O cultivo do milho safrinha tem se expandido consideravelmente nos ultimos anos. Porem devido as condicoes ambientais contrastantes entre safra e safrinha, deve ocorrer pronunciada interacao genotipos x safras, indicando que os genotipos selecionados para a safra podem nao ser adequados para a safrinha. Assim, os objetivos deste trabalho foram investigar a magnitude e natureza da interacao genotipos x safras e a necessidade de conduzir programas de melhoramento de milho distintos para o cultivo de safra e safrinha. Foram utilizadas 100 progenies S1fs derivadas do cruzamento entre as linhagens endogamicas L08- 05F e L38-05D, as quais foram retrocruzadas com ambas as linhagens genitoras e, posteriormente, estas foram cruzadas com a linhagem endogamica L02-03D para obtencao de 200 testecrosses (100 RC1 TC e 100 RC2 TC ). Os testecrosses obtidos foram avaliados nos ambientes de safra e safrinha em dois anos agricolas no municipio de Piracicaba/SP no delineamento ?¿-latice com duas repeticoes por ambiente por ano. Os caracteres analisados foram producao de graos (PG), prolificidade (PROL), acamamento e quebramento de plantas (ACQ), altura de planta (AP) e de espiga (AE), posicao relativa da espiga (PRE), florescimento masculino (FM) e feminino (FF) e intervalo entre florescimentos (IF). Diferencas entre as medias nas duas safras foram significativas para todos os caracteres, reduzindo significativamente a PG na safrinha devido a menor disponibilidade hidrica. A variancia genetica dos testecrosses para PG na safrinha foi inferior a da safra para os testecrosses RC2 TC , enquanto que para os testecrosses RC1 TC a variancia genetica nao diferiu de zero na safrinha. Consequentemente, o coeficiente de herdabilidade na safrinha para os RC1 TC nao diferiu de zero e os RC2 TC apresentaram coeficientes de herdabilidade de magnitudes similares nas duas safras. Para os demais caracteres as magnitudes das variancias geneticas e coeficientes de herdabilidade diferiram entre as safras e os dois grupos de testecrosses. As magnitudes dos componentes da interacao testecrosses x safras mostraram que o tipo complexa explica a maior parte da interacao, sendo causada pelas baixas magnitudes de correlacoes entre os caracteres nas diferentes safras. As respostas diretas e indiretas a selecao mostraram que a selecao direta foi mais eficiente que a indireta em todas as situacoes, com excecao daquelas em que nao foi detectada variancia genetica na safrinha. As respostas a selecao baseadas nas medias das safras resultaram em progressos na safra e na safrinha proximos aos observados pela selecao direta. Porem, a coincidencia de testecrosses selecionados em ambas as safras foi muito baixa para os dois testecrosses. Assim, os resultados deste estudo sugerem que os programas de melhoramento de milho devem ser distintos para safra e safrinha.

Maize second crop season, known as gsafrinhah, has increased considerably in the last years in Brazil. However, because of the different environmental conditions between the two crop seasons strong genotype by environment interaction is expected to occur, and then maize genotypes selected in the first crop season (known as gsafrah) could not be those ones that would be selected in the safrinha. Thus, objectives of this research were to investigate the magnitude and the nature of the genotype x crop season interaction and whether separate maize breeding programs for each crop season should be conducted. One hundred S1Ls progenies developed from a population produced from the cross of inbred lines L08-05F (P1) and L38-05D (P2), were backcrossed to both parental inbred lines and, subsequently, these backcrosses were crossed to the inbred line L02-03D giving rise to the 200 testcrosses, thereafter named as TBC1 and TBC2 for the backcrosses to P1 and P2, respectively. The testcrosses were evaluated in two crop seasons in two years at the Piracicaba city, Sao Paulo State, in the ?¿-lattice designs with two replications per crop season and year. The traits analyzed were grain yield (GY), prolificacy (PROL), plant lodging (PL), plant (PH) and ear (EH) heights, ear placement (EP), days to anthesis (DA), days to silking (DS), and anthesissilking interval (ASI). Significant differences for the means of all traits in the two crop seasons were detected, and GY reduced significantly in the safrinha because of the moisture stress. The magnitude of the genetic variance for GY in the safrinha was inferior to that in the safra for the TBC2, and for the TBC1 this parameter did not differ from zero in the safrinha. Consequently, the heritability coefficient in the safrinha for the TBC1 did not differ from zero and the TBC2 presented similar magnitudes of the heritability coefficients in both crop seasons. For the other traits, the magnitudes of the genetic variances and of the heritability coefficients were different between the crop seasons and for the two sets of testcrosses. The magnitudes of the components of the interactions testcrosses x crop seasons showed that the complex type i.e., the cross-over interaction type, accounted for the major part of the interactions, which were due to the low magnitudes of the correlations of the traits in the two crop seasons. Estimates of the direct and indirect responses to selection showed that the direct selections were more efficient than the indirect selections, except to that trait in which the genetic variance did not differ from zero in the safrinha. The responses to selection based on the means of the two crop seasons presented similar magnitudes of the direct responses in both crop seasons. However, the coincidence of testcrosses selected in the two crop seasons was very low for both sets of testcrosses. Therefore, the results of this study suggest that separate maize breeding programs should be conducted for each crop season.