CF ‘ TH‘E'SYS L‘BMRY Kichigan State University This is to certify that the thesis entitled GENETIC DIVERSITY OF BEAN LANDRACES IN NORTHERN MALAWI presented by GREGORY B. MARTIN has been accepted towards fulfillment of the requirements for Masters degree in Crop & Soil Science l . /f " >41” Wczi/m c. 42/2145..— Major professor Date M" J; #51 M- Wayne Adams \£\ 0-7639 MSU is an Affirmative Action/Equal Opportunity Institution MSU LIBRARIES “ \— RETURNING MATERIALS: Place in book drop to remove this checkout from your record. _EL3§§ will be charged if book is returned after the date stamped below. qr- —- GENETIC DIVERSITY OF BEAN LANDRACES IN NORTHERN MALAWI BY GREGORY B. MARTIN A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Crop and Soil Science 1984 3/7WL/r ABSTRACT GENETIC DIVERSITY OF BEAN LANDRACES IN NORTHERN MALAWI BY GREGORY B. MARTIN The observation that diverse bean mixtures (Phaseolus yulgaris LJ are maintained by Malawian farmers prompted this study into the extent, the underlying patterns, and the generation of variability in this species in northern Malawi. The outcrossing rate, estimated by using flower color and hypocotyl pigmentation as dominant gene markers, was found to be 0 - 2.23%. Considerable variability both within and between landraces was documented for numerous phenological, morphological, agronomical and qualitative characters. A principal components analysis revealed a clinal pattern, with the northern and southern areas forming the extremes. Several statistically significant correlations were found between environmental variables and plant character measurements, suggesting that site-specific adaptation is occurring. Separate principal component analyses of each landrace revealed genetic intergradation involving major seed classes in these bean mixtures. It is postulated that a low level of outcrossing exerts a profound impact on the generation and maintenance of variability in the Malawian bean landraces. ACKNOWLEDGEMENTS I wish to express my appreciation to my major professor, Dr. M. Wayne Adams, for providing me with the opportunity to conduct this research in Malawi. His perceptive observations initiated this study and inspired me to follow it through. .His challenging discussions, friendly manner, and encouragement were instrumental in the completion of this thesis. In sum, he has been one of those rare persons by whose contact a student grows - a mentor. I express my gratitude to: Dr. Jim Kelly and Dr. Jim Hancock, my committee members, for their critical evaluation of this thesis. Dr. Ardeshir Ghaderi, for his suggestions at the planning stage of this research and his assistance with the preliminary computer analysis. Dr. Eric Ayeh, for his companionship, support, and the many "porch-talks" we shared at the house in Lilongwe. Dr. 0. Todo Edje, for his patient guidance through administrative procedures and his special interest in this project. ii Special thanks to: Mr. Cohen Sichinga, for his assistance with land and irrigation facilities at a critical juncture in this study. Mr. Frederick Kalengamaliro and Mr. Augustine Mkangama, for their technical and linguistic assistance throughout the germplasm collection trips and during the course of this research. Four Bunda College students worked long, hot hours with me making thousands of measurements in the field. To Wellings Mwalabu, Fred Wachepa, David Chongo, and Office - Zikomo kwambiri! To the many Malawian farmers who greeted me with hospitality, patience, and generosity and allowed me to tramp through their fields, ask endless questions, and "borrow" their beans, I express my thanks to you and my admiration for your knowledge and fortitude. I express my appreciation to my family for their trustworthy correspondence while I was overseas — there were many days when a letter from home made the long Bunda road less dusty. Finallqn I am.most indebted to Susan who kept the faith and to a little one, Kelly, whose contribution to my life grows along with her. iii TABLE OF CONTENTS LIST OF TABLES O O O O O O O O O O O O O O O O O O 0 LIST OF FIGURES O O O C O O O O O O O O O O O O O O P ROLOG UE O O O O O O O O O O O O O O O O O O O O O 0 PART I : NATURAL HYBRIDIZATION IN MALAWIAN BEAN LANDRACES CHAPTER 1 : OUTCROSSING ESTIMATION . . . . . . Literature review . . . . . . . Introduction . . . . . . . . . . Materials and Method . . . . . Results . . . . . . . . . . . . Discussion . . . . . . . . . . . PART II : DIVERSITY IN MALAWIAN BEAN LANDRACES CHAPTER 2 . LITERATURE REVIEW . . . . . . . . . CHAPTER 3 . MATERIALS AND METHODS . . . . . . . Germplasm collection . . . . . . Experimental procedures . . . . . Character measurements and scores Phenological characters . . Morphological characters Agronomic characters . . Qualitative characters . Data Analysis . . . . . . . . . CHAPTER 4 : RESULTS 0 O O O O O O 0 O O O O O O Variability in the landraces . . . . . . . . Environmental / Plant character associations Landrace populational structure . . . . . . . iv 12 13 17 25 28 43 43 47 56 56 59 61 62 66 70 71 92 TABLE OF CONTENTS (Continued) Page CHAPTER 5 : DISCUSSION . . . . . . . . . . . . . . 107 Variability in the landraces . . . . . . . . . 107 Environmental / Plant character associations . 118 Landrace populational structure . . . . . . . . 121 SUMMARY AND CONCLUS IONS O O I O O O C O O O O O O l 2 8 LITERATURE CITED 0 O O O O O O O O O O O O O O O O 133 APPENDIX A : Character measurements and scores for the 37S Malawian bean lines, and the 4 Michigan cultivars included in the study . . . . . . . 138 APPENDIX B : Plots of principal component's I and II for the 15 collection sites . . . . . . . . . . 147 TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE LIST OF TABLES Growth habit, flower color, stem color and days of flowering coincidence in the six populations . . . . . . . . . . Estimates of outcrossing rates for the Six populations 0 O O O O O I O O O 0 Number of thrips per flower. Average of 25 random of each color . . . . . . . . Names and locations of the 15 COlIeCtion Sites I O O O O I O O O O O O 0 Temperature, rainfall and daylength for the five collection areas . . . . . . Soil characteristics of the 15 collection sites . . . . . . . . . . . Character codes and summary of the quantitative and qualitative character measurements . . . . . . . . . . Banner and wing color frequencies in the 15 landraces . . . . . . . . . . . Growth habit and hypocotyl color frequencies in the 15 landraces . . . . . Mean values and ranges for the 21 quantitative variables for the northern region of Malawi . . . . . . . . Analysis of variance for the 18 replicated character measurements . . . . Mean differences between collection areas for the 21 quantitative characters . . . . . . . . . . . . . . . Mean differences between collection sites for the 21 quantitative characters . . . . . . . . . . . . . . . . Distances between areas based on the first 6 principal components derived from the 21 quantitative plant characters . . . vi 14 18 23 SO 51 52 57 72 76 79 81 82 83 85 TABLE TABLE TABLE TABLE TABLE TABLE TABLE TABLE 15 l6 17 18 19 20 21 22 LIST OF TABLES (Continued) Distances between sites within areas based on the 21 quantitative plant characters . Correlations between the 21 quantitative characters and selected site variables Minimum and maximum distances between two-line pairs at each collection site, based on the 21 quantitative plant characters . . . . Analysis of variance for the 18 replicated plant characters in three homogenous seed types from collection site 1 Analysis of variance for the 18 replicated plant characters in two homogenous seed types from collection site 6 Mean distance and standard deviation between lines within each of five homogenous seed types Distances between Michigan lines, based on the 21 quantitative plant characters . Normalized scores for the first six principal components, and distances between three pairs of brown-seeded lines from site 1 vii 88 91 93 95 96 98 98 99 Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 10 ll: 12: 13: LIST OF FIGURES A collection of insects observed visiting bean flowers in Malawi . . . A bee from the genus Xylocopa visiting a bean flower . . . . . . . The array of new seed types recovered from the segregation Of 888d C110-24 o o o o o o o o o o o The five geographical areas from which bean landraces were collected . . . . . . . . . . . An example of a bean germplasm collection form . . . . . . . . . . . The fifteen collection sites in northern Malawi . . . . . . . Seed types of the 375 bean lines included in the study . . . . . . . . The seed color 'descriptor chart' . . Frequencies of the banner colors for the five geographical areas in the northern region of Malawi . . . . . . Frequencies of the wing colors for the five geographical areas in the northern region of Malawi . . . . . . Frequencies of the hypocotyl colors from the five geographical areas in the northern region of Malawi . . . . Frequencies of determinate and indeterminate growth habits in the five geographical areas in the northern region of Malawi . . . . . . Plot of the first two standardized principal components based upon the area means of the 21 quantitative characters . . . . . . . . . . . . . Plot of the first two standardized principal components based upon the site means of the 21 quantitative characters . . . . . . . . . . . . . viii 21 22 24 44 46 48 54 64 74 75 77 78 84 87 Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 17: 18: 19: 20: 21: 22: 23: 24: LIST OF FIGURES (Continued) Plot of the first two standardized principal components based upon the site means of the 5 phenological characters . . . . . . . . . . . . Distances between sites within areas and between areas based upon the standardized and normalized scores of the first 6 principal components derived from the site means of the 21 quantitative characters . . . . Principal component plot of the landrace from site 8 - an example of a simple mechanical mixture . . Principal component plot of the landrace from site 11 - an example of a simple mechanical mixture . . Principal component plot of the landrace from site 6 - an example of incipient genetic intergradation Principal component plot of the landrace from site 1 - an example of incipient genetic intergradation Principal component plot of the landrace from site 5 - an example of incipient genetic intergradation Principal component plot of the landrace from site 14 - an example of incipient genetic intergradation Principal component plot of the landrace from site 4 - an example of a later stage of genetic inter- gradation where a 'hybrid swarm' has formed . . . . . . . . . . . . Principal component plot of the landrace from site 10 - an example of very extensive genetic intergradation . . . . . . . . . . ix 89 90 101 101 102 102 104 104 105 105 PROLOGUE The common bean, Phaseolgg vulgaris L., is a cosmopolitan species. It grows throughout Latin America, from the Andean highlands to the hot, dry plains of northern Mexico. Beans are an important cash crop in Michigan, California, New York, Colorado, Nebraska, Idaho, North Dakota and Minnesota. In recent years their production has increased in Canada. But beans are not confined to the New World. They are cultivated throughout Europe and are found in the Eastern African highland areas from EthiOpia to South Africa. This wide range is directly attributable to the tremendous genetic diversity encompassed by this species. Beans were first domesticated in Latin America. Three theories exist regarding their origin. Miranda-Colin (1967) and Gentry (1969) believe that beans were domesticated in the Mexico-Guatemala region of Mesoamerica. However, Berglund-Brucher and Brucher (1976) have presented evidence for a South American (Andean) origin. Kaplan (1981) believes the archaeological record supports an independent origin in each of these areas. Beans were disseminated into the American southwest at least 2300 years ago and were in the northeast by 1000 AJL (Kaplan, 1971). By the time of Columbus, beans had been established in North America for over 1500 years. According to Baker (1971), there is no evidence for a pre-Columbian dispersal of beans to the Old World. In 1498, Vasco da Gama, the Portuguese explorer, first made contact with the Eastern African coast (Binns, 1976). During the next 60 years, Portuguese traders called at the ports of Sofala, Mozambique, Zanzibar and Mombassa (Oliver and Mathew, 1963). Although the Portuguese rarely ventured inland, their influence was carried to the interior by the Arab slave traders and Swahili merchants active in the region (Merril, 1954). This influence consisted of European and Asian goods, and most significantly, of many new food plants from Latin America. Merril (1954) states, "There is no doubt that the people who brought maize into Nyasaland (now Malawi) were Arabs of the coast. This community is known as 'Manga', which in Swahili means 'Arabia”2 The Malawians today still refer to maize as 'Chimanga'. By the early 1500's beans had been established in South America for several thousand years (Kaplan, 1971). The Portuguese, having established an outpost in Brazil, likely carried beans from this area during their voyages through the Indian Ocean. These voyages, besides promoting trade, were instrumental in setting up missions along the coast (Coupland, 1938). In 1542, St. Francis Xavier, on his way to Goa, visited missions at Mozambique, Malindi, and Socatra (Oliver and Mathewy 1963). These ports are within 150 miles of the lower Rift Valley in Malawi and the surrounding highlands. The highlands would prove ideal for many Latin American plant introductions, including beans. Several Jesuit missions were founded in 1560 in what is now Mozambique (Coupland, 1938). These missions were likely an important catalyst to plant introductions in East Africa. David Livingstone provides the first mention of beans in what is now Malawi. During his 1858-59 expedition along the lower Shire River, Dr. Livingstone recorded, "Many gardens of maize, pumpkins, and tobacco fringed the marshy banks as we went on u. and then the people brought meal, maize, fowls, batatas, yams, and beans for sale" (Livingstone, 1865). That beans were fully integrated into Malawian culture by the early 1900's is shown in the striking lexical diversity by which specific bean seed types were known (Williamson, 1975). Observations by the Malawi Bean/Cowpea Collaborative Research Support Project team in 1981-84 confirm this. Bean names may refer to seed color, (Kabaya = white). Plant architecture, (Mwikala = "just sits there" - a non-climber), seed pattern, (Mwasipengele = spotted) and origin (Mine = from the Zambian c0pper mines). Such lexical differentiation has also been noted in beans in the municipio of Tenejapa, Chiapas, Mexico by Berlin, Breedlove and Raven (1974). ‘They discovered that the degree of linguistic differentiation in a species is directly related to the importance of that species to the cultivator. This is consistent over both wild and culti- vated plants. The common bean is known by at least six different names in Tenejapa, whereas a culturally unimportant species, Clematis dioca L., is grouped with several other Clematis species (and even other genera) under a common name . Thus the remarkable array of local names given to beans in Malawi indicates that the farmers perceive qualitative (and perhaps quantitative) differences between genotypes and that these differences are culturally important. This thesis is the result of a detailed study of the qualitative and quantitative variability in 15 bean landraces collected in the northern region of Malawi. Five hypotheses will be addressed: 1. There exists in the Malawian bean populations some degree of outcrossing. This outcrossing generates new seed types which often do not undergo selection by the Malawian farmers. The pollination vectom(s) may be unique to Eastern Africa. 2. Extraordinary variability is present in Malawian bean landraces for many morphological, phenological, agronomic and qualitative characters. 3. The diversity within these landraces may lower the risk of crop loss and thus may have distinct advantages for the Malawian farmer. 4. There has been adaptation in these landraces to specific environmental variables such as soil nutrients, photoperiod, and altitude. 5. The populational structune of these landraces* is profoundly influenced byioccasional outcrossing and the resulting recombination and segregation. Outcrossing in fact, may maintain polymorphisms in bean mixtures to a degree beyond even the farmer's intention. The term Plandrace' is used in this thesis to refer to a heterogeneous mixture of a predominantly self-pollinating species which is maintained by a subsistence farmer. These mixtures, however, are much more than mere composites of pure lines. Rather, natural hybridization and recombination in a mixture result in the development of numerous “hybrid swarms". The differential adaptation of the segregants in such a swarm confers a genetic dynamism to these populations. The occurrence of genetic drift and individual farmer selection, as well as possible benefits both in performance and stability from their heterogeneity and homozygosity, serve to make each population genetically unique. PART I NATURAL HYBRIDIZATION IN MALAWIAN BEAN LANDRACES Chapter 1 OUTCROSS ING EST IMATI ON Literature review The common bean (Phaseolus vulgaris L.), is considered to be a predominantly self-pollinating species. However, Rick (1950) has observed that ”Intrinsic pollination relations of a Species are often assumed to be the same under foreign conditions, but this is not necessarily the case". Since beans are an introduced species in Eastern Africa, the presence of unique and perhaps more effective pollinating vectors is entirely possible. Even with a fairly low rate of outcrossing, the potential for hybridization is significantly increased when beans are grown in genotypical 1y complex mixtures as they are in Malawi. The garden pea (Pisum satim), commonly accepted to be completely self-pollinating, has been found to be 3.3% cross-pollinating in Peru - a region outside the native habitat of this species (Harland, 1948). Tomato (Lycopersicon esculentum), is highly self-pollinating in California but in its native habitat of Peru, some genotypes were found to be 40% cross-pollinating (Rick, 1950). 'The difference is thought to be due to the presence of several bee species which occur in Peru but not in California. 7 Darwin (1877), was probably the first to note cross- pollination in beans. He found that Phaseolus vulgaris L. was highly self-fertile, but that occasional outcrosses arose when black, white and brown-seeded beans were grown together. He remarked “The diversity of characters in the seedlings of the second generation was .... wonderful”. Mackie and Smith (1935), working in California, grew out seed of individual plant selections of the California Pink bean (Phaseolus vulgaris L.). They found over 15% of the plants were heterozygous for seed color, size or shape. Between the homozygous lines a wide variation appeared in seed color and maturity date. In another experiment they planted alternating rows of white-seeded and color-seeded varieties. By scoring the recessive progenies for heterozygosity they estimated 0.73% outcrossing. However, they did not adjust for intra-genotypic crosses and since the varieties were planted in a 1:1 ratio their estimate should have been doubled: 1.46%. .Allard (1954) used four monogenically'inherited gene pairs to estimate outcrossing in California lima beans (Phaseolus lunatus LJ. He discovered rates between 1% and 15%, the difference being attributed to three factors: distance between rows, wind direction and varietal differences. Plant habit, growing season and different flowering dates had little effect on outcrossing. 8 In Alabama, Barrons (1938) estimated outcrossing in beans (Phaseolus vulgaris L.) to be between 2.63% and 8.26%. Outcrossing rates decreased as spacing between rows was increased. He found that outcrossing among bush beans was much lower (2.7%) than among climbing beans (5.11%). .He attributed this to the longer flowering period of climbing beans. Tucker and Harding (1975) scored 20,000 progenies of 12 common bean cultivars in Davis, California and found zero outcrosses. They calculated the maximum outcrossing rate which was compatible with their data, given the sample size, and found it to be 3.0 x 10'4. They suggest that no isolation is necessary to obtain selfed seed at Davis. Outcrossing is said to reach 30% in parts of Puerto Rico (Freytag and Vakili, 1977; pers. comm. to M. W. Adams) but this has not been independently confirmed. Carpenter bees are believed to be the vector. Antunes, da Costa, and Oliviera (1973) reported outcrossing rates of 6.2% to 10.6% in Brazil. They, however, did not adjust for allelic frequencies in their two experiments. When adjusted, based on their cultivar ratio's, their estimates range from 7.47% to 12.75%. If similar levels of outcrossing have existed over the past few hundred years, the beans introduced into Malawi from Brazil can probably be assumed to have consisted of some heterozygous seeds. 9 There are no published estimates of bean outcrossing in Eastern Africa. However, Stoetzer (1984) has recently reported that bean outcrossing in Ethiopia may reach 4.8%. He attributes the outcrossing to carpenter bees (Xylocopa spp.). Onim (1981a) has investigated pigeon pea outcrossing in Kenya and has found it to vary from 25.2% to 94.5%. The level of outcrossing was directly related to the number and activity of insect pollinators at each site (r = 0.994). Pigeon pea (ngagug gajgg L.) is considered to be an outcrossing species so these high estimates are not surprising. They do, however, clearly establish that effective legume pollinators are present in Eastern Africa. Onim (1981b) observed seven insect genera, comprising 24 species, pollinating pigeon peas in Kenya. *Three genera: Megachile, Xylocopa, and Chalicodoma accounted for the majority of flower contacts. ‘Williams (1977) counted on one Xylocopa and one Megachile individual reSpectively, in India, 5,500 and 107,333 pollen grains. Rick (1950) has observed bees in the Xylocopa and AnthOphora genera pollinating tomatoes in Peru. These genera, along with Megachile, also occur in Malawi. Thus they appear to be present throughout the bean-growing regions of the tropics and are likely candidates for outcrossing activity. Several investigators believe various species of thrips (Thysanoptera) are the primary pollinating vectors of beans (Mackie and Smith, 1935; Allard, 1954; Antunes et. a1., 10 1973). Free (1970) observes “Thrips are able to enter legume flowers before they open and so may effect early pollination". Annand (1926) found six species of thrips in the 11 plant species which he observed. The average pollen grains per thrip was 5.8. However, large differences occured between thrips on different species: Acacia, 1 grain/thrip: Alfalfa, 2.3 grains/thrip; Lupin, 26.5 grains/thrip. He states that the pollination effectiveness of thrips depends on, 1) size of the pollen grains; 2) viscosity of the grains; 3) number and species of thrips present; and 4) the attractiveness of the flowers. Attractiveness of flowers appears to be an important factor to bees as well. Leleji (1973) has studied the apparent preference by bees for different flower colors in cowpeas. He observed that bumble bees (Bombus sppJ preferred purple flowers over white flowers in a 3:1 ratio. Honey bees (Apis mellifera), however, preferred white flowers over purple flowers 2:1. These observations confirm that outcrossing by bees is not always a random process. If there is more intra-color than inter-color pollen transfer as these findings suggest, then outcrossing estimates based on flower color markers would tend to be underestimates. The estimation of outcrossing rates play an important role in understanding the genetic structure and evolutionary dynamics of pOpulations. It has been demonstrated that ll considerable variability may exist both within and between lines in self-pollinating species as a result of low levels of outcrossing (Allard, Jain, and Workman, 1969). In Malawi, beans are maintained in often exceedingly complex mixtures. The role of outcrossing in generating and maintaining this variability is a vital component to the understanding of the evolutionary dynamics of these populations. 12 INTRODUCTION Beans (ghgsgglgg yglggris L.) are maintained by Malawian farmers as extremely heterogeneous mixtures. These mixtures may consist of 20 or more different seed types. However, four or five seed types often account for 80% or more of the variability in many mixtures. The remainder is made up of ”off-types", present in lower frequencies. These rare types are suspected to be the result of occasional outcrossing. If outcrossing is occurring in Malawi it could, in part, account for the striking variability which is observed in bean landraces in this country. Rick (1950) has observed that ”Intrinsic pollination relations of a species are often assumed to be the same under foreign conditions, but this is not necessarily the case". Since beans are an introduced species in Eastern Africa, the occurrence of different pollinating insects than are present in other bean-growing regions is entirely possible. Even with a fairly low rate of outcrossing, the potential for wide hybridization is significantly increased when beans are grown in complex mixtures as they are in Malawi. This study was undertaken to estimate the level of outcrossing which occurs among beans in Malawi. Observations and collections were made of suspected pollinating insects, and in the case of thrips, possible flower color preference was investigated. In addition, the l3 segregation from one heterozygous seed, collected from a farmers field, was observed in order to quantify the number of seed types which may arise from outcrossing. MATERIALS and METHODS Two genotypes, differing in flower color, stem pigment, or growth habit were used in each of six isolation blocks (Table 1L. Each line was increased for two generations in isolation in order to ensure homozygosity for the genetic markers. Purple flowers, purple hypocotyl, and indeterminant growth habit were used as dominant gene markers. In blocks 1, 2, 3, 4, and 6 the populations were formed by mixing 300 seeds of each genotype. The seeds were then sown randomly in 10 meter rows with 91 cm. between rows and 15 cm. between seeds within the rows. In block 5, the two lines were planted in alternating, 50 meter rows; spacing was the same as in the other blocks. A11 blocks were isolated from other beans either spatially (by at least 300 meters) or by flowering time. Block 6 was planted at Bembeke research station in Dedza district on 5 January, 1983; flowering occured in mid-Februaury. The other blocks were planted at Bunda College of Agriculture in Lilongwe district on 23 February, 1983 and the plants flowered in mid-April. No insecticides were applied from two weeks before until one week after flowering. Days to first Table Growth habit, flower color, stem color and days of flowering coincidence in the six populations._ Block Line Growth Flower Stem Flowering Habit * Color Color Coincidence 1 C86-8 I white green 15 days CSS-l I purple green 2 C61-4 IV white green 20 days C81—23 IV purple purple 3 C86-8 I white green 11 days C81-23 IV purple purple 4 C61-4 IV white green 10 days CSS-l I purple green 5 P643 I white green 12 days Ntcheu I purple green Khaki 6 Seafarer I white green 4 days Domino II purple purple * Growth habit according to Singh (1981). 15 flowering and days to end of flowering were recorded for each line and the overlap was calculated for each population. During flowering in blocks 1, 2, 3, and 4, notes were taken on the number of thrips in 25 randomly-selected flowers from each line. A t-value was calculated to test if the thrip number means differed on white versus colored flowers. Collections were made of insects observed visiting bean flowers. The thrip species was identified at the Commonwealth Institue of Entomology, London. The bees were identified at the Department of Entomology, Michigan State University. Since the germination in blocks 1 thru 4 was uneven, the number of seeds produced by each line was calculated by dividing their yield by single seed weight, and the proportions were used as estimates of the dominant fin and recessive (q) allelic frequencies. Stands were uniform in blocks 5 and 6 and allelic frequencies were assumed to be the same as the original seed proportions (p = q = 0.5). The progeny of parents carrying the recessive allele from blocks 2, 3, and 6 were sown in sand flats where germination was virtually 100%. The frequency of heterozygotes as determined by those plants showing stem pigment was recorded. Progeny of recessive parents from blocks 1, 4, and 5 were sown in the field and the frequency 16 of the heterozygotes was determined by flower color. Germination in the field was higher than 95%. Outcrossing rates (t) were estimated using the method suggested by Jain (1979): t=H/p where H = the number of heterozygotes (a) divided by the total progenies observed (a + b), and p = the dominant allelic frequency. Jain (1979) gives the variance of this estimate (32) as: 521 =th1 / (a + b)p)'(l - p / p ). Heterogeneity of the various t estimates was evaluated using the test-statistic given by Jain (1979): Chi-square = §:(ti - t*)Szi where t"r =E(ti / $21) /Z(1/szi), and chi-square has k - 1 degrees of freedom. In block 3, Harding and Tucker's (1975) formula was used no calculate the maximum outcrossing estimate that is compatible with this sample size, given that no outcrosses were observed: tmax = Tmax / 1 ’ q ’ 2(Q(1 ”Q) / R) 1/2 where Tmax = 1 - eunl1 " Pl) / N, q is the frequency of the l7 recessive allele and k = sample size from which q was determined. In a related study, the segregation was observed of a suspected heterozygous seed collected from a Malawian farmers' field. This seed, C110-24, was planted at East Lansing in the summer of 1982 and its selfed progeny were subsequently increased at Bunda College of Agriculture. In turn, progenies were grown from each of these plants. The various seed colors, shapes and sizes were recorded as an indication of the array of seed types which may arise from a single heterozygous seed. RESULTS With the exception.of block 6, flowering coincidence was longer than 10 days in all blocks (Table 1). Outcrosses were observed in all but block 3 (Table 2). The maximum outcrossing rate compatible with the block 3 sample size, given that no outcrosses were observed, was calculated to be 0.0032% according to Harding and Tucker's method (1975). Rates of outcrossing varied from this low up to 2.23% in block 2. The average outcrossing rate for all blocks was 0.813%. The highest rate was observed where both parents were climbing beans and the flowering coincidence was longest. Standard errors were relatively small for all block estimates. The chi-square test for heterogeneity of estimates was not significant (.25 < P<.50). 18 Table 2: Estimates of outcrossing rates for the six papulations. Total Number Frequency Outcrossing Standard Block progeny of of dominant rate error scored Heterozygotes allele (t) (s) 1 3295 6 0.466 0.39 0.017 2 6727 50 0.333 2.23 0.045 3 1502 0 0.635 0.003* 0.000 4 499 3 0.605 0.99 0.046 5 1062 4 0.500 0.75 0.038 6 4966 13 0.500 0.52 0.015 * Maximum outcrossing rate compatible with zero observed heterozygotes and sample size of 1502, according to method of Harding and Tucker (1975). 19 The insect collections revealed a diverse array of vectors (Figure 1). Thrips, Megalurothrips sjostedti Trybom. (det. J. Palmer) were present in the flowers of all blocks and indeed, are observed in bean flowers throughout Malawi. Seven bee species, from three genera, Anthophora (1 species), Xylocopa (3 species), and Megachile (3 species), (det. M. Arduser), were observed visiting bean flowers throughout flowering. Figure 2 depicts a female from the genus Xylocopa visiting a bean flower. The thrip counts revealed no significant differences for flower color preference by these insects in the four blocks observed (Table 3). In addition, there was no significant correlation (r = -0.5487) between average thrip number per flower and the outcrossing rates for the blocks. The segregation of C110-24 resulted in over 60 unique seed types based on color and pattern (Figure 3). The single seed weight ranged from 0.237 gms. to 0.562 gms. in the first generation and from 0.286 gms. to 0.771 gms in the second generation. Undoubtedly, many more types would have been retrieved had the progeny size been larger. Xylocopa (species female Xylocopa (species male Xylocopa (species female Eumenine 20 Bee genera - corresponding to Figure 1. spp. 1) spp. 1) spp. 2) wasp Xylocopa (species female Xylocopa (species female spp. 1) spp. 3) Anthophora spp. (species female Xylocopa (species female 1) spp. 3) Xylocopa (species female Xylocopa (species female Megachile spp. spp. 2) spp. 3) (species 2) female Xylocopa spp. (species female 3) (determinator: M. Ardeuser) Xylocopa spp. (species 2) female Megachile spp. (species 1) female Megachile spp. (species 3) female Cleridae beetle 21 Figure 1: A collection of insects observed visiting bean flowers in Malawi. (See key on facing page.) 22 Figure 2: A bee from the genus Xylocopa visiting a bean flower. 23 Table 3: Number of thrips per flower. Average of 25 random flowers of each color. Plot Flower color Plot t - value Number Average Purple White 1 1.80 1.64 1.72 0.35 ns 2 1.00 0.84 0.92 0.517 ns 3 1.24 1.56 1.40 0.745 ns 4 0.36 0.56 0.46 1.157 ns (956,656,4770A/ a: €//O"24~ @ _ j} %% ooepeeoocoO aoowobococe coooooooogo 900-09.006'90 anoooco Co a 09 «no Figure 3: The array of new seed types recovered from the segregation of seed C110-24. 25 DISCUSSION These experiments provide evidence that a low level of outcrossing is occurring in Malawian beans. Low to moderate outcrossing in beans has also been reported in the UJLA. and Brazil. However, these countries grow homogeneous varieties of beans and the outcrossing thus would unite similar genotypes in most fields. In contrast, Malawian farmers often grow extremely heterogeneous mixtures of beans and an outcross is much more likely to unite dissimilar types. If only one genotype is growing in a field, as is often the case in an American bean field, then outcrosses are in effect equivalent to self-pollinations. If two types are present, in equal proportions, then the probability of an outcross uniting the two types is 50%. However, if 10 types are present in equal proportions the chance of uniting different types, should an outcross occur, is 90%. With 30 seed types the probability is 97%. Many farmers in Malawi maintain mixtures of 10 or more seed types, and while these are not present in equal proportions, the probability of joining dissimilar types remains very high. Thus a low level of outcrossing can have a magnified effect in complex genotypic mixtures. This fact, coupled with the wide array of seed types which may arise from.just oneroutcross, as is shown by the segregation of C110-24, may provide a partial explanation of the seed diversity in Malawian bean landraces. Heterozygote advantage, {if present, would increase the level of heterozygosity'in these populations 26 and raise the probability of heterozygotes being joined by outcrossing. Clearly, natural outcrossing, even at the low levels observed here, could be a potent mechanism for generating landrace variability. Several authors have stated that outcrossing in beans is effected by certain species of thrips (Allard, 1954; Mackie & Smith, 1935). Thrips are observed in bean flowers throughout Malawi, however in this experiment the average number of thrips in a field did not correlate with the level of outcrossing. This suggests that other vectors are responsible for outcrossing. Three genera, comprising seven bee species were identified visiting flowers in the test plots. Two genera, Xylocopa and Megachile, are believed to effect cross-pollination of pigeon peas in Eastern Africa (Onim, 1981b). Large quantities of pollen grains have been observed on members of these genera in India. Another genus, Anthophora, also identified in the outcrossing test blocks, is known to be an effective tomato pollinator in Peru (Rick, 1950). These reports lend support to the hypothesis that the seven species observed from these genera may be the important pollinators of beans in Malawi. The highest rate of outcrossing was observed in the test block comprised of two climbing bean genotypes. Barron (1938) observed twice as much outcrossing in plots combining two climbing types as opposed to two bush bean types. This rate may be due to the longer flowering coincidence of 27 climbing beans. .A longer flowering time may, perhaps, allow more pollinators to discover the field and visit the flowers. Since beans in Malawi are known to be 75% to 100% climbing types depending on the district, it may be that the outcross rates observed here, where only one plot included two climbers, are underestimates of the true outcrossing rates. Preliminary data on heterozygosity levels from five Northern region bean-growing areas range from 2.3% to 46.7%. The average percentage of heterozygous plants per farm site was 17.3%. There is a positive, though not significant, correlation (r = 0.439), between frequency of indeterminate types and the level of heterozygosity. The high heterozygosity levels at these sites suggests either LJ The present estimates of outcrossing rates are underestimates, or 2.) There exists some degree of heterozygote advantage for fitness. PART II DIVERSITY IN MALAWIAN BEAN LANDRACES Chapter 2 GENETIC DIVERSITY IN LANDRACES Literature review Multivariate analyses, particulary principal component analysis and canonical variate analysis, of a complex of metrical and qualitative traits have been successfully used to establish diversity patterns in.the germplasm collections of many crop species. Two approaches to this numerical taxonomic technique have been considered by Gilmartin (1967). One is the utilization of large numbers of characters on relatively few individuals, and the other deals with a smaller number of characters pertaining to large numbers of individuals. There is a continuum, of course, between these two extremes but to date most biologists have used the latter to study population structure, geographical variation, and infra-specific variability. Gilmartin suggests this approach be termed ”quantitative taxonomy". A review of the application of various multivariate techniques to populations of several crOp species should serve to clarify the information which can be expected through the use of quantitative taxonomy. Goodman (1967) measured over 1000 plants belonging to 15 races of maize (Zea mays) indigenous to southeastern 28 29 South America. Sixteen morphological traits were observed, six pertaining to the ear, four to leaf size and number and the remainder to the length of various portions of the plant - peduncle, spike etc.. Using Mahalanobis' generalized distance, D2, he clustered the races into groups which corresponded well with divisions based on classical taxonomic techniques. Narayan and Macefield (1976) evaluated the world germplasm collection of 5477 line of chickpea (£193; arietinum) for eight characters: plant type, flower color, days to flowering, days to seed maturity, pod size, number of seeds per pod, weight of 100 seeds and seed color. From the total collection they randomly chose 100 lines representing eleven geographical areas of the world. Utilizing Mahalanobis' D2, they were able to cluster the populations into six groups, with plant type, seed color, and 100 seed weight contributing the most to total divergence. Interestingly, the lines from Mexico, where chickpeas were introduced by the Spaniards just 400 years ago, fell into a separate cluster. This is supportive of the view that human and natural selection under totally different agro-ecological conditions is such that considerable genetic divergence may occur over a relatively short evolutionary period. Hancock and Bringhurst (1978) evaluated thirteen (populations of strawberry (Fragaria vesca), from distinct 30 ecological sites in California, for 25 quantitative traits and two enzyme systems. They were able to show, using analysis of variance and genetic identity techniques (Hedrick's Index, (Hedrick, 1971)), that considerable adaptation to specific environments had evolved. They concluded that Fragaria vesca is a "specialist" - a species of numerous variant ecotypes - which enables it to adapt to a wide range of ecological habitats. A discriminant analysis of 254 individuals belonging to three species of California strawberries assigned each individual to the correct taxon with a 99.61 percent success rate (Jensen & Hancock, 1982). A Mahalanobis' D2 analysis of days to flowering, days to physiological maturity, height, lodging, and grain yield of ten Durum wheat cultivars (Triticum turgidum L. var. durum), revealed no specific association between types from the same geographic area (Lee & Kaltsikes, 1973). The analysis did, however, clearly separate the cultivars into trOpical (short-daylength), and temperate (long-daylength) clusters. A study of 190 rice types from Assam and the northeast Himalayas and four standard varieties - three ”indica" types and one "japonica' type - revealed considerable diversity among both existing cultivars and landraces (Vairavan, Siddiq, Arunachalam, and Swaminathan, 1973). Canonical 31 analysis reduced the 190 types to 42 groups, and Mahalanobis' D2 further reduced these to nine groups. The ”japonica" type formed a separate cluster by itself. The three ”indica” types were found in three separate clusters. A majority of the the primitive cultivars clustered with the ”indica' types, however, several were intermediate between these and the "japonica' type and others were very close to the "japonica" type. This appeared to indicate that a series of intergrades bridging these two rice types has evolved in the northeast Indian rice populations. Malhotra and Singh (1971), using Dz, were able to discern eight clusters among 25 strains of black gram (Phaseolus mungg Roxb.) collected from various geographic areas in India. Genetic diversity did not appear associated with geographic origin. They concluded that considerable diversity exists in this species in India. A principal component analysis of 18 characters measured on 640 lines of finger millet (Eleusine coracan LJ revealed twelve broad groups as a result of plotting the first two standardized principal components (Hussaini, Goodman & Timothy, 1977). The lines from India showed clinal variation, with southern and eastern samples forming the extremes. Finger millets from Uganda formed two distinct groups, which were close together, but isolated from the other material. The authors suggested that the gap between these two groups from the same country may be due to 32 a sampling error or may represent a true biological phenomenon. They stated further that ”It is likely that natural geographical barriers, proximity to the equator, and associated climatic and ecological conditions may have restricted gene flow from adjoining areas. Perhaps of more importance is that distribution of crops in Africa is determined more by the influence of tribes and cultures than by ecological adaptation of plants (Harlan, 1975)”. Interestingly, they found a resemblance between the millets from Africa and southern India, which they suggest may be the result of trade relationships between these two regions tracing back to the Arab slave trade. Nevo, Zohary, Brown and Haber (1979) have studied 28 populations of wild barley (Hordeum spontaneum) collected from various ecogeographical areas in Israel. They concluded that wild barley in Israel is extremely rich genetically, but because of predominant self-pollination the variation is carried primarily by different homozygotes in the pOpulations. This finding, in general, agrees with Imam and Allard's (1965) observations of wild oats in California. They, however, found that considerable variability may also exist within families in self- pollinating species populations under very low outcrossing levels. Nevo, et al. also found that allozymic diversity displayed clinal, regional, and local patterns which were often significantly correlated with environmental variables. 33 Distinct geographic variation for four morphological traits was observed between sites within areas and between areas. Several workers have analysed limited sets of materials of the common bean (Phaseolus vulgaris). Rathjen (1965) identified five ”botanical races” which he felt were genetically and physiologically distinct. Froussious (1970) evaluated 100 bean cultivars for adaptation to the growing conditions of England. He concentrated on characters which have agricultural value: determinancy or indeterminancy, number of internodes, seed size, pod characteristics and flowering behaviour, and arrived at four main plant types. Subsequently, these were modified by Singh (1981) to allow for subdivision according to pod load distribution, branch length and main stem strength. Walters and Evans (1978) used a canonical variate analysis to study eight characters of bean lines from twelve countries. A plot of the first two variates revealed five distinct clusters. One cluster included varieties from Latin American countries (Guatemala, Colombia, Nicaragua, Mexico, and El Salvador) and another cluster consisted of lines from Turkey, Chile, Ethiopia, and the U.S.A.. Cultivars from Iran, Afghanistan, and Malawi each clustered separately, though in the same general area of the plot. The preponderance of climbing beans from these latter three countries was felt to be the reason for this separation. 34 Adams (1977) evaluated 22 commonly grown North American bean cultivars for a diverse set of agronomic and chemical attributes. Employing principal components analysis, he identified each cultivar, on the basis of 36 metrical traits, as a point in a standardized multi- dimensional space. Distances were then calculated between the cultivars and, after weighting by the acreage of each variety, an assessment of genetic vulnerability of the germplasm pool was obtained. The distances between varieties, based on the first six principal components, were significantly inversely correlated (r = -0.68) with the pedigree relationships estimated by Wright's path coefficient method. Bassiri and Adams (1978), using banding patterns from just two polymorphic isozymes, peroxidase and esterase, successfully established the relationships between 34 [L8. bean cultivars which corresponded well with known pedigrees. More recently, Weeden (1984) used 10 polymorphic isozyme loci to discriminate among 88 cultivars of white-seeded snap beans. These polymorphisms allowed the unique identification of 60 percent of the cultivars. To date, no study has been published on the extent and underlying pattern of genetic diversity in landrace populations of the common bean. 35 Beans are apparently maintained as mixtures throughout their center(s) of origin in the Andes (Kaplan, 1981; Zimmerer, 1984 personal communication) and in Mesoamerica (Kaplan, 1981). They are maintained also as mixtures in what may be termed their "secondary center“ of diversity in Eastern Africa. These mixtures are composed of diverse seed types and, in general, fit Harlan's (1975) description of landraces: . . . often highly variable in appearance. Landraces are adapted to conditions of traditional agriculture, they are adapted to low soil fertility, low plant populations, and low yield. Genetic variation within a landrace may be considerable, but it is far from random. The various component genotypes have survived in a region for a long period or they are offspring of lines that have undergone local selection for many generations. Landraces are built up and the components are selected, reassorted, recombined, and rearranged but the local materials are constantly being adjusted to local conditions”. Harlan (1975) has stated that the secondary center of many crops holds more diversity than the primary center, and that this diversity is generated as much by people as by ecological or climatic diversity. Stebbins (1957) reiterates this view for autogamous species: "When a self- pollinating species is invading a new habitat, a larger number of new lines might become established than in the case of a species population maintaining itself in the same habitat". These new lines may arise for two reasons. First, the secondary center may present a new order of diverse ecological sites to the species which then has the 36 opportunity of generating genetic variation sufficiently rich enough to occupy these sites. Secondly, the natural (or human) selection constraints which suppressed variability in the primary center may not do so to the same extent in the secondary center and thus allow the species to expand and exploit the new ecological diversity. In either case, the genetic system of the species must be equal to the task. That is, it must be capable of generating and releasing variants into the heterogeneous environment. In the case of the Malawian bean populations this "capability" would, in part, consist of the existence of outcrossing. In addition, it must be reasoned that "off-types" are not selected out as they arise by the Malawian farmer. However, since the genetic expression of seed coat lags behind by one generation, it is doubtful whether a farmer could ever rid the population of recombinants. The concept of a heterogeneous environment offering different "adaptive peaks” to a population was put forward by Wright (1931). He theorized that the structure which would best equip a population for continuous evolution in either a stable or an unstable environment would have three attributes. It would be: 1. A population of intermediate size, divided into subgroups, each separated spatially from each other but capable of occasional gene exchange. 37 2. Because of this exchange, the subgroups would share a common gene pool, and valuable new mutations could spread through the population. 3. Inbreeding within the subgroups, however, enables different subgroups to occupy slightly different “adaptive peaks”. Population stucture in Malawian bean landraces is analogous to this model. The landraces are usually of intermediate size (10 - 15,000 seeds). Although their subcomponents are not separated spatially, they are isolated by the prevalence of autogamy. However, this isolation is not complete due to occasional outcrossing. Thus, the subgroups (the individual lines in a mixture) do share a common gene pool and valuable new gene combinations may be generated and spread through the population. Finally, self- fertilization reproduces the new recombinations again and again over successive generations and endows the population with the potential to adapt to a wider array of environmental niches. Hence, the Malawian bean populations may provide a paradigm of Wright's Optimal breeding structure. Stebbins (1957) cites several characteristics of self- fertilizing species populations: 1. They consist of several morphological types each of which is represented by many similar or identical types. 2. These types can be recognized by distinctive characteristics so that the population appears like a cluster of closely related species. 38 3. In addition to these dominant types, there may exist several other recognizable types which are represented by only one or two individuals. 4. The dominant types breed true. 5. Occasionally, outcrossing occurs between biotypes resulting in vigorous F1 hybrids. 6. In some species these hybrids are more vigorous than their parents; in others they are not. 7. These hybrids segregate, and the poorly adapted genotypes are eliminated. These postulates were derived from the study of natural populations such as Bromus carinatus (Harlan, 1945), Senecio vulgaris (Haskell, 1953), and Clarkia purpgrea (Lewis & Lewis, 1955). Similar conclusions have been drawn from barley composite crosses, wild oats, and lima bean populations by Allard and his colleagues (Jain &rAllard, 1960; Allard & Jain, 1962; Jain & Marshall, 1967; and Imam & Allard, 1965). Jain and Allard (1960) discovered that the changes in genotypic frequencies occuring over 18 generations of a barley composite cross indicated that the decay in heterozygosity was much slower at several marker loci than that expected for the low level of outcrossing observed. This persistence of heterozygosis, they believe, is due in part to an advantage of the many new heterozygous combinations of genes available upon segregation from the initial hybrids, i.e. heterozygote superiority. They state 39 “This situation is in sharp contrast to that of experimental populations made up of mixtures of homozygous lines; there the number of ”adaptive peaks” is small as shown by Harlan and Martini (1938) and the genetic potential of the population is limited". They conclude that the store of variability in a self-pollinating species is apparently not entirely associated with genetic diversity among multiple homozygous lines, since the storage potential of even a few superior heterozygotes could be substantial. In subsequent work, Allard and Jain (1962) studied the quantitative genetic changes in the composite-cross of barley. They found that the population remained extremely variable even in the nineteenth generation. In addition, they discovered larger within family variances than in the homozygous parent lines. They suggest that ”outcrossing at levels lower than those frequently observed in the so-called self-pollinating species can provide for high genetic variability in populations”. This population structure, however, was not encountered only in the rather artificial populations of the composite- crosses. Imam and Allard (1965) later documented the level of outcrossing and the‘variability'in populations of wild oats (Avena fatua) in California - a predominantly self- pollinating species. Estimates of outcrossing, based on lemma color and rachilla pubescence, varied from one to 40 twelve percent. They again found that the frequencies of genotypes for these monogenic traits indicated that a high reproductive advantage was associated with heterozygosity for the chromosome segments marked by these loci. Their measurements of genetic variability for days to flowering, plant height, number of tillers and seed number per panicle revealed a mosaic of highly localized differentiation. Considerable variation within each population existed and they suggest that ". . . a mixture of genotypes differing in morphology, physiological requirements and growth rhythyms may exploit the available moisture, light, and nutrients more efficiently than uniform populations where competition is between individuals of similar morphology and simultaneous requirements". Kaplan (1981), in a recent paper, has raised the question of bean mixture advantage. He notes that varietal mixtures are seen in collections throughout the highland regions of Mesoamerica and Andean America, the postulated centers of origin of the common bean. While these mixtures appear to be less diverse than the Malawian mixtures, similar adaptive mechanisms may be at work. The bean mixtures which Kaplan has observed consist of four to six distinct types. Mixtures around Cajamarca, Peru collected by Zimmerer (1984, personal communication) consist of between 14 - 16 lines. Berlin, Breedlove and Raven (1974) have collected six different seed types within a single 41 municipio in Chiapas, Mexico, but it is unclear if these were grown as mixtures by the local farmers. Kaplan (1981) states that a mixture dating over 1000 years B.P. (before present) has been excavated from an archaelogical site in the Mexico's Tehuacan valley. He claims its components are identical to contemporary market mixtures in the'Tehuacan valley village of Ajalpan. If confirmed, this indicates that bean mixtures possess a: unique intrapopulational balance which results in their persistence and stability over long periods of time. Kaplan (1981) states that the ”specific set of human and plant ecological factors which account for the existence and stability of mixed plantings is not clear". JHe found no nutritional advantage for rat growth as a result of feeding a mixture versus a single varietyu He did, however, find evidence that the bean lines in a mixture germinate at different rates thereby extending the seedling establishment period. Such an extension would have obvious advantages if the first rains were interrupted. Kaplan also pointed out that mixtures in Latin America are confined to climbing beans. He reasoned this to be so because cultivators of climbing beans are used to harvesting over an extended period - climbing beans mature their pods over a two to three week period anyway. Cultivators of bush beans harvest their crop in a single operation. Thus a staggered ripening time due to differential emergence lends itself more to climbing genotypes. Interestingly, the majority of bean 42 mixtures in Malawi also consists of climbing beans. Kaplan contends "Humans have originated not the components of mixtures but instead have assembled them from pre-existing variants into the mixtures that are adapted to particular production and land use systems. And, "Mixed planting may have been an idea that diffused from one area to another, but it need not have been so. The practice may be polyphyletic in origin . . ." Whether the original beans introduced into Malawi were mixtures, or whether they were single lines later mixed by the Malawian farmers remains moot. What may be profitably investigated is the contemporary variability of these mixtures and the probable mechanisms by which this variability arises and persists. Chapter 3 DIVERSITY IN MALAWIAN BEAN LANDRACES Materials and Methods Germplasm collection The bean lines included in this study were collected in the northern region of Malawi between 25 February and 5 March, 1983 by a Bean/Cowpea C.R.S.P. team consisting of Leo Mapondo, Eric Ayeh and Greg Martin. This trip flollowed a preliminary collection effort in March 1982 by another Bean/Cowpea C.R.S.P. team consisting of Drs. M.W. Adams, (LT. Edje, P. Barnes-McConnell, D. Greenberg and Mr. Augustine Mkangama. Based on the observations of this earlier team, six areas were selected as being representative of the geographical, topographical and cultural diversity in the northern region. These areas, Chitipa North, Chitipa Central, Misuku Hills, Livingstonia Hills, South Rukuru Plain, and Mabulabo Traditional Authority were subsequently revisited. Five of these areas were chosen for the present study (Figure 4). Chitipa Central was excluded due to its proximity to Chitipa North and the desire to have an even distribution of geographical areas . The 1983 team visited each area, contacted the local agricultural officers, and with their guidance collected 43 44 FIGURE 4: Tl-E FIVE GEOGRAPHICAL AREAS FROM w—ncn BEIW LANDRACES wear COLLECTED. l. CHITIPA ~0er ll. SourH RUKLRU PLAIN 2. MISUKU HILLS 5. MABLLABO TRADITIONAL 3. LIvmesromA HILLS “mm” 45 beans and supporting information from each of five farms. These farm sites were not selected with any particular criteria, but an attempt was made to sample sites which were evenly distributed throughout each area. The scenario of a typical visit was as follows. Through the interpretation of an agricultural officer, we would introduce ourselves and the purpose of the visit. Next, we would ask the farmer a series of questions about the village and her farming practices, especially those practices pertaining to bean landraces. This information, along with our own observations, was entered on a collection form (Figure 5). Special attention was given.1x> obtaining accurate information on the location of the site so that , if required, it could be revisited. After these formalities, we asked if we could collect a sample of the farmer's beans. The Malawian farmers, without exception, responded with generosity and helpfulness at this point, allowing us to tramp through their fields, or if the beans had already been harvested, offering us beans from their storage pots. Thus, two distinct collecting methods were used. If the beans were still in the field, we would walk a transect across the field taking one pod from each plant that we passed. If the field was too small for this to yield a sufficient amount of seed (most were), then we would repeat this procedure leaving a few meters between the new transect and theiold. The average sample size was 80 - 100 pods with this technique. Many farmers, eSpecially as we proceeded 46 34.5.0. I BL'NDA COLLEGE - BEAN ENLASM COLLECTION FOP." CollectIOn .'. :31 :31: a! Caller-sorts) .- GEM Runner 2. Date or Coiiecusnidsy.xnt.t.yee:;: 75-2-3T J. xenon a: coueeuen: Northern J. Dresser.- gmn'gg morrnL 5. ?:a¢.::;ons1 Aussor;:y(:'.a.1: 6. 'I:‘.‘.'sger:.'osesr;: _Smbi 7'. Far-rs Nssu: 51“.; 3“...”- feu.’e :- sale aescngsasn a! Site “3143453 . auatdemrid): 9'32'5 is. soc-1 sales.- ‘ .onqisudeIgvz-ad): ALL-11‘“? :9. div-restarts! seed) : 0.. " J . ' u:.-.:de(-n J“) ' 4240 e- aaontsa or nose seed rapes) Topography.- b- .diuollo so 29 ' I plan (givgg plain) :- .‘ovlless can .‘0 ‘ J o- ”112:9 :- 3.111: 20. Grove: sears.- d- noun 9 bush 6 unreliable; s- North :- lass e- . 5- 5mm: a. a.“ .(._,_ m; 31- Cups.” seem: used (In. :5 was) ' . Urumqi: nndfwita: } s- some]. a pure stead :- poor 6 droughts; (sandy) 22. Pleasure Method: 24. 5011 Erosion: e on :s‘dges ’- m' rsdges 6‘ amt: ” yes 9 so . General Site Description: 23' m” .‘ieid - not: d- as: valleys bliss: e- var-as :- mits: :- cam. :- onto: d- 0850:: 24. Pests: ' .I'e:u.-ser Applied: s- nevus :- speeds en ;es: em 5- beetles :- shale; .- aw.- . Stones.- “ maesous b- ocssssonal 25. l-Fisssson (nodules) : - 6 absent s- present (strove) b- ;rosen: (moss-n) L :- absent cmm's: Planted late December. Hgvested end of Februiv. Randomly collected one pod per plant in field. (O.P.P.) . "e " v 2: 339182 FIGURE 5: AN EXNPLE OF A BEAN GEFN’LASM COLLECTION FORM. 47 southwards, had already harvested their crop. In this case we would take several "random grabs“ of seed from the sack or pot that was shown to us. With this technique we usually' obtained 400 - 500 seeds. Both sampling strategies provided a random collection which,‘we felt, was representative of the farmer's landrace. Before leaving the field a soil sample was collected, following the procedure recommended by Shickluna (1981). The soil was subsequently analysed at Bunda College of Agriculture for percentage of sand, silt and clay. Further analysis was carried out at Michigan State University for nitrate, potassium, phosphorous, pH, calcium, magnesium, zinc, manganese and organic matter. Finally, an altitude reading, in meters, was taken at each site using a Thommen altimeter. The seed samples were then placed in cloth bags, labelled by farm site numbers, and returned to Bunda College of Agriculture. Experimental Procedures For this study, three of the five farm sites from each of the five geographical areas were selected (Figure 6). These fifteen sites were chosen by refering to the collection map and selecting sites which were evenly 48 '5 6 3 2 1 \I\ - 7 8 ‘1) I 9 I \ ( g I N \ l) ’7 ,' l." ,1? 4' I"! -v/' ’7 \ l 15 \ ) ,/ l| ‘; FIGURE 6: THE 15 COLLECTION SITES IN NORTl-Em MLAWI. l.Mweneloza.vi11age 2.8yumbi village 3.1penza village 4.Kasambra village 5.Kasumbi village 6.Kawotola village 7.Msuka village 8.Salawe village 9. 10. 11. 12. 13. 14. 15. Chaminono village Chisusu village Zebedea village Kumwenda village Chirwa village Mafundeya village Jengolima village 49 distributed within each area. No attention was given to the relative seed diversity of the particular collections. The fifteen sites are described and characterized in Tables 4 and 5. The results of the soil anaylsis for each site are given in Table 6. It was decided, based on seed type variability, that 25 bean lines from each site would be sufficient to portray the landrace diversity and yet would still be manageable experimentally (Figure 7). Since 30 seeds were required of each of the 375 total lines, a seed increase was undertaken at Bunda College between 24 March and 10 July, 1983. A 66 seed random sample was taken from each of the 15 collections for this purpose. Two 33 seed plots for each site were established, with the seeds being sown randomly at 15 centimeter spacings in rows 90 centimeters apart. To minimize the chance of outcrossing in these plots the insecticide Sevin was applied every four days during flowering; no pollination activity was observed. At maturity, the first 25 plants in each site-plot were individually harvested and the seed was threshed, treated with Dieldrin and Thiram and placed into envelopes. The 375 lines were grown at Bunda College of Agriculture between 26 July and 14 November, 1983 in a randomized complete block design with three replications. Also included in this study were four Michigan lines: Seafarer, Domino, Black Magic and Swan Valley. The maximum, minimum 50 Table 4: Names and locations of the 15 collection sites. Area Site Altitude Latitude Longitude Village No. (meters) Chitipa North 1288 Mweneloza 1 1280 9°36'S 33°03'E Syumbi 2 1292 9°32'S 33°07'E Ipenza 3 1292 9°28'S 33°02'E Misuku Hills 1560 Kasambra 4 1494 9°39'S 33°33'E Kasumbi 5 1609 9°42‘S 33°34'E Kawotola 6 1578 9°44'S 33°31'E Livingstonia Hills 1159 Msuka 7 1152 10°35'S 34°07'E Salawe 8 1139 10°40'S 34°04'E Chaminono 9 1188 10°46'S 34°02'E South Rukuru Plain 1299 Chisusu 10 (1396 11°59'S 33°37'E Zebedea 11 1292 11°56'S 33°27'E Kumwenda 12 1210 11°52'S 33°23'E Mabulabo T. A. 1276 Chirwa 13 1402 12°21'S 33°34'E Mafundeya 14 1219 12°34'S 33°37'E Jengolima 15 1207 12°36'S 33°34'E Table 5: Temperature, rainfall and daylength for the five collection areas*. 51 Temperature** Rainfall Day- . 1ength*‘* (Celsius) (millimeters) Area Low High § Dec. Jan. Feb. i Chitipa North 15.7 29.7 21.9 1790 2037 2189 2005 12h.37' Misuku Hills 15.7 29.7 21.9 2940 2583 2571 2698 12h.38' Livingstonia 15.7 30.0 20.8 2314 2556 2236 2368 12h.42' South Rukuru 13.3 29.3 21.0 1875 2495 2026 2132 12h.45' Mabulabo T.A. 13.3 29.3 21.0 1905 2465 2492 2287 12h.47' * Source: National Meteorological Service, Chileka Airport, Blantyre, Malawi. ** Temperature figures are means for December, and February. January *** Daylength is for 15 January, which would be the approximate date of flowering for beans planted in early December. 52 Table 6: 5011 characteristics of the 15 collection sites. E A A U 0) OJ 9.". L A L- \ U C. U 0) (U L- G: c\2 .5 \ U \ A v-1 0) t'.‘ (it E L- sne v a E f. .5: 3 E No. 9 V' f V' a. V' S 3 O E V E E Q) E V L: :3 5 c. m 0 -¢ 8 *4 C- 0 ° 9’ N as N .r: a: :3 U) v c: 01-! u c. U: v" Q C t: m 'U >. U U) H: U I: U at 9.1 H C: '0 v-‘ C u ---I 00 c: C 00 u C: .— -~ :: .c c I: r: -- r: L- -—4 m L: U: c. a. a. U 2 N I o z 1 68 13. 19 5.7 24 185 1173 126 0.9 28.0 1.90 0.008 2 48 31 21 6.2 92 396 2600 480 3.6 38.0 3.60 0.012 3 78 9 13 6.4 136 244 1387 216 3.6 35.0 2.24 0.017 4 ll 62 27 5.5 24 379 1387 149 0.9 48.6 4.50 0.210 5 45 40 15 7.0 201 729 4267 514 7.3 62.2 5.40 0.174 6 48 34 18 5.9 15 362 1700 320 0.9 47.6 5.52 0.375 7 22 4O 38 5.6 11 488 1900 514 1.8 35.0 3.67 0.454 8 38 36 26 5.9 105 738 1387 370 1.8 53.3 4.95 0.174 9 58 16 26 5.7 98 354 1387 273 1.8 15.5 4.29 0.413 10 80 10 10 6.1 75 111 640 69 0.9 18.2 1.79 0.158 11 74 20 6 5.8 141 185 747 80 0.9 31.0 2.62 0.281 12 80 10 10 6.1 145 168 747 103 0.9 18.2 2.26 0.309 13 56 31 13 6.0 39 362 1387 198 3.6 39.0 3.22 0.606 14 80 10 10 5.8 189 111 640 57 0.9 20.0 2.17 _0.211 15 80 6 14 7.7 172 111 2400 216 0.9 47.6 3.83 0.454 CA* 58 24 18 6.1 98 328 1583 246 2.0 35.8 3.46 0.257 MA** 59 17 24 5.4 70 405 n/a n/a n/a n/a 2.24 n/a * Average for collection sites ** Average for Malawi. (Sillanapa, M., 1982). Note: 1. 53 Figure 7 Seed types of the 375 bean lines included in the study. Each column constitutes the 25 lines collected from the farm site indicated at the top of the chart. Two mistakes were made in mounting the seeds to the board. Lines 16 and 17 from site 3 should be reversed. Lines 11 and 12 from site 5 should also be reversed. 54 Md/dfi/z'an dean i779: 74'0”?" A{M%%wwéym.flmm& l23456789M”flfl”fl ““Veneubx \\\ F\o .0.V'~-.;g-.50 I I- l o l «In .LI. /7 I.) 4.) u - J 1' 71 I.) to - (9;, ll ) ' i 21. - 1 23 . “) iu.m l“ ,, .j’ .I. ‘3 3 ,J l ‘1' I) 1.593;).. 1%.)- , "' J 7 J. . J C U 9‘ 9329:9990 J’ww'OJOOJou 5 "3‘93 99 . Q's-1m. “$99®I.JJCOCQ c.020thwoooa ’ 799’..$.J.J "-9 '90 39 11;).g..- ,J ” 7391.4...OV ’ '-.Q 1323...)..wfl. , 4.9.1). J. J :19 1)“) I J.J.Vj. ) 45:2: , )Q . 0. J-.- p -" ' ‘2 ’39 3.9)- 1.. W990'OOJOO-M 94134 0!)-..» erba'OOJ-.. . ')“ j. t ..\.,_’i. )- 9. " “°--"..oo ’94'000--- 55 and mean air temperatures in degrees Celsius during the the egrowing season were: July - 23.7, 11.1, 14.8; August - 24.8, 11.1, 15.3; September - 28.9, 14.1, 20.2; October - 29.9, 15.5, 22.2; November - 31.6, 18.2, 24.4. Total rainfall over the growing season was 10.6 millimeters. The seeds of each line were sown 10 centimeters apart, with 10 seeds per plot, in rows 10 meters long and 90 centimeters apart. Thus, each row contained 8 plots and a block consisted of 50 ten—meter rows. Germination was greater than 95% in the field overall" Three weeks after planting, in the process of taking hypocotyl length and seedling dry weight measurements, the plots were thinned to six plants. Thinning was accomplished by removing the first, fourth, seventh, and tenth plant from each plot. This left three groups of two plants each, the two plants being separated by 10 centimeters and the two-plant pairs being separated by 20 centimeters. A pole was then placed between each two-plant pair. Coupled with the 90 centimeter row spacing, this planting arrangement minimized competition between plants and lessened the number of poles required for staking. Standard bean cultural methods as practiced at Bunda College of Agriculture were followed throughout this study. Each 10 meter ridge was fertilized with 273 grams (300 kg/ha) of 20-20-0, N-P-K. The plots were furrow irrigated 56 (one or two blocks at a time) whenever required - usually once a week. Leafhoppers were controlled with alternate applications of Sevin and Lebicyd. Diseases were not a problem since this was the dry season, however, Plantvax was sprayed late in the season to control a slight rust infestation. The plots were kept entirely weed-free. Four weeks after planting, and occasionally throughout the season, the ridges were 'rebanked" to aid in supporting the plants. The plots were harvested one block at a time beginning on 31 October and continuing until 14 November, 1983. Character measurements and scores Measurements and scores were recorded for each line during the growing season. The recording was done one block at a time to minimize within block variation. An attempt was made to choose characters which would portray a broad picture of the architecture of the germplasm. Detailed descriptions and comments for each character follow. Summaries of the quantitative and qualitative characters are given in Table 7. Phenological characters 1. Days 59 emergence Days were counted from the date of first irrigation (27 July for replications 1 and 2, 28 July for replication 3) to 57 Table 7: Character cooes and summary of quantitative and qualitative character measurements. Code Character Experimental Scale unit Phenological characters P1 Days to emergence plot P2 Days to first flower " P3 Days to end of flowering " P4 Flowering duration " P5 Days to physiolog. maturity ' Morphological characters M2 Leaflet length M3 Leaflet width M4 Leaflet shape M1 Leaflet area 3 plants M5 Petiole length ” M6 Hypocotyl length 1 - 3 plants M7 Seedling dry weight " M8 Hypocotyl diameter 3 plants M9 Nodes on main stem 2 - 3 plants M10 Seed length 1 seed M11 Seed width Agronomic characters A1 Single seed weight 100 seeds A2 Seeds per pod 15 pods A3 Pods per plant plot A4 Seed yield plot/9Plants I A5 Seeds per plant Qualitative characters Ql Hypocotyl color plot 02 Standard color ' Q3 Wing color Q4 Seed primary color 1 seed 05 Growth habit plot number centimeters2 ' length / width centimeters centimeters grams 1/32 inch number centimeters OI ‘ See text for explanation of qualitative scales. 58 the date when at least 5 of the 10 seedlings had emerged and their primary leaves had unfolded. 2. Q§y§ 59 first flowering This was scored on a plot basis. The days were counted from the date of first irrigation up until at least 50% of the plants had opened at least one flower. Data were taken every other day. 3. Egyg 59 Egg 93 flowering Recorded on a plot basis. When 50% or more of the plants had finished flowering the number of days from first irrigation was recorded. The type 1 beans could be easily scored for this character. The indeterminate lines presented greater difficulty as these were occasionally still flowering at harvest. For these, an arbitrary criterion was set. When flowering had ceased below the tenth node, this date was recorded as end of flowering. For a type 4b this may be misleading, but as a whole it should accurately indicate differences among lines. 4. Duration 9f flowering The difference between days to end of flowering and days to first flower was calculated. 5. Days 39 physiological maturity This was recorded on a plot basis, the interval being 59 figured from the first day of irrigation to that when 90% of the pods had turned yellow or to an intermediate color (e.g. purple). Morphological characters 6. Leaflet area (cm.2) Leaflet area, in centimeters squared, was measured using a Li-Cor portable leaf area meter, model LI-3000. The measurements were taken when the third trifoliate had unrolled on 90% of the plants in the field. The measurements were taken on the first trifoliate leaf, the terminal leaflet. In most cases, the first three plants encountered in each plot were the one's measured. However, if one (or two) of these plants were diseased, stunted or otherwise grossly abnormal, the next normal plant was chosen. Three plants were measured for each entry in each replication. After measuring, a hole was made in the leaflet with a paper punch. 7. and 8. Leaflet length Egg giggg (cm.) Using the punched hole as identification, these measurements were taken on the same leaflet as the leaf area. The length of the leaflet was measured from the point where the pedicel joins the leaflet to the tip. The width was measured at the widest point. 6O 9. ngflet shape Leaf shape in general is deltoid, but considerable variation may be observed from long, narrow leaflets to short, broad leaflets. The ratio of length/width was calculated to portray this variation. 10. Petiole length (cm.) This measurement was taken on the same leaflet as the leaflet measurments. The petiole length was taken from the leaflet base to the abscission line. 11. Hypocotyl length (cm.) Hypocotyl length was recorded on three plants removed during thinning. In general, these were the first, fourth, and seventh seedlings. The length was recorded from the end of the hypocotyl to the cotyledonary scar. 12. Seedling dry weight g; four weeks (gms.) The three plants which were used for the hypocotyl length measurments were then taken, their roots rinsed in water to remove excess soil, and placed in paper bags. These were then placed in a drying oven at 110 °C for 24 hours. The dried weight of all three seedlings was then recorded, and the average weight per seedling was calculated. 61 13. Hypocotyl diameter (1/32 inch) This measurement was taken on the first three plants in each plot, just before flowering. A pair of calipers, with increments of 1/32 inch was used, although readings were taken to 1/64 inch. The calipers were placed one centimeter below the cotyledonary scar. The number was entered as 1/32 inch, thus 8.5, for example, is actually'l7/64 inch. 14. Node number pg main stem 35 physiological maturity This was recorded in replication 1 only. Two plants per plot were counted for indeterminate types, three plants per plot for determinate types. 15. and 16. Seed length and seed width (cm.) Seed length and width were recorded for one random- normal seed per entry. Agronomic characters 17. Single seed weight (gms.) Seed weight was calculated from a loo-seed sample of each line. 'The fresh seeds were first weighed, then oven- dried at 110 °C for 48 hours, then weighed again. Seed weight was adjusted to 10% moisture. 18. Seeds per pod This was calculated from a 25-pod sample from each plot. 62 19. Pods per plant This was calculated from the total seed yield, the single seed weight, and the number of seeds per pod. 20. Yield per plant (gms.) Seed yield per plant was calculated from the seed yield per plot divided by the number of plants per plot, and adjusted to 10% moisture. 21. Seeds per plant Fecundity was calculated by dividing adjusted yield by the adjusted single seed weight. Qualitative characters 22. Seed primary color Seed color was scored using a modification of the classification Scheme in Descriptors for Phaseolus vulggris g; (Anon., 1982) . 1. Black 9. White with purple tinge 2. Brown 10. Blue 3. Maroon 11. Green 4. Grey 12. Red 5. Yellow 13. Pink 6. Cream/Buff 14. Purple 7. Pure white 15. Other 8. Whitish These color categories were considered to encompass a range of hue. A descriptor chart, with examples of the 63 range of colors ”permitted" under each category, was used as an objective criterion (Figure 8). Thus, 'yellow' ranges from pale yellow, bright yellow, to a sort of greenish- yellow. Likewise 'brown' encompassed everything from light brown to sepia. A distinction was made between 'red' and 'maroonf, both common colors and easily distinguished in the Malawian germplasm. The standard for 'white' was Seafarer, the standard for 'whitish' was Swan Valley. Considering the complexity of seedcoat color genetics in P; vulgaris LL, this latitude in color categories probably disallows a Specific color-to-genotype correspondence. Still, the beans within one category, though varying in hue, likely have some genes in common. 23. Banner color The banner (or standard) color of each line was scored when the 'days to first flower' was recorded. This was taken on a plot basis and only for replication 1. An arbitrary scale was followed: 1. White 2. Greenish-white 3. Pink 4. Lilac 5. Purple There was little ambiguity using this scale. Flower color was always scored early in the morning to avoid mistakes due to fading or darkening. 64 SEED DESCRIPTORS Fore, PHASEOLuS VuLGARIS L. SEED COLOuR/ _ M; / fl/ac/c - 9 ,Wfff/y Z {Bros/n, 59 Q :___.:.J L... /0 B/ue, 3 Ma-..» 15:? Q // are... 4- 6m] :9 £3 E... 9 /Z fled 5- Ye//o./ Ga- it. A, L. /3 Pin/(9 6 ("m/flax; it. L. L, /4‘/0cn;o/e, 7/00reM‘I/e t. L /5 0M.»- 8 ”ff/IA, g 73.: E... Q- ‘0 $3117 19‘ Figure 8: The seed color 'descriptor chart'. 65 24. Wing color It was decided to take wing color along with banner color Since the two are rarely the same. The same 1 - 5 Scale was used. 25. Hypocotyl color Hypocotyl color was scored using an arbitrary scale: 1. Green 2. Pink 3. Purple 26. Growth habit This was actually a score for determinancy (score = 1) versus indeterminancy (score = 2). It was decided not to score for specific growth types (e.g. types l,2,3,4) Since the staking and "training” made it difficult to consistently differentiate between types 3 and 4. 27. Segreggtion Several lines were observed to be segregating for hypocotyl color, flower color, and growth habit during the course of the study. At harvest several more were seen to be segregating for seed color, pattern and Size. Segregation for each character was recorded in the data book. Samples of some of the seed types which segregated from some of the lines were saved and used in constructing seed charts depicting probable recombinational events at representative farms. 66 Data Analysis A) Variation pf the bean landraces The objective of this section of the analysis was to provide a quantitative description of the variability between areas, between sites within areas, and between lines within sites. The frequency of each character state “Lg. green, pink, purple hypocotyl) for the qualitative traits was calculated for each site» Histograms were constructed to depict the area-to-area variation across the northern region. Means for the 21 quantitative variables were calculated for the northern region as a whole, the five geographical areas, and the 15 sites. Separate analyses of variance for the 18 replicated character measurements were performed to test for significant differences between areas, between sites within areas, and between lines within Sites. Least significant differences (LSD) were calculated for those characters having significant differences between areas and between Sites within areas. Those traits which were statistically significant were used in a principal components analysis (PCA) . 67 A modification of a PCA program contained in the Genstat Package of Statistical Programs, Rothamstead Agricultural Experimental Station, was used toicalculate distances between areas and between Sites within areas. This modified program, originally used to calculate distances between a sample of U.S. bean cultivars by Dr. M;W..Adams (1977), is described in more detail in a later report (Adams & Wiersma, 1978). The program results in principal component scores (PC'S) based on standardized data. These scores are normalized, and then used in calculating the distances between taxonomic units. It was decided t0ma xuwawnnnOHQ Ho.o on» an ucmuouuwo haucnofiuwcmwm one muouonunzo Add a h~.vm oo.om~ o~.mma oh.mma cm.m>a o>.mma cm.mma om.HNH cm.mo~ om.vo om.mo o~.mam 05.5mm ov.vma om.m- oN.NNH m< mu.m hm.~m om.mm ma.>o mo.o> hm.~b cm.mo mu.am >~.hm wm.ov v~.ov vm.no vo.vm cm.mm ob.om mh.mm v< om.m o¢.om mm.m~ mc.om hv.Hv m>.om vm.vv Hm.m~ vm.vv v~.om om.o~ mm.mv ~m.vm om.o~ vm.o~ mN.mN md mm.o om.v mm.v om.v mN.v nv.v oa.v o~.v mm.m ma.v ~v.v hm.v mv.v wo.v mv.v H~.v ~< cc.o mv.° vv.o ov.c av.c mm.o ov.o ~v.o v~.o mv.o mv.o om.o mm.c vv.c no.0 no.0 H< + m>.o oh.o sh.o uh.o om.o $5.0 mn.o hm.c mb.o mh.o ch.o v5.9 ob.o Nh.o mh.o HHS + om.a vm.a wv.~ ~m.~ ~m.a bv.H mv.H mo.H mo.a mm.a m~.H ~m.~ ~m.~ m¢.A. mm.a ca: + om.om o~.mH om.w~ om.- o~.ma oo.m~ om.HH oo.wa oo.h ow.m oc.o~ oo.m~ cm.ha oo.m~ om.m~ m2 ov.o vm.m mv.m mm.o mo.» mm.» om.h v~.m mm.» av.m mm.m mo.h mm.n mm.h ~m.n vo.m m: ma.o Nv.~ w~.~ m~.a wa.a ov.H mo.H mm.~ mn.o mm.a ~m.a No.a om.o mH.H ma.H mH.H h: mv.o No.5 wa.w do.» No.5 m>.c mm.o ~o.p mo.w mo.o mm.o mw.w mv.o Ho.» vm.w va.» ox wv.o hN.m o~.m mm.m mw.m mo.m mm.m Hm.m mm.m vh.m -.m ma.m vH.m mm.m mv.m va.w m: mc.o cm.~ m¢.H oo.~ ov.H mm.H vm.H Hm.H mv.~ cm.~ Hm.a mm.a wv.H vm.H ~o.~ om.H ex mv.o om.h cm.» ~n.o ma.h ma.> Nw.o mm.o om.m Nv.h w~.n Hm.o Ho.o mm.o Hm.o oh.o m2 on.o mm.HH w~.oa o~.~a mm.oa hh.oa Ho.o~ mm.o~ He.c mm.- mo.o~ am.m No.m mm.oa vv.o~ hm.o~ N: mm.m Nv.mm mn.ov mH.om ma.mv mm.mv on.~v mm.hv om.~m mo.mm v~.ov mn.~v mv.~v nm.mv mo.ov hm.mv Hz mm.~ ~m.mm ov.mm em.hm vo.~o~ mm.ooa vm.voa o~.mm vq.ma om.mm mo.om m~.om mm.ma mo.~m mm.am Hm.mm mm mm.a om.ma ~m.ma mH.hH om.- vm.ma om.o~ vu.m~ vo.a~ ov.~a vm.v~ om.mu cm.n~ mo.- vo.- o>.- cm vv.~ Nm.mn Ho.ah ~m.on m~.wh oo.vh av.oh Nm.vo vm.mn Ho.Hw oo.aw mo.an 0H.mm mw.mo co.mm mu.mo mm vo.H oh.vm nm.~m mh.mm mm.mm NH.vm em.mm on.mv em.~m w~.mv mm.mv a~.mm mm.om mo.ov om.mv mm.ov mm >>.o mh.~a ~m.~H vw.NH hm.~H mm.~a Ah.ma mm.~a mv.~a ww.ma Hm.ma mm.~H Hm.~H om.- mo.~a mv.~a Hm ma va ma NH Ha oH m m n o m o m N H “Ho.o n my ++ mooo eomq mumm uwuonumco .mumuomumco o>fiumuaucmam Am uou mouflw cofiuowaaoo cmwzumn mwocoumuuwc com: and manna 84 113:! .mmwhu_b_hu000nonOum 00.0 was 00.0 on» an ucoo0u0cm0m a. . c 000.0: 000.0: 500.0 000.0 550.01 000.0: 000.0: a 000.0 uce0m mom 00000 a 000.0 050.0 050.0: 550.0: 000.0 000.0 4 000.0 000.0 0.0.1.:0 0000 000.0 a 050.0 000.0: 500.0 000.0 000.0 000.0 000.0 uce0m gem noon 050.0: 000.0: 000.0: a. 000.0 000.0 000.0 000.0: 000.0 005 non 00000 000.0 000.0: 000.0 a 000.0: 000.0: 000.0 500.0 000.0: unu0oa coe- 000000 000.0 000.0 000.0 a. 005.0: 000.0: 000.0: 000.0 000.0: nuo03 ovum 000.0 000.0 050.0 a. 000.0: 000.0: 050.0: 000.0 000.0: camco0 0000 000.0 a 000.0 000.0: 000.0: 000.0 000.0 a 000.0 000.0: eeuu c0ee :0 00002 000.0 000.0i a 000.0 a 000.0: 000.0: 000.0 000.0 000.0: u0u0fl~0o 00uoooa>= 000.0 500.0: a 000.0 a 000.0: 000.0: 000.0 000.0 000.0: u000ea ago 00000000 000.0 000.0 050.0: ac 005.0: 000.0 000.0 a. 050.0 000.0: numce0 05uooom>0 500.0 000.0: 000.0 000.0: 500.0 000.0 050.0 000.0: numco0 o0o0uom 000.0 000.0: 000.0 000.0: 000.0: 000.0 000.0 050.0 amen» u00uaeq 000.0 050.0 000.0 a 000.0! 500.0 500.0 050.0 000.01 nuv0a ue0ue00 000.0 000.0: 000.0 050.0: 000.0: 000.0 550.0 000.0: cuoco0 ue0unoq 050.0 500.0: 000.0 a 000.0: 000.0: 000.0 000.0 000.0: eoue 0000000 4 500.0 a. 000.0 000.0 000.0: 000.0 000.0 050.0 000.0: .0u0uauel .mo0o0nanm cu ammo 000.0 050.0 a. 005.0: 000.0: 000.0 000.0 000.0 000.0: co0uouao 0:0ueao0m a 000.0 a 000.0 050.0: 000.0: 000.0 000.0 a 000.0 000.0: 0:0uoao0u 00 one 00 05.5 a. 005.0 a 500.0 000.0 000.0: 000.0 000.0 500.0 000.0: uoao0u uau0u cu exec 000.0i a 000.0 050.0 000.0: 000.0: 000.0: 500.0 000.0: cocoouoao 00 0000 0o>o0 . neuueueao eo:U0ueq 000000000ueue: euauu0z aa0eauuom maOuonmnosm an 0 0:00 ausu0u0< .000na0uua eu0e 00000000 one nuouoeuono o>0uou0ucmav 00 can cooauon aco0ue0ouuou «o0 o0noe 92 by the Pcianalysis which separated sites into northern and southern groups. Seven out of the eleven morphological traits were negatively correlated with potassium content of the soil, as was single seed weight. The number of seeds per pod was positively correlated with potassium content. Heterozygosity level exhibited several positive correlations with phenological characters. Seed yield was positively correlated with sand percentage, soil phosphorous, and latitude. Altitude, to some extent an indication of temperature in the tropics, showed a significant positive correlation with fecundity, and a significant negative correlation with petiole length. C) Landrace population structure The separate principal component analyses of the 15 landraces revealed an extraordinary complexity in the populational structure of most mixtures. Each landrace exhibited a wide range in relationships between its constitutive components. The distances for the two lines showing the greatest and the least degree of relationship from each site are presented in Table 17. In three cases the two lines most closely related were not similar for seed type, although similar seed types were present at these sites. In only one case were the two lines that were most distantly related actually the same seed type. Distances ranged from a low of 0.0749 to a high of 1.3215. Since 93 Table 17: Minimum and maximum distances between two-line pairs at each collection Site, based on the 21 quantitative plant characters. Site Minimum Lines Maximum Lines 1 0.1898 9,10 1.1184 2,17 2 0.1886 9,10 1.1542 20,23 3 0.1616 12,22 1.1973 16,25 4 O.23l7+ 1,4 1.1211 20,23 5 O.1612+ 13,24 1.1821 6,25 6 0.1385 1,23 1.2102 10,22 7 0.1896+ 15,24 1.2653 13,23 8 0.2261 4,12 1.2477++ 20,21 9 0.1230 3,13 1.1969 10,22 10 0.0749 23,25 1.2707 5,8 11 0.0930 1,24 1.3215 6,13 12 0.1796 11,19 1.1898 10,17 13 0.1480 2,14 1.2662 10,25 14 0.2411 19,21 1.2137 2,6 15 0.1681 7,21 1.1557 4,25 + Seed types of these pairs are different. ++ Seed types of this pair are the same. 94 beans are predominantly self-pollinating it would be expected that at least some lines would be identical by descent, i.e. pure lines. In the case of line pair 23 - 25, from site 10 and line pair 1 - 24, from site 11, for example, this is very likely the case. However, there were other seed types which, though appearing homogeneous did not appear closely related according to the PCA distance. Five of these homogeneous seed classes were analysed further. Table 18 and 19 present the separate analyses of variance for the 18 replicated characters measured on homogeneous seed classes from two sites. The brown and purple seed types, from site 1, were found to be significantly different for three traits and one trait, respectively. The brown types differences were all in agronomic traits which are inter-related by mathematical transformations and are thus highly associated with each other. Thus, the brown and purple seed classes from this site do display a high level of homogeneity for quantitative, as well as qualitative traits. Homogeneous seed classes, however, did not always appear similar for metric traits. The yellow seeded class from both site 1 and site 6 and the tan/red class from site 6 displayed many significant differences in all three character categories, thereby suggesting that these homogeneous types are, in fact, not pure lines. The average distances between individual lines in each of the five phenotypically similar seed types are given in 95 Table 18: Analysis of variance for the 18 replicated plant characters in three homogeneous seed types from collection site 1. Character Seed type code ++ Brown Purple Yellow Df 11 6 4 Mean squares M1 M2 M3 M4 M5 M6 M7 M8 0.51 0.27 1.07 2.76 3.10 * 2.27 * 2.82 6.76 8.23 5.06 16.33 5.57 14.13 11.83 8.23 90.50 88.64 268.88 * 1.10 2.45 4.78 * 1.03 0.83 1.18 0.02 0.09 0.04 * 1.28 0.23 0.04 0.37 0.93 1.39 0.10 0.06 0.31 * 0.59 0.55 1.00 * 0.005 ** 0.001 0.0001 0.10 0.16 0.25 75.80 20.05 78.36 * 424.36 * 150.40 432.00 1272.80 * 316.49 2539.69 8* ** Significantly different at the 0.05 and 0.01 probability levels, respectively. ++ See table 7 for description of character codes. 96 Table 19: Analysis of variance for the 18 replicated plant characters in two homogeneous seed types from collection site 6. Character Seed type code ++ Tan / red Yellow Df 10 13 Mean squares Pl 2.80 1.50 * P2 3.21 2.86 * P3 2.32 18.54 ** P4 4.63 29.92 ** PS 12.34 * 8.29 M1 186.18 * 41.50 M2 1.67 0.76 M3 1.32 0.82 M4 0.04 0.06 ** MS 0.79 0.98 * M6 0.74 0.95 * M7 0.21 * 0.28 ** MB 0.30 0.56 Al 0.004 ** 0.008 ‘* AZ 0.33 * 0.29 ** 83 64.21 ** 55.94 ** A4 194.94 * 114.36 A5 997.88 ‘ 742.93 * *, *‘ Significantly different at the 0.05 and 0.01 probability levels, respectively. ++ See table 7 for description of character codes. 97 Table 20. The brown and purple classes, which displayed few differences for the individual traits, also appear most closely related and have the smallest standard deviation. The remaining three groups have higher mean intra—class distances and higher standard deviations. Since the usual range of distances for different degrees of relationship is unknown based on these 21 traits, the distances between the four Michigan lines included in this study whose pedigrees are known are presented in Table 21. Black Magic, Domino, and Swan Valley were all derived from one original cross and are thus full sibs. They have, however, undergone several cycles of selection for slightly different characteristics. Nevertheless, the small distance between Black Magic and Domino (0.2595) testifies to their close relationship and similar black seed type. The distances between these two and Swan Valley, a navy bean, is expectedly a little greater. These three lines are distantly related to Seafarer according to this analysis, an observation supported by their known lack of pedigree relationship. The brown-seeded types from site 1, based on the AOV's and their average intra-group distance, are closely related, suggesting that they may be sister progenies of a single pure line. Table 22 presents the PC scores for several two- line pairs from this group. It can be seen that even though 98 Table 20: Mean distance and standard deviation between lines within each of five homogeneous seed types. Site Number Mean Standard Number of significantly / of distance deviation different traits in Seed type lines the AOV's. 1 Brown 12 0.5573 0.1540 3 ( A1, A4, A5 ) 1 Purple 7 0.6246 0.1441 1 ( P2 ) 1 Yellow 5 0.7459 0.1600 8 ( P2, M1, M2, M4, M7, M8, A2, A5 6 Tan / red 11 0.6629 0.1801 8 ( P5, M1, M7, A1, A2, A3, A4, A5 6 Yellow 14 0.6486 0.2307 12 ( P1, P2, P3, P4, M2, M5, M6, M7, A1, A2, A3, A5 Table 21: Distances between Michigan lines, based on the 21 quantitative plant characters. Domino Black Magic Swan Valley Seafarer Domino 0 0.2595 0.7335 1.1827 Black Magic 0 0.5802 1.0601 Swan Valley 0 1.100 Seafarer 0 99 Table 22: Normalized scores for the first 6 principal components, and distances between three pairs of brown-seeded lines from site 1. Distance Principal component Line 1 2 3 4 5 6 7 -0.3130 0.1528 -0.1961 -0.0305 0.0779 -0.0417 0.2215 22 -0.2870 0.1213 -0.0159 -0.0826 -0.0015 -0.1186 Difference 0.0260 0.0315 0.1802 0.0521 0.0794 0.0769 3 -0.1340 0.1545 0.3293 0.2387 0.2325 0.1457 0.4697 18 -0.1918 0.1273 0.0982 0.0291 -0.0224 -0.0874 Difference 0.0578 0.0272 0.2311 0.2096 0.2576 0.2331 12 -0.0066 -0.l673 0.1508 -0.1360 -0.1251 0.3792 0.5785 13 -0.0631 -0.1078 0.1072 -0.2161 -0.2804 -0.1644 Difference 0.0565 0.0595 0.0436 0.0801 0.1553 0.5436 100 two lines may appear closely related based on the AOV‘s and the PCA, they may still display differences on individual PC axes e The scatter plots of the first two PC's for each site are presented in Appendix B. These plots accounted for an average of 52.0% of the variance. In general, distinct clusters.were observed which consisted of phenotypically similar seed types. The landraces, taken as a group, appear to represent a continuum, beginning with simple mechanical mixtures, proceeding to landraces exhibiting incipient genetic intergradation of major seed classes, then to landraces showing distinct "hybrid swarms” and finally to landraces appearing to consist of predominantly recombinant types. Examples of each of these stages are depicted in Figures 17 - 24. Segregating lines are denoted by a dash. Figure 17 and 18 (sites 8 and 11), depict simple mechanical mixtures. None of the lines from these landraces segregated and distinct clusters of fairly uniform seed types were observed. Sites 2 and 7 are also examples of simple mixtures (Appendix B). Figures 19 through 22 portray the first step towards introgression. In each of these landraces at least one line segregated, and in each case the positioning and the phenotypes of the recombinant seed types suggests the putative parents of the segregating line. Thus for example, in Figure 20, site 1, line 12 appears to have been the 101 Figure 17: Principal component plot of the landrace from site 8 - an example of a simple mechanical mixture. » ,4:,._l, Figure 18: Principal component plot of the landrace from site 11 - an example of a simple mechanical mixture. 102 ,g“ 0,. -.._ __ 0.1.0: _ Figure 19: Principal component plot of the landrace from site g - an example of incipient genetic intergradation. ~ 1 Figure 20: Principal component plot of the landrace from site 1 - an example of incipient genetic intergradation. 103 result of a natural outcross between a yellow and the brown- seeded type. Line 9, at the same site, appears to be the result of a cross between a brown-seeded and a purple- speckled type. Figure 21 (site 5) depicts this incipient genetic intergradation even more clearly. In this landrace there are recombinants which appear to have resulted from three crosses between three major seed classes at this site. Line 14 was probably a cross between a zebra-striped bean and the tan/red mottle bean (line 6). Line 9 seems to be a cross between one of the small, hilum-spotted types and a zebra-striped bean. Line 16 was likely'a cross between a solid brown bean and a zebra type. Line 20 may be another segregant from the tan/red mottle bean and a zebra type. For Figure222 (site 14) the segregating progeny from lines 15 and 16 were unavailable, however this landrace too exhibits a pattern of nascent genetic intergradation. Other examples of this early stage are sites 3, 9, 13 and 15 (Appendix B). Figure 23 (site 4), appears to be an example of a later stage of genetic intergradation. A distinct cluster, actually a 'hybrid swarm', is apparent in this landrace, consisting of lines 5, 8, 9, 18, 19, and 24. Progeny of four of these segregating lines were available and they clearly show that a unique seed class has arisen. This farm also appeared to have two distinct yellow seeded types. Figure 21: Principal component plot of the landrace from site 5 - an example of incipient genetic intergradation. Figure 22: Principal component plot of the landrace from site 14 - an example of incipient genetic intergradation. PRINCIPAL COMPONENT I 0: .Ag» ..8 9m\ ‘. v 1 .la\ -0 g 011‘.) Ly' _\ a C I S a .. "’ - ’ : 5 fig.) UL” ‘ '2 .2: *4 4 1'2“ 3 g :3 ~ " :J J. '10 ,4 'IP g E U .3 3 E I: .. Figure 23: Principal component plot of the landrace from site 4 - an example of a later stage of- genetic inter- gradation where a 'hybrid swarm' has formed. :‘(ZSCIPAL COMPONENT [ 4.ensre 00 N O '0 A -HISUSUAV k. U a N .0 -J M R \Lfib PJMN PRINCIPAL COMPONENT [I i Figure 24: Principal component plot of the landrace from site 10 - an example of very extensive genetic intergradation. 106 Figure 24 (site 10), depicts a truly complex array of new seed types which have arisen as a result of outcrossing. Eighteen of the twenty-five lines in this landrace segregated. Viewed broadly, this landrace appears to have consisted, at one time, of reddish and purple lines (to the right in the plot) and brown and white types (to the left). Considerable genetic intergradation has already occurred as shown by the many "off-types”. Site 12, where 17 of the 25 lines segregated also represents this later stage of genetic intergradation (Appendix B). Chapter 5 DIVERSITY IN MALAWIAN BEAN LANDRACES Discussion A) Variability 1Q the landraces. This study arose from a simple observation - bean landraces in Malawi were found to be extremely diverse for seed color, pattern, size and shape. As far back as 1938 Harlan and Martini observed that mixtures of self- pollinating varieties quickly became dominated by one or two genotypes. More recently, Vieira (1975) found that bean mixtures consisting of six homozygous lines were dominated by a single line after just four generations. Yet bean mixtures are present in Malawi and apparently have existed for a very long time. How can these conflicting observations be reconciled? What forces are generating the variability in these bean landraces, and what factors are influencing its maintenance? This study has addressed four specific questions. First, what is the extent of the genetic diversity in the Malawian bean landraces and is there a pattern to the variation over the northern region? Secondly, what, if any, are the advantages of variability for the specific characters measured in these landraces? Thirdly, is there evidence for site-specific:or regional adaptation to certain environmental variables? And lastly, what role does outcrossing play in generating and maintaining the variability in the landraces? 107 108 This study has confirmed that a vast store of genetic variability is extant in the Malawian bean landraces for both qualitative and quantitative characters. Seed characteristics - visually the most striking indication of this variability - were found to be extremely diverse within individual landraces. However, a broad regional pattern of variation is evident. The landraces from the northernmost sites consisted of predominantly yellow, brown, or tan/red mottled beans. The more southernly sites grew primarily reddish beans. Whether this demarcation in types is due primarily to environmental forces or to consumer preferences cannot be answered here. Kaplan (1981) has observed that ”seed variations serve as genetic markers that enable humans to supply the energy for selection that maintains the observed diversity". The existence of a vernacular for specific seed variation in Malawi indicates that preferences have developed for certain seed types. The presence of cosmopolitan seed types (tan/red mottle) suggests that a distribution network does or has existed and that types having general consumer acceptance (or environmental adaptation) may spread over a wide area. Thus, either consumer prejudice or environmental factors likely do play a role in the present-day distribution of seed types. The possibility that the introduction of seed types into new areas has been completely random should, in general, be discounted. The Livingstonia Hills area, however, where several anomolous mixtures occurred,may be an area where 109 there has been a more random introduction of seed types. These unusual mixtures, perhaps not coincidentally, were collected within 20 miles of the Livingstonia Mission. This mission was founded in 1881 and has had a considerable impact on the education of the local people. A distinct possibility exists that this mission has been responsible for the introduction of beans from other areas of the country and even from British colonies in other parts of Africa. The preponderance of white pea-beans at site 8, similar to those produced on a large scale in Tanzania in the late 1930's, would tend to support this hypothesis. Segregation of bean lines was observed in all but four landraces. In two landraces, those from sites 10 and 12, 70% of the lines were found to be segregating for seed type, flower color or hypocotyl color; other quantitative traits were also likely segregating. The great range of seed types which may arise from a single heterozygous seed was documented in Part I of this thesis. Thus, it may not be necessary to assume an inordinately high rate of outcrossing in order to account for this great seed diversity. :Although the outcrossing estimate of 1-2% arrived at in the outcrossing estimation plots may be low due to the low proportion of climbing-longer flowering types in the study, even this rate would have a dramatic effect in a population of 5,000 to 10,000 plants. Finally, since the genetic expression of seedcoat lags behind one generation, a farmers attempt at removing I'off-types" from.his population would 110 be, in good part, thwarted. And even if a farmer were to attain a uniform seed type an occasional outcross from his neighbors' field - often only a few meters away - would again contaminate his stock. Indeed, from this point of view, the maintenance of mixtures appears less puzzling than the maintenance of pure lines. The lack of any distinct pattern for flower colors and hypocotyl color probably indicates that there is little, if any, human or environmental selection for these traits. Segregation of both white and colored flowers from several lines would suggest that if there is a preference for flower color by pollinating bees, it is not exclusive. Relatively high prOportions of the three hypocotyl colors in every area may suggest the existence of a balanced polymorphism for this trait in many landraces. Indeterminate beans predominate in the northern region. This is not surprising since the vast majority of farmers grow their beans in association with maize, and climbing bean are probably better adapted to this system. However, it is interesting that all but three landraces also contained determinate beans. Vieira (1975) in a study of a six component mixture, found that climbing beans quickly dominated the mixture: the bush beans were reduced to 0.5% by the fourth generation. He believed that the determinate 111 types would have been eliminated in one more generation. Vieira, however, grew his mixtures at a population of 250,000 plants per hectare. Under this high density, competition between bean plants would be extreme. ‘Vieira mentions that Brazilian farmers usually'maintain beans at between 15,000 and 75,000 plants per hectare associated with 20,000 to 45,000 maize plants. In Malawi, beans are grown at between 20,000 and 60,000 plants per hectare in association with maize (ije et. a1., 1980). It is clear that the the competition in a Malwian bean/maize field, where there is an average of 3-4 bean plants per square meter, is not intraspecific. That is, beans do not compete with beans so much as they compete with maize. As the ecological niches of these two species are decidedly different, it is understandable how both bush and climbing types may coexist in a landrace. Indeed, they have different ”strategies” for competing with maize. The bush beans are generally earlier and thus mature before the maize has shaded the lower zones in a field. Indeterminate beans, in contrast, climb up the maize, thereby avoiding shading stress to some degree. Ultimately, since direct competition between bean types is avoided, both may survive. Several sites, (e.g. sites 4, 9, 14) contain high proportions of both bush beans and climbing beans. This suggests that the maize landraces at these sites are perhaps less vigorous, that they are planted later than beans, or that the farmer plants maize at relatively low density. 112 Phenological characters, in general, show a high level of variability within landraces, between sites and even between areas. Days to first flowering, days to end of flowering, and days to physiological maturity were significantly different between areas at the 0.01 probability level. Kaplan (1981) found that bean mixtures from Oaxaca, Mexico contain components with markedly different germination rates. He observed that a mixture may extend the germination period 2.3 times over that of a single variety. The components of Malawian landraces studied here differed in the rate of emergence from 10 to 21 days. In Malawi, where the first rains are often interrupted, this emergence differential could have obvious advantages. The variability in other phenological traits, by spreading out the occurrence of the critical flowering and seed-filling periods, would similarily reduce the risk of crop loss. This, in fact, has a direct corollary to Imam and Allard's (1965) wild oat populations wherein they believed the variability increased the adaptive peaks which the population could occupy and allowed a more efficient use of the available moisture, light and nutrients. Great morphological diversity is evident in these landraces for all 11 traits measured. Considerable adaptive importance may be associated with leaf size when a genotype is competing with maize for available light. Similarly, petiole length differences may play a role in the 113 positioning of the leaf to the incidence of sunlight, an important capability under shaded conditions. Greater hypocotyl diameter might be expected to be an advantage where beans are grown up on ridges, as they are in Malawi, where a tendency toward lodging may exist. Despite these supposed advantages for a certain morphology, the landraces are rarely uniform for a particular trait. This diversity, again, is likely a result of the presence of many adaptive peaks in these populations. Thus, although small leaves may more easily orient to the sunlight, they are associated with small seeds and thin hypocotyls which may have disadvantages with resPect to beanfly (Ophiomyia phaseoli) attack. Conversely, large leaves and long petioles may hinder leaf orientation, but they are associated with large seededness which generally results in early vigorous seedling growth. Thus, over the course of the year from sowing, seed-filling, and harvesting to storage and consumption, each morphological configuration may possess certain advantages which permit it a selective edge. Yield and yield components, of great importance to modern plant breeders, may be of less interest to Malawian farmers. Occasionally farmers will mention that a certain type is ”high yielding” but much more often qualitative reasons for growing a certain seed type will be offered (e.g. ”good taste”, ”store well", ”like the color“). It is 114 striking, and puzzling, that great differences in fecundity exist between lines in individual landraces. Small seeded types generally produce more seeds per plant than large seeded types. And yet, in most mixtures, large seeded types predominate. Part of the explanation may be that farmers say they prefer large seeds and hence they may be applying a strong selective pressure for these. Why do they maintain small seeded types at.all then? This again may fall into the elusive arena of consumer preferences. Farmers will occasionally cite qualitative reasons for growing certain small seeded types (64% small, hilum-spotted types are said to be drought resistant). llverall though, Malawian farmers simply may not care if their landrace is uniform for seed size; they appear to apply very little selection for other traits either. One dichotomy may be presented here. It is interesting that in maize, a difficult crop in which to maintain purity, Malawian farmers desire (and succeed in maintaining) a very uniform grain type. A whitish, floury seed which has the "right" qualities for making gsima is preferred. Now, maize and beans form the primary carbohydrate and protein source in Malawi and are thus culturally important. Why the farmers insist on great uniformity in one and seem to prefer great diversity in the other is a puzzle. Especially since selection for uniformity in a self-fertilizing species, such as the common bean, should be relatively easy. 115 Storage is likely a critical stage in the maintenance of bean mixtures. Roos (1984) observed differences in viability in an eight-variety mixture after storage under high temperature and high humidity - conditions similar to those on a Malawian farm. Using computer simulation, he found that after five generations the mixture was dominated by Cherokee Wax, a black-seeded bean. However, even after fifteen generations, two»other varieties still persisted. The two white-seeded lines in his study were rapidly eliminated in the first few generations. His study may provide a clue to the persistence of certain types in the Malawian mixtures even though they are less fecund than other lines. High temperature, high humidity, and perhaps most importantly bean weevil attack, may be very potent differential selective mechanisms which are operating to maintain the polymorphisms in these bean populations. To sum up this discussion of individual landrace diversity, it may be said that the presence of the extraordinary variability in Malawian landraces documented in this study indicates that specific advantages must exist for bean mixtures. These advantages may be due to a buffering effect which enhances disease and insect tolerance, or to a differential response to drought or light stress. The ability of a landrace to spread its nutrient, moisture, and light requirements over time by having differing phenological "schedules” may also confer specific 116 advantages. Finally, the advantages perceived by the farmer are very centrally involved here. Consumer preferences are often fickle, but they too may be based on very perceptive observations of the crop ecosystem. .Still, it is not beyond reason that farmers have simply grown to like the colors and patterns of their diverse landraces and thus maintain them, in part, for aesthetic purposes. The introduction of pure lines or homogeneous seed types that may replace these complex mixtures should be viewed with great circumspection. Although considerable variability may exist within individual landraces, there are regional patterns of diversity which were uncovered by the principal component analyses. Surprisingly, areas which shared similar seed types were not always similar when viewed multivariately. For example, area 2 (sites 4, 5, 6) and area 5 (sites l3, 14, 15) were most closely related according to the principal component distance analysis. Area 2, however, consists of primarily yellow, mottled, and small seeded white beans and tan beans. Area 5 contains mostly large seeded reddish beans. Apparently these areas share many similarities for other agronomical or phenological traits. This observation reinforces the fact that seed characteristics alone are but a small indication of the genetic architecture of these landraces. The principal component analysis of site means uncovered a fairly sharp differentiation between the nine, 117 more northernly sites and the six southernmost sites. The southern sites formed a fairly tight cluster, while the northern sites, though more heterogeneous, formed another cluster. Days to first flowering, days to end of flowering and days to physiological maturity were significantly different between areas and the PCA of the five phenological traits alone exhibited a similar and even clearer differentiation between the northern and southern landraces. The possibility that there has been specific adaptation to daylength in these landraces will be discussed in the section on environmental / plant character associations. The PCA of site means also revealed a pattern of diversity within areas. The landraces within areas may be very similar (e.g. sites 1, 2, 3; sites 10, 11, 12; and sites 13, 14, 15), but they may also be extremely dissimilar as in sites 4, 5, 6 and sites 7, 8, 9. The possibility that landraces 7, 8, and 9, which were collected near the Livingstonia Mission, may have been influenced by recent European introductions has already been mentioned. Why sites 4, 5, and 6 are so different from each other when they were collected within a relatively small area is unclear. Genetic drift may provide one explanation. Another possibility is that the extreme ethnic diversity in the Misuku Hills (over six language groups occupy this area) has given rise to varied cultural preferences for seed types. 118 To sum up, there appears to be evidence for broad regional adaptation in these landraces. However, individual farmer preferences, recent outside influences of Europeans, and ethnic differences within certain areas have also likely played a role in combining the contemporary components of the mixtures. B) Environmental 1 Plant character associations The many significant correlations between environmental variables and plant characters indicate that there is some site-specific adaptation occurring in the Malawian bean landraces. Several phenological traits (days to first flowering, days to end of flowering and days to physiological maturity) were significantly positively correlated with latitude, a finding supported by the PCA. The more southernly landraces were generally later for these traits. Daylength difference between the northernmost and the southernmost areas is only 10 minutes on January 15, which would be the approximate date of flowering for beans planted in early December. That the landraces are adapted to such small photoperiodic differences is not wholely unlikely. Bunting (1975) in a paper on the phenology of crOps, has stated that very specific adaptation to photoperiod exists in sorghums in northern Nigeria. The sorghums there flower progressively later as one proceeds northwards over a fairly short distance - 250 to 300 miles. 119 There may be other reasons beyond photoperiodic adaptation which are involved with these phenological traits, however. It is possible that the farmers prefer earlier beans in the north because they take better advantage of the rainfall regime or they fit in better with their farming system. This correlation with latitude may then be the result of agricultural practices of the various ethnic groups rather than a: direct environmental response. Another significant, positive correlation is that between soil phosphorous level and seed yield. Haag, Adams, and Wiersma (1978) found differential reSponse for phosphorous use efficiency in 124 dry bean genotypes from North and Central America. They suggested that "The response of a genotype and the nutrient supplying capacity of its native soil may not be unrelated”. ‘This appears to be the case in these 15 landraces. The landraces from high phosphorous soils are higher yielding, at least when grown under high phosphorous conditions as they were in this experiment. If a high phosphorous level does indeed exert a positive selective force for yield, it would be necessary to assume that the landraces were grown on high phosphorous soils continually for directed selection to occur. Since most farms in northern Malawi are quite small, it is likely that considerable soil homogeneity does exist from year to 120 year as a farmer rotates her fields and that a consistent selection pressure is present. Many morphological characters were significantly negatively'correlated with potassium content of the soil. High potassium was associated with small leaves, small seeds, narrow hypocotyl, short hypocotyl, short petioles and few nodes on the main stem. No obvious connection is known to explain these correlations, but the high number of correlations suggests that potassium level may be exerting a strong selective force on the morphology of individual landraces. The percentage of lines segregating at each farm ("heterozygosity level") is positively correlated with days to emergence, days to first flower, days to end of flowering, days to physiological maturity, nodes on the main stem, and pods per plant. Seed yield is also»positively, though not significantly, correlated with heterozygosity level. Interestingly, seeds per plant is not correlated with heterozygosity (r = -0.014). These correlations indicate that the landraces with a greater occurrence of recombination generally contain later maturing bean lines. It was observed during the course of this study that segregating lines often produced progeny which were taller and much more vegetative than the norm. The lack of correlation between fecundity and heterozygosity level 121 suggests that heterozygote advantage for fitness is not present in the landraces studied here. C)Landrace populational structure The separate principal component analyses of the 15 landraces, in many cases, revealed a considerable complexity in the interrelationships between the individual bean lines. Although seed characteristics were not used in these analyses, similar seed types, in general, clustered together indicating a degree of homogeneity within seed classes. Since beans are a predominantly self-fertilizing species it is to~be expected that a landrace will contain some lines which are identical by descent, i.e pure lines. In many instances this is undoubtedly the case. However, very large PC distances were observed between many line pairs which shared a similar seed type. This suggests that these lines, though phenotypically the same for seed type, actually carry different genes for the array of other characters recorded in this study. Before discussing quantitative differences within such homogeneous seed classes, one example of a striking qualitative difference will be noted. The landrace from site 3 contains two lines which are identical for seed type - medium-large, oval and solid white. Line 2 however, is a determinate bean and line 25 is indeterminate and a strong climber. Such gross qualitative differences within a seed type were observed for flower color as well. 122 It would be valuable to know what PC distance is expected between two entries of the same pure line in a study such as this. In other words, what sort of error exists. IHowever, since several pure lines pairs were not included in this study, such a "benchmark“ distance is unavailable. Lacking this, it may be highly subjective to set a specific distance below which lines may be considered pure lines and above which they may be assumed to be different. Adams (1977), in his principal component analysis of U.S. bean cultivars, found an average distance of 0.2048 between Red Mexican lines known to be closely related (Wright's relationship coefficient, R = 0.8750). This analysis, however, was based on very different morpho- agronomic and chemical traits than were used in this study and it may be inappropriate to use this distance in evaluating relationships among components in the Malawian landraces. The distances between the four Michigan lines included in this study are probably the closest standard available that can serve this purpose. Domino and Black Magic, phenotypically very similar, were derived from the same original cross, as was Swan‘Valley. Although these lines have undergone selection for slightly different traits, they can be expected to share many genes in common. The relatively small distance of 0.2595 between Domino and Black Magic indicates a close relationship. The distances between Black Magic and Swan Valley (0.5802) and Domino and Swan Valley (0.7335), however, indicate that selection 123 within a full-sib family can result in considerable differences between lines and hence relatively large distances. The minimum distance between line pairs from each landrace ranged from 0.0749 to 0.2411. In three cases these pairs did not share a similar seed type. Nevertheless, a distance of 0.0749 (site 10, lines 23-25) or 0.0930 (site 11, lines 1-24) very likely indicates that the members of these pairs are actually derived from the same pure line. Many homogeneous seed classes were found to contain lines, between which the average distance was considerably higher than these pairs. Five such classes were analysed further. Site 1 contains 12 lines which shared an identical brown, oval seed type, 7 lines which shared a purple- speckled seed type, and five which shared a solid yellow seed coat. Site 6 contains 11 lines of tan/red mottled type and 14 lines of a similar yellow oblong type as that from site 1. Separate AOV‘s for the 18 replicated, quantitative characters for each of these seed classes revealed both homogeneity and heterogeneity within different classes. For example, the brown class and purple-speckled class from site 1 showed very few significant differences between lines for the 18 traits, indicating that they may, in fact, be pure lines. The remaining three seed classes, however, had significant differences for very many traits suggesting they are far from being pure lines. The average distance between 124 the brown types was 0.5573, that between the purple types 0.6246. The more heterogeneous classes had larger distances: 0.6486 (yellow's, site 6), 0.6629 (tan/red mottle) and 0.7459 (yellow's, site 1). The standard deviations within the classes with larger distances were considerably higher than in the homogeneous classes. These analyses demonstrate two things. First, very homogeneous seed classes may have average intra-group distances much higher than the low distances which were mentioned earlier as being indicative of a pure line. Secondly, phenotypically similar seed classes may contain lines which are extremely variable for an array of quantitative traits. Thus, similar seed type among a group of lines in a landrace should not be taken as evidence that they are pure lines. Bean researchers in Malawi should not rely solely on seed diversity when evaluating the potential variability in a landrace. Indeed, even within the very homogeneous brown seed class from site 1, where few significant differences were observed, considerable variability may exist. The normalized scores for the first six PC’s revealed that although two lines may be closely related, as judged by the PC distance (eug lines 7 and 22, with a distance of 0.2215), their scores may vary widely on individual axes. Since a PC axis, in this study, is a multivariate portrayal of a particular germplasm aspect, these data suggest that these 125 brown lines, though assumed to be pure lines, may actually differ in many ways not uncovered by a univariate analysis. At the beginning of this discussion two conflicting observations were presented. One is that experimental mixtures of varieties of self-pollinating Species are generally found to be quickly dominated by the most competitive line. The other observation - that which has been documented in this thesis - is that mixtures of diverse bean lines are present in Malawi and have apparently persisted there for a very long time. The principal component analyses of the individual landraces, I believe, reveal a possible explanation for this paradox. The principal component analyses of the 15 landraces uncovered what appear to be several stages of genetic intergradation*. This process probably prevents one line from dominating the landrace. The degree of genetic intergradation likely depends on three factors: the number of seed types originally composited by the farmer, the rate of outcrossing which is present in the area, and the survival rate of the recombinants. Sites 8 and 11 are examples where distinct groups of more or less pure types are being maintained. These types may have recently been The term 'genetic intergradation' refers to the gradual process of gene exchange resulting from natural hybrid- ization between genotypically different entities in ‘ a landrace. - 126 composited by the farmer or there simply may be little or no outcrossing occurring in these areas. Sites 1, 5, 6, and 14 exhibit incipient genetic intergradation of the major types being grown at these farms. .As the new recombinant lines begin to segregate, and if the segregants are well adapted, we would expect ”hybrid swarms" to develop and to occupy presently unoccupied ecological niches in the population. If a principal component plot may be viewed as a multivariate portrayal of the present environmental niche of the bean mixture, then the gaps between seed type clusters would represent as yet unoccupied adaptive peaks within the niche. Site 4, then, is an example where a hybrid swarm has developed in an otherwise unoccupied adaptive peak. This swarm, at present consisting of a small number of segregating progeny, may be expected to expand and along with recombinants from other natural crosses, eventually "fill in" the gaps in the environmental niche of the bean mixture at this particular site. The principal component plot of site 10 is perhaps a portrayal of such exploitation. Here the many recombinants, along with the nonsegregating and apparently "original“ lines, are well distributed throughout the plot and analogously throughout the environmental niche at this site. Thus, in a sense, a continuum of genetic intergradation is present in these bean landraces. However, it should be said that it is probably not likely that all landraces are 127 progressing towards the extensive genetic intergradation found at site 10. Rather, this site, along with site 12, may be the result of an inordinately high outcrossing rate or a relatively diverse original population. Landraces such as those from sites 8 and 11 may remain as simple mixtures as the result of a lack of outcrossing or possibly due to a very vigilant farmer who is selecting out off-types. In conclusion, it appears that outcrossing is a powerful mechanism by which variability is generated and maintained in these populations. Although not all landraces seem to be influenced by outcrossing, those that are not are generally much less diverse for seed types and a range of other characters. Outcrossing is probably a recurrent force in many bean populations and is likely reSulting in variability beyond even the original intention of the farmer when she first mixed several bean types. The result of this outcrossing is not only a colorful and diverse bean mixture but also a genetically dynamic and possibly adaptive one. SUMMARY AND CONCLUSIONS This thesis has documented the variability in 15 bean landraces and has attempted to address the question: ”What forces generate and maintain variability in the Malawian bean landraces”. A summary of the salient findings follows. 1. A low level of outcrossing (l - 2%) was found in experimental plots in the central region of Malawi. Many lines collected from the northern region were found to be segregating and it is likely that outcrossing in this region is as high or higher than that observed in the experimental plots. 2. Seven bee species, from three genera, were observed visiting bean flowers. The same species are known to effect pollination in pigeon peas in Kenya and it is likely that they are the primary pollinators of beans in Malawi. 3. The segregation of a single heterozygous seed collected in the northern region was documented. Over 60 unique seed types were recovered. It is clear that even a single outcross may generate a tremendous diversity of seed types in a landrace. 128 129 The Malawian bean landraces are composed of many lines which are often extremely diverse for phenological, morphological, agronomical and qualitative traits. It was hypothesized that variability in phenological traits, by extending critical growth phases for the landrace as a whole, lessens the risk of crop loss to the farmer. The fact that both determinate and indeterminate beans coexist in these landraces is seen as an indication that very little intraspecific competition occurs within a Malawian bean landrace. .Rather, as beans and maize are generally grown in association, the primary competition confronting beans is that against maize. The fact.that bean plant populations are very low in Malawian fields was suggested as a possible reason that mixtures containing components of various competitive abilities are able to persist. The principal component analysis revealed a north - south clinal pattern in the 15 landraces. This cline appeared to be attributable to the sharp differences observed in flowering and maturing times for the beans from these two zones. 10. 11. 12. 130 The principal component analysis resulting in between- area distances indicated that geographical proximity was unrelated to the distance between landraces based on 21 quantitaive traits. Many statistically significant correlations were observed between environmental variables and plant characters. Several phenological traits were associated with latitude, indicating a possible photoperiodic adaptation. Numerous negative correlations between soil potassium and morphological traits suggest that this nutrient may be exerting a selective force on the morphology of the landraces. The PC distances between lines ranged from 0.0749 to 1.3215. The distance between two closely related Michigan linse, Domino and Black Magic, was 0.2595. It was suggested that lines separated by distances as low as 0.0749 (site 10, lines 23-25) or 0.0930 (site 11, lines 1-24) probably represent a pure line. Some highly uniform seed classes were found to have statistically significant differences between lines for only a few traits. Although the average distance separating these lines was considerably higher than that suggested as being indicative of a pure line, it was postulated that these too may constitute pure lines. 131 13. Some uniform seed classes were found to have significant differences for very many traits, and correspondingly higher average PC distances. These classes are very likely not pure lines. 14. It was concluded, based on the analyses of uniform seed classes, that bean researchers in Malawi should not evaluate the relative diversity of a landrace on the basis of seed characteristics alone. Highly uniform seed classes may contain enormous variability for other quantitative traits. 15. The low level of outcrossing present in these landraces was found to have a profound impact on the generation of variability. A continuum of genetic intergradation was observed, progressing from simple mechanical mixtures to hybrid swarms and eventually to very extensive recombination. 16. This genetic intergradation between the original farmer-composited bean lines in a mixture was postulated to result in a fuller exploitation of the environmental niche filled by this species in the maize/bean association. The Malawian bean landraces represent a vast and unique store of genetic variability which could be utilized by bean 132 researchers throughout the world. To the Malawian farmer this variability probably offers a degree of insurance against crop loss. The recombination occurring in these landraces due to outcrossing, combined with the various selection pressures exerted by the rigorous environment, serves to continually improve the adaptation of these landraces. Thus, the Malawian landraces are genetically dynamic populations which are shaped both by the local people and the environment. The replacement of these complex entities by uniform pure lines should be viewed with great circumsPection. LITERATURE CITED Literature cited Adams, M.W. 1977. An estimation of homogeneity in crOp plants with special reference to genetic vulnerability in the dry bean, Phaseolus vulgaris L.. Euphytica 26:665-679. Adams, M.W. and J.V. Wiersma. 1978. An adaptation of principal component analysis to an assessment of genetic distance. M.S.U. Ag. Ex. St. Res. Bull. E-347. Allard, R.W. 1954. Natural hybridization in lima beans in California. Proc. Amer. Soc. Hort. Sci. 64:410-416. Allard, R.W. 1962. Population studies in predominantly self-pollinated species. II: Analysis of quantitative genetic changes in a bulk hybrid p0pulation of barley. Evol. 16:90-101. Allard, ILW., S.K. Jain and P.L. Workman. 1968. The genetics of inbreeding populations. Adv. in Gen. 14:55-131. Annand, P.N. 1926. Thysanoptera and the pollination of flowers. Amer. Nat. 60:177-182. Anon. 1982. Descriptors for Phaseolus vulgaris L. Int'l Board for Plant Genetic Resources. FAQ. Rome. Antunes, I.F., J.G.C, da Costa, and E.A. Oliviera. 1973. Natural hybridization in Phaseolus vulgaris L. in Pelotas, Brasil. Bean Imp. Coop. Ann. Rep. 16:61-62. Baker, HAL 1971. Commentary, Section III. lg: Man across the sea. Ed. C.L. Riley. Texas Univ. Press. Barrons, K.C. 1938. Natural crossing in beans at different degrees of isolation. Proc. Amer. Soc. Hort. Sci. 36:637-640. Bassiri, A. and M.W. Adams. 1978. Evaluation of common bean cultivar relationships by means of isozyme electro- phoretic patterns. Euphytica 27:707-720. Berglund-Brucher, O. and H. Brucher. 1976. The South American wild bean (Phaseolus aborigineus BurkJ as ancestor of the common bean. Econ. Bot. 30:257-272. Berlin, B., D.E. Breedlove, and PuH. Raven. 1974. Principles of Tzeltal plant classification. An introduction to the botanical ethnography of a Mayan-speaking community in highland Chiapas. Academic Press. New York. 133 134 Berlin, B., DNE. Breedlove, and POH. Raven. 1966. Folk taxonomies and biological classification. Science 154:273-275. Binns, BJL 1976. Ethnobotany of plant names in Malawi: Their origins and meanings. Soc. of Malawi Journ. 29:46-55. Bunting, A.H. 1975. Time, phenology. and the yields of crOps. Weather 30:312-325. Coupland, R. 1938. East Africa and its invaders. Oxford at Clarendon Press. Darwin, C. 1877. The effects of cross and self fertil- ization in the vegetable kingdom. D. Appleton & Co., N.Y. Edje, O.T., L.K. Mughogho, Y.P. Rao and W.A. Msuku. 1981. Bean production in Malawi. 12: Potential for field beans in Eastern Africa. Centro Internacional de Agricultura Tropical. Cali, Colombia. Free, J.B. 1970. Insect pollination of crops. Ch. 18: Papillionaceae: Phaseolus. pp.257-264. Acad. Press N.Y. Freytag, G. and T. Vakili. 1977. Personal communication to M.W. Adams. Froussios, G. 1970. Genetic diversity and agricultural potential in Phaseolus vulgaris L. Exp. Agric. 6:129-141. Gentry, H.S. 1969. The origin of the common bean, Phaseolus vulgaris L" Econ. Bot. 23:55-69. Gilmartin, A.J. 1967. Numerical taxonomy - an eclectic viewpoint. Taxon 16:8-12. Goodman, M.M. 1967. The races of maize. I: The use of Mahalanobis' generalized distance to measure morphological similarity. Fitotechnia Lat. Amer. 4:1-22. Haag, W.L., M.W. Adams and J.V. Wiersma. 1978. Differential reSponses of dry bean genotypes to N and P fertilization of a Central American soil. Agronomy J. 70:565-568. Hancock, J.F. and R.S. Bringhurst. 1978. Inter-populational differentiation and adaptation in the perennial, diploid species Fragaria vesca L. Amer. J. Bot. 65(7):795-803. Harlan, H.V. and M.L. Martini. 1938. The effect of natural selection in a mixture of barley varieties. Journ. Agric. Res. 57:189-199. Harlan, J.R. 1975a. Geographic patterns of variation in some cultivated plants. Journ.«of Hered. 66:184-191. 135 Harlan, J.R. 1975b. Crops and Man. Amer. Soc. Agron. and Crop Sci. Soc. Amer. Madison, Wis.. Harlan, J.R. 1945. Natural breeding structure in the Bromus carinatus complex as determined by population analyses. Amer. Journ. Bot. 32:142-147. Harland, S.C. 1948. Inheritance of immunity to mildew in Peruvian forms of Pisum sativum. Heredity 2:263-269. Haskell. G. 1953. Adaptation and the breeding system in groundsel. Genetica 26:468-484. Hussaini, S.H., M.M. Goodman, and D.E. Timothy. 1977. Multivariate analysis and the geographical distribution of the world collection of finger millet. Cr0p Sci. 17:257-263. Imam, A.G., and R.W. Allard. 1965. Population studies in predominantly self-pollinated species. VI: Genetic variability between and within natural populations of wild oatsfrom differing habitats in California. Genetics 51:49-62. Jain, S.K. 1979. Estimation of outcrossing rates: Some alternative procedures. Crop Sci. 19:23-26. Jain, S.K. and R.W. Allard. 1960. Population studies in predominantly self-pollinated species. 1: Evidence for heterozygote advantage in a closed population of barley. Proc. Nat. Acad. Sci. 46:1371-1377. Jain, S.K. and D.R. Marshall. 1967. Population studies in predominantly self-pollinated species. x: Variation in natural populations of Avena fatua and Avena barbata. Amer. Nat. 101(17):l9-33. Jensen, R.J. and J.F. Hancock. 1982. Multivariate relation- ships among California strawberries. Bull. Torrey Bot. Club 109:136-147. Kaplan, L. 1981. What is the origin of the common bean? Kaplan, L. 1971. Phaseolus: Diffusion and centers of origin. 13: Man across the sea. Ed. C.L. Riley. Univ. of Texas Press. Lee, J. and P.J. Kaltsikes. 1973. The application of Mahalanobis' generalized distance to measure genetic divergence in durum wheat. Euphytica 22:124-131. 136 Leleji, O.I. 1973. Apparent preference by bees for different flower colors in cowpeas. Euphytica 22:150-153. Lewis, H. and M.E. Lewis. 1955. The genus Clarkia. Univ. of Calif. Publ. Bot. 20:241-392. Livingstone, D. and C. Livingstone. 1865. Narration of an expedition to the Zambesi and its tributaries,and the discovery of the lakes Shira and Nyasa, 1858-1864. London, J. Murray Co. Mackie, W.W. and F.L. Smith. 1935. Evidence of field hybridization in beans. J. Amer. Soc. Agron. 27:903-909. Malhotra, R. and K.S. Singh. 1971. Multivariate analysis in black gram. Ind. J. Agric. Sci. 41:757-760. Merril, E.D. 1954. The botany of Cook's voyages and its unexpected significance in relation to anthropology, biogeography, and history. Chronica Botanica Co., Waltham, Mass. Miranda-Colin, S. 1967. Origen de Phaseolus vulgaris]; Agrocienca 1:99-109. Oliver, R. and G, Mathew. 1963. History of East Africa. Vol. 1. Oxford at Clarendon Press. Onim, J.F. 1981a. Influence of insect pollinators on the degree of outcrossing in pigeon pea in Kenya. Z. Pflan. 86:317-323. Onim, J.F. 1981b. Some factors which encourage outcrossing in pigeon pea. z. Pflan. 87:79-85. Narayan, R.K. and A.J. Macefield. 1976. Adaptive responses and genetic divergence in a world germplasm collection of Cicer arietinum. Theo. Appl. Gen. 47:179-187. Nevo, E., D. Zohary, A. Brown, and M. Haber. 1979. Genetic diversity and environmental association of wild barley (Hordeum spontaneum) in Israel. Evol. 33(3):815-833. Rathjen, A.J. 1965. Genetic differentiation in Phaseolus vulgaris L. Ph.D. Thesis. Cambridge Univ., England. Rick, C.M. 1950. Pollination relations of Lycopersicon esculentum in native and foreign regions. Evol. 4:110-122. Roos, E.E. 1984. Genetic shifts in mixed bean populations. I: Storage effects. Crop Sci. 24:240-244. Shickluna, J.C. 1981. Sampling soils. M.S.U. Ag. Exp. St. Bull. E-498. 137 Sillanpaa, M. 1982. Micronutrients and the nutrient status of soils: a global study. FAO Bull. 48. Rome. Singh, S.P. 1981. A key for identification of different growth habits of Phaseolus vulgaris L. Bean Improv. Coop. 24:92-95. Stebbins, G.L. 1957. Self-fertilization and population variability in the higher plants. Amer. Nat. 91:337-354. Stoetzer, H.A.I. 1984. Natural cross-pollination in bean in Ethiopia. Bean Improv. Coop. Ann. Rep. 27:99-100. Tucker, C.L. and J. Harding. 1975. Outcrossing in common bean Phaseolus vulgaris L. Amer. Soc. Hort. Sci. 100(3):283-285. Vairavan, S., E.A. Siddiq, V. Arunnchalam, and M.S. Swaminathan. 1973. A study on the nature of genetic divergence in rice from Assam and northeast Himalayas. Theor. Appl. Gen. 43:213-221. Vieira, C. 1975. Change in varietal composition of bean mixtures after successive plantings. Bean Improv. Coop. Ann. Rep. 18:85-87. Walters, D.E. and A.M. Evans. 1978. The use of multi- variate statistical techniques in the evaluation of bean characteristics. Bean Improv. Coop. Ann. Rep. 21:63-66. Weeden, N.F. 1984. Identification of white-seeded snapbean cultivars with the aid of isozyme phenotype. Bean Improv. Coop. Ann. Rep. 27:22. Williams, I.H. 1977. Behaviour of insects foraging on pigeon peas in India. Trop. Agric. 54(4):355-363. Williamson, J. 1975. Useful plants of Malawi. Univ. of Malawi Press. Zomba, Malawi. Wright, S. 1931. Evolution in Mendelian populations. Genetics 16:97-159. Zimmerer, K. 1984. Personal communication. APPENDIX A Character measurements and scores for the 375 Malawian bean lines, and the 4 Michigan cultivars included in the study. 138 :uonuuoc 0:0 :0 00000 lueu 00 Baku 000000000 06000 050 «:0 no 000000 000 nucosuuamema 060600000 0 0 0 0 0 0 0500 0000 0000 000 000 5 00 000 00 00 00 00 000 00 000 0000 5000 0000 0000 0000 0000 00 0 0 0000 0 0 0 0 0 0 0000 0000 0000 000 000 5 00 000 05 000 05 00 500 00 00 0000 0000 0000 0000 0000 0000 00 0 0 0000 0 0 0 0 0 0 050 0000 5000 000 000 0 50 000 00 000 05 00 000 00 00 ,0000 0000 5000 0000 5000 0000 00 0 0 0000 0 0 0 0 0 0 000 5000 0050 000 000 5 00 000 00 000 55 05 000 00 50 0000 0000 0000 0000 0050 0000 00 0 0 0000 0 0 0 0 0 0 0000 5000 0000 050 000 0 00 000 05 000 00 50 500 50 00 0500 0000 0000 0005 5050 0000 00 0 0 0000 0 0 0 0 00 0 5000 0000 0500 000 000 0 00 000 00 000 50 00 000 00 000 0000 0000 0000 0000 0000 0000 00 0 0 0000 0 0 0 0 00 0 0000 0000 0000 000 050 0 00 000 05 000 05 00 000 00 000 0000 5000 5000 5000 0000 0000 0 0 0 0000 0 0 0 0 00 0 5000 0000 0000 000 000 5 00 000 05 000 05 00 500 05 000 0000 0000 5000 5000 0050 5000 0 0 0 0000 0 0 0 0 0 0 0000 0005 0050 000 500 0 00 000 00 000 05 00 000 00 00 0500 5000 0000 0005 0000 0000 5 0 0 5000 0 0 0 0 0 0 0000 0000 0000 000 000 5 00 000 50 000 05 00 000 05 500 0050 0000 0000 0050 5000 5000 0 0 0 0000 0 0 0 0 0 0 000 0000 0000 050 000 5 00 050 00 000 05 00 050 05 000 0000 5000 0000 5000 0000 5000 0 0 0 0000 0 0 0 0 00 0 0000 0000 0000 000 000 0 00 000 55 000 00 00 000 00 000 0500 0000 0000 5050 5000 0000 0 0 0 0000 0 0 0 0 0 0 0000 0005 0050 050 000 5 00 050 50 000 05 00 000 05 000 0500 0000 0000 0005 0000 0000 0 0 0 0000 0 0 0 0 00 0 0000 0000 0000 000 000 0 00 000 55 000 05 00 000 05 000 5000 0000 0000 0000 0000 0000 0 0 0 0000 0 0 0 0 0 0 500 0000 0000 500 000 5 00 000 00 000 05 00 000 05 000 0000 5000 5000 0000 0000 0000 0 0 0 0000 0 0 0 0 0 5 0 0 0 0 5 0 0 0 0 0 0 0 0 5 0 0 0 0 0 5 0 0 0 0 .oz 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 nm<0 900u0osuea .000000050 00 when 0 00 00000.00 mc0uoao0h 0 00 0c0uozo0u uo ecu Ou 0000 0 00 2:63 use: 3 13 0 50 uocomuoao 00 when 0 00 . 0000 920:90029 0 00 2.26: .6 e33: 008 0.30 0 00 (004 0 0 139 0 0 0 0 o 0 0000 0000 5000 000 500 0 N0 000 05 00 05 00 000 05 000 0000 0000 0000 0005 0000 0000 00 m 0 0000 N 0 m 0 0 0 5000 0500 0500 000 000 0 00 000 00 000 05 00 500 05 000 0050 00000 5000 0005 0050 0000 00 m 0 0000 0 0 0 N 0 0 0000 0005 0050 000 000 o 50 com 05 000 50 o0 500 00 000 0550 0000 0000 0000 0000 5000 00 0 0 0000 N 0 N o o 0 0000 0500 0000 000 000 0 00 000 55 000 00 00 500 05 500 0000 0000 5000 0000 0000 0000 00 0 0 0000 0 0 0 N 0 0 0000 0000 0000 000 000 o 00 000 00 000 N5 00 000 00 000 5000 5000 comm 0005 0000 0000 00 0 0 0000 N 0 0 N 00 0 000 0000 0000 000 000 5 00 000 55 05 05 00 000 00 000 0000 5000 5000 0000 5000 0000 00 0 0 0000 0 m 0 o o 0 0000 0000 5050 000 050 5 m0 000 00 000 N5 00 000 05 000 0500 5000 0000 0000 5000 5000 00 0 0 0000 0 0 0 0 o 0 0000 0000 0000 000 000 0 00 000 00 000 50 00 500 00 000 0000 0000 0000 5050 5000 5000 50 0 0 5000 0 0 0 o o m 0000 0000 0000 000 000 0 00 000 05 000 00 00 000 00 000 0000 5000 5000 5000 0000 5000 00 0 0 0000 0 0 0 o o 0 0000 0050 0000 000 000 5 00 000 05 000 O5 00 000 00 00 0000 0000 0000 0°00 0050 5000 00 m 0 0000 0 0 0 0 m 0 0500 5000 0000 000 000 0 00 000 05 00 00 00 N00 00 00 0500 0000 0000 5005 0000 0000 00 m 0 0000 N 0 0 0 0 0 0000 0000 0000 000 000 o 50 000 cm 000 00 50 000 m0 000 0500 0°00 0000 0005 0050 5000 mm m 0 0000 0 0 N o o 5 000 0000 5500 500 000 5 00 50 No 000 05 N0 000 m5 000 0000 0000 0000 0°00 0050 5000 00 0 0 0000 N 0 0 N m 0 0000 0000 0000 000 500 o 00 050 05 000 00 05 000 55 000 0000 5000 5000 5050 0000 5000 00 0 0 0000 N n 0 N 0 0 05m 0000 0000 000 000 a 50 000 05 000 05 00 000 50 000 0500 5000 0000 0000 0000 5000 00 N 0 0000 0 0 0 N m 0 0000 0000 0000 500 000 5 00 050 05 000 05 50 N00 00 50 0000 0000 0000 5000 0000 0000 00 0 0 0000 0 0 0 c o 0 000 0000 0000 000 000 0 00 00 00 000 00 o0 000 05 000 0000 5000 5000 0000 0000 0000 00 0 0 0000 N 0 0 o 0 m 0000 0050 0000 000 ouv 5 00 500 55 000 00 o0 000 m0 000 0000 0000 0000 5000 5000 0000 50 0 0 5000 N 0 0 o o m 0000 0500 0000 000 000 5 00 com 05 05 m5 00 050 50 000 5000 0000 0000 5000 5050 0000 00 N 0 0000 0.0 0 0 N0 0 0000 0000 005m 000 can 0 00 000 No 000 00 00 000 05 000 0000 0000 0000 0005 0000 0000 00 N 0 0000 0 n 0 o o m 0000 0000 0000 000 000 0 00 000 00 500 00 on 050 o0 000 0000 0000 5000 0000 0000 0000 00 N 0 0000 N 0 0 N 00 00 0000 0055 0000 M50 000 5 m0 000 on 000 05 00 000 00 000 0050 0000 0000 0000 5000 0000 00 0 0 0000 N m 0 o o 0 050 0000 0000 000 000 5 00 000 05 000 m5 00 000 N0 000 0000 0000 5000 0000 5000 5000 00 N 0 0000 0 0 0 0 0 0 5000 0000 0000 050 500 5 00 000 N5 mm 00 50 500 00 000 0000 0050 CONN 5005 5000 0050 00 N 0 0000 0 0 0 0 0 0 0000 0000 0000 000 000 5 00 000 05 000 05 00 000 00 00 0000 0000 5000 0000 0000 5000 00 0 0 0000 0 0 0 c o 0 0000 0000 0000 000 050 5 00 000 cm 50 00 00 000 00 000 0000 5000 0000 0050 0000 5000 mm 0 0 0000 0 m 0 c o 0 005 0000 0000 000 000 0 00 000 05 000 v5 00 000 05 500 0500 5000 0000 0050 0000 5000 00 0 0 0000 N 0 0 N 00 0 0000 0000 0000 000 000 5 00 000 05 000 00 50 000 00 000 0050 5000 0000 0005 5000 0000 50 0 0 5000 0 0 0 o c 0 000 5050 0000 000 000 0 00 050 05 000 00 00 000 00 00 0500 5000 0000 0050 5000 0000 00 N 0 0000 0 0 0 o o 0 0000 5050 000m 050 000 5 00 000 05 00 00 m0 050 00 000 0500 0000 0000 0050 5000 0000 mm 0 0 0000 0 m 0 o o 0 0000 0000 5500 050 000 0 00 000 N5 000 05 m0 000 00 000 5000 0000 0000 0000 0000 0000 00 N 0 0000 N 0 0 c o m 0000 5000 0000 000 000 5 00 000 05 00 00 cm 000 05 000 0000 505m 0000 5000 5000 0000 00 N 0 0000 0 0 0 0 00 0 0000 0000 5000 000 500 5 00 com 00 000 05 00 000 05 000 0000 5000 0000 0000 0000 0000 Nm 0 0 0000 0 0 0 o o 0 055 0000 0000 500 000 0 00 000 00 000 00 50 000 05 000 0000 5000 0000 5000 5000 0000 00 N 0 0000 0 0 0 o 0 00 0000 0005 0000 00m 000 5 00 000 05 mm 05 00 000 50 00 0000 0000 5000 0005 0000 5000 cm 0 0 0000 0 0 0.0 o 0 0500 0000 5050 000 000 0 00 000 00 000 00 om 050 00 000 0500 0000 0000 0000 0000 5000 00 0 0 0000 0 0 0 o o 0 000 0000 0000 000 000 0 00 000 05 000 05 00 000 N0 000 0000 0000 0000 0050 0000 0000 00 N 0 0000 0 m 0 o c m 0000 0000 0000 000 000 0 00 050 00 000 00 00 000 00 00 0000 0000 0000 0000 0000 0000 50 N 0 5000 N 0 0 N 00 0 000 0500 0000 000 500 0 00 com 05 mm 05 00 M00 00 00 0050 0000 comm 5000 5000 0000 00 N 0 0000 N m 0 o o N 0000 0000 0000 000 000 0 00 000 00 000 00 50 000 00 000 5500 0000 5000 0005 0050 0000 00 0 0 0000 N 0 0 o 0 0 0000 0000 0050 000 000 5 00 000 05 000 05 00 000 00 000 0000 5000 0000 0000 0000 0000 00 0 0 0000 0 0 0 o o 0 0000 0050 0000 000 000 0 00 050 00 000 50 00 000 00 00 0000 5000 0000 5005 5050 0000 00 0 0 0000 0 0 0 o o 0 0000 0005 0000 05m 550 0 00 000 00 00 50 m0 000 m0 000 0000 0000 0000 0005 0°00 0000 00 0 0 0000 0 0 0 0 00 0 0000 0500 5000 000 000 5 00 000 00 000 on 00 000 05 000 0500 0050 0000 0005 5000 0000 00 0 0 0000 0 0 0 o o N 0000 0050 0000 000 000 0 50 000 00 50 00 05 000 N5 500 0000 0000 0000 5000 5000 5000 00 0 0 0000 0 0 N 0 00 0 0000 0000 0000 000 000 0 00 000 N5 00 50 N0 050 50 500 5000 0000 5000 0050 5050 5000 00 0 0 0000 0 0 0 0 c 0 0000 5000 5500 000 000 0 00 000 50 000 05 00 000 00 000 0000 5000 0000 0005 0000 5000 m0 0 0 0000 0 0 0 o o 0 0000 0005 0050 000 000 5 00 050 cm 00 05 mm 000 00 00 0000 0000 0000 0005 0050 5000 50 0 0 5000 0 0 0 o o 0 5000 0000 0000 050 000 5 00 000 05 000 N5 00 000 00 00 0000 0000 0000 5005 5000 5000 00 0 0 0000 0 0 o 0 0 5 0 0 0 o 5 o 0 0 0 m 0 0 0 5 0 0 0 0 0 5 0 0 m o .02 0 0 0 0 0 0 m 0 0 0 0 0 0 0 0 0 0 0 m o o 0 0 0 0 N 0 0 v o mmdu 00200 «om mo 05202000000: 005000000 ”m0005 Humnmomum «000 mama 140 NNNNNNNNNNNNHNNNNNNNNNNNHHHNNHHHHNNNNNNNHNNNNNNN ammmmmmmmmmamaammvaaaaammmmaavmmmaammmamvmmammmm Nmmmmmmmmmvavamvvomaaaavmvvmamvmvmammvamovavvmm ONNNNNNNmmOOOOMOOMMOOOOONOOMHHONOMOOO‘DOGNOO‘DOWOO N o nmvN mmvo omov mNm moN n ca mNN mn Nn mo cm Nma vo om monm ccnm nooa comn mmom cha maa m N mama N o cmam ovoo vmvo oov onN o ca mma on mm vo ov mva mo om vmcv mmom mmna mmmn ooom mmNa Naa m N Nama N o nnoa vomm NmmN Nov Nom o ma caN co mNa No no oma Nn Naa oavm mmam oooN cono ocnv mmNa aaa m N aama o N mNca oomm omoN omm mnm o ma ch No aaa mo vm mma ao mm mnom mmmm mmnN mmmn ocov noNa oaa m N oama N o aaoN amNn mmmm mnv mmN o aa oNN on mma ao am oma no ooa man oomm mmoa mmmn comm ooNa moa m N mcma N o ona ovmo momN aam mmv n ma coa ao oNa an oo mva mo oaa ommm noom ocna chn comm noNa oca m N ocma c N nmNa mvoo amNm mom mmm n va oNN mn noa om mo oma ao vm vonm mmom mmvN ccan noov coma noa m N ncma va o mooa vomv anvN va acv o na mo oo vca oo mo vva nn aaa omam mmNm nova ooao mmov noNa oca m N ooma N o NooN momo nmmm cov mmN o ca moN mn oo mo Nv vva ao oo aNmm comm mmoa nomn mmmm ocNa moa m N mcma o n onmN maon ovnm mam mmN n ca mma mn maa oo mv mva an Nca Nonv non mmoa nomn mmnm noaa vca m N vcma N o amNN onnn van nNm mvm n Na oNN on mca mo mm mva oo aca oovv mmom coma mmNn mmmm ooNa mca m N moma N o vNom omon mNNo vov NmN n ca caN mn moa on Nm ama Nn moa omnv mmom mmna mmmn ooom mmaa Nca m N Noma N o ovNN mNom ommv mmv omN n m oNN mo No mn ov oma No vo amNm comm ccoa mmmn mmmm mmNa aca m N acma c n ooma oNam Nmov amv mmN o m cma on Nm co vm ova om vo movm mmnm nooa noNn covm mmaa coa v N coma N o maNa maom mmcm Nov vnv o ma ooN on no vn vm mva no nm oacv nomm nomN comn mmmv coma mm v N mmNa c m mmn nmov mmoa mmv omm m oa co no vNa mo ov ova vo mm oan mmno coma ooao ccov oova om v N omNa m o vaaa mnmv oomN mmv vmm o oa no mo mma on om oma mo oca nmmv oomo coma ooao ocov ooNa nm v N nmNa c m mao ooam mcva ovv mom o na mo mo mva co Nm mva mo moa ovvv nooo nova ooao mmov mmNa om v N omNa m o vNaa oanm vvoN mmm omm n Na nm mn mo on mo ooa oo maa Namv oomo nooa ooao mmvv mmNa mm v N mmNa va o mmNN onco voam mvv com o va omN on Noa mo nm mva an aca mmmv mmNoa mmoN nomn mmnv noNa vm v N vaa N o mmm monv amoN mom onv m va omN on mm no am mva No cm ooom comm coma nomo noom mmma mm v: N mmNa o m ovn Nonm aona omv ncm o Na co vm mNa om om mma oo Nm onom mmoo mmma ooao nonv mmma Nm v N NmNa c m va Ncov oaoN omv mam o na no vm nm nm mm Nva on aaa movm mmoo mmva ooao noov mmma am v N amNa m o Nnm manm mmma omv nom o ma nm on vaa cm mv nma nn voa moov oomo ooNa ooao ccmv noma cm v N omNa c m onma onnm nmvm amv nom n va coa mn om ao ov Nma on oca nmov mmvm nona oomo mmam noNa mo v N moNa o n mvoN Nomo aaam oam moN o ca moa oo cm mm ov nva co no NmNm noom nooa nomn oomm coaa oo v N ooNa o n oooN mmmm moov mam onN n aa maN an ao oo nv Nva om co moom mmnm coma noan noom coaa no v N noNa o n omvN momo oomm nov voN o ca moa mo mn on mv oma mo no ommm nomm noma oomo mmmm coaa oo v N ooNa c n mmvN nnvm vamv amv oNN n ca coN vn Nm oo vv ava vo cm nvmm noom mmma mmNn oomm coaa mo v N moNa N o moma ovoo mnnm onm cmv m ma moN ao ao ao mm mva mo om ooom noom ooNN mmon mmov mmoa vo v N voNa N o ovma mamo nnvm nom oam o ma mNN vn vo mo mv oma ao vo naNm noom noaN coNn mmcm nooa mo v N moNa o m oova Non monm omm vmm n ma ona on moa mn nv moa vo moa ova oomo nona mmoo noom ooNa No v N NoNa c m Nona oovo mnom nov mom n ma ooN vn mm vn mm ava mo nm vmam mmmm mmmN ocon novv mmaa ao v N aoNa N o Nova omoo cmcv Nom mov m va mNN on om vo nm ova oo mm vmav mmaca mmmN nomn mmov coma co v N ooNa o n noma mnom moav onv va o oa cna mn mm mo cm Nma mo om monm con nooa mmNn nomm noaa mn v N mnNa o m mvma mmmm nva onv mvm n ma nv nm mca No om nma nn oca moam mmmm mmva ooao noov cha on v N onNa o n mamN oNoo aon mmm oNN m ca coa an nm mm ov vma ao mm oonm mmvm mmna mmNn comm coaa nn v N nnNa o m Nvoa nvmm aNmm ch Nom o va mma on nm No cm nva mo om momm noom mmoa mmon con noNa on v N onNa m o mNm oNNv omma mnv omv n ma ooa oo nma mo oo aoa nn vNa oovo ooom mmaN mmno ooov ooNa mn m a mnNa m m amaa omnv vva mnv oav o ma coa on mNa oo oo oma on nm ommv mmam mmNN mmoo ocov oova vn m a vnNa m o aoca nvmm omvN mmv nov o na cna n naa an Nm mma on oaa mvam noom nomN nooo ccmv noNa mn m a mnNa m o mona vomo mmom oov Nnv m oa ch mn oaa an mm Noa oo oca anv mmvm nomN nomo coov mmma Nn m a NnNa m o omna mnmo oomm nov mnv n va ch oo oca Nn mm mma vn maa coam mmnm mmmN coon noov coma an m a anNa m o Noma cmNo vmcm nvv vmv n oa moa co moa mn mm coa no oca ooov ocvm coaN nooo nonv oova on m a onNa m o omNa canm mmnN mvv omv n ma mma co oNa oo mm ona vo oca oNnv non ooNN nooo noov noNa mo m a moNa m o Noaa vnam nonN NNv omv m na coa ao vma on am mna No noa novv mmom ooNN ocoo coov coma oo m a ooNa m o mnaa oamm oooN mmv oov o oa coa vo mma on mm ooa cn naa ooom coam ooNN oomo conv noma no m a noNa c ma momN omNn Nmav oao NoN o aa mvN mo oca an mo oNa vn mm mmav mmom noaN mmNn noom coaa oo m a ooNa m o maNa monm mvoN va oov o ma cma on noa mn mm mna om nm nnom cccm mmvN nomo mmmv noma mo m a moNa o n m o a c n c m a m m v v a n o N v m o n o N m o .02 a a m m v v m N a N a N a N m o c N m m N N N v v o mmov mwzHa 40m m0 mfizmimambmé mm80§<20 "wauae Humcmomum mafia mama 141 I NNNNNNNNNNNNHHHHHHHNHNHHHHHo-Ou-lHHHHHHHHHHHHHHHHHHHH HN I MLDMMHHMIDMU‘HMMQ’MMMMMMMMMMMMMMMMMQMMMMMMMMMMHMMMMMM HO l vmvvaavmvmavvmvmvvvvmvvvvvvvmvvmmmvovvvmvamvvvvv POI-i o va o oco nvmm ncma mov nmm n oa no mo mma No oo mma ao mNa omao mmom cha ooao ccmv oova Noa n m Noma o va o man vovv booa onm mNo m cN co mo moa an nm vva vo aNa cho mmNoa noaa ooao mmmv ccva aoa n m aoma o va o moo ommm naoa onm cNm o oa co mo nNa co vm ova nn vaa nmom coaca mmma oomo nomv noma coa n m coma o va o Nao cmom mcoa nom Nmm o ma no mo aNa mn co ova co naa omvm mmcca ooNa ooao ccmv nova mma n m mmma o va o aon oonv vNcN onm vNo n oa cm oo oNa no vo ova co naa nvom mmcca mmNa ooao noov oova oma n m omma N m o mmo mnmm omNN onm mov o oa no on Nma no No Nma mn oNa avnm nomo noma ooao mmnv noNa nma n m nmma c c va mav onm nNaa vom aaN o ma no nn co om mm ama an noa omvv noNaa com ooao chm ccva oma n m omma N m o omca noov mmNN amv an n ma nm co maa vo mm ova an mca vmnv chm ooNa ooao ccmv mmNa mma n m mmma o va o cvn mva aoma onm onm m ma nm no vma oo om voa mn maa ammm ooom ocNa ooao ccmv mmma vma n m vmma o va o avn cvcv oona oav vvm o na no vo Noa mo mo Nma on vaa oaom comm ooNa ooao ccmv noNa mma n m mmma o va o moo oaav oavN mom Nov o oa mo vo mma ao om voa mn oaa mmov nomm oova oovo ooom coma Nma n m Nmma o va ma mmm aavv anmN mom ovv o na nm oo moa vo mv Noa on maa onom mmaca noNa ccao mmov coma ama n m amma N m o ano mmom mvoa nov an o oa nm cm maa No cm aoa mn naa NNmm nomo oova ooao oonv noNa cma o N omma N Na o nmma omvm omcm omv aav o na no no vva co Nm oma nn aNa omco noom oova ooao oonv mmma mva o N mvma o o m ooo vaom nmva mov avm o na cm am nNa mo cm mva an mca mvvv nono mmva ooao noov cha ova o N ovma N m o oNo maam mvma va com n va mm vo am ao cm aNa mo oca aamm mmoo mmva oomo noov noma nva o N nvma N m o aoca oNnv ommN va nmv o oa no mo om oo mm ama ao NNa aano oovm noaa ooao mmmv coma ova o N ovma N va o Nmaa aamv novN anv aom o va mo mo vNa mo om mva mn mca mnov ooom noma ooao mmnv noNa mva o N mvma c o m vvoa cva ammN nov ocv o oa on om vma om Nm vma vn maa navm nooo noma ooao mmnv mmNa vva o N vvma c c m aan nvom ooma mmv mvm m oa mo Nm mna oo ov nma on om amov nomo mmma ooao nonv mmNa mva o N mvma N m o vmo vav oaaN NNv vmv o oa nm am nma oo nv moa mo Naa avov mmNm noma ooao mmnv noNa Nva o: N Nvma c c m moo cmcv omoa an vcm o oa co co voa vo om ova oo aca vav nooo mmva ooao noov noaa ava o N avma o c m mao cvnm oNcN mav vmv n oa nm vo oo an ov mva an oca vmov nooo mmma oomo nonv mmva ova o N cvma c c m mmoa Namv mmnN amm ovv n va nm No caa mn mv ava mn mca vav con mmva ooao noov coma mma o N mmma c o m vmm mmmv mmma Nov omm n oa nm no oaa no mm vva oo om Nomv noam oova ooao oonv noNa oma o N omma c o m mmo mvmm moNN nov mmm n va nm vo vNa mo mo vma cn noa moov nomo coma ccao coov noNa nma o N nmma N m o ooo nonm vNaN aav vmv o oa co mo Naa no am nma mn maa ommv nomo mmoa mmmo ccnv mmNa oma o N omma c o m avoa aomv chN mmv vnv o oa moN ao ama cn ov vna mm Noa oamm mmNm och mmoo mmvv ooNa mma o N mmma N m o moaa cncm vvoN cvv nmv n ma mo mo Nm om Nv nma oo mca ommm mmvm coma ooao coov nova vma o N vmma o c m mmm vnnN NNma ocv mmv o ma moa mo vNa no ao aoa ao mca onv noom noaN mmno nomv noNa mma o N mmma N m o cvo nmmm cmma mvv oNv n oa nm ao coa vo am oma an aaa omNm ocam oova ooao oonv oova Nma o N Nmma c c m ovn aoom oNna mmv amv o oa mm vo moa mn cm mNa mn nm aNcm oomo coma ooao coov noaa ama o N amma o o m mvn mNmm maoa Nov on o oa mo am vma mm mm oma vn Nca Ncnv mmmo mmma ooao nomv cha cma o N omma N m o amn mmnN nmma onv mom n ma co mo mca nm ov oma no coa nNov mmoo mmma ccao nonv noma mNa o N mNma c c m mno moom mnna omv oav o ma mm on vva oo ov oma an oca moov oomo mmma ooao nomv noaa oNa o N ona N va.o Nnm voom NnNN an omm n ma co mo cma vn mo mma vn naa oomm cccm noma ccao mmmv ooNa nNa o N nNma c c m oan mmnm omoa va mNm o na mm cm ana no Nm ama oo om mcav ccoo mmma ooao nonv noNa oNa o N ona o N o oooa Ncoo ooam mmm Nmm o Na cmN no Nm oo am moa om am aomm noom mmoa nomn mmmm noNa mNa m N mNma o c n amvN aNNo ommv amm mmN o ca coa mn am oo cm mva on oca nNmm nomm nona mmmn nomm ooNa vNa m N vaa m N o ommN moan coco Nov mvN n oa ch oo cm ao ov mva oo mm oocv nomm mmoa noan mmmm ooNa mNa m N mNma o N o nvm chm omoN mmv Nmm o oa moa mo oma ao ov moa No maa mmom mmmm noma nooo ccnv coma NNa m N NNma m N o nomm mmvo amvo NNm ovN o ca oaN oo an on nv moa vm cm onam nomm cooa novn noom mmma aNa m N aNma o N o moNa mmmm mmoN mmv va o va ch nn vm oo am vma mo oca nmmv nonm mmNN ooNn nomv noNa oNa m N ona c o n mvoN oooo omcm on omN o aa oaN on vn co ov ova on vca cmvv mmom mmoa mmvn ooom mmNa maa m N mama c c n nmmm mon amoo anm omN n m maN mn .mv mo mm mNa no no cnvm mmcca mmna mmmn ooom noaa oaa m N oama o N o mmma onan oNNm mmv mvv n ma mma mn nn vn ov vma oo Noa aocm mmom nocN chn mmam coma naa m N nama o N o mnma omcn nmcm omv aam n ma cmN oo ana mo Nm nna ao noa ooNv comm ocmN nooo nomv mmaa oaa m N oama m N o vavN moao maov NNm omN n aa ch Nn noa No am ama No mm cmom mmom noma noan ooom coaa maa m N mama o va m moma nmon momN nNm mom o oa ooN mn aaa on nv ama on mca Nomm noom coma oomo ooom noma vaa m N vama o o n m o a c n c m a m m v v a n o N v m o n o N m o .02 a a a m m v v m N a N a N a N m o c N m m N N N v v c mmco mmzaa com mo mazmzmmamamz mmsu¢mczu "wauaa Humomoaum mafia mama 142 HN I HHHHHHHHHHNHHNNNNNNNNNNNNNNNHNNNNNNNHNNHNNHHNNNNN r-lo-I I MMMMMMMMMMHMMHHHHHVHHHHHHHHHMHHHHHHHMHHMHHMMHHHMH m c o aa Nvma moov aoam m N va o mooa NnNm ova m o o aa Nnaa mmav amoN m c c aa ovma aomm oanm m o o aa moNa ovov mamN m N m o ovo omvv omma m N m o omm vvcv mmmN m c c aa coma avmm moom m o c aa ooNa moam omoN m N m o mcoa oamm mmvN m c c aa nmNa coov mNmN a c c n mvna anm omvm v o o m nmo avnN ccoa a c o n mnNN Ncan ova a o c n onmm ommo vnmm a c o n mmnN vomm nnom a o o n mmma cNmm amav a c o n cva omvv omvv a c c n von mmmm ooom a o o n mmmm nmon ammm v c c m vno mNom nooa a c c n ooma vmmm naov a c o n nnmm ano mmvm a c c n Nmom mmon mmom a c o n nmoN omvm ommv a c c n mnmN aamo Nmmv a c c n mvoa aomm aNNm a o c n nnma omao man a c o n noam oomo voom a c o n oNam vam maov m c c m Nooa NNmm oomN a o o n vam comm mon a c c n aoam mmom cnam a o c n movm ooco vmom a o o n monN ooom vav a c c n oomN mcoo mmov v o va o nmn coov vvma v o va ma Nnn vaom onoa N c c Na Nama vam mNom v o va o ono omov nooa m N m o omm amnm omoN m N m o mvaa monv mavN v o va o nnn mmnm mmoN v o va o cmn mNNv onna v o va o cam vcmm noaN v o va o Nmn ova moma m N m o ovm maom och v o va o mmoa vav vnmN v o va o voo ooom vaoa o o o n m o a a a a a m m v an omv va oav mmv va ocv mom vvv mov va mom mom mmm amo amv mmv mNm mvo cvo mov mov oao Noo oao vvm Ncm vov cvm vmo nom nNo oao onm nvo mmm mov aav mNo mnm omv onv com vov NNv com amv nvv com o v Nom mmv nmm aom Nom mNm mmv mmm mmm omm Nmm oma aav vam ooN oaN onN ama aoa mNN nmv NcN oma voN mma NnN avm oam oaN noa on Noa mom n m ma mca na oNN ma cna coa ma mma na no oa mo na mNa Na ova Na mma oNa Na coa m caN ca mma mma ca co o coa ch oa co oa co oa mo oN mm moooooonooonmomoooonnnoooonoooomnnonnoononooooooo 0‘ o m a N a N mo m a Nva mm maa m N mn no Nma oo mv nma vn mo moa on No vma on ao ooa mo om moa mo mm ova on mv mva vn vo mva No oo ona nn nm aoa ao vm oNa oo co ova co am oma ao nm ova mo mm vna Nm mv ava mo cm ova om am ova ao om nva no ao mna mo nm vva oo mm mma om om mma mm nv ova om om oma om mm Nva mm nv vma nm vm Nva mm Nm mma oo ao Noa om mm nma Nm Nm Nva ao om nva mo vm ova mo vv ooa vo om mma cn vm ooa on ao mNa oo oo oma mo mm mva mo co ova nn om moa vm om oma vn ao ooa oo om mna mn nm mma om nv ooa vo mm aoa v v a a N m Noa mca mca oca oaa mca mm mo oo nm maa no noa co nn Naa mm mn mm mn vm mn Nm Nn mn on oNa vo mca on oo mn mNa vn caa on maa mNa maa oaa vn nNa moa vo aoa No cca n o c o noov oamm mmmv mnmm Nva mnNm nnov vnmm aan amnv naov moom nvom mmom oooN nmmm caNm oooN oNoN oomm mmam NamN aomN mmmN onvN mmmN oovm aomm mmoN nnnN anv onvN anN maoN omNm mcmN onm vmov chv ooom moom nNnm oovo vmmm moam cmoo mvov mnmv Nomv N N mmzHa 40m ho mazmtmmnm<fl2 mmau