Progress in common bean breeding requires the exploitation of genetic variation among market classes, races and gene pools. The present study was conducted to determine the amount of genetic variation and the degree of relatedness among 192 selected common bean advanced cultivars using 58 simple-sequence-repeat markers (SSR) evenly distributed along the 11 linkage groups of the Phaseolus reference map. All the lines belonged to commercial seed type classes that are widely grown in the USA and include both dry bean and snap beans for the fresh and processing markets. Through population structure, principal components analyses, cluster analysis, and discriminant analysis of principal components (DAPC), Andean and Mesoamerican genotypes as well as most American commercial type classes could be distinguished. The genetic relationship among the commercial cultivars revealed by the SSR markers was generally in agreement with known pedigree data. The Mesoamerican cultivars were separated into three major groups - black, small white, and navy accessions clustered together in a distinct group, while great northern and pinto clustered in another group, showing mixed origin. The Andean cultivars were distributed in two different groups. The kidney market classes formed a single group, while the green bean accessions were distributed between the Andean and Mesoamerican groups, showing inter-gene pool genetic admixture. For a subset of 24 SSR markers, we compared and contrasted the genetic diversity of the commercial cultivars with those of wild and domesticated landrace accessions of common bean. An overall reduction in genetic diversity was observed in both gene pools, Andean and Mesoamerican, from wild to landraces to advanced cultivars. The limited diversity in the commercial cultivars suggests that an important goal of bean breeding programs should be to broaden the cultivated gene pool, particularly the genetic diversity of specific commercial classes, using the genetic variability present in common bean landraces.

Plant materials

A total of 192 common bean advanced cultivars representing the most popular commercial industrial seed types classes in the USA and released in the last ca. 50 years were included in this study. These genotypes were obtained from 20 American public and private breeding programs and have all been extensively used by breeders and geneticists in the USA and around the world. More specifically, the sample included 21 black, 29 navy and two small-white cultivars with small seeds (<25 g 100 seed weight^-1), belonging to race Mesoamerica; 61 pinto and 27 great northern cultivars with medium -seeds (25-40 g 100 seed weight^-1), belonging to the race Durango; 18 light red kidney, 11 dark red kidney, 3 white kidney, and 20 green beans with large-seeds (>40 g 100 seed weight^-1), belonging to the Andean race Nueva Granada.

As standard genotypes, three common bean accessions were also included in the analysis: Midas, a typical Andean green bean type; BAT93, a typical Mesoamerican breeding line with multiple disease resistance [(Veranic 2 x Tlalnepantla 64) x (Jamapa x Tara)], which is also the source of the Mesoamerican genome reference sequence [2]; and Jalo EEP558, a typical Andean cultivar, which was released by a breeding program in Brazil (EPAMIG, Patos de Minas, Minas Gerais). BAT93 and Jalo EEP558 were also used as parents to obtain a recombinant inbred population that was used to develop a consensus molecular linkage map of common bean [26], in which over 170 SSR markers have been mapped [27]. A complete list of the lines, including information on pedigree, market class, and/or origin can be found in Table S1.

Seeds were grown in a greenhouse at the University of California in Davis. After 15 days, the primary leaves were harvested and stored overnight at – 80 °C.

Genomic DNA extraction, PCR and SSR genotyping

The frozen leaf tissue samples were lyophilized for around 48 hours using VirTis Sentry 2.0. and ground to a fine powder. Genomic DNA was then extracted from the young leaf tissue using the Qiagen DNeasy Plant Kit (Qiagen, Valencia, CA), and following the protocols provided by the manufacturer. DNA was quantified with a DyNA Quant 200 fluorometer (Hoefer Pharmacia Biotech, San Francisco, CA) and diluted to a concentration of approximately 10 ng µL^–1 for polymerase chain reaction (PCR) amplification. The amount of genetic diversity in the 192 samples was then assessed with molecular marker analysis. A set of 58 SSR markers were selected based on their wide distribution over the Phaseolus genome and their high polymorphism information content (PIC) values [27–29]. More information about the SSR loci used, including the primer pair sequences, the repeat motif, and the chromosomal locations, can be found in Table S2.

The SSRs analysis was conducted using an economical fluorescent tagging method described by Schuelke [30] in which an M13 reverse sequence tail (TGTAAAACGACGGCCAGTATGC) was added to the 5’ end of each forward SSR primers. The fluorescent dyes, 6-FAM, PET, and VIC, were attached to the 5’ end of the complementary (TGTAAAACGACGGCCAGT) M-13 universal primer sequence. For amplification, PCR reaction consisted of about 30 ng of genomic DNA, 200 µM dNTP (New England Biolabs), 0.04 µM forward primer with M-13 universal sequence tail, 0.16 µM reverse primer, 0.16 µM M-13 labeled fluorescent dye (Sigma Life Science), one unit of standard ThermoPol (Taq) reaction buffer with 2 mM MgSO4, and one unit of Taq polymerase (New England Biolabs). The PCR program consisted of 5 minutes at 94 °C, 30 cycles of 30 seconds at 94 °C, 45 seconds at 56 °C, and 45 seconds at 72 °C followed by 8 cycles of 30 seconds at 94 °C, 45 seconds at 53 °C, and 45 seconds at 72 °C with a 10 minutes final extension period at 72 °C. After dilution to a standard concentration, the amplified DNA samples were separated and sized in multiplex fashion depending on their expected size variation and were analyzed on Applied Biosystems 3730 DNA automatic analyzers.

Genotypes of markers were determined using the GeneMarker program (version1.95; SoftGenetics). When markers produced more than one peak, the peaks with clearly separated size ranges were scored independently as a different locus. This was noticed for three of the SSRs (PVag004, BM188, and BMd1). For each SSR locus and for the whole set of accessions, the total number of alleles observed were recorded and reported in a data set. Missing data were recorded when there was no detectable peak in the target size region for the marker. The data set was converted into Powermarker [31] and STRUCTURE [32] input format using the program Convert [33].