New biologically relevant resistance to bacterial wilt in heat-stressed tomato revealed by two-reference Genome Wide Association Authors: Adrien Belny, Henri Desaint, Mylène Rigal, Sébastien Carrère, Jérôme Gouzy, Gregori Bonnet, Céline Labourey, Elise Albert, Laurent Grivet, Fabrice Roux, Fabienne Vailleau, Tristan Mary-Huard, Richard Berthomé, Angela Hancock, Tatiana Giraud, Angela Hancock, Tatiana Giraud, Angela Hancock, Tatiana Giraud, Angela Hancock, Tatiana Giraud Published: 2026 ABSTRACT Ongoing climate change is driving unprecedented environmental fluctuations, with extreme events predicted to increase in frequency, intensity and duration. Among climatic parameters, temperature is expected to fluctuate the most by the end of the century and its elevation has already demonstrated to pose a major challenge to plant health. In this context, understanding how temperature modulates plant-pathogen interactions and their underlying genetic architecture is critical. Bacterial wilt, caused by strains of the Ralstonia solanacearum species complex, is a devastating disease affecting many plant species. Genetic resistance remains the most effective control strategy. In tomato, resistance is quantitative and mostly relies on quantitative trait loci (QTLs) bwr-6 and bwr-12. However, as in other crops, high temperature and humidity can compromise this resistance in commercial tomato cultivars. We investigated temperature-dependent quantitative disease resistance (QDR) using a panel of 189 wild tomato accessions, predominantly Solanum pimpinellifolium, representing genetic diversity from contrasting ecological conditions. Disease progression was monitored from three to ten days post-inoculation at 28 °C and 32 °C, using a time-course phenotyping approach. Genome-wide association (GWA) analyses were performed, based on daily symptom scores and two reference genomes, to account for structural variations and improve QTL detection. This strategy identified 44 candidate genes and revealed a temporally dynamic genetic architecture of the plant response. Strikingly, no candidate genes were shared between temperatures, supporting distinct genetic determinants under different temperature conditions. Many candidate genes were expressed in roots and belong to gene families involved in plant immunity, with two candidates co-localizing with bwr-6 and bwr-12, whose causal genes remain unknown. Altogether, our findings demonstrate that resistance to bacterial wilt in wild tomato is flexible, environment-dependent, and temporally dynamic, highlighting the importance of integrating environmental context and genomic diversity to better understand plant-pathogen interactions. Introduction A growing number of studies have assessed various combinations of plant-pathogen interactions under abiotic stress conditions (cold, high temperature, low nitrogen, drought) [1,2], showing that many plant defense responses are negatively affected by such constraints. However, the underlying mechanisms remain largely unknown, and most studies are descriptive and primarily rely on phenotypic and transcriptomic analyses [1,3]. Among climatic parameters, temperature is predicted to fluctuate the most by the end of the century [4], and a meta-analysis showed that 55% of 145 genetic resistance sources are compromised under heat stress, leading to increased disease severity [3]. In this context, understanding how climatic parameters affect host and pathogen responses as well as identifying novel robust resistance or mechanisms that ensure plant resilience under fluctuating environmental conditions is crucial. Bacteria belonging to the Ralstonia solanacearum species complex (RSSC) are able to infect a wide range of plant species from 53 botanical families and are responsible for various bacterial diseases that can lead to considerable economic losses in major crops (tomato, potato, eggplant and other Solanaceae species), crops of local importance (banana, peanuts, etc), and ornamentals (pelargonium, rose) [5,6]. Strains of the RSSC are soil-borne pathogens that enter plant roots, colonize and multiply in the xylem, clogging vessels and leading to characteristic wilting symptoms [7]. These strains originate from tropical and sub-tropical regions and are organized into four major genetic clusters, called phylotypes. Interestingly, the geographical origins of the strains are correlated with their phylotype [5]. To limit the occurrence of bacterial wilt and overcome disease outbreaks, integrated pest management combines multiple strategies [8]. Promising developments in control methods are based on, or use, living organisms [9,10] and can complement traditional cultivation practices, such as crop rotation [11] and pesticide use [8]. However, multiple studies have shown that genetic solutions remain the most efficient, cost-effective and eco-friendly means of tackling these diseases [12,13]. To date, more than forty QTLs underlying plant defense responses to different strains of RSSC have been mapped in A. thaliana, tobacco, potato, peanut, pepper, eggplant and tomato, mainly through traditional QTL mapping approaches [13,14]. Analyses of host resistance together with studies of the RSSC type III effectors (T3E) repertoire, a major determinant of pathogenicity, indicate that plant immunity to RSSC is generally polygenic and strain-specific [12,13]. Despite the identification of numerous resistance loci, their underlying molecular mechanisms remain poorly understood. One exception is the A. thaliana nucleotide-binding, leucine-rich repeat (NLR) receptor pair RPS4/RRS1 [15–17], which confers qualitative resistance to several RSSC strains [15,18], as well as to other pathogens [19] and contributes to QDR against Xanthomonas campestris pv. campestris strain CFBP6943 and GMI1000 at 27 °C [20,21]. Importantly, both RPS4/RRS1-mediated resistance and broader defense responses are compromised in case of GMI1000 at 30 °C, highlighting the strong impact of elevated temperature on plant immunity to this pathogen [21]. Natural variation in plant-pathogen interactions coupled with GWAS have proven effective in detecting QTL underlying disease resistance [22–25]. In Arabidopsis, GWAS analyses of responses to GMI1000 and four related T3E mutant strains at 27°C detected several susceptibility genes, some of which were functionally validated [21,26,27]. Nevertheless, such strategies investigating plant responses under combined biotic and abiotic stress conditions remain scarce. Using 176 A. thaliana accessions inoculated with GMI1000 under controlled conditions at 27 °C and 30 °C, with different root-wounding regimes, recent GWAS revealed distinct resistance mechanisms and temporal plasticity in plant responses. Functional validations further identified Strictosidine Synthase-Like Proteins (SSL4 and SSL5), Solo Dancers (SDS), and Cellulose Synthase A3 (CESA3, subunit 3 of the primary cell wall cellulose complex) genes as susceptibility factors contributing to bacterial wilt development under elevated temperature conditions [21,28,29]. Tomato (Solanum lycopersicum) is one of the main Solanaceae species cultivated worldwide and serves as a model for fruit-bearing crops. As natural hosts of R. solanacearum, most tomato cultivars have been described to be susceptible to a wide range of strains belonging to the four phylotypes of the RSSC [30]. Solanum pimpinellifolium, a wild relative of cultivated tomato, has contributed to the identification of resistance to over 45 different diseases [31,32] including resistance to Phytophthora infestans [33,34], Fusarium oxysporum (I2 gene) [35], gray leaf spot disease (Sm gene) [36] and bacterial speck (Pto gene) [37]. For bacterial wilt, the S. lycopersicum ‘Hawaii 7996’ cultivar is still the most tolerant to several strains of R. solanacearum [38] and is widely used in breeding programs. Its tolerance originates from S. pimpinellifolium accession PI127805A and relies on at least seven QTLs, of which bwr-6 and bwr-12 are the most studied [39]. However, none of the genes underlying these QTLs have been cloned and the genetic architecture of this QDR appears complex: bwr-6 and bwr-12 each comprise multiple smaller QTLs, four for bwr-6 and three for bwr-12 [40]. As for other Solanaceae species defense responses, the “Hawaii 7996” cultivar tolerance appears sensitive to elevated temperatures [41,42]. Increased soil temperatures can impair the bacterial wilt QDR for some tomato varieties [43]. In Taiwan and other parts of South-East Asia, varieties with bwr-12-based QDR tend to be more susceptible to the bacteria at high temperatures. When phenotyped under heat stress, the ‘Hawaii 7996’ cultivar exhibited a wilting rate ranging from 28% to 52%, depending on the trial [38]. These observations demonstrate the need to identify and characterize new sources of resistance to bacterial wilt that remain effective under elevated temperature or changing environmental conditions. Tomato wild relatives originated from tropical and equatorial regions and evolved under diverse and often extreme environmental conditions, probably in contact with R. solanacearum species [2]. While optimal growth conditions for tomato vary between 13 °C and 33 °C depending on the variety studied [43], the optimal growth temperature for the R. pseudosolanacearum GMI1000 strain is 28 °C [44]. As tomato domestication led to a genetic bottleneck and reduced intraspecific diversity [45], exploring closely related wild species from contrasting habitats offers the opportunity to uncover adaptive traits that may have been lost during domestication. We therefore hypothesized that accessions naturally exposed to the bacterium could have evolved defense mechanisms shaped by their native environmental conditions. To investigate how temperature modulates QDR to R. pseudosolanacearum and identify its genetic bases, we implemented a combined phenotyping and genomics approach using contrasting temperature conditions. A panel of wild tomato accessions, enriched in S. pimpinellifolium, was challenged with the GMI1000 strain under permissive and elevated temperatures. Disease progression was monitored through a time-course experiment, enabling the capture of both early and late responses to infection. All accessions were sequenced and analysed using a GWAS approach based on two distinct reference genomes. These references were selected to reflect the genetic diversity of a previously described panel [46], for their contrasting biomes of origin and responses to the bacteria at 28 °C. This strategy allowed us to dissect the contribution of genetic variation, including structural differences between genomes, to the architecture of resistance across environments. Our study highlights the temporal dynamic and different genetic architectures of the plant defense response at 28 °C and 32 °C, with a strong impact of the temperature elevation. Our findings also contribute to refining the detection of two previously detected major QTLs: bwr-6 and bwr-12, for which genes have not been identified. Results Panel characteristics Most accessions of the panel originate from Peru and Ecuador, the native regions of S. pimpinellifolium and S. lycopersicum var. cerasiforme, with 73% from Peru and 14% from Ecuador. The S. lycopersicum var. cerasiforme accessions originated from Central America and from the western side of the Andes (Fig 1A and 1B). Some S. pimpinellifolium accessions were also found in this area, but most were sampled from the eastern side, either in the mountainous areas of southern Ecuador/northern Peru or in the coastal valley near the Pacific Ocean in southern Peru (Fig 1A and 1B). Probably due to this distribution on the slopes of the mountains, the accessions were collected at different altitudes with the majority (45%) found below 500 m. 14.3% found in a range of altitudes between 500 m and 1,000 m and finally 5.8% above 1,000 m. As Blanca et al. [46] showed that genetic diversity in their smaller panel of S. pimpinellifolium accessions was structured into two clusters, we selected two representative accessions from each cluster as internal references for HiFi sequencing. LA1670 and LA1246 were chosen not only for originating from habitats with contrasting climates, as determined from their GPS coordinates [48] (S1 Table, S1 Fig), but also for exhibiting differential responses to R. pseudosolanacearum inoculation at 28 °C and 32 °C (Fig 2A). Geographic distribution and population structure of accessions based on LA1670 and LA1246 reference genomes. A and C: Geographic distribution of accession clusters identified by kinship and hierarchical clustering (K = 5) on a map of northwestern South America using LA1670 (A) and LA1246 (C). B and D: PCA based on SNPs called from LA1670 (B) and LA1246 (D), with clusters overlaid. E and F: Mean annual surface air temperature (E) and precipitation (F) at sampling sites (1981-2010 period), projected onto the PCA based on LA1246. Corresponding results for LA1670 are shown in S4A and S4B Fig. Geographic origin of accessions was mapped onto both geographic (A) and PCA (B) plots to illustrate spatial distribution. The geographical distribution of the panel was plotted using passport data (S2 Table) and the worldmap() function from the rnaturalearth [47] R package. Basemap data (country boundaries) were obtained from Natural Earth: https://www.naturalearthdata.com/downloads/. Terms of use: https://www.naturalearthdata.com/about/terms-of-use/. Exploration of the phenotypic response of the panel following infection by R. pseudosolanacearum at 28 °C and 32 °C. A: Kinetics of the response of panel accessions to R. pseudosolanacearum at 28 °C (left) and 32 °C (right). The y-axis represents LS-means of disease index (Di; 0-4), over days after inoculation (Dai). Each line represents the disease progression curve of one accession. Mean panel response is shown in green, with reference accessions ‘Hawaii 7996’ (yellow), LA1246 (blue) and LA1670 (purple). B: Difference in disease index between 32 °C and 28 °C at 10 Dai (ΔDi = 32 °C-28 °C). Values represent LS-means per accession. Accessions are grouped and differentially colored according to ΔDi. Dark gray: accessions more susceptible at 32 °C (ΔDi > 0.5); light gray: accessions with similar response at both temperatures (|ΔDi| < 0.5); white: accessions less susceptible at 32 °C (ΔDi < -0.5). Genomic data and comparative analysis of reference genomes Reference genomes were sequenced using PacBio technology and reached a depth of 82x for LA1246 and 80x for LA1670 (S2 and S3 Tables). The high sequencing depth obtained supports the accuracy of both reference genomes. Interestingly, genome assembly highlighted chromosomal rearrangements between the two reference genomes, especially on chromosomes 6, 7 and 8 (Fig 3A and 3B). The final assembled genome size was 815.7 Mb for LA1670 and 808.6 Mb for LA1246. Gene prediction revealed 46,202 (LA1670) and 45,746 (LA1246) genes (S2 and S3 Tables). Illumina sequencing of the panel revealed a depth of between 15x and 25x for raw sequenced data which varied among accessions. For SNP (Table 1, filter A), the mean coverage at a locus was between 13.66x (LA1246) and 13.81x (LA1670) (S4 and S5 Tables). 10.1371/journal.pgen.1011820.t001Table 1Number of SNPs in matrices per reference genome after sequential filtering and their use in downstream analyses.Filter usedReference Genome (number of SNP)AnalysesLA1670LA1246A58 353 01558 606 561A + biallelic47 430 64447 486 684A + B25 622 12525 929 267A + B + C13 541 87214 551 270PCAA + B + C + D1 836 1482 567 085KinshipA + B + C + E3 728 9463 646 481GWAS SNPs matrix Number of SNPs in matrices per reference genome after sequential filtering and their use in downstream analyses. Collinearity between LA1246 and LA1670 genomes and QTLs detected either at 28 °C or 32 °C. A and B: Synteny anchor BLAST hits colored by block ID between Heinz (y-axis; 25,829 genes with BLAST hits, ITAG4.1 [50]) and LA1670 (x-axis; 25,829 genes) (A) or LA1246 (x-axis; 25,829 genes) (B). C: Synteny anchor BLAST hits colored by block ID between LA1246 (y-axis; 34,134 genes) and LA1670 (x-axis; 34,135 genes). Chromosome position is shown by gene rank order. Gridlines indicate every 1,000 genes. D: Global overview of collinearity between LA1246 and LA1670 genomes. QTLs specific at 28 °C or 32 °C are indicated. QTLs unique to LA1246 are shown on the right, those unique to LA1670 on the left, and shared QTLs are highlighted in green. Analyses, localization of QTLs and visualization were performed using GENESPACE [49]. Whole-genome dot plot comparisons were performed between the S. lycopersicum ‘Heinz’ reference genome [50] and those of LA1670 and LA1246, as well as between LA1670 and LA1246 (Fig 3A). All comparisons showed a nearly continuous main diagonal, indicating strong chromosome-level collinearity. Minor deviations and local breaks in the diagonal (Heinz - LA1670: chromosomes 2, 5, 12; Heinz - LA1246: chromosomes 1, 2, 3; LA1246 - LA1670: chromosomes 5, 12) suggest limited structural variation, likely reflecting small-scale rearrangements and/or local size differences (Figs 3B, S2). Consistent with the larger genome sizes of LA1670 and LA1246 relative to Heinz, these local expansions could correspond to repetitive or intergenic regions, as previously reported [51]. Orthogroup analysis performed between LA1246 and LA1670 (S6 and S7 Tables) classifyed genes according to their presence/absence and copy number variations (CNVs) across genomes (0–1, 1–0, 1–1, 0-n, n-0, 1-n, n-1, n-n). A total of 34,801 1–1 orthologs were identified, confirming a high level of gene conservation. Most differences between genomes were associated with CNVs, (1-n expansion in LA1670; n-1 expansion in LA1246) or presence/absence (0–1, 1–0), in a limited set of orthogroups. Gene ontology enrichment analyses of presence-absence orthogroups (S7 Table) revealed contrasting functional signatures between genomes. Orthogroups specific to LA1670 were strongly enriched for genes involved in chromosome segregation, mitotic division and cell cycle processes, whereas orthogroups specific to the LA1246 genome were predominantly associated with transcriptional regulation and metabolic processes. The number of specific orthogroups was slightly higher in LA1246 (2,953) than in LA1670 (2,606) corresponding to a 0.8 ratio, suggesting relatively more gene loss or contraction in LA1670. Notably, orthogroups with CNVs corresponding to genes involved in defense responses and stimulus perception were significantly overrepresented, consistent with the rapid evolution of these gene families often located in repeat-rich genomic regions as previously demonstrated [51]. Reference genomes were used to call SNPs obtained by panel sequencing. We assessed the gene SNP coverage by using filtered GWAS SNP matrices (Table 1; filters A, B, C + E). Results indicated that at least one SNP was found in 80% of genes with, on average, 9.6 SNPs/gene whichever reference genome was chosen. This coverage varied between 7.5 and 13.2 SNPs/gene depending on the chromosome (S3 Fig). A continuous population structure shaped by geographic and environmental gradients and a LD suitable for GWAS Population structure was studied through principal component analysis (PCA) using SNPs called from the two reference genomes. (Table 1; filters A - C). The first two PCA axes (Fig 1), explaining cumulatively around 10% of the genetic variation (LA1670: Dim1 6.53% - Dim2 4.07%; LA1246: Dim1 5.53% - Dim2 3.82%) revealed the limited global genetic structure within the panel, with individuals distributed along an arch pattern indicative of a continuous gradient. To determine the information that could be described by both axes, we first report manually the provinces of origin on the PCA results and on a map of northwestern South America to visualize their geographical distribution (Fig 1A and 1B). We also report on PCA results obtained with both reference genomes i) species distribution (S4C and S4D Fig), ii) mean annual surface air temperature (Figs 1E, S4A) and mean annual precipitation (Figs 1F, S4B) at the sampling site, when available, during the 1981–2010 period. The spatial distribution of individuals along the PCA axes seemed to reflect a geographic east-west gradient on the first axis, reflecting genetic variation from populations along the Pacific coast, across the western slopes of the Andes, to more inland regions of central Peru. The second axis mostly captured both a south-north geographic gradient and climatic variation, ranging from colder and drier to warmer and more humid conditions. These results support a scenario of continuous genetic variation shaped by both geographic and environmental factors. Hierarchical clustering using hclust applied to the kinship-derived distance matrix identified five groups (K = 5). When projected onto PCA results and on a map of northwestern South America (Fig 1A and 1D), clusters aligned with the main genetic gradient and corresponded to the geographic distribution of accessions. Pairwise F ST estimates revealed moderate to high genetic differentiation among clusters, confirming that these groups reflect non-random population structure. Overall, genetic variation appears structured along continuous geographic and environmental gradients rather than discrete populations. Using LA1246 as an alternative reference genome produced slightly different hierarchical clusters, despite high overall genome similarity (Fig 1C and 1D). The extent of LD was investigated using the filtered GWAS matrix (Table 1; filters A, B, C + E); its half-decay was calculated per chromosome and is of the order of a few kb for most chromosomes, with the exception of chromosomes 5, 6 and 8 for which LD decay was lower (S5 Fig, S8 Table). Specifically, LD visualization on these latter chromosomes showed large blocks of LD, located in central chromosomal regions, and low LD levels in telomeric regions (S6 Fig). Panel responses to R. pseudosolanacearum reveal three temperature-dependent profiles Panel accessions were phenotyped at 28 °C and 32 °C for bacterial wilt disease severity for ten days following pathogen inoculation (Fig 2). Overall, the 189 accessions were on average more susceptible at 32 °C than at 28 °C. At 28 °C, the two reference accessions displayed contrasting responses to R. pseudosolanacearum, with LA1246 remaining symptomless. This difference disappeared at 32 °C, where both accessions developed wilting symptoms. The cultivar ‘Hawaii 7996’, included as a known tolerant control at 28 °C, began to wilt at the end of the infection kinetic at 32 °C (Fig 2A). At 28 °C, no genetic variation was detected from 3 to 6 dai, whereas H² reached 0.19 and 0.24 at 7 and 10 dai, respectively (Table 2). In contrast, at 32 °C, genetic variation was detected at all time points except at 3 dai, with H² values fluctuating from 0.11 to 0.29. At both temperatures, H² was the highest at 10 Dai, with slightly higher values at 32 °C. Notably, at 32 °C, H² was already of a similar magnitude at 4 dai, suggesting greater genetic variability both at early stages of the plant-pathogen interaction and during disease progression. Although experiments were conducted under controlled conditions, the dynamic nature of disease development may have led to disparity in symptom onset between blocks phenotyped at different times. Such temporal shifts may reduce heritability estimates when blocks are analysed jointly. However, the higher heritability observed along the infection kinetic at 32 °C may reflect a stronger and more synchronous expression of plant responses, facilitating QTL detection. 10.1371/journal.pgen.1011820.t002Table 2Natural variation among the accessions of the panel for disease index at 28 °C and 32 °C.TemperatureModel termsDai 3Dai 4Dai 5Dai 6Dai 7Dai 10F/LRTp-valueF/LRTp-valueF/LRTp-valueF/LRTp-valueF/LRTp-valueF/LRTp-value28 °CBlock4.870.027312.140.000515.70.000121.83<0.000130.21<0.000192.38<0.0001Accession0.910.77100.970.59300.950.67001.110.18001.30.00871.490.0001Interaction1.380.2400116.36<0.000137.85<0.000130.19<0.000116.43<0.000125.98<0.0001 H 2 0.0000.0000.0000.0000.1820.24232 °CBlock15.070.000165.66<0.000191.29<0.0001123.82<0.0001121.65<0.0001119.88<0.0001Accession1.070.28101.480.00021.330.00521.380.00181.350.00371.64<0.0001Interaction44.610.00009.620.00198.770.003116.930.003915.9<0.000112.750.0004 H 2 0.0060.2470.1230.1110.1110.293 Natural variation among the accessions of the panel for disease index at 28 °C and 32 °C. Based on adjusted means disease severity ratings, a cross-temperature genetic Pearson correlation was calculated, at 10 dai between 28 °C and 32 °C, time point at which H² was the highest. The low correlation (r = 0.3) indicates that the genetic control of disease progression is only partially shared between the two temperatures, suggesting a substantial genotype-by-temperature interaction and a flexible genetic architecture underlying the response of accessions to inoculation with R. pseudosolanacearum GMI1000 strain. Several accessions remained symptomless until the end of the experiment, at both temperatures (Di < 0.2) (S7 Fig), suggesting the presence of strong resistance mechanisms in the panel. To explore temperature-dependent responses, reaction norms were plotted at 10 Dai between both temperatures, revealing four different profiles (S7 Fig). Finally, by comparing the adjusted mean disease severity ratings, three temperature-dependent response profiles were defined: (1) accessions that were more susceptible at 32 °C than at 28 °C (133 out of 189 accessions), (2) accessions displaying similar responses at both temperatures (49 out of 189), and (3) accessions showing lower susceptibility at 32 °C (7 out of 189) (Fig 2B). GWAS and post-GWAS analyses identify 44 candidate genes putatively involved in bacterial wilt defense response at 28 °C and 32 °C GWAS results are presented in (Tables 3 and S9, S8 Fig). After QTL filtering, 21 “strict candidate genes” corresponding to 18 QTLs were selected. QTL intervals did not exceed a few kb, with most loci underlying a single candidate gene, indicating near gene-level resolution. Exceptions included QTL 1–3-LA1246 (12 kb, high gene density) and QTL 1–4-BOTH, which encompassed two genes; however, the most significant SNPs in both references (SpimpiLA1670Chr01_88223852_T for LA1670 and SpimpiLA1246Chr01_89956087_T for LA1246) mapped to an exon of the Solyc01g105270 homolog. Considering “intergenic candidate genes” (Table 3) added 23 genes related to 10 oher QTLs. Although these did not meet proximity criteria, some were recurrently detected along the kinetic of infection, notably QTL IG12–1-BOTH identified over four consecutive days (S8 Fig), the maximum observed. 10.1371/journal.pgen.1011820.t003Table 3Synthetic overview of the QTL and corresponding candidate genes identified in this study at 28 °C (A) and 32 °C (B).AQTL name (a)Reference genomes (b)Dai of detection LA1246/ LA1670 (c)S. lycopersicum homolog (d)Expression pattern (e)A. thaliana ortholog (f)Expression pattern (g)Functional category (h)References1-1-LA1670LA1246*/LA16707/ 7, 10Solyc01g006660 Subtilisin-like serine proteaseRoots; leaves (S.pimpinellifolium) - nucleusAt1g20160 ATSBT5.2,CRSP,S CO2 RESPONSE SECRETED PROTEASE; Subtilisin-like serine endopeptidase family proteinRoots, stem, extracellular - endoplasmic reticulum, defense reponse to bacteria476,773-2-LA1670LA1246*/LA167010/ 7, 10Solyc03g093130 xyloglucan-modifying enzymeMature green, breaker fruits, rootsMultigene familynana3-4-LA1670LA1246*/LA16707/ 7Solyc03g019730 Sumo activating enzyme 1b, Molybdenum cofactor biosynthesisUnopened flower buds, roots, leaves - mitochondrionAt5g50580, At5g50680 AT-SAE1–2, SAE1B, sumo-activating enzyme 1BSeeds - cytosol44-1-LA1246LA1246/LA1670*7/ 7Solyc04g008310. UDP-glucuronosyl/UDP-glucosyltransferaseBreaker fruits - cytosol At2g15490 UDP-glycosyltransferase 73B4 Seeds, roots - cytosol, cell wall, response to biotic and abiotic stresses1667-3-LA1670LA1246*/LA16707/ 7Solyc07g044730 3-hydroxyisobutyryl- CoA hydrolase-like proteinOpened flowers At1g06550 Multigene family na51-4-BOTHLA1246/LA16707,10/ 7,10Solyc01g105260 Integral membrane protein Protein of unknown function DUF125, vacuolar iron transporter homolog 4-likeRoots - plasma membranenana11-4-BOTHLA1246/LA16707,10/ 7,10Solyc01g105270 LeTH2 tobamovirus multiplication 1 homolog 2Roots,fruits, flowersnana1682-1-BOTHLA1246/LA167010/ 10Solyc02g067460 Mitochondrial outer membrane protein porinYoung fruits, roots, leaves (S.pimpinellifolium) - mitochondrionMultigene familyna54-2-BOTHLA1246/LA16707,10/ 7,10Solyc04g078590 Receptor kinase; serine/threonine kinaseYoung, fruits - plasma membraneMultigene familyna325IG1–1-LA1246LA1246/LA1670*7,10/ 7,10nananananaIG1–1-LA1670LA124*/LA16707,10/ 7,10Solyc01g016350 unknown proteinExtracellularnana5IG2–1-LA1670LA1670na/ 7 Solyc02g084670 Unopened flower buds - mitochondrionMultigene familyna5IG2–1-LA1670LA1670na/ 7Solyc02g084680 Protein SHI RELATED SEQUENCE 5Flowers, roots, leavesMultigene familyna5IG3–1-LA1246LA124610/ nananaAt4g12560 (?)na5IG3–1-LA1246LA124610/ naSolyc03g093770 histone-lysine methyltransferaseLeavesnana5 B QTL name (a) Reference genomes (b) Dai of detection LA1246/ LA1670 (c) S. lycopersicum homolog (d) Expression pattern (e) A. thaliana ortholog (f) Expression pattern (g) Functional category (h) References 1-2-LA1246LA1246/LA1670*10/ 10Solyc01g086890 Splicing factor 4-like protein G-patchRoots, fruits, flowersAt3g52120 CIS1, CRY2 interacting splicing factor 1, regulate thermosensory flowering via FLM alternative splicing.Everywhere - nucleus, cytosol51-3-LA1246LA1246/LA1670*6, 7/ 6Solyc01g100650 NHL repeat-containing protein 2-likeLeaves - mitochondrionAt4g35090 (?) Multigene familyEverywhere - plastids; response to pathogen infection51-3-LA1246LA1246/LA1670*6, 7/ 6Solyc01g100630 CatalaseLeaves, roots - peroxisomeAt4g35090 (?) Multigene familyEverywhere - peroxysome, mitochondrion, response to infection21-3-LA1246LA1246/LA1670*6, 7/ 6Solyc01g100640 CatalaseBreaker fruits - peroxisomeAt4g35090 (?) Multigene familyEverywhere - peroxysome, mitochondrion, response to infection23-3-LA1670LA1246*/ LA167010/ 10Solyc03g097470 3-Beta-ketoacyl-acyl carrier protein synthase III (FabH)Breaker fruits, mitochondrionAt1g62640 3-ketoacyl-acyl carrier protein synthase IIIEverywhere - response to infections5657-1-LA1670LA1670na/ 4, 5Solyc07g005040, Solyc00g257110 Plasma membrane ATPaseFlowersMultigene familyna510-1-LA1246LA1246/LA1670*6, 7/ 7Solyc10g081260 Multidrug resistance protein mdtK Protein Multi antimicrobial extrusion protein MatEYoung and breaker fruits - plasma membraneAt5g17700 MATE efflux family proteinRoots, petals - plasma membrane - response to infections1653-1-BOTHLA1246/LA16707, 10/ 7, 10Solyc03g123790 Vacuolar calcium cation/proton exchangerRoots, mature green fruits, plasma membraneAt1g55730 (?) Multigene familyna27-2-BOTHLA1246/LA16704, 5, 6/ 4, 5, 6nananana8-1-BOTHLA1246/LA16705, 7/ 5, 7nananana9-1-BOTHLA1246/LA16707, 10/ 7, 10Solyc09g009700 GDSL esterase/lipaseRoots, opened flowers, extracellularAt5g03610 GGL25 - GDSL-motif esterase/acyltransferase/lipaseStem, embryo, primary root - endoplamsic reticulum, extracellular - response to pathogen and nitrogen1679-2-BOTHLA1246/LA16707, 10/ 7, 10Solyc09g011630 Glutation-S-transferaseRoots, cytosolAt2g29420 (?) Glutathione S-transferase 25, Glutathione S-transferase TAU 7, GST25, GSTU7Roots, petals - Response to hormones and infection49-2-BOTHLA1246/LA16707, 10/ 7, 10Solyc09g011640 Glutation-S-transferaseRoots, cytosolAt2g29420 Glutathione S-transferase 25, Glutathione S-transferase TAU 7, GST25, GSTU7Roots, petals - Response to hormones and infection4IG4–1-LA1246LA1246/LA1670*7/ 7nanananaIG4–1-LA1246LA1246/LA1670*7/ 7Multigene familynaMultigene familynaIG4–1-LA1246LA1246/LA1670*7/ 7Solyc00g028960 Glutaredoxin family proteinLeaves (S.pimpinellifolium)Multigene familyna3IG6–1-BOTHLA1246/LA16707, 10/ 7, 10Multigene familynaMultigene familynaIG6–1-BOTHLA1246/LA16707, 10/ 7, 10Solyc06g051510 Protein kinase superfamily proteinRoots, mature green fruits, nucleusMultigene familyna3IG6–2-BOTHLA1246/LA16704/ 4Solyc06g065060 FAD-binding domain-containing proteinMature green fruits, vacuoleMultigene familyna5IG6–2-BOTHLA1246/LA16704/ 4Solyc06g065070 FAD-binding domain-containing proteinMature green fruits, vacuoleMultigene familyna5IG7–1-LA1670LA1246*/LA16704/ 4Solyc07g007600 vacuolar-type H + -pyrophosphataseRoots, mature grenn fruits, leaves (S. pimpinellifolium), nucleus At1g15690 Everywhere - vacuole and plasma membrane - response to some infections5IG7–1-LA1670LA1246*/LA16704/ 4Solyc07g007610 At2G21500 protein (Fragment)Opened and unopened flowers, nucleusnanaIG7–2-BOTHLA1246/LA167010/ 10Solyc07g007940 Small nuclear RNA activating complex (SNAPc), subunit SNAP43 proteinRoots, immature fruits, cytosol At3g53270 Everywhere - cytosol, nucleus - response to infections565IG7–2-BOTHLA1246/LA167010/ 10nanananaIG12–1-BOTHLA1246/LA16704, 5, 6, 7/ 4, 5, 6, 7Solyc12g008950 DEAD-box ATP-dependent RNA helicaseRoots, young and breaker fruits - nucleus At1g72690 (?) Nucleus; response to infections1 IG12–1-BOTH LA1246/LA1670 4, 5, 6, 7/ 4, 5, 6, 7 Solyc12g008960 Calmodulin-binding protein Roots - nucleus At4g33050 (?) Everywhere - cytosol, mitochondrion, nucleus - response to some infections 1 (a) name of the QTL. (b) The reference genome(s) on which the QTL was detected. An asterisk (*) stipulates that the QTL underlying the candidate gene was detected with one reference genome and was below the threshold p-value of detection with the other reference genome, even if an association peak was observed (see S8A-S8F Fig for details). (c) Dai at which QTL were detected with LA1246 or LA1670. (d) S.lycopersicum homologous gene ID and its related annotation found for the ‘Heinz’ reference [50]. (e) Summary of known gene expression patterns information in organs - cell - experimental conditions, retrieved from eplant tomato visualization tools (https://bar.utoronto.ca/eplant_tomato/). (f) Corresponding A. thaliana orthologous gene ID and its related annotation when available; (?) corresponds to genes with low-confidence BLAST results. (g) Summary of known gene expression patterns information in organs - cell - experiment conditions retrieved from eplant A. thaliana visualization tools (https://bar.utoronto.ca/eplant/). Order of organs and subcellular compartments indicate the range of upregulation, the first being the one in which the gene is the most highly overexpressed. (h) Functional category: 1-immunity, 2-calcium signaling related, 3-RLKs, 4- may be involved in plant defense, 5-link to be demonstrated. na: not available. Synthetic overview of the QTL and corresponding candidate genes identified in this study at 28 °C (A) and 32 °C (B). (a) name of the QTL. (b) The reference genome(s) on which the QTL was detected. An asterisk (*) stipulates that the QTL underlying the candidate gene was detected with one reference genome and was below the threshold p-value of detection with the other reference genome, even if an association peak was observed (see S8A-S8F Fig for details). (c) Dai at which QTL were detected with LA1246 or LA1670. (d) S.lycopersicum homologous gene ID and its related annotation found for the ‘Heinz’ reference [50]. (e) Summary of known gene expression patterns information in organs - cell - experimental conditions, retrieved from eplant tomato visualization tools (https://bar.utoronto.ca/eplant_tomato/). (f) Corresponding A. thaliana orthologous gene ID and its related annotation when available; (?) corresponds to genes with low-confidence BLAST results. (g) Summary of known gene expression patterns information in organs - cell - experiment conditions retrieved from eplant A. thaliana visualization tools (https://bar.utoronto.ca/eplant/). Order of organs and subcellular compartments indicate the range of upregulation, the first being the one in which the gene is the most highly overexpressed. (h) Functional category: 1-immunity, 2-calcium signaling related, 3-RLKs, 4- may be involved in plant defense, 5-link to be demonstrated. na: not available. GWAS using two reference genomes help to detect QTLs Among the 21 strict candidates, only one was found specific to LA1670, nine were detected in both reference genomes, while 11 showed significant association in one genome with an association peak slightly below the detection threshold in the other (indicated by a * in Tables 3 and S9), and were thus considered nearly shared. Differences in QTL detection can be explained by three main factors. First, statistical threshold effects led to marginal QTLs being retained in only one reference, as illustrated by QTL 1–2-LA1246 (Tables 3 and S9, Fig 4A). The GWAS model used was based on a statistical test using both kinship and population structure. Despite strong correlations between kinship matrices (>0.9), differences at the whole genome level may impact its coverage and the ability to detect associations. Second, differences in variant calling at the locus level due to sequence variations and local sequence divergence between reference genomes. This is illustrated by the QTL 7–1-LA1670, underlying a gene encoding an APT8 Plasma membrane-like protein where SNP coverage differs depending on the reference genome used (Fig 4B). In LA1246, the exon containing the top SNP is not covered by any variant call, preventing identification. Third, structural variation between genomes contributed to differential detection. BLAST comparisons revealed indels and inversions affecting QTL regions. For example, QTL 4–1-LA1246 underlying a gene encoding a potential UDP-glycosyltransferase is located near two indels in the LA1246 genome that may have affected its detection in LA1670 (Fig 4C). Other genomic inversions were detected near QTL 3–2-LA1670, and near QTL IG3–1-LA1246, and a 200 kb indel separates the two genes underlying QTL IG8–1-LA1670. Additionally, BLAST analysis revealed that the gene closest to QTL IG2–1-LA1670, annotated as a putative serine/threonine phosphatase in LA1670, is absent in the LA1246 genome. Three different cases for which the use of two reference genomes improved QTL detection. A: Effect of the reference genome on QTL detection resulting from the selection criteria used. GWAS results at 32 °C and 10 Dai using LA1246 (left) and LA1670 (right) reference genomes. The light-blue peak on chromosome 1 corresponds to QTL 1-2-LA1246, detected in LA1246 but not in LA1670 due to differences in the detection p-value threshold (* in Table 3). The x-axis represents marker chromosomal positions and the y-axis shows local score values. The maximal (red line) and minimal (blue line) detection thresholds for QTLs with the local score method are indicated. B: Example of SNP detection failure using an alternative reference genome. QTL 7-1-LA1670 is detected using the LA1670 (right) but not identified using LA1246 (left) due to the absence of SNP detection in the exon. SNPs (dots) are positioned along the gene sequence, whose structure (exon in red, 3’ UTR in blue) is shown at the top of each graph. Arrows indicate the exon where the QTL is detected in LA1670. The y-axis represents -log10 (p-values). Random noise was added to the x-axis for vizualization purposes for all p-values in both conditions. * indicate the most significant SNPs. C: Structural variation between the LA1670 and LA1246 reference genomes. Two indels (genomic regions delimited by dotted lines) that may explain the absence of detection of QTL 4-1-LA1246 counterparts in LA1670. Conserved genomic regions are shown in light green and divergent regions in light grey. GWAS highlight temperature and time-dependent biologically relevant candidate genes Although similar numbers of QTL underlying « strict candidate genes » (8 vs 10) or « intergenic candidate genes (4 vs 6) were detected at 28 °C and 32 °C, no QTL was shared between the two conditions, indicating temperature-specific genetic control of the response. QTL detection also varied over time, with peaks appearing transiently across phenotyping days (S8 Fig). Most QTLs were detected over one or two consecutive days, with few persisting longer (up to four days only for IG12–1-BOTH at 32 °C), suggesting dynamic defense response at elevated temperature. At 28 °C, most QTLs were detected at the end of the kinetic (8/12 at 10 Dai), whereas QTLs detected at 32°C were distributed throughout the infection process. Three detection profiles could be defined: QTLs detected mainly at the beginning (3/16), throughout (3/16), or at the end (10/16) of the infection process. To test the hypothesis that defense - response - associated loci may be structure by environmental conditions, we examine allelic effects and phenottypic variance explained by three top SNPs represenative of three QTLs characterized by their temporal stability, reference genome specificity and temperature-dependent association patterns (Fig 5). The SNP SpimpiLA1670Chr12_2297876_A (QTL IG12–1-BOTH; detected at 32 °C, 5 Dai; Fig 5A) showed no clear association between allele state and geographic origin of accessions, suggesting an absence of local adaptation at this locus. In contrast, the two reference-specific SNPs SpimpiLA1670Chr07_95430_C (QTL 7–1-LA1670; identified at 32 °C, 4 Dai; Fig 5B) and SpimpiLA1246Chr03_47847419_G (QTL IG3–1-LA1246; detected at 28 °C, 10 Dai; Fig 5C) displayed structured distributions with specific genotypic states at these SNPs predominantly observed in accessions originating from contrasting climatic regions. Putative biome-dependent adaptation of accessions associated with allelic effects of three selected top SNPs underlying three distinct QTLs. For each SNP, allelic effects, phenotypic variance explained, and geographic distribution of alleles on a map of northwestern South America. A: SpimpiLA1670Chr12_2297876_A (QTL IG12-1-BOTH; at 32 °C, 5 Dai). B: SpimpiLA1670Chr07_95430_C (QTL 7-1-LA1670; at 32 °C, 4 Dai). C: SpimpiLA1246Chr03_47847419_G (QTL IG3-1-LA1246, 28 °C, 10 Dai). The geographical distribution of the panel was plotted using passport data (S2 Table) and the worldmap() function from the rnaturalearth [47] R package. Basemap data (country boundaries) were obtained from Natural Earth: https://www.naturalearthdata.com/downloads/. Terms of use: https://www.naturalearthdata.com/about/terms-of-use/. To further explore the biological relevance of the strongest associations, homology analyses were performed and expression profiles were investigated for « strict candidate genes » (Tables 3 and S9). Out of the 21 « strict candidate genes », information relative to their homolog expression profile in S. lycopersicum was retrieved for 19 genes, with 12 of them exhibiting upregulated expression in roots (9 strongly regulated), but also in other organs and different cell types. Five of them have already been described in previous studies (Solyc01g006660, Solyc01g105270, Solyc04g008310, Solyc09g009700, Solyc10g081260) as associated with plant immunity. Only eight high-confidence BLAST A. thaliana orthologs were identified, reflecting phylogenetic distance. Among them, five were also found to be expressed in roots and six had their expression regulated under biotic stress. Although functional data remain limited for several genes, overall patterns indicated enrichment in root-expressed, defense-related candidates responsive to pathogen challenge. Regarding “intergenic candidate genes,” six genes located near five QTLs, including IG12–1-BOTH detected between 4 and 7 Dai, are also reported to be expressed in the roots. Discussion Initially, our goal was not only to identify elevated temperature resistance but also to uncover genetic mechanisms that remain stable across temperatures, as many loci effective under standard conditions are negatively affected by heat stress [3] and some can be detected through GWAS [21,28,29]. To this end, we implemented a GWAS approach using a selected panel of wild tomatoes, with a higher representation of S. pimpinellifolium accessions, sequenced on this occasion, and including two internal reference genomes. To investigate the stability of resistance mechanisms to R. pseudosolanacearum effective at elevated temperatures and identify novel mechanisms, we studied the tomato/R. solanacearum pathosystem under two contrasted temperature conditions. While 28 °C is commonly used for phenotyping tomato genotypes in controlled conditions [52], we included 32 °C to represent an extreme warming scenario (SSP5-8.5; 4–5 °C increase by the end of the century; [4]) and because this temperature has been reported to compromise the defense response of the most tolerant Hawaii 7996 cultivar in the field [42]. High temperature influences the genetic architecture of tomato defense responses This study supports a strong impact of temperature on the genetic architecture of the plant defense response. The absence of shared QTLs between the two temperature conditions indicates that most loci are temperature-specific rather than temperature-stable, suggesting that robustness of plant defense may be a rare and complex trait. This interpretation is validated by the limited number of symptomless accessions observed, and is consistent with previous studies in A. thaliana, where resistance responses to bacterial infection were also highly temperature-dependent [21,53], with only a single accession remaining without symptoms at both 27 °C and 30 °C [21]. It supports the view that plant immunity is dynamic, depending on the temporal expression of bacterial pathogenic determinants during the infection progress, both being sensitive to environmental conditions, particularly temperature [28,38,54]. This conclusion is reinforced by the phenotypic diversity observed within the panel at elevated temperature. Indeed, we detected increased variability in infection dynamics, earlier and stronger heritability of the response to the bacterium, and an overall shift toward higher susceptibility at 32 °C for the majority of accessions. This pattern is confirmed by the low genetic correlation between conditions, indicating distinct genetic determinisms. This contrasts markedly with prior observations in A. thaliana, in which response variability was greater at 27 °C whereas most accessions became uniformly susceptible at 30 °C [21,28]. Such differences suggest that wild tomato relatives may harbor a broader repertoire of adaptive responses to combined biotic and heat stress. Finally, analysing the phenotypic responses at the endpoint of the infection kinetics revealed three different groups of accessions across temperatures: those exhibiting increased susceptibility at 32 °C, those showing reduced susceptibility and those with a stable response across conditions. This diversity in reaction norms supports the existence of diverse environment-specific genetic architectures underlying combined stress responses. Traces of local environmental adaptation in the panel Our results provide partial support for the hypothesis of biome-dependent adaptation of resistance mechanisms. Indeed, examination of allelic effects and phenotypic variance explained by three top SNPs corresponding to the three selected QTLs highlights contrasting geographic patterns. The QTL IG12–1-BOTH showed no clear association between allele distribution and geographic origin, suggesting a generalist effect whereby the locus contributes to resistance independently of the environmental or climatic context. In contrast, the LA1670-specific QTL 7–1-LA1670 exhibited a structured distribution, with heterozygous genotypes at the top SNP predominantly originating from equatorial warm and humid regions, consistent with a putative role in adaptation to high-temperature environments. The LA1246-specific QTL IG3–1-LA1246 displayed an opposite pattern, with the resistance-associated allele mainly found in accessions originating from cooler and drier environments, reflecting distinct selective pressures across climatic gradients. These patterns suggest that local environmental conditions may have contributed to shaping resistance to R. solanacearum, in agreement with the distribution of RSSC phylotype II strains in Central and South America [5,55]. A continuous diversity structured along geographic and climatic gradients Even if the panel spans a broad geographic range, population structure analyses revealed limited genetic subdivision, raising questions about species delimitation within S. pimpinellifolium and closely related taxa. This pattern, already described [46], may reflect ongoing gene flow, incomplete lineage sorting, and/or the impact of domestication-related introgressions. Rather than forming discrete genetic clusters, the observed diversity appears largely continuous and structured along geographic and climatic gradients, supporting potentially adaptive variation influenced by multiple environmental factors. Notably, the clustering of ‘Hawaii 7996’ with a subset of accessions including the reference LA1246 in the PCA may reflect introgression from wild relatives, as previously proposed [56], although uneven sampling of S. lycopersicum accessions could also contribute to this pattern. Beyond a single reference: capturing structural variation with two genomes A key methodological aspect of our study was the use of two internal reference genomes for GWAS. Most tomato GWAS used the ‘Heinz’ cultivar reference genome [57,58] whose domestication history has reduced genetic diversity, particularly in genomic regions associated with immunity [59]. Although pangenomes provide an ideal alternative for capturing structural variation [60], their construction remains computationally demanding [61]. Here, we adopted an intermediate strategy, using two reference genomes of S. pimpinellifolium accessions selected for their contrasting genetic and environmental backgrounds. Comparative analyses revealed high overall collinearity with the ‘Heinz’ reference and the presence of a large number of one-to-one orthologues, indicating strong genome-wide conservation. We also identified localized structural variations, including rearrangements on chromosomes 6, 7, and 8, as well as CNVs and gene presence/absence differences. In these variable regions, genes associated with cell division (LA1670) and transcriptional or metabolism regulation (LA1246) were significantly more frequent. Notably, genes related to stress responses and defense mechanisms were overrepresented in CNVs, corroborating previous observations in S. pimpinellifolium [62] and confirming their relevance for our study. While powerful, the multi-reference strategy introduces complications that must be taken into account, as previously reported in maize or fungi [63,64]. Some QTLs, such as QTL 7–1-LA1670, were detected in only one reference, putatively due to variants that could not be reliably called in the homologous region of the other genome. Small differences in kinship and population structure matrices influenced marker significance, affecting QTL detection. Additionally, structural variations altered local linkage disequilibrium patterns, impacting the detection of QTLs using local score methods. Nevertheless, this strategy enabled us to capture a broader spectrum of genetic variation than a single reference, increasing the likelihood of identifying functionally relevant loci and providing a more comprehensive view of the genetic architecture underlying tomato immunity under heat stress. Temperature-dependent temporal deployment of resistance Beyond these aspects, our analyses support a dynamic and context-dependent architecture of resistance, shaped by both temperature and infection stage. The identification of temperature-specific QTLs, combined with the identification of root-expressed candidate genes, suggests that plant responses to combined stress involve a temporally coordinated deployment of defense mechanisms. This is consistent with an emerging view in which pattern-triggered immunity (PTI) and effector-triggered immunity (ETI), the two layers of immune response induced upon pathogen attack in the zig-zag model [65], are interconnected and differ in their temporal dynamics [66]. Moreover, experimental evidence from the A. thaliana - Pseudomonas syringae pathosystem suggests that temperature modulates both pathogen virulence and plant immune responses. At permissive temperatures, which support optimal growth of A. thaliana, ETI is strongly activated, limiting bacterial proliferation. In contrast, higher temperatures tend to suppress ETI and favor PTI, leading to a defense response that is less specific but broader. These findings highlight how temperature fluctuations can shift the balance between distinct immune pathways, influencing both disease outcome and the effectiveness of plant defense [21,53]. In parallel, R. solanacearum infection progression from root invasion to vascular colonization was described to involve distinct host responses at each stage [54]. Together, our results suggest that temperature may impact the temporal deployment of these different immune responses. Interconnected pathways underlying tomato responses At the molecular level, the identified candidate genes point toward multiple and interconnected biological pathways. These include classical immunity-related processes [67–72], calcium signaling components involved in stress signal integration [73–75], receptor-like kinases mediating pathogen perception [76,77], and broader metabolic and regulatory pathways such as detoxification and protein turnover [78–80]. A final functional category contains remaining genes for which little functional information is currently available or whose link with immunity remains to be demonstrated. Our results align with previous studies that showed that QDR does not rely on a single class of genes but rather emerges from the integration of multiple physiological processes [25,66]. This complexity is likely amplified under combined stress conditions, as each specific stress combination may trigger distinct plant responses [3]. Importantly, our results also provide new insights into previously described major QTLs such as bwr-6 and bwr-12, which have been extensively studied for resistance to bacterial wilt. The identification of candidate genes in these regions, including receptor-like kinases and calcium-related signaling components, suggests potential mechanisms underlying these loci, although further validation is required. Limitations and next steps Although the candidate genes identified in this study appear relevant, several limitations should be considered. First, the use of two reference genomes, while improving variant discovery, does not fully capture the structural diversity present in the panel, leaving some genomic regions inaccessible to SNP calling. Second, GWAS results are sensitive to population structure and marker density. Even if we used high-quality reference genomes and dense SNP panels with short linkage disequilibrium, these factors cannot entirely eliminate confounding effects, and the continuous, gradient-like population structure observed in our wild tomato panel may have limited the power to detect some associations or inflated the detection of small-effect loci [81]. Third, the absence of shared QTLs between temperature conditions may also partly reflect limited statistical power or environmental variability, rather than entirely distinct genetic architectures. Finally, all candidate genes identified in this study remain putative. Therefore, future work will focus on the functional validation of selected candidate genes and the assessment of their stability across environments and genetic backgrounds, these validations being essential to evaluate their potential for breeding programs aimed at improving resistance to bacterial wilt under climate change. Beyond tomato, our findings have broader implications for crop improvement. As elevated temperatures are expected to alter plant-pathogen interactions across agroecosystems, identifying resistance mechanisms that are effective under heat stress is critical. In this context, the exploitation of natural diversity from wild relatives represents a promising strategy to uncover adaptive traits for future climates. Altogether, this study demonstrates that resistance to R. solanacearum is highly plastic and environmentally contingent, and highlights the value of integrating genomic diversity, ecological context, and environmental factors to better understand plant immunity. Materials and methods Biological material Tomato panel composition and R. pseudosolanacearum strain. An initial collection of 199 tomato accessions, available in the Tomato Genetic Resource Center (University of California, Davis campus, USA), was assembled, propagated by SYNGENTA and phenotyped. Of the original collection, a panel of 189 accessions was selected and sequenced. It includes 167 accessions of S. pimpinellifolium species and was further enriched by adding a smaller number of accessions from related species (20 of S. lycopersicum var. cerasiforme and one of S. cheesmaniae) in order to capture the domestication gradient and enhance ecological and genetic diversity. Compared to cultivated tomato, which exhibits limited intraspecific DNA variation, S. pimpinellifolium has the lowest estimated linkage disequilibrium (LD) [46], making it well suited for GWAS. Numerous QTLs associated with pathogen resistance have been identified in this species [31–38,82], and its genome has been reported to be enriched in genes related to biotic stress responses compared to S. lycopersicum [62]. The cultivated S. lycopersicum cv. ‘Hawaii7996’ was also included as a tolerant control for bacterial wilt [82]. S. pimpinellifolium species is considered one of the closest wild relatives of cultivated tomato [83], with which it can be easily crossed. It has also been proposed as the wild progenitor of S. lycopersicum var. cerasiforme, itself considered the most likely ancestor of modern large-fruited cultivated tomatoes [56]. Furthermore, climate data were retrieved from CHELSA (Climatologies at High Resolution for the Earth’s Land Surface Areas; https://chelsa-climate.org/, S10 Table). Accessions of the panel, originating from different regions of northwestern South America, likely experienced significant temperature variations, in some cases reaching up to 36,60 °C (Bio7, S10 Table), with temperatures at the collection sites ranging from -28.55 °C (BIO6, S10 Table) during the coldest months to 33.25 °C (BIO5, S10 Table) during the hottest months. BIO1 and BIO12 variables were extracted to characterize climatic conditions at each site. Passport data (S1 Table) for the accessions analyzed in this study include the accession identifier, species, country and province of origin, geographic coordinates (168 accessions) altitude (125 accessions) and BIO1/BIO2 of collection sites when available. The whole panel was inoculated using the wild type R. pseudosolanacearum GMI1000 strain classified by Hong et al. [84] as a phylotype I, race 1, biovar 3 strain. It originates from South America (French Guiana) [84] and was isolated in tomato [85]. Plant phenotyping conditions. For each accession, seeds were surface sterilized by immersion (5–25 seeds per 15 mL tube) in 5 mL of 20% (v/v) commercial bleach solution supplemented with 200 µL of TWEEN 20 for 10 min. Seeds were then rinsed four times with an equal volume of sterilized water and subsequently stratified in sterile water at 4°C for 48 h. Seeds were sown on rich BG solid medium [86] and allowed to germinate for 48 h at 28°C in the dark. Petri dishes were inverted during incubation to facilitate the recovery of germinated seeds. A maximum of two germinated seeds per accession were transplanted into pots (7 x 7 x 6.4 cm) filled with loam (Proveen potting soil Semi bouturage 2 with perlite, Gergy Fourniture, France). The plants were then grown for two weeks at the Toulouse Plant Microbe Phenotyping platform (INRAE-LIPME, Toulouse, France) under controlled conditions in a growth chamber [25°C day/ 24°C night, 75% RH, 12 h light, LED lights (intensity in μmol m − 2 s − 1 per range of wavelengths respectively at 100% Aria: 176; 400–700/ 1; 380–399/ 61; 400–499/ 20; 500–599/ 95; 600–699/ 4; 700–780/ blue, green and peak at 444, 597 and 661 nm)]. Plants were randomized at the beginning of the second week. Five days prior to inoculation, they were moved to a growth chamber in the quarantine laboratory for acclimatization under controlled conditions (28 °C day/ 27 °C night, 75% RH, 12 h light, 50 µmol m − 2 s − 1). Inoculations were carried out according to Morel et al. [52] by soil drenching, with 40 mL of a solution at 5 x107 colony-forming unit (CFU)/mL/pot. Immediately after R. pseudosolanacearum inoculation, plants were transferred to growth chambers maintained at either 28 °C/27 °C (day/night) or 32 °C/31 °C (day/night). Disease progression was assessed using an established 0–4 wilting scale previously described [52], based on the proportion of wilted leaves, with the score 0 (0% of leaves wilted) and 4 (100% of leaves wilted) corresponding to healthy and dead plants, respectively. Symptoms were recorded from two to ten days after inoculation (Dai). Tomato genomic resource creation and analysis Panel sequencing and reference genome assembly. All panel accessions were sequenced using Illumina technology by SYNGENTA (GENEWIZ, US), targeting a raw sequencing coverage of 15x to 25x per accession (S4 and S5 Tables). The LA1246 and LA1670 accessions, representative of the two major genetic groups identified in Blanca et al [46], were selected for the de novo reference genome assembly. They were sequenced with a PacBio Sequel Sequencer (Pacific Biosciences, Menlo Park, CA, USA) to achieve a minimum depth of 50x. We followed a meta-assembly strategy adapted from the Rosa genome [87]. Briefly, initial contigs were generated using CANU 1.8 [88], then scaffolded with Bionano optical maps (DLE-1 enzyme, Saphyr system) via the hybridScaffold.pl pipeline. Detailed protocols for DNA sample preparation, sequencing and post-sequencing processing are provided in (S1 Text, S2 and S3 Tables). Gene prediction and functional annotation. For both reference genomes, protein and non-protein coding gene models were predicted using the integrative EuGene pipeline (release 1.5; http://eugene.toulouse.inra.fr/Downloads/egnep-Linux-x86_64.1.5.tar.gz; [89]). Five protein databases were aligned with NCBI-BLASTx to detect translated regions. Protein coding genes were annotated, integrating data from all databases, depending on the accuracy of all sources of information. Priority was given to BLAST-P results from the “Solanum” database (S2 and S3 Tables). Comparative genomic analyses of reference genomes. Analyses were performed using GENESPACE version 1.2.3 [49] to investigate genome collinearity and orthologous relationships among tomato genomes. The analysis included three annotated genome assemblies: the cultivated tomato S. lycopersicum ‘Heinz’ reference genome (ITAG-4.1; [48]), LA1246 and LA1670 reference genomes generated in this study. Genome assemblies and their corresponding gene annotations (GFF3 format) and predicted peptide sequences (FASTA format) were used as input. Prior to synteny inference, gene annotations were processed using the GENESPACE preprocessing workflow to retain representative gene models and ensure consistent formatting across genomes. Synteny detection and block construction were performed using the GENESPACE pipeline and default parameters. Genome collinearity and structural variation were visualized using multi-genome dot plots and riparian plots generated within the GENESPACE framework (Figs 3, S2). All analyses were conducted within the R programming language environment using GENESPACE and its associated dependencies. Orthogroup analysis between LA1246 and LA1670 reference genomes. Comparative gene analyses between LA1246 and LA1670 reference genomes were performed using OrthoFinder v3.1.0 [90]. The coding sequences (CDSs) from each accession were used as datasets. Sequence alignment was performed using BLASTN with an identity threshold of 90%. The identified orthogroups were classified based on variations in the number of gene copies between the two accessions into the following classes: absent, one copy, or multiple copies in each genome (0–1, 0-n, 1–0, 1–1, 1-n, n-0, n-1, n-n) (S6 Table). Each orthogroup was annotated with Gene Ontology (GO) terms based on existing CDS annotations via Blast2GO, and enriched with parent terms from the Gene Ontology DAG for the Molecular Function and Biological Process categories. GO enrichment analysis by class was performed using topGO with Fisher’s exact test [91] (S7 Table). Variant calling and single nucleotide polymorphisms filtering. Variant calling was carried out using FreeBayes [92] and the two reference genomes. Single Nucleotide Polymorphism (SNP) matrices were filtered sequentially using PLINK1.07 [93] to retain high-quality, informative and independent markers for downstream GWAS (Table 1). Filter A corresponds to initial variant calls obtained from FreeBayes, while filter B was used to remove low-quality SNPs (minQ40), non-biallelic markers, or SNPs with >5% missing genotypes. SNPs with low allele frequency were excluded to ensure a sufficient representation for reliable association testing, retaining in filter C only SNPs with a minor allele frequency (MAF) ≥ 0.032 (corresponding to at least six to seven individuals) as described previously [94]. Then filter D corresponds to a linkage disequilibrium pruning (r² < 0.2), performed to reduce redundancy and false-positive associations by removing highly correlated SNPs. Since preliminary tests showed an overrepresentation of heterozygous SNPs at extreme GWAS p-values, they were excluded with filter E to prevent spurious associations, ensuring a balanced representation of allelic states. After filtering, 3,728,946 SNPs (LA1670) and 3,464,481 SNPs (LA1246) were retained for GWAS (Table 1). Kinship estimation and population structure. The kinship matrix was computed using the Balding-Nichols method [95] implemented in Efficient Mixed-Model Association eXpedited (EMMAX; [96]) on SNP matrices filtered for this purpose (Table 1, filters A-D). Population structure was analysed using PCA with gdsfmt [97] R package and its snpgdsPCA() function. Results are available in (S9 Fig). Additionally, hierarchical clustering was performed using the hclust function in R on kinship-derived distance matrices for both reference genomes, identifying K = 5 distinct clusters (Figs 1A and 1D, S10A). The choice of K was based on preliminary analysis that revealed over-fragmentation with alternative K values that could separate biologically coherent groups. To assess genetic differentiation among clusters, the fixation index (F ST; [98]) was calculated in 10 kilobase (kb) windows using vcftools on the GWAS SNP matrix (Table 1, filters A-C + E; S10B Fig). Using passport data, the geographical distribution of the panel was plotted on a map of northwestern South America with the worldmap() function from the rnaturalearth [47] R package. Basemap data (country boundaries) were obtained from Natural Earth (public domain): https://www.naturalearthdata.com/downloads/. Terms of use: https://www.naturalearthdata.com/about/terms-of-use/. Linkage disequilibrium analysis. The LD extent (r2 measure) was investigated in two complementary approaches. First, LD half-decay was computed in a three-step procedure: GWAS matrices were sub-sampled (10% of markers per chromosome), r2 was calculated for SNP pairs from 1 bp to 1 Mbp in distance and then the mean LD was summarized within 1 kb windows. Then, to visualize chromosome-wide LD patterns, a kinship-corrected LD was computed as described in [99] with LDcorSV [100] R package using a two-step procedure. For each chromosome, 1,000 SNPs were randomly sampled from GWAS matrices, then a kinship-corrected r2 was calculated for all pairwise comparisons, providing a corrected view of LD along chromosomes (S5 Fig). Natural variation of QDR in the tomato panel Experimental design. To ensure balanced genotype representation and enable variance partitioning of genetic and environmental effects, phenotyping was performed using a randomized complete block design (RCBD). Five independent blocks of 450 plants were established, each comprising two technical replicates. Each replicate consisted of nine trays of 25 plants (225 plants per replicate). Within each replicate, the 198 panel accessions were randomly distributed in the nine trays with one plant per accession per replicate. The tolerant control accession ‘Hawaii 7996’ [82] was included three times per tray at fixed positions (top-left, center, bottom-right), resulting in 27 control plants per replicate. Phenotypic modelization and broad-sense heritability estimation over the infection kinetic. For each temperature x Dai combination hereafter referred to as “condition”, phenotypic values were adjusted using model (1) where Yibr corresponds to the phenotype of the ith genotype in the rth technical replicate of the bth block, μ is the intercept, Gi~N(0,σG2) is the random genetic effect, Bb~N(0,σB2) is the random block effect, Iib~N(0,σI2) is the random block x genotype interaction effect and εibr~N(0,σϵ2) is the residual error. Random effects were assumed independent and normally distributed. Models were fitted by restricted maximum likelihood (REML) as implemented in the lme4 [101] R package. Broad-sense heritability was estimated applying the following formula (2) adapted from [102]: where nB is the number of blocks and nR the number of technical replicates. Conditions with H2 > 0.1 were retained for subsequent analyses (S11 Fig), to ensure that phenotypes values included sufficient genetic signal for association mapping. At 28 °C, 7 and 10 Dai conditions were kept whereas the 4, 5, 6, 7 and 10 Dai ones were selected at 32 °C. Earlier time points were excluded due to limited disease symptom development and consequently low genetic variance. For the retained conditions, accession-specific adjusted means (least-squares means; LS-means) were estimated using the lmerTest package in R [103] and used as phenotypic inputs for GWAS. For the models, fixed effects were tested by ANOVA, and random effects by likelihood ratio tests using the lrtest() function (Table 2). As 10 Dai was the time point at which the H² was the highest (Table 2, S11 Fig), we selected this time point to calculate the cross-temperature genetic correlations between accessions using Pearson’s correlation in order to assess the consistency of their responses across temperatures. The reaction norms were also calculated at that time point between the two temperatures on the five blocks to mitigate the potential delay in kinetics between individual blocks (S7 Fig). The three groups of phenotypic profiles observed across temperatures within the panel are presented in (Fig 2B). GWAS and QTL identification. GWAS was performed separately for each selected condition using model (3) described in [102] accounting for both population structure and relatedness: where Y is the phenotype vector; S is the SNP incidence matrix (-1,0,1 coding); τ is the fixed SNP effect; X is the structure matrix (corresponding to the first three principal component (PC) axes); β is the fixed structure effect; K is the kinship matrix; g~N(0,Kσg2) is the random genotypic effect vector and ε~N(0,σε2) is the residual error. The number of PCs was selected based on the part of the variance explained for both reference genomes (S9 Fig), with the first three PCs capturing 13.8% and 11.42% of the total variance for LA1670 and LA1246, respectively. Model (3) was fitted using EMMAX [96], applying the variance approximation [96] to cope with the large number of markers while maintaining appropriate control of type I error. Analyses were performed with the rrBLUP package [104]. To control False Positive Error Rate and refine QTL delineation, a local score approach was applied to SNP-level p-values [105]. This procedure aggregates statistical evidence across physically adjacent markers, allowing the identification of significant genomic intervals rather than single SNPs, thereby helping their biological interpretability. According to Bonhomme et al. [105] recommendations, the tuning parameter ξ, which controls the balance between sensitivity and specificity, was set to 3. Significant intervals were defined as putative QTLs. QTLs detected at consecutive time points and overlapping physically were merged to define consistent loci involved across infection kinetics. Underlying gene positions were retrieved from the reference genome annotations. Candidates were defined as “strict” for genes overlapping QTL intervals extended by +/-1000 bp to account for the promoter regions. Alternatively, when QTLs spanned regions outside annotated genes, the closest downstream and upstream genes were selected and defined as “intergenic candidates”. Candidate genes relevance. Because gene coordinates differed between the two reference genomes, candidate genes were reciprocally compared to assess their homology. Coding sequences were aligned against the alternate reference genome using BLASTN. Genes were considered homologous when they (i) mapped to the same chromosome; (ii) showed BLAST-N E-values close to zero, (iii) shared the same functional annotation; and (iv) displayed conserved genomic environments, defined by similar upstream and downstream genes. To assess the functional relevancy of candidate genes underlying QTLs, Gene Ontologies annotations and predicted biological functions were investigated. Coding sequences were extracted and queried by BLASTN via the Sol Genomics network platform (https://solgenomics.net/tools/blast/) against the tomato cDNA dataset from the International Tomato Genome Sequencing Project (ITAG 4.0, [48]). Orthologs in the A. thaliana genome were identified using the BLASTX tool on The Arabidopsis Information Resource website (https://www.arabidopsis.org/Blast/; TAIR), selecting hits with the lowest E-values. Spatiotemporal expression profiles were explored in A. thaliana and tomato using the ePlant tools available on the Bio-Analytic Resource for Plant Biology website (https://bar.utoronto.ca). To determine whether detected QTLs were shared or reference genome-specific and to assess potential structural variation, surrounding genomic environments, starting 1 kb upstream the start codon of the candidate genes were compared between reference genomes. Sequences, considering gene models (exons, 5’ UTR, 3’ UTR) for candidate genes, extracted from annotations of both reference genomes, were aligned using MultiAlin [106] after standardization of the start codon positions of candidate genes to zero with a custom script. Thirteen kb windows were visualized in Integrative Genomics Viewer [107]. Finally, SNP positions and corresponding p-values were added to the custom alignment script to assess SNP coverage equivalence, exon localization of top SNPs, and peak concordance between reference genomes. QTLs were named according to the chromosome number, the sequential order along the chromosome from the proximal end and the presence of a suffix that indicates whether the QTL was detected in both reference genomes (‘BOTH’) or was specific to one of them (pedigree indicated). ‘IG’ prefixes indicate QTLs classified as intergenic. A summary of all identified QTLs is provided in (Table 3), with detailed information available in (S9 Table). Comprehensive information on the top SNPs identified on the LA1670 and LA1246 reference genomes is provided in S11 and S12 Tables, respectively. Additionally, a dataset available via a DOI given in the data availability statement section enables visualization of the allelic effects associated with each SNP. Supporting information Supporting methods. (DOCX) Passport data for the accessions analyzed in this study. Accession identifier, species, country and province of origin, geographic coordinates (latitude and longitude) and altitude at collection sites were retrieved when available. CHELSA (https://chelsaclimate.org/) BIO1 and BIO12 values were extracted for each accession, when available, and used to characterize climatic conditions at each collection site. BIO1 (in °C) corresponds to mean annual temperature over the 1981–2010 period calculated as the average of mean monthly temperatures over the year. BIO12 (in mm) corresponds to the sum of monthly precipitation totals across the year during the 1981–2010 period. na: not available. (XLSX) Solanum pimpinellifolium reference accession LA1670 PacBio sequencing and analysis information. (XLSX) Solanum pimpinellifolium reference accession LA1246 PacBio sequencing and analysis information. (XLSX) Number of polymorphisms identified using LA1246 and mean sequencing depth per accessions. (XLSX) Number of polymorphisms identified using LA1670 and mean sequencing depth per accessions. (XLSX) Classification between LA1246 and LA1670 reference genomes of orthogroups by copy number and functional annotation. Orthogroups were classified into eight categories based on gene copy number variation between reference accessions LA1246 and LA1670: 0–1 (absent/one copy), 0-n (absent/multiple copies), 1–0 (one copy/absent), 1–1 (one copy/one copy), 1-n (one copy/multiple copies), n-0 (multiple copies/absent), n-1 (multiple copies/one copy), and n-n (multiple copies/multiple copies). Each orthogroup was annotated with Gene Ontology (GO) terms, categorized by biological process or molecular function. Comparative gene analyses between LA1246 and LA1670 reference genomes were performed using OrthoFinder v3.1.0 [90]. (XLSX) Gene Ontology (GO) enrichment analysis by orthogroup class. GO enrichment was performed for each genome and orthogroup class identified in S6 Table, using topGO with Fisher’s exact test [91]. (XLSX) Half-decay in base pairs for all chromosomes in both reference genomes. (XLSX) Detailed overview of the QTLs and corresponding candidate genes identified in this study. QTL name, the temperature condition (Tmp, °C) at which the QTL was detected, and the reference genome(s) in which the QTL was identified are indicated. An asterisk (*) indicates that the QTL underlying the candidate gene was detected with one reference genome but did not reach the detection p-value threshold with the other reference genome, although an association peak was observed (S8A-S8F Fig). The best S. lycopersicum homologous gene and corresponding annotation based on the ‘Heinz’ reference genome (ITAG4.0, 48), as well as a summary of known gene expression patterns (organs, cell types, and experimental conditions) retrieved from the ePlant tomato visualization tool (https://bar.utoronto.ca/eplant_tomato/) are provided. When available, the corresponding A. thaliana orthologous gene and annotation are reported. (?) indicates genes with low-confidence BLAST results. A summary of known gene expression patterns in A. thaliana (organs, cell types, and experimental conditions) retrieved from the ePlant visualization tool (https://bar.utoronto.ca/eplant/) is also given. Additional information includes the day after inoculation (Dai) at which the QTL was detected, gene IDs located within the identified QTL regions, the corresponding significant genomic regions (SigZones), and the top SNP names characterizing each QTL in both reference genomes (LA1246 and LA1670). na: not available. (XLSX) Climatologies at High Resolution for the Earth’s Land Surface Areas (CHELSA; https://chelsa-climate.org/) data were retrieved, when available, for each accession. Descriptions of all parameters are available on the website. (XLSX) Top SNPs identified along the infection kinetics by GWAS with the LA1670 reference genome. For each time point of the infection kinetics, the most significant SNPs for each QTL detected are reported, together with their minor allele frequency (MAF), the corresponding local score peak value, SNP association significance [−log10 (p-value)] and percentage of phenotypic variance explained (PVE). (XLSX) Top SNPs identified along the infection kinetics by GWAS with the LA1246 reference genome. For each time point of the infection kinetics, the most significant SNPs for each QTL detected are reported, together with their minor allele frequency (MAF), the corresponding local score peak value, SNP association significance [-log10 (p-value)] and percentage of phenotypic variance explained (PVE). (XLSX) Average climate data recorded monthly between 1990 and 2019, relative to the GPS coordinates of LA1246 (A: Loja; Ecuador; 3,99S 79,36W; elevation 1259m, Köppen climate classification: Cfb) and LA1670 (B: Sama Grande; Tacna; Peru; 17,833S 70,51W; elevation 452m; Köppen climate classification: Bwh). The red line corresponds to the average temperature recorded while the average amount of precipitation is represented by blue histograms. (https://climatecharts.net/; [48]). (TIF) Global overwiev of collinearity between the reference genomes (LA1246 and LA1670) used in this study and the commonly used refernce genome ‘Heinz’. Comparisons were performed between the de novo assembled genome of the ‘Heinz’ reference (ITAG4.1, [50]) and our two wild accessions reference genomes generated in this study, using GENESPACE for analyses and vizualization [49]. (TIF) Gene coverage by chromosome for each reference genome. Bar height indicates the mean number of SNPs per gene for each chromosome. The red line indicates the fraction of the gene containing at least one SNP. A: LA1246, B: LA1670. (TIF) Mean annual surface air temperature (A) and mean annual precipitation (B) at the sampling site, when available, during the 1981–2010 period reported on PCA results obtained with the LA1670 reference genome. Species distribution reported on PCA results obtained with the LA1670 (C) and the LA1246 (D) reference genomes. (TIF) Evolution of linkage disequilibrium (LD; r2) along chromosomes for both reference genomes. A: LA1246; B: LA1670. X-axis: positions on chromosomes are indicated in base pairs. Order of magnitude of half-LD is a few kb, except for chromosomes 5, 6 and 8. (TIF) LD (r2) between pairs of SNPs for chromosome 1 (A and B) and 8 (C and D). Raw LD is represented on the left (A and C) and LD corrected with kinship on the right (B and D). Small blocks of SNPs in LD appear on chromosome 1. However, large blocks appear on chromosome 8, which is consistent with previous results on the LD half-decay. (TIF) Effect of a 4 °C temperature elevation on plant response to R. pseudosolanacearum GMI1000 across the accession panel at 10 days after inoculation (Dai). Reaction norms showing the disease index (Di) at 10 Dai for each accession following inoculation at 28 °C and 32 °C. Accession-specific LS-means estimated from phenotypic values were used at each temperature. Four response profiles were defined according to the difference in disease index between 32 °C and 28 °C at 10 Dai (ΔDi = 32 °C-28 °C): (1) more susceptible at 32°C (ΔDi > 0.5); (2) less susceptible at 32°C (ΔDi < -0.5); (3) similar response at both temperatures (−0.5 < ΔDi < 0.5); and (4) symptomless accessions (−0.5 < ΔDi < 0.5 and Di 28 °C < 0.2 and DI 32 °C < 0.2), representing a subset of profile (3). (TIF) Manhattan plots, their related local score and QQ-plots. For each panel (A-F), data on the left correspond to results obtained with the LA1246 reference genome while data on the right correspond to LA1670. Manhattan plots obtained with the local score method (ξ = 3) are shown at the top of each panel. Red and blue lines represent the maximal and the minimal significance thresholds computed per chromosome with autocorrelation between SNPs. The Manhattan plots (with p-values; the dotted line represents the Bonferroni threshold) are displayed in the middle of each panel, and the corresponding QQ plots (with p-values) are displayed at the bottom. Each panel corresponds to a temperature condition and a Dai: A: 28 °C-Dai7; B: 28 °C-Dai10; C: 32 °C-Dai4; D: 32 °C-Dai5; E: 32 °C-Dai6; F: 32 °C-Dai7. Manhattan plots obtained with local score method at 32 °C and 10 Dai are shown in Fig 4. (TIF) Part of variance explained by PCA axes for both reference genomes. A: LA1246. B: LA1670. A slight decrease is observed when the number of axes exceeds three. (TIF) Hierarchical clustering of the panel and fixation index (F ST) within identified clusters based on the LA1670 reference genome Kinship. A: Kinship matrix and representation of population structure: individuals were ordered following their clustering based on PCA axes to allow blocks corresponding to population structure to appear. B: F ST was computed using vcftools (using a window size of 10 kb on the GWAS SNP matrix). Then the weighted means (by the number of markers within these windows) were computed between all combinations of the two clusters. Results are consistent between the two figures: high kinship coefficients between individuals of two different clusters lead to small F-ST values, indicating that clusters have similar genetic diversity (for example clusters 4–5). (TIF) Heritability (H2) calculated per day after infection for experiments performed at 28 °C and 32 °C. Blue bars indicate heritability for the 28 °C experiment and orange bars for the 32 °C one. (TIF) 9 Oct 2025 PGENETICS-D-25-00847 New resistance to bacterial wilt in heat-stressed tomato is revealed by two-reference Genome Wide Association PLOS Genetics Dear Dr. Berthomé, Thank you for submitting your manuscript to PLOS Genetics. After careful consideration, we feel that it has merit but does not fully meet PLOS Genetics's publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. Please submit your revised manuscript within 60 days Dec 08 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosgenetics@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pgenetics/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. Please include the following items when submitting your revised manuscript: * A rebuttal letter that responds to each point raised by the editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. This file does not need to include responses to any formatting updates and technical items listed in the 'Journal Requirements' section below. * A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. * An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, competing interests statement, or data availability statement, please make these updates within the submission form at the time of resubmission. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. We look forward to receiving your revised manuscript. Kind regards, Tatiana Giraud Academic Editor PLOS Genetics Angela Hancock Section Editor PLOS Genetics Aimée Dudley Editor-in-Chief PLOS Genetics Anne Goriely Editor-in-Chief PLOS Genetics Additional Editor Comments: We have now received three referee reports for your manuscript on the identification of loci for resistance to bacterial wilt in heat-stressed tomato by GWAS. All referees found the manuscript valuable and original; however, concerns were raised, in particular about the lack of clarity in the experimental design, the choice of accessions, statistical analyses, and data presentation, as well as lack of discussion of the study limitations. In addition, the referees suggest to tone down claims about gene functions, the existence of thermostable resistance and the utility of the two reference genomes, highlighting that the inference about so called "thermostable resistance" should actually be the opposite. They also recommend to discuss the findings in relationship with the accession origins, to better present the QTL results, with additional figures, and to perform a more comprehensive comparison of the two reference genomes. I agree that the study would be a nice addition to PLoS Genetics but that the manuscript needs to be clarified and claims toned down. In addition, it would be better to begin the abstract and the introduction by general statements on the scientific topics rather than on specificities about the model organisms. Also, there are too many abbreviations, this renders the text harder to read without providing benefits. Please include page numbers. The analysis of population structure needs improvements and there are awkward statements on population genetics, see below some examples. I would encourage resubmission if you are able to revise the manuscript along these lines. The referees also provide a list of excellent additional suggestions and questions, which should also be addressed. More specific comments: -L87: define PAMP -P11-12: there are too many abbreviations and specific details, this renders the text harder to read -L133-134; L166-167: unclear what you mean -P14: unclear if these "wild relatives" all belong to a single species? a single population? -L168: cut the sentence into two -L375-7: the species involved should be clear from the introduction; the text mentions 5 species but only four are listed; why including different species, especially the two with a single accession? -L386-390: unclear, two references are said to be needed because one species from different clusters, but what about the other species? The rationale for choosing different environments for a reference is unclear. -Figure 1: unclear if these includes the five species and the correspondence with the clusters; the species delimitation could even be discussed given the lack of clear genetic subdivision; the figure and the text are unclear about how the clusters were delimited. Analyses such as STRCUTRE and Splitstree should be run. The delimitation into these 5 clusters is not convincing at all based on the PCAs. -Figure 2: not sure this is needed as a main figure -L431: it does change the separation of the five so called clusters -L433: the term diversified is not correct here; give numbers of differences -L436: you cannot say this without proper analysis and without convincingly showing genetic subdivision in the first place; the PCA lets rather think of a lack of clear genetic subdivision -L439: the term "share" is not correct here, FST is a value for a pair -Define all abbreviations in all figure legends -L455-456: this is just the disease progressing, not susceptibility increasing -L488: high, not important -P33: a more comprehensive analysis of structural variants and better representation are needed -L588: not exactly, the variation in the sample for the response to the disease was different at the two temperatures, but there may be a shared regulation that is not variable among accession -L603: delete "interestingly"; based on what evidence was the previous annotation? -L608: what does "particularly expressed » means? upregulated ? -L632 : « assumption » is not the right term here, it is more a « hypothesis » -L634 : unclear ; what is accentuated ? -Do not refer to figures or Tables in the discussion, they should have explained well enough in the result section -L637 : who is « they » ? -L638, L674, L729: delete "interestingly"; -L639-643: this is a bit misleading, the results do not show well defined clusters in the response to disease, it looks more like a continuous trait homogeneous, saying it has lost diversity does not make sense, especially when using a single genome as a reference anyway -L650: reduction in what? -L614, L662: your question is unclear: do you look for resistance at high temperature, or at both temperature? This is different: clarify and discuss the results given the previse question -L664: why is geographical distance important? -L672: what do you mean by unique here? -L671, 713, 736, 762, 771, 781: remove bold -The discussion should end by broadening the context to other crops Journal Requirements: If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise. 1) Please upload all main figures as separate Figure files in .tif or .eps format. For more information about how to convert and format your figure files please see our guidelines: https://journals.plos.org/plosgenetics/s/figures 2) We have noticed that you have cited Table 2 in the manuscript file but there is no corresponding table in the manuscript. Please amend your manuscript to include this table noting that tables should not be uploaded as individual files. 3) We have noticed that you have a list of Supporting Information legends in your manuscript. However, there are no corresponding files uploaded to the submission. Please upload them as separate files with the item type 'Supporting Information'. 4) Some material included in your submission may be copyrighted. According to PLOSu2019s copyright policy, authors who use figures or other material (e.g., graphics, clipart, maps) from another author or copyright holder must demonstrate or obtain permission to publish this material under the Creative Commons Attribution 4.0 International (CC BY 4.0) License used by PLOS journals. Please closely review the details of PLOSu2019s copyright requirements here: PLOS Licenses and Copyright. If you need to request permissions from a copyright holder, you may use PLOS's Copyright Content Permission form. Please respond directly to this email and provide any known details concerning your material's license terms and permissions required for reuse, even if you have not yet obtained copyright permissions or are unsure of your material's copyright compatibility. Once you have responded and addressed all other outstanding technical requirements, you may resubmit your manuscript within Editorial Manager. Potential Copyright Issues: i) Figure 1. Please (a) provide a direct link to the base layer of the map (i.e., the country or region border shape) and ensure this is also included in the figure legend; and (b) provide a link to the terms of use / license information for the base layer image or shapefile. We cannot publish proprietary or copyrighted maps (e.g. Google Maps, Mapquest) and the terms of use for your map base layer must be compatible with our CC BY 4.0 license. Note: if you created the map in a software program like R or ArcGIS, please locate and indicate the source of the basemap shapefile onto which data has been plotted. If your map was obtained from a copyrighted source please amend the figure so that the base map used is from an openly available source. Alternatively, please provide explicit written permission from the copyright holder granting you the right to publish the material under our CC BY 4.0 license. If you are unsure whether you can use a map or not, please do reach out and we will be able to help you. The following websites are good examples of where you can source open access or public domain maps: * U.S. Geological Survey (USGS) - All maps are in the public domain. (http://www.usgs.gov) * PlaniGlobe - All maps are published under a Creative Commons license so please cite u201cPlaniGlobe, http://www.planiglobe.com, CC BY 2.0u201d in the image credit after the caption. (http://www.planiglobe.com/?lang=enl) * Natural Earth - All maps are public domain. (http://www.naturalearthdata.com/about/terms-of-use/). 5) Please amend your detailed Financial Disclosure statement. This is published with the article. It must therefore be completed in full sentences and contain the exact wording you wish to be published. 1) State the initials, alongside each funding source, of each author to receive each grant. For example: "This work was supported by the National Institutes of Health (####### to AM; ###### to CJ) and the National Science Foundation (###### to AM)." 2) State what role the funders took in the study. If the funders had no role in your study, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript." 3) If any authors received a salary from any of your funders, please state which authors and which funders.. If you did not receive any funding for this study, please simply state: u201cThe authors received no specific funding for this work.u201d 6) Kindly revise your competing statement in the online submission form to align with the journal's style guidelines: 'The authors declare that there are no competing interests.' Reviewers' comments: Reviewer's Responses to Questions Comments to the Authors: Please note here if the review is uploaded as an attachment. Reviewer #1: Summary and significance: This manuscript presents an ambitious GWAS of quantitative disease resistance (QDR) to Ralstonia solanacearumspecies complex in a panel of 189 wild tomato accessions. By phenotyping across two temperature regimes (28 °C vs 32 °C) and leveraging two de novo reference genomes (LA1246 and LA1670), the authors aim to identify candidate loci that may contribute to thermostable resistance under climate change. The dual-reference approach is innovative and has clear potential to complement pangenome efforts, and the daily scoring design provides temporal resolution rarely attempted in plant–pathogen GWAS. The work generates a valuable resource and important hypotheses. That said, several aspects of the experimental design, statistical analysis, and data presentation require clarification or revision before the claims can be fully supported. My comments are divided into Major and Minor issues. Major comments: 1. The authors emphasize thermostable resistance, yet no QTL were common to 28 °C and 32 °C. How do they justify the term “thermostable candidate loci” (see Abstract line ~57, Author summary) As field temperatures fluctuate below 28 °C and above 32 °C, it is unclear how results restricted to condition-specific QTL translate to breeding for stable resistance. Please clarify this interpretation. 2. Although the Results mention cases where SNPs could not be detected in LA1246, the Discussion does not explicitly address limitations. Please add a dedicated limitations paragraph. 3. In the Author summary, the authors state that some candidate loci are “currently undergoing functional validation.” If this is work planned by the authors, please move this point to a future directions statement in the Discussion. 4. The Introduction (lines ~152) states that optimal growth for tomato is 13–33 °C, yet 32 °C is used as the “heat stress” condition. Evidence is needed that the specific accessions studied have optimal growth below 32 °C. For example, a map of mean annual temperatures at accession origins would substantiate that 32 °C imposes stress relative to their natural ranges. 5. The phenotyping protocol (Materials & Methods) describes seed sterilization and transplanting but does not state clearly how many biological replicates per accession were used. 6. The inoculation method references Morel et al. (49) but does not specify whether root-dip, soil-drench, or stem injection was used. Please clarify. And the 0–4 disease index scale is not explained. A brief description of what each score corresponds to should be added. 7. The manuscript shows a Bonferroni line in Manhattan plots but does not explain the correction method. 8. Significant SNPs’ p-values are not reported in supplementary tables. 9. Given >3 million SNPs, Bonferroni is extremely strict (p ≈ 1.3 × 10⁻⁸). Consider also reporting FDR thresholds. 10. MAF was set to 0.032 (≈12 minor alleles in 189 diploids). This is low and risks unstable estimates. Please show the formula cited, explain why this cutoff was chosen. 11. Kinship estimation used filters A+B+C+D while GWAS used A+B+C+E. Please explain the rationale for using different SNP sets for kinship vs association. 12. Three PCs were included in GWAS; please report the variance explained by these PCs. 13. In the LA1246-based PCA, LA1246 and ‘Hawaii 7996’ cluster together. Is this due to introgression from S. pimpinellifolium into ‘Hawaii 7996’ or lack of sampling from S. lycopersicum? Please discuss. 14. Please provide a supplementary table with ancestry proportions of the five clusters, and consider comparing phenotypes across clusters to relate genetic diversity to resistance. Minor comments: o “Solanaceaous” → Solanaceae. o Lines 86–93 (extended PTI vs ETI zig-zag model explanation) feel redundant and distract from the main story. Consider shortening. o T3E is introduced at line 113 without prior definition. Please define “type III effectors (T3E)” when first mentioned (~line 103). o Example: line 98 cites (3,2); line 108 cites (17,14). Please fix order and format. o Figure 1 legend: clarify “A–C” should read “A & C” and “B–D” should read “B & D” to match panels. Also, each figure should indicate clearly which reference genome is used. o Figure 3: the three response groups should be more clearly distinguished; adding representative plant images would strengthen the figure. o Figure 5: Panel A y-axis should be labeled “local score.” Panel B legend refers to arrows that are not visible; please correct. o Table 3: add a column with p-values for candidate SNPs/QTL. o Results line 375: the panel included four species, not five (167 S. pimpinellifolium, 20 S. lycopersicum var. cerasiforme, one S. cheesmaniae, one S. lycopersicum cultivar). Please correct. o Please state whether any non-inoculated wild accessions were grown at both temperatures to confirm symptoms were disease-related and not physiological stress. o “GWA” (line 483) should be written consistently as “GWAS.” Overall assessment: This is an important and timely study. The innovative use of two panel-derived reference genomes to improve QTL detection is a real strength. However, clarification of experimental replication, inoculation methods, statistical thresholds, and explicit acknowledgment of limitations will be required to ensure reproducibility and to temper the claims about thermostable resistance. With these revisions, the manuscript will be a valuable contribution to the field. Reviewer #2: This study presents a high-quality genomic and phenotypic analysis of 189 wild tomato accessions, combining two newly sequenced reference genomes with a temperature-controlled GWAS approach to investigate resistance to R. solanacearum. The work reveals substantial natural variation in bacterial wilt responses, identifies temperature- and time-dependent QTLs, and highlights candidate genes with potential roles in immunity, calcium signaling, and receptor-mediated defense. Whreas true validation of the results is lacking, overall, the study provides a valuable genomic resource and a well-characterized panel for uncovering thermostable resistance mechanisms in tomato. I only have some minor comments: -Not sure why the data set was reduced from 199 to 189 -Percentages of the altitude of origin don’t add up. -It is extremely unlikely to find Pimpinellifolium or Cerasiformae above 3000 m. Please check your records again. -You first report 21 candidates, later only 19 homologs? -Can you authors explain a bit more why it is not weird that there was no overlap between QTL? Only 3 degrees difference? -Did the authors test whether the plants were stressed beforeand like by measuring Fv/Fm or other measures of respiratory fitness? -Whereas the study is useful, claiming using two reference for different GWAS is not the same as offering alternatives for Pangenomes, the relative sections should be tuned down -The functional claims of the GWAS candidates are solely based on Arabidopsis annotations. This should be properly placed in context without overstatement - The discussion could be tuned down. There were no actually validated candidate genes. I also don’t think that is needed for this study, but until real validation happens… I also think the time and temperature dependent results should be discussed a bit more in light of possible false positives or why the authors think all these loci are real. Reviewer #3: The review is in attachment ********** Have all data underlying the figures and results presented in the manuscript been provided? Large-scale datasets should be made available via a public repository as described in the PLOS Genetics data availability policy, and numerical data that underlies graphs or summary statistics should be provided in spreadsheet form as supporting information. Reviewer #1: No: The SNPs P-value from GWAS study is missing Reviewer #2: None Reviewer #3: Yes ********** PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy. Reviewer #1: No Reviewer #2: No Reviewer #3: No [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] Figure resubmission: While revising your submission, we strongly recommend that you use PLOS’s NAAS tool (https://ngplosjournals.pagemajik.ai/artanalysis) to test your figure files. NAAS can convert your figure files to the TIFF file type and meet basic requirements (such as print size, resolution), or provide you with a report on issues that do not meet our requirements and that NAAS cannot fix. After uploading your figures to PLOS’s NAAS tool - https://ngplosjournals.pagemajik.ai/artanalysis, NAAS will process the files provided and display the results in the "Uploaded Files" section of the page as the processing is complete. If the uploaded figures meet our requirements (or NAAS is able to fix the files to meet our requirements), the figure will be marked as "fixed" above. If NAAS is unable to fix the files, a red "failed" label will appear above. When NAAS has confirmed that the figure files meet our requirements, please download the file via the download option, and include these NAAS processed figure files when submitting your revised manuscript. Reproducibility: To enhance the reproducibility of your results, we recommend that authors of applicable studies deposit laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols Submitted filename: PGENETICS-D-25-00847_review.docx Submitted filename: PGENETICS-D-25-00847_reviewer.pdf 15 Apr 2026 Submitted filename: Responses to Reviewers.docx 13 May 2026 PGENETICS-D-25-00847R1 New biologically relevant resistance to bacterial wilt in heat-stressed tomato revealed by two-reference Genome Wide Association PLOS Genetics Dear Dr. Berthomé, Thank you for submitting your manuscript to PLOS Genetics. After careful consideration, we feel that it has merit but does not fully meet PLOS Genetics's publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. Please submit your revised manuscript within by Jun 12 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosgenetics@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pgenetics/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. Please include the following items when submitting your revised manuscript: * A letter that responds to each point raised by the editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. This file does not need to include responses to formatting updates and technical items listed in the 'Journal Requirements' section below. * A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. * An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, competing interests statement, or data availability statement, please make these updates within the submission form at the time of resubmission. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. As the corresponding author, your ORCID iD is verified in the submission system and will appear in the published article. PLOS supports the use of ORCID, and we encourage all coauthors to register for an ORCID iD and use it as well. Please encourage your coauthors to verify their ORCID iD within the submission system before final acceptance, as unverified ORCID iDs will not appear in the published article. Only the individual author can complete the verification step; PLOS staff cannot verify ORCID iDs on behalf of authors. We look forward to receiving your revised manuscript. Kind regards, Tatiana Giraud Academic Editor PLOS Genetics Aimée Dudley Editor-in-Chief PLOS Genetics Anne Goriely Editor-in-Chief PLOS Genetics Additional Editor Comments: I was pleased to see that this manuscript has been carefully revised. The referees also found the revisions overall satisfactory, but they had additional suggestions left. Other than that, I congratulate the authors for this nice piece of work that will be a great contribution to the journal. Journal Requirements: If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise. 1) Your manuscript's sections are not in the correct order. Please amend to the following order: Abstract, Introduction, Results, Discussion, and Methods. 2) ) Thank you for including the following in your data availability statement: 'The two reference genomes sequenced and annotated in this study have been deposited in the GenBank database under the accession numbers JASTWH000000000 (LA1246) and JASTWI000000000 (LA1670).' However, the accession number provided is currently inaccessible. Please verify and provide an active link to facilitate access to the data. Reviewers' comments: Reviewer's Responses to Questions Comments to the Authors: Please note here if the review is uploaded as an attachment. Reviewer #1: The manuscript presents valuable results and makes a meaningful contribution to our understanding of plant–pathogen interactions in tomato, an important model crop. The authors have also responded to the reviewers’ comments fairly. My recommendation is acceptance with minor revisions. My remaining comments are listed below: 1. Introduction, paragraph 3: This paragraph would benefit from revision for clarity and logical flow. At present, it includes many details about Arabidopsis defense mechanisms, which seem somewhat unnecessary in the context of this manuscript. More importantly, the main purpose of the paragraph is not clear. The reader may not understand whether the paragraph is intended to introduce QTLs/resistance genes or to justify the use of GWAS as an appropriate approach. Please consider rewriting this paragraph with a clearer focus. 2. Line 166: The text refers to “both temperatures,” but only 28°C appears to have been introduced in the paragraph. Please clarify or revise. 3. Table S1 and line 192: In Table S1, the BIO6 column lists the lowest temperature as -28°C, whereas line 192 states -8°C. Please clarify which value is correct. 4. Line 320: Please clarify whether the correct number of accessions is 189 or 198. 5. Line 365: Please specify what percentage of variation was considered. 6. Line 413: Please remove the extra “S8.” 7. Table 3: The temperature columns appear to be redundant, since panels A and B already correspond to specific temperatures. Please consider removing these columns or revising the table format. 8. Organization of figures and tables: In general, figures should be introduced in sequential order in the manuscript. However, the Results section begins by discussing Figure 3. Please consider reordering the figures and tables accordingly. In addition, Figure 5 is introduced for the first time in the Discussion; it should be presented and discussed in the Results section first. 9. Lines 614–626: Please add appropriate supporting citations, figures, or tables for the statements made in this section. 10. Figure 3, panels E and F: I am not sure whether panels E and F add substantial information to the figure. Since temperature and precipitation variation do not appear to cluster according to population structure, these panels seem somewhat redundant. Please consider moving them to the supplementary materials. 11. Figure legends: Please use a consistent format for introducing and describing figure panels across all figure legends. Overall, the manuscript is valuable and suitable for publication after these minor revisions are addressed. Reviewer #2: The manuscript is much improved. The new discussion nicely incorporates comments from all reveiwers and editors. It looks like the genbank numbers have not been updated. Similarly the links to the dois in the data statement are not active. This should be done before the paper can be accepted. l450, percentages still dont add up to 100%. Does that mean the rest has missing data? It may be that the altitude data are wrong in the database and for the conclusions of the manuscript it probably makes little difference, but the authors should mark this in their meta tables and not propagate error. Species ranges for pimpi en cerasiformae don't reach above 1500m. Reviewer #3: Thank you to the authors for their hard work on this update. All requested modifications have been successfully integrated, and I am pleased to accept the document in its current form. ********** Have all data underlying the figures and results presented in the manuscript been provided? Large-scale datasets should be made available via a public repository as described in the PLOS Genetics data availability policy, and numerical data that underlies graphs or summary statistics should be provided in spreadsheet form as supporting information. Reviewer #1: Yes Reviewer #2: No: Reviewer #3: Yes ********** PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy. Reviewer #1: No Reviewer #2: No Reviewer #3: No [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] Figure resubmission: While revising your submission, we strongly recommend that you use PLOS’s NAAS tool (https://ngplosjournals.pagemajik.ai/artanalysis) to test your figure files. NAAS can convert your figure files to the TIFF file type and meet basic requirements (such as print size, resolution), or provide you with a report on issues that do not meet our requirements and that NAAS cannot fix. After uploading your figures to PLOS’s NAAS tool - https://ngplosjournals.pagemajik.ai/artanalysis, NAAS will process the files provided and display the results in the "Uploaded Files" section of the page as the processing is complete. If the uploaded figures meet our requirements (or NAAS is able to fix the files to meet our requirements), the figure will be marked as "fixed" above. If NAAS is unable to fix the files, a red "failed" label will appear above. When NAAS has confirmed that the figure files meet our requirements, please download the file via the download option, and include these NAAS processed figure files when submitting your revised manuscript. Reproducibility: To enhance the reproducibility of your results, we recommend that authors deposit laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols 12 Jun 2026 Submitted filename: Responses to Reviewers_second revision.docx 17 Jun 2026 Dear Dr Berthomé, We are pleased to inform you that your manuscript entitled "New biologically relevant resistance to bacterial wilt in heat-stressed tomato revealed by two-reference Genome Wide Association " has been editorially accepted for publication in PLOS Genetics. Congratulations! Before your submission can be formally accepted and sent to production you will need to complete our formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Please note: the accept date on your published article will reflect the date of this provisional acceptance, but your manuscript will not be scheduled for publication until the required changes have been made. Once your paper is formally accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you’ve already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosgenetics@plos.org. In the meantime, please log into Editorial Manager at https://www.editorialmanager.com/pgenetics/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production and billing process. Note that PLOS requires an ORCID iD for all corresponding authors. Therefore, please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. If you have a press-related query, or would like to know about making your underlying data available (as you will be aware, this is required for publication), please see the end of this email. If your institution or institutions have a press office, please notify them about your upcoming article at this point, to enable them to help maximise its impact. Inform journal staff as soon as possible if you are preparing a press release for your article and need a publication date. Thank you again for supporting open-access publishing; we are looking forward to publishing your work in PLOS Genetics! Yours sincerely, Tatiana Giraud Academic Editor PLOS Genetics Angela Hancock Section Editor PLOS Genetics Aimée Dudley Editor-in-Chief PLOS Genetics Anne Goriely Editor-in-Chief PLOS Genetics www.plosgenetics.org BlueSky: @plos.bsky.social ---------------------------------------------------- Comments from the reviewers (if applicable): ---------------------------------------------------- Data Deposition If you have submitted a Research Article or Front Matter that has associated data that are not suitable for deposition in a subject-specific public repository (such as GenBank or ArrayExpress), one way to make that data available is to deposit it in the Dryad Digital Repository. As you may recall, we ask all authors to agree to make data available; this is one way to achieve that. A full list of recommended repositories can be found on our website. The following link will take you to the Dryad record for your article, so you won't have to re‐enter its bibliographic information, and can upload your files directly: http://datadryad.org/submit?journalID=pgenetics&manu=PGENETICS-D-25-00847R2 More information about depositing data in Dryad is available at http://www.datadryad.org/depositing. If you experience any difficulties in submitting your data, please contact help@datadryad.org for support. Additionally, please be aware that our data availability policy requires that all numerical data underlying display items are included with the submission, and you will need to provide this before we can formally accept your manuscript, if not already present. ---------------------------------------------------- Press Queries If you or your institution will be preparing press materials for this manuscript, or if you need to know your paper's publication date for media purposes, please inform the journal staff as soon as possible so that your submission can be scheduled accordingly. Your manuscript will remain under a strict press embargo until the publication date and time. This means an early version of your manuscript will not be published ahead of your final version. PLOS Genetics may also choose to issue a press release for your article. If there's anything the journal should know or you'd like more information, please get in touch via plosgenetics@plos.org. PGENETICS-D-25-00847R2 New biologically relevant resistance to bacterial wilt in heat-stressed tomato revealed by two-reference Genome Wide Association Dear Dr Berthomé, We are pleased to inform you that your manuscript entitled " New biologically relevant resistance to bacterial wilt in heat-stressed tomato revealed by two-reference Genome Wide Association " has been formally accepted for publication in PLOS Genetics! Your manuscript is now with our production department and you will be notified of the publication date in due course. The corresponding author will soon be receiving a typeset proof for review, to ensure errors have not been introduced during production. Please review the PDF proof of your manuscript carefully, as this is the last chance to correct any errors. Please note that major changes, or those which affect the scientific understanding of the work, will likely cause delays to the publication date of your manuscript. Soon after your final files are uploaded, unless you have opted out or your manuscript is a front-matter piece, the early version of your manuscript will be published online. The date of the early version will be your article's publication date. The final article will be published to the same URL, and all versions of the paper will be accessible to readers. For Research Articles, you will receive an invoice from PLOS for your publication fee after your manuscript has reached the completed accept phase. If you receive an email requesting payment before acceptance or for any other service, this may be a phishing scheme. Learn how to identify phishing emails and protect your accounts at https://explore.plos.org/phishing. Thank you again for supporting PLOS Genetics and open-access publishing. We are looking forward to publishing your work! With kind regards, Sharmila Kamatchi PLOS Genetics On behalf of: The PLOS Genetics Team Carlyle House, Carlyle Road, Cambridge CB4 3DN | United Kingdom plosgenetics@plos.org | +44 (0) 1223-442823 plosgenetics.org | Twitter: @PLOSGenetics