Introduction

Cowpea (Vigna unguiculata L.) (2n = 2x = 22) is an important grain legume, fodder and vegetable crop in many tropical and subtropical countries. Cowpea grain is a staple crop on the vast drought-prone savanna and Sahelian zones of sub-Saharan Africa, and widely grown in lower elevation areas of eastern and southern Africa, in northeastern Brazil, parts of the Middle East, India, and the southeastern and southwestern regions of North America (Ehlers and Hall 1997; Langyintuo et al. 2003). Cowpea hay plays a critical role in feeding animals during the dry season in West Africa (Tarawali et al. 2002).

Cowpea is presumed to have evolved in Africa, because wild cowpeas only exist in Africa and Madagascar (Steele 1976). Although the center of diversity of wild Vigna species is in southeastern Africa (Padulosi and Ng 1997), West Africa is a major center of diversity of cultivated cowpea (Faris 1965; Ng and Padulosi 1988) and several studies have concluded it was probably domesticated by farmers in this region (Faris 1965). There is also evidence that domestication occurred in northeastern Africa based on studies of amplified fragment length polymorphism (AFLP) markers (Coulibaly et al. 2002) and could have occurred simultaneously with the domestication of sorghum (Sorghum bicolor) and pearl millet (Pennisetum typhoides) in the third millennium B.C. (Steele 1976). The wild cowpea, V. unguiculata ssp. unguiculata var. spontanea is the likely progenitor of cultivated cowpea (Pasquet 1999).

Despite considerable phenotypic diversity that exists in cultivated cowpea germplasm, limited genetic diversity in cowpea breeding programs is of special concern because cowpea appears to have lower inherent genetic diversity than other cultivated crops as a result of a hypothesized single domestication event (Pasquet 1999, 2000). This genetic bottleneck may account for the low levels of polymorphism in isozyme markers (Panella and Gepts 1992; Vaillancourt et al. 1993), seed storage proteins (Fosto et al. 1994), random amplified polymophic DNA (RAPD) markers (Ba et al. 2004), and AFLP markers (Coulibaly et al. 2002) that have been observed.

Cultivated cowpeas have been divided into five cultivar groups (Unguiculata, Melanophthalmus, Sesquipedialis, Biflora, and Textilis) based mainly on pod and seed characteristics (Pursglove 1966; Pasquet 1998). Subspecies status has been given to the Sesquipedialis group by some taxonomists (e.g. USDA GRIN Taxonomy http://www.ars-grin.gov/cgi-bin/npgs/html/taxon.pl?41646).

Significant long-term cowpea genetic improvement efforts have taken place in several West African countries and the United States, as well as at the International Institute of Tropical Agriculture (IITA) based in Ibadan, Nigeria. Cowpea breeding programs have been active for many years in West Africa, particularly by the Savanna Agricultural Research Institute (SARI) in Ghana, and by the Institute of Agricultural Research (IAR) in Nigeria. For more than 50 years cowpea breeding has been conducted by the Institut Senegalais Recherche Agronomique (ISRA) in Senegal, for nearly 30 years by the Institut de l’Environment et Recherche Agronomique (INERA) in Burkina Faso, and for more than 20 years by the Institut de la Recherche Agronomique pour le Développement (IRAD) in Cameroon. At the University of California and the United States Department of Agriculture (USDA) U.S. Vegetable Lab in South Carolina, cowpea breeding has been conducted for more than 60 years. Active germplasm exchanges between the University of California and several African programs has resulted in development of improved African varieties with parentage from the California program (Elawad and Hall 2002; Padi et al. 2004a, b) and California varieties having African parentage (Ehlers et al. 2000). IITA has had an active program to distribute advanced breeding lines to ISRA, INERA, SARI, IRAD and US programs for over 30 years (Singh et al. 2002); and many of these IITA breeding lines have been used as parents to develop new breeding lines and varieties in Senegal (Cisse et al. 1995a, b), Ghana (Padi 2004a), Burkina Faso and Cameroon (Hall et al. 2003) and recently in the US (see Table 1). The extent to which this interchange has helped to maintain/augment genetic diversity in these West African and US cowpea breeding programs is not known.

Table 1 List genotypes, origin, type, utilization, parental origina and cultivar group of cowpea accessions used in the study
Full size table

In general, a very low frequency of usable segregates is obtained from crosses with ‘exotic’ genotypes of cowpea (Ehlers and Foster 1993). Thus, breeding programs, which must focus most of their efforts on rapid delivery of varieties with a specific range of production and quality traits, tend to cross and re-cross cultivars containing these production and quality traits, and many of them are related to some degree. This leads to reduced genetic variability among cultivars that are released and among advanced breeding lines in the program, both of which are likely to be used as parents in new breeding cycles. Similarly, when a new trait (e.g. pest resistance) is discovered in an unadapted or ‘exotic’ source of germplasm, multiple backcrosses to elite lines are used to introgress the new trait into an adapted background but with little effect on increasing overall genetic diversity in the program. The backcrossing approach, while necessary to meet the short-term objectives of the breeding program, can lead to reduced genetic variability that can limit long-term selection gains (Simmonds and Smartt 1999). In contrast, development of genetically diverse cultivars can help provide production stability in case of unforeseen pest outbreaks that can devastate genetically similar crop varieties, such as occurred to the US maize crop in 1970 with an outbreak of Southern corn leaf blight, Helminthosporium maydis (Hooker 1973).

An understanding of the genetic variation within and among breeding programs involved in active sharing of germplasm will provide useful information on integrating new germplasm into the programs. Populations from crosses between genetically diverse parents are expected to have greater genetic variation than populations developed from less diverse parents. However, phenotypic assessment of underlying diversity of parental lines is problematic. Molecular markers such as AFLP are useful tools to assess underlying genetic variability not apparent from the phenotype (Vos et al. 1995; Zeid et al. 2003; Charcosset and Moreau 2004).

The extent of genetic variation at the DNA level among cowpea cultivars and breeding lines in US and West African programs is not known. Analysis of AFLP (Fatokun et al. 1997) and microsatellite marker (Li et al. 2001) polymorphisms from 90 IITA breeding lines, indicated relatively low genetic diversity, despite 18 of the 90 lines having been developed from crosses with wild cowpea accessions. However, 51% of the lines had one or more parents in common in these studies. Most other studies of molecular diversity in cowpea have focused on crop evolution (Panella and Gepts 1992; Vallincourt and Weeden 1992), cowpea taxonomy (Fatokun et al. 1993; Pasquet 1999), introgression of wild cowpea, or assessment of diversity in landrace populations (Nkongolo 2003).

The objective of our research was to (1) assess the genetic variation within and among several cowpea breeding programs in West Africa and the USA and (2) determine the need for introgression of new genetic diversity into each program to ensure long-term selection gains.

Materials and methods

Plant materials

Cultivars and breeding lines from IITA, four West African national cowpea breeding programs (ISRA-Senegal, INERA-Burkina Faso, SARI-Ghana, IRAD-Cameroon), and two USA cowpea breeding programs (the USDA Vegetable Laboratory, South Carolina, the University of California at Riverside), and a set of six cultivar group Sesquipedialis cultivars from Asia were used in this study. The cultivars and breeding lines were chosen to represent promising lines developed by each program. A diverse set of four landrace accessions from Botswana in Southern Africa, and landrace accessions or cultivars from Afghanistan (1), Brazil (2), Ghana (3), India (3), Iran (1), and Nigeria (3) were also included to represent a larger sampling of the cowpea gene pool (Table 1).

Fluorescent-AFLP analysis

Total DNA was extracted from young leaves of 87 samples using the DNeasy system (Qiagen, Valencia, CA, USA). After DNA quantification using a Hoefer DyNA Quant 200 (Pharmacia Biotech, Piscataway, NJ, USA), AFLP analysis was conducted using the GIBCO BRL AFLP System I (Life Technologies, Grand Island, NY, USA) and visualized with the LI-COR IR automated sequencer 4000-L (Li-COR Inc., Lincoln, NE, USA). Total DNA (150 ng) from all 87 samples was digested with 1 μl of mixture of EcoRI/MseI (1.25 units/μl) at 37 °C for 6 h, and ligated to EcoRI/MseI adapters with 1.5 μl (1 unit/μl) of T4 DNA ligase at 25 °C for at least 3 h. The digested-ligated DNA was diluted 1:10 with ddH2O. The adaptor-ligated DNA was amplified in a total volume of 42 μl containing 5.0 μl of DNA from the ligation reaction, 28.3 μl of Pre-amp mix I, 4.2 μl of 10  ×  PCR buffer, 4.2 μl of MgCl2 (2.5 mM) and 0.3 μl of Taq DNA polymerase (5 units/μl, Promega, WI, USA). The pre-amplification reactions were performed on a MJR Cycle LRTM (MJ Research, Inc., Watertown, MA, USA) using the following cycling parameters: 30 cycles at 94 °C for 30 s, 56 °C for 1 min, and 72 °C for 1 min, then 1 cycle at 72 °C for 10 min. The pre-amplified PCR product was diluted 1:35 with ddH2O for subsequent selective PCR. Selective amplification was performed using a reaction mix composed of 2 μl of DNA from pre-amp, 1.2 μl of MseI primer (8.3 μM), 0.6 μl of IRD700-labeled EcoRI primer (1 μM), 0.6 μl of IRD800-labeled EcoRI primer (1 μM), 1.2 μl of 10  ×  PCR buffer, 1.2 μl of MgCl2 (2.5 mM), 5.1 μl of H2O, and 0.1 μl of Taq DNA polymerase (5 units/μl, Promega, Madison, WI, USA). Selective amplification PCR was performed by another touchdown program as follows: 10 cycles at 94 °C for 30 s, 63 °C for 30 s−1 °C per cycle, and 72 °C for 1 min, then 28 cycles at 94 °C for 30 s, 56 °C for 30 s, 72 °C for 1 min, then 72 °C for 10 min. The products from the selective amplification were electrophoresed on 25 cm  ×  0.25 mm 6.4% denaturing polyacrylamide Long Ranger® Gel Solution (BMA, Rockland, ME, USA) in 0.8  ×  TBE buffer using a LI-COR automated sequencer 4000-L. The gel was pre-run for 30–50 min at 1,500 V, 40 mA, and 40 W until the gel temperature reached 50 °C. The samples were denatured at 95 °C for 3 min and immediately placed on ice. Electrophoresis was performed at 1,500 V, 50 °C for 3.5 h after 0.9–1.0 μl samples and 1 μl of a mixture of IRD700 and IRD800 size markers were loaded (LI-COR Inc., Lincoln, NE, USA). Six EcoRI + 3 bases/MseI + 3 bases (E + ___/M +___) primer sets (IRD700 E+ACG/M+CAC; IRD800 E+AAC/M+CAG; IRD700 E+ACG/M+CTA; IRD800 E+AGG/M+CAC; IRD700 E+ACT/M+CAG; and IRD800 E+AAC/M+CTA) that showed clearly scoreable and polymorphic bands were selected for fluorescent-AFLP reactions with all the cowpea genotypes.

Data analysis

For the genetic similarity analysis, AFLP bands were scored visually as present (1) or absent (0) to create a binary dataset. The data were entered into a binary data matrix as discrete variables. Jaccard’s coefficient of similarity (Sneath and Sokal 1973) was calculated for all pair-wise comparisons among the cowpea accessions as follows: Jaccard =  \(N_{\rm AB}/(N_{\rm AB}+N_{\rm A}+N_{\rm B})\), where N AB is the number of bands shared by two cultivars (A and B), N A represents amplified fragments in cultivar A and N B represents fragments in cultivar B. Principal coordinates analysis (PCOA) was also carried out to reveal multiple dimensional distributions of the accessions in a scatter-plot (NTSYS-pc, version 2.1) (Rohlf 2000).

Results and discussion

A total of 382 bands were scored among the accessions with 207 polymorphic bands (54.2%) (Table 2). Despite a diverse origin, the 87 cowpea accessions shared a minimum 86% genetic similarity. The primer sets produced from 41 to 81 fragments and the percentage of polymorphic fragments per primer set ranged from 47.8 to 70.7 (Table 2).

Table 2 AFLP primer combination, total number of fragments generated by each primer set, number of polymorphic fragments per primer set, and percentage of polymorphic fragments in the study of 87 cowpea accessions
Full size table

Examples of AFLP profiles of 45 cowpea accessions using primer set IRD800 E+AAC/M+CTA are shown in Fig. 1.

Fig. 1
figure 1

Fluorescent-AFLP profiles of 45 cowpea accessions using primer set IRD800 E+AAC/M+CTA. The samples are arranged from left to the right in the order of (M) size marker, (1) UCR 162, (2) UCR516, (3) UCR 730, (4) UCR 739, (5) UCR 779, (6) Montiero, (7) UCR 282, (8) KVx421-25, (9) KVx442-3, (10) KVx525, (11) IAR-7/180-4-5-1, (12) KVx544-6-151, (13) KVx61-1, (14) SuVita2, (15) 24-125B-3, (16) IRAD12-58, (17) IRAD.7-29, (18) Lori Niebe, (19) B22-Val., (20) Apagbaala, (21) Marfo-Tuya, (22) IT82E-18, (23) IT84S-2049, (24) IT85F-3139, (25) IT85F-867-5, (26) IT86F-2014-1, (27) IT89KD-288, (28) IT90K-284-2, (29) IT93K-2046, (30) IT93K-503-1, (31) IT93K-693-2, (32) IT93K-93-10, (33) IT95K-1090-2, (34) IT95K-1093-5, (35) IT95K-1479, (36) IT95K-1491, (37) IT95M-190, (38) IT95M-303, (39) IT95M-309-1, (40) IT95M-311, (41) IT97K-207-15, (42) IT97K-207-21, (43) IT97K-499-39, (44) IT97K-556-4, (45) IT97K-819-132, and (M) size marker

Full size image

Landrace accessions

Landrace accessions from different origins were scattered throughout the PCOA diagram, indicating that this selected group served as a good sample of diversity with which to compare the relative diversity of breeding lines within and among cowpea breeding programs (Fig. 2). For example, landrace accessions from Botswana (Nos. 2–5) were grouped in the upper left of the PCOA diagram, while the accessions from Iran (No. 53) and Afghanistan (No. 1) were located in the upper center and right, respectively, while accessions from India (Nos. 51 and 52), and Nigeria (Nos. 54 and 55) were dispersed (Fig. 2).

Fig. 2
figure 2

Distribution of 87 cowpea genotypes based on the plot of the first and second coordinates of variation, generated from Principal Coordinates Analysis of AFLP profiles of the cowpea genotypes. The shape codes of the data points represent genotype country of origin. The color codes represent regional groupings, i.e. genotypes denoted by red symbols are from Asia and the Americas, genotypes noted by green colored symbols are from Africa. Genotypes with known parentage from both Africa and the Americas are denoted in blue. African genotypes with probable Asian parentage are denoted in yellow with green outline

Full size image

Asian and American genotypes

Without exception, genotypes from Asia and the US were located on the upper half of the PCOA diagram (Fig. 2), while most West African breeding lines were restricted to the lower half. This clustering along the second coordinate would suggest that the germplasm base of Asian and US cultivars is relatively narrow, and that US and Asian germplasm is derived from African regions outside of West Africa. In addition, the data indicated that there has been virtually no introgression of West African cowpea germplasm into Asian and US germplasm. The only exception to this was for lines CC-110 and 99-8-321-2 (Nos. 66 and 67, respectively) that were recently developed at the University of California-Riverside with known West African parentage. CC-110 is a line developed for use as a forage or cover crop and is a cross between Botswana landrace UCR 779 (No. 5) and IITA breeding line IT89KD-288 (No. 26). Line 99-8-321-2 is a breeding line developed for resistance to lygus bug (Lygus hesperus) from a cross of IITA line IT93K-2046 (No. 28) and CB27 (No. 68). (Fig. 2). For both these lines, the PC values are midway between the parental line PC values, as would be expected for lines with such parentage. The position of the two accessions from Brazil (Montiero-No. 6 and UCR 282-No. 7) indicates that these two cultivars have different origins. The AFLP profile suggests that Montiero’s origins lie in West Africa. Morphological and phenological data are consistent with this hypothesis. Montiero is a photoperiod sensitive cultivar with large white seed with rough seedcoat similar to cultivars that are popular in West Africa, and it is an old landrace cultivar from Northeastern Brazil similar to other cultivars thought to have been introduced from West Africa in the 16th century (F. Friere-Filho, personal communication). UCR 282 is an improved breeding line from Brazil (F. Friere-Filho, personal communication), and has an AFLP profile similar to Asian/US germplasm. Consistent with other genotypes from Asia, its seed is small, with a smooth textured seedcoat, similar to many Asian/US cultivars.

The AFLP profiles show extensive overlap in distribution of the US and Asian germplasm. Small, smooth-seeded genotypes from Asia and even genotypes of the morphologically distinct Sesquipedialis cultivar group showed extensive overlap with US germplasm, especially with respect to US cultivar group Melanophthalmus genotypes (Fig. 2), that are characterized by large grains with rough-textured seedcoat. A possible explanation for this relationship is that cowpea was taken from Eastern or Southern Africa to Asia and/or Europe and then more recently to the Americas with European settlement. Germplasm with large sized grain and ‘rough’ seedcoat texture from West Africa have distinct AFLP profiles from cultivar group Melanophthalmus members that have similar large grain size and rough textured seedcoat. Therefore, our data do not support that US germplasm came directly from West African cowpea germplasm as some have postulated (Steele 1976; Ng 1995).

Introgression of Asian/US germplasm into breeding lines developed in West Africa by IITA, ISRA-Senegal, SARI-Ghana, and by the Agricultural Research Corporation-Sudan was evident in the AFLP profiles of several breeding lines from these programs. IITA lines IT82E-18, IT85F-867-5, IT90K-284-2, IT91K-93-10, IT97K-556-4, IT98K-1105-5 (Nos. 21, 24, 27, 31, 43, and 45, respectively) were clustered in the upper region of the PCOA diagram (Fig. 2) indicating probable introgression of Asian/US or southern African germplasm into the IITA breeding program. Bambey 21 (No. 58) from ISRA-Senegal, two cultivars from Ghana, ‘B22/Vallenga’ (No. 18) and ‘Marfo-Tuya’ (No. 20), and Ein-El-Ghazal (No. 63) from Sudan had ‘Asian’ or intermediate (‘hybrid’) AFLP profiles that were consistent with their pedigrees. ‘B22/Vallenga’ was developed from a cross of an early maturing Ghana landrace ‘Bun 22’ and the improved cultivar Vallenga. Vallenga was developed by IITA and released as breeding line IT82E-16. IT82E-16 is an early, erect genotype with small, smooth red seeds typical of many Asian cowpea cultivars bred for high grain yield. Ghana cultivar ‘Marfo-Tuya’ (Padi et al. 2004a, b) was developed from a cross of Ghana landrace ‘Sumbrisogla’ and University of California-Riverside blackeye breeding line ‘518-2’. Similarly, the cultivar ‘Ein-El-Ghazal’, recently released in Sudan (Elawad and Hall 2002) was developed from a cross between a University of California blackeye bean variety ‘California Blackeye No. 5 (No. 70) and Senegalese variety ‘Bambey 23’. As expected, based on this pedigree information, Ein-El-Ghazal clustered in the upper-right quadrant of the PCOA diagram (Fig. 2).

Long bean cultivars from Thailand and other parts of Asia (cultivar group Sesquipedialis) were clustered in the upper-right corner of the PCOA diagram (Fig. 2). This clustering indicated limited genetic diversity among accessions of this type, supporting the prevailing view that this form of cowpea arose in Asia through a genetic bottleneck at two separate points. The first bottleneck was the introduction of cowpea into Asia and the second was due to the fact that this form arose from a single or limited number of mutation/selection events.

‘Blackeye-type’ (cultivar group Melanophthalmus) cowpea cultivars and breeding lines from the University of California were clustered in the upper right quadrant and overlapped with Asian long bean cultivars (Fig. 2). Breeding lines and cultivars from the US Vegetable Laboratory (Nos. 76–79) and other genotypes chosen randomly from both the US and Asia, Iron-Clay (US, No. 75), UCR 27 (US, No. 65), Minnesota 13 (US, No. 80), UCR 11 (Iran, No. 53), UCR 193 (India, 50), PI205140 (India, No. 51), UCR 1340 (India, No. 52), UCR 162 (Afghanistan, No. 1), were grouped in the upper half of the PCOA diagram (Fig. 2). These results indicated that much of the germplasm adapted to the US may have come via Asia or Europe, rather than from Africa directly. This is possible because cultivated cowpea was first introduced to India during the Neolithic period and India appears to be a secondary center of genetic diversity (Steele 1976; Pant et al. 1982), and cowpea has been cultivated in southern Europe since at least the 8th Century BC (Tosti and Negri 2002) and a diverse collection of these resources has been made by the Istituto di Genetica Vegetale del CNR in Bari, Italy. Cowpea was introduced to the West Indies in the 16th Century by the Spanish and was taken to the United States about 1700 (Purseglove 1968), and it appears that the germplasm they brought with them was from Europe, rather than collected in West Africa with the slave trade. Perrino et al. (1993) evaluated 452 cowpea accessions from 19 countries for morphological and developmental traits and came to a similar conclusion. A more detailed study using representative landrace accessions from several parts of Africa, Asia, Europe and the Americas is being undertaken to develop firmer conclusions about the diffusion of cowpea out of Africa, and the origin and diversity of cowpea germplasm in different regions of the world.

Genetic variation within West African breeding programs

Definite clustering of breeding lines by program and accessions by country of origin were observed based on PCOA of AFLP profiles (Fig. 2). In contrast to the dispersed landrace accessions from Nigeria (Nos. 56 and 57), 13 of the 27 breeding lines from IITA were clustered in one of two groups in the lower center-right area of the PCOA diagram. In West Africa, cultivars with large white or brown grains having a rough textured seedcoat are preferred, whereas in India, small red and brown seeded cultivars with smooth seedcoats are common. Most lines falling outside the two main clusters of IITA lines had at least one or more characteristics typical of Asian cultivars, suggesting that these lines were derived from crosses where at least one parent originated from Asia. For example, genotypes from IITA with principle coordinates similar to Asiatic germplasm such as IT82E-18, IT85F-867-5, and IT91K-93-10, are early maturing breeding lines with erect growth habit, small, smooth red or brown colored seeds and thus phenotypically similar to many modern cultivars developed in India. Lacking pedigree information, we infer that these lines were derived from crosses of erect early-maturing (ca 60-day) cowpeas originally developed in India, then brought by IITA scientists to Tanzania as part of a USAID-sponsored cowpea improvement program in that country in the late 1970s (P.N. Patel, personal communication). Subsequently, these early maturing cowpea genotypes were used by breeders at IITA to develop erect, ’60-day’ cowpea cultivars with seed quality characteristics preferred in West Africa, such as large size and rough seedcoat that is brown or white in color (Singh and Ntare 1985). It is likely that several other erect early-maturing IITA breeding lines such IT98K-1105-5 (No. 45), IT97K-556-6 (No. 43), IT95K-1479 (No. 34), and IT95K-1491 (No. 35) also have Asian parentage because they are all located in the upper area of the PCOA diagram (Fig. 2). IT86F-2014-1 (No. 25) is a breeding line bred for green pod production. Its AFLP profile suggests that this line resulted from a cross between cultivar group Sesquipedialis and representatives of the main IITA group of genotypes because it is located approximately midway between the Sesquipedialis cluster and main IITA group of 13 lines on the PCOA 2 axis. Three other IITA lines (Nos. 35, 36, and 37) that fall outside the main clusters of IITA lines were developed from crosses between wild and cultivated cowpeas.

The relatedness of the AFLP profiles observed among the IITA lines indicated that a restricted set of parental lines were used in the crosses that produced the lines included within these IITA clusters. The use of few parental lines in breeding programs is often driven by the need to maintain useful combinations of genes that have been pyramided over time in a limited number of genetic backgrounds. The use of backcrossing to maintain desirable gene combinations also could have reduced genetic variability among the clustered lines.

Breeding lines and cultivars developed by ISRA in Senegal (Nos. 56–62) were clustered in the center-right region of the PCOA diagram (Fig. 2), with the exception of Bambey 21 (No. 58), that is located in the upper-right region, similar in location to several US and Asian genotypes. Bambey 21 is an unusual cultivar in Senegal in that it is an erect early-maturing cultivar unlike most sprawling landrace or semi-erect improved cultivars. Bambey 21 was developed in the early 1960s from a cross of Senegal landrace 58–57 and a US breeding line coded as 58–40 in the Senegalese National Cowpea Germplasm Collection (N. Cisse, personal communication), so its position is consistent with other lines having hybrid origins between African and US or Asian germplasm.

Breeding lines from the INERA (Burkina Faso) breeding program with ‘Burkina parentage’ (Nos. 8, 10, 12, 13, 14) are clustered in the lower central area of the PCOA diagram in close proximity to the IITA ‘clusters’ (Fig. 2). This clustering may indicate that a limited number of related parental lines were used in crosses to develop theses lines. The proximity of many of the INERA breeding lines to the IITA clusters reflects the fact that IITA lines were used as parents in the crosses used to develop these lines when an IITA cowpea breeder was based in Burkina Faso during the 1980s. KVx442-3 (No. 9) and IAR-7/180-4-5-1 (No. 11) are exceptions and well outside the main ‘INERA cluster’. Interestingly, IAR-7/180-4-5-1 was developed from a cross of a Nigerian cultivar IAR-7 (developed by the Institute of Agricultural Research in Nigeria) and a breeding line from Burkina Faso and this hybrid origin would explain its position outside the main cluster of lines from the INERA program.

Because only three cowpea cultivars or breeding lines were sampled from Ghana and Cameroon, only limited inferences can be made concerning the genetic variability within these breeding programs. Based on the AFLP profiles of cultivars from Ghana (Nos. 18–20), relatively wide diversity was observed, especially along the C2 axis of the PCOA (Fig. 2). This range of diversity is due in part to the introgression of Asian and US germplasm as described above. In contrast, genotypes from Cameroon were obviously clustered and located in the middle right area of the PCOA diagram.

Theoretical arguments suggest that populations derived from genetically diverse parents should have higher genetic variance than populations derived from more closely related parents. Therefore, the AFLP profiles generated in the current study can be used as one criterion to be considered in selection of parental lines for crossing to increase genetic variability in a breeding program. For example, the current study indicates that California blackeye genotypes CB27 (No. 68), 524B (No. 69), CB5 (no. 70), have a very limited number of molecular polymorphisms between them, likely indicating that they are highly related. Therefore, crosses between these lines would be expected to generate sets of lines with less genetic variability than crosses between either of these lines and CB46 (No. 71) or other blackeye type lines. Neutral genetic markers such as AFLP can provide an estimate of the genetic relatedness of genotypes (Paczos-Grzeda 2004), but experimental evidence is inconclusive as to whether the degree of marker polymorphism between two parents is correlated with genetic variance of derived populations from these parents (Charcosset and Moreau 2004).

Understanding the relatedness of varieties and improved breeding lines in US and West African cowpea breeding programs will assist breeders in assessing the need to undertake efforts to introgress exotic germplasm into these programs. There was a clear separation of genotypes from Asia and Africa in the PCOA. The results obtained by PCOA of the AFLP profiles matched our expectations of the level of polymorphism for those breeding lines whose parents were included in this study. Specific lines could be identified that differed greatly in AFLP profile from materials in a breeding program and these can be hybridized to introduce additional genetic variation into the program. Our results suggest that genetic variation within breeding programs is limited relative to that which exists in cultivated cowpea germplasm. In particular, genetic variation as determined by AFLP analysis among cultivar group Melanophthalmus at the University of California appears to be quite limited as had been hypothesized earlier (Ehlers and Foster 1993).

The AFLP data support historical records that indicate introgression of germplasm from India probably played a major role in increasing the genetic diversity of cowpea in the IITA breeding program. Varieties such as ‘Marfo-Tuya’ and ‘Ein-El-Ghazal’ that were derived from crosses between African and US parents are also unique, adapted sources of increased genetic variation for West African breeders to use in their breeding programs.

West Africa is considered to be the primary center of diversity of cultivated cowpea. As such, this region should contain most of the genetic variation present in the species. Based on our limited sample of breeding lines, it appears that cowpea genetic resources from this region have not contributed significantly to the genetic background and genetic improvement of this crop outside the West Africa region. A more comprehensive study of AFLP profiles of landrace accessions from around the world is needed to confirm this. Despite substantial phenotypic variability observed in US germplasm for many morphological and developmental traits, US cowpea breeding programs appear to lack genetic diversity. This generalization may also be true for Asian cowpea germplasm, especially in the case of cultivar group Sesquipedialis germplasm. Further studies of cowpea germplasm resources can help reveal the need for introgression of additional genetic variability into cowpea breeding programs.