Comprehensive Analysis of Full-Length Transcriptome Profiling, Genetic and Phenotypic Variation in Multiplier Onion (Allium cepa var. aggregatum) Accessions in China

Abstract

Multiplier onion (Allium cepa L. var. aggregatum) is an important bulbous vegetable widely utilized for culinary, condimental, and medicinal purposes. However, limited research on its genetic diversity and phenotypic variation has hindered the development and utilization of superior cultivars. In this study, we conducted full-length transcriptome profiling to obtain unique transcripts and develop large-scale simple sequence repeat (SSR) markers. Subsequently, we employed integrative analysis to characterize the genetic and phenotypic variation of 263 multiplier onion accessions in China. Full-length transcriptome sequencing utilizing PacBio technology generated 61,108 high-quality non-redundant transcripts with an average length of 1816 bp, from which we developed 4124 SSR markers encompassing 100 motif types. Population structure, principal component analysis, and neighbor-joining phylogenetic analysis classified the 263 multiplier onion accessions into two distinct subpopulations: Pop1, consisting of 236 accessions primarily from Heilongjiang Province, and Pop2, comprising 27 accessions mostly from Shaanxi Province. Phenotypic evaluation demonstrated significant variation in bulb traits, with single bulb weight (SBW) exhibiting the highest variability (0.75–29.94 g; CV = 70.10%), followed by total bulb weight per plant (BW) (5.00–168.83 g; CV = 58.34%), indicating considerable potential for breeding high-yield varieties. Correlation analysis indicated that the SBW and BW had significantly positive correlations with multiple traits, including bulb height, bulb transverse diameter, diameter of basal plate of bulb, diameter of bulb neck, and number of cloves per bulb. Our findings provide a valuable genetic and phenotypic resource for the conservation and utilization of multiplier onion germplasms.

1. Introduction

Multiplier onion (Allium cepa L. var. aggregatum, 2n = 2x = 16) is an important bulbous vegetable widely utilized for culinary, condimental, and medicinal purposes. It features clusters of bulblets that develop from lateral buds located in the axils of certain inner blade-bearing leaves on the basal plate. Typically, it produces between 3 and 20 small bulbs per plant and primarily propagates vegetatively through bulbs [1]. A limited number of varieties can be reproduced through botanical seeds [2]. Compared with common onion (Allium cepa L. var. cepa), multiplier onion is characterized by a short growth cycle, extended storage life, and enhanced resistance to pests, diseases, and environmental stressors [3,4]. Multiplier onion is extensively cultivated in Asia, Africa, Europe, and South America [5,6]. Current commercial-scale production occurs in countries such as China, Thailand, Ethiopia, the Philippines, Sri Lanka, Indonesia, and India. Notably, this crop is considered a specialty agricultural product in Northeast China, where yields can reach approximately 6000 kg per acre [6]. Due to its unique flavor and pronounced pungency, multiplier onion is often preferred over common onion and commands higher market prices [7].

Several studies have assessed the genetic and morphological characterization of multiplier onion in Europe and India. For example, the genetic characterization of 264 multiplier onion accessions from Europe including the Nordic countries, Baltic countries, Czech Republic, and Croatia suggested historical movement in North-Eastern Europe [8]. Rich diversity was found in cultivated Finnish multiplier onion [9]. Morphological and molecular characterization of 36 multiplier onion genotypes in India was investigated and found the genotype with the highest yield and the genetic diversity in the genotypes [10]. In China, multiplier onion is predominantly cultivated in Northeast China. The genetic diversity of 49 multiplier onion cultivars from northeast China was analyzed using seventeen ISSR primers and classified into six groups [11]. However, due to the lack of effective markers and the limitation of sample size, the genetic composition and the phenotypic variation in Chinese multiplier onion have not been well elucidated.

Previous studies of genetic diversity and variation in multiplier onion resources were evaluated using restriction fragment length polymorphism, random amplified polymorphic DNA, inter simple sequence repeat, intron length polymorphic markers, or microsatellite (SSR) markers developed in common onion [8,9,10,11,12]. With the rapid development of high-throughput sequencing technology, genome sequencing and large-scale marker development have become increasingly efficient and cost-effective. The genome of the common onion has been sequenced at the chromosomal level, with a genome size of 15.78 Gb [13]. Nevertheless, the multiplier onion currently lacks genomic and transcriptomic data. This absence of information significantly hinders research into its genetic variation and molecular biology. PacBio SMRT sequencing generates extremely long reads of up to 60 kb, facilitating accelerated genome and transcriptome sequencing [14,15]. This sequencing technology has been employed across various species to enable de novo genome assembly, identify genetic variations, and analyze transcriptome complexity [16,17,18]. By utilizing PacBio sequencing, researchers can efficiently obtain transcripts from the multiplier onion, thereby enabling the exploration of molecular markers and elucidating regulatory pathways associated with specific traits.

In this study, we utilized PacBio SMRT transcriptomics to obtain full-length transcripts and identify a substantial number of microsatellite markers. The genetic diversity and population structure of 263 multiplier onion accessions from China were analyzed using newly developed simple sequence repeat (SSR) markers. The phenotypic traits of bulbs of these accessions were measured. The purpose of this study is to elucidate the comprehensive transcriptomic profiling and development of SSR markers to investigate the genetic diversity of multiplier onion, thereby providing a crucial foundation for the conservation and utilization of multiplier onion germplasm resources.

2. Materials and Methods

2.1. Plant Materials and Phenotypic Measurement

A total of 263 multiplier onion accessions were subjected to genetic and phenotypic variation analysis. During the 2018–2020 period, the majority of these accessions (238) were collected from Heilongjiang Province, while the remaining 25 were collected from Shaanxi Province. These multiplier onion accessions have been conserved in the National Germplasm Repository for Perennial and Vegetatively Propagated Vegetables. Additionally, eight Allium chinese accessions were selected as an outgroup. These materials were planted in the experimental fields of the Vegetable Research Center located within the International Agricultural High-tech Industrial Park of the Chinese Academy of Agricultural Sciences (39°36′9.781″ N, 116°36′30.157″ E). For each accession, thirty uniformly sized bulbs were sown in the soil in late March 2023. The planting was configured with 10 cm between rows and 20 cm between columns. Following a four-month growth period, all materials were harvested in late July of the same year. The study employed a completely randomized block design, which included three replications. Field cultivation followed standard agricultural practices.

Phenotypic measurements of nine bulb characteristics were conducted on 263 multiplier onion accessions at the harvesting stage, using the descriptors defined by Zhan and Li [19]. Twelve randomly selected plants per accession were assessed for phenotypic traits. The total bulb weight per plant (BW) and the single bulb weight (SBW) were measured using an electronic scale. The diameter of the bulb’s basal plate (DBPB), the diameter of the bulb neck (DBN), the bulb height (BH), and the bulb transverse diameter (BTD) were measured with a vernier caliper. The spherical index was calculated as the ratio of BH to BTD. The number of bulbs at the base (NBB) and the number of cloves per bulb (NCB) were counted manually.

2.2. RNA Extraction and PacBio SMRT Sequencing

The elite multiplier onion accession Aca-148 was selected for PacBio SMRT sequencing (Pacific Biosciences of California, Menlo Park, CA, USA) to acquire full-length transcripts. The Aca-148 plants were propagated asexually through bulbs and grown under optimal conditions. Root, pseudostem, leaf, and bulb tissues were sampled during the bulb expansion stage. Following sampling, total RNA was extracted from each of these tissues using the Tiangen RNA preparation kit, adhering to the manufacturer’s instructions. The RNA integrity and concentration were assessed using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and an Agilent 2100 bioanalyzer (Agilent, Palo Alto, CA, USA). For the creation of cDNA libraries, equal quantities of RNA extracted from the four tissues of Aca-148 were subsequently combined. Full-length cDNA was synthesized utilizing the PacBio SMARTer PCR cDNA Synthesis Kit. After the PCR amplification process, the full-length cDNA underwent end repair. The SMRT dumbbell adapters were ligated to the cDNA with the aid of the SMRTbell template preparation kit. Full-length cDNA fragments were screened using BluePippin, and the cDNA libraries were constructed according to the Iso-Seq protocol. The final product of the reaction was sequenced using a PacBio Sequel II sequencer.

2.3. Pacific Biosciences Long Read Processing and Remove Redundant

The raw sequencing reads were processed into circular consensus sequencing (CCS) reads by employing adapter sequences, followed by a correction process to obtain quality information of sequences. Full-length non-chimeric (FLNC) reads were recognized by searching for the 5′ and 3′ adapter sequences along with the poly-A tail signals within the CCS. The full-length sequences derived from the same transcript were clustered, and the similar full-length sequences were clustered into a cluster. Further correction of consistent sequences resulted in the production of high-quality sequences suitable for subsequent analysis. The longest sequences were selected as the final transcripts. To eliminate redundancy, high-quality FL transcripts from Iso-Seq were filtered using cd-hit (identity > 0.99). The obtained transcript sequences/unigenes were then employed for additional analyses concerning isoforms, homologous genes, gene families, and SSR markers.

2.4. Gene Functional Annotation, CDS Predication, and SSR Identification

To predict gene functions, the unigenes obtained from PacBio SMRT sequencing were annotated via eight different databases, including NCBI NR (https://www.ncbi.nlm.nih.gov/refseq/about/nonredundantproteins/ accessed on 2 December 2024), Pfam (https://pfam.xfam.org/ accessed on 4 December 2024) [20], Swiss-Prot (https://www.uniprot.org/ accessed on 10 December 2024) [21], KEGG (https://www.genome.jp/kegg/ accessed on 10 December 2024) [22], GO (http://geneontology.org/ accessed on 10 December 2024) [23], KOG [24] and COG [25] (https://www.ncbi.nlm.nih.gov/research/cog/ accessed on 15 December 2024), and eggNOG (http://eggnog5.embl.de/ accessed on 15 December 2024) [26]. The ANGLE pipeline was utilized to predict the coding sequence (CDS) for each full-length transcript. In addition, the MISA tool, available at http://pgrc.ipk-gatersleben.de/misa/ accessed on 25 December 2024, was employed to detect potential SSRs [27]. Di-nucleotide motifs were required to be repeated at least six times, while tri-, tetra-, penta-, and hexa-nucleotide motifs needed to show repetition of a minimum of five times [28]. Moreover, Primer 3 software was applied for designing primers with lengths between 18 and 27 bp, targeting PCR product sizes ranging from 100 to 280 bp, and an annealing temperature established between 57 and 63 °C.

2.5. DNA Extraction and SSR-Seq Detection

Young leaves of 263 multiplier onion accessions and eight A. chinense accessions were sampled at the seedling stage with scissors, placed in 5 mL centrifuge tubes, and flash-frozen in liquid nitrogen. The samples were then promptly transported to the laboratory and preserved at −80 °C in a refrigerator for DNA extraction. Three individual plants per accession were sampled, with each sample consisting of a pool of 2–3 young leaves from a single plant. Genomic DNA was extracted via the cetyltrimethylammonium bromide method. The concentration and quality of the extracted DNA were assessed using a Nanodrop 2000 spectrophotometer and 1.0% agarose gel electrophoresis.

SSR loci were chosen based on specific requirements: the repeat unit had to consist of 3 to 10 repetitions, should not consist solely of GC or AT, and should not be in close proximity to other SSR loci. A total of 200 SSR primers were synthesized by Thermo Fisher Scientific. Six multiplier onion accessions (Aca-9, Aca-37, Aca-82, Aca-248 Aca-260, and Aca-274), which displayed notable phenotypic variation in bulbs (Table S1), were identified to evaluate the polymorphism of the SSR primers with the use of fluorescence-labeled TP-M13-SSR [24]. PCR was carried out in a 20 μL reaction mixture containing 100 ng template DNA, 1× PCR buffer, 30 μM MgCl₂, 100 μM dNTPs, 1.0 U Taq DNA polymerase (Takara Dalian, China), 8 pmol of each reverse primer and fluorescently labeled (FAM/HEX) M13 universal primer, and 2 pmol of M13-tailed forward primer. Amplification was conducted under the following conditions: initial denaturation at 94 °C for 5 min; 30 cycles of 94 °C for 30 s, 53 °C for 30 s, 72 °C for 30 s; then 8 cycles of 94 °C for 30 s, 57 °C for 30 s, 72 °C for 30 s; and a final extension at 72 °C for 10 min. The PCR products from each sample were separated by capillary electrophoresis on an ABI 3730xl DNA Analyzer (Applied Biosystems, Foster City, CA, USA). Data analysis was performed using GeneMarker software (v1.65) to determine alleles and assess primer polymorphism.

Subsequently, the polymorphic primers were employed to amplify the DNA from all accessions. The selected primers were combined into multiplex PCR primer panels, with each panel containing 20 primer pairs. Using the multiplex PCR technology, amplification was performed with DNA as the template. Utilizing sequencing-based optimization, the efficiency and specificity of each primer pair in the multiplex system were evaluated to guide the adjustment and refinement of the primer composition and concentrations in the multiplex PCR panel. The optimized multiplex PCR primer panels were used to amplify target fragments from sample genomic DNA. After quality control, the amplification products from all multiplex PCR primer panels for the same sample genomic DNA were pooled, ensuring that the amount of amplification product from each primer pair was approximately equivalent. Using primers containing index sequences, PCR amplification was performed to introduce specific barcode sequences compatible with the Illumina platform at the ends of the library. The reaction employed an 11-cycle PCR program to minimize amplification bias. The Index PCR amplification products from all samples are mixed in equal amounts, and the final FastTarget™ sequencing library is obtained through gel extraction. The fragment size distribution of the library is verified using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). After precise quantification of the library concentration, high-throughput sequencing is ultimately performed on the Illumina HiSeq platform (Illumina, San Diego, CA, USA) with a paired-end 2 × 150 bp sequencing mode to generate Fastq data [29]. Raw sequencing data were processed for quality assessment with FastQC (v0.11.9). The paired-end reads were then merged using the Flash software, and the successfully merged sequences were retained for further genotyping. These sequences were aligned against the reference sequences of primer-captured targets via BLAST+ (v2.12.0) to filter qualified reads and evaluate the enrichment efficiency of the target regions. Genotyping was ultimately conducted employing the SSRseq (v1.1) program, which utilizes SSR frequency information.

2.6. Analysis of Genetic Diversity and Population Structure

To assess genetic diversity, POPGENE v1.32 software [30] was utilized to compute various parameters, such as the observed number of alleles (Na), effective number of alleles (Ne), observed homozygosity (Ho), observed heterozygosity (He), expected homozygosity, expected heterozygosity, xpected heterozygosity, and Shannon’s information index (I). The analysis of population structure was conducted using STRUCTURE v2.3.4 software [31]. An admixture model incorporating correlated allele frequencies across populations was employed, with ten iterations performed for each population number (K), varying from two to eight. The length of the burn-in period and the count of MCMC replications after the burn-in phase were established at 10⁶. The optimal K value was determined using Structure Harvester software (v0.6.94) [32]. Genetic distances among accessions were calculated with Populations v1.2.31 software, and this genetic distance matrix was subsequently used to create a phylogenetic tree through the neighbor-joining method in MEGA v6.06 [33]. Principal coordinate analysis (PCA) was carried out using GenAlEx v6.5 [34], and a two-dimensional scatter plot was generated using the R package v1.0. 0.

2.7. Analysis of Phenotypic Data of Bulbs

Basic statistical and correlation analysis of the phenotypic traits was performed using SPSS v 22.0 software. Correlation analysis was employed to evaluate the associations between traits by calculating Pearson’s correlation coefficients (r) and determining their statistical significance. Correlation plotting was conducted using the PerformanceAnalytics package in R v4.3.1, while principal component analysis (PCA) was carried out with the prcomp function from the stats package in R4.3.1. Comparative analysis of nine agronomic bulb traits between the subpopulations of multiplier onion was performed using the t-test, and bar plots were generated with GraphPad Prism v 9.0 software.

3. Results

3.1. PacBio SMRT Sequencing and Functional Annotation

The multiplier onion accession Aca-148 was chosen for PacBio SMRT sequencing to obtain full-length transcripts. The integrity and concentration of RNA extracted from four tissues of Aca-148 were presented in Figure S1. The sequencing produced a total of 356,943 polished circular consensus sequences, yielding 268,081 full-length non-chimeric (FLNC) reads. Following the clustering of FLNC reads, 102,387 high-quality consistent sequences were obtained, resulting in the identification of 61,108 transcripts through elimination of redundancy among these consistent sequences. The average length of the unique transcripts was 1816 bp, with lengths ranging from 50 bp to 10,906 bp. For the annotation of unique transcripts, a sequence similarity search was conducted across eight public databases (

Table 1. Summary of the functional annotation of unique transcripts in multiplier onion.

Annotation Database	Isoform Number	Percentage (%)
Nr (non-redundant protein sequences)	44,771	73.27
GO (Gene Ontology)	27,677	45.29
COG (Clusters of Orthologous Groups)	16,759	27.43
KEGG (Kyoto Encyclopedia of Genes and Genomes)	21,591	35.33
KOG (euKaryotic Ortholog Groups)	29,164	47.73
eggNOG (evolutionary genealogy of genes: Non-supervised Orthologous Groups)	43,147	70.61
Pfam (protein family)	33,190	54.31
Swiss-Prot	31,863	52.14
All	45,423	74.33

and Table S2, Figure 1). The results indicated that 44,771 transcripts (73.27%) had significant matches in the NR database, 27,677 (45.29%) in the GO database, 16,759 (27.43%) in the COG database, 21,591 (35.33%) in the KEGG database, 29,164 (47.73%) in the KOG database, 43,147 (70.61%) in the eggNOG database, 33,190 (54.31%) in the Pfam database, and 31,863 (52.14%) in the Swiss-Prot database. Overall, 45,423 transcripts (74.33%) were annotated in at least one of these databases, with 7556 transcripts (12.36%) being annotated across all eight databases (Table 1 and Table S2, Figure 1). Among the NR database matches, 53.93% of the transcripts were matched in Asparagus officinalis, 7.09% in Elaeis guineensis, and 5.44% in Phoenix dactylifera (Figure S2A). For GO analysis, 27,677 transcripts were classified into three categories: biological process (primarily annotated as “cell”, “cell part”, “membrane”, “organelle”), molecular function (primarily annotated as “catalytic activity”, “binding”), and cellular component (primarily annotated as “metabolic process”, “cellular process”, “single-organism process”) (Figure S2B). Additionally, 29,164 transcripts were assigned to KOG classification and divided into 25 specific categories. Notably, the category of “general functional prediction only” associated with basic physiological and metabolic functions was the largest group (Figure S2C).

3.2. Frequency and Distribution of Simple Sequence Repeat Markers

All unique transcripts were utilized for the subsequent development of SSRs. Out of 61,108 transcripts, 3784 were identified as containing 4124 SSR loci (

Table 2. Frequencies of different SSR motif types in multiplier onion.

Repeat Number	Di-	Tri-	Tetra-	Penta-	Hexa-	Total	Percentage (%)
5	0	1276	77	10	36	1399	33.92
6	652	410	16	3	11	1092	26.48
7	295	168	10	0	1	474	11.49
8	164	69	11	0	3	247	5.99
9	99	24	2	0	0	125	3.03
10	53	30	0	0	0	83	2.01
11	45	6	4	0	0	55	1.33
12	31	13	0	0	0	44	1.07
13	40	14	2	0	1	57	1.38
14	27	9	6	0	0	42	1.02
15	37	9	0	0	0	46	1.12
16	33	3	1	0	0	37	0.9
≥17	413	9	1	0	0	423	10.26
Total	1889	2040	130	13	52	4124
Percentage (%)	45.8	49.47	3.15	0.32	1.26

Note: These SSR markers were developed from the 61,108 transcripts obtained via full-length transcriptome profiling of the elite multiplier onion accession Aca-148.

and Table S3). Among these developed SSRs, 293 (7.01%) contained two or more SSRs, while 1033 (25.05%) comprised compound SSRs (Table 2 and Table S3). Among these 4124 SSRs, tri-nucleotide motifs were the most abundant, accounting for 2040 (49.67%), followed by di-nucleotide motifs with 1,889 (45.81%), tetra-nucleotide motifs with 130 (3.15%), hexa-nucleotide motifs with 52 (1.26%), and penta-nucleotide motifs with 13 (0.32%) (Table 2 and Table S5). A total of 100 SSR motif types were identified. Among these, di-, tri-, tetra-, penta-, and hexa-nucleotide repeats comprised 8, 30, 25, 9, and 28 types, respectively (Table S5). The most prevalent type was TA/TA (483, 11.71%), followed by AT/AT (463, 11.23%), CA/TG (335, 8.12%), AC/GT (302, 7.32%), and GAA/TTC (201, 4.87%) (Table S5).

3.3. Genetic Diversity, Population Structure, PCA, and NJ Phylogenetic Analysis

A total of 200 SSR markers containing di- and tri-nucleotide repeats were selected randomly, and primers were designed according to their flanking sequences. Out of these 200 SSR primers, 178 pairs were successfully amplified. Ultimately, 35 pairs exhibiting stable amplification and polymorphic alleles were chosen for the analysis of genetic diversity and population structure. A summary of the genetic characteristics of these 35 markers was presented in Table S6. A total of 126 alleles were amplified, resulting in a mean of 3.6 observed alleles per locus. The expected heterozygosity (H_E) ranged from 0.040 to 0.597, with an average of 0.334. The Shannon-Wiener index varied from 0.118 to 1.049, with an average of 0.541. Notably, the marker Aca-SSR2964 exhibited the highest genetic diversity, with a Shannon-Wiener index of 1.049.

Population structure analysis was conducted using the STRUCTURE software v2.3.4, with the optimal value of K determined to be three. The eight A. chinense accessions were categorized into a separate group, serving as an outgroup. The 263 A. cepa var. aggregatum accessions were divided into two subpopulations, Pop1 and Pop2 (Figure 2A, Table S7). Pop1 comprised 236 A. cepa var. aggregatum accessions, with the vast majority (233) from Heilongjiang province, while the remaining three were from Shaanxi province. In contrast, Pop2 included 27 A. cepa var. aggregatum accessions, with most (22) from Shaanxi province and five accessions from Heilongjiang province (Table S7).

The phylogenetic tree and PCA results were further well supported by the population structure analysis (Figure 2). In the phylogenetic tree, the Pop1 and Pop2 accessions of A. cepa var. aggregatum were clearly segregated into two separate groups, whereas A. chinense accessions formed a distinct cluster (Figure 2B). In the PCA, the first two principal coordinates accounted for 51.28% and 28.62% of the total variance, respectively, together explaining 79.90% of the overall variation. The distribution of accessions in the PCA plot further confirmed the genetic separation of A. cepa var. aggregatum, with Pop1 and Pop2 forming two distinct clusters. In addition, A. chinense accessions were clearly separated from A. cepa var. aggregatum accessions, reinforcing the genetic distinctions identified in the population structure analysis (Figure 2C).

3.4. Phenotypic Variation and Correlation Analysis for Bulb Traits

Nine bulb traits of 263 A. cepa var. aggregatum accessions were detected, with detailed descriptive statistics presented in

Table 3. Statistical analysis of nine traits of bulbs of 263 multiplier onion accessions.

Traits	Range (Min-Max)	Mean ± SD	Coefficient of Variation (%)
Total bulb weight per plant (BW, g)	5.00–168.83	33.37 ± 19.47	58.34
Single bulb weight (SBW, g)	0.75–29.94	6.53 ± 4.58	70.10
Number of bulblets per plant (NBB)	1.00–14.33	5.56 ± 1.79	32.18
Diameter of basal plate of bulb (DBPB, cm)	0.24–1.47	0.94 ± 0.21	22.53
Diameter of bulb neck (DBN, cm)	0.22–1.33	0.74 ± 0.19	26.42
Bulb height (BH, cm)	1.08–5.06	3.19 ± 0.59	18.35
Bulb transverse diameter (BTD, cm)	1.20–3.89	2.26 ± 0.47	20.60
Spherical index (SI)	0.56–2.58	1.43 ± 0.26	17.99
Number of cloves per bulb (NCB)	1.00–5.17	2.29 ± 0.46	20.12

. The single bulb weight (SBW) exhibited the highest variation, ranging from 0.75 to 29.94 g (coefficient of variation, CV: 70.10%), followed by the total bulb weight per plant (BW), which varied from 5.00 to 168.83 g (CV: 58.34%). The number of bulblets at the base (NBB) varied from 2.00 to 14.33, with an average of 5.56 bulblets per plant and a CV of 32.18%. In contrast, the spherical index (SI) and bulb height (BH) exhibited the lowest phenotypic variation, with CVs of 17.99% and 18.35%, respectively. Correlation analysis indicated that the SBW and BW had significantly positive correlations with the BH, bulb transverse diameter (BTD), diameter of basal plate of bulb (DBPB), diameter of bulb neck (DBN), and number of cloves per bulb (NCB). The NBB demonstrated significant negative correlations with the SBW, whereas positive correlations with the BW (Figure 3).

Two groups of variables were identified as principal component 1 (PC1) and principal component 2 (PC2), which accounted for 40.10% and 15.71% of the total phenotypic variation, respectively (Figure 4). With the exception of the NBB and SI, all other traits were positively weighted for PC1 (Figure 4). The principal component loadings matrix indicated that the variables BTD, SBW, BH, and BW had high contributions to PC1, with high loadings (0.482, 0.472, 0.420, and 0.391, respectively), indicating that PC1 was primarily defined by these variables. The NBB, SI, BH, BW, NCB, and SBW were positively weighted for PC2, whereas BTD, DBN, and DBPB exhibited negative relationships with PC2 (Figure 4). The principal component loadings matrix indicated that SI, NBB, BW, and BH had high contributions to PC2, with high loadings (0.632, 0.474, 0.380, and 0.377, respectively), indicating that PC2 was mainly defined by SI, NBB, BW, and BH.

Population structure, principal component analysis, and neighbor-joining phylogenetic analysis classified the 263 multiplier onion accessions into two distinct subpopulations. The comparative analysis of nine agronomic bulb traits between these two subpopulations revealed significant differences in seven traits. Among these, BW, SBW, BH, and BTD exhibited the most pronounced differences between Pop1 and Pop2 (p < 0.0001), followed by DBN (p < 0.001), SI (p < 0.01), and DBPB (p < 0.05) (Figure 5). In contrast, no significant differences were observed for NBB and NCB (Figure 5).

4. Discussion

Due to its diverse applications in culinary and medicinal purposes, the multiplier onion is highly valued for both its economic and health benefits. However, the advancement of molecular biology research and precision breeding programs for this crop has been historically hampered by a scarcity of comprehensive genomic and transcriptomic resources. In this study, we employed PacBio SMRT sequencing to generate a high-quality, full-length transcriptome. This approach yielded 61,108 full-length transcripts, of which 45,423 (74.33%) were functionally annotated against eight major public databases (Table 1 and Table S2, Figure 1). This full-length transcriptome dataset offered a valuable resource for elucidating gene function, characterizing genetic variation, and exploring complex biological pathways in multiplier onion.

Leveraging this rich transcriptomic data, we developed a suite of 4124 SSR markers. SSRs are markers widely valued in molecular genetics for their co-dominance, high polymorphism, good stability, high accuracy, straightforward operation, and excellent repeatability and interspecific transmission [35,36]. The development of these markers represents a significant toolset for the scientific community, as they will facilitate research on genetic diversity and population structure, molecular-assisted selection in breeding programs, genetic map construction, variety identification, and purity assessment for multiplier onion. This foundational work thus establishes critical resources to accelerate future genetic research and breeding innovation in multiplier onion.

Currently, some studies have assessed the genetic diversity of multiplier onion [8,9,10,11,12]. However, a comprehensive large-scale understanding of its population structure across major growing regions in China remains limited. To address this gap, our study conducted a detailed genetic analysis of 263 multiplier onion accessions collected in China. These results revealed a clear division into two distinct subpopulations, Pop1 and Pop2, which exhibited strong correlations with their geographic origins. Specifically, the majority of accessions in Pop1 were traced to Heilongjiang Province in Northeast China, whereas Pop2 consisted predominantly of accessions from Shaanxi Province in Northwest China (Table S7). This pronounced genetic clustering provides strong evidence for a significant correlation between genetic composition and geographic distribution, a pattern likely driven by long-term geographic isolation and localized domestication processes [36,37,38]. Furthermore, the genetic analysis identified a small number of accessions that did not cluster with their expected geographic groups. These outliers suggested potential human-mediated exchange and introduction of germplasm between regions. Although multiplier onion accessions are typically cultivated stably in local areas for decades, our findings indicate that some accessions have been transferred to new cultivation zones. Similar events are observed in crops like Finnish multiplier onions [9], demonstrating how human activities (e.g., domestication) interact with natural selection to shape genetic diversity.

Among the two subpopulations identified by genetic analysis, significant divergences were observed specifically in bulb characteristics. A comparative analysis revealed statistically significant differences in bulb weight (BW and SBW) and size (BH and BTD) between Pop1 and Pop2. In contrast, traits related to the reproductive propagation coefficient, namely the number of bulblets, showed no significant inter-subpopulation variation. This distinct phenotypic pattern suggests that while the fundamental reproductive architecture remains conserved, the primary divergence lies in traits directly influencing bulb development and enlargement. Consequently, these morphological disparities indicate that multiplier onions cultivated in different geographical regions exhibit significant differences in potential bulb yield. This correlation between genetic subpopulation and yield-related traits underscores the profound impact of regional growing conditions—likely shaped by local environmental factors and selection pressures—on the expression of key agronomic characteristics.

Phenotypic variation serves as a fundamental prerequisite for crop genetic improvement and breeding programs. In this study, an evaluation of important bulb traits in 263 multiplier onion accessions revealed extensive phenotypic diversity, particularly in BW and SBW. The remarkably high coefficients of variation (CV) for these traits—70.10% for BW and 58.34% for SBW—highlight a broad genetic base and significant potential for selective breeding. As the primary edible and economically valuable organ of the multiplier onion, the bulb plays a crucial role in determining overall yield. The substantial variation observed in BW and SBW indicates considerable potential for breeding high-yield varieties.

Further correlation analysis indicated that SBW and BW had significantly positive correlations with BH, BTD, DBPB, DBN, and NCB. This complex network of correlations suggests that bulb weight is not governed by a single factor but is rather an integrated trait co-modulated by multiple component characteristics. Therefore, future research should prioritize elucidating the genetic architecture underlying bulb weight and its related traits. Unraveling the quantitative trait loci, or key genes, controlling these traits will be crucial, as such knowledge will enable the application of molecular marker-assisted selection and other advanced breeding strategies, ultimately accelerating the development of new multiplier onion varieties with superior yield and optimized bulb morphology.

5. Conclusions

This study performed an integrative analysis of the genetic diversity and phenotypic variation in multiplier onion. A total of 61,108 unique transcripts were obtained through redundancy elimination via PacBio sequencing. Additionally, 4124 SSR markers were developed through full-length transcriptome data. Population structure, PCA, and NJ phylogenetic analysis revealed that 263 multiplier onion accessions could be categorized into two distinct subpopulations. Pop1 comprised 236 accessions, with the vast majority from Heilongjiang province, while Pop2 included 27 accessions, with most from Shaanxi province. This suggests distinct regional characteristics in the genetic composition of the multiple onions. Phenotypic measurement indicated significant variations in bulb weight among these accessions. Single bulb weight exhibited the highest variation, followed by the total bulb weight per plant, indicating considerable potential for breeding high-yield varieties. These findings not only provide valuable molecular markers and insights into genetic diversity for resource conservation and utilization but also offer a crucial foundation for the conservation and utilization of multiplier onion germplasm resources.