Graph-based pangenomes and pan-phenome provide a cornerstone for eggplant biology and breeding

Article https://doi.org/10.1038/s41467-025-64866-1 Graph-based pangenomes and pan-phenome provide a cornerstone for eggplant biology and breeding Luciana Gaccione 1 ,LauraToppino 2 , Marie Bolger 3 , Maximilian Schmidt 4 , Maria Rosaria Tassone 2 ,MariaSulli 5 , Emily Idahl 6 , David Alonso 7 , Giuseppe Aprea 5 , Paola Ferrante 5 , Véronique Lefebvre 8 , Hatice Filiz Boyaci 9 , Roland Schafleitner 10 , Stefano Gattolin 11 , Richard Finkers 12,13 , Matthijs Brouwer 12 ,ArnaudBovy 12 , Jaime Prohens 7 ,EzioPortis 1 , Sergio Lanteri 1 , Giuseppe Leonardo Rotino 2 , Giovanni Giuliano 5,14 , Björn Usadel 3,6 & Lorenzo Barchi 1 Eggplant (Solanum melongena L.) is a major Solanaceous crop of Asian origin, but genomic resources remain limited compared to related species. Here, a core collection of 368 accessions spanning global diversity of S. melongena and wild relatives is phenotyped for agronomic, disease resistance and fruit metabolomic traits and resequenced. Additionally, 40 chromosome-level assemblies of S. melongena, its progenitor S. insanum and the allied species S. incanum enable the construction of two graph-based pangenomes, capturing broad genetic variation. We demonstrate the power of these datasets by identifying major loci controlling prickliness and resistance to Fusarium oxysporum f. sp. melongenae, driven by SVs affecting the LONELY GUY 3 gene and a resistance gene cluster, respectively, as well as a mutation in a GDSL-like esterase/lipase gene altering the levels of dicaffeoyl-quinic acids. These findings provide a cornerstone for pangenome-assisted breeding, enabling detailed analyses of genetic diversity, domestication history, and trait evolution in eggplant. Eggplant (Solanum melongena L.) is a member of the Solanaceae family and a widely cultivated crop, with over 59 Mt produced globally1. China and India are the top global producers, while Egypt, Türkiye, and Italy lead the production in the Mediterranean region. Unlike most Solanaceous crops, which are native to the New World2–6, eggplant originates in Asia and belongs to the subgenus Leptostemonum, also known as the “spiny Solanum”group7, phylogenetically distant from other Solanum crops like tomato and potato. Solanum melongena, its direct wild progenitor S. insanum, and the sister closely related species S. incanum form a well-supported clade, providing a rich source of genetic diversity for research and breeding efforts7–9. Received: 31 May 2025 Accepted: 30 September 2025 Check for updates 1 DISAFA, Plant Genetics, University of Turin, Grugliasco, TO, Italy. 2 CREA, Research Centre for Genomics and Bioinformatics, Montanaso Lombardo, LO, Italy. 3 IBG-4,CEPLAS, Forschungzentrum Jülich, Jülich, Germany. 4 Plant Breeding Department, GeisenheimUniversity, Geisenheim, Germany. 5 ENEA, Casaccia Res Ctr, Roma, Italy. 6 Faculty for Mathematica and Natural Sciences, Institute for Biological Data Sciences, Heinrich Heine University, Düsseldorf, Germany. 7 Universitat Politècnica de València, Valencia, Spain. 8 INRAE, GAFL, Montfavet, France. 9 Faculty of Agriculture, University of Recep Tayyip Erdogan, Rize, Türkiye. 10 The World Vegetable Center, Tainan City, Taiwan. 11 National Research Council of Italy, Institute of Agricultural Biology and Biotechnology (IBBA-CNR), Milan, Italy. 12 Plant Breeding, Wageningen University & Research WUR, Wageningen, The Netherlands. 13 Present address: GenNovation B.V, Wageningen, The Netherlands. 14 Present address: Pandora Research Srls, Rome, Italy. e-mail: gi[email protected];[email protected]; [email protected] Nature Communications | (2025) 16:9919 1 1234567890():,; 1234567890():,; Our previous studies on 3400 eggplant accessions provided a snapshot of the worldwide genetic diversity of this species, and supported the hypothesis of independent domestication events in two distinct centers: Southeast Asia and the Indian subcontinent, where both anthropic and environmental selection acted on different genomic regions8,10.Afirst pangenome, constructed using short-read resequencing of 23 representative eggplant accessions, and one accession each of S. insanum and S. incanum, added 51.5 Mb and 816 genes to the reference genome; this study also identified 53 selective sweeps related to key domestication traits such as fruit color, prickliness, and fruit shape11. The advent of graph-based pangenomes and super-pangenomes, representing single or multiple species, respectively12,13, has provided deeper insights into the evolutionary history, domestication, and genetic relationships within and between species/taxa. These reference-unbiased frameworks enable more accurate comparisons by identifying core, dispensable, and private genomes, as well as a better understanding of the structural variation at nucleotide-level resolution14. In this study, we describe the construction of a core collection (CC) of 368 accessions, including 321 accessions that capture the worldwide genetic and phenotypic diversity of cultivated eggplant and 47 that represent its wild relatives. The full collection is phenotyped in multiple locations and resequenced using short reads. Longread sequencing is performed on 33 S. melongena, three S. insanum, and four S. incanum accessions (Fig. 1a) to generate chromosome level-assemblies, which are integrated into a reference-unbiased graph-based pangenome. A second pangenome graph, limited to the 33 S. melongena accessions, is used for pangenome-wide association (Pan-GWA) analysis to identify genes/quantitative trait loci (QTLs) associated with key agronomic and metabolic traits, as well as in responses to biotic and abiotic stresses. We provide examples how these pangenomes and pan-phenome can be used to explore eggplant evolution and demography, and to identify the genomic basis of important traits, such as prickle development, resistance to Fusarium wilt and variation in fruit chlorogenic/isochlorogenic acids content. Results Construction of a worldwide eggplant core collection CCs are essential for simplifying phenotyping and phenotypegenotype association studies, by reducing redundancy and maximizing diversity, thus identifying selection targets and enhancing the effectiveness of breeding efforts15. We have previously described and genotyped a worldwide collection of 3412 worldwide accessions from 105 countries8,16. Using the genotyping data, we established a CC of 368 accessions (see “Methods”; Supplementary Fig. 1), of which 321 represent the worldwide genetic diversity of cultivated eggplant, and 47 are wild relatives from the primary, secondary, and tertiary gene pools (Supplementary Data 1). The latter comprised 13 accessions of S. insanum (the direct progenitor of cultivated eggplant) and 7 of the sister species S. incanum. Genome sequencing and chromosome-scale genome assemblies The 368 accessions were sequenced using short reads (2 × 150 bases), providing an average coverage of 20× per accession (Supplementary Data 1). Additionally, 40 accessions (33 of S. melongena, including GPE001970 (67/3), previously used to generate the v4.1 of the reference genome11,threeofS. insanum and four of S. incanum) representing the genetic diversity of the species (Fig. 1b) were sequenced using Oxford Nanopore Technology (ONT), generating almost 66 Gb of data per individual, with an overall read N50 of 32.2 kb (Supplementary Data 2). The genomes of each of the 40 accessions were assembled and polished, yielding an average genome size of 1.13 Gb, and contig N50 up to 83 Mb (Supplementary Data 3 and Fig. 1c). To further enhance assembly quality, six S. melongena assemblies (including GPE001970) and two each of S. insanum and S. incanum were subjected to high-throughput chromosome conformation capture technique (Hi-C) scaffolding (Supplementary Data 4 and Supplementary Fig. 2). The remaining 30 assemblies were scaffolded using as a guide an accession from the same species (see “Methods”). After the scaffolding steps, on average 99.54% of the contig sequences were anchored on the 12 eggplant chromosomes. The average genome sizes were 1.106 Gb (S. incanum), 1.115 Gb (S. insanum), and 1.137 Gb (S. melongena) and the average scaffold N50 were 94.67 Mb (S. incanum), 90.76 Mb (S. insanum), and 93.43 Mb (S. melongena) (Supplementary Data 5). The quality and completeness of the reference genomes was assessed using several parameters: Benchmarking Universal SingleCopy Orthologs (BUSCO) scores17 ranged between 98.3 and 98.7% (98.7% for version 5 of the reference GPE001970 genome vs 96.9% in the previously published v4.1)11. All 40 assemblies reached the “reference”level (LAI > 10) of the LTR Assembly Index (LAI)18, with 16 reaching a “gold standard”level (LAI > 17), while the average QV score19 was 49.26 (less than 1 error in 63,131 bases, Supplementary Data 5). Repetitive elements covered 72.8% (GPE013320; S. insanum)to 75.3% (GPE008640; S. incanum) of the total genome sequences, with long terminal repeat retrotransposons (LTR-RTs) being the predominant class (Supplementary Data 6 and Fig. 1d). Solanum insanum and S. incanum exhibited a higher percentage of LTR-RTs than S. melongena, compensated by a lower percentage of DNA transposons (Supplementary Data 6 and Fig. 1d). Gene prediction was conducted by combining Helixer and BRAKER3 (see “Methods”). The number of protein-coding gene models per genome ranged from 30,886 to 33,449, with an annotation completeness using BUSCO ranging between 95.3% and 97.2% (Supplementary Data 5). Phylogeny and synteny analyses To clarify the phylogenetic relationships, we identified 69 single-copy and 9323 low-copy gene orthogroups, respectively, across all 40 accessions and 6 other plant species (see “Methods”). The two approaches produced similar phylogenetic tree structures (Supplementary Fig. 3a, b), with S. melongena accessions clustering together, and S. insanum and S. incanum positioned as a separate group. Minor topological differences were observed in both the single-copy and lowcopy trees; in the former, a S. incanum accession (GPE008640) was positioned in the S. melongena group, reflecting an ambiguous genetic background for this accession. Consequently, we excluded S. incanum GPE008640 from the species-level comparison, adopting the resulting species trees for the subsequent analyses (Supplementary Fig. 4a, b). In both trees, S. insanum and S. incanum accessions appeared intermingled. These patterns likely reflect a combination of evolutionary processes, notably hybridization20, introgression, and incomplete lineage sorting, leading to shared genetic variation and blurred species boundaries in the reconstructed trees. The estimated divergencetimes obtained using the single-copy tree, revealed that S. incanum diverged from the common ancestor of S. melongena and S. insanum approximately 0.75 million years ago (Mya), at the end of the Mid-Pleistocene climatic transition21, while the split between S. insanum (the direct progenitor) and S. melongena dated around 0.29 Mya (Pleistocene) (Fig. 2a). Theanalysisofgenefamilyevolutionhighlightedhowdomestication and ecological adaptation shaped distinct gene family trajectories across the “Eggplant”clade (Fig. 2b, Supplementary Fig. 5a, b, and Supplementary Data 7). In S. melongena,expansionsweremainly associated with stress adaptation (gene silencing, desiccation tolerance, cuticular wax biosynthesis), while contractions affected sugar transport and circadian metabolism. Solanum insanum showed expansion in detoxification and terpenoid biosynthesis, with Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 2 Fig. 1 | 40 eggplant reference genomes. a Fruit phenotypes of the 40 chromosome-scale accessions. In clockwise direction: fruits of the four S. incanum, the three S. insanum, and the 33 S. melongena accessions. Solanum incanum and S. insanum pictures are not in scale. bPrincipal component analysis (PCA) of component 1 (EV1) vs component 2 (EV2) for the 368 eggplant accessions in the core collection; the 40 accessions with reference genomes are represented by solid dots; dots are colored by species. cAssessment of genome contiguity by contig N50. The Nx value represents x% of the total contigs length that is covered by the shortest contig length. dFractions of transposable elements, low-complexity (simple repeats), and non-repetitive sequences in the 40 reference genomes. Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 3 contraction of growth-related pathways. In S. incanum,genefamily expansions predominantly involved growth and reproductive programs, whereas contractions were enriched in ethylene metabolism and host–pathogen responses. These patterns suggest that domestication reshaped gene family content towards stress adaptation in S. melongena, metabolic plasticity in S. insanum, and structural investment in S. incanum. Synteny analysis of the 40 eggplant genome assemblies revealed strong overall collinearity, with limited structural variations such as inversions on chromosomes 1 and 10, and a few translocations (e.g., between chromosomes 7 and 9) within S. melongena (Fig. 3aand Supplementary Data 8). More rearrangements were observed in S. insanum and S. incanum, including a large translocation between chromosomes 1 and 5 in GPE008290 (S. incanum). In addition, expansion or contraction of syntenic regions is evident in some accessions, suggesting segmental duplications or deletions. Paranome (i.e., paralogous within a genome) analysis identified 4456, 4757, and 4818 tandem duplications in S. melongena,S. incanum, and S. insanum, respectively (Fig. 3b). Tandem duplicates were consistently enriched for genes involved in secondary metabolite biosynthesis, with lineage-specific differences observed: lipid localization and transport terms were reduced in S. incanum and S. insanum,while phenylpropanoid metabolism was enriched (Supplementary Data 9). Conversely, terpene-related pathways were enriched in S. melongena. Segmental duplications were more abundant (7925, 7398, and 7810 for S. melongena,S. insanum, and S. incanum) and enriched across all three species for core functions such as cell growth, transcriptional activation, and translation. On the other hand, developmental processes were particularly enriched in S. incanum and S. insanum, while in S. melongena, there was a peculiar enrichment for energy metabolism and abscisic acid response (Supplementary Data 9). A graph-based pangenome of the eggplant clade To access the genetic diversity of S. melongena and its allied species, and avoid any bias towards a specific reference, a reference-free pangenome graph (“PG-SMA”) was constructed with PGGB14 using the 40 chromosome-level assemblies (Fig. 4a). PG-SMA exhibited a total length of 2.94 Gb, i.e., over twofold that of our previously published reference-based pangenome (1.21 Gb)11. It exhibited 141.8 M nodes, 194.7 M edges, an average node degree of 2.75 (Supplementary Data 10) and a total of 83,147 structural variants (SVs; Fig. 4a, b). The Jaccard similarities of the embedded paths in PG-SMA graph structure were used to investigate the phylogenetic relationship among the 40 accessions. In agreement with the results from both single-copy and low-copy species trees, S. incanum and S. insanum accessions appeared intermingled rather than forming clearly separated groups (Supplementary Fig. 6). The PG-SMA variation graph revealed a predominance of soft-core nodes, highlighting a substantial dispensable genome content (~40%) as well as core genomic sequences (~50%, Supplementary Fig. 7). The PG-SMA growth curves modeled on protein families indicated a closed pangenome with a γ-value of 0, reaching essentially a plateau at 40 genomes (Fig. 4c). In contrast, the same curves modeled on genome sequences had a γ-value of 0.16, suggesting that additional genomic sequences remain to be characterized (Supplementary Fig. 8). Population structure of cultivated eggplant and its allied species Mapping of short reads from the resequenced CC accessions on the PG-SMA graph yielded 34,576,891 polymorphic sites. After quality filtering, 12,893,056 SNPs and 12,597 SVs were retained for phylogenetic and population structure analyses. The maximum likelihood (ML) tree obtained using the SNP markers identified two main branches, including: (i) S. melongena accessions; (ii) S. insanum,S. incanum, and all the other species (Fig. 5a). Five S. insanum accessions grouped within S. melongena, likely due to feral escapes and/or admixture events22. Within the broader group, three sub-clusters were identified: (i) the “Eggplant”clade, which includes S. insanum,S. incanum,S. linnaeanum,andS. campylacanthum Hochst. ex A. Rich.; (ii) the “Anguivi”grade, comprising S. aethiopicum and its wild progenitor S. anguivi,S. macrocarpon,alongwithS. humile Lam. Fig. 2 | Divergence times of eggplant and related species and gene family evolution. a Time-calibrated single-copy species tree illustrating the estimated divergence times among S. melongena,S. insanum,S. incanum, and six other plant species. Calibration points are marked as black dots. Gene family expansions and contractions are indicated by magenta and blue values, respectively, along the branches. bGO enrichment analysis of the biological processes for expanded and contracted gene families in S. melongena,S. insanum,andS. incanum. The enriched GO categories were determined using the hypergeometric test, followed by the Benjamin–Hochberg correction to obtain adjusted Pvalues for multiple testing. Circle size represents the statistical significance (–log₁₀ p-value), magenta color represents gene family expansions and blue represents contractions. Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 4 and S. tomentosum L.; and (iii) a cluster consisting of East African (S. tettense Klotzsch and S. schimperianum Hochst. ex A. Rich), Madagascar (S. myoxotrichum Baker, S. pyracanthos Lam.), and “Lasiocarpa” clade species (S. sisymbriifolium and S. torvum). These results were consistent with previous reports8,16,23,24, and confirmed coherent phylogeny obtained using different genetic markers (SNPs identified with SPET and GBS methods) for S. melongena and its wild relatives. In a PCA based on SNPs (Fig. 5b, d), the first two components explained 28.66% and 6.05% of the total genetic variance, respectively. Solanum melongena accessions formed a compact cluster, while its wild progenitor, S. insanum, showed a broader dispersion with several accessions clustering near or within S. melongena,reflecting their genetic affinity. Solanum incanum partially overlapped with other wild species, such as S. linnaeanum and S. campylacanthum, while accessions of the “Anguivi”grade clustered closely together. Tertiary genepool species like S. torvum and S. sisymbriifolium were distinctly divergent. A SV-based PCA (Fig. 5c, e) showed a different structure, despite a lower explained variance (EV1: 8.89%, EV2: 6.35%). Solanum melongena accessions dispersed broadly along both components, whereas the “Eggplant”clade and “Anguivi”grade species clustered distinctly, with limited overlap. Solanum insanum occupied an intermediate position, consistent with its progenitor status, with some accessions again mapping near or within S. melongena,whileS. incanum formed a separate cluster. The differences of clustering in the two PCAs is expected, given the different nature (SNPs vs SVs) and numbers (23,600 vs 151 after filtering) of the variants used to produce them. Geographical analysis (Fig. 5d, e) revealed two major clusters corresponding to accessions from Southeast Asia and the Indian subcontinent, both closely related to S. insanum,confirming the identification of these regions as primary domestication centers for S. melongena8. A model-based clustering analysis using the pruned SNPs from PGSMA, with membership proportions exceeding 0.70, was used to infer the genetic ancestry of S. melongena groups originating from diverse geographical regions and their allied species, setting the optimal number of ancestral populations (K) to 6 (Fig. 5f and Supplementary Fig. 9a, b). Solanum incanum and S. insanum exhibited minimal admixture, highlighting their distinct genetic identities (Fig. 5fand Supplementary Fig. 9c). The latter showed an additional genetic substructure based on its geographical origin: Solanum insanum from Southeast Asia and India are clearly separated, while those from Africa (Madagascar) are an admixture of the two (Fig. 5f), possibly resulting from past hybridization in Madagascar of S. insanum introductions from different Asian provenances25. Southeast Asian S. insanum genetic signatures remain detectable in local S. melongena germplasm, whereas Indian eggplants largely lack genomic traces of Indian S. insanum,potentiallyreflecting post-domestication selection linked to regional phenotypic preferences10. These patterns align with PCA results and further support separate domestication events in Southeast Asia and the Indian subcontinent8. Chinese and Japanese accessions predominantly reflect Southeast Asian ancestry, while accessions from the Middle East and Africa display signals from both domestication centers (Fig. 5f), also observed in European (EEU, SEU, NCU) and American (NAM, MAC) accessions. In contrast, wild species retained mainly unique, divergent components, highlighting their value for conservation and future eggplant breeding efforts. Population dynamics and inference of demographic history To investigate the demographic history and population structure of S. melongena, we integrated multiple population genetic approaches using both SNP and SV data. Pairwise fixation index (F ST ), a metric for evaluating genetic differentiation between populations and lineagespecific genetic drift, revealed the highest genetic differentiation between S. melongena and its wild relative S. incanum, followed by S. insanum (Fig. 6a and Supplementary Fig. 10a). Among S. melongena groups, accessions from Southeast Asia, India, and Japan showed the lowest F ST values relative to S. insanum, indicating a close genetic affinity consistent with recent domestication events8. Outgroup f3statistics, using S. insanum as the outgroup, supported Southeast Asia as the primary domestication center, with this group showing the lowest shared drift with other S. melongena populations (Fig. 6b and Supplementary Fig. 10b). D-statistics further confirmed Fig. 3 | Synteny and gene duplications across the 40 eggplant reference genomes. a Synteny map of the 40 chromosome-scale eggplant accessions. On the left is shown the species tree of the 40 eggplant accessions (S. melongena in purple, S. insanum in pink, and S. incanum in brown) constructed based on low-copy (1 to 5 genes) orthogroups. On the right is reported the synteny map, arranged according to the species tree, based on genome similarities across the eggplant accession panel. Synteny blocks are represented by color-coded bands that connect various genomes, with each color corresponding to a distinct chromosome according to the first accession reported at the top of the species tree. The scale bar represents 100 Mb. bCounts of duplicated genes across the 40 eggplant accessions, classified into dispersed duplications (DD), proximal (PD), tandem (TD), and segmental (SD). Bar order corresponds to the species tree shown in (a). Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 5 asymmetric gene flow from S. insanum into Southeast Asian, Indian, and Japanese accessions, while European varieties showed limited introgression (Fig. 6c and Supplementary Fig. 10c). The domestication and migration history were also explored using TreeMix26 and the OptM package27,withS. incanum as outgroup and grouping accessions according to their geographical origin. OptM determined that the most plausible scenario included two migrations (Fig. 6d). The first involved gene flow from S. insanum into Southeast Asian S. melongena9. The second suggested back-introgression from Japanese/Korean accessions into S. insanum, indicating bidirectional gene flow across generations. Residual matrices (Supplementary Fig. 11) suggested an additional admixture event between Chinese and European accessions. Most observations confirm our previous study conducted on a much larger set of accessions using SPET genotyping, with the exception of the back-introgression of Japanese/Korean accessions into S. insanum, which was discovered due to the high resolving power of whole genome resequencing, and is consistent with the S. melongena-S. insanum admixture observed in other geographical areas. Fig. 4 | Graph-based pangenome of S. melongena and its allied species. aNumber of structural variants (SVs) in each accession of the PG-SMA graph. bLength distribution of SVs in PG-SMA graph. cAccumulation curves showing pan and core gene families as a function of the number of genomes incorporated into the PG-SMA graph. Dots represent permutations of genome addition order; solid lines indicate model fits. Source data are provided as a Source Data file. Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 6 Demographic modeling suggests that a reduction in effective population size (N e ) for both S. insanum and S. incanum already occurred ~40–45 thousands years ago (kya), during a glacial maximum, while a more recent bottleneck (~ 15 kya) was observed in S. melongena accessions from Southeast Asia and India (Fig. 6e), in agreement with previous studies8,25 and compatible with the timing suggested for the domestication of eggplant. A pan-phenome of S. melongena The CC was cultivated in a randomized block design in three different locations (Antalya, Türkiye; Montanaso Lombardo, Italy; Valencia, Spain, Supplementary Fig. 12) and phenotyped for 46 agronomic, 10 biotic/abiotic stress, and 162 fruit metabolomic (80 peel and 81 flesh) traits. In the present study, we focused on a selected set of representative traits from the broader phenotyping dataset, including fruit calyx and leaf prickliness as an example of agronomic variation, resistance to Fusarium wilt as a key biotic stress trait, and the contents of chlorogenic and isochlorogenic acids in fruit peel and flesh as metabolic features. To facilitate alignment of the reads of the 321 S. melongena accessions, polymorphism identification and Pan-GWA on this species, a second pangenome graph (“PG-SM”) was built with MinigraphCactus28,usingonlythe33S. melongena accessions, with the GPE001970 genome sequence as the reference. The metrics of this pangenome graph are reported in Supplementary Data 10. Both graphs were used to carry out Pan-GWA studies using selected traits (see “Methods”; Supplementary Data 11–16). Eggplant Pan-GWA analysis clarifies the genetic mechanisms regulating prickliness Eggplant is unique among Solanaceous crops, showing prickles on leaves, stems, and fruit calyxes, that provide adaptive advantages, including herbivore deterrence and water retention, but also complicate harvesting, storage, and transport of fruits. For this reason, prickles have been selected against in modern varieties, although some regions, close to the domestication centers, still prefer prickly varieties for their better perceived organoleptic qualities29.UsingSNPand short indel-based Pan-GWA analyses, we confirmed the LONELY GUY (LOG) 3 gene (SMEL5_06g025740) on chromosome 6 as the strongest candidate gene associated with prickle formation on both leaves and calyxes (Fig. 7a, Supplementary Figs. 13a and 14a, and Supplementary Data 11 and 12). Recently, a splice-site mutation in LOG3 (A “prickleless”>G“prickly”)wasidentified as a key variant causing the prickleless phenotype30. However, in our genome this mutation is located in the second intron at 90,042,124 bp (Fig. 7b). Our Pan-GWA analysis reported an identical SNP (A/G) at a slightly different position 90,042,191 bp, annotated as a splice-site variant at the end of exon 3. Examination of the PG-SMA graphshowedthatinallspeciesexamined, Fig. 5 | Phylogeny and population structure of S. melongena and wild relatives. aMaximum likelihood (ML) tree using the SNP dataset from PG-SMA graph. The colors in the outer circle correspond to species, in the inner circle to geographical origin. Branch support is indicated by bootstrap values, ranging from electric blue (43) to magenta (100). b,dPrincipal component analysis (PCA) using SNPs colored according to species (b) and origin (d). c,ePCA obtained using SVs colored according to species (c) and origin (e). fsmnf clustering analysis using PG-SMA graph SNPs at K= 6. Accessions are grouped according to geographical origin (INC S. incanum;INSS. insanum from: SEA Southeast Asia, IND India, AFR Africa; S. melongena from: MES Middle East, SEA Southeast Asia, SEU South Europe, UNK Not available, CHN/JPN China and Japan, IND India, AFR Africa, NCE North and Central Europe, NAM Americas and Oceania, EEU East Europe). White dashedlines separate the groups according to geographical origin. The full snmf analyses, with K=2–20, are shown in Supplementary Fig. 7. Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 7 the “A”allele is responsible for the prickly phenotype, while prickleless accessions harbor the “G”allele (Fig. 7b). Indeed, the splice-site mutation leads to the loss of exon 3 (30 bp), resulting in a 10 amino acid deletion, which however occurs only in a subset of accessions carrying the “G”allele, indicating possible context-specificalternative splicing or other genetic mechanisms, in agreement with previous observations30. Accessions carrying the “G”allele showed strong reduction of LOG3 gene expression (Supplementary Fig. 15a). We found no evidence for the alternative splicing observed by Satterlee et al.30, possibly due to different sensitivity of the methods used (RNASeq vs RT-PCR) or genotypeor tissue-specific differences. Interestingly, heterozygous accessions at this position exhibit enhanced prickliness, suggesting hyperdominance of the prickly “A”allele. We also confirmed that in accession GPE003520 the prickleless phenotype is associated to a 226 bp deletion that disrupts exon 6 of LOG3, although it carries the “A”allele (Fig. 7b). The presence of S. incanum accessions in the PG-SMA pangenome helped decipher additional mechanisms involving LOG3 and controlling prickliness. As expected, the S. incanum accession (GPE008290) exhibiting a highly prickly phenotype (Fig. 7c) contains a functional LOG3 gene (Fig. 7d). On the other hand, accession GPE008350 (S. incanum) is prickleless (Fig. 7c) in spite of the fact that it carries the wild-type “A”allele. Inspection of the pangenome revealed a translocation of the region containing LOG3 (about 22 Mb) from one arm of chromosome 6 to the heterochromatic, pericentromeric region of chromosome 1 (Fig. 7d), which is probably causing the prickleless phenotypes. To validate this translocation, we aligned ONT reads from GPE008350 (carrying the putative rearrangement) to the HiC–scaffolded S. incanum GPE008290 and GPE016100 genomes. The analysis confirmed the presence of the translocation, observable by regions of zero coverage coinciding with the translocation breakpoints on chromosomes 1 and 6 (Supplementary Fig. 16). The molecular basis of Fusarium resistance in eggplant We analyzed the resistance to Fusarium oxysporum f. sp. melongenae (Fom;Fig.8a, b), a pathogen of significant economic importance in eggplant cultivation. Two strong associations on chromosomes 2 and 11 were identified, co-localizing with previously identified regions, conferring partial resistance to Fom31,32 (Fig. 8c, d, Supplementary Figs. 13b and 14b, and Supplementary Data 13 and 14). The QTL on chromosome 11 contained a cluster of RPP13-like disease resistance genes, known for their role in plant immune responses31–35.Onchromosome 2, a QTL was identified at 44.7 Mb, in correspondence with a 260 bp insertion (Fig. 8c, d, f, Supplementary Figs. 13b and 14b, and Supplementary Data 13 and 14). This genomic region contains an AThook motif nuclear-localized (AHL) protein 10-like gene (SMEL5_02g002450) located at 24.78 kb upstream of the insertion and involved in plant stress resistance. In tomato, overexpression of CaATL1 (Capsicum annuum AHL gene) significantly enhanced resistance to oomycete and bacterial pathogens36,37. Additional minor QTLs were identified in single-environment analyses on chromosomes 6, 7, 9, and 11 (Supplementary Data 13 and 14, and Supplementary Figs. 13b and 14b). The pangenome graph revealed an insertion in the first exon of SMEL5_11g023850,encodingaRPP13-like gene, present exclusively in Fom-susceptible accessions and absent in resistant accessions such as the reference line “67/3”(Fig. 8e). An additional deletion of about 31 kb was identified in the region, affecting a gene (SMEL5_11g023860) encoding a ginkbilobin-like antifungal protein and a second disease resistance proteinRPP13-like gene (SMEL5_11g023870)(Fig.8e). This Fig. 6 | Eggplant domestication from Solanum insanum based on PG-SMA SNPs. aF ST values of S. melongena from different geographical areas, S. insanum and S. incanum.bOutgroup f3-statistic for all possible admixture populations using S. insanum as outgroup. The higher the value, the more recently the two populations diverged. The error bars indicate ±1 standard error (SE). A total of 1,249,951 SNPs were used to compute the f3-statistic. cD-statistic to detect admixture according to W,X;Y,Z model. World regions of origin are coded as: MES Middle East, SEA Southeast Asia, SEU South Europe, CHN China, IND India, NAM Americas and Oceania, JPN Japan, AFR Africa, NCE North and Central Europe, EEU East Europe. The error bars indicate ±1 standard error (SE). A total of 1,249,951SNPs were usedto compute the D-statistic. dMaximum likelihood (ML) tree with bootstrap support values with migration events from TreeMix, indicated by colored arrows. The color scale shows the migration weight. The scale bar shows 10 times the average standard error of the estimated entries in the sample covariance matrix. eSMC++ plot showing changes in effective population size (N e )ofS. melongena accessions grouped according to their origin (SEA: Southeast Asia, IND: India) through relative time. Gray bar delimits the time of N e decrease (8–15 kya). Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 8 deletion showed a peculiar pattern, since it contains a relic fragment of about 2400 bp corresponding to a portion of the second exon of SMEL5_11g023870 (Supplementary Fig. 15b, c). Transcriptomic evidence highlighted altered expression of SMEL5_11g023870 in the accessions carrying the 31 kb deletion (Supplementary Fig. 15b). The presence of this deletion is strongly (but not exclusively) associated withthesusceptiblestateinthethreespeciesincludedinthegraph. This was also confirmed by the PanSel analysis, which identified a divergent bin subjected to selective pressure (i.e., selective sweep) at this genomic position (Supplementary Data 17). Further analysis of the PG-SMA graph revealed that the deletion on chromosome 11 is present in 5 out of the 7 S. insanum and S. incanum accessions used for graph construction, suggesting that this polymorphism is ancient (Fig. 8f). The chromosome 2 insertion is consistently present in our PG-SMA S. incanum and S. insanum accessions, with the exclusion of GPE008370 and GPE008470 (Fig. 8f). The former lacked both QTLs and was thus susceptible to the pathogen, while the latter, although lacking the insertion in chromosome 2, showed the presence of the RPP13-like gene on chromosome 11 and was fully resistant to Fusarium (Fig. 8e, f). The molecular basis of chlorogenic acid metabolism in eggplant Chlorogenic acid (caffeoyl-quinic acid; CQA) and its derivatives isochlorogenic acid A and B (3,5and 4,5-DiCQA, respectively), are abundant phenolic compounds present in eggplant fruits, responsible both for their antioxidant properties and for their browning through oxidation by polyphenol oxidases. CQA is the major compound, accounting for more than 90% of these compounds38.Pan-GWAanalyses for CQA highlighted the presence of eight weak QTLs, of which the strongest, on chr 6, (p-value 2.34 × 10⁻11; PVE 45%; Supplementary Data 15 and 16), did not contain any evident candidate genes. We acknowledge that environmental conditions (temperature, light exposure, and water availability) in the two cultivation locations may have influenced the biosynthesis of chlorogenic acids, as previously reported39,40. In contrast, we identified a major QTL for isochlorogenic acids content in both fruit peel and flesh on chromosome 4 (4.88–5.55 Mb), with highly significant marker-trait associations (pvalues from 1.2 × 10⁻¹⁹to 1.15 × 10⁻⁷⁶; PVE up to 74%; Fig. 9a, b, Supplementary Figs. 13c, d, and 14c, d, Supplementary Data 15 and 16). Within this region, three GDSL lipase-like genes were Fig. 7 | Mutations influencing leaf and calyx prickliness in eggplant. a Multienvironment Manhattan plot from the PG-SM-based Pan-GWA analyses for leaf and calyx prickliness, together with the corresponding box plot of the number of accessions carrying distinct alleles for the splice-site mutation SNP in LOG3 (GG, accessions with homozygous reference type of allele; AA, accessions possessing homozygous alternative allele; GA, accessions possessing both alleles). In Manhattan plots, the colors refer to SNP (magenta), SV (dark blue), short indel (light blue), while the shape refers to leaf prickles (lpr: solid dot) and fruit calyx prickles (fcp: triangle). Genome-wide thresholds are marked using dashed (minimum) and continuous (maximum) lines, and estimated based on q-values. For box plots, the central mark indicates the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. Data beyond the end of the whiskers are displayed as black dots. bSequence Tube Map representation of the paths embedded in the PG-SMA graph highlighting the splice-site mutation SNP in SMEL5_06g025740 (LOG3) in and around exon 3 (top), as well as the GPE003520 deletion in exon 6 (bottom). Solanum incanum and S. insanum accessions are represented as brown and pink paths, respectively, while S. melongena accessions are colored based on the allele at the locus (purple–prickly; green–prickleless). Gene structure created in BioRender. Gaccione, L. (2025) https://BioRender.com/ pf32ps3.cPhenotypes of the prickleless GPE008350 (left) and prickly GPE008290 (right) S. incanum accessions. dSynteny map of the translocations of LOG3 in S. incanum (GPE008350) with respect to S. melongena (GPE001970) and S. incanum GPE008290. Synteny blocks are represented by color-coded bands that connect various genomes, with each color corresponding to a distinct chromosome. The scale bar represents 100 Mb. Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 9 Pan-GWA studies Adjusted means for each accession and for each resistance score were estimated using a best linear unbiased prediction (BLUP) model with the “inti”R package (v.0.6.6)127 and were used as the phenotypic input for the GWAS. The BLUPs were calculated for single environments (correcting for block and genotype–block interaction effects when the number of measurements was sufficient) and across environments/ trials (correcting for block and environment–block interaction effects). The broad-sense heritability was computed using the “inti”R package (Supplementary Data 24). For the GWA studies, we used SNPs, short indels, and SVs derived from the genotyping of the eggplant CC based on the PG-SM pangenome graph. The whole SNPs set was filtered using bcftools (v1.21)112 to retain bi-allelic SNPs with a MAF > 0.05, a mean read depth > 15, < 10% missing data, and a heterozygosity rate <20% of the accessions. Accessions with a heterozygosity level > 5% and/or a missing rate > 20% were removed. Comparable filtering criteria were applied to the SVs, selecting variants that were longer than 50 bp. SnpEff (v5.0)128 was used to annotate and predict the effects of SNPs and SVs. SNPs and SVs were named according to their assigned chromosomes in the variation graph, followed by the base pair start position relative to the reference path in the PG-SM pangenome. After excluding accessions that exhibited large phenotypic variation within the same accession, as well as those with >20% missing data or a heterozygosity rate >20%, we selected 308 accessions forSNP and 311 for short indels and SV-based Pan-GWA studies, which were carried out using 2,133,370 high-quality SNPs,186,343 short indels and 4480 SVs. GWA studies on single-environment and multi-environment BLUPs were carried out using the R package GAPIT-3 (v3)129 and the Bayesian-information and Linkage-disequilibrium Iteratively Nested Keyway (BLINK) model130. To reduce the number of false positives caused by population stratification, up to three principal components (PCs)wereaddedascovariatevariablestotheBLINKmodelforeach trait. The significance threshold for each trait was determined by estimating the FDR with the qvalue (v1.2) R package131. CMplot (v4.2) was used to produce rectangular Manhattan plots and QQ-plots132.The QTL confidence intervals were defined according to the LD decay distances estimated for each chromosome using SNPs from the PG-SM pangenome, extending the region around the significant marker in the upstream and downstream directions. Each QTL name was defined as the trait code, the chromosome number, and the QTL position. The gene models of the S. melongena genome of the reference line “67/3” (GPE001970) was then used to retrieve annotations for genes within the mostpromising QTL regions. Genes found within each QTL interval were investigated in literature to identify putative causal genes. To visualize specific regions of the pangenome graphs corresponding to variants involved in the resistance to fungal wilt, Sequence Tube Map (v0.8.1)133 and odgi viz command were used. Conserved and divergent regions in the PG-SM pangenome were identified using PanSel134.Afixed sliding window of 100 kb was chosen to estimate the average number of edit distance between each haplotype per bin by computing the number of paths that are found in each window between two anchor nodes. Bins with low numbers of paths were designated as under selection genomic regions. The divergent regions in the PG-SM graphwerethenintersectedwiththe whole set of QTLs identified by the Pan-GWA analyses using bedtools intersect (v.2.31)135. Transcriptomic data for GPE001970 and GPE001490 were used to confirm the presence and potential transcriptional impact of mutations identified through Pan-GWA analyses. Total RNA was extracted from leaves of the selected accessions using the Qiagen RNeasy Plant Kit and sequenced on the DNBseq BGI platform at BGI group. Raw RNA-seq reads were trimmed using fastp (v0.24.0)136 and aligned to the eggplant reference line (GPE001970) genome sequence using HISAT2 (v2.2.1)137. The visual inspection of loci harboring candidate mutations was carried out using the Integrative Genomics Viewer (IGV; v2.18.4)138,139. To validate the presence of rearrangements identified through both graphs and the Pan-GWA analyses, the ONT reads of the S. incanum accessions GPE008350, GPE008290, and GPE016100 were aligned against the Hi-C-guided genome sequences of S. incanum (GPE008290 and GPE016100) and the GPE008530 genome sequence using minimap264. Loci harboring candidate variants were visually inspected using the Integrative Genomics Viewer (IGV; v2.18.4)138,139. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Data availability The sequencing raw data generated in this study have been deposited in the NCBI database under BioProject accession PRJNA1276259 and in the ENA database under BioProject accession PRJEB89803. The genome assemblies generated in this study are also available in the ENA database under BioProject accession PRJEB89803. Additionally, the genome and pangenome assemblies and annotations are available at Sol Genomics Network (https://solgenomics.net/). Source data are provided with this paper. References 1. FAOSTAT 2023. https://www.fao.org/faostat/en/#home (accessed 30 March 2023). 2. Spooner, D. M., McLean, K., Ramsay, G., Waugh, R. & Bryan, G. J. A single domestication for potato based on multilocus amplified fragment length polymorphism genotyping. Proc. Natl. Acad. Sci. USA 102, 14694–14699 (2005). 3. Portis,E.,Nervo,G.,Cavallanti,F.,Barchi,L.&Lanteri,S.Multivariate analysis of genetic relationships between Italian pepper landraces. Crop Sci. 46,2517–2525 (2006). 4. Bai, Y. & Lindhout, P. Domestication and breeding of tomatoes: what have we gained and what can we gain in the future? Ann. Bot. 100,1085–1094 (2007). 5. Razifard, H. et al. Genomic evidence for complex domestication history of the cultivated tomato in Latin America. Mol. Biol. Evol. 37, 1118–1132 (2020). 6. Tripodi, P. et al. Global range expansion history of pepper (Capsicum spp.) revealed by over 10,000 genebank accessions. Proc. Natl. Acad. Sci. USA 118, e2104315118 (2021). 7. Vorontsova, M. S., Stern, S., Bohs, L. & Knapp, S. African spiny Solanum (subgenus Leptostemonum, Solanaceae): a thorny phylogenetic tangle. Bot. J. Linn. Soc. 173,176–193 (2013). 8. Barchi, L. et al. Analysis of >3400 worldwide eggplant accessions reveals two independent domestication events and multiple migration-diversification routes. Plant J. 116,1667–1680 (2023). 9. Barchi, L. et al. A chromosome-anchored eggplant genome sequence reveals key events in Solanaceae evolution. Sci. Rep. 9, 11769 (2019). 10. Omondi, E. et al. Association analyses reveal both anthropic and environmental selective events during eggplant domestication. Plant J. 121, e17229 (2025). 11. Barchi, L. et al. Improved genome assembly and pan-genome provide key insights into eggplant domestication and breeding. Plant J. 107,579–596 (2021). 12. Golicz,A.A.,Bayer,P.E.,Bhalla,P.L.,Batley,J.&Edwards,D. Pangenomics comes of age: from bacteria to plant and animal applications. Trends Genet. 36,132–145 (2020). Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 16 13. He, W., Li, X., Qian, Q. & Shang, L. The developments and prospects of plant super-pangenomes: Demands, approaches, and applications. Plant Commun. 6, 101230 (2025). 14. Garrison, E. et al. Building pangenome graphs. Nat. Methods 21, 2008–2012 (2024). 15. Gu, R. et al. Developments on core collections of plant genetic resources: do we know enough? Forests 14,926(2023). 16. Barchi, L. et al. Single primer enrichment technology (SPET) for high-throughput genotyping in tomato and eggplant germplasm. Front. Plant Sci. 10,1005(2019). 17. Simão,F.A.,Waterhouse,R.M.,Panagiotis,I.,Kriventseva,E.V.& Zdobnov, E. M. BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics 31, 3210–3212 (2015). 18. Ou, S., Chen, J. & Jiang, N. Assessing genome assembly quality using the LTR Assembly Index (LAI). Nucleic Acids Res. 46, e126 (2018). 19. Rhie,A.,Walenz,B.P.,Koren,S.&Phillippy,A.M.Merqury: reference-free quality, completeness, and phasing assessment for genome assemblies. Genome Biol. 21,245(2020). 20. Zhou, Y. et al. Importance of incomplete lineage sorting and introgression in the origin of shared genetic variation between two closely related pines with overlapping distributions. Heredity 118, 211–220 (2017). 21. Herbert, T. D. The mid-pleistocene climate transition. Annu. Rev. Earth Planet. Sci. 51,389–418 (2023). 22. Page, A., Gibson, J., Meyer, R. S. & Chapman, M. A. Eggplant domestication: pervasive gene flow, feralization, and transcriptomic divergence. Mol. Biol. Evol. 36,1359–1372 (2019). 23. Acquadro, A. et al. Coding SNPs analysis highlights genetic relationships and evolution pattern in eggplant complexes. PLoS ONE 12,e0180774(2017). 24. Gramazio, P. et al. Development and genetic characterization of advanced backcross materials and an introgression line population of Solanum incanum in a S. melongena background. Front. Plant Sci. 8, 1477 (2017). 25. Arnoux,S.,Fraïsse,C.&Sauvage,C.Genomicinferenceofcomplex domestication histories in three Solanaceae species. J. Evol. Biol. 34,270–283 (2021). 26. Pickrell, J. K. & Pritchard, J. K. Inference of population splits and mixtures from genome-wide allele frequency data. PLoS Genet. 8, e1002967 (2012). 27. Fitak, R. R. OptM: estimating the optimal number of migration edges on population trees using Treemix. Biol. Methods Protoc. 6, bpab017 (2021). 28. Hickey, G. et al. Pangenome graph construction from genome alignments with Minigraph-Cactus. Nat. Biotechnol. 42, 663–673 (2024). 29. Ke, C. et al. Map-based cloning of LPD, a major gene positively regulates leaf prickle development in eggplant. Theor. Appl. Genet. 137, 216 (2024). 30. Satterlee, J. W. et al. Convergent evolution of plant prickles by repeated gene co-option over deep time. Science 385, eado1663 (2024). 31. Barchi, L. et al. QTL analysis reveals new eggplant loci involved in resistance to fungal wilts. Euphytica 214,20(2018). 32. Tassone, M. R. et al. A genomic BSAseq approach for the characterization of QTLs underlying resistance to Fusarium oxysporum in eggplant. Cells 11, 2548 (2022). 33. Thanyasiriwat, T. et al. Genetic loci associated with Fusarium wilt resistance in tomato (Solanum lycopersicum L.) discovered by genome-wide association study. Plant Breed. 142,788–797 (2023). 34. Miyatake, K. et al. Detailed mapping of a resistance locus against Fusarium wilt in cultivated eggplant (Solanum melongena). Theor. Appl. Genet. 129,357–367 (2016). 35. Bittner-Eddy, P. D., Crute, I. R., Holub, E. B. & Beynon, J. L. RPP13 is a simple locus in Arabidopsis thaliana for alleles that specify downy mildew resistance to different avirulence determinants in Peronospora parasitica.PlantJ.CellMol.Biol.21,177–188 (2000). 36. Kim,S.-Y.etal.ThechilipepperCaATL1:anAT-hook motif-containing transcription factor implicated in defence responses against pathogens. Mol. Plant Pathol. 8,761–771 (2007). 37. Martinčová, M. & Soukup, A. Twenty years of AT-HOOK MOTIF NUCLEAR LOCALIZED (AHL) gene family research –Their potential in crop improvement. Curr. Plant Biol. 42, 100460 (2025). 38. Whitaker, B. D. & Stommel, J. R. Distribution of hydroxycinnamic acid conjugates in fruit of commercial eggplant (Solanum melongena L.) cultivars. J. Agric. Food Chem. 51,3448–3454 (2003). 39. Sharma, M. & Kaushik, P. Biochemical composition of eggplant fruits: a review. Appl. Sci. 11,7078(2021). 40. Stommel,J.R.,Whitaker,B.D.,Haynes,K.G.&Prohens,J.Genotype × environment interactions in eggplant for fruit phenolic acid content. Euphytica 205,823–836 (2015). 41. Miguel, S. et al. A GDSL lipase-like from Ipomoea batatas catalyzes efficient production of 3,5-diCQA when expressed in Pichia pastoris.Commun. Biol. 3,1–13 (2020). 42. Li, N. et al. Super-pangenome analyses highlight genomic diversity and structural variation across wild and cultivated tomato species. Nat. Genet. 55,852–860 (2023). 43. Zhou, Y. et al. Graph pangenome captures missing heritability and empowers tomato breeding. Nature 606,527–534 (2022). 44. Liu, F. et al. Genomes of cultivated and wild Capsicum species provide insights into pepper domestication and population differentiation. Nat. Commun. 14,5487(2023). 45. Cheng, L. et al. Leveraging a phased pangenome for haplotype design of hybrid potato. Nature 1–10 https://doi.org/10.1038/ s41586-024-08476-9 (2025). 46. Guo, L. et al. Super pangenome of Vitis empowers identification of downy mildew resistance genes for grapevine improvement. Nat. Genet. 57,741–753 (2025). 47. Miyatake, K. et al. Fine mapping of a major locus representing the lack of prickles in eggplant revealed the availability of a 0.5-kb insertion/deletion for marker-assisted selection. Breed. Sci. 70, 438–448 (2020). 48. Stravato, V. M., Cappelli, C. & Polverari, A. Attacks of Fusarium oxysporum f. sp. melongenae, causing vascular wilt of aubergine in central Italy. Inf. Fitopatologico 43,51–54 (1993). (in Italian). 49. Monma,S.,Akazawa,S.,Simosaka,K.,Sakata,Y.&Matsunaga,H. ‘Diataro’, a bacterial wilt-and Fusarium wilt-resistant hybrid eggplant for rootstock. Bull. Natl Inst. Veg. Ornam. Plants Tea 12,73–83 (1997). in Japanese with English summary. 50. Rizza, F. et al. Androgenic dihaploids from somatic hybrids between Solanum melongena and S. aethiopicum group gilo as a source of resistance to Fusarium oxysporum f. sp melongenae. Plant Cell Rep. 20,1022–1032 (2002). 51. Toppino, L., Vale, G. & Rotino, G. L. Inheritance of Fusarium wilt resistance introgressed from Solanum aethiopicum Gilo and Aculeatum groups into cultivated eggplant (S. melongena)and development of associated PCR-based markers. Mol. Breed. 22, 237–250 (2008). 52. Zhang, X. et al. A truncated CC-NB-ARC gene TaRPP13L1-3D positively regulates powdery mildew resistance in wheat via the RanGAP-WPP complex-mediated nucleocytoplasmic shuttle. Planta 255, 60 (2022). 53. Chen, Y. et al. Grapevine VaRPP13 protein enhances oomycetes resistance by activating SA signal pathway. Plant Cell Rep. 41, 2341–2350 (2022). 54. Yang, H. et al. A new adenylyl cyclase, putative disease-resistance RPP13-like protein 3, participates in abscisic acid-mediated resistancetoheatstressinmaize.J. Exp. Bot. 72,283–301 (2021). Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 17 55. Kwon, Y.-I., Apostolidis, E. & Shetty,K.Invitrostudiesofeggplant (Solanum melongena) phenolics as inhibitors of key enzymes relevant for type 2 diabetes and hypertension. Bioresour. Technol. 99,2981–2988 (2008). 56. R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Available at https://www.R-project.org/ (2024). 57. Brouwer, M. & de Blok, R. coreCollection: Creating a Core Collection. (Wageningen University and Research, Department Plant Breeding, Wageningen, The Netherlands, 2022). 58. Jansen, J. & van Hintum, Th. Genetic distance sampling: a novel sampling method for obtaining core collections using genetic distances with an application to cultivated lettuce. Theor. Appl. Genet. 114,421–428 (2007). 59. Vilanova, S. et al. SILEX: a fast and inexpensive high-quality DNA extraction method suitable for multiple sequencing platforms and recalcitrant plant species. Plant Methods 16,110(2020). 60. Gramazio, P. et al. Whole-genome resequencing of seven eggplant (Solanum melongena) and one wild relative (S. incanum) accessions provides new insights and breeding tools for eggplant enhancement. Front. Plant Sci. 10, 1220 (2019). 61. Wick, R. R., Judd, L. M. & Holt, K. E. Performance of neural network basecalling tools for Oxford Nanopore sequencing. Genome Biol. 20, 129 (2019). 62. Hu, J. et al. NextDenovo: an efficient error correction and accurate assembly tool for noisy long reads. Genome Biol. 25,107(2024). 63. Medaka. Oxford Nanopore Technologies.https://github.com/ nanoporetech/medaka (2022). 64. Li, H. Minimap2: pairwise alignment for nucleotide sequences. Bioinformatics 34,3094–3100 (2018). 65. Mapping pipeline for data generated using Arima-HiC. Arima Genomics, Inc.https://github.com/ArimaGenomics/mapping_ pipeline (2025). 66. Zhou, C., McCarthy, S. A. & Durbin, R. YaHS: yet another Hi-C scaffolding tool. Bioinformatics 39,btac808(2023). 67. Durand, N. C. et al. Juicebox provides a visualization system for HiC contact maps with unlimited zoom. Cell Syst. 3,99–101 (2016). 68. Coombe, L., Kazemi, P., Wong, J., Birol, I. & Warren, R. L. Multigenome synteny detection using minimizer graph mappings. Preprint at https://doi.org/10.1101/2024.02.07.579356 (2024). 69. Alonge, M. et al. Automated assembly scaffolding using RagTag elevates a new tomato system for high-throughput genome editing. Genome Biol. 23, 258 (2022). 70. Ou, S. & Jiang, N. LTR_retriever: a highly accurate and sensitive program for identification of long terminal repeat retrotransposons. Plant Physiol. 176,1410–1422 (2018). 71. Gremme, G., Steinbiss, S. & Kurtz, S. GenomeTools: a comprehensive software library for efficient processing of structured genome annotations. IEEEACM Trans. Comput. Biol. Bioinforma. 10, 645–656 (2013). 72. Ou, S. et al. Benchmarking transposable element annotation methods for creation of a streamlined, comprehensive pipeline. Genome Biol. 20, 275 (2019). 73. Tarailo-Graovac, M., & Chen, N. Using RepeatMasker to identify repetitive elements in genomic sequences. Curr. Protoc. Bioinformatics.Chapter4:4.10.1–4.10.14 (2009). 74. Stiehler, F. et al. Helixer: cross-species gene annotation of large eukaryotic genomes using deep learning. Bioinformatics 36, 5291–5298 (2021). 75. Gabriel, L. et al. BRAKER3: fully automated genome annotation using RNA-seq and protein evidence with GeneMark-ETP, AUGUSTUS, and TSEBRA. Genome Res. 34,769–777 (2024). 76. Kim,D.,Langmead,B.&Salzberg,S.L.HISAT:afastspliced aligner with low memory requirements. Nat. Methods 12, 357–360 (2015). 77. Stanke, M. et al. AUGUSTUS: ab initio prediction of alternative transcripts. Nucleic Acids Res. 34, W435–W439 (2006). 78. Brůna, T., Lomsadze, A. & Borodovsky, M. GeneMark-ETP significantly improves the accuracy of automatic annotation of large eukaryotic genomes. Genome Res. 34,757–768 (2024). 79. Dainat J. AGAT: another Gff analysis toolkit to handle annotations in any GTF/GFF format. (Version v0.7.0). Zenodo. https://doi.org/ 10.5281/zenodo.3552717. 80. Buchfink, B., Reuter, K. & Drost, H.-G. Sensitive protein alignments at tree-of-life scale using DIAMOND. Nat. Methods 18, 366–368 (2021). 81. Jones, P. et al. InterProScan 5: genome-scale protein function classification. Bioinforma. Oxf. Engl. 30,1236–1240 (2014). 82. Emms, D. M. & Kelly, S. OrthoFinder: phylogenetic orthology inference for comparative genomics. Genome Biol. 20, 238 (2019). 83. Katoh,K.,Misawa,K.,Kuma,K.&Miyata,T.MAFFT:anovelmethod for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 30,3059–3066 (2002). 84. Capella-Gutiérrez, S., Silla-Martínez, J. M. & Gabaldón, T. trimAl: a tool for automated alignment trimming in large-scale phylogenetic analyses. Bioinformatics 25,1972–1973 (2009). 85. Price,M.N.,Dehal,P.S.&Arkin,A.P.FastTree2–approximately maximum-likelihood trees for large alignments. PLoS ONE 5, e9490 (2010). 86. Zhang, C., Scornavacca, C., Molloy, E. K. & Mirarab, S. ASTRAL-Pro: quartet-based species-tree inference despite paralogy. Mol. Biol. Evol. 37,3292–3307 (2020). 87. Zhang, C. & Mirarab, S. ASTRAL-Pro 2: ultrafast species tree reconstruction from multi-copy gene family trees. Bioinformatics 38,4949–4950 (2022). 88. Yang, Z. PAML 4: phylogenetic analysis by maximum likelihood. Mol. Biol. Evol. 24,1586–1591 (2007). 89. Zharkikh, A. Estimation of evolutionary distances between nucleotide sequences. J. Mol. Evol. 39,315–329 (1994). 90. Yang, Z. Estimating the pattern of nucleotide substitution. J. Mol. Evol. 39,105–111 (1994). 91. Kumar, S. et al. TimeTree 5: an expanded resource for species divergence times. Mol. Biol. Evol. 39, msac174 (2022). 92. Puttick, M. MCMCtreeR: a package to estimate distribution values for temporal node constraints and prepare input files for MCMCtree. https://github.com/PuttickMacroevolution/ MCMCtreeR. 93. Han, M. V., Thomas, G. W. C., Lugo-Martinez, J. & Hahn, M. W. Estimating gene gain and loss rates in the presence of error in genome assembly and annotation using CAFE 3. Mol. Biol. Evol. 30,1987–1997 (2013). 94. Yu, G., Wang, L.-G., Han, Y. & He, Q.-Y. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS J. Integr. Biol. 16,284–287 (2012). 95. Coombe, L., Warren, R. L. & Birol, I. ntSynt-viz: visualizing synteny patterns across multiple genomes. Preprint at https://doi.org/10. 1101/2025.01.15.633221 (2025). 96. Almeida-Silva, F. & Van de Peer, Y. doubletrouble: an R/Bioconductor package for the identification, classification, and analysis of gene and genome duplications. Bioinformatics 41, btaf043 (2025). 97. Kille,B.,Garrison,E.,Treangen,T.J.&Phillippy,A.M.Minmersare a generalization of minimizers that enable unbiased local Jaccard estimation. Bioinformatics 39,btad512(2023). 98. Guarracino, A., Mwaniki, N., Marco-Sola, S. & Garrison, E. wfmash: whole-chromosome pairwise alignment using the hierarchical wavefront algorithm. https://github.com/waveygang/wfmash 99. Garrison, E. & Guarracino, A. Unbiased pangenome graphs. Bioinformatics 39, btac743 (2023). Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 18 100. smoothxg: linearize and simplify variation graphs using blocked partial order alignment. https://github.com/pangenome/ smoothxg. 101. Doerr, D. & Marijon, P. GFAffix: GFAffix identifies walk-preserving shared affixes in variation graphs and collapses them into a nonredundant graph structure. https://github.com/codialab/GFAffix. 102. Guarracino, A., Heumos, S., Nahnsen, S., Prins, P. & Garrison, E. ODGI: understanding pangenome graphs. Bioinformatics 38, 3319–3326 (2022). 103. Ondov, B. D. et al. Mash: fast genome and metagenome distance estimation using MinHash. Genome Biol. 17,132(2016). 104. Garrison, E. et al. Variation graph toolkit improves read mapping by representing genetic variation in the reference. Nat. Biotechnol. 36,875–879 (2018). 105. Novak,A.M.etal.Efficient indexing and querying of annotations in a pangenome graph. Preprint at https://doi.org/10.1101/2024.10. 12.618009 (2024). 106. Garrison, E., Kronenberg, Z. N., Dawson, E. T., Pedersen, B. S. & Prins, P. A spectrum of free software tools for processing the VCF variant call format: vcflib, bio-vcf, cyvcf2, hts-nim and slivar. PLoS Comput. Biol. 18, e1009123 (2022). 107. Romain,S.,Dubois,S.,Legeai,F.&Lemaitre,C.Investigatingthe topological motifs of inversions in pangenome graphs. Preprint at https://doi.org/10.1101/2025.03.14.643331 (2025). 108. Parmigiani, L., Garrison, E., Stoye, J., Marschall, T. & Doerr, D. Panacus: fast and exact pangenome growth and core size estimation. Bioinformatics 40,btae720(2024). 109. Vorbrugg,S.,Bezrukov,I.,Bao,Z.&Weigel,D.Gretl-Variation GRaph evaluation TooLkit. Bioinformatics 41, btae755 (2024). 110. Zhao, Y. et al. PanGP: a tool for quickly analyzing bacterial pangenome profile. Bioinforma.Oxf.Engl.30,1297–1299 (2014). 111. Revell, L. J. phytools: an R package for phylogenetic comparative biology (and other things). Methods Ecol. Evol. 3,217–223 (2012). 112. Danecek, P. et al. Twelve years of SAMtools and BCFtools. GigaScience 10, giab008 (2021). 113. Zheng, X. et al. A high-performance computing toolset for relatedness and principal component analysis of SNP data. Bioinformatics 28, 3326–3328 (2012). 114. Chang, C. C. et al. Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience 4, 7 (2015). 115. Minh, B. Q. et al. IQ-TREE 2: new models and efficient methods for phylogenetic inference in the genomic era. Mol. Biol. Evol. 37, 1530–1534 (2020). 116. Letunic, I. & Bork, P. Interactive Tree of Life (iTOL) v6: recent updates to the phylogenetic tree display and annotation tool. Nucleic Acids Res. 52,W78–W82 (2024). 117. Frichot,E.&François,O.LEA: an R package for landscape and ecological association studies. Methods Ecol. Evol. 6, 925–929 (2015). 118. Francis, R. M. pophelper: an R package and web app to analyse and visualize population structure. Mol. Ecol. Resour. 17, 27–32 (2017). 119. Cattell, R. B. The scree test for the number of factors. Multivar. Behav. Res. 1,245–276 (1966). 120. Maier, R. et al. On the limits of fitting complex models of population history to f-statistics. Elife 12, e85492 (2023). 121. Weir, B. S. & Cockerham, C. C. Estimating F-statistics for the analysis of population-structure. Evolution 38,1358–1370 (1984). 122. Milanesi, M. et al. BITE: an R package for biodiversity analyses. 181610 Preprint at https://doi.org/10.1101/181610 (2017). 123. Terhorst, J., Kamm, J. A. & Song, Y. S. Robust and scalable inference of population history from hundreds of unphased whole genomes. Nat. Genet. 49,303–309 (2017). 124. Cappelli, C., Stravato, V. M., Rotino, G. L. & Buonaurio, R. Sources of resistance among Solanum spp. to an Italian isolate of Fusarium oxisporum f. sp. melongenae. In EUCARPIA, Proceedings of the 9th Meeting of Genetics and Breeding of Capsicum and Eggplant.(eds Andràsfalvi, A., Moòr, A. & Zatykò, L.) 221–224 (SINCOP, Budapest, Hungary, 1995). 125. Sulli, M. et al. An eggplant recombinant inbred population allows the discovery of metabolic QTLs controlling fruit nutritional quality. Front. Plant Sci. 12, 638195 (2021). 126. Ranawaka,B.etal.Amulti-omicNicotiana benthamiana resource for fundamental research and biotechnology. Nat. Plants 9, 1558–1571 (2023). 127. Lozano-Isla, F. inti: tools and statistical procedures in plant science. R package version 0.5.6. https://CRAN.R-project.org/ package=inti. 128. Cingolani, P. et al. A program for annotating and predicting the effects of single nucleotide polymorphisms, SnpEff. Fly. (Austin) 6, 80–92 (2012). 129. Wang, J. & Zhang, Z. GAPIT Version 3: boosting power and accuracy for genomic association and prediction. Genomics Proteom. Bioinforma. 19,629–640 (2021). 130. Huang, M., Liu, X., Zhou, Y., Summers, R. M. & Zhang, Z. BLINK: a package for the next level of genome-wide association studies with both individuals and markers in the millions. GigaScience 8, giy154 (2019). 131. Storey, J., Bass, A., Dabney, A., & Robinson, D. Package ‘qvalue’. https://bioconductor.org/packages/qvalue. 132. Yin, L. et al. rMVP: a memory-efficient, visualization-enhanced, and parallel-accelerated tool for genome-wide association study. Genomics, Proteom. Bioinforma. 19,619–628 (2021). 133. Beyer, W. et al. Sequence tube maps: making graph genomes intuitive to commuters. Bioinformatics 35,5318–5320 (2019). 134. Zytnicki, M. Assessing genome conservation on pangenome graphs with PanSel. Bioinforma. Adv. 5, vbaf018 (2025). 135. Quinlan, A. R. & Hall, I. M. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26,841–842 (2010). 136. Chen, S., Zhou, Y., Chen, Y. & Gu, J. fastp: an ultra-fast all-in-one FASTQ preprocessor. Bioinformatics 34,i884–i890 (2018). 137. Kim,D.,Paggi,J.M.,Park,C.,Bennett,C.&Salzberg,S.L.Graphbased genome alignment and genotyping with HISAT2 and HISAT-genotype. Nat. Biotechnol. 37,907–915 (2019). 138. Thorvaldsdóttir, H., Robinson, J. T. & Mesirov, J. P. Integrative Genomics Viewer (IGV): high-performance genomics data visualization and exploration. Brief. Bioinform. 14,178–192 (2013). 139. Robinson, J. T. et al. Integrative genomics viewer. Nat. Biotechnol. 29,24–26 (2011). Acknowledgements This work has been funded by the European Commission, Horizon 2020 G2P-SOL project (grant no. 677379 to G.G.) and by the Horizon Europe “Promoting a Plant Genetic Resource Community for Europe (PROGRACE)”project (grant agreement no. 101094738 to G.G.). The overall work also partially fulfills some goals of the Agritech National Research Center and received funding from the European Union Next-Generation EU (PIANO NAZIONALE DI RIPRESA E RESILIENZA (PNRR)–MISSIONE 4 COMPONENTE 2, INVESTIMENTO 1.4—D.D. 1032 17/06/2022, CN00000022). In particular, this study covers some activities comprised in: Spoke 4 (Task 4.1.1.) “Next-generation genotyping and -omics technologies for the molecular prediction of multiple resilient traits in crop plants”); Spoke 1 (Task 1.2.1 Linking phenotype and genotype: discovery of loci/genes/alleles for traits of interest) and Spoke 2 (Task 2.2.1: “Improved genetic materials to reduce the use of agrochemicals”). The World Vegetable Center also acknowledges the long-term strategic donors, including Taiwan, the United Kingdom, the United States, Australia, Germany, Thailand, South Korea, the Philippines, and Japan. Furthermore, Worldveg acknowledges the funding from NSCT Taiwan: Promoting a Plant Genetic Resource Community (NSTC 112-2923-B-125Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 19 002-MY2 to R.S.). B.U. received funding from the DFG under Germany’s Excellence Strategy (CEPLAS Cluster, EXC 2048/1, Project ID: 390686111). We finally thank Manuela Costanzo (ENEA, Casaccia Res Ctr, 00123 Roma, Italy) for help with metabolite extractions. Author contributions B.U., L.B., and G.G. conceived the study. E.I., M. Schmidt, S.L., and E.P. produced sequencing data. R.F. and M. Brouwer constructed the eggplant CC. L.G. and M. Bolger performed the genome assemblies and annotations. L.G. and L.B. performed pangenome graphs construction and downstream analyses. L.G., L.B., and G.A. analyzed resequencing data and GWA analyses. L.T., G.L.R., D.A., V.L., R.S., A.B., J.P., and H.F. B. provided plant materials. L.T., M.R.T., G.L.R., and H.F.B. performed fungal tests. L. T., M.R.T., G.L.R., H.F.B., J.P., D.A., and P.F. provided field data. M. Sulli produced and analyzed metabolic data. L.T. and S.G. performed real time analyzes on Fusarium candidate genes. L.G., L.B., B.U., and G.G. wrote the manuscript. All authors critically revised and approved the manuscript. Competing interests The authors declare no competing interests. Additional information Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41467-025-64866-1. Correspondence and requests for materials should be addressed to Giovanni Giuliano, Björn Usadel or Lorenzo Barchi. Peer review information Nature Communications thanks Kozo Nakamura, Junpeng Shi and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Reprints and permissions information is available at http://www.nature.com/reprints Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creativecommons.org/licenses/by-nc-nd/4.0/. © The Author(s) 2025 Article https://doi.org/10.1038/s41467-025-64866-1 Nature Communications | (2025) 16:9919 20