Chemical profiling and chemotypic diversity of basil (Ocimum basilicum L.) genotypes: a large-scale evaluation of bioactive compounds and antioxidant capacity

Abstract

Background

Basil (Ocimum basilicum L.) is an economically significant medicinal and aromatic species valued for its diverse phytochemical profile and essential oil (EO) composition. Despite its economic importance, there is a critical need to characterize genetic resources to identify stable, high-yielding genotypes adapted to specific environmental conditions. This study aimed to evaluate the phytochemical diversity and industrial potential of 90 basil genotypes under Mediterranean climatic conditions, focusing on identifying superior chemotypes for pharmaceutical and cosmetic applications.

Methods

A field experiment was conducted using an augmented randomized complete block design at the Çukurova University Research and Application Area during the 2023 and 2024 growing seasons. Phytochemical screening included the determination of total phenolic content (TPC) and antioxidant capacities via DPPH (2,2-diphenyl-1-picrylhydrazyl) and ABTS [2,2’-azino-bis (3-ethylbenzothiazoline-6-sulfonic acid)] radical scavenging assays. The chemical composition of the EOs was determined using gas chromatography–mass spectrometry (GC–MS).

Results

Significant biochemical variation was observed among the 90 genotypes. Genotypes MT08, MT86, MT38, and MT67 exhibited the highest bioactive potential, characterized by superior TPC and antioxidant activities. EO yields were notably higher in genotypes MT63 and MT74. Chemical profiling revealed five distinct chemotypes within the population: linalool, methyl chavicol, citral, methyl cinnamate, and methyl eugenol. Pearson correlation, PCA, and cluster analyses supported that these genotypes represent a broad phytochemical basis for selection and breeding.

Conclusions

This study provides a detailed characterization of basil germplasm, identifying specific genotypes with high industrial value. The classification of five major chemotypes offers a strategic foundation for targeted breeding programs aimed at developing high-quality cultivars. These findings contribute to the standardization of raw materials for the global pharmaceutical and fragrance industries.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-026-09147-9.

Introduction

Basil (Ocimum basilicum L.), also known as reyhan in Turkiye, is one of the most prominent medicinal and aromatic plants within the Lamiaceae family, thriving primarily in tropical and subtropical climates due to its distinctive aromatic profile [1, 2]. While it is not a naturally occurring species in the Turkish flora, various cultural forms are extensively cultivated across the country [3–5]. This widespread cultivation reflects basil’s versatility, serving multiple sectors including the pharmaceutical, food, and perfumery industries [6, 7]. For instance, its aromatic leaves, used either fresh or dried, are staple ingredients in numerous culinary applications, most notably in Italian pesto and Thai curries. Interest in culinary and medicinal spices has increased markedly in recent years within developed nations; in the North American market, basil ranks among the most sought-after spices alongside coriander, fennel, and marjoram [8, 9]. Consequently, dried basil leaves, their essential oils (EOs), and chemical derivatives constitute a significant export commodity to European markets [6].

The commercial value of these exports is fundamentally derived from the EOs found in basil leaves, which comprise a wide array of chemical constituents, such as linalool, methyl cinnamate, citral, methyl chavicol (estragole), and eugenol, depending on chemotypic variations [10]. Within this chemical profile, specific components like chavicol and linalool provide potent antiviral and antibacterial properties [11–14]. Furthermore, eugenol, an allylphenolic derivative, imparts a sharp, clove-like spicy aroma and is utilized as a dental analgesic and disinfectant. Methyl chavicol, together with linalool, has been shown to significantly inhibit in vitro fungal growth [15]. Often referred to as estragole, methyl chavicol is also utilized in the cosmetics industry for its fennel- and anise-like scent [16]. However, the industrial use of methyl chavicol is subject to regulatory scrutiny due to its potential genotoxic and carcinogenic effects in high doses [17].

Beyond their industrial utility, these volatile secondary metabolites play crucial ecological roles, acting as allelopathic agents, defense compounds against phytopathogens, and chemical signals for pollinators [18]. From a biological perspective, EOs are synthesized and accumulated within glandular trichomes (GTs), which are specialized epidermal structures primarily located on leaves and floral parts in Ocimum species [19]. Damage to these glandular cells results in the evaporation of volatile compounds such as eugenol, chavicol, methyl chavicol, and methyl eugenol [20]. These are the primary phenylpropenes derived from phenylalanine, which also serves as a precursor in general phenylpropanoid metabolism.

The quality and quantity of these EOs are determined not only by these internal genetic factors but are also heavily influenced by cultivation ecology and agronomic management [21]. Environmental factors, including temperature, precipitation, irrigation, and fertilization, directly impact both herb yield and the synthesis of bioactive compounds. Nevertheless, genetic structure remains a fundamental determinant of yield and quality [2, 8, 21]. Basil exhibits various varieties categorized by leaf color, aroma, and morphology, such as sweet, purple, lemon, cinnamon, anise, and fine-leaf basil [22]. Research indicates significant differences in dry herb yield among various genetic resources, with certain landraces and genebank accessions demonstrating higher productivity than standard commercial cultivars [23]. In this context, studies conducted in Turkiye have demonstrated substantial variations in the EO compositions of local basil genotypes, suggesting a rich gene pool available for breeding programs [24].

To capitalize on this diversity and achieve sustainable basil cultivation, breeders must develop improved cultivars characterized by disease resistance, high EO yield, and resilience to abiotic stressors such as drought and salinity, alongside diverse flavor profiles. Consequently, evaluating basil germplasm for morphology, growth habit, leaf traits, EO content, and resistance to biotic and abiotic stresses is crucial for selecting promising parents for breeding programs. While some genotypes are valued as ornamentals for their attractive purple (such as ‘Dark Opal’ or ‘Purple Ruffles’) or variegated foliage, others are selected for the high EO content (such as Mrs. Burns’ Lemon Basil) required in perfumery and aromatherapy [25–28]. From a commercial standpoint, linalool- and citral-rich types are generally preferred [14].

Building on the need for targeted selection, the southern and western coastal regions of Turkiye, characterized by a typical Mediterranean climate with hot and dry summers, present a challenging yet high-potential ecology for basil cultivation. Despite Turkiye’s significant role in the medicinal and aromatic plant trade, there remains a notable lack of standardized, region-adapted basil genotypes [4, 10, 29]. High temperatures and summer drought stress are decisive factors in determining the adaptation potential of genotypes. This research is based on the hypothesis that the extensive genetic diversity within the selected basil germplasm results in significant variations in secondary metabolite accumulation and essential oil (EO) composition, facilitating the identification of chemotypes capable of maintaining high bioactive potential under Mediterranean environmental stressors. Therefore, the primary objective of this study was the evaluation of phytochemical diversity, specifically total phenolic content and antioxidant capacities, across 90 basil genotypes over two consecutive growing seasons. Furthermore, the study aims to characterize and classify the population into distinct chemotypes based on primary EO constituents via GC–MS analysis to identify rare and industrially significant variants. Additionally, the stability of biochemical and quality traits was assessed to determine the specific impact of Genotype × Environment interactions in a climate change hotspot. Through this comprehensive evaluation, superior genotypes with high industrial value were identified for potential standardization in the global pharmaceutical and fragrance sectors.

Materials and methods

Plant material and experimental procedure

The plant material used in this study consisted of 94 basil (Ocimum basilicum L.) genotypes from diverse geographical origins. These genotypes were obtained from the collections of the United States Department of Agriculture (USDA-ARS, USA), Dicle University Faculty of Agriculture, Çukurova University Faculty of Agriculture, AG Tohum San. ve Tic. A.Ş., and Kemal Cüce Tarım. Taxonomic identity was verified by Prof. Dr. Leyla Sezen Tansi at Çukurova University. To calculate the error variance and facilitate genotypic comparison within the augmented randomized complete block design, four commercial genotypes representing different morphological characteristics were utilized as replicated controls (C1: Mrs. Burns’ Lemon Basil, C2: Morfes, C3: Dino, and C4: Largesweet). The remaining 90 genotypes served as the primary test material. Although 94 genotypes were initially sown, four genotypes (coded MT70, MT82, MT83, and MT90) failed to germinate or emerge during both experimental years (2023 and 2024). Consequently, these four genotypes were excluded from phenological observations and subsequent analyses, and the study proceeded with 90 genotypes. Detailed information regarding the genotype numbers (Gen No), material names, origins, and sources is presented in Table 1.

Table 1.

Genetic identity and geographical origin of the 94 basil accessions

Gen. No.	Accession Name	Plant Name	Origin	Improvement status	Source
C1	Mrs. Burns’ Lemon Basil	-	New Mexico, USA	Cultivar	USDA
C2	Morfes	-	Turkiye	Cultivar	AG Tohum San. ve Tic. A. Ş.
C3	Dino	-	Turkiye	Cultivar	AG Tohum San. ve Tic. A. Ş.
C4	Largesweet	-	Turkiye	Cultivar	Vilmorin Mikado Tohumculuk A.Ş.
MT1	Ames 29,184	GSMO2-19	South Ossetia, Georgia	Landrace	USDA
MT2	Ames 32,309	GE.2013-21	Georgia	Landrace	USDA
MT3	Ames 32,310	GE.2013-37	Georgia	Landrace	USDA
MT4	Ames 32,311	GE.2013-43	Georgia	Landrace	USDA
MT5	Ames 32,312	GE.2013-50	Georgia	Landrace	USDA
MT6	Ames 32,313	GE.2013-68	Georgia	Landrace	USDA
MT7	Ames 32,314	GE.2013-78	Georgia	Landrace	USDA
MT8	PI 170,578	Fesligen	Aydın, Turkiye	Landrace	USDA
MT9	PI 170,579	226-3	İzmir, Turkiye	Landrace	USDA
MT10	PI 170,581	3031	Çanakkale, Turkiye	Landrace	USDA
MT11	PI 172,996	Reyhan	Kars, Turkiye	Landrace	USDA
MT12	PI 172,997	Reyhan	Kars, Turkiye	Landrace	USDA
MT13	PI 172,998	Reyhan	Van, Turkiye	Landrace	USDA
MT14	PI 173,746	Reyhan	Malatya, Turkiye	Landrace	USDA
MT15	PI 174,284	Zetrun	Van, Turkiye	UIS	USDA
MT16	PI 174,285	Reyhan	Elazığ, Turkiye	Landrace	USDA
MT17	PI 175,793	Festagan	Çanakkale, Turkiye	Landrace	USDA
MT18	PI 176,646	Reyhan	Tokat, Turkiye	Landrace	USDA
MT19	PI 182,246	Reyhan	Kahramanmaraş, Turkiye	Landrace	USDA
MT20	PI 190,100	1	Iran	UIS	USDA
MT21	PI 197,442	10,126	Ethiopia	UIS	USDA
MT22	PI 207,498	B-44,178	Kabul, Afghanistan	UIS	USDA
MT23	PI 211,586	12,832	Kunduz, Afghanistan	UIS	USDA
MT24	PI 253,157	Rayhoon	Isfahan, Iran	Landrace	USDA
MT25	PI 263,870	-	Greece	UIS	USDA
MT26	PI 296,390	-	Iran	Landrace	USDA
MT27	PI 296,391	-	Iran	Landrace	USDA
MT28	PI 358,463	Obicen bosilok	North Macedonia	Landrace	USDA
MT29	PI 358,464	Edrolisten	North Macedonia	Landrace	USDA
MT30	PI 358,465	Srednolisten	North Macedonia	Landrace	USDA
MT31	PI 358,466	Krstaten	North Macedonia	Landrace	USDA
MT32	PI 358,467	Domasen siten	North Macedonia	Landrace	USDA
MT33	PI 358,468	Krupnoüsten	North Macedonia	Landrace	USDA
MT34	PI 358,469	Siten	North Macedonia	Landrace	USDA
MT35	PI 358,470	Lokaîen	North Macedonia	Landrace	USDA
MT36	PI 358,471	S’ıtnofcten	North Macedonia	Landrace	USDA
MT37	PI 358,472	Bitolski	North Macedonia	Landrace	USDA
MT38	PI 368,695	Zelen	North Macedonia	Landrace	USDA
MT39	PI 368,697	Vladimirska	North Macedonia	Landrace	USDA
MT40	PI 368,698	Srecno listen	North Macedonia	Landrace	USDA
MT41	PI 368,699	Edar	North Macedonia	Landrace	USDA
MT42	PI 368,700	Edar	North Macedonia	Landrace	USDA
MT43	PI 379,412	Krupen bel	North Macedonia	Landrace	USDA
MT44	PI 379,413	Krupnolsen	North Macedonia	Landrace	USDA
MT45	PI 379,414	Bel kripen	North Macedonia	Landrace	USDA
MT46	PI 414,193	B 49,939	Maryland, USA	UIS	USDA
MT47	PI 414,194	B 49,926	Maryland, USA	UIS	USDA
MT48	PI 414,195	B 49,927	Maryland, USA	UIS	USDA
MT49	PI 414,196	B 49,928	Maryland, USA	UIS	USDA
MT50	PI 414,197	B 49,929	Maryland, USA	UIS	USDA
MT51	PI 414,198	B 49,930	Maryland, USA	UIS	USDA
MT52	PI 414,199	B 49,931	Maryland, USA	UIS	USDA
MT53	PI 414,200	B 49,932	Maryland, USA	UIS	USDA
MT54	PI 531,396	1420	Hungary	Cultivated material	USDA
MT55	PI 584,458	Sweet Dani	USA	Cultivar	USDA
MT56	PI 652,065	Genovese	Veneto, Italy	Cultivar	USDA
MT57	PI 652,070	Sweet Basil	Pennsylvania, USA	Cultivar	USDA
MT58	PI 652,071	Dark Opal	California, USA	Cultivar	USDA
MT59	Arapgir	-	Turkiye	Landrace	Dicle University
MT60	Şeyh Kendir	-	Turkiye	Landrace	Dicle University
MT61	Diyarbakir	-	Turkiye	Landrace	Dicle University
MT62	Iraq	-	Iraq	Landrace	Iraq Spice Dealer
MT63	Limoni	-	Turkiye	Cultivar	AG Tohum San. ve Tic. A. Ş.
MT64	Compact	-	Turkiye	Cultivar	Vilmorin Mikado Tohumculuk A.Ş.
MT65	Midnight	-	Turkiye	Cultivar	Vilmorin Mikado Tohumculuk A.Ş.
MT66	Miracle Saksilik Top F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT67	Arzuman Top F.	-	Turkiye	UIS	Arzuman Tohumculuk
MT68	Arzuman Mor F.	-	Turkiye	UIS	Arzuman Tohumculuk
MT69	Red Basil Boza Kırmızı Tatlı F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT70	Red Freddy Kırmızı Tatlı F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT71	Miracle Mor F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT72	Tatli Geniş Yaprakli	-	Turkiye	UIS	Kemal Cüce Agriculture
MT73	Miracle Tarçın Kokulu F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT74	Miracle Limon Kokulu F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT75	Miracle Marul Yapraklı F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT76	Iraq 2	-	Iraq	UIS	Iraq Spice Dealer
MT77	Miracle Iri Yapraklı Genovese F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT78	Miracle Tatlı Geniş Yapraklı F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT79	Doğal Green Flesh Geniş Yapraklı F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT80	Iri Yapraklı Genovese F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT81	Koyu Opal Siyat Tatlı	-	Turkiye	UIS	Kemal Cüce Agriculture
MT82	Dik Gelişen Limon F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT83	Doğal Lime F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT84	Doğal Küçük Yapraklı Saksılık F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT85	Saksılık Tatlı F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT86	Mor F.	-	Turkiye	UIS	Kemal Cüce Agriculture
MT87	Limon F.	-	Turkiye	UIS	Çukurova University
MT88	Mor F.	-	Turkiye	UIS	Çukurova University
MT89	Tarçın F.	-	Turkiye	UIS	Çukurova University
MT90	Fesleğen	-	Turkiye	UIS	Çukurova University

UIS Uncertain improvement status

The study was conducted at the experimental field site of the Department of Field Crops, Faculty of Agriculture, Çukurova University, during the 2023 and 2024 growing seasons. The experimental area was deep-plowed with a moldboard plow in winter and prepared for planting by disking and harrowing prior to transplanting. A basal fertilizer dose of 2.5 kg/da of DAP (18-46-0) from Toros Tarım was applied to the soil before sowing. Seeds were sown in the greenhouse in trays containing a mixture of soil, peat, and perlite (1:1:1) on March 27, 2023, and March 20, 2024. The same original seed stock from 2023 was used for the 2024 season to ensure genetic consistency. Off-type plants were removed during the seedling stage. Once the seedlings reached the appropriate size, they were transplanted to the field on May 15, 2023, and May 9, 2024. The experiment was laid out in an augmented randomized complete block design consisting of 5 blocks, with the four control genotypes replicated in each block. Each genotype was planted in 3-meter rows with a spacing of 30 × 30 cm (between and within rows). To meet the plant water requirements, initial irrigation was applied immediately after transplanting, followed by a sprinkler irrigation system throughout the vegetation period, depending on plant development and meteorological conditions. Weed control was performed manually using hand hoes twice each year. Harvesting was conducted at the onset of flowering (10–50% flowering) using pruning shears at a height of 10 cm above the soil surface. Based on the length of the vegetation period, three cuttings were obtained in each experimental year. In 2023, the cultivation process began with sowing in trays on March 27, leading to emergence on April 5, and transplanting on May 15. The maintenance phase included the first hoeing on June 19 and the second on July 10. The harvest cycle consisted of three cuttings, which took place on June 23, July 13, and August 7. In contrast, the 2024 season followed a slightly accelerated schedule. Sowing commenced earlier on March 20, with emergence observed by March 28 and transplanting completed by May 9. Hoeing activities were moved forward to June 10 and July 1, respectively. Consequently, the harvest periods were also earlier than the previous year, with the first cutting on June 14, the second on July 5, and the final cutting on July 29.

Analysis of the climatic data for the experimental years (2023 and 2024) and the long-term averages (1984–2024) in Adana reveals a typical Mediterranean climate (Figure S1). A steady increase in temperature (minimum, average, and maximum) was observed from March to August. Notably, the maximum temperatures in July and August during the experimental years (approx. 36 °C) were higher than the long-term average (approx. 35 °C). Regarding precipitation, spring rainfall was followed by drought starting in June, with precipitation levels reaching a minimum during the summer months. Precipitation in April 2023 (approx. 130 mm) was significantly higher than in other years and the long-term average. Soil elemental content and pH were analysed before fertilization. The soil at the experimental site had a clay texture and a slightly alkaline pH of 7.69. Additionally, the soil was characterized by low salinity (0.19 mmhos/cm), high lime content (26.17%), low organic matter (1.39%), low available phosphorus (3.91 kg/da), and high available potassium (75.70 kg/da). Irrigation water pH was 7.2.

Plant extraction

Leaf samples were shade-dried until a constant weight was achieved (approx. 10% residual moisture) and ground to pass through a 0.5 mm mesh sieve. Following a modified method based on Kwee and Niemeyer [30] and Izadiyan and Hemmateenejad [31], 0.2 g of the ground sample was mixed with 20 mL of methanol (Isolab, ≥ 99.9% purity) and extracted in an ultrasonic bath (Caliskan, Turkiye) at 57 °C for 20 min. After filtration (Whatman No. 1), the solvent was evaporated (DLAB RE100-pro) at 40 °C, and stock extracts were prepared by diluting the residue to a final concentration of 3 mg/mL.

Total phenolic content (mg GAE/g)

Total Phenolic Content (TPC) was determined using the Folin-Ciocalteu method according to Slinkard and Singleton [32]. Briefly, 0.5 mL of the plant extract was mixed with 2 mL of diluted (1:9) Folin-Ciocalteu (Sigma-Aldrich) reagent and 1.5 mL of 1% Na₂CO₃ (Sigma-Aldrich) solution. After 30 min of incubation in the dark, absorbance was measured at 765 nm. The results were expressed as Gallic Acid Equivalents (mg GAE/g).

DPPH radical scavenging activity (mg TE/g)

DPPH radical scavenging activity was determined according to Gülçin et al. [33]. One milliliter of plant extract was mixed with 4 mL of methanolic DPPH (0.004% w/v) (Sigma-Aldrich) solution. Following 30 min of incubation in the dark at room temperature, absorbance was measured at 517 nm. The results were expressed as Trolox (Sigma-Aldrich) Equivalents (mg TE/g).

ABTS cation radical scavenging activity (mg TE/g)

Based on the method by Re et al. [34], a radical solution was generated by mixing 7.4 mM ABTS (Sigma-Aldrich) and 2.45 mM potassium persulfate (K₂S₂O₈) (Sigma-Aldrich). The solution was diluted to an absorbance of 0.70 ± 0.02 at 734 nm. 1 mL of extract was mixed with 2 mL of ABTS solution, and absorbance was measured at 734 nm after 30 min. The results were expressed as Trolox (Sigma-Aldrich) Equivalents (mg TE/g).

Essential oil analysis

Dried leaf samples (30 g) were placed in a 1000 mL round-bottom flask containing distilled water. Following the methodology of Yaldiz et al. [35], hydrodistillation was performed using a Clevenger-type apparatus for 3 h. The isolated essential oil, which was collected above the aqueous phase, was measured via the graduated portion of the apparatus. The final yield was calculated as a volume-to-weight% (v/w) based on the dry tissue mass. Prior to GC–MS analysis, EOs were dried over anhydrous sodium sulfate, transferred to amber glass vials with PTFE-lined caps, and stored at 4 °C in the dark.

Gas chromatography–mass spectrometry (GC–MS) analysis

The chemical composition of the essential oil was analyzed using an Agilent GC–6890 II series gas chromatograph coupled with an Agilent 5975 C mass spectrometer at Kahramanmaras Sutcu Imam University. Briefly, 10 µL of essential oil was diluted in 250 µL of dichloromethane, with 1 µL of the mixture injected into an HP-88 capillary column (100 m × 250 μm × 0.20 μm). Helium was used as the carrier gas at a constant flow of 1.0 mL/min. The oven program started at 70 °C (1 min), increased to 220 °C at 10 °C/min (10 min hold), and finally reached 230 °C at 10 °C/min (10 min hold). The injector temperature was maintained at 250 °C with a 20:1 split ratio. The MS operated in EI mode (70 eV) scanning a range of 35–400 m/z. Retention indices (RI) were determined relative to a homologous series of n-alkanes (C₈-C₄₀) injected under identical chromatographic conditions. Compounds were identified by comparing their mass spectra against the Flavor2, HPCH1607, and Wiley7Nist05 libraries. GC–MS analysis was performed using two analytical replicates.

Statistical analysis

Statistical analyses were performed according to the augmented randomized complete block design principles described by Federer [36]. For each year, analysis of variance (ANOVA) was conducted for each measured trait using a model including genotype and block effects, with the replicated check cultivars used to estimate experimental error. For the combined analysis across years, year, genotype, and year × genotype interaction effects were included in the model. Differences between replicated controls and unreplicated genotypes were determined using specific standard error formulations. Means were compared using the Least Significant Difference (LSD) for general differences and the Least Significant Increase (LSI) to identify genotypes superior to controls according to Peterson [37]. Relationships between traits were analyzed via correlation using the metan package [38]. Pearson correlation analysis was conducted on the pooled genotype dataset to identify overall trait associations across the evaluated germplasm panel. These correlations are interpreted as panel-level co-variation patterns, not genotype-specific mechanistic relationships. In addition, Variance Inflation Factor (VIF) values were calculated for the main biochemical and essential oil-related variables as a diagnostic indicator of redundancy among variables. VIF values were interpreted according to commonly used thresholds, with values above 5 indicating moderate multicollinearity and values above 10 indicating severe multicollinearity. Since no predictive multiple regression model was used to estimate response factors from correlated metabolites, VIF analysis was used only as a diagnostic and interpretive tool. Highly correlated variables were retained in PCA and HCA because these methods were used for exploratory chemotypic classification and dimensionality reduction, but their biological interpretation was made cautiously. This approach follows the general recommendations for diagnosing multicollinearity in multivariate datasets as described by Montgomery et al. [39]. Principal component analysis (PCA) was performed on the column-standardized dataset to evaluate multivariate relationships among genotypes based on the measured quality traits. PCA and biplot visualization were generated using JMP (Version 17.0, SAS Institute Inc.). The first two principal components were interpreted based on their explained variance and trait loadings. Hierarchical clustering was conducted using Ward’s linkage on column-standardized trait data in the same software. To visualize chemical profiles, a profile network of major EO constituents was generated using tidyverse [40], igraph [41], and ggraph [42] packages in R (Version 4.3.0, R Foundation for Statistical Computing) with the viridis [43] color palette. Standard errors (SE) for comparisons in the augmented randomized complete block design were calculated as follows:

Between two controls:

Between two genotypes in the same block:

Between two genotypes in different blocks:

Between a control and a genotype:

Combined analysis over the years:

Results

Biochemical profiling

The analysis of variance (ANOVA) for quality traits of the evaluated basil genotypes is presented in Tables 2 and 3. Significant genotypic differences were detected for total phenolic content (TPC), antioxidant activities (DPPH and ABTS), and leaf essential oil (EO) content in both growing seasons. The combined analysis also showed significant year effects and significant year × genotype interactions for most antioxidant-related traits. For EO content, the year effect was non-significant for the first cutting but significant for the second cutting, whereas the year × genotype interaction was non-significant in the first cutting and significant in the second cutting. Comparisons between test genotypes and check varieties revealed significant variation in biochemical and EO-related traits (Table 4).

Table 2.

Analysis of variance (ANOVA) results for biochemical/antioxidant-related traits (TPC, DPPH, ABTS) of basil genotypes

Traits	Source of Variation	df	2023 Growing Season		2024 Growing Season		df (Comb.)	Combined Years
Total Phenolic Content	Year	-	-		-		1	10172.99	**
	Genotype	89	212.48	**	94.18	**	89	164.11	**
	Genotype: Check	3	652.56	**	241.25	**	-	-
	Genotype: Test vs. Check	1	0.03	ns	250.2	**	-	-
	Genotype: Test	85	199.45	*	87.15	**	-	-
	Year × Genotype	-					89	99.23	**
	Block	4	187.15	ns	87.57	*	8	137.36	**
	Error	12	57.99		21.92		24	39.96
Antioxidant Activity (DPPH Assay)	Year	-	-		-		1	22354.96	*
	Genotype	89	624.7	**	622.63	**	89	744.12	*
	Genotype: Check	3	1791.57	**	3682.84	**	-	-
	Genotype: Test vs. Check	1	656.5	ns	1673.93	**	-	-
	Genotype: Test	85	583.14	*	502.25	**	-	-
	Year × Genotype	-					89	324.24	*
	Block	4	679.22	*	763.77	**	8	721.49	**
	Error	12	171.17		133.89		24	152.53
Antioxidant Activity (ABTS Assay)	Year	-	-		-		1	12450.23	**
	Genotype	89	263.82	**	357.95	**	89	274.53	**
	Genotype: Check	3	402.62	*	407.22	*	-	-
	Genotype: Test vs. Check	1	503.96	*	3742.15	**	-	-
	Genotype: Test	85	256.09	*	316.39	**	-	-
	Year × Genotype	-					89	148.12	*
	Block	4	263.29	*	118.48	ns	8	190.89	*
	Error	12	75.52		83.22		24	79.37

Degrees of freedom, ^ns non-significant (P > 0.05), ^*Significant at P < 0.05, ^**Significant at P < 0.01, Check: Standard control varieties, Test: Evaluated basil genotypes. Values in the table represent the Mean Squares (MS)

Table 3.

Analysis of variance (ANOVA) results for the leaf EO content of basil genotypes

Source of Variation	df	2023 Growing Season				2024 Growing Season				df (Comb.)	Combined Years
		1st Cutting		2nd Cutting		1st Cutting		2nd Cutting			1st Cutting		2nd Cutting
Year	-	-		-		-		-		1	0.18	ns	0.48	*
Genotype	89	0.59	*	0.39	**	0.33	**	0.33	*	89	0.77	**	0.53	**
Genotype: Check	3	7.33	**	1.2	**	1.29	**	0.83	**	-	-		-
Genotype: Test vs. Check	1	3.29	**	0.003	ns	0.17	ns	0.14	ns	-	-		-
Genotype: Test	85	0.32	ns	0.36	**	0.29	**	0.32	*	-	-		-
Year × Genotype	-									89	0.12	ns	0.17	*
Block	4	0.13	ns	0.14	ns	0.11	ns	0.17	ns	8	0.12	ns	0.15	ns
Error	12	0.18		0.05		0.04		0.13		24	0.11		0.09

Degrees of freedom; ns: non-significant (P > 0.05), *Significant at P < 0.05, **Significant at P < 0.01, Check: Standard control varieties, Test: Evaluated basil genotypes. Values in the table represent the Mean Squares (MS)

Table 4.

Adjusted mean values for quality traits of basil genotypes

Genotypes	TPC (mg GAE/g)			DPPH Assay (mg TE/g)			ABTS Assay (mg TE/g)
	2023	2024	Mean	2023	2024	Mean	2023	2024	Mean
C1	59.67	35.16	47.42	128.90	95.55	112.22	107.15	93.80	100.48
C2	36.13	38.52	37.33	89.62	112.94	101.28	88.85	93.89	91.37
C3	39.07	27.40	33.24	98.40	64.36	81.38	94.52	84.42	89.47
C4	35.82	23.39	29.60	88.32	54.47	71.39	87.46	75.00	81.23
MT01	46.08	24.25	35.17	95.30	63.60	79.45	87.63	66.03	76.83
MT02	41.59	19.48	30.53	75.69	45.32	60.51	54.04	53.27	53.65
MT03	43.30	18.54	30.92	70.79	51.38	61.09	60.31	49.93	55.12
MT04	42.41	22.78	32.60	63.15	62.75	62.95	40.49	57.42	48.96
MT05	61.01	22.63	41.82	88.05	60.57	74.31	74.79	61.76	68.27
MT06	57.07	25.09	41.08	82.36	66.82	74.59	72.36	57.99	65.18
MT07	71.98	23.69	47.83	113.15	60.66	86.90	90.23	56.51	73.37
MT08	113.26	29.92	71.59	149.81	90.11	119.96	89.65	74.77	82.21
MT09	50.19	21.38	35.79	85.30	56.50	70.90	75.73	47.38	61.56
MT10	35.06	20.70	27.88	53.34	52.33	52.84	39.77	45.44	42.60
MT11	38.44	25.60	32.02	90.28	74.68	82.48	87.78	57.54	72.66
MT12	41.09	27.91	34.50	100.08	77.23	88.66	94.25	55.71	74.98
MT13	39.14	36.35	37.74	93.07	115.78	104.42	89.79	74.31	82.05
MT14	46.28	29.23	37.76	102.87	80.36	91.62	93.69	72.26	82.97
MT15	41.67	27.43	34.55	95.46	79.51	87.49	89.63	59.36	74.49
MT16	42.60	26.74	34.67	95.35	70.98	83.17	90.48	64.15	77.32
MT17	30.18	23.92	27.05	68.40	60.47	64.44	89.26	54.00	71.63
MT18	34.49	25.55	30.02	78.14	73.26	75.70	79.24	56.05	67.65
MT19	72.82	35.90	54.36	113.32	105.43	109.37	84.79	86.81	85.80
MT20	38.24	29.22	33.73	72.46	89.24	80.85	71.78	70.72	71.25
MT21	34.51	25.96	30.24	54.38	76.74	65.56	86.53	67.86	77.20
MT22	50.65	24.92	37.78	101.17	84.88	93.03	88.64	64.10	76.37
MT23	47.26	31.05	39.15	102.91	89.62	96.26	91.79	78.36	85.07
MT24	55.03	24.01	39.52	115.82	75.41	95.62	99.40	56.00	77.70
MT25	63.21	26.37	44.79	135.88	75.22	105.55	111.75	60.56	86.16
MT26	44.68	25.13	34.90	96.68	73.42	85.05	89.61	61.36	75.49
MT27	52.14	37.14	44.64	115.82	106.85	111.34	100.32	89.09	94.71
MT28	50.83	28.63	39.73	121.32	83.93	102.63	106.75	80.99	93.87
MT29	44.30	36.45	40.38	100.90	111.11	106.00	92.56	92.63	92.60
MT30	19.05	23.10	21.08	42.93	65.47	54.20	60.32	61.02	60.67
MT31	26.55	24.44	25.49	47.42	71.53	59.47	57.77	65.47	61.62
MT32	39.73	26.11	32.92	84.05	73.23	78.64	81.47	70.03	75.75
MT33	43.34	25.16	34.25	92.65	72.95	82.80	85.88	66.49	76.19
MT34	45.39	22.07	33.73	94.58	65.75	80.17	87.70	56.22	71.96
MT35	36.16	27.60	31.88	82.12	82.99	82.56	82.33	70.94	76.64
MT36	40.01	20.74	30.37	92.90	60.92	76.91	89.45	52.80	71.12
MT37	84.45	27.51	55.98	136.77	81.26	109.02	97.40	75.48	86.44
MT38	69.21	37.56	53.39	135.30	111.09	123.19	106.16	94.65	100.41
MT39	32.23	31.88	32.06	93.44	94.90	94.17	94.99	85.52	90.26
MT40	31.59	31.75	31.67	86.48	88.93	87.70	90.00	80.84	85.42
MT41	24.51	25.22	24.87	61.87	73.30	67.59	73.06	73.20	73.13
MT42	38.70	32.06	35.38	103.34	89.97	96.66	98.44	87.58	93.01
MT43	28.99	27.74	28.36	90.79	77.09	83.94	94.76	77.53	86.15
MT44	21.82	23.70	22.76	52.07	55.98	54.02	68.37	64.53	66.45
MT45	17.65	23.31	20.48	51.97	56.26	54.11	66.95	65.32	66.14
MT46	24.64	29.02	26.83	69.42	81.16	75.29	86.03	78.45	82.24
MT47	33.03	26.04	29.54	101.48	66.39	83.93	101.48	74.68	88.08
MT48	43.85	17.12	30.48	125.50	42.15	83.82	115.49	56.42	85.96
MT49	37.72	12.15	24.93	106.87	27.66	67.27	102.72	40.11	71.41
MT50	25.13	16.04	20.58	74.12	40.54	57.33	83.62	57.79	70.71
MT51	22.74	12.28	17.51	79.91	32.02	55.96	90.34	42.84	66.59
MT52	22.58	14.83	18.70	59.22	35.90	47.56	70.74	51.52	61.13
MT53	23.04	25.17	24.10	77.26	65.45	71.35	87.02	72.06	79.54
MT54	27.64	21.23	24.43	70.50	66.30	68.40	93.01	68.86	80.93
MT55	44.63	26.05	35.34	126.12	98.99	112.55	102.40	72.20	87.30
MT56	56.73	20.34	38.54	144.94	55.64	100.29	122.92	67.64	95.28
MT57	52.60	24.41	38.50	132.98	63.07	98.03	115.24	76.99	96.12
MT58	50.57	25.44	38.01	135.63	62.87	99.25	120.19	77.56	98.88
MT59	45.06	43.11	44.08	121.61	105.64	113.63	105.09	95.71	100.40
MT60	34.24	28.78	31.51	92.00	76.12	84.06	91.76	89.20	90.48
MT61	35.38	12.19	23.78	97.39	26.43	61.91	95.79	44.36	70.07
MT62	46.41	22.92	34.66	126.22	65.48	95.85	112.92	73.57	93.24
MT63	33.79	24.33	29.06	99.35	64.68	82.02	99.22	73.00	86.11
MT64	44.05	57.02	50.53	107.78	113.57	110.68	98.92	95.71	97.31
MT65	41.95	35.80	38.87	101.09	83.55	92.32	93.81	94.00	93.90
MT66	37.91	30.81	34.36	93.42	74.12	83.77	91.22	94.34	92.78
MT67	49.39	40.69	45.04	135.81	105.84	120.83	114.03	95.82	104.93
MT68	41.09	32.92	37.00	111.74	87.07	99.40	106.24	95.48	100.86
MT69	38.18	28.75	33.47	101.74	77.43	89.58	97.87	82.70	90.28
MT71	32.12	22.07	27.09	83.09	54.94	69.01	85.79	69.58	77.69
MT72	37.71	29.65	33.68	92.07	66.08	79.08	89.47	72.20	80.84
MT73	38.05	34.59	36.32	84.20	75.30	79.75	82.42	91.86	87.14
MT74	45.48	29.71	37.60	104.13	73.07	88.60	95.20	75.80	85.50
MT75	40.15	24.21	32.18	87.91	59.73	73.82	85.09	69.74	77.42
MT76	42.74	42.15	42.45	99.20	101.60	100.40	93.50	91.97	92.74
MT77	59.01	37.86	48.43	85.77	86.44	86.11	70.40	92.08	81.24
MT78	31.37	27.80	29.59	63.42	55.12	59.27	69.13	90.83	79.98
MT79	34.80	30.43	32.62	68.13	62.15	65.14	89.91	91.74	90.83
MT80	37.10	35.24	36.17	91.85	72.09	81.97	77.73	91.63	84.68
MT81	52.52	38.66	45.59	125.26	98.69	111.97	108.88	91.40	100.14
MT84	38.15	23.84	30.99	83.57	46.68	65.12	72.15	83.18	77.67
MT85	35.15	23.07	29.11	72.26	40.86	56.56	76.45	76.22	76.34
MT86	64.15	47.27	55.71	145.89	106.52	126.20	119.45	91.97	105.71
MT87	39.15	24.66	31.91	91.89	56.32	74.10	88.27	74.40	81.34
MT88	48.19	33.59	40.89	103.26	68.77	86.01	93.24	80.22	86.73
MT89	36.23	19.47	27.85	78.96	43.37	61.16	79.43	53.29	66.36
Mean	42.49	27.58	35.03	94.87	72.77	83.82	89.01	72.51	80.76
Min.	17.65	12.15	17.51	42.93	26.43	47.56	39.77	40.11	42.60
Max.	113.26	57.02	71.59	149.81	115.78	126.20	122.92	95.82	105.71
SE1	4.82	2.96	4.00	8.27	7.32	7.81	5.50	5.77	5.63
SE2	10.77	6.62	8.94	18.50	16.36	17.47	12.29	12.90	12.60
SE3	12.04	7.40	9.99	20.69	18.30	19.53	13.74	14.42	14.09
SE4	9.33	5.73	7.74	16.02	14.17	15.13	10.64	11.17	10.91
LSD1	10.49	6.45	8.25	18.03	15.94	16.12	11.98	12.57	11.63
LSD2	23.46	14.43	18.45	40.31	35.65	36.05	26.78	28.11	26.00
LSD3	26.23	16.13	20.63	45.07	39.86	40.30	29.94	31.43	29.07
LSD4	20.32	12.49	15.98	34.91	30.88	31.22	23.19	24.34	22.52
LSI	16.62	10.22	13.24	28.56	25.26	25.88	18.97	19.91	18.67
C1Mean + LSI	76.29	45.38	60.66	157.45	120.81	138.10	126.12	113.71	119.15
C2Mean + LSI	52.75	48.74	50.57	118.18	138.19	127.16	107.82	113.81	110.04
C3Mean + LSI	55.69	37.62	46.48	126.96	89.61	107.26	113.49	104.33	108.14
C4Mean + LSI	52.44	33.61	42.85	116.88	79.73	97.27	106.43	94.91	99.90
CF1.Block (MT01-MT18)	4.85	6.13	5.49	5.99	15.85	10.92	2.15	6.30	4.23
CF2.Block(MT19-MT36)	-2.53	3.22	0.34	-13.28	9.96	-1.66	-9.76	4.99	-2.39
CF3.Block(MT37-MT54)	-10.76	-1.90	-6.33	-13.91	-6.45	-10.18	-5.08	-2.12	-3.60
CF4.Block(MT55-MT72)	2.50	-1.68	0.41	15.53	-0.08	7.72	11.69	-2.66	4.52
CF5.Block(MT73-MT89)	5.94	-5.77	0.09	5.68	-19.28	-6.80	1.00	-6.51	-2.76

Thresholds for selection were calculated as Check Mean + LSI

SE_1–4 Standard errors for various comparisons, LSD_1–4 Least significant differences at P < 0.05, CF Correction Factor, Checks (C1-C4) Standard control varieties, Test Genotypes (MT) Evaluated basil genotypes

In the 2023 growing season, the highest TPC value among the controls was recorded in C1, giving a selection threshold of 76.29 mg GAE/g. Under these conditions, MT08 (113.26 mg GAE/g) and MT37 (84.45 mg GAE/g) exceeded the threshold. In 2024, the threshold was determined from C2 as 48.74 mg GAE/g, and only MT64 (57.02 mg GAE/g) exceeded this level. Based on the two-year average, MT08 had the highest mean TPC (71.59 mg GAE/g) and was the only genotype above the combined threshold of 60.66 mg GAE/g.

For DPPH radical scavenging activity, the 2023 threshold was 157.45 mg TE/g based on C1 (128.90 mg TE/g). Although no genotype exceeded this threshold, MT08 (149.81 mg TE/g), MT86 (145.89 mg TE/g), and MT56 (144.94 mg TE/g) showed the highest values among the test genotypes. In 2024, MT13 (115.78 mg TE/g) and MT64 (113.57 mg TE/g) were the best-performing genotypes relative to C2 (112.94 mg TE/g). In the two-year mean, MT86 (126.20 mg TE/g), MT38 (123.19 mg TE/g), and MT08 (119.96 mg TE/g) had the highest DPPH values.

For ABTS radical scavenging activity, the 2023 threshold was 126.12 mg TE/g, with C1 showing 107.15 mg TE/g. MT56 (122.92 mg TE/g), MT58 (120.19 mg TE/g), and MT86 (119.45 mg TE/g) had the highest values in 2023. In 2024, MT67 (95.82 mg TE/g) and MT64 (95.71 mg TE/g) exceeded the check standards. In the two-year average, MT86 (105.71 mg TE/g) and MT67 (104.93 mg TE/g) were the most prominent genotypes.

Regarding EO content, in the first cut of 2023 the highest values were obtained from MT87 (3.36%) and MT74 (3.25%) (Table S1). In the second cut of 2023, MT20 reached the maximum EO content with 3.76%, exceeding the 2.71% selection threshold. In 2024, MT74 and MT87 remained among the leading genotypes in the first cut, whereas MT63 (3.35%) showed the highest EO content in the second cut. Across all observations, the highest single EO value was 3.76% in MT20 during the second cut of 2023.

Essential oil composition

The chemical composition of the essential oils (EOs) obtained from the shade-dried leaves of the basil genotypes was characterized via GC–MS, with detailed constituent data and genotype information presented in Table S2 and Table S3. Analysis of the EO profiles revealed that the population is structured into five distinct chemotypes based on primary component variations. The most dominant group was the Linalool chemotype, accounting for the majority of the population (n = 57). Considering the floral, fresh, and slightly spicy aromatic character of linalool, this distribution indicates that the sensory profile of the studied population is predominantly clustered in a floral direction.

The second-largest group, the Methyl Chavicol chemotype, encompassed 24 genotypes and is distinguished by its sweet, sharp, and characteristic aroma reminiscent of anise. The presence of other primary chemotypes, Methyl Cinnamate, Citral-Neral, and Methyl Eugenol, in a more limited number of genotypes reveals the existence of rare but aromatically significant variants within the population. These minority chemotypes, carrying cinnamon-like (methyl cinnamate), lemon-like (citral-neral), and warm spicy notes (methyl eugenol), hold strategic importance for the development of unique aromatic types in selection and breeding programs. The distinction between these clusters, as visualized in the network profile (Fig. 1), confirms that the chemical fingerprints of the genotypes possess rich chemotypic diversity.

Fig. 1.

Profile network of major EO constituents across basil genotypes (Sub-chemotypes are defined by components exceeding a 10% threshold)

Hierarchical cluster analysis of genotypes based on quality traits

The hierarchical cluster analysis (HCA) and the generated heatmap, conducted to evaluate the similarities and differences among genotypes in terms of EO composition and antioxidant capacity, are presented in Fig. 2. The dendrogram revealed that the examined genotypes exhibit distinct biochemical variations and are partitioned into several major sub-clusters based on their chemical profiles. Upon examining the heatmap color scale, it is observed that the antioxidant parameters, specifically Total Phenolics (TP), DPPH, and ABTS radical scavenging activities, are interrelated and exhibit a highly synchronized distribution pattern, with the highest values concentrated in the central and middle-right clusters. In terms of chemical markers, the genotypes on the far-left (such as C1, MT63, and MT74) are characterized by the highest concentrations of Citral, Neral, and EOC. The central groups are predominantly defined by elevated levels of Eucalyptol, Linalool, and Methyl chavicol. Conversely, the cluster on the far-right (including C3, C4, MT31, and MT35) is distinguished by superior values of Methyl cinnamate and Eugenol. This metabolic fingerprinting provides significant data for identifying specific chemotypes with potentially valuable chemical characteristics for future breeding programs and for characterizing the chemical variation within the population.

Fig. 2.

Hierarchical cluster analysis and heatmap illustrating the classification of genotypes based on agronomic and quality traits (EOC: Essential Oil Content, TP: Total Phenolic Content, DPPH: DPPH radical scavenging activity, ABTS: ABTS radical scavenging activity

Principal component analysis (PCA) of genotypes based on quality traits

The PCA Biplot, conducted to visualize the multidimensional relationships between the examined chemical traits and genotypes, is presented in Fig. 3. The first two principal components (PC1 and PC2) explained 43.634% of the total variation, with PC1 accounting for 22.8% and PC2 for 20.8%. The corresponding eigenvalues were 3.2 for PC1 and 2.9 for PC2, respectively, and the complete eigenvalue structure is provided in Table S4. Upon examining the directions of the vectors and the angles between them, distinct groupings among the traits are evident. Specifically, the narrow angles between antioxidant parameters, such as TP (Total Phenolics), DPPH, and ABTS, indicate a very strong positive correlation among these activities. However, the close positioning of TP, DPPH, and ABTS vectors also indicates partial statistical redundancy among antioxidant-related variables. Therefore, these traits were interpreted as complementary indicators of a shared antioxidant-response axis rather than as completely independent sources of variation. Similarly, Citral, Neral, and EOC (Essential Oil Content) form a separate cluster in the upper-left quadrant, showing a significant relationship between these components. In contrast, terpenoid compounds like Linalool, Eucalyptol, and tau-Cadinol are grouped on the right side of the plot, representing a different chemical profile. Regarding the genotypes, those situated in the upper-left section, such as MT55, MT74, and MT24, are characterized by high concentrations of Citral and Neral. Genotypes like MT78, MT80, and C4 on the right side are distinguished by their high Linalool and tau-Cadinol content, while those in the lower-left quadrant, including MT11, MT12, and MT75, are primarily associated with Methyl chavicol content.

Fig. 3.

Principal Component Analysis (PCA) Biplot illustrating the distribution of genotypes based on major EO constituents and antioxidant activities

Pearson correlation matrix of the investigated quality traits

The results of the Pearson correlation analysis, visualized as a heatmap in Fig. 4, reveal several highly significant relationships among the examined traits. A very strong and statistically significant positive correlation was observed between Neral and Citral (r = 0.91, p < 0.001), indicating that these isomers accumulate in close synchronization. Regarding biochemical activities, strong positive correlations were identified between TP and DPPH (r = 0.83, p < 0.001) and TP and ABTS (r = 0.79, p < 0.001), suggesting that phenolic compounds are major contributors to antioxidant capacity in this germplasm panel, while also indicating partial redundancy among antioxidant-related variables. Furthermore, EO constituents such as Eugenol and Eucalyptol (r = 0.49, p < 0.001) and Linalool and tau-Cadinol (r = 0.45, p < 0.001) showed significant positive associations. In contrast, noteworthy negative relationships were detected, most notably between Linalool and Methyl chavicol (r = -0.75, p < 0.001), as well as between Methyl chavicol and tau-Cadinol (r = -0.43, p < 0.001). These negative correlations may reflect alternative chemotypic backgrounds or differential pathway flux among genotypes rather than direct biochemical antagonism. Therefore, these associations should be interpreted as population-level co-variation patterns that require further targeted biochemical or transcriptomic validation.

Fig. 4.

Pearson correlation matrix illustrating the direction and significance level of the relationships between the major EO constituents and antioxidant activities (EOC: Essential oil content, TP: Total phenolics, DPPH and ABTS: Radical scavenging activities)

Multicollinearity assessment among biochemical and essential oil traits

Variance Inflation Factor (VIF) analysis was performed to diagnose multicollinearity among the biochemical and essential oil-related traits used in the multivariate interpretation. The results indicated that several variables exhibited considerable statistical interdependence (Table S5). Severe multicollinearity was detected for Methyl chavicol (VIF = 25.531), Linalool (VIF = 14.648), DPPH (VIF = 14.121), and Citral (VIF = 12.227), whereas Neral (VIF = 7.349), ABTS (VIF = 7.089), and TP (VIF = 6.037) showed moderate multicollinearity. In contrast, Methyl cinnamate, Eucalyptol, tau-Cadinol, Eugenol, Methyl eugenol, Geraniol, and EOC had VIF values below 5, indicating no serious multicollinearity.

The high VIF values observed for DPPH, ABTS, and TP indicate partial statistical redundancy among antioxidant-related variables. Therefore, these parameters were interpreted as complementary indicators of a shared antioxidant-response axis rather than as fully independent traits. Similarly, the elevated VIF values for Methyl chavicol, Linalool, Citral, and Neral reflect strong covariance among major chemotypic constituents, probably arising from coordinated or alternative accumulation patterns within the essential oil profiles. Accordingly, PCA and hierarchical clustering were used as exploratory dimension-reduction and classification approaches to summarize the main patterns of trait covariation, while biological interpretations based on highly collinear variables were made cautiously.

Discussion

The significant genotype, year, and genotype × year effects detected for most biochemical traits indicate that antioxidant capacity and EO content in basil are controlled by both genetic background and environmental conditions. This pattern suggests that superior performance in a single season may not necessarily translate into stable expression across years; however, because the present study was conducted at a single location, further multi-location testing is required to confirm genotype stability across contrasting environments.

The marked seasonal fluctuation observed for TPC, particularly the decline in several high-performing genotypes from 2023 to 2024, may reflect environmental sensitivity during secondary metabolite biosynthesis. Variations in solar radiation, temperature, or other seasonal factors could have influenced phenolic accumulation, which is consistent with previous reports emphasizing the environmental plasticity of phenolic metabolism in basil and related Ocimum species [8, 11]. Despite this fluctuation, MT08 remained the most promising genotype in terms of two-year average TPC, indicating its potential as a candidate for phenolic-rich basil improvement. The TPC values obtained here are within the broad range reported in the literature, although MT08 exceeded the values reported for ‘Genovese’ by Mounira et al. [44] and also surpassed the range reported by Romano et al. [45] for Italian-origin materials. However, higher values have also been reported in Turkish basil by Mulugeta et al. [46], confirming the wide biochemical variability of basil germplasm.

The DPPH and ABTS results further support the presence of substantial variation in antioxidant capacity among the evaluated genotypes. The consistency between DPPH and ABTS assays and their strong association with TPC suggests that phenolic compounds are important contributors to antioxidant activity in this material. However, because DPPH and ABTS evaluate related radical-scavenging mechanisms and are both strongly associated with phenolic accumulation, these variables should not be interpreted as fully independent antioxidant indicators. Instead, they likely represent a shared antioxidant-response axis, with TPC acting as a major but not exclusive biochemical contributor. Therefore, the correlation structure was interpreted as evidence of coordinated variation among antioxidant-related traits rather than direct causal proof. Genotypes such as MT86, MT08, MT67, and MT64 appear particularly valuable because they combined relatively high antioxidant values with better performance across years. The magnitude of antioxidant activity observed in the present material is comparable to previously reported variability in basil, where values differ widely depending on genotype, extraction method, and assay conditions [47–49]. From a breeding perspective, these findings indicate that targeted selection can improve antioxidant-related quality traits in basil.

EO content is a major commercial trait in basil, and the present results revealed both high-performing and seasonally unstable genotypes. MT20 showed the highest single EO content (3.76%), indicating strong production potential under favorable conditions, whereas MT74, MT87, and MT63 displayed more repeatable performance across cuts or years. These values exceed the EO levels commonly reported for many commercial or local basil populations [50, 51] and are also competitive with elite breeding materials described in previous studies [20, 23, 52, 53]. The peak EO level recorded for MT20 approaches the upper range reported for the species by Pavan et al. [54], suggesting that this germplasm contains highly promising material for industrial EO production.

The chemotype analysis demonstrated that the population is dominated by linalool-rich genotypes, followed by methyl chavicol-rich types, with smaller groups representing methyl cinnamate, citral-neral, and methyl eugenol chemotypes. This distribution is important because chemotype determines aroma profile and end-use suitability in basil. The predominance of the linalool chemotype indicates that much of the population may be suitable for floral and fresh aromatic markets, whereas methyl chavicol, citral-rich, and methyl cinnamate types may be valuable for more specialized flavor and fragrance applications. The observed chemotypic structure is consistent with previous reports that identified the linalool/methyl chavicol axis as a major pattern of basil chemical diversity [55, 56]. Likewise, the occurrence of high linalool and methyl chavicol concentrations agrees with earlier reports from diverse basil germplasm [57–59], while the citral-rich types are comparable to lemon basil profiles described in the literature [60].

The observed divergence between phenylpropanoid-dominant and terpenoid-dominant chemotypes may be explained by differences in underlying biosynthetic pathways. Phenylpropenes such as eugenol, methyl chavicol, and methyl eugenol are derived from phenylalanine through cinnamic acid metabolism, whereas linalool and citral arise mainly through terpene biosynthesis pathways involving the MVA and MEP routes. Therefore, the clear chemotypic separation found in this study likely reflects differential pathway flux and enzyme activity among genotypes, as previously described for basil essential oil biosynthesis [61, 62].

The multivariate analyses supported the single-trait findings. In the HCA and PCA, antioxidant-related traits clustered together, confirming their close association, while EO components formed distinct groups corresponding to different chemical types. The grouping of Citral, Neral, and EO content suggests that citral-rich types may also express favorable oil accumulation patterns. In contrast, linalool-, eucalyptol-, and tau-cadinol-associated genotypes formed a separate chemical profile. These patterns demonstrate that the population includes multiple quality ideotypes that could be exploited depending on breeding objectives, whether for antioxidant enrichment, EO yield, or specific aroma composition.

The correlation analysis reinforced these interpretations. The very strong positive relationship between Neral and Citral is expected because they are closely related isomeric components. Similarly, the strong positive associations of TP with both DPPH and ABTS confirm that phenolic accumulation is a central determinant of antioxidant capacity in this basil material. The strong negative correlation between Linalool and Methyl chavicol suggests that these compounds tend to dominate alternative chemical backgrounds rather than co-accumulate at high levels. Even so, these correlations should be interpreted cautiously at the biological level, because genotype-specific relationships may require larger replicated datasets for more robust mechanistic inference. The evaluated basil population contains biochemical and chemotypic diversity. Certain genotypes stand out for high phenolic content and antioxidant activity (especially MT08 and MT86), whereas others are notable for EO productivity or distinct chemotype profiles (such as MT20, MT63, MT74, and MT87). This diversity provides a strong basis for selecting parental material suited to different breeding targets in medicinal, nutraceutical, flavor, and EO-oriented basil improvement programs.

Conclusions

In this study, conducted under Mediterranean climatic conditions (Adana) during the 2023 and 2024 growing seasons, the quality traits, EO productivity, and chemotypic diversity of 90 different basil genotypes were examined. The results obtained and the recommendations based on these findings are presented below:

• Significant variation in phytochemical capacity was detected among the genotypes, though secondary metabolite accumulation was strongly influenced by environmental fluctuations, as evidenced by the high significance of the “Year” and “Year × Genotype” interactions. Despite a general decline in phenolic levels in the second year, the MT08 genotype consistently exceeded the calculated breeding potential thresholds (LSI), maintaining the highest two-year average for total phenolic content (71.59 mg GAE/g). Similarly, MT86 and MT38 emerged as the most reliable sources of antioxidant activity (DPPH and ABTS). These genotypes are recommended as candidate genotypes for further nutraceutical evaluation.

• Exceptional EO productivity was a hallmark of the evaluated “MT” population. The genotypes exhibited a wide variation in EO content, reaching as high as 3.76% in MT20, which approaches the known biological upper limits of the species. Elite genotypes such as MT20, MT63, MT87, and MT74 significantly outperformed standard commercial varieties (0.5–1.5%) across multiple harvests and years. These high-yielding genotypes should be prioritized for variety registration and industrial-scale oil extraction, as they possess the genetic architecture required for superior terpene and phenylpropene biosynthesis.

• Rich chemotypic polymorphism was observed, with the population clustering into five distinct chemical fingerprints. While the linalool chemotype (57 genotypes) dominated the population, signifying a floral and aromatic sensory profile, the presence of rare citral (lemon-scented), methyl cinnamate (cinnamon-scented), and methyl eugenol (clove-scented) types provides strategic opportunities for niche market development. It is recommended that the citral-rich genotypes be utilized in the aromatherapy and cosmetic sectors, while the methyl chavicol and methyl eugenol types should be integrated into the pharmaceutical industry due to their well-documented antimicrobial properties.

• Although the two-year evaluation provided useful evidence of seasonal variation in quality traits, the study was conducted at a single experimental site; therefore, conclusions regarding environmental plasticity should be interpreted cautiously. Future breeding strategies should include multi-location trials under contrasting ecological conditions to more accurately quantify genotype × environment interactions, environmental plasticity, and genotype stability. Such trials would strengthen the selection of superior genotypes with stable chemical fingerprints across diverse production environments.

Authors’ contributions

Conceptualization, M.B.; methodology, M.B., and L.S.T.; software, M.B.; validation, M.B., and L.S.T.; formal analysis, M.B.; investigation, M.B.; resources, M.B.; data curation, M.B., and L.S.T.; writing original draft preparation, M.B.; writing, review and editing, M.B., and L.S.T.; visualization, M.B.; supervision, L.S.T. All authors read and approved the final manuscript.

Funding

Part of this research received financial support from the Scientific Research Project Unit of Çukurova University in Adana, Turkiye, under Project No. FDK-2024-16597.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.