Pigeon pea [Cajanus cajan (L.) Millspaugh] is a very important grain legume crop for food, cash and firewood in Malawi. However, its production is affected by Fusarium wilt (Fusarium udum Butler), which causes up to 100% yield loss. The deployment of resistant varieties would be an economical way to manage the disease, and for this, more information is needed on farmers' preferences for local landraces, how farmers and consumers can be involved in developing new varieties resistant to wilt disease, and the genetics of inheritance of resistance. This information would be used to devise a breeding strategy. A participatory rural appraisal was used in the southern region of Malawi to identify pigeonpea production and marketing constraints. Results showed that Fusarium wilt was the most prevalent and destructive disease of pigeonpea in the area. Other constraints included pests, flower abortion, low yields, and low soil fertility. Local landraces accounted for 84% of the pigeonpea production in Malawi. Local landraces were preferred due to their fast cooking time, taste, and the high prices they earn the farmer. Participatory variety selection was used to identify landraces with desirable traits that could be used in the breeding programme. Farmers and buyers selected ten local landraces which were used in the genetic improvement programme. Pigeonpea local landraces and international Crops Research Institute for the Semi-Arid Tropics (ICRISAT) genotypes were evaluated for wilt resistance, yield, and secondary traits at three sites over three seasons. Most of the landraces were susceptible to wilt and late maturing. However, AP10, a local landrace, was high yielding and resistant to wilt and outperformed ICRISAT varieties. This local landrace showed promise for use as a source material for Fusarium wilt resistance in other locally adapted farmer-preferred varieties lacking resistance. The local landraces needed genetic improvement in wilt resistance, yield, early maturity, number of branches and seeds pod. Laboratory and screenhouse studies were performed to develop a new Fusarium wilt screening technique. Grains of finger millet, sorghum, and wheat were tested as media for multiplying F. udum isolates. Pathogenicity tests were done on Bunda College and Bvumbwe Research Station isolates. The Bunda isolate was then used in an infested-seed inoculation technique against eight differential cultivars. The results showed that finger millet, sorghum and wheat were equally effective for rapid multiplication of F. udum isolates. Wheat grain showed the best results for pathogen multiplication and inoculation, due to the large seed size for easy handling. The inoculation process involved placing infested wheat grain on bruised pigeonpea roots and transplanting into soil in pots. The infested seed inoculation technique, which is the first of its kind for pigeonpea, was effective in screening pigeonpea for wilt resistance. The selected landraces were crossed with wilt resistant testers in a 12 lines x 4 testers mating scheme, and 48 F1 crosses were generated. These F1 crosses were evaluated for wilt resistance, yield, and secondary traits. The variations among F1 crosses for wilt and secondary traits were due to additive gene action in both parents and the dominance effects arising from the interactions of parents. Parental lines, with good combining ability effects for wilt resistance (AP2, AP3, and AP4), days to 50% flowering, seed pod, plant height, stem diameter, and number of primary and secondary branches were identified, while ICEAP00554 (tester) was a good general combiner for wilt resistance and days to 50% flowering. These lines would be useful in breeding for Fusarium wilt resistance in farmer-preferred pigeonpea genotypes in Malawi or similar environments. Specific F1 crosses were identified with significant SCAs for wilt resistance, days to 50% flowering, and secondary branches. The significance of GCA and SCA effects, which indicated importance of both additive and non-additive gene effects, respectively, suggested that both selection and hybridisation would be useful to improve the resistance in farmer-preferred varieties. Segregation analyses were conducted on F2 populations to determine the resistance to susceptibility phenotypic ratios. The Chi-square analyses showed that resistance to wilt was dominant over susceptibility in most F2 populations. The segregation ratios of 3:1, 13:3, 15:1, and 9:7 (R:S) indicated that either one dominant gene, or two inhibitory genes, or two independent dominant genes, or two complementary genes, respectively, were conferring wilt resistance in these crosses. Involvement of only a few genes governing wilt resistance suggested few complications, if any, in breeding for this trait in these locally adapted pigeonpeas. The Pedigree breeding method would be recommended for incorporating these traits. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2007.

The common bean (Phaseolus vulgaris L.) is an important crop grown widely in Uganda. It is also an important source of income for smallholder farmers particularly women. Despite its importance, production in the cool highland regions is constrained by anthracnose disease which causes losses in both the quantity and the quality of beans produced. The principal aim of this research was to elucidate on the status of dry bean anthracnose and the genetics governing its resistance. A participatory rural appraisal study was conducted to explore farmers' knowledge, experience, problems and cultivar preferences in association with managing dry bean anthracnose disease. This study revealed that anthracnose is an important constraint to production which is not controlled in any way. Although farmers have varying cultivar preferences, they use mostly home saved seed and only 1% could access improved seed. The study suggested the need for practical approaches in the provision of quality anthracnose resistant seed in consideration of farmers' preferences and the dynamics of their rural livelihoods. A study was conducted to determine the variability of the anthracnose (Colletotrichum lindemuthianum) pathogen in some of the major bean growing regions of Uganda. Use was made of a set of 12 internationally accepted anthracnose differential cultivars to identify the physiological races present. The results obtained indicated the presence of eight races with one race (767) being dominant and most aggressive. Differential cultivars AB 136 and G2333 were resistant to all the eight races, and can be utilised as potential sources of resistant genes. A germplasm collection of mostly Ugandan accessions was screened for anthracnose resistance. Using the area under disease progression curve as the tool for assessing disease severity, eleven accessions were identified that posses good levels of anthracnose resistance. The yield loss attributed to the anthracnose disease was determined on three susceptible Ugandan market-class dry bean cultivars and two resistant cultivars. The results showed that the yield of susceptible cultivars was reduced by about 40% and an almost equivalent yield was lost due to poor quality seed. In comparison, the yield lost by the resistant cultivars was not significant. The study suggested the use of resistant cultivars as the best solution in combating anthracnose resistance. Three susceptible Ugandan market class dry bean cultivars and six resistant cultivars were used for the study of the inheritance of resistance to the anthracnose pathotype 767 in a complete 9x9 diallel design. The results clearly indicated that the resistance was predominately conditioned by additive gene action. It was also established that epistatic gene action was important. More than one pair of genes displaying partial dominance were responsible for determining resistance and the maternal effect did not have an influence on resistance. Additionally, the result showed that some of susceptible cultivars combined very well with the resistant cultivars and that anthracnose resistance heritability estimates in both the narrow and broad sense were high. These results suggested that the use of simple pedigree breeding procedures such as backcross selection could be useful in improving anthracnose resistance levels in the Ugandan market class varieties. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2006.

In Burkina Faso, the existence of different races of Striga gesnerioides (Willd.) Vatke, with apparent variable aggressiveness on cowpea (Vigna unguiculata (L.) Walp) renders the breeding task very complex. Therefore, a number of studies was carried out from 2006 to 2009 in field, pot and ‘’in-vitro’’ to identify new sources of resistance to three prevailing Striga races, SR 1, SR 5 and a newly occurring Striga race named SR Kp and to understand the genetic pattern of the underlying resistance of cowpea germplasm to Striga races found in Burkina Faso. To achieve these objectives, the following investigations were initiated: (i) a participatory rural appraisal (PRA), a participatory variety selection (PVS) and grain quality survey were implemented to identify cowpea breeding priorities for Burkina Faso Striga hot-spots; (ii) the identification of sources of resistance in Burkina Faso germplasm, using three prevailing Striga races of S. gesnerioides as sources of inoculum; (iii) the identification of the mechanisms of resistance underlying the resistance to Striga in such genotypes; (iv) a study of combining abilities of selected parents through a diallel cross; (v) a study of the segregation patterns in crosses involving resistant and susceptible sources and a study of the allelic relationships between different resistance sources. The participatory studies conducted in 2007 and 2008 over three districts in Striga hotspots; there was no effective control method against Striga at farmers’ level. These investigations highlighted the importance of cowpea across all sites. Rain decline over time, low input use coupled with a poor extension system were the major constraints mentioned by farmers. Differential reactions of genotype KVx61-1 for Striga resistance suggested that different Striga races were prevailing in different areas. Farmers’ preferred traits in cowpea genotypes were oriented towards grain quality such as big sized grain, white seed colour and rough texture of cowpea grain, except in Northern-Burkina Faso, where farmers preferred brown-coloured grain for food. Cowpea was also seen as an income generating crop. An evaluation of 108 genotypes was done in 2007 in the field (rainy season) and in pots (off-season) for Striga resistance assessments. The screening trials enabled the identification of sources of resistance to S. gesnerioides. Genotypes KVx771-10, IT93K- 693-2, KVx775-33-2, Melakh and IT81D-994 are potential sources of resistance to all three Striga races with acceptable yield. Landraces were susceptible and late-maturing whilst most wild species were resistant but with unwanted shattering traits. A combining ability study for Striga resistance parameters conducted in pots and a resistance mechanism study conducted ‘’in-vitro’’ were performed using F1 populations from a 10 x 10 diallel cross. The general combining ability (GCA) effects were significant for the resistance parameters Striga emergence date (DSE), Striga height above soil (SH), cowpea grain weight (CGW), hundred grain weight (HGW) for all Striga races involved and Striga vigour (SVIG) for SR 5 and SR Kp. The pot-screening showed that, regardless of the SR used as inoculum, the additive genes were important in conferring Striga resistance for parameters DSE, SH, CGW and HGW. The selection of parents could therefore result in breeding advance. Complete dominance, partial, over-dominance and non-allelic interactions (epistasis or failure of some assumptions) were present for some parameters. The ‘’in-vitro’’ screening showed that additive genes were important, with high narrow sense heritability values for the resistance mechanisms Striga seed germination frequency (GR) for SR 1 and SR Kp, the frequency of Striga radicle necrosis before the penetration in cowpea rootlet (NBP) for SR 5, the frequency of Striga radicle necrosis after the penetration in cowpea rootlet (NAP) for SR 1 and SR Kp and the susceptibility ‘’in-vitro’’ (SIV) for SR 5 and SR Kp. The selection of parents can be useful in accumulating the genes for Striga resistance mechanisms in progenies. The F2 populations derived from crosses between Striga-resistant x susceptible genotypes were evaluated in Striga infested benches in 2008 and 2009. The segregation patterns suggest that single dominant genes govern Striga resistance. The test for allelism showed that two non-allelic genes were responsible for the resistance to S. gesnerioides in cowpea. A new Striga resistance gene seems to be involved in genotype KVx771-10 resistance to S. gesnerioides, which confers resistance to all studied Striga races. Gene 994-Rsg in genotype IT81D-994 which confers Striga resistance to SR 1 and gene Rsg 3 also conferring Striga resistance to SR 1 segregated differently for the resistance to SR 5 suggesting that they were different but both confer resistance to SR 5.

Maize ear rots are among the most important impediments to increased maize production in Africa. Besides yield loss, they produce mycotoxins in their host whose contamination has been linked to several human and animal mycoses. The main objectives of the studies reported on in this thesis were (i) to investigate farmer perceptions of maize ear rot disease and prospects for breeding for host plant resistance in Zambia; and (ii) to establish the levels of incidence and extent of maize ear rot infection as well as the level of mycotoxins in the maize crops of smallholder farms in central and southern Zambia; (iii) to appraise the field inoculation techniques and assess them for their suitability for the Zambian environmental conditions, (iv) to determine the combining ability of Zambian maize populations for resistance to ear rot and investigate the genetic basis of this resistance; and (v) to investigate both direct and indirect responses to full-sib selection for ear rot resistance in Zambian maize populations. A participatory rural appraisal (PRA) was conducted in four communities, involving a total of 90 farmers. Participatory methods were used, such as focused group discussions, group interviews, participant scoring and ranking. Farmers ranked and scored the various constraints affecting their maize production in general and the maize ear rots in particular. Ear rots were ranked as the third most important biotic stress and it was evident that although farmers were aware of the disease, they were not aware of mycotoxins. This was reflected in the way they disposed of rotten maize: either by feeding livestock or eating it in periods of hunger. The survey of ear rots and mycotoxins was carried out in the Southern and Central Provinces of Zambia. A total of 114 farms were covered in the survey: maize samples were collected and both ear rot fungi and mycotoxins were isolated. Fusarium and Stenocarpella were the most frequently isolated fungi from smallholder farms. The levels of fumonisins on these farms ranged from 0.05 to 192 ppm, while those of aflatoxins were between 1.5 and 10.6 ppb. In 50% of the farmsteads surveyed, the mycotoxins, i.e. fumonisins and aflatoxins, exceeded the recommended FAO/WHO 1limits of 2 ppm and 2 ppb, respectively. Five field inoculation techniques namely, colonised toothpick, leaf whorl placement, ear top placement, spore suspension spray, and silk channel injection, were evaluated over three seasons in a series of experiments. It was found that the leaf whorl placement of inoculums, followed by colonized toothpick method, gave a constant ranking of genotypes across locations and years compared to the other three methods. In addition, the use of a mixture of ear rots as inoculum was as effective as its principal single species constituents. In the population diallel analysis, five broad-based maize populations were crossed in a diallel and evaluated under artificial ear rot inoculation using an inoculum mixture of three ear rot fungi, Aspergillus flavus, Fusarium verticilloides and Stenocarpella maydis at four locations in Zambia. The purpose was to estimate general (GCA) and specific combining ability (SCA) and investigate genotype x environment interaction. GCA effects were found not to be significant for disease severity but were significant for grain yield across environments. Populations with a strong GCA effect for disease severity across sites included PRA783244c3, Pop25, MMV600, and ZUCASRc2. Across sites, the F1 combinations, MMV600 x Pop25, ZUCASRc2 X Pop25, and Pop25 x PRA783244c2 had strong SCA effects for root lodging, ear drooping, husk cover and ear insect damage. In a related diallel analysis of 10 full-sib families derived from these populations, it was observed that resistant x susceptible families and their reciprocal crosses performed better than their resistant parents, suggesting an over dominant expression of resistance. Both maternal and non maternal effects were observed to be influencing resistance to ear rots. There was a preponderance influence of non-additive gene action. A response to full-sib recurrent selection was conducted in four locations in Central Zambia. Out of the 343 families created in 2005/6 season, 10% were selected from each population and recombined to create five new populations. These, with the original populations, were evaluated in four sites during the 2007/8 season. There was a net reduction in ear rot incidence and rot severity in the new synthetic population. Pop10 had the largest reduction in disease severity. The predicted gain per cycle was -4.1% and realized gain was -2.5% for disease incidence, and 0.19% and 19.4% for grain yield. Genetic variability was maintained though with low heritability estimates. Negative but at times strong association between grain yield and ear rot disease severity was detected suggesting that in general selecting for ear rot resistance would enhance grain yield in the five populations. Overall the importance of the ear rots and mycotoxins in compromising yield and health of the communities in Zambia, respectively, were confirmed and support the call to improve maize varieties for resistance to ear rots. The results indicate that the five populations could be enhanced for ear rot resistance through population improvement procedures such reciprocal recurrent selection that exploit both additive and non-additive variation. Selection might be compromised by the large genotype x environment interaction effects, and large reciprocal effects and their interaction with the environments. To enhance repeatability genotypes should be artificially inoculated, by placing the inoculum in the leaf whorl followed by colonized toothpick inoculation, and screened in many environments to identify genotypes with stable resistance to ear rots. / Thesis (Ph.D) - University of KwaZulu-Natal, Pietermaritzburg, 2009.

Bean fly (Ophiomyia spp.) is a major pest of common bean (Phaseolus vulgaris L.) throughout eastern and southern Africa. In the semi-arid areas, apart from drought, the insect pest is reported to cause high crop losses up to 100%, particularly when drought occurs and under low soil fertility. Host-plant resistance is part of the integrated pest management strategies that have been widely employed against major insect pests of tropical legumes. However, information regarding its use in control of bean fly in common bean is limited. Therefore, the objectives of this study were to: (1) validate farmers’ perceptions of major constraints responsible for yield losses, particularly the major insect pests of beans; (2) asses the level of adoption of improved bean varieties and determine factors that influence farmers’ preferences of the varieties and criteria for selection; (3) identify sources of resistance to bean fly available in landraces; (4) determine the nature of gene action controlling bean fly resistance and seed yield in common bean; (5) describe a procedure for generating optimal bean fly populations for artificial cage screening for study of the mechanisms of resistance available in common bean against bean fly. Farmers considered drought and insect pest problems as main causes for low yields. The adoption rate for improved varieties was high but self-sufficiency in beans stood at 23% in the dry transitional (DT) agro-ecology and at 18% in the dry mid-altitude (DM) agroecology, respectively. Drought, earliness, yield stability, and insect pest resistance were the factors determining the choice of varieties by farmers. Bean fly (Ophiomyia spp.), African bollworm (Helicoverpa armigera) and bean aphid (Aphis fabae) were identified as key crop pests of beans limiting yield. The study to identify new sources of resistance included 64 genotypes consisting of landraces, bean fly resistant lines and local checks. The experiment was done under drought stressed (DS) and non-stressed (NS) environments and two bean fly treatments (insecticide sprayed and natural infestation) for three cropping seasons between 2008 and 2009. Genotypes differed in their reaction to natural bean fly attack under drought stressed (DS) and non-stressed environments (NS) over different cropping seasons. However, the effect of bean fly appeared to vary between the long rains (LR) and short rains (SR). It was observed that an increase in the number of pupae per stem resulted in a higher plant mortality. The range of seed yield was from 345 to1704 kg ha-1 under natural infestation and from 591 to 2659 kg ha-1 under insecticide protection. Seed yield loss ranged from 3 to 69 %. The resistance of most of the bean fly resistant lines seemed to be ineffective in presence of DS. To determine the nature of gene action controlling the inheritance of resistance to bean fly, four parents with known reaction to bean fly were crossed with four locally adapted genotypes in an 8 x 8 half-diallel mating design. Similarly, two resistant and two susceptible parents were selected and crossed to produce populations for generations means and variance components analysis. Results revealed that both general combining ability (GCA) and specific combining ability (SCA) mean squares were significant (p A 0.05) for all four traits studied, except SCA for stem damage during one cropping season. Among the parents, GBK 047858 was the best general combiner for all the traits studied across seasons except for stem damage during LR 2009. Genotypes GBK 047821 and Kat x 69 (a locally adapted variety) were generally good general combiners for resistance traits as well as seed yield. General predictability ratio values ranging from 0.63 to 0.90 were obtained for plant mortality, stem damage, pupae in stem and seed yield across cropping seasons. These results established the predominance of additive gene effects (fixable variation) over the non-additive effects in controlling the traits. Low to moderate narrow sense heritability values ranging from 0.22 to 0.45 were obtained for pupae in stem. Such heritability estimates indicate that although additive gene components were critical in the inheritance of resistance for the trait, non-additive gene action was also important in addition to the environmental effects. A major disadvantage in screening for resistance to bean fly in common bean by controlled means in net cages has been the lack of a method to use for raising adequate fly populations for screening. Due to this problem, a simple procedure for raising sufficient numbers of adult bean flies required for screening was described. Through this method, up to 62 % emergence of the adult flies was achieved. Moreover, the flies retained their ability to infest bean plants. To determine the presence of antibiosis and antixenosis mechanisms of resistance in common bean, five genotypes [CC 888 (G15430), GBK 047821, GBK 047858, Ikinimba and Macho (G22501)] and two local check varieties (Kat B1 and Kat B9) were screened under free-choice in outdoor net cages and no-choice conditions in net cages placed in a shadehouse. All the five resistant genotypes tested had relatively long internodes. It was established that long internode was a morphological trait associated with reduced pupation rate in bean stems, hence an antixenosis component of resistance. Both ovipositional non-preference and antibiosis mechanisms were found to exist in three genotypes namely CC 888 (G45430), GBK 047858 and Macho (G22501). These genotypes were resistant when they were subjected to bean fly under both free-choice and no-choice conditions. They had fewer feeding/oviposition punctures, low number of pupae in the stem, reduced damage to the stems and low percent plant mortality. The remaining genotypes, Ikinimba and GBK 047821 only expressed antixenosis. To maximize the effectiveness of host-plant resistance against bean fly, multiple insect resistances should be incorporated into a single bean genotype in order to ensure durability. However, this should be within the background of integrated pest management strategy. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2010.

Maize (Zea mays L.) is the principal crop of Southern Africa but production is threatened by gray leaf spot (Cercospora zea-maydis L.) and phaeosphaeria leaf spot (Phaeosphaeria maydis L.) diseases, drought and the use of unadapted cultivars, among other constraints. There are few studies of gray leaf spot (GLS) and Phaeosphaeria leaf spot (PLS) resistance, drought tolerance, yield stability and maize cultivar preferences in Southern Africa. The objective of this study was to: a) determine farmers’ preferences for cultivars; b) investigate the gene action and heritability for resistance to GLS and PLS, and drought tolerance; and c) evaluate yield stability and its relationship with high yield potential in Southern African maize germplasm. The study was conducted in South Africa and Zimbabwe during 2003 to 2004. A participatory rural appraisal (PRA) established that farmers preferred old hybrids of the 1970s because they had better tolerance to drought stress. Farmers also preferred their local landrace because of its flintier grain and better taste than the hybrids. The major prevailing constraints that influenced farmers’ preferences were lack of appropriate cultivars that fit into the ultra short seasons, drought and low soil fertility. Thus they preferred cultivars that combine high yield potential, early maturity, and drought tolerance in all areas. However, those in relatively wet areas preferred cultivars with tolerance to low soil fertility, and weevil resistance, among other traits. A genetic analysis of 72 hybrids from a North Carolina Design II mating revealed significant differences for GLS and PLS resistance, and drought tolerance. General combining ability (GCA) effects accounted for 86% of genetic variation for GLS and 90% for PLS resistance indicating that additive effects were more important than non-additive gene action in controlling these traits. Some crosses between susceptible and resistant inbreds had high resistance to GLS suggesting the importance of dominance gene action in controlling GLS resistance. Resistance to GLS and PLS was highly heritable (62 to 73%) indicating that resistance could be improved by selection. Also large GCA effects for yield (72%), number of ears per plant (77%), and anthesis-silking interval (ASI) (77%) under drought stress indicated that predominantly additive effects controlled hybrid performance under drought conditions. Although heritability for yield declined from 60% under optimum to 19% under drought conditions, heritability for ASI ranged from 32 to 49% under moisture stress. High heritability for ASI suggested that yield could be improved through selection for short ASI, which is positively correlated with high yield potential under drought stress. The stability analyses of the hybrids over 10 environments indicated that 86% had average stability; 8% had below average stability and were adapted to favourable environments; and 6% displayed above average stability and were specifically adapted to drought stress environments. Grain yield potential and yield stability were positively correlated. In sum, the study indicated that farmers’ preferences would be greatly influenced by the major prevailing constraints. It also identified adequate genetic variation for stress tolerance, yield potential and yield stability in Southern African maize base germplasm, without negative associations among them, suggesting that cultivars combining high yield potential, high stress tolerance and yield stability would be obtainable. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2005.

Sorghum (Sorghum bicolor) is an important cereal crop in sub-Saharan Africa. The parasitic weed Striga hermonthica is a major biotic constraint to sorghum production. A novel technology where planting seeds are coated with herbicide to kill Striga that attach to the roots of the host has been shown to be effective in protecting the cereal crop from Striga damage. However, the host plant must have herbicide tolerance. This technology has not been tested in sorghum because there are no herbicide tolerant sorghum varieties available in Kenya and is, therefore, unavailable for subsistence farmers. One of the ways in which genetic variation can be enhanced and herbicide resistance developed is through chemical mutagenesis with ethyl methane sulfonate (EMS). The objectives of this project, therefore, were to: 1) identify sorghum production constraints through farmer PRA in order to determine breeding priorities.in two Striga endemic districts in western Kenya; 2) develop an EMS mutagenesis protocol for sorghum and to enhance the genetic variability of the crop using chemical mutagenesis; 3) evaluate EMS-derived sorghum mutants for improved agronomic performance; 4) develop acetolactate synthase (ALS) herbicide resistance in sorghum and to characterize the mode of inheritance of the trait; 5) determine the effect of herbicide coating of seed of herbicide tolerant sorghum on Striga infestation. In order to determine breeding priorities and constraints in sorghum production and the likelihood of adoption of herbicide seed coating technology, a survey involving 213 farmers was conducted in two Striga endemic rural districts of Nyanza province in Kenya. Results indicated that local landraces like Ochuti, and Nyakabala were grown by more farmers (> 60%) than the improved varieties like Seredo and Serena (48%). Popularity of the landraces was linked to Striga tolerance, resistance to drought, bird damage and storage pests, yield stability and high satiety value. Major constraints to sorghum production were drought, Striga weed, storage pests, bird damage and poverty among the rural farmers. Important characteristics farmers wanted in new varieties were Striga and drought resistance, earliness, resistance to bird and weevil damage and good taste. Striga infestations in sorghum fields were > 70%. Cultural Striga control options were considered inadequate while inorganic fertilization and chemical control were considered effective but unaffordable. Farmers’ willingness to pay a premium of over 30% for a Striga solution gave indication that herbicide seed coating if effective could be adopted by farmers. As a prerequisite to development of herbicide resistance, a comparative study was carried out to determine optimum conditions for mutagenesis and to induce genetic variation in the sorghum. Two sorghum varieties were mutagenized using varying concentrations (0.1 to 1.5% v/v) of EMS and two exposure times (6h and 12h). In laboratory and greenhouse experiments, severe reduction of sorghum root and shoot lengths indicated effective mutagenesis. The LD50 based on shoot length reduction was 0.35% and 0.4% EMS for 6h for Seredo and Kari/mtama-1, respectively. The highest mutation frequency based on chlorophyll abnormalities was 56% for 0.3% EMS for 6h. In the M2 generation, phenotypic variances for panicle characteristics were increased on treatment with EMS. However, significant effects of exposure time and variety indicated the necessity of genotype optimization for some traits. In order to determine the significance of mutation breeding in sorghum, 78 mutant lines derived from EMS mutagenesis, their wild type progenitor (Seredo) and two local checks (Kari/mtama-1 and Serena) were evaluated for agronomic performance in two locations in Kenya. There were significant (P = 0.05) effects among entries for grain yield, 1000-seed weight and visual scores for height uniformity, head exertion, head architecture and overall desirability. The highest yielding entry-mutant line “SB2M13” had a yield of 160% and 152% relative to the wild type (Seredo) and the best check Kari/mtama-1, respectively. Mutant line “tag27” had the highest 1000-seed weight which was 133% relative to the wild type. Seven mutant lines were rated superior to the wild type for panicle characteristics, head exertion and overall desirability. However, the majority of mutants were inferior to the wild type for most characteristics. Superior mutant lines may be developed into direct mutant varieties after multi-location trials or used as breeding material for sorghum improvement. In order to develop acetolactate synthase (ALS) herbicide resistance in sorghum, over 50,000 seeds of Seredo were mutagenized with 0.3% EMS. Over four million M2 plants were screened using 20g ha-1 of the ALS herbicide, sulfosulfuron. Five mutants (hb46 hb12, hb462, hb56 and hb8) survived the herbicide treatment and were confirmed to be tolerant. Mutant lines displayed differential herbicide tolerance, and the general order of tolerance after spray or seed coat application was hb46 > hb12 > hb462 ~ hb56 > hb8. The LD50 values for herbicide application as a spray, or seed coat, showed mutant lines to be up to 20 and 170 fold, respectively, more resistant than the wild type. Chi square analysis of data from herbicide screening of F2 generation of mutant X wild type crosses indicated no difference from the Mendelian segregation of 1:2:1 indicating the herbicide tolerance was inherited as a single semi-dominant gene. Mutant X mutant crosses did not show allelism indicating that the tolerance in all five mutants could be a result of the same gene mutation. To determine effect of herbicide seed coating on Striga infestation, the five herbicide tolerant mutant lines, hb46, hb12, hb462, hb56 and hb8 and the wild type progenitor Seredo were coated with varying concentrations (0.5-1.5% g ha-1) of sulfosulfuron and planted in a Striga endemic field. There were significant (P=0.05) effects of herbicide concentration on Striga density, Striga flowering and seed set, and sorghum plant stand and biomass. All treatments with herbicide coated on sorghum seeds had lower Striga emergence. Coating sorghum seed with 1g ha-1 sulfosulfuron reduced Striga infestation, Striga flowering and Striga seed set by 47%, 52% and 77%, respectively, and was considered the most effective rate as it did not result in sorghum biomass reduction. Mutants displayed differential herbicide tolerance and Striga resistance. Combining seed coating with high herbicide tolerance and inherent Striga resistance would be most effective for Striga control. Overall, the study showed that EMS mutagenesis is effective in inducing variation in sorghum for several traits including herbicide resistance. The mutants developed in this study will be important for sorghum breeding and for protection of sorghum against the Striga weed. / Thesis (Ph.D.) - University of KwaZulu-Natal, Pietermaritzburg, 2009.

Reduced banana yield owing to black Sigatoka Mycosphaerella fijiensis Morelet is a threat to the livelihoods of Ugandan subsistence farmers who depend entirely on the banana crop for food security. The objectives of this investigation were to: (i) assess farmers’ knowledge of black Sigatoka disease in central Uganda; (ii) document the qualities farmers would desire in the banana genotypes to be developed for black Sigatoka resistance; (iii) appraise the methods for assessing black Sigatoka resistance in diploid banana populations; (iv) determine the phenotypic variation for black Sigatoka resistance and agronomic traits in diploid and tetraploid bananas; (v) determine the influence of tetraploid and diploid parents on the black Sigatoka resistance and agronomic traits in the triploid progenies; and (vi) evaluate 2x by 2x banana progenies for yield and black Sigatoka resistance. A survey that focused on low and medium banana production zones in Uganda established that there was limited awareness of black Sigatoka disease as a constraint on banana production in the areas surveyed. It was also established that farmers liked local bananas because of their superior taste, early maturity, and marketability. There were farmers who had been exposed to new black Sigatoka resistant materials but never liked these new banana materials because of poor taste and lack of market. Farmers desired new banana materials with good taste on cooking, heavy bunches, resistance to pests and diseases, drought tolerance, and early maturing capacity in that order. The results indicated that the banana farmers in Uganda attached more importance to food quality attributes than to production attributes especially when considering new banana materials. This suggested that farmers mainly grow bananas for consumption. Three black Sigatoka assessment methods, youngest leaf spotted, disease development time and area under disease progress curve (AUDPC) were appraised using a diploid population. All the three methods were able to classify the diploid accessions into resistant and susceptible clones. The cultivar rankings of AUDPC correlated strongly with the rankings of disease development time. The cultivar rankings of AUDPC correlated positively with the rankings of youngest leaf spotted method. The youngest leaf spotted at flowering and AUDPC predicted significantly total number of leaves at flowering (R2 = 0.53). Overall AUDPC had the highest coefficient of determination (R2=0.84) in assessment of banana diploids for black Sigatoka resistance indicating that it accounted for the highest variation in disease response observed among the diploid clones. From this investigation it was recommended that AUDPC should be used to assess resistance on black Sigatoka in Musa species. A phenotypic analysis on the diploid and synthetic tetraploids, and a molecular analysis using RAPD markers on the tetraploid population were conducted. Results indicated that the diploid population had significant (P<0.001) variation for plant height, plant girth, days from flowering to harvest, bunch weight, number of suckers, youngest leaf spotted, total leaves at flowering, area under disease progress curve, and number of functional leaves at harvest. Principal component analysis showed that plant height and girth explained most of the variation observed in the diploid population. In the tetraploid population, significant differences were observed for plant height, plant girth, and number of suckers (P<0.05). In the tetraploids principal component analysis, indicated that youngest leaf spotted and total leaves at flowering had higher loadings on principal component one. Genetic distances computed from RAPD markers indicated limited genetic variability in the tetraploid population. Another investigation was also carried out to determine the influence of tetraploid and diploid parents on black Sigatoka resistance and agronomic traits in the triploid progenies generated from tetraploid-diploid crosses. The results indicated that diploids transferred black Sigatoka resistance to triploid progenies as measured by disease development over time, the number of functional leaves at flowering and at harvest. On the other hand, the female synthetic tetraploids influenced plant height and bunch weight in the triploid progenies generated from tetraploid-diploid crosses as observed from triploid progeny correlations and parent-offspring regressions. Therefore, it is important to select tetraploids with heavy bunch weights to generate high yielding triploids in tetraploid-diploid crosses. Lastly, this thesis investigated the relationship between bunch weight and black Sigatoka resistance traits in 2x by 2x progenies generated using a random polycross design. Phenotypic correlations revealed strong positive relationships between bunch weight with total leaves at flowering, youngest leaf spotted, plant girth, and days from planting to flowering among the 2x by 2x progenies. Linear regression analysis indicated that girth, total fingers and finger length significantly predicted bunch weight (R2=0.67). However, days from planting to flowering, and total leaves at flowering had strong indirect effects on bunch weight via plant girth. The results imply that selection for parents with good combining ability for girth, finger length and total fingers can improve bunch weight in a diploid population. / Thesis (Ph.D.) - University of KwaZulu-Natal, Pietermaritzburg, 2009.

Maize (Zea mays L.) production is constrained by a number of stresses, amongst the most important are gray leaf spot (GLS) caused by a fungus Cercospora zeaemaydis Tehon and E.Y. Daniels and Phaeosphaeria leaf spot (PLS) caused by Phaeosphaeria maydis (Henn.). The diverse germplasm comprising farmer collections and exotic material used in the medium and highland altitudes maize breeding programmes in western Kenya has not been improved for resistance to the two diseases. Heterotic patterns of germplasm from this region have also not been studied. Therefore, the objectives of this study were to (i) assess the prevalence, importance, and farmers’ perceptions of GLS and PLS, (ii) characterize maize germplasm collections into their heterotic groups and (iii) improve four maize populations for GLS and PLS resistance through recurrent selection. The participatory rural appraisal (PRA) was conducted at three sites in western Kenya during the 2005/2006 cropping season. Data was generated using a checklist in group discussions with 109 male and 123 female farmers as well as key informants. Constraints were identified and prioritised. The five most limiting, in order of importance, were low soil fertility, poor varieties and seed, drought, Striga, pests and diseases (GLS and PLS). Gray leaf spot and PLS were reported in all sites but farmers did not know the causes of these diseases. Farmers preferred local varieties Tiriki, Anzika and Kipindi due to their greater resistance to diseases than commercial hybrids. Farmer criteria for variety selection were low fertilizer, Striga and disease resistance, drought tolerance, closed tips, and high yield potential. Due to the high cost of hybrid seed farmers selected and planted their own seed from advanced generations from previous seasons. Across all the sites, yield gap between on-farm and expected yield potential was estimated as ranging from 4.73t ha-1 to 5.3t ha-1 mainly due to the identified constraints. Therefore maize breeding should focus on addressing important maize production constraints and farmers’ preferences identified in this study in developing varieties that will increase maize yields on-farm. During 2005/2006, seventy 77 testcrosses were developed through crossing 47 germplasm collections with four population testers, Kitale synthetic II (KSII), Ecuador 573 (EC 573), Pool A and Pool B. Crosses and testers were evaluated at Kakamega during 2006/2007 in a 9 x 9 triple lattice design. Significant (p < 0.05) differences in grain yield, ear height, days to 50% anthesis, GLS and PLS resistance were observed. Both general and specific combining ability effects (GCA and SCA, respectively) were significant (p < 0.01), with SCA accounting for more than 50% of the variation for GLS, PLS and yield and less than 50% for ear height, days to 50% anthesis and silk. This indicated that both additive and non-additive gene effects were important but non-additive gene effects were more important in conditioning these traits. High SCA effects indicated high heterosis between collections and populations. Both yield heterosis and SCA were used to study heterotic patterns, but percentage yield heterosis data was used to classify these materials into heterotic groups. Based on significance (p < 0.05) of percentage yield heterosis as a primary factor for classification, seven collections were classified to Pool A, 17 to Pool B, 12 to KSII and 6 to EC 573 heterotic groups. The study indicated that germplasm collections belong to distinct heterotic groups therefore they can be infused into these populations (Pool A, Pool B, KSII and EC 573). Four populations, KSII, EC 573, Pool A and Pool B were subjected to one cycle of reciprocal recurrent selection (RRS) and two cycles of simple recurrent selection (SRS) during the 2004-2006 cropping seasons at Kakamega. Response to selection was assessed by evaluating C0, C1 and C2 and four commercial checks in a randomised complete block design in three replications at Kakamega and Kitale during 2007. All cycles except C0 of Pool A were more resistant to GLS than the three checks, H623, KSTP94 and PHB3253. Response to selection for GLS was significant (p < 0.01) in the desired direction. Gains ranged from -32.2% to 6.4% cycle-1 for RRS and 0.0% to -61.3% cycle-1 for SRS. Heritability estimates of between 59% and 76.3% for GLS and 39% and 80% for PLS were observed indicating that both GLS and PLS can be improved through selection. Significant negative correlations between GLS and yield were observed in Pool A C0 (r = -0.947, p < 0.01) and between yield and PLS in Pool A C0 (r = -0.926, p < 0.01). These indicated gain in yield as GLS and PLS were selected against. Generally, SRS out performed RRS method both in genetic gain and time, as indicated by gain of -61% for SRS and -32.2% for RRS, respectively. Two cycles of selection were achieved in two years with SRS as compared to only one with RRS. These results clearly demonstrated that it is possible to improve for GLS resistance using simple and reciprocal recurrent selection methods. The main constraints to maize production in Western Kenya were low soil fertility, Striga, drought, lack of seed and diseases. Farmers preferred varieties that can do well under the constraints mentioned. Local collections belonged to distinct heterotic groups with good resistance to GLS and PLS and were highly heterotic to four maize population testers with both SCA and GCA effects being important in conditioning GLS and PLS resistance. Recurrent selection methods were found to improve maize resistance to GLS and PLS. Breeding should therefore, focus in development of hybrids and improvement of populations using these local collections by employing SRS and RRS selection methods with identified constraints and farmer preferences in mind. / Thesis (Ph.D.) - University of KwaZulu-Natal, Pietermaritzburg, 2007.

Viral diseases are a major limiting factor to cowpea production in many countries of Africa. In Uganda, studies indicated that the cowpea aphid-borne mosaic virus (CABMV) is common and a potential threat to cowpea production in the region. There have been no efforts to develop cowpea cultivars with resistance to CABMV in Uganda. This work focused on the development of cultivars resistant to CABMV. Production of cowpea in Uganda is constrained by several factors, including a lack of awareness of diseases among the majority of farmers. A participatory rural appraisal (PRA) was conducted to elicit farmers’ indigenous knowledge of cowpea production and also to gain insight into their understanding of viral diseases affecting cowpea in Uganda. PRA tools such as group discussions, transect walks, problem listing and ranking were used to gather information. Insect pests, diseases, low yielding cultivars and the high cost of pesticides were perceived to be the most important production constraints. Farmers were not aware of the problem of virus diseases, but provided descriptive names of symptoms. Only three cowpea cultivars (Ebelat, Ecirikukwai and Blackcowpea) were produced in the area. Seed size and colour were seen as important traits in new varieties. Information about the occurrence, distribution and identity of cowpea viruses is limited in Uganda. The objective of this study was to identify the important cowpea virus diseases occurring naturally in the major cowpea growing regions of Uganda. Surveys were conducted to determine the incidence and severity of virus symptoms in four districts (Soroti, Kumi, Pallisa and Tororo) in 2004 and 2005. The incidence ranged from 40.5 to 94.4% and severity ranged from 15.0 to 30.6% (for Kumi and Pallisa districts, respectively) during the 2004 surveys. In 2005, the incidence ranged from 55.9 to 85.4% and severity ranged from 4.7 to 14.5% (for Tororo and Soroti districts, respectively). The CABMV, cowpea mild mottle virus (CPMMV), cowpea severe mosaic virus (CPSMV) and cucumber mosaic virus (CMV) were serologically detected by double antibody sandwich enzyme- linked immunosorbent assay (DAS-ELISA). Fifty four improved cowpea genotypes were screened for resistance to CABMV during the first season of 2004 at Serere Agricultural and Animal Production Research Institute in Uganda. Further screening was conducted in the second season of 2004 using 27 genotypes. The genotypes were planted in single rows between the rows of the susceptible cultivar, Ebelat. This was to provide high pressure of aphid vector (Aphis craccivora Koch) and CABMV inoculum. In addition, the test genotypes were artificially inoculated with a CABMV extract on fully expanded primary leaves of fourteen day-old seedlings. The CABMV incidence and severity was assessed. Disease severity was assessed on a 0-60% visual estimation scale where 0 = with no symptoms and 60 = with severe symptoms. Serological analysis was conducted using DAS-ELISA. Five genotypes showed good levels of resistance to CABMV, namely MU-93, IT82D-889, IT82D-516-2, IT85F-2841 and SECOW-2W. These resistant lines were crossed with three susceptible local landraces, namely Ebelat, Ecirikukwai and Blackcowpea in a North Carolina II mating design. The F1, F2 and BC1F1 populations and the parents were evaluated in the field to assess the response to CABMV and to study the inheritance of resistance to CABMV. The general combining ability (GCA) and specific combining ability (SCA) effects were significant, indicating that both additive and non-additive genetic factors are important in determining the control of CABMV in cowpea. The proportions (%) of the sum of squares for crosses attributable to GCA and SCA for CABMV severity were 51.4% for GCA due to females, 8.4% for GCA due to males and 40.2% for the SCA. The narrow-sense heritability estimates, obtained by regressing F1 on mid-parents was 0.87 and 0.84, F2 on F1 progenies 0.49 and 0.48, and F2 progenies on mid-parents 0.63 and 0.79, for AUDPC and final disease severity, respectively. Single gene conditioned resistance in seven populations, but resistance was quantitatively inherited and involved many genes in eight populations. Observation of transgressive segregation and moderate to high heritability suggests a quantitative mode of gene action and the importance of additive effects. The predominance of GCA variance, high heritability estimates and observation of transgressive segregation suggested that resistance could be improved by selection. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2007.