Genomics, genetics and breeding of common bean in Africa: A review of tropical legume project

Abstract Common bean (Phaseolus vulgaris L.) is an important legume crop worldwide. The International Centre for Tropical Agriculture (CIAT) and its national partners in Africa aim to overcome production constraints of common bean and address the food, nutrition needs and market demands through development of multitrait bean varieties. Breeding is guided by principles of market‐driven approaches to develop client‐demanded varieties. Germplasm accessions from especially two sister species, P. coccineus and P. acutifolius, have been utilized as sources of resistance to major production constraints and interspecific lines deployed. Elucidation of plant mechanisms governing pest and disease resistance, abiotic stress tolerance and grain nutritional quality guides the selection methods used by the breeders. Molecular markers are used to select for resistance to key diseases and insect pests. Efforts have been made to utilize modern genomic tools to increase scale, efficiency, accuracy and speed of breeding. Through gender‐responsive participatory variety selection, market‐demanded varieties have been released in several African countries. These new bean varieties are a key component of sustainable food systems in the tropics.

bean varieties (Buruchara et al., 2011). There is a notable increase in bean production in most African countries in the last 10 years most likely as a result of an increase in the area planted ( Figure 1). However, on-farm productivity remains low averaging of 850 kg/ha (Figure 1; FAOSTAT, 2016) compared to 2.5-5 t/ha that is achievable (Muthoni et al., 2017). More market-driven African countries report higher productivity (yield/ha) probably because they are able to adopt and use improved crop technologies (Table 1) due to the assurance of market.

| AD DRESSING MAJOR CONSTRAINTS TO BEAN PR ODUCTION
The low yield growth rates shown in Figure 1 could be attributed to a number of field-based production constraints .
Common bean is typically not well adapted to extreme environments of heat, drought and excessive rainfall. Impacts of climate change on agriculture and bean productivity in particular have been discussed by Beebe et al. (2011), Boko et al. (2007), Christensen, Carter, Rummukainen, and Amanatidis (2007) and IFAD (2011) among other authorities. Crop improvement through breeding brings immense value relative to investment and offers an effective approach to improving food security (Tester & Langridge, 2010). Since 1996, CIAT's bean research and in particular the development of improved bean varieties for the smallholder farmers in SSA have been coordinated through the Pan Africa Bean Research Alliance (PABRA (www.pabra-africa.org) (Buruchara et al., 2011). Common bean breeding programmes in PABRA are hinged on three thematic areas: (i) improved dry bean varieties resistant to multiple environmental (biotic/ abiotic) and climate-change related stresses; (ii) micronutrient-rich bean varieties and (iii) high value bean varieties targeted to niche markets. Three cross-cutting themes are emphasized, that is, yield potential, multiple stress tolerance and enduser traits such as cooking time and canning quality. This study presents a review of the progress made in addressing major production constraints of common bean, hinging on the achievements of the Tropical Legumes (TL) project http://tropicallegumes.icrisat.org/ supported by the Bill and Melinda Gates Foundation (BMGF) being conducted by CIAT and the national bean programmes of Ethiopia, Tanzania and Uganda. The study also includes research conducted by other institutions on subjects relevant to the TL project. Three breeding pipelines F I G U R E 1 Common bean production (tonnes) vs. area under beans (ha) worldwide and in Africa over a 15 year period (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015) [Colour figure can be viewed at wileyonlinelibrary.com] T A B L E 1 Dry bean production in selected African countries FAOSTAT database (Source: FAOSTAT, 2014).
are guiding the products being developed under the TL project. They include (i) bush and climbing bean breeding lines bred for drought tolerance, high mineral content, low P or N tolerance, (ii) bush and climbing bean breeding lines with heat and/or drought tolerance, and (iii) bush and climbing bean breeding lines for insect pest and disease resistance.
Constraints to bean production are addressed through conducting strategic research to understand and utilize the available genetic diversity of the common bean (prebreeding), elucidate and exploit the biological basis of the bean plant for productivity gains, understand and utilize knowledge of the genetics and physiology to inform breeding strategies, and use this knowledge to develop superior client-demanded varieties adaptable to environments in target regions.

GENETIC DIVERSITY FOR COMMON BEAN IMPROVEMENT
CIAT holds in trust the international Phaseolus genebank at its headquarters in Colombia, South America, and also maintains continuously increasing active collections of bean germplasm at the bean programme headquarters in Colombia and its Africa regional offices in Uganda (4,000 accessions currently) and Malawi (3,000 accessions currently).
This huge diversity is tapped by plant breeders all over the world to continually improve specific traits. Four Phaseolus sister species, P. coccinues, P. acutifolius, P. dumosus and P. costaricensis, have been exploited to access unique genes for local selection .
P. dumosus and P. coccineus and P. costaricensis are native to competitive environments like that of wild common bean, but typically in somewhat more humid ecologies and are of interest as sources of resistance to diseases associated with humid environments (Singh, 2001). Tepary bean (Phaseolus acutifolius) evolved in the semi-arid to arid environment where light is abundant and competition less intense, but moisture is severely limiting. Tepary bean also presents resistance to common bacterial blight (Xanthomonas axonopodis), leaf miner (Empoasca kraemeri) and bruchids (Acanthoscelides obtectus) (Singh, 1992).

FOR PRODUCTIVITY GAINS
Understanding the physiology of key traits helps breeder to promote faster breeding progress, improve selection methodologies and inform phenotyping protocols. In addition, enhancing the basic understanding of the biology of beans in general, traits of interest, for example, mechanisms of drought tolerance, yield, and virus resistance informs the breeding process. A study by Aruajo and Teixeria (2008) showed that grain yield of different common bean cultivars was not intrinsically associated with vegetative vigour at flowering and that mechanisms during pod filling could strongly influence the final crop yield. The establishment of a profuse root system during pod setting, associated with the continuous N and P acquisition during early pod filling, seemed to be relevant for higher grain yields of common bean.

| Genetics, physiology and breeding for drought tolerance in common bean
Water scarcity, abundance, variability, date of onset of rains and length of the growing seasons or their combinations have direct impact on bean productivity. Drought limits the productivity of 50% of the arable land prompting competition for water (Cattivelli et al., 2008;Rosegrant, Ringler, & Zhu, 2009). Drought tolerance improvement will likely benefit 3.8 million ha in the 2020s . When the Tropical legume project was initiated, breeding for drought tolerance in common bean that had received only sporadic attention gained prominence and moved into the research agenda. thates from stems to pods and from pod walls to grain , pod partitioning index (PPI), harvest index (HI) and pod harvest index (PHI) (Polania et al., 2016) and basal root whorl number (BRWN) (Lynch, 2011) have also been identified as some of the physiological mechanisms governing drought tolerance. Breeders select for high yield potential under drought and irrigated conditions and also consider secondary traits such as Pod Harvest Index (PHI) in the selection index. Several drought QTL have been identified by CIAT Scientists (Blair et al., 2010;Diaz et al., in preparation), and at other institutes (Mukeshimana, Butare, Cregan, Blair, & Kelly, 2014;Trapp, Urrea, Cregan, & Miklas, 2015). Candidate QTL linked to PHI are being validated through additional phenotyping . GWAS analysis of an 8-parental MAGIC population revealed yield QTL on three chromosomes (Izquierdo et al., in preparation) which are in the validation process. However, other factors such as soil factors and poor soil fertility limit the expression of drought tolerance as they do not permit adequate plant development for crops to sustain additional physiological stress imposed by drought .

| Genetics, physiology and breeding for heat tolerance
Common bean is adapted to relatively cool climatic conditions, and temperatures of >30°C during the day or >20°C at night result in yield reduction (Porch, 2006). High temperatures were shown to aggravate the stress imposed by drought, and combinations of stress tolerance would be necessary in the near future .
Heat stress manifests in decline of photosynthetic leaf area, death of flowers, flower abortion, shortening of grain-filling period, reduced pollen viability and seed weight, impaired development of yield components including ovaries (Hatfield et al., 2011) resulting in few seeds and decline in grain yield potential. Heat tolerance indices, geometric mean (GM) and stress tolerance index (STI) (Porch, 2006), extent of abscission of reproductive organs (Rainey & Griffith, 2005), chlorophyll a fluorescence (Stefanov, Petkova, & Denev, 2011) (Blair, Iriarte, & Beebe, 2006) have been developed. Introgression of heat tolerance from P. acutifolius in backgrounds of more acceptable seed types and evaluating newly developed lines confirmed that P. acutifolius is an important and useful genetic resource for improving heat tolerance in common bean Polania et al., 2017).

| Genetics, physiology and breeding for low soil fertility tolerance
Considerable genetic variability has been detected from field evaluations, and genotypes with specific single or multiple edaphic stress tolerance (low N, low P and soil acidity with the associated Al and/ or manganese (Mn) toxicities) have been identified (Lunze et al., 2012). Long-term research using common bean has contributed to defining root phenes and their role in enhanced soil exploration and P acquisition (Lynch, 2011). The genetics of N fixation and low P tolerance was evaluated by Diaz et al. (2017). Traits such as greater BRWN (Lynch, 2011), percentage of nitrogen derived from atmosphere (%Ndfa) (Mehdi, 2015;Rao, Miles, & Beebe, 2016), biological nitrification inhibition (BNI) (Subbarao, Yoshihashi, & Worthington, 2015), receptor kinases, transmembrane transporters, and transcription factors (Kamfwa, Zhao, Kelly, & Cichy, 2017) have been employed in selecting for tolerance to specific mineral deficiencies.
Lines developed by CIAT for combined drought and low soil fertility tolerance are being evaluated in Ethiopia, Uganda and Tanzania.
Some of these lines have been found to be resistant to Pythium and Fusarium root rot .

RESISTANCE TO MAJOR FIELD AND POSTHARVEST INSECT PE STS
Although a multitude of insect pests attack beans, bean stem mag- Bean stem maggot (BSM) is generally regarded as the principal insect pest of beans throughout Africa causing up to 50%-100% yield losses especially when seedlings are attacked (Songa, 1999).
BSM infestation is aggravated by drought (Ojwang et al., 2010) and also occurs in association with bean root rots (Ochilo, 2013). The use of host plant resistance against BSM is supposed to be more effective in the management of BSM (Abate, 1990;Murenju, 2015) though not absolute (Belmain, Haggar, Holt, & Stevenson, 2013). Conde-Petit, Cardona, & Dorn, 2008) in wild bean accessions has been exploited in developing bruchid-resistant common bean germplasm. Bean genotypes with arcelin based resistance have been developed (Cardona, 2004;Beneke, 2010) and markers tagging this resistance also developed.

RESISTANCE TO KE Y DISEASES IN AFRICA
Diseases are the second most important constraints to bean production, after abiotic factors in Africa causing up to 80%-100% yield loss (Wortmann et al., 1998). Major success has been in breeding for resistance to angular leaf spot, anthracnose, and common bacterial blight, bean root rot and bean common mosaic virus. Although resistance to angular leaf spot (ALS) (Psuedocercospora griseola) is mostly a monogenic trait, the pathogen is highly variable with many different races (Mahuku, Henriquez, Munoz, & Buruchara, 2002). Three ALS resistance genes are mapped and named following the guidelines for gene . Gene pyramiding has been suggested to provide resistance to a wide range of the ALS pathotypes (Miklas, Kelly, Beebe, & Blair, 2006;CIAT, 2007). Bean anthracnose is a highly variable pathogen, new pathotypes reportedly keep emerging time after time (Leaky and Simbwa-Bunya, 1972;Nkalubo, 2006;Pastor-Corrales & Tu, 1989). Resistance to this pathogen is conditioned by nine independent resistances (Co1-Co-10). Information on pathogenic variability present in production areas is essential in designing effective gene pyramids in addition to continued evaluation of resistance sources as the genes differ in their effectiveness in controlling variable races. The genotype G2333 which possesses Co-4 2 , Co-5 and Co-7 resistance genes (Young, Melotto, Nodari, & Kelly, 1998) has been utilized routinely in introgressing Anthracnose resistance. Gene pyramiding is suggested to provide efficient long-term control of bean anthracnose (Balardin & Kelly, 1998).  (Mukankusi et al., 2011). QTL related to FRR resistance and root/ shoot biomass were identified in RIL populations of MLB-49-89A (Weijia et al., in press) and Puebla 152 (Nakedde et al., 2016). Bean common mosaic virus (BCMV) and bean common mosaic necrosis virus (BCMNV) are the most widespread and important viral diseases affecting production of common beans in Africa (Spence & Walkey, 1995) causing up to 80% yield loss (Morales, 2003;Wortmann, 1998).
A number of BCMV and BCMNV resistance genes have been identified and tagged. They include the single dominant I and the recessive bc-u, bc-1, bc-1 2 , bc-2, bc-2 2 and bc-3 genes (Drijfhout, 1978;Miklas & Kelly, 2002). The dominant I gene inhibits all known strains of the BCMV (Drijfhout, 1978). However, due to occurrence of BCMNV (a fact with consequences that were unknown when the "I" gene was introduced in bean lines in Africa) germplasm containing the I gene (introduced or developed in the region) faced an unanticipated problem.
When the "I" gene containing material is invaded by the BCMNV strains, the "I" gene responds by producing excess phaseolin in the vascular system of the inoculated leaf and in the plant apex. This results in the death of the apex (a condition known as systemic top necrosis or black root) and discoloration of the vascular tissue due to downward movement of phaseolin, and eventually death of the plant (Kelly, 1997). This limits their usefulness of germplasm with the dominant I gene in the presence of the BCMNV strains. Protection of the "I" gene by combining it with race-specific resistance recessive genes (typically bc-3 or bc-2 2 ) and introgressing this resistance into key materials that neither have "I" gene nor any other type of resistance against BCMV or BCMNV have been used as the most suitable strategies to provide stable and broad-based resistance.

| Biofortification in common bean
Biofortification research was initiated following justification of the prevalence of high levels of undernutrition due to nutrient deficiencies including iron deficiency anaemia (Petry et al., 2015;Mulambu et al., 2017;Blair et al., 2013). Grain mineral levels ranging from 30 to 110 ppm for iron and 25-60 ppm for zinc have been found from screening of bean germplasm accessions from the global gene bank and local collections from ten African countries Mukamuhirwa, Tusiime, & Mukankusi, 2015;Mulambu et al., 2017). The highest concentrations were often found in progenitors or wild relatives of common bean (Beebe et al., 2002;Islam, Basford, Jara, Redden, & Beebe, 2002). Substantial positive associations (60%-80%) were discovered between iron and zinc levels, which provided an opportunity for raising levels of both micronutrients simultaneously . Early product development involved identifying parental genotypes for use in crosses and understanding the genetics of the trait (Mulambu et al., 2017). High-iron genotypes were used to conduct crosses (including double-crosses with two or three high-iron parents) to combine the high-mineral trait with acceptable grain types and agronomic MUKANKUSI ET AL.
characteristics (Beebe et al., 2000) including wide crosses with P. dumosus and P. acutifolius (Beebe, 2012a;Beebe et al., 2012b). Genotype-by-environment (GxE) tests were conducted to verify that mineral accumulation was stable across sites and generations (Blair et al., 2010;Mukamuhirwa et al., 2015). The small-seeded Mesoamerican bush bean lines emerging from the breeding programme in Colombia have 80% higher iron and drought resistance that was equal to or superior to the tolerant check . The improvement of mineral levels in climbing bean materials has also been most successful and had an added advantage of increased productivity per unit area . Further improvements can be achieved because nutrient content was shown to be positively correlated with high yield potential and genotype x environment effects were small (Bationo, Waswa, Kihara, & Kimetu, 2007).

| Marker-assisted selection in common bean
Most progress with marker-assisted selection (MAS) in common bean breeding has been with disease resistance. Through the TL project, communication was established between CIAT and USDA to access sequence data to identify SNP markers. Under the Generation Challenge Program (GCP),~1,500 SNPs available through the BeanCAP project were converted to the KasPAR system at a genotyping outsourcing service provider. SNP markers for major disease resistance genes (for BCMNV, BGMV, CBB, bruchids, ALS) were developed, and markers of other classes (SCARS, SSRs) have been converted to a SNP platform for ready to use in gel-free systems for in-house genotyping or through the genotyping service provider (Table 2) The bc-3 gene is the only allele with a known mechanism of resistance to bean common mosaic virus (BCMV) disease and its necrotic strain, bean common mosaic necrosis virus (BCMNV). The bc-3 gene is also identified as the eIF4E allele carrying a mutated eukaryotic translation initiation factor gene (Naderpour et al. 2010).
To date, a CAPS and SNP marker based on the eIF4E gene have been developed for utilization using the Intertek and LGC platforms (  (2013) found a significant major QTL for resistance to Fusarium root rot in the resistant line MLB-49-89A. The study also found that the two markers PVBR87 and PVBR109 spanning the QTL are found on B3 of the common bean core map close to the region where resistance to root rots, anthracnose, common bacterial blight and bacterial brown spot have been previously mapped (Kamfwa et al., 2013).
RAPD markers that are tightly linked to angular leaf spot resistance genes were identified and some successfully converted to SCAR markers (CIAT, 2003;Namayanja et al., 2006). The protocol for their use in marker-assisted selection breeding was also developed (Mahuku, Jara, Cajiao, & Beebe, 2003 Awash-1 9 RAZ42; Awash-1 9 RAZ120) originating from Ethiopia using a real-time PCR platform. traits under local field conditions. Tools needed to make these measurements, process the data and extract useful information have been out of reach of most researchers, extension agents and farmers in Africa for a long time. Accurate cultivar performance data over a period of time are essential for making predictions for the future. The breeding management system (https://www.integratedbreeding.net/)

| Genomic selection
is a suite of interconnected software designed to help breeders manage day-to-day activities through all phases of their breeding programmes: from straightforward phenotyping to complex genotyping, providing necessary tools to conduct modern breeding in one comprehensive package, local database for a breeder to track. Under the Tropical legume III project, three countries Ethiopia, Uganda and Tanzania are uploading bean breeding data into the BMS. A server has been installed at CIAT HQ, and the first field books have been uploaded with the plan to follow some breeding generations with BMS in 2016 in parallel with the current data management system.
The use of electronic data collection gadgets is also being streamlined across these countries to help speed up the process of data collection and reduce errors. Servers of the BMS have also been installed at NaCRRI and EIAR.

RELE ASE
The CIAT bean programme is promoting principles that drive success in demand-led breeding that include (i) target-driven breeding approach; (ii) demand-led variety development strategy; and (iii) performance indicators to measure progress towards the adoption and widespread use of new plant varieties (Persley & Anthony, 2017). Product profiling is one of the best practices under a target-driven approach to variety development. This includes defining the type of product being developed and the market. In addition, factors that would affect the development of that product are also outlined and a breeding scheme designed with a timeline (Tropical legume III report, 2017). Product profiles for seven grain market classes that include large white, large red, small white, small red, large red speckled (sugar bean), large red mottled (calima type) and medium-to-large yellow beans were developed across the three countries, Ethiopia, Tanzania and Uganda. An example of a product profile for small white beans for Ethiopia bean programme is shown in Table 3. Under the TL project, a total of 71 market-demanded varieties with on-farm yield advantage of 10%-40% over the commercial varieties and additional traits of resistance to key pests and diseases and/or high grain Fe and Zn content were released in six countries, Ethiopia, Kenya, Malawi, Tanzania, Uganda and Zimbabwe over a 10-year period (Table 4).

| GENDE R-RE SPONSIVE BE AN BREEDING
The primary goal of bean breeding is to increase production in highly heterogeneous environments. Under the PABRA framework, effort has been made to integrate gender in the breeding process especially in participatory variety selection. Farmer participatory variety selection (PVS) is a step included in the later stages of the bean breeding process to ensure acceptability and eventual adoption (Gyawali, Sunwar, & Subedi, 2007)  depending on the gender roles within the value chain. For example, empirical evidence shows that criteria such as texture of bean leaves, keeping quality and cooking time are more important to women than men (Katungi et al., 2011). In Kenya, men are more likely than women to reject varieties with climbing habits as they interfere with the growth of their maize crop (Katungi et al., 2011).
Consequently, breeding has maintained its focus on achieving key acceptable traits (i.e., yield, resistance, marketability and taste) while minimizing those that will lead to rejection.
T A B L E 3 Product profile for Small white bean for export market with drought and disease tolerance in Ethiopia

CONFLI CT OF INTEREST STATEMENT
This paper is a review of the major outputs of the Tropical legume (TL) project Common bean breeding, genomics and genetics. It includes findings of the project but also highlights findings of other researchers on specific topics. As such, there is no conflict of interest.