The common bean (Phaseolus vulgaris L.) is a widely cultivated legume that serves as a significant source of protein and other essential nutrients for human consumption. These include dietary fiber, complex carbohydrates, minerals, vitamins and phenolic compounds (Bitocchi et al., 2017a; Ganesan and Xu, 2017; Carbas et al., 2020). It is a self-pollinating annual plant with a relatively small diploid genome of 587 Mbp, cultivated primarily for its mature seeds and also for its immature (green) pods, which are utilized as a vegetable (Schmutz et al., 2014; Ntatsi et al., 2018). Following the process of domestication, which led to a reduction in genetic diversity and gene expression, the global dissemination of the common bean in the 16^th century gave rise to further changes associated with adaptations to different environmental conditions and cultivation practices (Bitocchi et al., 2017a; Cortinovis et al., 2020; Nadeem et al., 2021). The common bean is currently cultivated in many regions of the world, including Europe, where it plays an important role in countries such as Spain, Italy and France (CBI, Ministry of Foreign Affairs, 2022). There is considerable variation in the preferences for certain bean varieties between countries and even within regions. The continuous process of breeding and developing new varieties is driven by the necessity to enhance plant productivity, quality and resistance to pests and environmental stressors. Moreover, the newly developed varieties must meet the needs of producers and consumers (Assefa et al., 2019). Each new variety must be distinct, uniform and stable (DUS), maintaining the genetic and morphological characteristics that are unique to it. Moreover, the characteristics of the new variety, including yield, resistance, and other quality traits, must be superior to those of existing varieties (Council of the European Union, 1994, 2002; UPOV, 2002; Yang et al., 2021). One of the most significant characteristics of dry common beans for consumers is the color of the seed coat, which can vary considerably. The seed coat can be uniform in color, covering the entire seed (primary seed coat color) or it may have a secondary seed coat color, also known as a pattern, which overlays a light tan background and produces stripes or mottling patterns. Furthermore, seeds may exhibit partial coloration, wherein discrete white and colored zones manifest in various shapes across the seed coat (Bassett, 2007; Lioi and Piergiovanni, 2013; García-Fernández et al., 2021).

The observed seed coat color in common bean is the result of the accumulation of specific compounds in seed coats, namely flavonols (usually colorless to light yellow pigments), anthocyanins (variation of red, purple and black pigments) and proanthocyanidins (colorless pigments that turn brown after post-harvest darkening) (Freixas Coutin et al., 2017). The flavonoid biosynthetic pathway is organized into two distinct categories of genes: structural genes, which encode enzymes involved in different steps of the pathway; and regulatory genes, which govern the activity of structural genes by influencing their transcription process. Three families of transcription factors are involved in the regulation of structural genes: MYB, bHLH (basic helix-loop-helix) and WDR proteins. In dicotyledonous plants, the early biosynthetic genes (EBG) are regulated by MYB, while the late biosynthetic genes (LBG) are regulated by all three families of transcription factors, resulting in the formation of MBW (MYB, bHLH, WD40) complexes. MYB transcription factors play a crucial role in DNA binding and subsequent activation of DNA transcription. The bHLH transcription factor facilitates precise tuning of gene expression and the third component of the MBW complex the WD40 protein, stabilizes the MYB-bHLH interaction and provides structural support for the complex (Freixas Coutin et al., 2017; Paauw et al., 2019; García-Fernández et al., 2021).

The main regulator for the gene expression of seed coat color and flower color is the P gene. In plants with a homozygous recessive form of the P gene (pp), all seeds and flowers are white, irrespective of the presence of other genes responsible for the pigmentation of the seed coat. The protein is a member of clade B of subclass IIIf of plant bHLH proteins (Bassett, 2007; McClean et al., 2018). The influence of additional genes on seed coat color can be classified into two categories. The first group of genes, including Bic, C (which encodes a MYB transcription factor, PvMYB113), J (which is suspected to act as a MYB transcription factor), Prp ⁱ -2, R-2 and Sal, determines the color of the seed coat. The second group of genes, including B, G, Rk (possibly a MYB transcription factor) and V (encoding flavonoid 3´5´hydroxylase), influences the intensity of the seed coat color. The primary color, which encompasses the entire surface of the seed coat and can be observed as black, yellow, red, etc., is determined by genes for both seed coat color and color intensity (Bassett, 2007; McClean et al., 2018; Erfatpour and Pauls, 2020; García-Fernández et al., 2021; Celebioglu et al., 2023; Parker et al., 2024). However, numerous genes possess multifunctional characteristics. For instance, locus C comprises a set of closely linked genes, including Prp, Prp ⁱ, and Gy, which influence on seed coat color. Additionally, M, Pr, Acc and R, have been demonstrated to exert an influence on seed coat pattern. Patterns observed include a lighter background color and a darker pattern color, which are classified as pinto, striped and mottled types (Bassett, 2007; Wu et al., 2019). The dominant form of the T gene is responsible for the production of fully colored seeds and flowers, whereas the homozygous recessive form (tt) results in partial coloration. The gene encodes a transcription factor, WD40. Different mutations and deletions may result in the loss of its function, leading to the different partial coloration of the seeds and it can also influence flowers. The distribution of the colored part is also influenced by the action of other genes, including Bip (identified as the PvMYC1 transcription factor), Fib, J and Z (which encodes the MYB transcription factor, MYB113) (Bassett, 2007; Wu et al., 2019; García-Fernández et al., 2021; Parker et al., 2024). A considerable number of these genes possess multiple alleles, exert influence on both primary and secondary seed coat coloration, as well as on the coloration of different plant parts (pod, flower, stem), and exhibit epistatic interactions that result in a wide variety of seed coat colors (Bassett, 2007; McClean et al., 2018, 2022; García-Fernández et al., 2021).

The results of recent studies have provided greater insight into the specific functions of certain genes, however, a significant number of genes remain without a clearly defined function. Despite this progress, the relationship between seed coat color variation and underlying genomic factors remains poorly understood, especially in composite populations where phenotypic diversity is high. In some cases, seeds from the same plant may exhibit variation in seed coat color as a consequence of adaptation to unfavorable environmental conditions, increased outcrossing, mutation or genetic drift (Tiranti and Negri, 2007; Klaedtke et al., 2017; Šajgalík et al., 2019). An understanding of the genetic control of seed coat color variation and how it correlates with population structure could provide critical insights into the adaptive potential of common bean populations. This study evaluated composite populations of common bean, which exhibited differences in primary and/or secondary seed coat color. Each composite population exhibited between two and five distinct seed phenotypes, which were evaluated individually. By focusing on composite populations, this study offers a unique perspective on the genetic diversity within populations that are subject to evolutionary forces such as selection, gene flow and genetic drift. The objective of this research was to examine the genetic diversity and population structure of composite common bean populations, with a particular focus on identifying genetic variations and specific genomic regions associated with seed coat color variation. A comprehensive understanding of the genetic diversity and structure of these populations is essential for elucidating the principles of adaptive evolution and for developing effective breeding strategies. Additionally, our findings provide a foundation for future studies aiming to explore the ecological and evolutionary significance of seed coat color in Phaseolus vulgaris.

This study represents the first genomic characterization of composite populations of the common bean. Given their diverse origin and phenotypic seed characteristics, these populations represent a unique set with unknown traits and potential advantages, particularly in the context of organic breeding and climate change. The study included four composite populations of common bean, comprising 17 individual phenotypes that exhibited the highest level of segregation following the initial year of field evaluation, as well as two standard varieties. The two-year phenotypic analysis revealed considerable diversity in seed coat color within and among the four composite populations, with notable differences in growth habit and plant type observed among populations. In 2023, changes in seed coat color were observed in the populations KIS_Amand, SRGB_00189 and SRGB_00366, whereas population INCBN_03048 and the standard varieties ETNA and Golden_Gate remained phenotypically stable in both years. The results of multiple correspondence analyses (MCA) indicated that seed coat color was a significant discriminating factor, with the populations clustering into two main groups based on their dark or light seed coat. The highest degree of variability was observed in KIS_Amand and SRGB_00366, while SRGB_00189 exhibited more uniform characteristics, particularly in regard to growth habit. The changes in seed coat color observed in KIS_Amand, SRGB_00189 and SRGB_00366 between years 2022 and 2023 suggest the potential influence of genetic or environmental factors. In contrast, the stability of seed coat color in INCBN_03048 and both standard varieties ETNA and Golden_Gate suggests the possibility of genetic resistance to phenotypic shifts. These findings underscore the significance of investigating the genetic basis of seed coat color variation, as this may elucidate key adaptive traits for incorporation into breeding programs. The variability observed in common bean composite populations can be attributed to a number of factors, including mutations, genetic drift, environmental conditions, and local adaptations, which can lead to an increase in the outcrossing rate. Historical landraces have demonstrated notable shifts in genetic and phenotypic traits after only a few generations in differential environments, in part due to gene flow from proximate varieties (Klaedtke et al., 2017). While ecological and geographical factors contribute to differentiation, evidence suggests that a significant portion of seed coat color variation in common beans is genetically determined, indicating the potential for targeted optimization through selection (Bassett, 2007; McClean et al., 2018; García-Fernández et al., 2021; Parker et al., 2024).

Building on these findings, we conducted whole-genome sequencing to generate a comprehensive genetic and genomic dataset for the four composite populations and the two standard varieties of Phaseolus vulgaris. This analysis revealed a high level of genetic diversity, providing previously unavailable insights into the genomic architecture and evolutionary history of the species. The high-quality sequencing data (427 GB) and depth of coverage (30 ×) represent a significant advancement over previous studies in common bean genomics (Raggi et al., 2019; García-Fernández et al., 2021), facilitating the identification of subtle genetic variations that may have been previously overlooked. The remarkably high mapping rate (97.71%) to the reference genome indicates robust genomic conservation across populations, despite observed phenotypic variation. The distribution of over 8.6 million high-confidence SNPs across the genome provides a detailed representation of the genetic variation observed in the studied germplasm. The observed SNP density (16.68 SNPs per kilobase) is consistent with the general levels of genetic variation reported in previous studies on common bean populations. Additionally, distinct chromosomal patterns in SNP density were observed, potentially reflecting the evolutionary history and population dynamics of these populations (Raggi et al., 2019; Delfini et al., 2021). The higher concentrations of SNPs on chromosomes four and one in comparison to the lower densities on chromosomes three and six indicate differential selection pressure across the genome. These patterns may be the consequence of historical breeding practices or natural selection acting on specific genomic regions associated with agronomically important traits. In accordance with the findings of Valdisser et al. (2017) regarding a Brazilian core collection of common bean, our study similarly revealed a prevalence of SNPs in non-coding regions, offering valuable insights into the regulatory architecture of the common bean genome. The prevalence of variants in intergenic, upstream and downstream regions indicates that adaptations may be predominantly attributable to changes in gene regulation rather than alterations in protein-coding sequences (Li et al., 2023). The structural categorization of SNPs as predominantly modifiers with limited high-impact variants reflects a balance between conservation and innovation within the genome. Moreover, the observed transition-to-transversion ratio of 1.97, coupled with the prevalence of C/T transitions, provides insights into the mutational mechanisms influencing these populations. This findings are in accordance with those reported by Valdisser et al. (2017) for a Brazillian common bean collection, Gelaw et al. (2023) for an Ethiopian collection, and Lioi et al. (2019) for an Italian common bean collection.

Furthermore, despite the relatively limited sample size (1 – 7 individuals per population), we found considerable levels of genetic diversity (He = 0.047 – 0.205), comparable to those reported in larger studies on domesticated common bean collections, such as those by Valdisser et al. (2017) (He = 0.168, n = 111) and Rodriguez et al. (2016) (He = 0.157, n = 100). Even in smaller but phenotypically diverse collections, such as the one studied by Cichy et al. (2015) (He = 0.233, n = 21), the genetic diversity observed here is consistent with the broader diversity across common bean varieties. Interestingly, the genetic diversity patterns observed in this study revealed significant correlations between genomic variation and seed coat color change across diverse populations. Indeed, the composite populations KIS_Amand and SRGB_00366, which both exhibited seed coat color change, demonstrated the highest levels of nucleotide diversity (π) and allelic richness (Ar), which align with their high morpho-phenotypic diversity. This suggests that the genetic diversity observed in these populations is associated with their capacity to manifest diverse phenotypes, such as variations in seed coat color. The elevated genetic variation observed in these populations is likely to facilitate their adaptability to different selection pressures, whether occurring naturally or as a result of human intervention. These patterns may be influenced by agricultural practices, whereby the selection of traits such as seed coat color may be combined with the preservation of genetic diversity at other loci, thereby promoting a wide range of adaptive potential (Khoury et al., 2022; Alam and Purugganan, 2024). Conversely, population SRGB_00189, which exhibited a change in seed coat color, showed a discrepancy between the observed heterozygosity (Ho = 0.097) and the expected heterozygosity (He = 0.058). This was accompanied by a negative Fis value, indicating an excess of heterozygotes within the population. Two distinct patterns are observed in the seed coat coloration. The primary pattern exhibits darker mottling on a lighter-colored background, while the alternate pattern reverses this, with the background appearing darker and the mottling lighter. This patterning appears to be under the control of the C locus. Specifically, a heterozygous form of the C gene (Cc) exerts an influence on the pattern, with additional modifier genes within the C locus affecting the specific appearance of the pattern. This type of seed coat variation is observed at low frequencies in the majority of bean cultivars with C-locus patterns (Bassett, 2007). The heterozygote excess and distinctive seed coat pattern observed in SRGB_00189 may be indicative of genetic drift or a distinctive demographic history, such as population bottlenecks or admixture events. Such events could have resulted in the preservation of heterozygosity in specific genomic regions (such as the C locus) while leading to genetic uniformity (fixation) in others. The lower nucleotide diversity in SRGB_00189 (π = 0.067) provides further evidence that this population may have undergone recent population expansions or been subject to genetic drift, which may have impacted loci controlling seed coat color, leading to phenotypic fixation. It is noteworthy that the two standard varieties, ETNA and Golden_Gate, which also exhibited minimal variation in seed coat color, demonstrated comparable levels of genetic diversity (He = 0.047 – 0.108). The low nucleotide diversity and fixed phenotypes observed in these varieties are likely the result of intensive selective breeding, where uniformity in traits such as seed coat color is prioritized, potentially at the cost of broader genomic variation. This lends support to the notion that rigorous selection for particular traits, such as seed coat color, can result in a reduction of genetic diversity across the genome. Moreover, previous studies have shown that high-frequency ROH regions shared by individuals frequently indicate the presence of selection pressure in populations (Pemberton et al., 2012; Caivio-Nasner et al., 2021). In our study, the composite population SRGB_00189 and the standard variety ETNA, both of which exhibited low genetic diversity and a stable seed coat color, demonstrated the highest ROH coverage. This indicates that there has been a significant focus on selective breeding with the objective of maintaining specific traits, including seed coat color. In contrast, the extensive ROHs observed in KIS_Amand indicate the potential for inbreeding and selection pressure, particularly in relation to the observed phenotypic changes in seed coat color. Furthermore, the genetic structure analysis yielded a clustering pattern that was consistent with the morpho-phenotypic data, thereby reinforcing the assumption that seed coat color variation is predominantly driven by genetic factors. These findings highlight the pivotal role of selection in shaping both the genetic and phenotypic diversity within the analyzed germplasm. The differential nucleotide diversity (π) between populations with altered and stable seed coat color provides compelling evidence that selection is acting on seed coat color-related genomic regions in common bean. The 15.69% reduction in diversity (π = 0.30 vs 0.36) observed in populations displaying alterations in seed coat color is consistent with selective sweep patterns, whereby positive selection results in a reduction in genetic variation within regions linked to advantageous alleles. Similar findings have been documented in chickpea, where a 61% reduction in nucleotide diversity was observed in light brown/yellow-brown ‘desi’ chickpea accessions relative to ‘kabuli’ accessions with beige seed coats. This highlights the influence of strong purifying selection on seed coat color (Bajaj et al., 2015). Comparable signatures of selection have also been documented in chickpea (Varshney et al., 2013) and in rice (Sweeney et al., 2007), which further emphasizes the evolutionary pressures shaping seed color variation across crop species.

It is important to acknowledge that evolutionary and demographic processes, such as the bottleneck effect at the outset of breeding and the different effective population sizes throughout the improvement process, can give rise to discrepancies in genetic composition among populations. This, in turn, may give rise to false positives (Zhang et al., 2024). However, the combination of XP-EHH and XP-CLR methods employed in this study effectively mitigates the confounding effects. This robust analytical approach enabled the identification of divergent genomic regions that are likely to harbor candidate genes or loci associated with phenotypic changes in seed coat color within the studied populations. Previous research has successfully utilized these methods to detect significant genomic regions in various legumes, including common bean (M Rendón-Anaya et al., 2023) and soybean (Zuo et al., 2022; Cho et al., 2024). Notably, XP-EHH is particularly adept at identifying selective genomic regions even with small sample sizes (Pickrell et al., 2009), thereby enhancing the reliability of our analysis and strengthening our confidence in the identified candidate regions. For example, Zhang et al. (2021) used 39 samples, while Kumar et al. (2020) used only nine samples. Our sample size is also consistent with several pivotal studies on common bean genomics, including Bitocchi et al. (2017b; n=45), which have made important contributions to the field.

Previous studies have shown that seed coat color in legumes is predominantly determined by the accumulation of phenolic compounds, including flavonoids, phenolic acids, and proanthocyanidins, which are synthesized through the shikimate/phenylpropanoid pathway and specifically accumulate in the seed coat tissue (Ranilla et al., 2007; Singh et al., 2017). While traditional genetic studies in common bean have focused on genes directly controlling pigment biosynthesis and spatial distribution (Bassett, 2007; Parker et al., 2024), our results demonstrated that there is an additional layer of complexity in determining seed coat color. Using complementary selection scans (XP-EHH and XP-CLR) led to the identification of a total of 118 candidate regions were identified, 29 of which exhibited a high degree of enrichment within the phosphatidylinositol pathway. These findings indicate that cellular transport mechanisms may be a significant factor influencing the observed variation in seed coat coloration. The connection between phosphatidylinositol signaling and pigmentation is consistent with findings from other legumes, where anthocyanin production is induced via the phosphatidylinositol 3-kinase/Akt pathway in black soybean seed coats (Tsoyi et al., 2008) and phosphatidylinositol 4-phosphate 5-kinase is upregulated in faba bean seed coats (Ray et al., 2015). The significant enrichment of phosphatidylinositol-related functions and vesicle-mediated transport (GO:0016192) in the identified selection signatures indicates the presence of a coordinated system in which the cellular transport machinery works in conjunction with pigment biosynthesis to determine seed coat color phenotypes (Wan et al., 2016; Ren et al., 2017; Campa et al., 2023).

The integration of high-coverage whole-genome sequencing with detailed phenotypic analysis in this study has revealed novel insights into the genetic architecture of seed coat color variation in common bean. The identification of 8.6 million high-confidence SNPs revealed distinct patterns of genetic diversity across composite populations. It is noteworthy that KIS_Amand exhibited the highest nucleotide diversity (π = 0.222) and allelic richness (Ar = 1.380), which contrasts sharply with the limited diversity observed in SRGB_00189 (π = 0.067, Ar = 1.327). This suggests that the two populations have undergone divergent evolutionary trajectories due to different selection pressures. Our analysis revealed two key findings regarding population structure and selection. First, the strong genetic differentiation between standard varieties (ETNA and Golden_Gate, Fst = 0.704) and composite populations illustrates the profound impact of selective breeding on genetic architecture. Second, the reduced genetic diversity in populations with modified seed coat colors points to ongoing positive selection, particularly evident in the 118 selective sweeps identified primarily on chromosomes four, nine, ten and eleven. The pathway analysis revealed a critical role for phosphatidylinositol signaling in seed coat pigmentation. This finding extends beyond the traditional pigment biosynthesis pathways to implicate cellular transport mechanisms in color development, suggesting a more complex regulatory network than was previously recognized. The enrichment of genes involved in vesicle-mediated transport (P < 0.001) and exocytosis (P < 0.01) within selective sweep regions provides substantial support for this novel mechanistic model. These findings have direct implications for the development of breeding programs. The identified selective sweeps provide precise genomic targets for breeding programs focused on seed coat color modification. The high diversity observed in composite populations such as KIS_Amand represents a valuable genetic resource for maintaining crop resilience while improving aesthetic traits. The newly discovered role of transport pathways offers alternative approaches of modifying seed coat color characteristics, extending beyond the scope of traditional pigment pathway manipulation. With further aim of validating the identified regions and conducting transcriptomic analysis to better understand the underlying regulatory mechanisms, this research advances both theoretical understanding and practical applications in common bean improvement, thereby contributing to broader goals of crop enhancement and agricultural sustainability.