Construction of a High-Density Genetic Map and QTL Mapping Analysis for Yield, Tuber Shape, and Eye Number in Diploid Potato

Abstract

Potato (Solanum tuberosum L.) is a globally important food crop, but its tetrasomic inheritance and diploid self-incompatibility have limited the discovery of potato genes and progress in breeding. Here, we developed an F₂ segregating population consisting of 174 lines by crossing a self-compatible genome-homozygous diploid line (Y8, female parent) with a heterozygous diploid line (IVP101, male parent), followed by selfing. Using whole-genome resequencing, we constructed a high-density genetic map containing 4464 recombinant bin markers with an average physical distance of 165.51 Kb. Phenotypic evaluation of 8 traits related to yield, tuber shape, and tuber eye number across three environments revealed significant parental differences and wide phenotypic variation within the F₂ population. QTL (Quantitative trait loci) mapping using this genetic map and multi-environment phenotypic data identified 89 QTLs, including 7 previously reported QTLs/genes. In addition, 10 QTLs were stably detected across multiple seasons (stable QTLs). Further genetic effect analysis showed that favorable alleles of these stable QTLs significantly enhanced phenotypic values. Notably, two pleiotropic QTLs were identified on chromosomes 5 and 12; the major-effect QTL on chromosome 12 (qTY-12-6, qTS-12-3, and qTE-12-4) exhibited high phenotypic variance explained (PVE). Its favorable allele from Y8 significantly increased mean tuber weight, tuber number per plant, and promoted rounder tuber shape while reducing eye number, simultaneously improving yield and quality. Collectively, this study provides a reference for genetic mapping using homozygous and heterozygous diploid parents, and the identified QTLs offer valuable genetic resources for potato breeding and molecular mechanism research, enhancing our understanding of the genetic regulation of yield, tuber shape, and eye number in potato.

1. Introduction

Potato (Solanum tuberosum L.), as the world’s fourth most important food crop, with a global annual production exceeding 383 million tons across approximately 16.8 million hectares in 2023, plays a critical role in ensuring food security and nutritional supply worldwide [1]. However, the intensification of global climate change in recent years has posed severe threats to global food security [2]. Additionally, most cultivated potatoes are autotetraploids (2n = 4x = 48) with a narrow genetic background. These factors contribute to challenges such as unstable yields and slow improvement of quality traits in potato production. Among these, tuber yield, tuber shape, and eye number are core agronomic traits determining commercial value. For instance, irregular tuber shapes increase processing losses, while excessive eyes directly affect tuber appearance and reduce market value. Therefore, deciphering the genetic basis of these traits and developing molecular marker-assisted breeding technologies are of great significance for advancing potato genetic improvement.

Compared with major cereal crops such as rice and maize, potato research lags significantly, primarily due to its tetraploid genetic characteristics. The tetraploid potato genome is highly heterozygous, with frequent recombination events between the four homologous chromosomes, leading to complex segregation patterns in progeny [3]. Consequently, most genetic maps constructed to date are based on diploid materials, with only rare reports on tetraploid potato genetic maps [4,5,6,7,8,9,10]. In nature, almost 75% of potato wild relatives are diploid [11], originating from diverse habitats ranging from southwestern North America to central Chile and Argentina. These wild species possess high genetic diversity and adaptability to harsh environments [12], providing a robust genetic resource base for potato breeding improvement. However, introgressing a new trait from diploid wild species into tetraploid potato cultivars can take 15–50 years [13]. Diploid hybrid breeding represents a revolutionary breakthrough in potato breeding in recent years [14]. Compared with traditional tetraploid breeding, diploid hybrid breeding offers distinct advantages: shorter fixation time for favorable alleles, feasibility of marker-assisted backcrossing, greater genetic variation for selection, and systematic utilization of heterosis [15].

In summary, breeding and genetic analysis at the diploid level are pivotal directions for future research. Nevertheless, most diploid potatoes exhibit self-incompatibility and inbreeding depression, hindering the development of homozygous inbred lines through multiple generations of selfing, even for F₂ mapping populations derived from F₁ selfing [16]. Recently, the Joint Academy of Potato Science of Yunnan Normal University successively cloned the S-RNase gene [17] and Sli gene [18], overcoming self-incompatibility and successfully creating a set of self-compatible diploid potato advanced inbred lines [19], which provide material foundations for constructing secondary segregation populations. Furthermore, the completion of high-quality genome sequencing of diploid potatoes (e.g., DM8.1 and C88) [20,21] and advancements in resequencing technologies have further facilitated the construction of high-density genetic maps and fine QTL (Quantitative trait loci) mapping.

QTL mapping remains a traditional and effective approach for mining causal genes [22], yet research on genetic map construction and QTL mapping in potatoes is severely delayed. Early genetic analyses predominantly relied on crosses of heterozygous diploid parents, resulting in complex genetic backgrounds that hindered the decomposition of polygenic traits into individual mendelian factors for chromosomal localization. Additionally, markers such as SSR (Simple sequence repeat), AFLP (Amplified fragment length polymorphism), and SCAR (Sequence-characterized amplified region) yielded low-density maps [23,24,25,26,27], leading to overly broad QTL intervals that impeded gene cloning. With the development of high-throughput sequencing, SNP (Single nucleotide polymorphism) markers have been widely adopted for constructing high-density genetic maps in potatoes [13,28,29,30,31,32,33]. However, only a few studies have utilized optimally selected parental materials. For example, Meijer et al. [13] constructed an F₂ population of 272 lines using a homozygous diploid carrying the Sli gene and a heterozygous diploid. Endelman and Jansky [29] developed the first F₂ population from two homozygous diploids: DM (a doubled monoploid potato from anther culture) and M6 (an advanced S₇ inbred line with Sli), which was later advanced to an 87-line RIL population [34]. Most recently, Li et al. [33] generated an F₂ population of 1064 individuals from two homozygous inbred lines (A6-26 and E4-63), performing whole-genome resequencing to construct a high-density genetic map for QTL mapping.

Yield is the most economically significant potato trait, determined by tuber number per plant and tuber size. Hurtado-Lopez et al. [35] identified 4 QTLs for tuber number and 6 for yield using a backcross population. Manrique-Carpintero et al. [30] detected 3 QTLs for total tuber yield, 2 for tuber number per plant, and 6 for average tuber weight in an F₁ population. Hara-Skrzypiec et al. [26] reported QTLs for tuber weight on chromosomes 1, 4, 5, and 6 in an F₁ population. Using a diploid diversity panel of 246 breeding lines with genotyping-by-sequencing and GWAS (Genome-wide association study), Śliwka et al. [36] identified 2 loci associated with tuber weight per plant and 26 with average tuber weight. However, no major yield-related genes have been cloned via QTL mapping to date.

Tuber shape and eye traits also critically influence quality appearance and processing suitability [22]. Extensive QTL mapping has been conducted to identify shape-regulating genes [25,26,28,29,37,38], culminating in the cloning of StOFP20 as a major causal gene in 2023 [39]. Nevertheless, numerous shape-related genes remain uncloned, such as a novel locus on chromosome 6 (~58 Mb) reported by Huang et al. [40]. Prashar et al. [28] mapped a major QTL explaining 20% of phenotypic variation on chromosome 2 using a dense SNP map, while Lindqvist-Kreuze et al. [25] identified secondary QTLs on chromosomes 5 and 12 alongside the major chromosome 10 locus. For eye depth, Fan et al. [22] recently identified a major QTL qEyd10.1 on chromosome 10 explaining 55.0% of phenotypic variation, refined to a 309.10 kb interval via recombination analysis. However, genetic studies on eye number remain limited, with no systematic reports of stable QTLs. Thus, mining stably expressed major QTLs through multi-environment trials combined with high-density genetic maps is crucial for deciphering these traits’ genetic mechanisms.

This study utilized an F₂ segregating population of 174 lines derived from a cross between two diploid potato materials, Y8 and IVP101. Y8 is a homozygous line carrying the Sli gene, developed through hybridization of CIP151 (StGp Goniocalyx) and CIP65 (StGp Phureja) followed by 6 generations of selfing. IVP101 (StGp Phureja), a widely used inducer for ploidy reduction in potato wide crosses [41], has a heterozygous genome. The two parents exhibit significant differences in multiple traits. We constructed a high-density genetic map using bin markers developed via whole-genome resequencing and phenotyped tuber yield, shape, and eye number across three environments to perform multi-environment joint QTL analysis for stable genetic loci. This study aims to: (1) explore resequencing-based high-density genetic map construction in an F₂ population derived from homozygous × heterozygous parents; (2) clarify the genetic basis of yield, tuber shape, and eye number in diploid potatoes; (3) enhance QTL mapping precision by integrating high-density genetic maps with multi-environment phenotypic data, laying foundations for dissecting complex trait genetic networks and accelerating potato breeding.

2. Materials and Methods

2.1. Plant Material

In this study, two diploid potato lines, IVP101 (male parent) and Y8 (female parent), were crossed to produce F₁ progeny, which were further selfed to generate an F₂ segregating population consisting of 174 lines. IVP101 (StGp Phureja) is a widely used inducer for ploidy reduction in potato wide crosses, characterized by a heterozygous genome, short growth period, concentrated flowering time, long tubers with purple skin and white flesh, numerous deep eyes with non-dormant tubers, purple self-incompatible flowers, seeds with embryo spots, and stems with abundant purple pigmentation. Y8 is an advanced generation inbred line developed by the Joint Academy of Potato Science of Yunnan Normal University through hybridization of CIP65 and CIP151 followed by 6 generations of selfing. It exhibits high genomic homozygosity, round tubers with yellow skin and yellow flesh, few shallow eyes, white self-compatible flowers, seeds without embryo spots, and green stems without purple pigmentation. F₁ seeds were obtained from the Y8 × IVP101 cross. Seeds were germinated with gibberellin and sown in a greenhouse, yielding 49 F₁ plants. We referenced the population construction method described by Meijer et al. [13], two vigorous and easily selfed F₁ plants (designated IYY30 and IYP31) were selected. Selfed seeds from these two F₁ plants were randomly mixed and sown, ultimately generating an F₂ genetic segregating population with 174 lines.

2.2. Ploidy Analysis and Parentage Identification of F₁ Plants

Healthy, disease-free leaves of moderate size (0.3–0.5 g) were collected. The leaves were placed into a 1.5 mL centrifuge tube, followed by the addition of 1 mL pre-chilled lysis buffer and three 3 mm steel beads. The mixture was ground at 50 Hz for 30 s using a tissue grinder and then incubated on ice for 30 min. The lysate was filtered through a 400-mesh filter into a new 1.5 mL centrifuge tube. The tube was centrifuged at 1500 rpm for 5 min at 4 °C in a pre-chilled centrifuge. After discarding the supernatant, the pellet (intact nuclear fraction) was collected. Next, 1 mL of pre-chilled lysis buffer containing 1% NP-40 was added, and the mixture was vigorously vortexed. Subsequently, 20 μL of pre-chilled propidium iodide (PI) dye was added, and the mixture was incubated at 4 °C in the dark for 15 min. After filtration through a 400-mesh nylon cell strainer, the filtrate was transferred to a light-protected centrifuge tube, and ploidy was determined using a flow cytometer (Agilent Technologies, NovoCyte 2000R, Santa Clara, CA, USA).

Sample DNA extraction and quality assessment were performed following the method described by Yang et al. [42]. A total of 231 SSR molecular markers were designed based on reference genome sequences, and primers were synthesized by Sangon Biotech (Sangon Biotech, Shanghai, China). These 231 primer pairs were used to amplify DNA samples, with reactions conducted on a Veriti 384 PCR instrument (Applied Biosystems, Foster City, CA, USA). The PCR amplification program was as follows: initial denaturation at 95 °C for 5 min; followed by 10 cycles of denaturation at 95 °C for 30 s, touchdown annealing from 62 to 52 °C (30 s), and extension at 72 °C for 30 s; then 25 cycles of denaturation at 95 °C for 30 s, annealing at 52 °C for 30 s, and extension at 72 °C for 30 s; final extension at 72 °C for 20 min; and storage at 4 °C. After PCR, amplification products were analyzed by fluorescent capillary electrophoresis (ABI 3730xL DNA Analyzer, Applied Biosystems, Waltham, MA, USA). Results were processed using GeneMarker software (Vision 2.2.0, Developed by Soft Genetics LLC, State College, PA, USA) to determine the number of alleles, electropherograms, and genotypes for each sample.

2.3. Phenotyping for Yield, Tuber Shape, and Eye Number

The two parental lines (IVP101 and Y8), two F₁ plants (IYY30 and IYP31), and 174 F₂ lines were grown across three environments (E1–E3). E1 corresponded to March–July 2024, and E3 to September 2024–January 2025, both conducted in a greenhouse at the Songming Research Base of Yunnan Academy of Agricultural Sciences (N 25°21′, E 103°6′). E2 was conducted in a walk-in growth chamber at the Joint Academy of Potato Science of Yunnan Normal University from June to October 2024, with environmental conditions set as follows: photoperiod of 7:00–19:00 (light, 23 °C) and 19:00–7:00 (dark, 19 °C), and light intensity of 261 PPFD. For E1 and E2, uniform tissue culture seedlings were transplanted; for E3, uniformly sized and germinated seed tubers were sown. In E1 and E3, plants were grown in large pots (29 cm diameter × 23 cm height), with 5 plants per line; in E2, small pots (14 cm diameter × 13 cm height) were used, with 3 plants per line. Crop management and pest control followed local recommendations.

After plant maturation, eight phenotypic traits were investigated: (1) Tuber number per plant (TNP): Individual counts of all tubers harvested per pot. (2) Tuber fresh weight per plant (TFWP): Total weight of all tubers per pot, measured using an electronic balance. (3) Mean tuber weight (MTW): Five large, uniform tubers (labeled 1–5) were selected from each pot, weighed individually, and the mean was calculated. (4) Tuber length (TL): Measured on labeled tubers 1–5 using a vernier caliper. (5) Tuber width (TW): Measured on labeled tubers 1–5 using a vernier caliper. (6) Tuber length/width ratio (TLWR): Calculated as tuber length divided by tuber width. (7) Tuber eye number (TEN): Visually counted on labeled tubers 1–5. (8) Eye number-to-tuber volume ratio (ENTVR): Calculated as the number of eyes divided by tuber volume; volume of the five tubers was measured via water displacement using a graduated cylinder. All phenotypic data were collated using Microsoft Excel 16. Statistical analyses were performed using SAS (Statistical Analysis System, version 8.01, SAS Institute Inc., Cary, NC, USA). Best linear unbiased estimators (BLUEs) were calculated for root phenotypic traits in the F₂ population using QTL IciMapping v4.2 software [43].

2.4. Sample Genomic Resequencing, Quality Control, and SNP Identification

After passing quality control, DNA samples of the two parental lines and 174 F₂ lines were entrusted to Beijing Biomarker Technologies Co., Ltd. (Beijing, China) for library construction and whole-genome resequencing. Library construction was performed using the VAHTS™ Fg DNA Library Prep Kit for Illumina. Briefly, 10 ng of Qubit-quantified genomic DNA was used and supplemented with FEA Buffer (2 μL) and FEA Enzyme Mix (4 μL) in a total volume of 20 μL. This mixture was subjected to PCR thermal cycling for enzymatic fragmentation and end repair with the following cycling conditions: 4 °C for 15 s, followed by 37 °C for 8 min, 65 °C for 30 min, and a final hold at 4 °C (with a 75 °C heated lid). Fragmented products (20 μL) were then mixed with Rapid Ligation Buffer 3 (10 μL), DNA Adapter (2 μL), Rapid DNA Ligase (2 μL), and ddH₂O (6 μL) in a 40 μL reaction system for ligation at 20 °C for 15 min (with the heated lid turned off). Ligation products were purified using 0.8× VAHTS™ DNA Clean Beads (Cat. No. N411-03), followed by size selection via adding 0.485× beads (supernatant collected) and 0.1× beads (pellet retained). The beads were washed twice with 80% ethanol, and DNA was eluted in 10 μL nuclease-free water; 8 μL of the eluate was used for PCR amplification. The PCR reaction mixture (20 μL) contained PCR Primer Mix 3 (2 μL), VAHTS HiFi Amplification Mix (10 μL), and size-selected product (8 μL). PCR products were purified using 0.6× beads to obtain the final library. Libraries were quality-controlled using a Qsep-400 (Qsep Inc., Taiwan), with a fragment peak at 430–530 bp, average fragment size of 420–580 bp, a normal distribution, and absence of unspecific peaks. Library concentration was quantified using a Qubit 3.0 Fluorometer (Life Technologies) (concentration ≥ 1 ng/μL). Adapter sequences were as follows: adapter3 = “AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC”; adapter5 = “AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT”. Libraries were sequenced on a BGI DNBSEQ-T7 sequencer.

Raw Data were generated from the original sequencing output. To ensure the quality of bioinformatics analysis, Raw Reads were filtered to obtain Clean Reads for subsequent analyses. The main filtering steps were as follows: (1) removal of adapter sequences; (2) filtering out paired-end reads where the proportion of N (undetermined base calls) exceeded 10%; and (3) removal of low-quality reads (reads with >50% of bases having a quality score ≤ 10).

For variant analysis, alignment of clean reads to the DM8.1 reference genome (http://www.bioinformaticslab.cn/pubs/dm8/, accessed on 10 May 2025) was performed using bwa-mem2 (v2.2) [44] with parameters set as “bwa-mem2 mem -t 4 -M”. Aligned results were sorted using samtools (https://sourceforge.net/projects/samtools/files/samtools/, accessed on 22 May 2025) sort, followed by calculation of alignment rate, sequencing depth, and coverage using samtools flagstat/depth.

Detection of SNPs and small INDELs was conducted using the GATK [45] toolkit: first, redundant reads were filtered with samtools to ensure accuracy. Then, GATK HaplotypeCaller (local haplotype assembly algorithm) was used to generate gVCF files for each sample, followed by population-level joint genotyping.

Variant results were rigorously filtered to ensure reliability: (1) SNPs within 5 bp of INDELs and adjacent INDELs within 10 bp were filtered using bcftools vcfutils.pl (varFilter -w 5 -W 10); (2) clusterSize 2 and clusterWindowSize 5 were set (no more than 2 variants within a 5 bp window); (3) GATK hard filtering parameters were applied: QUAL < 30 (Phred-scaled quality score for variant reliability), QD < 2.0 (ratio of variant quality to coverage depth), MQ < 40 (root mean square of mapping quality), and FS > 60.0 (strand bias estimated by Fisher’s exact test). Other parameters followed GATK official default values. The final set of variant sites was stored in VCF format.

2.5. Linkage Map Construction and QTL Analyses

To ensure marker quality, the final SNP set was filtered and screened as follows: (1) retaining markers that were homozygous and polymorphic in parents (aa × bb); (2) filtering out markers with a depth < 4×; (3) removing markers not mapped to chromosomes. After obtaining high-quality SNPs, a sliding window approach was applied across chromosomes using a 15-SNP window and 1-SNP step size. If ≥13 SNPs in a window were genotyped as “aa”, the window was assigned “aa”; if ≥13 SNPs were “bb”, it was assigned “bb”; otherwise, genotypes were imputed and corrected as “ab” [46]. Following SNP imputation and correction, samples were aligned by chromosomal physical positions. A recombination breakpoint was defined as a genotype transition in any sample, and SNPs between breakpoints were clustered into Bins (non-recombinant intervals), generating unfiltered Bin markers. Further filtering excluded polymorphic Bin markers with length < 5 kb or severe segregation distortion (χ² test, p < 0.001).

Filtered Bin markers were subjected to multi-round deduplication using HighMap software [47]. Optimal recombination rates between markers were calculated, and genetic distances between adjacent Bins were estimated to construct a linkage map. QTL IciMapping v4.2 software [43] was used for QTL analyses. Additive QTLs were calculated using the ICIM-ADD mapping method. The threshold for logarithm of odds (LOD) scores was set as 2.5.

3. Results

3.1. Feasibility Evaluation of Genetic Map Construction Using Resequencing Data from Two Parental Lines

SNPs that are homozygous and polymorphic (aa × bb) between parents can exhibit simple segregation patterns in F₂ populations, making them the most ideal marker type for genetic map construction. Theoretically, two homozygous diploid parents can generate a large number of aa × bb-type SNP markers with high genome coverage. However, in potato, we constructed an F₂ segregating population using one genomically homozygous (Y8) and one genomically heterozygous (IVP101) diploid for the first time, and whole-genome resequencing is planned to develop genome-wide SNP markers for genetic map construction.

To verify the feasibility and quality of map construction, genome resequencing was first performed on the two parents. Their clean data were aligned to the DM8.1 reference genome, yielding a total of 9,530,402 SNPs. After filtering out SNPs with parental missingness, non-polymorphism, or sequencing depth < 4×, 8,149,439 SNPs remained for genotyping (Figure 1B). These markers were primarily classified into four types, with the nn × np type being the most abundant (5,601,059), while our target aa × bb type accounted for 1,860,910 markers (Figure 1A). Markers were more densely distributed on chromosomes 1, 4, 9, and 10, and relatively fewer on chromosomes 2, 3, 5, 7, 8, 11, and 12 (Figure 1C). To analyze their chromosomal distribution in detail, chromosome density plots were generated using a 1 Mb window size (Figure 1D and Table S1). SNP markers were present in all windows with uniform distribution, meeting the requirements for subsequent map construction.

3.2. Construction of F₂ Genetic Segregating Population

F₁ seeds were obtained by crossing IVP101 (male parent) and Y8 (female parent). After randomly sowing a portion of the seeds, 49 F₁ plants were generated. To further construct an F₂ segregating population, two vigorously growing and highly self-fertile F₁ plants (designated IYY30 and IYP31) were selected for selfing, ultimately generating an F₂ genetic segregating population with 174 lines. Prior to this, given IVP101’s capacity to induce ploidy reduction, ploidy analysis of IYY30 and IYP31 was performed using a flow cytometer, with the tetraploid cultivar Cooperation-88 and the two parents (Y8 and IVP101) as controls. As shown in Figure 2C, both IYY30 and IYP31 were diploid and had not undergone ploidy reduction influenced by the male parent IVP101.

Subsequently, preliminary screening for SSR marker polymorphism was conducted using DNA from IVP101 (male parent) and Y8 (female parent) as templates. From 231 primer pairs, two SSR markers with good polymorphism, Sot071 and Sot077, were identified (Table S2). These markers were used for parentage identification of IYY30 and IYP31. Capillary electrophoresis results revealed that both IYY30 and IYP31 exhibited specific genetic loci from both parents (Figure 2D), confirming they were true hybrids.

3.3. Resequencing and Genotyping of Samples

All samples were subjected to resequencing and data quality control filtering. The male parent IVP101 generated 45.71 Gb of clean data, the female parent Y8 yielded 46.59 Gb, and 174 progeny produced a total of 1693.04 Gb of clean data. All sequencing data achieved a Q30 score of over 85%. Alignment with the reference genome DM8.1 showed that the percentage of clean reads mapped to the reference genome exceeded 97% of all clean reads, indicating high alignment efficiency. The calculated genome sequencing depths for IVP101 and Y8 were 56× and 58×, respectively, with average genome coverages of 93.93% and 91.34% (at least 1× coverage). The average sequencing depth of F₂ progeny reached 12.58×, with coverage exceeding 87.97% (at least 1× coverage) (

Table 1. Resequencing data statistics and alignment with reference genome for parents and F₂ population.

Sample ID	Total Clean Reads a	Total Clean Bases b	Q30 Percentage (%) c	GC Percentage (%) d	Mapped (%) e	Properly_Mapped (%) f	Ava-depth g	Cov_ratio_1X (%) h	Cov_ratio_5X (%)	Cov_ratio_10X (%)
A056	311,734,376	46,587,204,544	96.38	36.04	97.93	88.81	58	91.34	87.66	85.15
IVP101	304,744,996	45,711,749,400	85.79	35.71	99.12	88.89	56	93.93	91.07	89.1
Offspring	11,340,051,656	1,693,037,684,560	96.32	36.21	98.11	88.61	12.58	87.97	72.11	44.05

^a Total clean reads: Number of filtered reads for each sample; ^b Total clean bases: Number of filtered bases for each sample; ^c Q30 percentage: Percentage of bases with sequencing quality value ≥ 30; ^d GC percentage: Percentage of G and C bases in the total bases; ^e Mapped (%): Percentage of clean reads mapped to the reference genome relative to total clean reads; ^f Properly mapped (%): Percentage of paired-end sequencing reads both mapped to the reference genome with distances consistent with the fragment length distribution; ^g Ave-depth: Average coverage depth of the sample; ^h Cov_ratio: Proportion of bases with coverage depth ≥ specified depth relative to the total bases of the reference genome.

). These results confirm high sequencing quality and uniform genome coverage distribution, which fully meets the quality requirements for subsequent studies such as genetic map construction.

SNP calling was performed on all samples using GATK, yielding a total of 11,073,260 SNP loci. Statistical analysis of homozygous and heterozygous loci revealed that the proportion of homozygous SNPs in Y8 was as high as 95.26%, whereas that in IVP101 was only 53.49% (Table S3). These results were consistent with our expectations: Y8 is a highly homozygous inbred line, while IVP101 is a diploid with a highly heterozygous genome.

3.4. Bin Map Construction and Characteristics

Of the 11,073,260 SNPs, a total of 6,588,620 SNPs were detected between the two parents. After filtering based on thresholds, a total of 1,552,920 SNPs were retained. Using a 15-SNP window and 1-SNP step size, a sliding scan across chromosomes generated 63,595 unfiltered Bin markers. Further filtering removed Bin markers shorter than 5 kb and those with severe segregation distortion (χ² test, p < 0.001), resulting in 5125 filtered Bin markers for subsequent linkage analysis (

Table 2. Statistics of bin markers for genetic map analysis.

LG	SNP Markers	Bin Marker	Average Length (Mb)
LG01	23,773	114	0.13
LG02	74,385	364	0.092
LG03	99,340	367	0.156
LG04	86,461	780	0.045
LG05	58,453	466	0.056
LG06	27,995	139	0.075
LG07	63,335	374	0.103
LG08	55,097	329	0.097
LG09	157,167	645	0.075
LG10	90,145	807	0.046
LG11	45,399	369	0.055
LG12	50,740	371	0.052
Total	832,290	5125	0.082

Note: LG: Linkage group number; SNP markers: Number of SNP markers identified on the linkage group; Bin marker: Number of Bin markers on the linkage group; Average Length (Mb): Average number of base pairs per bin.

).

After multiple rounds of redundancy removal for 5125 Bin markers using HighMap software, the optimal recombination rates between markers were calculated, and genetic distances between adjacent Bin markers were estimated, yielding the linear arrangement of Bin markers within linkage groups. A total of 4464 mapped Bin markers were ultimately obtained (

Table 3. Distribution of genetic markers across the 12 chromosomes in potato.

Chromosome	Total Bin Marker	Total Distance (cM)	Average Distance (cM)	Max Gap (cM)	Gaps < 5 cM (%)	Average Physical Distance Between Markers (kb) a
1	265	75.02	0.28	2.65	1	335.87
2	336	137.59	0.41	6.77	0.9821	138.13
3	358	116.53	0.33	6.47	0.9944	171.35
4	459	75.98	0.17	1.78	1	152.31
5	470	127.02	0.27	7.06	0.9957	117.92
6	231	72.63	0.32	4.50	1	258.78
7	347	99.01	0.29	7.41	0.9971	168.78
8	342	108.94	0.32	9.54	0.9971	175.74
9	430	106.02	0.25	5.48	0.9977	158.46
10	427	104.42	0.25	7.18	0.9953	145.10
11	382	109.91	0.29	16.27	0.9948	124.36
12	417	106.67	0.26	7.53	0.9952	145.07
Overall	4464	1239.74	0.28	16.27	0.9957	165.51

^a annotation on DM8.1 (http://www.bioinformaticslab.cn/pubs/dm8/, accessed on 10 May 2025).

and Figure 3A). This map contained 12 linkage groups, corresponding exactly to the chromosome number of diploid potato. Chromosome 5 harbored the most markers (470), while chromosome 6 contained the fewest (231). The total map length of the 12 linkage groups was 1239.74 cM, with an average genetic distance of 0.28 cM between adjacent markers and an average physical distance of only 165.51 kb. Statistical analysis of genome-wide genetic distances between all markers showed that 99.57% of genetic gaps were smaller than 5 cM, with the largest gap (16.27 cM) located on chromosome 11.

The Spearman correlation coefficient was calculated for all the linkage groups. Its value near to 1 showed an improved collinearity between the physical and genetic maps. The order of most markers on every chromosome of the project is consistent with the genome, indicating that collinearity is good and the calculation accuracy of the genetic recombination rate is high (Figure 3B).

3.5. Phenotypic Analysis of F₂ Population and Parents

We investigated eight phenotypic traits (Table S4) in parental lines IVP101 and Y8, two F₁ hybrids (IYY30 and IYP31), and 174 F₂ lines across three environments. These included three yield-related traits: Tuber Number per Plant (TNP), Tuber Fresh Weight per Plant (TFWP), and Mean Tuber Weight (MTW); three tuber-shape-related traits: Tuber Length (TL), Tuber Width (TW), and Tuber Length-to-Width Ratio (TLWR); and two eye-related traits: Tuber Eye Number (TEN) and Eye Number-to-Tuber Volume Ratio (ENTVR). Best linear unbiased estimates (BLUEs) were calculated for the eight traits based on phenotypic values across the three environments. The results revealed highly significant differences in all eight traits between the two parents and between the two F₁ hybrids (Figure 2A), indicating substantial genetic differences between the two parents, which are beneficial for QTL identification. For the F₂ population, all eight traits exhibited wide variation (Figure 2B), with high coefficients of variation observed for all traits except TLWR. Additionally, the phenotypic values of all eight investigated traits showed continuous unimodal distributions, consistent with quantitative traits controlled by multiple genes (

Table 4. Analysis of phenotypic data for yield, tuber shape, and eye number in parents, F₁, and F₂ populations across three seasons.

Traits Name a	Env b	Parents c		F1 d		F2 Population						Heritability (%) f
		IVP101	Y8	IYY30	IYP31	Mean ± SD	Minimum	Maximum	Skewness	Kurtosis	CV (%) e
TNP	E1	14.66 ± 1.53 a	8.33 ± 1.62 b	10.00 ± 0.58 b	9.67 ± 1.00 b	15.35 ± 9.11	1.00	51.00	0.88	0.89	59.36	58.51
	E2	12.67 ± 1.53 a	7.33 ± 1.53 b	9.33 ± 2.08 ab	7.33 ± 0.58 b	17.22 ± 10.86	1.50	57.33	1.15	1.67	63.05
	E3	15.67 ± 2.08 a	9.67 ± 1.15 b	12.67 ± 2.08 ab	11.33 ± 1.52 ab	22.57 ± 15.40	1.00	117.00	2.47	11.22	68.25
	BLUE	14.37	8.38	10.43	9.44	17.28 ± 7.73	2.00	38.93	0.61	0.23	44.74
TFWP (g)	E1	148.75 ± 15.45 b	88.93 ± 7.76 c	177.02 ± 8.67 a	79.12 ± 3.29 c	45.32 ± 42.17	0.08	205.11	1.32	1.43	93.07	71.72
	E2	23.18 ± 1.49 b	12.16 ± 1.31 c	26 ± 1.67 c	11.84 ± 2.44 b	18.99 ± 12.47	0.77	53.18	0.61	−0.5	65.7
	E3	180.08 ± 16.14 b	115.6 ± 9.74 c	207.02 ± 8.67 a	107.65 ± 3.92 c	105.15 ± 61.78	1.37	282.27	0.28	−0.42	58.76
	BLUE	30.38	17.38	35.42	15.6	23.53 ± 22.36	0.79	199.43	4.18	27.16	95.02
MTW (g)	E1	18.08 ± 0.60 b	12.71 ± 0.53 c	25.46 ± 0.68 a	9.54 ± 0.54 d	5.49 ± 5.07	0.32	32.78	2.27	7.81	92.37	62.89
	E2	3.71 ± 0.42 ab	2.83 ± 0.39 bc	4.50 ± 0.54 a	2.19 ± 0.59 c	2.44 ± 1.60	0.20	7.63	0.93	0.29	65.61
	E3	38.75 ± 1.1 b	32.72 ± 0.55 c	45.82 ± 1.02 a	29.67 ± 0.49 d	10.73 ± 6.90	0.00	30.25	0.48	−0.09	64.31
	BLUE	13.36	10.36	17.06	8.6	3.03 ± 2.79	0.21	19.77	2.92	11.79	91.82
TL (mm)	E1	50.77 ± 2.53 a	33.20 ± 2.06 b	57.60 ± 13.38 a	29.08 ± 0.65 b	23.04 ± 9.95	8.12	58.93	0.9	0.25	43.18	72.26
	E2	29.86 ± 2.51 a	15.47 ± 0.55 c	22.21 ± 3.72 b	15.95 ± 1.55 c	16.39 ± 4.03	7.40	28.81	0.08	−0.35	24.58
	E3	53.24 ± 2.72 a	36.43 ± 2.51 b	58.88 ± 11.67 a	32.43 ± 0.67 b	26.22 ± 6.86	9.44	43.10	−0.35	−0.23	26.17
	BLUE	33.55	18.7	28.18	18.44	18.58 ± 4.77	9.11	40.83	0.83	2.7	25.69
TW (mm)	E1	25.77 ± 1.53 a	30.67 ± 1.33 a	28.81 ± 7.5 a	25.93 ± 0.95 a	15.01 ± 4.27	4.95	26.78	0.16	−0.37	28.42	76.09
	E2	15.70 ± 1.21 a	14.33 ± 0.81 a	18.41 ± 3.93 a	13.18 ± 0.55 a	12.54 ± 3.31	4.88	20.28	−0.05	−0.63	26.41
	E3	28.96 ± 0.92 a	33.48 ± 2.65 a	31.55 ± 8.06 a	29.05 ± 0.53 a	21.56 ± 5.49	6.85	32.08	−0.48	−0.16	25.48
	BLUE	19.81	20.47	22.57	18.23	13.84 ± 3.46	5.28	23.98	0.03	−0.09	24.99
TLWR	E1	1.98 ± 0.18 a	1.11 ± 0.04 b	2.01 ± 0.12 a	1.12 ± 0.03 b	1.50 ± 0.35	1.08	2.79	0.88	0.25	23.05	56.37
	E2	1.90 ± 0.12 a	1.08 ± 0.03 b	1.22 ± 0.14 b	1.21 ± 0.07 b	1.34 ± 0.26	1.03	2.69	2.52	7.85	19.12
	E3	1.84 ± 0.09 a	1.09 ± 0.01 b	1.89 ± 0.12 a	1.12 ± 0.01 b	1.23 ± 0.07	1.07	1.46	0.84	0.75	5.71
	BLUE	1.89	1.09	1.72	1.14	1.28 ± 0.12	1.14	2.18	3.23	16.87	9.65
TEN	E1	20.00 ± 5.00 a	10.33 ± 2.08 b	25.67 ± 4.93 a	7.67 ± 1.53 b	11.96 ± 3.91	1.29	29.52	1.06	2.64	32.69	53.86
	E2	17.33 ± 6.03 a	9.00 ± 2.00 a	8.33 ± 1.53 a	9.67 ± 2.52 a	8.10 ± 2.05	4.80	19.00	1.62	4.59	25.27
	E3	20.33 ± 3.79 a	12.67 ± 1.53 b	21.33 ± 2.08 a	10.67 ± 2.52 b	9.15 ± 1.55	4.52	13.16	−0.02	−0.13	16.89
	BLUE	19.44	11.03	19.12	9.55	8.76 ± 1.49	5.13	14.54	0.69	0.9	16.96
ENTVR	E1	1.12 ± 0.06 a	0.81 ± 0.13 b	0.81 ± 0.13 b	0.72 ± 0.13 b	2.75 ± 1.18	0.00	6.08	0.1	0.46	43.09	75.43
	E2	3.96 ± 1.18 a	2.5 ± 0.34 ab	1.68 ± 0.14 b	3.05 ± 0.83 ab	2.85 ± 1.17	1.21	6.61	0.78	0.1	41.08
	E3	0.55 ± 0.09 a	0.40 ± 0.05 a	0.49 ± 0.04 a	0.37 ± 0.09 a	1.33 ± 0.82	0.34	4.83	1.72	3.42	61.6
	BLUE	0.71	0.52	0.57	0.48	2.24 ± 0.92	0.85	5.28	0.88	0.28	40.99

^a Traits Name, TNP, Tuber Number per Plant; TFWP, Tuber Fresh Weight per Plant; MTW, Mean Tuber Weight; TL, Tuber Length; TW, Tuber Width; TLWR; Tuber Length-to-Width Ratio; TEN, Tuber Eye Number; ENTVR, Eye Number-to-Tuber Volume Ratio. ^b Env, E1 denotes March–July 2024, E3 denotes September 2024–January 2025, and E2 denotes June–October 2024 in a walk-in growth chamber. BLUE is calculated based on the phenotypic data from three environments. ^c,d Parents and F1 refer to the mean ± standard deviation (SD) of the data in four materials. Letters from a to d indicate significantly different values according to statistical analysis using Duncan’s multiple range test (α = 0.05). ^e CV, coefficient of variation. ^f Heritability (%), broad sense heritability cross three environments.

and Figure 4). We also analyzed broad-sense heritability across four environments, which ranged from 53.86% to 76.09% (Table 4), indicating that these traits are primarily genetically controlled despite environmental influences.

Correlations between pairs of the eight traits were calculated using BLUE values, with results presented in

Table 5. Correlation analysis of 8 traits related to yield, tuber shape, and eye in the F₂ population.

Traits Name	TNP	TFWP	MTW	TL	TW	TLWR	EN	ENTVR
TNP
TFWP	0.49 **
MTW	0.27 **	0.88 **
TL	0.26 **	0.89 **	0.85 **
TW	0.28 **	0.89 **	0.90 **	0.91 **
TLWR	−0.01 ns	0.11 ns	0.03 ns	0.32 **	−0.07 ns
TEN	0.12 ns	0.41 **	0.39 **	0.55 **	0.42 **	0.34 **
ENTVR	−0.24 *	−0.74 **	−0.76 **	−0.71 **	−0.82 **	0.19 *	−0.01 ns

*, ** significant at the 0.05 and 0.01 levels, respectively.

. TFWP, a direct indicator of yield, can be decomposed into two yield components: TNP and MTW. In this study, TFWP showed a correlation coefficient of 0.49 with TNP and 0.88 with MTW. Additionally, TL and TW effectively characterize tuber size; notably, pairwise correlation coefficients among TFWP, MTW, TL, and TW exceeded 0.85, indicating a greater contribution of MTW to yield. TLWR was uncorrelated or weakly correlated with other traits. TEN exhibited moderate positive correlations with TL, TW, and MTW, while TLWR showed strong negative correlations with TL, TW, and MTW.

3.6. QTL Mapping in the F₂ Population

Using the inclusive composite interval mapping method, a total of 148 additive QTLs were detected with IciMapping v4.2 software based on the high-density genetic map (Table S5). These included 39 in E1, 31 in E2, 36 in E3, and 42 based on BLUE values. QTLs were distributed across all chromosomes, with chromosome 9 harboring the fewest (1 QTL) and chromosome 12 the most (33 QTLs). The phenotypic variation explained (PVE) of all QTLs ranged from 0.16% to 30.45%. By trait category, 50 QTLs were identified for tuber yield traits (13 for TNP, 19 for TFWP, and 18 for MTW), 60 for tuber shape traits (23 for TL, 20 for TW, and 17 for TLWR), and 38 for tuber eye traits (17 for TEN and 21 for ENTVR). After merging QTLs by physical position, 89 unique genetic loci were obtained, comprising 32 for tuber yield, 33 for tuber shape, and 24 for tuber eye traits.

To further authenticate the correctness of these results, loci of the present study were compared with reported QTLs previously identified by the linkage or association mapping approach. We discovered seven loci that have been reported previously (Table S5). qTY-2 was detected by both TFWP and MTW. This locus was also significantly associated with mean tuber weight in another diploid GWAS study [36]. qTY-5-4, identified by MTW in this study, was significantly associated with mean tuber weight in a diploid GWAS study [36]. qTY-8-3 was detected by both TFWP and MTW in this study and was also significantly associated with mean tuber weight in another diploid GWAS study [36]. StOFP20 [39], the only cloned and validated tuber shape regulatory gene to date, was notably detected in our study as qTS-10-2 via genetic mapping combined with TLWR phenotypic data. Furthermore, qTS-5-2, detected by both TL and TLWR in this study, was identified by Meijer et al. [13]. using a high-density genetic map of an F₂ segregating population, explaining 6.3% of phenotypic variation. In this study, qTS-1-2 was detected by TW and co-localized with the tuber shape-associated significant SNP locus chr01_84614836 [36]. For tuber eye traits, Endelman and Jansky [29] identified a QTL at chr05: 47,991,614–49,804,489 using eye tubers (yes/no) as a trait and a high-density genetic map, which was also detected by TEN in our population (qTE-5-4). Overall, our QTLs mapped using this high-density bin map are reliable.

3.7. Identification of Stable QTLs

QTL mapping was conducted for 8 traits across three environments, resulting in the identification of 89 loci (Table S5). Some loci were detected by two or more traits; notably, 10 loci were mapped by traits in at least two environments. These QTLs were repeatedly detected across environments, exhibited stable expression, and thus had higher reliability, leading to their designation as stable QTLs. Three stable QTLs were associated with tuber yield: qTY-1-3, qTY-5-2, and qTY-12-6. Four stable QTLs were associated with tuber shape: qTS-2-1, qTS-5-1, qTS-8-3, and qTS-12-3. Three stable QTLs were associated with tuber eye number: qTE-10-2, qTE-12-2, and qTE-12-4 (

Table 6. Information on the 10 stable QTLs.

Traits	QTLs	TraitName	Environment	Chromosome	Marker Interval	Physical Interval (bp)	LOD a	PVE (%) b	Add c	Dom d
Tuber Yield	qTY-1-3	TNP	E3	1	Block1809-Block2598	32,322,452–43,216,548	11.17	7.13	35.63	−35.23
		TFWP	BLUE	1	Block1809-Block2598	32,322,452–43,216,548	13.34	7.61	62.19	−62.46
		MTW	E1	1	Block1809-Block2598	32,322,452–43,216,548	7.95	9.90	11.25	−10.64
	qTY-5-2	MTW	E1	5	Block26017-Block26034	28,815,367–30,380,884	4.62	3.82	−0.21	−3.53
		MTW	E2	5	Block26017-Block26034	28,815,367–30,380,884	5.27	8.63	−0.74	−0.33
		MTW	BLUE	5	Block26017-Block26034	28,815,367–30,380,884	3.49	2.73	−1.09	−0.40
	qTY-12-6	TFWP	E1	12	Block63545-Block63552	58,272,481–58,483,696	2.62	6.68	12.87	8.17
		TNP	E3	12	Block63557-Block63558	58,846,303–59,038,431	7.65	4.22	9.10	0.23
		TFWP	E3	12	Block63562-Block63563	59,191,799–59,272,168	17.22	25.26	51.52	12.23
		MTW	E3	12	Block63562-Block63563	59,191,799–59,272,168	11.22	20.35	4.06	3.40
		MTW	E2	12	Block63568-Block63586	59,356,926–59,931,216	9.41	15.40	0.97	0.66
		MTW	BLUE	12	Block63568-Block63586	59,356,926–59,931,216	4.62	3.62	1.25	0.50
Tuber Shape	qTS-2-1	TL	E3	2	Block6751-Block6786	20,496,610–20,799,674	2.52	4.85	2.49	−3.10
		TL	E2	2	Block6805-Block6812	20,865,801–20,964,210	2.77	5.00	1.21	−1.04
		TL	BLUE	2	Block6805-Block6812	20,865,801–20,964,210	2.58	4.22	1.40	−1.19
		TLWR	BLUE	2	Block6812-Block6831	20,937,218–21,011,445	5.49	7.82	0.01	−0.08
	qTS-5-1	TW	E2	5	Block26017-Block26034	28,815,367–30,380,884	6.80	10.09	−1.94	0.11
		TW	BLUE	5	Block26017-Block26034	28,815,367–30,380,884	7.03	9.59	−1.59	−0.45
		TL	E2	5	Block26017-Block26034	28,815,367–30,380,884	5.73	11.65	−2.04	−0.73
		TL	BLUE	5	Block26017-Block26034	28,815,367–30,380,884	7.70	14.26	−2.59	−1.39
		TW	E1	5	Block26017-Block26034	28,815,367–30,380,884	7.18	8.30	−0.40	−2.41
	qTS-8-3	TW	E2	8	Block37053-Block37061	47,632,918–47,845,728	3.21	4.85	−0.01	1.56
		TL	E2	8	Block37053-Block37061	47,632,918–47,845,728	3.47	7.25	−0.14	2.31
		TW	E3	8	Block37066-Block37067	48,093,484–48,266,777	3.59	7.92	−0.54	2.90
		TL	E3	8	Block37066-Block37067	48,093,484–48,258,086	3.58	5.95	−0.34	3.76
	qTS-12-3	TW	E3	12	Block63562-Block63563	59,191,799–59,272,168	8.86	20.11	3.44	2.15
		TL	E3	12	Block63562-Block63563	59,191,799–59,272,168	9.04	15.35	4.28	2.83
		TL	BLUE	12	Block63562-Block63563	59,191,799–59,272,168	2.70	4.38	1.01	1.76
		TW	E2	12	Block63563-Block63564	59,228,476–59,288,325	11.02	16.92	2.10	0.03
		TW	BLUE	12	Block63568-Block63586	59,356,926–59,931,216	14.91	22.88	2.39	0.02
		TW	E1	12	Block63586-Block63594	59,867,456–60,213,650	9.69	10.72	1.93	2.14
		TL	E1	12	Block63586-Block63594	59,867,456–60,213,650	4.13	3.77	3.11	1.54
		TLWR	E2	12	Block63586-Block63594	59,867,456–60,213,650	24.63	30.45	−0.21	−0.22
		TLWR	BLUE	12	Block63586-Block63594	59,867,456–60,213,650	14.18	18.05	−0.07	−0.07
Tuber Eye Number	qTE-10-2	TEN	E2	10	Block54399-Block54400	50,222,323–50,306,461	3.00	4.17	−0.24	0.97
		TEN	E3	10	Block54400-Block54431	50,294,024–50,583,496	8.05	18.18	−0.81	0.84
		TEN	BLUE	10	Block54400-Block54431	50,294,024–50,583,496	3.53	6.20	−0.31	0.66
	qTE-12-2	ENTVR	E2	12	Block61447-Block61456	38,024,947–42,352,859	12.35	11.37	−0.63	−0.12
		TEN	E3	12	Block61452-Block61460	38,725,985–42,461,859	3.12	5.94	−0.62	0.11
	qTE-12-4	ENTVR	E3	12	Block63562-Block63563	59,191,799–59,272,168	21.37	16.92	−0.70	−0.53
		TEN	E2	12	Block63586-Block63594	59,867,456–60,213,650	11.26	17.80	−1.27	−1.30
		TEN	BLUE	12	Block63586-Block63594	59,867,456–60,213,650	4.41	7.06	−0.23	−0.81
		ENTVR	E1	12	Block63586-Block63594	59,867,456–60,213,650	9.29	23.90	−0.76	−0.53
		ENTVR	E2	12	Block63586-Block63594	59,867,456–60,213,650	25.49	27.74	−0.80	−0.98
		ENTVR	BLUE	12	Block63586-Block63594	59,867,456–60,213,650	18.57	26.63	−0.73	−0.11

^a LOD, logarithm of odds. ^b PVE (%), phenotypic variation explained (%). ^c Add, additive effect. ^d Dom, dominant effect.

and Figure 5).

Notably, among the 10 stable QTLs, two exhibited pleiotropy. The locus on chromosome 5 (qTY-5-2 and qTS-5-1) co-regulates TL, TW, and MTW. The pleiotropic locus at the terminal region of chromosome 12 (qTY-12-6, qTS-12-3, and qTE-12-4) regulates all eight traits, with high phenotypic variation explained: a maximum of 25.26% for yield (TFWP-E3), 30.45% for tuber shape (TLWR-E2), and 23.90% for tuber eye traits (ENTVR-E2). These results confirm it as a major-effect locus with significant research and application value.

3.8. Validation of the Effects of the 10 Stable QTLs

To further clarify the effects of the ten stable QTLs, without considering interactions between loci, we summarized phenotypic differences among the three genotypes of each locus in the F₂ population (

Table 7. Summary of the phenotypic effects of ten stable QTLs.

Stable QTLs	Traits Type	Traits	Number of Lines a			Donor of Positive Allele b	Phenotypic Value c
			Paternal Genotype (IVP101)	Maternal Genotype (Y8)	Heterozygous Genotype (H)		Paternal Genotype (IVP101)	Maternal Genotype (Y8)	Heterozygous Genotype (H)
qTY-1-3	Tuber Yield	TNP	73	15	83	Y8	16.76 b	19.45 a	16.98 b
		TFWP (g)					21.69 b	25.66 a	24.31 a
		MTW (g)					2.80 b	3.27 a	3.18 a
qTY-5-2		TNP	56	8	89	IVP101	16.96 a	16.99 a	17.27 a
		TFWP (g)					30.77 a	15.23 b	20.68 ab
		MTW (g)					3.96 a	1.66 b	2.71 ab
qTY-12-6		TNP	29	39	84	Y8	13.25 b	19.65 a	17.75 a
		TFWP (g)					13.08 b	27.67 a	25.48 a
		MTW (g)					1.74 b	3.64 a	3.29 a
qTS-2-1	Tuber Shape	TL (mm)	31	15	70	Y8	18.58 b	21.46 a	17.86 b
		TW (mm)					13.46 a	14.88 a	13.71 a
		TLWR					1.30 b	1.42 a	1.26 b
qTS-5-1		TL (mm)	56	8	89	IVP101	20.94 a	15.66 b	17.36 b
		TW (mm)					15.52 a	12.63 b	13.06 b
		TLWR					1.31 a	1.22 b	1.26 ab
qTS-8-3		TL (mm)	39	25	84	IVP101	17.96 ab	16.68 b	19.61 a
		TW (mm)					13.29 ab	12.44 b	14.52 a
		TLWR					1.28 a	1.29 a	1.29 a
qTS-12-3		TL (mm)	29	39	84	Y8	17.47 a	19.28 a	18.93 a
		TW (mm)					11.08 b	14.66 a	14.72 a
		TLWR					1.40 a	1.26 b	1.25 b
qTE-10-2	Tuber Eye Number	EN	31	36	82	IVP101	8.61 a	7.97 b	9.10 a
		ENTVR					2.01 a	2.19 a	2.23 a
qTE-12-2		EN	91	18	59	IVP101	9.04 a	8.30 b	8.46 ab
		ENTVR					2.43 a	1.95 b	2.02 ab
qTE-12-4		EN	29	39	84	IVP101	9.64 a	8.73 b	8.40 b
		ENTVR					3.24 a	1.97 b	1.91 b

^a Number of lines, the F₂ population was classified into paternal, maternal, and heterozygous genotypes based on the markers flanking the QTL. ^b Donor of positive allele, a positive value indicates an increasing effect from parent Y8. ^c Phenotypic value, Different letters (e.g., a, b) denote statistically significant differences as determined by Duncan’s multiple range test (α = 0.05).

). After segregation and recombination, each locus in the F₂ population exhibits homozygous paternal (II), maternal (YY), and heterozygous (IY) genotypes. Therefore, we first classified F₂ individuals into IVP101-type, Y8-type, and H-type based on marker genotypes of identified loci, then compared phenotypic differences in corresponding traits calculated using BLUE values.

Results showed that both parents contributed favorable alleles. For most QTLs, the average phenotypic values of all corresponding traits in F₂ lines carrying favorable alleles were significantly (p < 0.05) higher than those carrying unfavorable alleles. For example, at the qTY-5-2, the IVP101-derived allele increased average tuber weight by ~2.30 g and total tuber weight per plant by 15.54 g compared to the Y8-derived allele. At qTS-5-1, the IVP101 allele increased tuber length and width by 5.28 mm and 2.89 mm, respectively, with a corresponding 0.09 increase in TLWR. At qTE-12-2, the Y8-derived allele reduced eye number by ~0.48 per cm³. Regarding the major-effect QTL regulating all eight traits (qTY-12-6, qTS-12-3, and qTE-12-4), the allele derived from parent Y8 effectively enhances yield, promotes rounder tuber shape, and reduces tuber eyes. Specifically, it increases mean tuber weight by ~1.90 g and tuber number per plant by ~6.40 tubers; simultaneously, it increases tuber length and width by 1.81 mm and 3.58 mm, respectively, while reducing tuber eyes by 1.27 per cm³.

Additionally, some loci exhibited dominant heterozygous advantage, where heterozygous genotypes showed significantly higher phenotypic values than both parental genotypes. For instance, qTS-8-3 heterozygotes displayed the longest tuber length and width compared to both parental genotypes.

4. Discussion

4.1. Appropriate Genetic Populations and High-Density Genetic Maps Can Effectively Improve QTL Mapping Efficiency

Potato research lags behind major crops (e.g., rice, wheat) primarily due to cultivated potatoes being autotetraploids with complex progeny segregation. Most genetic studies on key traits use diploid materials [13,25,27,28,29,30,33,34], but self-incompatibility and inbreeding depression in diploids hinder genetic population development. Recent advances, including S-RNase gene editing [17] and cloning of the self-compatibility gene Sli [18], have overcome self-incompatibility. Concurrently, large-scale genome resequencing has enabled analysis of deleterious mutation distribution and frequency, mitigating inbreeding depression and facilitating development of genome-homozygous advanced diploid lines [19]. These homozygous diploid inbred lines enable the construction of primary populations (e.g., F₂, BC populations) and secondary mapping populations, such as near-isogenic lines (NILs), single-segment substitution lines (SSSLs), analogous to strategies in rice and Arabidopsis. Notably, secondary mapping populations like NILs and SSSLs offer key advantages: uniform genetic background, high mapping precision, and the ability to validate QTL effects. However, to date, genome-homozygous diploids generated via Sli introgression remain limited. Consequently, some mapping populations consist of F₁ hybrids from heterozygous diploid crosses [26,27,28,31,32], only a few utilize optimally selected homozygous parents to generate F₂ or RIL populations [29,33,34]. In this study, we developed highly homozygous self-compatible diploid lines through Sli introgression followed by multiple generations of selfing (Y8). Crosses between Y8 and the heterozygous parent IVP101 (paternal) produced an F₂ segregating population comprising 174 lines.

Although the Sli gene was genetically mapped as early as 1998 [48], most natural potato populations exist as heterozygous diploids, and Sli-mediated genome-homozygous lines remain scarce. While populations derived from two homozygous diploids are ideal, their development requires substantial time and resources. Thus, an efficient alternative involves crossing a self-compatible homozygous diploid (recipient) with a heterozygous diploid carrying elite alleles (donor) to generate F₂ populations. Although Meijer et al. [13] attempted this approach, their map included only 90 SNP markers, insufficient to evaluate whether heterozygous parents could provide adequate homozygous markers for map construction. Therefore, we first resequenced both parents to estimate homozygosity via SNPs. Consistent with expectations, Y8 exhibited 95.26% homozygosity, while IVP101 showed only 53.49% (Table S3). We then developed “aa × bb” type SNPs between parents, whose quantity and distribution were fully sufficient for subsequent high-density genetic map construction (Figure 1). Ultimately, we constructed a genetic map containing 4464 markers with an average physical distance of 165.51 kb, corresponding to ~17 genes in the DM8.1 reference genome. Combined with eight traits across three seasons, we detected 149 QTLs, including previously reported QTLs/genes (Table S5). In summary, the use of appropriate genetic populations and high-density genetic maps significantly improved the efficiency and precision of our genetic mapping.

4.2. Pyramiding Breeding Is an Essential Approach in the Hybrid Breeding System of Diploid Potatoes

At present, the potato (Solanum tuberosum L.) of international commerce is autotetraploid, and the complexity of this genetic system creates limitations for breeding. A diploid, inbred-hybrid breeding system offers many advantages to the current breeding system in potato: it takes less time to fix favorable alleles, marker-assisted backcrossing is possible, there is greater genetic variance for selection, and heterosis can be exploited systematically [15]. Pyramiding QTLs using marker-assisted selection (MAS) is effective for achieving the desirable phenotypic level of a quantitative trait in plant breeding programs [49,50,51]. QTL mapping analysis not only facilitates the mining of elite genetic resources but also guides pyramiding breeding. This is particularly relevant for environmentally sensitive traits in potato, such as yield, quality, and abiotic stress resistance. Additionally, few major effect genes associated with important agronomic traits have been cloned in potato, making direct pyramiding using major genes challenging. Integrating QTL mapping results with molecular marker-assisted selection (MAS) enables rapid and accurate selection of lines with target traits, thus rendering QTL pyramiding breeding of great value.

However, this technique remains largely unexploited in diploid potato breeding systems. Recently, Song and Endelman [52] successfully pyramided the Sli gene and the early-maturing allele CDF1.3 via backcrossing combined with molecular markers. In our study, a total of 89 genetic loci were mapped, 10 of which were repeatedly detected in two or more environments and defined as stable QTLs (Table 6). Further analysis of the genetic effects of these 10 stable QTLs revealed that favorable alleles could effectively increase phenotypic values, indicating that these QTLs have high heritability, are not easily affected by the environment, and represent valuable genetic resources for pyramiding breeding (Table 7). In particular, the pleiotropic loci on chromosome 5 (qTY-5-2 and qTS-5-1) and chromosome 12 (qTY-12-6, qTS-12-3, and qTE-12-4) can simultaneously control multiple traits. Additionally, the pleiotropic locus on chromosome 12 exhibits a high phenotypic variance explained (PVE) and is a major QTL. Their pyramiding has effectively improved our breeding efficiency.

Balancing the yield, quality and resistance to disease is a daunting challenge in crop breeding due to the negative relationship among these traits [53]. Notably, among the pleiotropic loci on chromosome 12 (qTY-12-6, qTS-12-3, and qTE-12-4) (Figure 5), the favorable allele derived from Y8 can increase the mean weight per tuber, the number of tubers per plant, and the tuber weight per plant, while simultaneously resulting in a rounder tuber shape and significantly fewer eyes (Table 7). This achieves the effect of simultaneously improving potato yield and quality, thus holding substantial research and application value. In conclusion, these QTLs provide excellent targets for subsequent gene cloning, functional studies, and breeding applications.

5. Conclusions

In this study, we constructed an F₂ population consisting of 174 lines using a genome-homozygous diploid inbred line and a heterozygous diploid material, and constructed a high-density genetic map containing 4464 bin markers by resequencing. A total of 8 traits related to yield, tuber shape, and eye number were investigated under 3 different environments. Using the high-density genetic map combined with phenotypic data for QTL mapping, a total of 89 QTLs were obtained, 7 of which were co-localized with previous studies; 10 were repeatedly detected in two or more environments and were defined as stable QTLs. One pleiotropic QTL was obtained on chromosomes 5 and 12, respectively. In particular, the QTL at the end of chromosome 12 (qTY-12-6, qTS-12-3, and qTE-12-4) had a high contribution rate and was a major QTL. In addition, genetic effect analysis showed that this QTL could effectively increase yield, make tuber shape rounder, and reduce the number of eyes, which has important research and application value.