Quatro variedades de feijão Phaseolus vulgaris foram estudadas com respeito a dois tipos de seleção para capacidade competitiva alta e baixa, dois tipos de competição em monocultura e em competição com mistura varietal, estando cada uma representada por dez linhagens, por tipo de seleção. Chegou-se as seguintes conclusões: 1 - Não se observou efeito de seleção nas cinco análises de variância efetuadas. 2 - O efeito de competição e suas interações, principalmente variedades por competições, foi significativo para produção e número de vagens por cova e não significativo para número de sementes por vagem e peso médio de uma semente. 3 - O efeito das linhagens dentro de variedades, para cada tipo de seleção, foi significativo a 1 ou 5% em todas as análises. A interação linhagens por competição, dentro de variedades, dentro de seleções somente foi significativa para um dos componentes da produção, ou seja, número de sementes por vagem. 4 - A influência do tipo parcela usada na eficiência experimental foi fonte de erros elevados para produção e número de vagens por cova e satisfatórios para número de sementes por vagem e peso médio de uma semente. 5 - O ganho percentual de seleção foi obtido por comparação, dentro de cada variedade, das médias das produções um ou outro caráter das linhagens selecionadas para capacidade competitiva alta e baixa, sob efeito de competição da mistura varietal. Estes dados refletiram a variabilidade existente no material básico do qual as linhagens foram tiradas. Eles devem ser apreciados em termos de tendências, uma vez que a baixa eficiência experimental do ensaio não permitiu que se detectas sem efeitos de seleç6es mesmo em presença de diferenças entre médias de até 28,4%. Utilizando dados de produção, em ordem de crescente de variabilidade, as variedades assim se apresentaram: Costa Rica, Iguaçu, Preto G₁ e Cuva 168 N. 6 - As herdabilidades ao nível de médias de linhagens foram elevadas, especialmente para a produção, obtendo-se o valor médio de 53,8%, e amplitude de 25,3 a 73,8%. As herdabilidades a nível de plantas individuais foram baixas, com média de 4,9% e amplitude de 1,1 a 9,2%. Os mais baixos valores ocorreram para a produção; os mais altos para o número de sementes por vagem. 7 - Nas discussões feitas sobre relações de desenvolvimento, importantes para explicar o surgimento de correlações negativas entre componentes de produção, com reflexos na competição e na produção, concluiu-se que o rendimento de 2.607 kg/ha obtido no ensaio possibilitou a expressão do teto genético na realização do desenvolvimento e na manifestação da competição.

The estimation of the relativa competitiva abilities of fifteen varieties of common bean Phaseolus vulgaris L. was obtained by GUAZZELLI (1975) in a dissertation dealing with intergenotypic competition. Highly significant effects were identified in one variety for competitiva ability as well as varietal tendencies to produce better in pure stand, or as in competition with a mixture of fifteen varieties. A complementary interaction was detected, indicating that the gains obtained in some associations nulified the lasses found in others. In the present study, it was hoped: 9. 1. - To verify the effect of a disruptiva selection technique (B) for high and low competitiva ability praticed in four bean varieties (A₁ to A₄). In the basic population of these varieties two them exibited a general effect for competition, the other two for pure stand. The analysis of variance in a split split split-plot design was unable to detect a significant selection for yield (W) using either the four central replicates or the total of sixteen replicates in the experiment. The sarne result was obtained, whem yield in four replicates was split into its yield components, number of pods per hill (X) number of seeds per plant (Y) and unit seed weight (Z). 9. 2. - To observe the performance of each selection technique, pure stand (B₁) and in mixture (B₂) were sown in pure stand (C₁) and in competition (C₂) with a mixture of the fifteen varieties used by GUAZZELLI (1975). The analysis of variance showed that this source of variation, and principally its interation AC, was significant with regard to yield and number of pods per hill. No significant diferences were obtained for number of seeds per pod and unit seed weight. 9. 3. - To determine line effects (D). 10 lines were selected from the basic material in each branch of the selection process. Twenty lines per variety were tested. A hierarchical classification was considered for the following sources of variation : lines ( D), the interaction DC within varieties (A), and withing selection techniques (B). In all the analyses of variance, line effects were significant at the 1% ar 5% levels for all characters studied, namely X. Y. Z and W. The interaction of lines and competition was significant only for Y. 9. 4. - The influence of plot area is discussed with regard to experimental efficiency. Because no other adequate plot types were found in the literature, the same plot type was used to evaluate competition namely, one center hill per plot, in spite of deficiencies pointed out in previous work. The adoption of GUAZZELLI's (1975) recommendation of using 10 hill plots would have increased the experimental area twenty times, and it would have covered an area of 3.6 ha. Another reason for using the one hill plot was to confer the maximum efficiency at the level of split Split split plot. Contrary to the expected results, high experimental errors were observed at all levels of the analysis of variance for production and number of pods per hill, satisfactory erros were found for number of seeds per pod and unit seed weight. 9. 5. - The selection gain in percentage was estimated by comparing within eacb variety the means of lines selected for high and low competitive ability . In each case the comparison was made under competition with the variety mixture. This estimation reflects the existing variability in the basic material from which the selections were drawn. Data presented shows that differences of up to 28.4 % were detected in the Iguaçu (A₄) variety in spite of non significant diferences for selections. The varieties showed the following decreasing arder of variability for production, using sixteen replicates: Costa Rica (A₂ ), Iguaçu (A₄ ), Preto G₁ (A₁) and Cuva 168 N(A₃). It was observed that selection gain tended to increase with the number of replicates. In Costa Rica a reversion favouring pure stand instead of competition occured. Its probable causes were discussed. 9. 6. - The estimation of genetic parameters for yield and the components of yield and its relationship to intergenotypic competition was the major objective of this paper. It has been observed that mean values of lines were not so uniformilly distributed, as to justify the estimation of parameters, from the mean values of the componente of variance in table 4. The componente of variance were estimated by computing D/A/B/C. Negative estimates of the genetic componentes of variance were considered null according to NELDER (1954). Likewise heritabilities for mean values of lines and individual plants were estimated. Wherever the interaction DC/A/B was non significant average values of heritability were taken between pure stand (C₁) and competition (C₂ ). Heritabilities for the mean values of lines were high especially for yield in sixteen replicates, with a range of 25.3 to 73.8 % with an average of 53.8%. Heritabilities for individual plants (hills) ranged from 1.1 to 9.2 % with an average of 4.9%. It was observed that the heritabilities for individual plant were 1.6 to eleven time lower than those for the mean values of lines. This reflects the power of progeny tests. 9. 7. - Yield camponents X, Y. and Z determined in four replicates have been discussed. The lower estimate of heritability at the individual plant level was related to yield and the higher, number of seeds per pod (Y). This is in accord with DENI's (1967). 9. 8. - Developmental relations are briefly discussed. It is a subject which has recently attracted much attention for its value in explaining the onset of negative correlations between yield components (ADAMS and GRAFIUS. 1970). There is a genetic ceiling for the utilization of avaliable environmental resources. In the case of limitation of these resources, either in a continuous form ar in an oscillating pattern, this limitation may modify the yield components and yield itself and the genetic potential of the plant may not be attained.