Genotype-by-Environment Interaction and Stability Analysis for Four Functional Compounds in Tea Chrysanthemums: A Three-Year Study

Abstract

Chrysanthemum contains numerous active compounds, including flavonoids and phenolic acids, with its dried capitula widely used for tea and medicinal applications. The content of functional compounds is readily influenced by environmental factors, and the use of varieties with high-level and stable bioactive compounds is essential for sustainable cultivation. However, a key challenge is identifying genotypes that consistently perform well for functional-component traits in contemporary breeding activities. This study aimed to evaluate the performance and stability of functional components in tea chrysanthemums across multiple years. Total flavonoids, chlorogenic acid, luteoloside, and isochlorogenic acid A were investigated in 24 tea chrysanthemum accessions across three growing years of 2018, 2021, and 2022. The additive main effects and multiplicative interaction (AMMI) model analysis revealed significant genotype (G), environment (E), and genotype-by-environment interaction (GEI) effects for all functional traits across three growing years. The GEI accounted for 63.58% to 80.82% of the variation across the four components in the AMMI model. Based on the AMMI stability value (ASV) parameter, the tea chrysanthemums showing the most stable concentrations of total flavonoids, chlorogenic acid, luteoloside, and isochlorogenic acid A were identified. Based on phenotypic values and stability results, Suju-6, Hongxinju, Wangongju, and Baixiaoxiangju performed relatively well across the functional components investigated, making them promising candidates for future breeding and promotion programs. These findings provide valuable insights into the genetic basis of functional elements in tea chrysanthemum and will contribute to further genetic improvement.

1. Introduction

Chrysanthemum (Chrysanthemum morifolium Ramat.) is a renowned ornamental plant native to China, with a cultivation history of more than 3000 years. It displays a substantial decorative effect and serves as an edible and medicinal herb, thereby dramatically boosting the commercial benefits of the flower industry chain [1,2]. Chrysanthemum flowers have many functional compounds, i.e., flavonoids, organic acids, etc. These ingredients function in heat-clearing, detoxification, liver protection, and vision improvement, while modern research supports their antiviral, antitumor, antioxidant, immunomodulatory, and cardiovascular benefits [3,4]. Due to their health-promoting properties, dried capitula harvested at the optimal flowering stage are widely used in traditional Chinese medicine formulas and as a flower tea in daily life [5,6]. According to the Chinese Pharmacopeia [7], chrysanthemum must contain minimum levels of chlorogenic acid (≥0.2%), luteoloside (≥0.08%), and isochlorogenic acid A (≥0.7%). Enhancing bioactive compounds is therefore a key goal for the functional chrysanthemum industry.

Despite their functional and commercial importance, the inheritance pattern of chrysanthemum active compounds is scarcely explored thus far. Recently, Ning [1] identified genetic variation in major active compounds in an F₁ segregating population and screened for elite hybrids with superior levels of these compounds for breeding. Zhang [6] identified the excellent alleles and candidate genes responsible for the major active compounds. These findings demonstrated that the functional compounds, as secondary metabolites, are complex quantitative traits with moderate heritabilities in chrysanthemums, mainly determined by polygenes and also affected by environmental factors, i.e., soil fertility, inorganic elements, and climatic conditions. The most widely cultivated tea chrysanthemum varieties in China include Chuju, Boju, Gongju, Huaiju, and Hangju, which differ in appearance, aroma, and chemical composition due to genetic and environmental factors [8]. However, current production still relies heavily on classic varieties, which suffer from limited high-quality germplasm, inconsistent quality, and insufficient focus on functional components [1]. Moreover, morphological and chemical traits, influenced by secondary metabolites, vary with genetic, physiological, and ecological factors, leading to variability even within the same variety [9]. Current research on tea chrysanthemum functional components primarily focuses on comparative analyses among classic varieties [8,10] or on new cultivar development [11], with limited attention to varietal stability. Stability refers to a genotype’s ability to perform consistently across a wide range of environments [12]. In agricultural production, assessing varietal stability is a crucial step prior to introducing new varieties. Genotype-by-environment interaction (GEI), the differential response of genotypes to varying environments, is inevitable and significantly influences stability [13]. The additive main effects and multiplicative interaction (AMMI) model, which accounts for GEI, facilitates accurate cultivar recommendation for target environments [14] and has been successfully applied to evaluate stability in many crops such as barley [15], faba bean [16], grain sorghum [17], maize [18] and wheat [19]. This provides an important reference for assessing the genetic stability of bioactive compounds in tea chrysanthemums across environments.

In this study, we examined the contents of total flavonoids, chlorogenic acid, luteoloside, and isochlorogenic acid A in 24 tea chrysanthemum varieties and elite hybrids over three years: 2018, 2021, and 2022. The AMMI model was employed to analyze GEI effects on functional quality, aiming to estimate tea chrysanthemums with high quality and greater stability across multiple years. The findings are significant for understanding the genetic basis of functional elements in tea chrysanthemums and for further genetic improvement.

2. Materials and Methods

2.1. Plant Materials and Field Trial Design

Twenty-four tea chrysanthemums, including eighteen varieties and six elite hybrids (

Table 1. Basic information on the 24 tea chrysanthemums used in this study.

Accession	Abbreviation	Germplasm Type	Flower Type	Flower Color
CH1-44	CH1-44	Elite hybrid	Single	Yellow
CH7-18	CH7-18	Elite hybrid	Semi-double	White
GC1-5	GC1-5	Elite hybrid	Double	White
GC1-16	GC1-16	Elite hybrid	Double	White
GH1-3	GH1-3	Elite hybrid	Double	Yellow
GH1-8	GH1-8	Elite hybrid	Double	Yellow
Suju-6	Sj6	Variety	Incurve	White
Huangju	Hj	Variety	Pompon	Yellow
Jinsihuangju	Jshj	Variety	Double	Yellow
Chuju	Cj	Variety	Double	White
Hangbaiju	Hbj	Variety	Full-double	White
Danyanghua	Dyh	Variety	Semi-double	White
Fubaiju	Fbj	Variety	Full-double	White
Wangongju	Wgj	Variety	Full-double	Yellow
Hongxinju	Hxj	Variety	Full-double	White
Huangxiangli	Hxl	Variety	Double	Yellow
Jinju-3	Jj3	Variety	Full-double	Yellow
Jiuyueju	Jyj	Variety	Semi-double	White
Sheyang Dabaiju	SyDbj	Variety	Full-double	White
Xiaohuangju	Xhj	Variety	Single	Yellow
Xiuning Dabaihua	XnDbh	Variety	Double	White
Zaohua-1	Zh1	Variety	Single	White
Huangxiaoxiangju	Hxxj	Variety	Anemone	Yellow
Baixiaoxiangju	Bxxj	Variety	Anemone	White

), were obtained at the Chrysanthemum Germplasm Resource Preserving Center of Nanjing Agricultural University (Nanjing, China). In 2018, 2021, and 2022, the tea chrysanthemums were cultivated at Hushu experiment bases (119.12° N, 31.80° E), Nanjing, Jiangsu Province, P.R. China. The annual average temperature is 17.0 °C, 17.6 °C, and 17.6 °C; the annual precipitation is 1267.1 mm, 1267.1 mm, and 819.8 mm; the annual sunshine duration is 2035.6 h, 1955.6 h, and 2138.6 h, respectively. Field trials were conducted using a completely randomized block design with three replications. In each replication, the 24 accessions were randomly assigned to individual plots, with 40 cm between plots, and both plant and row spacing were set at 30 cm. Field management followed conventional practices. Field trials were conducted with a randomized block design, and field management followed conventional practices. In autumn, the capitula at approximately 70% flowering were randomly harvested for each accession. According to Ning et al. [1], the gathered capitula were first dried at 37 °C for 30 h in an automatic oven with a circulating hot-air system, then at 60 °C for a further 5 h until completely dry; after natural cooling, the dried flowers were ground into powder using a tissue grinder, passed through a 20-mesh sieve, and stored in sealed containers in a well-ventilated area at room temperature for subsequent analysis.

2.2. Assay of Functional Compounds

In this study, four functional compounds, including total flavonoids, chlorogenic acid, luteoloside, and isochlorogenic acid A, were determined for three years. Each accession is measured three times.

2.2.1. Total Flavonoids

A 0.1 g sample of accurately weighed capitula powder was extracted with 70% aqueous ethanol (v/v) by ultrasonication at 60 °C for 30 min. Extracts were analyzed using the Al(NO₃)₃-NaNO₂ spectrophotometric method. Absorbance at 510 nm was measured with a multifunctional microplate reader. A standard curve was generated using rutin as the reference standard at concentrations of 0, 50, 100, 150, 200, and 250 μg∙mL⁻¹. A linear regression equation was established by plotting absorbance against concentration. Total flavonoid content was calculated as:

Content of total flavonoids = (C × V₂ × V₀)/(V₁ × M)

(1)

where C is the concentration from the regression equation (μg∙mL⁻¹), M is the sample mass (0.1 g), V₀ is the test extract volume (10 mL), V₁ is the initial sample volume (1 mL), and V₂ is the diluted volume (10 mL).

2.2.2. Chlorogenic Acid, Luteoloside, and Isochlorogenic Acid A

According to the Chinese Pharmacopeia [7], the contents of chlorogenic acid, luteoloside, and isochlorogenic acid were simultaneously determined using a Ultra-Performance Liquid Chromatography (UPLC) system (Thermo, Waltham, MA, USA). Briefly, 0.1 g of the capitula powder was extracted with 70% HPLC-grade methanol by ultrasonication at 60 °C for 30 min.

Chromatographic separation was performed on an ACQUITY UPLC HSS T3 (Waters, Milford, MA, USA) column (10 × 2.1 mm, 1.8 μm). The mobile phase consisted of acetonitrile (A) and 1% formic acid aqueous solution (B) with gradient elution: 0–2.0 min (10–18% A), 2.0–6.0 min (18–35% A), 6.0–6.1 min (35–10% A), and 6.1–8.0 min (10% A). The flow rate was 0.3 mL∙min⁻¹, column temperature was 40 °C, detection wavelength was 348 nm, and injection volume was 2 μL. Standard calibration curves were established for each analyte. Concentration gradients of the reference solution standards for the three components were: chlorogenic acid (0.4, 0.8, 2, 4, 20, 40 μg∙mL⁻¹), luteoloside (0.4, 0.8, 2, 4, 20, 40 μg∙mL⁻¹), and isochlorogenic acid A (0.21, 0.42, 4.2, 8.4, 42, 84 μg∙mL⁻¹). Linear regression equations were generated by plotting peak areas against concentrations. The content (%) of the three compounds was calculated as:

Content = (C × V₁ × N)/M

(2)

where C is the concentration from the linear regression equation, V₁ is the extraction volume (1 mL), N is the number of times of dilution, and M is the sample mass (0.1 g).

2.3. AMMI Model

The genotype-by-environment interaction (GEI) for the four functional compounds was analyzed using the AMMI model to assess cultivar stability. The model was implemented with the agricolae package in R, yielding the AMMI stability value (ASV) for each accession. Lower ASVs indicate greater phenotypic stability. The average AMMI stability value (AASV) was calculated as per Scavo [20]—genotypes with ASV values lower than AASV were classified as relatively stable. Visualization was performed using ggplot2 to generate AMMI biplots.

The AMMI model structure [21]:

$$Y_{ij}=\mu +g_i+e_j+{\sum }_{n=1}^N{\lambda }_n{\gamma }_{in}{\delta }_{jn}+{\rho }_{ij}$$

(3)

where Y_ij is the trait value of genotype i under environment j; $\mu$ is the grand mean of the trait; $g_i$ is the main effect of genotype i; e_j is the main effect of environment j; ${\lambda }_n$ is the eigenvalue of the n-th interaction principal component axis (IPCA); ${\gamma }_{in}$ and ${\delta }_{jn}$ are the eigenvector scores of genotype i and j on the n-th IPCA; and ${\rho }_{ij}$ is residua. The ASV formula [22]:

$$ASV=\sqrt{{\displaystyle \frac{SSPC1}{SSPC2}}{\left(PC1\right)}^2+{\left(PC2\right)}^2}$$

(4)

where SS denotes the sum of squares, and PC1 and PC2 are the scores of the first and second principal components.

2.4. Data Statistics

Descriptive statistical analysis was conducted using SPSS Statistics 21.0 to obtain mean, range, kurtosis, and skewness. Duncan’s new multiple-range test was employed for multiple comparisons. R v4.5.2 software packages were used to generate standard curves and boxplots, as well as to calculate p-values.

3. Results

3.1. Establishment of the Standard Curve

The standard curve for total flavonoids (Figure 1a) yielded a linear regression equation of y = 0.0055x + 0.027 (R² = 0.9939), indicating excellent linearity within 0–250 μg∙mL⁻¹. Chromatographic analysis for the three phenolic compounds also showed high linearity: chlorogenic acid: y = 0.0699x − 0.0003 (R² = 0.9994, 0.4–40 μg∙mL⁻¹); luteoloside: y = 0.1761x − 0.0583 (R² = 0.9997, 0.4–40 μg∙mL⁻¹); isochlorogenic acid A: y = 0.1076x − 0.1412 (R² = 0.9978, 0.84–84 μg∙mL⁻¹) (Figure 1b). All R² values exceeded 0.997.

3.2. Phenotypic Variation in Functional Components

The variation in the contents of the four components across 24 accessions over three years is presented in

Table 2. Descriptive statistics of functional compounds in tea chrysanthemums across three years.

Trait	Year	Mean ± SD	CV (%)	Range	Kurt	Skew
Total flavone (TF, mg∙g−1)	2018	78.91 ± 31.35	39.73	29.61~195.46	1.47	1.00
	2021	60.51 ± 30.72	50.76	18.18~145.64	0.87	1.07
	2022	65.56 ± 29.61	45.17	14.06~156.39	−0.05	0.60
	Average	68.33 ± 28.23	41.31	19.82~132.29	−0.70	0.47
Chlorogenic acid (CA, %)	2018	0.48 ± 0.21	43.44	0.11~1.01	0.12	0.78
	2021	0.46 ± 0.3	64.85	0.08~1.17	−1.05	0.34
	2022	0.25 ± 0.09	34.58	0.12~0.48	0.02	0.80
	Average	0.40 ± 0.23	57.50	0.10~1.05	0.02	0.87
Luteoloside (Lut, %)	2018	0.17 ± 0.11	65.48	0.05~0.59	2.09	1.44
	2021	0.32 ± 0.29	90.75	0.02~1.44	4.68	2.00
	2022	0.27 ± 0.21	79.16	0.03~1.06	5.91	2.17
	Average	0.25 ± 0.19	76.00	0.05~1.03	6.46	2.25
Isochlorogenic acid A (ICA, %)	2018	1.28 ± 0.57	44.26	0.27~2.82	−0.13	0.56
	2021	1.11 ± 0.73	65.94	0.16~3.42	1.95	1.33
	2022	1.18 ± 0.48	41.05	0.55~2.33	0.18	1.05
	Average	1.19 ± 0.47	59.50	0.46~2.45	0.70	1.04

. The coefficients of variation (CV) over the three years ranged from 34.58% to 90.75%. Based on three-year averages, the CVs ranked as: luteoloside (76%) > isochlorogenic acid A (59.5%) > chlorogenic acid (57.5%) > total flavonoids (41.31%), indicating considerable variation.

The average content of total flavonoids was 68.33 mg·g⁻¹. The average contents of chlorogenic acid, luteoloside, and isochlorogenic acid A were 0.40%, 0.25%, and 1.19%, respectively. Significant yearly differences were observed for three components (Figure 2): total flavonoids were significantly higher in 2018 (78.91 mg·g⁻¹) than in 2021 (60.51 mg·g⁻¹) and 2022 (65.56 mg·g⁻¹); chlorogenic acid was higher in 2018 (0.48%) and 2021 (0.46%) than in 2022 (0.25%); luteoloside was higher in 2021 (0.32%) and 2022 (0.27%) than in 2018 (0.17%). In contrast, isochlorogenic acid A content showed no significant differences among the three years (1.28%, 1.11%, 1.18%). In addition, there is a clear trend indicating no significant differences between 2021 and 2022 for most components, except for chlorogenic acid. Descriptive statistics alone cannot elucidate genotype-by-environment interaction (GEI). Therefore, further analysis was conducted to investigate the main effects of genotype (G), environment (E), and their interaction (GEI) on functional qualities.

3.3. AMMI Model Analysis

The additive main effects and multiplicative interaction (AMMI) model was employed to dissect the contributions of genotype (G), environment (E), and genotype-by-environment interaction (GEI) to the variation in functional components. Analysis of variance (ANOVA) results showed that the genotype sum of squares (SS) accounted for 43.60%, 34.69%, 69.17%, and 59.03% of the total variation for total flavonoids, chlorogenic acid, luteoloside, and isochlorogenic acid A, respectively (

Table 3. AMMI model analysis of the four functional compounds in tea chrysanthemums.

Trait	Item	Df	SS	Proportion of SS	F Value
TF (mg∙g−1)	Treatment	71	169,770.1
	Genotype(G)	23	92,441	43.60%	13.69 **
	Environment(E)	2	13,012.25	6.14%	22.17 **
	G × E	46	64,316.8	30.33%	4.76 **
	IPCA1	24	40,894.29	63.58% a	1.60 **
	Error	144	42,263.4
	Total	215	212,033.5
CA (%)	Treatment	71	11.43
	Genotype(G)	23	4.23	34.69%	35.38 **
	Environment(E)	2	2.25	18.49%	216.85 **
	G × E	46	4.96	40.68%	20.75 **
	IPCA1	24	3.85	77.66% a	3.19 **
	Error	144	0.75
	Total	215	12.18
Lut (%)	Treatment	71	10.37
	Genotype(G)	23	7.41	69.17%	135.04 **
	Environment(E)	2	0.78	7.28%	163.37 **
	G × E	46	2.18	20.35%	19.86 **
	IPCA1	24	1.51	69.10% a	2.05 **
	Error	144	0.34
	Total	215	10.72
ICA (%)	Treatment	71	73.52
	Genotype(G)	23	46.41	59.03%	56.89 **
	Environment(E)	2	0.99	1.26%	14.01 **
	G × E	46	26.11	33.21%	16.00 **
	IPCA1	24	21.1	80.82% a	3.86 **
	Error	144	5.11
	Total	215	78.62

Note: AMMI, additive main effects and multiplicative interaction; G, genotype; E, environment; G × E indicates genotype × year interaction; IPCA1, first interaction principal component axis; Df, degrees of freedom; SS, sum of squares. ^a represents the proportion of SS of IPCA1 relative to SS of G × E; ** indicates a significant difference at the level of p < 0.01.

). Environmental SS was lowest for isochlorogenic acid A (1.26%), followed by total flavonoids (6.14%) and luteoloside (7.28%), and highest for chlorogenic acid (18.49%). The GEI exceeded 20% for all traits, ranging from 20.35% (luteoloside) to 40.68% (chlorogenic acid). For each trait, the percentage contributions are ranked as: genotype > GEI > environment.

Although genotype had the predominant influence, GEI also played a substantial role in the four functional components. As shown in Table 3, GEI was significant (p < 0.01) for all investigated traits. The considerable contribution of the first interaction principal component axis (IPCA1) confirmed the suitability of the AMMI model. IPCA1 contributed significantly (p < 0.01) and explained 63.58%, 77.66%, 69.10%, and 80.82% of the GEI for total flavonoids, chlorogenic acid, luteoloside, and isochlorogenic acid A, respectively, indicating high explanatory power.

3.4. Stability Analysis and Selection of High-Quality Tea Chrysanthemums

Based on the AMMI model, the AMMI stability value (ASV) was used to evaluate stability for each trait (

Table 4. The AMMI stability value (ASV) and content level of four functional components in 24 tea chrysanthemum accessions.

Accessions	TF (mg∙g−1)		CA (%)		Lut (%)		ICA (%)
	ASV	Content	ASV	Content	ASV	Content	ASV	Content
CH1-44	4.61	53.81	0.30	0.22	0.26	0.27	0.81	0.76
CH7-18	4.10	59.97	0.45	0.26	0.14	0.13	1.05	0.95
GC1-5	0.90	44.70	0.97	0.30	0.33	0.13	1.81	1.03
GC1-16	3.62	67.16	1.00	0.29	0.49	0.18	1.67	0.92
GH1-3	5.77	67.00	0.31	0.17	0.13	0.11	0.15	0.62
GH1-8	2.83	58.09	0.52	0.19	0.13	0.16	0.20	0.79
Suju-6	4.66	100.44	0.37	0.52	0.14	0.43	1.53	1.93
Huangju	3.91	116.65	1.15	0.70	1.25	0.95	1.35	2.27
Jinsihuangju	9.23	87.62	0.53	0.36	0.57	0.41	0.87	0.85
Chuju	6.17	70.21	1.08	0.51	0.11	0.15	3.31	2.34
Hangbaiju	2.28	112.83	1.18	0.51	0.62	0.46	1.44	1.25
Dayanghua	5.69	89.26	1.17	0.50	0.45	0.36	1.84	1.24
Fubaiju	3.96	81.80	0.12	0.66	0.45	0.29	1.49	1.36
Wangongju	4.47	72.42	0.51	0.42	0.16	0.24	1.05	1.43
Hongxinju	2.59	39.64	0.55	0.55	0.17	0.19	0.67	1.41
Huangxiangli	4.31	60.25	0.84	0.54	0.43	0.06	1.46	1.35
Jinju-3	5.12	52.42	0.74	0.46	0.33	0.38	0.69	0.95
Jiuyueju	1.06	50.47	1.33	0.38	0.33	0.27	3.00	1.64
Sheyang Dabaiju	2.11	61.62	0.77	0.43	0.10	0.18	1.18	1.04
Xiaohuangju	2.27	49.97	0.88	0.29	0.31	0.16	0.77	0.49
Xiuning Dabaihua	3.93	44.91	0.41	0.27	0.17	0.08	0.60	1.12
Zaohua-1	5.30	69.56	0.26	0.33	0.26	0.15	0.44	1.01
Huangxiaoxiangju	2.30	48.02	0.82	0.29	0.30	0.06	1.08	0.77
Baixiaoxiangju	2.54	81.03	0.39	0.41	0.28	0.28	0.22	1.06
AASV/Average	3.91	68.33	0.69	0.40	0.33	0.25	1.20	1.29

Note: TF, total flavonoids; CA, chlorogenic acid; Lut, luteoloside; ICA, isochlorogenic acid A. Bold values in the ASV columns indicate ASV < AASV (favorable stability). Bold values in the Content columns indicate content > overall average (high content).

). For total flavonoids, GC1-5 had the lowest ASV (0.9), indicating the most stable content, followed by Jiuyueju, Sheyang Dabaiju, Xiaohuangju, and Hangbaiju. For chlorogenic acid, Fubaiju showed the lowest ASV (0.12), followed by Zaohua-1, CH1-44, GH1-3, and Suju-6. For luteoloside, Sheyang Dabaiju ranked first (ASV = 0.10), followed by Chuju, GH1-3, GH1-8, and Suju-6. For isochlorogenic acid A, GH1-3 was the most stable (ASV = 0.15), followed by GH1-8, Baixiaoxiangju, Zaohua-1, and Xiuning Dabaihua.

Based on mean performance (Table 4), the top five genotypes for each functional component were identified. The top five for total flavonoids were Huangju (116.65 mg·g⁻¹), Hangbaiju (112.83 mg·g⁻¹), Suju-6 (100.44 mg·g⁻¹), Dayanghua (89.26 mg·g⁻¹), and Jinsihuangju (87.62 mg·g⁻¹). Huangju (0.70%), Fubaiju (0.66%), Hongxinju (0.55%), Huangxiangli (0.54%), and Suju-6 (0.52%) ranked in the top five for chlorogenic acid. For luteoloside, Huangju (0.95%), Hangbaiju (0.46%), Suju-6 (0.43%), Jinsihuangju (0.41%), and Jinju-3 (0.38%) were the top five accessions. For isochlorogenic acid A, Chuju (2.34%), Huangju (2.27%), Suju-6 (1.93%), Jiuyueju (1.64%), and Wangongju (1.43%) performed better. Overall, Huangju, Suju-6, Hangbaiju, and Fubaiju showed high average contents across years.

Accessions with ASV < AASV and content > overall average were considered as stable and high-quality. AMMI biplots (Figure 3) visualized average content (x-axis) against GEI-PC1 (y-axis). A higher x-coordinate indicates higher content; a y-coordinate closer to zero indicates smaller ASV and greater stability. Hangbaiju and Baixiaoxiangju performed well for total flavonoids; Fubaiju, Suju-6, and Hongxinju for chlorogenic acid; Suju-6, CH1-44, and Baixiaoxiangju for luteoloside; Hongxinju and Wangongju for isochlorogenic acid A. No genotype exceeded the average across all four components. However, Suju-6, Hongxinju, Wangongju, and Baixiaoxiangju demonstrated favorable overall performance, with only isolated indicators falling short.

4. Discussion

Quality evaluation of tea chrysanthemums involves external indicators (e.g., yield, morphology) and internal indicators (e.g., functional components, aroma, sensory properties) [23]. Few studies have examined the genetic stability of chrysanthemum functional quality traits. Total flavonoids, chlorogenic acid, luteoloside, and isochlorogenic acid A are the most essential functional compounds of chrysanthemum and are crucial for health-promoting and disease-preventive effects [24]. Our present study quantified the four functional components in 24 tea chrysanthemum accessions across three years. It revealed wide variation and significant year-to-year differences, highlighting the need to investigate stability.

Plant phenotypic traits are influenced by genotype, environment, and their interaction [25,26]. In chrysanthemum, the accumulation of functional components is susceptible to external factors, such as soil conditions [27]. Hangbaiju from different origins showed varying levels of compounds [10]. Significant compositional variation among chrysanthemum genotypes leads to divergent pharmacological efficacies [8,28]. In the current work, ANOVA within the AMMI model revealed significant effects of genotype, environment, and GEI across all four functional components, indicating substantial variation in mean performance among genotypes across multiple years. The difference in the core climate factors, i.e., annual mean temperature, precipitation, and sunshine duration across 2018, 2021, and 2022, jointly exerted significant regulatory effects on the accumulation of functional components in tea chrysanthemums. In 2018, with suitable annual mean temperature, abundant precipitation, and moderate sunshine duration, the accumulation of total flavonoids and chlorogenic acid was promoted, resulting in the highest contents among the three years. In 2022, the annual mean temperature was similar to that in 2021, but annual precipitation was significantly lower, and sunshine duration was the highest. Drought stress markedly inhibited the accumulation of chlorogenic acid, leading to a sharp decline in its content, whereas strong light promoted the synthesis of luteoloside, resulting in a significantly higher content than in 2018. The higher luteoloside content in 2021 and 2022 indicated that it was more adapted to relatively high temperatures and strong light. Isochlorogenic acid A showed no significant differences across the three years, suggesting it was less affected by annual climatic fluctuations and exhibited high genetic stability. The insignificant differences in most components between 2021 and 2022 were consistent with similar temperatures but with varying precipitation and sunshine duration. These findings underscore that functional quality performance is not solely determined by genetics but is substantially modulated by the growing environment. The variance analysis results demonstrated that genotypic and GEI effects were the primary sources of variation, while the environmental effect alone contributed the least. This highlights the critical need to account for GEI when evaluating genotype performance. As emphasized by Van Eeuwijk et al. [29], significant interaction for quantitative traits can constrain the selection of superior genotypes and limit the utility of subsequent analyses. While this study confirms the predominant role of GEI, precisely delineating its complex influence on functional components in tea chrysanthemum remains challenging and warrants targeted future investigation.

The ideal tea chrysanthemum genotype should possess both high functional-component traits and stable performance across diverse environments. However, as noted by Kuang et al. [30], genotype-by-environment interaction in multi-environment trials complicates analysis and interpretation, reducing the efficiency of identifying stable genotypes. Various statistical methods have been developed to evaluate genotype stability and adaptability. Among these, the AMMI model is one of the most widely applied approaches [15], which combines analysis of variance and principal component analysis with fixed effects [31]. Its value lies in addressing the complexity of GEI, enabling researchers to better understand and interpret genotype performance across diverse environments [32]. The concept of stability is agronomically relevant only when coupled with satisfactory mean performance [33]. In AMMI analysis, stability can be quantified using ASV, with genotypes having lower ASVs considered more stable. For instance, Scavo et al. [20] successfully used ASV in combination with mean performance to select stable, high-yielding potato varieties for traits such as dry matter and total phenolics. An AMMI biplot visualizes yield potential and stability, serving as a key tool for evaluating significant multiplicative GEI for agronomic traits [34].

In practice, AMMI model analysis advocates selecting genotypes that exhibit not only minimal variability (high stability) but also desirable mean performance, ensuring both reliability and high quality. In this study, the ASV and the mean functional-component content across all tested materials were used as selection thresholds to identify relatively stable cultivars with high functional-component content. The AMMI biplot results indicated that Suju-6, Hongxinju, Wangongju, and Baixiaoxiangju were ideal candidates, a finding corroborated by their ASVs. Among these, Suju-6 showed relatively lower stability for total flavonoids and isochlorogenic acid A, while Hongxinju had room for improvement in the content of total flavonoids and luteoloside. Wangongju was slightly less stable for total flavonoids, and Baixiaoxiangju could be enhanced for isochlorogenic acid A content. Additionally, Huangju and Fubaiju were notable for their high functional-component content across the three years, although their stability was comparatively lower. Therefore, the ASV and the AMMI biplots may serve as valuable tools for assessing the genetic stability of tea chrysanthemums. Further research should integrate more genotypes across multiple locations and employ additional mathematical methods to dissect GEI effects and ultimately identify genetically stable tea chrysanthemum cultivars with high quality and yield.

5. Conclusions

This study found substantial phenotypic variation in total flavonoids (TF), chlorogenic acid (CA), luteoloside (Lut), and isochlorogenic acid A (ICA) across 24 tea chrysanthemum accessions over three years. Variance and AMMI model analyses identified genotypic (G) and genotype-by-environment interaction (GEI) effects as the primary sources of this variation. Based on the integration of AMMI stability value (ASV) and mean content analysis, the optimal genotypes for each functional trait (i.e., Huangju for TF, CA, and Lut, and Chuju for ICA) and the genotypes (i.e., Suju-6, Hongxinju, Wangongju, and Baixiaoxiangju) exhibiting relatively high stability and content for most functional components were identified, making them promising candidates for future breeding and promotion programs.