Genetic parameters and agronomic characterization of elite barley accessions under irrigation in the Cerrado

There is a demand for barley cultivars that are suitable for the malt industry as well as genotypes that are stable and better adapted for irrigation in the Cerrado. This study aimed to estimate the genetic parameters by assessing 69 barley genotypes from different countries, from 2012 to 2014, under irrigation in the Cerrado. Six agronomic characteristics were assessed: grain yield, plumpness kernel, thousand seeds weight, plant height, degree of plant lodging and days to heading. Analysis of variance, cluster test and phenotypic, genotypic and environmental correlations were performed. Significant effects were observed for genotypes, years and the G x E interaction. High values of broadsense heritability (> 86%) were found for all the characteristics, which enabled direct selection. The Colombian accession MCU363PI402112 stood out for its agronomic characteristics. Genotype selection based on the phenotypic evaluations was possible due to their good experimental accuracy and precision. Precocious genotypes with high grain yields and homogeneous grain sizes were selected. Due to the environmental influence on the grain yield, additional studies concerning the components of yield in this environment are necessary to facilitate the selection of more productive genotypes.


Introduction
The search for a germplasm that can be introduced into plant breeding programs is indispensable for developing new cultivars that meet the demands of the productive sector in terms of the agronomic traits of interest. The constant search for selection and recombination strategies to increase the plants' efficiencies and genetic gains is equally important. In general, breeders seek to incorporate those genetic resources that have favorable alleles for the agronomic traits of interest to increase their frequencies in breeding populations.
To aid the barley breeding program based on the selection of superior genotypes that are more adapted to the Cerrado irrigated conditions, the agronomic characterization of accessions from the germplasm bank is of fundamental importance, providing knowledge about the full productive potential and agronomic traits of commercial interest (Amabile, Faleiro, Capettini, & Sayd, 2017).
The different responses of the genotypes according to changes in environmental conditions represent a problem to producers and a great challenge to breeders. It is in the interest of both that plants present stability in both quality and yield both in different locations and over years, in addition to high yields. An alteration of the relative performance of genotypes due to different environments is called the genotype by environment interaction (G x E) (Borém & Miranda, 2005).
The planning and performance of experiments in breeding programs are based on interpretations and estimates of the genetic parameters that indicate the experimental precision and accuracy, the proportion of total variance due to the genetic differences, and the prediction of gains with plant selection (Cruz, Regazzi, & Carneiro, 2004). Experiments on the selection of genotypes for at least two years are necessary to segregate the environmental effects in the response of the genotypes. These effects can be minimized by adding sites and evaluating genotypes in different years.
interactions on the agronomic characteristics assessed. The effect of the G x E interaction is particularly important to show the variations in the agronomic performance of the elite accessions when they are evaluated in different years.
For all the characteristics evaluated, the values of F were above 5.26, which is the value suggested by Steel and Torrie (1980) that will allow one to make a safe statistic inference and to indicate a high selective accuracy. The values of F for the effect of genotypes ranged from 7.6 (Yield) to 50.4 (TSW), thus showing the existence of differences among the means of the accessions. The values of F for the effect of environment ranged from 17.4 for DH to 162.7 for TSW, and the values for the G x E interaction ranged from 5.2 (Yield) to 24.5 (Height) ( Table 1). Table 1. Analysis of variance, mean, and genetic parameters (heritability -H 2 ; coefficients of variation genetic -CGV; environmental -CEV; and relative -CRV) of the characteristics grain yield (Yield), plumpness kernel (PK), thousand seeds weight (TSW), plant height (Height), degree of plant lodging (LOD), and days to heading ( The F values for environment and the G x E interaction were lower than those obtained by Sayd, Amabile, Faleiro, Montalvão, and Coelho (2017) for the same characteristics with 113 genotypes in two locations. Therefore, we inferred that for the characteristics evaluated in the irrigated conditions of Cerrado, the local factor had a greater influence on the phenotypic variation of the genotypes than the year factor. These results are consistent with those of Borém and Miranda (2005), who concluded that the G x E interaction generally has less relevance for monogenic characteristics and more homogeneous sites and perennial species compared to polygenic characteristics, heterogeneous sites and annual species.
According to Robertson (1959), there are two types of G x E interactions: simple and complex. The simple interaction indicates the presence of genotypes adapted to a wide range of environments. The complex interaction indicates the presence of materials adapted to environments. To identify the nature of the interaction, it is necessary to analyze the adaptability and stability in several sites.
High values of heritability indicate that the direct selection can successfully be used under these experimental conditions. Heritability is one of the most important genetic parameters that a breeder uses to assess a metric characteristic and to select subjects that can obtain genetic gains in barley and other crops (Amabile & Faleiro, 2014). Therefore, due to the high computations of H 2 obtained in the experiments, a high selective accuracy is expected, which will maximize the genetic gains for the next selection and recombination cycles. The estimate of heritability (H 2 ) also indicates the reliability with which the phenotypic value represents the genotypic value and determines the proportion of gain obtained by selection (Falconer, 1989). Sayd et al. (2017) obtained wide heritability values of the same magnitude in 113 elite genotypes of barley in two sites under irrigated conditions in the Cerrado. Addisu and Shumet (2015), evaluating 36 genotypes from Ethiopia, reported high values of H 2 for the same quantitative traits in barley. Amabile and Faleiro (2014) conducted an extensive review on heritability in barley and concluded that for most agronomic traits, the H 2 values are generally high (above 80%). The trait grain yield, due to its quantitative profile, is reported as the experimental condition with the lowest H 2 values (Delogu et al., 1988;Nadziak, Kudła, & Małysa, 1994;Pesaraklu, Soltanloo, Ramezanpour, Kalate, & Nasrollah, 2016), although depending on the population or experimental conditions, the trait grain yield can present high values (Jalata, Ayana, & Zeleke, 2011;Addisu & Shumet, 2015).
In addition to the heritability, another important parameter that indicates experimental accuracy is the coefficient of genotypic variation (CGV), which quantifies the magnitude of the genetic variability present in the genotypes evaluated in different treatments (Resende, 2002). It is important to consider the CGV/CEV ratio, as represented by the CRV (coefficient of relative variation). When the value of the CGV is superior to the value of the CEV, the genotypic contribution is higher than the environmental effect in the phenotypic expression. Thus, the conditions of selection for such characteristics are favored for breeding. Table 1 shows that the values of CRV were higher than those for all characteristics, except Yield. This value was expected due to the higher environmental variance in the composition of the phenotypic variance of Yield. An alternative for an efficient selection of genotypes for Yield would be the indirect selection through characteristics correlated to productivity. Previous experiments conducted in the Cerrado under irrigation have demonstrated favorable selection conditions Amabile et al., 2015). In trials conducted under rainfed conditions, the CVR values were less than one unit for agronomic traits in barley (Adissu & Shumet, 2015;Yadav, Singh, Pandey, & Singh, 2015;Ahmadi, Vaezi, & Pour-Aboughadareh, 2016).
Another characteristic that indicates the appropriate environmental control in these experiments is the coefficient of environmental variation (CEV). The estimates of this parameter ranged from 2.34% (DH) to 33.59% (LOD). According to the evaluation criteria of Pimentel-Gomes (1985), there were four characteristics with a low CEV (PK, TSW, Height and DH), one characteristic with a medium CEV (Yield), and another characteristic with a high CEV (LOD). However, the magnitudes of these CEVs are similar to those found by other authors Sayd et al. 2017) and are within the acceptable values in agricultural experimentation.
The main characteristic for barley, as well as for most crops, is Yield. The values for Yield presented the greatest variations in the years 2014 and 2012, with CEVs of 18.51% and 14.7%, respectively, while in 2013, the CEV was 4.84%. The whole CEV presented in Table 1 shows a value of 14.02%. Sayd et al. (2017) obtained a low CEV value (3.97%) from a comparison of 113 elite genotypes of barley in two sites under irrigation in the Cerrado. In this case, there was a better environmental control to make the selection when the genotypes were compared in the same year and at different sites.
The genotypic, phenotypic and environmental correlations between the evaluated characteristics were mostly non-significant and of low magnitude. Generally, pleiotropy and close linkage are the two major reasons for genetic trait correlations and are often confounded at the level of genes (Gardner & Latta, 2007). The highest correlation was observed between TSW and PK, which indicates that the larger the grains were, the higher the weight of the grain mass was. An important absence of correlation was observed for DH x Yield, which shows the potential for obtaining genotypes that are both early and of high yield (Table 2). Weak correlations but with the same signal of this experiment between traits were obtained by Raham (2015). Other studies under irrigated conditions of Cerrado showed correlations similar to those obtained in this study Sayd et al., 2017). Table 2. Genotypic, phenotypic, and environmental correlation coefficients among the characteristics estimated grain yield (Yield), plumpness kernel (PK), thousand seeds weight (TSW), plant height (Height), degree of plant lodging (LOD), and days to heading (DH) in 69  Similar to Yield and as a characteristic influenced by the environment, the characteristics PK and LOD showed high values of CEV (Table 1) compared to those observed by Sayd et al. (2017). These contrasting values indicated the necessity of data-gathering studies concerning genotypes at different sites and, primarily, in different years of growth. These studies are important for obtaining a wider knowledge of elite and promising barley in irrigated cultivation in the Cerrado. For the characteristics TSE, Height and DH, the CEV values were close to those obtained by Sayd et al. (2017) and Sayd, Amabile, Faleiro, and Bellon (2015), who only collected the parameters at different sites.
The agronomic characteristics assessed based on the main demands from the producers and the industry served to verify the stability and potentiality of the genotypes along three years under irrigated conditions in the Cerrado. Among the most productive (means above 5,200 kg ha -1 ) and stable genotypes along the three years of evaluation are three Canadian genotypes (CI 15560 QB 136-4-1, CI 15591 QB 139-1, and CI 15580 QB 136-41), four Colombian genotypes (MCU 3634 PI 402112, MCU 3449 PI 401927, MCU 3870 PI 402348, and CI 10022), one Ethiopian genotype (CI 12918), one American genotype (CI 13683 NUMAR), and the Brazilian control BRS 180, which were all six-row and cream-colored genotypes. The control BRS 180, a cultivar adapted to the conditions of the Cerrado, had the highest value for Yield (7,318 kg ha -1 ) (  (Tables 3, 4, and 5) and could be selected as potential cultivars for the Cerrado. Values of this magnitude indicate the potential of these genotypes as an option for the current cultivars that average 6,000 kg ha -1 in irrigated conditions in the Cerrado . Table 3. Means of the genotypes, characteristics, heritability, and variation coefficients of the estimated characteristics, including grain yield (Yield -kg ha -1 ), plumpness kernel (PK -%), thousand seeds weight (TSW -g), plant height (Height -cm), degree of plant lodging (LOD -%), and days to heading (DH -days), in 69 barley genotypes submitted to the Scott-Knott test at 1%, the estimates of heritability at the level of the mean (h 2 ) and the coefficient of environmental variation (CV) in 2012.  Table 4. Means of the genotypes, characteristics, heritability, and variation coefficients of the estimated characteristics, including grain yield (Yield -kg ha -1 ), plumpness kernel (PK -%), thousand seeds weight (TSW -g), plant height (Height -cm), degree of plant lodging (LOD -%), and days to heading (DH -days), in 69 barley genotypes submitted to the Scott-Knott test at 1%, the estimates of heritability at the level of the mean (H 2 ), and the coefficient of environmental variation (CV) in 2013.  Table 5. Means of the genotypes, characteristics, heritability, and variation coefficients of the estimated characteristics, including grain yield (Yield -kg ha -1 ), plumpness kernel (PK -%), thousand seeds weight (TSW -g), plant height (Height -cm), degree of plant lodging (LOD -%), and days to heading (DH -days), in 69 barley genotypes submitted to the Scott-Knott test at 1%, the estimates of heritability at the level of mean (H 2 ), and the coefficient of environmental variation (CV) in 2014. For the characteristic PK, the value used by the breeding program of Embrapa Cerrados and recommended by the Brazilian Ministry of Agriculture, Livestock, and Supply (MAPA) is >80% (Brasil 1996). The larger the grain is, the higher the proportion of starch and, consequently, the higher the efficiency of sugar usage in beer production. Among the 69 genotypes assessed, 27 were classified above 80% in 2012, and only 12 and five genotypes were classified above 80% in 2013 and 2014, respectively (Tables 3, 4, and 5). The mean PK along the three years had the same performance, with the highest mean observed in 2012 (73. For the characteristic thousand seeds weight (TSW), the Iranian genotypes I 0996 and CI 09961 and the Moroccan CI 09958 had the highest weights, with values typically above 50 g. The genotype with the lowest value was also of Iranian origin (CI 06946), with a weight ranging from 33 to 34 g within the three years. The yearly means were similar, with the highest value reported in 2012 (45.1 g). In 2013, the lowest mean was recorded (41.1 g), and in 2014, the mean was 43.3 g (Tables 3, 4, and 5). A great variation in relation to this trait, with values between 33 and 64 g among the 18 Ethiopian genotypes, was obtained by Adissu and Shumet (2015). Values with great variation were also described by Amabile et al. (2017) and Sayd et al. (2017) (Tables 2 and 4). The highest genotypes were of Canadian and Colombian origin, with values above 80 cm.
In the year 2012, the lowest mean (75.7 cm) was recorded, followed by those recorded in 2013 (77.0 cm), and 2014 (80.1 cm). Values between 70 and 80 cm are considered ideal for the conditions of the Cerrado, thus contributing to the lower lodging of plants . In breeding programs worldwide, genotypes with low size and greater resistance to lodging are sought (Ordon, Ahlemeyer, Werner, Köhler, & Friedt, 2005). Thus, the values of plant height obtained in this study showed that most of the elite accessions evaluated are adequate for the cropping system in the Cerrado.
The German two-row cultivars CARINA PI 371632 and CI 11493 FRUGHERSTE STANKAS and the six-row Colombian genotypes MCU 3832 PI 402310 and MCU 3452 PI 401930 were highly resistant to lodging in all trials (Tables 3, 4, and 5). Among the 69 genotypes evaluated, 28 had LOD values below 30%. The selection of shorter and more lodging-resistant genotypes reduces the production costs. This reduction can be attributed to not using growth reducers for crop management. The characteristic LOD had great variation among the trials due to the climate differences among the years evaluated. In 2013, when there was the highest rainfall, the lowest CEV was observed, and in general, the genotypes presented the highest percentage of lodging. In 2012 and 2014, the LOD values were lower, allowing a higher variation of the genotype performances. It is important to emphasize that the correlation between lodging and Yield was low and not significant (Table 2), as observed by Amabile et al. (2015). This result indicates the feasibility of selecting genotypes with high yield and low lodging levels.
The feasibility of barley in the Cerrado is primarily correlated with the irrigated cropping system. In this environment, it is essential to search for earlier genotypes with heading cycles and, consequently, to select those genotypes with the most efficient use of water. In the irrigated system, the relevance of early genotypes is also due to the economic bias because the irrigated area is cleared as quickly as possible for the next crop. The Colombian genotype MCU 3634 PI 402112 was an area that presented a lower DH, ranging from 51 to 55 days over the three years (Tables 3, 4, and 5). This genotype, in addition to being early, was one of the outstanding genotypes in terms of grain yield, and held potential to be selected for future evaluations and to compose the breeding basis of the breeding program. In addition to this genotype, other Colombian genotypes were highlighted for their precocity.
The means of DH among the years were close, at 57.6 days in 2012, 60.3 days in 2013, and 57.9 days in 2014 (Tables 3, 4, and 5). The evaluated genotypes obtained heading dates of same magnitude as those found by Amabile et al. (2015), who reported a mean of 59 days for 39 elite barley genotypes under irrigation in the Cerrado.
Based on the genetic parameters obtained for the 69 elite barley accessions evaluated under an irrigated system in the Cerrado during three consecutive years of cropping, the experimental precision and accuracy for most agronomic characteristics were considered adequate. In addition, the effect of the G x E interaction was observed for all agronomic characteristics evaluated, which indicated the importance of the evaluation of elite accessions in different years and environments to subsidize the selection of more adapted and stable genotypes in the conditions of the Cerrado.

Conclusion
The Colombian accession MCU 3634 PI 402112 was the main agronomic characteristic evaluated. Moreover, we selected early genotypes with high grain yields and homogeneous grain sizes, which are important characteristics for irrigated systems in the Cerrado and for use in the brewing industry. Therefore, due to the great environmental influence on the grain yield, it is necessary to evaluate the genotypes in different environments and years to promote the selection of the best materials with experimental precision and accuracy.