A total of 61 accessions, representing twenty-two traditional Italian potato varieties/ecotypes, have been collected at the Department of Agriculture, Food, Environment, and Forestry (DAGRI) at the University of Florence (Italy) since the late 1990s ( Table 1 , Figure 1 ). Tubers were characterized for pulp and skin color according to the cultivated potato descriptors (Huaman et al., 1977), with minor modifications. On the collection day, thanks to the interaction with the custodian farmers, information on the traditional culinary destination was gathered. These varieties underwent an in vitro introduction process, virus testing, and virus elimination, following the methodology previously described (Andrenelli et al., 2002, 2018; Natale et al., 2024), as detailed in the next paragraph. Additionally, six commercial cultivars (Alba, Blu di San Gallo, Desiree, Institute de Beauvais, Kennebec, and Magenta) were included in the analysis as reference standards, based on previously detected genetic identities with some local accessions using microsatellite (SSR) markers (Mandolino et al., 2015). Furthermore, sequencing data from approximately 106 genotypes from various regions worldwide were retrieved from the NCBI database (PRJNA723185 - Sood et al., 2023) for comparative analysis.

Collected tubers were carefully washed and allowed to sprout. Sprouts measuring 1–2 cm were harvested and disinfected by soaking in 70% ethanol for 30 seconds, followed by immersion in 2.5% sodium hypochlorite for 15 minutes. After multiple rinses with sterilized water, the sprouts were transferred to tubes containing Murashige and Skoog (MS) medium (Murashige and Skoog, 1962) supplemented with 25 g L^-1 sucrose and 2 g L^-1 Phytagel for growth and propagation ( Figure 1 ). All varieties underwent virus molecular detection testing (RT-PCR) for the main potato viruses (PVX, PVY, PVS, PLRV) according to Mortimer-Jones et al. (2009), with minor modifications. The analyses were carried out by the “Regional Phytosanitary Service” of the Tuscany region, in Italy, which exercises control and official supervision over the phytosanitary status of plants and their products during cultivation, storage, and marketing stages, to verify the possible presence of harmful organisms to plants. Those testing positive were subjected to a sanitization protocol. Two weeks-old in vitro plants were placed in a climate chamber for thermotherapy treatment, following a one-month program: 4 hours at 35°C (daylight) and 4 hours at 30°C (darkness - Griffiths et al., 1990). After this period, the apical meristem (0.1 to 0.4 mm in size) of each plant was excised and cultured on a modified MS medium containing 0.1 mg L^-1 gibberellic acid (GA₃) and 0.04 mg L^-1 kinetin to stimulate root development and growth. The regenerated plants were subsequently tested to confirm virus eradication. The complete process is illustrated in Figure 2 .

Genomic DNA was isolated from in vitro plantlets using a modified CTAB method (Doyle, 1991; Li et al., 2024). The purified DNA was resuspended in 100 µL of deionized water. To remove residual RNA, DNase-free RNase was added to the sample and incubated at 35°C for one hour. DNA concentration and purity were assessed using a Qubit™ 2.0 Fluorometer and by measuring absorbance ratios at 260/280 nm and 260/230 nm with a NanoDrop™ 2000 spectrophotometer. The integrity and size distribution of the genomic DNA were evaluated by agarose gel electrophoresis. Finally, samples were digested with MseI and EcoRI restriction enzymes to assess their compatibility with subsequent library construction and sequencing.

For genotyping-by-sequencing (GBS), libraries were prepared using a dual-enzyme strategy, following a modified version of the protocol by Poland et al., 2012. Genome complexity was reduced using the restriction enzymes MseI and EcoRI, and the resulting libraries were sequenced on an Illumina NovaSeq X system at the Biotechnology Center of the University of Wisconsin-Madison.

Sequencing data from both the Italian local varieties/ecotypes and the dataset from Sood et al., 2023 were quality-filtered using fastp (Chen et al., 2018) and aligned to the DM-13 Potato reference genome (v6.1; Pham et al., 2020) using the GATK pipeline (Poplin et al., 2018). After alignment, variants were filtered based on the following criteria: a minimum depth of 15 (min DP15), a minor allele frequency threshold of ≤0.5 (MAF), and a missing data rate of ≤0.05. Pruning was carried out with the SNPrelate R package (Zheng et al., 2012), using a linkage disequilibrium (LD) threshold of 0.1. The resulting set of filtered SNPs was then used for Neighbor-Joining phylogenetic analysis in MEGA11 software (Tamura et al., 2021) with the p-distance method.

Italy provides a favorable environment for the genetic diversification of potatoes, resulting in the selection of numerous local ecotypes adapted to the specific pedoclimatic conditions of its diverse regions (Perrino and Perrino, 2024). The geographic distribution analysis of these ecotypes reveals a predominance in the northern and central regions, with some varieties also present in southern Italy ( Figure 3 ). In the northwestern Alps, particularly in Valle d’Aosta and Piedmont, varieties such as Piatlina, Walser, Rosa, and Morella della Valsesia exhibit good adaptability to mountain environments. These varieties produce medium-small tubers, with skin colors ranging from yellow to red. In Veneto region, Mora della Reggenza and Cinquantina Vicentina are traditionally cultivated in hilly and pre-alpine environments, with tubers having either dark or yellow skin ( Table 2 ).

Along the Apennine ridge, Liguria region hosts ecotypes like Quarantina and Giana Riunda, while Tuscany showcases genetic diversity with varieties such as Rossa di Cetica, Castagno d’Andrea, and Estatina, possessing flesh colors that range from white to deep yellow. In the central-southern regions, Fiocco di Neve in Abruzzo and Viola Calabrese in Calabria feature skin colors ranging from pure white to deep purple.

Italian local potato ecotypes are distinguished not only by their geographic distribution but also by a range of agronomic and morphological characteristics. Many of these ecotypes have been selected over time for their resistance to harsh environmental conditions, such as poor soil, low temperatures, and seasonal drought. Mountain varieties tend to have shorter growth cycles and smaller tubers compared to those from temperate regions, which exhibit higher productivity and a greater variety of shapes and colors ( Table 2 ). The preservation of these ecotypes by local farmers, particularly in extreme marginal areas like the upper Valsesia valley in Piedmont, is crucial. Here, the Walser population, who migrated from Switzerland at the end of the 14th century, selected varieties like Formazza, Morella della Valsesia, and Verrayes for their adaptability to extreme pedoclimatic conditions. These varieties became integral to the seasonal diet of the Walser community, composed of small farmers and shepherds. Preserving such genetic material is vital for maintaining genetic diversity, ensuring the resilience of potato crops to environmental changes, and safeguarding Italy’s rich agricultural heritage.

Viral infections pose a significant challenge to the conservation and cultivation of local potato landraces, often resulting in reduced yields and, in severe cases, the loss of valuable genetic material. Aiming at assessing the phytosanitary state of the Italian landraces, the collected material was screened at molecular level, revealing a widespread presence of PVX, PLRV, PVY, and PVS, with some ecotypes exhibiting multiple infections ( Table 3 ).

Notably, Rossa di Cetica, Cannellina Nera, Giana Riunda, and Morella della Valsesia were among the most affected, carrying several viral strains simultaneously. This widespread viral incidence underscores the challenges associated with the long-term vegetative propagation of traditional varieties without systematic phytosanitary management. To restore the health status of infected accessions, a sanitation protocol combining thermotherapy and meristem culture was applied. Post-treatment diagnostic tests confirmed the complete elimination of viral infections in all ecotypes, demonstrating the effectiveness of this method. These findings align with previous studies on the effectiveness of meristem culture in the recovery of vegetatively propagated crops (Wang et al., 2018; Bhat and Rao, 2020; Alavijeh et al., 2025). The availability of virus-free planting material is crucial for both conservation and sustainable seed tuber production. Healthy accessions not only preserve genetic diversity but also improve agronomic performance, particularly in low-input and organic farming systems (Tegen, 2016). Moreover, the recovery and maintenance of these local varieties play a key role in safeguarding Italians potato genetic heritage, facilitating their reintroduction into traditional cultivation systems. This effort is essential for maintaining culinary traditions, regional identity, and biodiversity, while also supporting the adaptation of landraces to changing environmental conditions.

Overall, approximately 17 million reads (PE150) were generated, resulting in a total of about 56.5 gigabytes of raw data. The cleaned reads were aligned to the Solanum tuberosum reference genome (v6.1; Pham et al., 2020), leading to the identification of 8.3 million SNPs. After filtering, this number was reduced to 15.872.

The resulting NJ dendrogram ( Figure 4 ) provides a distinct molecular fingerprint for each analyzed local variety/ecotype, organizing them into three main clusters (A, B, and C), each further divided into smaller subclusters (three or four per cluster). Although selections/provenances within each variety fall within the same subcluster, they are generally distinguishable, highlighting varying levels of within-variety diversity. This diversity reflects both genetic variation and potential influences from local adaptations or microenvironmental conditions. The five commercial varieties, used as reference material, are distributed across five different subclusters (B1, B2, C1, C2, and C3), each exhibiting a unique molecular profile. This contrasts with the findings of Mandolino et al., 2015, who used 12 SSR markers for molecular characterization of a small subset of the DAGRI collection and reported perfect identity between the SSR profiles of some commercial varieties and certain local varieties included in this analysis. This discrepancy highlights the enhanced resolution of the NGS-based approach employed in our study.

Branch A includes Mora della Reggenza (Vellar’ and ‘Muraro’ selections) and Morella accessions clustered together in A₁, along with Turchesa and Viola Calabrese, suggesting a shared genetic background or conserved selection practices among these ecotypes. In contrast, the Mora della Reggenza ‘Rigoni’ and ‘Cantele’ selections are separated, clustering independently in A₄ and showing clear genetic differentiation from other ecotypes with similar phenological traits. This differentiation could reflect distinct historical selection pressures or genetic drift specific to its location of origin. A₂ comprises a group including Unica, Salecchio, and Walser accessions, along with the three Verrayes accessions, all characterized by a single molecular profile. This uniformity across multiple accessions may indicate a stable clonal lineage, possibly conserved due to its adaptability or favorable traits. A₃ contains Rosa and Occhi Rossi, which, though grouped together, remain well distinguished from each other, suggesting that these varieties, while similar, have retained distinct genetic identities.

Branch B’s sub-cluster B₁ confirms the previously detected similarity between Castagno d’Andrea and the commercial variety Kennebec, consistent with the findings of Mandolino et al. (2015). However, while the SSR analysis showed complete identity between these two varieties, the higher resolution of the NGS approach used here allows us to distinguish them. This finding emphasizes the sensitivity of NGS in revealing subtle genetic differences that SSR markers may overlook. Similarly, the relationship between the Quarantina ecotype and the commercial variety Institute de Beauvais, as well as with the Fiocco di Neve accessions in B₂, was also observed by Mandolino et al. (2015), who nonetheless reported complete identity between Quarantina and the French-origin commercial variety. The distinctions observed here reinforce the potential of high-resolution markers to refine our understanding of genetic relationships. Finally, Branch B includes subcluster B₃, where Rossa di Cetica and Giana Riunda selections are grouped together but remain distinguishable. This indicates that, while these selections may share some genetic similarities, they have diverged enough to maintain unique profiles, showcasing how local practices can introduce variability even within closely related selections.

In Branch C, the commercial varieties Magenta and Alba cluster in C₁ with Estatina accessions, each retaining distinct profiles. This demonstrates that even when commercial and local varieties cluster together, they maintain unique molecular signatures, likely reflecting the influence of their respective breeding histories. In sub-cluster C₂, the commercial variety Desiree clusters with Rossa di Sulcina and Cinquantina Vicentina, each maintaining distinctive molecular profiles. This contrasts with the findings of Mandolino et al. (2015), who reported complete identity among these varieties. Within this cluster, it is interesting to note that the Rossa di Sulcina accessions show significant internal variability, with two accessions exhibiting greater genetic distance from each other than from the Cinquantina Vicentina ecotype. This internal diversity could suggest either a broader genetic base within Rossa di Sulcina or the influence of localized environmental adaptation. Finally, the commercial variety Blu di San Gallo clusters with Piatlina, Cannellina Nera, and Formazza, with the Formazza accessions well-separated from the others in the final sub-cluster (C₃) of Figure 2 , underlining the diversity present even within this closely grouped set.

Notably, the resolution power of the NGS technique was sufficient to distinguish only a limited number of accessions sampled from the same provenance. For instance, this was the case for two accessions of Morella della Valsesia and Giana Riunda “Mele” selection, two of the three accessions of Rossa di Sulcina, and two of the four accessions of Formazza. Similarly, it could not distinguish between different provenances within the Quarantina ecotype, even though they are characterized by different skin colors. This phenotypic difference may be due to a point mutation not captured by our SNP marker dataset. The ability of the NGS technique to detect some but not all intraspecific variations highlight both its strengths and limitations in detecting subtle genetic differences within clonally propagated crops.

Building on the work of Sood et al. (2023), we examined the genetic relationships between Italian potato local varieties/ecotypes and genotypes from around the world. In their study, the neighbor-joining (NJ) method was used to categorize 222 accessions and varieties into three main clusters (A, B, and C), which were further divided into subclusters. Their results indicated that, although no strict geographic pattern was evident, Indian varieties grown in the subtropical plains tended to cluster with a subset of andigena accessions, as well as accessions from Australia, New Zealand, and Afghanistan. In contrast, the other two clusters comprised a mix of varieties from Europe, North America, and South America, suggesting some level of geographic association with notable overlaps.

In our research, we compared a collection of Italian potato varieties with approximately 106 accessions from various parts of the world, as included in Sood et al. (2023 - Supplementary Table 1 ). Raw sequence reads were retrieved from the NCBI database (PRJNA723185), cleaned and aligned to the reference genome (Pham et al., 2020). SNP calling and filtering were then performed, yielding a set of 10,197 high-quality SNPs for further analysis. By integrating these global accessions with our Italian dataset, we aimed to explore whether Italian varieties would exhibit similar clustering patterns or if they would stand out genetically from other regional groups.

The generated NJ dendrogram ( Figure 5 ) reveals a marked separation between the Italian collection and the global accessions, highlighting a unique genetic signature among the Italian varieties. Unlike the findings of Sood et al. (2023), where some regional clustering patterns were observed but not strictly defined, our analysis demonstrates a clearer phylogenetic distinction. Italian varieties consistently formed a separate group, suggesting unique genetic characteristics likely shaped by local breeding practices and adaptation to regional conditions. This pronounced clustering of Italian varieties contrasts with the more intermixed clusters of international accessions, where European, North American, and South American varieties often appeared within the same groups, as reported by Sood et al. (2023).

Overall, our expanded approach, which combines Italian and international accessions, underscores a distinct phylogenetic identity for the Italian varieties. This separation may reflect historical, ecological, and selective pressures unique to Italian agricultural systems, offering valuable insights into the genetic diversity and potential breeding value of these local varieties within a global context.