ABSTRACT Maize is importance in the semiarid region due to its forage potential. However, the large number of hybrids on the market hampers the selection of the hybrid by the producer. Given the above, the objective of this paper is to identify genotypes with productive potential for forage and grain production according to the characteristics of the semiarid region of Brazil. Twenty genotypes with different genotypic classes were evaluated in two years (2018 and 2019) in the municipality of Gracho Cardoso, state of Sergipe. The experiments were carried out in a randomized complete block design (RCBD), with two replicates. Genotype characteristics, grain yield, and forage mass were evaluated and the results were estimated using mixed models and GT Biplots. The genotypic classes consisting of topcrosses and intervarietal hybrids showed high average grain yield and the three classes produced equally on average for forage mass. It was observed that the HI, HTC, and V genotypic classes were the most responsive for grain yield, forage mass, and dry matter content, respectively. Therefore, it is concluded that the genetically broad-based hybrids (HTC and HI) were the most promising for the semiarid of region of Sergipe, while interspecific hybrids were the most productive and stable.

RESUMO O milho é bastante cultivado na região semiárida brasileira e contém alto valor energético na dieta animal. Sabendo disto, objetivou-se caracterizar diferentes classes genéticas de milho quanto ao seu potencial forrageiro nas condições semiáridas de Sergipe. Para isso, foram avaliados 36 genótipos em blocos incompletos em Látice quadrado reticulado 6x6, com duas repetições, na fazenda experimental da Embrapa – Semiárido, em Graccho Cardoso-SE. As variáveis analisadas foram: altura da planta, diâmetro do colmo, peso de planta, massa fresca, massa seca, porcentagem de massa seca e massa de forragem. Com os dados obtidos foi realizada análise de variância com auxílio do software SAS 9,4 e, para a geração dos componentes principais (CP) e gráficos, foi utilizado o software R com o pacote “GGEBiplotGUI”. Todas as variáveis apresentaram diferenças significativas entre classes genéticas. Os dois primeiros componentes principais contemplaram 83,9% da variação total. O autovalor do primeiro CP (CP1) foi de 4,11, considerado muito significativo, já que a soma de todos os autovalores foi de 6,99. A porcentagem de massa seca, diâmetro do colmo, peso de planta, massa fresca, seca e de forragem obtiveram maior capacidade de discriminar as classes genotípicas. Os híbridos intervarietais ficaram mais próximos do genótipo “ideal”. A maior correlação positiva foi entre massa de forragem e massa seca. Os híbridos simples e triplos destacaram-se individualmente na altura da planta e porcentagem de massa. A massa seca e de forragem obtiveram maiores médias em todas as classes, exceto nos híbridos simples. As variedades foram as menos estáveis nas avaliações. Os híbridos intervarietais apresentaram maior potencial forrageiro e os duplos, topcross e variedades maior variabilidade genética, as demais classes não são indicadas para agricultores de baixa e média tecnologia quando cultivados em região semiárida para forragem de milho.

ABSTRACT Corn is widely grown in the semi-arid region and contains high energy value in the animal diet. Thus, the aim of this study was to characterize different genetic classes of maize cultivated for forage potential in the semiarid conditions the Sergipe. For this, 36 genotypes were evaluated in incomplete blocks in lattice square reticulate 6x6, with two replications, in the experimental farm of Embrapa - Semiárido, in Graccho Cardoso-SE. The variables analyzed were: plant height, stem diameter, plant weight, fresh weight, dry weight, percentage of dry weight and forage weight. With the data obtained was accomplished an analysis of variance with the help of software SAS 9,4 and, for generation the principal component analysis (PCA) and graphics, was used the software R with the package “GGEBiplotGUI”. All the variables showed significant differences between genetic classes. The first two main components included 83.9% of the total variation. The eigenvalues of the first PC (PC1) were 4,11, considering more significant, since the sum of all was of 6,99. The percentage of dry mass, stem diameter, plant weight, fresh, dry and forage mass obtained greater hability to discriminate genotypic classes. The intervarietal hybrids were closer to the “ideal” genotype. The greatest positive correlation was between forage weight and dry weight. The single and triple hybrids stood out individually in plant height and percentage of mass. The dry and forage mass obtained higher averages in all classes, except for simple hybrids. The varieties were the least stable in the evaluations. The intervarietal hybrids showed greater forage potential and the double, topcross and varieties had greater genetic variability, already to the genetic classes they are not indicated for farmers of low and mean tech in the semiarid region for forage of maize.