Genetic Diversity Analysis of American Ginseng (Panax quinquefolius L.) Accessions Based on Phenotypic Traits and SSR Markers

Abstract

American ginseng (Panax quinquefolius L.) is an important medicinal crop, but its improvement in China is limited by variety degeneration and a shortage of elite cultivars. In this study, phenotypic traits and simple sequence repeat (SSR) markers were integrated to evaluate the genetic diversity of 51 selected accessions from major Chinese production regions. Phenotypic analysis showed that five of the 18 quantitative traits had phenotypic coefficients of variation exceeding 40%, mainly involving root traits such as fresh root weight and lateral root number. Broad-sense heritability for these root traits ranged from 61.70% to 74.80%, indicating substantial genetic contribution under standardized conditions. Principal component analysis identified five candidate elite accessions: CY3 and KD1 for tall stature and high yield, DH1 and LH2 for high ginsenoside content, and AT1 for well-developed lateral roots. A 12-accession representative subset was further proposed for conservation and pre-breeding. SSR-based clustering showed weak geographic differentiation, and Mantel analysis revealed no significant correlation between phenotypic and SSR-based genetic distances. These materials, together with the proposed accession-level conservation strategy, provide useful resources for germplasm preservation, parental selection, QTL mapping, and marker-assisted breeding.

1. Introduction

American ginseng (Panax quinquefolius L.), a perennial herbaceous medicinal plant of the family Araliaceae native to North America, is globally renowned for its potent bioactive properties and nutritional benefits. It produces diverse secondary metabolites, primarily tetracyclic triterpenoid saponins (ginsenosides), along with polysaccharides, polyacetylenes, amino acids, and volatile oils. These phytochemicals contribute to its pharmacological effects, including immunomodulation, anti-fatigue effects, anti-aging properties, and metabolic regulatory activities [1]. In Traditional Chinese Medicine (TCM), American ginseng is classified as a “Yin-nourishing” and “Qi-tonifying” herb, contrasting with the “heating” nature of Panax ginseng, and is used to alleviate fatigue and restore physiological balance [2].

Driven by these attributes, P. quinquefolius has evolved into a high-value economic commodity with expanding applications in functional foods, nutraceuticals, and cosmetics [3]. Global cultivation is concentrated in China, Canada, and the United States. Although China is the largest producer and consumer, with an annual output of approximately 3667 tonnes, domestic supply remains insufficient, requiring substantial imports from Canada and the United States [4,5]. However, expanding global production to close this market deficit is becoming increasingly difficult. On the supply side, wild P. quinquefolius populations are depleted and protected under the Convention on International Trade in Endangered Species (CITES), shifting reliance entirely to cultivated sources [6]. Unfortunately, cultivation is severely constrained by biological limitations. The species requires strict geo-environmental conditions (cool, shaded climates, and organic-rich soil) and has a lengthy growth cycle (3–4 years). Furthermore, it suffers from severe continuous cropping obstacles (replant disease), which render land unsuitable for replanting for decades, effectively making suitable arable land a non-renewable resource in the short term [7]. This mounting environmental and spatial pressure on cultivation highlights an urgent need for robust, resilient, and high-yielding cultivars to maximize land utilization efficiency.

Yet, germplasm improvement has lagged behind the expansion of cultivation scale. Since its successful introduction from North America in the 1980s, major cultivation bases have been established in provinces such as Jilin, Shandong, and Liaoning [8]. While Canada has developed varieties based on ginsenoside and sucrose levels, these are often not fully aligned with China’s distinct ecological environments and domestic market preferences for specific root morphologies. Consequently, China’s industry relies heavily on landraces with unclear genetic backgrounds [9]. To date, officially registered cultivars remain scarce in China, with Jilin Province being a primary contributor of varieties based on phenotypic characteristics, notably ‘Zhongnongyangshen No. 1’ (ZNYS-1), ‘Jinongyangshen No. 1’ (JNYS-1), and ‘Zhongnongyangshen No. 2’ (ZNYS-2) [10]. However, long-term unregulated introduction and the lack of standardized accessions management have led to serious issues, including germplasm homogenization, genetic degradation, and unstable root quality [11]. These issues not only reduce the economic value of the crop but also escalate the risk of genetic diversity loss, which is a core limiting factor in breeding high-yielding and stress-resistant varieties [12].

Given these challenges, characterizing genetic diversity in P. quinquefolius is fundamental to deciphering the hereditary architecture of complex agronomic traits, thereby enabling the strategic selection of superior accessions. Traditional diversity assessments for this species rely primarily on morphological traits, which are intuitive and cost-effective [13]. For instance, Abaya et al. [14] assessed phenotypic traits of P. quinquefolius in Ontario, Canada. However, morphological traits are strongly influenced by environmental plasticity, often resulting in limited discriminatory power among closely related accessions. To address these limitations, molecular markers have been widely adopted. Among them, Simple Sequence Repeat (SSR) markers are favored for their codominance, high reproducibility, and genome-wide coverage [15,16,17]. While SSRs have been utilized for species identification in the Panax genus [18,19], comprehensive genome-wide SSR studies specifically focused on the genetic diversity of P. quinquefolius accessions in China remain scarce.

Against this backdrop, this study seeks to bridge the gap between phenotypic variability and genomic diversity to support the genetic improvement of American ginseng. By comprehensively evaluating 51 selected P. quinquefolius accessions from all major Chinese production regions based on 28 integrated agronomic traits and novel genome-wide SSR markers, we identified superior breeding resources and clarified the genetic relationships among accessions. This integrated phenotypic–genomic analysis provides the first systematic profile of selected American ginseng accessions in China, offering a robust framework for germplasm identification and sustainable breeding.

2. Materials and Methods

2.1. Plant Material and Field Experiment

A total of 51 selected P. quinquefolius accessions (Table S1), comprising three registered cultivars (‘ZNYS-1’, ‘ZNYS-2’, ‘JNYS-1’) and 48 breeding lines from long-term cultivation sites, were evaluated. The 51 accessions were selected from an original collection of 153 P. quinquefolius accessions using a stratified sampling strategy that considered geographic origin, germplasm type, and phenotypic variation. Specifically, the selected panel was designed to cover all major Chinese production regions, represent the principal resource types maintained in the collection, and include accessions with distinct or extreme performance for major agronomic traits, including plant architecture, phenological traits, root morphology, fresh root weight, and ginsenoside content. This sampling strategy ensured broad representation of the available germplasm resources while retaining materials with potential breeding value. For the subsequent field experiments, all 51 selected accessions were cultivated at the Germplasm Resource Garden of Jilin Agricultural University (Changchun, China; 43°48′ N, 125°25′ E, 220 m a.s.l.) from 2021 to 2024, and all phenotypic evaluations were uniformly conducted on 4-year-old plants during the 2024 growing season. The genetic experimental site features a temperate continental monsoon climate with flat terrain. To minimize soil heterogeneity and ensure uniform seedling establishment, the cultivation substrate was standardized as a 1:1 (v/v) mixture of humus and peat soil. The experiment was arranged in a randomized complete block design (RCBD) with three replications [20,21]. Each plot consisted of five 1.5 m long rows. A standardized planting geometry of 10 cm × 20 cm (intra-row × inter-row) was maintained, ensuring a population of at least 50 plants per accession per plot. Agronomic management was strictly performed in accordance with the standard technical regulations of Jilin Province (DB22/T 1066-2018). Specifically, plants were grown under polyethylene shade nets allowing approximately 20–35% light transmittance. Manual weeding was performed monthly to prevent herbicide damage, and supplementary irrigation was applied during drought periods to maintain soil moisture at 40–50% of field capacity. The collection and experimental research of all Panax quinquefolius accessions were conducted in compliance with relevant institutional and national agricultural guidelines.

2.2. Accession Phenotyping

Twenty-eight agronomic traits were evaluated based on standard morphological descriptors for P. quinquefolius (

Table 1. Detailed descriptions, coding systems, observation stage and measurement criteria for the 28 agronomic traits evaluated in P. quinquefolius accessions.

No.	Trait Code	Trait Name	Observation Stage	Recording Criteria
1	IT	Inflorescence type	Initial flowering stage	Simple = 1, compound = 2
2	BB	Bud bracts	Initial flowering stage	Absence = 1, presence = 2
3	FC	Fruit color	Fruit ripening stage	Yellow = 1, red = 2
4	ES	Ear shape	Fruit ripening stage	Umbrella type = 1, spherical = 2
5	PGT	Petiole growth type	Fruit ripening stage	Erect = 1, semi-erect = 2, horizontal = 3
6	LESD	Leaf edge sawtooth degree	Fruit ripening stage	Weak = 1, medium = 2, strong = 3
7	LC	Leaf color	Fruit ripening stage	Light green = 1, green = 2, dark green = 3
8	LS	Leaf shape	Fruit ripening stage	Elliptical = 1, obovate = 2
9	SC	Stem color	Fruit ripening stage	Green = 1, purple with green = 2, purple = 3
10	ST	Stem type	Fruit ripening stage	Single stem = 1, double stem = 2
11	IFP	Initial flowering period	Initial flowering stage	The number of days from the emergence stage to the time when 1–3 whorls of flower buds in the outer circle of the inflorescence of 25% of the plants in the plot start to bloom
12	FRP	Fruit ripening period	Fruit ripening stage	The number of days from the emergence date to the time when the fruit color of 90% of the plants in the plot turns into the mature color
13	SH	Stem height	Fruit ripening stage	The height from the base of the stem to the attachment point of the petiole, measured with steel ruler
14	SD	Stem diameter	Fruit ripening stage	The diameter of the stem base, measured with digital vernier calipers
15	PL	Petiole length	Fruit ripening stage	The length from the base of the petiole to the attachment point of the leaf blade, measured with a steel ruler
16	LL	Leaf length	Fruit ripening stage	The length of the middle leaflet of the compound leaf, measured with a steel ruler
17	LW	Leaf width	Fruit ripening stage	The width of the middle leaflet of the compound leaf, measured with a steel ruler
18	LLWR	Leaf length to width ratio	Fruit ripening stage	The ratio of leaf length to leaf width
19	PL′	Peduncle length	Fruit ripening stage	The length from the base of the pedicel to the attachment point of the inflorescence, measured with a steel ruler
20	TSW	Thousand seed weight	Fruit ripening stage	The weight of 1000 seeds
21	RL	Rhizome length	Plant senescence stage	The length from the base of the rhizome to the junction with the stem, measured with digital vernier calipers
22	RT	Rhizome thickness	Plant senescence stage	The thickness at the widest part of the rhizome, measured with digital vernier calipers
23	TL	Taproot length	Plant senescence stage	The length from the base of the rhizome to the forking point of the main root, measured with a steel ruler
24	TD	Taproot diameter	Plant senescence stage	The diameter at the widest part of the main root, measured with digital vernier calipers
25	NLR	Number of lateral roots	Plant senescence stage	The number of lateral roots
26	NFR	Number of fibrous roots	Plant senescence stage	The number of fibrous roots
27	FRW	Fresh root weight	Plant senescence stage	The weight of fresh roots
28	GC	Ginsenoside content	Plant senescence stage	The sum of the contents of ginsenosides Rg1, Re, and Rb1

Note: The traits are categorized into two groups: Qualitative traits (No. 1–10), which are scored based on distinct morphological categories, and Quantitative traits (No. 11–28), which are measured biological parameters.

). To ensure statistical representativeness, 20 healthy individuals were randomly selected per plot (60 plants per accession). Sampling was initiated at the seedling emergence stage (early May, for 4-year-old plants), where plants in the central rows were indexed and selected using a random number generator, strictly excluding marginal rows to avoid edge effects. These plants were tagged for continuous monitoring. Agronomic trait measurements were anchored to distinct developmental stages: inflorescence-related traits (e.g., IT and IFP) were recorded at the initial flowering stage (mid-June, approximately 40–55 days after emergence); vegetative traits (e.g., SH and LL) were measured during the fruit ripening stage (mid-August, approximately 100–115 days after emergence) when fruit turned fully red; root-related traits (e.g., NLR and TL) and ginsenoside content were quantified at the plant senescence stage (early October, approximately 150 days after emergence), characterized by aerial part withering and maximum root biomass accumulation (Table 1).

Measurement protocols were trait-specific: linear dimensions (e.g., SH, LL, and PL) were measured with a steel ruler (precision: 1 mm), diameters (e.g., SD, RT, and TD) were measured with digital vernier calipers (precision: 0.01 mm), and biomass traits (TSW and FRW) were weighed with an electronic analytical balance (Mettler Toledo, Zurich, Switzerland, precision: 0.01 g). Detailed definitions, including specific measurement loci for quantitative traits and scoring criteria for qualitative traits, are presented in Table 1. Root ginsenoside extraction and quantification followed a previously described protocol [22]. Briefly, dried root samples were ground to a fine powder, homogenized, extracted via ultrasonication in methanol, and analyzed using high-performance liquid chromatography (HPLC, Waters Corporation, Milford, MA, USA) to quantify ginsenosides Rg1, Re, and Rb1.

2.3. Development and Screening of SSR Markers

SSR loci were identified from the published P. quinquefolius genome [23] using Krait v1.5.1 software [24]. Search parameters targeted mono- to hexanucleotide motifs with minimum repeat thresholds of 10, 6, 5, 5, 5, and 5, respectively, and a maximum interval of 100 bp between adjacent loci. Primers were designed using specific constraints: a length of 18–27 bp (optimum 20 bp), a melting temperature (T_m) of 59–61 °C, a GC content of 40–60%, and an expected polymerase chain reaction (PCR) product size of 150–500 bp. Primers were synthesized by Sangon Biotech Co., Ltd., Shanghai, China.

To screen for effective markers, a random subset of 100 primer pairs was tested on five phenotypically diverse selected accessions. While 76 pairs (76%) successfully amplified genomic DNA, only 35 produced clear amplicons, and 12 exhibited polymorphism across the accessions (

Table 2. Primer sequences and amplification characteristics of 12 polymorphic SSR markers developed for P. quinquefolius.

Primers	Repetitive Motif	Primer Sequence (5′–3′)	Product Size (bp)	Annealing Temperature (°C)
Pq29	(GTA)11(CAG)5	F: GGGCAAATCAAATAAAGGAG R: GAGTGATTGCAGAGCAGGGT	153	62
Pq164	(AT)12	F: ACCTAGAGATGGATCCCGGT R: GATGCATGTGCGGCTAGATA	180	62
Pq298	(CGA)6	F: TGCGCAGATACCTACTCGTG R: TCCACGGTCTCAAATCCTTC	206	60
Pq303	(GCT)9	F: AAAGCAGGGTGTGCTCACTT R: AGGGAGACCGGAGCATTATT	198	55
Pq346	(ATC)8	F: TAGGCACCCTCGACTTCTGT R: TATTTCCGCTGTGCGTTTCT	200	60
Pq455	(AT)12	F: AAGGGGCACTGTGTAAGAAAA R: CCGTAGGGGAGAAAATGAAT	226	62
Pq460	(ATT)15	F: CGTTGCTTCCCTTGCCTTTGAT R: GCCGTGCGCCATCTTTTATTT	305	62
Pq584	(GAT)6	F: ACCTCTGGCTGGGCTGGTCAAT R: TCGGGGAGGAGTCGAGCTAGGG	224	62
Pq601	(TC)7	F: GGGTACCATCGAGATGTGCT R: GATCTCCTGTCTGCCCTACG	195	62
Pq807	(TG)8	F: CCGGGGATTGAAGCAGGAAT R: CACCCGGCCCACACATATAT	261	56
Pq827	(TA)11	F: TTATCCAGGTGGAGGTATGG R: CCACTGATGATCATTTCCTT	208	56
Pq1038	(AGC)7	F: AAAGGAGATAGTAGATGTTATGTGGTT R: CAAACCAGGCATCCCTAATC	186	62

). These 12 polymorphic markers were selected for the subsequent genetic diversity analysis. This functional polymorphism rate (12%) aligns with prior reports on polyploid plant systems [25,26], reflecting the inherent genomic complexity of the species.

2.4. DNA Extraction, PCR Amplification and Polyacrylamide Gel Electrophoresis

Genomic DNA was extracted from fresh, young leaves of 4-year-old plants collected during the vigorous vegetative growth stage (late May, approximately 20–30 days after emergence). For each accession, young leaves were collected from three healthy plants, with one plant sampled from each of the three replicate plots, and pooled in equal amounts to form a single accession-level composite sample. This pooling strategy was adopted to capture the genetic profile of each accession while minimizing the influence of variation among individual plants within the accession. All 51 accessions were thus represented by 51 DNA samples for SSR analysis. Samples were processed using the Hi-DNAsecure Plant Kit (Tiangen Biotech Co., Ltd., Beijing, China), and DNA concentration and integrity were assessed using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and 1% agarose gel electrophoresis. PCR amplification was performed in 20 µL reactions containing 1 µL of genomic DNA (standardized to 50 ng/µL), 1 µL of each forward and reverse primer (10 µM), 10 µL of 2× PCR Master Mix (Beyotime Biotechnology Co., Ltd., Shanghai, China), and 7 µL of ddH₂O. The thermal cycling profile was: initial denaturation at 94 °C for 5 min; 35 cycles of 94 °C for 30 s, primer-specific annealing temperatures (Table 2) for 45 s, and 72 °C for 1 min; followed by a final extension at 72 °C for 10 min. Amplicons were separated on a 6% polyacrylamide gel (180 V, 60 min), visualized via silver nitrate staining, and digitally documented.

2.5. Statistical Data Analysis

All phenotypic data processing and statistical analyses were performed using R software version 4.4.2 (R Foundation for Statistical Computing, Vienna, Austria). Descriptive statistics, including mean, range, and standard deviation, were computed using the psych package. The Shannon–Weaver diversity index (H′) for phenotypic traits was calculated according to the formula:

$$H^{\prime } = -{\displaystyle {\sum }_{i=1}^S}Pi\ln Pi,$$

where Pi represents the proportion of accessions belonging to category i within the total population, and S denotes the total number of phenotypic categories [27]. To evaluate the genetic potential of the accessions, analysis of variance (ANOVA) was performed based on the RCBD. Variance components were estimated from the expected mean squares following the method of Singh and Chaudhary [28]. Genotypic variance (V_g), phenotypic variance (V_p), and environmental variance (V_e) were derived as follows:

$$V_g = ({MS}_g-{MS}_e)/r, V_e = {MS}_e, \mathrm{and} V_p ={ V}_g + V_e,$$

where ${MS}_g$ is the mean square of genotypes, ${MS}_e$ is the mean square of error, and r is the number of replications. Coefficients of variation were calculated to assess the magnitude of variability:

$$(1) GCV = \left(\sqrt{V_g}/\overline{X}\right) \times 100; (2) PCV = (\sqrt{V_p}/\overline{X}) \times 100; \left(3\right) ECV = \left(\sqrt{V_e}/\overline{X}\right)\times 100,$$

where $\overline{X}$ is the grand mean of the trait [28]. Broad-sense heritability (H²) was computed as $H^2 = \left(V_g/V_p\right) \times 100$. Mean separation for quantitative traits was performed using Fisher’s least significant difference (LSD) test at p < 0.05 using the agricolae package. Pairwise Pearson correlation coefficients were calculated and visualized using the corrplot package. Prior to multivariate analysis, phenotypic data were standardized (Z-score normalization) to eliminate dimensional differences. Hierarchical clustering was performed based on Euclidean distance matrices using the unweighted pair group method with arithmetic mean (UPGMA) algorithm. Principal component analysis (PCA) was conducted using the FactoMineR package to reduce dimensionality. Mathematically, PCA transforms the original correlated variables into a new set of uncorrelated variables (principal components) through linear combinations:

$$Y_k={\displaystyle {\sum }_{j=1}^p}{w}_{kj}{X}_j,$$

where Y_k is the k-th principal component, X_j represents the original variables, and w_kj are the connection weights (loadings) [29]. To quantify the overall agronomic performance of each accession, a composite score (F) was calculated based on the weighted contribution of the principal components using the following equation [30]:

$$F={\sum }_{i=1}^m\left({\lambda }_i \times F_i\right)/{\sum }_{i=1}^m{\lambda }_i,$$

where F represents the comprehensive evaluation score, m is the number of extracted principal components (eigenvalues > 1.0), ${\lambda }_i$ is the variance contribution rate (eigenvalue) of the i-th principal component, and $F_i$ is the individual score of the accession on the i-th principal component. For SSR genotyping, data were encoded as a binary matrix (1 for presence, 0 for absence). Genetic diversity parameters—including observed alleles (N_a), effective alleles (N_e), Nei’s gene diversity (H), and Shannon’s Information Index (I)—were computed using POPGENE32 v1.32 [31]. The polymorphism information content (PIC) was calculated using PowerMarker v3.25 [32] based on the formula:

$$PIC = 1-{\displaystyle {\sum }_{i=1}^n}{p}_i^2-{\displaystyle {\sum }_{i=1}^{n-1}}{\displaystyle {\sum }_{j=i+1}^n}{2p}_i^2{p}_j^2,$$

where p_i and p_j are the frequencies of the i-th and j-th alleles. Nei’s genetic distance [33] was calculated based on allele frequencies. A Mantel test [34] with 999 permutations was performed to assess the correlation between the standardized phenotypic Euclidean distance matrix and the SSR-based Nei’s genetic distance matrix. The UPGMA dendrogram was constructed using MEGA 11 [35] and visualized with iTOL v6 [36].

3. Results and Discussion

3.1. Analysis of Morphological Genetic Diversity in P. quinquefolius Accessions

3.1.1. Differences in Morphological Characteristics of P. quinquefolius

Significant morphological variation was observed in the aerial parts of P. quinquefolius from flowering to fruit ripening (Figure 1). Stem morphology was distinctively categorized into single- and double-stem types, with the latter identified in accessions WD3, JY2, SYB, and LH1. Notably, multi-stem accessions exhibited generally higher fresh root weights compared to single-stem types (Table S1), indicating superior biomass accumulation and yield potential. This observation aligns with commercial agronomic practices, where farmers artificially induce multi-stem formation to maximize production [37]. Regarding reproductive traits, accessions QY4, JA1, and DH3, characterized by compound inflorescences and spherical fruits, displayed significantly higher fruit counts. These accessions serve as valuable parental lines for hybridization with genotypes possessing high thousand-seed weight to systematically enhance seed yield. Additionally, the identification of accessions with yellow fruits and inflorescence bracts not only enhances the ornamental potential of P. quinquefolius but also provides unique materials for investigating flavonoid metabolic pathways [10]. For accessions with an upright petiole growth form, there was less leaf shading between adjacent lines in the field. Further purification of such accessions is expected to allow increased planting density.

3.1.2. Phenotypic Variability and Genetic Potential of P. quinquefolius Accessions

Agronomic traits provide a direct and accessible measure of germplasm diversity. For the 10 qualitative traits, Shannon’s diversity indices (H′) varied from 0.22 (BB and LS) to 1.05 (LESD) (

Table 3. Frequency distribution and genetic diversity indices (H′) of 10 qualitative morphological traits across the 51 P. quinquefolius accessions.

Traits	Frequency Distribution (%)			H′
	1	2	3
Inflorescence type	84.31	15.69		0.43
Bud bracts	94.12	5.88		0.22
Fruit color	9.80	90.20		0.32
Ear shape	25.49	74.51		0.57
Petiole growth type	72.55	17.65	9.80	0.76
Leaf edge sawtooth degree	35.29	45.10	19.61	1.05
Leaf color	11.76	72.55	15.69	0.77
Leaf shape	5.88	94.12		0.22
Stem color	58.85	23.53	17.65	0.96
Stem type	86.27	13.73		0.40

Note: The numbers 1, 2, and 3 correspond to the specific phenotypic categories consistent with the recording criteria detailed in Table 1. H′: Shannon–Weaver genetic diversity index.

), revealing a landscape where specific morphological types have become predominant while key traits maintain variability. Notably, leaf edge sawtooth degree (LESD) and stem color exhibited the highest polymorphism (H′ = 1.05 and 0.96, respectively), reflecting a rich phenotypic variation available for breeder selection. For instance, stem color ranged from green (58.85%) to purple (17.65%), offering distinct morphological markers. In contrast, traits such as bud bracts (BB) and leaf shape (LS) displayed lower diversity indices (H′ = 0.22), indicating a high degree of morphological uniformity within the population. This stability suggests that these traits have been fixed through long-term adaptation or artificial selection, serving as consistent diagnostic features for the species.

To elucidate the heritability and breeding potential of quantitative traits, comprehensive genetic parameters—including variance components (V_g, V_p, V_e), coefficients of variation (ECV, GCV, PCV), and broad-sense heritability (H²)—were systematically evaluated (

Table 4. Estimates of genetic parameters, variance components, and heritability for 18 quantitative agronomic traits in P. quinquefolius.

Traits	Grand Mean	Range	Vg	Vp	Ve	R2	ECV (%)	GCV (%)	PCV (%)	H2 (%)
Initial flowering period (day)	50.04	41.33–55.67	10.81	14.91	4.10	0.73	4.05	6.57	7.72	72.50
Fruit ripening period (day)	110.29	104.67–116.00	9.47	12.78	3.31	0.74	1.65	2.79	3.24	74.10
Stem height (cm)	27.15	11.94–46.50	67.11	91.68	24.57	0.73	18.26	30.17	35.27	73.20
Stem diameter (mm)	6.54	4.49–9.68	1.12	1.50	0.38	0.75	9.42	16.18	18.74	74.50
Petiole length (cm)	9.83	5.29–15.45	3.20	4.45	1.25	0.72	11.37	18.22	21.46	71.90
Leaf length (cm)	12.73	9.06–20.34	7.37	10.45	3.08	0.71	13.78	21.32	25.40	70.50
Leaf width (cm)	7.01	5.15–10.50	1.59	2.18	0.59	0.73	10.95	17.95	21.07	72.80
Leaf length to width ratio	1.84	1.35–2.63	0.15	0.22	0.07	0.69	14.38	20.89	25.50	68.90
Peduncle length (cm)	38.24	32.06–44.54	6.26	8.76	2.50	0.72	4.14	6.54	7.74	71.50
Thousand seed weight (g)	13.22	5.55–22.11	5.26	7.09	1.83	0.74	10.23	17.34	20.14	74.20
Rhizome length (mm)	9.99	4.40–16.70	2.23	3.40	1.17	0.66	10.82	14.95	18.46	65.50
Rhizome thickness (mm)	10.22	4.44–17.20	14.89	21.24	6.35	0.70	24.66	37.74	45.10	70.10
Taproot length (cm)	10.09	4.81–17.21	13.04	19.41	6.37	0.67	25.02	35.78	43.64	67.20
Taproot diameter (mm)	19.06	11.83–32.48	11.95	17.19	5.24	0.70	12.01	18.13	21.75	69.50
Number of lateral roots	2.50	0.87–4.13	0.80	1.07	0.27	0.75	20.78	35.89	41.38	74.80
Number of fibrous roots	19.92	7.37–33.83	40.29	65.29	25.00	0.62	25.10	31.87	40.54	61.70
Fresh root weight (g)	26.43	17.25–48.86	84.24	112.92	28.68	0.75	20.27	34.72	40.20	74.60
Ginsenoside content (%)	2.71	1.68–3.67	0.21	0.31	0.10	0.68	11.66	16.89	20.55	68.40

Note: V_g: Genotypic variance; V_p: Phenotypic variance; V_e: Environmental variance; R²: Coefficient of determination; ECV: Environmental Coefficient of Variation; GCV: Genotypic coefficient of variation; PCV: Phenotypic coefficient of variation; H²: Broad-sense heritability.

). The coefficient of determination (R²) estimates ranged from 0.62 to 0.75, indicating that the statistical model accounted for a considerable proportion of the phenotypic variation, although the influence of environmental variance (V_e) was notable. Based on the classification criteria established by Nardino et al. [38], the experimental coefficient of variation (ECV) for specific root traits such as TL (25.02%) and NFR (25.10%) fell into the high range (>20%), reflecting the significant impact of micro-environmental factors on root plasticity under field conditions [39,40]. In contrast, traits like TD (12.01%) and SH (18.26%) showed moderate ECV, suggesting relatively better experimental stability. According to Drisya Ravi et al. [41], phenotypic coefficients of variation (PCV) and genotypic coefficients of variation (GCV) values are classified as low (<10%), moderate (10–20%), and high (>20%). In this study, PCV values were consistently higher than GCV across all traits, indicating the presence of environmental influence. However, as noted by Palange et al. [42], the magnitude of the difference between PCV and GCV is critical for understanding the stability of trait expression. For traits like SD and PL′, the gap between PCV and GCV was relatively narrow (e.g., PL′: PCV = 7.74%, GCV = 6.54%), suggesting that genetic factors (V_g) play a predominant role in trait expression and that these traits are less sensitive to environmental fluctuations. In contrast, traits with larger gaps (e.g., NFR) implied a greater environmental contribution [43]. Analysis revealed that most root morphological traits (RT, TL, NLR, NFR and FRW) exhibited high PCV values (>40%), highlighting abundant genetic variability among accessions and offering a wide scope for yield improvement through selection. Conversely, phenological traits (IFP and FRP) exhibited low coefficients of variation (PCV < 8%). However, this statistical stability is largely attributable to the calculation method based on total days from emergence, in which the large baseline values dilute the relative variance. Given the substantial heritability of these traits (H² > 70%), even small numerical gains represent significant biological shifts, allowing breeders to fine-tune crop cycles for regional adaptability [44].

More broadly, to further validate the genetic potential of the population, robust estimates of broad-sense heritability (H²) are fundamental. According to Terfa and Gurmu [45], H² estimates are classified as low (<30%), moderate (30–60%), and high (>60%). In our study, calculated H² values ranged from 61.70% (NFR) to 74.80% (NLR), uniformly falling into the high category across all evaluated traits. This indicates that, despite environmental noise, the phenotype remains a robust indicator of the genotype within this cohort [46]. The consistently high H² across vegetative, root architectural, and quality traits highlights a strong genetic control over these phenotypes. Nevertheless, the expression of these traits in American ginseng may still be affected by local ecological conditions and human-mediated agronomic practices, including temperature, shade intensity, soil properties, moisture availability, replant history, fertilization, disease management, and pathogen pressure. Thus, the phenotypic variation observed here should be interpreted within the context of the present standardized field environment. While the single-location nature of this trial limits the full dissection of genotype-by-environment (G × E) interactions, conducting systematic testing and validation of promising accessions identified from this population across major American ginseng production regions in China will be essential to ensure their consistent performance under diverse ecological conditions.

Collectively, the genetic parameter analysis demonstrates abundant genetic variability (high PCV/GCV in key yield traits) coupled with consistently high heritability, providing a robust foundation for phenotypic selection and subsequent genetic improvement of high-yielding American ginseng varieties.

3.2. Correlation Among Traits

Multivariate correlation analysis of 18 quantitative traits across 51 P. quinquefolius accessions revealed distinct association patterns (Figure 2). Among 153 pairwise comparisons, 27 trait pairs exhibited highly significant associations (p ≤ 0.01), consisting of 22 positive and 5 negative correlations, while 13 pairs showed significant relationships (p ≤ 0.05; 9 positive and 4 negative). Notably, several pairs of traits demonstrated strong positive correlations, reflecting a coordinated developmental pattern among different plant organs. For instance, both SH and SD were highly significantly positively correlated with PL (r = 0.701 and 0.713, p ≤ 0.01), and both SH and SD were strongly correlated with TD (r = 0.713 and 0.799, p ≤ 0.01). This positive association reflects the allometric scaling between vegetative traits and underground biomass, a pattern consistent with findings reported by Freschet et al. [47], suggesting that stronger aboveground growth is associated with higher root biomass accumulation. Consistent with the findings of Abaya et al. [14], a particularly significant association was observed between LL and FRW (r = 0.816, p ≤ 0.01), suggesting that the size of the leaf, as a key photosynthetic organ, might be a reliable indicator of root biomass accumulation, which is a crucial trait for increasing yields in cultivation. In contrast, significant negative correlations may reflect a potential allocation relationship between vegetative growth and reproductive investment [48]. Specifically, thousand-seed weight (TSW) was negatively correlated with LL and FRW (r = −0.619 and −0.577, p ≤ 0.01), suggesting a possible negative association between reproductive investment and vegetative growth under the present experimental conditions. This finding aligns with observations in other Panax species, where excessive reproductive expenditure often constrains underground biomass accumulation [49]. Crucially, GC, a key determinant of medicinal quality, showed moderate positive correlations with vegetative traits (e.g., LL and NLR; 0.4 < r < 0.7, p ≤ 0.01). This suggests that while vigorous vegetative growth provides the photosynthate source for secondary metabolite biosynthesis, the relationship between yield and quality is complex and not strictly linear [50].

Root phenotypes and GC represent the primary agronomic targets in P. quinquefolius. The demonstrated correlations between these subterranean traits and aerial morphological features suggest that these aboveground traits may help identify accessions combining favorable root biomass and quality-related characteristics. These correlations suggest that aboveground traits may serve as useful auxiliary indicators for early-stage screening, although their predictive value should be validated across environments. Conversely, the observed negative associations may require trait-specific selection strategies, such as prioritizing either seed production or root biomass. Overall, these findings provide a phenotypic framework for understanding trait integration in P. quinquefolius, guiding the strategic utilization of germplasm for diverse agronomic and market demands.

3.3. Principal Component Analysis and Comprehensive Evaluation

A total of six principal components (PCs) with eigenvalues greater than 1.0 were extracted, explaining a cumulative variance of 78.50% (Figure 3A and Table S2). This indicates that these six PCs effectively captured the majority of the phenotypic variation present in the original dataset. The biplot of the first two principal components visually confirmed the trait relationships and accession distribution (Figure 3B). The first principal component (PC1) explained 23.60% of the total variance, characterized by strong covariance between aerial traits (SH, SD and PL) and subterranean dimensions (RT and TD), reflecting a coordinated allometric growth pattern where robust stems and petioles support thicker roots and rhizomes. The second principal component (PC2) accounted for 20.57% of the variance, primarily representing leaf morphology (LL and LLWR), thousand-seed weight (TSW) and fresh root weight (FRW). The third principal component (PC3) explained 13.92% of the variance, highlighting the contribution of developed lateral/fibrous roots (NLR and NFR) and ginsenoside content (GC). The remaining components captured specific morphological nuances: the fourth principal component PC4 (7.92%) represented phenological timing (IFP and FRP); the fifth principal component PC5 (6.74%) focused on leaf width (LW); and the sixth principal component PC6 (5.75%) reflected taproot length (TL). Consequently, the traits dominating the first two principal components (SH, SD, PL, RT, TD, LL, LLWR, TSW, and FRW) were identified as the primary discriminatory indicators for distinguishing P. quinquefolius accessions, while traits from subsequent components (e.g., NLR, NFR, and GC) served as complementary descriptors to resolve finer phenotypic differences.

Based on the eigenvector matrix and standardized phenotypic data, a comprehensive score (F) was calculated for each accession. The scores ranged from −1.22 (SYB) to 1.67 (CY3) (

Table 5. Comprehensive evaluation and ranking of 51 P. quinquefolius accessions based on principal component scores.

Accession	Principal Component Value						F-Value	Ranking
	F1	F2	F3	F4	F5	F6
CY3	2.63	4.29	1.44	−0.30	0.50	3.15	1.67	1
KD1	1.39	3.96	0.35	0.52	0.75	−0.74	1.19	2
DH1	−0.89	2.15	0.37	1.11	1.36	0.24	0.86	3
LH2	−2.91	3.81	−1.61	1.91	−0.19	−0.23	0.82	4
AT1	−1.44	2.11	1.38	0.35	−0.18	−0.03	0.79	5
QY2	−3.81	2.21	1.62	−0.18	0.55	−2.24	0.67	6
LH1	−1.02	2.05	−0.29	0.53	1.73	−0.37	0.64	7
RH3	3.26	2.60	−0.46	0.26	0.48	−1.12	0.64	8
JY1	−1.47	3.36	−0.51	−1.41	0.63	−0.69	0.63	9
KD2	−2.44	1.79	−0.82	2.80	0.97	−0.31	0.62	10
JZ2	0.38	2.33	0.20	−0.50	0.28	−0.34	0.60	11
DHC	1.65	0.74	1.36	0.67	−1.47	0.42	0.43	12
TH4	0.33	1.60	1.00	−1.42	0.02	−0.71	0.41	13
TH2	2.73	1.14	1.26	−1.70	−0.47	0.59	0.40	14
TH3	1.03	0.41	1.40	−0.89	−0.33	1.70	0.38	15
WQ1	−0.16	−0.52	3.17	0.27	−0.17	−0.99	0.37	16
QY1	2.20	−0.90	2.57	0.44	0.50	0.27	0.36	17
QY3	−1.64	1.63	−1.17	2.42	−2.10	1.28	0.35	18
JZ1	3.44	0.65	0.30	−0.14	1.08	−0.44	0.33	19
ZNYS-2	5.82	−1.01	0.23	1.67	2.02	−0.76	0.17	20
RH1	−2.60	−0.39	0.05	1.00	1.40	0.61	0.13	21
WD3	−2.51	1.72	−1.71	−1.58	1.13	0.34	0.07	22
LB3	0.64	−0.78	2.41	−0.23	−1.35	−0.92	0.03	23
RH2	2.34	0.15	0.27	−0.78	−1.68	0.95	−0.03	24
JY2	−1.65	−0.07	1.00	−1.95	0.07	−0.06	−0.07	25
BQ3	0.82	−1.35	2.28	−1.07	−0.18	−0.42	−0.09	26
TH1	−2.38	0.24	1.11	−1.59	−1.77	−0.39	−0.12	27
BQ1	1.82	−1.22	0.14	2.65	−1.29	−0.43	−0.14	28
JA1	−0.58	−1.02	1.91	−0.23	−1.37	−0.79	−0.14	29
LB1	−2.16	−3.32	2.38	0.80	0.75	1.96	−0.20	30
WD1	1.51	0.72	−2.23	−1.57	−0.31	2.22	−0.20	31
DHB	−0.77	−0.62	−0.66	0.71	−1.26	1.53	−0.22	32
LB2	−2.57	−1.27	0.87	−0.74	0.54	0.40	−0.22	33
CY4	−2.25	−0.36	−0.72	0.21	−1.25	1.16	−0.25	34
CY1	−1.07	−1.22	0.29	−0.38	0.83	−0.44	−0.29	35
LH3	0.82	−0.63	−0.75	1.35	−1.69	−0.22	−0.31	36
JA2	1.14	−2.48	1.47	−0.19	0.26	−0.13	−0.38	37
ST2	−0.18	−0.24	−1.18	−0.96	−0.30	0.23	−0.38	38
DHA	−1.27	−1.41	−0.79	0.06	0.91	−0.43	−0.48	39
BQ2	−2.59	−1.71	−0.70	0.42	0.27	0.64	−0.51	40
ZNYS-1	2.53	−1.70	−1.56	2.25	−0.71	0.03	−0.51	41
ST3	1.27	−1.83	−0.30	−1.07	0.93	−0.19	−0.55	42
DH3	0.79	−2.69	0.89	0.50	0.03	−1.64	−0.60	43
QY4	−2.13	−1.95	0.67	−0.82	−1.15	−0.48	−0.65	44
JNYS-1	1.54	−2.43	−1.67	−0.30	2.56	0.12	−0.70	45
CY2	−1.06	−1.48	−1.32	0.49	−0.98	−1.07	−0.75	46
ST1	0.98	−0.90	−2.79	0.09	−0.01	−1.07	−0.78	47
WD4	2.44	−0.54	−3.53	−1.08	−0.21	0.34	−0.83	48
DH2	−1.62	−3.57	−1.03	0.81	0.80	0.96	−0.93	49
WD2	1.34	0.68	−4.11	−1.56	−2.22	−1.87	−1.01	50
SYB	−1.67	−2.77	−2.47	−1.63	1.31	0.32	−1.22	51

Note: F1–F6: Principal component scores calculated for each accession corresponding to the six extracted components (PC1–PC6 defined in Table 5). A positive score indicates a strong expression of the traits associated with that component. F-Value: The comprehensive evaluation score, calculated as the weighted sum of the six component scores using their respective variance contribution rates as weights. Ranking: Accessions are ranked in descending order based on the F-Value. A higher ranking indicates superior overall agronomic performance.

). The top five accessions were CY3, KD1, DH1, LH2, and AT1, with F values of 1.67, 1.19, 0.86, 0.82, and 0.79, respectively. CY3 and KD1 showed tall stature and high yield potential, DH1 and LH2 had relatively high ginsenoside content, and AT1 exhibited well-developed lateral and fibrous roots (Table S3). These accessions may therefore serve as candidate materials for targeted breeding.

3.4. Cluster Analysis Based on Phenotypes

Phenotype-based clustering classified the 51 accessions into four groups (Groups I–IV), reflecting differences in major agronomic trait combinations (Figure 4). Group I, the largest cluster comprising 29 accessions (56.86%), was distinguished by delayed flowering, elevated thousand-seed weight, and elongated taproots (Table S4). However, these traits were associated with compromised overall root yield, suggesting a possible negative association between reproductive investment and root yield, consistent with findings by Zhao et al. [51]. Group II (12 accessions, 23.53%) was characterized by early phenological development, indicating its potential value for selecting early-maturing materials. In contrast, Group III (3 accessions, 5.88%) featured tall plant architecture, thick taproots, and well-developed lateral/fibrous root systems. These characteristics align closely with current market preference for premium whole-root products, suggesting their potential value for whole-root product-oriented breeding. Finally, Group IV (7 accessions, 13.73%) represented a superior agronomic cluster, characterized by large leaf area, high root yield, and elevated ginsenoside content. Consequently, accessions in this group may provide useful parental materials for breeding programs aiming to simultaneously improve yield and quality. These group-level trait profiles provide a practical reference for selecting accessions according to different breeding objectives.

3.5. Genetic Polymorphism of the SSR Markers

Genome-wide mining of SSRs in P. quinquefolius revealed extensive microsatellite resources for future genetic studies and marker development. Screening of approximately 3.57 Gb of genomic sequences identified 976,584 SSR loci (density: 267.14 loci/Mb), and 148,756 compound SSRs, indicating a high microsatellite density characteristic of the P. quinquefolius genome (Figure 5A). Structurally, dinucleotide repeats dominated the landscape (47.32%), followed by mononucleotide (26.66%), compound (13.22%), and tri- to hexanucleotide motifs (11.12–0.37%). Notably, motif abundance showed an inverse correlation with repeat unit length (Figure 5B), a pattern commonly observed in plant genomes [52,53]. Motif analysis identified AT/TA as the predominant type (43.82%), followed by A/T (28.59%) and AAT/ATT (8.51%) (Figure 5C), reflecting the inherent genomic AT-bias characteristic of the P. quinquefolius genome. This SSR resource provides a useful basis for developing additional polymorphic markers for future genetic analysis and breeding studies in P. quinquefolius.

3.6. Polymorphism and Genetic Diversity Analysis of SSR Markers

Genetic characterization using the 12 polymorphic SSR markers revealed substantial discriminatory power across the 51 P. quinquefolius accessions (

Table 6. Genetic diversity parameters and polymorphism information content (PIC) of the 12 selected SSR markers across the 51 P. quinquefolius accessions.

Primers	Na	Ne	H	I	PIC
Pq29	4.00	3.01	0.31	0.49	0.34
Pq164	7.00	6.53	0.46	0.66	0.53
Pq298	5.00	4.98	0.08	0.11	0.46
Pq303	4.00	3.99	0.26	0.37	0.55
Pq346	16.00	7.92	0.27	0.43	0.83
Pq455	6.00	5.44	0.17	0.25	0.49
Pq460	8.00	7.27	0.45	0.64	0.61
Pq584	5.00	4.91	0.49	0.68	0.42
Pq601	6.00	5.00	0.36	0.53	0.29
Pq807	11.00	4.64	0.31	0.48	0.86
Pq827	6.00	4.97	0.39	0.57	0.61
Pq1038	8.00	6.68	0.40	0.58	0.63
Mean	7.17	5.44	0.33	0.48	0.55

Note: N_a: Observed number of alleles, N_e: Effective number of alleles, H: Nei’s gene diversity index, I: Shannon’s information index, PIC: Polymorphism information content.

and Figure S1). The N_a ranged from 4.00 to 16.00 (mean = 7.17), while N_e varied from 3.01 to 7.92 (mean = 5.44). Genetic diversity indices were moderate, with H averaging 0.33 (range: 0.08–0.49) and I averaging 0.48 (range: 0.11–0.68). PIC values ranged from 0.29 to 0.86, with a mean of 0.55, indicating that the selected markers were informative for distinguishing the tested accessions. However, the moderate values of H and I suggest that the detected molecular diversity was limited, which may be related to the restricted marker number and the cultivated nature of the tested materials. Given the limited number of polymorphic SSR markers, the molecular diversity estimates obtained here should be interpreted as a preliminary assessment rather than a high-resolution characterization of population structure.

3.7. Genetic Distance and Cluster Analysis

Genetic distances among the 51 P. quinquefolius accessions ranged from 0.04 to 0.49 (mean = 0.26) (Table S5), suggesting detectable genetic differentiation that may be useful for preliminary parental selection. The closest relationship (D = 0.04) was identified between TH3 and TH4, both originating from the same municipality and exhibiting highly similar phenotypes (FRW: approximately 33–35 g; GC: approximately 2.6–2.9%) (Table S3). Conversely, the greatest divergence (D = 0.49) occurred between WQ1 (Jilin) and ST2 (Anhui). These two accessions displayed distinct phenotypic complementarity: WQ1 was characterized by high ginsenoside content (3.12%) but lower root weight, whereas ST2 exhibited superior root biomass (32.07 g) but lower phytochemical content. Such genetically distant pairs with complementary traits may represent promising parental combinations for future crossing experiments. Notably, the genetic distances among the five candidate elite accessions selected by PCA (CY3, KD1, DH1, LH2, and AT1) ranged from 0.19 to 0.34. This moderate genetic divergence suggests that these accessions may provide useful parental variation for future crossing and segregation studies while maintaining favorable agronomic traits, making them suitable candidates as parents for hybridization. The UPGMA dendrogram based on SSR markers classified the population into five clusters (Groups I–V) (Figure 6). Group I contained five accessions from three provinces, while Group II (12 accessions) was dominated by accessions from Jilin and Liaoning. Notably, Group III comprised only two cultivars, ‘ZNYS-1’ and ‘ZNYS-2’, suggesting a shared pedigree or selection history. Groups IV and V were larger, mixed clusters containing accessions from diverse geographical origins.

Combining genetic distance and cluster analysis revealed a weak association between SSR-based clustering patterns and geographical origin. While localized sub-clusters existed (e.g., Group II), most groups showed extensive geographical mixing. Similar weak associations between genetic clustering and geographic origin have been reported in other long-cultivated perennial crops, probably reflecting human-mediated germplasm exchange [54,55]. For P. quinquefolius in China, this weak geographic pattern is likely associated with anthropogenic factors: frequent trans-provincial seed exchange, unregulated introductions, and cultivation-based selection over the past 40 years may have facilitated extensive gene flow and weakened natural geographical differentiation [56]. This finding has practical implications for germplasm management, indicating that geographic origin or traditional provenance labels should not be used alone as proxies for accession identity. Instead, reliable identification and utilization of American ginseng germplasm should rely on standardized molecular fingerprints, passport information, phenotypic records, and, where available, pedigree or propagation history.

Furthermore, the discordance between SSR-based (Figure 6) and phenotype-based (Figure 4) clustering suggests that phenotypic divergence was not solely determined by neutral molecular variation, but may also reflect environmental plasticity, artificial selection under cultivation, and the polygenic nature of complex traits, such as root yield and ginsenoside content [14,57,58]. Consistently, the Mantel test showed no significant correlation between phenotypic and SSR-based distances (Mantel r = 0.018, p = 0.294; Table S6), indicating that phenotypic similarity does not necessarily correspond to SSR-based genetic similarity. In addition, the limited number of putatively neutral SSR loci used in this study may not fully capture genome-wide variation associated with agronomic traits. Therefore, integrating phenotypic and SSR information may improve parental selection by identifying accessions with complementary traits and relatively divergent molecular profiles. Given the limited marker density, model-based population structure analyses, such as STRUCTURE, were not performed in this study.

3.8. Proposal of a Representative Accession Subset for Future Breeding and Conservation

To facilitate future breeding and conservation, a representative subset of 12 accessions, accounting for 23.53% of the evaluated germplasm, was proposed based on phenotypic clustering, PCA-based comprehensive evaluation, and SSR-based genetic diversity [59]. Three accessions were selected from each of the four phenotypic clusters to ensure phenotypic representativeness, with priority given to accessions showing high PCA-based performance and desirable agronomic traits. SSR cluster membership and pairwise Nei’s genetic distances were further used to reduce genetic redundancy. For instance, CY4 was excluded because of its short genetic distance from CY3, whereas JZ1 was retained to improve SSR-based genetic coverage. The final subset included DH1, AT1, JZ1, KD1, RH3, JZ2, CY3, RH2, TH2, LH2, KD2, and WD3. Among them, CY3, KD1, DH1, LH2, and AT1 represented elite accessions with superior comprehensive performance, while WD3 was retained for its distinctive double-stem phenotype. This subset covered all phenotypic clusters and different SSR-based profiles, providing a practical working panel for parental selection, germplasm management, genetic analysis, and pre-breeding studies [60,61]. Details are provided in Table S7.

Beyond its use for breeding and genetic analysis, this subset provides a practical basis for conserving valuable germplasm [59]. Because these accessions are breeding materials rather than released cultivars, their identity should be maintained through standardized field preservation in a dedicated germplasm nursery at Jilin Agricultural University, with each accession linked to passport information, phenotypic records, and SSR fingerprints [62]. Separate handling during propagation, together with duplicate field collections or backup seed stocks and periodic phenotypic and molecular verification, would help minimize admixture, reduce material loss, and preserve these accessions as reliable parental resources for future evaluation and breeding [63,64].

4. Conclusions

This study integrated phenotypic traits and SSR markers to evaluate genetic diversity among 51 selected P. quinquefolius accessions. Substantial phenotypic variation was observed across 28 agronomic traits, particularly in root-related traits such as fresh root weight, taproot diameter, and lateral root number. These traits, together with ginsenoside content, showed high broad-sense heritability, indicating their potential for effective improvement through phenotypic selection. The positive correlations between aerial vegetative traits and root biomass also provide a basis for non-destructive early-stage screening.

SSR-based analysis revealed weak geographical differentiation and limited concordance with phenotypic clustering, suggesting that long-term germplasm exchange, cultivation history, and human selection may have reshaped the genetic composition of local materials. Based on the integrated evaluation, five candidate elite accessions and a 12-accession representative subset were identified. Together with the proposed accession-level conservation strategy, these materials provide a practical basis for future breeding, germplasm management, and genetic analysis. Further multi-year and multi-location validation across major American ginseng production regions will be necessary to assess genotype-by-environment interactions and confirm the stability and adaptability of these materials under diverse ecological conditions. High-density genome-wide marker analysis will also help clarify population structure more precisely and support future marker-assisted improvement.