Mapping and identification of QTLs for low light tolerance traits in grapevine

Abstract

Background

Low light is a major limiting factor in greenhouse grapevine cultivation, particularly during winter and early spring. To investigate the genetic basis of low light tolerance, a population of 198 F₁ individuals derived from a cross between low light-tolerant and low light-sensitive grapevine varieties was used in this study.

Results

SNP markers were converted into bin markers using the self-developed VCF_to_binMap analysis pipeline, and a genetic map was constructed containing 19 linkage groups with a total length of 1161.95 cM and an average marker distance of 0.71 cM. Phenotypic analysis revealed significant segregation for chlorophyll content and light compensation point (LCP) in the F1 population, with traits approximately following a normal distribution. QTL mapping identified three significant QTLs associated with chlorophyll a content, four with chlorophyll b content, and one with total chlorophyll content, all located on linkage group LG17. Additionally, one significant QTL related to LCP was mapped to LG8. Transcriptome sequencing of extreme phenotype plants and correlation analysis identified two MYB transcription factors (Vitvi17g00232 and Vitvi17g00309) and one STN8-encoding gene (Vitvi08g02097) within the QTL intervals, suggesting their potential roles in low light tolerance.

Conclusions

This study provides insights into the genetic mechanisms underlying low light tolerance in grapevines and identifies candidate genes for further functional validation. The findings contribute to the development of grapevine varieties with improved low light tolerance, supporting sustainable greenhouse cultivation practices.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-06968-y.

Introduction

Grapevines are light-loving crops, and their growth and development are closely related to light intensity [45]. In early-season (December to April) cultivation in protected environments in northern China, light availability is limited due to insufficient sunlight during the growing season. The daily sunlight hours are only approximately 65% of those in open-field cultivation [65]. In addition, factors such as shading by the greenhouse structure, film aging, and dust accumulation further reduce solar radiation and ultraviolet radiation intensity, both of which are much lower than those in open-field cultivation. Moreover, inappropriate variety selection exacerbates this problem, leading to poor flower bud differentiation, reduced flower formation, and significant declines in yield and quality [54, 64].

According to years of field surveys, significant differences in the flower bud differentiation rate are observed among different grapevine varieties under the early-season cultivation model in greenhouses, even with the same level of fertilizer and water management and identical environmental conditions. This indicates genetic diversity. Monitoring the light and temperature conditions within a greenhouse revealed that temperature is not the primary environmental factor limiting flower bud differentiation in greenhouse grapevines; rather, deteriorating light conditions are the fundamental cause [65]. Studies have shown that the light compensation point and chlorophyll content are highly correlated with the continuous high-yielding capacity of grapevines [57]. Research on the impact of different rootstocks on the photosynthetic characteristics and chlorophyll fluorescence properties of scion varieties has shown that different rootstocks can influence the low light tolerance of scions to some extent [17, 18]. In recent years, the low light tolerance traits of grapevines have attracted increasing attention. Li et al. [29] evaluated the low light tolerance of 47 grapevine varieties by analyzing various photosynthetic parameters, such as the light compensation point (LCP), maximum net photosynthetic rate (Pm), dark respiration rate (Rd), apparent quantum efficiency (AQY), and chlorophyll fluorescence parameters (Fo, Fm, φPSII, qP, and NPQ). They identified 36 grapevine varieties that are suitable for early-season cultivation in protected environments, as these varieties all presented low LCP values. Lou et al. [30] analyzed the low light tolerance characteristics of eight grapevine varieties based on the basis of photosynthetic parameters such as the LCP, light saturation point, Pm, Rd, and AQY. These authors reported that varieties such as JinShouZhi and RuiDuHongYu presented lower LCP and higher net photosynthetic rates, indicating strong low light tolerance, whereas varieties such as like LanManHongYan and JuMeiGui presented higher LCP and lower net photosynthetic rates, suggesting poor low light tolerance and suitability for greenhouse cultivation.

Genetic map construction is an important tool in biological and genetic research. This approach helps to identify genes or loci associated with specific traits (such as disease resistance, yield, and fruit quality), providing clues for further gene cloning and functional analysis. In recent years, sequencing-based genotyping technologies, such as RAD-Seq [2], SNP microarrays [53], AmpSeq [58], and rhAmpSeq [66], have rapidly advanced, making single nucleotide polymorphisms (SNPs) and their derived markers the mainstream for constructing high-density genetic maps. However, a mapping population typically yields hundreds of thousands to tens of millions of SNP markers through genotyping-by-sequencing (GBS) or restriction-site associated DNA sequencing (RAD-seq) approaches. Preprocessing the data to address issues such as missing data and sequencing errors caused by low-coverage sequencing has become critical. Otherwise, the genetic map would experience severe inflation. Grapevine, as a perennial crop, has a life cycle of at least 3–4 years, and its regeneration under tissue culture systems is challenging. Commonly used population types, such as BC₁, DH, Hap and RIL, are not suitable for genetic map construction in grapevine. Due to the highly heterozygous nature of the grapevine genome [49], its F₁ population exhibits extensive segregation and can be used for genetic map construction using a pseudo-testcross strategy [16].

To uncover the genetic mechanisms underlying low light tolerance in grapevines, this study conducted low-coverage whole-genome sequencing on a population of 198 F₁ individuals derived from a cross between the low light-tolerant variety ZhuoSeXiang and the low light-sensitive variety HeiCuiWuHe. A self-developed VCF_to_binMap analysis pipeline was employed to impute missing data and correct sequencing errors based on the linkage between adjacent loci. Subsequently, genetic map construction and QTL mapping were performed using JoinMap 5 and MapQTL 6 software. To identify differentially expressed genes within the QTL intervals, transcriptome sequencing was conducted on extreme phenotype progeny individuals. Candidate genes were further analyzed for variant site correlations and validated using qPCR.

Materials and methods

Experimental design

This study was conducted from 2022 to 2024 at the grape vineyards of the Institute of Pomology, Chinese Academy of Agricultural Sciences, in Xingcheng, Liaoning Province. The F₁ hybrid population (named 2021 A) consisted of a cross between the female parent ZhuoSeXiang, which reliably sets fruitful buds year after year under protected cultivation (low light conditions), and the male parent HeiCuiWuHe, which exhibits instability in this trait. Genetic screening with SSR markers (Table S1), with allele sizing determined by capillary electrophoresis (CE), enabled stringent selection of 198 F₁ progeny plants exhibiting the expected heterozygous genotype pattern (one allele from each parent). This selection strategy effectively excluded potential self-pollinated offspring and pollen contaminants, ensuring the genetic purity of the experimental population for subsequent analyses. The male, female, and offspring plants (single plant per genotype) were all planted in a greenhouse. The sampling and measurement of the parent plants (both male and female) were conducted three years after they were grafted onto Beta rootstocks, whereas the sampling and measurement of the offspring were conducted three years after the own-rooted seedling-derived progeny were planted. The planting pattern was set with a spacing of 0.5 m between plants and 2 m between rows, using a horizontal cordon training system combined with a V-shaped canopy. The temperature was maintained at 18–28 °C during the day and 11–19 °C at night. Leaf sampling for chlorophyll content measurement was conducted during the flowering stage. Healthy leaves from the 4th to 6th nodes of vigorously growing shoots were selected for sampling, and measurements were performed immediately afterward. Leaves for light compensation point (LCP) measurements were selected following the same standardized criteria as those used for chlorophyll content determination. Both chlorophyll content and light compensation point (LCP) measurements were conducted in 2024. For chlorophyll analysis, nine leaves were collected and equally divided into three biological replicates. For LCP determination, only one leaf per plant was measured due to the prohibitively lengthy measurement process that precluded biological replication.

DNA extraction

Young, tender leaves were ground into a powder in liquid nitrogen. A 0.1 g sample of the powder was used for DNA extraction following Protocol-II of the TaKaRa MiniBEST Plant Genomic DNA Extraction Kit (9768).

Library construction and sequencing

For each sample, 0.2 µg of DNA was used as the starting material for library preparation. The Rapid Plus DNA Library Preparation Kit (Illumina) was used to generate sequencing libraries according to the manufacturer’s instructions, and an index barcode was added to each sample. Briefly, genomic DNA samples were fragmented into 350 bp fragments via sonication. The DNA fragments were then end-repaired, an A-tail was added, and Illumina full-length adapters were ligated. PCR amplification was performed, and the PCR products were purified via the AMPure XP system (Beckman Coulter, Beverly, USA). The DNA concentration was measured via a Qubit^® 3.0 fluorometer (Invitrogen, USA). The library was analyzed for fragment size distribution via the Agilent 2100 Bioanalyzer, and quantified by real-time PCR. Indexed samples were clustered on the cBot Cluster Generation System via the Illumina PE Cluster Kit (Illumina, USA) according to the manufacturer’s instructions. After clustering, the DNA libraries were sequenced on the Illumina platform, generating 150 bp paired-end reads. The amount of sequencing data for the parents was no less than 25 Gb, whereas the amount of sequencing data for the offspring was 3 Gb.

Variant detection

The raw sequencing data were filtered via fastp software [9] with default parameters. The hap1 haplotype genome of ZhuoSeXiang grape (NCBI accession number PRJNA1185455) was used as the reference genome. Clean reads were aligned to the reference genome via the bwa mem command [25], and a BAM file was generated. The BAM file was then sorted by chromosome via the samtools sort command [26]. Next, PCR duplicates were marked via the gatk MarkDuplicates command [34]. The gatk HaplotypeCaller command was used to generate a GVCF file, which was then merged for the offspring and parents via the gatk CombineGVCFs command. The GATK GenotypeGVCFs command was applied to generate the VCF file, and GATK SelectVariants was used to extract SNP variant sites.

Heterozygosity analysis of the parents

The filtered second-generation whole-genome resequencing data of the male and female parents were used as input files. Jellyfish software [31] with the parameter -m 21 was used to count k-mers, and GenomeScope software [52] was employed to calculate heterozygosity.

Chlorophyll content measurement

The extraction solvent was absolute ethanol. A 1 g sample of leaf tissue was accurately weighed and added to 15 mL of extraction solution. The sample was incubated at 60 °C for 6 h and diluted 10 times for colorimetric analysis. Absolute ethanol was used as the blank. The absorbances at 645 nm and 663 nm were measured, and the chlorophyll content was calculated via the following formula. Three replicates were performed for each individual plant.

where V is the total volume of the sample extract, D is the dilution factor, and M is the sample mass (g).

Light compensation point measurement

The light response curve was measured via a Li6400 photosynthesis system (LI-COR Biosciences, Lincoln, NE, USA). The instrument was preheated, and measurements were started once the zero points for carbon dioxide and water were accurately calibrated. The carbon dioxide concentration was set at approximately 420 ppm, and the relative humidity (RH) was set between 40% and 50%. The ambient temperature was controlled at approximately 25 ℃. The automatic measurement program for the light response curve was then set, with the artificial light intensity adjusted to 1500, 1200, 1000, 800, 400, 200, 100, 50, 20, and 0 µmol m⁻² s⁻¹. Measurements were conducted between 8:00 AM and 12:00 PM on clear days. The calculation of the light response curve parameters and the light compensation point followed the methods of Ye et al. [61] (http://photosynthetic.sinaapp.com/calc.html).

Genetic map construction

The genotyping was performed by processing the VCF files via the self-developed VCF_to_binMap analysis pipeline (https://github.com/CAAS000JXH/Genetic-Linkage-Map-Construction-and-QTL-Mapping-for-Grape-Hybrid-Populations). The pipeline primarily consists of the following steps:

1. SNP Filtering

   • Parental SNPs: Only SNPs with a coverage depth of 6× were retained for the parents.
   • Progeny SNPs: Only SNPs with a coverage depth of 3× were retained for the progeny.
   • Intersection of SNPs: Only SNPs shared between the parents and progeny were retained.
   • Missing Data Filtering: SNPs with missing data in more than 25% of the progeny individuals were filtered out. This step applies only to the progeny, as the parents were sequenced at high coverage and do not have missing SNP data.
2. Variant Classification

   • Maternal Genetic Map Construction: Variants where the male parent was homozygous (e.g., 0/0 or 1/1) and the female parent was heterozygous (e.g., 0/1) were extracted for maternal genetic map construction (nn×np).
   • Paternal Genetic Map Construction: Variants where the male parent was heterozygous (e.g., 0/1) and the female parent was homozygous (e.g., 0/0 or 1/1) were extracted for paternal genetic map construction (lm×ll).
   • Pseudo-Testcross Principle: This approach transforms the genetic map construction of the F1 population into the construction of two BC₁ populations. Each locus is simplified to the segregation type AB×AA, with an expected segregation ratio of 1:1. One parent is entirely heterozygous, while the other is entirely homozygous.
3. MNP Marker Construction

   • SNP Grouping: SNPs were sorted by physical position, and groups were formed at 5,000 bp intervals along each chromosome, starting from 1 bp. Only the first 5 SNPs in each group were retained.
   • Haplotype Formation: For the grape genome, 5,000 bp is a small distance, ensuring tight genetic linkage. These 5 SNPs were concatenated with commas to form a multi-nucleotide polymorphism (MNP), e.g., 0/1,0/1,0/0,0/1,0/1.
4. MNP Genotyping

   • Genotype Counting: The types and counts of progeny genotypes for each MNP marker were statistically analyzed. In theory, only two genotypes should exist in a 1:1 ratio. However, due to missing data or sequencing errors, additional genotypes may appear, albeit in small numbers.
   • Dictionary Construction: The top two genotype types were assigned as the theoretical values for the locus, labeled as nn and np. A dictionary variable was created, e.g., {'0/0,0/0,0/1,0/0,0/0':'nn','0/1,0/1,1/1,0/1,0/1':'np'}.
   • Fuzzy Matching: Progeny numerical genotypes were converted to character genotypes using a fuzzy matching function. If at least 3 out of 5 SNP genotypes matched completely, the assignment was made. For example,'0/0,0/0,0/1,0/0,./.' or'0/0,0/0,0/1,./.,0/1' could both be assigned as nn. Assuming a 1% error rate for each SNP and allowing for a maximum 25% random missing data rate at each site, the probability of 5 independent SNPs being correct and nonmissing is 31.85%. If we treat these 5 SNPs as one MNP marker, with at least 3 correct and nonmissing SNPs, the accuracy of the MNP marker is 88.84%.
   • MNP Filtering: MNP markers with a missing rate exceeding 25% of the progeny or failing the chi-squared test (p >0.05) were filtered out.
5. Correction of MNP Linkage

   • Recombination Principle: Due to random assignment of character genotypes, adjacent MNP markers may lose linkage. Using the principle that recombinant gametes are fewer than parental gametes during meiosis, adjacent MNP markers were corrected.
   • Correction Process: MNP markers on each chromosome were sorted by physical distance. Starting from the second MNP marker, progeny genotypes were swapped (e.g., nn replaced with np and vice versa). The number of recombinants between the first and second MNP markers was counted before and after swapping, and the configuration with the fewest recombinants was selected. This process was repeated sequentially for all adjacent MNP pairs until the last marker.
6. Bin Marker Construction

   • Bin Formation: Similar to MNP construction, bins were formed at 50 kb intervals along each chromosome, starting from 1 bp.
   • Genotype Assignment: Within each bin, the counts of the two genotype types (e.g., nn and np) were tallied. If the count of nn exceeded two-thirds of the total, the bin was assigned as nn; otherwise, it was assigned as --.
7. Bin Genotype Correction

   • Error Correction: Missing or erroneous bin genotypes were corrected using a combination of scripts and manual inspection. For example:

     • nn,nn,nn,nn,np,nn,nn,nn could be corrected to nn,nn,nn,nn,nn,nn,nn,nn (assuming np is an error).
     • nn,nn,--,nn,nn,nn could be corrected to nn,nn,nn,nn,nn,nn (filling in missing data).
   • Since the low light tolerance trait originates from the female parent and exhibits segregation in the progeny, only SNPs where the female parent was heterozygous (e.g., 0/1) and the male parent was homozygous (e.g., 0/0 or 1/1) were retained. After manual correction, the data for 19 chromosomes were imported into JoinMap 5 software [47] for maternal genetic map construction.The parameters were set as follows:

     • Grouping: independence LOD;
     • Independence LOD: 2, 10, 1;
     • Recombination frequency: 1e-3, 1e-4, -5e-5;
     • Recombination frequency: 0.25, 0.05, -0.05;
     • Linkage LOD: 2, 10, 1;
     • Mapping algorithm: maximum likelihood (ML) mapping.
   • Other parameters were set to the software's default values. The genetic map was visualized via MapChart software [51]

QTL analysis

QTL analysis was performed via MapQTL 6 software [46]. The population type was set as BC₁, and the nn and np genotypes were converted to a and h. The analysis method used was interval mapping [22], with other parameters set to the software’s default values. The LOD threshold was calculated via the permutation test [10].

Transcriptome analysis

To construct two extreme phenotype pools, RNA was extracted from the functional leaves of three single individuals with low chlorophyll contents and low light compensation points and three single individuals with high chlorophyll contents and high light compensation points using the RNAprep Pure Polysaccharide Polyphenol Plant Total RNA Extraction Kit (DP441; TianGen, China). Transcriptome libraries were constructed, with three independent libraries per pool as biological replicates. Sequencing was performed on the Illumina HiSeq 2000 platform via the PE150 sequencing mode, with sequencing data for each library not less than 10 Gb. Fastp software was used for quality control of the raw sequencing data. The reference genome was the self-assembled hap1 genome of ZhuoSeXiang, and homology annotation was conducted via Liftoff software based on the PN40024 V4 annotation. Clean reads were mapped to the reference genome via Hisat software. Gene expression quantification was performed via FeatureCount software, differential gene expression analysis was carried out via DESeq2 software, and GO analysis was conducted using the online tool at https://www.omicshare.com/tools/.

Correlation analysis

Correlation analysis was conducted by extracting variant information from candidate gene regions using bcftools [12], based on gene location data from the GFF3 annotation file of the hap1 genome of ZhuoSeXiang and the corresponding VCF file. Sites with identical genotypes in parents were excluded. The remaining variant sites were converted into numeric values, and the correlation between the genotype and phenotype values for each site across all samples was calculated via the Pearson correlation coefficient method.

Promoter sequence analysis

Promoter sequence analysis was performed by extracting the 1700–1800 bp upstream region of the start codon of the candidate gene based on the hap1 genome of ZhuoSeXiang. The sequence was submitted to the PlantPAN online analysis tool (http://plantpan.itps.ncku.edu.tw/plantpan4/index.html) for visualization of potential MYB transcription factor binding sites.

Quantitative real-time PCR validation

Quantitative real-time PCR validation was performed using RNA samples that were previously employed for transcriptome library construction, with the TianGen^® FastKing RT Kit (KR116) for reverse transcription and the TianGen^® SuperReal PreMix Kit (FP205-02) for qt-PCR. The instrument used was the Bio-Rad CFX96 real-time PCR system. The primer information is shown in Table S1. Relative expression levels were calculated via the 2^ΔΔCt method. Three biological replicates were performed for each sample.

Results

Genetic map construction

The resequencing of the female parent ZhuoSeXiang yielded a total of 54 Gb of clean reads, with an average Q20 value of 97.8% and an average Q30 value of 93.5%. The resequencing of the male parent, HeiCuiWuHe, generated 29 Gb of clean reads, with a Q20 value of 97.4% and a Q30 value of 91.7%. On the basis of k-mer calculations from the resequencing data, the heterozygosity of the female parent ZhuoSeXiang was 2.73% (Fig.S1), whereas the heterozygosity of the male parent HeiCuiWuHe was 1.36% (Fig.S2). The resequencing of 198 individual offspring generated a total of 911 Gb of clean reads, with an average data volume of 4.6 Gb per offspring. The average Q20 value was 99.4%, and the average Q30 value was 97.8%. The clean reads from the offspring were aligned to the ZhuoSeXiang haplotype genome, resulting in 25,996,820 SNP variant sites, whereas the clean reads from the parents were aligned to generate 14,324,575 SNP variant sites.

The genetic map of ZhuoSeXiang was constructed via the self-developed VCF_to_binMap analysis pipeline (Fig. 1), as the trait for low light tolerance was inherited from the female parent ZhuoSeXiang. Initially, SNP variant sites were filtered to retain only those where the female parent was heterozygous (e.g., 0/1) and the male parent was homozygous (e.g., 0/0 or 1/1). This resulted in 5,095,551 SNP sites, with an average of 268,186 SNPs per chromosome. A 5,000 bp window was subsequently used to extract 5 SNPs to construct MNP molecular markers, resulting in 57,584 MNP sites, with an average of 3,030 MNPs per chromosome. Next, a 50 kb window was used to divide the data into bins, yielding 7,972 bin sites, with an average of 419 bins per chromosome. These data were then imported into JoinMap 5 software for genetic map construction, resulting in 19 linkage groups. A total of 1,646 nonredundant bins were identified on the genetic map, with an average of 86 bins per chromosome. The total map distance was 1,161.95 cM, with an average of 61.2 cM per chromosome, and the average map distance of the bin markers was 0.71 cM (Table 1; Fig. 2).

Fig. 1.

Schematic of the VCF_to_binMap analysis pipeline. This pipeline is suitable for constructing genetic maps in F₁ hybrid populations. It involves mining and filtering SNP sites from whole-genome sequencing data of offspring and their parents. SNP sites that conform to the lmxll or nnxnp segregation types are selected to construct MNP molecular markers. For each chromosome, bins are defined according to predetermined lengths, and the segregation types of MNP molecular markers in each bin are assigned accordingly. A genetic map is then constructed on the basis of the segregation type data of each bin. The principle of this method is based on the ‘double pseudotestcross’ theory for CP populations and the linkage of SNP molecular markers. By converting SNP markers to MNP markers, errors and gaps in the SNP data can be corrected

Table 1.

Statistical information on the construction of the maternal genetic map

Chromosome	SNP (n)	MNP (n)	bin (n)	Nonredundant Bin (n)	Map Distance (cM)	Average Marker Distance (cM)
chr1	305,208	3463	440	102	67.72	0.66
chr2	209,922	2718	353	83	62.72	0.76
chr3	181,819	2430	354	81	55.00	0.68
chr4	265,172	2342	371	77	74.31	0.97
chr5	317,989	3375	449	79	52.02	0.66
chr6	147,377	2641	378	92	68.89	0.75
chr7	375,632	4279	561	128	90.71	0.71
chr8	271,515	3626	446	82	57.61	0.70
chr9	252,387	2659	366	75	54.04	0.72
chr10	292,036	3078	450	82	57.65	0.70
chr11	216,490	2729	357	84	60.14	0.72
chr12	248,452	2908	418	74	45.81	0.62
chr13	292,949	3579	485	99	61.07	0.62
chr14	348,835	3911	531	106	74.89	0.71
chr15	151,550	1871	262	54	46.55	0.86
chr16	229,836	2493	370	80	53.48	0.67
chr17	237,556	1917	293	73	45.28	0.62
chr18	456,005	4570	651	121	77.88	0.64
chr19	294,821	2995	437	74	56.18	0.76
Mean	268,186	3030	419	86	61.2	0.71
Total	5,095,551	57,584	7972	1646	1161.95	-

Fig. 2.

Genetic Map of the Maternal Parent Zhuosexiang

Chlorophyll content and light compensation point measurement

The range of chlorophyll a in the grape hybrid population was 0.4–2.34 mg·g⁻¹, with an average value of 1.56 mg·g⁻¹ and a coefficient of variation (CV) of 18.70% (Table 2). The variation range of chlorophyll b was 0.07–0.63 mg·g⁻¹, with an average value of 0.40 mg·g⁻¹ and a CV of 22.11%. The total chlorophyll content ranged from 0.47 to 2.95 mg·g⁻¹ with an average value of 1.96 mg·g⁻¹ and a CV of 19.24%. The variation range of the light compensation point was 8.18–81.79 µmol m⁻² s⁻¹, with an average value of 31.84 µmol m⁻² s⁻¹ and a CV of 45.51%. The frequency histograms revealed that the distributions of the chlorophyll content and light compensation points were both approximately normal, indicating that they are quantitative traits. (Fig. 3).

Table 2.

Statistical analysis of chlorophyll content and light compensation point measurements in the F₁ mapping population derived from ZhuoSeXiang × HeiCuiWuHe under greenhouse conditions

Trait	Sample Size	Minimum Value	Maximum Value	Mean Value	Standard Deviation	Coefficient of Variation (%)
Chlorophyll a (mg·g−1)	178	0.40	2.34	1.56	0.29	18.70
Chlorophyll b (mg·g−1)	178	0.07	0.63	0.40	0.09	22.11
Total Chlorophyll (mg·g−1)	178	0.47	2.95	1.96	0.38	19.24
Light Compensation Point (µmol m⁻² s⁻¹)	183	8.18	81.79	31.84	14.49	45.51

Fig. 3.

Frequency histogram of chlorophyll content and light compensation point. The red solid line represents the maternal content value, and the black dashed line represents the paternal content value

QTL mapping of chlorophyll content and light compensation points

QTL mapping revealed distinct genomic patterns for different photosynthetic traits. For chlorophyll a, while signals were detected across 11 linkage groups (LG3, LG7-12, LG14-15, LG17-18; Fig. S6), only the LG17 signal surpassed the significance threshold (Fig. 4a). We identified a putative QTL region on LG17 (combined PVE = 19.3%; Table 3), where initial multi-peak detection (3.0%, 11.6%, 4.7% PVE) likely reflected methodological artifacts in LOD profiling rather than distinct loci, leading us to conservatively interpret this as a single QTL (Table 3). Similarly, chlorophyll b showed signals across nine LGs (LG1, LG3, LG8, LG10, LG12, LG14-17; Fig. S7), with only LG17 exceeding threshold (Fig. 4b). The LG17 QTL (combined PVE = 21.8%) exhibited multiple sub-peaks (0.8–10.4% PVE) that we attribute to technical fragmentation. Total chlorophyll patterns mirrored these findings, with detectable signals across nine LGs (LG3, LG8-12, LG14-15, LG17; Fig. S8) but only the LG17 QTL reaching significance (PVE = 27.2%; Fig. 4c). For the light compensation point, among five detected LGs (LG8-10, LG15, LG19; Fig. S9), only LG8 showed a significant QTL (PVE = 7.3%; Fig. 4d).

Fig. 4.

QTLs for Chlorophyll a (a), Chlorophyll b (b), Total Chlorophyll Content (c), and Light Compensation Points (d) on linkage groups with signals exceeding the LOD threshold. The LOD threshold was calculated via the permutation test

Table 3.

QTLs for chlorophyll content and light compensation points

Trait	Linkage Group	Region Type	Position (cM)	Position (Mb)	PVE (%)
Chlorophyll a	LG17	Main QTL	0-26.448	0-5.85	19.3
Chlorophyll a	LG17	Sub-peak 1	0-4.077	0-0.9	3.0
Chlorophyll a	LG17	Sub-peak 2	5.092–16.798	1.2-3.0	11.6
Chlorophyll a	LG17	Sub-peak 3	20.865–26.448	3.75–5.85	4.7
Chlorophyll b	LG17	Main QTL	5.092–33.068	1.2–8.15	21.8
Chlorophyll b	LG17	Sub-peak 1	5.092–8.148	1.2–1.55	3.0
Chlorophyll b	LG17	Sub-peak 2	9.656–16.798	1.85-3.0	7.6
Chlorophyll b	LG17	Sub-peak 3	18.321–18.829	3.25–3.5	0.8
Chlorophyll b	LG17	Sub-peak 4	20.865–33.068	3.75–8.15	10.4
Total Chlorophyll	LG17	Single QTL	0-32.561	0-7.8	27.2
LCP	LG8	Single QTL	45.404–52.019	6.1-13.35	7.3

PVE: Phenotypic variance explained, calculated as the percentage of total phenotypic variation attributable to the QTL. Main QTL: The primary genetic region with the highest phenotypic variance explained (PVE), used when sub-peaks are present within a broader QTL interval. Sub-peak: Secondary LOD score peaks within a Main QTL region, potentially representing technical artifacts or linked loci (PVE < Main QTL). Single QTL: A genetically independent QTL with no detectable sub-peaks (continuous significant LOD interval)

Candidate gene identification for low light tolerance

Genes with an absolute log2 fold change ≥ 1 and an adjusted p-value ≤ 0.05 were considered significantly differentially expressed. Transcriptome analysis of pooled samples from low chlorophyll content/low light compensation point (LL) and high chlorophyll content/high light compensation point (HH) groups identified 2,377 differentially expressed genes, including 953 upregulated and 1,424 downregulated genes (Table S2). GO enrichment analysis revealed that most of the DEGs were related to photosynthesis, with significant enrichment in biological functions such as membrane, photosystem, protein phosphorylation, oxidoreductase activity, and chloroplast thylakoid (Fig. 5). Moreover, compared with those of the high-chlorophyll/high-light compensation point plants, the low-chlorophyll/low-light compensation point plants were predominantly downregulated.

Fig. 5.

GO enrichment analysis of differentially expressed genes. The vertical axis represents -log10 (Q/P-value). A larger value on the vertical axis indicates a smaller Q/P-value, meaning the corresponding bubble is more significant. The horizontal axis represents either up-down normalization or z-score value (the proportion of the difference between the number of upregulated and downregulated genes relative to the total number of differentially expressed genes). A value further to the right indicates that the difference between upregulated and downregulated genes enriched in the pathway or GO Term is larger, with more upregulated genes. Conversely, a value further to the left indicates that the difference between downregulated and upregulated genes enriched in the pathway or GO Term is larger, with more downregulated genes

On the basis of the GO analysis results and genome annotation information, we analyzed the DEGs within the QTL intervals. In the chlorophyll content QTL interval, a total of 78 DEGs were identified (Table S3), with no structural genes found related to chlorophyll metabolism. However, two MYB transcription factors, Vitvi17g00232 and Vitvi17g00309, were discovered. Among all the DEGs (Table S2), three key enzyme-encoding genes in the chlorophyll metabolism pathway—Vitvi12g00346, Vitvi05g00350, and Vitvi07g02114—were differentially expressed (Table 4). The promoters of these three key enzyme-encoding genes in the chlorophyll metabolism pathway contain numerous MYB cis-regulatory elements (Fig.S3; Fig.S4; Fig.S5). Therefore, the two differentially expressed MYB transcription factors within the QTL interval can be considered as candidate genes for further functional validation. In the light compensation point QTL region, 23 differentially expressed genes were identified (Table S3). On the basis of functional annotation, we found that Vitvi08g02097 encodes a serine/threonine protein kinase, which was significantly downregulated in low light compensation point plants (Table 4). This gene belongs to the STN protein kinases family, which regulates the protein of the PSII reaction center and controls the electron transfer process in photosynthesis. STN protein kinases can phosphorylate the D1 protein of the PSII reaction center, thereby affecting the stability and function of PSII [4]. Phosphorylation not only helps stabilize the PSII reaction center but also plays a protective role in the repair process. Further qPCR validation confirmed the accuracy of the transcriptome results (Fig. 6).

Table 4.

Candidate genes for low light tolerance in grapes

Gene ID	Gene Name	log2fc	Gene Annotation
Vitvi17g00232	MYB6	−5.62	Transcription repressor MYB6
Vitvi17g00309	MYB108	2.23	Transcription factor MYB108
Vitvi12g00346	POR	−2.13	Protochlorophyllide reductase
Vitvi05g00350	CVR	−2.68	Divinyl chlorophyllide a 8-vinyl-reductase
Vitvi07g02114	CHL1	−7.39	Chlorophyllase-1
Vitvi08g02097	STN8	−1.26	Serine/threonine-protein kinase STN8

Fig. 6.

Quantitative real-time PCR validation of candidate genes for low light tolerance in grapes. The x-axis labels “LL” and “HH” represent experimental groups with specific phenotypic traits: “LL” denotes the group exhibiting both low chlorophyll content and low light compensation point, while “HH” denotes the group exhibiting both high chlorophyll content and high light compensation point. Three biological replicates were performed. *, ** denote significant differences at P < 0.05 and P < 0.01, respectively, according to the Student’s t-test

We performed correlation analysis of the variant sites in the three candidate genes within the QTL regions with the phenotype values (Table S4). The variant site of Vitvi17g00232 presented the highest correlation coefficient of 0.32 with total chlorophyll content, while the variant site of Vitvi17g00309 had the highest correlation coefficient of 0.20 with total chlorophyll content. The variant site of Vitvi08g02097 showed a strong negative correlation with the light compensation point, with the highest absolute correlation coefficient of 0.28. Variations in these candidate genes may lead to alterations in gene regulation or translation, which could indirectly explain the differences in expression and phenotypes observed between the parents. This aspect warrants further study.

Discussion

Effects of low light intensity on crop yield

Numerous factors influence flower bud differentiation, but light conditions constitute one of the most significant and direct environmental regulators. In nectarine trees, the number of flower buds in the lower canopy zone is significantly lower than in the upper zone, correlating positively with the progressive decline in light intensity toward the canopy base, though fruit set remains unaffected [38]. Optical fiber insertion into the buds of Scots pine (Pinus sylvestris L.) and Norway spruce (Picea abies (L.) Karst.) successfully induced female strobili formation by delivering natural light to apical meristems [21]. Increased plant density has been shown to reduce individual reproductive potential in cucurbits. Greenhouse experiments with 30%, 60%, and 80% reduced incident light demonstrated decreased male and female flower production in zucchini (Cucurbita pepo), with complete suppression of female flowers under 80% shading [37].

Grapevine flower bud differentiation, spanning two growing seasons through three developmental phases (anlage formation, inflorescence primordia initiation, and floral organogenesis), is strongly influenced by light irradiance. Controlled-environment studies confirm that inflorescence primordia quantity increases with light intensity [5, 6], consistent with field observations in Vitis vinifera cultivars including Cabernet Sauvignon, Chardonnay, Flame Seedless, and Thompson Seedless [40, 43]. In contrast, low light or complete shading during primordia initiation and differentiation reduces both the number and size of inflorescence primordia [20, 32].

The mechanistic link between light and bud fertility likely involves photosynthetic activity and carbohydrate allocation. Light limitation diminishes photoassimilate production, restricting carbon supply to developing buds [15, 23, 48]. Based on this physiological framework, we selected chlorophyll content and light compensation point as key indicators to investigate the genetic mechanisms underlying low light tolerance in grapevines.

Challenges and solutions for using SNP molecular markers in genetic map construction

This study provides an efficient method for genotyping low-coverage whole-genome sequencing data of progeny segregation populations, based on high-quality parental genome physical maps. This method can convert millions or even tens of millions of SNPs into thousands of bin markers, significantly reducing the computational burden for downstream genetic map construction. Although SNPs have been widely used in genetic map construction, such as RAD-Seq [2], SNP microarrays [53], AmpSeq [58], and rhAmpSeq [66], these genotyping technologies have limitations. For example, reduced-representation sequencing technologies like RAD-Seq are constrained by restriction enzymes, where factors such as DNA purity, chromosomal structure, digestion conditions, and methylation modifications can affect enzyme efficiency. In contrast, our whole-genome sequencing approach provides comprehensive genome coverage without enzyme-related biases, enables uniform detection across all genomic regions, and offers higher resolution for variant detection while maintaining consistent data quality independent of restriction site distribution or methylation status [13]. The cross-species applicability of SNP microarrays is concerning. Vezzulli et al. [50] demonstrated that SNPs from Vitis vinifera are almost unusable in non-vinifera species. AmpSeq and rhAmpSeq are recently developed amplicon-based genotyping technologies, but they require primer screening and design, as well as multiplex PCR, which increases technical complexity and introduces uncertainty due to variable amplification efficiency, even with rhPCR technology [14]. Whole-genome sequencing provides comprehensive genomic information and does not require consideration of marker transferability, as markers are derived de novo from the sample. However, it is important to note that SNP discovery in whole-genome sequencing relies on a reference genome. Thanks to advances in third-generation sequencing and genome assembly technologies, many species have achieved telomere-to-telomere (T2T) genome assemblies [24, 41, 63], and the cost of next-generation sequencing has significantly decreased. This makes the technical approach adopted in this study widely applicable in the plant breeding community.

To address challenges such as massive data volume, missing data, and errors in genotyping-by-sequencing (GBS), several softwares have been developed, including MSTmap [56], Lep-MAP3 [39], and HetMappS [19]. These tools not only perform genotyping but also include functions for genetic map construction and QTL mapping. The method proposed in this study focuses solely on genotyping, while genetic map construction and QTL mapping are performed using specialized commercial software, JoinMap 5 and MapQTL 6. Our method is independently developed but shares similarities with previous approaches, such as SNP classification and clustering. However, we place greater emphasis on logical reasoning based on genetic principles rather than rigorous mathematical calculations. We set the MNP segment size to 5 kb and the bin segment size to 50 kb. Similar parameter settings were used in the development of the apple 8 K SNP array, where the minimum distance between SNPs was set to 2 kb to minimize the detection of redundant haplotypes during genotyping. Clustering four to ten SNP markers typically corresponds to a non-recombinant physical distance of approximately ± 50 kb [8].

Genetic map construction and QTL mapping can only analyze loci that follow Mendelian inheritance. However, non-Mendelian genetic events, such as transposons and structural variations, are widespread in genomes and can affect marker ordering and map distances. We employed a semi-automated approach to correct for these issues. Typically, only one to three recombination events occur per chromosome, and sudden double crossover events are likely due to errors. The final genetic map we obtained has a total length of 1,161.95 cM, consistent with previously reported grape genetic maps ranging from 1,172 to 1,406 cM [1]. The choice of reference genome is crucial. In this study, the maternal genetic map was constructed using the maternal genome as the reference. The syntenic core genome of Vitis is only 39.8 Mb (8.67% of the genome) [66].

QTLs for low light tolerance traits

Low light reduces the photosynthetic efficiency of plants, which in turn decreases their growth, negatively affecting crop production [60]. Chlorophyll is responsible for absorbing and utilizing light energy, and there is a correlation between chlorophyll content and photosynthetic efficiency [28]. Low light leads to a reduction in chlorophyll content, which subsequently reduces photosynthetic efficiency. Different plants have varying abilities to adapt to low light, with plants having a lower light compensation point exhibiting greater tolerance to low light [11].

Currently, the genetic basis and molecular mechanisms of low light tolerance in grapes are not well understood. However, studies in other crop species, such as cucumber, suggest that low light tolerance is a complex quantitative trait controlled by multiple genes. These genes are distributed across many loci and are controlled by multiple minor genes that are similar, independent, and have no dominant-recessive relationships. Low light tolerance is also highly influenced by environmental factors [24].

In this study, we found chlorophyll content and light compensation point exhibited wide segregation in the F₁ population of low light-tolerant and low light-sensitive varieties, with distributions close to normal. The significant QTLs for chlorophyll a, chlorophyll b, and total chlorophyll were all located on chromosome 17, suggesting that chromosome 17 may play a role in the synthesis and degradation of chlorophyll in grape leaves. Only one significant QTL related to the light compensation point was identified, and it was located on chromosome 8. Similarly, in rice, association mapping and genetic mapping have also identified QTLs related to chlorophyll content [35]. In barley, 29 QTLs associated with chlorophyll content were detected through genetic mapping via recombinant inbred lines [36]. In canola (Brassica napus), three chlorophyll-related genes were finely mapped based on near-isogenic lines and the BC3F2 population [60].

Protochlorophyllide reductase (POR) is a key enzyme in the chlorophyll biosynthesis pathway, and is responsible for the reduction of protochlorophyllide (Pchlide) to chlorophyllide (Chlide), a core step in chlorophyll synthesis. Divinyl chlorophyllide a 8-vinyl-reductase (VCR) is an enzyme that plays a crucial role in chlorophyll metabolism, particularly in the synthesis of chlorophyll a and chlorophyll b, by converting the 8-vinyl group into an ethyl group. Chlorophyllase-1 (CHL1) is an enzyme involved in chlorophyll degradation, and is responsible for removing the phytol chain from the chlorophyll molecule and breaking down chlorophyll into smaller products [44].

Transcriptome analysis of extreme phenotypes revealed that the genes encoding these three enzymes were all downregulated in plants with low chlorophyll contents and low light compensation points, which was consistent with the phenotypic measurements. Bioinformatics analysis revealed numerous MYB binding elements in the promoters of these three genes. Yang et al. [59] reported that MYB108 was involved in reducing the leaf chlorophyll content under magnesium deficiency in bananas. In this study, we identified two differentially expressed MYB transcription factors (MYB6: Vitvi17g00232 and MYB108: Vitvi17g00309) within the QTL intervals, and qPCR validation confirmed these findings.

In this study, we identified an STN8-encoding gene (Vitvi08g02097), which encodes STN protein kinases, within the QTL interval on linkage group 8. STN protein kinases regulates protein phosphorylation in the PSII reaction center, controlling electron transfer processes in photosynthesis [4]. It can phosphorylate the D1 protein of the PSII reaction center, thereby affecting the stability and function of PSII. Phosphorylation not only helps stabilize the PSII reaction center but also plays a protective role during repair. Betterle et al. [3] demonstrated that OsSTN8 is responsible for high-light-induced reversible phosphorylation of the PSII inner antenna subunit CP29 in rice and that the reversible phosphorylation of PSII core proteins such as CP43, D1, D2, and LHCII plays a significant role in leaf adaptation to light stress. We speculate that the STN8-encoding gene (Vitvi08g02097) may be involved in the regulation of photosynthesis in grape leaves under low light stress. This hypothesis requires further biological experimentation for validation.

Conclusion

In this study, we developed an analysis software, VCF_to_binMap, suitable for constructing bin genetic maps for highly heterozygous CP populations. The quantitative trait loci (QTLs) associated with chlorophyll content and light compensation points were identified via a bin genetic map of F1 plants constructed by crossing the low light-tolerant grape variety ‘ZhuoSeXiang’ and the low light-sensitive grape variety ‘HeicuiWuhe’. Transcriptome sequencing of extreme phenotypic individuals and correlation analysis revealed that two MYB transcription factors (Vitvi17g00232 and Vitvi17g00309) and one STN8-encoding gene (Vitvi08g02097) within the QTL intervals may be key genes associated with low light tolerance in grapes.

Authors’ contributions

J.X.H., L.F.Z., and W.H.B.planned and designed the research; J.X.H. and S.M. performed the research and analyzed the data; J.X.H. wrote the paper; J.X.H. and S.M. performed genotyping, processed data, SNP calling; J.X.H, and W.Z.Q. collected data in the field; W.X.D. edited the paper. All authors approved the manuscript.

Data availability

The codes and scripts used for parent heterozygosity analysis, variant detection, genetic map construction, and QTL mapping, as well as the result files, can be accessed at: https://github.com/CAAS000JXH/Genetic-Linkage-Map-Construction-and-QTL-Mapping-for-Grape-Hybrid-Populations. In addition, the WGS and RNA-Seq data have been deposited into the NGDC Genome Sequence Archive (https://ngdc.cncb.ac.cn/) with accession number PRJCA032719.

Declarations

Competing interests

The authors declare no competing interests.