Mapping of Candidate Genes Involved in Bud Dormancy and Flowering Time in Sweet Cherry (Prunus avium)

The timing of flowering in perennial plants is crucial for their survival in temperate climates and is regulated by the duration of bud dormancy. Bud dormancy release and bud break depend on the perception of cumulative chilling during endodormancy and heat during the bud development. The objectives of this work were to identify candidate genes involved in dormancy and flowering processes in sweet cherry, their mapping in two mapping progenies ‘Regina’ × ‘Garnet’ and ‘Regina’ × ‘Lapins’, and to select those candidate genes which co-localized with quantitative trait loci (QTLs) associated with temperature requirements for bud dormancy release and flowering. Based on available data on flowering processes in various species, a list of 79 candidate genes was established. The peach and sweet cherry orthologs were identified and primers were designed to amplify sweet cherry candidate gene fragments. Based on the amplified sequences of the three parents of the mapping progenies, SNPs segregations in the progenies were identified. Thirty five candidate genes were genetically mapped in at least one of the two progenies and all were in silico mapped. Co-localization between candidate genes and QTLs associated with temperature requirements and flowering date were identified for the first time in sweet cherry. The allelic composition of the candidate genes located in the major QTL for heat requirements and flowering date located on linkage group 4 have a significant effect on these two traits indicating their potential use for breeding programs in sweet cherry to select new varieties adapted to putative future climatic conditions.


Introduction
The life cycle control of plants is complex, especially for perennial species with repeating cycles under contrasting environments. This is particularly true for flowering as it is one of the most fundamental and complex developmental processes in plants and needs to be synchronized with climatic conditions especially in temperate regions with alternating well-differentiated Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Regarding perennial plants, bud dormancy and flowering have been extensively studied in poplar (Populus ssp.), the model woody plant, and a strong involvement of photoperiod, temperature and gibberellic acid was reported [20,21]. However, alternative mechanisms of flowering regulation were also described in other woody perennials. In particular, dormancy release for several Rosaceae trees was shown to rely mainly on temperature sensing mechanisms [22,23] making them vulnerable to global warming. In Prunus, several studies suggest that chilling requirements affect flowering date more strongly than heat requirements: sweet cherry [24,25], apricot (Prunus armeniaca L.) [26], almond (Prunus amygdalus L. syn. Prunus dulcis M.) [27], and peach (Prunus persica L.) [28]. Using highly dense linkage maps [29], quantitative trait loci (QTL) for flowering date, chilling and heat requirements were detected in sweet cherry [25]. In this species, QTLs with major effects for chilling requirements and flowering date were found on linkage group (LG) 4 [25]. They were highly stable between the years of evaluation (three for the chilling requirements and five for the flowering date) and with a narrow confidence interval (less than 1 cM). This result was in agreement with previous works in almond [27], apricot [30][31][32] and peach [31,33]. Additionally, a QTL for flowering date colocalizing with the evergrowing (EVG) region was detected on LG1 in peach [33,34]and sweet cherry [25]. In peach, six tandemly repeated MADS-box genes, named DORMANCY ASSOCI-ATED MADS-box1-6 (DAM1-6) genes, were located in the EVG region and two of them (DAM5 and DAM6) were reported to be the strongest candidate genes (CGs) for the major QTL for chilling requirements at the lower end of LG1 [35][36][37]. These genes are not expressed in the peach mutant evergrowing, which fails to both cease growth and enter dormancy under dormancy-inducing conditions [38]. More recently, histone modifications in the DAM6 gene were investigated and results suggested the role of epigenetic mechanisms in DAM6 regulation and bud dormancy release [39]. Modification of DAM6 chromatin in a similar way to FLC, as a consequence of the prolonged exposure to low temperatures [40], suggests the existence of mechanistic similarities between vernalization and bud dormancy processes. Several additional genes were also reported to play an important role in dormancy processes in Rosaceae species. In Japanese apricot (Prunus mume Sieb. et Zucc.) and in raspberry (Rubus idaeus L.) the SHORT VEGETATIVE PHASE (SVP) genes are down-regulated during dormancy release [41,42]. All these results suggest a complex genetic and molecular determinism of dormancy and flowering processes.
Following our work that led to the identification of QTLs for flowering, chilling and heat requirements in sweet cherry [25], we aimed to further investigate these traits and the genes and pathways potentially involved. The main objectives of this work were to identify and map genes involved in the flowering processes. Thus, genes known to be involved in flowering processes in model (Arabidopsis and poplar) and related species (peach and blackcurrant) were selected. These genes were then mapped in two sweet cherry mapping progenies and those that co-localized with flowering date QTLs already identified in these progenies were selected. This work provides further information on genetic and molecular determinism of flowering phenology and its findings will help cherry breeders to implement efficient strategies focused on the adaptation to future climatic conditions. In this work, selected varieties, 'Garnet', 'Lapins' and 'Regina' differ for traits involved in flowering phenology. 'Lapins' and 'Garnet' are considered early flowering varieties (before reference cultivar 'Burlat') and 'Regina' displays a late flowering phenotype (after 'Burlat').
All trees were cultivated in the experimental orchard of INRA-Bordeaux research center located at Toulenne, near Bordeaux, France. The trees were grown on their own roots, on deep loamy soil on the bank of Garonne River at 15 m above sea level, lat. 44.57 N, long. 0.28 W, in a mild winter area with a yearly average rainfall of 825 mm.
R×L and R×G mapping progenies were previously used for a QTL detection study for the three traits, chilling requirements, heat requirements and flowering date [25].

Sampling and DNA extraction
For genotyping, random samples of 1 g of young leaves were collected from 'Regina', 'Garnet', 'Lapins' and all individuals of both progenies. Samples were frozen with liquid nitrogen, grounded into a fine powder and stored at -80°C. DNA extraction was from 100 mg of frozen leaves powder, with the DNeasy Plant Mini Kit (Qiagen) following manufacturing instructions and stored at -20°C.

Candidate genes selection
CGs involved in dormancy and flowering processes were selected from already published data on Prunus species [25,35,43], black currant (Ribes nigrum L.) [44], poplar (Populus trichocarpa) [20] and Arabidopsis [45][46][47][48], model species for the study of dormancy or flowering mechanisms (S1 Table). Genes controlling dormancy and vernalization and the different flowering pathways (temperature, photoperiod, gibberellin, autonomous) were analyzed. For each selected gene, the peach orthologous sequence was identified in the Genome Database for Rosaceae (GDR). As the analyses were performed before the release of the second version of the peach sequence, the peach CGs were referred according to the prediction gene code of the Peach genome v1.0 (named ppaxxx) [49]. Following the release of the Peach genome v2.0.a1, the corresponding sequences were identified in this new version (named Prupe.XXX). Sweet cherry orthologs were searched in the sweet cherry 'Regina' transcribed sequences [50]. Orthologs were identified using BLASTp on peach and sweet cherry with an E value threshold of 10−8. Bidirectional best hit was used to test each sweet cherry sequence identified against the genome from which the original CG was defined to ensure that it was the accurate ortholog.
Candidate genes mapping SNP identification. In order to map the CGs, heterozygous SNPs are required for at least one of the parents. Sequence alignments were visualized using the software BioEdit Sequence Alignment Editor [51]. Primers were designed using the software online Primer3Plus (http:// www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi/) (S2 Table). PCR amplifications were performed in a final volume of 36 μl including 0.2 μM of each primer, 200 μM of each dNTP, 1.5 mM MgCl2, 10 ng of DNA of each parent and 0.4 U Taq polymerase (Invitrogen). PCR products were controlled on agarose gel and sent for sequencing to GENOME Express (http://www.beckmangenomics.com/). Heterozygous SNPs in each parent ('Garnet', 'Lapins' and 'Regina') were identified by visual inspection of DNA sequence chromatogram (BioEdit Sequence Alignment Editor).
SNP genotyping. Sequences of a minimum of 50 bp up and downstream from the SNP were used for primer design using Sequenom MassARRAY Assay Design 3.1 software with default parameters (S3 Table). The genotyping for the two progenies and their parents was performed according to the iPLEX protocol from Sequenom (http://www.sequenom.com/Home) using the MassARRAY Analyzer 4 System (Sequenom, San Diego, USA) at the Genomic and Sequencing Facility of Bordeaux, France. Data were analyzed with Typer4 software (Sequenom, San Diego, USA).
In order to identify if the SNPs associated amino acid change, the DNA sequences of the CGs were translated into protein using MEGA6 [52].
CGs mapping in the two progenies. The CGs were then mapped on the available linkage maps constructed for the two progenies R×L and R×G [25,29] using JoinMap1 4.0 [53]. A minimum LOD value of 3 and a recombinant threshold of 0.35 were chosen for mapping. Kosambi's mapping function was used to convert recombination frequency into map distance [54]. The SNP markers were named according to the original full names used in the RosBREED cherry 6K SNP array, including the physical and genetic positions and the targeted genome [55]. To facilitate reading of data, SNP names were contracted as follows: 'RosBREED_snp_s-weet_cherry_Pp1_00493090' is written 'Rsweet_1_00493090'. Mapped CGs in sweet cherry progenies were named preceded by the prefix "Pav". The graphical presentation of linkage maps was performed using the MapChart software version 2.2 [56].
CGs that could not be genetically mapped were mapped in silico using the peach genome v2.0.a1 sequence. This was performed using the peach physical position of the SNP markers located in the sweet cherry maps flanking each CG. In silico mapped CGs were preceded by the prefix "is".
In order to select the CGs located in the QTLs controlling the flowering date identified in the two progenies (R×L and R×G) and in QTLs for chilling and heat requirements identified in R×G [25], the physical positions of the confidence interval of the QTLs was estimated from the physical position of the closest SNP markers from each side of the confidence interval according to the Peach genome v2.0.a1.

Allelic effect of candidate genes on the phenological traits
Three SNPs were used for haplotype construction, including SNPs located in NUA, EMF2 and ARP4. T-tests (P>0.05) were used to determinate significant differences between means of each haplotype. The statistical analyses were performed using the R Commander package of the software R 2.15.0 (R_Development_Core_Team, 2012).
The phenotypic traits were evaluated according to procedures described by Castède et al. [25]. The Dynamic model [57] was used to calculate chilling accumulation as the chill portion (CP) and the model proposed by Richardson et al. [58] was chosen to calculate the heat accumulation as growing degree hours (GDH). Flowering dates were evaluated when 70% of floral buds reached the full bloom stage in the field.

Candidate genes identification
A list of 79 functional CGs was established based on available data on flowering processes (S1 Table). Among these 79 genes, 20 have been shown to be involved in the photoperiod pathway, 25 in the temperature pathway, 15 in dormancy processes, five in the gibberellin pathway, two in the autonomous pathway and three were floral integrators. Following this selection, both peach and sweet cherry orthologous sequences were identified for each gene (S1 Table). Peach CG orthologous sequences were identified using the two versions of the genome sequence (Prunus Persica v1.0 and v2.0.a1). Sweet cherry orthologous cDNA sequences were found for each CG (S1 File). Sweet cherry 'Regina' transcribed sequences of candidate genes are available in the GDR http://www.rosaceae.org (Accession number tfGDR1019).

Candidate genes mapping
The primers used to amplify sweet cherry CGs are indicated in S2 Table. For CGs for which no sweet cherry accurate ortholog was found by bidirectional best hit, primers were designed from the peach sequence (S2 File). Sequence of the DAM6 genomic amplicon from the three parents revealed a strong homology with a cDNA fragment of the 'Regina' unigene corresponding to part of the peach DAM6 gene (S2 File). Amplified sequences were obtained for 76 genes, 67 of them with a single amplified fragment (S3 Table). For all candidate genes, the amplified products from the three parents 'Regina', 'Lapins' and 'Garnet' were sequenced and aligned with the peach orthologous sequence (S2 File). Multiple amplifications were obtained for nine genes, probably due to the presence of multigenic families (S3 Table). SNPs were detected for 39 genes. For two of them (FRI and GH17-39), two SNPs were identified. Twenty one SNPs were found in intronic regions (within 19 genes), 19 in exons and one in the 5'UTR region. Among the 19 SNPs found in exons, 6 induced nonsynonymous substitutions (S3 Table). Twenty nine SNPs were transitions (19 were C/T polymorphisms while 10 were A/G polymorphisms) and 12 were transversions (4 A/T, 3 G/T, 3 C/G, 2 A/C). PavVRN2 was placed on the 'Regina' linkage map in the R×L progeny but could not be mapped in this parent in the R×G progeny due to an ambiguous genotyping. In addition, for four genes (FIO1, VIL1, GH17-39, GH17-101), ambiguous genotyping in both progenies did not allow their mapping.
They were all mapped in sweet cherry at the same location than their peach orthologs, except one, orthologous of SUPPRESSOR OF AUXIN RESISTANCE 3 (PavSAR3), mapped on LG8 in sweet cherry while it is located on LG2 in peach. Twenty eight genes could not be mapped because no polymorphic SNP was detected among the parents of the mapping progenies. However, for each CG, the physical position of the peach ortholog was determined on the Prunus persica v2.0.a1 genome using GDR generic Genome Browser (GBrowse) (S3 Table). Subsequently, all CGs that were not genetically mapped were localized in silico in sweet cherry based on the high synteny between sweet cherry and peach as shown in the high density sweet cherry map using Peach genome v1.0 markers [29].
The mapping data are available in S4 Table and in the GDR http://www.rosaceae.org (Accession number tfGDR1019).

Positioning of CGs within QTL intervals
Among the 79 CGs analyzed in this study, 57 CGs co-localize with QTLs for chilling or heat requirements or for flowering date, 21 CGs being genetically mapped and 38 CGs being in silico mapped (Fig 2, S5 Table).
On LG1, the QTL regions for heat and chilling requirements are very large, in particular for the 'Garnet' map, covering nearly all the linkage group especially for heat requirements. Consequently, nearly all CGs are located in QTLs (16 among the 18 CGs located in this LG1). The QTL for flowering date detected in 'Lapins' is much smaller, covering a region of 5.4 cM (S5 Table, Fig 3). PavHYL1, isDAM5 and isDAM6 are located in this QTL region, explaining 8% of The SNP maps were already reported for R×L [29] and for R×G [25]. The SNP markers are named according to the original full names used in the RosBREED cherry 6K SNP array, including the physical and genetic positions and the targeted genome [55] but with contraction: the variation in the R×L progeny (Fig 3, S5 Table). This genomic region is conserved between sweet cherry and peach (Fig 3).
On LG2, PavLHY, isSOC1, isCRP, isSPA1, isPRR9, isFLD are located within the region of the QTLs for chilling requirements and flowering date, except PavLHY located only in the QTL for chilling requirements.
for example 'RosBREED_snp_sweet_cherry_Pp1_00493090' is written 'Rsweet_1_00493090'. The candidate genes are indicated in red and were genetically mapped using SNP markers described in S3   On LG3, PavFRS6, PavFRS8, isMBF1, isSDG2, isCAC, isCALd are located within the region of the QTL for flowering date, and isCALd is also localized within the QTL for heat requirements.
From the 14 CGs located on LG4, 8 are mapped within the small interval of QTLs showing the highest effect for flowering date and for chilling requirements, explaining 36.3% and 17.5% of the phenotypic variation in 'Regina' from the R×G progeny, respectively [25]: PavEMF2, PavARP4, isGA2ox, KAURENE SYNTHASE (KS) CGs (isKS, isKSb, isKSc) and isPIE1 co-localize with the QTL for flowering date and isKS, isKSb, isKSc and isPIE1 are also located within the QTL for chilling requirements (Fig 4, S5 Table). All these CGs and PavNUA are located within the QTL for heat requirements.
On LG6, PavSVP, isPRR7 and isFT co-localized with the QTL for flowering date and isPRR7 and isFT also located in the QTL region for chilling requirements.
On LG7, PavGH17, PavMBF1C and isVIL1 located in the QTL region for chilling requirements and flowering date and PavCLF located only in the QTL region for flowering date.
On LG8, PavGI, PavSWN, PavSAR3, isVRN1 and isFIO1, co-localized with the QTL for flowering date and the three last CG also for the heat requirements QTL.

Allelic effect of candidate genes on the phenological traits
The allelic effects of the CGs located in the QTLs with the main effect for chilling and heat requirements, and for flowering date on LG4, were analyzed. Haplotypes were constructed with the three genetically mapped CGs: PavNUA, PavEMF2 and PavARP4 in R×G and with PavEMF2 and PavARP4 in R×L (Table A in S6 Table). The mean values of chilling requirements, heat requirements and flowering date for the two haplotypes were calculated for each year of evaluation. Significant differences between the two haplotypes were found for each year for chilling requirements and flowering date ( Table B in S6 Table). For heat requirements, significant differences between the two haplotypes were only found for two of the three years (2010, 2011 but not 2012) ( Table B in S6 Table).
The allelic effects of PavNUA, PavEMF2 and PavARP4 in R×G are significant for all traits and for all years of evaluation, except for heat requirements in 2012 (Table C in S6 Table). The allelic effects of PavEMF2 and PavARP4 in R×L progeny are also significant for the flowering date for the 6 years of evaluation. For all traits, the 'Regina' haplotype has higher values than the 'Lapins' and 'Garnet' haplotype. For chilling requirements, differences ranging from 4.17 to 5.32 CP were observed according to the year of evaluation. For heat requirements, the highest difference was observed for PavEMF2 in 2011: 490 GDH. For flowering date, differences ranging from 1.44 (2011 for R×L) to 3.13 (2008 for R×L) or 4.75 (2009 for R×G) days were observed.

Discussion
Data on the molecular and genetic control of flowering are scarce in sweet cherry. Consequently, we investigated CGs based on genes involved in dormancy and flowering processes in model plant species such as Prunus persica, Ribes nigrum L., Populus trichocarpa, and Arabidopsis thaliana. Starting from a wide list of CGs, we identified and mapped orthologs in sweet cherry and we were able to co-localize a subset of these genes with QTLs for flowering time and chilling and heat requirements. Candidate genes mapping A set of 79 CGs were selected, already reported to be involved in dormancy and flowering processes in several species. Several of them are part of multigenic families (DAM, GH17, KS), increasing the difficulty of designing specific primers to amplify a single CG. For the DAM genes, we focused on the DAM5 and DAM6 that were shown to be the strongest CGs in peach. Unfortunately no polymorphism was detected within the parental sequences. Concerning the GH17 family, over 100 putative genes were identified in the Populus trichocarpa genome but only a subset of the 10 were analyzed for their effect on dormancy processes [59]. The seven most promising GH17 genes were analyzed in this study, and six primer pairs were designed, the Populus GH17-33 and GH17-65 genes being too close to identify sweet cherry orthologs. Among the six GH17 genes that could be amplified, three were segregating and mapped on different LGs, LG1, LG6 and LG7. Among all the SNPs detected, 19 SNPs were located in exonic regions. No nonsense mutations were detected but nine of the SNPs induced nonsynonymous substitutions (S3 Table). Three of them induced amino acid polarity changes that may affect the function of the proteins (CLF, VIM1, DDL).
The sequences around the SNPs obtained for each candidate genes (S3 Table) are now available in the GDR and will be used to include these SNPs in the new cherry SNP array that will be constructed with the support of the RosBREED 2 project (https://www.rosbreed.org/). Consequently, it will be soon possible to screen cherry trees for their allelic pattern on these genes.
Thirty four CGs, among the 35 mapped CGs, colocalized with peach orthologs confirming the high synteny between sweet cherry and peach. In addition, these results give good feedback regarding the synteny for the in silico mapped CGs, for which no SNPs were detected in the parents of the mapping populations. PavSAR3 was the only gene that was not mapped in the same LG in cherry (LG8) than in peach (LG2), which may suggest that this gene is present in several copies and the polymorphism was detected only in one locus in sweet cherry. Another hypothesis could be the presence of a translocation between peach and sweet cherry in this region. However, if it is the case, the translocation concerns only a fragment within the two flanking markers Rsweet_8_10229141 and Rsweet_8_10482115, both located on LG8 in peach.

CG-QTL co-localizations
Most of the analyzed CGs (56 among 79) co-localized within QTLs for chilling or heat requirements, or flowering date (S5 Table). This result can be expected if we consider the wide regions covered by several QTLs (Fig 2). However, the important results are for QTLs with small confidence intervals such as QTLs for the flowering date in LG1 of 'Lapins' and in LG4 of 'Regina' (Fig 4, S5 Table) covering each 5 and 4 cM, respectively.
At the bottom of LG1, isDAM5, isDAM6 and PavHYL1 are located within the QTL for flowering date found for the R×L mapping progeny, explaining 8% of the variation of this trait. This region is the same as the EVG peach region suggesting a common genetic control between these two Prunus species. In peach, DAM genes have been proposed as regulators of developmental processes, including seasonal elongation cessation, terminal bud formation [60] and especially dormancy [61]. In particular, DAM5 and DAM6 were shown to be strongly regulated by exposure to chilling temperatures over the dormancy period [62].
On LG4, isKS, isKSb, isKSc and isPIE1 are located within the 4 cM covered by the QTL for chilling requirements detected in 'Regina' (S3 Table, Fig 4)'. The QTL for heat requirements covers a larger region (16 cM) that includes isKS, isKSb, isKSc and isPIE1 but also PavNUA, PavEMF2, PavARP4 and isGA2ox. For flowering date, a total of eight CGs co-localized with this major QTL in LG4. PavEMF2, PavARP4 and isGA2ox mapped within the 3.3 cM QTL region of 'Regina' (from the R×L progeny) possibly explaining 21.2% of the phenotypic variation for flowering date, and isKS, isKSb, isKSc and isPIE1 mapped within the 4.1 cM QTL region of 'Regina' (from R×G progeny) possibly explaining 36.3% of the phenotypic variation [25].
These genes are orthologs of genes known to be involved in the temperature pathway (ARP4, EMF2, NUA and PIE1) and gibberellins pathway (KS, KSb, KSc and GA2ox). In Arabidopsis thaliana, ARP4, EMF2 and PIE1, were reported to be involved in epigenetic regulation [17,63,64]. GA2ox is an enzyme related to gibberellin inactivation [65]. Increases in gibberellin levels were reported to be necessary for endodormancy release [66]. Transcriptome analysis of Japanese pear flower buds at endodormancy and endodormancy release stages revealed that transcripts of GA2ox were up-regulated in ecodormancy [67].
No CG of the photoperiod pathway was found in the major QTL of LG4, which might suggest that photoperiod has little or no effect on flowering time in sweet cherry as it was already reported in several genus of the Rosaceae family: Malus, Pyrus and Sorbus [22,68]. However, as CGs involved in the photoperiod pathway were mapped in QTLs on other LGs (ELF3, FRS5 and FYPP3 in LG1, CRP in LG2, FRS6 in LG3, COP1 in LG5) the effect of the photoperiod on flowering time cannot be excluded. Heide [23] showed a pronounced photoperiod/temperature interaction on the control of growth cessation and resumption in several Prunus species including P. avium.
No CG was located in the QTLs for flowering date in 'Garnet' LG2 and LG6 and in the QTL for heat requirements in 'Garnet' LG6 suggesting that additional CGs may be involved in the flowering and dormancy processes. Another hypothesis could be that genes not already reported to be involved in dormancy or flowering processes may play a significant role in sweet cherry. However, interestingly in 'Regina' LG2, SOC1 was located in silico in the QTL for chilling requirements and for flowering date. This gene was reported to be involved in chilling requirements in apricot [69].

CGs and sweet cherry breeding programs
The allelic compositions of PavNUA, PavEMF2 and PavARP4 had a significant effect on the three traits, chilling and heat requirements in the R×G progeny and on flowering date on the two progenies (R×L and R×G). Consequently, we showed that it is possible to select individuals for low/high chilling requirements and early/late flowering time in these progenies by using the SNP markers identified in these CGs. These markers represent a potential tool for breeding programs using marker assisted selection. This selection could be promising for regions where chilling units are not sufficient to fulfill chilling requirements of sweet cherry cultivars. Also, selection for late flowering could help to mitigate spring frost risk in cold winter regions.

Conclusions
In this study we analyzed a large set of CGs already reported to be involved in flowering and dormancy processes in several species. SNP markers were identified and used to map 39 CGs in sweet cherry progenies. Based on the high synteny between peach and sweet cherry, CGs that were not polymorphic in the parents were mapped in silico using the peach genome sequence. We showed the first co-localization between CGs and QTLs for temperature requirements and flowering date in sweet cherry. Most of the selected CGs were located in QTL regions, making them a useful resource for future bud dormancy and flowering network analyses. Those located in QTLs of high effect with a small confidence interval, in the LG4 QTLs for example, are very promising and may be useful to select new varieties adapted to diverse climatic conditions. Moreover, further analyses including expression analyses of CGs located in QTL regions and association genetics could also provide new understanding on processes involved in the control of dormancy and flowering time.
Supporting Information S1 Fig. Linkage maps of 'Regina' and 'Lapins' constructed based on the segregation analysis in the R×L progeny. Linkage maps were constructed using JoinMap1 4.0 [53]. A minimum LOD value of 3 and a recombinant threshold of 0.35 were chosen for the mapping. Kosambi's mapping function was used to convert recombination frequency into map distance [54]. The graphical presentation of linkage maps was performed using the MapChart software version 2.2 [56]. The candidate genes are indicated in red. The genetics distances are indicated in cM.  [53]. A minimum LOD value of 3 and a recombinant threshold of 0.35 were chosen for the mapping. Kosambi's mapping function was used to convert recombination frequency into map distance [54]. The graphical presentation of linkage maps was performed using the MapChart software version 2.2 [56]. The candidate genes are indicated in red. The genetics distances are indicated in cM. (TIF) S1 File. Sweet cherry cDNA sequences of candidate genes referred as PRUAVXXXXX in S1 Table. (TXT) S2 File. Alignment of candidate genes sequences: peach genomic (ppa), sweet cherry  Table. Primers used to amplify sweet cherry candidate genes. (XLSX) S3 Table. Candidate genes mapping in peach and sweet cherry. SNPs used for the genetic mapping in sweet cherry and the genotype for each parent are detailed-The in silico position of each candidate gene in the Prunus persica physical map is referred according to the Peach genome v2.0.a1. The linkage group and the parent where the candidate gene is genetically mapped are reported for Prunus avium.