ESTIMATES OF MEANS COMPONENTS FOR MAIZE GRAIN-FILLING TRAITS

– One of the strategies to increase maize yield is to select lines that present a higher matter accumulation rate in the grains. This work was carried out to identify the population with the most significant potential to obtain this line type. For this, nine commercial hybrids’ F 1 and F 2 generations were evaluated in four environments, involving two years and two seasons each year. The traits evaluated were the number of days to female flowering (NDF) and physiological maturity (NDPM), dry matter accumulation rate (RATE), and grain yield (YLD). For each trait, the contribution of the loci in homozygosity ( m+a ) and heterozygosis (d) was estimated. NDF, NDPM, and RATE predominated an additive effect with a higher m+a estimate. Regarding YLD, the estimate of d was much higher than m+a, indicating the greater importance of dominance. Hybrid 5 is the most promising for obtaining the segregating population because it associates a high m+a , and d estimate that, although it was not among the highest, represented 56.03% of the average. For this hybrid, considering the YLD, the mean of the lines in F∞ will be the highest and associated with sufficient variability in the population to enable successful selection.

Maize yield in Brazil has increased significantly over the last 40 years (Ramalho et al., 2017(Ramalho et al., , 2021. This increase is particularly significant because, in the last 30 years, maize has grown at two seasons per year. In the first season, the sowing extends from September to November, depending on the region, and in the second season, sowing occurs primarily from January to March. This second season is grown under unfavorable conditions due to several factors. The rainfall amount and distribution is the predominant factor in southern Minas and other regions of Brazil (Andrea et al., 2019).
Additionally, the presence of maize plants in the field throughout the entire year has created substantial problems with pests and pathogens that were previously less significant. Despite these issues, the success of this second season has been substantial and currently accounts for more than 50% of the maize production in Brazil (CONAB, 2017).
A considerable research effort has been made to continue the increase in yield.
Breeding programs are notable for many lines acquired and hybrids tested annually, primarily by private companies. Among the traits that affect maize yield those related to the number of days to flowering (NDF), the number of days to physiological maturity (NDPM), and the rate of dry matter accumulation (RATE) in the grains are essential.
Grain filling is divided into three phases.
Initially, there is a relatively short period in which the dry matter accumulation is still incipient. This stage is characterized by rapid cell division that aims to increase the number of cells in the grain (Yadegari & Drews, 2004). In the literature, this period is referred to as the lag phase (Gasura et al., 2014). This phase begins with fertilization, but its end is more challenging to determine. When dry matter accumulation begins, the second period is the linear phase. Ultimately, the duration and rate of dry matter accumulation during this phase determine the grains' final dry weight and yield.
This period begins 7 to 15 days after fertilization and ends when the black layer is formed. This black layer indicates the termination of dry matter accumulation in the grain and corresponds to the end of the grain-filling stage (Gasura et al., 2014). The final period, called the maturation drying phase, corresponds to the loss of water by the grain until the moment of harvest.
One of the alternatives is to seek more information about genetic control. Then, depending on the type of allelic interaction, the breeder can direct his work towards improving the performance of the lines per se or in heterosis.
Different methodologies can be applied to study the genetic control of a trait (Bernardo, 2020;Hallauer et al., 2010;Ramalho et al., 2012a).
These methodologies include the use of variances or means. In the latter case, the estimates of m+a (the contribution of homozygous loci) and d (the contribution of heterozygous loci) have some advantages. Furthermore, due to its feasibility and less associated error genetic control of traits can be used to identify new lines in populations with a higher mean in F∞ generation and more significant variability for selection (Vencovsky, 1987;Ramalho et al., 2012b).
Some studies have estimated m+a and d for grain yield in maize (Souza Sobrinho, 2001;Viana et al., 2009;Ribeiro et al., 2014). Overall, the authors report an expressive dominance effect on the phenotypic expression of this trait. It is crucial to assess whether these estimates behave similarly for traits related to the blooming and grain-filling periods directly associated with yield.
This study aimed to obtain information on genetic control through estimates of the contribution of homozygous loci (m+a) and heterozygous loci (d) of the traits: the number of days for female flowering, the number of days to physiological maturity, the dry matter accumulation rate and grain yield, aiming to identify promising commercial hybrids for extraction of lines, focusing on those traits.

Materials and Methods
The study was conducted in the   (Darroch & Baker, 1990;Borrás et al., 2003;Prado et al., 2013;Gasura et al., 2013Gasura et al., , 2014, and the one adopted in this study was the logistic equation used by Darroch and Baker (1990). The daily rate per plot was obtained by dividing the dry matter at physiological maturity by the number of days to physiological maturity; d) grain yield (YLD). The YLD was obtained by weighing the threshed grains (g plant -1 ) from one of the plot rows corrected to 13% moisture, in which no ear sampling was performed.
The individual variances analysis was realized for each trait, and afterward, the joint variance analysis was carried out (Ramalho et al., 2012a).

Results and Discussion
All sources of variation for the evaluated traits were significant (p≤0.01), except for the F 1 vs. F 2 contrast x environment interaction for NDPM (Table 2). However, it is essential to emphasize that although the hybrid x F 1 vs. F 2 contrast and the hybrid x environment interactions were significant, the focus was on the overall mean because there was relatively good agreement in the comparison between the F 1 and F 2 generations for the different hybrids, the primary aiming of this study. Regarding the significant interaction, the objective was not to study the interaction by the environment but to have more replications of the results to have greater security in the inferences. Although, the issue of hybrid x contrast interaction for different traits will be further discussed in some situations.
The NDF was very similar among the four seasons. The range in the NDF was only three days, i.e., 4.5% of the overall mean (66 days). Although the F test was significant for the NDF, the averages in the NDF between the environments were very similar. The significance was due, probably, to the greater precision with which they evaluated this characteristic.
The mean number of days from flowering to physiological maturity (the mean of the fourseason, nine hybrids, and two generations) was 62 days (Table 3).
About the RATE, it is evident that the daily dry matter accumulation rate for the second season was lower than in the first season (Table   3). On average for the two years, the RATE for the first season was 12.6% higher than the average of the second season.
The effect of the environment on YLD was very similar to the results for RATE. The YLD was consistently higher in the first season.
On average, the yield of the first season (sown in October and November) was 35.4% higher than the second season (sown in January/March).

Results obtained previously in this region
indicate that the delay in sowing maize afterward in October resulted in a reduction of 27 kg ha -1 day -1 ; this was true for sowing until December . Similarly, Ribeiro   (Table 4). The ideal hybrid would combine the earliest flowering and most extended period of dry matter accumulation in the grains.
The hybrids were also classified into five groups for RATE (Table 5). Hybrids 6, 8, and 9 showed the highest mean dry matter accumulation rate. For YLD, the variability was even more remarkable, and the hybrids were classified into seven groups. The highest mean yield was exhibited by hybrid 6 ( Table 5).
Hybrid 6 was also classified in the group with the highest dry matter accumulation rate. Gasura et al. (2014) reported a positive association (r gg' = 0.61) between these two traits, explaining why the same hybrid combines a high dry matter accumulation rate and a high grain yield.
It is important to note that there was a significant difference between the F 1 vs.
F 2 contrast, which is essential for achieving the objectives of this study. Among the traits evaluated, the NDF showed a distinct response, i.e., the mean for the F 2 generation, for all hybrids and environments, was higher than the F 1 generation ( Table 4). The effect of inbreeding depression was similar among the hybrids for RATE and YLD. However, the overall mean inbreeding depression for YLD (36%) was higher For the NDPM, the dominance effect was also less relevant than the additive effect.
In the literature, there are few reports of this phenomenon. However, using variance components, Wang et al. (1999) and Gasura et al. (2013) showed that the dominance variance was more significant than the additive variance.
As expected, the dominance effect was also less critical than the additive effect for the RATE. In the mean of the nine hybrids, m+a contributed almost 83% to the mean performance (Table 5). Similar results were obtained by Gasura et al. (2013).  Table 5. Mean of F 1 and F 2 generations and estimates of m+a, d, and inbreeding depression for the dry matter accumulation rate (RATE; 10 -3 g grain -1 day -1 ) and grain yield (YLD; g plant -1 ). Means were calculated for the four environments (Years/season) of nine hybrids evaluated in Lavras, MG.   hybrid 5 is promising for line selection because this hybrid has an estimate of m+a higher than the others and has a relatively high d magnitude for YLD. Emphasizing, the highest estimate of m+a for hybrid 5 indicates that if it is used in the breeding program, the mean of the lines derived from it will be higher than that of the others.

RATE
The major magnitude of d makes it possible to comparatively infer the number of segregating locos. Therefore, the population derived from this hybrid must have sufficient variability for success in selecting the best lines.

Conclusions
The additive effect predominates for the traits NDF, RATE, and NDPM, as indicated by the higher estimate of m+a.
However, for YLD, the estimate of d was more expressivesignificantly more significant than the estimate of m+a, thus indicating the greater importance of the dominance effect and heterosis.
Considering a breeding program, the population derived from hybrid 5 is the most promising because it associates a high average expectation of the lines to be obtained and sufficient variability, among them, for the success of the selection.