Genetic Diversity and Distinctiveness of Common Beans (Phaseolus vulgaris L.) Between Landraces and Formal Cultivars Supporting Ex Situ Conservation Policy: The Borlotti Case Study in Northern Italy

Abstract

The common bean (Phaseolus vulgaris L.) arrived in Europe in the sixteenth century from the American continent and spread across the old continent as a result of adaptation to different climatic and geographical conditions, as well as selection for different cultivation techniques and eating habits. This expansion gave rise to a huge number of landraces, which have now been largely replaced by more productive cultivars. To avoid losing this genetic diversity heritage, it is essential to characterize the genetic resources at risk of extinction for correct in situ/ex situ conservation and as a first step toward promoting and relaunching their exploitation. In this article, we report the study of a group of Borlotti beans, both landraces and modern cultivars, which are typically cultivated in Northern Italy. The aim was to explore the variability within the assembled Borlotti panel to assess whether landraces have maintained their morphological and genetic identity over time during on-farm cultivation. In particular, we investigated whether it is possible to distinguish between landraces and commercial accessions, a topic that has so far been understudied, but in the field of conservation, it is particularly important because it allows for the prioritization of preserving genetic resources. We found distinctive traits among the various accessions, with only a few landraces maintaining their identity, many of which should more properly be defined as obsolete cultivars. Taken together, these data indicate that it is possible to establish a priority scale for in/ex situ conservation by comparing landraces and cultivars using morphological and genetic data. Furthermore, thanks to geographical isolation, on-farm conservation has proven to still be successful in maintaining the identity of landraces.

1. Introduction

The common bean (Phaseolus vulgaris L.) is a widely consumed legume with a rich history and cultural significance. This species belongs to the family Fabaceae, subfamily Faboideae, genus Phaseolus. Although the common bean is predominantly a self-fertilizing species, the wild and cultivated forms can easily hybridize [1].

Mesoamerica and South America are recognized as the two centers of origin of wild common beans, where two independent processes of domestication took place, originating in the Mesoamerican and Andean gene pools. The first gene pool introduced to Europe was the Andean, likely from Francisco Pizarro’s expedition to northern Peru in 1528 [2]. Mesoamerican types were eventually spread to Europe following the introgression of Andean genome traits into Mesoamerican genotypes [3]. The Andean gene pool is the most frequent in European common beans, especially in the Iberian Peninsula, Italy, and Central–Northern Europe [4].

Italian Borlotti beans of Lamon, typical of the Belluno area in the north-east of the Veneto region, are considered the descendants of the first beans that arrived in Italy from Pizzarro’s trip to Peru via the Spanish emperor Charles V [5,6], and still today, their common name recalls their Spanish origin (i.e., Spagnol, Spagnolet) [7].

Borlotti and Borlotti-type beans are widely cultivated and consumed in various parts of the world, where they are known by different names such as Cranberry, Cargamanto, Rosecoco, Roman, and Pinto/Judias Pinto beans [1,2,6]. They are considered a high-value variety from both an agronomic and nutritional point of view and are of great commercial interest in Italy [8,9,10]. Borlotti beans are a typology of common bean with particular characteristics: well-mottled seeds; dwarf (with determined development) or climbing plant habitus (indeterminate development), characterized by suitability for mechanical harvesting; resistance to rust; and early defoliation and tolerance of high temperatures. In particular, in the fresh market, the pods are characterized by a long length, straight profile, and marked red streaks. These characteristics are found in local Italian cultivars, such as the four types of the Lamon bean, “Calonega”, “Canalino”, “Spagnolet”, and “Spagnol” [10,11].

Differences in climate, geography, cultivation techniques, and diet in Italy have led, over the centuries, to the differentiation of hundreds of common bean landraces [12]. Although the common bean is a predominantly autogamous species, genetic diversity within populations and landraces can be high. This reservoir of allele diversity helps natural populations cope with stresses and allows adaptation to specific, local environmental conditions [13].

In the early 20th century, a total of 472 Italian landraces of common beans had been documented [14]. Between 1950 and 1993, approximately 60% of these common bean types had disappeared and were replaced by a small number of modern cultivars, which had a higher and more stable yield [15]. Consequently, these traditional varieties, strongly linked to the local environment and culture in which they evolved, were relegated to marginal lands, associated with small-scale and low-input production for family consumption or local markets, and therefore at a high risk of extinction [13,14,15]. A remarkable portion of crop genetic diversity is currently represented by landraces and obsolete cultivars that have not yet been integrated into gene bank collections [13]. Their acquisition is an important goal for ex situ conservation, as is the evaluation of their genetic background [16]. Nevertheless, given the always rather limited resources available to gene banks for ex situ conservation, it is essential to focus conservation efforts on high-quality plant genetic resources. From this perspective, landraces represent genetic resources of greater value compared to obsolete cultivars, while conserving modern cultivars is not relevant to the objectives of gene banks. To this end, the ability to distinguish landraces [17] from cultivars, both modern [18] and obsolete [19], is of key importance. Obsolete cultivars [20] are cultivars that were cultivated in the past and which have fallen into disuse [21]. Unlike landraces, obsolete cultivars underwent formal crop improvement but were replaced by more productive, modern cultivars. They are considered of great importance for cultivation since many of them are in the pedigrees of modern cultivars [22]. The distinction between landraces and obsolete cultivars is, in many cases, particularly difficult, as both are defined by current growers as “old family varieties” [19].

Morphological characterization is a fundamental step in the process aimed at the unambiguous identification of different accessions, although it is not always sufficient to distinguish landraces from modern cultivars and provide a definitive identification.

The use of molecular markers, enabling genome-wide characterization, can resolve ambiguities remaining after morphological identification. Among the different molecular markers used for genomic characterization, SSR (Simple Sequence Repeat) markers are a powerful tool for dissecting the extent and configuration of genetic diversity for evolutionary and phylogenetic studies [13,15,23,24,25], including for landraces.

Next-Generation Sequencing (NGS) has further facilitated the development of efficient methods for comprehensive genome-wide characterization. The substantial number of genetic markers that GBS can discern in a resource-efficient and cost-effective manner allows the assessment of genetic diversity within the targeted cultivars. Furthermore, it enables the identification of distinct genotypes, the exploration of phylogenetic relationships, and the analysis of population structures among the examined varieties [26,27,28].

The promotion of landraces as unique local products has the potential to bring advantages not only for consumers but also for the community that has conserved these landraces from generation to generation [29,30,31]. There are many examples of characterization aimed at the conservation and valorization of common bean landraces in the literature, particularly landraces typical to the central and southern regions of Italy [15,24,25,29,32]. To our knowledge, there are only a few papers that refer to landraces in Northern Italy, with little data regarding Borlotti beans [33,34], particularly in the Lombardy and Emilia–Romagna regions [2].

In this paper, we report the phenotypic and genetic characterization of 28 accessions of the common bean, 25 being Borlotti or Borlotti-type (17 putative landraces and 8 cultivars) typical to the regions of Northern Italy (Emilia–Romagna, Lombardy, Veneto, and Piedmont), and 3 that are non-Borlotti landraces, as reference accessions.

The aim of this study is to characterize these common bean accessions, as a fundamental step for their identification, conservation, and promotion as unique entities linked to the territory of origin. In particular, this characterization has focused on the possibility of distinguishing true landraces, which have maintained their distinctive characteristics over generations, from cultivars, modern or obsolete, cultivated in the same areas.

In the area considered in this study—Northern Italy—where intensive cultivation of cultivars has promoted hybridization phenomena with traditional accessions, leading to genetic erosion, distinguishing these landraces from obsolete and modern cultivars remains a challenge.

Overall, the results shown in the present study will contribute to gathering data for the registration of some accessions in special conservation lists as “Conservation varieties” or for their designation as Protected Designation of Origins (PDO) or Protected Geographical Indication (PGI) products, thus potentially promoting their use and contributing to their conservation.

Finally, the characterization of these genetic resources can provide a greater understanding of traits that can help improve the resilience of crops to climate change [35,36].

2. Materials and Methods

2.1. Plant Material

We identified all the accessions conserved at the germplasm bank of the University of Pavia corresponding to putative landraces of Borlotti and Borlotti-type beans cultivated in Northern Italy, particularly in the regions of Lombardy, Veneto, and Emilia–Romagna, together with some non-Borlotti accessions as an outgroup. These accessions were collected from small farms that continue to cultivate them despite the availability of more productive, modern cultivars. For each accession, a record is available in the germplasm bank database indicating the origin and the number of the seeds, the year in which cultivation began, its general agronomic and morphological characteristics, its use, and any peculiarities of the accession. Often, the sample consists of only very few seeds; therefore, accessions where no seeds germinated could not be included in the analyses. We also used cultivars purchased from seed stores, for which the origin is the location where they are marketed.

We selected 46 Phaseolus vulgaris accessions, 38 of which were Borlotti or Borlotti-type and 8 reference (i.e., non-Borlotti) entities.

We identified an accession as “Borlotti” either when it was identified as such in records from farmers or when it was clearly recognizable by the described morphology. Borlotti-type accessions were identified as those with similarities in some traits to Borlotti.

Among the 46 accessions, 18 did not germinate or did not complete their life cycle, although these materials were cultivated locally and produced by farmers. Therefore, we performed morphological and genetic characterization on the remaining 28 accessions (

Table 1. List of accessions analyzed. ID: identification number; Bank Code: University of Pavia (from ID1 to ID36 and from 44 to 47) or University of Torino (from ID37 to ID43) germplasm bank identification code; Accession Name: traditional name of the entity under examination; Borlotti: “Yes” indicates Borlotti beans, “Yes*” is for Borlotti-type beans, and “No” is for non-Borlotti beans; Region: region of origin and/or collection of each accession; Type: putative landrace or cultivar; GBS: n° of plants analyzed; Phenotyping: n° of plants analyzed. N/A: data not available.

ID	Bank Code	Accession Name	Borlotti	Region	Type	GBS	Phenotyping
1	2241	Borlotto	Yes	Emilia–Romagna	putative landrace	2	1
3	2285	Borlotto	Yes	Emilia–Romagna	putative landrace	2	1
4	2249	Borlotto	Yes	Emilia–Romagna	putative landrace	3	2
5	3074	Borlotto	Yes	Emilia–Romagna	putative landrace	3	1
7	1804	Brunetto	Yes*	Emilia–Romagna	putative landrace	3	2
8	1358	Borlotto	Yes	Emilia–Romagna	putative landrace	3	1
9	2663	Borlotto	Yes	Lombardy	putative landrace	2	1
10	2885	Borlotto	Yes	Emilia–Romagna	putative landrace	3	2
11	2891	Borlotto	Yes	Emilia–Romagna	putative landrace	3	2
12	2884	Borlotto	Yes	Emilia–Romagna	putative landrace	3	3
14	1828	Borlotto	Yes	Lombardy	putative landrace	3	2
15	1817	Dorato di Codera	No	Lombardy	putative landrace	2	1
17	2401	Borlotto	Yes	Lombardy	putative landrace	3	2
18	1824	Borlotto	Yes	Lombardy	putative landrace	3	N/A
19	2468	Borlotto della Valchiavenna	Yes	Lombardy	putative landrace	3	2
20	1913	Fagiolo di San Giacomo Filippo	No	Lombardy	putative landrace	2	1
21	1785	Fagiolo di Lamon	Yes*	Veneto	putative landrace	3	3
22	1789	Fagiolo di Lamon della vallata bellunese	Yes*	Veneto	putative landrace	3	N/A
23	1781	Fagiolo “bala rossa”	Yes*	Veneto	putative landrace	3	2
24	1922	Spagnolet di Lamon	Yes*	Veneto	putative landrace	3	1
25	2171	Spagnolet di Lamon	Yes*	Veneto	putative landrace	3	N/A
26	2825	Borlotto	Yes	Lombardy	putative landrace	2	1
28	3080	Borlotto	Yes	Lombardy	cultivar	2	N/A
29	3078	Anellino d’Oltrepò	No	Lombardy	putative landrace	2	1
35	3988	Fagiolo di Saluggia	Yes*	Veneto	cultivar	3	1
36	3989	Rossano	Yes*	Piedmont	cultivar	3	3
37	VLA042	Sanguigno	Yes*	N/A	cultivar	3	3
38	VLA046	Regina rossa di Boves	Yes*	Piedmont	cultivar	3	2
39	VLA047	Borlotto di Entraque	Yes	Piedmont	cultivar	3	2
40	VLA050	Regina precoce di Roccavione	Yes*	Piedmont	cultivar	2	1
41	VLA054	Borlotto gigante	Yes	Piedmont	cultivar	3	2
42	VLA056	Regina rossa di Centallo	Yes*	Piedmont	cultivar	3	N/A
43	N/A	Billò	Yes*	Piedmont	cultivar	2	1
44	3458	Fagiolo del Belesi (rosso)	Yes*	Lombardy	putative landrace	1	N/A
45	3535	Borlotto della Nina Laini	Yes	Lombardy	putative landrace	1	N/A
46	3457	Borlotto della Nina Laini	Yes	Lombardy	putative landrace	1	N/A
47	2410	Borlotto nano di Pietragavina	Yes	Lombardy	putative landrace	2	N/A

, Supplementary Figure S1).

To gain a more comprehensive understanding of the genetic diversity along with the 28 selected accessions, we also included 5 Borlotti and 4 Borlotti-type bean accessions in this analysis (7 putative landraces and 2 modern cultivars). These accessions belonged to the group of 46 initially selected accessions; however, their morphological analysis could not be completed because they did not complete their life cycle. Nevertheless, their DNA could still be extracted. This resulted in a total of 37 accessions characterized at the DNA level, bringing the number of samples characterized by GBS to 94 (Table 1, Supplementary Figure S1).

Of the 25 Borlotti and Borlotti-type accessions, 16 are classified as Borlotti beans, and the remaining 9 accessions are Borlotti-type (Table 1, Supplementary Figure S1). The accessions selected are traditionally cultivated in 4 regions of Northern Italy: Emilia–Romagna, Piedmont, Lombardy, and Veneto (Figure 1).

The 3 non-Borlotti varieties, very different in morphology compared to the Borlotti and Borlotti-type accessions (#15: Dorato di Codera; #20: Fagiolo di San Giacomo Filippo; #29: Anellino d’Oltrepo’), were included in the analysis as out-of-type references (Table 1 and Figure 2).

For simplification, in the following paragraphs, the different accessions will be indicated by their corresponding ID numbers (Table 1).

2.2. Phenotyping of Plant Material

During the 2021 growing season, the plants representing the 28 accessions included in the panel (Table 1, Supplementary Figure S1) were grown in greenhouses and in the CREA-GB experimental field in Montanaso Lombardo, Lodi, Italy (latitude 45.3425 N, longitude 9.4492 E), with sowing in April and harvesting in August. Due to the availability of material and germination of the seeds, a variable number of plants per accession (from one to five) were transplanted in the field, arranged randomly in double rows, and morphologically characterized. A total of 42 traits were recorded, including the type of growth, poise, coiling speed, flowering time, and the position, color, shape, size and roughness of the plant, leaf, pod, flower, and seed. All traits were measured under field conditions using the score values and following the criteria reported in the “Protocol for tests on distinctness, uniformity and stability Phaseolus vulgaris L. French bean, CPVO-TP/012/4” [37] (Supplementary Tables S1 and S2). Phenotypic data were employed for MCA (multiple correspondence analysis) in R, using the FactoMineR package, version 2.11.

2.3. DNA Extraction and Genotyping with ddRAD Sequencing

For each genotype examined in the present study, nucleic acids were purified from young leaves sampled at the beginning of the vegetative season using the GenElute™ Plant Genomic DNA Miniprep Kit (Sigma-Aldrich, St. Louis, MO, USA) following the manufacturer’s protocol.

The TapeStation system (Agilent, Santa Clara, CA, USA) along with Genomic ScreenTape was used to assess the quality control of the nucleic acids purified from plants. Moreover, the purity and concentration of the extracted nucleic acids were measured using a NanoDrop spectrophotometer and the dsDNA assay kit (Qubit BR; Life Technologies, Carlsbad, CA, USA), respectively.

Double-digest restriction-site-associated DNA (ddRAD) sequencing was conducted at the Functional Genomics Laboratory (University of Verona, Italy) to fingerprint DNA samples following published protocols [38], with the following modifications. For each sample, 200 ng of gDNA was digested using the following protocol: 2 h at 37 °C, 2 h at 65 °C and 20 min at 80 °C with 2 U of TaqI-v2 and 2 U of MseI (New England Biolabs, NEB, Ipswich, MA, USA) in 1× CutSmart buffer, in a final volume of 20 μL. The products of the digestion reactions were examined by comparing the digested DNA and the non-digested gDNA on a 4150 TapeStation using a Genomic DNA assay (Agilent Technologies, Santa Clara, CA, USA). The double-stranded barcoded adaptor (previously annealed, 0.025 μM final concentration) and the double-stranded common adapter Y with a biotin tail (previously annealed, 0.025 μM final concentration) were ligated to the digested DNA products using 1 U of T4 DNA ligase (Invitrogen, Carlsbad, CA, USA) in a 1× ligase buffer at a final volume of 50 μL.

A set of 24 different barcoded adaptors was used to uniquely label 24 samples at a time. Ligation reactions were performed in a thermocycler using the following thermal protocol: 10 min at 30 °C and 4 h at 22 °C. A final inactivation step of 30 min at 65 °C was finally applied. Barcoded DNA samples were subsequently pooled using the same final concentration per sample and purified using beads (0.75× volume of AMPure XP; Beckman Coulter, Brea, CA, USA) following the manufacturer’s instructions. The pool of purified DNA was subsequently resuspended in 25 μL of water. A total of 24 µL of biotinylated DNA fragments was captured in a reaction mixture containing a 1x volume of Dynabeads M-270 streptavidin (Invitrogen, Waltham, MA, USA), according to the manufacturer’s instructions. Captured DNA tied with beads was resuspended in 60 µL of water, 30 μL of which was subsequently amplified in a final volume of 50 μL in a reaction mixture containing 2 U of Taq Phusion polymerase in the presence of 1× Taq Phusion HF buffer, 0.3 mM dNTPs, and two different primers: Primer PCR1 (0.5 μM) and PPIX Illumina Index (0.5 μM), the latter of which includes the index for Illumina sequencing. Four PPI Illumina Index primers with four different Illumina indexes were used to multiplex four pools of 24 samples (96 total samples) at a time. Amplification was performed using the following thermal protocol: 30 s at 98 °C, 15 cycles of 10 s at 98 °C, 30 s at 65 °C, 30 s at 72 °C, and 5 min at 72 °C for final elongation. The resulting ddRAD libraries were purified with beads (0.65× AMPure XP; Beckman Coulter, Brea, CA, USA) and their size distribution was examined on a 4150 Tape Station using a D1000 HS Assay (average insert size was ~512 bp). ddRAD libraries passing quality controls were quantified by qPCR, pooled at an equimolar concentration, and sequenced on a NovaSeq 6000 platform Illumina with 2× 150 bp reads, generating 2.3 million fragments per sample on average.

2.4. SNP Calling Procedure

Raw read quality was assessed using FastQC v0.11.5 [39]. Subsequently, low-quality 3′ ends of the reads were trimmed, removing those bases below the Q20 threshold, and the sequencing adapter was removed using cutadapt v3.2. Filtered reads were aligned to the common bean reference genome ‘Phaseolus vulgaris v2.1’, downloaded from the Phytozome website [40], using bwa v0.7.17, and the resulting alignments were converted to bam files and sorted using samtools v1.13. Alignment files were cleaned to remove overlapping portions of the paired sequences using fgbio v1.3.0. Genomic variants were identified using GATK HaplotypeCaller v4.1.9 with the parameters “-min-base-quality-score 20 -ERC GVCF”. Individual gVCF files were merged using GATK GenomicsDBImport v4.1.9.0, and the final VCF file was generated using GATK GenotypeGVCFs v4.1.9.0. Variant filtration was achieved using GATK hard filters [41].

2.5. SNP Calling Procedure and Clustering Analysis

Raw read sequencing data were filtered using cutadapt 3.2 to remove adapter contamination and low-quality reads. Subsequently, for each sample, filtered reads were mapped with BWA v0.7.17 against the reference genome of the common bean [40]. The resulting BAM files were pre-processed with BAM clipping to remove duplicated reads. Processed BAM files were used for SNP calling using GATK 4.1.9 following GATK best practices for germplasm variant calling [42]. The resulting variants detected with GATK were hard-filtered by removing low-quality SNPs and those with over 90% of missing data, meaning sites not genotyped in more than 90% of common bean samples were removed from the SNP dataset using bcftools. Principal component analysis (PCA) was carried out using the full set of polymorphic SNPs identified with GBS, using base functions implemented in R [43], without applying either pruning or clumping procedures. PCA plots were created in R using the ggplot2 package [44]. To investigate the phylogenetic relationships among the common bean genotypes examined in the present study, SNP data were imported into R to compute the Euclidean distance. A phylogenetic tree based on the Euclidean genetic distance and the neighbor-joining method was constructed and plotted in R using the ggtree package [45].

Pairwise F_ST values among common bean populations and the genomic relationship matrix were computed using the R package dartR version 3.0 [46]. The R package dartR was also used to plot the genomic relationship matrix (GRM) as a network.

3. Results

3.1. Morphological Characterization

Of the 28 Phaseolus vulgaris L. accessions, one or three plants for each ID reached maturity, and it was possible to complete the morphological characterization, taking the traits described in the “Protocol for tests on distinctness, uniformity and stability Phaseolus vulgaris L. French bean, CPVO-TP/012/4” as a reference [37] (Supplementary Tables S1 and S2). The traits reported in CPVO-TP/012/4 [37] allow the characterization, in the open field, of discrete phenotypic data describing the plant, leaf, flower, pod, and seed characteristics of common beans to identify the individual accessions. We described the 42 morphological traits observed on dwarf plants (Supplementary Table S1) and climbing beans (Supplementary Table S2).

Among the seven dwarf accessions, it was possible to identify differences in plant height and in leaf and pod traits, which were useful for discriminating between different accessions (Supplementary Table S1).

Concerning the 21 climbing bean accessions, the morphological analyses highlighted differences at the level of plant architecture, climbing (start and speed), and leaf, pod, and flower traits (Supplementary Table S2). In particular, for flower traits, both flower color and size of the flower bracts were polymorphic among the different accessions. Most of the plants had pink or pinkish flower coloration; four accessions displayed white flowers; three accessions had violet flowers; and only one accession (#2) produced completely red flowers, a characteristic not mentioned in the reference phenotyping tables [37].

To analyze the phenotypic diversity among these 28 common bean accessions, the phenotypic data collected were subjected to an MCA. This analysis shows that the first and second MCA components explain 5.9% and 4.9% of the total phenotypic variability, respectively (Figure 3). The accessions on the right side of the graph shown in Figure 3 are all dwarf plants (Supplementary Figure S2), among which there is only one commercial variety (#35), along with a well-separated group consisting entirely of putative landraces (#1, #4, #5, #8, #12, #17). On the left-hand side is a larger group of climbing plants (Supplementary Figure S2), which includes both putative landraces and cultivars. Within this group, a subgroup is more widely distributed at the top of the graph, consisting exclusively of putative landraces (#3, #7, #15, #19, #20, #23, #29).

Considering the nine seed traits analyzed for each accession, all were polymorphic among both climbing and dwarf lines, with the exception of the trait “number of colours” (the seeds of all accessions analyzed have two colors) (Figure 4, Supplementary Tables S1 and S2). In particular, the morphological data indicated that considering only the traits relating to the seed, it is possible to distinguish 28 morphotypes, each describing a single accession (Figure 4, Supplementary Table S3). Overall, except for the veining trait, the phenotypic distribution of these ordinal categorical traits did not show substantial change in the cultivars vs. putative landraces examined in the present study (Figure 4).

3.2. Genetic Diversity Analysis

ddRAD sequencing was applied to a panel of 94 samples of common beans collected from 38 different accessions. For most of the samples, the plant material was taken from plants in the greenhouse and from at least one in the field. For accessions #44, #45 and #46, only one individual grown in the greenhouse was analyzed. To examine the diversity and population stratification of this panel of common beans, the resulting 13,070 polymorphic SNPs identified using GBS were subjected to a PCA. This analysis shows that the first and second PCA components explain 13.18% and 11.73% of the total genetic variability, respectively (Figure 5A,B).

From the PCA results, the genotypes appear to be roughly clustered into six groups (Figure 5B). Starting from the top left, “cluster 1” (Figure 5B) includes a first, very compact group of genotypes consisting mainly of accessions registered as putative landraces. Among these plants, three belong to accession #10, a Borlotti putative landrace native to the Emilia–Romagna region, together with accessions #45 and #46. These two accessions are putative landraces of Borlotti beans cultivated in the Lombardy region by different farmers within the same village (Tremosine, BS), a rather isolated location on a natural terrace overlooking Lake Garda. As indicated by the farmer who donated these accessions to the germplasm bank, these accessions were originally from Modena in Emilia–Romagna. Individuals from accessions #4, #5, #11, and #47, all registered as putative landraces, also belong to this group. A few individuals of two cultivars (#35B and #36D) can also be found in this portion of the chart. These two accessions appear non-uniform as the individuals are scattered in very different and distant positions on the PCA graph (Figure 5).

Immediately below, there is a second, rather compact group, “cluster 5” (Figure 5B). It is interesting to note that plants from accession #38, a Borlotti bean cultivar from the Piedmont region, differed morphologically from the other cultivars of the same region, mainly due to their dark pod and seed color (Supplementary Tables S1–S3). Furthermore, in this cluster, plants from the commercial accessions (#35 and #37) and from the putative landraces (#5, #8, #15, #17, #22, #23 and #25) appear to be closely associated.

In the bottom left-hand corner of Figure 5, “cluster 6” accessions registered as putative landraces (#21, #22, #23, #24, #26, #29) and two plants of the cultivar #28 are shown. This group includes some individuals of non-uniform accessions whose plants appear to cluster in distant areas of the PCA plot (#21, #22, #23, #24, #26). It is worth underlining the presence in this group of the two plants of the non-Borlotti reference accession #29.

On the right-hand side of the PCA plot, individuals of “cluster 4” are clearly distinguishable, and this is the only group made up of putative landraces; here, we find plants of four accessions originating in Lombardy and, with the exception of the plant of accession #17 (from Gambolò, PV), all the other plants (accessions #14, #18, and #19) are cultivated in a mountainous and very isolated area, Val Codera (Valchiavenna, SO).

Finally, cluster 2 includes the Borlotti bean commercial accession #39 and other non-homogeneous Borlotti bean accessions (both cultivars and putative landraces). The remaining plants are positioned along the central horizontal axis of Figure 5 and were grouped together in “cluster 3” according to our analysis. Interestingly, within “cluster 3”, other groups could be identified: a lower and broader area consisting only of landraces, and another area formed of a large group slightly higher up on the graph with a few individual of cultivars. Included in “cluster 3”, there is a small sub-cluster of three plants belonging to cultivars #40 and #41 originating in the Piedmont region (Figure 5).

SNPs obtained from ddRAD sequencing were used to build a phylogenetic tree based on the Euclidean genetic distance and the neighbor-joining method (Figure 6). The tree has three monophyletic groups, each containing Lamon beans. The beans from Val Codera (Lombardy, #14, #18, #19) were all grouped on the same branch. Furthermore, considering the geographic origin of the putative landraces, the first branch of the tree predominantly contains putative landraces from Lombardy (14 from Lombardy, 9 from Emilia–Romagna, and 3 from Veneto). The second branch, moving clockwise, mostly contains putative landraces from Emilia–Romagna (14 from Emilia–Romagna, 5 from Lombardy, and 4 from Veneto), while the third branch is a cluster of genotypes from different regions (6 putative landraces from Lombardy, 4 from Emilia–Romagna, and 8 from Veneto) (Figure 6).

The analysis of the observed heterozygosity (Ho) averaged over the 13,070 polymorphic SNPs detected with ddRAD analysis showed a value of Ho = 0.34.

The first attempt to cluster the common bean genotypes examined in the present study was carried out by plotting the genomic relationship matrix, computed with polymorphic SNPs, as a network. This analysis showed five main groups, each containing more than three common bean individuals, including both putative landraces and cultivars (Figure 7). In particular, one cluster is composed of a clearly defined subcluster, consisting of putative landraces (#14, 17#, #18, #19) exhibiting relatedness values between 0.4 and 0.5, and a second subcluster, consisting of cultivars (#36, #37, #40, #41) and putative landraces (#4, #44, #47), with lower values of genomic relationships. Interestingly, several genotypes showed values of genomic relationships below 0.25 and were consequently not connected with other genotypes in the network (Figure 7). Overall, the genomic relationship-based network indicates the existence of clusters composed of cultivars and putative landraces along with more diverse genotypes, most of which are classified as putative landraces (Figure 7).

As a further attempt to examine the genetic relationship in our panel of common beans, we considered putative landraces and cultivars sampled in the same location as belonging to the same population (Table 1). Pairwise analyses among these common bean populations showed that, overall, F_ST values varied from 0 to 0.3, indicating the existence of populations showing a limited degree of differentiation and significant differentiation, respectively (Figure 8). The common bean population sampled in Entracque (CN) showed a higher degree of genetic differentiation compared with the other populations examined in the present study. Although geographically close, the common bean populations sampled in Boves (CN) (#38) and Entracque (CN) (#39) showed the highest pairwise F_ST values identified in the present study.

4. Discussion

The morphological and genotypic characterization carried out in the panel of Borlotti beans assembled in the present study allowed the unique identification of each accession, a fundamental step for the correct conservation and valorization of plant genetic resources. Particularly, with a view to exploiting this genetic material, a specific challenge of this analysis was to find out if it was possible to distinguish true landraces from cultivars. This classification appears particularly challenging for genotypes sampled in the area under consideration, Northern Italy, where intensive cultivation of modern cultivars has led to hybridization phenomena with the traditional varieties still grown locally.

4.1. Morphological Characterization

Among the 28 accessions analyzed, the 25 accessions of the Borlotti or Borlotti-type beans that we phenotyped in the experimental field displayed peculiar traits of Borlotti beans. For instance, regardless of whether the plants were dwarf or climbing, they produced long pods, which in most cases showed a green color mottled with red or violet, and seeds exhibited the typical colored spots scattered over the entire surface of Borlotti beans [10,11]. However, the different accessions showed variability for many morphological traits, which allowed us to distinguish them from each other both at morphological and phenological levels.

High genetic variability between different local bean varieties, often reflecting variation in phenotypic traits, has been previously reported, not only in those with different geographical origins but also in those originating in the same area [24,27,47,48,49]. It has therefore been proposed that Europe might be considered a secondary diversification center for common beans [50]. Interestingly, our study examined accessions belonging to the subgroup of Borlotti or Borlotti-type beans, which already have a well-defined morphology, and consequently, morphological variation among entities was not expected.

The accessions included in the panel, in general, showed uniformity for the traits analyzed, with the exception of #12 (Borlotti). This accession was the least homogeneous, not only in characteristics for which it is difficult to attribute a value (e.g., leaf shape and color intensity) but also for two easily classifiable traits: flower color and pod curvature (Supplementary Table S1). The variation at the morphological level was also corroborated by GBS data. The PCA plot showed that plants belonging to accession #12 were scattered in the central part of the PCA within the heterogeneous cluster 3 of the putative landraces (Figure 5). However, a high level of genetic variability within accessions, expected for allogamous plants, has also been observed in autogamous plants such as the common bean. In fact, in a study on bean landraces from the Lazio region (Italy), high genetic variability was observed in 49% of the landraces analyzed, suggesting the presence of distinct genotypes and highlighting the effectiveness of on-farm conservation in preserving the allelic diversity of landraces [23].

Flowering time and traits tied to seed and pod morphology showed the greatest diversity in our panel of common bean accessions. Flowering time, in particular, appeared to be one of the most polymorphic traits among the climbing beans in our experimental field (Supplementary Table S2). This corroborates previous findings reported in the literature, which show a high level of variability for this trait [23,51].

Flowering time is known to be influenced by a gene-regulatory network responding to both endogenous and exogenous environmental factors, including plant age, hormones, photoperiod, light intensity, temperature, vernalization, nutrients, drought, and salinity [52,53]. The multiplicity of factors involved in determining this trait, together with the genotype-dependent effect, may explain the high variability we observed among the bean accessions in this study.

All phenotypic pod-related traits of the climbing bean plants in our field were polymorphic, indicating that the analysis of the size, shape, color and curvature of the pods can be a good starting point for identifying the different accessions under study.

The seed traits analyzed in the present study were also found to be mostly polymorphic. Seed color is perhaps the most attractive trait for the first macroscopic analysis of the plant material and as a marker in breeding programs [54]. In our selection of Borlotti and Borlotti-type beans, it was possible to detect differences in the color pattern. In particular, we found accessions whose seeds were characterized by an “inversion” in the color of the background and the spots (beige spots in accession #7, and black spots in accession #2 on a red background). A large variation in seed size and color among common bean landraces has previously been reported in a study of bean landraces from the Italian region Lazio, which considered five qualitative seed descriptors to evaluate 114 accessions belonging to 66 landraces. The authors identified 32 different morphotypes, with frequencies ranging from 1.52 to 12% [24]. In our case, the analysis of nine seed-related traits allowed us to identify 28 specific morphotypes, one for each accession analyzed (Figure 4, Supplementary Table S3). Similar results were shown in another study, which included 57 common bean landraces from the Italian region of Sicily. The authors identified 46 morphotypes based on seed traits, one describing three landraces, nine describing two landraces each and thirty-six describing unique landraces [25].

The higher number of morphotypes identified in our study compared to the aforementioned studies carried out on Lazio and Sicily bean collections might be explained by considering the higher number of traits we examined, which allowed a more complete characterization and therefore distinction of the different accessions [24,25]. This result also underlines how within a group of beans, characterized by a common morphology at the seed level (Borlotti beans), it is still possible to distinguish different morphotypes using traits at the seed level, as a consequence of different selection and adaptation processes for the different accessions.

The regulation of seed coat color is complex and involves many genes. Color-related genes are grouped into three categories: a ground factor gene (a transcription factor necessary for seed color), the color genes (C, Z, and J), postharvest darkening genes (J and sd), and color-modifying genes that intensify the expression of the color genes. The local environment can further influence seed coat color [55]. Therefore, considering the complexity in determining the seed color and the number of genes and alleles involved, the diversity in seed coat color we observed, not only among different varieties of common beans but also within the Borlotti bean group, is not surprising. In fact, the selection of this easily distinguishable trait by farmers could have led to the selection of different morphological variants in different landraces, as previously reported in the highly polymorphic morphological traits among tomato landraces [56].

4.2. Genetic Diversity

A PCA based on more than 13,000 SNPs identified using ddRAD sequencing depicted the diversity and population stratification on the panel of 94 samples of common beans examined in the present study. This analysis showed that the panel of common beans clusters into six main groups. Moreover, these genomic analyses highlighted limited intra-accession uniformity, as only plants of accessions #7, #10, #12, #28, #38, and #39 fell within narrow and close areas on the PCA graph (Figure 5).

We found that the three plants of accession #7 (a Borlotti-type accession) almost totally overlapped in the central right-hand part of the PCA. This accession appeared to be very uniform, and the genetic proximity to some cultivars, in particular the two plants of accession #37, suggest that accession #7 could be classified as an obsolete cultivar. Obsolete cultivars are cultivated varieties (i.e., the product of breeding programs) that are no longer marketed because they have been replaced by more productive modern cultivars [19]. They are still cultivated by small-scale farmers, who save some of the harvested seeds for the next season. Often, the memory of their ancient commercial origin has been lost, and they are considered landraces.

The presence of clusters of highly related landraces and cultivars is also highlighted by the phylogenetic tree and the genomic relationship-based network (Figure 6 and Figure 7).

Similar results were previously reported on the characterization of over 1200 accessions of traditional and modern European tomato varieties. Blanca and colleagues observed that approximately 25% of the varieties cataloged as landraces mapped closely to modern cultivars on a PCA and therefore re-classified them as belonging to this latter group [56].

In one of the mixed groups, containing both landraces and modern cultivars, it is interesting to note the presence of the Gambolò bean (#17). This bean has been registered as a “Conservation Variety” (Commission Directive 1998/95/EC), but considering its proximity to cultivars (#36, #37, #41, #42), it appears more likely to be an obsolete cultivar.

In another heterogeneous group identified using PCA, there are accessions classified as landraces (#4, #5, #10, #11 and #45) and cultivars (#35, #36 and #31). In this group, the first four landraces were originally sampled in Emilia–Romagna, while accession #45 was sampled in Lombardy (Tremosine, BS). This cluster can therefore be considered as a group of accessions from the macroareas encompassing the Lombardy and Emilia–Romagna regions.

Furthermore, it is also worth noting the proximity of sample #43A (Billò bean) to #21 (Lamon bean). This appears to confirm a relationship between a Borlotti bean from Piedmont and one from Veneto (Lamon), which cannot be explained by geography, but rather by the origin of the Billò bean. In fact, it is known that the Billò bean derives from the Lamon bean varieties [2].

The PCA analysis showed a peculiar heterogeneous group composed of a cultivar of Borlotti bean (#28) and six putative landraces (#21, #22, #23, #24, #26, #29), including two cultivated in Lamon (the area where the first Borlotti beans arrived in Italy in 1528).

The hybridization of different entities was hypothesized to explain the composition of mixed groups resulting from a PCA carried out on a group of European landraces and cultivars of tomato (Solanum lycopersicum), an autogamous species like the common bean [56].

The possibility of hybridization between different landraces has also been reported in the case of maize, whose allogamous nature makes this type of hybridization quite easy. In this case, only genetic analysis was able to highlight the hybridization; the landraces maintained their distinctive traits at the morphological level [57].

According to PCA and MCA analyses, two groups appear to be made up of true landraces: the widely distributed central group 3 of PCA (accessions #1, #12, #20), which includes an accession that is not a Borlotti-type (#20), and group 4 in the top right corner of the PCA composed of accessions #14, #18, and #19 (Figure 5A,B). These three accessions are cultivated in Val Codera, a very isolated valley in the Alps of Lombardy, which, for geographical reasons, may have avoided hybridization with other bean varieties. This observation is also supported by the analysis of the phylogenetic tree (Figure 6) and the genomic relationship matrix (Figure 7), which showed that all the plants belonging to the accessions cultivated in Val Codera were grouped in the same cluster. Distance and geographical barriers therefore appear to contribute to maintaining the identity of the different accessions, while political distinctions (i.e., regional borders) do not play a role in avoiding hybridization or limiting seed exchanges between farmers. The high conservation value of these landraces, which have retained their uniqueness thanks to isolation, suggests that these important genetic resources should be preserved through the Register of “Conservation Varieties” (Commission Directive 1998/95/EC) [58].

A different situation, however, was observed for the two cultivars #38 (Regina Rossa di Boves) and #39 (Borlotto di Entraque), which, despite being sampled in very close locations, maintained the highest level of genetic diversity in the group of beans analyzed (Figure 8). This result is also confirmed by the phylogenetic tree in which the two varieties are found in distinct monophyletic groups (Figure 6). In the case of cultivars, seeds are purchased for sowing and are not stored from one generation to the next, as is carried out with landraces. Consequently, the effects of possible hybridization with other accessions are not carried forward from generation to generation.

The values for Ho reported in the literature for common bean landraces are quite different, varying from 0 to 0.779 for landraces from Slovakia and Ukraine [13], from 0 to 0.050 for landraces sampled in the Italian region Campania [15], and from 0 to 0.067 for landraces sampled in the Italian region Lazio [24], while a value of 0 is reported for landraces from Croatia [12]. The value we obtained, 0.34, suggested a good level of genetic diversity, especially considering that we are studying an autogamous species and, apart from the four out of type, all the accessions share the Borlotti morphotype.

The commonly found isolation of mountain areas makes the maintenance of agrobiodiversity easier than in the plains, as has previously been reported in the literature. In particular, a survey on the biodiversity of Italian mountain areas has shown a wealth of landraces that led the authors to suggest that these areas could be considered hot spots of agrobiodiversity [59].

Our analysis of this group of landraces has shown that on-farm conservation can be effective in maintaining the identity of accessions, even allowing their evolution in the environment where they developed their unique characteristics. This is particularly effective in situations of geographic isolation. Considering the initial classification of putative landraces compared to cultivars (Table 1), the results of our genetic analysis lead us to suggest the identity of landrace for 6 out of 20 accessions analyzed (#1, #12, #14, #18, #19, #20). These data confirmed the difficulty in finding pure landraces after the introduction of modern cultivars. They can still be found, but only in isolated areas, where large distances or geographical barriers prevent hybridization phenomena with modern cultivars [16]. This suggested the adoption of a new concept of landraces, defined as “plant materials consisting of cultivated varieties that have evolved and may continue evolving, using conventional or modern breeding techniques, in traditional or new agricultural environments within a defined ecogeographical area and under the influence of local human culture” [60].

However, examples of successful on-farm conservation of landraces have been reported for both autogamous and allogamous species [61,62,63]. In particular, it was shown that on-farm conservation by farmers resulted in the maintenance of (in the case of maize landraces) or increase (in the case of rice landraces) in genetic diversity in comparison to the variation in accessions conserved ex situ [61,63].

A separate consideration should be made for the Lamon bean accessions (#21, #22, #23, #24, #25), present in different groupings, including together with cultivars. In this case, the proximity could be explained by the fact that Lamon beans are considered the descendants of the first beans that arrived in Italy in 1538 [5,6,7]. Therefore, considering even subsequent introductions, it is plausible that Lamon beans are genetically close to many of the bean varieties present in the area today. This observation seems to be confirmed by the phylogenetic tree, which shows accessions of Lamon beans in each of the three monophyletic branches, supporting the hypothesis that the Borlotti beans analyzed in this study derive from the first beans introduced to Italy.

It is worth emphasizing the importance of the conservation of obsolete cultivars, as they, like landraces, have adapted to the environment in which they have been cultivated for generations. Therefore, they can act, like landraces, as a reserve of useful genes for breeding programs aimed at selecting plants capable of adapting to the ongoing environmental changes.

The morphological characterization we have performed has highlighted differences between accessions, which is useful for their correct identification. Comparing the results of the PCA on the genetic data and MCA on the phenotypic data, it emerges that the analysis based on the genetic data is more discriminating, as it is able to define a greater number of clusters. Phenotypic uniformity within the clusters identified by the genetic analysis with SSR has previously been reported in the study by Santalla and colleagues [64], who characterized Spanish and Portuguese common bean landraces. These authors suggested a two-step process to classify traditional bean landraces, a first classification based on molecular data, then a morphological description of each group defined by the molecular approach [64].

In our study, the results of the PCA and MCA analyses are in agreement; in both, accessions #1, #12, #20, and #19 appear to be well separated from the cultivars and can therefore be considered true landraces. However, it was only through genomic analysis that we were able to reveal possible hybridization events or common origins and, for some accessions, to suggest an obsolete cultivar classification rather than that of a true landrace. Therefore, GBS analysis appears to be an efficient technique for the genomic characterization of common bean landraces, as previously reported in many other crop species [27,49,63,64,65,66,67,68,69].

5. Conclusions

Current-day Italian Borlotti beans are considered to derive from the first beans that were introduced to Italy in 1528 [2]. Adaptation to different geographical, climatic, and cultural conditions has led to the appearance of a myriad of landraces, while breeding programs have produced cultivars. Our results showed a clear differentiation in morphology between the accessions analyzed. Genetic analysis revealed that the distance between landraces and cultivars was particularly evident in the group of landraces originating from the Codera Valley (a remote area in the Alps of Lombardy), where strong geographical isolation has prevented hybridization. Conversely, a common origin and possible hybridization may explain the genetic proximity between putative landraces and cultivars, along with the possibility that some entities, which are classified as landraces, should instead be considered obsolete cultivars.

Therefore, genetic analysis, particularly GBS analysis, has proven to be extremely useful in assessing genetic relationships within the common bean germplasm. It is also a fundamental tool for valorizing landraces as entities linked to specific territories and their cultural traditions. In addition to genetic analysis, further information to discriminate landraces from cultivars could be found by searching for varietal registry entry data of the cultivars under consideration, particularly the date of entry. Data could be cross-referenced with the testimonies of farmers growing putative landrace cultivars and with data available in the literature.

A correct identification of landraces and the ability to distinguish them not only from modern varieties but also from obsolete cultivars are crucial for proper in situ and ex situ conservation, as well as for optimizing the resources available for conservation. In particular, these steps are fundamental for germplasm banks, which must cope with limited resources and long-term ex situ conservation. It is, therefore, essential to determine the priority ranking for each accession, with landraces being the top priority. These genetic resources, now cultivated by only a few farmers on a small scale and thus at risk of extinction or genetic erosion, represent a valuable repository of unique genotypes adapted to their environment, with high potential for agriculture in the face of ongoing climate change.