EVALUATION OF CORN HYBRIDS UNDER CONTRASTING WATER AVAILABILITY CONDITIONS

The current study aimed to assess adaptability and stability of corn hybrids regarding grain yields when sown in three different seasons in the Brazilian cerrado. The research was carried out at the Experimental Farm of the IF Goiano, campus in Ceres GO. Pre-sowing fertilization was made with 20 kg ha of nitrogen, 150 kg ha of phosphorus and 80 kg ha of potassium (04-30-16 formulation). Data were analyzed in a random block experimental design. We assessed tem corn hybrids (Truck, Fórmula, P30F53, P3646H, P30F35H, AGN 30A77H, AGN 30A37H, AG 8088PRO, DKB 390 and DKB Bi9440) in three sowing seasons (Nov. 18, 2011; Jan. 31, 2012 and Feb. 20, 2012) with three replications. Harvests were held on Apr. 4, 2012; Jun. 10, 2012 and Jul. 1, 2012. Each hybrid were assessed on yielding; and a graphical analysis was made to contrast hybrids and sowing seasons regarding stability and adaptability. Results showed a significant effect of the interaction between hybrids and sowing times. Therefore, we may state that the best hybrid in an environment would not be necessarily good in another. Thereby, we can infer that drought and corn genetic variability have diverse behaviors in each season. Over the two-dimensional graphics generated by GGE Biplot method, we observed an increased adaptability of AGN 30A77H when sown on Nov. 18, 2011 and on Jan. 31, 2012, followed by Fórmula; however, for P30F35H, it was on Feb. 20, 2012.


INTRODUCTION
Corn (Zea mays L.) is widely adapted and can be grown from north to south of Brazil, in every month of the year.This crop is the most used in rotation and succession systems with other crops, with so prominent role in national agriculture.However, until nowadays, corn has been grown under poor technological levels, and using outdated spacing and fertilization recommendations (MENDES et al., 2012).Compared to other countries, average corn yield in Brazil is low.The United States and France have the highest productivity ranging from eight to nine tons ha -1 (USDA, 2014).In Brazil, annual corn yield for the 2014/ 2015 season was of 5382 kg ha -1 (CONAB, 2016).
Grain yield potential of a crop depends on genetic and management factors as well as favorable environmental conditions.Variations in air temperature, radiation and water availability influence plant phenology, growth and development.Thus, as a genotype-environment interaction, grain yield potential can be maximized by choosing a proper time for sowing without burdening significantly the production costs (TOLLENAAR; LEE, 2002).
Water deficit can lead to marked declines in plant performance, depending on its intensity and in which phenological phase it occurs.In this sense, searching for varieties of plants that are tolerant to drought stress is still the best alternative to agricultural production maintenance, seeking for better crop performance and stability (SOUZA et al., 2015).
In accordance with Forsthofer et al. (2006), evaluations of corn grain yield potentials at different management levels and in various sowing times assists in identifying the limiting environmental factors in each season.Based on this knowledge, management strategies can be devised, enabling the adoption of feasible recommendations to minimize or even overcome weaknesses.By the definition of proper managements and suitable sowing times, farmers can optimize the use of existing resources on farm, maximizing gross income, besides protecting the environment.
With new improved cultivars being launched into the market, new management strategies are required to increase corn yields, such measures would be to narrow spacing between rows and consequent increase in plant stand.Nevertheless, one of the major emerging issues is related to inconsistent behavior of corn hybrids against environmental changes, which is expressed by the interaction between genotypes and environments.This interaction has a key role when recommending a new cultivar; thus, effects should be mitigated by identifying cultivars with higher phenotypic stability (CARVALHO et al., 2002).Hence, researches have focused on improved cultivar adaptations to soil and climate conditions at diverse regions (LOPES et al., 2007;SOUZA et al., 2015).
The expression of plant traits is linked to genetic control, the environment and the interaction between both aspects.The responses of genotypes, under different environmental conditions, reduce correlations between phenotypic and genotypic values, hindering selection and recommendation of adapted and stable genotypes.There are several theories to evaluate plant adaptability and stability, among them the GGE biplot model (genotype main effects + genotype environment interaction), which considers the main genotype effect plus the interaction genotype x environment (YAN; HOLLAND 2010; SILVA; BENIN, 2012).
Given the above, the current study aimed at assessing adaptability and stability of corn hybrids grown in three sowing seasons in the city of Ceres-Go, Brazil.

MATERIAL AND METHODS
Experiments were carried at outbuildings of an Experimental Farm in the IF Goiano -Campus Ceres, which is located in the city of Ceres, Goiás state.The area lies at 15º 21' 02'' south latitude, 49º 35' 36'' west longitude and at altitude of 564 m, in a land under no-tillage system.Rainfall distribution data are shown in Figure 1.
Four planting rows were sown along the experiment edges to provide a border effect.Each plot consisted of two 5-m rows, which were harvested disregarding 0.50 m edges on boh sides.The harvests were held on April 04, 2012; June 10, 2012 and July 1, 2012; respectively, for the three sowing dates.
After harvesting, ears were dehusked and threshed, using a manual thresher.Then, kernels were weighed on a digital scale and the moisture of each plot was adjusted to 13%, calculating thus corn yields in kg ha -1 .
The grain yield data underwent variance analysis (individual and joint) and the means were compared by Scott-Knott testing at 5% significance level.We also performed a graphical analysis comparing the hybrids and environments to assess each hybrid stability and adaptability, using the GGE Biplot method (FRUTOS et al. 2013).The statistical analysis were accomplished using the R software (R DEVELOPMENT CORE TEAM, 2014) with "easyanova" package (ARNHOLD, 2013).

RESULTS AND DISCUSSION
The three experiments were jointly evaluated (three sowing dates), since the relationship between the highest and lowest crop residue of the individual analyzes was 2.77 (Table 2).For Cruz et al. ( 2004), joint analysis of experiments can be performed when the ratio highest/ lowest crop residue does not exceed 7: 1.
The variation coefficient of joint analysis was 14%.In accordance with Scapim et al. (1995), variation coefficients between 10 to 22% are considered average values.However, there was greater variations in experiments where water deficit was higher, with values ranging of 9, 16 and 23% for the sowing dates of Nov. 18, 2011;Jan. 31, 2012 andFeb. 20, 2012,  Table 2. Overview of the variance analysis for grain yield (kg ha -1 ) of ten corn hybrids under three contrasting seasons regarding water availability in the north cerrado of Goiás state, in Brazil.
The significant interaction between hybrids and seasons suggested that hybrids performed inconsistently across the three periods, i.e. there were changes in the means of corn hybrids (Table 2).Accordingly, these hybrids were individually evaluated at each sowing time, since their yield was variable (Table 3), indicating that each genotype has a distinct performance at each season.This indicates that the recommendation of a certain hybrid determined for different sowing dates in a generalized way can be misleading; therefore, regional results should also be considered, otherwise cultivars of most stable productive behavior have to be selected to face the environmental variations (OLIVEIRA et al., 2004).According to Souza et al. (2015), hybrids feature different behavior as they are subjected to varied water deficits due to the genetic variability among cultivars.
In the statistical breakdown of hybrids into the environments, means of AGN 30A77H were superior on Nov. 18, 2011 and Jan.31, 2012 (Table 3).However, on Feb. 20, 2012, when occurred the highest drought, the hybrids showed no statistically significant difference.
Table 3. Grain yield averages (kg ha -1 ) for ten corn hybrids under three contrasting seasons regarding water availability in the north cerrado of Goiás state, Brazil.
1 means followed by the same lowercase letter within the column and uppercase within line are statistically similar by the Scott-Knott's test.
Sowing on Nov. 18, 2011 reached most favorable results likely because water was properly supplied meeting crop requirements (Figure 1).Similarly, we can infer that low yields on Feb. 20, 2012 are related to poor water availability, which was one of the major yield-reduction factors (Figure 1).Moreover, another factor that may be related to such low yields in this season is temperature (Figure 1), which have reduced and may have generated losses on productive potential of In this research, soil type was the same for the three sowing dates, and plots were allocated side by side.As a result, we can infer that climatic factors, particularly drought and temperature, in addition to genetic differences of the hybrids were the parameters influencing crop performance (Table 3).Figure 2 illustrates the superiority of AGN 30A77H on Nov. 18, 2011, reaching a yield of 10,959 kg ha -1 (Table 3), and on Jan.31, 2012, compared to the other seasons.In contrast, sowing on Feb. 20, 2012 may be considered unfavorable due to drought, as shown in Figure 1, resulting in poor performance of all hybrids.(Nov. 18, 2011), E2 (Jan. 31, 2012) and E3 (Feb. 20, 2012)].
The hybrid AGN 30A37H (g7) showed the best stability as it is closest to the center of the biplot (SILVA, BENIN, 2012), while the DKB bi9440 (g10) was the least stable, as shown in Figures 2,3,4 and 5. -

CONCLUSIONS
The hybrid AGN 30A77H is best suited for summer sowing (Nov. 18, 2011) and interim-harvest in the end of January (Jan. 31, 2012).

Figure 1 .
Figure 1.Averages of rainfall and temperature throughout the experimental period.Source: weather station from the IF Goiano Campus Ceres.

Table 1 .
Characteristics of the assessed hybrids. respectively.