A backcross breeding program between commercial common bean cultivars (VC3 and M20) and sources of resistance (Ex-Rico 23 and G122) was conducted with a view toward selecting carioca (beige with brown stripes) progenies resistant to white mold. Forty-eight progenies (27 F2:6 BC1 and 21 F1:5 BC2) were evaluated for yield, growth habit, grain type and pathogen response by two methodologies for assessment ("straw test" and "oxalic acid"). The methods were effective in discriminating the progenies, showing differing results, for they may assess different resistance mechanisms. Thus, they should be used together. Simultaneous selection for yield, growth habit, grain type and white mold resistance proved to be viable. The most appropriate strategy was to use lower intensities of selection and prioritize traits such as resistance and grain type, which are essential for commercial acceptance. Two progenies proved their superiority for the breeding program for they combine the traits of resistance and favorable grain type.

This study was carried out in order to verify whether natural selection acts on segregating common bean populations grown in the presence and absence of N, and to verify that natural selection affects the frequency of microsatellite alleles specifically, and in order to identify those that can assist in selection. Four populations from the crosses Ouro Negro x CI-107 and VC-5 x IAPAR-81 were carried forward in bulk to F8 at both levels of N. Ouro Negro and VC-5 are considered stress tolerant to nitrogen and IAPAR-81 and CI-107 responsive. One hundred progenies were obtained from each cross and grown in the presence and absence of N. DNA was extracted from 400 progenies, of which 194 were evaluated in the rainy season and 79 in the winter crop for grain yield. These progenies were evaluated in two separate experiments with and without N, in a 14 x 14 lattice design during rainy season and 9 x 9 in the winter season. We selected 35 pairs of polymorphic primers from the parents, with 20 being common from the two crosses. Using DNA of the 400 progenies, it was found that natural selection acted in 33 loci, in at least one environment and cross, and that alleles of the four parents were favoured by natural selection. QTLs were identified for the response rate of N and for grain yield and the linked markers are potential for assisted selection, especially the two most stable; BMD-20, in response to the use of N and PVBR-93 for grain yield.

O objetivo deste trabalho foi selecionar famílias de feijoeiro-comum, com alta produtividade de grãos, por meio de seleção fenotípica e de seleção assistida por marcadores moleculares (SAM). Foram avaliadas 394 famílias, de quatro populações, e seus seis genitores, no Município de Lavras, em dois experimentos: um na geração F3:4, na safra das águas de 2005/2006, em látice simples 20x20; e outro na geração F3:5, na safra da seca de 2006, em látice triplo 20x20. Foram estimados parâmetros genéticos e fenotípicos, e foi realizada a genotipagem das famílias, com marcadores microssatélites associados a QTL controladores da produção de grãos, previamente identificados. Também foram realizadas análises de associação por marcas simples, entre os marcadores e a produção de grãos, e foi obtido um índice para a SAM. A ampla variabilidade entre famílias e as altas estimativas de herdabilidade possibilitaram obter elevados ganhos com a seleção fenotípica. Os marcadores explicaram pequena percentagem da variação fenotípica e apresentaram alta interação QTL x ambiente e QTL x população. A SAM gerou baixos ganhos e a coincidência de famílias selecionadas pelas duas metodologias foi baixa, o que evidencia, neste caso, a ineficiência da SAM, principalmente pela pouca disponibilidade de marcadores ligados a QTL.

The objective of this work was to select common bean families with high grain yield, using phenotypic and marker-assisted selections (MAS). Evaluations were made for 394 families of four populations and their six parents, in Lavras county, Minas Gerais, Brazil. Two field experiments were conducted: one in the F3:4 generation, in the rainy season of 2005/2006, using a 20x20 simple lattice design; and other in the F3:5 generation, in the dry season of 2006, using a 20x20 triple lattice design. Genetic and phenotypic parameters were estimated, and the families were genotyped with microsatellite markers for grain yield quantitative trait loci (QTL), that had been identified previously. Simple marker association analysis was performed between markers and grain yield, and an index was estimated for MAS. The high variability between families and the heritability estimates made high genetic gains possible through phenotypic selection. The markers explained a small percentage of phenotypic variation, and presented high QTL x environment and QTL x population interactions. The MAS achieved small gains, with low coincidence of families selected by both methodologies, showing, in this case, that the MAS inefficiency was mainly due to the low number of available markers.