UNDERSTANDING THE GENETIC COMPONENTS OF DROUGHT TOLERANCE IN COMMON BEAN (PHASEOLUS VULGARIS L.) By Isaac Onziga Dramadri A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Plant Breeding, Genetics and Biotechnology-Crop and Soil Sciences -Doctor of Philosophy 2018 ABSTRACT UNDERSTANDING THE GENETIC COMPONENTS OF DROUGHT TOLERANCE IN COMMON BEAN (PHASEOLUS VULGARIS L.) By Isaac Onziga Dramadri Common bean (Phaseolus vulgaris. L.) is a cheap source of protein, energy, and micronutrients including iron, zinc, vitamins, and dietary fiber for millions in the developing world. Bean production is constrained by a number of abiotic and biotic stresses, and drought stress has become the most important abiotic stress negatively affecting subsistence farming systems particularly in Africa. The overall goal of this study was to explore additional genetic diversity for drought tolerance within the Andean gene pool, and understand the genetic control of drought responses using genomic and phenometric tools. The first objective focused on the determination of the genetic architecture of yield component and photosynthetic traits at pod filling stage under terminal drought stress using a genome-wide association study (GWAS). This study involved 256 common bean genotypes from the Andean gene pool that were evaluated for two seasons (2016 and 2017) under stress and non-stress at two locations (Kasese and Namulonge) in Uganda. Several significant marker-trait associations were identified for agronomic and photosynthetic traits under drought stress conditions. Overlapping GWAS signals were detected for yield components and partitioning traits on chromosomes Pv02, Pv06, and Pv11 respectively, while colocalized signals for photosynthetic traits were identified on Pv03, Pv04, and Pv11 under drought stress conditions. Positional candidate genes, including Phvul.006G117500 encoding, plant invertase/pectin methylesterase inhibitor (INH/PMEI) superfamily protein, a positive regulator of carbon partitioning and sink strength in the reproductive stage, and Phvul.011G210000, encoding DNAJ heat shock N-terminal domain-containing protein, were identified. The second objective focused on identifying quantitative trait loci (QTL) associated with drought tolerance in a Portillo x Red Hawk recombinant inbred line (RIL) mapping population. 97 F4:7 RILs were evaluated under drought stress and non-stress conditions in the field for two seasons (2016 and 2017), at Kasese and one season (2017) at Namulonge in Uganda. Thirty-two significant QTL signals were identified for phenology, yield components and partitioning traits measured under drought and non-stress conditions. Colocalized QTL signals were identified for phenology, yield component traits and partitioning traits on Pv01, Pv02, Pv03, Pv04, Pv06 and Pv11. Functional annotation of positional candidate genes identified within the colocalized peak regions were consistent with the functional similarity or pleiotropic effects of those genes on the associated traits, and could be validated for marker assisted breeding. The third objective focused on the use photosynthetic traits and visual scores for high throughput screening for drought tolerance at the seedling stage. Three hundred genotypes were evaluated under stress and non-stress conditions in the green house at Namulonge, Uganda. Drought stress was applied by withholding water 21 days after planting. Significant genetic variability was observed among accessions for photosynthetic traits during early stages of drought stress, suggesting that substantial genetic variability exists among common bean genotypes from the Andean gene pool for photosynthetic traits under drought conditions at the seedling stage. Slow wilting was positively associated with recovery and could be used to identify drought tolerant genotypes at the seedling stage. The results from GWAS, QTL analysis and high throughput screening for drought tolerance at seedling stage will facilitate the screening of germplasm, and use of marker assisted breeding for drought tolerance, to improve the genetic gain for yield under drought stress in the Andean gene pool of common bean. Copyright by ISAAC ONZIGA DRAMADRI 2018 iv This dissertation is dedicated to my parents Andema Rofino (Late) and Helen Acidriru, my wife Annet Maliko, and son Gabriel Aita , and finally to my siblings for their encouragement and support. v ACKNOWLEDGEMENTS I am greatly thankful to the almighty God for protection. I would like to thank the BHEARD program for the scholarship opportunity with support from the United states Agency for International development, as part of the Feed the Future initiative, under the CGIAR Fund, award number BFS-G-11-00002, and the predecessor fund the Food Security and Crisis Mitigation II grant, award number EEM-G-04-00013. I wish to take this chance to express my humble and sincere gratitude, and appreciation to my advisor Dr. James D. Kelly for giving me an opportunity to work with him, and for the countless efforts, support and encouragement. Your dedication and commitment has forever changed my entire life. I thank and appreciate my committee members Drs. Wayne Loescher, Dechun Wang, and Cholani K. Weebadde for their wiliness to serve on the committee. I thank Dr. David Kramer lab and Dr. Karen Cichy for their support and encouragement. I thank the Legume Innovation Lab, the BHEARD and The Elmer C. Rossman Endowed Graduate Student Support Award for providing me with funding for field research I thank the past and present members of Kelly and Cichy‘s Labs for their support. In particular, I thank Evan Wright, Halima Awale for their help. I thank Makerere University, Dr. Stanley Nkalubo, Program leader, Legume program in Uganda, Technicians at Kasese Experimental Station, and Dr. Clare Mukankusi, CIAT Uganda for providing conducive environment for the field research. vi TABLE OF CONTENTS LIST OF TABLES ......................................................................................................................... ix LIST OF FIGURES ....................................................................................................................... xi CHAPTER 1: GENERAL INTRODUCTION ............................................................................... 1 CHAPTER 2: DETERMINATION OF THE GENETIC ARCHITECTURE FOR DROUGHT ADAPTIVE ALLELES IN THE ANDEAN DIVERSITY PANEL USING GENOME WIDE ASSOCIATION STUDY................................................................................................................ 9 Abstract ........................................................................................................................................... 9 Introduction ................................................................................................................................... 11 Materials and Methods .................................................................................................................. 14 Plant Materials........................................................................................................................... 14 Field Phenotyping ..................................................................................................................... 14 Genotyping ................................................................................................................................ 16 Data Analysis ............................................................................................................................ 17 Population Structure and Marker-Trait Association Analysis .................................................. 17 Heritability ................................................................................................................................ 18 Results ........................................................................................................................................... 19 Rainfall ...................................................................................................................................... 19 Field Experiments ..................................................................................................................... 20 Population Structure .................................................................................................................. 21 Marker-Trait Associations......................................................................................................... 21 Number of pods per plant (NP) ................................................................................................. 21 Pod weight per plant (PW) ........................................................................................................ 22 Number of seeds per plant (SN) ................................................................................................ 23 Seed yield per plant (SY) .......................................................................................................... 23 Harvest index (HI) ..................................................................................................................... 24 PhiNPQ...................................................................................................................................... 24 Linear Electron Flow (LEF) ...................................................................................................... 25 Phi2............................................................................................................................................ 25 PhiNO ........................................................................................................................................ 26 Allelic Effects of Significant SNPs on Yield Component Traits .............................................. 26 Discussion ..................................................................................................................................... 27 Marker-Trait Associations......................................................................................................... 28 Yield Component Traits ............................................................................................................ 29 Number of pods per plant .......................................................................................................... 29 Pod weight per plant .................................................................................................................. 31 Number of seeds per plant (NSP) .............................................................................................. 32 Seed yield per plant (SY) .......................................................................................................... 34 Harvest index (HI) ..................................................................................................................... 39 Colocalization of GWAS Signals for Photosynthetic Traits ..................................................... 41 Conclusion .................................................................................................................................... 44 vii APPENDIX ................................................................................................................................... 46 CHAPTER 3: IDENTIFICATION OF QUANTITATIVE TRAIT LOCI ASSOCIATED WITH DROUGHT TOLERANCE IN PORTILLO X RED HAWK RECOMBINANT INBRED LINE MAPPING POPULATION .......................................................................................................... 61 Abstract ......................................................................................................................................... 61 Introduction ................................................................................................................................... 63 Materials and Methods .................................................................................................................. 66 Plant Materials........................................................................................................................... 66 Phenotypic Data Analysis ......................................................................................................... 68 Heritability ................................................................................................................................ 68 Genotyping ................................................................................................................................ 69 Genetic Map Construction ........................................................................................................ 69 Quantitative Trait Loci (QTL) Analysis.................................................................................... 69 Results ........................................................................................................................................... 70 The effect of drought stress on performance of the RIL population ......................................... 70 Genetic Map and QTL Analysis................................................................................................ 72 Days to flowering ...................................................................................................................... 73 Days to harvest maturity ........................................................................................................... 74 Pod weight per plant .................................................................................................................. 75 Seed yield per plant ................................................................................................................... 76 Harvest index (HI) ..................................................................................................................... 78 Pod partitioning index (PPI) ...................................................................................................... 79 Colocalized QTL Regions ......................................................................................................... 80 Discussion ..................................................................................................................................... 81 Putative candidate genes in genomic regions associated with colocalized QTL peaks ............ 88 Conclusion .................................................................................................................................... 91 APPENDIX ................................................................................................................................... 93 CHAPTER 4: DETERMINATION OF THE USE OF PHOTOSYNTHETIC TRAITS AS DROUGHT SENSITIVITY INDICATORS OF DROUGHT TOLERANCE IN COMMON BEAN .......................................................................................................................................... 110 Abstract ....................................................................................................................................... 110 Introduction ................................................................................................................................. 111 Materials and Methods ................................................................................................................ 113 Statistical Analysis ...................................................................................................................... 114 Results ......................................................................................................................................... 115 Photosynthetic Traits ............................................................................................................... 115 Determination of sensitivity of photosynthetic traits to drought stress ................................... 115 Correlation analysis for Phi2 and PhiNPQ among the high and less sensitive genotypes ...... 116 Discussion ................................................................................................................................... 117 Conclusion .................................................................................................................................. 119 APPENDIX ................................................................................................................................. 120 REFERENCES ........................................................................................................................... 130 viii LIST OF TABLES Table 2.1a. Mean, minimum and maximum phenotypic values associated with agronomic traits and partitioning in Andean Diversity panel of 256 common bean genotypes grown under drought stress and non-stress conditions in Kasese for two seasons, 2016 and 2017 in Uganda .............. 47 Table 2.1b. Mean, minimum and maximum phenotypic values associated with agronomic traits and partitioning in Andean Diversity panel of 256 common bean genotypes grown under drought stress and non-stress conditions in Namulonge for two seasons, 2016 and 2017 in Uganda ....... 48 Table 2.2. Mean and Variance associated with photosynthetic traits in Andean Diversity Panel (ADP) of 256 common bean genotypes grown under drought stress and non-stress conditions in Kasese for two seasons, 2016 and 2017 in Uganda. ..................................................................... 49 Table 2.3. Pearson correlation coefficients among nine agronomic and three photosynthetic traits measured on 256 common bean genotypes in the Andean diversity panel grown at the Kasese, Uganda drought stress conditions in 2017. ................................................................................... 50 Table 2.4a. Location, chromosome, position, P-values, proportion of phenotypic variation explained (R2), and minor allele frequency of two most significant single nucleotide polymorphisms (SNPs) for five agronomic traits and four photosynthetic traits measured on 256 common bean genotypes in the Andean diversity panel grown under DS Conditions at Kasese and Namulonge, Uganda in 2016 and 2017. ................................................................................. 51 Table 2.4b. Location, chromosome, position, P-values, proportion of phenotypic variation explained (R2), and minor allele frequency of significant single nucleotide polymorphisms (SNPs) for four agronomic traits with 265K SNPs measured on 256 common bean genotypes in the Andean diversity panel grown under drought stress (DS) and non-stress (NS) conditions at Kasese and Namulonge, Uganda in 2016 and 2017. .................................................................... 52 Table 2.5. Eleven genotypes with good performance based on geometric mean for seed yield from the Andean Diversity Panel evaluated under DS and NS at Kasese and Namulonge in 2016 and 2017. ....................................................................................................................................... 56 Table 3.1 Mean, range for yield component traits, phenology and portioning indices for Portillo x Red Hawk common bean recombinant inbred line (RIL) population evaluated under drought stress (DS) and non-stress (NS) in Kasese and Namulonge, Uganda in 2016 and 2017. ............. 97 Table 3.2 Pearson correlation coefficients among mean variables for days to flowering, days to maturity, harvest index, pod partitioning index , seed yield per plant, and pod weight per plant for the Portillo x Red Hawk recombinant inbred line (RIL) population evaluated under drought stress and non-stress in Kasese and Namulonge, Uganda during 2016 and 2017 growing seasons. ......................................................................................................................................... 98 ix Table 3.3 Single nucleotide polymorphism (SNP) marker distribution and distance on individual chromosomes of the genetic linkage map of Portillo x Red Hawk recombinant inbred line (RIL) population ..................................................................................................................................... 99 Table 3.4 Quantitative trait loci (QTL) for days to 50% flowering(DF), days to maturity(DM) and pod weight per plant(PW) in the Portillo x Red Hawk recombinant inbred line (RIL) evaluated under drought stress and non-stress in Kasese and Namulonge, Uganda during 2016 and 2017 growing seasons. ......................................................................................................... 100 Table 3.5 Quantitative trait loci (QTL) for seed yield in the Portillo x Red Hawk recombinant inbred line (RIL) evaluated under drought stress and non-stress in Kasese and Namulonge, Uganda during 2016 and 2017 growing seasons. ...................................................................... 101 Table 3.6 Quantitative trait loci (QTL) for harvest index(HI) and pod partitioning index(PPI) in the Portillo x Red Hawk recombinant inbred line (RIL) evaluated under drought stress and non- stress in Kasese and Namulonge, Uganda during 2016 and 2017 growing seasons................. 102 Table 4.1. Mean and variance for photosynthetic traits of 294 Andean common bean genotypes measured at 22, 24, and 26 days after planting (DAP) evaluated under water stress conditions. ..................................................................................................................................................... 121 Table 4.2. Mean and variance for visual scores of 294 Andean common bean genotypes measured at 24, 26 and 32 days after planting (DAP) evaluated under water stress conditions. 122 Table 4.3. Correlations among photosynthetic traits Phi2, PhiNO, PhiNPQ, slow wilting at 26 DAP and recovery at 32 DAP for high and low sensitive common bean genotypes (60) genotypes selected using PhiNPQ at 26 DAP under drought condition ...................................................... 126 Table 4.4. Mean photosynthetic traits, visual scores, and geometric mean of seed yield for the 10 best individuals based on PhiNPQ score determined at 26 days after planting. ......................... 129 x LIST OF FIGURES temperature (°C) rainfall (mm) and mean for Kasese_2016(a), Figure 2.1. Mean Kasese_2017(b); Namulonge_2016(c) and Namulonge_2017(d) growing seasons. .................. 54 Figure 2.2. The plot of principle component analysis (PCA) PC1 against PC2, illustrating the population structure in subset of the Andean Diversity Panel used in this study with 265K SNPs ....................................................................................................................................................... 55 Figure 2.3. Manhattan plots showing candidate single nucleotide polymorphism (SNP) markers, and their P- values from genome wide association using mixed linear model for number of pods per plant on Pv02, Pv03, Pv06, Pv07, andPv08, pod weight per plant on Pv06 in Kasese for two seasons under DS conditions. ....................................................................................................... 57 Figure 2.4. Manhattan plots showing candidate single nucleotide polymorphism (SNP) markers, and their P- values from genome wide association using mixed linear model for and seed yield per plant on Pv02, Pv03, Pv05, Pv07, Pv08,andPv10 under DS and harvest index on Pv02, and Pv04 under NS conditions. .......................................................................................................... 58 Figure 2.5. Manhattan plots showing candidate single nucleotide polymorphisms (SNPs) and their P- values from genome wide association using mixed linear model for LEF on Pv04 andPv011, for PhiNPQ on Pv04, Pv05, Pv09 and Pv011 and for Phi2 on Pv011 under drought conditions. ..................................................................................................................................... 59 Figure 2.6. Manhattan plots showing candidate single nucleotide polymorphisms (SNPs) and their P- values from genome wide association using mixed linear model for PhiNO on Pv03 and for Phi2 on Pv03 and Pv05 under drought conditions. ................................................................. 60 Figure 3.1a Histograms of days to flowering and days to maturity for Portillo x Red Hawk common bean recombinant inbred line (RIL) population evaluated under drought stress (DS) and non-stress (NS) in Kasese and Namulonge, Uganda in 2016 and 2017 ....................................... 94 Figure 3.1b Histograms of seed yield per plant and pod weight per plant for Portillo x Red Hawk common bean recombinant inbred line (RIL) population evaluated under drought stress (DS) and non-stress (NS) in Kasese and Namulonge, Uganda in 2016 and 2017 ....................................... 95 Figure 3.1c Histograms of harvest index and pod partitioning index for Portillo x Red Hawk common bean recombinant inbred line (RIL) population evaluated under drought stress (DS) and non-stress (NS) in Kasese and Namulonge, Uganda in 2016 and 2017 ....................................... 96 Figure 3.2 Common bean chromosomes Pv01, Pv012, Pv03, Pv04, Pv06, Pv07, Pv08, Pv010, and Pv11 showing quantitative trait loci (QTL) for phenology, yield component traits, and partitioning traits in Portillo x Red Hawk recombinant inbred line (RIL) grown under drought stress and non-stress in Kasese and Namulonge, Uganda during 2016 and 2017 growing seasons ..................................................................................................................................................... 103 xi Figure 4.1. The boxplot for the means of photosynthetic traits PhiNO, Phi2 and PhiNPQ of 294 Andean common bean genotypes measured at 22 and 26 days after planting (DAP) evaluated under water stress conditions. ..................................................................................................... 123 Figure 4.2. The effect of drought stress on photosynthetic traits. (A) The average Phi2, PhiNO and PhiNPQ measured drought and control conditions measured on 22, 24 and 26DAP. (B) The effect of drought stress on photosynthetic traits based on the difference between the means of drought stress and control treatments ......................................................................................... 124 Figure 4.3. The distribution of Phi2 and PhiNPQ of 294 Andean common bean genotypes measured at 26 days after planting (DAP) under water stress conditions. ................................. 125 Figure 4.4. The variation in photosynthetic traits PhiNPQ, Phi2, and PhiNO among high sensitive low sensitive genotypes under drought conditions on 22, 24 and 26 DAP ................. 127 xii CHAPTER 1: GENERAL INTRODUCTION Common bean (Phaseolus vulgaris .L) is the most important grain legume consumed worldwide (Beebe et al., 2014; Bellucci et al., 2014). Nutritionally, bean is a cheap source of protein, energy, and micronutrients such as iron and zinc, vitamins, and dietary fiber for millions in Africa and Latin America (Broughton et al., 2003; Akibode and Maredia, 2011). In the great lakes region of eastern Africa, bean provides up to 65% of protein, 32% of energy, and vital micro nutrients (Welch et al., 2000). Highest per capita consumption of 60kg have been reported in Rwanda and western Kenya (Blair et al., 2010; Akibode and Maredia, 2011). In Uganda, common bean is a food security crop particularly for low income families, and provides up to 25% of energy; 45% of total protein, micro nutrient and dietary fiber (Namugwanya et al., 2014). Common bean has become an important component of diets in developed countries because of the health benefits associated with disease prevention namely, diabetes, mellitus, coronary heart disease and colon cancer (Thompson et al., 2009). In 2014, the total world production of common bean was 26.5 million metric tons on 30.62 million hectares of land. In comparison to the production levels in 1990, this was a 51.2% increase in production volume and about 15% increase in area cultivated (FAOSTAT, 2014). The increase in global production over the last two decades has been attributed to genetic gains and improved management practices rather than increase in area under bean production, which has expanded to marginal areas prone to drought and soil nutrient deficiencies (Siddiq and Uebersax, 2012). Latin America and sub-Saharan Africa produce the largest volume of common beans, representing more than 60% of the world production (Beebe et al., 2013). Common bean is cultivated throughout eastern and southern Africa mainly by smallholder farmers who depend on beans for their livelihoods and household 1 incomes. Major producers include Kenya, Uganda, Tanzania, and Rwanda (Beebe et al., 2013; Beebe et al., 2014). A number of abiotic and biotic stresses limit crop productivity, and drought stress is the most complex and devastating abiotic stress worldwide (Pennisi, 2008). Drought events have become recurrent due to changes in global climatic conditions. These events have threatened world agriculture and global food security (Borém et al., 2012; Cavatte et al., 2012). Drought stress affects over 60% of common bean production areas, and makes it the most important yield limiting abiotic stress of common bean (Beebe et al., 2012). Drought is particularly endemic in northeastern Brazil, the north-central highlands of Mexico, and eastern Africa (Broughton et al., 2003; Beebe et al., 2012; Beebe et al., 2014). In addition, bean production in these regions continues to expand into marginal areas with greater risk of drought stress and soil nutrient deficiencies (Beebe et al., 2014). In the semi-arid areas of western US, supplementary irrigation is required for common bean production without significant yield losses (Singh, 2007). In sub- Saharan Africa, drought events are widespread, and affect over one-third of bean production areas annually (Beebe et al., 2011). This is exacerbated by the lack of irrigation infrastructure to supplement natural rainfall (Katungi et al., 2011; Beebe et al., 2014). Furthermore, current and future predictions for climate change indicate reduced amount of rainfall in major common bean production regions worldwide (Battisti and Naylor, 2009; Lobell and Gourdji, 2012). The development of high yielding, and drought tolerant common bean cultivars has become critical for improving performance under stressed and non-stressed environments (Miklas et al., 2006; Beebe et al., 2014). Recent genomic studies support the Mesoamerican origin of common bean (Bellucci et al., 2014) with two distinct gene pools, the Middle American and Andean gene pools following 2 domestication (Kwak and Gepts, 2009; Brinez et al., 2012; Schmutz et al., 2014). The two gene pools have been further categorized into races based on plant morphology, adaption range, and agronomic traits. The Middle American gene pool contain races Durango, Jalisco, Mesoamerica and Guatemala (Beebe et al., 2000). The Andean gene pool contain races Peru, Nueva Granada and Chile (Díaz and Blair, 2006; Blair et al., 2007). Race Durango from the semiarid central and northern highlands of Mexico has been identified with the highest levels of drought tolerance, followed by races Mesoamerica and Jalisco (White and Singh, 1991; Singh, 2007). Introgression between Mesoamerican (type II) and Durango races (type III) have been used as source to improve and develop drought tolerant cultivars for low land tropical environments with local adapted genetic backgrounds (Frahm et al., 2004; Beebe et al., 2008; Mukeshimana et al., 2014). Most drought tolerant cultivars from these two races have shown high photosynthate remobilization potential to grain under terminal drought (Beebe et al., 2008). Although races Durango and Mesoamerica have been a reliable source of drought tolerance, transfer of drought tolerance and other favorable genes into the large seeded Andean gene pool that is widely cultivated in Africa and South America is limited (Singh and Gutiérrez, 1984; Gepts and Bliss, 1985; Beebe et al., 2013). This is due to cross incompatibility and linkage drag between the gene pools that results in low rates of valuable recombinant progenies (Beebe et al., 2013). Additional genetic diversity for drought tolerance needs to be identified to improve the larger seeded Andean beans that are highly susceptible to drought stress. Plant response to drought stress is physiologically complex and involves a wide range of modifications, from rapid responses to long term adjustments (Galmes et al., 2013). These responses depending on timing, duration and severity of drought stress have been broadly classified as drought escape and drought resistance (Levitt, 1972; Beebe et al., 2013; Blum, 3 2014). Drought resistance is further subdivied into two components namely, dehydration avoidance and dehydration tolerance (Blum, 2014). Recent studies have reported drought escape strategies in common bean and these involved greater assimilate remobilization potential of photosynthate to pod filling. Major traits contributing to better adaption to terminal drought stress include high pod harvest index, high pod partitioning index, lower proportion of pod wall biomass and pod load (Assefa et al., 2013; Beebe et al., 2013; Mukeshimana et al., 2014). In addition, differences in photosynthetic remobilization capacity among genotypes were used to improve drought tolerance in small red and black seeded classes by selecting for improved seed filling and seed quality under terminal drought (Beebe et al., 2008). Similarly, dehydration avoidance maximizes water uptake through increased rooting depth and hydraulic conductance from the deeper soil layers while balancing water loss by reducing leaf conductance, and evaporative surface (Beebe et al., 2013). However the root to shoot biomass ratio is important to define the functional balance between water uptake by the root and photosynthesis by the shoot. In rice, alteration of root architecture using the Deeper Rooting 1 (DRO1) genes has significantly improved yield performance under drought conditions (Uga et al., 2013). Positive yield advantages associated with rooting depth under drought conditions have been reported in grain legumes such as common bean, soybean, and cowpea (Devi et al., 2010; Beebe et al., 2013; Belko et al., 2013). Furthermore, another widely used mechanism by plants to protect cell integrity under drought stress is dehydration tolerance that involves accumulation of molecules with detoxifying capabilities or antioxidants. These molecules include superoxide dismutase, glutathione reductase and ascorbate peroxidase that detoxify and limits reactive oxygen species (ROS) that damage sensitive organelles like chloroplasts, mitochondria and peroxisomes during drought 4 stress (Farooq et al., 2009). Other improtant molecules include late embryonic abundant (LEA) dehydrin proteins with hydrophilic properties that protects DNA, stabilizes cytoskeletal filament, and act as molecular chaperone and anti aggregants, enhancing structural and stability to proteins (Moore et al., 2009). Another strategy involves compartmentalization of ions and compatible solutes such as soluble sugars, sugar alcohols, amino acid proline, glycinebetaine, organic acids, and potassium and calcium in the cytoplasm to maintain water potential equilibrium (Taiz and Zeiger, 2010). Osmotic adjustment in the root meristems enables plants to extract more of the tightly held water in order to maintain root turgor and root growth (Taiz and Zeiger, 2010; Blum, 2011) . These responses tend to overlap under drought stress; however, the physiological and molecular basis of these interacations is not well understood. In order to develop common bean cultivars with improved performance under drought stress, knowledge of physiological mechanisms and genetic control of the traits contributing drought tolerance at the different stages of plant growth is critical. Previous studies identified traits associated with drought tolerance in common bean. These include deep and balanced root systems, phenology traits, biomass accumulation, accelerated partitioning of photosynthate towards seed yield components under drought stress (Acosta-Gallegos and Adams, 1991; Schneider et al., 1997; Ramirez-Vallejo and Kelly, 1998; Beebe et al., 2008). Selection for high yield potential and effective drought resistance mechanisms in the same genotype in the target environment has been the most practical way to achieve yield stability under drought stress while also maintaining high yields in higher rainfall seasons (Schneider et al., 1997; Beebe et al., 2008; Mukeshimana et al., 2014). However, this selection approach has been difficult to implement in highly unpredictable target environments, and in situations where interactions between heat and drought are difficult to separate (Boyer et al., 2013). 5 The sequencing of the common bean reference genome has led to recent advances in development of single nucleotide polymorphisms (SNP) marker technologies and genotyping arrays that have enhanced the possibility of generating dense genome-wide markers. This creates new opportunities for dissecting quantitative traits into their single genetic determinants through quantitative trait loci (QTL) and association mapping so that individual locus can be targeted for marker-assisted selection (MAS) with repeatability (Mukeshimana et al., 2014; Schmutz et al., 2014) . Previous studies have identified QTL for drought tolerance traits in common bean. QTL for rooting patterns and root architectural traits have been reported in the DOR364 x BAT477 inter gene pool population (Asfaw et al., 2012; Blair et al., 2012) . In addition, a number of QTL analysis have shown the co-localization of phenology, yield and yield components, and root traits on the same chromosome regardless of mapping population and marker system used (Cichy et al., 2009; Asfaw et al., 2012; Mukeshimana et al., 2014). However, most QTL analysis studies for drought tolerance in beans have been conducted within the Middle American gene pool and inter-gene pool crosses (Schneider et al., 1997; Asfaw et al., 2012; Mukeshimana et al., 2014). In order to explore additional diversity for drought tolerance, QTL analyses for drought tolerance is critical in the Andean gene pool to identify the genetic potential and map genomic regions for populations developed from crosses within the Andean gene pool. This will provide an opportunity to analyze the effect of genetic backgrounds on QTL expression and accelerate identification of alternative sources of drought tolerance genes within the Andean gene pool. The availability of the SNP markers has enabled high throughput genotyping of populations, generating high-density maps, and precise identification of QTL that will enhance understanding of underlying genetic basis and physiological mechanism. Developing a strategy to improve 6 large seeded Andean types widely grown in East Africa and South America is critical (Bitocchi et al., 2013; Mukeshimana et al., 2014). Genome wide association studies (GWAS) have become another powerful tool to identify marker allele association with quantitative traits and for gene discovery. GWAS takes advantage of early recombination events in a diverse population to identify the genetic loci underlying traits with relatively high resolution, thus reducing research time with greater allele numbers (Huang and Han, 2014). Recent GWAS studies in common beans have identified genomic regions associated with agronomic traits, phenology traits, yield traits and cooking time (Cichy et al., 2015; Kamfwa et al., 2015; Hoyos-Villegas et al., 2017). With the available and ever increasing number of SNP markers the easy of identifying genomic regions associated with important traits with high resolution is enhanced in common bean. Additional traits associated with critical physiological processes such as photosynthesis needs to be investigated in common bean. Drought stress restricts CO2 assimilation stimulating reduction in the photochemical Chl fluorescence quenching, photosystem II quantum yield in common bean (Dias and Brüggemann, 2010). The use of photosynthetic traits associated with drought sensitive components of photosynthetic process indicators of stress sensitivity among genotypes will improve the efficiencies of breeding for abiotic stress tolerance. The role of photosynthetic traits that relate to photosystem II quantum yield (ΦII), non- photochemical excitation quenching (NPQ), total chlorophyll content, stomatal conductance, transpiration rate, leaf temperature, and leaf water potential needs to be investigated and better understood in common bean (Lizana et al., 2006; Dias and Brüggemann, 2010). The availability of versatile and robust next generation phenotyping technology such as the MultispeQ makes it possible to measure photosynthetic traits (Kuhlgert et al., 2016). Previous studies reported significant natural variation in photosynthetic 7 rates per unit leaf area as well as tolerance to drought (Wentworth et al., 2006). The MultispeQ can be deployed under a range of environmental conditions (temperature, humidity, CO2 levels, intensity and quality, time, and location), which could improve phenotype-assisted breeding approach to transfer drought resistance traits into productive lines. Additionally, the importance of photosynthetic electron transport as a drought tolerance trait for common bean will be critical to breed for yield under drought conditions (Asfaw et al., 2012). Therefore, breeding for drought tolerance in common bean will require identifying critical plant tolerance responses, understanding of physiological mechanisms, and genetic control of traits contributing to critical plant developmental stages in the same environment. This study seeks to determine the genetic architecture of drought adaptive alleles in the Andean diversity panel using genome wide association study, to identify QTL associated with drought tolerance in Portillo x Red Hawk recombinant inbred line mapping population, and determine the use of photosynthetic traits as drought sensitivity indicator of adaptive physiological response under drought conditions in common bean. 8 CHAPTER 2: DETERMINATION OF THE GENETIC ARCHITECTURE FOR DROUGHT ADAPTIVE ALLELES IN THE ANDEAN DIVERSITY PANEL USING GENOME WIDE ASSOCIATION STUDY Abstract A genome-wide association study (GWAS) was conducted to determine the genetic architecture of yield components and photosynthetic traits in 256 genotypes from the common bean (Phaseolus vulgaris L.) of Andean gene pool grown under drought conditions. The panel was evaluated under drought stress and non-stress conditions at two locations for two seasons in Uganda and genotyped with 5398 single nucleotide polymorphism (SNP) markers on the BARCBean6K_3 BeadChip (6K Beadchip) and 265,704 (265K) SNPs generated using genotyping by sequencing (GBS). We identified 44 significant marker-trait association signals for agronomic and photosynthetic traits using best linear unbiased predictors, 24 for agronomic traits, and 20 for photosynthetic traits under drought stress conditions with the 6K Beadchip. In addition, 56 association signals were detected for agronomic with 265K SNP data. Colocalized GWAS signals were detected for pod weight per plant and harvest index on chromosome Pv02, number of pods per plant, pod weight per plant, and seed yield per plant on Pv06, and for pod weight per plant, number of seeds per plant, and 100 seed weight on Pv11, respectively. Colocalized signals for photosynthetic traits were also identified for Phi2 and PhiNO on Pv03, for PhiNPQ and linear electron flow (LEF) on Pv04, and for PhiNPQ, LEF and Phi2 drought stress on Pv11 under drought stress conditions. Positional candidate genes, including Phvul.006G117500 on Pv06, associated with pod weight per plant, and seed yield per plant and Phvul.011G210000 on Pv11, associated with LEF, Phi2, and PhiNPQ, were identified. SNP ss715640054 was located 8.7kb downstream of Phvul.006G117500, annotated as Plant invertase/pectin methylesterase inhibitor (INH/PMEI) superfamily protein, recognized as a 9 positive regulator of carbon partitioning and sink strength during seed development in beans. SNP marker ss715640406 was located 28.3kb upstream of Phvul.011G210000, annotated as DNAJ heat shock N-terminal domain-containing protein, which acts as a molecular chaperone, protecting plants cells against drought and heat effects. This study provides insights into the genetic control of drought stress responses in Andean beans and significant genomic regions were identified that could be useful for marker assisted breeding to improve drought tolerance in the large seeded beans. 10 Introduction Common bean (Phaseolus vulgaris. L) is the most important grain legume consumed globally (Bellucci et al., 2014). Common bean is considered a food and nutritional security crop, and provides a cheap source of proteins, energy, and micro nutrients such as iron and zinc, vitamins, and dietary fiber for millions of people in sub Saharan Africa and Latin America (Broughton et al., 2003; Akibode and Maredia, 2011). However, common bean production is constrained by a number of abiotic, and biotic stresses, and soil nutrient deficiencies (Beebe et al., 2012). Drought stress is the most important yield limiting abiotic stress of common bean, and affects over 60% of common bean production areas worldwide (Beebe et al., 2013). Drought causes significant reduction in biomass, pod set, and seed yield of common bean (Nielsen and Nelson, 1998; Ramirez-Vallejo and Kelly, 1998). These effects have become very pronounced in major bean production areas because of continued expansion of production into marginal areas with greater risk of drought stress (Beebe et al., 2011). In sub-Saharan Africa, drought events are widespread, and affect over one-third of bean production areas annually (Beebe et al., 2014). Developing high yielding, and drought tolerant common bean cultivars is the most sustainable approach to improve productivity under stressed and non-stressed environments particularly for smallholder farmers in developing countries (Miklas et al., 2006; Beebe et al., 2012). Plant responses to drought stress are physiologically complex and involves a wide range of adaptive mechanisms, from rapid responses to long term adjustments (Galmes et al., 2013). These adaptive mechanisms depend on timing, duration and severity of drought stress and have been classified as escape, avoidance, tolerance and recovery strategies (Levitt, 1980; Blum, 2011; Beebe et al., 2013). In common bean, adaptive strategies that enhance drought tolerance have been reported and these include deeper rooting systems, efficient use of water for 11 photosynthesis and biomass accumulation, and efficient remobilization of photosynthate to seed under drought stress (White et al., 1994; Rao, 2001; Beebe et al., 2008; Beebe et al., 2014). Recent studies reported that greater assimilate remobilization potential of photosynthate to seed as an important strategy to escape terminal drought (Beebe et al., 2014). Major determinant traits associated with this strategy are yield related and include high seed yield, pod harvest index, pod load, and lower proportion of pod wall biomass (Beebe et al., 2013; Mukeshimana et al., 2014). Breeding and selection for greater remobilization potential of photosynthate to seed has resulted in the identification of superior landraces from race Durango and development of drought tolerant breeding lines such as SEA5, and SEA15 and a number of advanced breeding lines in small red and black bean market classes (Terán and Singh, 2002; Beebe et al., 2008). In common bean, race Durango from the semiarid central and northern highlands of Mexico has been identified with the highest levels of drought tolerance, followed by races Mesoamerica and Jalisco (White and Izquierdo, 1991; Singh, 2007). However, moving drought tolerance and favorable genes from races Durango and Mesoamerica into the large seeded Andean gene pool has been limited (Singh and Gutiérrez, 1984; Gepts and Bliss, 1985; Beebe et al., 2011). This is due to cross incompatibility and linkage drag between the gene pools that results in low rates of recovery of valuable recombinant progeny (Beebe et al., 2013). Additionally, in east and southern Africa, Europe and some parts of Latin America, over 70% of common beans cultivated are large seeded types of Andean origin, and are generally susceptible to many abiotic and biotic stresses (Wortmann, 1998; Beebe et al., 2013). Genetic improvement of Andean beans for yield, abiotic and biotic stress has lagged behind the Mesoamerican beans because of narrow genetic diversity and a lack of breeding efforts (Gepts and Bliss, 1988; Bellucci et al., 2014; Cichy et al., 2015). The assemblage of global Andean diversity panel 12 (ADP) that consists of common bean breeding lines and landraces should enhance the exploration of genetic diversity and identification of unique sources of valuable genes for different traits in Andean beans (Cichy et al., 2015). Previous studies identified traits associated with drought tolerance in common beans of diverse genetic backgrounds (Acosta-Gallegos and Adams, 1991; Schneider et al., 1997; Ramirez- Vallejo and Kelly, 1998; Beebe et al., 2008). Additional traits associated with critical physiological processes such as photosynthesis need to be investigated in common bean. Previous studies reported significant natural variation in photosynthetic rates per unit leaf area as well as tolerance to drought (Wentworth et al., 2006). With the invention of the MultispeQ, a versatile and robust high throughput, hand-held, photosynthetic phenotyping tool, assessing photosynthetic traits under multiple environmental conditions will improve phenotype-assisted breeding by quickly identifying drought resistance traits (Kuhlgert et al., 2016). Additionally, understanding photosynthetic dynamics that relates to adaptation in multiple environments should assist decision making to improve common bean yields under drought conditions. Furthermore, the sequencing of the common bean reference genome has led to recent advances in development of SNP marker technologies and genotyping arrays that have significantly enhanced the possibility of generating dense genome-wide markers (Schmutz et al., 2014; Song et al., 2015). The availability of dense genome wide markers has enabled the identification of marker alleles that contribute to causative variation in quantitative traits using genome wide association studies. GWAS analysis takes advantage of early recombination events in a diverse population to identify the genetic loci underlying traits with relatively high resolution, thus reducing research time with greater allele numbers (Huang and Han, 2014). In common bean, GWAS has been used to 13 identity genomic regions associated with agronomic traits, phenology traits, yield traits, and cooking time in Andean and Mesoamerican gene pools (Cichy et al., 2015; Kamfwa et al., 2015; Hoyos-Villegas et al., 2017). In order to explore the genetic diversity and understand the genetic control of agronomic and photosynthetic traits contributing to drought tolerance in common bean of Andean gene pool, a GWAS was initiated to determine the genetic architecture of terminal drought tolerance in the Andean diversity panel grown under drought stress at two locations in Uganda. Plant materials Materials and Methods The plant materials used in this study consisted of 256 bean genotypes from the ADP. The panel is comprised of a diverse collection of varieties, and landraces with varied seed types and market classes mainly from Africa, South America, Central America, and North America (Cichy et al., 2015). The panel is a global representative sample of the genetic diversity present in the Andean gene pool. Field phenotyping The ADP panel was evaluated for drought stress in the field at Mubuku Irrigation Experimental Station, Kasese and at the National Crop Resources Research Institute (NaCRRI), Namulonge in 2016 and 2017 growing seasons in Uganda. The two seasons hereafter referred to as Kasese_2016, Kasese _2017 and Namulonge_2016, Namulonge_2017, respectively. In 2016, 247 genotypes were evaluated; and the number was increased in 2017 to 256 after additional ADP lines were provided by CIAT Uganda. The Mubuku Irrigation Experimental Station is located in Kasese Western Uganda at 0o16‘10‖ N, 30o6‘9‖ E at an elevation of 1330 m above sea 14 level and receives 1000 mm of rainfall annually that is irregularly distributed across seasons (Figure.1). The soils were characterized as sandy loam with a pH of 5.7. The National Crop Resources Research Institute (NaCRRI), Namulonge is located in Central Uganda at 0o 31‘30‖ N, 32o36‘54‖ E at an elevation of 1160 m above sea level. The average temperatures are 21.8 oC, ranging from 15.8 oC to 27.9 oC. Namulonge receives 1300 mm of rainfall annually (Figure.1), and soils are characterized as oxisols with a pH of 5.8. The results of soil nutrient analysis indicated an average low available Phosphorus and Nitrogen where there were below the 15ppm and 0.2% respectively in both sites and NPK was applied at a rate of 50 kg/ha to enhance soil nutrient levels. The ADP lines were evaluated side by side under non-stress (NS) and drought stress (DS) conditions. In 2016, the experiments were planted in one row plots of 2 m long and 0.5 m between rows. In 2017, experiments were two rows 2 m long and 0.5 m between rows. For the two seasons, the experiments were planted in 16x16-lattice design with two replications. Planting was done during the mid-rainy to end of season to target terminal drought stress at the reproductive stage of plant growth. In Kasese, the first experiment was planted on 28 Nov 2016 and harvested on 26 Feb 2017 and the second field experiment on 15 May 2017 and harvested on 11 Sept 2017. Drip irrigation was used to irrigate both DS and NS to ensure good germination and plant establishment until flowering time. Irrigation was discontinued at flowering following emergence in the DS plots while supplemental irrigation was continued in the NS plots twice a week up to harvest maturity. The first experiment in Namulonge was planted on 19 Oct 2016 and harvested on 16 Feb 2017 and second season experiment was planted on 18 June 2017 and harvested on 30 Sept 2017. Overhead sprinklers were used to irrigate both DS and NS to ensure good germination and plant establishment until flowering time. Supplemental irrigation was continued in the NS plots twice a week up to harvest maturity while irrigation was discontinued 15 in the DS plots following emergence. In both locations and seasons, recommended agronomic procedures were observed. Data was collected for daily weather conditions including daily rainfall (mm) and temperature (0C) during the growing seasons. Phenotypic data was collected on days to 50% flowering (DF), days to maturity (DM). In order to determine shoot biomass, yield and yield component traits, six plants were sampled from every plot for every genotype and sun dried. These were then used for measuring shoot biomass (BM), number of pods (PN), pod weight (PW), seed weight per plant (SY), Number of seeds per plant (SN), and 100 seed weight (100SWT). In both seasons, SY and 100SWT were calculated after adjusting for seed moisture to 16%. Harvest index (HI), pod harvest index (PHI), and pod partitioning index (PPI) were also computed. The severity of drought stress in each environment for yield was quantified using drought intensity index (DII) as 1– (Xds/Xns) where Xds and Xns are the mean value for yield under drought stress and non-stress, respectively. Geometric mean (GM) of seed yield as (NS x DS)1/2 where DS and NS are seed yield for drought stress and no stress treatments, respectively (Fischer and Maurer, 1978). In addition, photosynthetic traits included linear electron flow (LEF), quantum yield of photosystem II (Phi2), ratio of incoming light that is lost via non- regulated processes (PhiNO), and ratio of incoming light that goes towards non-photochemical quenching (PhiNPQ) measured during pod filling stage using the MultispeQ (Kanazawa and Kramer, 2002; Kuhlgert et al., 2016). Data for photosynthetic traits was taken at 8-9 weeks after planting and used for analysis. Genotyping The ADP panel consisting of 396 accessions was genotyped using Illumina (Illumina Inc., San Diego, CA) BARCBean6K_3 Beadchip (6K Beadchip) with 5398 SNP markers distributed across the 11 the chromosomes of common bean (Song et al., 2015). The second set of marker 16 data consisted of 265704 (265K) SNPs generated by Oladzad and McClean (unpublished) for 325 ADP accessions using GBS. Data Analysis Statistical analysis for agronomic traits was performed using best linear unbiased predictors (BLUPs) which minimizes environmental, and seasonal effects, were used for GWAS analysis. The mixed model in R statistical package ‗lme4‘(Bates et al., 2014) was used to determine and extract the random effects (BLUPs) of the phenotypic traits. BLUPs for traits were calculated based on following statistical model; Yijk = µ +αi +βj+γk+ αβij+ βγjk+ εijk Where Yijk is the response variable, µ is the mean, α is the effect of the ith genotype, β represents the effects of the jth treatment, γ corresponds to the kth season and εijk is the random error. Pearson correlation among variables was analyzed using the PROC CORR command using SAS 9.4 (Institute, 2011). Population Structure and Marker-Trait Association Analysis A total of 5326 SNPs remained after removing low quality and monomorphic SNPs on the 6K Beadchip. These were then filtered for minor allele frequency (MAF≥ 0.05) leaving 4393 SNPs. The 265K markers were filtered for minor allele frequency (MAF≥ 0.05) leaving 228K markers. The two data sets were later used for kinship, population structure and association analysis. Population stratification can confound GWAS results causing spurious association. Population structure in the association panel was estimated using the principal component analysis (PCA) method in TASSEL (Bradbury et al., 2007). The number of principal components used was based on the proportion of variation accounted for by the principal components. This was 17 determined using the Genome Association and Prediction Integrated Tool (GAPIT) (Lipka et al., 2012), and the suggested number of principal components was equal to the number of subpopulations (K). Relatedness within the panel was determined by calculating the kinship matrix using identity by descent method in TASSEL (Bradbury et al., 2007). Mixed Linear Model (MLM) (Zhang et al., 2010) was used to determine marker trait associations implemented in TASSEL (Bradbury et al., 2007) explained as; Y=Xα +Pβ+Kμ+ε Where Y is the vector of observed phenotype; X is the fixed effects of SNPs; P is the fixed effects of population structure; K is the random effects of kinship; ε is the random error. Significance threshold for p-values resulting from association analysis was determined using the false discovery rate (FDR) criteria (Benjamini and Hochberg, 1995). SNPs were declared significant after applying an FDR adjusted p value of less than 0.05 as a threshold. Candidate genes associated with significant SNPs were identified using the Jbrowse on Phytozome v12 (Goodstein et al., 2011) to browse the common bean genome v2 (Schmutz et al., 2014). TASSEL (Bradbury et al., 2007) was used to determine linkage disequilibrium (LD) among or between pairs of markers in the genomic regions associated with significant SNPs. A gene was considered a candidate gene if it was within LD block of genomic region associated with the significant SNP and the protein coded by the candidate gene plays a role in regulating drought stress response and seed yield under drought stress. Heritability Heritability estimate for traits was determined using mixed models in GAPIT R package (Lipka et al., 2012), and variation due to the genetic component was calculated by considering genotype 18 as a random effect in the model. In addition, the genetic variance for a diversity panel was calculated as the ratio of phenotypic variance associated with family relatedness (kinship) among accessions in the collection. Heritability was estimated as; Where σa 2 is the additive genetic variance associated with variance component of the vectors u and e from Where G= σa 2K with σa 2 as the additive genetic variance and K as the kinship matrix, homogenous variance is assumed as the residual effect represented as R= σe 2I, where σe 2 is the residual variance (Lipka et al., 2012). Rainfall Results The average temperature and daily rainfall recorded during the 2016 and 2017 growing season at Kasese and Namulonge are reported in Figure1.1. In Kasese, a total of 144 mm was recorded during the 2016 (October-February) and 180 mm of precipitation recorded for 2017 (May- September) growing season. In both growing seasons, the rainfall patterns were inconsistent and most rains were recorded during the vegetative stage of growth (122 mm and 90 mm) in 2016 and 2017 respectively, which is below the minimum 400 mm required for normal growth of common bean (Wortmann et al., 1998). In addition, only 22 mm and 16 mm of precipitation were recorded during the reproductive and pod filling stages (before 90 days after planting), 19 subjecting the field experiments to severe terminal drought stress (Figure 2.1 a&b). In Namulonge higher rainfall was recorded and relatively uniformly distributed during the growing season of 2017 than in 2016, a total of 160 and 216 mm of precipitation were recorded during 2016 and 2017 growing season respectively (Figure 2.1c &d). In addition, greater precipitation was recorded during the vegetative stage (129 mm and 101 mm) and only (28 mm and 52 mm) was recorded during the reproductive and pod filling stage (before 90 DAP). Field Experiments The overall drought severity observed in this study was moderate based on the DII for SY (0.40) measured across the two locations for the two years. More severe effect of drought on seed yield was observed in Namulonge_2016 where DII for SY was 0.58 compared to 0.30 observed in Kasese_2016. Highly significant differences (P<0.001) were observed among 256 genotypes for agronomic traits measured under NS and DS conditions in Kasese and Namulonge for the two seasons of 2016 and 2017 (Table 2.1 a&b). Higher mean performance was observed in 2017 at both sites compared to 2016 (Table 2.1 a&b). This was due to a severe bean fly infestation at the onset of germination that affected plant establishment in 2016 at both sites. Significant differences (P<0.01) were observed among 256 genotypes for photosynthetic traits such as LEF, SPAD, Phi2, PhiNO, and PhiNPQ (Table 2.2). In addition, highly significant and positive correlations were observed among yield component traits such as, PN, PW, SY, SN, and among partitioning traits such as HI, PHI, and PPI. Significant positive correlations were observed between BM, HI, PHI, PPI and yield component traits under DS conditions in Kasese_2017. Highly significant positive correlations were observed among photosynthetic traits such as Phi2 and PhiNO. However, these traits were negatively correlated with PhiNPQ. Additionally, most 20 yield component traits were not correlated with photosynthetic traits with exception of relative chlorophyll content under DS conditions in Kasese_2017 (Table 2.3). Population Structure The results of the principal component analysis showed a subpopulation structure among genotypes in the panel used in this study. The principal components were determined using the 228K markers. The first four principal components accounted for 41.2, 5.2, 4.6, and 3.5% variation observed among genotypes in the panel respectively. A plot of PC1 against PC2 indicated the presence of three clusters (Figure.2). The smallest cluster consisted of 11 individuals; seven were landraces from Africa, three from South America, and one from North America. The second cluster comprised of 23 genotypes, of which seventeen were landraces from Africa, seven from South America, and three from North America. The third cluster consisted of 222 genotypes mainly landraces, cultivars, and elite lines of diverse origin. The first three PCs accounted for 51 % of variation in the genotypes and were considered in MLM for association tests in TASSEL (Bradbury et al., 2007). Marker-Trait Associations Significant marker trait associations were identified for agronomic and photosynthetic traits using both the 6K Beadchip and 265K GBS marker data. The agronomic traits measured in this study included NP, PW, SN, SY, and HI, and the photosynthetic traits included LFF, Phi2, PhiNPQ, and PhiNO measured under drought stress conditions (Table1 4a & b). Number of pods per plant (NP) Several significant marker trait associations were detected for NP under drought stress conditions using both the 6K Beadchip and the 265K GBS marker data. Three SNPs were significantly 21 associated with the NP under drought stress conditions in Kasese with the 6K Beadchip. SNP (ss715641140; P=7.5x10-06) at 32.72Mb on Pv10 with minor allele frequency (MAF) of 0.12 was the most significant and explained about 9% of the variation. Two significant SNPs were detected on Pv06, and marker (ss715639365; P=4.2x10-05) at 31.37Mb with MAF of 0.11 was the most significant and explained 8% of the variation (Table 2.4a). In addition, 14 significant trait associations for NP were detected using the 265K SNP data in combined data under DS conditions on Pv02, Pv03, Pv06, Pv07 and Pv08 (Figure 2.3, Table 2.4b). Marker (S08_55438500; P=7.3x10-09) at 55.4Mb with MAF of 0.12 was the most significant and explained 16.6% of the variation, while SNP (S07_11078838; P=1.1x10-08) at 11.07Mb with MAF of 0.07 was the most significant on Pv07 and explained 16.2 % variation in NP under DS conditions. No significant associations for NP were identified in Namulonge under DS conditions (Figure 2.3, Table 2.4b). Pod weight per plant (PW) Nine SNPs were significantly associated with PW under DS conditions in Kasese and Namulonge using combined data analysis with the 6K Beadchip. Five SNPs all on Pv06 were associated with PW under drought stress in Kasese. The most significant SNP (ss715640054; P=4.2x10-06) at 23.23Mb with MAF of 0.12 accounted for 10% of the variation in PW (Table 2.4a). In Namulonge, four SNPs were significantly associated with PW. These included two on Pv02 and marker (ss715641175; P=4.1x10-06) located at 24.35 Mb with MAF of 0.09 was the most significant and explained 9% of the variation. SNP (ss715646672; P=2.0x10-05) at 6.12Mb on Pv11with MAF of 0.10 was significant and accounted for 8% of the variation (Table 2.4a). In addition, five significant markers were detected on Pv06 for PW with the 265K SNP data using combined data under DS. The most significant SNP (S06_23021418; P=8.6x10-08) at 23.02Mb 22 with MAF of 0.08 accounted for 14.7% of the variation in PW under DS conditions. (Figure 2.3, Table 2.4b). Number of seeds per plant (SN) Four SNPs were significantly associated with SN under DS conditions in Namulonge in combined data using the 6K Beadchip (Table 2.4a). These included two markers on Pv01 with marker ss715646300 located at 48.06 Mb with MAF of 0.32 being the significant and explained 12% of the variation. Two significant markers were identified on Pv11 with SNP (ss715649043; P=1.12x10-06) located at 2.01 Mb with MAF of 0.17 was most significant and accounted for 12% of the variation in number of seeds per plant. No significant associations were detected for NS under drought stress in Kasese (Table 2.4a) Seed yield per plant (SY) Significant SNPs associated with SY were identified in Kasese under DS conditions in the combined data using the 6K Beadchip and under NS and DS using the 265K GBS data. Four markers were significant on Pv06, and two on Pv04. The most significant SNP (ss715640054; P=2.0x10-05) on Pv06 with MAF of 0.12 was located at 23.23 Mb and explained 9% of the variation in SY. In addition, SNP (ss715640054) on Pv06 was the most significant SNP for PW. Interestingly all the significant SNPs associated with PW on Pv06 were also significant for SY. Two SNPs were identified on Pv04 and the most significant SNP (ss715645803; P=6.0x10-05) with MAF of 0.15 located at 4.45Mb explained about 8% of the variation (Table 2.4a). In addition, 14 markers were significantly associated with SY under both NS and DS conditions using the 265K SNP data. Four markers all on Pv10 were significantly associated with SY under NS conditions using combined data. SNP (S10_27780836; P=3.5x10-09) with a physical position of 27.7Mb and MAF of 0.08 was the most significant and explained 17.1% variation in SY under 23 NS conditions. An additional 10 markers were significantly associated with SY under DS conditions in Kasese on Pv02, Pv03, Pv05, Pv07, Pv08, and Pv10 (Figure 2.4, Table 2.4b). SNP (S03_51268099; P=3.07x10-09) at 51.2Mb with MAF of 0.06 on Pv03 was the most significant and explained 16.5% variation in SY under DS conditions (Figure 2.4, Table 2.4b). Harvest index (HI) Significant SNPs associated with HI were identified in Kasese under DS conditions and in Namulonge NS conditions using the 6K Beadchip and 265K GBS marker data set. Two significant SNPs all on Pv02 were detected for HI under DS conditions in Kasese using the combined data analysis. The most significant SNP (ss715648530; P= 9.1x10-05) with MAF of 0.14 and physical position of 24.23 Mb accounted for 6% of the variation in HI (Table 2.4a). Four significant SNPs on Pv09 were detected for HI under NS conditions in combined data using the 265K SNP data. The most significant SNP (S09_28676522; P= 5.2x1007) at 28.6Mb with MAF of 0.06 accounted for 11.1% of the variation in HI. In addition, two significant SNPs were identified for HI under NS conditions in Namulonge on Pv02 and Pv04 respectively. Marker (S02_30630638; P=3.1x10-07) at 30.6Mb with MAF of 0.10 was the most significant and explained 13% variation in HI while SNP (S04_31117609; P=4.6x10-07) at 31.1Mb with MAF of 0.11 on Pv04 explained 12% variation in HI under NS conditions in Namulonge (Figure 2.4, Table 2.4b). PhiNPQ A total of four SNPs were significantly associated with the photosynthetic trait PhiNPQ in Kasese under DS conditions in the combined data. Marker (ss715650748; P=1.1x10-06) located at 48.77 Mb on Pv11 with MAF of 0.11 was the most significant and explained of the variation 24 in PhiNPQ. Other significant markers included (SNP ss715648421; P=3.8x10-05) located at 0.54 Mb on Pv09 with MAF of 0.17, SNP (ss715648989; P= 5.1x10-05) located at 39.78 Mb on Pv05 with MAF of 0.19 and SNP (ss715645811; P= 2.7x10-05) with a physical position of 44.33 Mb on Pv04 with MAF of 0.10 respectively (Figure 2.6, Table 2.4a). Linear Electron Flow (LEF) Seven significant SNPs for LEF were identified in Kasese_2016 on Pv04 and Pv11 under DS conditions. Three SNPs were significant within a 355.2 kb region on Pv04, and the most significant SNP (ss715645811; P=7.1x10-06) on Pv04 was located 44.40 Mb with MAF of 0.12 and explained 12 % variation in LEF. Four SNPs were significant for LEF on Pv11, and the most significant SNP (ss715650748; P=1.6x10-07) located at 48.77 Mb with MAF of 0.11 explained 14 % of variation in LEF under DS conditions. No significant associations for LEF were identified in Kasese_2017. Additionally, SNP (ss715650748; P=1.60x10-07) on Pv11 was most significant for both LEF and PhiNPQ (Figure 2.6, Table 2.4a). Phi2 Significant SNPs for Phi2 were identified in both Kasese_2016 and Kasese_2017 under DS conditions. In Kasese_2016, three SNPs were significant on Pv11 and the most significant SNP (ss715640406; P=1.35x10-05) was located at 52.21 Mb with MAF of 0.13, and explained 11% variation in Phi2 (Table 2.4a). In addition, SNP, (ss715640406; P=1.35x10-05) was significant for both LEF and Phi2 (Table 2.4a). In Kasese_2017, five SNPs were significant, four on Pv03 and one on Pv05. The most significant SNP (ss715646087; P=9.64x10-06) for Phi2 on Pv03 was located at 52.21 Mb with MAF of 0.13 accounted for 10% variation in Phi2. SNP (ss715649491; P=5.26x10-05) located at 30.43 Mb on Pv05 with MAF of 0.15 explained 7% variation for Phi2 in Kasese_2017 (Figure 2.6 & 7, Table2.4a). 25 PhiNO In Kasese_2017, one SNP was significantly associated with PhiNO on Pv03. Marker (ss715646087; P=8.67x10-05) located at 52.21 Mb with MAF 0.13 was the only significant SNP detected under DS conditions, and explained for 10% variation PhiNO. Interestingly, the same marker was detected for Phi2 in Kasese_2017 (Figure 2.7, Table 2.4a). Allelic Effects of Significant SNPs on Yield Component Traits The allelic effects associated with significant SNPs from the 6K Beadchip for yield component traits namely PW and SY on Pv06 were assessed. Significant marker (ss715646419l; P= 2.2x10- 05) located at 24.9 Mb was significantly associated with PW, and explained 7% variation under DS conditions in Kasese. Interestingly, individuals carrying the marker allele G allele, with MAF of 0.14 had relatively higher PW under DS. Similarly, another SNP (ss715644727; P=4.2x10-05) at 23.3 Mb on Pv06 was significantly associated with PW had the minor C allele (0.16). These two significant SNPs had substantially large positive effects of 44.9 and 44.1g on PW under DS conditions while the major alternate T allele in both SNPs had zero effect. Similarly, SNP (ss715646419; P= 8.0x10-05) had large positive effect of 25.5g on SY and zero effects from the major T allele. In addition, 37 accessions in the panel were carrying the same G and C minor alleles for the two markers that came from diverse geographical and market classes. The red mottled cultivar PR9745-232 (ADP-432) from the Caribbean had the highest PW of 41.1g and SY of 34.7g under drought stress conditions among genotypes with these marker alleles. Interestingly, most genotypes carrying these loci have previously been reported to express drought tolerance under DS conditions. In addition, it is worth noting that the traits associated with these positive allelic effects are yield related implying that breeders have over time have 26 been accumulating favorable loci with positive effects on actual seed yield through indirect selection under DS conditions (Beebe et al., 2008; Assefa et al., 2013). Discussion The objective of this study was to use GWAS analysis to identify genomic regions and advance knowledge on genetic control of yield component traits and photosynthetic traits particularly at pod filling stage of common bean from the Andean gene pool grown under drought stress conditions. This study was conducted to explore the genetic diversity for drought tolerance and enhance genetic improvement for the trait in large seeded beans popularly cultivated in Africa. The severity of terminal drought stress observed in the current study, was moderate based on the DII of 0.4 for SY across locations and years. In addition, high variability in DII was observed across locations and years for SY and ranged from 0.30 to 0.58. The high variability in DII for SY could be attributed to the negative effects of drought stress on highly drought sensitive reproductive stage from flowering formation to seed development, with a cumulative negative effect on overall seed yield, although dependent on the severity and duration of the stress. This is consistent with previous studies that reported significant reduction in seed yield, harvest index, and pod number in common beans grown under drought stress (Schneider et al., 1997; Ramirez- Vallejo and Kelly, 1998; Rosales-Serna et al., 2004). Significant phenotypic variability was observed among the accessions for phenology, yield component, partitioning and photosynthetic traits measured during the pod filling stage with much variability observed under DS conditions (Table 1.2&3). In addition, significant positive correlations were observed among yield component and partitioning traits (Table 1.3), which is consistent with previous studies that reported significant positive correlations among yield component and partitioning traits DS (Asfaw et al., 2012; Mukeshimana et al., 2014). Such correlations imply that the ability to 27 remobilize photosynthate to seed yield under moderate drought conditions could be a useful for indirect selection for seed yield (Porch et al., 2009; Asfaw et al., 2012). This approach, has been used to develop and select superior common bean lines with greater assimilate remobilization potential of photosynthate to seed yield under terminal drought (Beebe et al., 2008; Klaedtke et al., 2012; Assefa et al., 2013). In this study, GM for SY was used to identify individuals that performed well under both DS and NS conditions across sites in the two locations. Accession AC Calmont (ADP-0670), a dark red kidney from Canada had the highest GM for SY (Table 1.5). In this study, significant positive correlations were observed among photosynthetic traits such as LEF, Phi2 and PhiNO and negative correlation was observed with PhiNPQ. However, yield component and partitioning traits were not correlated with photosynthetic traits under DS conditions (Table 3), suggesting that the likelihood of using photosynthetic traits as predictors of yield trait components under DS conditions using field experiments is minimal. However, photosynthetic traits are useful for understanding the mechanistic and physiological aspects of drought stress during the critical pod filling stage. Marker-Trait Associations In this study, 44 significant marker-trait association signals were detected among agronomic and photosynthetic traits using the 6K Beadchip. Twenty-four significant associations for yield component and partitioning traits, such as NP, PW, SY, and HI were detected on Pv02, Pv04, Pv06, Pv09, Pv10, and Pv11. In addition, 20 significant associations were detected for photosynthetic traits on Pv03, Pv04, Pv05, Pv09 and Pv11 for PhiNPQ, LEF, Phi2 and PhiNO. Colocalized signals were detected on Pv03 for Phi2 and PhiNO, on Pv04 for PhiNPQ and LEF, and on Pv11 for PhiNPQ, LEF and Phi2 under DS conditions (Table 2.4a). 28 In addition, a total of 39 significant marker-trait association signals were detected for agronomic traits such as NP, PW, SY, and HI using the 265K GBS marker data (Table1.4b). Colocalized GWAS signals were detected on Pv02 for PW and HI, on Pv06 for PN, PW, and SWT (Table 2.4a) using the 6K Beadchip. This is consistent with the high significant positive correlations observed among the yield component traits, suggesting pleiotropic effects of those genomic regions on yield component traits (Table 2.3). Yield Component Traits Number of pods per plant In the current study, three SNPs were significantly associated with NP on Pv06 and Pv10 under DS conditions in Kasese using the 6K Beadchip, whereas 14 significant association signals were detected for NP on Pv02, Pv03, Pv06, Pv07 and Pv08 using the 265K in combined data under DS conditions (Table 2.4a,b). The heritability for NP was 43.2% under NS and 41.5% under DS conditions respectively. This is consistent with previous studies that reported several QTL for NP across the genome of common bean. QTL for NP was previously mapped to Pv02 (Tar'an et al., 2002; Beattie et al., 2003), Pv03 (Beattie et al., 2003; Mukeshimana et al., 2014), Pv06 (Beattie et al., 2003; González et al., 2016; Diaz et al., 2017), Pv07 (Blair et al., 2006; Kamfwa et al., 2015), Pv08 and Pv10 (Koinange et al., 1996; Blair et al., 2012; Mukeshimana et al., 2014) in diverse mapping populations. However, the lack of physical positions for several markers linked to significant QTL reported in previous studies has limited comparisons of similar QTL in subsequent studies. Interestingly the most significant SNP S08_55438500; P=7.3x10-09 at 55.4 Mb on Pv08 was detected under DS was in LD (r2 > 0.6; D' =0.96) with markers S08_56044860 and S08_56048573 located at 56.0 Mb. These markers are within the same genomic region to SNP ss715648562 (56.0 Mb) associated with NP8.1SC identified in F5:7 SEA5 x CAL96 29 population under NS conditions (Mukeshimana et al., 2014), suggesting that the genomic region associated with NP detected in the two studies is the same regardless of the population and environment. However, other significant signals detected on Pv03 at 52.3 Mb and Pv07 at 11.0 Mb and 26.0 Mb were different from those reported on Pv03 at 2.2 Mb (Mukeshimana et al., 2014) and on Pv07 at 40.0Mb (Kamfwa et al., 2015) (Table 2.4b). Two markers on Pv10 namely ss715641139 and ss715640368 were downstream of significant SNP ss715641140 detected using 6K Beadchip were in strong LD (r2 > 0.8; D' =1), and spanned 198 kb region with four genes. Positional candidate gene Phvul.010G087800 is 2.4kb downstream of marker ss715641140, and is annotated as ―GLUTAREDOXIN-RELATED (GRX)‖. GRXs are small molecular weight oxidoreductase glutaredoxin_related proteins that have antioxidant activity, and are associated with detoxifying reactive oxygen species (ROS) (Rouhier et al., 2006; Rouhier et al., 2008; Li, 2014), and floral organ development especially petal and anther initiation and differentiation in Arabidopsis and rice (Rouhier et al., 2008; Xing and Zachgo, 2008; Hofmann, 2009). Plants produce ROS under DS conditions, which at low concentrations are important signaling molecule mediating early plant response to drought stress (Farooq et al., 2017). However, high concentrations of ROS produced under continued drought conditions could result in oxidative damage and eventually cell death (Rouhier et al., 2006). The high expression of GRX proteins could be associated with the maintenance of cellular equilibrium through scavenging activity on ROS. The second positional candidate genes was within the 198kb region flanked by significant SNP ss715641140 and ss715640368 in LD (r2 > 0.8; D' =1) were three genes Phvul.010G087900, Phvul.010G088000, and Phvul.010G088100 that encode ―AGAMOUS-LIKE (AGL)‖ genes. These genes, namely AGL6, AGL20, and AGL14 are members of the MAD-box transcription factor family. AGL6 and AGL20 are 30 positive regulators of flowering time (Yu et al., 2002; Yoo et al., 2011), and are associated with floral initiation and organ development in soybean and Arabidopsis (Tapia-López et al., 2008; Wang et al., 2009; Wong et al., 2013). The functional annotation of the positional candidate gene reflects the importance of proper floral development in determining pod formation and development, and consequently overall yield under drought conditions. Pod weight per plant Significant association signals for PW were identified on Pv02 in Namulonge and on Pv06 in Kasese, respectively. In Kasese, the significant SNPs associated with PW were detected on Pv06 under DS conditions using both the 6K Beadchip and 265K SNP data. Interestingly, four of the significant markers (Table 2. 4b) identified using the 265K SNP data on Pv06 were in LD (r2 > 0.88; D' >0.96), and spanned, 0.97Mb (23.0-23.9Mb) region. In addition, significant markers ss715640054 (23.2 Mb) and ss715644685 (23.1 Mb) associated with PW using 6K Beadchip were within the same genomic region. The heritability for PW was 46.6% under NS and 42.5% under DS conditions respectively. Previous studies mapped QTL associated with pod related traits on Pv06 (Beattie et al., 2003; Diaz et al., 2017) but the physical position of the markers associated with these QTL were not determined. The most significant SNP (ss715640054; P=4.2x10-06) located at 23.23 Mb on Pv06 with MAF of 0.12 accounted for 10% of the variation in PW (Table 1.4a). In addition, four significant SNPs (ss715640054, ss715644685, ss715645159, and ss715644727) were in strong LD (r2 > 0.85; D' =1), and spanned 399 kb region with 45 genes on Pv06. However, marker ss715640054 is 8.7kb downstream of the positional candidate gene Phvul.006G117500, annotated as ‗PLANT INVERTASE/PECTIN METHYLESTERASE INHIBITOR (INH/PMEI) SUPERFAMILY PROTEIN‘, that has been associated with the regulation of carbon partitioning and sink strength 31 particularly during seed development in plants (Hothorn et al., 2010; Ruan, 2014). Consistent with this finding is a recent transcriptome analysis of the drought tolerant common bean cultivar, BAT 477 that reported an up regulation of INH/PMEI at flowering and grain filling stage (de Faria Müller et al., 2014). Similarly, a recent study in soybean reported the involvement of INH/PMEI complex is determining seed size in soybean. The INH/PMEI complex was significantly up-regulated in a large seeded soybean cultivar during seed filling stage (20-22 days after flowering) than in a small seeded cultivar (Du et al., 2017). These results seems to suggest that the INH/PMEI complex is conserved in legumes and consistently expressed during pod filling and seed development stages of growth. The primary role of INH/PMEI complex appears to be related to the balancing of starch and sucrose metabolism in determining seed size, and subsequently seed weight at pod filling stage and could be targeted for marker assisted breeding to improve seed yield. In addition, four SNPs were significantly associated with PW on Pv02, Pv09, and Pv11 under drought stress conditions in Namulonge. The SNP ss715641175 located at 24.35 Mb on Pv02 was the most significant and explained 9% of the variation. The two significant markers were in complete LD (r2 >1; D' =1), and spanned 338kb region with 18 genes on Pv02. However, within the 5 kb region upstream of ss715641175 were two genes Phvul.002G120100, annotated ―2-OXOGLUTARATE (2OG) and FE(II)- DEPENDENTOXYGENASE SUPERFAMILY PROTEIN‖, and Phvul.002G120200, annotated as ―KINESIN-LIKE CALMODULIN-BINDING PROTEIN (ZWICHEL)‖. The function of the two genes under drought stress remains unclear. Number of seeds per plant (NSP) Four SNPs were significantly associated with NSP on Pv01 and Pv11 under drought stress conditions in Namulonge. The heritability for NSP was 54.8% under NS and 43.5% under DS 32 conditions, respectively. SNP (ss715646300: P=6.44x10-06) located at 48.06 Mb on Pv01 with MAF of 0.32 was the most significant SNP and explained 12 % of variation. Previous studies mapped QTL for number of seeds per pod, related to NSP on Pv01. SP1PP for seeds per pod was identified in PMB0225 x PHA1037 mapping population at 49.04 Mb (Yuste-Lisbona et al., 2014). QTL NSP-1AM for NSP was reported on Pv01 in PHA0419 x Beluga mapping population at 44.97- 45.46 Mb flanked by markers BMD045-FIN, while NSP-1MA was identified in Beluga x PHA0419 mapping population flanked by markers FIN (45.46 Mb) and BMC224 (48.11 Mb), respectively (Gonzalez et al., 2016). Interestingly, NSP-1MA colocalized with seed yield QTL SY-1MA. In addition, marker ss715646300 (48.06 Mb) for NSP was in LD (r2 > 0.5; D' =0.9) with ss715646315 (48.11 Mb) from 6K Beadchip, and in LD (r2 > 0.94; D' =1) with SNPs, S01_48060395 (48.06 Mb) and S01_48119839 (48.11 Mb) from the 265K GBS data. This is consistent with previous studies that mapped QTL for seed yield in the same genomic region using Buster x Roza mapping population evaluated under multiple stress conditions (Trapp et al., 2015). However, the positional candidate gene Phvul.001G217800 is 7.9kb upstream of ss715646300, annotated as ―Ca2+ ACTIVATED OUTWARD RECTIFYING K+ CHANNEL 5‖, and plays a critical role in abscisic acid (ABA) mediated opening and closure of stomata during drought stress (Osakabe et al., 2014b; Cai et al., 2017). SNP (ss715649043; P=1.12x10-06) located at 2.01 Mb on Pv11 was the most significant SNP and accounted for 12% of the variation in NSP. In addition, the two significant SNPs (ss715649043and ss715649044) on Pv11 were in strong LD (r2 > 0.9; D' =1), and spanned 30.9 kb region with four genes. Two genes within this same region could be considered candidates in seed development. Marker ss715649044 is within the genic region of Phvul.011G024100, annotated as ―SCARECROW-LIKE 3 (SCL3)‖. SCL3 is a member of GAI (GIBBERELLIC 33 ACID INSENSITIVE), RGA (REPRESSOR OF GAI) and SCR (SCARECROW), GRAS transcription factor family. In Arabidopsis, SCL3 is a positive regulator of gibberellin (GA) signaling and interacts with DELLA proteins to control GA responses and biosynthesis (Zhang et al., 2011). A recent study reported significant enrichment of SCL3 in developing seed of soybean especially late in the developmental stage, suggesting its role in seed development (Jones et al., 2010; Redekar et al., 2015). Phvul.011G024000 is 7.3kb upstream of ss715649044, annotated as ―BIFUNCTIONAL INHIBITOR/LIPID-TRANSFER PROTEIN/SEED STORAGE 2S ALBUMIN SUPERFAMILY PROTEIN‖. Members of this gene family have been associated with seed development and abiotic responses in plants. For instance, in rice lipid-transfer protein OsLTPL36, is essential in determining seed setting rate, 1000-grain weight, and seed quality (Wang et al., 2015). However, lipid transfer proteins have been reported to exhibit differentially expression under drought, salt, and cold conditions in Arabidopsis, wheat, and maize (Wei and Zhong, 2014; Safi et al., 2015) . In addition, seed storage 2S albumin proteins, a member of this gene complex is an important component of seed development in common bean and soybean (Parreira et al., 2016; Du et al., 2017). Seed yield per plant (SY) Significant signals associated with SY were detected in Kasese under DS conditions in combined data on Pv04, and Pv06 using the 6K Beadchip (Table 2.4a), and under NS and DS conditions with the 265K marker data on Pv02, Pv03, Pv05, Pv07, Pv08, and Pv10 (Figure 2.4, Table 2.4b). Heritability estimates for SY was 29.6% under NS and 22.5% under DS conditions, respectively. Several studies reported QTL for SY across the genome of common bean. QTL for SY was mapped to Pv02 (Blair et al., 2006; Blair et al., 2012; Trapp et al., 2015; Gonzalez et al., 2016; 34 Diaz et al., 2017). The QTL YD2.1DB tagged by marker BMa16 (33.1 Mb) on Pv02 in 98 RILs from a cross between DOR 364 × BAT 477 was detected under low phosphorus condition (Diaz et al., 2017). This QTL was within the 3.5 Mb region flanked by the two significant markers S02_34117091 (34.1 Mb), and S02_30580034 (30.5 Mb) detected for SY in Kasese under DS. However, the genomic region described above was different from QTL SY2.1BR at 11.8 Mb (SNP40055) identified in 140 F7:9 RILs from Buster x Roza population (Trapp et al., 2015), and SY-2AM flanked by markers BMB097 (13.65 Mb) and BMB080 (18.55 Mb) in F2:8 Beluga x PHA0399 population (Gonzalez et al., 2016). Significant SNPs for SY were identified on Pv03 under DS conditions in Kasese. The two most important significant SNPs were S03_51268099 at 51.2 Mb and S03_28782488 at 28.7 Mb. Several studies consistently reported QTL for SY on Pv03 (Beattie et al., 2003; Blair et al., 2006; Wright and Kelly, 2011; Checa and Blair, 2012; Mukeshimana et al., 2014; Kamfwa et al., 2015; Gonzalez et al., 2016). The yield QTL SY3.3SC was initially reported on Pv03 at ss715649325 (39.64 Mb) in SEA5 x CAL96 population (Mukeshimana et al., 2014). Subsequent studies reported yield QTL SY3.3 flanked by markers ss715646621(39.4 Mb) and ss715639345 (39.6 Mb) in Puebla 152 x Zorro population (Heilig et al., 2017). The same QTL was reported between markers ss715647671(37.06 Mb) and ss715639244 (45.59 Mb) a nested association mapping (NAM) population consisting of three small red-seeded bean lines (S48M, S94M, and S95M) crossed to Merlot evaluated under DS conditions (Hoyos-Villegas et al., 2015). In another study using the Andean diversity panel significant marker association for seed yield was detected on Pv03 for marker ss715648538 (38.27 Mb) (Kamfwa et al., 2015). The consistency in these studies suggests the existence of stable QTL for yield on Pv03. Another seed yield QTL, SY-3AM was identified in Beluga x PHA0399 population and flanked by BMD036 (46.37 Mb) and 35 BM189 (47.39 Mb) (Gonzalez et al., 2016). However the significant markers S03_51286099 (51.2 Mb) and S03_28782488 (28.7 Mb) associated with SY in this study were not in LD with markers previously reported, suggesting that the genomic regions detected for SY on Pv03 in this study could be different from those previously reported. Significant signals associated SY were identified on Pv04 in Kasese under DS conditions. Previous studies reported QTL for SY on Pv04 (Blair et al., 2006; Blair et al., 2012; Checa and Blair, 2012). Four QTL for SY (yld4.1, yld4.2, yld4.3, and yld4.4) were mapped to Pv04 using advanced backcross BC2F3:5 population from a cross between ICA Cerinza (the cultivated recurrent parent) and G24404 (the wild donor parent) (Blair et al., 2006). Another SY QTL yld4.1 on Pv04 was identified in 97 RILs from DOR 364 × BAT 477 population under DS condition (Blair et al., 2012), and HP yld4.1 on Pv04 in F5:8 population from G2333 × G19839 evaluated under high phosphorus condition (Checa and Blair, 2012). However, validation of these seed yield QTL on Pv04 is not possible due to unknown physical positions of the prior markers tagging these QTL. In this study, two significant SNPs were identified on Pv04 and the most significant SNP (ss715645803; P=6x10-05) and explained about 8% of the variation (Table 2. 4a). The two SNPs (ss715645803 and ss715645801) were in strong LD (r2 > 0.9; D' =0.98), and spanned 53.7kb region with four genes. Marker ss715645803 is 0.46kb upstream of Phvul.004G163300, annotated as ―HEAT SHOCK TRANSCRIPTION FACTOR C1 (HSFC1)‖. HSFC1 is a member of the plant heat stress transcription factors (HSFs) that regulate the expression of stress responsive genes such as heat shock proteins under adverse environmental conditions such as heat, drought and cold stress in soybean and common bean (Li et al., 2014; Wang et al., 2014; Guo et al., 2016). A recent proteomic analysis of the stem of drought-tolerant common bean 36 cultivar Tiber detected increased levels of 70k dalton heat shock proteins under drought condition. This demonstrates the importance of heat shock proteins in protecting and maintaining protein conformations and cellular homeostasis during drought stress (Zargar et al., 2017). Furthermore, Phvul.004G163600 is 1.8kb upstream of significant marker ss715645801 and annotated as ―POLYOL/MONOSACCHARIDE TRANSPORTER 5 (PMT5)‖. PMT5 is a member of monosaccharide transporter (MST) family that are mainly involved in the long distance transport of photosynthate, primarily sugars and sugar alcohols such as mannitol, sorbitol, and xylitol in tree plants of the Rosaceae family (Noiraud et al., 2001; Slewinski, 2011; Reuscher et al., 2014) . In Arabidopsis AtPMT5 was partially characterized as a broad-spectrum H+symporter for polyols, cyclitols and several monosaccharaides (Klepek et al., 2005; Klepek et al., 2009; Reuscher et al., 2014). However, the physiological functions of PMT5 remain unclear because of wide specificity of substrates, and non-responsive mutant phenotype under drought stress (Klepek et al., 2005; Klepek et al., 2009; Reuscher et al., 2014). Significant signals for SY was also detected on Pv05 at S05_18002393 (18.0 Mb). QTL for seed yield were previously mapped to Pv05 (Tar'an et al., 2002; Wright and Kelly, 2011; Trapp et al., 2015; Gonzalez et al., 2016) . QTL SY5.1BR for SY at 38.7 Mb was linked to marker SNP45307 and identified in 140 F7:9 RILs from Buster x Roza population grown under NS conditions (Trapp et al., 2015) , while QTL SY-5AM flanked by IAC044(1.6Mb) and IAC010Pv05(2.7 Mb) was identified in F2:8 RIL population from Beluga x PHA0399 cross, respectively (Gonzalez et al., 2016). Significant QTL signals for seed yield were detected on Pv06 consistent with previous studies (Cichy et al., 2009; Blair et al., 2012).The QTL YD6.1DB on Pv06 was identified from a cross between DOR 364 × BAT 477 under DS condition (Blair et al., 2012), However the physical 37 distance of markers tagging this particular yield QTL on Pv06 are not known. In addition, colocalized GWAS signals were detected for PW and SY on Pv06. Significant SNP (ss715640054; P=2x10-05) with MAF of 0.12 located at 23.23 Mb on Pv06 was the most significant for PW, described earlier and explained 9% of the variation in SY. The positional candidate gene associated with this SNP is Phvul.006G117500, annotated as ‗PLANT INVERTASE/PECTIN METHYLESTERASE INHIBITOR (INH/PMEI) SUPERFAMILY PROTEIN‘. Significant marker associations were detected on Pv07 and several QTL for SY have been reported on Pv07 (Cichy et al., 2009; Wright and Kelly, 2011; Asfaw et al., 2012; Blair et al., 2012; Hoyos-Villegas et al., 2016; Diaz et al., 2017). Interestingly, QTL for seed yield reported on Pv07 using different set of markers in previous studies appear to overlap within the same region. The QTL SY7.3 reported on Pv07 flanked by ss715650404 (4.59 Mb) and ss715640392 (5.03 Mb) detected in a NAM population grown under DS conditions (Hoyos-Villegas et al., 2015).QTL YD7.1DB and YD7.2DB on Pv07 flanked by BMc72 (0.28 Mb) and BMb1536 (4.6 Mb) from DOR 364 × BAT 477 population evaluated under low and medium phosphorus conditions (Diaz et al., 2017). In the current study, significant signals for SY on Pv07 were detected for markers S07_17054621 (17.0 Mb) and S07_12460824 (12.4 Mb), respectively distant from regions reported in previous studies. Quantitative trait loci for SY were reported on Pv08 (Perez-Vega et al., 2010; Asfaw et al., 2012; Blair et al., 2012; Kamfwa et al., 2015). The QTL YLD8.1DB, and YLD8.2DB (Blair et al., 2012) and several SY QTL reported by Asfaw et al. (2012) on Pv08 were identified using the DOR 364 × BAT 477 population in multiple environments. However the physical positions of 38 the markers linked to these QTL were unknown preventing comparisons with significant marker S08_57056693 (57.0 Mb) detected in Kasese under DS conditions. In the current study, five significant association signals were detected on Pv10 for SY under both DS and NS conditions. The four significant markers S10_27780836 (27.7 Mb), S10_27883169 (27.8 Mb), S10_28437825 (28.4 Mb) and S10_30139711 (30.1 Mb) all detected under NS conditions were in LD (r2 > 0.65; D' =1), with exception of marker S10_39085960 (39.0 Mb) identified under DS in Kasese. QTL for SY have been consistently reported on Pv10 in previous studies (Wright and Kelly, 2011; Blair et al., 2012; Checa and Blair, 2012; Trapp et al., 2015; Hoyos-Villegas et al., 2016). The QTL, YLD10.1GG on Pv10 was reported under high phosphorus condition in G2333 × G19839 population tagged by marker BM157 at 8.9 Mb (Checa and Blair, 2012). SY10.1BR was identified in Buster x Roza population on Pv10 under DS linked to SNP46337 (39.9 Mb) (Trapp et al., 2015). The same QTL SY10.1 was detected between markers ss715646348 (40.1 Mb) and ss715645508 (40.95 Mb) in NAM population of three small red-seeded beans evaluated under DS conditions (Hoyos-Villegas et al., 2015). Harvest index (HI) Four significant SNPs were detected for HI under NS conditions in combined data on Pv09 and under DS in Namulonge on Pv02 and Pv04. The heritability estimate for HI was 48.5% under NS and 25.7 under DS. Quantitative trait loci for HI were reported on Pv02, Pv04, Pv05, Pv06, Pv09, Pv10, and Pv11 in DOR 364 × BAT 477 RIL population under DS condition in multiple environments (Asfaw et al., 2012). Two SNPs all on Pv02 were significantly associated with HI under drought stress conditions in Kasese, consistent with previous studies that reported significant QTL signal for HI and pod partitioning index QTL on Pv02 in DOR364 x BAT477 population grown under stress conditions (Asfaw et al., 2012). The most significant SNP 39 (ss715648530; P=9.13x10-05) was located at 24.2 Mb on Pv02 with MAF 0.14 and explained 6% of the variation in HI. In addition, the two SNPs (ss715648530 and ss715648525) were in strong LD (r2 > 0.9; D' =1) with this SNP, and spanned 87.9 kb region with seven genes. Positional candidate gene Phvul.002G119500 is 29.7kb upstream of ss715648530, and annotated as ―DIACYLGLYCEROL ACYLTRANSFERASE FAMILY (DGATs)‖. In soybeans, DGAT activity has been reported as the rate-limiting step in triacylglycerol (TAG) biosynthesis, the most common storage lipids in seeds, suggesting its critical role in accumulating seed oils during seed development (Lardizabal et al., 2008; Li et al., 2013; Chen et al., 2016). In addition, a recent study reported the involvement of GmDGAT2D a member of the monoacylgycerol acyltransferase (MGAT) family DGAT2, in plant response to environmental and hormonal cues in soybean (Chen et al., 2016). For instance, GmDGAT2D activity was up- regulated by plant hormone ABA under cold and heat stress, suggesting its role in responses to abiotic stresses and hormonal cues (Chen et al., 2016). Another candidate gene Phvul.002G119000 is 6.6kb downstream of ss715648525, annotated as ―NADPH-DEPENDENT THIOREDOXIN REDUCTASE A (NTRA)‖. In Arabidopsis and rice, NTRA activity is associated with enhanced tolerance to drought stress by acting as an antioxidant and scavenging ROS to prevent oxidative damage (Serrato et al., 2004; Cha et al., 2014; Kim et al., 2017). Recent study in wheat reported the localization of NTRs in developing and germinating wheat seed, and NTR dependent antioxidant activity in seed cells suffering from oxidative stress, suggesting that NTRs play an important role as antioxidant during reproductive stage of wheat (Pulido et al., 2009). 40 Colocalization of GWAS Signals for Photosynthetic Traits In this study, significant association signals were detected for photosynthetic traits namely LEF, Phi2, PhiNPQ and PhiNO on chromosomes Pv03, Pv04, Pv05, Pv09, and Pv11 under drought stress conditions with multiple traits exhibiting colocalized signals. Heritability estimates for photosynthetic traits were LEF (22.5%), Phi2 (21.6%), PhiNPQ (14.4%), and PhiNO (8.1%) respectively. Significant QTL signals associated with photosynthetic traits listed above have not been reported previously in common bean, but QTL for relative chlorophyll content and photosynthetic efficiency (FV/FM) as fv′/fm′ (Beebe et al., 2013a) have been identified in common bean (Asfaw et al., 2012; Kamfwa et al., 2015b; Diaz et al., 2017). Previous studies reported significant QTL signals for relative chlorophyll content on Pv03, and Pv04 in DOR 364 × BAT 477 population at pod filling stage under drought and low phosphorus conditions (Asfaw et al., 2012; Diaz et al., 2017). QTL for relative chlorophyll content SCMR_f3.1DB and SCMR_f3.2DB were mapped to Pv03, and linked to markers BMb1008 (25.5 Mb) and BMb37 (22.2 Mb) under low phosphorus conditions, while SCMR_f4.1DB mapped to Pv04 linked to X701 and flanked by BMb353 (28.6 Mb) and BMa20 (35.4 Mb) under low and moderate phosphorus conditions (Diaz et al., 2017). In addition, QTL for relative chlorophyll content were reported on Pv09 and associated with markers ss715648916 (20.0 Mb), and ss715647747 (26.6 Mb) in ADP panel evaluated under field and greenhouse conditions in Michigan (Kamfwa et al., 2015). Furthermore, QTL FVFM11.3DB for photosynthetic efficiency (FV/FM) as fv′/fm′ was reported on Pv11 and linked to marker BMb10 at 40.0 Mb under low phosphorus conditions (Diaz et al., 2017). In this study, photosynthetic traits measured exhibited colocalized QTL signals. For instance, four significant SNP associations were detected on Pv11 for LEF, Phi2 in Kasese_2016 and 41 PhiNPQ. Marker (ss715650748; P=1.60x10-07) located at 48.78 Mb on Pv11 with MAF of 0.11 was the most significant SNP for LEF and PhiNPQ and in strong LD (r2 =0.65; D' =0.83) with the most significant SNP (ss715640406; P=1.04x10-05) for Phi2 in Kasese_2016 located at 49.24 Mb with MAF of 0.13. This region spanned 462.3kb with 29 genes. Marker ss715650748 is within promoter region (0.24kb) of gene Phvul.011G207400, and encodes a ―TRANSDUCIN/WD40 REPEAT-LIKE SUPERFAMILY PROTEIN (WD40-REPEAT)‖. The WD40-REPEAT proteins are key regulators of plant-specific developmental process including response to abiotic stresses such as drought (Van Nocker and Ludwig, 2003; Shaar-Moshe et al., 2015). SNP ss715640406 (most significant for Phi2 in Kasese_2016) is 1.9kb downstream of Phvul.011G210200 and encodes ―RING/U-BOX SUPERFAMILY PROTEIN‖, a member of ubiquitin ligase (E3) that regulate specific molecular processes by acting as target recognition sites for recruiting specific target proteins for degradation in response to various biotic and environmental cues (Yee and Goring, 2009; Lee and Kim, 2011). For instance, RING/U-BOX SUPERFAMILY PROTEINs has been found to be a negative regulator of ABA mediated responses to drought stress by modulating ABA-signaling pathway (Yee and Goring, 2009; Sadanandom et al., 2012). Another candidate gene Phvul.011G210000 located 28.3kb upstream of marker ss715640406 encodes ―DNAJ HEAT SHOCK N-TERMINAL DOMAIN- CONTAINING PROTEIN‖. Heat-shock proteins are molecular chaperones that protect plants cells against adverse effects of abiotic stresses such as drought and heat by maintaining stability in protein conformation and cellular homeostasis (Takayama et al., 2003). However, the activities of heat shock proteins are regulated by heat shock transcription factors (Hsf), that have been up regulated in plants overexpressing Hsf under drought stress (Li et al., 2014; Jacob et al., 2017). Recent studies in Arabidopsis and soybean showed that plants overexpressing heat shock 42 transcription factor had enhanced tolerance to drought stress, by activating stress responsive genes in an ABA dependent manner (Lim et al., 2006; Hwang et al., 2014; Li et al., 2014). Additionally, three SNPs were significantly associated with LEF in Kasese_2016 and PhiNPQ on Pv04. Two SNPs, ss715645808 at 44.40 Mb and ss715645811 at 44.33 Mb in strong LD (r2 =0.75; D' =0.96) spanned 71.7kb region with four genes. Candidate gene Phvul.004G161500 is 14.4 kb downstream of ss715645811, annotated as ―LEUCINE-RICH REPEAT (LRR) FAMILY PROTEIN‖. LRR are members of receptor-like kinase (RLK) family that acts as a signaling molecule in modulating plants response to biotic and abiotic stress (Ouyang et al., 2010; Osakabe et al., 2014; Dievart et al., 2016). In Kasese_2017, colocalized signals were detected for Phi2 and PhiNO on Pv03. SNP (ss715646087; P=3.5x10-06) located at 52.22 Mb with MAF of 0.11 was the most significant SNP for Phi2 and in strong LD (r2 =0.70; D' =0.93) with marker (ss715646085; P=8.6x10-05) located at 51.73 Mb, the most significant SNP for PhiNO and for Phi2. The two markers flanked a 484.2 kb region with 56 genes. Candidate gene Phvul.003G291500 is 5.3kb downstream of ss715646085 and encodes ―GRAS FAMILY TRANSCRIPTION FACTOR FAMILY PROTEIN (GRAS)‖. GRAS transcription factors modulate expression of plant-specific GRAS proteins that are important in plant development (Hirsch and Oldroyd, 2009; Cenci and Rouard, 2017). A recent study in rice reported improved tolerance to drought and oxidative stress in mutants overexpressing GRAS transcription factor gene, OsGRAS23, suggesting its role in modulating plant response to abiotic stress (Xu et al., 2015). Another gene Phvul.003G291800 is 25kb downstream of ss715646085 encoded ―ABSCISIC ACID RESPONSIVE ELEMENTS- BINDING FACTOR 2 (AREB1/ABF2)‖. AREB1/ABF2 regulates the expression of ABA inducible genes that are critical in improves tolerance of plants to drought and osmotic stress, 43 and enhanced seed development (Fujita et al., 2005; Nakashima and Yamaguchi-Shinozaki, 2013). Conclusion The objective of this study was to conduct a GWAS to determine the genetic architecture of terminal drought tolerance in the Andean common bean diversity panel evaluated under drought conditions in Uganda. A substantial number of significant QTL signals associated with agronomic and photosynthetic traits were identified under both drought stress and non-stress conditions using 6K Beadchip and 265K SNP data from GBS used in this study. Genetic variance explained by the association signals ranged from 6% to 14% for 6K Beadchip, and 11.1 % to 17.1% with 265K SNP data implying that most of the genomic regions detected had relatively small but cumulative genetic effects on these traits, which is consistent with previous QTL signals identified using bi-parental mapping populations grown under drought conditions. Colocalized GWAS signals were detected for yield component traits and photosynthetic traits. Important positional candidate genes identified included Phvul.006G117500 8.7kb upstream of SNP ss715640054 (23.23 Mb) on Pv06, annotated as ‗PLANT INVERTASE/PECTIN METHYLESTERASE INHIBITOR (INH/PMEI) SUPERFAMILY PROTEIN‘ that is involved in starch and sucrose metabolism during seed development under DS. Phvul.011G024000 7.3kb upstream of ss715649044 on Pv04, annotated as ―BIFUNCTIONAL INHIBITOR/LIPID- TRANSFER PROTEIN/SEED STORAGE 2S ALBUMIN SUPERFAMILY PROTEIN‖, which is a seed storage protein important in seed development. Meanwhile positional candidate genes associated with colocalized GWAS signals or photosynthetic traits include Phvul.011G210200 located 1.9kb upstream of SNP ss715640406, 44 annotated as ―RING/U-BOX SUPERFAMILY PROTEIN‖, Phvul.011G210000, located at 28.3kb upstream of marker ss715640406 and encodes ―DNAJ HEAT SHOCK N-TERMINAL DOMAIN-CONTAINING PROTEIN‖ all on Pv11 for Phi2, LEF, and PhiNPQ under DS. Phvul.003G291800 located 25kb downstream of ss715646085, annotated as ―ABSCISIC ACID RESPONSIVE ELEMENTS-BINDING FACTOR 2 (AREB1/ABF2)‖ for Phi2 and PhiNO under DS. Furthermore, these candidate genes are involved in partitioning of photosynthate to seed yield and regulating plant responses to drought stress at pod filling stage. The SNPs tagging these genomic regions could be used in further studies with bi parental mapping to determine their usefulness for marker assisted breeding to improve genetic gain for seed yield under drought stress. In conclusion, this study demonstrates the effectiveness of GWAS in uncovering genomic regions that contribute to yield under drought stress in Andean common bean germplasm. 45 APPENDIX 46 Table 2.1a. Mean, minimum and maximum phenotypic values associated with agronomic traits and partitioning in Andean Diversity panel of 256 common bean genotypes grown under drought stress and non-stress conditions in Kasese for two seasons, 2016 and 2017 in Uganda Kasese Non stress (NS) Year Trait 2016 100SWT (g) 2017 2016 Biomass per plant (g) 2017 2016 Pod weight per plant (g) 2017 2016 Seed yield per plant (g) 2017 2016 Harvest index (HI) 2017 2016 Pod harvest index (PHI) 2017 Pod Partitioning index (PPI) 2016 2017 2016 Days to flowering (DF) 2017 2016 Days to maturity (DM) 2017 2016 Pod number per plant 2017 Seed number per plant 2016 2017 Drought stress (DS) Min. Max Variance Mean Mean 25.8±0.5 14.3 29.6.0±0.5 16.7 47.3±3.2 15.0 11.5 49.1±2.9 5.4 17.4±2.0 5.3 16.9±1.8 11.8±1.1 2.0 2.5 12.2±1.3 2.0 17.1±0.8 31.0±1.1 3.9 3.0 40.8±1.7 14.4 39.9±2.1 3.0 37.6±1.1 48.5±1.3 6.4 32.0 39.9±0.3 30.0 36.7±0.2 64.0 75.7±0.8 72.7±0.9 61.0 2.7 9.6.4±1.5 2.7 10.6±1.8 18.2±3.1 3.0 3.7 20.1±3.3 27.6±0.5 33.2±0.5 61.8±3.2 68.2±2.9 21.2±2.0 24.0±1.8 16.9±1.1 19.2±1.3 20.7±0.8 39.1±1.1 48.4±1.7 45.5±2.1 40.4±1.1 52.2±1.3 35.9±0.3 36.1±0.2 77.9±0.8 74.3±0.9 14.5±1.5 16.1±1.8 24.0±3.1 29.5±3.3 *** *** *** *** *** *** ** *** *** *** *** *** *** *** *** *** ** *** ** * *** *** Min. Max. 64.0 12.5 65.4 14.4 26.8 93.0 99.2 16.3 66.1 4.0 64.0 2.6 4.5 54.0 56.2 5.5 65.6 3.3 4.2 70.1 80.0 2.9 81.5 6.3 67.6 3.9 4.7 77.1 40.0 32.0 45.0 30.0 93.0 67.0 69.0 95.0 21.5 4.0 24.0 2.0 7.0 47.9 54.5 8.0 Variance ** *** *** *** ** ** *** ** *** ** ** *** *** ** *** *** ** ** * * *** *** 41.7 61.7 74.2 81.7 50.0 55.0 48.3 50.8 61.5 50.8 60.9 63.4 50.5 57.3 46.0 44.0 91.0 89.0 15.5 16.2 37.3 41.2 47 Table 2.1b. Mean, minimum and maximum phenotypic values associated with agronomic traits and partitioning in Andean Diversity panel of 256 common bean genotypes grown under drought stress and non-stress conditions in Namulonge for two seasons, 2016 and 2017 in Uganda Drought stress (DS) Namulonge Non stress (NS) Trait 100SWT (g) Biomass per plant (g) Pod weight per plant (g) Seed yield per plant (g) Harvest index (HI) Pod harvest index (PHI) Pod Partitioning index (PPI) Days to flowering (DF) Days to maturity (DM) Pod number per plant Seed number per plant Year Mean 2016 2017 2016 2017 2016 2017 2016 2017 2016 2017 2016 2017 2016 2017 2016 2017 2016 2017 2016 2017 2016 2017 24.8±0.8 26.6±0.5 36.6±2.9 46.3±3.5 8.0±1.1 14.3±1.6 5.6±1.4 9.2±1.2 10.5±1.1 22.7±1.3 39.1±2.1 42.4±1.8 19.8±1.7 21.5±1.4 38.4±0.4 36.3±0.6 73.1±0.3 77.0±0.8 8.1±1.0 9.2±1.3 13.9±3.8 18.5±2.9 Variance Mean *** *** *** *** *** *** *** *** *** *** *** *** *** *** ** ** *** ** * * *** *** 32.8±0.8 33.1±0.5 51.3±2.9 58.2±3.5 16.0±1.1 19.3±1.6 13.6±1.4 14.9±1.2 25.7±1.1 31.0±1.3 47.9±2.1 48.5±1.8 34.7±1.7 39.5±1.4 38.8±0.4 37.6±0.6 78.3±0.3 79.0±0.8 12.0±1.0 13.2±1.3 20.0±3.8 23.4±2.9 Min Max 12.5 45.4 15.9 47.0 11.0 68.1 18.6 100.2 35.1 2.5 37.3 5.2 23.8 1.3 29.0 2.5 39.8 3.0 50.9 8.6 73.2 2.5 67.0 3.5 51.0 4.5 2.5 53.1 31.0 45.0 32.0 43.0 67.0 91.0 69.0 94.0 14.4 3.0 15.5 4.0 3.0 39.1 40.9 6.0 48 Min Max Variance 11.5 13.3 14.0 20.4 6.1 7.5 4.0 6.3 3.0 7.5 4.0 5.5 6.0 4.3 32.0 31.0 67.0 66.0 4.0 5.0 5.0 7.0 57.6 62.7 88.7 113.6 56.5 60.0 45.2 40.3 45.3 66.3 76.3 75.1 73.3 68.1 46.0 44.0 86.0 92.0 19.1 22.1 43.5 48.7 ** ** *** *** *** ** *** ** *** ** ** *** *** ** ** ** *** ** * * *** *** Table 2.2. Mean and Variance associated with photosynthetic traits in Andean Diversity Panel (ADP) of 256 common bean genotypes grown under drought stress and non-stress conditions in Kasese for two seasons, 2016 and 2017 in Uganda. Kasese Year 2016 2017 2016 2017 2016 2017 2016 2017 2016 2017 Trait LEF SPAD Phi2 PhiNO PhiNPQ Rainfed Mean 217.7±3.2 133.7±1.5 37.9±0.3 37.7±0.4 0.35±0.004 0.39±0.003 0.27±0.005 0.22±0.004 0.38±0.007 0.39±0.006 Irrigated Variance Mean Variance 206.6±3.2 140.5±1.5 36.1±0.3 39.9±0.4 0.34±0.004 0.40±0.003 0.28±0.005 0.24±0.004 0.39±0.007 0.36±0.006 * *** *** *** ** *** ** *** *** *** ** *** *** *** *** *** *** *** *** *** 49 Table 2.3. Pearson correlation coefficients among nine agronomic and three photosynthetic traits measured on 256 common bean genotypes in the Andean diversity panel grown at the Kasese, Uganda drought stress conditions in 2017. Traits DTM BM PD PW SD SY 100SWT PHI HI PPI Phi2 PhiNO PhiNPQ PD BM 0.18* 0.13* 0.46*** PW 0.16* 0.54*** 0.84*** SD 0.13* 0.46*** 0.96*** 0.84*** SY 0.12* 0.45*** 0.89*** 0.87*** 0.89*** 100SWT 0.13* -0.03ns 0.03ns 0.19* 0.03ns 0.24** PHI -0.07ns -0.18* 0.22** 0.15* 0.22** 0.28** 0.25** HI -0.02ns -0.09ns 0.41*** 0.30** 0.41*** 0.40*** 0.30** 0.60*** PPI 0.09ns -0.09ns 0.42*** 0.43*** 0.43*** 0.38** 0.27** 0.19* 0.74*** Phi2 0.01ns -0.01ns 0.01ns 0.03ns 0.01ns -0.01ns 0.01ns -0.16* -0.11* 0.01ns PhiNO -0.07ns 0.08ns 0.03ns 0.01ns 0.03ns 0.01ns -0.04ns -0.16* -0.08ns 0.02ns 0.66*** PhiNPQ 0.04ns -0.05ns -0.02ns -0.02ns -0.02ns 0.01ns 0.02ns 0.18* 0.09ns -0.01ns -0.88*** -0.93*** BM Biomass, PD Pod number per plant, PW Pod weight per plant, SD Seed number per plant, SY Seed yield per plant.100SWT 100 seed weight, PHI Pod harvest Index, HI Harvest Index, PPI Pod Partitioning Index. Photosystem II (Phi2), Ratio of incoming light that is lost via non-regulated processes (PhNO), Ratio of incoming light that goes towards non-photochemical quenching (PhiNPQ), * Significant at the 0.05 probability level, ** Significant at the 0.01 probability level, *** Significant at the 0.001 probability level, † ns, not significant 50 Table 2.4a. Location, chromosome, position, P-values, proportion of phenotypic variation explained (R2), and minor allele frequency of two most significant single nucleotide polymorphisms (SNPs) for five agronomic traits and four photosynthetic traits measured on 256 common bean genotypes in the Andean diversity panel grown under DS Conditions at Kasese and Namulonge, Uganda in 2016 and 2017. Location Kasese Kasese Trait Number of pods per plant Pod weight per plant (g) Number of seeds per plant Seed yield per plant (g) Kasese Harvest index (HI) PhiNPQ LEF Phi2 Phi2 PhiNO Namulonge Namulonge Chromosome SNP Pv10 Pv06 Pv06 Pv06 Pv02 Pv02 Pv11 Pv01 Pv06 Pv04 Pv02 Pv02 Pv11 Pv09 Pv11 Pv11 Pv11 Pv11 Pv03 Pv03 Kasese_2017 Kasese_2017 Pv03 ss715641140 ss715639365 ss715640054 ss715644685 ss715641175 ss715645824 ss715649043 ss715646300 ss715640054 ss715645803 ss715648530 ss715648525 ss715650748 ss715648421 ss715650748 ss715640406 ss715640406 ss715649384 ss715646087 ss715646085 ss715646087 Kasese Kasese Kasese_2016 Kasese_2016 Position (bp) 32728023 31379011 23234133 23135033 24351389 24689544 2018589 48065649 23234133 44509705 24230627 24142672 48779978 539280 48779978 49242263 49242263 49159380 52218575 51734329 52218575 P-value 7.5x10-06 4.2x10-05 4.2x10-06 1.8x10-05 4.1x10-06 4.1x10-06 1.1x10-06 6.4x10-06 2.0x10-05 6.0x10-05 9.1x10-05 9.7x10-05 1.2x10-06 3.9x10-05 1.6x10-07 1.0x10-06 1.0x10-05 8.8x10-05 3.6x10-06 5.4x10-06 8.7x10-05 R2 0.09 0.09 0.10 0.07 0.09 0.09 0.12 0.12 0.09 0.08 0.06 0.06 0.13 0.11 0.14 0.11 0.11 0.08 0.10 0.11 0.10 MAF 0.12 0.11 0.12 0.14 0.09 0.09 0.17 0.32 0.12 0.15 0.14 0.15 0.11 0.17 0.11 0.13 0.13 0.13 0.13 0.13 0.13 51 Table 2.4b. Location, chromosome, position, P-values, proportion of phenotypic variation explained (R2), and minor allele frequency of significant single nucleotide polymorphisms (SNPs) for four agronomic traits with 265K SNPs measured on 256 common bean genotypes in the Andean diversity panel grown under drought stress (DS) and non-stress (NS) conditions at Kasese and Namulonge, Uganda in 2016 and 2017. Treatment Location Chromosome Trait DS Number of pods DS Number of pods DS Number of pods DS Number of pods DS Number of pods DS Number of pods DS Number of pods DS Number of pods DS Number of pods DS Number of pods DS Number of pods DS Number of pods DS Number of pods Number of pods DS Pod weight per plant DS Pod weight per plant DS Pod weight per plant DS Pod weight per plant DS Pod weight per plant DS Combined 8 Combined 7 Combined 7 Combined 3 Combined 3 Combined 3 Combined 3 Combined 3 Combined 3 Combined 2 Combined 2 Combined 2 Combined 2 Combined 6 Kasese 6 6 Kasese 6 Kasese Kasese 6 6 Kasese SNP Position(bp) P_value S08_55438500 55438500 S07_11078838 11078838 S07_26035465 26035465 S03_52310519 52310519 S03_52310559 52310559 S03_52310595 52310595 S03_52310665 52310665 S03_52310683 52310683 S03_52310685 52310685 S02_22564824 22564824 S02_22564882 22564882 S02_22564888 22564888 S02_22564914 22564914 S06_23054750 23054750 S06_23021418 23021418 S06_23996747 23996747 S06_23996725 23996725 S06_22229432 22229432 S06_23996801 23996801 7.30x10-09 1.06x10-08 1.10x10-08 1.10x10-08 1.10x10-08 1.10x10-08 1.10x10-08 1.10x10-08 1.10x10-08 2.36x10-07 2.36x10-07 2.36x10-07 2.36x10-07 1.29x10-06 8.61x10-08 3.08x10-07 4.92x10-07 7.44x10-07 8.03x10-07 MAF R2 12 7 22 9 9 9 9 9 9 11 11 11 11 7 8 8 9 8 8 16.6 16.2 16.2 16.2 16.2 16.2 16.2 16.2 16.2 13.3 13.3 13.3 13.3 11.7 14.7 13.6 13.2 12.8 12.8 52 Table 2.4b (cont’d) Trait Treatment Location Chromosome SNP Position(bp) P_value MAF R2 Seed yield per plant NS Seed yield per plant NS Seed yield per plant NS Seed yield per plant NS Seed yield per plant DS Seed yield per plant DS Seed yield per plant DS Seed yield per plant DS Seed yield per plant DS Seed yield per plant DS Seed yield per plant DS Seed yield per plant DS Seed yield per plant DS Seed yield per plant DS Harvest Index Harvest Index Harvest Index Harvest Index Harvest Index Harvest Index NS NS NS NS NS NS Combined Combined Combined Combined Kasese Kasese Kasese Kasese Kasese Kasese Kasese Kasese Kasese Kasese Combined Combined Combined Combined 10 10 10 10 3 7 2 5 2 2 3 10 8 7 9 9 9 9 Namulonge 2 Namulonge 4 S10_27780836 27780836 S10_27883169 27883169 S10_30139711 30139711 S10_28437825 28437825 S03_51268099 51268099 S07_17054621 17054621 S02_34117091 34117091 S05_18002393 18002393 S02_30580034 30580034 S02_49146965 49146965 S03_28782488 28782488 S10_39085960 39085960 S08_57056693 57056693 S07_12460824 12460824 S09_28676522 28676522 S09_28676493 28676493 S09_28676508 28676508 S09_28676553 28676553 S02_30630638 30630638 S04_31117609 31117609 3.55x10-09 8 2.24x10-08 7 1.46x10-07 4 1.57x10-07 7 3.07x10-09 6 1.13x10-08 6 3.00x10-07 11 3.37x10-07 6 4.53x10-07 8 4.56x10-07 34 5.37x10-07 16 5.50x10-07 5 7.94x10-07 14 8.38x10-07 10 5.21x10-07 6 5.21x10-07 6 5.21x10-07 6 5.64x10-07 5 3.18x10-07 10 4.64x10-07 11 17.1 15.2 13.3 13.2 16.5 15.3 12.4 12.3 12.1 12.1 11.9 11.9 11.6 11.5 11.1 11.1 11.1 11.1 13.0 12.6 53 Figure 2.1. Mean rainfall (mm) and mean temperature (°C) for Kasese_2016(a), Kasese_2017(b); Namulonge_2016(c) and Namulonge_2017(d) growing seasons. a c b d 54 Figure 2.2. The plot of principle component analysis (PCA) PC1 against PC2, illustrating the population structure in subset of the Andean Diversity Panel used in this study with 265K SNPs 55 Table 2.5. Eleven genotypes with good performance based on geometric mean for seed yield from the Andean Diversity Panel evaluated under DS and NS at Kasese and Namulonge in 2016 and 2017. ADP(ID) ADP-0670 ADP-0009 ADP-0659 ADP-0054 ADP-0432 ADP-0119 Rust resistant ADP-0648 ADP-0468 ADP-0061 ADP-0660 ADP-0430 Grand Mean SED Cultivar name Country(region) Seed color AC Calmont Maalasa USLK-1 W6 16447 PR0637-134 A193 Red Kloud PI527538 Maulasi Krimson PR1013-3 Canada Tanzania USA Tanzania Caribbean Africa USA East Africa Africa USA Caribbean DRK Red mottled LRK Cranberry Red mottled spec red LRK yellow Cranberry Cranberry Pink mottled Traits High yield High yield Roza HBB resistant High yield Rust resistant Drought Tolerant Drought Tolerant CBB GM SY_DS(g) SY_NS(g) 54.5 42.6 40.5 39.7 37.5 33.9 33.5 32.3 31.0 30.8 30.2 15.5 1.65 57.4 45.1 54.7 47.8 40.5 46.1 42.4 36.4 57.1 46.0 29.5 18.3 1.9 51.7 40.2 30.0 32.9 34.7 24.9 26.4 28.6 16.7 20.6 30.8 11.5 1.01 ADP Andean Diversity Panel identity, GM Geometric mean, SY_DS Seed yield under drought stress, SY_NS Seed yield under non- stress, DRK Dark red kidney, LRK Light red kidney, CBB, Common bacterial blight, HBB, Halo bacterial blight, SED standard error of difference 56 Figure 2.3. Manhattan plots showing candidate single nucleotide polymorphism (SNP) markers, and their P- values from genome wide association using mixed linear model for number of pods per plant on Pv02, Pv03, Pv06, Pv07, andPv08, pod weight per plant on Pv06 in Kasese for two seasons under DS conditions. Number of pods per plant _Kasese Pod weight per plant (g) _Kasese 57 Figure 2.4. Manhattan plots showing candidate single nucleotide polymorphism (SNP) markers, and their P- values from genome wide association using mixed linear model for and seed yield per plant on Pv02, Pv03, Pv05, Pv07, Pv08,andPv10 under DS and harvest index on Pv02, and Pv04 under NS conditions. Seed yield per plant (g) _Kasese Harvest Index 58 Figure 2.5. Manhattan plots showing candidate single nucleotide polymorphisms (SNPs) and their P- values from genome wide association using mixed linear model for LEF on Pv04 andPv011, for PhiNPQ on Pv04, Pv05, Pv09 and Pv011 and for Phi2 on Pv011 under drought conditions. LEF_Kasese_2016 PhiNPQ Phi2_Kasese_2016 59 Figure 2.6. Manhattan plots showing candidate single nucleotide polymorphisms (SNPs) and their P- values from genome wide association using mixed linear model for PhiNO on Pv03 and for Phi2 on Pv03 and Pv05 under drought conditions. PhiNO_Kasese_2017 Phi2_Kasese_2017 60 CHAPTER 3: IDENTIFICATION OF QUANTITATIVE TRAIT LOCI ASSOCIATED WITH DROUGHT TOLERANCE IN PORTILLO X RED HAWK RECOMBINANT INBRED LINE MAPPING POPULATION Abstract Drought stress is increasingly becoming the most important abiotic stress limiting productivity of common bean (Phaseolus vulgaris. L.) globally. The objective of this study was to conduct QTL analysis for drought tolerance in Portillo x Red Hawk recombinant inbred line (RIL) mapping population genotyped using single nucleotide polymorphism (SNP) markers from the BARCBean6K_3 Beadchip. The RIL population was evaluated for drought tolerance in Uganda for two seasons in Kasese and for one season at Namulonge under drought and non-stress conditions. The drought intensity index for seed yield per plant was moderate although seed yield per plant was significantly reduced under drought stress. A linkage map of the Portillo x Red Hawk mapping population was constructed with 206 markers that spanned 766.5 cM on ten of the eleven chromosomes of common bean with an average distance of 3.7 cM between markers. Thirty-two significant QTL signals were identified for phenology, yield component traits and partitioning traits measured under drought and non conditions. Colocalized QTL signals were consistently detected for phenology, yield component and partitioning traits on Pv01, Pv02, Pv03, Pv04, Pv06 and Pv11. Functional annotation of positional candidate genes identified within the colocalized peak regions were consistent with the functional similarity or pleiotropic effects of those genes on the associated traits. For instance, we identified positional candidate genes on Pv01 that were significantly associated with QTL SY1.1 and PPI1.2 under both DS and NS conditions. Gene Phvul.001G186400, encode ―AGAMOUS-LIKE 62 (AGL62)‖, a type I MADS-box domain protein involved in endosperm cellularization, directly influencing seed size, sink strength and seed weigh and gene Phvul.001G241200, encodes 61 ―PLASMA MEMBRANE INTRINSIC PROTEIN 1;4 (PIP1;4,PIP1E) involved with movement of water and small molecules across membrane. Therefore, validation of markers linked to these QTL could increase genetic gain for the seed yield and partitioning through marker assisted breeding 62 Introduction Common bean (Phaseolus vulgaris L) is the most important grain legume consumed worldwide (Beebe et al., 2013). Common bean is an important food and nutritional security crop providing a cheap of source proteins (65%), energy (32%), and vital micro nutrients (Welch et al., 2000), in many developing countries. Bean is particularly important in Rwanda and western Kenya where highest global per capita consumption of 60kg have been reported (Akibode and Maredia, 2011; Namugwanya et al., 2014). Common bean production is constrained by many abiotic and biotic stresses. Drought stress is one of the most devastating cause of crop failures worldwide (Pennisi, 2008). It is estimated that over 60% of common bean production areas are prone to drought stress globally, and this makes drought the most important yield limiting abiotic stress of common bean (Beebe et al., 2013). Many bean-producing areas in sub-Saharan Africa are prone to both intermittent and terminal drought stress. In addition, common bean is cultivated by smallholder farmers who do not have irrigation infrastructure to supplement natural rainfall thus exacerbating the effect of drought (Beebe et al., 2011; Katungi et al., 2011; Beebe et al., 2014). In recent decades, drought stress is widely recognized as an important production constraint and improving drought tolerance in common beans has been a major breeding goal for both international and national bean improvement programs worldwide (Beebe et al., 2012). However, progress towards the development of high yielding, and drought tolerant common bean cultivars has been limited because most traits that confer drought tolerance are quantitatively inherited with the low heritability, and significant genotype by environment (GxE) interactions that results in low predictability of genotype performance under variable environments (Cattivelli et al., 2008). 63 Developing common bean cultivars with improved tolerance to multiple stresses such as drought, diseases, and broad adaptability can be achieved by acquiring and transferring sufficient genetic diversity for the contributing traits into locally adapted genetic backgrounds. In common bean, the highest levels of drought tolerance has been identified among the race Durango from the semiarid central and northern highlands of Mexico (Singh et al., 1991; Terán and Singh, 2002; Muñoz-Perea et al., 2006), followed by races Mesoamerica and Jalisco (Singh, 2007). Under drought stress, Durango race genotypes exhibit both higher harvest index and seed yield (Singh, 2007), while the Mesoamerican race genotypes has improved seed filling traits in addition to seed yield (Beebe et al., 2008). The combination of Durango and Mesoamerican races have been a consistent genetic resource to improve and develop adapted drought tolerant germplasm for low land tropical environments (Frahm et al., 2004; Beebe et al., 2008; Mukeshimana et al., 2014). A number of breeding lines with high yield potential were selected for their ability to remobilize photosynthate more efficiently to grain under terminal drought (Beebe et al., 2008), including SEA5 that is widely used as parent (Mukeshimana et al., 2014). Although races Durango and Mesoamerica genotypes have been a reliable and widely used source of drought tolerance, introgression of drought tolerance and other favorable genes into the large seeded Andean gene pool widely cultivated in Africa and South America has been limited (Gepts and Bliss, 1985; Mumba and Galwey, 1999). This is due to cross incompatibility and linkage drag that results in low rates of recovery of valuble recombinant progenies (Beebe et al., 2013). Recent studies identified quantitative trait loci (QTL) associated with rooting patterns and root architectural traits, phenology, seed yield and yield components in the DOR364 x BAT477 intra- gene pool RIL population. Mukeshimana et al. (2014) identified QTL associated with seed yield, and phenology traits in the SEA5 x CAL96 inter-gene pool RIL population. More recently a 64 number of QTL with varying size effects associated with seed yield and yield components were identified under drought stress using a nested association mapping (NAM) population (Hoyos- Villegas et al., 2016) The NAM population was developed from crossing three small red-seeded beans namely SER48, SER94, and SER95 with the common parent ‗Merlot‘. Interestingly, previous QTL analysis have consistently reported the co-localization of phenology, yield and yield components, and root traits on the same chromosome regardless of mapping population and marker system, thus providing evidence of functionality for specific genes contributing to complex traits such as drought (Asfaw et al., 2012; Blair et al., 2012; Mukeshimana et al., 2014). However, most QTL analysis studies on drought tolerance in common bean have been conducted within the Middle American gene pool and inter-gene pool crosses (Schneider et al., 1997; Asfaw et al., 2012; Blair et al., 2012; Mukeshimana et al., 2014; Hoyos-Villegas et al., 2016). In order to explore additional diversity for drought tolerance alleles, QTL analyses for drought tolerance is critical in the Andean gene pool to unlock the genetic potential and map genomic regions for populations developed from crosses within the Andean gene pool. This research will provide opportunity to analyze the effect of genetic backgrounds on QTL expression and accelerate identification of alternative sources of drought tolerance genes within the Andean gene pool (Beebe et al., 2013). In addition, these efforts should facilitate developing a strategy for improving the large seeded Andean types widely grown in East Africa and South America (Mukeshimana et al., 2014). The objective of this study was to identify QTL for drought tolerance and related traits in a Portillo x Red Hawk mapping population and identify recombinant lines for use in future breeding of drought tolerant Andean genotypes for Uganda. 65 Materials and Methods Plant Materials A RIL population consisting of 97 F4:7 lines derived from crossing Portillo a red mottled, drought tolerant variety from Ecuador with Red Hawk, red kidney bean variety, susceptible to drought developed from Michigan State University. The 97 RILs were evaluated for drought stress in the field at the Mubuku Irrigation Experimental Station in Kasese for two seasons 2016 and 2017, and at the National Crop Resources Research Institute Namulonge in 2017 off-season planting in Uganda. The two seasons hereafter referred to as Kasese_2016, Kasese_2017, and Namulonge_2017 respectively. Mubuku Irrigation Experimental Station is located in Kasese Western Uganda at 0o16‘10‖N, 30o6‘9‖E at an elevation of 1330 m above sea level and receives 1000mm of rainfall annually that is unevenly distributed between months. The soils are characterized as sandy loam soils with a pH of 5.68, above the 5.5 critical pH levels for common beans. The National Crop Resources Research Institute, Namulonge located in Central Uganda at 0o31‘30‖N, 32o36‘54‖E at an elevation of 1160 meters above sea level. The average temperatures are 21.80C ranging from 15.80C to 27.90C. Namulonge receives 1300 mm of rainfall annually and soils are characterized as oxisols with a pH of 5.8,respectively. The results of soil nutrient analysis indicated an average low available phosphorus and nitrogen below the 15 ppm and 0.2% respectively in both sites and NPK was applied at a rate of 50 kg/ha to enhance soil nutrient levels. The RIL population was evaluated side by side under non-stress (NS) and drought stress (DS) conditions respectively. The 97 RILs were planted on one row plot of 2 m long and 0.5 m between rows for two seasons in Kasese and for one season at Namulonge in 2017. The experiments were planted in 10x12lattice design with two replications. Planting was done during mid-rainy to end of season to target terminal drought stress at the reproductive stage 66 of plant growth. In Mubuku, the first experiment was planted on 28 October 2016 and harvested on 15 February 2017 and the second field experiment on 15 May 2017 and harvested on 11 September 2017. In Mubuku, drip irrigation was used to irrigate both DS and NS to ensure good germination and plant establishment until flowering time. Irrigation was discontinued in the DS plots at flowering while supplemental irrigation was continued in the NS plots twice a week up to harvest maturity. The field experiment in Namulonge was planted on 18 June 2017, and harvested on 30 September 2017. Overhead sprinklers were used to irrigate both DS and NS to ensure good germination and plant establishment until flowering time. Supplemental irrigation was continued in the NS plots twice a week up to harvest maturity while irrigation was discontinued in the DS plots at flowering. In both locations and seasons, recommended agronomic procedures were followed. Data was collected for daily weather conditions including daily rainfall (mm) and temperature (C) during the growing seasons. Phenotypic data was collected on days to flowering (DF), days to maturity (HM). In order to determine shoot biomass, yield and yield component traits, six plants were sampled from every plot for every genotype and sun dried and used for measuring shoot biomass, pod weight per plant (PW), seed yield per plant. In both seasons, seed yield per plant (SY) was calculated after adjusting for moisture to 14%. Partitioning traits for each genotype were estimated as a percentage as follows. Harvest index (HI) for each genotype was calculated in percentage as the ratio of total seed weight to total shoot biomass at harvest. In order to ensure uniform sample size for data collection, pod- partitioning index (PPI) was estimated using shoot biomass at harvest instead of total shoot biomass dry weight at mid-pod filling as the ratio of pod weight at harvest to total shoot biomass. The severity of drought stress in each environment for yield was quantified using drought intensity index (DII) as 1–(Xds/Xns) where Xds and Xns is the mean value for yield under 67 drought stress and non-stress. Geometric mean (GM) of seed yield as (NS x DS)1/2 where DS and NS are seed yield for drought stress and no stress treatments, respectively (Fischer and Maurer, 1978). Phenotypic Data Analysis Statistical analyses for field data were performed using mixed models in SAS 9.4 (SAS Institute 2002). Normality tests were conducted for each trait measured in the field using PROC UNIVARIATE command in SAS. Analysis of variance was performed to determine the effects of water treatment on the RIL population using PROC MIXED command in SAS where genotype, water treatment, location, and seasons were considered fixed effects while replication and blocks were random effects. Pearson correlations among variables were determined using the PROC CORR command and histograms using R package. Heritability Broad sense heritability (H2) was calculated for days to flowering, days to maturity, shoot biomass per plant, pod weight per plant, seed yield per plant, harvest index, pod harvest index, pod-partitioning index on entry mean basis using the variance component method (Fehr et al., 1987) described as; Where σg 2 is the genetic variance, σge 2 variance associated with genotype x environment interaction, σe 2 experimental error, t number of environments, r number of replications. 68 Genotyping DNA was isolated from the leaves of the F4 RILs and parents in the greenhouse at MSU using a protocol described by (Cichy et al., 2015a). The DNA from the samples used in this study were genotyped using an Illumina BARC_Bean6K_3BeadChip with 5398 SNPs (Song et al., 2015) in the Soybean Genomics and Improvement USDA Laboratory (USDA-ARS, Beltsville Agricultural Research Center) in MD, USA. Genome Studio software from the Illumina, Inc was used to call SNPs. Genetic Map Construction The 5398 SNPs were filtered for polymorphism leaving 810 polymorphic markers. The SNPs were then used to construct a genetic linkage map with the JoinMap software v4.1 (Van Ooijen, 2011). In JoinMap additional filtering was conducted to remove SNPs showing severe segregation distortion and markers mapping to the same position. Markers were then assigned to linkage groups using a logarithm of odds (LOD) score threshold varying from 5 to 10. Markers were ordered within linkage groups using the regression mapping algorithm and map distances between markers was estimated using Kosambi mapping function implemented in JoinMap and linkage maps were displayed using Map Chart software (Voorrips, 2002) Quantitative Trait Loci (QTL) Analysis Quantitative trait loci analysis was implemented using composite interval mapping (CIM) method in Win QTL Cartographer V2.5–011 (Wang et al., 2012). The CIM mapping control parameters included model 6 (standard model) with five markers as control/ background markers. Window size of 10cM and 1 cM walk speed. Analysis was conducted using forward and backward multiple regression method for every trait. In addition, a QTL was declared 69 significant based on 1000 permutation test (Churchill and Doerge, 1994) for each trait in Win QTL Cartographer to determine the LOD threshold at P=0.05 and the peak LOD score was considered the location of the QTL. The proportion of phenotypic variance accounted for by the QTL was determined using the coefficient of determination (R2) value. Quantitative trait loci were named using guidelines provided by the Genetics Committee of Bean Improvement Cooperative for common bean QTL nomenclature (Miklas and Porch, 2010) Results The effect of drought stress on performance of the RIL population The average daily precipitation and temperature recorded for 2016 growing season at Kasese, and 2017 growing season at Kasese and Namulonge were reported in Figure (2.1 a, b&c). In 2016 growing season (October-February), Kasese received a total of 150 mm of precipitation. Meanwhile during the 2017 (May-September) growing season a total of 180 mm of precipitation was recorded in Kasese. In Namulonge 2017, higher rainfall was recorded during the off-season planting with a total of 216 mm of precipitation (Figure 2.1c). In both growing seasons, the rainfall patterns were erratic and on average, most rains were recorded during the vegetative stage of growth (130 mm and 98mm) in 2016 and 2017 respectively. The results also indicated that little rains (22 mm and 16 mm) were recorded during reproductive and pod filling stages, and hence for conducive for conducting the field experiments on terminal drought stress (Figure 2.1 a&b). In addition, the total amount of precipitation recorded in this study is far below the minimum 400 mm required for normal growth of common beans. In this study, the overall DII for SY was 0.40, which is moderate. In Kasese, DII for SY was 0.44 in 2016 and 0.40 in 2017 respectively. In Namulonge 2017, DII for SY 0.36 in 2017, implying 70 that drought effects on seed yield per plant among the RIL population were slightly higher in Kasese than in Namulonge. In 2016, significant differences (P<0.01) were observed for drought effects in Kasese for DM, HI, PPI, PW, and SY (Table 3.1). Average HM were 73 d under DS and 75 d under NS, implying that plots under DS matured 3d earlier than in NS (Figure 3.1a). In this study, yield component traits were significantly affected by drought stress. The average PW were 21.9g and 32.9g under DS and NS resulting in 33.4% reduction in PW, while average seed SY were 9.9g and 18.1g under DS and NS conditions respectively, resulting in 45.3% reduction in SY under terminal drought (Figure 3.1b). . Similarly, partitioning indices were affected by drought stress resulting in 26% reduction in HI and PPI under DS. HI ranged from 2.1% to 37.2% under DS while it ranged 3.6 % to 52.6% under NS. Mean HI of 15.3% and 20.9% were reordered under DS and NS respectively. In addition, PPI ranged from 3.1% to 56.2%, average of 33.0% under DS while it ranged from 10.5% to 80.9 with an average of 44.6% under NS (Figure 3.1c). . In 2017, significant differences (P<0.01) for drought effects were observed for DM, HI, PPI, PW, and SY, and non-significant difference were observed for DF in Kasese and Namulonge (Table 3.1). In Kasese 2017, the mean HM was 73 d under DS and 77 d under NS, implying that DS plots matured 4 d earlier than NS. Yield component traits were significantly affected by drought stress. The mean seed yield per plant was 10.5g and 17.7g under DS and NS resulting in 40.7% reduction in SY due to drought stress. Average PW were 26.2g and 36.1g under DS and NS resulting in 27.4% reduction in PW. In Namulonge 2017, mean HM were 74 d under DS and 77 d under NS and suggesting that DS plots matured 3 d earlier than NS in this location. Significant drought effects were observed for 71 yield component traits such as SY and PW. The mean SY were 12.6g and 19.1g under DS and NS causing a 34 % reduction in SY due to drought stress. Average PW were 21.4g and 28.3g under DS and NS resulting in 24 % reduction in PW. Similarly, partitioning indices were affected by drought stress in Namulonge, HI were generally low with mean values of 13.2 and 16.5 under DS and NS resulting in 20% reduction in HI while average PPI were 25.5 and 28.7% under DS and NS, respectively causing 10% decrease due to DS (Table 3.1). Significant genotype by water treatments effects (GxE) across the two locations were observed for HM, HI, PW, BM and SY while non-significant GxE effects were observed for DF and PPI. This implies that the three traits namely DF and PPI were independent of the water treatments effects across the two locations. Correlation analysis revealed high significant and positive correlations among related traits under DS. As expected positive correlations were observed among yield component traits such as SY and PW (r=0.84, p<0.001). Positive correlations were observed between HI and PPI (r=0.82, p<0.001). In addition, yield component traits were positively correlated with partitioning indices. SY was correlated with HI (r=0.62, p<0.001) and PPI (r=0.5, p<0.001). PW was moderately correlated with HI (r=0.29, p<0.01) and PPI (r=0.23, p<0.05) as in Table 3.2. Genetic Map and QTL Analysis The Portillo and Red Hawk parents and the RIL population were genotyped using BARC_Bean6K_3BeadChip with 5398 markers. A total of 810 SNP markers were polymorphic between the parents. Additional filtering was conducted to remove markers with severe segregation distortion and those that mapped to the same position using JoinMap 4.1. A total of 206 markers were used to build the linkage map. In this study, 10 out of 11 linkage groups representing the 10 chromosomes in common bean were constructed with exception of 72 chromosome Pv05 that lacked polymorphic markers between the two parents and was excluded from analysis. The Portillo x Red Hawk map spanned 766.5 cM and number of markers per linkage group ranged from 12 on Pv08 to 31 on Pv11. The largest linkage group was Pv01 with 167.3 cM and the shortest linkage group was Pv10 with 30.07 cM. The average number of SNPs per linkage group was 20 and the mean genetic distance between markers was 3.7 cM (Table 3.3). Generally the order of linkage groups and orientation agreed with those in the Stampede x Red Hawk linkage map (Song et al., 2015). In this study, 32 QTL were identified using Composite Interval Mapping (CIM) method of Win QTL Cartographer with LOD thresholds calculated using 1000 permutations. Significant QTL signals for phenology traits (DF and HM), yield component traits (PW and SY), and partitioning traits (HI and PPI) were detected on eight chromosomes namely Pv01, Pv02, Pv03, Pv04, Pv06, Pv07, Pv08, Pv10, and Pv11 (Tables 3.4, 5 &6, Figure 3.2). Quantitative trait loci associated with phenology traits mapped to Pv03, Pv06, Pv07, Pv10, and Pv11, while yield components and partitioning traits mapped to Pv01, Pv02, Pv03, Pv04, Pv06, Pv08, Pv10, and Pv11. It is worth noting that colocalized QTL signals were detected on five chromosomes namely Pv01, Pv02, Pv03, Pv04, Pv06, and Pv11. Days to flowering Four QTL for DF were identified in this study, three under DS and one under NS conditions on Pv03, Pv06, Pv10, and Pv11 respectively. The first QTL identified on Pv03 under NS condition, in Namulonge_2017 spanned a region of 56.5 to 69.0 cM. This QTL was flanked by markers ss715639424 (40.16 Mb) and ss715639424 (41.29 Mb) with a peak region at 67.0 cM, and the nearest marker was ss715639424 (40.16 Mb). The LOD score associated with the QTL was 3.9, additive allelic effect of 1.3 days, and explained 48% variation in DF under NS conditions. The 73 additive allelic effect associated with this QTL on DF was 1.3 days (Table 3.4). The second QTL for DF identified from combined analysis under DS data on Pv06 had a peak at 6.4 cM. The nearest marker associated the QTL was ss715647843 with a physical position of 6.53 Mb and accounted for 6.1% overall variation in DF under DS. The LOD score for the QTL was 3.05 and additive allelic effect on DF was -0.2 day. The third QTL signal for DF identified under DS in Kasese_2016 on Pv10 spanned a region of 18.07 to 19.7 cM, and was flanked by markers ss715649029 (37.74 Mb) and ss715639432 (38.17 Mb), with a peak position at 19.4 cM. The nearest marker was ss715646963 located at 38.09Mb. The QTL had a LOD score of 3.19 and explained 6.4% variation in DF under DS and additive allelic effect of0.1 day on DF. The fourth QTL identified under DS from combined data and in Kasese_2017 on Pv11, was detected from 10.6 to 19.69 cM with a peak at 15.7 cM, and was flanked by markers ss715649044 (1.98 Mb) and ss715646273 (2.64 Mb) in Kasese_2017. A final QTL signal for DF was detected on Pv11 in combined data and spanned 14.6 to18.6 cM interval with a peak region at 16.7 cM. The nearest marker to the peak regions was ss715646273 with a physical position of 2.64 Mb. The LOD scores associated with the QTL were 3.54 and 3.19, with R2 value of 19.4% and 16.4%, and the additive allelic effects on DF were -1.0 and -0.7 days respectively under DS conditions in Kasese_2017 and the combined analysis, respectively (Table 3.4, Figure 3.2). Days to harvest maturity Two significant QTL signals for HM were identified under DS and NS conditions on Pv03 and Pv07. In Namulonge_2017, a QTL for HM was detected on Pv03 under NS condition, and spanned a region from 65.0 to 67.0 cM with a peak at 65.0 cM. The nearest marker was ss715639424 with a physical position of 40.16 Mb. This QTL had a LOD score of was 3.40 and 74 explained 31.4% variation, with an additive allelic effect of 1.2 day on HM under NS conditions. The second QTL was identified under DS in combined data and Kasese_2016 on Pv07 at 46.7 cM between markers. ss715649609 (11.06 Mb), and ss715646773 (44.62 Mb).The nearest marker to the peak region was ss715646773 (44.62 Mb). The LOD scores associated with this QTL on Pv07 were 3.64 and 3.24, explaining 7.5% and 1.48% variation, the additive allelic effects associated with HM7.1 on HM were 0.03days and -1.2days under DS conditions in combined data analysis and Kasese_2016 respectively (Table 3.4, Figure 3.2). Pod weight per plant Six QTL were identified for PW on Pv01, Pv02, Pv03, Pv04, and Pv08. Significant QTL signal for PW was detected on Pv01 in Kasese_2016 under DS at 88.5 cM with a physical position of 45.15Mb. The LOD score associated with this QTL was 3.29 and explained 18.7% of the variation in PW with an additive allelic effect of -5.01g (Table 2.4).The second QTL signal was detected on Pv02 under DS in combined data, and spanned 0.01 and 4.01 cM with a peak region at 0.01cM. The nearest marker to the peak was ss715649478 with a physical position of 0.12 Mb. The LOD score for the QTL was 3.17 and explained 14.6% variation. In addition, the additive allelic effect associated with the QTL on PW was 4.11g (Table 3.4, Figure 3.2). Another significant QTL signal for PW was detected on Pv03 in combined data under NS conditions. This QTL was located at 38.0cM tagged by marker ss715646441 at 47.81Mb, with a LOD score of 3.86, and accounted for 10.7% variation in PW. The additive allelic effect associated with this QTL signal was 3.7g. Two significant QTL signals were detected for PW on Pv04 under both DS and NS conditions in Kasese_2016. The first QTL signal was detected at 38.9 cM near marker ss715646215 at 2.67 Mb under NS conditions. This QTL had a LOD score of 3.12 with an additive allelic effect of -24.3g and explained 13.3% variation in PW. The second 75 QTL on Pv04 identified under DS conditions, spanning 53.7 to 54.9 cM with a peak at 53.8cM. The nearest marker to the peak region was ss715646781 (3.95Mb), with a LOD score of 4.77. The additive allelic effect associated with this signal was 1.6g and explained 5.9% variation in PW. Another QTL signal for PW was detected on Pv08 under DS conditions in combined data and spanned 32.8 to 34.7 cM with a peak at 34.3 cM tagged by marker ss715646880 with physical position of 1.21 Mb. The LOD score for this QTL was 3.33, with an additive allelic effect of 5.3g and explained 2.8% variation in PW. Seed yield per plant Seven QTL signals were detected for SY in this study on Pv01, Pv02, Pv03, Pv04, and Pv06. The first QTL for SY was detected on Pv01 in Kasese_2016 under both DS and NS conditions. The SY QTL detected under DS conditions spanned a region between 79.9 and 86.5 cM, and was flanked by markers ss715648564 and ss715646076, respectively. The peak region was located at 82.5cM and the nearest maker was ss715646076 with a physical position of 45.15 Mb. The LOD score for the QTL was 3.7 with an additive allelic effect of -3.21g and explained 25.1% variation in SY under DS conditions (Table 3.5, Figure 3.2). A second QTL signal for SY was detected on Pv01 under NS conditions. This QTL had a peak located at 79.9 cM and the nearest marker was ss715648564 located at 44.50 Mb. The LOD score for this peak was 3.15, and the additive allelic effect was -6.09g and explained 21.6% variation in SY under NS conditions. Two QTL for SY were detected on Pv02 under NS and DS conditions, respectively. The first QTL was identified under DS from combined data, and spanned a region of 0.01 to 4.0 cM with a peak at 0.01 cM. The nearest marker to the peak region was ss715649478 with a physical position of 0.12 Mb. The LOD score associated the QTL was 3.3, and the additive allelic effect of 2.36g on SY, accounting for 14.03 % variation. The second QTL for SY was identified on 76 Pv02 under NS conditions in Kasese_2017 and spanned a region of 72.3 to 74.39 cM with a peak at 73.4 cM. The nearest marker to the peak region was ss715649647 with a physical position of 39.15Mb. The LOD score associated with the QTL was 3.27 and explained 1.9% variation in SY with an additive allelic effect of 0.61g. Two QTL signals for SY were detected on Pv03 under both NS and DS conditions. The first QTL spanned 0.01 to 4.2 cM under NS in Namulonge_2017 and combined data under DS. The peak regions associated with these regions were 4.21 cM in and 0.01 cM respectively (Table 2.5). The QTL detected under NS conditions in Namulonge_2017 spanned 4.01cM to 4.24 cM was flanked by markers ss715648183 (0.56 Mb) and ss715649302 (0.06 Mb), the LOD score associated with this QTL signal was 3.11 and additive allelic effect of -0.07g, explaining 3.96% variation in SY. Meanwhile the QTL signal SY under DS in combined data spanned 0.01 to 4.24 cM with a peak region at 0.01 cM tagged by marker ss715648183 at 0.56 Mb. The LOD score for this peak was 11.59, with an additive allelic effect of 2.4g, explaining 13.7% variation in SY. The second QTL signal for SY on Pv03 was detected in Kasese_2017 under DS and spanned 59.5 to 73.2 cM with a peak region at 70.0 cM. The nearest marker to this peak region was ss715639424 (40.16 Mb). The LOD score associated with this QTL signal was 3.45 and additive allelic effect of 2.38g explaining 9.7% variation in SY under DS. One QTL signal was detected for SY on Pv04 under DS in Kasese_2017 and spanned a region of 32.2 to 57.3 cM with a peak at 50.9 cM. The nearest marker to the peak region was ss715640609 located at 29.17 Mb. The LOD score associated with this QTL was 7.04 and explained 36.87% variation in SY under DS condition with additive allelic effects of 6.4g on SY (Table 3.5, Figure 3.2). 77 One QTL was detected for SY on Pv06 under DS conditions in Kasese_2016 and spanned a region of 8.3 to 15.9 cM flanked by markers ss715647843 (6.53 Mb) and ss715649019 (17.92 Mb) with a peak region at 8.9 cM. The nearest marker to the peak region was ss715649019 located at 17.92 Mb with a LOD score of 4.10, and a large R2 value of 26.5% and additive effect of 2.6g on SY. Harvest index (HI) Six QTL signals for HI were identified in this study on Pv01, Pv03, Pv04, Pv06, Pv08, and Pv11. The first QTL for HI was detected under DS in combined data and Kasese_2017 on Pv01. In combined data, the QTL region spanned 137.4 to 151.5 cM with a peak at 148.4 cM, whereas the same QTL under DS in Kasese_2017 spanned 130.1 to 149.5 cM with a peak region at 143.4 cM. The QTL signal was flanked by markers ss715650354 (52.15 Mb) and ss715645295 (51.55 Mb), respectively, and the nearest marker to the peak regions was ss715650354 located at 52.15 Mb. The LOD scores for the QTL signals were 4.91 and 6.28 with R2 value of 6.76% and 4.7% under DS in combined and Kasese_2017 respectively. The additive effects associated this QTL on HI were 1.06 and 1.69 respectively (Table 3.6, Figure 3.2). The second QTL detected on Pv03 in Kasese_2017 under DS conditions, spanned 71.0 to78.8 cM with a peak located at 74.8 cM. The nearest marker to this QTL was ss715649460 at 3.29 Mb, and the LOD score of 3.94 with R2 of 5.12%. The additive allelic effects associated with the QTL on HI was -3.74. One QTL for HI was detected on Pv04 under DS in Kasese_2016, and spanned a region of 60.9 to 62.9 cM with a peak at 61.9 cM. The nearest marker to the peak region was ss715648923 with a physical location of 45.79 Mb. The LOD score associated with the peak region was 4.22 and R2 value of 2.53, with additive effects of -0.99 on HI. One of the QTL signal was detected on Pv06 for HI in Kasese_2016 under DS condition, and spanned a region of 8.9 to 10.9 cM with a peak 78 region at 8.9 cM. The nearest marker to this region was ss715649019 located at 17.92 Mb, with a LOD score of 3.01 and explained 23.3% variation in HI, the additive allelic effect of this QTL on HI was 3.09 (Table 3.6, Figure 3.2). Another significant QTL signal for HI was detected on Pv08 under NS conditions in Kasese_2016 and spanned a region between 2.3 and 8.3 cM with a peak at 4.3 cM. The nearest marker to the peak region was ss715641300 with a physical location of 19.36 Mb. The LOD score for this QTL was 3.35 and explained 2.58% variation in HI, and additive allelic effect of 1.72 on HI. A significant QTL signal for HI was detected on Pv11 under NS conditions in Kasese_2017, located between 24.9 and 26.9 cM with a peak region at 24.9 cM. The nearest marker to the peak region was ss715645541 with a physical position of 4.31 Mb. The LOD score associated with this QTL signal was 3.31 and explained 7.05% variation in HI, with an additive allelic effect of 2.7 (Table 3.6, Figure 3.2). Pod partitioning index (PPI) Six QTL were detected for PPI on Pv01, Pv02, Pv04, and Pv11 under DS and NS conditions. The first QTL signal for PPI was detected in Namulonge_2017 under DS conditions and spanned a region of 59.6 to 63.1 cM, and was flanked by markers ss715648275 and ss715647963 with a peak at 61.0 cM. The nearest marker to the peak region was ss715648382 located at 6.67 Mb. The LOD score for this QTL was 3.52 and explained 11.45% variation in PPI with additive allelic effect of -4.44. The second QTL for PPI on Pv01 was identified in combined data under DS conditions and was located between 135.4 and 140.0 cM with a peak region at 138.4 cM. The nearest marker to the peak region was ss715645258 located at 50.15 Mb, and the LOD score was 3.09 with R2 value of 3.18, and additive allelic effects of 2.92 on PPI (Table 3.6, Figure 3.2). 79 A third QTL for PPI was detected on Pv02 under DS in Kasese_2017 and spanned a region of 19.1 to 21.7 cM, flanked by markers ss715639436 and ss715647234 with a peak region located at 20.1 cM. The nearest marker to the peak region was ss715647234 with a physical position of 2.52 Mb. The LOD score associated was 3.04 and explained 10.44% of the variation in PPI and the additive allelic effect of 2.06 under DS conditions. Another QTL signal was detected on Pv04 under DS conditions in Kasese_2017 and spanned a region of 59.9 to 60.9 cM, and was located between markers ss715648130 and ss715648923 with a peak at 59.9 cM. The nearby marker to the peak region was ss715648923 located at 45.79 Mb. The LOD score associated this QTL was 3.70 and R2 value of 1.8%. The additive allelic effect associated the QTL was -3.9 (Table 3.6, Figure 3.2). Two QTL signals were detected on Pv11 for PPI were identified under NS conditions in Kasese_2016 and Kasese_2017respectively. The first QTL was detected in Kasese_2017 and spanned a region between 20.6 and 23.4 cM flanked by markers ss715646273 and ss715648851. The peak region was 23.5 cM with marker ss715648851 located at 5.19 Mb as the nearest marker. This QTL had a LOD score of 3.68 and explained 17.56 % variation in PPI and with additive allelic effect of 4.66 on PPI. The second QTL detected on Pv11 for PPI was located at 45.63 cM as the peak region, and the nearest marker was ss715650599 with a physical position of 45.63 Mb. This QTL had a LOD score of 3.26 and explained 14% variation in PPI. The additive allelic effect on PPI was -3.42 under NS conditions (Table 3.6, Figure 3.2). Colocalized QTL Regions In this study, colocalized signals were detected for phenology, yield component traits and partitioning indices on Pv01, Pv02, Pv03, Pv04, Pv06, and Pv11. Interestingly, the lower arm of 80 chromosome Pv01 was the peak region for QTL associated with yield component traits namely PW and SY under NS conditions and for partitioning traits HI and PPI under DS. The nearest markers flanking the peak regions were ss715648564 at 44.50Mb and ss715646075 at 45.15 Mb, respectively. In addition, the upper arm of chromosome Pv02 was the peak region for significant QTL signals for PW and SY detected under DS conditions in combined analysis, and tagged by SNP ss715649478 with a physical position of 0.12 Mb. Another colocalized QTL region was identified on Pv03 for DF and HM under NS conditions in Namulonge_2017 and for SY under DS conditions in Kasese_2017.The nearest common SNP to the peak region was ss715639424 located at 40.16 Mb. Colocalized QTL signals for partitioning traits namely HI and PPI were identified on Pv04 under DS conditions and the common marker to the peak region was ss715648923 located at 45.79 Mb. In addition, another colocalized QTL signal was detected on Pv06 HI and SY under DS conditions and the common marker associated with the peak region was ss715649019 located at 17.92 Mb (Figure 3.6, Figure 3.2). Discussion In the current study, a RIL mapping population was developed from a cross between Portillo a red mottled, drought tolerant Andean variety from Ecuador and Red Hawk, red kidney bean variety, susceptible to drought. Both parents from the Andean gene pool were contrasting for drought tolerance and for seed color patterns. The population was evaluated to identify QTL associated with drought tolerance traits at the pod filling stage. Significant differences (P<0.01) drought effects were observed for all agronomic traits except for DF. Significant genotype by water treatments effects (GxE) were also observed across the two locations except for DF and PPI. This implies that the three traits namely DF, PHI and PPI were independent of the water 81 treatments effects across the two locations. Mukeshimana et al. (2014) observed similar GxE effects on DF and partitioning indices in SEA5 x CAL96 mapping population evaluated in Rwanda and Colombia. As expected positive correlations were observed among yield component traits and partitioning traits, consistent with previous studies (Ramirez-Vallejo and Kelly, 1998; Mukeshimana et al., 2014), suggesting that mechanisms associated with drought responses have pleotropic effects on traits contributing to performance under drought in this population. In this study, QTL for phenology traits mapped to chromosomes Pv03, Pv06, Pv07, Pv10 and Pv11. The QTL for DF designated as DF3.2PR was detected on Pv03 NS condition in Namulonge_2017 and explained 48% variation. The nearest marker to the peak region was ss715639424 located at 40.16 Mb. Previous studies mapped QTL for DF on Pv03 (Checa and Blair, 2012; Brinez et al., 2017). The QTL DF3.1GG flanked by markers BM197 and BM181B on Pv03 for DF was identified in F5:8 derived G2333 × G19839 RIL population evaluated under high phosphorus in Darién (Checa and Blair, 2012). Another QTL DF3.1AS was reported in SEA 5 x AND 277 population under NS conditions (Brinez et al., 2017), and spanned 74.5 to159.2 cM on Pv03 tagged by marker BAR3097. However, comparison of DF3.1GG and DF3.1AS QTL reported on Pv03 was not possible due to lack of physical positions for markers BM197, BM181B, and BAR3097. Additionally, QTL for DF mapped to Pv06, Pv10, and Pv11 under DS conditions. Previous studies reported QTL for DF on Pv06 (Blair et al., 2006; Perez-Vega et al., 2010; Blair et al., 2012), and Pv11 (Tar'an et al., 2002; Blair et al., 2006; Kwak et al., 2008; Blair et al., 2012). The QTL DF6.1CG and DF6.2CG on Pv06 flanked by markers BM170 (25.09 Mb)-V and V-BM187(20.25 Mb) were reported in advanced backcross a mapping population derived from Andean cultivar ICA Cerinza and G24404, a wild species of common bean (Blair et al., 82 2006). Other QTL DF6.1DB and DF6.2DB were identified in DOR364 x BAT477 intra-gene pool mapping population under DS and NS conditions. These QTL were tagged by marker M1601 located between markers BM278 (2.69 Mb) and BMb1157 (12.11 Mb). The QTL DF6.1PR identified in this study was near marker ss715647843 at 6.53 Mb under DS in combined data was within the marker interval for QTL reported by Blair et al. (2012) although the physical position of marker M1601 was not known. The QTL DF11.1DB and DF11.2DB for DF were detected in DOR364 x BAT477 under DS and NS conditions (Blair et al., 2012), and under moderate to low phosphorus condition (Diaz et al., 2017). These QTL were tagged by markers AB502 and P302 whose physical positions were not known but flanked by BMd43 (1.71 Mb) and BMb1074 (48.04 Mb) respectively. In addition, QTL DF11.1PR for DF tagged by ss715646273 at 2.64 Mb was most likely the same QTL DF11.1DB reported in previous studies (Blair et al., 2012; Diaz et al., 2017). In this study, QTL for DM were identified on Pv03 and Pv07 under NS and DS conditions. Previous studies consistently reported QTL for DM on Pv03 (Hoyos-Villegas et al., 2015) and on Pv07 (Blair et al., 2006; Mukeshimana et al., 2014). The QTL DM3.1PR was near marker ss715639424 at 40.16 Mb identified in this study was the same QTL DM3.1AP linked to marker ss715646619 at 39.50 Mb reported in AN-37 x P02630 population under white mold pressure (Hoyos-Villegas et al., 2015). The second QTL DM7.1PR was identified on Pv07 DS conditions and was tagged by marker ss715646773 at 44.6 Mb. The QTL DM7.1SC identified on Pv07 linked to marker ss715645222 at 51.21 Mb in SEA5 x CAL96 inter-gene pool cross under NS and DS conditions (Mukeshimana et al., 2014) were 6.6 Mb apart, implying that these are the same QTL. 83 In addition, marker ss715646773 (44.6 Mb) on Pv07 is 1.025 kb upstream of gene Phvul.007G207300 annotated as ―TETRATRICOPEPTIDE REPEAT (TPR)-LIKE SUPERFAMILY PROTEIN‘. Members of TPR regulate plant hormonal responses especially ethylene biosynthesis that directly regulates transition from reproductive stage to maturity (Schapire et al., 2006; Iqbal et al., 2017). In addition, Arabidopsis, TETRATRICOPEPTIDE- REPEAT THIOREDOXIN-LIKE 1 (TTL1) has been reported as a positive regulator of ABA signaling under drought and osmotic stresses (Rosado et al., 2006). Six QTL for PW were detected on Pv01, Pv02, Pv03, Pv04, and Pv08. Furthermore, correlation between PW and number of pods per plant in this study were positive and significant (r=0.67; p<0.001) (data not presented). Several QTL associated pod traits were previously reported on Pv01, Pv02, Pv03, and Pv04 (Beattie et al., 2003; Checa and Blair, 2012; Gonzalez et al., 2016; Diaz et al., 2017), and PW was mapped to Pv08 (Kamfwa et al., 2015). Interestingly, the marker ss715646880 at 1.21 Mb linked to QTL PW8.1PR on Pv08 in this study were in LD (r2 > 0.65; D' >0.85), with two marker ss715649359 (4.74 Mb) and ss715639408 (5.15 Mb) that were significantly associated with PW in Andean diversity panel (Kamfwa et al., 2015), suggesting the same genomic region is involved. Seven QTL signals were detected for SY in this study on Pv01, Pv02, Pv03, Pv04, and Pv06 under DS and NS conditions. This is consistent with previous studies that reported significant QTL signals for SY on the same chromosomes. In this study, QTL SY1.1PR was detected on Pv01 under both NS and DS condition near markers ss715646076 at 45.15 Mb and ss715648564 at 44.50 Mb respectively. SY QTL were consistently reported on Pv01 and Pv02 in Buster x Roza and DOR364 x BAT477 mapping populations evaluated under NS and DS conditions in multi-locations for several years (Asfaw et al., 2012; Trapp et al., 2015). Trapp et al. (2015) 84 reported SY1.1BR at 47.7 Mb near marker SNP50809 in Buster x Roza mapping population under multiple stress environments, suggesting these two QTL were the same. In addition, Asfaw et al. (2012) reported the same QTL signal for SY on Pv1 under NS tagged by marker BM200 at 30.8 Mb. The QTL SY2.1BR for SY on Pv02 was identified in Buster x Roza mapping population at 11.8 Mb near marker SNP40055 (Trapp et al., 2015). This QTL was 11.7 Mb downstream of SY2.1PR on Pv02 near marker ss715649478 (0.12 Mb) identified in the current study under DS in combined data. In addition, the second QTL SY2.2PR for SY identified on Pv02 near marker ss715649647 (39.15 Mb) in this study was near QTL YD2.1DB reported under low phosphorus condition tagged by BMa16 (33.18 Mb) in DOR364 x BAT477 mapping populations (Diaz et al., 2017). The marker ss715649647 (39.15 Mb) was not in LD with markers within the same genomic region as BMa16 (33.18 Mb) implying that SY QTL SY2.1BR and YD2.1DB were different as verified in this study. Two QTL signals were detected for SY on Pv03 in the current study. This is consistent with several previous studies that reported QTL for yield on Pv03 (Wright and Kelly, 2011; Blair et al., 2012; Checa and Blair, 2012; Mukeshimana et al., 2014; Hoyos-Villegas et al., 2015; Kamfwa et al., 2015; Heilig et al., 2017). The QTL YLD3.1CG linked to marker BM172 and YLD3.2CG near marker BM98 for yield were detected in Popayán using Cerinza x G24404 advanced backcross population (Blair et al., 2006). Yield QTL YLD3.1GG, YLD3.2GG linked to markers BM197 and AG01 were reported in G2333 x G19839 mapping population in Popayán (Checa and Blair, 2012), and QTL YLD3.1DB, YLD3.2DB , YLD3.3DB , YLD3.4DB near markers AE103, Q1001, BM181, and AD1801 respectively were detected on Pv03 under NS conditions in DOR364 x BAT477 mapping population (Blair et al., 2012). However, yield QTL reported in 85 many of the previous studies could not verified due to unknown physical positions of markers near those QTL (Blair et al., 2006; Wright and Kelly, 2011; Blair et al., 2012; Checa and Blair, 2012). The SY3.3PR for SY on Pv03 was identified under DS conditions in Kasese_2017 linked to ss715639424 at 40.16 Mb. The SY QTL on Pv03 has been consistently reported in several studies (Kelly, 2018). The QTL SY3.3SC on Pv03 was initially reported in SEA5 x CAL96 mapping population under DS and NS conditions (Mukeshimana et al., 2014). A subsequent study reported significant SNP association for SY on Pv03 associated with marker ss715648538 at 38.27 Mb using Andean beans under low-N conditions in Michigan (Kamfwa et al., 2015). Interestingly, the significant marker ss715648538 was in strong LD with markers within SY3.3SC interval, implying that the gene(s) underlying the QTL for seed yield reported in the two studies was the same (Mukeshimana et al., 2014; Kamfwa et al., 2015; Kelly, 2018). In addition, two separate studies mapped SY3.3 on Pv03 near markers ss715639345 (39.6 Mb) and ss715646621 (39.4 Mb) in Puebla 152 x Zorro mapping population (Heilig et al., 2017). The same QTL was mapped between markers ss715647671 (37.06 Mb) and ss715639244 (45.59 Mb) using a NAM population of Merlot with three small red-seeded bean lines (S48M, S94M, and S95M) under DS conditions (Hoyos-Villegas et al., 2016), further validating SY QTL SY3.3. The second SY QTL SY3.4PR was identified on Pv03 under DS and NS conditions near marker ss715648183 at 0.56 Mb. This QTL SY3.4PR at the upper arm of Pv03 could have been reported in previous studies but could not be compared due to lack of physical positions of markers linked to QTL reported in previous studies. The QTL SY4.1PR and SY6.1PR for SY were identified on Pv04 and Pv06 under DS conditions in Kasese_2016,and 2017 respectively, which is consistent with previous studies (Cichy et al., 2009; Pérez-Vega et al., 2010; Asfaw et al., 2012; Blair et al., 2012). The QTL SY6.1PR on Pv06 86 near marker ss715649019 at 17.92 Mb identified in this study was the same to YLD6.1DB detected in DOR364 x BAT477 mapping population for SY under DS on Pv06 tagged by marker M501, located between flanking markers BMb342 (13.44 Mb) and IAC115 (15.45 Mb). Significant QTL signals for partitioning traits namely HI and PPI were detected on Pv01, Pv02, Pv03, Pv04, Pv06, Pv08, and Pv11 under DS and NS conditions. Six QTL for HI were identified on Pv01, Pv03, Pv04, Pv06, Pv08, and Pv11. Asfaw et al., (2012) reported significant QTL signals for HI on Pv03, Pv04, Pv06 and Pv11 in DOR364 x BAT477 mapping population. Interestingly, the QTL HI6.1PR on Pv06 near ss715649019 at 17.92 Mb detected in the current study was within the marker interval for HI was reported on Pv06 near marker Y501 flanked by IAC115 (15.45 Mb) and BM187 (20.25 Mb) in DOR364 x BAT477 mapping population (Asfaw et al., 2012). Similarly, QTL H11.1PR on Pv11 tagged by ss715645541 at 4.31 Mb was near marker BMd43 (1.73 Mb) that was flanking the peak marker AB502 in the same mapping population (Asfaw et al., 2012), implying that the QTL reported in the two studies were the same. The QTL HI6.1PR on Pv06 is near ss715649019 at 17.92 Mb. Six significant QTL signals for PPI were detected on Pv01, Pv02, Pv04, and Pv11. Several significant QTL signals for PPI were reported on Pv02, Pv04 and Pv11 using DOR364 x BAT477 mapping population (Asfaw et al., 2012). The QTL PPI11.1PR and PPI11.2PR detected for PPI were linked to markers ss715648851 at 5.19 Mb and ss715650599 at 45.63 Mb. The location is consistent with previous studies that detected QTL for PPI on Pv011 near marker AB502 flanked by BMd43 (1.73 Mb) and BMb1074 (48.04 Mb) in DOR364 x BAT477 mapping population (Asfaw et al., 2012), and near peak regions detected in this study. 87 Putative candidate genes in genomic regions associated with colocalized QTL peaks In this study, QTL signal for phenology trait colocalized on the lower arm of Pv03. DF3.2PR and DM3.1PR QTL were identified under NS conditions in Namulonge_2017 and the nearest common SNP to the peak region was ss715639424 located at 40.16 Mb. Within 31kb region upstream of ss715639424 SNP were two genes that could be considered positional candidates. Gene Phvul.003G189100 was 30.3kb upstream ss715639424 and encodes ―AGAMOUS-LIKE 65 (AGL65)‖, a MADS-box gene family protein. AGL65 has been reported as a regulator of pollen development through activation of mature pollen genes (Verelst et al., 2007). Additional, previous studies in soybean reported several AGAMOUS-LIKE genes as key regulators of floral organ identity, flowering time and seed development (Fan et al., 2013; Chen et al., 2014). Another important gene in this region was Phvul.003G189300, 17.9kb upstream of ss715639424 that encodes ―CYCLING DOF FACTOR 3 (CDF3)‖, a member of the DOF family transcriptional factor that regulates flowering time by transcriptional repressing the activity of ―CONSTANS (CO)‖ and ―FLOWERING LOCUS (FT)‖(Fornara et al., 2009; Fornara et al., 2015). The significant QTL signals detected Pv03 for DF and HM may contain genes that regulate the expression of plant flowering and maturity under NS conditions. In common bean, significant progress has been made to improve seed yield under drought stress particularly within the Middle American gene pool. However, selecting for stable seed yield under DS and NS conditions in the Andean gene pool has been a challenge. Recent studies reported the use of yield component traits and partitioning indices that positively contribute to seed yield stability under DS as surrogate traits for indirect selection of superior drought tolerant cultivars under NS and DS (Ramirez-Vallejo and Kelly, 1998; Rosales-Serna et al., 2004; Miklas et al., 2006; Beebe et al., 2008). 88 In addition, several studies reported that partitioning indices especially harvest index and pod harvest index as primary indicators of photosynthate remobilization potential from shoot biomass to yield under drought stress conditions (Rosales-Serna et al., 2004; Beebe et al., 2008; Assefa et al., 2013). In this study, significant and positive correlations were observed between and among yield component traits and partitioning indices (Table 7) under DS conditions, which is consistent with previous studies (Klaedtke et al., 2012; Mukeshimana et al., 2014). Significant QTL signals for yield component traits and partitioning traits colocalized on Pv01, Pv02, and Pv04 suggesting that genomic regions condition these traits were linked and could be used for indirect selection for yield in marker assisted breeding. . The seed yield QTL SY1.1PR that mapped to Pv01 was detected under NS and DS conditions in Kasese_2016. This QTL colocalized with partitioning trait QTL HI1.1PR identified under DS in the combined data and in Kasese_2017 and PPI1.2PR detected under DS in combined study. The significant SNPs associated with the peak QTL signals were ss715646076 (45.15 Mb) for SY1.1PR ss715650354 (52.15 Mb) for HI1.1PR and ss715645258 (50.15 Mb) for PPI1.2PR respectively. The SNPs were not in LD and spanned 1.99Mb (45.15 to 52.15 Mb) region containing 275 genes with diverse functional annotations. In order to gain insights into the genes associated with these traits in these common genomic regions, we performed functional annotations of positional candidate genes tagged by the significant SNP ss715646076 (45.15 Mb) for SY1.1PR. Two genes within 35kb region upstream of this marker were considered candidates. Gene Phvul.001G186400 is 19.2kb upstream of SNP ss715646076, annotated as ―AGAMOUS-LIKE 62 (AGL62)‖. AGL62 is a type I MADS-box domain protein that acts as a key regulator of endosperm cellularization, directly influencing seed size, sink strength and seed weight (Kang et al., 2008). The second gene is Phvul.001G186300 32kb upstream of SNP 89 ss715646076, annotated as ―HISTONE DEACETYLASE 3 (HDA3, HD2A)‖ a plant specific histone that has been reportedly involved in regulating gene expression during seed development and maturation in Arabidopsis (Zhou et al., 2004). While within 60kb region upstream of the significant marker ss715645258 (50.15 Mb) associated with QTL PPI1.2PR were three positional candidate genes. Gene Phvul.001G241700 is 18kb upstream of marker s71 5645258, annotated as ―ANTHRANILATE SYNTHASE ALPHA SUBUNIT 1 (AMT1, ASA1)‖, an enzyme catalyzing the rate limiting stage in the biosynthesis of tryptophan (Kreps et al., 1996; Tien Lea et al., 2016). Gene Phvul.001G241500 is 29kb upstream of marker s715645258 and encodes ―TRANSMEMBRANE AMINO ACID TRANSPORTER FAMILY PROTEIN (AUX1, MAP1)‖ an auxin-influx carrier protein that is associated with transport of auxin hormone (Yang et al., 2006; Basu et al., 2013). Candidate gene Phvul.001G241200 is 59kb upstream of SNP ss715645258 and encodes ―PLASMA MEMBRANE INTRINSIC PROTEIN 1;4 (PIP1;4,PIP1E)‖, a member of aquaporin family protein that is associated with movement of water and small molecules across membrane (Gomes et al., 2009). However, in Arabidopsis PIP1;4 has been reported to play a dual function in water movement especially under drought stress and C02 permeability across plasma membrane hence increasing photosynthetic output (Alexandersson et al., 2005; Li et al., 2015). Ariani and Gepts (2015) recently characterized aquaporin‘s in common bean into five sub families and (Hoyos-Villegas et al., 2017) reported candidate gene Phvul.011G102700 that encodes PvSIP1;3 gene, aquaporin on Pv11 during wilting in common bean suggesting their involvement in water transport. A second region of interest located on Pv02 was the peak region for QTL SY2.1BR and PW2.1PR detected under DS conditions in the combined analysis. This region was tagged by SNP ss715649478 with a physical position of 116.65 kb. SNP ss715649478 is within the genic region 90 of Phvul.002G001800 that is annotated as ―CATIONIC AMINO ACID TRANSPORTER 6 (CAT6)‖, a putative sink-specific amino acid transporter that is involved in phloem loading of amino acids (Su et al., 2004; Hammes et al., 2006). Recent studies reported the association of CAT6 in source–sink partitioning through import of amino acids into developing seed (Rentsch et al., 2007), suggesting possible involvement in partitioning of assimilates to yield components. The upper arm of chromosome Pv04 is the peak region for QTL PW4.2 and SY4.1 detected in combined data under DS. This region was tagged by marker ss715646781 located at 3.95 Mb within genomic region containing cluster of genes that encode UDP- GLYCOSYLTRANSFERASE SUPERFAMILY PROTEIN (UGT) that are involved in glycosylation of plant compounds (Wang and Hou, 2009). Conclusion The objective of this study was to map QTL associated drought tolerance within the Andean Portillo x Red Hawk mapping population. Significant differences were observed among the mapping population for phenology, yield component and partitioning traits. Thirty-two QTL were identified using CIM method for agronomic traits measured under DS and NS conditions in Uganda for two seasons and locations. Consistent with correlation analysis for agronomic traits, colocalized QTL signals were also identified for yield component traits and partitioning traits on Pv01, Pv02, Pv03, Pv04, Pv06 and Pv11. We performed functional annotation of positional candidate genes tagged by significant SNP markers within the colocalized peak region. We report two candidate genes, Phvul.003G189100, 30.3 kb upstream ss715639424, and encodes ―AGAMOUS-LIKE 65 (AGL65)‖, a MADS-box gene family protein a positive regulator of pollen development on Pv03. In addition, Phvul.003G189300, 17.9 kb upstream of ss715639424 and encodes ―CYCLING DOF FACTOR 3 (CDF3)‖, a member of the DOF family 91 transcriptional factor which acts as transcriptional repressor of two flowering time genes namely ―CONSTANS (CO)‖ and ―FLOWERING LOCUS (FT)‖. These candidate genes are involved in regulating flowering time; floral organ identity and development were associated with phenology and detected in the peak region for DF3.2PR and DM3.1PR under NS conditions In addition, we identified positional candidate genes on Pv01 that were significantly associated with QTL SY1.1PR and PPI1.2PR under both DS and NS conditions. Phvul.001G186400, 19.2 kb upstream of SNP ss715646076, annotated as ―AGAMOUS-LIKE 62 (AGL62)‖, a type I MADS- box domain protein that regulates endosperm cellularization, directly influencing seed size, sink strength and seed weight. Candidate gene Phvul.001G241500 annotated as ―TRANSMEMBRANE AMINO ACID TRANSPORTER FAMILY PROTEIN (AUX1, MAP1)‖ is involved in the transport of auxin hormone, and Phvul.001G241200, annotated as ―PLASMA MEMBRANE INTRINSIC PROTEIN 1;4 (PIP1;4,PIP1E) is associated with movement of water and small molecules across membrane. Interestingly the colocalized QTL signal for SY2.1BR and PW2.1PR on Pv02 was tagged by SNP ss715649478 (116.65 kb), within the genic region of Phvul.002G001800 and annotated as ―CATIONIC AMINO ACID TRANSPORTER 6 (CAT6)‖, a putative sink-specific amino acid transporter that is involved in phloem loading of amino acids into developing seed. The results from this study improve our understanding genetic control and association between seed development, sink-specific amino acid transport and biosynthesis, and water movement during reproductive stage of common bean under DS and the pleiotropic effects of genes associated with colocalized regions on the traits. 92 APPENDIX 93 Figure 3.1a Histograms of days to flowering and days to maturity for Portillo x Red Hawk common bean recombinant inbred line (RIL) population evaluated under drought stress (DS) and non-stress (NS) in Kasese and Namulonge, Uganda in 2016 and 2017 Po RH RH Po RH Red hawk, Po Portillo RH RH Po Po 94 Figure 3.1b Histograms of seed yield per plant and pod weight per plant for Portillo x Red Hawk common bean recombinant inbred line (RIL) population evaluated under drought stress (DS) and non-stress (NS) in Kasese and Namulonge, Uganda in 2016 and 2017 RH Po RH Po 95 RH Po RH Po Figure 3.1c Histograms of harvest index and pod partitioning index for Portillo x Red Hawk common bean recombinant inbred line (RIL) population evaluated under drought stress (DS) and non-stress (NS) in Kasese and Namulonge, Uganda in 2016 and 2017 RH RH Po Po RH Po RH Po 96 Table 3.1 Mean, range for yield component traits, phenology and portioning indices for Portillo x Red Hawk common bean recombinant inbred line (RIL) population evaluated under drought stress (DS) and non-stress (NS) in Kasese and Namulonge, Uganda in 2016 and 2017. Kasese _2016 Kasese _2017 Namulonge _2017 Treatment effects DS NS DS NS DS NS ANOVA Trait mean (range) mean (range) mean (range) mean (range) mean (range) mean (range) G T G*T H2 DF 36.7(33-40) 36.4(33-40) 38.7(32-43) 38.2(31-45) 38.1(33-43) 37.4(34-41) *** ns ns 0.87 HM 72.5(65-93) 75.5(67-91) 73.2(69-91) 77.5(68-85) 74.2(67-93) 77.1(70-88) ** * PW 21.9(7-70) 32.9(8-86) 26.2(5-86) 36.1(7- 45) 21.4(7-74) 28.3(7-87) * ** * * 0.79 0.74 9.9(3-32) 18.1(4-79) 10.5(4-51) 17.7(6-63) 12.6(4-60) 19.4(4-65) *** *** ** 0.61 SY HI 15.3(2-37) 20.9(4-53) 27.7(7-53) 30.4(12-44) 13.2(4-35) 16.5(4-29) PPI 33.0(3-56) 44.6(11-83) 47.2(15-72) 59.4(28-87) 25.6(6-61) 28.7(26-64) * * ** * * 0.52 ns 0.63 DF days to 50% flowering(d), HM days to maturity (d), PW pod weight per plant(g), SY seed yield per plant(g), HI harvest index, PPI pod partitioning index, DS drought stress, NS non-stress, G genotype, T treatment, G*T genotype x treatment interaction, H2 broad sense heritability . * Significant at the 0.05 probability level, ** Significant at the 0.01 probability level, *** Significant at the 0.001 probability level, and ns not significant. 97 Table 3.2 Pearson correlation coefficients among mean variables for days to flowering, days to maturity, harvest index, pod partitioning index , seed yield per plant, and pod weight per plant for the Portillo x Red Hawk recombinant inbred line (RIL) population evaluated under drought stress and non-stress in Kasese and Namulonge, Uganda during 2016 and 2017 growing seasons. DF DM HI PPI SY PW DF DM HI PPI SY PW 0.42*** -0.03ns -0.08ns -0.04ns -0.15ns 0.82*** -0.03ns -0.05ns 0.62*** 0.50*** -0.11ns 0.03ns 0.29** 0.24** 0.84*** DF days to 50% flowering(d), HM days to maturity(d), HI harvest index, PPI pod partitioning index, PW pod weight per plant(g), SY seed yield per plant(g). * Significant at the 0.05 probability level, ** Significant at the 0.01 probability level, *** Significant at the 0.001 probability level, and ns not significant. 98 Table 3.3 Single nucleotide polymorphism (SNP) marker distribution and distance on individual chromosomes of the genetic linkage map of Portillo x Red Hawk recombinant inbred line (RIL) population Chromosome No. of markers Length Mean distance(cM) Pv01 Pv02 Pv03 Pv04 Pv06 Pv07 Pv08 Pv09 Pv10 Pv11 Entire genome 20 16 22 23 14 22 12 31 20 26 206 167.33 103.39 86.51 87.46 56.89 68.88 46.15 71.79 30.07 48.03 766.5 8.4 6.5 3.9 3.8 4.1 3.1 3.8 2.3 1.5 1.8 3.7 99 Table 3.4 Quantitative trait loci (QTL) for days to 50% flowering(DF), days to maturity(DM) and pod weight per plant(PW) in the Portillo x Red Hawk recombinant inbred line (RIL) evaluated under drought stress and non-stress in Kasese and Namulonge, Uganda during 2016 and 2017 growing seasons. Trait QTL Chromosome Site Year Treatment Peak(cM) Near marker Peak(Mb) LOD Additive R2 DF3.2PR DF DF6.1PR DF10.1PR DF11.1PR DF11.1PR DM DM3.1PR DM7.1PR DM7.1PR PW PW1.1PR PW2.1PR PW3.1PR PW4.1PR PW4.2PR PW8.1PR Pv03 Pv06 Pv10 Pv11 Pv11 Pv03 Pv07 Pv07 Pv01 Pv02 Pv03 Pv04 Pv04 Pv08 Namulonge 2017 Combined Kasese 2016 Combined Kasese 2017 Namulonge 2017 Combined Kasese 2016 Kasese 2016 Combined Combined Kasese 2016 Kasese 2016 Combined NS DS DS DS DS NS DS DS DS DS NS DS NS DS 67.01 ss715639424 40.16 6.41 ss715647843 6.53 19.41 ss715646963 38.09 3.96 3.05 3.19 1.34 48.3 -0.23 0.10 6.1 6.4 16.71 ss715646273 15.71 ss715646273 2.64 2.64 3.19 -0.67 16.4 3.54 -1.04 19.4 65.01 ss715639424 40.16 46.71 ss715646773 44.62 46.61 ss715646773 44.62 3.40 3.64 3.24 1.19 0.03 -1.20 31.4 7.5 1.5 88.51 ss715646076 45.15 3.29 -5.01 18.7 0.01 ss715649478 0.12 38.01 ss715646441 47.81 53.81 ss715646781 38.91 ss715646215 34.31 ss715646880 3.95 2.67 1.21 3.17 3.86 4.77 4.11 3.74 1.61 14.6 10.7 5.9 3.12 -24.33 13.3 3.33 5.39 2.6 100 Table 3.5 Quantitative trait loci (QTL) for seed yield in the Portillo x Red Hawk recombinant inbred line (RIL) evaluated under drought stress and non-stress in Kasese and Namulonge, Uganda during 2016 and 2017 growing seasons. Trait QTL SW SY1.1PR SY1.1PR SY2.1BR SY2.2BR SY3.3PR SY3.4PR SY3.4PR SY4.1PR SY6.1PR Chromosome Site Year Treatment Peak(cM) Near marker Peak(Mb) LOD Additive R2 Pv01 Pv01 Pv02 Pv02 Pv03 Pv03 Pv03 Pv04 Pv06 Kasese 2016 Kasese 2016 Combined Kasese 2017 Kasese 2017 Namulonge 2017 Combined Kasese 2017 Kasese 2016 DS NS DS NS DS NS DS DS DS 82.51 ss715646076 45.15 79.91 ss715648564 44.50 0.01 ss715649478 0.12 73.41 ss715649647 39.15 70.01 ss715639424 40.16 4.21 0.01 ss715648183 ss715648183 0.56 0.56 50.91 ss715640609 29.17 8.91 ss715649019 17.92 3.65 3.15 3.30 3.27 3.45 3.11 11.59 7.04 4.10 -3.21 25.1 -6.09 21.6 2.36 0.61 2.38 -0.07 2.40 6.43 2.60 14.0 1.9 9.7 4.0 13.7 36.9 26.5 101 Table 3.6 Quantitative trait loci (QTL) for harvest index(HI) and pod partitioning index(PPI) in the Portillo x Red Hawk recombinant inbred line (RIL) evaluated under drought stress and non-stress in Kasese and Namulonge, Uganda during 2016 and 2017 growing seasons. Trait QTL Chromosome Site Year Treatment Peak(cM) Near marker Peak(Mb) LOD Additive R2 HI1.1PR HI1.1PR HI3.1PR HI4.1PR HI6.1PR HI8.1PR HI11.1PR PPI1.1PR PPI1.2PR PPI2.1PR PPI4.1PR PPI11.1PR PPI11.2PR HI PPI Pv01 Pv01 Pv03 Pv04 Pv06 Pv08 Pv11 Pv01 Pv01 Pv02 Pv04 Pv11 Pv11 Combined Kasese 2017 Kasese 2017 Kasese 2016 Kasese 2016 Kasese 2016 Kasese 2017 Namulonge 2017 Combined Kasese 2017 Kasese 2017 Kasese 2017 Kasese 2016 DS DS DS DS DS NS NS DS DS DS DS NS NS 148.41 ss715650354 52.15 143.41 ss715650354 52.15 74.81 ss715649460 3.29 61.91 ss715648923 45.79 8.91 4.31 ss715649019 17.92 ss715641300 19.36 24.91 ss715645541 61.01 ss715648382 4.31 6.67 138.41 ss715645258 50.15 20.11 ss715647234 2.52 59.91 ss715648923 45.79 23.51 ss715648851 5.19 45.61 ss715650599 45.63 4.91 6.28 3.94 4.22 3.01 3.35 3.31 3.52 3.09 3.04 3.70 3.68 3.26 1.06 1.69 -3.74 -0.99 3.09 1.72 2.66 6.8 4.7 5.1 2.5 23.3 2.6 7.1 -4.44 11.4 2.92 2.06 3.2 10.4 -3.90 1.8 4.66 17.6 -3.42 14.2 102 Figure 3.2 Common bean chromosomes Pv01, Pv012, Pv03, Pv04, Pv06, Pv07, Pv08, Pv010, and Pv11 showing quantitative trait loci (QTL) for phenology, yield component traits, and partitioning traits in Portillo x Red Hawk recombinant inbred line (RIL) grown under drought stress and non-stress in Kasese and Namulonge, Uganda during 2016 and 2017 growing seasons 103 ss7156493060.0ss7156403565.7ss71564452028.3ss71564626028.9ss71564626230.1ss71565006039.9ss71564177543.8ss71564683744.7ss71564827560.7ss71564838261.0ss71564796367.6ss71563949367.9ss71564856479.9ss71564607680.5ss71564630094.3ss71564659195.9ss715645258106.4ss715650354148.5ss715645295161.0ss715645293162.3ss715645250169.0HI1.1PPI1.1PPI1.2SY1.1PW1.10 2 4 6 8 10 Pv01 Figure 3.2 (cont’d) 104 ss7156494780.0ss71563943612.0ss71564723420.1ss71563986333.1ss71564202436.7ss71564083337.5ss71564098937.8ss71564050242.2ss71563951549.7ss71564850653.6ss71564964760.4ss71564613978.0ss71564404490.0ss71563990690.3ss71563974590.6ss71563974691.4ss715646145105.0PPI12.1PW2.1SY2.1SY2.20 1 2 3 4 5 6 7 8 9 10 Pv02 Figure 3.2 (cont’d) 105 ss7156481830.0ss7156493024.2ss71564558127.6ss71564557428.0ss71564644141.8ss71564644048.9ss71565056650.1ss71564594150.4ss71564595152.9ss71564950655.0ss71564803956.5ss71564704157.0ss71564704057.2ss71564694660.1ss71563942461.0ss71564946072.8ss71565019081.6ss71564076985.3ss71564417485.6ss71564287385.9ss71564234886.2ss71564992486.5ss71564340488.2HI3.1DF3.2PW3.1SY3.4SY3.3DM3.10 1 2 3 4 5 6 7 8 9 10 11 Pv03 Figure 3.2 (cont’d) 106 ss7156509200.0ss71564678627.8ss71564728634.0ss71564900634.3ss71564680436.5ss71564678436.9ss71564678137.3ss71564621239.4ss71564812441.1ss71564812141.5ss71564953241.6ss71564622442.0ss71564621549.0ss71564060954.4ss71564813059.1ss71564892359.9ss71564830465.0ss71564736065.8ss71564043866.6ss71563980867.0ss71564097467.3ss71563963567.6ss71563950768.2ss71564579889.3PPI14.1PW4.1PW4.2HI4.1SY4.10 1 2 3 4 5 6 7 8 9 10 Pv04 Figure 3.2(cont’d) 107 ss7156501710.0ss7156397812.3ss7156509872.5ss7156474232.9ss7156478436.3ss7156490198.9ss71565028628.7ss71564643743.3ss71564512444.0ss71564848646.6ss71564578948.1ss71564710855.0ss71564710956.4ss71564567456.9ss71564566657.3HI6.1DF6.1SY6.10 1 2 3 4 5 6 7 8 9 10 Pv06ss7156400290.0ss71564635010.8ss71564635211.1ss71564635411.8ss71564635612.2ss71564635812.5ss71564927612.9ss71564888513.9ss71564039214.5ss71564888615.1ss71564715924.4ss71564715825.3ss71564017430.9ss71563972431.1ss71564010631.4ss71564167831.6ss71564201031.8ss71565066333.9ss71564960934.2ss71564772743.6ss71564677346.7ss71564858060.9ss71564601569.6DM7.10 1 2 3 4 5 6 7 8 9 10 Pv07 Figure 3.2(cont’d) 108 ss7156426750.0ss7156413000.3ss71565070010.5ss71564665016.8ss71564652721.8ss71564688034.2ss71564608835.2ss71564611140.8ss71564675841.4ss71564677041.6ss71564676441.9ss71564739346.1ss71564739246.5HI8.1PW8.10 1 2 3 4 5 6 7 8 9 10 Pv08ss7156407940.0ss71565029415.1ss71564902918.1ss71564697418.9ss71564696719.2ss71564696319.4ss71563943219.7ss71564695219.9ss71564694820.1ss71564720720.4ss71564720220.7ss71564720321.2ss71565002122.7ss71564859323.0ss71564632123.6ss71564631624.4ss71564632425.0ss71564633225.3ss71564633326.0ss71564791930.1ss71564550931.8DF10.10 1 2 3 4 5 6 7 8 9 10 Pv10 Figure 3.2(cont’d) 109 ss7156454910.0ss7156490448.9ss71564627310.7ss71564555922.6ss71564885123.5ss71564554124.9ss71564801931.5ss71564318137.6ss71564233337.7ss71564152338.8ss71563947340.3ss71564820140.4ss71564067241.4ss71563965441.8ss71564425341.9ss71564860142.0ss71564800242.1ss71564101642.3ss71565094542.6ss71564880642.7ss71564755143.4ss71564754243.6ss71565059945.6ss71564420747.6ss71564989547.8ss71565013048.0ss71564981348.5HI11.1PPI11.1PPI11.2DF11.10 1 2 3 4 5 6 7 8 9 10 Pv11HI1.1PPI1.1PPI1.2SY1.1PW1.1PPI12.1PW2.1SY2.1SY2.2HI3.1DF3.2PW3.1SY3.4SY3.3DM3.1PPI14.1PW4.1PW4.2HI4.1SY4.1HI6.1DF6.1SY6.1DM7.1HI8.1PW8.1DF10.1HI11.1PPI11.1PPI11.2DF11.1 CHAPTER 4: DETERMINATION OF THE USE OF PHOTOSYNTHETIC TRAITS AS DROUGHT SENSITIVITY INDICATORS OF DROUGHT TOLERANCE IN COMMON BEAN Abstract In this study, 294 common bean genotypes from the Andean gene pool were evaluated at the seedling stage to determine the genetic variability and sensitivity of photosynthetic traits to drought stress. Drought stress was applied by withholding watering for seven days at 21 days after planting before reapplying water to determine recovery while maintaining watering in the control treatment. Photosynthetic traits namely, Phi2, PhiNPQ, and PhiNO were measured using the MultispeQ Beta device while visual scores were assessed using standard protocols. Significant differences were observed among genotypes for photosynthetic traits and visual scores under drought and control conditions, suggesting that genetic variability exists for photosynthetic traits under drought conditions at the seedling stage. Photosynthetic traits were severely affected with the progression of drought stress in the highly sensitive genotypes compared with the less sensitive genotypes, suggesting that alternative mechanisms exist that regulate plant response to drought stress. In addition, slow wilting was positively associated with recovery and could be used to identify drought tolerant common bean genotypes at the seedling stage. 110 Introduction Common bean (Phaseolus vulgaris. L.) is a major component of food and nutrition for millions in the developing world (Broughton et al., 2003; Gepts et al., 2008). Common bean is a staple and source of income in many countries in sub Saharan Africa, and is cultivated under complex farming systems (Broughton et al., 2003). In Uganda, common bean is also an important non- traditional agricultural export crop. However, drought stress caused by erratic rainfall patterns affects over 60% of the bean production areas globally (Beebe et al., 2011), and one-third of the crop production areas in sub-Saharan Africa. The prevalence of both terminal and intermittent drought stress in major common bean production areas in sub-Saharan Africa has been widely reported by farmers, so assessing drought tolerance is an important attribute of common bean varieties (Katungi et al., 2011; Beebe et al., 2014). Therefore, identifying and developing common bean cultivars that are tolerant to both intermittent and terminal drought stress will be critical to sustain yield stability in these drought prone areas of Uganda. Plant responses to drought stress are physiologically complex and difficult to characterize because a wide range of modifications occur within a short period of time (Galmes et al., 2013). In addition several combinations of inherent plant response factors determine the plant‘s abilty to sustain growth under water deficit (Farooq et al., 2009). Drought stress causes stomatal closure in plants in order to prevent excessive water loss, but this subsequently restricts diffusion of CO2 to the sites of carboxylation inside the chloroplast (Lizana et al., 2006; Dias and Brüggemann, 2010; Ninou et al., 2013). These limitations depend on the degree of water deficit which significantly inhibits CO2 assimilation and photosynthetic performance (Flexas et al., 2012). Previous studies have reported that photosynthetic acclimation to drought and its recovery from drought stress depends on the capacity of plants to adjust their stomatal and mesophyll 111 conductance to prevent photo damage from excessive light and elevated temperatures. Therefore, an efficient photo protection mechanism is an essential component of stress resistant plants (Lizana et al., 2006; Chaves et al., 2009; Flexas et al., 2012). Similar trends have been reported in common bean under drought stress resulting in decreases in photosystem II quantum yield (ΦII), non-photochemical excitation quenching (NPQ), total chlorophyll content, stomatal conductance, transpiration rate, leaf temperature, and leaf water potential (Lizana et al., 2006; Dias and Brüggemann, 2010; Ninou et al., 2013; Traub et al., 2017). However, the effect of drought stress on sensitive components of the photosynthetic process are not well understood, thus limiting their use in breeding for tolerance to drought stress. The availability of versatile and robust next generation phenotyping technologies such as the MultispeQ Beta device increases the opportunity for direct measurement of photosynthetic traits (Kuhlgert et al., 2016). MultispeQ Beta device can be deployed under multiple environmental conditions (temperature, humidity, CO2 levels, intensity and quality, time, and location) and could boost phenotype-assisted breeding approaches designed to transfer drought resistance traits to more productive lines. Therefore, breeding for drought tolerance in common bean will require identifying critical plant tolerance responses, and better understanding of physiological mechanisms and genetic controls of contributing traits at critical plant developmental stages. This study seeks to (i) determine genetic variability for photosynthetic traits in Andean beans evaluated under drought stress at the seedling stage; and (ii) determine the sensitivity of photosynthetic traits to drought stress and their subsequent use as an indicator of adaptive physiological responses for identifying drought tolerant common bean genotypes. 112 Materials and Methods In this study a total of 294 genotypes of the Andean Diversity Panel (ADP) belonging to the Andean gene pool including five checks. These included TARS-Tep 32 (PI 666351), an improved Tepary (TB) line tolerant to drought (Porch et al., 2013; Traub et al., 2017), SEN 80, SEN 70, SCN 11, drought tolerant lines from CIAT, and NABE 14 ( local check variety) were evaluated for drought stress under greenhouse conditions at the National Crop Resources Research Institute Namulonge (NaCRRI) in Uganda. The panel consisted of a global collection of Andean beans, mainly varieties, landraces, and breeding lines with different seed types and market classes from Africa, Central America, South America, and North America (Cichy et al., 2015). The experiments were set up modifying protocols previously used in common bean (Mukeshimana et al., 2014). The experiment was planted in a 7x42 incomplete block design (lattice) with two replications. The experiment was repeated twice for both drought stress and control. The genotypes were planted in a 1500 cm3 plastic pots that were filled with a mixture of soil and lake sand in a ratio of 3:1(v/v). The pots were watered to field capacity and allowed to drain overnight before planting. Four seeds were planted in every pot for each genotype to ensure germination and seedlings were thinned to two plants per pot at 7 days after planting (DAP). Approximately 20g of NPK fertilizer was applied to every pot at 10 DAP. Drought stress was imposed at 21 DAP by withholding watering for 7 days for drought treatment while watering was continued for the control treatment. In order to assess the recovery of genotypes after stress, water was then applied to the drought treatment at 7 days (28 DAP) after drought application and recovery was evaluated at 32 DAP. In these experiments photosynthetic traits Phi2, PhiNPQ, and PhiNO were measured using the MultispeQ Beta device (Kuhlgert et al., 2016), on the middle and mature trifoliate leaf of two plants at 22, 24, and 26 DAP in every pot. 113 In the drought stress experiments, wilting score was determined at 24 and 26 DAP; while recovery was scored at 32 DAP. A visual rating for drought stress was performed using a wilting score on a scale of 0 to 5, where 0 shows no wilting and a 5 indicates the plant is completely wilted. The rating score for recovery was modified from the previous 0 to 1 scale used by (Mukeshimana et al., 2014), to a 0 to 5 scale, where 0 means full recovery and 5 means no recovery. Moisture content in every pot was measured in the drought stress experiment at 22, 24 and 26 DAP using soil moisture meter model (PMS-714). Statistical Analysis Data analysis for photosynthetic traits and visual scores was performed using SAS (Institute, 2011) and R statistical packages. Both the photosynthetic traits, Phi2, PhiNPQ, PhiNO and visual rating scores were analyzed using the PROC MIXED procedure. During the analysis for photosynthetic traits, Photosynthetically Active Radiation (PAR), and time of the day were used as covariates to account for diurnal changes in light intensity and time of the day, while the soil moisture content was used to correct for visual scores. The mean values obtained from the PROC MIXED procedure were used for subsequent analysis. The sensitivity of photosynthetic traits to drought stress was tested using a t-test by comparing the difference between the means of drought and control. A normality test was performed to check the distribution of photosynthetic traits under drought treatment at 26 DAP in SAS using the PROC MIXED procedure. In addition, two groups represented by 10% from each side of distribution designated as less and highly sensitive genotypes were used to determine the effect of drought stress on adaptive physiological responses and use for comparing and identifying drought tolerant common bean genotypes. Correlations between photosynthetic traits and visual scores were determined for the 114 selected group using PROC CORR procedure. Visual scores were then used to compare means of the two groups. Photosynthetic Traits Results Overall, significant variation (p<0.05) was observed among the genotypes for photosynthetic traits, Phi2, PhiNPQ, and PhiNO at 22 and 24 DAP under drought conditions, and at 22 and 26 DAP under non-stress conditions (Table 4.1, Figure 4. 1). Under drought conditions the average Phi2 value decreased significantly from 0.48 on 22 DAP to 0.22 on 26 DAP with an overall reduction of 0.26 in photosynthetic efficiency compared to a 0.2 increase observed under control conditions during the same period (Table 4.1, Figure 4.2). Differences in PhiNO was non- significant among genotypes at 26 DAP as the drought stress progressed, and the average PhiNO decreased by 0.7 from 0.19 on 22 DAP to 0.12 on 26 DAP compared to 0.2 decrease under control conditions. Mean values for PhiNPQ increased consistently from 0.32 on 22 DAP to 0.66 on 26 DAP compared with a relatively constant PhiNPQ observed under non-stress conditions (Table 4.1, Figure 4.1). In addition, significant differences (p<0.05) were observed among genotypes for wilting on 24 and 26 DAP, and following recovery on 32 DAP (Table 4.2). Determination of sensitivity of photosynthetic traits to drought stress Significant differences (P<0.001) were detected among photosynthetic traits, namely Phi2 and PhiNPQ for their sensitivity to drought stress. Mean differences were compared for each photosynthetic trait on 22 DAP, 24 DAP, and 26 DAP (data not shown). Relative to the control treatment, the PhiNPQ increased significantly from 0.14 to 0.49 while Phi2 decreased from -0.05 to -0.31, and PhiNO decreased from -0.09 to -0.16 (Figure 4.3). The two photosynthetic traits considered highly sensitive to drought stress were PhiNPQ and Phi2 evaluated at 26 DAP. In 115 addition, the normality test for the two traits showed that both Phi2 and PhiNPQ were normally distributed at 26 DAP under drought stress (Figure 4.3). Correlation analysis for Phi2 and PhiNPQ among the high and less sensitive genotypes As expected, highly significant correlations were observed among photosynthetic traits under drought stress conditions using both most and least sensitive genotypes at 26DAP. Phi2 was negatively correlated with PhiNPQ (r = -0.93, p<0.001), slow wilting at 26 DAP (r=- 0.43, p<0.001), and recovery at 32 DAP (r=-0.37, p<0.001). Meanwhile, PhiNPQ was positively correlated with slow wilting at 26 DAP (r=031, p<0.01), and recovery at 32 DAP (r=0.21, p<0.01). A highly significant and positive correlation was observed between slow wilting at 26 DAP and recovery at 32 DAP (r = 0.83, p<0.001) (Table 4.3). However, it is worth noting that the correlations between photosynthetic traits and visual scores were moderate for both the highly and less sensitive genotypes (Table 4.3). In addition, significant differences in response were observed between high and low sensitive genotypes for Phi2 and PhiNPQ as drought stress progressed from 22 DAP to 26 DAP. On 22 DAP, there were no differences between the means of high and low sensitive genotypes, while on 24 DAP, both highly and less sensitive genotypes responded in the same direction to drought stress. As expected, there was a decrease in Phi2 and an increase in PhiNPQ. The largest difference between the contrasting genotypes was observed on 26 DAP. At 24 DAP, there was a significant decrease in Phi2 and increase in PhiNPQ for highly sensitive genotypes whereas, there was no observable change for these traits among the less sensitive genotypes, implying some inherent mechanism could be responsible for keeping the functionality of the photosynthetic machinery under water deficit. A comparison of the less sensitive genotypes 116 identified using PhiNPQ and slow wilting at 26 DAP and recovery at 32 DAP indicated that TB1was the best candidate under drought stress (Table 4.3, Figure 4.4). Discussion Significant differences were observed among genotypes for photosynthetic traits namely Phi2, PhiNPQ, and PhiNO under drought and control conditions. As expected, Phi2 and PhiNO consistently decreased under the drought treatment while PhiNPQ increased significantly as stress progressed and photosynthetic traits remained stable under the control treatment. These results seem to suggest the existence of genetic variation for photosynthetic traits and drought induced differences among common bean genotypes from the Andean gene pool. These results were not, however, surprising because previous studies reported significant variation among common bean lines for photosynthetic traits that decreased significantly with increase in drought stress (Lizana et al., 2006; Traub et al., 2017). The significant decline in Phi2 from 22 DAP to 26 DAP could be associated with the limiting amounts of CO2 diffusing into the carboxylation sites inside the chloroplast. Drought stress results in stomatal closure, which acts as a preventive measure to prevent excessive water loss. This, however, reduces the evaporative cooling ability for normal function of photosynthetic apparatus and could led to photo damage (Ninou et al., 2013). The significant increase in PhiNPQ from 22 DAP to 26 DAP under drought conditions observed here could be associated with several ways in which plants actively deal with the excessive captured light in order to prevent photo damage. One interesting observation in this study was the difference in PhiNPQ and Phi2 between the highly and less sensitive genotypes from 24 DAP to 26 DAP. Under moderate drought stress (22 to 24 DAP), there were no observable differences between the less and highly sensitive genotypes in Phi2 and PhiNPQ suggesting that PhiNPQ was still effective in dealing with 117 excessive captured light and reduced the risk of photo damage without necessarily affecting Phi2. The significant decrease in Phi2 and PhiNPQ between these genotypes at 26 DAP seems to suggest that PhiNPQ is not the only process or mechanism but the involvement of other non- PhiNPQ processes in regulating drought response in less sensitive genotypes. A similar process appears to be lacking in the highly sensitive group, thus resulting in wilting that was observed in the highly sensitive group. This observation was consistent with previous studies that reported adjustments in stomatal and mesophyll conductance as an important strategy in acclimatizing the photosynthetic apparatus to drought stress in order to prevent photo damage (Lizana et al., 2006; Chaves et al., 2009; Flexas et al., 2012) . A recent study reported lower stomatal conductance for drought tolerant SER 16 and TB1 compared to the more drought susceptible Zorro and Jaguar bean varieties (Traub et al., 2017). It was unclear however, if the difference between the less and highly sensitive genotypes detected in this study was associated with adjustments in stomatal and mesophyll conductance or the involvement of inherent factors, since stomatal conductance was not measured in this study. A highly significant (p<0.001) and positive correlation was observed between slow wilting at 26 DAP and recovery at 32 DAP, suggesting that genotypes that wilted slowly under drought stress had better chances of recovery following stress. A previous study consistently reported slow wilting trait was important for plant survival during drought stress at the seedling stage (Mukeshimana et al., 2014), implying that slow wilting could be a useful trait for identifying drought tolerant lines of common bean at the seedling stage. In this study, the 10 least sensitive genotypes based on low PhiNPQ ranking were generally slow wilting and exhibited higher chances of recovery after stress although correlations between PhiNPQ and the visual scores 118 were low. As expected, the tepary bean was the most drought tolerant genotype with an exceptional balance between photosynthetic traits under drought stress. Conclusion The objective of this study was to determine the genetic variability for photosynthetic traits in Andean beans, and to determine the sensitivity of photosynthetic traits to drought stress and their subsequent use as an indicator of adaptive physiological responses for identifying drought tolerant common bean genotypes. Significant differences were detected among genotypes for photosynthetic traits namely Phi2, PhiNPQ, and PhiNO under drought and control conditions, suggesting that substantial genetic variability exists among common bean genotypes from the Andean gene pool for photosynthetic traits measured under drought conditions at the seedling stage. The significant and irreversible changes in photosynthetic traits under severe drought stress is indicative that photosynthetic traits alone are not the only regulators of drought responses in plants. Genotypes that wilted slowly under drought stress had higher chances of recovery after the removal of the drought stress suggesting that slow wilting could be a useful trait for identifying drought tolerant common bean lines at seedling stage. In addition, low PhiNPQ appears to be associated with slow wilting and increased chances of recovery after stress although correlations between these traits were moderate. 119 APPENDIX 120 Table 4.1. Mean and variance for photosynthetic traits of 294 Andean common bean genotypes measured at 22, 24, and 26 days after planting (DAP) evaluated under water stress conditions. Treatment Drought Control 22DAP 24DAP 26DAP 22DAP 26DAP Trait Mean Variance Mean Variance Mean Variance Mean Variance Mean Variance Phi2 0.48±0.05 *** 0.32±0.06 ** 0.22±0.08 * 0.52±0.04 *** 0.54±0.04 *** PhiNO 0.19±0.03 *** 0.17±0.05 * 0.12±0.05 Ns 0.28±0.04 *** 0.25±0.03 *** PhiNPQ 0.32±0.07 *** 0.51±0.09 ** 0.66±0.11 ** 0.18±0.06 *** 0.19±0.06 *** * Significant at the 0.05 probability level, ** Significant at the 0.01 probability level, *** Significant at the 0.001 probability level, and ns not significant. 121 Table 4.2. Mean and variance for visual scores of 294 Andean common bean genotypes measured at 24, 26 and 32 days after planting (DAP) evaluated under water stress conditions. Visual score (Drought) Mean Variance WLT24DAP(0_5) WLT26DAP(0_5) 3.18±0.81 4.31±0.67 RECV_32DAP(0_5) 4.17±0.84 * *** *** WLT Slow wilting score, RECV Recovery , * Significant at the 0.05 probability level, ** Significant at the 0.01 probability level, *** Significant at the 0.001 probability level, and ns not significant. 122 Figure 4.1. The boxplot for the means of photosynthetic traits PhiNO, Phi2 and PhiNPQ of 294 Andean common bean genotypes measured at 22 and 26 days after planting (DAP) evaluated under water stress conditions. 123 0.10.20.30.40.52226DAPPhiNOTreatmentControlDrought0.00.20.40.60.82226DAPPhi2TreatmentControlDrought0.000.250.500.752226DAPPhiNPQTreatmentControlDrought Figure 4.2. The effect of drought stress on photosynthetic traits. (A) The average Phi2, PhiNO and PhiNPQ measured drought and control conditions measured on 22, 24 and 26DAP. (B) The effect of drought stress on photosynthetic traits based on the difference between the means of drought stress and control treatments A B 124 Figure 4.3. The distribution of Phi2 and PhiNPQ of 294 Andean common bean genotypes measured at 26 days after planting (DAP) under water stress conditions. Tepary (TARS-Tep 32) Tepary (TARS-Tep 32) 125 Table 4.3. Correlations among photosynthetic traits Phi2, PhiNO, PhiNPQ, slow wilting at 26 DAP and recovery at 32 DAP for high and low sensitive common bean genotypes (60) genotypes selected using PhiNPQ at 26 DAP under drought condition Traits Phi2_26DAP PhiNPQ_26DAP RECV_32DAP WLT_26DAP Phi2_26DAP -0.93*** -0.37*** -0.43*** PhiNPQ_26DAP RECV_32DAP WLT26DAP 0.21** 0.31** 0.83*** WLT Slow wilting score, RECV Recovery , * Significant at the 0.05 probability level, ** Significant at the 0.01 probability level, *** Significant at the 0.001 probability level, and ns not significant. 126 Figure 4.4. The variation in photosynthetic traits PhiNPQ, Phi2, and PhiNO among high sensitive low sensitive genotypes under drought conditions on 22, 24 and 26 DAP 127 Figure 4.5. The distribution of wilting at 26 DAP and recovery on 32 DAP for genotypes evaluated under drought conditions. Tepary (TARS-Tep 32) Tepary (TARS-Tep 32) 128 Table 4.4. Mean photosynthetic traits, visual scores, and geometric mean of seed yield for the 10 best individuals based on PhiNPQ score determined at 26 days after planting. Genotype Cultivar name TARS-Tep 32 PI 203934 USDK-4 G 13167 Chijar (ID) Tepary ADP 379 ADP 654 ADP 271 ADP 445 ADP 062 Maulasi G 10994 ADP 255 Badillo ADP 626 G 1368 ADP 186 SEN 80 SEN 80 Mean SED Country (region) USA USA USA CIAT Caribbean Tanzania CIAT Caribbean CIAT core CIAT Seed color Phi2 PhiNPQ RECV WLT GM Cream Cream DRK White Red mottled Red Tan LRK Red Red 0.35 0.33 0.37 0.33 0.29 0.27 0.28 0.23 0.24 0.25 0.29 0.05 0.43 0.47 0.50 0.52 0.57 0.59 0.60 0.61 0.61 0.61 0.55 0.07 0.13 1.97 2.68 2.09 1.81 1.66 2.63 2.97 2.60 2.85 2.14 0.84 1.17 4.38 2.34 3.46 3.48 2.69 3.79 4.26 4.86 3.47 3.39 1.09 11.6 20.1 Na 10 7.2 Na Na 10.7 19.8 1.65 RECV= recovery at 32 DAP; WLT= wilting at 26 DAP; GM= Geometric mean of seed yield under field conditions; ADP Andean diversity panel identity; LRK Light red kidney; DRK Dark red kidney; SED Standard error of difference. 129 REFERENCES 130 REFERENCES Acosta-Gallegos J, Adams M (1991) Plant traits and yield stability of dry bean (Phaseolus vulgaris) cultivars under drought stress. The Journal of Agricultural Science 117: 213- 219 Akibode S, Maredia M (2011) Global and Regional Trends in Production, Trade and Consumption of Food Legume Crops. Alexandersson E, Fraysse L, Sjövall-Larsen S, Gustavsson S, Fellert M, Karlsson M, Johanson U, Kjellbom P (2005) Whole gene family expression and drought stress regulation of aquaporins. Plant molecular biology 59: 469-484 Ariani A, Gepts P (2015) Genome-wide identification and characterization of aquaporin gene family in common bean (Phaseolus vulgaris L.). Molecular genetics and genomics 290: 1771-1785 Asfaw A, Blair MW, Struik PC (2012) Multienvironment Quantitative Trait Loci Analysis for Photosynthate Acquisition, Accumulation, and Remobilization Traits in Common Bean Under Drought Stress. G3: Genes|Genomes|Genetics 2: 579-595 Assefa T, Beebe SE, Rao IM, Cuasquer JB, Duque MC, Rivera M, Battisti A, Lucchin M (2013) Pod harvest index as a selection criterion to improve drought resistance in white pea bean. Field crops research 148: 24-33 Basu MM, González-Carranza ZH, Azam-Ali S, Tang S, Shahid AA, Roberts JA (2013) The manipulation of auxin in the abscission zone cells of Arabidopsis flowers reveals that indoleacetic acid signaling is a prerequisite for organ shedding. Plant physiology 162: 96- 106 Bates D, Mächler M, Bolker B, Walker S (2014) Fitting linear mixed-effects models using lme4. arXiv preprint arXiv:1406.5823 Battisti DS, Naylor RL (2009) Historical warnings of future food insecurity with unprecedented seasonal heat. Science 323: 240-244 Beattie AD, Larsen J, Michaels TE, Pauls KP (2003) Mapping quantitative trait loci for a common bean (Phaseolus vulgaris L.) ideotype. Genome 46: 411-422 Beebe S, Ramirez J, Jarvis A, Rao IM, Mosquera G, Bueno JM, Blair MW (2011) Genetic improvement of common beans and the challenges of climate change. Crop adaptation to climate change: 356-369 131 Beebe S, Rao I, Blair M, Acosta J (2013) Phenotyping common beans for adaptation to drought. Frontiers in physiology 4: 35 Beebe S, Skroch PW, Tohme J, Duque M, Pedraza F, Nienhuis J (2000) Structure of genetic diversity among common bean landraces of Middle American origin based on correspondence analysis of RAPD. Crop science 40: 264-273 Beebe SE, Rao IM, Blair MW, Acosta-Gallegos JA (2013) Phenotyping common beans for adaptation to drought. Front Physiology 4: 35 Beebe SE, Rao IM, Cajiao C, Grajales M (2008) Selection for Drought Resistance in Common Bean Also Improves Yield in Phosphorus Limited and Favorable Environments. Crop Science 48 Beebe SE, Rao IM, Devi MJ, Polania J (2014) Common beans, biodiversity, and multiple stresses: challenges of drought resistance in tropical soils. Crop and Pasture Science 65: 667-675 Beebe SE, Rao IM, Mukankusi C, Buruchara R (2012) Improving resource use efficiency and reducing risk of common bean production in Africa, Latin America, and the Caribbean. In. Centro Internacional de Agricultura Tropical (CIAT) Belko N, Zaman‐Allah M, Diop N, Cisse N, Zombre G, Ehlers J, Vadez V (2013) Restriction of transpiration rate under high vapour pressure deficit and non‐limiting water conditions is important for terminal drought tolerance in cowpea. Plant Biology 15: 304-316 Bellucci E, Bitocchi E, Rau D, Rodriguez M, Biagetti E, Giardini A, Attene G, Nanni L, Papa R (2014) Genomics of Origin, Domestication and Evolution of Phaseolus vulgaris. In Genomics of Plant Genetic Resources, pp 483-507 Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the royal statistical society. Series B (Methodological): 289-300 Bitocchi E, Bellucci E, Giardini A, Rau D, Rodriguez M, Biagetti E, Santilocchi R, Spagnoletti Zeuli P, Gioia T, Logozzo G (2013) Molecular analysis of the parallel domestication of the common bean (Phaseolus vulgaris) in Mesoamerica and the Andes. New Phytologist 197: 300-313 Blair MW, Diaz J, Hidalgo R, Diaz L, Duque M (2007) Microsatellite characterization of Andean races of common bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics 116: 29-43 Blair MW, Galeano CH, Tovar E, Munoz Torres MC, Castrillon AV, Beebe SE, Rao IM (2012) Development of a Mesoamerican intra-genepool genetic map for quantitative trait loci 132 detection in a drought tolerant x susceptible common bean (Phaseolus vulgaris L.) cross. Molecular Breeding 29: 71-88 Blair MW, Iriarte G, Beebe S (2006) QTL analysis of yield traits in an advanced backcross population derived from a cultivated Andean x wild common bean (Phaseolus vulgaris L.) cross. Theoretical and Applied Genetics 112: 1149-1163 Blair MW, Prieto S, Díaz LM, Buendía HF, Cardona C (2010) Linkage disequilibrium at the APA insecticidal seed protein locus of common bean (Phaseolus vulgaris L.). BMC plant biology 10: 79 Blum A (2011) Drought resistance–is it really a complex trait? Functional Plant Biology 38: 753- 757 Blum A (2011) Plant Water Relations, Plant Stress and Plant Production. 11-52 Blum A (2014) Genomics for drought resistance–getting down to earth. Functional Plant Biology 41: 1191-1198 Borém A, Ramalho MAP, Fritsche-Neto R (2012) Abiotic stresses: challenges for plant breeding in the coming decades. In Plant Breeding for Abiotic Stress Tolerance. Springer, pp 1-12 Boyer J, Byrne P, Cassman K, Cooper M, Delmer D, Greene T, Gruis F, Habben J, Hausmann N, Kenny N (2013) The US drought of 2012 in perspective: A call to action. Global Food Security 2: 139-143 Bradbury PJ, Zhang Z, Kroon DE, Casstevens TM, Ramdoss Y, Buckler ES (2007) TASSEL: software for association mapping of complex traits in diverse samples. Bioinformatics 23: 2633-2635 Brinez B, Blair MW, Kilian A, Carbonell SAM, Chiorato AF, Rubiano LB (2012) A whole genome DArT assay to assess germplasm collection diversity in common beans. Molecular breeding 30: 181-193 Brinez B, Perseguini J, Rosa JS, Bassi D, Goncalves JGR, Almeida C, Paulino JFC, Blair MW, Chioratto AF, Carbonell SAM, Valdisser P, Vianello RP, Benchimol-Reis LL (2017) Mapping QTL for drought tolerance in a SEA 5 x AND 277 common bean cross with SSRs and SNP markers. Genetics Molecular Biology 40: 813-823 Broughton WJ, Hernandez G, Blair M, Beebe S, Gepts P, Vanderleyden J (2003) Beans (Phaseolus spp.)–model food legumes. Plant and soil 252: 55-128 Cattivelli L, Rizza F, Badeck F-W, Mazzucotelli E, Mastrangelo AM, Francia E, Mare C, Tondelli A, Stanca AM (2008) Drought tolerance improvement in crop plants: an integrated view from breeding to genomics. Field Crops Research 105: 1-14 133 Cavatte PC, Martins SC, Morais LE, Silva PE, DaMatta FM (2012) The physiology of abiotic stresses. In Plant breeding for abiotic stress tolerance. Springer, pp 21-51 Cenci A, Rouard M (2017) Evolutionary analyses of GRAS transcription factors in angiosperms. Frontiers in Plant Science 8: 273 Cha J-Y, Kim JY, Jung IJ, Kim MR, Melencion A, Alam SS, Yun D-J, Lee SY, Kim MG, Kim W-Y (2014) NADPH-dependent thioredoxin reductase A (NTRA) confers elevated tolerance to oxidative stress and drought. Plant physiology and biochemistry 80: 184-191 Chaves MM, Flexas J, Pinheiro C (2009) Photosynthesis under drought and salt stress: regulation mechanisms from whole plant to cell. Annals of botany 103: 551-560 Checa OE, Blair MW (2012) Inheritance of Yield-Related Traits in Climbing Beans ( L.). Crop Science 52 Chen B, Wang J, Zhang G, Liu J, Manan S, Hu H, Zhao J (2016) Two types of soybean diacylglycerol acyltransferases are differentially involved in triacylglycerol biosynthesis and response to environmental stresses and hormones. Scientific reports 6: 28541 Chen X, Chen Z, Zhao H, Zhao Y, Cheng B, Xiang Y (2014) Genome-wide analysis of soybean HD-Zip gene family and expression profiling under salinity and drought treatments. PloS one 9: e87156 Cichy KA, Caldas GV, Snapp SS, Blair MW (2009) QTL Analysis of Seed Iron, Zinc, and Phosphorus Levels in an Andean Bean Population. Crop Science 49 Cichy KA, Porch TG, Beaver JS, Cregan P, Fourie D, Glahn RP, Grusak MA, Kamfwa K, Katuuramu DN, McClean P (2015) A Phaseolus vulgaris diversity panel for Andean bean improvement. Crop Science 55: 2149-2160 Cichy KA, Porch TG, Beaver JS, Cregan P, Fourie D, Glahn RP, Grusak MA, Kamfwa K, Katuuramu DN, McClean P, Mndolwa E, Nchimbi-Msolla S, Pastor-Corrales MA, Miklas PN (2015) A Diversity Panel for Andean Bean Improvement. Crop Science 55: 2149 Cichy KA, Wiesinger JA, Mendoza FA (2015) Genetic diversity and genome-wide association analysis of cooking time in dry bean (Phaseolus vulgaris L.). Theoretical and applied genetics 128: 1555-1567 de Faria Müller BS, Sakamoto T, Silveira RDD, Zambussi-Carvalho PF, Pereira M, Pappas GJ, do Carmo Costa MM, Guimarães CM, Pereira WJ, Brondani C (2014) Differentially expressed genes during flowering and grain filling in common bean (Phaseolus vulgaris) grown under drought stress conditions. Plant molecular biology reporter 32: 438-451 134 Devi MJ, Sinclair TR, Vadez V (2010) Genotypic variability among peanut (Arachis hypogea L.) in sensitivity of nitrogen fixation to soil drying. Plant and Soil 330: 139-148 Dias M, Brüggemann W (2010) Limitations of photosynthesis in Phaseolus vulgaris under drought stress: gas exchange, chlorophyll fluorescence and Calvin cycle enzymes. Photosynthetica 48: 96-102 Díaz L, Blair MW (2006) Race structure within the Mesoamerican gene pool of common bean (Phaseolus vulgaris L.) as determined by microsatellite markers. Theoretical and Applied Genetics 114: 143-154 Diaz LM, Ricaurte J, Cajiao C, Galeano CH, Rao I, Beebe S, Raatz B (2017) Phenotypic evaluation and QTL analysis of yield and symbiotic nitrogen fixation in a common bean population grown with two levels of phosphorus supply. Molecular Breeding 37: 76 Dievart A, Perin C, Hirsch J, Bettembourg M, Lanau N, Artus F, Bureau C, Noel N, Droc G, Peyramard M (2016) The phenome analysis of mutant alleles in leucine-rich repeat receptor-like kinase genes in rice reveals new potential targets for stress tolerant cereals. Plant Science 242: 240-249 Du J, Wang S, He C, Zhou B, Ruan Y-L, Shou H (2017) Identification of regulatory networks and hub genes controlling soybean seed set and size using RNA sequencing analysis. Journal of experimental botany 68: 1955-1972 Fan C-M, Wang X, Wang Y-W, Hu R-B, Zhang X-M, Chen J-X, Fu Y-F (2013) Genome-wide expression analysis of soybean MADS genes showing potential function in the seed development. PloS one 8: e62288 Farooq M, Gogoi N, Barthakur S, Baroowa B, Bharadwaj N, Alghamdi SS, Siddique KHM (2017) Drought Stress in Grain Legumes during Reproduction and Grain Filling. Journal of Agronomy and Crop Science 203: 81-102 Farooq M, Wahid A, Kobayashi N, Fujita D, Basra S (2009) Plant drought stress: effects, mechanisms and management. In Sustainable agriculture. Springer, pp 153-188 Fehr WR, Fehr EL, Jessen HJ (1987) Principles of cultivar development. Fischer R, Maurer R (1978) Drought resistance in spring wheat cultivars. I. Grain yield responses. Australian Journal of Agricultural Research 29: 897-912 Flexas J, Barbour MM, Brendel O, Cabrera HM, Carriquí M, Díaz-Espejo A, Douthe C, Dreyer E, Ferrio JP, Gago J (2012) Mesophyll diffusion conductance to CO2: an unappreciated central player in photosynthesis. Plant Science 193: 70-84 135 Fornara F, Montaigu A, Sánchez‐Villarreal A, Takahashi Y, Ver Loren van Themaat E, Huettel B, Davis SJ, Coupland G (2015) The GI–CDF module of Arabidopsis affects freezing tolerance and growth as well as flowering. The Plant Journal 81: 695-706 Fornara F, Panigrahi KC, Gissot L, Sauerbrunn N, Rühl M, Jarillo JA, Coupland G (2009) Arabidopsis DOF transcription factors act redundantly to reduce CONSTANS expression and are essential for a photoperiodic flowering response. Developmental cell 17: 75-86 Frahm MA, Rosas JC, Mayek-Pérez N, López-Salinas E, Acosta-Gallegos JA, Kelly JD (2004) Breeding beans for resistance to terminal drought in the lowland tropics. Euphytica 136: 223-232 Fujita Y, Fujita M, Satoh R, Maruyama K, Parvez MM, Seki M, Hiratsu K, Ohme-Takagi M, Shinozaki K, Yamaguchi-Shinozaki K (2005) AREB1 is a transcription activator of novel ABRE-dependent ABA signaling that enhances drought stress tolerance in Arabidopsis. The Plant Cell 17: 3470-3488 Galmes J, OCHOGAVÍA J, Gago J, ROLDÁN EJ, Cifre J, CONESA MÀ (2013) Leaf responses to drought stress in Mediterranean accessions of Solanum lycopersicum: anatomical adaptations in relation to gas exchange parameters. Plant, cell & environment 36: 920- 935 Gepts P, Aragão FJ, De Barros E, Blair MW, Brondani R, Broughton W, Galasso I, Hernández G, Kami J, Lariguet P (2008) Genomics of Phaseolus beans, a major source of dietary protein and micronutrients in the tropics. In Genomics of tropical crop plants. Springer, pp 113-143 Gepts P, Bliss F (1985) F1 hybrid weakness in the common bean: Differential geographic origin suggets two gene pools in cultivated bean germplasm. Journal of Heredity 76: 447-450 Gepts P, Bliss F (1988) Dissemination pathways of common bean (Phaseolus vulgaris, Fabaceae) deduced from phaseolin electrophoretic variability. II. Europe and Africa. Economic Botany 42: 86-104 Gomes D, Agasse A, Thiébaud P, Delrot S, Gerós H, Chaumont F (2009) Aquaporins are multifunctional water and solute transporters highly divergent in living organisms. Biochimica et Biophysica Acta (BBA)-Biomembranes 1788: 1213-1228 Gonzalez AM, Yuste-Lisbona FJ, Saburido S, Bretones S, De Ron AM, Lozano R, Santalla M (2016) Major Contribution of Flowering Time and Vegetative Growth to Plant Production in Common Bean As Deduced from a Comparative Genetic Mapping. Front Plant Science 7: 1940 136 Goodstein DM, Shu S, Howson R, Neupane R, Hayes RD, Fazo J, Mitros T, Dirks W, Hellsten U, Putnam N (2011) Phytozome: a comparative platform for green plant genomics. Nucleic acids research 40: D1178-D1186 Guo M, Liu J-H, Ma X, Luo D-X, Gong Z-H, Lu M-H (2016) The plant heat stress transcription factors (HSFs): structure, regulation, and function in response to abiotic stresses. Frontiers in plant science 7: 114 Hammes UZ, Nielsen E, Honaas LA, Taylor CG, Schachtman DP (2006) AtCAT6, a sink‐tissue‐localized transporter for essential amino acids in Arabidopsis. The Plant Journal 48: 414-426 Heilig JA, Beaver JS, Wright EM, Song Q, Kelly JD (2017) QTL Analysis of Symbiotic Nitrogen Fixation in a Black Bean Population. Crop Science 57: 118-129 Hirsch S, Oldroyd GE (2009) GRAS-domain transcription factors that regulate plant development. Plant signaling & behavior 4: 698-700 Hofmann NR (2009) Glutaredoxin Functions in Floral Development. The Plant Cell 21: 363-363 Hothorn M, Van den Ende W, Lammens W, Rybin V, Scheffzek K (2010) Structural insights into the pH-controlled targeting of plant cell-wall invertase by a specific inhibitor protein. Proceedings of the National Academy of Sciences 107: 17427-17432 Hoyos-Villegas V, Mkwaila W, Cregan PB, Kelly JD (2015) Quantitative Trait Loci Analysis of White Mold Avoidance in Pinto Bean. Crop Science 55: 2116-2129 Hoyos-Villegas V, Song Q, Kelly JD (2017) Genome-wide Association Analysis for Drought Tolerance and Associated Traits in Common Bean. Plant Genome 10:1-17 Hoyos-Villegas V, Song Q, Wright EM, Beebe SE, Kelly JD (2016) Joint Linkage QTL Mapping for Yield and Agronomic Traits in a Composite Map of Three Common Bean RIL Populations. Crop Science 56: 1-18 Huang X, Han B (2014) Natural variations and genome-wide association studies in crop plants. Annual review of plant biology 65: 531-551 Hwang SM, Kim DW, Woo MS, Jeong HS, Son YS, Akhter S, Choi GJ, Bahk JD (2014) Functional characterization of Arabidopsis HsfA6a as a heat‐shock transcription factor under high salinity and dehydration conditions. Plant, cell & environment 37: 1202-1222 Iqbal N, Khan NA, Ferrante A, Trivellini A, Francini A, Khan M (2017) Ethylene role in plant growth, development and senescence: interaction with other phytohormones. Frontiers in plant science 8: 475 137 Jacob P, Hirt H, Bendahmane A (2017) The heat‐shock protein/chaperone network and multiple stress resistance. Plant biotechnology journal 15: 405-414 Kamfwa K, Cichy KA, Kelly JD (2015) Genome-wide association analysis of symbiotic nitrogen fixation in common bean. Theoretical and Applied Genetics 128: 1999-2017 Kamfwa K, Cichy KA, Kelly JD (2015) Genome-Wide Association Study of Agronomic Traits in Common Bean. The Plant Genome 8:1-12 Kanazawa A, Kramer DM (2002) In vivo modulation of nonphotochemical exciton quenching (NPQ) by regulation of the chloroplast ATP synthase. Proceedings of the National Academy of Sciences 99: 12789-12794 Kang I-H, Steffen JG, Portereiko MF, Lloyd A, Drews GN (2008) The AGL62 MADS domain protein regulates cellularization during endosperm development in Arabidopsis. The Plant Cell 20: 635-647 Katungi E, Sperling L, Karanja D, Beebe S (2011) Relative importance of common bean attributes and variety demand in the drought areas of Kenya. Journal of Development and Agricultural Economics 3: 411-422 Kelly JD (2018) Developing improved varieties of common bean. In: Achieving sustainable cultivation of grain legumes Volume 2. Burleigh Dodds Science Publishing, pp 25-40 Kim MR, Khaleda L, Jung IJ, Kim JY, Lee SY, Cha J-Y, Kim W-Y (2017) Overexpression of chloroplast-localized NADPH-dependent thioredoxin reductase C (NTRC) enhances tolerance to photo-oxidative and drought stresses in Arabidopsis thaliana. Journal of Plant Biology 60: 175-180 Klaedtke SM, Cajiao C, Grajales M, Polanía J, Borrero G, Guerrero A, Rivera M, Rao I, Beebe SE, Léon J (2012) Photosynthate remobilization capacity from drought-adapted common bean (Phaseolus vulgaris L.) lines can improve yield potential of interspecific populations within the secondary gene pool. Journal of Plant Breeding and Crop Science 4: 49-61 Klepek Y-S, Geiger D, Stadler R, Klebl F, Landouar-Arsivaud L, Lemoine R, Hedrich R, Sauer the transporter-like superfamily, mediates H+-symport of numerous N monosaccharide substrates, including myo-inositol, glycerol, and ribose. The Plant Cell 17: 204-218 (2005) Arabidopsis POLYOL TRANSPORTER5, a new member of Klepek Y-S, Volke M, Konrad KR, Wippel K, Hoth S, Hedrich R, Sauer N (2009) Arabidopsis thaliana POLYOL/MONOSACCHARIDE TRANSPORTERS 1 and 2: fructose and xylitol/H+ symporters in pollen and young xylem cells. Journal of experimental botany 61: 537-550 138 Koinange EMK, Singh SP, Gepts P (1996) Genetic Control of the Domestication Syndrome in Common Bean. Crop Science 36: 1037-1045 Kreps JA, Ponappa T, Dong W, Town CD (1996) Molecular basis of [alpha]-methyltryptophan resistance in amt-1, a mutant of Arabidopsis thaliana with altered tryptophan metabolism. Plant physiology 110: 1159-1165 Kuhlgert S, Austic G, Zegarac R, Osei-Bonsu I, Hoh D, Chilvers MI, Roth MG, Bi K, TerAvest D, Weebadde P (2016) MultispeQ Beta: a tool for large-scale plant phenotyping connected to the open PhotosynQ network. Royal Society open science 3: 160592 Kwak M, Gepts P (2009) Structure of genetic diversity in the two major gene pools of common bean (Phaseolus vulgaris L., Fabaceae). Theoretical and Applied Genetics 118: 979-992 Kwak M, Velasco D, Gepts P (2008) Mapping homologous sequences for determinacy and photoperiod sensitivity in common bean (Phaseolus vulgaris). Journal of heredity 99: 283-291 Lardizabal K, Effertz R, Levering C, Mai J, Pedroso M, Jury T, Aasen E, Gruys K, Bennett K (2008) Expression of Umbelopsis ramanniana DGAT2A in seed increases oil in soybean. Plant physiology 148: 89-96 Lee J-H, Kim WT (2011) Regulation of abiotic stress signal transduction by E3 ubiquitin ligases in Arabidopsis. Molecules and cells 31: 201-208 Levitt J (1972) Response of plants to environmental stresses. Academic Press, 698 pp. Levitt J (1980) Responses of Plants to Environmental Stress. Academic Press.2nd Edition, Volume 1: Chilling, Freezing, and High Temperature Stresses. 497 pp Li L, Wang H, Gago J, Cui H, Qian Z, Kodama N, Ji H, Tian S, Shen D, Chen Y (2015) Harpin Hpa1 interacts with aquaporin PIP1; 4 to promote the substrate transport and photosynthesis in Arabidopsis. Scientific reports 5: 17207 Li P-S, Yu T-F, He G-H, Chen M, Zhou Y-B, Chai S-C, Xu Z-S, Ma Y-Z (2014) Genome-wide analysis of the Hsf family in soybean and functional identification of GmHsf-34 involvement in drought and heat stresses. BMC genomics 15: 1009 Li R, Hatanaka T, Yu K, Wu Y, Fukushige H, Hildebrand D (2013) Soybean oil biosynthesis: role of diacylglycerol acyltransferases. Functional & integrative genomics 13: 99-113 Li S (2014) Redox modulation matters: emerging functions for glutaredoxins in plant development and stress responses. Plants 3: 559-582 139 Lim CJ, Yang KA, Hong JK, Choi JS, Yun D-J, Hong JC, Chung WS, Lee SY, Cho MJ, Lim CO (2006) Gene expression profiles during heat acclimation in Arabidopsis thaliana suspension-culture cells. Journal of plant research 119: 373-383 Lipka AE, Tian F, Wang Q, Peiffer J, Li M, Bradbury PJ, Gore MA, Buckler ES, Zhang Z (2012) GAPIT: genome association and prediction integrated tool. Bioinformatics 28: 2397-2399 Lizana C, Wentworth M, Martinez JP, Villegas D, Meneses R, Murchie EH, Pastenes C, Lercari B, Vernieri P, Horton P (2006) Differential adaptation of two varieties of common bean to abiotic stress: I. Effects of drought on yield and photosynthesis. Journal of Experimental botany 57: 685-697 Lobell DB, Gourdji SM (2012) The influence of climate change on global crop productivity. Plant Physiology 160: 1686-1697 Miklas PN, Kelly JD, Beebe SE, Blair MW (2006) Common bean breeding for resistance against biotic and abiotic stresses: From classical to MAS breeding. Euphytica 147: 105-131 Moore JP, Le NT, Brandt WF, Driouich A, Farrant JM (2009) Towards a systems-based understanding of plant desiccation tolerance. Trends in plant science 14: 110-117 Mukeshimana G, Butare L, Cregan PB, Blair MW, Kelly JD (2014) Quantitative Trait Loci Associated with Drought Tolerance in Common Bean. Crop Science 54: 923-938 Mukeshimana G, Lasley AL, Loescher WH, Kelly JD (2014) Identification of shoot traits related to drought tolerance in common bean seedlings. Journal of the American Society for Horticultural Science 139: 299-309 Mumba L, Galwey N (1999) Compatibility between wild and cultivated common bean (Phaseolus vulgaris L.) genotypes of the Mesoamerican and Andean gene pools: Evidence from the inheritance of quantitative characters. Euphytica 108: 105-119 Muñoz-Perea CG, Terán H, Allen RG, Wright JL, Westermann DT, Singh SP (2006) Selection for Drought Resistance in Dry Bean Landraces and Cultivars. Crop Science 46 Nakashima K, Yamaguchi-Shinozaki K (2013) ABA signaling in stress-response and seed development. Plant cell reports 32: 959-970 Namugwanya M, Tenywa JS, Otabbong E, Mubiru DN, Masamba TA (2014) Development of Common Bean (Phaseolus Vulgaris L.) Production Under Low Soil Phosphorus and Drought in Sub-Saharan Africa: A Review. Journal of Sustainable Development 7 Nielsen DC, Nelson NO (1998) Black Bean Sensitivity to Water Stress at Various Growth Stages. Crop Science 38: 422-427 140 Ninou E, Tsialtas J, Dordas C, Papakosta D (2013) Effect of irrigation on the relationships between leaf gas exchange related traits and yield in dwarf dry bean grown under Mediterranean conditions. Agricultural water management 116: 235-241 Noiraud N, Maurousset L, Lemoine R (2001) Transport of polyols in higher plants. Plant Physiology and Biochemistry 39: 717-728 Osakabe Y, Osakabe K, Shinozaki K, Tran L-SP (2014) Response of plants to water stress. Frontiers in plant science 5: 86 Ouyang SQ, Liu YF, Liu P, Lei G, He SJ, Ma B, Zhang WK, Zhang JS, Chen SY (2010) Receptor‐like kinase OsSIK1 improves drought and salt stress tolerance in rice (Oryza sativa) plants. The Plant Journal 62: 316-329 Pennisi E (2008) Getting to the root of drought responses. Science 320: 173-173 Perez-Vega E, Paneda A, Rodriguez-Suarez C, Campa A, Giraldez R, Ferreira JJ (2010) Mapping of QTL for morpho-agronomic and seed quality traits in a RIL population of common bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics 120: 1367-1380 Porch TG, Beaver JS, Brick MA (2013) Registration of tepary germplasm with multiple-stress tolerance, TARS-Tep 22 and TARS-Tep 32. Journal of Plant Registrations 7: 358-364 Porch TG, Ramirez VH, Santana D, Harmsen EW (2009) Evaluation of Common Bean for Drought Tolerance in Juana Diaz, Puerto Rico. Journal of Agronomy and Crop Science 195: 328-334 Pulido P, Cazalis R, Cejudo FJ (2009) An antioxidant redox system in the nucleus of wheat seed cells suffering oxidative stress. The Plant Journal 57: 132-145 Ramirez-Vallejo P, Kelly JD (1998) Traits related to drought resistance in common bean. Euphytica 99: 127-136 Rao IM (2001) Role of physiology in improving crop adaptation to abiotic stresses in the tropics: The case of common bean and tropical forages. Handbook of plant and crop physiology: 583-613 Rentsch D, Schmidt S, Tegeder M (2007) Transporters for uptake and allocation of organic nitrogen compounds in plants. FEBS letters 581: 2281-2289 Reuscher S, Akiyama M, Yasuda T, Makino H, Aoki K, Shibata D, Shiratake K (2014) The sugar transporter inventory of tomato: genome-wide identification and expression analysis. Plant and Cell Physiology 55: 1123-1141 141 Rosado A, Schapire AL, Bressan RA, Harfouche AL, Hasegawa PM, Valpuesta V, Botella MA (2006) The Arabidopsis tetratricopeptide repeat-containing protein TTL1 is required for osmotic stress responses and abscisic acid sensitivity. Plant Physiology 142: 1113-1126 Rosales-Serna R, Kohashi-Shibata J, Acosta-Gallegos JA, Trejo-López C, Ortiz-Cereceres Jn, Kelly JD (2004) Biomass distribution, maturity acceleration and yield in drought-stressed common bean cultivars. Field Crops Research 85: 203-211 Rouhier N, Couturier J, Jacquot J-P (2006) Genome-wide analysis of plant glutaredoxin systems. Journal of Experimental Botany 57: 1685-1696 Rouhier N, Lemaire SD, Jacquot J-P (2008) The role of glutathione in photosynthetic organisms: emerging functions for glutaredoxins and glutathionylation. Annual Review of Plant Biology. 59: 143-166 Ruan Y-L (2014) Sucrose metabolism: gateway to diverse carbon use and sugar signaling. Annual review of plant biology 65: 33-67 Sadanandom A, Bailey M, Ewan R, Lee J, Nelis S (2012) The ubiquitin–proteasome system: central modifier of plant signalling. New Phytologist 196: 13-28 SAS Institute (2002) he SAS system for Windows. Release 9.4. SAS Institute Inc., Cary, NC. Schapire AL, Valpuesta V, Botella MA (2006) TPR proteins in plant hormone signaling. Plant signaling & behavior 1: 229-230 Schmutz J, McClean PE, Mamidi S, Wu GA, Cannon SB, Grimwood J, Jenkins J, Shu S, Song Q, Chavarro C, Torres-Torres M, Geffroy V, Moghaddam SM, Gao D, Abernathy B, Barry K, Blair M, Brick MA, Chovatia M, Gepts P, Goodstein DM, Gonzales M, Hellsten U, Hyten DL, Jia G, Kelly JD, Kudrna D, Lee R, Richard MM, Miklas PN, Osorno JM, Rodrigues J, Thareau V, Urrea CA, Wang M, Yu Y, Zhang M, Wing RA, Cregan PB, Rokhsar DS, Jackson SA (2014) A reference genome for common bean and genome-wide analysis of dual domestications. Nature Genetics 46: 707-713 Schneider KA, Brothers ME, Kelly JD (1997) Marker-assisted selection to improve drought resistance in common bean. Crop Science 37: 51-60 Schneider KA, Rosales-Serna R, Ibarra-Perez F, Cazares-Enriquez B, Acosta-Gallegos JA, Ramirez-Vallejo P, Wassimi N, Kelly JD (1997) Improving Common Bean Performance under Drought Stress. Crop Science 37 Serrato AJ, Pérez-Ruiz JM, Spínola MC, Cejudo FJ (2004) A novel NADPH thioredoxin reductase, localized in the chloroplast, which deficiency causes hypersensitivity to abiotic stress in Arabidopsis thaliana. Journal of Biological Chemistry 279: 43821-43827 142 Shaar-Moshe L, Hübner S, Peleg Z (2015) Identification of conserved drought-adaptive genes using a cross-species meta-analysis approach. BMC plant biology 15: 111 Siddiq M, Uebersax MA (2012) Dry beans and pulses production and consumption—an overview. Dry Beans and Pulses Production, Processing and Nutrition: 1-22 Singh SP (2007) Drought resistance in the race Durango dry bean landraces and cultivars. Agronomy journal 99: 1219-1225 Singh SP, Gepts P, Debouck DG (1991) Races of common bean (Phaseolus vulgaris, Fabaceae). Economic Botany 45: 379-396 Singh SP, Gutiérrez JA (1984) Geographical distribution of the DL1 and DL2 genes causing hybrid dwarfism in Phaseolus vulgaris L., their association with seed size, and their significance to breeding. Euphytica 33: 337-345 Slewinski TL (2011) Diverse functional roles of monosaccharide transporters and their homologs in vascular plants: a physiological perspective. Molecular Plant 4: 641-662 Song Q, Jia G, Hyten DL, Jenkins J, Hwang E-Y, Schroeder SG, Osorno JM, Schmutz J, Jackson SA, McClean PE (2015) SNP assay development for linkage map construction, anchoring whole-genome sequence, and other genetic and genomic applications in common bean. G3: Genes, Genomes, Genetics 5: 2285-2290 Su Y-H, Frommer WB, Ludewig U (2004) Molecular and functional characterization of a family of amino acid transporters from Arabidopsis. Plant physiology 136: 3104-3113 Taiz L, Zeiger E (2010) Plant physiology 5th Ed. Sunderland, MA: Sinauer Associates Takayama S, Reed JC, Homma S (2003) Heat-shock proteins as regulators of apoptosis. Oncogene 22: 9041 Tapia-López R, García-Ponce B, Dubrovsky JG, Garay-Arroyo A, Pérez-Ruíz RV, Kim S-H, Acevedo F, Pelaz S, Alvarez-Buylla ER (2008) An AGAMOUS-related MADS-box gene, XAL1 (AGL12), regulates root meristem cell proliferation and flowering transition in Arabidopsis. Plant physiology 146: 1182-1192 Tar'an B, Michaels TE, Pauls KP (2002) Genetic Mapping of Agronomic Traits in Common Bean. Crop Science 42 Terán H, Singh SP (2002) Selection for drought resistance in early generations of common bean populations. Canadian journal of plant science 82: 491-497 Thompson MD, Brick MA, McGinley JN, Thompson HJ (2009) Chemical composition and mammary cancer inhibitory activity of dry bean. Crop Science 49: 179-186 143 Tien Lea D, Duc Chua H, Quynh Lea N (2016) Improving Nutritional Quality of Plant Proteins Through Genetic Engineering. Current genomics 17: 220-229 Trapp JJ, Urrea CA, Cregan PB, Miklas PN (2015) Quantitative Trait Loci for Yield under Multiple Stress and Drought Conditions in a Dry Bean Population. Crop Science 55 Traub J, Kelly JD, Loescher W (2017) Early Metabolic and Photosynthetic Responses to Drought Stress in Common and Tepary Bean. Crop Science 57: 1670 Traub J, Kelly JD, Loescher W (2017) Early Metabolic and Photosynthetic Responses to Drought Stress in Common and Tepary Bean. Crop Science 57: 1670-1686 Uga Y, Sugimoto K, Ogawa S, Rane J, Ishitani M, Hara N, Kitomi Y, Inukai Y, Ono K, Kanno N (2013) Control of root system architecture by DEEPER ROOTING 1 increases rice yield under drought conditions. Nature genetics 45: 1097 Van Nocker S, Ludwig P (2003) The WD-repeat protein superfamily in Arabidopsis: conservation and divergence in structure and function. BMC genomics 4: 50 Verelst W, Twell D, de Folter S, Immink R, Saedler H, Münster T (2007) MADS-complexes regulate transcriptome dynamics during pollen maturation. Genome biology 8: R249 Wang J, Hou B (2009) Glycosyltransferases: key players involved in the modification of plant secondary metabolites. Frontiers of Biology in China 4: 39-46 Wang Y, Lin S, Song Q, Li K, Tao H, Huang J, Chen X, Que S, He H (2014) Genome-wide identification of heat shock proteins (Hsps) and Hsp interactors in rice: Hsp70s as a case study. BMC genomics 15: 344 Wang Z, Xing S, Birkenbihl RP, Zachgo S (2009) Conserved functions of Arabidopsis and rice CC-type glutaredoxins in flower development and pathogen response. Molecular plant 2: 323-335 Welch RM, House WA, Beebe S, Cheng Z (2000) Genetic selection for enhanced bioavailable levels of iron in bean (phaseolus v ulgaris L.) seeds. Journal of Agricultural and Food Chemistry 48: 3576-3580 Wentworth M, Murchie EH, Gray JE, Villegas D, Pastenes C, Pinto M, Horton P (2006) Differential adaptation of two varieties of common bean to abiotic stress: II. Acclimation of photosynthesis. Journal of experimental botany 57: 699-709 White JW, Castillo J, Ehleringer J, Garcia J, Singh SP (1994) Relations of carbon isotope discrimination and other physiological traits to yield in common bean (Phaseolus vulgaris) under rainfed conditions. The Journal of Agricultural Science 122: 275-284 144 White JW, Izquierdo J (1991) Physiology of yield potential and stress tolerance. Common beans: Research for crop improvement. CAB International., Wallingford, UK: 287-383 White JW, Singh SP (1991) Sources and inheritance of earliness in tropically adapted indeterminate common bean. Euphytica 55: 15-19 Wong CE, Singh MB, Bhalla PL (2013) Novel members of the AGAMOUS LIKE 6 subfamily of MIKC C-type MADS-box genes in soybean. BMC plant biology 13: 105 Wortmann CS (1998) Atlas of common bean (Phaseolus vulgaris L.) production in Africa. CIAT Wright EM, Kelly JD (2011) Mapping QTL for seed yield and canning quality following processing of black bean (Phaseolus vulgaris L.). Euphytica 179: 471-484 Xing S, Zachgo S (2008) ROXY1 and ROXY2, two Arabidopsis glutaredoxin genes, are required for anther development. The Plant Journal 53: 790-801 Xu K, Chen S, Li T, Ma X, Liang X, Ding X, Liu H, Luo L (2015) OsGRAS23, a rice GRAS transcription factor gene, is involved in drought stress response through regulating expression of stress-responsive genes. BMC plant biology 15: 141 Yang Y, Hammes UZ, Taylor CG, Schachtman DP, Nielsen E (2006) High-affinity auxin transport by the AUX1 influx carrier protein. Current Biology 16: 1123-1127 Yee D, Goring DR (2009) The diversity of plant U-box E3 ubiquitin ligases: from upstream activators to downstream target substrates. Journal of Experimental Botany 60: 1109- 1121 Yoo SK, Wu X, Lee JS, Ahn JH (2011) AGAMOUS‐LIKE 6 is a floral promoter that negatively regulates the FLC/MAF clade genes and positively regulates FT in Arabidopsis. The Plant Journal 65: 62-76 Yu H, Xu Y, Tan EL, Kumar PP (2002) AGAMOUS-LIKE 24, a dosage-dependent mediator of the flowering signals. Proceedings of the National Academy of Sciences 99: 16336- 16341 Yuste-Lisbona FJ, Gonzalez AM, Capel C, Garcia-Alcazar M, Capel J, De Ron AM, Lozano R, Santalla M (2014) Genetic analysis of single-locus and epistatic QTL for seed traits in an adapted x nuna RIL population of common bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics 127: 897-912 Zargar SM, Mahajan R, Nazir M, Nagar P, Kim ST, Rai V, Masi A, Ahmad SM, Shah RA, Ganai NA (2017) Common bean proteomics: present status and future strategies. Journal of proteomics 145 Zhang Z, Ersoz E, Lai C-Q, Todhunter RJ, Tiwari HK, Gore MA, Bradbury PJ, Yu J, Arnett DK, Ordovas JM (2010) Mixed linear model approach adapted for genome-wide association studies. Nature genetics 42: 355 Zhou C, Labbe H, Sridha S, Wang L, Tian L, Latoszek‐Green M, Yang Z, Brown D, Miki B, Wu K (2004) Expression and function of HD2‐type histone deacetylases in Arabidopsis development. The Plant Journal 38: 715-724 146