Introduction
Common bean (Phaseolus vulgaris L.) was domesticated over a period of 7000 years and evolved from a wild vining plant (Broughton et al., Reference Broughton, Hernandez, Blair, Beebe, Gepts and Vanderleyden2003). This New World crop is grown in a rangeof environments and cropping systems (Gepts and Debouck, Reference Gepts, Debouck, Schoonhoven and Voysest1991; Singh et al., 1991c). Cultivated germplasm is well characterized with divisions of genotypes into two major gene pools: the Andean originating from the Andes mountains of South America and the Mesoamerican from Mexico and Central America (Blair et al., Reference Blair, Giraldo, Buendia, Tovar, Duque and Beebe2006).
Microsatellite markers have been used to determine the population structure of various cereal and legume crop species (e.g. Ferguson et al., Reference Ferguson, Bramel and Chandra2004; Garris et al., Reference Garris, Tai, Coburn, Kresovich and McCouch2005) with properties of detecting polymorphism in simple sequence repeat (SSR) loci that usually have many alleles (Powell et al., Reference Powell, Machray and Provan1996; Mitchell et al., Reference Mitchell, Kresovich, Jester, Hernandez and Szewc-McFadden1997). Microsatellite analysis in common bean genotypes has revealed wide variation and confirmed the existence of races within each gene pool (Blair et al., Reference Blair, Giraldo, Buendia, Tovar, Duque and Beebe2006, Reference Blair, Díaz, Hidalgo, Díaz and Duque2007, Reference Blair, Díaz, Buendia and Duque2009; Díaz and Blair, Reference Díaz and Blair2006; Kwak et al., Reference Kwak, Kami and Gepts2009; Becerra et al., Reference Becerra, Paredes, Rojo, Díaz and Blair2010). Microsatellites have also been successfully used to study beans grown in different parts of the world, including Africa (Díaz and Blair, Reference Díaz and Blair2006; Asfaw et al., Reference Asfaw, Blair and Almekinders2009; Blair et al., 2010), Brazil (Burle et al., Reference Burle, Fonseca, Kami and Gepts2010), China (Zhang et al., Reference Zhang, Blair and Wang2008), Europe (Métais et al., Reference Métais, Hamon, Jalouzot and Peltier2002; Masi et al., Reference Masi, Spagnoletti and Donini2003), Central America (Goméz et al., Reference Goméz, Blair, Frankow-Lindberg and Gullberg2004), Colombia (Díaz et al., Reference Díaz, Buendía, Duque and Blair2011) and Mexico (Payró de la Cruz et al., Reference Payró de la Cruz, Gepts, Garcia Marín and Villareal2005, Blair et al., Reference Blair, Díaz, Gil-Langarica, Mayek Perez and Acosta2011).
In India, common bean has been cultivated for around four centuries with an emphasis on cultivation in the north-western region. This part of the Indian subcontinent has accumulated and developed a large amount of diversity (Vavilov, Reference Vavilov1951; Singh et al., 1991a, b; Pathania et al., Reference Pathania, Sharma and Sharma2006). India holds the fourth position in ‘statistically accounted for’ common bean production with an area of about 11 M ha (Food and Agriculture Organization (FAO), 2010) out of a total of 30 M ha for all legumes. Neighbouring countries such as Myanmar and Nepal are also large producers with an area of over 1 M ha each. However, the area reported by FAO statistics may include other bean species such as mung bean (Vigna radiata), so the effective area under common bean in the region will be at a maximum around 1 M ha for winter or summer cultivation. Historical cultivation records are sparse, although diversity is found in the Himalayas (Tiwari et al., Reference Tiwari, Singh, Rathore and Kumar2005; Sharma et al., Reference Sharma, Rana, Sharma, Rathore and Sharma2006) as well as in other regions of India (Kumar et al., Reference Kumar, Sharma, Sharma, Kumar, Sharma, Malik, Singh, Sanger and Bhat2008).
In spite of the significant production of common bean in India, practically no information about the amount of genetic diversity within landraces exists. Diversity analysis would be a prerequisite for the improvement as well as conservation of germplasm from this region. The goals of the present study were to (1) determine the level of genetic diversity in Indian landraces, (2) elucidate their gene pool identity and (3) postulate the sources of germplasm found in India.
Materials and methods
Plant material and DNA extraction
A total of 153 common bean accessions were used in this study (Supplementary Table S1, available online only at http://journals.cambridge.org). These included 149 Indian landraces and four control genotypes for each gene pool, with two from the Mesoamerican gene pool and two from the Andean gene pool. The Indian landraces consisted of two sets; set I (the HPK collection) contained 67 landrace accessions and cultivars from Himachal Pradesh (HP), a north-western Himalayan state of India that is a major traditional common bean-producing area of the country.
Meanwhile, set II (the FAO collection) contained 82 accessions from the Genetic Resource Unit (GRU) of International Center for Tropical Agriculture that are part of the FAO in-trust collection of facilitated-access, food crop species (http://isa.ciat.cgiar.org/urg/main.do). This latter group of accessions represented several diverse common bean-growing regions of India, but also included a majority of the landraces from the north-western Himalayan mountain region. Passport information for the second set was obtained from the GRU.
After recording the seed characteristics of each landrace, three seeds of each accession were grown under screen-house conditions in plastic pots to record data on vegetative morphological traits and to provide young trifoliate leaf samples for DNA extraction that was according to a Cetyl trimethylammonium bromide (CTAB)-based mini-prep extraction protocol from Afanador et al. (Reference Afanador, Hadley and Kelly1993).
DNA quality was checked on 1.0% agarose gels using lambda DNA as the standard for quantification. Based on this evaluation, DNA concentrations were diluted to 5 ng/μl for PCR amplification.
Microsatellite analysis
Twenty-four genic and genomic common bean microsatellite (SSR) markers were selected for the evaluation of the genotypes from India based on their high polymorphism and stable amplification (Blair et al., Reference Blair, Giraldo, Buendia, Tovar, Duque and Beebe2006), and even distribution across the linkage groups of the common bean genome (Blair et al., Reference Blair, Pedraza, Buendia, Gaitán-Solís, Beebe, Gepts and Tohme2003). PCR amplification was carried out in 96-well PCR plates with a 15 μl reaction volume containing 15 ng of genomic DNA, 0.1 μM of forward and reverse primers, 10 mM of Tris–HCl (pH 7.2) PCR buffer, 50 mM KCl, 1.5–2.5 mM MgCl, depending upon the primers, 0.2 mM dNTPs and 1.0 unit of Taq polymerase enzyme (Promega, Inc., Madison, WI, USA). PCR amplifications were performed with an MJR-200 Peltier Thermal Cycler (MJ Research, Watertown, MA, USA) programmed with a hot start of 92°C for 3 min followed by 34 cycles of denaturation at 92°C for 30 s, annealing at 50, 60 or 65°C for 30 s, depending upon the melting temperature for the individual primer pair, and extension at 72°C for 45 s followed by the final extension at 72°C for 5 min.
After the PCRs, 5 μl of formamide, containing 0.4% bromophenol blue and 0.25% xylene cyanol FF, were added to stop the reaction mix. The mixture of the PCR product, dye and formamide was denatured for 5 min at 94° C and cooled in ice. Then, 3 μl of the denatured PCR product mixture were loaded onto a 4% denaturing polyacrylamide gel (29:1, acrylamide–bis-acrylamide) and run on Sequi-Gen GT electrophoresis units (Bio-Rad, Hercules, CA, USA) at 100 constant watts for about 1.5 h. During the sample loading, 2.0 μl of a 10 bp molecular weight ladder (Invitrogen, Carlsbad, CA, USA) were also loaded every 24th lane for fragment size determination.
After electrophoresis, the gels were silver-stained according to the manufacturer's instructions (Promega, Inc.) using a recirculating tank system described in Blair et al. (Reference Blair, Giraldo, Buendia, Tovar, Duque and Beebe2006).
Data analysis
Allele sizes for each of the 24 microsatellites were scored for all the test genotypes based on computer-scanned images using Quantity One software and by comparing the heaviest microsatellite band with the nearest 10 bp molecular weight ladders. The genetic similarity of the Indian landraces based on the proportion of shared alleles was estimated as recommended by Blair et al. (Reference Blair, Díaz, Buendia and Duque2009) using the software program Statistical Analysis Systems v. 9.1.3 (SAS Institute, 1996). The genetic similarity matrix was then used in the SAHN subprogram of NTSYS-pc software v. 2.10 (Rohlf, 2002), followed by sequential, agglomerative, hierarchical nested cluster analysis in the same program to construct dendrograms based on unweighted pair group method with arithmetic mean (UPGMA) algorithms. Popgene software v. 1.31 (Yeh et al., Reference Yeh, Boyle, Ye and Mao1997) was then employed to determine the genetic diversity and genetic relationships within and among the groups and subgroups in each identified gene pool. The parameters defined by this program were polymorphism percentage (P), observed heterozygosity (H o), expected heterozygosity (H e), coefficient of genetic differentiation (G ST), gene flow (N m), genetic distance (GD) and genetic identity (GI). The sizes, numbers and predominance of alleles across the genotypes and their polymorphism information contents (PIC) were determined with Power marker software (Liu and Muse, Reference Liu and Muse2005). Finally, the number of populations (K) and subgroups within each gene pool were estimated using the software program STRUCTURE (Pritchard et al., Reference Pritchard, Stephens and Donnelly2000) with the ideal number of populations estimated with the Evano et al. (Reference Evano, Regnaut and Goudet2005) test after a total of 15 independent runs each with 50,000 burn-ins and 100,000 repetitions performed for K= 1–10.
Results
Allele diversity in the two collections of the Indian landraces
Microsatellite analysis of the 153 common bean genotypes with the 24 genomic and genic microsatellites amplified a total of 107 alleles with an average of 4.5 alleles per marker (Supplementary Table S2, available online only at http://journals.cambridge.org). The number of alleles per microsatellite varied between 2 and 12, though the number was comparatively higher in genomic markers (average 5.1 alleles per marker across all genotypes) than in gene-based markers (average 2.6 alleles). This was true across both germplasm sets (data not shown). The overall average across all genotypes and all markers was 4.5 alleles per marker. Totals of 91 and 89 alleles were amplified in germplasm sets I and II, with averages of 3.8 and 3.7, respectively. A maximum of ten alleles were amplified by BM143 in germplasm set I and seven by BM139 and BM143 in germplasm set II.
The PIC value for all the microsatellites was 0.454 with a range of 0.067 (Pv-ag003) to 0.740 (BM143). Expected heterozygosity for the individual markers in the HPK germplasm set (I) ranged from 0.134 to 0.744 for genomic microsatellites and from 0.041 to 0.598 for genic microsatellites. Observed heterozygosity was very low for both types of markers and ranged between 0.000 and 0.085, with an average of 0.019. A similar pattern of expected and observed heterozygosity was observed in the FAO germplasm set (II); however, the average observed heterozygosity (0.012) was less than that in the HPK set (I). The allele size for a given SSR ranged between 84 bp in BM139 and 255 bp in BM160 for genomic markers, whereas it was 92 bp in BMd45 and 325 bp in BMd46 for gene-based markers. These values are useful for the design of fluorescent microsatellite panels for automated detection of diversity in further Indian germplasm.
Genetic relationships and gene pool identity of the Indian landraces
The genetic analysis of the two sets of the Indian common bean accessions based on SSR polymorphism categorized the landraces into the two common bean gene pools: namely the Andean and Mesoamerican groups as shown in the UPGMA dendrograms (Fig. 1(A) and (B)). The FAO collection was subdivided into three subgroups (M1, M2 and M3) in the Mesoamerican gene pool possessing 12, 35 and 4 genotypes, respectively, making a total of 51 individuals, while the HPK collection was subdivided into two subgroups for this gene pool with only 31 genotypes. The control genotypes DOR364 and G5773 were grouped in the M1 subgroup in both collections. The Andean gene pool group for the FAO germplasm set (II) contained only 35 accessions placed in two subgroups (A1 and A2), accommodating 26 and 9 landraces. The Andean gene pool group for the HPK collection contained three subgroups with a total of 40 genotypes with the majority in the first subgroup. Likewise, in the FAO collection, the majority of the accessions were part of the A1 subgroup along with the Andean control genotypes G19833 and G4494.
Fig. 1 (colour online) Dendrograms showing genetic similarity of the Indian common bean landraces based on SSR markers: (A) FAO collection and (B) HPK collection.
The data on morphological traits such as seed size agreed with the microsatellite-based gene pool distinction of the Indian landraces of common bean into the two gene pools, with large-seeded genotypes of the Andean gene pool and small-seeded genotypes of the Mesoamerican gene pool (Table 1). The genotypes in the Mesoamerican gene pool were principally comprised of cream- and red-coloured seeds with type II and IV growth habits. The genotypes in the M1 clade were grouped along with the Mesoamerican checks DOR364 and G5773, thus indicating the presence of race Mesoamerica. The landraces in the M2 subgroup shared some of the phenotypic traits characteristic of the Durango–Jalisco race complex.
Table 1 Seed colour, growth habit and seed size distribution in the Indian landraces of common bean among the Andean (A1, A2 and A3) and Mesoamerican (M1, M2 and M3) gene pools from the two collections evaluated
a Seed colour: primary and secondary colour designations as 1 = white, 2 = cream, 3 = yellow, 4 = brown, 5 = pink, 6 = red, 7 = purple, 8 = black and 9 = other.
b Growth habit: I = determinate bush, II = indeterminate bush, III = indeterminate prostrate and IV = indeterminate climbing beans. Growth habit data missing for M2 (8) and A1 (2).
c Seed size: S = small, M = medium, L = large and NA = no analysis.
The genetic diversity indices for the landraces that were placed in either gene pool are given in Table 2. The observed heterozygosity in both germplasm sets was low, especially for the FAO collection compared with the HPK collection. However, no defined pattern was observed in the values of this index among and between the gene pools. No observed heterozygosity was observed for the landraces placed in the M3 subgroup of the FAO germplasm set (II) and was also low in the M2 subgroup of the HPK set (I). The expected heterozygosity, number of alleles and PIC values were similar between the subgroups and between the germplasm collections and tended to be intermediate compared with the full collection of the 149 Indian landraces.
Table 2 Genetic diversity in the Indian common bean genotypes among the Andean (A1, A2 and A3) and Mesoamerican (M1, M2 and M3) gene pools from the FAO and HPK collections
N, genotype number; N a, observed number of alleles; H e, expected heterozygosity based on Nei's estimate; H o, observed heterozygosity, PIC, polymorphism information content.
The genetic diversity analysis indices for genetic differentiation (G ST), gene flow estimate (N m), Nei's genetic identity (GI) and genetic distance (GD) among and between the Andean and Mesoamerican gene pools, shown in Table 3, exhibited gene flow estimates within the gene pool groups: namely, 1.473 for M1–M2 and 2.023 for A2–A3 in the FAO collection versus 2.643 for M1–M2 and 1.716 for A1–A2 in the HPK collection of the landraces. Furthermore, gene flow was evident between the gene pools in both collections; however, the highest gene flow was always between the M1 and M2 genotypes showing the migration of almost three individuals per generation. The pattern of genetic differentiation indices was variable for both within- and between-gene pool comparisons in both germplasm collections. However, it was higher between the gene pools than between the subgroups and within the gene pools as was evident in the M1–M2 (1.096) and M2–A2 (1.046) comparisons. The groups classified in this study showed high genetic similarity for the M1–M2 and A1–A2 subgroups in both the FAO and HPK collections. However, the genetic identity values ranged between 0.334 and 0.781 in the FAO collection and between 0.511 and 0.670 in the HPK collection.
Table 3 Microsatellite genetic distance (GD), genetic differentiation (G ST), gene flow (N m) and genetic identity (GI) among the Andean (A1, A2, A3) and Mesoamerican (M1, M2, M3) gene pools of the Indian common bean genotypes from the FAO and HPK collections
a N m estimated from G ST= 0.25(1 − G ST)/G ST.
Principal coordinate analysis (PCoA) and population structure
PCoA of the two Indian common bean collections (Fig. 2, upper panel) separated the genotypes into the Andean and Mesoamerican gene pools. Each group was differentiated by the presence of the respective gene pool checks, i.e. G4494 and G19833 with large-seeded genotypes representing the Andean gene pool and DOR364 and ICA Pijao having small-seeded genotypes from the Mesoamerican gene pool. Very little or no introgression between the gene pools was observed in the dimension 1 versus dimension 2 comparisons in both sets of the genotypes representing various bean-growing states of India and the north-western region.
Fig. 2 (colour online) Principal coordinate analysis of the two sets of the common bean landraces of Indian origin (upper panels) and their corresponding grouping and gene pool introgression based on the population structure analysis (lower panels) at K= 3. The two groups of the genotypes are (A) the FAO collection and (B) the HPK collection.
The population structure analysis using STRUCTURE software (Fig. 2, lower panel) further confirmed the grouping of the test genotypes found in the principal component analysis and UPGMA clustering analyses. In this analysis, the two gene pools were separated initially at a value of K= 2 and then further subgroups were formed at a value of K= 3 to determine whether any intermediate groups were formed. Admixture of the genotypes was observed in the Mesoamerican and Andean gene pools, respectively, for the two sets of the genotypes. This indicated some level of subgrouping in each gene pool as was also observed in the UPGMA analysis. The subgrouping within each gene pool could reflect race differences, geographic structuring or adaptation in different seasons. However, the overlap of the UPGMA subgroups in the principal component analysis indicated that strong differentiation was not evident to support the biological meaning for the subgroups.
Geographic distribution of the two collections of common beans was predominantly from one large region of India, namely the Himalayan mountain range and adjacent plains of north-western India. A state-by-state distribution of the genotypes for the HPK set involved 64 genotypes from HP; 2 from Jammu and Kashmir (J&K) and 1 from Uttrakhand (UK), all north-western states. The FAP collection was from a wider range of states with 26 from HP, 28 from J&K, 14 from the UK, 5 from Maharashtra, 5 from Tamil Nadu and 1 each from Karnataka, Meghalaya, Orissa and Uttar Pradesh. The geographic positions of the collection sites are shown in Supplementary Fig. S1 (available online only at http://journals.cambridge.org).
Discussion
This study constitutes the first report on the gene pool structure of common bean landraces from India with microsatellite markers and among the first ten analyses of beans outside of the Americas. Previously, Southwest Europe (Rodiño et al., Reference Rodiño, Santalla, González, de Ron and Singh2006), Eastern Africa (Asfaw et al., Reference Asfaw, Blair and Almekinders2009), Central Africa (Blair et al., 2010) and China (Zhang et al., Reference Zhang, Blair and Wang2008) had been studied for common bean gene pool balance. The SSR markers used in this study revealed genetic diversity from two South Asian collections. Whereas here we analysed a global collection that was representative of the whole country of India, previously only Randomly-amplified polymorphic DNA (RAPD) markers have been used to characterize Indian germplasm from a limited number of states (Tiwari et al., Reference Tiwari, Singh, Rathore and Kumar2005; Sharma et al., Reference Sharma, Rana, Sharma, Rathore and Sharma2006; Kumar et al., Reference Kumar, Sharma, Sharma, Kumar, Sharma, Malik, Singh, Sanger and Bhat2008; Jose et al., Reference Jose, Mohammed, Thomas, Varghese, Selvaraj and Dorai2009).
As a novel outcome of this study, we detected the presence of both major gene pools of common bean in the domesticated landraces of India based on both microsatellite markers and some morphological traits. Similar dichotomy of germplasm representing both primary centres of diversity has been reported from various bean-growing regions of the world (Zhang et al., Reference Zhang, Blair and Wang2008; Asfaw et al., Reference Asfaw, Blair and Almekinders2009; Blair et al., 2010; Burle et al., Reference Burle, Fonseca, Kami and Gepts2010). The races that make Indian germplasm are still open to discussion as we found only partial evidence for races Mesoamerica, Durango–Jalisco, Nueva Granada and Peru. Gene pool identity was further substantiated by the comparison of marker profiles with seed traits, particularly seed size, an important criterion for gene pool identification. Genotypes grouped in the Andean gene pool possessed medium to large seeds, with a blend of various seed colours such as cream, brown, red and purple. Purple seeds were restricted only to the Andean gene pool. Meanwhile, the Mesoamerican group of Indian landraces in both collections also possessed variable coloured seeds, however, with the dominance of cream and red seeds.
The landraces of Indian origin in the FAO collection mainly belonged to the Mesoamerican gene pool (59.3%), and two subgroups of the Mesoamerican beans were the most diverse in the entire analysis; however, both groups did not present genetic differentiation from each other and, on the contrary, were very similar. The HPK germplasm collection was principally dominated by the Andean gene pool (56.3%). The main reason for the prevalence of the Andean accessions in north-western Himalayas could be due to consumer preference for the typical large red-seeded beans found in races Peru and Nueva Granada.
The population structure of common bean identified in this study is largely in agreement with earlier studies of global diversity sets for common bean (Blair et al., Reference Blair, Díaz, Buendia and Duque2009; Kwak et al., Reference Kwak, Kami and Gepts2009). The dendrograms (Fig. 1) and the PCoA (Fig. 2, upper panel) differentiated the Indian landraces into two clusters representing the Mesoamerican and Andean gene pools. The population structure analysis showed that gene pool introgression was present (Fig. 2, lower panel) as A1 and M1 subgroups of the HPK and FAO collections, but in neither case were the genotypes far away from their gene pool identity. The lack of higher introgression is different from other regions such as China (Zhang et al., Reference Zhang, Blair and Wang2008), Central Africa (Blair et al., 2010), Eastern Africa (Asfaw et al., Reference Asfaw, Blair and Almekinders2009), Southern Africa (Martin and Adams, Reference Martin and Adams1987) and Southwest Europe (Rodiño et al., Reference Rodiño, Santalla, González, de Ron and Singh2006). Introgression is also common in genotypes from some parts of the Americas where both gene pools overlap (Díaz and Blair, Reference Díaz and Blair2006; Blair et al., Reference Blair, Giraldo, Buendia, Tovar, Duque and Beebe2006, Reference Blair, Díaz, Hidalgo, Díaz and Duque2009; Kwak et al., Reference Kwak, Kami and Gepts2009).
The diversity in India may be explained by the circumstances of its agroecological adaptation. More specifically, common bean cultivation in India is spread over two regions of the Indian subcontinent as well as two cropping seasons, namely (1) the kharif (monsoon) season in the north-western Himalayan mountain region and (2) the Rabi (winter/spring) season in the plains of north-western and South India. A large diversity in plant types, grain colours and agroecological adaptation of landraces is especially evident in the north-western Himalayan mountain region. The Andean and climbing beans are mainly grown in this highland region while the bush Mesoamerican beans are mainly grown in the lowland regions along with some bush Andean genotypes.
Among the areas of production in India, the north-western Himalayan region concentrates the main sites of common bean cultivation with niche-based production areas for particular grain types. For example, large, red and white/cream mottled beans are predominant in almost the entire Kinnaur and Kullu regions of HP, while red-coloured beans are popular in the Chamba region of the same state. Small red-coloured beans are found in the Badharwah region of J&K states adjacent to HP. In this study, we found some distinction between all these commercial classes in molecular fingerprint.
One remaining question is: what route did common bean follow to India? Many crops that originated in Asia have been postulated to have arrived in India, the Middle East and Europe via the spice and silk routes probably during the 16th century. During this same period, crops from the New World could have entered India from the East through the same route. Alternatively, crops from the New World could have arrived with European colonization after being spread into Europe and Southern or Eastern Africa via the Cape of Good Hope. Common beans could have entered from that route but likewise seem to have a long history in China and the Far East (Zhang et al., Reference Zhang, Blair and Wang2008). Indeed, since European trade with India first started in the coastal tropics, the adaptation of beans to the northern mountain regions argues for an Asian rather than a European introduction via Africa. One New World crop that probably entered from Spanish colonization of the Far East could have been grain amaranth that has established itself in India, Pakistan and China even to a greater extent than in its Andean homeland. Alternatively, common bean could have accompanied the introduction of maize that was established early on in Europe and could have crossed over from Turkey into the present-day Iran, Afghanistan, Pakistan and India.
In summary, two non-mutually exclusive hypotheses exist for the introduction of beans to India: an eastward introduction or a westward introduction. The introduction of common bean to India might have been through both China and Europe (via Africa), as both regions have been discovered to have a high diversity of common beans with both gene pools present (Santalla et al., Reference Santalla, Rodiño and De Ron2002, Reference Santalla, Menéndez-Sevillano, Monteagudo and De Ron2003; Zhang et al., Reference Zhang, Blair and Wang2008; Blair et al., 2010). China, especially Tibet, could have been a source of Andean bean diversity (Zhang et al., Reference Zhang, Blair and Wang2008). Finally, in the dual source theory, there may have been some distinction between gene pools and growth types from China or Europe: the bush beans coming from Europe and adapting to lowland production in the irrigated, spring season, and the climbing beans coming from China for adaptation to the wetter summer season. Selection for new combinations of characteristics such as heat tolerance for the spring season and photoperiod insensitivity for the summer season is probable.
As a final conclusion, the diversity detected in the entire collection was moderate (H e= 0.525) and comparable with that in China (H e= 0.535) and the Iberian Peninsula (H e= 0.293–0.317) (Santalla et al., Reference Santalla, Rodiño and De Ron2002; Zhang et al., Reference Zhang, Blair and Wang2008). The discovery of large amounts of diversity in India is not surprising if we consider that this country is a large producer of dry beans and has multiple seasons and agroecological niches in which to grow the crop. Introgression between the gene pools seems to be lower than in these other regions, perhaps indicating selection for large-seededness in the Andean gene pool rather than intermediate seed size typical of inter-gene pool hybrids.
In practical terms, our study serves as a baseline for further analysis of Indian common beans and is the first report using molecular markers that showed the presence of both Andean and Mesoamerican gene pools in the Indian subcontinent. Further collection of germplasm in neighbouring countries such as Afghanistan, Bhutan, Nepal and Pakistan is warranted. These results should be useful for planning crosses with Indian germplasm and for plant breeding of the crop in general. Common bean has a large potential to meet the dietary needs of the Indian population and therefore is of interest in this region.
Acknowledgements
We are grateful to the Department of Biotechnology, Government of India for providing DBT Overseas Associateship to P.N. Sharma. We are also grateful to the Generation Challenge Program for establishing the SSR marker kit used in this study through a grant to M.W. Blair. The help of the GRU is gratefully acknowledged.
