; 75; 100 mg/mL; R² = 0.98; y = 0.0065x). Results were expressed as milligrams of starch per gram of dry mass.
The data were submitted to individual preliminary examinations for normality and homogeneity of variances. The diallel analysis was performed according to Model II by Griffing (1956), in which the parents and the F1's hybrids are included in the analysis, without considering the reciprocal hybrids. The following random statistical-mathematical model was Yijkl = μ + gi + gj + sij + lz + lgiz + lgjz + lsijz + bk(z) + ijkz being: Yijkz = average value to hybrid combination (i ≠j) or parent (i=j); μ = overall average of treatments; gi e gj = effect of the general combining ability of the i-th or j-th parent (i,j = 1,2 ...6); sij = effect of specific combining ability for crosses between parents of order i and j; lz = environment effect (1, 2); lgiz and lgjz = effect of the interaction of the general combining ability of parents i and j with the z-th environment; asijz = effect of the interaction of the specific combining ability of parents i and j with the z-th environment; bk(z) = effect of blocks within environments, and eijkz = average experimental error associated with order observation ijkz.  , where: Yij: mean of the parent, when i = j, or of the hybrid, when i ≠ j, of the selected variable.
All statistical and genetic analyzes were performed with the aid of the computer software GENES (Cruz, 2013).

Results and Discussion
In the unfolding of the effects of treatments on GCA and SCA of the diallel analysis, significant differences (p ≤ 0.05) were detected in the estimates of the effects of SCA for all characters and in the estimates of the effects of GCA for starch content and relationship sugars/starch. There was no significant effect of the GCA x Environment interaction (CGC x E) and the SCA x Environment interaction (SCA x E) ( Table 1).
The coefficients of variation (CVs) were 9.56% for total soluble sugars, 4.89% for starch content and 11.21% for the relationship sugars/starch (Table 1). According to the classification by Pimentel Gomes (1978), CVs were considered low for total soluble sugars and starch content (CV < 10%) and medium for the relationship sugars/starch (CV < 20%). These values were lower than those reported by Scapim et al. (1996), who reported CVs of 16.47% and 27.27% when evaluating the content of total sugars and reducing sugars in sweet corn grains, respectively. ¹ Percentage in grain dry matter. ² Variance components associated with the general ability (ĝi) and the specific ability (ŝij) of combination; ns: not significant; ** and * significant at 1 % and 5 %, respectively, by the F Test. Source: Authors.
With the exception of total soluble sugars, significant effects were detected for the general combining ability, indicating the differences between genotypes regarding the additive effects for starch and for the sugar/starch ratio (Table 1). Estimates of variance components associated with GCA effects (σ²ĝi), SCA (σ²ŝi), and the ratio between them (σ²ŝi/σ²ĝi) of 4.32 for total soluble sugars, 2.78 for starch content, and 2.17 for relationship sugars/starch, indicate that non-additive gene effects are the most important for these characters, mainly due to the starch content. When comparing the relative contribution of the components of σ²ĝi and σ²ŝi in the character control, for the characters that presented significance, the contribution was greater for the non-additive part. These results are in line with those reported in several studies carried out with sweet corn (Khanduri et al., 2010;Kumari et al., 2008;Sadaiah et al., 2013). When studying the genetic effects involved in the concentration of total sugars in sweet corn grains, Kumari et al. (2008) and Sadaiah et al. (2013) reported the predominance of non-additive gene action. Khanduri et al. (2010) also reported the greater importance of non-additive gene action for the concentration of total sugars, reducing sugars, non-reducing sugar, and grain yield in sweet corn. According to Cruz and Regazzi (2001), the significance of variations attributed to non-additive effects enables interpopulational breeding to obtain hybrids and the significance of variations attributed to additive effects enables the indication of populations to be used in intrapopulational breeding programs.
Although the environments where the experiments were conducted are at different altitudes, in two municipalities in the state of Santa Catarina, the differences between the environments were not enough to detect significance for the interactions G x E, GCA x E and SCA x E, possibly because Anchieta and Guaraciaba are located in the same agroecological micro-region.
The non-significant effects of the GCA x E and SCA x E interaction allowed the estimation of the GCA effects (ĝi) and SCA (ŝij), as well as the prediction of compounds based on the average of the sites.
For the total soluble sugars character, the greatest positive effect of ĝi was presented by the local variety 2276A (2.14%) and the greatest negative effect was presented by the variety Cubano (-2.66%). For the character starch content, the greatest negative effect was presented by 2255A (-3.00%) and the greatest positive effect was presented by Cubano (2.53%). For the relationship sugars/starch, the greatest positive effect of ĝi was presented by 2255A (0.11) and the greatest negative effect was presented again by the variety Cubano (-0.12) ( Table 2). The values obtained indicate the increase (positive) or reduction (negative) of the chemical quality indicators of the grains, according to the cross of diallel analysis. Considering that significant genetic differences between the local varieties associated with the general combining ability were due to starch content, the most promising genotypes on the basis of this parameter would be the ones whose estimated values of ĝi were negative for this trait and positive for the relationship sugars/starch. According to Cruz and Vencovsky (1989), the most favorable hybrid is the one with the highest SCA, in which one of the parents has the highest GCA. Therefore, we discussed the hybrid combinations with the best estimates of ŝij and that involve at least one of the parents that have presented the favorable effect of GCA. The best estimates of ŝij for total soluble sugars and relationship sugars/starch were from the hybrid combinations 2255A x 319A, 2255A x 2029A, 2255A x 2276A, Cubano x 2276A, 2276A x 2029A and 2255A x 741B (Table 3) Other combinations with the parent 2276A and 2255A, which showed positive effects closer to zero, as in the case of the hybrid 741B x 2276A (for total soluble sugars), or negative values, as in the case of the combination 2276A x 319A (for sugars/starch), indicate the same or lower than expected for the GCA.
Based on the set of characteristics evaluated, it is concluded that the hybrids that stood out from the other combinations were 2255A x 319A, 2255A x 2029A, 2255A x 2276A, Cubano x 2276A, 2276A x 2029A and 2255A x 741B. Souza (2019) also highlighted the combinations 2255A x 319A, 2255A x 2276A and 2255A x 741B, due to their potential for productivity (t ha -1 ) of sweet corn.
Mean predicted values for compounds derived from double, triple, and quadruple combinations for three characters of grains from six sweet corn parents in two FWSC environments are shown in  The concentration of carbohydrates (sugars and starch) is one of the most relevant quality parameters of grains for sweet corn, as it is correlated with sensory attributes, such as sweet taste and grain texture. In a sensory analysis study , Oliveira Jr. et al. (2006) showed that consumers prefer corn with the highest concentration of sugars in the samples. For this reason, the sweet corn industry prefers materials with a high concentration of sugars and low concentration of starch (Kwiatkowski & Clemente, 2007). The greatest commercial potential at the moment is the materials of the super-sweet group, which have between 15% and 25% of sugars in the grains (Pereira Filho & Teixeira, 2016). Thus, the combinations of local sweet corn varieties from the FWSC favor for increasing the percentage of sugars and reducing the percentage of starch and are, therefore, the most promising to be used in genetic improvement programs destined for this region of the Santa Catarina State.
In the context of the FWSC and aiming to support the conservation of the genetic variability of these local varieties, it is opportune that the interpopulation improvement strategy is focused on the identification of genotypes that present superiority of the effects of the specific combining ability and that intervarietal hybrids F1's can be recommended for direct use. On the other hand, compounds involving two, three and four parents can allow the formation of new populations with good prospects to be used in breeding programs with high potential variability available for selection, aiming at developing varieties with wide adaptation in different cultivation conditions in this region.

Conclusions
The non-additive genetic effects are more important for the grain chemical quality characteristics, percentage of total soluble sugars, percentage of starch and ratio between sugars and starch in grains.
Varieties 2255A and 2276A have a higher concentration of favorable alleles for increasing the chemical quality of grains.
The biparental compounds derived from the combinations 2255A x 319A and 2255A x 2276A, the triple compounds 2255A x 2276A x 2029A, and the quadruple compounds 2255A x 741B x 2276A x 319A and 2255A x 2276A x 2029A x 319A are the most suitable for the formation of composite populations followed by recurrent selection, aiming to increase the concentration of sugars in the grains.