Modern plant breeding based on the fundamental principles  of   inheritance   has   become    an   important component of agricultural science and technology. Conventional breeding methodologies  have proved to be broadly successful in development of plant cultivars and germplasm. However, conventional breeding is still dependent to a considerable extent on subjective evaluation and empirical selection using often highly variable and time-consuming phenotyping assays. Recent technological advances in molecular marker-assisted breeding (MAB) have brought new opportunities and prospects for optimizing the accuracy and efficiency of breeding systems through indirect selection.

In backcross breeding, the main objective is the introgression of one or more genes from a donor into the background of an elite variety and to recover the recurrent parent genome as rapidly as possible (Semagn et al., 2006). In the past, this was usually achieved by conventional backcross methods, but in many ways the same objective is being pursued through transgenic breeding, bypassing recombination altogether but introducing a value-added trait. Recurrent backcrossing is thus a traditional breeding method commonly employed to transfer alleles at one or more loci from a donor to an elite variety (Allard, 1960). This method requires many generations over several years before recovering the background of the elite cultivar. According to Semagn et al. (2006), the recovery of 99.2% of the elite cultivar could take at least six generations when there is no deviation. In cases involving a deviation, Young and Tanksley (1989) for example found an introgressed segment as large as 4 centiMorgans (cM) in tomato cultivars developed after 20 backcrosses, and one cultivar developed after 11 backcrosses still contained the entire chromosome arm carrying the gene from the donor parent.

During the past two decades, development of genomic resources has facilitated gene transfer using molecular markers. The use of SNP markers in MAB programs has progressed rapidly together with development of technologies and platforms for the discovery and high-throughput (HTP) screening of SNPs in many crops (Mammadov et al., 2012). The MAB technology allows transfer of target genome regions coupled with extensive genetic mapping and QTL discovery for the development of molecular markers for use in marker-assisted back-crossing (MABC)  and marker-assisted selection (MAS) (Semagn et al., 2006). Molecular markers are tools that can be used as chromosome landmarks to facilitate the introgression of genes associated with economically important traits. It is an approach that has been developed to avoid problems connected with conventional plant breeding by shifting the selection criteria from selection of phenotypes towards selection of genes that control traits of interest, either directly or indirectly. Unlike conventional backcross breeding, MAB can be viewed as a four-step selection process to quickly recover the recurrent parent genotype (Frisch et al., 1999). This includes: (1) selecting individuals carrying the targeted alleles, (2) selecting individuals homozygous for the recurrent parent genotype at loci flanking the target  locus,  (3)  selecting  individuals homozygous for recurrent parent genotype at remaining loci on the same chromosome comprising the targeted allele, and (4) selecting individuals that are homozygous for the recurrent parent genotype at most genotyped loci across the whole genome among those that remain.

The use of MAS for introgression of major quantitatively inherited trait loci for stress tolerance is increasingly being applied in crop improvement. However, the use of such technology has been slow in pulse breeding programmes (Kumar et al., 2011). Effort are made for the use of high-throughput genotyping platforms in pulses like chickpea, common beans and cowpea (Muchero et al., 2008; Muchero et al., 2009a; Muchero et al., 2009b).

In this study, the MABC method has been used to introgress drought-related QTLs (Muchero et al., 2009a; Muchero et al., 2009b), striga and nematode resistance genes from two IITA cowpea varieties into one local farmer-preferred variety.

Leaf sampling and DNA extraction 

A sampling kit (LGC Genomics, Teddington, UK) was used which is designed to facilitate both cutting of leaf discs and their transport and concomitant desiccation, for eventual DNA isolation. For each sample, 4 leaf discs of 6-mm-diameter were cut using the Harris Uni‐Core leaf-cutting tool supported by the Harris self‐healing cutting mat and placed into one well of a 96-well storage plate, whereupon the plate was sealed with a perforated (gas‐permeable) heat seal by applying a medium hot household iron to the top of the seal for about 2 seconds. The sealed plate was then placed in a heavy‐duty sealed bag in the presence of a desiccant to dehydrate and preserve the leaf tissue during transit at ambient temperature. Decontamination of the leaf-cutting tool between sampling different plants was achieved using 70% ethanol or 2% sodium hypochlorite. The leave samples were sent to LGC Genomics for DNA extraction and SNP genotyping using the KASP system.

Selection of markers

A set of 184 genome wide SNP markers spanning an average of 2-cM intervals was selected using the BreedIt^© SNP Selector tool (http://breedit.org/) developed at University of California – Riverside (UCR). The SNP Selector provides an interface to generate customized lists of SNPs based on cM distance between markers genome-wide, and markers flanking known trait positions. All SNP markers were developed from EST of drought-stressed tissues, so there is a chance the markers are associated with drought tolerance candidate genes (Muchero et al., 2009b).

Plant materials and QTL introgression procedures

Two drought-tolerant lines from IITA (IT93K-503-1 and IT97K-499-35) that were found to be drought tolerant in Burkina Faso (Sawadogo, 2009) based on their yielding and staying green abilities under water stress conditions and in which drought-tolerant QTL have been discovered and mapped (Muchero et al., 2008; Muchero et al., 2009a; Muchero et al., 2009b; Muchero et al., 2010; Muchero et  al., 2011)  were  used  as  donors  of  positive alleles of drought tolerance QTLs, and Striga and root-knot nematode resistance genes. ‘Moussa Local’, a local farmer-preferred purified variety from Burkina Faso, was used as the recurrent parent. The donor alleles for yield and nematode resistance were selected based on results from UCR/INERA collaborative on-going projects. The donor alleles for Striga resistance were selected based on synteny with the Striga resistance locus reported in Ouedraogo et al. (2002).

For the MABC scheme, IT93K-503-1 and IT97K-499-35 were crossed to Moussa Local to obtain F₁s. The F₁s were backcrossed to Moussa Local to obtain 95 BC₁F₁ seeds for each recurrent-donor combination. The BC₁F₁ seeds were planted in boxes in a greenhouse. Two weeks after planting leaf samples were collected from each plant and sent to LGC Genomics for genotyping with the 184 SNPs (supplemental Appendix 1). This allowed selection in each population of the BC₁F₁ individual plants that were heterozygous for SNPs associated with drought tolerance, Striga and/or nematode resistance (foreground SNPs) and carried as many recurrent-parent alleles as possible at other SNP loci (background SNPs) (supplemental Appendix 4).

The selected BC₁F₁ individual plants were backcrossed with Moussa Local to obtain 95 BC₂F₁ individuals that were SNP-genotyped. In the BC₂F₁ generation, the individual plants that were heterozygous for foreground SNPs and carried as many recurrent-parent alleles as possible at background SNPs were identified. Alleles A and B are designated for Moussa local and IT93K-503-1 or IT97K-499-35, respectively (supplemental, Appendix 2 and 3).

In the next cycle, each of the selected BC₂F₁ individual plants was backcrossed to Moussa Local to create BC₃F₁ lines. Four BC₃F₁ individuals from the cross Moussa Local/IT97K-499-35 and three BC₃F₁ individuals from the cross Moussa Local/IT93K-503-1 were selfed to obtain about forty BC₃F₂ seeds per line. Seed from BC₃F₂ were used for morphological characterization of the families and yield performance estimation. Ten entries (6 MABC lines and 4 controls) were planted using a randomized complete block design (RCBD) with two replications in two water regimes-water-stressed (WS) and well-watered (WW). The parents used in the introgression process  (IT97K-499-35, IT93K-503-1, and Moussa Local) and one known drought tolerant variety (Gorom Local) were used as checks. The trial was conducted during the off-season in 2014 from April to June under a drip-irrigation system and striga naturally infested field at the Kamboinsé Research Station, near Ouagadougou, Burkina Faso. 

QTLs introgression

Genotyping of the BC₁F₁ identified the plant named M503_BC1F1_31 carrying the donor IT93K-503-1 alleles for yield under drought, Striga and nematode resistance, and about 67% of recurrent parent Moussa Local alleles at background markers. In the cross from Moussa and IT97K499-35, the plant named M499_BC1F1_04 carried the donor IT97K-499-35 alleles for yield under drought and Striga resistance, and about 70% of Moussa Local alleles at background markers. In addition, some other BC₁F₁plants (M499_BC1F1_49, M499_BC1F1_48, M499_BC1F1_44, M503_BC1F1_54 and M503_BC1F1_92) carrying donor positive alleles but less Moussa Local background than M503_BC1F1_31 and M499_BC1F1_04 were also selected for backcrossing to Moussa  Local.   In total,  190  individuals  were  obtained from the BC₂ backcrosses. Genotyping of these BC₂F₁ plants identified 10 individuals carrying different combinations of donor IT93K-503-1 positive alleles and 80 – 97% of Moussa Local background. Three selected plants with highest Moussa Local background (M503_BC2F1_54P15, M503_BC2F1_54P8, and M503_BC1F2_92P27) were backcrossed to Moussa Local to generate the M503_BC3F1 families. Likewise, 21 plants were selected in the BC₂F₁ population; they carried different combinations of donor positive alleles for yield and Striga resistance and 69 – 93% of recurrent Moussa Local background. Five selected plants with highest Moussa Local background (M499_BC2F1_48P90, M499_BC2F1_44P19, M499_BC2F1_48P93, M499_BC2F1_48P85, and M499_BC2F1_4P67) were backcrossed to Moussa Local to generate the BC₃F₁ families. A total of six families derived from the two donors were retained and selfed (fiveM499_BC3F2s and oneM503_BC3F2s) to increase seed for further studies.

Morphological characterization of the MABC selected lines

Seed from BC₃F₂ were not enough to undertake a multi-location trial, so they were used for morphological characterization of the families in a single site yield trial. Table 1 shows the morphological characteristics of the selected families and the recurrent parent. Figure 1 also shows the plant type or growth habit of the lines and their dry pods form and color in comparison to Moussa Local. 

Yield performance of the MABC selected lines

The yields of the ten lines per water regime are represented in Figure 2. The yields ranged between 287.2 and 1184.70 Kg ha^-1 in the water-stressed environment, while in the well-watered environment the yields were higher, ranging from 272.3 to 1771.0 Kg ha^-1. Three BC3F2 families (M499_BC3F3_48P85, M499_BC3F3_4P67, and M499_BC3F3_48P90) yielded better than all parents and the local control Gorom Local. All BC3F2 families appeared to perform better than the recurrent parent Moussa Local under limited water conditions.

This study involving MABC methodology using SNPs breaks new ground for cowpea breeding, particularly in selection for drought tolerance. The methodology enabled rapid recovery of the recurrent-parent (Moussa Local) background (up to 97%) with only two backcross cycles (BC₂) by using ninety-five individuals for each set of backcrosses   in   the   BC₁   generation.   This  expedient recovery of the recurrent parent background allowed early selection and helped to minimize population sizes at each generation, therefore reducing the required time and work load. The levels of recovery of the recurrent parent background confirmed the findings reported by Jiang (2013) that revealed a percentage of recovery of 98% at BC3 with a number of 100 individuals selected at BC₂.

In the present study, several donor loci (yield under drought, stay-green, Striga resistance and root-knot nematode resistance) were introgressed simultaneously. This decreases the chance to identify a line carrying all donor alleles and high Moussa background and therefore, limits the number of offspring to be selected in the subsequent generations. Sebolt et al. (2000) also reported that the rate of success decreases when large numbers of QTLs are targeted for introgression; by using MABC for two QTLs for seed protein content in soybean introgression, they eventually found that only one QTL was confirmed in BC3F4:5. Compared to MABC, conventional backcross breeding, however, requires a much larger backcross population of 500 or more plants to be produced to ensure that there are sufficient plants for background selection after the foreground and recombinant selection have been performed. During this process, unless breeders screen the material to identify those that are carrying the gene of interest, they may need to conduct ‘blind crossing’ to the recurrent parent. In such   conditions    conventional    backcrossing   can   be extremely time-consuming and inefficient. By using a combination trait-flanking markers and evenly distributed markers across the recurrent parent genome, effective introgression can be done to avoid linkage drag and the use of large numbers of individuals.

In Burkina Faso, most of the cowpea landraces have a prostrate growth habit and are susceptible to Striga and are often grown with cereal crops like sorghum and millet. The prostrate growth habit allows soil conservation and maintenance of soil moisture by the cover of the vines. These morphological characteristics in the selected advanced BC lines are in accordance with the expectations of recovering the recurrent parent back-ground (Figure 1). Moussa Local, a farmer-preferred landrace has some purple pods which remain purple even for dry pods. The six selected lines had the purple green and dry pod color character confirming that they recovered this character from the recurrent parent. The prostrate (spreading) growth habit of Moussa Local was also found in the selected MABC lines. These results, therefore, confirmed the molecular results that showed a high level of recovery of the Moussa Local recurrent parent plant type. The same trend was observed for Striga resistance. Only one line (M499_BC3F3_48P93) that had no Striga donor allele had Striga emergence in the well-watered environment. The Striga-resistant checks did not emerge Striga confirming their resistance while Moussa local had emerged Striga confirming its susceptibility  to  Striga. This  result also confirms that the lines selected based on the presence of the Striga gene through the MABC introgression are effective in controlling Striga. 

Yield is by far the most important criterion for varietal selection by African farmers (Tignegre, 2010; Some, 2012; Traore, 2013). Promising results from the preliminary yield performance trial indicated that three lines yielded better than the parents and the drought-tolerant control Gorom Local (Figure 2). In addition, the general performance of these lines reached the potential yield of the recently released varieties in Burkina Faso which is around 1.5 t.ha^-1 (Ouedraogo et al., 2012). The low yields of certain lines could be attributed to the fact that there were drought spells due to water shortages during the growing period at Kamboinsé. However, other authors have reported in maize three QTLs for two traits (earliness and yield) were introgressed between maize elite lines with MABC but the results were influenced by the function of other genes controlling the traits (Bouchez et al., 2002). 

In the domain of molecular breeding, a lot of conventional breeding methods have been associated with markers to design a large number a marker-aided selection methods. Some examples are marker-assisted selection (MAS), marker-assisted recurrent selection (MARS), marker-assisted backcrossing (MABC), and ultimately genomic selection (GS). Among the molecular breeding methods, MABC has been the most widely and successfully used in plant breeding up to date. Marker-assisted backcrossing (MABC) is an effective method for developing improved versions of widely cultivated varieties, also referred to as Mega varieties (Neeraja et al., 2007). It has been applied to different types of traits (e.g. disease/pest resistance, drought tolerance and quality) in many species, e.g. rice, wheat, maize, barley, pear millet, soybean, tomato, etc. (Collard et al., 2005; Dwivedi et al., 2007; Xu, 2010).  

The improvement of cowpea varieties by adding traits through MABC is now becoming a significant advance-ment in Africa, and work is continuing to transfer drought tolerance QTL, Striga resistance, aphid and nematode resistance among other traits into a range of economically important parental lines at INERA. The use of the marker technology tested in this study has allowed rapid recovery of the background of a farmer preferred widely grown cowpea landrace. The morphological characteristics of the improved lines developed in this study indicated that they could be good candidates for production under intercropping in farmer fields. Until now, no Striga resistant variety combining preferences has been proposed by INERA for intercropping. Since Moussa Local is used in intercropping, these improved lines hold much promise for performance under this production system. From this study three of the six  MABC-improved lines are most promising based on good yield potential and their resistance to Striga. These three lines (M499_BC3F3_48P85, M499_BC3F3_4P67, and M499_BC3F3_48P90) are being seed-increased for multi-location trials to measure their performance and ability to withstand drought events and Striga attack. 

The authors have not declared any conflict of interests.

This work was supported in part by The Generation Challenge Program of the CGIAR, USAID Legume Innovation Lab, AGRA, Kirkhouse Trust and WACCI. The guidance of Dr. J. D. Ehlers is gratefully acknowledged.