This is the accepted version of the article: Mayobre, Carlos; Pereira, Lara; Eltahiri, Abdelali; [et al.]. «Genetic dis- section of aroma biosynthesis in melon and its relationship with climacteric ripening». Food Chemistry, Vol. 353 (August 2021), art. 129484. DOI 10.1016/j.foodchem.2021.129484 This version is available at https://ddd.uab.cat/record/238270 under the terms of the license 1 Genetic dissection of aroma biosynthesis in melon and its 2 relationship with climacteric ripening 3 4 Carlos Mayobre1$, Lara Pereira1,2$, Abdelali Eltahiri1, Einat Bar3, Efraim Lewinsohn3, Jordi 5 Garcia-Mas1,4*, Marta Pujol1,4* 6 1 7 Centre for Research in Agricultural Genomics (CRAG) CSIC-IRTA-UAB-UB, Edifici CRAG, 8 Campus UAB, 08193 Bellaterra, Barcelona, Spain. 2 9 Present address. Center for Applied Genetic Technologies, Genetics & Genomics, University 10 of Georgia, Athens, GA 30602, U.S. 3 11 Department of Vegetable Crops, Newe Ya’ar Research Center, Agricultural Research 12 Organization, Volcani Center, Ramat Yishay, Israel 4 13 Institut de Recerca i Tecnologia Agroalimentàries (IRTA), Edifici CRAG, Campus UAB, 08193 14 Bellaterra, Barcelona, Spain. 15 $ 16 These authors contributed equally to this work *Corresponding 17 authors addresses: 18 Jordi Garcia-Mas, jordi.garcia@irta.cat Phone: +34 935636600 19 Marta Pujol, marta.pujol@irta.cat Phone: +34 935636600 20 21 1 22 Highlights 23 • Two melon commercial varieties produced 88 different VOCs. 24 • Melon rind and flesh contributed differently to fruit aroma. 25 • ETHQV8.1, a climacteric ripening QTL, has an impact in the biosynthesis of VOCs. 26 • Candidate genes and QTLs have been identified as controlling VOCs production in 27 melon. 28 • New tools for breeding fruit flavor in melon have been developed. 29 30 Abstract 31 Aroma is an essential trait in melon fruit quality, but its complexity and genetic basis is still 32 poorly understood. The aim of this study was the identification of quantitative trait loci (QTLs) 33 underlying volatile organic compounds (VOCs) biosynthesis in melon rind and flesh, using a 34 Recombinant Inbred Line (RIL) population from the cross ‘Piel de Sapo’ (PS) x ‘Védrantais’ 35 (VED), two commercial varieties segregating for ripening behavior. A total of 82 VOCs were 36 detected by gas chromatography-mass spectrometry (GC-MS), and 166 QTLs were identified. 37 The main QTL cluster was in chromosome 8, collocating with the previously described 38 ripening-related QTL ETHQV8.1, with an important role in VOCs biosynthesis. QTL clusters 39 involved in esters, lipid-derived volatiles and apocarotenoids were also identified, and 40 candidate genes have been proposed for ethyl 3-(methylthio)propanoate and benzaldehyde 41 biosynthesis. Our results provide genetic insides for deciphering fruit aroma in melon and 42 offer new tools for flavor breeding. 43 44 Keywords: aroma, climacteric ripening, Cucumis melo L., GC-MS, QTL mapping, VOCs. 2 45 1. Introduction 46 Melon (Cucumis melo L.) is an important crop worldwide, with a global production of more 47 than 27 million tonnes in 2018 (fao.org/faostat). Fruit quality is a complex trait determined by 48 several fruit characteristics and is essential for consumer’s appreciation. Melon is a very 49 diverse species within the Cucurbitaceae family, comprising accessions that differ in color, 50 texture, aroma, flavor and nutritional value (Pitrat, 2017). Melon is also an interesting crop to 51 study fruit ripening, as the species includes climacteric and non-climacteric accessions. 52 Climacteric ripening consists in a concomitant increase of the respiration rate and ethylene 53 production at the onset of the ripening process, whereas in non-climacteric fruits both remain 54 at a basal levels during ripening (Lelievre, Latche, Jones, Bouzayen, & Pech, 1997). Ethylene 55 has a direct impact in aroma production (Manríquez et al., 2006; Shalit et al., 2001), which is 56 one of the most important determinants of flavor and consumer preferences. 57 Fruit aroma depends on the concentration and the combination of volatile organic compounds 58 (VOCs), which are genetically determined, but also highly influenced by environmental and 59 post-harvest factors (Wyllie, Leach, Wang, & Shewfelt, 1994). More than 500 different VOCs 60 have been identified in melon, and depending on the technique used more than 100 61 compounds can be found in a single accession (Esteras et al., 2018; Obando-Ulloa, Ruiz, 62 Monforte, & Fernández-Trujillo, 2010; Pang et al., 2012). VOCs can originate from amino acids, 63 fatty-acids or terpenes, and can be classified according to their chemical group. Esters, 64 alcohols and aldehydes are the most abundant VOCs in melon, but other types of compounds 65 such as benzenoids, ketones, furans, lactones, monoterpenes, sesquiterpenes or 66 apocarotenoids are also found (Gonda et al., 2016). Each volatile has a different impact in the 67 final aroma of the fruit depending on the individual odor threshold. Several melon accessions 3 68 have been studied combining gas chromatography-mass spectrometry (GC-MS) analyses with 69 odor evaluations to determine the compounds that contribute the most to the unique flavor 70 of melon varieties (Pang et al., 2012). Some key VOCs identified are nonanal, 6-nonenal (Z), 71 2,6-nonadienal-(E,Z), hexyl acetate, ethyl 2-methylpropanoate or phenylethyl alcohol. Each of 72 these VOCs provides specific aroma and flavor to melon fruits, some of them being 73 denominated as melon-like, cucumber-like, fruity, floral, sweet, green or even foul. For 74 instance, most esters give fruity and sweet notes, whereas aldehydes and alcohols give green, 75 fresh, cucumber-like notes (Gonda et al., 2016; Pang et al., 2012). 76 The volatile profile of climacteric and non-climacteric fruits differs both qualitative and 77 quantitatively (Esteras et al., 2018; Moing et al., 2020; Obando-Ulloa et al., 2008). For 78 instance, esters such as hexyl acetate and ethyl acetate are highly abundant in climacteric 79 melons, whereas aldehydes like pentanal are more common in non-climacteric melons 80 (Obando-Ulloa et al., 2008). Moreover, the diversity and amount of VOCs produced in 81 climacteric melons is usually higher than in non-climacteric ones. 82 Melon is a diploid species (2n=2x=24) with a relatively small genome (357Mbp). The 83 availability of a melon reference genome (Garcia-Mas et al., 2012), dense genetic maps 84 (Pereira et al., 2018), germplasm collections and mapping populations such as Recombinant 85 Inbred Lines (RILs) and Introgression Lines (ILs) have eased the identification of quantitative 86 trait loci (QTLs) associated to fruit traits such as morphology, carotenoid content or ripening 87 behavior (Obando-Ulloa et al., 2010; Pereira et al., 2018, 2020). RILs have shown to be 88 powerful populations to detect QTLs in melon and other species as they usually present a high 89 number of recombinations and represent diverse allelic combinations. Other powerful 90 available tools in melon are databases such as Melonet-DB (Yano, Nonaka, & Ezura, 2018), 4 91 fruitENCODE (http://www.epigenome.cuhk.edu.hk/encode.html) and Melonomics 92 (http://www.melonomics.net), which contain transcriptomic and epigenomic data, and the 93 melon variome map (Zhao et al., 2019). 94 Biosynthesis pathways to several volatiles have been partially elucidated in melon although 95 the complete biosynthetic pathways from amino acids, fatty acids and terpenes remain 96 unknown. Aldehydes are hydrogenated to form alcohols (Manríquez et al., 2006), and alcohols 97 are then acetylated to produce esters (Yahyaoui et al., 2002). The aminotransferases 98 CmBCAT1 and CmArAT1 have been described to catalyze amino acids deamination, the first 99 step of the production of several branched and sulfur VOCs (Gonda et al., 2010, 2013). In the 100 phenylpropanoid pathway, CmCNL and CmBAMT participate in the synthesis of cinnamoyl- 101 CoA and methyl benzoate, respectively (Gonda et al., 2018). In the fatty acid pathway, 102 lipoxygenases (LOX), hydroperoxide lyases, isomerases and dehydrogenases play a role in the 103 biosynthesis of C6 and C9 aliphatic aldehydes and alcohols from linoleic and linolenic acids 104 (Schwab, Davidovich-Rikanati, & Lewinsohn, 2008). Several LOX genes have been identified in 105 melon (Tang, Zhang, Cao, Wang, & Qi, 2015). For other aliphatic VOCs, α and β-oxidation have 106 been proposed as possible biosynthetic pathways (Schwab et al., 2008). In the terpenoid 107 pathway, the terpene synthases CmTpsDul and CmTpsNY modulate the accumulation of 108 sesquiterpenes (Portnoy et al., 2008), whereas the cleavage dioxygenase CmCCD1 cleaves 109 carotenoids to produce several apocarotenoid VOCs such as geranylacetone and -ionone 110 (Ibdah et al., 2006). 111 In this study, we analyzed the volatile profile of a RIL population developed from a cross 112 between two melon elite varieties, ‘Piel de Sapo’ (PS) and ‘Védrantais’ (VED) which segregates 113 for many fruit quality traits, including aroma and ripening behavior (Pereira et al., 2020). QTL 5 114 mapping was performed using metabolite data obtained from both rind and flesh tissues, 115 integrating phenotypic data from GC-MS and genotypic data previously obtained by 116 genotyping by sequencing (Pereira et al., 2018). We hypothesize that both PS and VED may 117 present functional alleles for the VOC synthesis, some of them related to ripening behavior. 118 The aim of this work is to identify genetic factors controlling aroma production in melon which 119 could be integrated into breeding programs for fruit flavor improvement. 120 121 2. Materials and methods 122 2.1. Plant material 123 The 89 RIL population, with 82 RILs in the F8 generation and seven in the F7, was developed 124 from a cross between the non-aromatic ‘Piel de Sapo’ T111 (PS) (C. melo ssp. melo group 125 inodorus) and the aromatic ‘Védrantais’ (VED) (C. melo ssp. melo group cantalupensis, French 126 Charentais-type) by Pereira et al. (2018). The plants were grown at Caldes de Montbui 127 (Barcelona) under greenhouse conditions, allowing only one melon per plant. The fruits were 128 harvested at the ripe stage as previously defined (Pereira et al., 2018). The fruits analyzed in 129 this study were harvested in summer 2015 (T3) and summer 2016 (T4 and T5), using a total of 130 three biological replicates for the 89 RILs (one per harvest time) and 9 biological replicates for 131 PS, VED and the hybrid PS x VED (three per harvest time) (Pereira et al., 2018). 132 2.2. Sample preparation for volatile extraction 133 Frozen samples of rind and flesh from the three biological replicates for each RIL were ground 134 up and stored at -80 C. All biological replicates were independently analyzed by GC-MS. Two 135 grams of tissue were weighed and added to frozen chromatography 20 ml vials (Thermo Fisher 136 Scientific®, Waltham, MA, U.S.) containing 7 ml of a saturated NaCl solution and 15 ppm of 3- 6 137 hexanone used as internal standard. Samples were sealed with silicone septa and stored at 4 138 C in the dark for no longer than a week, until GC-MS analysis was performed. 139 2.3. Aroma analysis by GC-MS 140 Solid-Phase Micro-Extraction (SPME) was carried out using a 7890A gas chromatograph 141 coupled to a 5975C mass spectrometer and a GC PAL 80 autosampler (Agilent Technologies®, 142 Santa Clara, CA, U.S.). The chromatograph was equipped with an SPME fiber (50/30 µm 143 DVB/CAR/PDMS, Merck®, Darmstadt, Germany) and a Sapiens-X5MS capillary column (30 144 m/0.25 mm/0.25 µm, Teknokroma®, Sant Cugat del Vallès, Spain). Samples were pre- 145 incubated 15 min to 50 ºC at 250 rpm, and then the SPME fiber was exposed 30 min to the 146 vial headspace. Sample injection was performed in splitless mode, starting at a temperature 147 of 50 C for 1 min, increasing 5 C/min to 280 C, and holding that temperature for 5 min (total 148 time 52 min). The carrier gas was helium at a head pressure of 13.37 psi. The source 149 temperature of the mass spectrometer was set at 230 C and the quadrupole temperature at 150 150 C, with no delay in the detection. 151 We performed an untargeted analysis for VOCs of rind and flesh samples. VOCs were 152 identified by comparison of their mass spectra with the NIST 11 library (NIST/EPA/NIH) and by 153 their Kovats retention index. The Kovats retention index was calculated using a mix of alkanes 154 (C7-C40 in hexane, Merck®, Darmstadt, Germany) under the same chromatographic conditions. 155 We created a list of compounds identified in several samples by applying two quality criteria: 156 a calculated Kovats index within a confidence interval of 10 comparing to tabulated standard 157 values, and a minimum MS quality of 80. Compounds not matching both criteria were 158 excluded from the analysis. For individual samples, when the proximity of two compounds 159 lowered the MS quality, a cutoff of 50 was considered. 7 160 The content of each compound was estimated by comparison to the 3-hexanone internal 161 standard peak area. Compounds were coded following a similar criterion to Freilich et al. 162 (2015), grouping them based on their chemistry, tissue and name. Data was organized in four 163 different subsets: each biological replicate individually and the mean of each line. Means were 164 only considered when data of the three replicates was available. Medians of the biological 165 replicates were also calculated. 166 Information about the odor of the compounds detected was extracted from the literature. 167 2.4. QTL analysis 168 QTL mapping was conducted with MapQTL 6® performing both interval mapping (IM) and the 169 Kruskal-Wallis test (K-W). The genetic map used was described in Pereira et al. (2018) and was 170 constructed using genotyping-by-sequencing data from the same RIL population. MapChart® 171 was used for the map representation. For the statistical support of the QTL mapping, a 172 permutation test with 1,000 iterations and 95% confidence was performed. QTLs with a LOD 173 score ≥2.75 in at least one of the four subsets were preselected. The final map included only 174 QTLs that reached the 2.75 threshold in the means subset, or a score ≥2.50 in two subsets. K- 175 W scores were used to confirm the QTLs detected by IM. Candidate genes were selected using 176 genome version v3.6.1 and according to functional annotation data from the Melonomics 177 database v4.0 (https://www.melonomics.net/), orthologs found in PLAZA 4.0 (Van Bel et al., 178 2018) and transcriptomic data from the fruitENCODE database 179 (http://www.epigenome.cuhk.edu.hk/encode.html). 180 2.5. Statistical analysis 181 The software R (v4.0.0) (https://www.r-project.org/) with the RStudio interface (v1.1.463) 182 (http://www.rstudio.com/) was used to perform statistical analysis and to represent the data. 8 183 A Shapiro-Wilk test was performed to check the normality of the data and a correlation test 184 was performed using the Spearman’s rank correlation coefficient. Z scores were calculated to 185 standardize the data since they deviated from normality. PCA was performed with “ggfortify”, 186 correlation was calculated with “Hmisc” and “corrplot” packages, and plots were generated 187 with “gplots”, “factoextra” and “ggplot2”. 188 189 3. Results 190 3.1. General VOC analysis 191 Melon fruits from the RIL population produced 82 different VOCs, 79 were present in rind and 192 all were present in flesh. According to their chemical structure, the list comprised 47 esters, 193 13 aldehydes, seven alcohols, six terpenes, four benzenoid derivatives, three ketones and two 194 furans (Table 1). The distribution of all VOCs in the RIL population was not normal (p<0.05) 195 but left-skewed. Both rind and flesh produced a great quantity of esters and aldehydes. Rind 196 samples contained a higher quantity of these compounds compared to flesh (Supplementary 197 Table S1-S4). 198 Table 1. VOCs identified by Kovats index and mass spectrum in the PS x VED RIL population. Compound NIST nomenclature Codea RIb CAS Alcohols 2-Methylbutanol 1-Butanol, 2-methyl- ALX_1B2M 729 137-32-6 1-Hexanol ALX_1HXO 864 111-27-3 1-Octen-3-ol ALX_1O3O 977 3391-86-4 Benzyl Alcohol ALX_BZOL 1034 100-51-6 3-Octen-1-ol, (Z)- ALX_3O1OZ 1066 20125-85-3 1-Octanol ALX_1OCOL 1069 111-87-5 Phenylethyl Alcohol ALX_PHEOL 1114 60-12-8 Aldehydes 3-Methylbutanal Butanal, 3-methyl- ADX_B3M 647 590-86-3 2-Methylbutanal Butanal, 2-methyl- ADX_B2M 658 96-17-3 Pentanal ADX_PNL 695 110-62-3 Hexanal ADX_HXL 800 66-25-1 Heptanal ADX_HPL 901 111-71-7 Benzaldehyde ADX_BZALD 959 100-52-7 Octanal ADX_OCL 1002 124-13-0 Benzeneacetaldehyde ADX_BZACALD 1042 122-78-1 9 2-Octenal, (E)- ADX_2OLE 1063 2548-87-0 6-Nonenal, (Z)- ADX_6NNLZ 1103 2277-19-2 Nonanal ADX_NNL 1104 124-19-6 2,6-Nonadienal, (E,Z)- ADX_26NDLEZ 1152 557-48-2 Decanal ADX_DCL 1205 112-31-2 Benzenoids Toluene BX_TOL 761 108-88-3 1,3-Dimethylbenzene Benzene, 1,3-dimethyl- BX_BZ13DM 862 108-38-3 Styrene BX_STYR 890 100-42-5 1-Methoxy-4-methylbenzene Benzene, 1-methoxy-4-methyl- BX_BZ1M4M 1019 104-93-8 Esters Ethyl Acetate EX_EAC 615 141-78-6 Ethyl propanoate Propanoic acid, ethyl ester EX_PEE 707 105-37-3 n-Propyl acetate EX_NPA 709 109-60-4 Methyl butanoate Butanoic acid, methyl ester EX_BME 717 623-42-7 Ethyl 2-methylpropanoate Propanoic acid, 2-methyl-, ethyl ester EX_P2MEE 751 97-62-1 2-Methylpropyl acetate Acetic acid, 2-methylpropyl ester EX_A2MPE 767 110-19-0 Methyl 2-methylbutanoate Butanoic acid, 2-methyl-, methyl ester EX_B2MME 772 868-57-5 Ethyl butanoate Butanoic acid, ethyl ester EX_BEE 801 105-54-4 Butyl acetate Acetic acid, butyl ester EX_ABE 812 123-86-4 Ethyl 2-methylbutanoate Butanoic acid, 2-methyl-, ethyl ester EX_B2MEE 846 7452-79-1 3-Methylbutyl acetate 1-Butanol, 3-methyl-, acetate EX_1B3MA 873 123-92-2 2-Methylbutyl acetate 1-Butanol, 2-methyl-, acetate EX_1B2MA 877 624-41-9 Propyl butanoate Butanoic acid, propyl ester EX_BPE 898 105-66-8 Ethyl pentanoate Pentanoic acid, ethyl ester EX_PNEE 901 539-82-2 Pentyl acetate Acetic acid, pentyl ester EX_APNE 913 628-63-7 3-Methyl-2-buten-1-ol acetate 2-Buten-1-ol, 3-methyl-, acetate EX_2B1O3MA 921 1191-16-8 Methyl hexanoate Hexanoic acid, methyl ester EX_HXME 923 106-70-7 Propyl 2-methylbutanoate Butanoic acid, 2-methyl-, propyl ester EX_B2M.PE 943 37064-20-3 2-Methylpropyl butanoate Butanoic acid, 2-methylpropyl ester EX_B2MPE 953 539-90-2 Ethyl (methylthio)acetate EX_EMTHAC 983 4455-13-4 Butyl butanoate Butanoic acid, butyl ester EX_BBE 995 109-21-7 Ethyl hexanoate Hexanoic acid, ethyl ester EX_HXEE 1000 123-66-0 3-Hexen-1-ol, acetate, (Z)- EX_3HX1OAZ 1007 3681-71-8 Hexyl acetate Acetic acid, hexyl ester EX_AHXE 1016 142-92-7 1,2-Propanediol, diacetate EX_12PLDA 1027 623-84-7 Cyclohexyl acetate* Acetic acid, cyclohexyl ester* EX_ACHXE 1039 622-45-7 2-Methylbutyl butanoate Butanoic acid, 2-methylbutyl ester EX_B2MBE 1057 51115-64-1 2,3-Butanedioldiacetate EX_23BOLDA 1061 & 1075 1114-92-7 Ethyl 2,4-hexadienoate 2,4-Hexadienoic acid, ethyl ester EX_24HXEE 1094 2396-84-1 Propyl hexanoate Hexanoic acid, propyl ester EX_HXPE 1094 626-77-7 Ethyl heptanoate Heptanoic acid, ethyl ester EX_HPEE 1098 106-30-9 Ethyl 3-(methylthio)propanoate 3-(Methylthio)propanoic acid ethyl ester EX_3MTHPEE 1100 13327-56-5 Heptyl acetate Acetic acid, heptyl ester EX_AHPE 1112 112-06-1 Ethyl (+-)-3-acetoxybutyrate* EX_E3AXB 1114 27846-49-7 1,3-Butanediol, diacetate EX_13BOLDA 1128 1117-31-3 2-Methylpropyl hexanoate Hexanoic acid, 2-methylpropyl ester EX_HX2MPE 1150 105-79-3 Phenylmethyl acetate Acetic acid, phenylmethyl ester EX_APHME 1165 140-11-4 Ethyl benzoate Benzoic acid, ethyl ester EX_BZEE 1171 93-89-0 Diethyl butanedioate Butanedioic acid, diethyl ester EX_BDEE 1181 123-25-1 Ethyl 4-octenoate 4-Octenoic acid, ethyl ester EX_4OEE 1186 138234-61-4 Butyl hexanoate Hexanoic acid, butyl ester EX_HXBE 1191 626-82-4 Ethyl octanoate Octanoic acid, ethyl ester EX_OEE 1198 106-32-1 Octyl acetate Acetic acid, octyl ester EX_AOE 1212 112-14-1 Ethyl benzeneacetate Benzeneacetic acid, ethyl ester EX_BZAEE 1245 101-97-3 2-Methylbutyl hexanoate Hexanoic acid, 2-methylbutyl ester EX_HX2MBE 1253 2601-13-0 2-Phenylethyl acetate Acetic acid, 2-phenylethyl ester EX_A2PHEE 1257 103-45-7 Ethyl decanoate Decanoic acid, ethyl ester EX_DCEE 1397 110-38-3 Furans 2-Ethylfuran Furan, 2-ethyl- FX_F2E 700 3208-16-0 2-Pentylfuran Furan, 2-pentyl- FX_F2PN 992 3777-69-3 10 Ketones 2-Methyl-3-pentanone 3-Pentanone, 2-methyl- KX_3PN2M 743 565-69-5 4-Heptanone* KX_4HPA 868 123-19-3 6-Methyl-5-hepten-2-one 5-Hepten-2-one, 6-methyl- KX_5HP2O 985 110-93-0 Terpenes Eucalyptol TX_EUL 1030 470-82-6 β-Cylohomocitral TX_ßCYHCL 1254 472-66-2 Megastigma-4,6(Z),8(Z)-triene TX_MEGA 1324 71186-25-9 Caryophyllene TX_ßCPHL 1421 87-44-5 Cis-Geranylacetone TX_CGERAC 1445 3879-26-3 δ-Cadinene TX_δCAD 1527 483-76-1 199 a X = R for rind, F for flesh 200 b RI = retention index 201 *Newly described in melon (to our knowledge) 202 We first analyzed the volatile profile of the parental lines. PS displayed a lower number of 203 VOCs than VED in both rind (30 and 66, respectively) and flesh (27 and 71) tissues 204 (Supplementary Table S1). The hybrid (Hyb) accumulated more compounds in rind (70) than 205 in flesh (58). In all three lines, benzaldehyde was one of the most abundant compounds in 206 rind, although the mean levels varied from 12914 ng g-1 of frozen tissue in PS to 56057 ng g-1 207 of frozen tissue in VED. VED and Hyb, both climacteric, also produced high amounts of the C6 208 lipid derived hexyl acetate (56132 ng g-1 and 38153 ng g-1 of frozen tissue, respectively). In 209 flesh, lipid derived aldehydes such as nonanal (PS, 1364 ng g-1) and octanal (Ved, 4220 ng g-1 210 and Hyb, 1530 ng g-1) were predominant. The climacteric lines presented in addition high 211 levels of ethyl hexanoate and butanoate derivatives in flesh. Some compounds were 212 overproduced in Hyb fruits compared to PS and VED, such as heptanal and methyl hexanoate 213 in rind, whereas other VOCs were underproduced like decanal in flesh. 214 VED produced significantly more esters, alcohols, aldehydes and terpenes than PS (p<0.05). 215 For instance, VED samples produced 15 times more octanal and 40 times more benzyl alcohol 216 in flesh, or 186 times more 2,3-butanedioldiacetate in rind. Furans and some ketones such as 217 2-ethylfuran and 2-methyl-3-pentanone were produced principally in PS fruits, whereas 218 benzenoids were more abundant in Hyb rind. The total amount of VOCs in flesh was more 11 219 variable than in rind, possibly due to a stronger environmental effect in flesh samples caused 220 by slight changes in ripening stage. VED main groups were esters (68-85 % in rind, 52-84 % in 221 flesh), followed by aldehydes (12-27 % in rind, 4-38 % in flesh) and alcohols (2-5 % in rind, 2- 222 14 % in flesh) (Supplemental Fig. S1). PS main groups were aldehydes (88-97 % in rind, 67-90 223 % in flesh), alcohols (1-3 % in rind, 1-4 % in flesh) and ketones (1 % in rind, 2-20 % in flesh). 224 Hyb melons had an intermediate phenotype between PS and VED but were more similar to 225 the VED profile. 226 The RIL population showed a left-skewed distribution and transgressive segregation for most 227 compounds. Interestingly, flesh was the only tissue to produce β-cyclocitral, cis- 228 geranylacetone and megastigma-4,6(Z),8(Z)-triene, the last one not being detected in any 229 parental line. The only VOCs found in all samples were benzaldehyde in rind and eucalyptol in 230 flesh, although nonanal and ethyl acetate were also produced by most lines in both tissues. 231 Ethyl decanoate in rind and pentanal in flesh were rare in the population, detected only in one 232 and 5 lines, respectively. 233 Correlations between VOCs in the RIL population were represented in a heatmap for both 234 tissues, rind and flesh (Fig. 1). Esters were positively correlated among them, forming two 235 clusters in flesh and 3 in rind. The cluster showing the strongest correlation comprised ethyl 236 esters. Another ester cluster included also some alcohols such as hexanol, 2-methylbutanol 237 and octanol, as well as eucalyptol in flesh and nonanal and decanal in rind. In addition, we 238 found clusters of positive correlation for aldehydes, phenylalanine derived and lipid derived 239 VOCs in rind, for acetates, nonanal derived VOCs and terpenes in flesh, plus some others with 240 diverse composition. Furans, ketones and benzenoids showed negative correlation with the 241 rest of the compounds. 12 242 3.2. QTL mapping 243 We found 166 QTLs for 63 different compounds, 77 of these QTLs were found in flesh and 89 244 in rind. Among them, 13 QTLs were coincident between tissues. All QTLs were confirmed by 245 both IM and K-W, except QTL EF_3MTHPEE_3.1 that did not fit the IM thresholds but had a 246 significant K-W value. QTLs were identified in all 12 melon chromosomes (Fig. 2 and Table 2). 247 Of the total number of QTLs, 18 were found in all four analyzed subsets (individual replicates 248 and mean), 28 were detected in three subsets, and 4 were detected in two subsets but not in 249 the mean subset. In addition, 72 QTLs appeared in only one subset and in the mean, and 44 250 were only found in the mean subset. The physical interval of the QTLs ranged from 5.9 Kb to 251 25.3 Mb, with a median of 9.2 cM and 1.96 Mb in genetic and physical distance, respectively. 252 The median number of genes within these intervals was 173. We defined 12 clusters of QTLS 253 which included multiple overlapping QTLs for compounds chemically similar and/or belonging 254 to the same biochemical pathway (Table 2 and Supplementary Table S5). One of them on 255 Chr08 contained up to 92 different QTLs. 256 Table 2. QTLs detected in the PS x VED RIL population. Genetic Positive position Size LOD LOD LOD LOD K-W % additive Chr Ca QTL (cM) Nearest markerb (Mb)b meanc T3c T4c T5c value Exp effect 1 EF_E3AXB_1.1 3.3 chr01_452338 1.2 2.54 0.79 2.53 1.35 <0.005 16.9 PS EF_NPA_1.1 122.2 chr01_36593213 0.7 2.75 1.66 1.30 1.31 <0.001 18.2 Ved 2 ER_BME_2.1 23.7 chr02_1937609 1.1 2.55 0.82 2.20 3.33 <0.005 16.7 PS EF_B2MME_2.1 50.8 chr02_6015657 3.7 2.64 0.77 2.82 1.31 <0.005 14.6 PS ER_1B2MA_2.1 66.4 chr02_13326331 9.8 3.04 0.62 1.91 2.08 <0.005 19.7 PS ER_B2MME_2.1 61.5 chr02_9079639 5.5 3.08 1.95 2.07 2.06 <0.005 19.9 PS C2 EF_B2MME_2.2 61.5 chr02_9079629 5.0 2.79 0.48 3.02 2.05 <0.0005 15.6 PS EF_B2M.PE_2.1 64.8 chr02_11199109 5.0 2.54 0.95 0.99 2.92 <0.005 17.0 PS ER_B2MBE_2.1 61.5 chr02_9079639 7.4 2.85 0.28 0.48 2.97 <0.0005 18.5 PS ADR_BZACALD_2.1 67.5 chr02_13326331 7.4 2.76 0.42 1.37 2.14 <0.001 18.0 PS 3 ALR_1HXO_3.1 38.9 chr03_6150294 4.5 2.78 3.90 0.94 1.26 <0.0001 18.1 PS FR_F2E_3.1 119.0 chr03_30900408 1.8 2.87 0.79 0.75 0.04 <0.0001 18.7 PS EF_EMTHAC_3.1 119.6 chr03_30962715 0.4 3.99 1.51 1.82 3.02 <0.0001 24.0 Ved C3 EF_3MTHPEE_3.1 119.6 chr03_30962715 0.7 2.64 1.77 1.91 1.20 <0.0005 16.6 Ved 13 4 EF_A2MPE_4.1 4.9 chr04_1127842 0.8 3.05 1.56 0.55 1.65 <0.0005 18.9 Ved ER_NPA_4.1 68.4 chr04_15256396 3.1 2.70 3.03 1.00 0.90 <0.005 17.7 PS ADR_BZALD_4.1 72.4 chr04_15922403 0.9 3.74 3.95 1.71 1.54 <0.0001 23.6 PS KF_3PN2M_4.1 75.0 chr04_16687925 1.5 2.90 2.12 0.96 1.70 <0.001 19.1 Ved EF_EMTHAC_4.1 75.0 chr04_16687925 3.0 2.80 0.94 1.29 3.07 <0.005 17.5 PS C4 ER_B2MME_4.1 75.0 chr04_16687925 1.5 3.20 2.32 2.09 1.82 <0.005 20.5 PS ER_3MTHPEE_4.1 75.0 chr04_16687925 3.0 2.96 2.15 2.11 1.32 <0.0005 19.2 PS ER_NPA_4.2 86.7 chr04_20479692 1.8 2.75 1.37 1.17 1.88 <0.005 18.0 PS ER_A2MPE_4.1 92.7 chr04_25893667 6.0 2.99 0.36 1.36 2.12 <0.005 19.3 PS 5 ADF_BZALD_5.1 0.0 chr05_182820 0.5 2.86 1.77 1.29 3.91 <0.0001 17.9 PS C5.1 ALF_BZOL_5.1 1.0 chr05_182820 0.5 4.28 3.27 3.21 3.23 <0.0001 20.6 PS EF_NPA_5.1 6.5 chr05_729528 1.2 2.51 2.99 1.84 0.37 <0.005 16.9 PS ALF_1B2M_5.1 10.9 chr05_1085286 0.6 2.55 2.80 1.43 0.85 <0.005 17.0 PS C5.2 ALR_1B2M_5.1 12.5 chr05_1085286 0.6 3.39 4.16 0.50 1.46 <0.0005 21.6 PS ER_NPA_5.1 8.9 chr05_911334 0.5 2.71 3.70 0.80 1.51 <0.0005 17.7 PS ADF_NNL_5.1 41.2 chr05_2516188 0.5 3.80 2.08 1.34 0.92 <0.0005 24.3 PS C5.3 ADF_NNL_5.3 48.2 chr05_2846615 1.7 3.51 1.92 0.56 1.79 <0.001 24.3 PS ADF_NNL_5.2 44.6 chr05_2849664 0.0 4.11 1.95 1.32 1.29 <0.0005 25.9 PS EF_BEE_5.1 93.6 chr05_25326402 0.9 2.74 2.91 1.09 1.88 <0.001 17.2 Ved 6 ALR_3O1OZ_6.1 21.7 chr06_2396961 1.8 2.83 2.42 2.03 1.16 <0.0001 18.7 Ved ER_3HX1OAZ_6.1 21.7 chr06_2396961 0.2 3.21 7.05 1.34 3.39 <0.0001 20.9 Ved C6.1 EF_3HX1OAZ_6.1 28.0 chr06_3059995 1.3 5.10 3.31 3.03 2.74 <0.0001 29.6 Ved ER_3HX1OAZ_6.2 28.0 chr06_3059995 0.9 3.41 7.21 1.47 3.98 <0.0001 22.1 Ved ALR_1B2M_6.1 143.3 chr06_34012274 0.8 3.29 1.34 0.82 1.52 <0.0005 21.1 Ved C6.2 ALF_BZOL_6.1 144.3 chr06_34012274 1.1 2.77 0.54 0.72 2.90 <0.005 18.4 Ved 7 EF_NPA_7.1 3.0 chr07_551647 1.1 2.76 0.63 3.21 1.10 <0.005 16.5 Ved ER_BBE_7.1 14.3 chr07_1055303 0.6 2.77 1.19 1.49 0.26 <0.005 18.1 Ved TF_βCYHCL_7.1 56.1 chr07_18142671 18.7 2.89 0.03 2.37 2.14 <0.0005 18.0 Ved 8 ER_24HXEE_8.2 102.7 chr08_29813774 25.3 2.47 3.02 2.17 2.77 <0.001 16.3 Ved ADR_B3M_8.1 84.2 chr08_9446475 3.1 2.59 3.44 2.12 0.81 <0.0001 17.0 Ved EF_B2M.PE_8.2 90.3 chr08_24466782 18.5 2.84 2.55 1.07 0.47 <0.0005 18.7 Ved ER_AOE_8.1 84.2 chr08_9446475 0.8 3.24 3.18 2.52 1.59 <0.0001 20.8 Ved ER_AHPE_8.1 85.2 chr08_9634968 1.1 3.13 3.63 2.19 2.81 <0.0001 20.2 Ved ER_ABE_8.1 86.2 chr08_9634968 1.1 3.81 0.42 2.70 2.73 <0.0001 24.0 Ved M_8.1 EF_B2M.PE_8.1 86.5 chr08_9745430 8.2 3.91 2.80 1.54 1.09 <0.0005 24.8 Ved ER_B2M.PE_8.1 86.5 chr08_9634968 1.1 3.36 2.74 1.39 0.48 <0.0001 21.5 Ved EF_3MTHPEE_8.2 90.3 chr08_24466782 16.6 3.83 1.84 1.80 2.10 <0.0001 32.7 Ved EF_24HXEE_8.1 87.5 chr08_9941727 18.3 2.53 3.50 0.43 0.91 <0.0001 15.9 Ved M_8.2 ER_ABE_8.2 90.3 chr08_24466782 18.3 3.26 1.00 1.86 2.65 <0.0001 20.9 Ved ALF_PHEOL_8.2 91.3 chr08_24466782 18.3 4.74 3.61 1.22 3.18 <0.0001 27.8 Ved ER_23BOLDA_8.1 87.6 chr08_9941727 20.3 2.82 2.03 3.65 2.12 <0.0001 18.3 Ved TR_EUL_8.1 86.2 chr08_9634968 0.5 6.70 7.53 3.52 0.12 <0.0001 38.3 Ved ALF_PHEOL_8.1 86.5 chr08_9634968 0.5 6.71 2.86 1.65 4.32 <0.0001 37.0 Ved M_8.3 EF_HXME_8.1 86.5 chr08_9634968 0.5 5.32 3.95 2.60 1.09 <0.0001 30.6 Ved ADR_OCL_8.1 86.5 chr08_9634968 0.5 4.26 3.40 2.12 1.77 <0.0001 26.4 Ved 14 ALR_OCOL_8.1 86.5 chr08_9634968 0.5 4.17 4.28 3.86 1.34 <0.0001 25.9 Ved ALR_1HXO_8.1 86.5 chr08_9634968 0.5 3.87 1.37 3.33 2.57 <0.0005 24.3 Ved EF_3MTHPEE_8.1 86.5 chr08_9634968 0.8 5.76 2.57 3.20 3.61 <0.0001 32.7 Ved ER_EMTHAC_8.1 86.5 chr08_9634968 0.8 5.05 4.71 4.63 3.40 <0.0001 30.5 Ved ER_BZAEE_8.1 86.5 chr08_9634968 0.8 4.95 4.44 3.58 5.19 <0.0001 30.0 Ved ER_B2MME_8.1 86.5 chr08_9634968 0.8 4.93 3.44 3.50 4.05 <0.0001 29.9 Ved ADR_DCL_8.1 86.5 chr08_9634968 0.8 4.79 3.29 3.11 4.45 <0.0001 29.2 Ved ER_3MTHPEE_8.1 86.5 chr08_9634968 0.8 4.64 2.41 3.68 3.54 <0.0001 28.4 Ved BR_STYR_8.1 86.5 chr08_9634968 0.8 3.73 0.15 3.50 1.90 <0.0005 23.9 Ved M_8.4 ADR_BZACALD_8.1 86.5 chr08_9634968 0.8 3.70 4.66 2.65 2.95 <0.0001 23.4 Ved ADR_NNL_8.1 86.5 chr08_9634968 0.8 3.68 2.78 2.58 2.75 <0.0001 23.2 Ved ER_AHXE_8.1 86.5 chr08_9634968 0.8 3.38 2.08 2.05 3.02 <0.0001 21.6 Ved ER_A2PHEE_8.1 86.5 chr08_9634968 0.8 2.60 0.74 2.44 2.75 <0.0001 17.0 Ved ER_E3AXB_8.1 87.5 chr08_9941727 0.8 4.29 2.70 2.61 3.37 <0.0001 26.6 Ved KR_4HPA_8.1 87.5 chr08_24466782 0.8 0.21 0.18 3.22 5.50 <0.0001 1.5 PS EF_BZAEE_8.1 86.5 chr08_9634968 7.8 4.31 2.70 1.60 0.88 <0.0001 25.6 Ved ALF_1B2M_8.1 87.5 chr08_9941727 12.3 4.09 3.68 2.77 1.41 <0.0001 25.8 Ved KF_3PN2M_8.1 88.6 chr08_10225136 15.0 3.03 1.33 2.45 0.85 <0.001 19.8 PS EF_1B2MA_8.1 86.5 chr08_9634968 16.3 3.68 1.91 2.46 1.98 <0.0001 22.3 Ved ER_24HXEE_8.1 86.5 chr08_9634968 16.3 3.30 3.51 2.98 4.06 <0.0005 21.1 Ved ER_HXEE_8.1 87.5 chr08_9941727 16.3 4.26 5.58 3.98 3.69 <0.0001 26.4 Ved EF_HXBE_8.1 87.5 chr08_9745430 16.3 3.07 2.50 1.37 0.77 <0.0001 20.1 Ved EF_BBE_8.1 88.6 chr08_10225136 16.3 3.78 1.65 1.40 2.06 <0.0001 24.1 Ved M_8.5 ER_B2MEE_8.1 88.9 chr08_10225136 16.3 4.58 2.68 1.93 5.67 <0.0001 28.1 Ved ALF_1HXO_8.1 88.9 chr08_10225136 16.3 2.50 2.19 3.79 0.08 <0.0001 15.8 Ved ER_BEE_8.1 90.3 chr08_24466782 16.3 4.96 1.93 3.94 5.68 <0.0001 30.0 Ved ER_PNEE_8.1 90.3 chr08_24466782 16.3 4.92 4.23 3.06 2.99 <0.0001 29.8 Ved ER_E3AXB_8.2 90.3 chr08_24466782 16.3 3.28 2.37 1.36 3.10 <0.0001 21.0 Ved ER_HXME_8.1 86.5 chr08_9634968 18.0 4.03 6.51 4.57 0.04 <0.0001 25.2 Ved EF_HXPE_8.1 86.5 chr08_9745430 18.0 2.79 3.15 1.27 0.95 <0.0001 18.4 Ved EF_HX2MBE_8.1 88.6 chr08_10225136 18.0 3.56 2.29 1.63 0.91 <0.0001 22.9 Ved M_8.6 ER_HPEE_8.1 90.3 chr08_24466782 18.0 3.35 2.88 2.44 2.38 <0.0001 21.4 Ved BR_STYR_8.2 90.3 chr08_24466782 18.0 2.85 1.02 2.97 1.16 <0.001 18.8 Ved ER_OEE_8.1 90.3 chr08_24466782 18.0 2.64 1.68 1.62 2.91 <0.0001 17.3 Ved ER_B2MME_8.2 91.3 chr08_25723466 18.0 4.25 3.50 2.24 4.00 <0.0001 26.3 Ved EF_HPEE_8.1 88.9 chr08_10225136 20.0 4.02 3.40 1.58 1.68 <0.0001 25.4 Ved EF_HX2MPE_8.1 90.3 chr08_24466782 20.0 2.88 1.82 1.11 1.15 <0.0001 19.0 Ved M_8.7 ALR_OCOL_8.2 90.3 chr08_24466782 20.0 1.44 4.53 3.19 0.08 <0.0001 9.8 Ved EF_DCEE_8.1 93.5 chr08_27430936 20.0 3.09 1.22 2.19 1.20 <0.0001 19.1 Ved EF_HPEE_8.2 103.0 chr08_29813774 22.4 3.31 3.40 2.29 1.60 <0.0001 21.5 Ved M_8.8 ALR_OCOL_8.3 103.0 chr08_29813774 22.4 1.26 4.19 2.82 0.03 <0.0001 8.7 Ved ER_PEE_8.1 90.3 chr08_24466782 16.1 4.02 1.68 2.23 3.18 <0.0001 25.1 Ved M_8.9 EF_B2MEE_8.1 90.3 chr08_24466782 16.1 3.96 1.87 1.81 1.31 <0.0001 23.8 Ved 15 ER_EAC_8.1 90.3 chr08_24466782 17.8 3.89 2.04 1.87 4.22 <0.0005 24.4 Ved EF_E3AXB_8.2 93.5 chr08_27430936 19.7 4.23 3.38 2.31 1.84 <0.0001 26.6 Ved EF_E3AXB_8.1 90.3 chr08_24466782 15.8 4.93 2.40 1.47 1.99 <0.0001 30.3 Ved M_8.10 EF_1B2MA_8.2 90.3 chr08_24466782 15.8 4.07 2.24 2.87 1.63 <0.0001 24.4 Ved EF_HXEE_8.1 96.6 chr08_27868043 19.5 5.39 3.55 4.01 2.89 <0.0001 31.0 Ved EF_B2MPE_8.1 88.9 chr08_17287431 19.9 2.84 1.44 1.87 0.95 <0.005 18.7 Ved M_8.11 BF_BZ13DM_8.1 96.6 chr08_27868043 19.9 2.97 2.02 1.60 2.33 <0.0005 18.5 PS EF_PNEE_8.1 97.6 chr08_27868043 21.4 4.90 3.94 3.28 2.94 <0.0001 28.6 Ved EF_BEE_8.1 91.3 chr08_24466782 22.5 3.06 1.21 2.38 1.79 <0.0005 19.0 Ved ADR_NNL_8.2 91.7 chr08_25723466 12.1 2.53 3.24 1.99 1.73 <0.001 16.6 Ved ER_A2MPE_8.1 100.7 chr08_29419309 8.0 2.90 1.90 1.30 1.19 <0.0005 18.8 Ved EF_HXME_8.2 95.6 chr08_27868043 5.0 5.12 4.97 2.38 1.40 <0.0001 29.7 Ved ADR_BZACALD_8.2 91.7 chr08_25723466 5.3 2.65 3.98 2.07 3.23 <0.0001 17.3 Ved EF_BZAEE_8.2 95.6 chr08_27868043 3.7 5.03 5.53 0.73 1.70 <0.0001 29.3 Ved M_8.12 ER_HXME_8.2 96.6 chr08_27868043 3.7 2.87 7.77 1.02 0.01 <0.0001 18.7 Ved ADF_OCL_8.1 102.7 chr08_29419309 6.2 3.13 3.63 2.56 1.10 <0.0001 20.4 Ved ER_ABE_8.3 100.7 chr08_29419309 2.4 3.77 1.65 1.65 1.86 <0.0001 23.8 Ved ER_AHXE_8.2 99.6 chr08_29419309 1.9 2.77 3.56 1.77 1.23 <0.0001 18.1 Ved ADR_BZACALD_8.3 100.7 chr08_29419309 1.9 4.77 6.68 2.41 4.27 <0.0001 29.1 Ved EF_1B2MA_8.3 100.7 chr08_29419309 1.9 3.70 3.38 2.66 1.23 <0.0001 22.4 Ved M_8.13 TR_EUL_8.2 100.7 chr08_29419309 1.9 3.31 4.71 1.51 0.00 <0.0005 21.2 Ved EF_AHXE_8.1 100.7 chr08_29419309 1.9 3.12 2.10 1.95 0.88 <0.0001 19.3 Ved EF_B2M.PE_8.3 100.7 chr08_29419309 1.9 3.01 3.16 1.25 1.25 <0.005 19.7 Ved EF_APNE_8.1 100.7 chr08_29419309 1.9 2.99 2.63 2.02 1.06 <0.0001 18.6 Ved ER_PNEE_8.2 102.7 chr08_29813774 4.0 4.14 4.50 2.70 2.09 <0.0001 25.8 Ved M_8.14 ER_HXEE_8.2 102.7 chr08_29813774 4.0 3.66 6.04 2.76 2.36 <0.0001 23.2 Ved ER_HPEE_8.2 102.7 chr08_29813774 4.0 3.30 2.34 2.76 1.90 <0.0001 21.1 Ved ER_BZAEE_8.2 103.0 chr08_29813774 4.0 3.93 5.90 2.72 3.10 <0.0001 24.7 Ved EF_E3AXB_8.3 103.0 chr08_29813774 2.5 3.46 2.54 0.85 4.13 <0.0005 22.3 Ved M_8.15 EF_BEE_8.2 104.5 chr08_31341018 2.5 4.26 0.86 5.01 3.73 <0.0005 25.4 Ved EF_BZAEE_8.3 113.8 chr08_32190395 0.4 3.90 2.44 0.96 1.33 <0.0005 23.5 Ved EF_BEE_8.3 114.3 chr08_32190395 0.4 4.50 1.66 4.24 3.90 <0.0001 26.6 Ved 9 ER_EAC_9.1 52.9 chr09_20327236 1.7 2.56 3.19 0.49 1.17 <0.001 16.8 PS ER_BEE_9.1 64.3 chr09_21685526 2.3 2.75 1.74 1.02 1.31 <0.005 18.0 PS TF_CGERAC_9.1 65.3 chr09_21754707 0.6 4.54 5.81 2.72 3.27 <0.0001 26.8 Ved C9 TF_βCYHCL_9.1 62.6 chr09_21387823 0.7 2.83 0.91 2.00 2.65 <0.0001 17.7 Ved KF_5HP2O_9.1 64.3 chr09_21685526 0.5 8.36 11.65 6.69 7.68 <0.0001 45.7 Ved 10 ER_PEE_10.1 9.1 chr10_1133000 1.3 2.52 3.17 0.20 1.19 <0.005 16.6 PS ADF_HXL_10.1 31.0 chr10_3152004 2.4 3.07 0.61 0.17 2.51 <0.001 20.1 Ved 11 BF_BZ13DM_11.1 3.9 chr11_7689143 11.1 4.82 0.44 3.17 2.86 <0.0001 28.2 PS 16 ER_B2MME_11.1 6.5 chr11_10453734 13.1 2.83 2.29 0.91 1.65 <0.0005 18.4 Ved C11.1 ER_BEE_11.1 5.9 chr11_10453734 11.5 3.54 1.52 1.66 2.11 <0.0001 22.5 Ved EF_HXEE_11.1 5.9 chr11_10453734 11.5 2.72 1.21 3.96 1.32 <0.0005 17.0 Ved ER_PEE_11.1 6.5 chr11_10453734 11.2 3.52 2.47 1.91 1.25 <0.0005 22.4 Ved ER_EAC_11.1 6.5 chr11_10453734 11.2 2.83 1.51 1.49 1.08 <0.0005 18.4 Ved EF_APNE_11.1 88.3 chr11_32731899 1.2 4.21 1.07 2.06 3.03 <0.0001 25.1 PS EF_AHXE_11.1 88.3 chr11_32731899 1.2 2.75 0.27 1.83 2.83 <0.005 17.2 PS EF_1B2MA_11.1 88.3 chr11_32731899 1.3 2.71 0.98 1.93 2.54 <0.005 17.0 PS ER_BPE_11.1 88.5 chr11_32755551 1.3 4.31 1.76 1.38 3.82 <0.0001 26.7 PS ER_AHXE_11.1 88.5 chr11_32755551 1.3 4.30 1.76 1.60 3.34 <0.0005 26.6 PS ER_ABE_11.1 88.5 chr11_32755551 1.3 3.97 1.17 1.29 2.68 <0.0005 24.9 PS C11.2 ER_1B2MA_11.1 88.5 chr11_32755551 1.6 3.47 2.02 1.56 1.81 <0.001 22.1 PS ER_12PLDA_11.1 88.5 chr11_32755551 1.9 3.21 2.74 0.35 1.14 <0.0001 20.6 PS ER_2B1O3MA_11.1 88.5 chr11_32755551 1.9 3.05 0.01 1.34 1.72 <0.005 19.7 PS ER_3HX1OAZ_11.1 90.5 chr11_32928874 2.0 2.55 1.51 0.98 3.72 <0.005 17.0 PS ER_APNE_11.1 88.5 chr11_32755551 0.6 4.30 1.59 1.92 3.00 <0.001 26.6 PS ER_BBE_11.1 88.5 chr11_32755551 1.1 2.73 0.31 0.69 3.43 <0.0005 17.9 PS EF_12PLDA_11.1 91.5 chr11_32928874 1.2 2.72 3.30 0.39 0.30 <0.0005 18.0 PS EF_3HX1OAZ_11.1 107.8 chr11_34107169 0.6 3.51 1.53 1.82 2.39 <0.0001 21.4 PS 12 FR_F2PN_12.1 0.0 chr12_399410 1.1 3.26 2.51 1.18 2.50 <0.0005 21.2 PS FR_F2PN_12.2 19.0 chr00_4930997 1.5 4.33 0.70 0.66 4.51 <0.0001 27.1 PS ALF_1B2M_12.1 94.0 chr12_25861436 0.7 2.52 3.28 1.08 2.34 <0.005 20.3 PS EF_PEE_12.1 96.2 chr12_26103502 0.4 3.80 1.33 2.69 2.22 <0.0001 24.2 PS EF_BZEE_12.1 99.6 chr12_26757080 1.3 2.76 0.79 1.32 1.25 <0.005 17.3 PS 257 a C = cluster, M = merged 258 b Physical positions and distances refer to the v3.6.1 melon reference genome 259 c LOD ≥2.50 are in bold, LOD ≥2.75 are underlined 260 The highest LOD values were found in clusters on Chr06, Chr08 and Chr09. On Chr06, we 261 detected a stable QTL cluster (C6.1) with a LOD score of 7.21 related to lipid-derived VOCs, 262 with three QTLs for 3-hexen-1-ol acetate (Z) at 19-32 cM in both flesh and rind, and one QTL 263 for 3-octen-1-ol (Z) for rind. The narrowest QTL interval included 26 annotated genes and the 264 positive additive effect of C6.1 was provided by the VED allele. The largest cluster was found 265 on Chr08, formed by 92 QTLs overlapping between 71-121 cM. Many of these QTL were stable 266 since they were detected in all subsets. We could identify 3 recurrent maximum LOD peaks, 267 which were located around 86, 91 or 100 cM. These QTLs were merged (M) in black rectangles 268 to help visualization (Fig. 2 and Table 2). The length of these QTLs ranged 0.41-25 Mb. The 17 269 positive effect was provided by the VED allele for all significant VOCs, except for 4-heptanone, 270 2-methyl-3-pentanone and 1,3-dimethylbenzene. QTLs in this cluster explained 15-30 % of the 271 variation. Another QTL cluster (C9) was found on Chr09 in flesh tissue for two terpenes and a 272 ketone derived from the carotenoid cleavage pathway. The cluster shares an interval of 0.51 273 Mb at 62-69 cM, a region that contains 73 annotated genes. The highest LOD value was 11.65 274 for 6-methyl-5-hepten-2-one and explained 45 % of the variance. 275 Other minor QTLs for compounds relevant for melon flavour were found on Chr02, Chr03, 276 Chr05 and Chr11. On Chr02, we found a cluster of 5 QTLs (C2) with a LOD peak of 3.08 between 277 61,5-67,5 cM. C2 included QTLs for isoleucine-derived esters, the PS allele contributing to 278 higher content of these VOCs. Another cluster (C3) was found at the end of Chr03, containing 279 two methionine-derived sulfur esters and overlapping in 0.39 Mb (containing 59 genes), with 280 a positive effect of the VED allele. The ethyl (methylthio)acetate QTL had a LOD above 3.00 in 281 one of the subsets, whereas the QTL for ethyl 3-(methylthio)propanoate was significant only 282 using K-W test. On Chr05, a cluster (C5.1) containing 109 genes was detected for benzaldehyde 283 and benzyl alcohol, two phenylalanine derived VOCs. The benzyl alcohol QTL was detected in 284 all subsets. Two more clusters were found on chromosome 5 (C5.2 and C5.3) for 1-butanol-2- 285 methyl and nonanal. In all these QTLs the positive additive effect was provided by the PS allele. 286 Two ester clusters were localized on Chr11, C11.1 and C11.2, containing predominantly ethyl 287 esters and acetates, respectively. The first cluster, with a positive effect of the VED allele, was 288 located between 0-11 cM and had a LOD peak of 3.54. The second cluster, with a positive 289 additive effect of the PS allele, was detected at 81-100 cM and had a LOD peak of 4.31. 290 3.3. VOCs and ripening 18 291 The RIL population segregates for climacteric ripening, as expected since the two parental 292 lines have opposite ripening behavior (Supplementary Table S2). The PS x VED RILs were 293 classified in a previous work as climacteric, non-climacteric and “unstable-climacteric”, the 294 last for those RILs that were climacteric only in some biological replicates (Pereira et al., 2020). 295 We first compared the three groups by doing a PCA using the mean subset (Fig. 3). Non- 296 climacteric RILs were grouped apart from the climacteric lines and showed less diversity, in 297 both rind and flesh. Climacteric RILs, which produced a more diverse profile and a higher 298 quantity of VOCs, presented higher dispersion. “Unstable-climacteric” RILs were distributed 299 within the two other groups. The same pattern was observed when each individual replicate 300 was analyzed (Supplementary Fig. S2). In rind tissue, the metabolites that represented the 301 largest positive and negative effects for each principal component were 2-ethylfuran and 302 hexyl acetate, respectively, in PC1 (X-axis), and ethyl octanoate and β-caryophyllene in PC2 (Y- 303 axis). In flesh, PC1 was determined principally by 2-methylbutyl acetate and 2-methyl-3- 304 pentanone, while PC2 was determined by 1-octen-3-ol and 2-methylpropyl butanoate. 305 The relative amount of each chemical group was calculated and compared between 306 climacteric and non-climacteric RILs for each subset (Fig. 4). Due to the dispersion of 307 “unstable-climacteric” lines in volatile content, they were excluded from the analysis. In 308 climacteric lines, esters were predominant (45-80% rind, 75-90% flesh), followed by aldehydes 309 (20-55% rind, 5-15% flesh). Non-climacteric lines had a more variable phenotype depending 310 on the tissue. In rind, aldehydes were the main group (50-95%), followed by esters (5-45%), 311 while in flesh both esters and aldehydes showed high and variable contents ranging from 20- 312 90% for esters and 5-60% for aldehydes. Therefore, an enrichment of esters was observed in 313 the flesh of non-climacteric RILs compared to rind tissue. Alcohols and the rest of the chemical 314 groups showed similar mass percentages in climacteric and non-climacteric RILs. 19 315 316 4. Discussion 317 Aroma is one of the main determinants of melon flavor. Insights into the genetic control of 318 VOC biosynthesis and regulation would facilitate its incorporation in breeding programs, 319 leading to varieties with superior flavor and better consumer’s acceptance. The composition 320 of rind aroma has been neglected in some studies, despite its relevance for fruit quality and 321 acceptability and divergent VOC profiling. In this study, we used PS and VED which are both 322 elite commercial varieties with good flavor but differ in ripening behavior, fruit color and 323 volatile profile, among other fruit traits. 324 4.1. Volatile profile 325 From the 82 VOCs detected, three had not been previously identified in melon, to our 326 knowledge (Table 1). The number of VOCs found was in agreement with other GC-MS 327 experiments in melon that used SPME fibers (Shi et al., 2020). We found compounds that had 328 been previously described as key compounds for flavor in muskmelons and honeydews. For 329 instance, 6-nonenal (Z) and nonanal, which give melon-like and citrus notes, and ethyl acetate, 330 which evokes butterscotch (Kemp, Knavel, & Stoltz, 1972; Lignou, Parker, Oruna-Concha, & 331 Mottram, 2013; Wyllie, Leach, Wang, & Shewfelt, 1995). Other compounds like 2-methylbutyl 332 acetate, methyl 2-methylbutanoate, ethyl 2-methylpropanoate, hexyl acetate, ethyl 333 butanoate and ethyl hexanoate, which were more abundant in VED melons, have been 334 reported as important fruity, floral and sweet compounds (Lignou et al., 2013; Pang et al., 335 2012; Wyllie et al., 1995). Another key compound found in the population was 2,6-nonadienal 336 (E,Z), which gives cucumber-like notes and is the main compound in honeydews (Perry, Wang, 20 337 & Lin, 2009). All the cited key VOCs were detected above its reported odor threshold, 338 suggesting that they play an important role in PS and VED final flavor. 339 Rind accumulated higher levels of VOCs, probably due to the lesser water content as 340 compared to flesh, and, remarkably, presented less variability than flesh tissues 341 (Supplementary Fig. S1 and Supplementary Table S1-2). The higher production of VOCs in VED 342 as compared to PS correlates with a more intense smell. Fruity aroma of VED melons comes 343 mainly from esters in both rind and flesh, in agreement with several studies that described 344 Cantaloupe aroma (Esteras, Rambla, Sánchez, Granell, & Picó, 2020; Esteras et al., 2018; 345 Obando-Ulloa et al., 2008). Conversely, the fresh aroma of PS comes from a high accumulation 346 of aldehydes, also shown in previous studies (Esteras et al., 2020, 2018; Obando-Ulloa et al., 347 2008). Hyb melons also displayed many volatile compounds and had an intense fruity aroma, 348 likely conferred by high amounts of esters. The transgressive segregation showed for 349 heptanal, methyl hexanoate and decanal in Hyb melons exposed the effect of heterosis. The 350 volatile profile of the RILs was very diverse, although most lines were more similar to VED than 351 to PS, as the Hyb. Therefore, some VED alleles probably have a dominant effect on the final 352 volatile profile. A similar pattern was observed for climacteric ripening, which determines to 353 a high extent the volatile profile (Pereira et al., 2020). 354 Ethyl esters, acetates, and compounds coming from the same biosynthetic pathway were 355 positively correlated (Fig. 1). The ethyl ester correlation cluster could be explained by the 356 activity of alcohol acyltransferases able to transform several substrates (El-Sharkawy et al., 357 2005; Yahyaoui et al., 2002). Another ester cluster including several alcohols could be related 358 to ethylene, as both alcohol dehydrogenases and acyltransferases are associated with 359 ethylene production (Manríquez et al., 2006; Shalit et al., 2001). The rest of correlation 21 360 clusters also grouped compounds from the same chemical group or the same biochemical 361 origin, being probably led by the same key enzymes. For instance, hexanoate or nonanal 362 derivatives are lipid derived VOCs (Schwab et al., 2008), β-caryophyllene and δ-cadinene are 363 terpenoids, and 2-methylbutanal or benzene derivatives belong to isoleucine and 364 phenylalanine pathways, respectively, through the CmBCAT1 and CmArAT1 enzymes (Gonda 365 et al., 2010). The inverse correlation of furans and certain benzenoids observed in PS, with 366 most esters abundant in VED, could be explained by the different ripening behavior of both 367 accessions. 368 4.2. QTL mapping 369 We found a similar number of QTLs (166) and segregating compounds (63) (Table 2 and Fig. 370 2) as Galpaz et al. (2018), which also used VOC profiling in a RIL population for QTL analysis 371 (FDR<10 %, 145 QTLs for 58 different compounds in 99 RILs). The median length of the QTLs 372 (9.2 cM and 1.96 Mb) is comparable with the 9.42 cM and 0.94 Mb obtained by Pereira et al. 373 (2018) using the same population for other traits. The maximum percentage of variance 374 explained by a QTL in our study was 45.7 %, near to a monogenic inheritance, but the majority 375 ranged from 15-25 %, which suggests a complex inheritance in VOCs synthesis. Similar values 376 have been reported in melon (Galpaz et al., 2018). Multiple QTLs on different chromosomes 377 controlled the amount of certain compounds, also indicating multigenic inheritance. 378 An important environmental effect was observed in the volatile composition of RILs, as 72 379 QTLs were found only in one year. Despite the environmental effect, 30.12 % of the QTLs were 380 detected in a minimum of two subsets. When using the mean subset, which should be more 381 robust, we found consistent results for the QTLs detected in more than one subset. We 22 382 decided to focus on the most reliable QTLs detected at least in two subsets and in those 383 forming clusters. 384 The detection of clusters of QTLs could be due to a common regulation, a common step in a 385 biochemical pathway or an enzyme capable of transforming several substrates, as had been 386 previously reported in melon (El-Sharkawy et al., 2005; Gonda et al., 2010). For instance, 387 several isoleucine derived VOCs were controlled by the QTL cluster C2 on Chr02. The length 388 of the confidence interval determined for each cluster was variable, ranging from 0.39 Mb to 389 22.44 Mb. Some of them contained 100 genes or less, such as the sulfur esters cluster on 390 Chr03 and the benzaldehyde and benzoic acid cluster on Chr05. 391 4.2.1. Branched-chain amino acid aminotransferases as first-step enzymes in the volatile 392 biosynthesis 393 The sulfur compounds ethyl(methylthio)acetate and ethyl 3-(methylthio)propanoate are 394 important odorants giving grassy and earthy aromas (Wyllie et al., 1994). A QTL cluster 395 modulating the production of these compounds was detected at the end of Chr03 partially 396 overlapping with a 2-ethylfuran QTL. This cluster includes the previously described CmBCAT1 397 gene (MELO3C010776) at 20-40 Kb from the maximum LOD peak. The aminotransferase 398 CmBCAT1 is involved in the first step of the L-leucine, L-isoleucine and L-valine catabolic 399 pathways when expressed in vitro (Gonda et al., 2010). Previous studies in Arabidopsis have 400 shown that BCAT4, a homologue of CmBCAT1, participates in the methionine elongation 401 pathway (Schuster, Knill, Reichelt, Gershenzon, & Binder, 2006). Therefore, the melon enzyme 402 may participate in the deamination of L-methionine to form the precursor of ethyl 3- 403 (methylthio)propanoate. According to transcriptomic data from fruitENCODE 404 (http://www.epigenome.cuhk.edu.hk/encode.html), MELO3C010776 is highly expressed in 23 405 VED flesh compared to PS. The two sulfur compounds were only detected in VED, and 406 interestingly, correlate with the ethylene production. Apart from MELO3C010776, an 407 ethylene-responsive transcription factor and an ACC synthase are present in this interval and 408 could also have an effect for this trait. In addition, this region overlaps with an aroma QTL 409 described by Pereira et al. (2020), which could be related to the CmBCAT1 activity. 410 A BCAT-like gene (MELO3C003454) was found in the EF_A2MPE_4.1 interval on Chr04. 2- 411 Methylpropyl acetate is an important odorant described by (Pang et al., 2012) and having a 412 floral aroma. In the presence of valine, higher quantities of 2-methylpropyl acetate were 413 produced (Gonda et al., 2010). This gene could have a similar activity to CmBCAT1. A BLAST 414 search revealed that MELO3C003454 is similar to predicted chloroplastic D-aminoacid 415 aminotransferases in peach, strawberry and apple. Thus, a hypothetical activity of this enzyme 416 could be the deamination of D-valine into the precursor of 2-methylpropanal. The gene 417 underlying 2-methylbutanal derivatives’ QTL cluster on Chr02 could also be an 418 aminotransferase, but no annotated BCAT-encoding gene has been found within this interval. 419 4.2.2. Aldehyde oxidases and dehydrogenases 420 Benzaldehyde and benzyl alcohol are also important odorants in melon, having almond and 421 floral-fruity aroma, respectively (Jordán, Shaw, & Goodner, 2001; Pang et al., 2012). These 422 compounds were controlled by a QTL on Chr05 in flesh, which includes three aldehyde 423 oxidases annotated less than 30 Kb downstream the maximum LOD peak. Although 424 benzaldehyde is produced at high quantities in rind, these enzymes could also be important 425 for flesh aroma. A BLAST analysis and a search for orthologues in PLAZA 4.0 (Van Bel et al., 426 2018) revealed that MELO3C014719 is an indole-3-acetaldehyde oxidase-like and is similar to 427 the A. thaliana AAO4 protein, which was reported to transform benzaldehyde into benzoic 24 428 acid in siliques (Ibdah, Chen, Wilkerson, & Pichersky, 2009). The transcriptomic data from 429 fruitENCODE supports this idea, as MELO3C014719 is more expressed in VED flesh than in PS. 430 Both functional annotation and transcriptomic data point out MELO3C014719 as a likely 431 candidate gene. 432 We identified another candidate gene within the 2-methylbutanol QTL interval, also on Chr05. 433 Although 2-methylbutanol is not a potent odorant, it is the precursor of 2-methylbutyl 434 acetate, described by Wyllie et al. (1995) and Lignou et al. (2013) as a key compound in melon 435 flavor giving banana and pear notes. MELO3C014601, an aldehyde dehydrogenase, could 436 catalyze the dehydrogenation of 2-methylbutanal to 2-methylbutanoate, and therefore 437 limiting the synthesis of the corresponding alcohol. However, the amount of alcohols and 438 acids also depends on the activity of alcohol acyltransferases (El-Sharkawy et al., 2005). 439 Functional validation would be needed to confirm this candidate gene. 440 4.2.3. Alcohol acyltransferases in ester formation 441 Ester synthesis is known to be driven by alcohol acyltransferases. The characterization of 442 AAT1, AAT2, AAT3 and AAT4 showed their ability to transform a great variety of substrates (El- 443 Sharkawy et al., 2005; Yahyaoui et al., 2002). These enzymes were described as key steps in 444 the unique aroma of melon (Liu et al., 2020). AAT1, AAT2 and AAT3 are located within the first 445 QTL cluster on Chr11 at 0-11.7 cM (C11.1). AAT1 is the most active enzyme and prefers C6 446 substrates, AAT3 prefers benzyl alcohol and acetate as substrates and AAT2 is inactive in VED 447 (El-Sharkawy et al., 2005; Yahyaoui et al., 2002). Interestingly, ethyl acetate and ethyl 448 propanoate were tested but not detected as products of these three enzymes (El-Sharkawy et 449 al., 2005). Other AATs in the interval such as MELO3C024764 and MELO3C024769 may be 450 involved in the synthesis of these compounds. 25 451 Within the QTL cluster on Chr11 at 81-100 cM (C11.2), some metabolic enzymes and 452 transcription factors could be involved in the synthesis of esters, but no clear candidate gene 453 could be identified. Due to the interest of ester compounds, a fine mapping of this region will 454 be performed to help identifying possible candidates within the 234 genes contained in the 455 cluster (Table 2). 456 4.2.4. Fatty-acid pathway 457 Fatty-acid-derived VOCs such as 2,6-nonadienal and 6-nonenal (Z), which give honeydew and 458 cucumber notes, form part of the key VOCs identified in melon (Pang et al., 2012). 459 Lipoxygenases increase their expression at the final stages of fruit ripening. These enzymes 460 can also be enhanced by the addition of fatty acid precursors (Tang et al., 2015). On Chr05, 461 EF_BEE_5.1, a QTL for ethyl butanoate, another relevant odorant with ripe melon aroma 462 (Lignou et al., 2013), contains a cluster of 10 lipoxygenases and a γ-aminobutyrate 463 transaminase that could be involved in the formation of the butanoate backbone. In the 464 nonanal cluster C5.3, we identified 2 lipid-transfer proteins that could be involved in the first 465 steps of the synthesis. On Chr06, the QTL cluster C6.1 for ER/F_3HX1OAZ_6.1/2 and 466 ALR_3O1OZ_6.1 contains β-oxidation-related genes such as a 3-ketoacyl-CoA thiolase and a 467 synthase (Goepfert & Poirier, 2007). These candidate genes have also been found within 468 ER_BBE_7.1 interval however its activity would need to be validated. 469 4.2.5. Carotenoid derivatives 470 Some terpenoids such as β-cyclocitral and 6-methyl-5-hepten-2-one are important odorants 471 in melon, giving tropical-like and peculiar aromas (Gonda et al., 2016; Pang et al., 2012). Their 472 biosynthesis starts from isopentenyl diphosphate (IPP) or dimethyl allyl diphosphate 473 (DMAPP), which can form chains to create the precursors of mono-, sesquiterpenes and 26 474 carotenoids. Two terpene synthases in ‘Dulce’ and ‘Noy Yizre’el’ accessions have been already 475 described in melon (Portnoy et al., 2008). Carotenoid cleavage dioxygenases such as CmCCD1 476 can cleave larger terpenes to generate VOCs such as geranylacetone, 6-methyl-5-hepten-2- 477 one and ionone (Ibdah et al., 2006; Vogel, Tan, McCarty, & Klee, 2008). In our study, we found 478 a QTL cluster with the highest LOD value for 6-methyl-5-hepten-2-one, and two other QTLs for 479 geranylacetone and β-cyclohomocitral. All of them are apocarotenoids, and the positive 480 additive effect is given by the VED allele. This QTL cluster is located on Chr09, and within its 481 interval we found the CmOr gene, MELO3C005449, responsible for the orange flesh color 482 (Tzuri et al., 2015). CmOr segregates in the RIL population, and we observed that only orange 483 fleshed lines (Pereira et al., 2018) accumulated apocarotenoids. Thus, CmOr is likely the 484 underlying gene for the apocarotenoid QTL cluster. This correlation between orange flesh 485 color and the presence of apocarotenoids was also seen in previous studies (Esteras et al., 486 2018; Ibdah et al., 2006). A new QTL mapping including only the orange fleshed RILs did not 487 reveal any new QTLs for those VOCs. 488 4.2.6. Aroma as an ethylene-dependent trait 489 The association between aroma profile and climacteric ripening in melon has been largely 490 studied (Flores et al., 2002; Li et al., 2016; Obando-Ulloa et al., 2008; Pereira et al., 2020), since 491 both the aroma and other traits associated to ripening have a major influence on fruit quality. 492 The ethylene peak characteristic of climacteric ripening promotes ester and alcohol formation 493 (El-Sharkawy et al., 2005; Li et al., 2016; Manríquez et al., 2006). Indeed, the presence of 494 aroma has previously been used as a climacteric trait (Pereira et al., 2020). 495 As previously reported by Obando-Ulloa et al. (2008) and Esteras et al. (2018), we observed a 496 predominance of esters in climacteric lines and aldehydes in non-climacteric lines (Fig. 4). 27 497 Moreover, esters were the principal components leading to the climacteric profile in the PCA 498 (Fig. 3). The high diversity in the VOC profile of climacteric lines had also been seen when 499 comparing climacteric and non-climacteric accessions (Esteras et al., 2018; Moing et al., 2020). 500 We have also noticed that the balance between esters and aldehydes remains constant, 501 summing up more than 90 % of the total amount of VOCs (Supplementary Table S2). This 502 balance could be explained by the ethylene-dependent nature of alcohol dehydrogenases and 503 alcohol acyltransferases (El-Sharkawy et al., 2005; Manríquez et al., 2006). Shi et al. (2020) 504 classified 39 melon cultivars according to their aldehyde/ester ratio and observed that several 505 cultivars had an intermediate phenotype. The evaluation of the aroma production in our RIL 506 population provides nuances to the previous classification between climacteric and non- 507 climacteric behavior, since some RILs classified as climacteric produce high levels of aldehydes 508 and others classified as non-climacteric produce abundant esters (Supplementary Table S2). 509 In addition, ester abundance was increased by PS alleles in Chr02 and Chr11. Esteras et al. 510 (2018) also detected a climacteric-like profile, linked to esters, in some inodorus accessions. 511 In fact, climacteric ripening is not an absolute trait, as it has been seen in the RIL population 512 showing different degrees of ethylene production (Pereira et al., 2020). 513 Many of our QTL intervals contain ethylene-related transcription factors that could play a role 514 in the production of VOCs. For instance, the cluster on Chr02 overlaps with ripening-related 515 QTLs that were previously described, such as QTLs for abscission layer formation, firmness, 516 ethylene production and earliness of aroma (Pereira et al., 2020). QTLs on chromosomes 3, 5, 517 6, 7, 8, 10 and 11 also overlap with harvest, fruit abscission, width of ethylene peak, 518 chlorophyll degradation and aroma QTLs (Galpaz et al., 2018; Obando-Ulloa et al., 2010; 519 Pereira et al., 2020). Specifically, 92 QTLs for very diverse compounds converged into the QTL 520 cluster on Chr08. These QTLs had high LOD scores and reproducibility in both 2015 and 2016 28 521 samples. In this region, ETHQV8.1 has been recently described as a major regulator of 522 climacteric ripening in melon (Pereira et al., 2020). Since 66 % of the Chr08 QTLs described in 523 this work overlap with ETHQV8.1, including esters, aliphatic alcohols and phenylalanine 524 derivatives, these compounds are most likely ethylene dependent. The region that does not 525 contain ETHQV8.1, from 89-120 cM, included QTLs for ethyl esters and aldehydes. This region 526 overlaps with many benzenoid, terpenoid and sulfur compounds QTLs identified in Galpaz et 527 al. (2018), together with octanal and 3,6-nonadienal-(Z,Z) QTLs in Obando-Ulloa et al. (2010). 528 Our results confirm that Chr08 is playing an important role in aroma production in our 529 population. 530 5. Conclusions 531 Our work was focused on the genetics behind the biosynthesis of VOCs in melon to shed light 532 to the complex metabolic pathways that affect its flavor. We evaluated two melon tissues at 533 ripen stage, rind and flesh, since both affect consumer preferences in the market, but rind 534 VOCs profile is still poorly known. In this study, we described the aroma profile in a PS x VED 535 RIL population and its relationship with ripening behavior, mapping 166 QTLs to the melon 536 genome. A QTL cluster on Chr08 was the most important contributor to volatile biosynthesis, 537 mainly ethylene-related and overlapping with the major ripening QTL ETHQV8.1 (Pereira et 538 al., 2020). Moreover, QTL clusters identified in Ch02, Chr05, Chr06 and Chr11 contributed to 539 decipher different VOCs synthesis pathways. These findings provide the ground for further 540 fine-mapping projects and the functional validation of the candidate genes proposed related 541 to aroma formation in melon. 542 543 CRediT authorship contribution statement 29 544 Carlos Mayobre: Methodology, Formal analysis, Writing - Original Draft. Lara Pereira: 545 Methodology, Writing – Review & Editing. Abdelali Eltahiri: Methodology, Formal analysis. 546 Einat Bar: Methodology. Efraim Lewinsohn: Writing – Review & Editing. Jordi Garcia-Mas: 547 Conceptualization, Writing – Review & Editing, Supervision, Funding Acquisition. Marta Pujol: 548 Conceptualization, Writing – Review & Editing, Supervision. 549 Declaration of Competing Interest 550 The authors declare that they have no known competing financial interests or personal 551 relationships that could have appeared to influence the work reported in this paper. 552 Acknowledgements 553 This work was supported by the Spanish Ministry of Economy and Competitiveness grants 554 AGL2015–64625-C2–1-R and RTI2018-097665-B-C2, the Severo Ochoa Programme for Centres 555 of Excellence in R&D 2016-2010 (SEV-2015-0533), the CERCA Programme/Generalitat de 556 Catalunya and 2017 SGR 1319 grant from the Generalitat de Catalunya to JGM. The authors 557 also wish to thank the support of the Secretaria d'Universitats i Recerca del Departament 558 d'Empresa i Coneixement de la Generalitat de Catalunya for having funded the 2019 FI_B 559 00124 project, and the co-funding of the European Social Fund (ESF – “ESF is investing in your 560 future”) from the European Union. Special thanks to Fuensanta García for technical assistance 561 in field and lab operations, as well as M. Bernardo for advice on data treatment and statistical 562 analysis. 563 564 Appendix A. Supplementary Data 30 565 Supplementary Fig. S1. Boxplot showing the amount of VOCs according to their chemical 566 group in rind (A) and flesh (B) samples of the parental lines (PS, VED) and the F1 (Hyb). 567 Supplementary Fig. S2. Principal component analysis (PCA) of the PS x VED RIL population 568 volatile profile. The upper part represents rind PCAs for T3 (A), T4 (B) and T5 (C) subsets; the 569 lower part represents flesh PCAs for T3 (D), T4 (E) and T5 (F) subsets. Different ellipses group 570 climacteric lines (C), non-climacteric lines (NC) and “unstable-climacteric” lines (UC). 571 Supplementary Table S1. Average content of volatile compounds detected in the parental 572 lines (PS, VED), the Hyb and in the RIL population for rind and flesh tissues. 573 Supplementary Table S2. Total VOC content according to the chemical classification and 574 ripening classification of RILs. 575 Supplementary Table S3. VOC content in ng g-1 frozen tissue of each VOC in rind. 576 Supplementary Table S4. VOC content in ng g-1 frozen tissue of each VOC in flesh. 577 Supplementary Table S5. Positions of the QTLs detected in the PS x VED RIL population. 578 579 References 580 El-Sharkawy, I., Manríquez, D., Flores, F. B., Regad, F., Bouzayen, M., Latché, A., & Pech, J. C. (2005). Functional 581 characterization of a melon alcohol acyl-transferase gene family involved in the biosynthesis of ester 582 volatiles. Identification of the crucial role of a threonine residue for enzyme activity. Plant Molecular 583 Biology, 59(2), 345–362. https://doi.org/10.1007/s11103-005-8884-y 584 Esteras, C., Rambla, J. L., Sánchez, G., Granell, A., & Picó, M. B. (2020). Melon Genetic Resources 585 Characterization for Rind Volatile Profile. Agronomy, 10(1512). 586 https://doi.org/10.3390/agronomy10101512 31 587 Esteras, C., Rambla, J. L., Sánchez, G., López-Gresa, M. P., González-Mas, M. C., Fernández-Trujillo, J. P., … Picó, 588 M. B. (2018). Fruit flesh volatile and carotenoid profile analysis within the Cucumis melo L. species reveals 589 unexploited variability for future genetic breeding. Journal of the Science of Food and Agriculture, 98(10), 590 3915–3925. https://doi.org/10.1002/jsfa.8909 591 Flores, F., El Yahyaoui, F., de Billerbeck, G., Romojaro, F., Latché, A., Bouzayen, M., … Ambid, C. (2002). Role of 592 ethylene in the biosynthetic pathway of aliphatic ester aroma volatiles in Charentais Cantaloupe melons. 593 Journal of Experimental Botany, 53(367), 201–206. https://doi.org/10.1093/jexbot/53.367.201 594 Freilich, S., Lev, S., Gonda, I., Reuveni, E., Portnoy, V., Oren, E., … Katzir, N. (2015). Systems approach for 595 exploring the intricate associations between sweetness, color and aroma in melon fruits. BMC Plant 596 Biology, 15(1), 1–16. https://doi.org/10.1186/s12870-015-0449-x 597 Galpaz, N., Gonda, I., Shem-Tov, D., Barad, O., Tzuri, G., Lev, S., … Katzir, N. (2018). Deciphering genetic factors 598 that determine melon fruit-quality traits using RNA-Seq-based high-resolution QTL and eQTL mapping. 599 Plant Journal, 94(1), 169–191. https://doi.org/10.1111/tpj.13838 600 Garcia-Mas, J., Benjak, A., Sanseverino, W., Bourgeois, M., Mir, G., Gonzalez, V. M., … Puigdomenech, P. (2012). 601 The genome of melon (Cucumis melo L.). Proceedings of the National Academy of Sciences, 109(29), 602 11872–11877. https://doi.org/10.1073/pnas.1205415109 603 Goepfert, S., & Poirier, Y. (2007). β-Oxidation in fatty acid degradation and beyond. Current Opinion in Plant 604 Biology, 10(3), 245–251. https://doi.org/10.1016/j.pbi.2007.04.007 605 Gonda, I., Bar, E., Portnoy, V., Lev, S., Burger, J., Schaffer, A. A., … Lewinsohn, E. (2010). Branched-chain and 606 aromatic amino acid catabolism into aroma volatiles in Cucumis melo L. fruit. Journal of Experimental 607 Botany, 61(4), 1111–1123. https://doi.org/10.1093/jxb/erp390 608 Gonda, I., Burger, Y., Schaffer, A. A., Ibdah, M., Tadmor, Y., Katzir, N., … Lewinsohn, E. (2016). Biosynthesis and 609 perception of melon aroma. In Biotechnology in Flavor Production (pp. 281–305). Chichester, UK: John 610 Wiley & Sons, Ltd. https://doi.org/10.1002/9781118354056.ch11 611 Gonda, I., Davidovich-Rikanati, R., Bar, E., Lev, S., Jhirad, P., Meshulam, Y., … Lewinsohn, E. (2018). Differential 612 metabolism of L–phenylalanine in the formation of aromatic volatiles in melon (Cucumis melo L.) fruit. 32 613 Phytochemistry, 148, 122–131. https://doi.org/10.1016/j.phytochem.2017.12.018 614 Gonda, I., Lev, S., Bar, E., Sikron, N., Portnoy, V., Davidovich-Rikanati, R., … Lewinsohn, E. (2013). Catabolism of l 615 -methionine in the formation of sulfur and other volatiles in melon (Cucumis melo L.) fruit. Plant Journal, 616 74(3), 458–472. https://doi.org/10.1111/tpj.12149 617 Ibdah, M., Azulay, Y., Portnoy, V., Wasserman, B., Bar, E., Meir, A., … Lewinsohn, E. (2006). Functional 618 characterization of CmCCD1, a carotenoid cleavage dioxygenase from melon. Phytochemistry, 67(15), 619 1579–1589. https://doi.org/10.1016/j.phytochem.2006.02.009 620 Ibdah, M., Chen, Y. T., Wilkerson, C. G., & Pichersky, E. (2009). An aldehyde oxidase in developing seeds of 621 arabidopsis converts benzaldehyde to benzoic acid. Plant Physiology, 150(1), 416–423. 622 https://doi.org/10.1104/pp.109.135848 623 Jordán, M. J., Shaw, P. E., & Goodner, K. L. (2001). Volatile components in aqueous essence and fresh fruit of 624 Cucumis melo cv. Athena (Muskmelon) by GC-MS and GC-O. Journal of Agricultural and Food Chemistry, 625 49(12), 5929–5933. https://doi.org/10.1021/jf010954o 626 Kemp, T. R., Knavel, D. E., & Stoltz, L. P. (1972). cis-6-Nonenal: A flavor component of muskmelon fruit. 627 Phytochemistry, 11(11), 3321–3322. https://doi.org/10.1016/S0031-9422(00)86399-2 628 Lelievre, J.-M., Latche, A., Jones, B., Bouzayen, M., & Pech, J.-C. (1997). Ethylene and fruit ripening. Physiologia 629 Plantarum, 101, 727–739. 630 Li, Y., Qi, H., Jin, Y., Tian, X., Sui, L., & Qiu, Y. (2016). Role of ethylene in biosynthetic pathway of related-aroma 631 volatiles derived from amino acids in oriental sweet melons (Cucumis melo var. makuwa Makino). 632 Scientia Horticulturae, 201, 24–35. https://doi.org/10.1016/j.scienta.2015.12.053 633 Lignou, S., Parker, J. K., Oruna-Concha, M. J., & Mottram, D. S. (2013). Flavour profiles of three novel acidic 634 varieties of muskmelon (Cucumis melo L.). Food Chemistry, 139(1–4), 1152–1160. 635 https://doi.org/10.1016/j.foodchem.2013.01.068 636 Liu, S., Gao, P., Zhu, Q., Zhu, Z., Liu, H., Wang, X., … Luan, F. (2020). Resequencing of 297 melon accessions 637 reveals the genomic history of improvement and loci related to fruit traits in melon. Plant Biotechnology 638 Journal, 1–14. https://doi.org/10.1111/pbi.13434 33 639 Manríquez, D., El-Sharkawy, I., Flores, F. B., El-Yahyaoui, F., Regad, F., Bouzayen, M., … Pech, J. C. (2006). Two 640 highly divergent alcohol dehydrogenases of melon exhibit fruit ripening-specific expression and distinct 641 biochemical characteristics. Plant Molecular Biology, 61(4–5), 675–685. https://doi.org/10.1007/s11103- 642 006-0040-9 643 Moing, A., Allwood, J. W., Aharoni, A., Baker, J., Beale, M. H., Ben-Dor, S., … Schaffer, A. A. (2020). Comparative 644 Metabolomics and Molecular Phylogenetics of Melon (Cucumis melo, Cucurbitaceae) Biodiversity. 645 Metabolites, 10(3), 121. https://doi.org/10.3390/metabo10030121 646 Obando-Ulloa, J. M., Moreno, E., García-Mas, J., Nicolai, B., Lammertyn, J., Monforte, A. J., & Fernández- 647 Trujillo, J. P. (2008). Climacteric or non-climacteric behavior in melon fruit. 1. Aroma volatiles. 648 Postharvest Biology and Technology, 49(1), 27–37. https://doi.org/10.1016/j.postharvbio.2007.11.004 649 Obando-Ulloa, J. M., Ruiz, J., Monforte, A. J., & Fernández-Trujillo, J. P. (2010). Aroma profile of a collection of 650 near-isogenic lines of melon (Cucumis melo L.). Food Chemistry, 118(3), 815–822. 651 https://doi.org/10.1016/j.foodchem.2009.05.068 652 Pang, X., Guo, X., Qin, Z., Yao, Y., Hu, X., & Wu, J. (2012). Identification of aroma-active compounds in Jiashi 653 Muskmelon juice by GC-O-MS and OAV Calculation. Journal of Agricultural and Food Chemistry, 60(17), 654 4179–4185. https://doi.org/10.1021/jf300149m 655 Pereira, L., Ruggieri, V., Pérez, S., Alexiou, K. G., Fernández, M., Jahrmann, T., … Garcia-Mas, J. (2018). QTL 656 mapping of melon fruit quality traits using a high-density GBS-based genetic map. BMC Plant Biology, 657 18(1), 1–17. https://doi.org/10.1186/s12870-018-1537-5 658 Pereira, L., Santo Domingo, M., Ruggieri, V., Argyris, J., Phillips, M. A., Zhao, G., … Garcia-Mas, J. (2020). Genetic 659 dissection of climacteric fruit ripening in a melon population segregating for ripening behavior. 660 Horticulture Research, 7(1), 187. https://doi.org/10.1038/s41438-020-00411-z 661 Perry, P. L., Wang, Y., & Lin, J. (2009). Analysis of honeydew melon (Cucumis melo var. inodorus) flavour and 662 GC–MS/MS identification of (E,Z)-2,6-nonadienyl acetate. Flavour and Fragrance Journal, 24, 341–347. 663 Pitrat, M. (2017). Genetics and Genomics of Cucurbitaceae. (R. Grumet, N. Katzir, & J. Garcia-Mas, Eds.) (1st ed., 664 Vol. 20). Cham, Germany: Springer International Publishing. https://doi.org/10.1007/978-3-319-49332-9 34 665 Portnoy, V., Benyamini, Y., Bar, E., Harel-Beja, R., Gepstein, S., Giovannoni, J. J., … Katzir, N. (2008). The 666 molecular and biochemical basis for varietal variation in sesquiterpene content in melon (Cucumis melo 667 L.) rinds. Plant Molecular Biology, 66(6), 647–661. https://doi.org/10.1007/s11103-008-9296-6 668 Schuster, J., Knill, T., Reichelt, M., Gershenzon, J., & Binder, S. (2006). BRANCHED-CHAIN AMINOTRANSFERASE4 669 is part of the chain elongation pathway in the biosynthesis of methionine-derived glucosinolates in 670 Arabidopsis. Plant Cell, 18(10), 2664–2679. https://doi.org/10.1105/tpc.105.039339 671 Schwab, W., Davidovich-Rikanati, R., & Lewinsohn, E. (2008). Biosynthesis of plant-derived flavor compounds. 672 The Plant Journal, 54(4), 712–732. https://doi.org/10.1111/j.1365-313X.2008.03446.x 673 Shalit, M., Katzir, N., Tadmor, Y., Larkov, O., Burger, Y., Shalekhet, F., … Lewinsohn, E. (2001). Acetyl-CoA: 674 Alcohol acetyltransferase activity and aroma formation in ripening melon fruits. Journal of Agricultural 675 and Food Chemistry, 49(2), 794–799. https://doi.org/10.1021/jf001075p 676 Shi, J., Wu, H., Xiong, M., Chen, Y., Chen, J., Zhou, B., … Huang, Y. (2020). Comparative analysis of volatile 677 compounds in thirty nine melon cultivars by headspace solid-phase microextraction and gas 678 chromatography-mass spectrometry. Food Chemistry, 316(January), 126342. 679 https://doi.org/10.1016/j.foodchem.2020.126342 680 Tang, Y., Zhang, C., Cao, S., Wang, X., & Qi, H. (2015). The effect of CmLOXs on the production of volatile 681 organic compounds in four aroma types of melon (Cucumis melo). PLoS ONE, 10(11), 1–18. 682 https://doi.org/10.1371/journal.pone.0143567 683 Tzuri, G., Zhou, X., Chayut, N., Yuan, H., Portnoy, V., Meir, A., … Tadmor, Y. (2015). A “golden” SNP in CmOr 684 governs the fruit flesh color of melon (Cucumis melo). Plant Journal, 82(2), 267–279. 685 https://doi.org/10.1111/tpj.12814 686 Van Bel, M., Diels, T., Vancaester, E., Kreft, L., Botzki, A., Van De Peer, Y., … Vandepoele, K. (2018). PLAZA 4.0: 687 An integrative resource for functional, evolutionary and comparative plant genomics. Nucleic Acids 688 Research, 46(D1), D1190–D1196. https://doi.org/10.1093/nar/gkx1002 689 Vogel, J. T., Tan, B. C., McCarty, D. R., & Klee, H. J. (2008). The carotenoid cleavage dioxygenase 1 enzyme has 690 broad substrate specificity, cleaving multiple carotenoids at two different bond positions. Journal of 35 691 Biological Chemistry, 283(17), 11364–11373. https://doi.org/10.1074/jbc.M710106200 692 Wyllie, S. G., Leach, D. N., Wang, Y., & Shewfelt, R. L. (1994). Sulfur Volatiles in Cucumis melo cv. Makdimon 693 (Muskmelon) Aroma. In Sulfur Compounds in Foods (pp. 36–48). https://doi.org/10.1021/bk-1994- 694 0564.ch004 695 Wyllie, S. G., Leach, D. N., Wang, Y., & Shewfelt, R. L. (1995). Key Aroma Compounds in Melons. In Fruit Flavors 696 (pp. 248–257). https://doi.org/10.1021/bk-1995-0596.ch022 697 Yahyaoui, F. E. L., Wongs-Aree, C., Latché, A., Hackett, R., Grierson, D., & Pech, J. C. (2002). Molecular and 698 biochemical characteristics of a gene encoding an alcohol acyl-transferase involved in the generation of 699 aroma volatile esters during melon ripening. European Journal of Biochemistry, 269(9), 2359–2366. 700 https://doi.org/10.1046/j.1432-1033.2002.02892.x 701 Yano, R., Nonaka, S., & Ezura, H. (2018). Melonet-DB, a Grand RNA-Seq Gene Expression Atlas in Melon 702 (Cucumis melo L.). Plant and Cell Physiology, 59(1), e4(1-15). https://doi.org/10.1093/pcp/pcx193 703 Zhao, G., Lian, Q., Zhang, Z., Fu, Q., He, Y., Ma, S., … Huang, S. (2019). A comprehensive genome variation map 704 of melon identifies multiple domestication events and loci influencing agronomic traits. Nature Genetics, 705 51(11), 1607–1615. https://doi.org/10.1038/s41588-019-0522-8 706 36 707 708 Fig. 1. Heatmap and dendrogram showing the correlation matrix by the Spearman’s coefficient between VOCs in the RIL population for rind (A) and flesh (B). 37 709 Fig. 2. Genetic map showing QTLs found in rind and flesh. QTLs for esters (red), alcohols (blue), aldehydes (green), terpenes (orange), benzenoids (brown), ketones (pink) and furans (purple) were represented selecting LOD-1 as upper and lower limit. Black QTLs represent merged (M) QTLs coincident for several compounds. 38 710 A B 1-Octen-3-ol β-Caryophyllene VED Hexyl acetate HYB 2-Ethylfuran HYB PS PS 2-Methyl-3-pentanone 2-Methylbutyl acetate 2-Methylpropyl butanoate Ethyl octanoate VED Fig. 3. Principal component analysis (PCA) of the PS x VED RIL population volatile profile in rind (A) and flesh (B). Different ellipses group climacteric lines (C), non-climacteric lines (NC) and “unstable- climacteric” lines (UC). The arrows represent the VOCs with the most positive and negative effect for the two principal components. 39 711 Climacteric RILs Non-climacteric RILs A 712 B Fig. 4. Percentage of VOCs in RILs according to their chemistry, in rind (A) and flesh (B). Data from three subsets with plants performed in 2015 (T3) and 2016 (T4, T5) are represented. 40 713 A B Supplementary Fig. S1. Boxplot showing the amount of VOCs according to their chemical group in rind (A) and flesh (B) samples of the parental lines (PS, VED) and the F1 (Hyb). 41 714 A B C D E F Supplementary Fig. S2. Principal component analysis (PCA) of the PS x VED RIL population volatile profile. The upper part represents rind PCAs for T3 (A), T4 (B) and T5 (C) subsets; the lower part represents flesh PCAs for T3 (D), T4 (E) and T5 (F) subsets. Different ellipses group climacteric lines (C), non-climacteric lines (NC) and “unstable-climacteric” lines (UC). 42