INCORPORATTON 0F GERMPLASM OF THE SEAFARER NAVY BEAN ENTO SEVERAL CENTRAL AND SOUTH AMERTCAN BEAN CULTIVARS, TN RELATTON T0 BREEDING FOR TMPR’OVEMENT IN ADAPTATION Dissertation for the Degree of Ph. D. MICHTGAN STATE UNIVERSITY GASPAR ALBERTO STU/ERA C. 1974 we»: LIBRARY Micaigan State. INCORPORATION “95¢.“ SEVERAL CENT AL AND SOUTH AMERI IN RELATION TO BREEDING FOR IMPRO '..' r: . ~ , 3 v -=.;_’_ ' Gaspar Alberto Silvera C. . - a - . - . ' ’ at t: :" fi'é} _'r' " has been accepted towards fulfillment ~ i L. . of the requirements for '1; ’ ‘ .j-Ij‘y Ph.D. degreem Crop 5 Soil Science ~+ ' l '4' - f "9 %W Majoxprofessor {. é 61 1- . “4% //47 .1. ?: F ‘ It ‘32., ‘ '; ".,-' , ~ .. f' ‘ 0-7639 ., . _: -' 0 at ABSTRACT INCORPORATION OF GERMPLASM OF THE SEAFARER NAVY BEAN INTO SEVERAL CENTRAL AND SOUTH AMERICAN BEAN CULTIVARS, IN RELATION TO BREEDING FOR IMPROVEMENT IN ADAPTATION BY Gaspar Alberto Silvera C. The identification of the traits that condition wide adaptation in field beans is necessary to the development of high-yielding, widely adapted varieties. The accomplish- ment of this goal in cereals, such as rice and wheat, was realized several years ago, but in beans it is a task that remains to be done. The Seafarer navy bean has been successfully grown in diverse areas of the world. The adaptive values of sev— eral of its characteristics, e.g. the determinate growth habit, early maturity, and small leaflets, were determined, in an attempt to relate them to wide adaptation. Those traits were incorporated into eight Central and South Amer- ican bean varieties by means of backcrosses. Selection was practiced in backcross self generations to recover the Seafarer traits and the seed type of the Central or South American variety. Six different "selection classes" were also established, composed of combinations of alternative §\ Gaspar Alberto Silvera C. states of the traits, e.g. determinate, early, small-leafed; determinate, late, large-leafed; etc. The adaptive value of each selection class and the contribution to adaptation of individual traits was determined. The methods proposed by Finlay and Wilkinson (1963), and Eberhart and Russell (1966) were followed. Two stabi— lity parameters were determined for each of the 64 genotypes evaluated: first, a regression coefficient of the varietal mean yield or yield component onto the respective environ— mental indexes (location mean minus the grand mean); and second, the varietal mean over all locations. A third esti- mator of adaptation for every genotype was the deviations— from-regression mean square. The ideal variety is con- sidered to be that with maximum yield or yield component value, a regression coefficient of 1.0 and a deviations from regression of zero. Data were collected from two locations in South America: La Molina, Peru and Palmira, Columbia, and from one location in North America, East Lansing, Michi- gan. Yield was the variable most affected by the environ- ment, as determined from the analysis of variance over the three locations, and seed weight was the most affected among the yield components. The number of seeds per pod was least subject to the influence of the environment. Gaspar Alberto Silvera C. Seafarer was the highest yielding genotype, showing wide adaptation to all locations. Genotypes with the determinate, early, small—leafed characteristics were found to be better adapted to all en- vironments, as determined from.the performance of the select— ion classes and from the contribution to adaptation of the individual traits. Lines selected from backcrosses to Sea— farer produced larger yields and greater number of pods per plant and seeds per pod. They showed a tendency to be widely adapted. It is suggested that the three traits present in Seafarer, plus daylength neutrality, are mainly responsible for its wide adaptation. The contribution to adaptation of daylength neutrality was not determined, but this trait directly conditions wide adaptation in other crops and its contribution might be a decisive factor in explaining wide adaptation in beans. The Seafarer germplasm, as a whole, is relatively widely adapted and it is possible that some other unidenti- fied traits may also be related to wide adaptation. The idea that it is the whole Seafarer genotype that conditions wide adaptation is not here suggested as an explanation of the excellent performance of this variety. INCORPORATION OF GERMPLASM OF THE SEAFARER NAVY BEAN INTO SEVERAL CENTRAL AND SOUTH AMERICAN BEAN CULTIVARS, IN RELATION TO BREEDING FOR IMPROVEMENT IN ADAPTATION BY Gaspar Alberto Silvera C. A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR.OF PHILOSOPHY Department of Crop and Soil Science 1974 To my wife, Flor To my parents, Gaspar and Rosa ii ACKNOWLEDGMENTS The author wishes to express his sincere apprec— iation to Dr. M.W. Adams for his valuable guidance and sug- gestions as the research advisor and for his help in the linguistic improvement of this dissertation. To the members of the Graduate Committee, Drs. W.J. Hanover, S. Honma and A. Saettler, for their cooperation and assistance. The benefits of a fine education would not have been possible for me without the support from the Department of Crop Science, through an assistantship, and from the Uni- versity of Panama, that provided me a sabbatical leave. Finally, I deeply appreciate the moral support given to me by my wife, Flor. Her dedication and help was most valuable during these years at Michigan State. iii TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . V LIST OF FIGURES . . . . . . . . . . . . . viii INTRODUCTION . . . . . . . . . . . . . . 1 REVIEW OF LITERATURE . . . . . . . . . . . 4 MATERIALS AND METHODS . . . . . . . . . . . 10 RESULTS AND DISCUSSION . . . . . . . . . . . l7 Pods Per Plant . . . . . . . . . . . . l8 Seeds Per Pod . . . . . . . . . . . . . 25 Seed Weight . . . . . . . . . . . . . . 30 Yield . . . . . . . . . . . . . . . . 33 Performance of the Parental Varieties . . . . . 41 Performance of the Selection Classes . . . . . 49 Contribution of Individual Traits to Adaptation . . 60 Performance of Backcross Populations . . . . . 63 SUMMARY AND CONCLUSIONS . . . . . . . . . . 67 LITERATURE CITED . . . . . . . . . . . . 71 APPENDIX . . . . . . . . . . . . . . . 74 iv Table 10 ll 12 LIST OF TABLES Characteristics of the selection classes. Characteristics of the progenitor varieties. Experimental error and coefficients of vari— ation for yield and yield components for 64 bean genotypes, measured at three locations in North and South America. Location means and environmental index for yield and yield components, when measured at three locations in North and South America. Analysis of variance of pods per plant (on loglo scale) of 64 bean genotypes grown in three different locations. Pods per plant stability parameters and average of every line when planted at three different locations. Analysis of variance for seeds per pod (on 10910 scale) of 64 bean genotypes grown in three different locations. Seeds—per-pod stability parameters and aver- age of every line when planted at three dif— ferent locations. Analysis of variance for seed weight (on loglo scale) of 64 bean genotypes grown in three different locations. Seed weight stability parameters and average in every line when planted at three different locations. Analysis of variance for the yields (on loglo scale) of 64 bean genotypes grown in three different locations. Yield stability parameters and average of every line when planted at three different locations. V Page 12 l6 l8 19 20 21 26 27 31 32 34 35 Table Page 13 Mean squares for regressions and devia- tions from regression, for yield and its components, from the analysis of variance of data collected at three different loc- ations. 39 14 Stability parameters and means of parental varieties. 43 15 Average number of days to harvest and leaflet size over all selection classes and for Seafarer. 50 16 Average stability parameters and means of every selection class. 51 17 Average regression coefficients over all entries for each selected trait. 61 18 Average values of yield and yield components over all entries for each selected trait. 61 19 Average deviations mean square over all entries for each selected trait (in loglo measure). 62 20 Average stability parameters over all sel- ected genotypes for each backcross population and Seafarer. 64 21 Genotypic constitution, plant type, and some maturity characteristics for each evaluated line. 75 22 Mean values for maturity and leaf charact~ eristics, and harvest index for each evaluated line. 78 23 Location mean number of pods per plant for every line evaluated. 81 24 Location mean number of seeds per pod for every line evaluated. 84 25 Location mean seed weight for every line evaluated. 87 26 Location mean yields (gm/lmt) for every line evaluated. 90 vi Table Page 27 Analysis of variance for log yield (gms/ lmt) values from data collected in an 8x8 lattice planted at East Lansing, Michigan and La Molina, Peru. 93 28 Analysis of variance for log of pods per plant values from data collected in an 8x8 lattice planted at East Lansing, Michigan and La Molina, Peru. 94 29 Analysis of variance for log seeds per pod values from data collected in an 8x8 lattice planted at East Lansing, Michigan and La Molina, Peru. 95 30 Analysis of variance for log (seed weight + 1.0) values from data collected at East Lansing, Michigan, and La Molina, Peru. 96 31 Analysis of variance for yield and yield components values from data collected at Palmira, Colombia. All data on loglo scale. 97 32 Weather data (1973) for the three locations for the period the experiments were in the field. 98 vii Figure 10 11 LIST OF FIGURES The relation of pods per plant and the regression coefficient for 64 bean lines grown in three different environments. The relation of seeds per pod and the regr ression coefficient for 64 bean lines grown in three different environments. The relation of seed weight and the regres— sion coefficient for 64 bean lines grown in three different environments. The relation of yield and the regression coefficient for 64 bean lines grown in three different environments. Regression lines showing the response (pods per plant) of nine parental varieties to three different environments. Regression lines showing the response (seeds per pod) of nine parental varieties to three different environments. Regression lines showing the response (seed weight) of nine parental varieties to three different environments. Regression lines showing the response (yield gm/lm) of nine parental varieties to three different environments. Fluctuation around the mean of the coeffi- cients of regression, for each selection class, calculated for yield and yield com— ponents. Regression lines of the average response (pods per plant) of six selection classes to three different environments. Regression lines of the average response (seeds per pod) of the six selection classes to three different environments. viii Page 24 29 38 45 46 47 48 53 54 55 56 Figure Page 12 Regression lines of the average response (seed weight) of six selection classes to three different environments. 57 13 Regression lines of the average yield (gm/ 1m) of six selection classes to three dif- ferent environments. 58 ix INTRODUCTION The narrow germplasm base of many modern crOp culti- vars has been recognized in recent years‘and the ensuing genetic vulnerability to parasites makes it a potential cause for crop failure. The genetic and phenotypic homo- geneity maximizes yield potentials in localized environments but reduces adaptation to varying conditions. Market prefer- ences encourage the breeder to develop uniform varieties but the wide genetic diversity available in most crop species is not fully exploited in the development of com- mercial varieties. Adaptation studies are valuable for the breeder in teaching him which plant traits have adaptive and in helping him to decide whether broad or narrow adapt— ation should be the goal of the breeding program. The determinate habit, early maturity, small leaflet, and daylength neutrality present in Seafarer navy bean were postulated as adaptive traits for this study. The contri— buted to adaptation of these morphological-physiological characteristics (except for daylength neutrality) was studied. Inheritance studies (Emerson, 1916, Duarte, 1961) show these traits to be heritable and therefore subject to l selection. Seafarer was chosen as the donor parent and selection was practiced in backcross-selfed generations. The rationale for this approach was inferred from the good yields obtained from Seafarer and Sanilac navy beans when grown in several areas of the world (e.g. Chile, Peru, South Africa, Eastern Africa, Australia, Rumania). An open, erect, small~leafed, determinate plant allows greater sunlight penetration into the internal leaf canopy and as a result yields can be expected to approach a maximum. All of these traits may be related to the capability of the plant to be better adapted to diverse environmental condi- tions. A better understanding of the traits under study may help in the development of high yielding varieties with greater genetic diversity, that can be grown in different environments. Recently developed wheat and rice cultivars with strong, short, erect stems, narrow vertical leaves, ferti— lizer responsiveness and daylength neutrality are now suc- cessfully grown. These varieties have a narrow germplasm base, but the traits were recognized as having wide adaptive value and their development was considered a breakthrough due to their high yielding capability and wide adaptation. Bean varieties in Central and South America have received some genetic improvement but many are narrowly adapted since there is variation in consumer preferences and local climatic conditions. There is a great natural diversity present in bean varieties in Central America (the bean center of origin) and some types, such as Rico—23, show wide adaptation. Little is known of the factors that con— dition adaptation in beans and more research in this field is needed to identify traits that may contribute to it. REVIEW OF LITERATURE Stern (1970), in a review of the meaning of adapt— ation, discussed the controversy over its definition and described "adaptation" in general usage as the relative state of success of the organism in dealing with its en- vironment. An adaptive trait is any characteristic of an organism or population that causes its possessors to be, on the average, at a higher level of "adaptedness" than would occur in its absence. Plant breeders use the genotype-environment inter— action as a measure of adaptation, to develop better vari— eties. The variety x location or variety x season inter- action mean square was the first measure of adaptability. They provided a useful estimate of stability of varietal performance but subsequent regression methods gave a better evaluation of the response of a variety to different en- vironments. Yates and Cochran (1938) presented a statistical model to analyze groups of experiments and to evaluate varieties grown at several locations. Finlay and Wilkinson (1963) and Eberhart and Russell (1966) developed a re- gression technique with barley and maize respectively, and stressed the plant breeding applications and adaptational 4 aspects. In their analyses, adaptation was measured by the average yield of the variety and by a regression coefficient. Finlay and Wilkinson used the location mean as a measure of the environment. Eberhart and Russell used the environ— mental index, calculated as the location mean minus the grand mean, as the environmental measure. In both methods a linear regression of the variety yield, at all locations, on the respective location mean or environmental index was calculated for each variety. The regression coefficient was then an average measure of the response of the genotype to the effect of the environment. Perkins and Jinks (1968) developed a genetic ap- proach to specify the contributions of genetic, environ- mental and genotype-environmental interactions to the gener— ation means and variances of multiple lines and crosses. Knight (1970) noted that the success of Finlay and Wilkinson's method depended largely on the linearizing nature of the logarithmic transformations employed in the analysis. Hardwick and Wood (1972) indicated that "where the regressions are found to account for a substantial part of the genotype-environment interaction variance, this empiri- cal approach to the problem of interactions has proved to have considerable value". They showed that there is a bias in the estimates of the coefficients of regression on the environmental mean and proposed an alternative method using multiple regressions on the levels of environmental variables. Allard and Bradshaw (1964) defined any variety that performs well in different environments as being "well buf— fered". Buffering is most important in heterogeneous pop— ulations, but it also takes place in homogeneous populations (hybrids, for example). Individual buffering occurs in the latter group, in addition population buffering is also pre- sent in the former, caused by interactions between different genotypes. Both types of buffering are reflected in the genotype-environment interaction. Allard and Bradshaw con- cluded that "genetic diversity either in heterozygotes or in mixtures of different genotypes often leads to stability under varying environmental conditions" and suggested that heterozygous and heterogeneous pepulations are more likely to show small genotyperenvironment interactions. Finlay and Wilkinson (loc.cit.) measured several morphological and physiological characteristics in their study of barley adaptation in Australia. Maturity was the most significant character; very early varieties appeared to be specifically adapted to lowaielding environments. They used the mean yield of 277 genotypes at each of several 10- cations during three years as the environmental measures. The yield of each genotype at several locations was regressed on the respective site mean yields, to calculate the re— gression equation. The regression coefficient was defined as an adaptation parameter. A variety with an average stab— ility had a regression coefficient of unity, those with a coefficient greater than unity were adapted to high yielding environments and those with a coefficient smaller than unity were adapted to low yielding environments. The other adapt- ation parameter was the variety mean yield over all environ~ ments. Eberhart and Russell‘s model, applied to corn data, made use of two "stability" parameters to describe genotypic performance in a series of environments. The first parameter is similar to Finlay and Wilkinson's, a regression coeffv icient of genotypic values on an environmental index (site mean minus the grand mean) over all environments. Their second parameter is the squared deviations from regression, a function of the number of environments. Significance tests for both parameters are presented. Camacho (1968) studied the genotypic performance (yield) of 26 homozygous bean lines, planted twice a year for a period of four years, in the same location in the Cauca Valley in Colombia. The variety x season variance was esti- mated for each line by successively omitting it from the analysis and estimating its contribution to the total variety x season variance when all lines were included in the an- alysis. The regression coefficient of variety means on the environmental index, according to Eberhart and Russell's model, was the other adaptation parameter estimated for all lines. Estimates of the regression coefficient for some lines were significantly different from the mean b=1.0. The interaction variances were also similar, within the two bean groups studied. Some genotypes showed adaptation to the unfavorable conditions of the first planting season and others were adapted to the conditions of both planting seasons. Parsons and Allard (1960) reported that seed size in segregating populations of lima beans (Phaseolus lunatus) was strongly affected by an apparently stable environment when planted over several years, if interplant competition was taking place. Small variations in seed size, a complex character, depend on interactions between genotype and en— vironment. When competition between plants (at commercial seeding rates) was eliminated, seed size was constant over years. Emerson (1916) studied the inheritance of plant height in beans. A simple 3:1 ratio was found in the F2 generation, the allele for the pole type (indeterminate) being dominant to the allele for bush (determinate) plants. Bliss (1971) studied the inheritance of growth habit and time of flowering in seven bean cultivars. He confirmed that indeterminate plant types were dominant over determinate and controlled by either a single gene or by two epistatic genes. Determinate plants flowered earlier than indeter— minate in segregating generations. Coyne and Mattson (1964) crossed three bean vari- eties to study the inheritance of time of flowering. Earli- ness was found to be dominant or recessive depending on the parental variety. They suggested that the trait was cen— trolled by three interacting genes. A late flowering plant must be homozygous dominant at the A and B loci regardless of the condition at the C— locus, or heterozygous at these loci in the presence of C, or homozygous dominant at one and heterozygous at the other in the presence of C. Duarte (1961) reported that number of leaflets per plant (N) and size of the leaflet (S) are influenced by dominance and additive genetic systems, respectively. Re— current selection methods were used to produce lines with high, intermediate and low levels of expression of both traits and were successful after two cycles of selection, in populations derived from crosses between two bean vari— eties. A genotype—environment interaction was suspected, since the Colombian, large—leafed, kidney type was prefer— entially recovered from a mixed population when planted in its country of origin (Palmira and Medellin, C01.) and the small-leafed navy types were also preferentially recovered when the mixed population of lines was planted in Michigan (Duarte, 1966). MATERIALS AND METHODS Eight Central and South American varieties were each crossed with Seafarer, a commercial Michigan navy bean (See Table 2 for the characteristics of each progenitor variety). The F1 hybrids were backcrossed both to Seafarer and to their respective Latin American parents. It was not possible to obtain all backcrosses due to the large number of plants involved. A variable number of backcross plants in each population was selfed one generation. Each F1 was selfed and the F2 plants grown to produce F3 seeds. All crosses, backcrosses and selfs were maintained in the Plant Science greenhouses at East Lansing during 1971 and 1972. The second backcross self generation was planted at the Inter- American Institute for Agricultural Sciences, Turrialba, Costa Rica, in February of 1972. The third selfing of all backcrosses took place at the Michigan State University Bean and Beet Research Farm, Saginaw, Michigan, during the summer of 1972. The selection of lines combining the traits under study was done in this generation. Fifty—five single—plant selections were chosen from the backcross selfed populations, or, in a few cases, from F3 families. These 55 selections comprised six different selection classes combining alternative states of the three traits 10 ll (determinate or indeterminate growth habit, early or late maturity, small or large leaflets) (Table l). A selection class is defined as a group of genotypes with a specific combination of the traits studied (e.g. det— erminate, early, small-leaflets; determinate, late, large— leaflets; etc.). The main objective of selection was to re~ cover the parental seed type of the Central and South American varieties, through backcrosses, and to obtain from Seafarer combinations of the presumed adaptive traits under study. Certain selection classes were not obtained for some pOpulations, eg.g vine types did not show up in crosses from two bush types, neither did large leafed types from crosses among two small leafed parents. Seeds of the 55 selections were planted in the green- house and selfed for one to three additional generations to increase the seed supply for testing purposes. An 8 x 8 lattice experiment with three replications, including the nine parental lines and the 55 selections, was planted at three locations, two of them in South America; La Molina, Peru, in March 1973 and Palmira, Colombia, in July, 1973, and the third one at East Lansing, in June, 1973. The effect of the environment on yield and yield components was measured from data collected at the three locations to obtain an estimate of environmental influences on the genotypes and on the different selection classes and to determine if Seafarer germplasm improved the yield and general performance in the selected lines. TABLE l.——Characteristics of the Selection Classes 12 Selection Class 1 Growth Maturityl Habit — Determinate Early Determinate Early Determinate Late Determinate Late Indeterminate Early Indeterminate Late Leaflet Size Small Large Large Small Small Large 2 5 25 34 52 l 17 53 4 36 6 30 51 3 48 8 Entry Numbers 16 28 41 59 2 23 54 14 46 9 37 7 50 19 18 31 43 60 10 32 55 21 49 13 38 12 56 24 22 33 47 61 11 35 63 27 57 20 42 40 58 15 39 29 64 26 44 45 62 lEarly=entries harvested before 95 days from planting at East Lansing, or before 77 days at Colombia (pop- ulation means at both locations). 35mall=entries with apical leaflets smaller than 63.5 cm2 (population mean at East Lansing). The populations in which selection was practiced were: (Seafarer (Seafarer (Seafarer (Seafarer Progenitors Col.l-63)xCol.l—63 Col.l-63)xSeafarer 27-R) x 27—R 27—R) x Seafarer Number of plants Number of selfed per BC or selected Fl population entries* 4 9 4 4 3 5 9 4 13 TABLE lr_Continued (Seafarer x R.O.) x R.O. 10 6 (Seafarer x R.O.) x Seafarer 2 3 (Seafarer x C.N.) x C.N. 15 5 (Seafarer x C.N.) x Seafarer 4 l (Seafarer x Sang.) x Sang. 4. 5 (Seafarer x Sang.) x Seafarer 6 6 (Seafarer x Lib.) x Liborino 2 l Seafarer x Cocacho 8 3 Seafarer x Canario 10 1 Seafarer x Liborino 7 l Seafarer x Sangretoro 12 1 Total 55 * Selection was practiced on BClS2 or F3 generations. An analysis of adaptation for all genotypes and for every selection class and trait was carried out according to the models of Finlay and Wilkinson and of Eberhart and Russell, with the data collected at the three locations. The variables measured were yield (gms per 1 mt row), number of pods per plant, number of seeds per pod and seed weight (measured from the weight of 100 seeds). Other variables recorded at East Lansing only were: days to first flower, days to last flower, length of the blooming period, days to first ripe pod, days to harvest, harvest index, leaflet size, l4 leaf area and leaf area ratio. All data were statistically analyzed on the CDC 6500 computer. It was not possible to perform the lattice analysis of the data collected at Palmira, due to several missing plots, so an analysis of variance for data with unequal cell frequencies was made, to obtain the within—line variance as an estimate of the error. All data were transformed to common logarithms, to reduce or avoid positive correlations between the mean and the variance and to normalize the distribution of values around the mean. The regression coefficient, symbolized by b1' and the variety mean over all locations are the two parameters considered to evaluate the adaptation of every line, as ori— ginally proposed by Finlay and Wilkinson (1963). Eberhart and Russell (1966) proposed another parameter, the squared deviations from regression, symbolized by s: and calculated as the deviation mean square minus the pooled error. ‘The location error mean squares in the data analyzed for this study were heterogeneous, so neither the pooled error nor the squared deviations from regression values could be estimated. The deviations from regression mean squares were calculated for every line, as an estimator of adaptation. The individual varietal or line yield and yield com— ponents values at each of the three locations were regressed onto their respective environmental index, calculated as the 15 location mean minus the grand mean. These indexes, being actually deviations from the grand mean, provide a useful gradation of the environments. The regression coefficient was estimated as the cor— rected sum of crossproducts of the variable means at each location and the environmental index, divided by the sum of squares of the environmental indexes, as follows: 2...I J 13 J b = XI? 3 J Yij = location mean of the ith variety at the jth environment. Ij = environmental index (location mean minus the grand mean. For a given variable, the environmental indexes are the same for all lines. The population mean has a regression coefficient of b=l.0. The analysis of variance for every variable when the data were collected at three locations was performed using the model proposed by Eberhart and Russell. The replication x lines mean square was the measure of eXperimental error. The respective pooled sums of squares and degrees of freedom provide an estimate of the pooled error, to be used in the analysis over locations. 6 l mummm ooa mo unmflmz m NED ca .ucMHm mom mumawmma M0 quESG x when umammoa "noun Moon m N EU CH umammma HMCHEHmu mcu mo down mmmum>¢ H dude mhma mcwmcdq when do ommdm e when: somam own xumo own emu unmflq ooh xumo cacao unmfla co G30HQ nmflo town xnoo emauuos mmflmmummsm soaam» unmflum om mm vm 5N Hm em om vm ow hm moa mm mm so mad ova vva hva HOHOU usmwwz umm>umm on mama comm comm mm mv wm mm mm om mu mo an mva wo.ov mmNN 0H.om mmNN hm.mm evwa mv.mm mmvm hm.m> wama om.mm hmom mm.moa ommm mN.Hm mmhv NN.hm umzoam umuflm on mmmo Nomad HmNHm swam mcfl> swam mcfl> gmsm swam :msm msH> mcd> uflnmm Emma umammmq ausouo cmmwnofls MUHm mumou MHMEmDMSU mmudocom nose>amm Hm mensoaou mansoaoo Show Show GHmHHO nonemmmm cummz cymmsmfiou ousomo cocflm Nlmmza COHUUGHOU mlhm ououmumddm oafluonflq onomoou oaumcmo mumflnm> .«mmflumflum> nouwcmmoum mcu mo mowumwumDUMH8£U|t.N mqmme RESULTS AND DISCUSSION The experimental errors from the analyses of vari- ance for each variable at each location are presented in Table 3. The errors and coefficients of variation for all variables measured at Colombia were larger (except for the seed size error) than those measured at East Lansing or Peru. The calculated within—line error in the data from Colombia was unaccountably large when compared to the repli— cations x lines error from East Lansing and Peru. The environmental index, calculated as the location mean minus the grand mean, shows the relative superiority of environments (locations). The highest yielding environ— ment was East Lansing and the lowest yielding was Peru. The best environment for the eXpression of high number of pods per plant was Colombia, and the best for the expression of greater number of seeds per pod and heaVier seed weight was Peru. No environment was consistently better for the expression of all variables. The performance of all geno- types whcr grown at the different locations was not the same, due to interaction with the environment. This geno- type environment interaction is analyzed separately for each variable. 17 18 TABLE 3.--Experimental error and coefficients of variation for yield and yield components for 64 bean genotypes, measured at three locations in North and South America. Location df Error Coefficient Mean Mean Mean of Variation (Log) (Actual) __ Square Log pods per plant East Lansing 126 .01373467 11.87 .9905 9.78 Peru 126 .00952182 9.85 .9821 9.59 Colombia 104 .02409950 21.49 1.0982 12.54 Log seeds per pod East Lansing 126 .00316053 8.52 .6341 4.31 Peru 126 .00347501 8.83 .6722 4.70 Colombia 104 .01615319 28.11 .5684 3.70 Log (seed weight + 1.0) East Lansing 126 .00014455 10.33 .1118 .2934 Peru 126 .00007875 7.92 .1120 .2942 Colombia 104 .00011299 19.15 .1097 .2874 Log yield (gms/lmt) East Lansing 126 .02233084 6.99 2.0760 119.1 Peru 126 .01490687 6.13 1.9762 94.7 Colombia 104 .06261118 15.75 2.0499 112.2 Pods Per Plant The results of the analyses of variance for the three locations are shown in Table 5. The F-test of lines DS/Deviations MS measures the differences between the means of different lines. The calculated value F=5.08 is sign- ificant at the 1% level. The F—test of Lines x Environments (linear) MS/Deviations MS measures the genetic differences between varieties for their regression on the environmental 19 TABLE 4.--Location means and environmental index for yield and yield components when measured at three locations in North and South America. Variable Grand Location Sum Mean Env. Mean ____ Index Log10 1.02360 1 63.38984 .99047 -.03313 Pods/plant 2 62.85708 .98215 —.04145 3 70.28377 1.09818 .07458 Loglo 0.62489 1 40.57994 .63406 .00917 Seeds/pod 2 43.02209 .67221 .04732 3 36.37767 .56840 —.05649 Log10 .11117 1 7.154670 .11178 .00061 Seed weight +1.0 2 7.170509 .11204 .00087 3 7.020089 .10969 -.00148 Loglo 2.035 1 132.91781 2.07603 .042 Yield 2 126.54999 1.97625 -.058 3 131.29224 2.04991 .016 *Location 1 2 3 East Lansing La Molina, Peru Palmira, Colombia index (the significance between the different regression co- efficients). The calculated F value (2.15) was significant at the 1% level. The pooled error could not be estimated; the error mean squares at each location were heterogeneous and it was not possible to pool the respective error sums of squares. This pooled error would normally be used to test the signi~ ficance of the squared deviations from regression of every variety, to determine if those deviations were significantly different from zero. 20 TABLE 5.-—Analysis of variance of pods per plant (on logA0 scale) or 64 bean genotypes t grown i hree different locations. Source SE. Mean Square E Total 191 Lines 63 .057950 5.08** Environments and Lines x env. 128 .021967 Regressions Env. (linear) 1 .536272 Lines x env. (linear) 63 .024523 2.15** Deviations from regression 64 .011414 Pooled errorl ** Significant at the 1% level 1 Could not be estimated due to heterogeneity of the respective location errors. The variance component due to deviations from reg— ression (.011414) is 46% as large as the linear regression variance, which indicates that the deviations are accounting for a substantial part of the lines x environments variance. Regression coefficients and deviation mean squares are shown in Table 6. A coefficient of 1.0 associated with a large mean and a deviation mean square as small as poss- ible (close to zero) are indicative of superior genotypic stability and general adaptation. The ideal variety with general adaptation should have a high mean (yield or yield component). Regression coefficients significantly greater 21 TABLE 6.--Pods per plant stability parameters and average of every line when planted at three different locations. Line Log Deviation Mealll0 mean Pods/plant b square 1 1.0512 0.98 .032655 2 0.9385 0.57 .002347 3 1.0582 1.17 .000908 4 1.0110 2.60 .017792 5 1.2066 0.18* .000032 6 1.1946 2.55* .000006 7 0.9886 1.66 .016827 8 1.0975 0.68 .018342 9 1.1995 1.89 .002077 10 0.8042 el.86 .002000 11 0.8597 ~0.80 .008325 12 1.0218 —0.18 .000194 13 1.0857 1.33 .000539 14 0.8807 1.17 .001573 15 0.7777 F0.98 .010375 16 1.2601 0.37 .025067 17 0.9134 0.31 .008594 18 0.9692 1.89 .062450 19 0.9893 1.43 .000121 20 1.2982 3.52** .000002 21 1.0820 —l.98 .010345 22 1.1706 1.00 .002495 23 1.0627 1.92 .015834 24 1.2421 2.72 .004926 25 1.0322 3.66 .063448 26 1.1533 ~0.71 .000365 27 0.7607 -4.43 .006561 28 0.9644 4.50 .048401 29 1.0729 —0.01 .002774 30 1.0947 0.89 .002573 31 1.1396 0.59 .000373 32 1.0563 1.14 .002742 33 1.1532 0.32 .001426 34 0.9717 3.02 .012551 35 0.8134 -1.20 .001758 36 0.7830 -2.64 .001264 37 1.1865 3.16* .000024 38 1.0451 0.56 .007627 39 0.8948 1.94 .000939 40 1.1121 1.44 .003462 41 1.1734 1.28 .004937 42 0.8280 2.56 . [.013910. TABLE 6.--Continued 22 43 1.1126 1.15 .002973 44 1.0921 1.98 .004848 45 1.1149 —0.89 .000381 46 0.9010 0.34 .002340 47 1.3320 1.40 .005385 48 1.0756 1.32 .001296 49 0.7143 5.30 .082427 50 1.0391 0.27 .001545 51 1.1251 2.37 .000823 52 1.0621 0.50 .000833 53 1.1294 1.32 .000943 54 0.9326 0.44* .000030 55 0.8209 0.97 .008108 56 0.9821 0.04 .007450 57 0.9754 1.81 .007007 58 0.8706 4.14 .106517 59 1.0752 2.26 .036239 60 1.0165 2.60 .024338 61 1.0237 0.44 .000124 62 0.9001 ' ~2.18 .002895 63 0.8523 90.57 .001994 64 0.9635 0.78 .012835 Mean 1.0236 1.00 .011414 Std. Dev. 0.1839 1.71 * Significant at the 5% level ** Significant at the 1% level than 1.0 indicate below average stability, or varieties adapted primarily to high yielding environments. Coeffi- cients smaller than 1.0 indicate above average stability, or varieties that perform the same in favorable or unfavor— able environments. This is usually associated with a low mean, since the varieties are not able to exploit favorable environments. Varieties with large positive b—values are characteristic of genotypes that perform poorly in unfavor— able conditions, but that respond better as the environment 23 improves. On the contrary, those varieties with large negative coefficients perform relatively better in poor environments. The overall mean for pods per plant was 10.56. Seafarer (entry 47) had the highest mean (21.48 pods per plant), a coefficient of 1.40 and a deviation mean square (.005385) a little more than 50% smaller than the average deviation (.011414). The mean number of pods per plant for Seafarer increased as the environment became increasingly better, its mean at every location being higher than the location mean over all varieties. There was a large variation in the values of the regression coefficient, as low as b=4.43 (entry 27) and as high as 5.30 (entry 49). This variation is believed due to the small number of locations tested, which resulted in a large deviation from regression for most entries. Whenever a particular location mean for a line was noticeably diff- erent from the means at the other two locations, a large positive or negative b_value resulted if the environmental index at that location was highly positive or negative. With data from more locations the effect of one strikingly low or high location mean for one line would be balanced by the weighting effect of the other environmental indexes. The significance of every regression coefficient was calculated with a t-test, to identify those coefficients significantly different from the population mean value of 24 049 5.0 : .20 w 059 .18 .30 50* ~54 47 41 4'-‘0 .‘ .14 fl 6: 20 l' 31 W '7 .35“ .q 7 5 L0 ~ M x“ .4, . 1 55 .‘ a t E o“ 3 .. a. E «-.,.,"T a 1': ,5 4‘ u. .5; '50 5. g 0 i on. '1‘ u z 0" ‘2‘ a ‘w L J, '4’ g 65' 0.9 b ‘92 -20. °'° 6" an 06: 46 .10b 0 "mum. van-nu .40. -27 .u C A L L 4 A . an as 0.9 L0 M L2 L5 L4 LOG” MEAN PODS PER PLANT Figure 1. The relation of pods per plant and the regression coefficient for 64 bean lines grown in three difv ferent environments . 25 1.0. Most coefficients were nonsignificant at the 5% level, due to the large standard error of the coefficients. The large deviations obtained explain the nonsignificance of the coefficient values. Lines with coefficients significantly different from unity had small deviations from regression and consequently a small standard error. The distribution of the coefficients, when plotted against the mean, for each genotype appears in Figure 1. There is a grouping of points along the mean b=l.0. The only variety that approaches the ideal of b=l.0 and high mean is Seafarer (entry 47). This reinforces the original assumption that the Seafarer type would be widely adapted. Seeds Per Pod The analysis of variance for seeds per pod set out in Table 7 shows significant F values for lines and lines x environments. The differences between lines for the mean number of seeds per pod and for the regression coefficient values were significant at the 5% level. The variance due to deviations from regression was approximately 40% smaller than the regressions (lines x environments) variance. The pooled error was not calculated because the (lines x repli- cations) errors for locations were heterogeneous. The regression coefficients, deviations mean squares and means for each line are presented in Table 8. The co— efficients varied, from a minimum of rl.14 to a maximum of 26 TABLE 7.—— Analysis of variances for seeds per pod (on logilio scale) of 64 bean genotypes t grown i hree different locations. Source ' df_ Mean Square 5 Total 191 Lines 63 .020655 6.48** Environments and Lines x env. 128 .007241 Regressions Env. (linear) 1 .217495 Lines x env. (linear) 63 .008025 2.52** Deviations from regression 64 .003185 Pooled errorl ** Significant at the 1% level Could not be estimated due to heterogeneity of the respective location errors. 4.41. Departures from the mean b=1.0 were not as great as for pods per plant. Some coefficients were significantly different from unity. The relationship between coefficients and means for each line, as seen in Figure 2, follows the same pattern as that for pods per plant, most values being grouped around the mean b=1.0 line. The highest value of mean seeds per pod is 5.85, for line 43, with a coefficient of b=0.15. Line 8 has the next highest mean of 5.64 seeds per pod, and a coeffiw cient of 0.92, showing excellent stability and general adapt— ation. The lowest mean value is 2.87, for line 59, with a co- 27 TABLE 8.——Seeds per pod stability parameters and average of every line when planted at three different locations. Line Log b Deviation MeaAo Mean Seeds[pod _’ Square 1 0.6852 0.94 .000895 2 0.5432 1.45 .003802 3 0.7074 0.19 .000942 4 0.5319 0.32 .003217 5 0.6667 1.60 .001650 6 0.4907 1.42 .003628 7 0.7299 1.26 .000430 8 0.7512 0.92 .000054 9 0.5832 1.04 .001728 10 0.5232 3.91 .008905 11 0.5759 1.71 .000180 12 0.6906 1.69 .000274 13 0.5543 0.96 .008696 14 0.5503 1.07 .002656 15 0.6515 1.29 .000110 16 0.6513 1.64 .001294 17 0.4758 4.41 .007588 18 0.7388 0.49 .001038 19 0.5581 0.80 .000076 20 0.5970 0.94 .008199 21 0.5513 3.49 .011586 22 0.7195 0.51* .0000003 23 0.5495 3.03 .014810 24 0.6791 0.61 .007335 25 0.7090 —0.17 .000069 26 0.5450 0.05 .015995 27 0.5379 1.16 .000406 28 0.7014 0.55 .000021 29 0.5017 1.64 .000105 30 0.5661 0.33 .005349 31 0.6790 0.12 .001876 32 0.5938 1.51 .000916 33 0.6587 1.45 .003792 34 0.7217 -0.30 .002051 35 0.6925 0.96 .000006 36 0.4918 4.07* .000484 37 0.6060 0.79 .002165 38 0.5766 1.80 .001043 39 0.6644 0.52 .000160 40 0.6887 0.64 .000440 41 0.6765 0.43* .000001 42 0.4726 —0.10 .019049 28 TABLE 8.-—Continued 43 0.7674 0.15 .000694 44 0.5116 1.55 .003237 45 0.7293 0.89 .000358 46 0.6498 0.05 .000231 47 0.7014 0.08 .000238 48 0.7087 1.92 .000541 49 0.5224 -0.36 .028547 50 0.7104 —0.21 .000541 51 0.5742 0.49 .001400 52 0.7221 1.10 .001100 53 0.7058 0.77 .001124 54 0.6176 0.13 .001111 55 0.6000 0.51 .001124 56 0.7206 0.54 .000305 57 0.6490 -l.14 .000992 58 0.7039 0.63 .009270 59 0.4576 0.96 .001614 60 0.7015 0.96 .001614 61 0.5736 0.22 .003406 62 0.6159 1.04 .000753 63 0.6326 1.64 .002657 64 0.5883 1.01 .001196 Mean 0.6249 1.00 .003185 Std. Dev. 1.1080 1.03 * Significant at the 5% level efficient of 0.96, indicating average stability but poor adaptation to all three environments. It is very interesting that Compuesto Negro (line 58) has a b-value near 1.0 and also a high mean, and that several of the other good lines (1,8,24, 52, 53) are from crosses of Seafarer and Compuesto Negro, and backcrosses to Com- puesto Negro. All the lines from backcrosses to Sangretoro, Liborino and Rifibn Oscuro show poor adaptation (low means, widely different coefficients) but the lines from back— crosses to Seafarer show a better adaptation (high means, or; 29 ab 40 no 16- l— 2.0 b 4.. '6 68 U '114‘ 0' .‘3 4‘ o, "'- 1: .5 '3 '52. '55 ‘3 0" ' ca 2 .5CD " 4'2 igto. wwuwwmmm .thI-aowooxn wwwu QOU'l Bamboo NHO‘O Loglo Mean Seed Weight +1.0 .0890 .1087 .1123 .1115 .0824 .1189 .1044 .0875 .1239 .1300 .1571 .0995 .1163 .1745 .1021 .0831 .1295 .0784 .1529 .0979 .0941 .0893 .1031 .0904 .0809 .0964 .1292 .0827 .1346 .1264 .0906 .1382 .0724 .0832 .1027 .1234 .1064 .1264 .1138 .0990 .0958 .1153 b 6.54 ~1.36 ~0.55 1.38 1.66 ~3.94 ~2.02 1.97 —7.89 19.87* 3.93 ”5.60 —2.97 19.15* 5.71 -2.04 13.01 ~4.84 1.91 —1.34 -6.74 -4.95 11.29* -0.60 ~2.32 -0.26 1.00 -6.55 -4.23 —2.84 -4.62 -5.60 -4.27 -5.19 5.85 -2.82 6.10 4.28 4.50 -3.85 Pl.43 -5.54 Deviation Mean Square .00000015 .00021347 .00000280 .00017900 .00001216 .00000210 .00000224 .00000176 .00019039 .00000895 .00062610 .00004627 .00000088 .00000010 .00001127 .00001524 .00058507 .00001289 .00013516 .00003118 .00001222 .00001789 .00001165 .00000138 .00000272 .00000070 .00001429 .00000010 .00014896 .00000696 .00004723 .00021027 .00006796 .00000295 .00006227 .00000326 .00002952 .00003796 .00000023 .00000210 .00015247 .00001741 TABLE 10.--Continued W .00001988 43 .0811 —3.08 44 .1261 1.82 .00028716 45 .1041 -l.63 .00007047 46 .1537 10.61 .00012602 47 .0811 —2.79 .00000016 48 .1090 -3.60 .00002163 49 .1635 21.81** .00001755 50 .1028 ~7.20 .00003851 51 .1052 —2.32 .00012272 52 .0937 7.31* .00000551 53 .1005 2.42 .00014372 54 .1388 1.62 .00022347 55 .1500 17.71 .00076451 56 .1085 —2.21 .00000924 57 .1380 5.16 .00014695 58 .0956 3.30 .00000755 59 .1261 2.68 .00000476 60 .0825 —5.20 .00001595 61 .1217 —3.86 .00002301 62 .1402 1.34 .00038618 63 .1017 4.64 .00002423 64 .1364 1.65 .00019447 Mean .11117 1.00 .00009180 Std. Dev. .0250 6.71 * Significant at the 5% level ** Significant at the 1% level X1319. In the analysis of variance of yield over locations (Table 11) the calculated F value was highly significant for differences between line means. The deviations—from—regres- sions variance (.0352) was larger than the variance due to regressions (.0214), so the calculated F value for differ- ences among regression coefficients is nonsignificant. The interaction lines x environments variance was partitioned into the regressions variance and the deviations from re- gressions variance. In the case of yield, the deviations 34 TABLE ll.——Analysis of variance for the yields (on loglo scale of 64 genotypes grown in three different locations. Source df Mean Square 3 Total 191 Lines 63 .05999830 1.70** Environments and Lines x env. 128 .03630887 Regressions Env. (linear) 1 .342075 Lines x env. (linear) 63 .021426 0.61 Deviations from regression 64 .03524831 Pooled errorl ** Significant at the 1% level Could not be estimated due to heterogeneity of the respective location errors. account for a large part of the interaction. Data from a higher number of environments would have reduced those de— viations. Some of the regression coefficients were signifi- cantly different from unity; those lines have a small de— viation from regression in each environment. Lines 10, 17, 21, 27, 34, 36, 42 and 49 had deviation mean square values three or four times larger than the mean deviation (.035248) the deviations of those lines account for 33% of the total deviations from regression. All those lines had a very low 35 TABLE 12.—-Yie1d stability parameters and average of every line when planted at three different locations. Line Log 0 b Deviation Mean iield Mean ____ gm/lm _ Square 1 1.9671 F2.62 .046262 2 1.8090 ~1.49 .021793 3 2.2125 0.50 .018286 4 1.9173 ~0.95 .014132 5 2.1563 0.60 .009057 6 2.1024 1.15 .037997 7 2.1042 2.97* .000397 8 2.1364 ~1.05* .000001 9 2.2323 1.44 .028140 10 1.7786 ~0.66 .124840 11 2.0246 2.81 .043772 12 2.0497 0.76 .018728 13 2.0225 0.77 .012497 14 2.0791 0.23 .008404 15 1.7927 0.70 .099011 16 2.1564 —0.48 .000391 17 1.8202 2.03 .146146 18 1.9785 4.41 .021108 19 2.1072 0.94 .001056 20 2.1500 0.28 .012653 21 1.9307 —0.61 .120003 22 2.2197 0.37 .017722 23 1.9471 2.68 .081140 24 2.2111 1.63 .031965 25 2.0091 6.40 .088392 26 1.9845 -0.57 .000429 27 1.7700 -0.18 .100389 28 1.8977 5.63 .056069 29 2.0476 1.49 .003248 30 2.1382 1.71 .019008 31 2.1566 0.99 .018778 32 2.1476 0.03* .000084 - 33 1.9818 2.09 .000717 34 1.9659 4.05 .124409 35 1.8587 -0.66 .085483 36 1.6812 —l.49 .131564 37 2.0920 1.69 .000504 38 2.0906 -0.33 .012572 39 1.9472 1.45 .000342 40 2.1733 0.60 .022515 41 2.1945 1.77 .014426 1.7138 —l.40 .118536 H b N TABLE 12.-—Continued 43 2.1589 2.56 .008172 44 1.9899 1.59 .001570 45 2.2455 -0.61 .002844 46 2.1390 1.36 .000940 47 2.2808 1.80 .047654 48 2.1230 0.16 .004736 49 1.8145 v3.62 .102446 50 2.1226 0.84 .001201 51 2.0701 0.78 .078597 52 2.1314 0.10 .009359 53 2.2613 0.35 .031231 54 2.0814 1.37 .001685 55 1.9762 4.00 .007989 56 2.0891 0.22 .000011 57 2.1282 1.93 .047791 58 1.8675 7.06 .000018 59 1.9806 0.40 .055967 60 2.0049 4.40 .027583 61 2.0437 2.36 .016254 62 2.0370 —1.96 .062033 63 1.8722 0.44 .021641 64 2.0787 'r1.42 .000173 Mean 2.0352 1.00 .035248 Std. Dev. 0.2100 1.99 * Significant at the 5% level ** Significant at the 1% level yield in one location, usually at Colombia, as the result of severe virus infection (bean common mosaic virus). The distribution of coefficients calculated for yield and for pods per plant, when plotted against the line mean yield and mean number of pods per plant form a triangular pattern, as first noted by Finlay and Wilkinson. In this triangular pattern, most values are clustered around the mean regression coefficient of 1.0. Lines with coefficients not significantly different from unity and high means (yield 37 or yield component) have average stability and general ad- aptation. Coefficients significantly greater or smaller than unity (excluding negative values) are indicative of genotypes with below or above average stability, respect" ively. The ideal variety was defined by Finlay and Wilkinson as that with a coefficient of 1.0 and a high mean yield. Lines with coefficients significantly larger than unity and low means are specifically adapted to high yielding envirr onments. Lines with coefficients significantly smaller than unity and low means are specifically adapted to low yielding environments. Most coefficients in Figure 4 are grouped around the mean b=1.0. Seafarer (entry 47) has the highest mean (190.9 gm) and a coefficient of 1.80, line 53 has the next highest mean yield (182.52 gm) and a coefficient of 0.35. Line 32 has a b=0.03 value significantly different from unity, a small deviation mean square and a relatively large mean (140.48 gm/lm), showing absolute phenotypic stability and good adaptation to the environments tested. Lines 18, 25, 28, 34, 60 all are from (Se xSa) x Se; and they have large positive coefficients and they cluster in the same area (center top, Figure 4). Line 43, also from this backcross, is positioned closer to Seafarer. Lines 2, 4, 21, and 26 from (Se xSa) x Sa, all have negative coeff— icients and are close to each other and to Sangretoro perfor— mance with a coefficient of 0.78. COEFFHHENT REGRESSWII 38 “’I 044 100 46 0M no . as mu) m- ol? “9* ~23 mm. to. '81 ' 39-31 60~ : 11 .. .7 49' 5‘ CM! 050 20* vs '55 :1 8. 0'9 5 4 ~27 61 is 0 ’ as 3 . '10 .602 ~zof “.45 4' 19:“ 4794.5 5' 43 3"30 0 .. .6-61 PARENTAL VAMETIB O. z . .‘m :4 2,: :2 47. 52 28 «60 ‘M A L _4 A A A ’1 A A A A A j .05 .08 40 .n. . 44 .I6 43 L06“, MEAN ' SEED WEIGHT + H) Figure 3. The relation of seed weight and the regression coefficient for 64 bean 1ines_grown in three different environments. 39 All variables analyzed (yield and yield components) showed large deviations from regression. A summary of those deviations is presented in Table 13. The largest deviations from regression were observed for yield, which were 164.51% greater than the Lines x Environments (linear) regression. Seed weight deviations were the largest among the yield com— ponents, but smaller than for yield. The deviations for seeds per pod were the smallest. According to the magnitude of the deviations from regression observed in the analysis of all variables, yield was most affected by the environment. TABLE 13.-—Mean squares for regressions and deviations from regression, for yield and its components, from the analysis of variance of data coll- ected at three locations. Source df Pods Seeds Seed Yield Per Per Weight .__ Plant Pod Regressions Environments (linear) 1 .536272 .217495 .000214 .342075 Lines x env. (linear) 63 .024523 .008025 .000149 .021426 Deviations from regression 64 .011414 .003185 .000092 .035248 DeViationS x 100 46.54 39.68 61.74 164.51 Lines x env. All variables analyzed (yield and yield components) showed large deviations from regression. A summary of those deviations is presented in Table 13. The largest deviations 40 from regression were observed for yield, which were 164.51% greater than the Lines x Environments (linear) regression. Seed weight deviations were the largest among the yield components, but smaller than for yield. The deviations for seeds per pod were the smallest. According to the magnitude of the deviations from regression observed in the analysis of all variables, yield was most affected by the environ— ment. 41 Performance of Parental Varieties The stability parameters and means of all parental varieties are shown in Table 14, and their performance when yield and yield components are considered is illustrated in Figures 5—8. Seafarer produced the highest number of pods per plant, exceeding all other varieties, with a regression coP efficient of 1.40 and a deviation mean square smaller than the mean deviation. Some varieties were strongly influenced by the environment, having large positive values (Liborino, b=5.30) or large negative values (Cocacho, b=2.18). Cocacho, a Peruvian variety, performed better in its native environ— ment. Canario, the other Peruvian variety, was able to ex- ploit the better environments and showed a positive slope, b=1.43. The responses between the varieties for number of pods per plant are clearly very different, as suggested by differences in the means and regression coefficients. Canario showed the most stable performance, with a value of b-1.43 and the smallest deviation mean square. Col.1—63 produced the largest mean number of seeds per pod (5.37), with a coefficient of b=1.40 and a deviation mean square smaller than the mean deviation. This variety has a short, 42 indeterminate plant type, and early maturity (76 days to harvest). Seafarer showed absolute phenotypic stability for production of seeds per pod (b=0.08), high mean (5.028 seeds per pod) and a low deviation mean square. Cocacho has a re- gression coefficient very close to unity (1.04), a small deviation mean square, but a low number of seeds per pod (4.130). Canario had a stability (b=0.80) close to the average and the smallest deviation mean square. In general, the response of the varieties to the environments, based on seeds per pod, was more uniform than for any of the other variables measured. The seed weight coefficients for the parental vari— eties showed large departures from the mean (Figure 7). Some varieties had high positive values (Rffion Oscuro and Liborino) and others had negative values (Seafarer and Col. 1-63). The coefficient for Cocacho (b=1.34) approximates most closely to unity, but its deviation mean square value is larger than the mean deviation. The responses of the varieties to environmental changes were more variable for seed weight than for other yield components. The coefficients of regression of the varieties for yield (Figure 8) showed large deviations from the mean in some cases. Seafarer had the highest mean yield and a co- efficient of 1.81, while Canario showed average stability (b=0.94), above average yield and a small deviation mean square. 43 m.omH o.mNH 0.0NH m.moa H.5NH N~.v hm.m ma.v mo.m mm.m mo.m Hm.m mm.m 5h.m hm.m CMQE wmmnvo. mmoaoo. vovmoo. thmvo. hmmooo. «aUHwflw mmamoo. mmHHoo. mmbooo. onmmoo. hemmmo. mmmooo. mnoooo. mmmmoo. omHooo. omeooo. Um N pom pom mommm Hm.H vm.o mm.o Hm.m hm.m oo.H Ho.H vo.H mm.o mm.oL mo.o om.o no.H ab.H mN.H mom. wa. mmv. mmv. mum. mm.oa mH.m vm.b N¢.h ma.m mv.HN mh.m 00.5 Av.h wh.m C605 maoooooo. mammaooo. oaoooooo. oammmooo. «mmooooo. .unmflmz,6mmm vavaao. mmmmao. mmmmoo. hammoa. hmvmmo. mmmmoo. Hmaooo. mnmaoo. mNMmoo. hummao. pm N Hanan ”mm mh.ml Hm.a mH.mH mm.m No.ml oo.H mh.o mH.NL va.v om.m ov.H mv.a hH.H om.ol mm.a fig ma «a HH vm mm Housmmmm oflnmcmo OHSOmO GOCflm Mlmmla MINN CHOU com: paduw ououmumCMm onumoou mm oummz oummdeoo mg mg ma ea HH 5 nonfidz muucm ocfluonfla Moummmmm oaumcmu oudomo coch [\1 mlbm wimmla .aou mumfium> .mmfiumflnm> anaconda mo mammE can mumumEdem muHaHQMpmll.vH mqm<9 44 «.moa m.mHH m.moa h.mh N.mm mvmmmo. mnaooo. mmommo. maoooo. mwgmoa. oo.H mv.Hl mm.al 00.5 mm.ml Nam. mmm. Hmm. mom. hmv. .uE A\mfim CH pamflw cmmz pmmm\mEb ca usmwmz comm cmmz .mamoooo.. hvvmaooo. mammmooo. mmnooooo. mmhaoooo. oo.H mm.H vm.a om.m Am.HN vm mm a; cam: pdduw ououmumGMm ocomoov mm oummz oummsmEoo mv cannonfla @wSCHuGOUII.VH mqmde REGRESSION COEFFIC1ENT 45 (35! .13 “F .3. 5.0 - 13.“, 4.0- it.“ 3.0. 01 1! 0n '10 .4; -GI 29. 47 'J5 57 .44 '37 5° 432.4 047 0" 2‘ 5‘ ’4‘ 0‘ .64. '00 D ‘ 0" 45 )P.anK250 4o "5 ‘5‘ 5 '? 4555 HC>5fltz 21 o > . .2 7 238 n .’O 0” 02' .15 ~10 - -4 .5 66""- .2. 064 .10. (362 ' (D a .10 . MRSNTM. vuu HIS 044 “F A A A A A L !A 1.6 L1 u M 2.0 24 1.2. 2-3 1.06,. new we» gun/1m Figure 4. The relation of yield and the regression coefe ficient for 64 bean lines grown in three difs ferent environments. 46 H I-S IQ. - des PIP Plan? Lo... / . . '0' " .030 . co 2- o O o ‘ ' PE. “5 E- .333.” O 4 0 07453 I j l . COLOMBM E N V! RONMENTAL ”IBEX Figure 5. Regression lines showing the response (pods per plant) of nine parental varieties to three difn ferent environments. SE EbS PER POD 106m 47 (13 03 ’O I I / I / / .I ,v .z” V’ I I 4’ 4 V9 ’ 4 hfiey" d6 ‘2 I o O z ” 5‘09 / . 9,0\ \l \0 ’01)" . c. ’ [I C o I o" a o /"' ‘ eat ._ a: k Ml 0.0564; A Jo (j 00;"? «104752 common. '0'“ ‘0' 2' humans pew ENVIRONMENTAL \NDE X Figure 6. Regression lines showing the response (seeds per pod) of nine parental varieties to three dif- ferent environments. 48 .mucmficoufl>cm ucmnmmwflp owns» on mmwumflud> Hmucmuwm one: «0 Aunmfim3 pmmmv oncommmu map mcwzonm mwcwa coflmmmummm .h musmflm xwa 2. draixouslu at: $1184...“ (aioaou 50609 :8. Odooo. O 8009.... 0‘68... 8...... 0300.... h . l . c0. 33... cu (IIIIJWIIIqIJfi.h flat! can -A c on o m .6: ll .0? 9 (L o .13 .. u A— ogrumox- 2 L? 0" O .1 0' '08 I- V «-° 0 6, \0” *~ 0 O 6’ Q L7“ 0* 4‘ 00 .6 c’0 "I" L 4 A J L A 4 j A A A mass ~04 ~03. o .016 0.042. PERU C OLOMIIA ELAN 3““ ENVHLONMEHTAL \NDEX Figure 8. Regression lines showing the response (yield, gm/ 1m) of nine parental varieties to three different environments. REGRESSION COEFFlQIENTS -|-0 -z, Pods r Seeds per seed 4 Pl.“ P06 was“ “dd Figure 9. Fluctuation around the mean of the coefficients of regression, for each selection class, calcun lated for yield and yield components. P0 93 Pill PLANT L 0‘0. 55 L2" o. ‘50 no» 6““ 1% “Liza/1.61199 “"5?" 03 A A L A A L 40.0414: 0.03313 0 .62. -04 «)5 4,7453 pun 5mm counfln ENV|RONMENTAL INDEX Figure 10. Regression lines of the average response (pods per plant) of six selection classes to three different environments. SEED PER POD LOG” 56 as) g’y . . . . -o.asz4q -o.oq ~40: 0 0.0901 004751- cownu A s. uusmr. "no ENVIRONMENTAL leE'X Figure 11. Regression lines of the average response (seeds per pod) of the six selection classes to three different environments. SEED WEWGHT +I-O L 57 .Mr' SILlcflON “ml? ""7 ; #- J}:- .‘R/ 4;. ‘1 s b cb’b.’ s #54 y 5110“ eb‘s’ up" ‘9 59* e15 ggfiaanflflflL_;gL————e—- _ / Han—b:'£--d__d_- gig/.4- - — - sin-“77°” “‘5‘ 5. L=-’°35- do» '0‘. Wfl\\ . "09'4‘ --OOIOO ~30” "006020 0 .00020 .0006! 0.09.7 cannon twosome Puw ENV|RONMENTAL \NDEX Figure 12. Regression lines of the average response (seed weight) of six selection classes to three dif— ferent environments. 58 22w )2 = °" 49'“ I/ C‘v‘ss V '3’} c1‘0 ‘ E ’9‘, ~ 0' 5", 2' \°é 0" c» 5 4 c,“ o.‘ / Q 9% 1'" c» I ’ EC. v; / sit ’ I I I .O ’ 3 “25’ 9.1 1A, , 7' 20L ‘I.’ ‘39 J I 33141710 I ‘35 a .fi5' 9 7' b, m" a” 30 55‘- '4 3 A A A A o A__ ”‘o" -GM *0007- O '0" .04} "RD COLOMBIA E. LANSING ENVtRONM ENTAL 1ND EX Figure 13. Regression lines of the average yield (gm/1m) of six selection classes to three different environments. 59 and seed weight. This relationship (large leaflets—heavy seeds) holds true in the stability analysis of seed weight. When two selection classes were considered as the better classes (in relation to a high mean yield or yield component, and regression coefficients closer to unity) for the variables studied, the traits that are common to both classes are as follows: Variable Best Selection Second Best Traits Common Class Selection to both selec- ClaSS' tion classes Pods per plant D—L—Sl D-E—Sl Determinate small-leaflets Seeds per pod I—E-Sl D—E-Sl Early, small leaflets Seed weight DrLrLl I—LPLl Late, large leaflets Yield I-E-Sl DuL-Sl Small leaflets The only trait common to all variables, except seed weight, is small leaflets. Both traits common to the better classes for seed weight (late, maturity, large—leaflets) are not common to any other two selection classes for the other variables. Summarizing over all traits, it is suggested that determinate growth, early maturity and small leaflets might give the better general adaptation. This is the precise com- bination of traits possessed by Seafarer, which among the parental lines was best adapted. 6O Contribution of Individual Traits to Adaptation Average stability parameters and means for each trait over all variables measured were calculated (Tables 17-19). The regression coefficients for all traits in the stability analysis of yield (Table 17) were closer to unity for determinate (b=1.01) than for indeterminate (b=0.93) plants. Determinate plants showed a very good average yield stability. In the stability analysis for yield, determinate, early, small-leafed plants showed coefficient values closer to unity than plants with the alternative traits, indicating a better average stability. The coefficients for determinate early, small-leafed plants were also closer to unity in the stability analysis of yield components, with the exception of the coefficients for indeterminate plants in the analysis of pods per plant and seeds per pod. The mean yield, the mean number of pods per plant and seeds per pod for early, small—leafed plants are higher than the means of late, large—leafed plants (Table 18). Det- erminate plants produced lower yields, number of pods per plant and seeds per pod. This group is composed of 39% of late, large-leafed genotypes that yielded below the average. 61 TABLE 17.--Average regression coefficients over all entries for each selected trait. Determinate Indeterminate Early Late Small Leaflets Large Leaflets Overall Mean Pods Per Plant 0.89 1.03 0.94 1.15 1.41 0.50 1.00 Seeds Per Pod 0.84 1.04 1.04 0.95 0.79 1.26 1.00 Seed Weight 1.62 —l|44 0.63 1.54 r2.3l 5.00 1.00 Yield 1.01 0.93 1.41 0.40 1.43 0.48 1.00 TABLE 18.--Average values of yield and yield com- ponents over all entries for each selected trait. Trait Pods Per Seeds Per Seed Yield Plant Pod Weight (gm/1m) (gms/seed) Determinate 10.48 4.05 .294 103.6 Indeterminate 10.91 4.92 .283 130.0 Early 10.60 4.56 .269 111.0 Late 10.50 3.76 .326 104.8 Small Leaflets 12.47 4.61 .269 122.0 Large Leaflets 8.63 3.78 .319 94.1 Overall Mean 10.56 4.22 .292 108.4 62 The yield of determinate plants was lower, but close to the over all mean for yield and pods per plant. Determinate, late, large-leafed plants produced heavier seeds. The deviation mean square values for early, small— 1eafed genotypes were smaller for yield and yield components except for the seed weight value for early genotypes (Table 19). The deviation values for determinate plants were larger than those of indeterminate plants, for yield and yield components. TABLE 19.--Average deviations mean square over all entries for each selected trait (in lag measure). Pods Per Seeds Per Seed Yield Plant Pod Weight Determinate .011020 .003578 .00010098 .041022 Indeterminate .012682 .001640 .00005579 .012599 Early .011053 .001748 .00009749 .034526 Late .011943 .005285 .00008349 .036304 Small Leaflets .010176 .002672 .00004614 .026295 Large Leaflets .012909 .003803 .00014690 .046055 Overall Mean .011414 .003185 .0000918 .03548 Summarizing, plants with the determinate, early small-leafed characteristics showed a better adaptation to the three environments tested. This combination of traits would produce a plant better adapted to diverse environments. 63 Performance of Backcross Populations Average regression coefficients and deviation mean squares were calculated for the genotypes selected in back- crosses to Seafarer and in backcrosses to the respective Central or South American variety. The means for yield and yield components in each group were also calculated (Table 20). The genotypes selected in backcrosses to Seafarer showed superior yield and higher number of pods per plant and seeds per pod. The mean seed weight was slightly smal— ler in backcrosses to Seafarer. This indicates that in backcrosses to Seafarer it was not possible to recover com— pletely the large seed weights of some of the varieties, nevertheless the means of both groups (.275 and .293 gm) are close. The regression coefficients and deviation mean squares for reciprocal backcross populations show differ- ences that are not the same for yield and yield components. The average regression coefficient for yield of genotypes from backcrosses to Seafarer is equal to 1.78 as compared to 0.37 in the other backcross populations. The deviation mean squares for yield in backcrosses to Seafarer is 50% smalbr than that of backcrosses to the Central and South American varieties. The regression coefficients for seeds per pod, seed weight and yield in backcrosses to the Central and South 64 mommuocmm Hm u c «c mommuocmo ma u c A vmwhvo. omod m.omH UHOHN maoooooo. ma.~n mom. unmnmz 666m mmmooo. mo.o mo.m Uom Hmm mUmmm mwmmoo. ov.H mvoHN #dwflm Hmm mdom umummmmm vmwmwo. hmco moeoH mmoHNo. mh.H m.hNH UHQHH m¢omoooo. mm.o mmm. mmmOHooo. Ho.o th. Ufimwm3 Ummm mthoo. VN.H mm.¢ whmfloo. mm.o mmcv 00m me mdwmm mommoo. mvoo mh.oa wOMHHo. HN.H mh.HH Madam Hmm mfiom mumnwm cum: mudmmm dam: coflumw>ma m cum: doauww>mn m com: mmwumwum> cmoflumad . . nusom can Hmuucmo on mommouoxomm “mummmmm on mommonoxomm .umanwmm can coaumasmom mmouoxomn comm Mom mommuocmm Umuumamm Add Hm>o mumumEdumm muHHHndum mmwum>¢ll.om Manda 65 American varieties are closer to unity than the coefficients of genotypes from backcrosses to Seafarer. Since the means of the latter group are larger, the genotypes selected from backcrosses to Seafarer show a tendency to be better adapted to all environments. Genotypes from backcrosses to the Central and South American varieties, with low means and coefficients close to unity, show a tendency to be poorly adapted to all environments. Seafarer was the highest yielding genotype, with acceptable stability and outstanding performance among all lines evaluated. It is suggested that the determinate, early, small-leafed complex of traits condition adaptation in Seafarer. In addition, its daylength neutrality may also be part of this complex of adaptive traits. Parsons and Allard (1960) found that particular genotypes in lima bean populations were characterized by unique responses in seed size to changes in the environment. The genotypic performance within a given population and year was consistent over several years, despite changes in popu- lation structure. It was concluded that the population genotype determined the reaction with the environment. The genotype of Seafarer, as a whole, is well adapted to diverse environments. Genotypes selected from backcrosses to Seafarer had a regression coefficient for yield similar to that of Seafarer, but an average yield one third lower than Seafarer. The contribution to adaptation 66 of daylength neutrality, present in Seafarer, could not be analyzed. This factor, in conjunction with the other three traits identified in Seafarer, might be decisive in explain- ing wide adaptation. The complex of the four morphologicalpphysiological traits present in Seafarer is suggested as mainly respon— sible for its wide adaptation. The possibility that some other unidentified traits are also partly responsible for wide adaptation is not excluded. The idea that it is the Seafarer genotype as a whole, that determines wide adapta— tion is not advanced. The Seafarer traits related to adapt- ation are heritable and subject to selection. The breeder, therefore, should be able to develop varieties with the adaptational characteristics of the Seafarer plant type and with any seed types preferred by the consumers. SUMMARY AND CONCLUSIONS The qualitative effects of growth habit, maturity, and leaflet size upon certain parameters of adaptation were determined in a bean population grown at three locations in North and South America. The population consisted of 9 parents and 55 lines derived from crosses and backcrosses (selfed), selected for different combinations of the components named above which were thought to have some influence upon adaptation. The 64 entries were grown in Lima Peru, Cali Colombia, and East Lansing, Michigan in 1973. Measurements taken, from which parameters of adaptation were calculated, included number of pods per plant or unit length of row, number of seeds per pod, 100-seed weight, and yield in grams per meter of row. Parameters of adaptation were determined for each of the 64 entries, and included the following: 1) the line mean over all locations. 2) the regression of the line mean at each location upon the environmental index for that location. 3) the mean squares due to deviation from linear re— gression. It is considered that lines are best adapted that have regression coefficients near unity, small deviations 67 68 from.regression, and high mean values for yield. Most regression coefficients were not significantly different from unity because of large standard errors. In terms of the three adaptation measures, vari- ation among varieties was greatest for seed weight and yield, less for pods per plant, and least for seeds per pod. The adaptation of different selection classes - groups of lines with particular combinations of the traits being studied - was evaluated. The class of lines with determinate plant habit, early maturity, and small leaflets showed better general adaptation, as measured by the para- meters related to pods per plant, seeds per pod, and seed yield. When the evaluation was based on lOO-seed weight, those lines that were determinate, late in maturity, and large leaved showed better adaptation. An attempt was also made to assess the contribution to adaptation by individual traits. Lines that were deter- minate, or were early in maturity, or were small-leaved, showed regression coefficients closer to unity than lines of contrasting characteristics, indicating a better average stability for yield and the components of yield. However, when adaptation was judged on the criterion of yield alone, the outcome was somewhat different. The best yield levels were reached by lines that were early and small—leaved, in- dependent of plant habit. These lines also were superior 69 in number of pods and seeds per pod. Determinate plants produced lower yields and this outcome was attributed to the fact that among the population of lines studied deter- minate plant habit was most frequently associated with late- maturity, large—leaved types. Lines selected from populations derived from back- crosses to Seafarer produced larger yields, higher number of pods and a higher number of seeds per pod. It is concluded that the complex of traits present in Seafarer, namely, determinate habit, early maturity, and small leaves — and daylength neutrality, though this trait was not accurately determined — is mainly responsible for the superior adaptation of this variety, just as these com- ponents were associated most frequently, except for seed weight, with the superior adaptation of certain of the num— bered selections. The data in this thesis could be interpreted as sup- port for the hypothesis that increasing levels of adapt- ation in lines are due to increasing proportions of Seafarer germplasm in those lines, leading to the general hypothesis that adaptation, good or poor, is a function of the total genotype. The implications of this interpretation for breeding for improved adaptation is not promising in that it suggests that the breeder must, to improve adaptation in a population incorporate a large amount of germplasm of the adapted 70 parent, irregardless of the particular characteristics or genes that may be introduced in the process. The view or interpretation of the findings toward which the present author is inclined is that adaptation is a function of some, perhaps several, morphological and physiological characteristics, that these are heritable, and that once identified as contributing to adaptation, they can be transferred into other populations or lines genetically. No claim is advanced that the four charact- eristics defined in Seafarer completely determine the high level of adaptation of that variety. The measurement of adaptation has proved to be a very difficult task. No research is known where plant traits have been associated with adaptation and analyzed by the methods of this study, in crop plants. The results presented may provide a starting point for further work. LITERATURE CITED 71 LITERATURE CITED ALLARD, R.W. and A.D. BRADSHAW. 1964. Implications of genotype-environmental interactions in applied plant breeding. Crop Science 4:503-508. BLISS, F.A. 1971. Inheritance of growth habit and time of flowering in beans, Phaseolus vulgaris L. J. Amer. Soc. Hort. Sci. 96(6Tt7l7. CAMACHO, L.H. 1968. Estabilidad y adaptabilidad de lineas homocigotas de frijol Phaseolus Vulgaris L. y su impli- cacion en la seleccion por rendimiento Revista Insti- tuto Colombiano Agropecuario 3:165—178. COYNE, D.P. and R.H. MATTSON. 1964. Inheritance of time of flowering and length of blooming period in Phaseolus vulgaris L. Proc. Amer. Soc. Hort. Sci. 85:366—373. DUARTE, R. 1961. Component interaction in relation to mean expression of complex traits in a field bean cross. M.S. Thesis, Michigan State University. 33p. 1966. Responses in yield and yield components from recurrent selection practiced in a bean hybrid pop— ulation at three locations in North and South America. PhD. thesis. Michigan State University. 93p. EBERHART, S.A. and W.A. RUSSELL. 1966. Stability parameters for comparing varieties. Crop Science 6:36—40. EMERSON, R.A. 1916. A genetic study of plant height in Phaseolu§_vulgaris. Agric. Exp. Sta. of Nebraska. Research Bulletin No. 7. 73p. FINLAY, K.W. and G.N. WILKINSON. 1963. The analysis of ad— aptation in a plant breeding program. Aust. J. of agric. Res. 14: 742-754. HARDWICK, R.C. and J.T. WOOD. 1972. Regression methods for studying genotype-environment interactions. Heredity 28:209-222. KNIGHT, R. 1970. The measurement and interpretation of genotype-environment interactions. Euphytica 19: 225-235. MATHER, K. and J.L. JINKS. 1971. Biometrical Genetics. Ithaca, Cornell University Press. 382p. 72 73 PARSONS, F.A. and R.W. ALLARD. 1960. Seasonal variation in lima bean seed size: an example of genotypic—environ— mental interaction. Heredity 14:115-123. PERKINS, J.M. and J.L. JINKS. 1963. Environmental and genotype-environmental components of variability III. Multiple lines and crosses. Heredity 23: 339-356. STERN, J.T. 1970. The meaning of adaptation and its re— lation to the phenomenon of natural selection. In Dobzhansky, Hecht and Steere (eds). N.Y., AppleESn— Century—Crofts. Evolutionary Biology 4: 39-66. YATES, F. and COCHRAN, W.G. 1938. The analysis of groups of eXperiments J. of Agric. Sci. 28: 556—580. APPENDICES 74 75 TABLE 21.—-Genotypic constitution, plant type, and some mat— urity characteristics for each evaluated line. Entry Genotypel Selection Growth Days Length Number — Class Habit First Blooming Number2 —~ Flower Period —- (dare?) 1 (Se x CN)xCN 2 D 42.9 26.8 2 (Se x Sa)xSa 2 D 39.8 28.9 3 (Se x C1)xC1 5 I 35.4 20.4 4 (Se x Sa)xSa 3 D 35.9 35.1 5 (Se x Co) 1 D 37.0 22.8 6 (Se x 27)x27 4 D 36.2 19.1 7 Coleccion l-63a 5 I 35.3 19.4 9 (Se x Sa) 4 D 42.8 40.2 10 (Se x RO)xRO 2 D 33.9 22.8 11 27-R 2 D 35.3 22.7 12 (Se x C1)xC1 5 I 34.9 16.2 13 (Se x 27)XSe 4 D 35.0 23.6 14 Ri‘fi‘on Oscuro 3 D 33 . 8 23 . 8 15 (Se x C1)xC1 2 D 31.1 24.4 16 (Se x C1)xSe l D 34.6 15.5 17 (Se x RO)xRO 2 D 35.0 27.0 18 (Se x Sa)xSe l D 36.8 18.0 19 Canario 6 I 71.3 59.6 20 (Se x Co) 4 D 36.5 32.1 21 (Se x Sa)xSa 3 D 37.0 42.3 22 (Se x CN)xSe l D 38.2 19.2 23 (Se x 27)xSe 2 D 36.3 20.0 24 (Se x CN)xCN 6 I 31.6 31.0 25 (Se x Sa)xSa 1 D 35.6 23.4 26 (Se x Sa)xSa 4 D 34.9 42.5 27 (Se x RO)xRO 3 D 34.4 23.1 28 (Se x Sa)xSe 1 D 35.3 24.1 29 (Se x 27)x27 3 D 34.5 23.0 TABLE 21.——Continued 30 31 32 33 34 35 36 37 38 39 4O 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 E7: 6 (Se (Se (Se (Se (Se (Se (Se (Se (Se (Se (Se (Se (Se (Se (Se (Se X XXXXXXXXXK‘XXXXX X 27)x27 C1)xSe RO)xRo Ca) Sa)xSe C1)xC1 RO)xRO Co) Li) C1)xC1 C1)xC1 27)xSe Li)xLi Sa)xSe 27)xSe C1)xSe RO)xSe Seafarer (Se x C1)xC1 Liborino (Se (Se (Se (Se (Se (Se (Se X X X X C1)xC1 Sa)xSa CN)xCN CN)xCN xRO)xSe X X RO)xSe C1)xSe (Se x RO)xRo Compuesto Negro (Se x 27)x27 (Se x Sa)xSe HHO‘WU‘INMNI—‘bmwml—‘LUUTvbi—‘IbHNNbAWNHHNl-‘b UUHUHUUUUUHUHUUHUUUUUUUUUUUUUUU 35.2 34.0 33.6 39.2 34.9 30.4 33.7 40.7 37.8 31.7 31.7 38.2 63.6 36.4 37.9 37.7 33.2 35.0 35.7 74.8 34.9 34.6 39.2 33.6 34.1 34.3 35.2 33.7 44.7 34.5 36.4 20.4 18.9 23.3 14.3 22.1 22.8 25.5 21.2 28.2 21.2 21.2 21.9 43.2 18.3 22.2 23.7 23.1 19.2 22.0 32.6 20.3 26.9 29.2 28.0 21.0 19.6 20.2 38.0 39.9 23.1 20.2 77 TABLE 21.-—Continued. 61 (Se x27)x27 l D 34.5 20.4 62 Cocacho 6 I 65.8 68.4 63 (Se x C1)x C1 2 D 31.5 20.4 64 Sangretoro 3 D 50.5 54.4 Mean 38.0 26.6 Standard Deviation 1.7784 3.8783 Coefficient of Variation 4.68 14.58 L Data collected in East Lansing, BC S or F l 6 6 generation. The parental varieties are: Ca = Canario Co = Cocacho C1 = Coleccion l—63a CN = Compuesto Negro Li = Liborino R0 = Rinon Oscuro Sa = Sangretoro Se = Seafarer 27 = 27—R 3 See Table 2, page for characteristics of each selection class. 1 D = Determinate I = Indeterminate 78 TABLE 22.——Mean values for maturity and leaf characteristics and harvest index for each evaluated line. Days to Harvest Entry Days Leaflet Leaf Harvest Leaf Number First E.L. Colombia Size Area Index Area Ripe cm2 cm2 Ratio Pod dcmZ/gmv-l 1 80.0 96.7 76.0 69.7 2671.6 .35 1.13 2 79.3 92.7 76.7 92.1 1205.1 .39 0.58 3 66.0 75.0 76.3 56.5 2015.3 .50 0.69 4 80.7 98.7 76.5 90.5 1810.0 .29 0.80 5 71.7 96.3 73.3 60.8 2087.3 .52 0.62 6 74.0 106.7 78.0 55.0 1631.8 .39 0.57 7 68.7 75.0 77.7 56.5 1845.9 .52 0.56 8 75.7 92.3 83.0 67.3 2019.0 .44 0.71 9 83.7 113.0 76.0 53.4 1869.0 .33 0.66 10 73.0 92.7 77.0 69.9 1980.3 .49 0.71 11 75.7 96.7 73.3 79.3 2438.5 .53 0.71 12 64.7 74.0 76.0 42.0 1722.0 .52 0.63 13 73.0 104.0 76.0 48.8 1772.9 .35 0.59 14 75.3 96.3 76.5 82.6 2285.5 .44 0.74 15 65.3 77.0 76.0 74.1 1358.2 .52 0.69 16 61.7 73.0 68.0 40.2 1246.2 .52 0.50 17 72.0 94.0 73.0 85.9 2405.2 .50 0.76 18 69.0 77.0 76.0 45.6 1428.6 .49 0.55 19 113.3 146.7 118.6 67.2 4793.4 .37 1.36 20 79.3 107.7 81.0 56.8 1988.0 .34 0.57 21 79.7 106.0 81.5 71.2 1851.2 .37 0.53 22 72.7 94.3 76.0 51.9 2266.5 .55 0.88 23 72.7 94.3 76.3 68.0 1360.0 .51 0.41 24 70.3 94.7 78.3 66.0 2376.0 .52 0.83 25 69.0 78.3 76.0 54.9 1592.1 .48 0.63 26 77.0 102.7 81.0 55.2 1472.2 .40 0.59 27. 73.7..107.3 ...81.0 .- 99.8 - 1963.1 -.33. ~.0.70< v TABLE 22.--Continued.. 79 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 71.3 72.7 75.3 68.0 73.0 72.7 70.3 61.3 75.7 78.7 84.0 67.0 65.0 73.7 104.7 69.3 79.0 69.7 73.3 67.7 69.7 113.0 67.7 76.3 78.0 71.3 69.0 71.7 67.3 76.7 83.0 79.3 103.3 104.7 74.0 95.3 78.0 77.0 74.0 101.3 101.7 114.7 77.0 73.0 97.3 137.3 78.3 106.3 83.7 98.3 77.0 80.3 140.3 76.0 100.0 90.0 93.7 94.0 82.7 74.0 104.7 104.7 76.3 76.7 77.0 68.0 75.5 76.3 76.7 76.5 82.0 78.0 78.3 82.0 78.0 72.0 80.7 75.0 78.0 76.3 76.0 68.0 82.7 79.0 76.7 76.3 76.0 76.0 72.0 76.0 76.0 82.0 44.8 69.0 61.0 46.8 66.6 39.1 45.4 75.1 87.0 53.0 54.3 87.9 43.8 46.6 60.8 48.6 48.5 43.1 87.0 40.1 45.6 103.5 51.4 56.2 63.1 90.1 71.6 68.5 40.3 98.9 60.1 1598.0 1816.8 2541.0 1014.2 1776.2 977.5 1801.0 1126.5 1798.3 2084.5 1321.1 1494.3 1197.0 1351.4 992.9 1944.0 1665.0 2772.6 2203.7 1550.7 1063.8 2034.8 1696.2 2004.6 1956.1 1921.8 1838.0 1666.6 1034.5 1681.3 2884.8 .53 .38 .48 .56 .52 .52 .48 .38 .44 No.79 0.53 0.88 0.39 0.56 0.40 0.86 0.49 0.77 0.67 0.37 0.72 0.39 0.44 0.55 0.63 0.54 0.87 0.72 0.47 0.37 0.81 0.61 0.84 . 0.61 0.71 0.78 0.55 0.39 0.55 0.89 80 TABLE 22.--Continued 59 73.7 '103.0 77.7 63.3 2552.9 .28 1.06 60 69.0 77.0 76.0 48.3 1803.0 .50 0.71 61 70.0 103.0 77.0 64.5 1806.0 .42 0.60 62 110.0 144.7 94.0 61.3 2860.9 .35 0.72 63 67.0 84.7 76.3 82.9 1685.4 .47 0.83 64 95.7 115.3 79.5 82.6 1514.1 .32 0.47 Mean 75.29 94.73 77.15 63.46 1851.35 .44 Std. Dev 2.49 4.8 5.03 18.32 593.69 ._.., 81 TABLE 23.——Location mean number of pods per plant for every line evaluated. Entry East Peru Colombia Lansing 1 7.702 13.599 13.600 2 7.658 8.866 9.630 3 9.938 10.717 14.033 4 6.710 9.848 16.265 5 16.020 15.674 16.600 6 12.938 12.222 24.250 7 10.681 6.788 12.750 8 9.461 14.501 14.310 9 14.793 12.302 21.800 10 7.920 7.098 4.600 11 8.978 6.778 6.237 12 10.415 10.935 10.210 13 11.450 10.351 15.257 14 6.500 7.230 9.335 15 7.672 5.615 5.000 16 13.551 22.508 19.770 17 9.353 6.879 8.547 18 12.301 5.259 12.500 19, 8.584 8.658 12.496 20 15.228 14.178 36.335 21 11.835 17.108 8.700 22 12.614 14.555 17.700 23 12.334 7.898 15.833 24 15.971 12.068 27.637 25 12.452 5.120 19.600 26 14.545 15.688 12.635 27 9.265 7.752 2.666 28 9.473 4.252 19.417 11.880 N ‘ \0 "10.833 .. 12.8589, rim TABLE 23.-—Continued 82 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 '45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 12.660 13.618 9.554 14.804 8.985 6.647 6.992 12.172 9.174 7.129 10.501 12.011 4.537 13.009 11.952 13.490 8.417 17.062 10.125 2.131 10.030 10.609 10.577 11.565 8.356 7.158 8.271 7.147 9.385 7.262 11.081 10.410 10.555 12.645 11.081 13.015 5.892 7.795 8.256 11.278 12.054 6.218 12.367 14.726 6.337 10.658 9.177 14.620 7.146 21.085 11.097 4.894 11.337 11.129 11.506 12.459 8.140 5.244 10.940 9.061 3.002 12.910 6.350 9.956 14.400 15.233 13.940 14.957 15.533 5.320 3.870 26.415 12.350 10.907 16.703 18.733 10.603 15.700 17.223 11.217 8.390 27.547 15.000 13.327 11.523 20.100 12.620 16.967 9.233 7.733 9.770 13.033 14.523 17.933 15.933 11.367 83 Variance LSD (P=.05) 10.640 8.058 10.194 10.299 4.97 3.642 84 TABLE 24.--Location mean number of seeds per pod for every line evaluated. 32 Entry East Peru Colombia Lansing 1 5.224 5.181 4.200 2 4.039 3.805 2.773 3 4.834 5.394 5.080 4 3.808 3.298 3.140 5 5.178 5.268 3.667 6 2.852 3.880 2.680 7 5.731 6.012 4.493 8 5.830 6.178 4.980 9 3.624 4.506 3.440 10 4.318 4.573 1.880 11 4.004 4.467 2.987 12 5.242 5.781 3.893 '13 3.075 4.438 3.373 14 3.997 3.757 2.980 15 4.697 5.098 3.760 16 4.959 5.134 3.533 17 3.861 4.368 1.587 18 5.216 6.007 5.253 19 3.736 3.902 3.240 20 3.409 4.875 3.720 21 4.679 4.586 2.100 22 5.293 5.545 4.907 23 4.737 4.272 2.200 24 4.126 5.643 4.680 25 5.179 4.976 5.200 26 2.776 4.092 3.800 27 3.407 4.012 3.007 28 5.044 5.369 4.693 29 3.350 3.749 2.547 TABLE 24.-—Continued 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 '45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 3.237 4.418 4.286 4.191 4.813 5.051 3.524 3.764 4.161 4.779 5.147 4.789 2.292 6.166 3.019 5.275 4.598 4.895 5.560 2.359 5.356 3.536 5.741 5.494 4.423 4.283 5.491 4.103 6.130 2.844 5.530 4.161 5.094 4.463 5.741 5.377 5.452 4.712 4.646 4.418 4.813 5.111 4.976 3.452 5.766 4.111 6.039 4.412 5.167 6.133 3.816 4.868 4.134 5.719 5.310 4.042 4.047 5.461 4.085 4.840 3.208 5.325 3.707 4.840 3.160 3.933 5.653 43.40 1.800 3.760 2.920 4.280 4.427 4.493 3.307 5.640 2.760 4.840 4.387 5.027 3.920 3.827 5.187 3.613 4.467 4.493 3.987 3.640 4.840 5.280 4.360 2.587 4.320 TABLE 24.--Continued 61 62 63 64 Mean Variance LSD (P=.05) 3.377 4.442 4.890 4.221 4.213 0.2055 0.740 4.111 4.478 4.830 4.152 4.756 0.2067 0.743 3.787 3.540 3.347 3.320 3.906 1.537 87 TABLE 25.—-Location mean seed weight for every line ‘l evaluated. Entry East Peru Colombia Lansing_ 1 .243 .240 .200 2 .250 .310 .293 3 .298 .290 .297 4 .266 .323 .290 5 .204 .220 .203 6 .305 .307 .333 7 .272 .263 .280 8 .228 .227 .215 9 .285 .337 .370 10 .395 .397 .260 11 .507 .393 .410 12 .263 .230 .280 '13 .304 .297 .320 14 .535 .553 .400 15 .283 .273 .240 16 .199 .213 .220 17 .430 .333 .283 18 .197 .180 .217 19 .398 .453 .416 20 .239 .260 .260 21 .236 .220 .270 22 .212 .223 .250 23 .288 .297 .220 24 .238 .223 .233 25 .197 .203 .215 26 .246 .250 .250 27 .340 .357 .343 28 .200 .193 .237 29 .384 .327 .380 TABLE 25.——Continued 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 ' 45 46 47 48 49 . 50 51 52 53 54 55 56 57 58 59 60 .314 .210 .330 .191 .201 .294 .336 .300 .360 .306 .245 .271 .286 .209 .302 .250 .419 .199 .289 .502 .268 .246 .256 .239 .345 .519 .286 .413 .246 .358 .208 .347 .233 .390 .157 .200 .267 .310 .283 .337 .313 .250 .220 .297 .190 .377 .283 .480 .200 .267 .523 .237 .290 .257 .290 .413 .403 .273 .363 .260 .330 .190 .353 .253 .405 .197 .233 .240 .340 .250 .317 .280 .273 .250 .330 .217 .333 .280 .377 .217 .300 .353 .297 .287 .210 .253 .373 .323 .293 .347 .233 .323 .230 TABLE 25.-—Continued 61 62 63 64 Mean Variance LSD(P=.05) .351 .338 .282 .340 .296 .00087 .048 .283 .427 .267 .403 .297 .00076 .045 .337 .380 .243 .365 .287 - .0039 _——- 90 TABLE 26.——Location mean yields (gm/1m) for every line evaluated. Entry East Peru Colombia Lansing 1 95.97 145.64 57.00 2 67.99 84.25, 46.67 3 142.89 143.08 212.33- 4 64.29 88.75 99.00 5 172.62 138.24 123.33 6 108.93 99.06 188.00 7 164.91 84.73 147.00 8 123.78 157.66 131.50 9 156.78 130.06 244.00 10 135.07 89.15 18.00 11 183.78 80.27 80.33 12 145.03 108.00 90.00 '13 97.66 90.16 132.67 14 138.76 121.47 102.50 15 101.23 65.53 36.00 16 133.30 151.39 146.00 17 156.52 63.63 29.00 18 119.96 49.34 145.67 19 134.15 111.25 140.52 20 124.86 128.97 175.00 21 127.88 108.86 44.50 22 143.80 148.28 214.00 23 168.09 70.80 58.33 24 149.73 120.27 238.67 25 127.25 37.78 221.50 26 93.86 105.21 91.00 27 105.51 74.44 26.00 28 99.17 33.32 149.33 29 139.15 93.92 106.33 TABLE 26.--Continued 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 '45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 134.89 131.34 139.18 121.64 85.22 100.26 79.22 141.24 138.68 104.45 129.19 158.10 28.47 163.66 108.10 177.50 163.75 169.64 161.62 25.14 137.34 87.02 155.61 148.99 130.36 157.22 125.94 120.74 145.14 72.39 123.85 102.47 117.82 139.28 73.48 45.60 90.55 73.47 97.58 135.72 73.59 128.17 116.78 53.05 98.12 77.50 196.87 116.46 135.32 138.50 85.63 116.61 92.81 139.70 160.24 98.48 57.84 119.48 93.72 28.65 81.11 51.98 188.00 190.67 143.00 98.67 203.33 41.50- 19.00 137.00 99.33 90.33 200.00 207.67 91.67 186.67 111.33 156.00 137.00 303.00 104.50 129.00 145.67 201.00 114.00 254.67 136.67 93.33 123.00 214.33 96.33 149.00 160.67 92 TABLE 26.--Continued 61 117.13 62 125.74 63 94.67 64 102.66 Mean 125.31 Variance 1075.91 LSD(P=.05) 53.56 75.99 159.17 75.32 144.06 101.64 604.95 40.16 152.00 64.50 58.00 116.50 131.29 6572.34 93 TABLE 27.——Ana1ysis of variance for log yield (gms/lmt.) values from data collected in an 8x8 lattice planted at East Lansing, Michigan and La Molina, Peru. MEAN SQUARE df_ East Lansing Peru Source of variance Replications 2 .03617473 .17653406 Entries unadjusted 63 .07724556** .09289412** adjusted 63 .08124066** .09656913** Blocks within reps. (adj.) 21 .03763240 .02060404 Error Randomized complete block 126 .02233084 .01490687 Effective 105 .02083763 .01452879 Total 191 ** Significant at the 18 level Lattice efficiency 107.17 102.60 LSD P=.05 .2357 .1968 Coefficient of variation 6.9948 6.1326 94 TABLE 28.-—Analysis of variance for log of pods per plant values from data collected in an 8x8 lattice planted at East Lansing, Mich— gan and La Molina, Peru. MEAN SQUARE df East Lansing Peru Source of Variance Replications 2 .01569055 .13357551 Entries unadjusted 63 .05702494** .08595206 adjusted 63 .05766791** .08837842 Blocks within reps. (adj.) 21 .01590987 .01364442 Error Randomized complete block 126 .01373467 .00952182 Effective 105 .01366329 .00922287 Total 191 ** Significant at the 1% level Lattice efficiency 100.52 103.24 LSD P=.05 .1909 .1568 Coefficient of variation 11.8714 9.8550 95 TABLE 29.——Ana1ysis of variance for log seeds per pod values from data collected in an 8x8 lattice planted at East Lansing, Michigan and La Molina, Peru. MEAN SQUARE df East Lansing Peru Source of Variance _ Replications 2 .00007150 .00072505 Entries unadjusted 63 ' .03059230** .01569158 adjusted 63 .03100152** .01569158** Blocks within reps. (adj.) 21 .00566145 .00328671 Error Randomized complete block 126 .00316053 .00328671 Effective 191 ** Significant at the 1% level Lattice efficiency 109.16 Less efficient LSD P=.05 .0879 Coefficient of variation 8.5189 96 TABLE 30.--Ana1ysis of variance for log (seed weight + 1.0) values frOm data collected in an 8x8 lattice planted at East Lansing, Mich— gan and La Molina, Peru. ' MEAN SQUARE d£_ East Lansing Peru Source of Variance Replications 2 .00097907 .00002037 Entries unadjusted 63 .00211906** .00226225** adjusted 63 .00211014** Blocks within reps. (adj.) 21 .00024596 .00005678 Error Randomized complete block 126 .00014455 .00007875 Effective 105 .00013452 Total 191 ** Significant at the 1% level Lattice efficiency 107.46 Less efficient LSD P=.05 .0189 .0145 Coefficient of variation 10.3308 7.9243 97 ms.ma . ma.ma Ha.m~ mw.a~ coflumaum> «6 pamaoaummoo mmvo.m nmoa. vmmm. Nmmo.H madmz H0>ma ma 0:0 pm unmofluflcmflm 44 mmH HMDOB mHHHNmo. mmmaaooo. mammamao. ommmovmo. moa mmcflq GHQDHZ mmmmowha. «tmmammooo. «smemmmavo. «ammmmmhoa. mm mwcfla dwmaumm .WMMMM ugmflmz doom U0m\mpmmm uddam Hmm moom .Ww mocmflwmw mo mundOm mm¢DOm Zémz .madom camoa co wumc Had .MHQEOHOU .wuflEHdm um cmuomaaoo dump Eoum modam> mucmcomfioo cams» can camflw How mocmwum> mo mfimwamcdll.am mqmda 98 TABLE 32.—~Weather data (1973) for the three loc~ ations for the period the experiments were in the field. East Lansing June July August September Peru March April May June Colombia July August September Average Maximum 26.8 28.6 28.5 24.3 27.1 25.2 22.9 19.1 29.8 29.5 28.5 Average Minimum 15.7 16.3 16.2 11.2 19.2 16.7 14.4 12.8 18.2 18.4 18.4 Average 21.3 22.5 22.3 17.8 22.2 20.2 17.8 15.3 23.6 23.2 22.6 Rainfall (mms) 91.5 21.7 56.0 82.0 52.7* 65.5 112.5 * The plots in Peru were irrigated as needed since rainfall is very scarce in that region. ombia were irrigated three times for a period of 14 days after planting, no irrigation was necessary for the rest of the season. Lansing. The plots in Col— No irrigation applied to the plots in East 174 8 20 m m m m m III I" um 3 1293 03 0111117 IIWI