vulgaris L.) populations in sole crop and in

Early generation testing of common bean (Phaseolus vulgaris L.) populations in sole crop and in maize/bean intercrop G. Atuahene-Amankwa, D. E. Falk, A. D. Beattie and T. E. Michaels University of Guelph, Guelph, Ontario, Canada N1G 2W1. Received 27 March 1997, accepted 13 July 1998. Atuahene-Amankwa, G., Falk, D. E., Beattie, A. D. and Michaels, T. E. 1998. Early generation testing of common bean (Phaseolus vulgaris L.) populations in sole crop and in maize/bean intercrop. Can. J. Plant Sci. 78: 583 588. Few plant-breeding studies have examined methodology for improving common bean (Phaseolus vulgaris L.) yields by selecting in an intercrop situation. We hypothesized that early-generation testing would be as useful in a maize (Zea mays L.)/bean intercrop as in sole crop for identifying superior bean populations for yield. F 2 to F 5 bulks of six selected crosses and their F 5 -derived advanced lines were evaluated in sole crop or intercrop. The F 2 and F 5 bulks were evaluated together in a preliminary trial in one location, while the advanced lines were evaluated with the F 3 s in one location, and with the F 4 s in two locations. Within sole crop, selection of the best three populations, based on F 2 performance, provided 67% of the top advanced lines. The rank correlation between average bulk yield across generations and the average line yield was positive and significant. Within intercrop, selection of the best three populations provided 56% of the top advanced lines. The rank correlation between advanced line yield and bulk yield across generations was positive but not significant. Also, the top three F 2 populations selected in sole crop produced 89% of the top advanced lines in intercrop. Advanced line performance showed a positive significant correlation with mean F 4 bulk performance for grain yield, 100-seed weight and seeds per pod within sole crop, while positive significant correlation was seen with pods per plant and seeds per pod in intercrop. Results indicate that F 2 bulk yields can be used to discard the least promising crosses in both cropping systems. Key words: Early generation testing, Phaseolus vulgaris, intercropping Atuahene-Amankwa, G., Falk, D. E., Beattie, A. D. et Michaels, T. E. 1998. Essai en génération précoce de populations de haricot (Phaseolus vulgaris L.) en culture pure et en culture intercalaire maïs-haricot. Can. J. Plant Sci. 78: 583 588. Peu de phytogénéticiens se sont intéressés jusqu ici à l amélioration des rendements du haricot (Phaseolus vulgaris L.) par la sélection de culture intercalaire. Notre hypothèse de départ était que les essais en génération précoce seraient aussi utiles en culture intercalaire avec le maïs (Zea mays L.) qu en culture pure pour identifier les populations de haricot de productivité supérieure. Les semences mélangées F 2 et F 5 de six croisements et de leurs lignées avancées issues de la F 5 ont été évaluées en culture pure et en culture intercalaire. Les mélanges F 2 et les mélanges F 5 ont d abord été évalués ensemble à un emplacement dans un essai préliminaire, tandis que les lignées avancées étaient évaluées avec les F 3 à un emplacement et avec les F 4 à deux emplacements. En culture pure, la sélection des trois meilleures populations d après les performances des F 2 a fourni 67 % des lignées avancées supérieures. La corrélation de rang entre le rendement moyen des populations, toutes générations confondues, et celui des lignées était positive et statistiquement significative. En culture intercalaire, la sélection des trois meilleures populations se concrétisait par 56 % des lignées avancées supérieures. La corrélation de rang entre le rendement des lignées avancées et celui des populations mélangées, toutes générations confondues était, là aussi positive mais non significative. En outre, les trois populations F 2 supérieures sélectionnées en culture pure produisaient 89 % des lignées avancées supérieures en culture intercalaire. En culture pure, on constatait une corrélation positive significative entre les performances des lignées avancées et celles des performances moyennes des populations F 4, quant au rendement grainier, au poids de 100 grains et au nombre de grains par gousse, tandis qu en culture intercalaire, pareille corrélation ne s observait qu entre le nombre de gousses par plante et celui de grains par gousse. Il ressort de ces observations que les deux systèmes culturaux permettent d utiliser les rendements des populations F 2 mélangés pour rejeter les croisements les moins prometteurs. Mots clés: Essai en génération précoce, Phaseolus vulgaris, culture intercalaire Common bean (Phaseolus vulgaris L.) is a major source of dietary protein in the tropics (Centro International de Agricultura Tropics [CIAT] 1993). It is usually grown by small-scale farmers in association with other crops, the predominant one being maize (Zea mays L.) (Singh 1992). In a maize/bean intercrop, the bean is usually the component negatively affected by competition (Fisher 1977a,b; Francis et al. 1978c; Davis and Garcia 1983; Davis and Woolley 1993), suggesting that beans are the critical factor in this 583 association. A positive correlation has been found between common bean yields in sole crop and in intercrop (Francis et al. 1978a,b; Davis and Garcia 1983; Zimmermann et al. 1984). Therefore, it was suggested that common bean performance in intercrop could be predicted in sole crop (Francis et al. 1978a,b; Francis 1981). At the same time, evidence of genotype cropping system interaction has been reported frequently (Francis et al. 1978a,b; Davis and Garcia 1983; Davis and Woolley 1983; Zimmermann et al.

584 CANADIAN JOURNAL OF PLANT SCIENCE 1984). Theoretically, this interaction suggests that plant breeders may be able to take advantage of favourable genotype cropping system interactions by selecting in intercrop. The ability to select in early generations superior populations that would give high-yielding lines has traditionally been of interest to plant breeders (Smith and Lambert 1968; Cregan and Busch 1977; Hamblin and Evans 1978; Nass 1979; Cooper 1990). If crosses with good potential for bean yield in intercrop could be identified in an early generation, attention could be focused on populations and families from these crosses. The desirability for using early-generation testing in breeding for common bean yield has already been emphasised (Singh et al. 1990; Singh 1991). Investigations conducted in sole crop by Quinones (1969), Hamblin and Evans (1976), and Singh et al. (1990) have demonstrated that early-generation yield tests can be used to estimate the potential of bean crosses for grain yield. Recently, a review by Singh (1992) reported that early-generation yield testing is either used or under study in bean programs to facilitate selection for adaptation to stress factors such as drought (Singh and White 1988; White and Singh 1991), and insects (Kornegay and Cardona 1990), in addition to yield per se (Singh 1989, 1991; Singh and Gutierrez 1990; Singh et al. 1990). Little research has been conducted on early-generation yield tests in maize/bean intercrop. Davis et al. (1983) studied the F 2 -derived F 3 and F 4 families of different bean crosses in sole crop and in maize/bean intercrop. In both cropping systems they found a significant positive linear relationship between bean yields in the F 3 and F 4. It is not known what the relationship is between early-generation bulks and advanced generation lines. The objective of this study was to determine if early-generation testing in sole crop and intercrop can be used to identify populations that will provide top-yielding advanced lines in either cropping system. MATERIALS AND METHODS The study consisted of two stages. The first involved three main activities: (a) selecting crosses with contrasting yield potential in each of the two cropping systems, (b) obtaining individual F 5 plants of these crosses and (c) multiplying their seeds to obtain F 5 -derived F 7 seed. The experimental material for this stage of the experiment consisted of F 2 and F 5 bulks of 10 crosses. These bulks were obtained from elite elite crosses in the bean-breeding program at the University of Guelph, Ontario, Canada (Table 1). The parents were white seeded, and consisted of determinate and indeterminate bush beans (i.e., types I and II plant growth habits) (CIAT 1987). The F 2 seed was residual seed of bulk populations that were in the F 5 generation. All the populations had been advanced to F 5 without artificial selection using a single pod descent method. The F 5 bulk seed had been produced in an off-season nursery in New Zealand during the 1993 winter. The 10 F 2 and 10 F 5 bulks were evaluated in the field in sole crop and in intercrop at the Elora Research Station, Table 1. Pedigree of crosses evaluated in the study Population Parents A OAC Laser W72988 (Unknown Belize selection) B OAC Gryphon W55788 (Midnight/Ex Rico 23//Domino/Neptune) C OAC Speedvale W55788 (Midnight/Ex Rico 23/Domino/Neptune) D OAC Laser W72188 (Kentwood/Seafarer Are) E OAC 88-2 W55788 (Midnight/Ex Rico 23//Domino/Neptune) F OAC 88-1 W72988 (Unknown Belize selection) G OAC Speedvale WO9788 (Swan Valley/Seafarer Are) Ontario, (43 38 N, 80 24 W; elevation 376 m). The soil was a Guelph loam (classified as orthic grey brown luvisol). The trial was planted on 4 June 1993. A split plot design with three replications was used. Cropping systems were assigned to main plots and the populations were assigned to the subplots. Single-row plots of 5 m length were used. Inter-row spacing in sole crop was 60 cm. In intercrop each row of common bean was planted between two rows of hybrid maize (Pioneer 3921) with a distance of 50 cm between the bean and maize rows. Plots were over-planted and thinned to approximately 10 plants per metre for bean rows and 5 plants per metre for maize rows. Ammonium nitrate was applied at the rate of 40 kg ha 1 30 d after planting. Weeds were controlled by inter-row mechanical cultivation and hand-weeding. At maturity, F 3 bulk seed was constituted from each F 2 bulk by harvesting two pods from each F 2 plant. In addition, two plants were visually selected, based on general vigour, and harvested individually from each F 5 plot in each replication and each cropping system. In all, 12 F 5 plants were selected for each F 5 bulk throughout the field. These plants were threshed individually to obtain F 6 seed. Grain yield was determined by harvesting the central 4 m of the bean plots. Pods per plant, 100-seed weight, and seed per pod were estimated from 10 random plants per sub-plot. In addition, days to maturity was also recorded on sub-plots. Due to limited production of F 1 seeds, the F 2 is the first generation in which a modest yield evaluation can occur in our bean-breeding program. The two highest and the two lowest crosses for F 2 grain yield in each cropping system were selected. The same F 2 population was poor in both cropping systems, thus in total, seven populations were selected. Subsequently, the F 6 seed of these crosses, involving 84 lines, was increased in the greenhouse from October 1993 to May 1994 to generate F 7 seed which was used in the second stage of the study. The second stage of the study involved a 2-yr evaluation of the F 5 -derived F 7 lines of the selected populations together with their parental F 3 bulks in the first year, and F 5 - derived F 8 lines and F 4 bulks in the second year. The evaluation was carried out in 1994 and 1995. The experimental design was a 9 9 lattice design with two replications. Seventy-two lines from six populations and their parental bulks were planted in each cropping system. Population F was grown only in sole crop and population G

ATUAHENE-AMANKWA ET AL. EARLY GENERATION TESTING OF COMMON BEAN 585 Table 2. Grain yield (t/ha) of early generation (F 2 to F 5 ) bulks and their F 5 -derived F 7 /F 8 lines of common bean populations evaluated in sole crop and in maize/bean intercrop Bulks Advanced lines Population F 2 (1993) F 3 (1994) F 4 (1995) F 5 (1993) Mean Mean Range Sole crop A 1.89 2.07 2.23 2.06 2.06 2.18 1.74 2.67 B 2.04 2.36 2.28 2.02 2.17 2.21 1.77 2.58 C 1.67 2.14 1.93 1.53 1.82 1.91 1.45 2.35 D 1.60 1.92 1.93 1.48 1.73 2.09 1.63 2.55 E 2.10 2.09 2.05 1.75 1.99 2.17 1.81 2.56 F z 1.57 2.40 1.71 1.64 1.83 2.05 1.25 2.70 LSD y (0.05) 0.32 0.45 0.47 0.32 Intercrop A 1.17 1.42 1.05 0.92 1.14 1.21 0.82 1.50 B 0.96 1.11 1.13 0.87 1.04 1.39 1.19 1.67 C 1.08 1.46 1.08 0.75 1.09 1.08 0.68 1.32 D 0.88 1.39 0.94 0.66 0.97 1.24 1.04 1.53 E 1.06 1.71 1.45 1.05 1.32 1.41 1.26 1.64 G z 0.66 1.13 1.06 0.53 0.84 1.08 0.54 1.37 LSD y (0.05) 0.24 0.39 0.38 0.24 z Population was evaluated in one cropping system. y Based on analyses of all genotypes evaluated in the indicated year. only in intercrop because there was insufficient seed to grow either population in both cropping systems. Three check cultivars were added to complete the 81 entries required for a 9 9 lattice design. Bulk populations of five crosses and related F 5 -derived lines (60) and the three check cultivars were common to both cropping systems. Row lengths of 5 m in 1994 and 6 m in 1995 were used with planting arrangements the same as those for the preliminary evaluation. Data were collected as in the preliminary evaluation, with grain yield from plants harvested from the middle 4 m in 1994 and 5 m in 1995. In 1994 the trial was planted in two replicates at the Elora Research Station on 2 June. Beans were sown at the rate of 11 seeds per metre in rows 5 m in length. At maturity, F 4 bulks were constituted from the F 3 bulks by harvesting two pods from each F 3 plant. F 8 lines were derived by harvesting the seed from F 7 lines. F 4 bulks and F 8 lines were used to repeat the evaluation in 1995 at two locations: the Elora Research Station and the Woodstock Research Station (43 13 N, 80 46 W; elevation 282 m). The soil at Woodstock is a Guelph silt loam (grey brown podzolic). Planting dates were 5 June at Woodstock and 20 June at Elora. Three replications were planted at each location in 6 m rows. The seeding rate for beans was 13 seeds per metre of row. Statistical Analyses Data for each cropping system were analysed separately to determine means for the traits in each population for each generation. Within each generation, means of bulk populations were compared using protected least significant difference (LSD) values. In 1995, one replication at Elora was discarded after flooding caused by heavy rains destroyed significant parts of that replication; hence two replications were used. The lattice design analyses of data were performed with a locally written Advanced Programming Table 3. The number of F 5 -derived F 7 /F 8 lines from each cross which were among the top 18 (25% selection level) lines from crosses evaluated in sole crop or intercrop Population Sole crop Intercrop A 4 4 B 4 6 C 3 0 D 2 2 E 4 6 F z 1 G z 0 z Population was evaluated in one cropping system. Language (APL) program (Petar Gostovick, personal communication). Adjusted plot values and means were determined where the efficiency of the lattice design compared with a randomized complete block design exceeded 110% (Federer 1955; Rodriguez and Hallauer 1991). The adjusted values were used in almost all trait analyses. The exceptions were seeds per pod (Elora 1995, data for both cropping sytems), pods per plant (Elora 1995, intercrop) and maturity (Elora 1994, sole crop). Grain yield was expressed in tons per hectare with the yields adjusted to 18% moisture. Within each cropping system, mean yields across environments (Elora 1994, Elora 1995, and Woodstock 1995) were determined and ranked using PROC MEANS (SAS Institute Inc. 1988). The top 18 lines among the 72 lines were identified (25% selection intensity). The rank orders of crosses in each generation (F 2 F 7 /F 8 ) were used to examine consistency of population performance over environments. Spearman s rank correlations were calculated using SAS Institute, Inc. (1988) procedures to determine the relationships between bulks and advanced lines for the measured traits. Within each cropping system, the data involving the 60 lines and three checks common to both intercrop and sole crop were used to conduct a combined analysis of variance over envi-

586 CANADIAN JOURNAL OF PLANT SCIENCE Table 4. Spearman s rank correlation coefficients (r s ) between average performance of F 5 -derived F 7 /F 8 lines and early-generation (F 2, F 3, F 4, and F 5 ) bulks of six common bean crosses evaluated in sole crop and intercrop for grain yield, pods per plant, 100-seed weight, seeds per pod, and maturity Population Grain yield Pods per plant 100 seed weight Seed per pod Maturity SC z IC z SC IC SC IC SC IC SC IC F 2 0.66 0.03 0.26 0.03 0.21 0.03 0.84* 0.65 0.06 0.12 F 3 0.14 0.09 0.03 0.49 0.25 0.60 0.28 0.88* 0.09 0.21 F 4 0.94** 0.43 0.66 0.83* 1.00** 0.77 0.84* 0.82* 0.41 0.51 F 5 0.77 0.60 0.29 0.60 0.58 0.71 0.58 0.52 0.12 0.56 F 2 to F 5 mean 0.83* 0.37 0.09 0.60 0.64 0.94** 0.75 0.88* 0.18 0.44 z SC = Sole crop, IC = intercrop. *, ** Indicate significance at P = 0.05 and P = 0.01 probability levels, respectively. ronments. This was done to investigate the extent of genotypic, environmental and genotype environment effects. RESULTS Sole Crop Within sole crop, population D was consistently low yielding across generations, in contrast to population B which was consistently high yielding (Table 2). Otherwise, differences among populations did not appear to be consistent over generations, and this could partly be due to interaction of bulks with the evaluation environments, and/or the number of entries involved in 1993 (when F 2 and F 5 bulks were grown) compared with those in 1994 to 1995 (when the F 3 and F 4 were evaluated). At the 25% selection level, 12 of the 18 highest yielding lines (67%) originated from the top three crosses in the preliminary evaluation, namely populations E, B, and A (Table 3). Only one line from population F ranked among the top-yielding lines in sole crop, which incidentally, was the highest-yielding entry. Population B consistently ranked high in all generations, and the performance of advanced lines from this population parrallelled those of the bulks. The results suggest that crosses with poorest overall bulk performance produced fewest superior lines. An examination of parents of all the populations (Table 1) shows that populations B, C and E have one parent in common, namely W55788, and, except for population C, these populations performed relatively well in the study. Populations A and D also have a common parent, OAC Laser, while populations A and F shared parent W72988. The rank correlation between average bulk yield across generations and the average line yield was positive and significant (r = 0.83*) (Table 4). Except for the F 4 generation, the rank correlation between bulk yield and yield of the F 5 - derived line was not significant (Table 4). This trend was not surprising since the F 4 bulks were evaluated with the advanced lines in the same two environments while other populations were not. The poor correlation between F 3 bulks and advanced lines was unexpected. This may be the result of evaluation in only one environment which provided a less-reliable assessment of genotypic effects unbiased by genotype environment interactions. In addition to yield, significant and highly significant rank correlations were found between F 4 bulks and their advanced lines for seeds per pod and 100-seed weight. Significant rank correlation was also found between F 2 bulk and advanced line for seeds per pod. Intercrop Within intercrop, population G showed below-average performance in all stages of the study, as might be expected from the F 2 evaluation, while population E consistently showed a good performance (Table 2). Population C showed satisfactory bulk yields, yet lines from this cross were not equally satisfactory (Table 3), but this case was the exception, rather than the general trend. Populations C and G shared OAC Speedvale as a common parent (Table 1), and neither population contributed a selected line. The three top crosses based on F 2 performance (A, C and E) provided 10 of the 18 highest-yielding lines (56%) at the 25% selection pressure. The two poorest-yielding populations provided only two superior lines (11%) (Table 3). The rank correlation between average bulk yield across generations and the average line yield was positive but not significant. However, the F 4 bulk was significantly correlated with advanced lines for pods per plant and seeds per pod (Table 4). Also, the F 3 bulk showed significant positive correlation with the advanced lines for seeds per pod. Though not significant, the F 5 bulk showed moderately high correlations with advanced lines for all traits. It would appear from these results that the relative rankings of successive bulk generations, particularly for yield and pods per plant, were slightly less consistent than those found in sole crop. Analyses of variance for grain yield involving the advanced lines common to both cropping systems in the 2-yr evaluation (Table 5) showed relatively higher genotype environment interaction in intercrop. DISCUSSION A major decision faced by breeders is whether to eliminate populations early in a breeding cycle because they show little promise in early generations. Selection of the three highest-yielding populations in the F 2 captured 67% of the top F 7 /F 8 lines in sole crop. As a method for yield selection, early-generation selection might be used to eliminate the two least-promising populations (3 of 18 lines). Yet this requires some caution as the lowest-yielding population in sole crop provided the highest-yielding advanced line. In

ATUAHENE-AMANKWA ET AL. EARLY GENERATION TESTING OF COMMON BEAN 587 Table 5. Mean squares from the analysis for grain yield (t/ha) of 63 genotypes of common bean evaluated in sole crop and intercrop across three environments in 1994 1995 Source df Sole crop Intercrop Environment 2 9.189 11.517** Replication 4 1.480** 0.223* Genotype 62 0.568** 0.313** Genotype Environment 124 0.168 0.111** Error 248 0.134 0.054 CV 17.2 18.9 R 2 0.72 0.81 *, **Significant at P < 0.05 and P < 0.01, respectively. intercrop, selection of the three highest-yielding populations captured 56% of the top F 7 /F 8 lines. However, it might be efficient to discard the two least-promising populations (2 of 18 lines). It is interesting to note that early-generation selection in the cropping system in which the advanced lines were also tested produced gains of 16% for sole crop and 6% for intercrop over random selection (i.e., 50% of top-yielding lines captured), but selection in sole crop for intercrop captured 16 of the 18 top-yielding advanced lines (89%), an improvement of 39% over random selection. This finding indicates that a separate selection program under intercropping conditions may not be justified when developing cultivars destined for intercrop production. The crosses used for this study involved elite germplasm. For such genetic material a breeder may be inclined to retain a relatively large proportion, especially if few genotypic differences can be detected among populations. In contrast, for wide crosses it might be necessary to discard at a much higher rate. This study provides some insight on the value of early-generation testing if the goal is to discard a small percentage of the populations during an initial evaluation, for example in the F 2. Thus, unless the populations available show an extreme range of yield potential, early generation tests should only be used to discard the worst-yielding populations or families. The results further demonstrate that F 4 generation testing could reliably be used for predicting grain yield, 100-seed weight and seed per pod within sole crop and, pods per plant and seeds per pod in intercrop. For a quantitative trait such as grain yield, it can be expected that interactions between environmental and genotypic effects could weaken the correlation between yields of bulks and those of advanced lines, particularly if the two were evaluated in different environments. Results reported by Singh et al. (1990) in common beans and Nass (1979) in wheat also demonstrated that changes in rank order of populations can affect performance of crosses across generations. This problem may be minimized through the use of multiple environments to evaluate bulks, if enough seed is available. This study shows that the value of early-generation testing varies with cropping system. The top two F 2 populations selected in the first part of the study as being best in intercrop produced only 22% of the top-yielding advanced lines in intercrop, while the top two F 2 selections for sole crop produced 44% of the top lines in sole crop. Also, these same two selections for sole crop captured 67% of the top-yielding lines in intercrop. If it is necessary to select in intercropping, early-generation tests can be used to discard the worst populations. The best populations can then be advanced through various generations using conventional methods such as single-seed descent or bulk method for final selection of pure lines. However, good progress may be made by selection of lines in sole crop for use in either cropping system. ACKNOWLEDGEMENTS Financial support by the Canadian Commonwealth Scholarship and Fellowship Plan towards the graduate studies of G. Atuahene-Amankwa, and the Natural Sciences and Engineering Research Council, Ontario Bean Producers Marketing Board and Ontario Coloured Bean Growers toward the studies of A. D. Beattie is gratefully acknowledged. We also acknowledge Thomas Smith for assistance in carrying out the field trials and Peter Gostovic for assistance on data analysis. Centro International de Agricultura Tropics. 1987. Standard system for the evaluation of bean germplasm. A van Schoonhoven and M. A. Pastor-Corrales, Compilers. Cali, Colombia. 54 pp. Centro International de Agricultura Tropics. 1993. CIAT at the threshold of sustainable development, 1992 93. CIAT, Colombia. 32 pp. Cooper, R. L. 1990. Modified early generations testing procedure for yield selection in soybean. Crop Sci. 30: 417 419. Cregan, P. B. and Busch, R. H. 1977. Early generation bulk hybrid yield testing of adapted red spring wheat crosses. Crop Sci. 17: 887 891. Davis, J. H. C. and Garcia S. 1983. Competitive ability and growth habit of indeterminate bean and maize for intercropping. Field Crops Res. 6: 59 75. Davis, J. H. C. and Woolley, J. N. 1993. Genotypic requirement for intercropping. Field Crops Res. 43: 407 430. Davis, J. H.C., Perez, A. V. and Hopmans, S. 1983. Early generation yield testing in beans in monoculture and intercropped with maize. Proc. EUCARPIA meeting on Phaseolus improvement, Hamburg, Germany. August 1983. pp. 54 65. Federer, W. T. 1955. Experimental design. The Macmillan Co. New York, NY. 544 pp. Fisher, H. M. 1977a. Studies in mixed cropping. I. Seasonal differences in relative productivity of crop mixtures and pure stands in Kenya highlands. Exp. Agric. 13: 177 184. Fisher, H. M. 1977b. Studies in mixed cropping. II. Population pressure in maize-bean mixtures. Exp. Agric. 13: 185 191. Francis, C. A. 1981. Development of plant genotypes for multiple cropping system. Pages 179 232 in K. J. Frey ed. Plant breeding. II. Iowa State University Press, Ames IA. Francis, C. A., Flor, C. A. and Prager, M. 1978c. Effect of bean association on yields and yield components of maize. Crop Sci. 18: 760 764. Francis, C. A., Prager, M., Laing, D. R. and Flor, C. A. 1978a. Genotype by environment interactions in bush bean cultivars in monoculture and associated with maize. Crop Sci. 18: 237 242. Francis, C. A., Prager, M., Laing, D. R. and Flor, C. A. 1978b. Genotype by environment interactions in climbing bean cultivars in monoculture and associated with maize. Crop Sci. 18: 242 246. Hamblin, J. and Evans, A. A. 1976. The estimation of cross yield

588 CANADIAN JOURNAL OF PLANT SCIENCE using early generation and parental yields in dry beans (Phaseolus vulgaris L.) Euphytica 25: 515 520. Kornegay, J. and Cardona, C. 1990. Development of an appropriate breeding scheme for tolerance to Empoasca kraemeri in common bean. Euphytica 47: 223 231. Nass, H. G. 1979. Selecting superior spring wheat cross in early generations. Euphytica 28: 161 167. Quinones, F. A. 1969. Relationship between parents and selections in dry beans (Phaseolus vulgaris L.) Crop Sci. 9: 673 675. Rodriguez, O. A. and Hallauer, A. R. 1991. Variation among full-sib families of corn for different generations of inbreeding. Crop Sci. 31: 43 47. SAS Institute, Inc. 1988. SAS user s guide: Statistics Release 6.03. SAS Institute, Inc., Gary, NC. Singh, S. P. 1989. Breeding for yield in common bean of Middle- American origin. Pages 248 263 in S. Beebe, ed. Current topics in breeding of common bean. Working Document 47. CIAT. Cali. Colombia. Singh, S. P. 1991. Breeding for seed yield. Pages 383 443 in A. van Schoonhoven and O. Voyset, eds. Common beans: Research for crop improvement. CAB International, Wallingford, UK and CIAT, Cali, Columbia. Singh, S. P. 1992. Common bean improvement in the tropics. Pages 199 269 in J. Janick, ed. Plant breeding reviews. Vol. 10. John Wiley and Sons, Inc., New York, NY. Singh, S. P. and White, J. 1988. Breeding common beans for adaptation to drought conditions. Pages 261 284 in J. W. White, G. Hoogenboom, F. Ibarra, and S. P. Singh, eds. Research on drought tolerance in common bean. Working document 41. CIAT, Cali, Columbia. Singh, S. P. and Gutierrez, J. A. 1990. Effect of plant density on selection for seed yield in two population types of Phaseolus vulgaris L. Euphytica 51: 173 178. Singh, S. P., Lepiz, R., Gutierrez, J. A., Urrea, C., Molina, A. and Teran, H. 1990. Yield testing of early generation populations of common bean. Crop Sci. 30: 874 878. Smith, E. L. and Lambert, J. W. 1968. Evaluation of early generation testing in spring barley. Crop Sci. 8: 490 493. White, J. W. and Singh, S. P. 1991. Breeding for adaptation to drought. Pages 501 560 in A. van Schoonhoven and O. Voyset, eds. Common beans: Research for crop improvements. CAB International, Wallingford, UK and CIAT, Cali, Colombia. Zimmermann, M. J. O., Rosielle, A. A., Waines, A. A. and Foster, K. W. 1984. A heritability and correlation study of yield, yield components and harvest index of common bean in sole crop and intercrop. Field Crop Res. 9: 109 118.