Genetic diversity among bambara groundnut accessions

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Bambara groundnut accessions, native to African countries, are evaluated for their genetic diversity to identify important descriptors for conservation and utilization existing genetic resources. Accessions obtained from the International Institute of Tropical Agriculture germplasm collection were analyzed based on 28 quantitative, 12 qualitative traits and DArT molecular markers (diversity arrays technology). The seed pattern diversity for a total of one thousand, nine hundred and seventy-three accessions was also considered. The highest diversity indices and occurrence of descriptor states in the Nigerian/Cameroon collections, relative to other regions confirmed earlier reports suggesting Northern Nigeria and Northern as the crop's centre of origin. Some of the Cameroon/Nigeria accessions also had open growth habit and thicker shells characteristic of the wild type, indicating the cultivated type in this region is closer to the wild form. Population structure analysis based on the genetic polymorphism by the DArT markers and phenetic tree based the descriptors, reveal similar trends pointing to Nigerian/Cameroon as centre of origin. Varying degrees of sensitivity to photoperiod was established in this germplasm, providing opportunities for improvement of crop seed yield, since photoperiodic regulation of its phenology must be well understood to capture the optimal daylenght in relation to the sowing season, flowering and fruit set. Better understanding of the organization of the genetic diversity from this study will ultimately promote their direct utility in breeding for improved lines.



Bambara groundnut [Vigna subterranea (L.) Verdc., syn. Voandzeia subterranea (L.) Thouars], a pulse with subterranean fruit-set, is cultivated by smallholders over much of semi-arid Africa (Linnemann and Azam-Ali 1993) and found intercropped with cereals, root and tuber crops. Although average yields (650 to 850 kgha-1) are less than other legumes (Stanton et al. 1966), under prevailing less favorable conditions, such as little rainfall and infertile soil, it yields better (National Research Council 1979). In much of Africa, it is the third most important legume after groundnut and cowpea (Sellschop 1962). As a preferred food crop of many local people, it is a good supplement to a cereal diet and nutritionally superior to other legumes (Azam-Ali et al. 2001).

Investigators interested in the origin of bambara (family: Leguminosae, subfamily: Papilionoideae) all agreed that the crop originated from the African continent. Suggestion by Jacques Felix (1946; 1950) pointed to Sudan as the centre of origin based diverse seed pattern diversity there. Rassel (1960) supported this hypothesis by reference to old Arab documents from the voyage of Ibn Batouta to Sudan dated 1380 (Mac Guckin De Slane 1843). De Candolle (1886) and Stuhlmann (1909) also stated that Schweinfurth had found the crop growing wild on the banks of Nile between Chartoum and Gondokoro in Sudan. However, a wild stage and centre of origin could not be confirmed in Sudan (Harms 1912).

Dalziel (1937) found bambara groundnut in its genuine wild state in the North Yola province of Nigeria and reported another finding by Ledermann near Garoua in northern Cameroon. Subsequently, Hepper (1963) undertook an expedition to West Africa and suggested the origin between Yola and Garuoa, as previously supposed. He proposed a new nomenclature and treated the forms as varieties (V. subterranean var. subterranea and V.subterrranea var. spontanea). Analyses of the seed pattern diversity within the large collection at IITA by Begemann (1988) showed that samples collected less than 200km from the putative centre of origin, between Yola and Garoua, consistently showed greater seed-pattern diversity. Characterization studies by Goli et al. (1997) revealed the frequency of the seed pattern states for 1384 accessions from the IITA (International Institute of Tropical Agriculture, Ibadan) collection but study on the geographical distribution was not considered.

To breed qualities of resistance, adaptability, high yield and better nutritive value into crops, sources of genetic variation are requisite. Old landraces are being displaced frequently by improved varieties leading to a genetic simplification in traditional cropping systems (Anchirinah et al. 2001). Together with promotion of modern varieties, factors such as a shift to monoculture, human settlements occupying the habitats of crop relatives, land clearing and the extinction of tribal cultures account for the genetic erosion. Consequently, the conservation of genetic diversity is crucial to breeding programs.

The study aims to promote conservation of the genetic resources and ultimately their direct utility to breeding for improved lines through a better understanding of the organization of the genetic diversity and variability. Based on molecular marker analysis and phenotypic descriptors, Nigeria and Cameroon was validated as the centre of origin as well as showing the highest phenotypic diversity.

Material and methods

Plant material

One thousand, one hundred and seventy-three accessions of bambara groundnut were obtained from the genebank collection at IITA. All 1173 accessions were sown on the 15th of April 2002 in Ibadan and scored for the number of days from sowing to the first flowering (DFF) and 50% flowering (D50). From this trial, 539 accessions were scored for DFF, while only 461 of these accessions were scored for D50. For comparison, data for the same set of accessions was obtained from a trial at Ikenne, planted on the 3rd of December 1986. A subset comprising 124 accessions were selected for further study based on their source of origin, latitude and longitude, growth habit and day to maturity so as to access their diversity, while a subset of the 124 accessions comprising 40 accessions were genotyped using a Diversity Arrays Technology (DArT). The seed pattern diversity was evaluated on all the accessions (1973) existing in the germplasm collection (Table 1).

Experimental site and conditions

The experimental site at IITA,Ibadan, is located within the transition zone between the humid and sub-humid tropical climates bordering to its south, and a dry savannah zone bordering to its north. It lies within latitude 70 30′N and longitude 30 54′E on an elevation of about 160m. The bimodal rainfall distribution peaks in June and September and separated by a period of lower precipitation in August, while December through to February constitutes the major dry season (Moormann et al. 1974). The average annual rainfall exceeds 1250mm and the yearly mean temperature is 26.60C (IITA 1984), with a minimum temperature range between 210C and 230C, while the maximum is between 280C and 340C. Radiation is approximately 5285 MJ/m2/year, while mean relative humidity is in the range of 64-83%. The soils at IITA are classified as ferric luvisols according to the FAO classification (Moormann et al. 1974). The longest daylength at the latitude 7°30'N is estimated at 12 hours 30minutes and the shortest daylength at 11hours 30minutes. The daylength is longest on the 21st or 22nd day of June and the shortest daylength on the 21st or 22nd day of December.

At sowing in April and at the initial stage of growing of the young plants, the field in Ibadan was irrigated with overhead sprinklers while the crop sown in December 1986 in Ikenne was entirely dependent on overhead sprinkler irrigation till the crops were harvested. Based on extrapolation from the findings of Doku and Karikari (1971a), the peak of the flowering of the crop planted in April coincides with the longest daylength in June while the peak of flowering of the crops planted in December should be at the end of January when the daylength was about the shortest during the year.

Phenotypic Data

The methods of scoring for forty of the traits were based on descriptors for bambara groundnut published by IPGRI et al. (2000) (a revision of the original IBPGR, IITA and GTZ publication, IBPGR et al. 1987), while 7 additional descriptors (flower colour, banner/flag petal length, wing length, gap between the banner and wing tips, ratio of banner to wing lengths, peduncle length, and pedicel length) were included. A total of twenty-eight quantitative and nineteen qualitative traits were used in this study.

DNA extraction and DArt Marker analysis

< DNA extraction, DArT analysis, DArT markers and scoring >

Statistical analysis

The program STRUCTURE version 2.3.3 (Pritchard et al. 2000) was used to test the hypotheses of K = 1 to K = 6 i.e. 1 to 6 populations using an admixture model, a burn-in phase of 1 x 104 and a sampling phase of 5 x 104 replicates. The optimal division of the population in to subpopulations was determined when the probability of K was very small for K less than the appropriate value (effectively zero) and then more-or-less plateaus for larger K (Pritchard et al. 2009). The structure analysis was performed at K = 4, a burn-in phase of 1 x 105 and a sampling phase of 5 x 105 replicates.

The means comparison among the 124 selected accessions was performed for quantitative traits, while frequencies of qualitative descriptor states were specified for each region. The Shannon-Weaver's diversity Index (, where pi is the proportion of accessions for a particular descriptor state) (Shannon and Weaver 1963) was calculated to compare diversity among regions. The diversity index takes into account the proportion of each state, as well as the abundance of accessions. The quantitative traits were correlated against each other, while the principal component analysis was performed to provide information on the relative importance of descriptors for germplasm characterization. The data was analysed using the SAS Version 8.0 (2000). The Hierarchical clustering of 40 accessions was based on the UPGMA (Unweighted pair-grouped method with arithmetic averaging) method using the NTSYS-pc software package version 2.02j.

Between the daylength durations from the two seasons, differences (DIFF) computed from the days to first flowering (DFF) was used to separate the accessions into classes. T-test procedure was used to compare the mean of the differences in the paired observations from the two trials (equal sample sizes and had a normal distribution). The group comparison t-statistic was computed based on the assumption that the variances are unequal (Cochran and Cox 1950; Satterthwaite 1946). The confidence limits were also computed according to Cochran and Cox (1950) approximation; an equal tailed and uniformly most powerful unbiased confidence intervals. The confidence limit for the standard deviation established how much departure away from the mean could be permitted within experimental error beyond which indicated a response to change in daylength duration. An analysis of variance was performed on values (DIFF) generated from the pairs of DFF and D50 between the two trials and between the two flowering parameters. The data sets were analyzed with SAS Version 8.0 (2000).


Genetic diversity, phenetics and population Structure

Based on a subset of 40 accessions and 544 DArT markers, 4 subpopulations were delineated with each of the subpopulations containing accessions from West Africa (Fig. 1). However, the East/South African accessions were restricted to subpopulation III, clustering with accessions from Cameroon. This indicates common alleles between a subset of the Cameroon and the East/South African accessions and that the East/South African accessions were probably derived from around this region in Cameroon. Only one accession from Madagascar was a result of hybridization between the subpopulation III and I (only accessions from Cameroon), with about equal contribution from both subpopulations.

Even though the phenetic tree does not necessarily reflect phylogeny or evolutionary relationship, the trend observed in the population structure analysis above was also mirrored in the phenetic tree (Fig. 2), which is based on the overall similarity between accessions using the phenotypic descriptors. Based on the phenetic tree, all the East/South African accessions clustered together with accessions from Nigeria and Cameroon.

Quantitative traits of bambara groundnut accessions

Comparatively, the Cameroon/Nigeria and West African regions (Republic of Benin, Burkina Faso, Cote d'Ivoire, Ghana, Gambia and Mali) had higher diversity indices for the quantitative traits (Table 2), with the average of the diversity indices in Cameroon/Nigerian region exceeding that of any other regions. The East African (Kenya and Tanzania) accessions were the least diverse i.e. low diversity indices for most of the descriptors and lowest average of the diversity indices.

Flowering in Cameroon/Nigeria, West and East African accessions were earlier than North/Central (Sudan and Central Republic of Africa) and South African (Madagascar, Zambia and Zimbabwe) accessions during the mid-April planting. Flowers of North/Central and South African accessions were smaller. South and East African accessions comprise of bigger plants reflected in larger plant spread, plant height, terminal leaflet length and petiole length, while Cameroon/Nigeria, North/Central and West Africa accessions had smaller plants. The East and South African accessions were characterized by low 100-seed weight, pod length and pod width. Mean yield values of seeds for Cameroon/Nigerian, West and North/Central African accessions at 457.22 kgha-1, 333.52 kgha-1 and 245.27 kgha-1 exceeded that of South and East African accessions at 75.31 kgha-1 and 27.86 kgha-1 respectively (Supplementary Table 1).

Three components generated from the principal component analysis explained 39.29% of the total variation contributed by all quantitative traits (Table 3). The banner length, gap between banner and wing tips, ratio of banner and wing lengths, peduncle length, petiole length, plant diameter, plant height, 100-seed weight, yield, seeds per pod, days to first flowering and day to maturity all had a high loading in the first component which accounted for 18.06% of the total variation. The second component accounts10.75% of the total variation, while the third component accounted for 10.48% of the total variation. The traits with consistently high loadings on more than one of the principal component will be very appropriate for the characterization bambara groundnut germplasm since they are better at delineating accessions.

Qualitative traits of bambara groundnut accessions

For the qualitative traits, the greatest diversity was also found in the Cameroon/Nigerian, West African and North/Central African regions (Table 4). Again diversity in the East Africa region is shown to be least with the lowest diversity index. On the overall, the Cameroon/Nigeria, West and South Africa regions were observed to have more occurrences of the descriptor states than the North/Central and East Africa regions.

Seed pattern diversity in bambara groundnut accessions

Collections from the individual countries have the round and oval seed shape except for South Africa and most countries in the East Africa (Burundi, Ethiopia and Kenya). Most West African countries (excluding Republic of Benin and Togo), South African countries (Madagascar, Swaziland, Zambia and Zimbabwe) and Central Republic of Africa had high proportions of both state. West African accessions had more variation in their seed testa pattern (especially Burkina Faso, Cameroon, Ghana, Nigeria and Togo) except in Cote d'Ivoire where the frequency was low (Supplementary Table 2). The seed testa pattern variations in Central Republic of Africa and some South African countries were also high. The highest variations of descriptor states (Fig. 3 and 4) were observed in Nigeria collections followed by Ghana, Cameroon, Burkina Faso, Central republic of Africa and Zimbabwe. Collections from the Cameroon/Nigeria showed the greatest diversity index in all seed characters except for the ground color of the testa (Supplementary Table 3).

Flowering response to variation in photoperiod

Comparison of data from the two trials using the t-test procedure indicated that data from the two seasons were significantly different (p ≤ 0.0001). The lower and upper confidence limits for DFF are 2.24 and 2.44 respectively, while those of the D50 are 4.00 and 4.35 respectively (Supplementary Tables 4). Hence, any DIFF value (i.e. difference between DFF or D50 values between the two seasons) outside these confidence limits were attributed to the response due to the variations in daylength rather than experimental error (Supplementary Tables 5 and 6).

After scoring 539 accessions for DFF, 127 accessions were sensitive to change in daylength duration, while 412 accessions were insensitive. Of the 461 accessions scored with D50, 134 accessions were sensitive, while 327 accessions were insensitive. Considering the 461 accessions common to both DFF and D50 scores, 337 accessions had a similar response to variation in daylength, while 124 accessions were inconsistent in their response (i.e. 42 sensitive accessions scored for DFF turned out to be insensitive when scored for D50, while 83 sensitive accessions scored for D50 turned out to be insensitive when scored for DFF) and were often observed to have a low sensitivity (low DIFF) to change in daylength.

None of the daylength durations recorded prevented flowering (qualitative long-day or short-day response), rather, a quantitative long day and short day was observed. Delayed response to flowering due to the short daylength was observed in as much as 100 accessions (quantitative long day); however, they flowered early under the long daylength duration. Quantitative short daylength accessions were found to outnumber the quantitative long day accessions (Supplementary Tables 7 and 8). Varying degrees of sensitivity to daylength durations were established among the quantitative long day and short day accessions, while accessions with a greater degree of sensitivity to daylength durations were observed to have been collected farther away from the equator.

The source of variation was observed among the country of origin and between the DFF and D50 for quantitative long day and short day accessions. For the quantitative short day accessions, no significant difference existed between the DFF and D50 except for the DIFF values of 4, 7 and 9 days. Only the 3 days difference was significantly different for the quantitative long day accessions. Among the countries of origin, the 12 and 5 days differences were significantly different for the quantitative short and long day accessions respectively.


We report on the validation of the centre of origin for bambara groundnut using molecular markers. This is particularly interesting, considering that bambara groundnut's generic name was changed from Voandzeia into Vigna, the same genus as Cowpea (Verdcourt 1980), since no morphological character of importance separated Voandzeia from Vigna. Although controversial, it is suggested that cowpea, cultigroup Unguiculata, had been introduced from West Africa to East Africa and Asia (Ng and Marechal 1985). Similarly, Hepper suggested that bambara groundnut was introduced to East Africa from West Africa (Hepper 1963).

Based on Vavilov's theory (Vavilov 1926), the centres of origin are considered the area of greatest diversity. Diversity study based on the phenotypic traits considered support the hypothesis of Hepper (1963), which suggested the area between Yola (Northern Nigerian) and Garuoa (Cameroon) as centre of origin for bambara groundnut. The distribution of accessions from the population structure analysis, phenetic tree and the diversity indices of the quantitative and qualitative traits reveal the greatest diversity in the Cameroon/Nigeria collection. For diversity in seed patterns of 1973 accessions, the Cameroon/Nigeria region was again obviously the most diverse.

Doku and Karikari (1971b) studied the evolution of bambara groundnut as a species and concluded that the cultivated ones originated from the wild form (Vigna subterranean var. spontanea) and evolved through series of gradual changes i.e. switching from open to bunch growth habit, from outbreeding to inbreeding, and a reduction in shell thickness. From the data gathered from this study, some Cameroon/Nigerian accessions are characteristic of the wild type since it's the only region that consist of thicker shells and open growth habit accessions. The centre of origin in the region has been verified by Dalziel's (1937) collections that contain many wild relatives, hence, the source of dominant genes.

With bambara groundnut's low average yields, ranging from 650 to 850 kgha-1 (Stanton et al., 1966), TVSu 23, 25, 32, 86, 231, 360, and 405 showed potentials for high yields (i.e. 1362.33 kgha-1, 1019.48 kgha-1, 972.16 kgha-1, 1025.74 kgha-1, 1189.50 kgha-1, 2473.38 kgha-1, 1983.10 kgha-1 respectively). Their seed sizes were relatively large with bunchy or semi bunchy growth habits, while susceptibility to Cercospora was relatively low. Information to enhance and reduce cost in germplasm characterization was provided by the relative importance of traits that have consistent high loadings in the principal components (Days to first flowering, 50% flowering and maturity; banner, wing and peduncle lengths; ratio of banner to wing lengths; plant diameter, plant height and yield).

We also report on the effect of photoperiod on growth and development of bambara groundnut considerable the relationship to effects on seed yield as shown in photoperiod sensitive field crops such as Soybean (Kantolic and Slafer 2007). Apparently, photo-regulation of development provides the plant with flexible mechanism to adapt to circumstances that create seasonal fluctuations in the length of the growing period. Though preliminary observations about the photoperiod insensitivity of bambara groundnut accessions have been reported, other studies show a delay in flowering and a strong effect on fruit set (Linnemann 1991; 1993). The onset of flowering, process of flowering, onset of podding and progress in pod growth have been well studied and shown to be impeded, with a stronger effect on podding than on flowering (Nishitani 1988; Linnemann 1993).

Our preliminary observations on the insensitivity of most bambara groundnut accessions to daylength in this study agree with findings by Linnemann (1991) and Nishitani (1988). In contrast to previous studies, response to daylength variation from the larger sample size in this study implies that insensitivity is a dominant trait within the population, with less than 30% of the accessions been quantitatively sensitive to long and short daylengths. This is similar in wheat genotypes, where insensitivity is dominant, while various degrees of sensitivity to short daylength were observed in some Chinese wheat genotypes (McIntosh et al. 1995). Apart from the quantitative short day behaviours found and supported by earlier results, a different photoperiodic behaviour of quantitative long day was also revealed (i.e. delayed response to flowering due to short daylength). Likewise, wide ranges of variations have also been observed in the flowering response of rice to photoperiods (Vergara and Chang 1985). The long day and short daylength accessions that are more sensitive to change in daylength are probably adapted to the short and long daylength durations that occur farther away from the equator. The genetic and environmental control of flowering time is a key determinant of the adaptation of crops to the environment. The ability to predict the time of flowering and harvest for optimal yield is crucial to matching crop germplasm to the resources available for crop growth within a given cropping system and to avoiding stressful environments. Since most of bambara groundnut accessions are daylength insensitive, most of the accessions can be adapted to different ecotypes with variations in daylength at anytime of the year.