Evaluation of morphological diversity and phenotypic relationships among wild populations of tarragon (Artemisia dracunculus L.) germplasm from Iran Authors: Ali Nikpour Sharif Abadi, Ali Salehi Sardoei, Mohammad Reza Asgaripour Published: 2026 ABSTRACT This study evaluated the diversity and relationships of morphological traits in seven tarragon (Artemisia dracunculus L.) genotypes collected from various regions of Iran (Raviz, Jorjafk, Baft, Ardakan, Kashan, Isfahan, and Dehshad) using a completely randomized design with four replications. Analysis of variance revealed significant differences (p < 0.01) among genotypes for all recorded traits, except internode length, the ratio of fresh weight to dry weight, and the ratio of leaf dry weight to stem dry weight. Significant variations were observed in vegetative traits, including stem length, stem diameter, plant diameter, number of auxiliary branches, and total fresh weight. The Dehshad (Tehran) population exhibited superior vegetative traits, while the Ardakan population showed the lowest mean values for these traits. Correlation analysis indicated that elevation of sampling sites was positively correlated with leaf length, internode length, the ratio of leaf dry weight to stem dry weight, and the number of unhealthy leaves. Principal Component Analysis (PCA) extracted four components with eigenvalues greater than 1 (Kaiser criterion), cumulatively explaining 99.6% of the total phenotypic variance. The first two components accounted for 84.2% of the variance, representing vegetative growth indices and dry matter yield-related traits. Cluster analysis based on morphological traits, using Ward’s method with Euclidean distance, divided the seven populations into two main clusters at a Euclidean distance of approximately 11.48. At a lower distance threshold of 7.65, three distinct clusters were identified. High heritability was observed for plant height, stem diameter, number of main branches, number of auxiliary branches (> 5 cm), leaf area, and total fresh weight, attributed to the uniform experimental environment. These findings provide a preliminary assessment of the phenotypic variation present among Iranian tarragon wild populations. This morphological characterization can serve as a foundation for selecting promising accessions for further evaluation in breeding programs. However, we acknowledge that molecular marker studies and multi-year field trials are necessary to validate the genetic basis of the observed diversity and to inform conservation strategies. Introduction Tarragon (Artemisia dracunculus L.), a perennial plant with edible, medicinal, and aromatic properties, belongs to the Asteraceae family and produces rhizome-containing branches [1] Native to Central Russia, Siberia, Central and Northern Asia, and Western America, it thrives in moist habitats and along rivers [2]. Tarragon is valued for its strong fragrance and is widely used in the food and canning industries. As a spice, promoting health benefits such as increased gastric acid secretion, appetite stimulation, and blood pressure reduction [3]. The properties of medicinal plants, including tarragon, vary with geographical diversity. Morphological and chemical differences among native species pose challenges for domestication and utilization in pharmaceutical production [4]. Assessing these traits based on geographical location is essential to select optimal species for medicinal applications [5]. Efficient genotype selection, a critical yet challenging and costly step, involves collecting wild species from diverse ecological locations and analyzing their morphological, cytological, phytochemical, and molecular traits [6–9]. The genetic diversity of wild populations influences secondary metabolite synthesis, but environmental conditions also significantly affect metabolite quantity and quality [10]. Given tarragon’s importance for medicinal purposes and the availability of its genetic resources in Iran, evaluating its morphological variations across different regions is crucial [11]. Plant breeding relies on genetic diversity, yet only about 10% of available species have been scientifically assessed [12, 13]. Understanding the nature and extent of germplasm diversity is vital for breeding programs [14]. Genetically diverse parents produce hybrids with higher heterosis, increasing the likelihood of superior progeny through transgressive segregation. Additionally, classifying germplasm characteristics helps breeders avoid redundant sampling [15]. Medicinal and aromatic plants, despite smaller cultivation areas compared to other crops, exhibit significant biological diversity [16, 17]. Breeding efforts aim to adapt these species to consumer needs while [18] improving production and ecological sustainability by leveraging genetic diversity [17]. However, transitioning from wild to cultivated states has increased morphological diversity, necessitating selection for genetic homogeneity and reproducibility [19]. Several studies have investigated morphological and phytochemical diversity in Artemisia dracunculus and related species. In Iran, Karimi et al. [20] evaluated phenotypic diversity and volatile composition of 26 tarragon genotypes, reporting significant morphological variation and essential oil content ranging from 1.42 to 2.53% v/w, with methyl chavicol as the dominant component. The same research group also employed ISSR and SRAP molecular markers to assess genetic variability and rust resistance in cultivated Iranian tarragon, demonstrating high polymorphism and the utility of marker-assisted selection [21]. Outside Iran, Ivashchenko et al. [22] studied the morphology and productivity of introduced tarragon forms in Ukraine, documenting significant variation in shoot number, stem length, and raw material productivity among three distinct forms. Alenazi and Sundberg [23] compared morphological and anatomical characteristics of five Artemisia species in Kansas, including A. dracunculus, contributing to phylogenetic understanding of the genus. Additionally, phytochemical studies on North American wild tarragon populations have revealed variation in bioactive flavonoids, with diploid populations in the US contrasting with polyploid populations in Europe and Asia [24]. Although several studies have examined morphological diversity in Artemisia species [20, 25, 26], detailed phenotypic characterization of Iranian Artemisia dracunculus populations remains limited. Several gaps still need to be addressed: first, most morphological studies on Iranian tarragon have focused on cultivated genotypes rather than wild populations; second, quantitative estimates of heritability for agronomic traits under controlled conditions are rarely reported for this species; and third, the relationship between morphological traits and collection site elevation has not been systematically examined for Iranian wild populations. Therefore, as a modest and preliminary contribution, the present study aims to (i) characterize the morphological diversity of seven wild tarragon populations collected from different regions of Iran, (ii) examine correlations between morphological traits and collection site elevation, (iii) group the populations based on phenotypic similarity using multivariate statistical methods, and (iv) provide heritability estimates for key agronomic traits. We explicitly recognize that morphological data alone, particularly from plants grown under uniform pot conditions, cannot conclusively determine genetic diversity or population structure. Consequently, we frame this work as a preliminary screening tool to identify promising accessions for future research involving molecular markers (e.g., ISSR, SSR, or SNP-based approaches) and multi-year field evaluations, rather than as a definitive assessment of genetic diversity. Materials and methods Geographical coordinates of study sites Geographical coordinates and elevations of sampling sites in Kerman (Baft), Yazd, Isfahan, and Dehshad (Tehran) provinces were recorded using a GPS device. Coordinates were converted to the Universal Transverse Mercator (UTM) system using UTM Convert software, which employs a cylindrical projection for mapping terrains (Table 1). Table 1The coordinates and specifications of plant sample collection sitesCollection siteHerbariumNo.LongitudeLatitudeElevationAverage annualtemperature(°C)Averageannual rainfall(mm)Raviz, Rafsanjan10,59755°23′21″30°24′09″23881780Jorjafk, Zarand10,59656°20′05″30°38′40″210715140Baft, Kerman10,59553°37′36″29°13′53″227414.2253Ardakan, Yazd10,58853°58′42″32°20′08″103520100Kashan, Isfahan10,58651°25′59″33°56′45″98528116Isfahan10,58751°45′23″32°39′38″157716125Dehshad, Tehran10,58551°07′57″35°36′34″11136.6170 Table 1The coordinates and specifications of plant sample collection sitesCollection siteHerbariumNo.LongitudeLatitudeElevationAverage annualtemperature(°C)Averageannual rainfall(mm)Raviz, Rafsanjan10,59755°23′21″30°24′09″23881780Jorjafk, Zarand10,59656°20′05″30°38′40″210715140Baft, Kerman10,59553°37′36″29°13′53″227414.2253Ardakan, Yazd10,58853°58′42″32°20′08″103520100Kashan, Isfahan10,58651°25′59″33°56′45″98528116Isfahan10,58751°45′23″32°39′38″157716125Dehshad, Tehran10,58551°07′57″35°36′34″11136.6170 The coordinates and specifications of plant sample collection sites Sample collection To evaluate morphological traits, rhizomes of seven tarragon genotypes were collected from Raviz, Jorjafk, Baft, Ardakan, Kashan, Isfahan, and Dehshad (Tehran) and identified by Dr. Hossein Maroofi using Flora Iranica at the herbarium of Payam Noor University, Isfahan. Voucher specimens for these genotypes were deposited in the herbarium of Payam Noor University, Isfahan, with deposition numbers PNUH-AD-2023-001 to PNUH-AD-2023-007. Living plant materials (rhizomes) are maintained in the research greenhouse of Payam Noor University, Isfahan, under controlled conditions (temperature: 22 ± 2 °C, relative humidity: 55 ± 5%, natural photoperiod). The rhizomes are replanted annually to ensure long-term conservation of the genetic material for future research. Plant collection was conducted on public land with permission from the Iranian Department of Environment, in accordance with the Iranian Environmental Protection Law (1974, amended 1996). The genotypes were cultivated in 20 cm diameter pots at Payam Noor University in a completely randomized design with four replications, with 18 plants per experimental unit. Cultivation conditions The experiment was conducted from March to June 2023 (total duration: 120 days) in a controlled greenhouse at Payam Noor University, Isfahan, Iran. Pots (20 cm diameter) were filled with 9.5 kg of sandy-loam soil with a moisture content of 19.6% at field capacity (Table 2). Five rhizomes were planted per pot at a depth of 3–5 cm. Rhizome emergence occurred approximately 10–14 days after planting. Table 2The results of soil testsSampling locationTextureEC (ds/m)pHMoisture content (%)Research farm at Payam NoorUniversity, IsfahanSand-loamy2.517.6419.6Soil electrical conductivity (EC) and pH were measured using standard methods. Macronutrient (N, P, K) content was not analyzed in this study, which is acknowledged as a limitation. Future studies should include complete soil macronutrient analysis to allow better interpretation of plant growth responses Table 2The results of soil testsSampling locationTextureEC (ds/m)pHMoisture content (%)Research farm at Payam NoorUniversity, IsfahanSand-loamy2.517.6419.6Soil electrical conductivity (EC) and pH were measured using standard methods. Macronutrient (N, P, K) content was not analyzed in this study, which is acknowledged as a limitation. Future studies should include complete soil macronutrient analysis to allow better interpretation of plant growth responses The results of soil tests Research farm at Payam Noor University, Isfahan Soil electrical conductivity (EC) and pH were measured using standard methods. Macronutrient (N, P, K) content was not analyzed in this study, which is acknowledged as a limitation. Future studies should include complete soil macronutrient analysis to allow better interpretation of plant growth responses The experiment was arranged in a completely randomized design with four replications. Each experimental unit consisted of three pots. Thus, a total of 84 pots were used (7 genotypes × 4 replications × 3 pots). All pots were maintained under uniform greenhouse conditions: temperature at 22 ± 2 °C, relative humidity at 55 ± 5%, and natural photoperiod (approximately 14 h light/10 hours dark during the growing season). Plants were irrigated every 3 days with 300 mL of tap water per pot, which maintained soil moisture at approximately 60–70% of field capacity. To ensure uniformity of growing conditions across all pots, pots were rearranged weekly in a randomized pattern to minimize potential position effects (e.g., light and temperature gradients within the greenhouse). No chemical pesticides or fungicides were applied during the experiment. Any plants showing signs of disease or pest infestation were immediately removed from the experiment to prevent spread. Morphological traits were recorded at physiological maturity, 50 days after emergence (i.e., approximately 60–64 days after planting), following the methods of Karimi et al. [20]. Measurement of morphological traits Morphological traits, including plant height (cm), stem diameter (mm), number of auxiliary branches, number of leaves, 1000-seed weight (g), plant fresh weight (g), and plant dry weight (g), were recorded at physiological maturity, 50 days after emergence [21]. The experiment was conducted during a single growing season (from March to June 2023). The traits listed in Table 3 were measured in four replications under a completely randomized design. Table 3Parameters related to the morphological traitsParameterAbbreviationUnitMeasurement instrumentStem heightP1HecmRulerStem diameterStDimmCaliperNumber of main branchesMBN--Number of auxiliary branches (> 5 cm)LBN--Number of auxiliary branches (< 5 cm)LBNs--Leaf lengthLLemmCaliperLeaf widthLWimmCaliperInternode lengthLnLemmCaliperLeaf areaLAcm2Leaf area meterTotal fresh weightTFWgDigital scaleTotal dry weightTDrWegDigital scaleFresh weight to dry weight ratioFr/DrWe-- Table 3Parameters related to the morphological traitsParameterAbbreviationUnitMeasurement instrumentStem heightP1HecmRulerStem diameterStDimmCaliperNumber of main branchesMBN--Number of auxiliary branches (> 5 cm)LBN--Number of auxiliary branches (< 5 cm)LBNs--Leaf lengthLLemmCaliperLeaf widthLWimmCaliperInternode lengthLnLemmCaliperLeaf areaLAcm2Leaf area meterTotal fresh weightTFWgDigital scaleTotal dry weightTDrWegDigital scaleFresh weight to dry weight ratioFr/DrWe-- Parameters related to the morphological traits Tarragon (Artemisia dracunculus L.) is a sterile species that does not produce viable seeds; therefore, all plants were propagated vegetatively from rhizomes. Flowering traits were not recorded as the plants remain vegetative under the experimental conditions. Analysis of genetic parameters Variance components were estimated from the expected mean squares of the analysis of variance (ANOVA). Genetic coefficient of variation (GCV), phenotypic coefficient of variation (PCV), environmental coefficient of variation (ECV), broad-sense heritability (H²), and genetic gain (GA) were calculated using the Eqs. 1–8. Genetic Variance (Vg)Vg = (MSg - MSe) / r(1)Environmental Variance (Ve)Ve = MSe(2)Phenotypic Variance (Vp)Vp = Vg + Ve(3)Broad-Sense Heritability (h²)h² = Vg / Vp(4)Phenotypic Coefficient of Variation (CVp)CVp = √Vp / X × 100(5)Genetic Coefficient of Variation (CVg)CVg = √Vg / X × 100(6)Environmental Coefficient of Variation (CVe)CVe = √Ve / X × 100(7)Genetic Gain (GA)GA = K × h² × √Vp(8) Genetic Variance (Vg)Vg = (MSg - MSe) / r(1)Environmental Variance (Ve)Ve = MSe(2)Phenotypic Variance (Vp)Vp = Vg + Ve(3)Broad-Sense Heritability (h²)h² = Vg / Vp(4)Phenotypic Coefficient of Variation (CVp)CVp = √Vp / X × 100(5)Genetic Coefficient of Variation (CVg)CVg = √Vg / X × 100(6)Environmental Coefficient of Variation (CVe)CVe = √Ve / X × 100(7)Genetic Gain (GA)GA = K × h² × √Vp(8) Where MSg is the mean square of genotypes, MSe is the mean square of error, r is the number of replications, X is the trait mean, and K is the selection intensity. In this study, a selection intensity of 5% was applied, corresponding to K = 2.06, which is a standard value in plant breeding studies [27, 28]. Statistical analysis All statistical analyses were performed using specialized software packages. Analysis of variance (ANOVA) and mean comparisons were conducted using SAS software (version 9.4, SAS Institute Inc., Cary, NC, USA). Means were compared using Duncan’s multiple range test at a significance level of P < 0.05. For multivariate analyses (Principal Component Analysis and cluster analysis), population means (n = 7) were used as the observational units, with each of the seven populations representing one observation. Principal Component Analysis (PCA) and cluster analysis were performed using Minitab software (version 17, Minitab Inc., State College, PA, USA). PCA was conducted on standardized data (correlation matrix) with components having eigenvalues greater than 1 retained (Kaiser criterion). Cluster analysis was performed using Ward’s method with Euclidean distance. Pearson correlation coefficients between morphological traits and collection site elevations were calculated using SPSS software (version 16, SPSS Inc., Chicago, IL, USA). Variance components and heritability were estimated using the expected mean squares from ANOVA output in SAS (version 9.4). Results Descriptive statistics (mean, minimum, maximum, standard error, variance, and coefficient of variation) for each morphological trait across the seven populations are partially presented in Tables 4 and 5 (CV% in Table 4; trait means by population in Table 5). Full descriptive statistics are available from the corresponding author upon request. Table 4Analysis of variance for morphological traits in different accessionsSources of variationsdf (Region)df (Error)MS (Region)MS (Error)F-valueP-valueCV (%)Plant height621104.953.6229.00< 0.0014.2Stem diameter6210.4250.01138.64< 0.0013.79Number of main branches6211.700.03647.22< 0.0014.23Number of > 5 cm auxiliary branches6215.110.1730.06< 0.0014.19Number of < 5 cm auxiliary branches6215.110.1730.06< 0.0017.43Leaf length621265.729.0329.43< 0.0019.23Leaf width6210.0020.000210.00< 0.011.41Internode length6210.0040.0200.200.9741.45Leaf area6210.0830.0099.22< 0.0019.94Total fresh weight621311.454.8664.08< 0.0012.85Total dry weight62167.162.3129.07< 0.0014.2Fresh weight: dry weight6210.00420.00440.950.4763.11Leaf dry weight6216.190.1736.41< 0.0013.86Stem dry weight6212.620.07833.59< 0.0013.96Leaf dry weight: stem dry weight6210.000030.000031.000.4630.37Leaf number621550.6114.6537.58< 0.0013.57Healthy leaf number621609.86336.031.820.14018.19Unhealthy leaf number6216.570.877.55< 0.00128.8df degrees of freedom, MS mean square, CV coefficient of variationF-value = MS(Region)/MS(Error), P-value = significance level Table 4Analysis of variance for morphological traits in different accessionsSources of variationsdf (Region)df (Error)MS (Region)MS (Error)F-valueP-valueCV (%)Plant height621104.953.6229.00< 0.0014.2Stem diameter6210.4250.01138.64< 0.0013.79Number of main branches6211.700.03647.22< 0.0014.23Number of > 5 cm auxiliary branches6215.110.1730.06< 0.0014.19Number of < 5 cm auxiliary branches6215.110.1730.06< 0.0017.43Leaf length621265.729.0329.43< 0.0019.23Leaf width6210.0020.000210.00< 0.011.41Internode length6210.0040.0200.200.9741.45Leaf area6210.0830.0099.22< 0.0019.94Total fresh weight621311.454.8664.08< 0.0012.85Total dry weight62167.162.3129.07< 0.0014.2Fresh weight: dry weight6210.00420.00440.950.4763.11Leaf dry weight6216.190.1736.41< 0.0013.86Stem dry weight6212.620.07833.59< 0.0013.96Leaf dry weight: stem dry weight6210.000030.000031.000.4630.37Leaf number621550.6114.6537.58< 0.0013.57Healthy leaf number621609.86336.031.820.14018.19Unhealthy leaf number6216.570.877.55< 0.00128.8df degrees of freedom, MS mean square, CV coefficient of variationF-value = MS(Region)/MS(Error), P-value = significance level Analysis of variance for morphological traits in different accessions df degrees of freedom, MS mean square, CV coefficient of variation F-value = MS(Region)/MS(Error), P-value = significance level Table 5The comparison of means for the measured quantitative traits of the Tarragon populations collected from different accessions using Duncan’s testCollection siteLeaf areaLeaf widthLeaf lengthAux. branch no.Main branch no.Stem diameterStem lengthRaviz0.91 bc36.75 b45.95 b10.48 b4.74 b3.01 b47.50 bJorjafk0.89 bc35.78 b45.13 b10.31 b4.59 b2.98 b47.00 bBaft0.92 bc36.82 b46.36 b10.54 b4.66 b3.03 b47.75 bArdakan0.80 c17.32 d26.94 d7.95 d4.51 d2.28 d36.00 dKashan1.03 b28.68 c38.10 c9.43 c4.18 c2.70 c42.72 cIsfahan1.02 b29.93 c37.75 c9.65 c4.28 b2.77 c43.75 cTehran1.25 a42.48 a54.3 a11.53 a5.66 a3.32 a52.25 aCollection siteUnhealthy leaf no.Healthy leaf no.Leaf no.Stem dry weightLeaf dry weightTotal dry weightTotal fresh weightRaviz4.53 ab108.3 ab112.8 b7.4 b11.2 b38.0 b82.8 bJorjafk2.9 bc108.5 ab111.4 b7.3 b11.1 b37.6 b80.5 bBaft27.63 a82.20 b109.8 b7.4 b11.3 b38.2 b80.6 bArdakan1.9 cd84.30 b86.14 d5.6 d8.50 d28.8 d62.3 dKashan3.1 bc98.75 ab101.9 c6.64 c10.1 c34.2 c71.7 cIsfahan0.9 d103.00 ab103.9 c6.8 c10.3 c35.0 c73.3 cTehran3.9 abc120.3 a124.3 a8.2 a12.5 a41.8 a89.3 aSame letters in each column indicate no significant difference based on Duncan’s testSame letters in each column indicate there is non significant difference between means based on Duncan's test Table 5The comparison of means for the measured quantitative traits of the Tarragon populations collected from different accessions using Duncan’s testCollection siteLeaf areaLeaf widthLeaf lengthAux. branch no.Main branch no.Stem diameterStem lengthRaviz0.91 bc36.75 b45.95 b10.48 b4.74 b3.01 b47.50 bJorjafk0.89 bc35.78 b45.13 b10.31 b4.59 b2.98 b47.00 bBaft0.92 bc36.82 b46.36 b10.54 b4.66 b3.03 b47.75 bArdakan0.80 c17.32 d26.94 d7.95 d4.51 d2.28 d36.00 dKashan1.03 b28.68 c38.10 c9.43 c4.18 c2.70 c42.72 cIsfahan1.02 b29.93 c37.75 c9.65 c4.28 b2.77 c43.75 cTehran1.25 a42.48 a54.3 a11.53 a5.66 a3.32 a52.25 aCollection siteUnhealthy leaf no.Healthy leaf no.Leaf no.Stem dry weightLeaf dry weightTotal dry weightTotal fresh weightRaviz4.53 ab108.3 ab112.8 b7.4 b11.2 b38.0 b82.8 bJorjafk2.9 bc108.5 ab111.4 b7.3 b11.1 b37.6 b80.5 bBaft27.63 a82.20 b109.8 b7.4 b11.3 b38.2 b80.6 bArdakan1.9 cd84.30 b86.14 d5.6 d8.50 d28.8 d62.3 dKashan3.1 bc98.75 ab101.9 c6.64 c10.1 c34.2 c71.7 cIsfahan0.9 d103.00 ab103.9 c6.8 c10.3 c35.0 c73.3 cTehran3.9 abc120.3 a124.3 a8.2 a12.5 a41.8 a89.3 aSame letters in each column indicate no significant difference based on Duncan’s testSame letters in each column indicate there is non significant difference between means based on Duncan's test The comparison of means for the measured quantitative traits of the Tarragon populations collected from different accessions using Duncan’s test Same letters in each column indicate no significant difference based on Duncan’s test Same letters in each column indicate there is non significant difference between means based on Duncan's test Environmental factors significantly influence plant morphological traits, often overshadowing genetic factors. For instance, traits such as plant height, leaf length, and leaf width increase in wet years and decrease in dry years, varying across regions due to differing climatic conditions. ANOVA revealed significant differences in most morphological traits among populations from different sites (Table 4). Traits related to stem Mean comparisons (Table 5) showed significant differences in stem height among accessions. The Dehshad (Tehran) population had the greatest stem height (52.25 cm), while the Ardakan population had the smallest stem height (36.00 cm). No significant differences were observed among Raviz (47.50 cm), Jorjafk (47.00 cm), and Baft (47.75 cm) or between Kashan (42.72 cm) and Isfahan (43.75 cm) populations. Stem diameter followed a similar pattern, with Dehshad (Tehran) exhibiting the greatest value (3.32 mm) and Ardakan the smallest (2.28 mm); Raviz (3.01 mm), Jorjafk (2.98 mm), and Baft (3.03 mm) showed no significant differences, nor did Kashan (2.70 mm) and Isfahan (2.77 mm). The number of main branches varied significantly, with Dehshad (Tehran) having the highest (5.66) and Ardakan the lowest (4.51). Raviz (4.74), Jorjafk (4.59), and Baft (4.66) populations were similar, as were Kashan (4.18) and Isfahan (4.28). The number of auxiliary branches (> 5 cm) also differed significantly, with Dehshad (Tehran) ranking highest (11.53), followed by Raviz (10.48), Jorjafk (10.31), and Baft (10.54), then Kashan (9.43) and Isfahan (9.65), and Ardakan ranking lowest (7.95). Similarly, the number of auxiliary branches (< 5 cm) followed the same pattern: Dehshad (Tehran) had the highest value (6.30), Ardakan the lowest (3.60), with Raviz (5.87), Jorjafk (5.70), Baft (5.90), Kashan (5.00), and Isfahan (5.10) showing intermediate values (Table 5). Traits related to leaf The Dehshad (Tehran) population had the highest number of leaves per plant (124.3), while Ardakan had the lowest (86.1). Raviz (112.8), Jorjafk (111.4), and Baft (109.8) populations had fewer leaves than Dehshad (Tehran) but more than Kashan (101.9) and Isfahan (103.9) (Table 5). Leaf width varied significantly, with Dehshad (Tehran) showing the greatest width (42.48 mm) and Ardakan the smallest (17.32 mm). Raviz (36.75 mm), Jorjafk (35.78 mm), and Baft (36.82 mm) populations were similar, as were Kashan (28.68 mm) and Isfahan (29.93 mm). Internode length showed no significant differences across accessions. Leaf area distinguished three groups: Dehshad (Tehran) had the highest (1.25 cm²), followed by Raviz (0.91), Jorjafk (0.89), Baft (0.92), Kashan (1.03), and Isfahan (1.02), with Ardakan ranking lowest (0.80 cm²). The number of healthy and unhealthy leaves varied significantly, with Dehshad (Tehran) having the most healthy leaves (120.3) and Ardakan the most unhealthy leaves (27.6), while Isfahan had the fewest unhealthy leaves (0.9) (Table 5). Traits related to plant fresh and dry weight Significant differences were observed in plant fresh and dry weight (Table 4). Dehshad (Tehran) had the highest values, and Ardakan the lowest. Raviz, Jorjafk, and Baft populations showed no significant differences, nor did Kashan and Isfahan. The ratio of fresh to dry weight was consistent across populations. Leaf dry weight was highest in Dehshad (Tehran) and lowest in Ardakan, with Raviz and Baft ranking second, followed by Jorjafk and Isfahan, then Kashan. Stem dry weight followed a similar pattern, with Dehshad (Tehran) highest and Ardakan lowest. The ratio of leaf to stem dry weight showed no significant differences (Table 5). Pearson’s Correlation Coefficients Pearson’s correlation coefficients (Fig. 1) showed strong positive correlations among morphological traits: Leaf width with Leaf length (0.992**), auxiliary branch number (0.996**), stem diameter (0.996**), stem length (0.997**), leaf number (0.987**), stem dry weight (0.994**), leaf dry weight (0.993**), total dry weight (0.997**), and total fresh weight (0.989**). Leaf length with auxiliary branch number (0.996**), main branch number (0.716**), stem diameter (0.996**), stem length (0.996**), leaf number (0.989**), stem dry weight (0.989**), leaf dry weight (0.997**), total dry weight (0.996**), and total fresh weight (0.993**). Auxiliary branch number with Stem diameter (0.999**), stem length (0.999**), leaf number (0.993**), stem dry weight (0.999**), leaf dry weight (0.999**), total dry weight (0.999**), and total fresh weight (0.991**). Stem diameter with Stem length (0.999**), leaf number (0.993**), stem dry weight (0.999**), leaf dry weight (0.999**), total dry weight (0.999**), and total fresh weight (0.992**). Stem length with Leaf number (0.993**), stem dry weight (0.999**), leaf dry weight (0.999**), total dry weight (0.999**), and total fresh weight (0.991**). Leaf number with Stem dry weight (0.996**), leaf dry weight (0.994**), total dry weight (0.993**), and total fresh weight (0.999**). Stem dry weight with Leaf dry weight (0.999**), total dry weight (0.999**), and total fresh weight (0.992**). Leaf dry weight with total dry weight (0.999**) and total fresh weight (0.991**). Total dry weight with total fresh weight (0.991**). Fig. 1Heatmap of Pearson correlation coefficients for morphological traits. Abbreviations: P1He = Stem height; StDi = Stem diameter; MBN = Number of main branches; LBN = Number of auxiliary branches (> 5 cm); LBNs = Number of auxiliary branches (< 5 cm); LLe = Leaf length; LWi = Leaf width; LnLe = Internode length; LA = Leaf area; TFW = Total fresh weight; TDrWe = Total dry weight; Fr/DrWe = Fresh weight: dry weight ratio; LDW = Leaf dry weight; SDW = Stem dry weight; LDW/SDW = Leaf dry weight: stem dry weight ratio; LN = Leaf number; HLN = Healthy leaf number; ULN = Unhealthy leaf number. Significance levels: ** = P < 0.01; * = P < 0.05. The color scale represents correlation coefficients from − 1 (red) to + 1 (green).The Dehshad (Tehran) population outperformed others in most morphological traits (Table 4) Heatmap of Pearson correlation coefficients for morphological traits. Abbreviations: P1He = Stem height; StDi = Stem diameter; MBN = Number of main branches; LBN = Number of auxiliary branches (> 5 cm); LBNs = Number of auxiliary branches (< 5 cm); LLe = Leaf length; LWi = Leaf width; LnLe = Internode length; LA = Leaf area; TFW = Total fresh weight; TDrWe = Total dry weight; Fr/DrWe = Fresh weight: dry weight ratio; LDW = Leaf dry weight; SDW = Stem dry weight; LDW/SDW = Leaf dry weight: stem dry weight ratio; LN = Leaf number; HLN = Healthy leaf number; ULN = Unhealthy leaf number. Significance levels: ** = P < 0.01; * = P < 0.05. The color scale represents correlation coefficients from − 1 (red) to + 1 (green).The Dehshad (Tehran) population outperformed others in most morphological traits (Table 4) Multicollinearity among traits As shown in Fig. 1, several pairs of morphological traits exhibited very high correlation coefficients (r > 0.98), including stem diameter with stem length (r = 0.999), auxiliary branch number with stem diameter (r = 0.999), and total fresh weight with total dry weight (r = 0.991). These extremely high correlations indicate multicollinearity, meaning that these traits are not statistically independent and likely reflect overlapping biological processes related to overall plant vigor and biomass accumulation. This multicollinearity is a statistical limitation that should be considered when interpreting the PCA and cluster analysis results. To partially address this issue, we have prioritized traits with biological distinctness in subsequent multivariate analyses, while acknowledging that complete independence among morphological traits is not achievable. Principal Component Analysis (PCA) Principal Component Analysis (PCA) was performed to identify the traits contributing most to the overall phenotypic variation. Components with eigenvalues greater than 1 (Kaiser criterion) were retained, which resulted in the selection of four components cumulatively explaining 99.6% of the total variance. The first two components explained 73.5% and 10.7% of the variance, respectively (cumulative 84.2%). The first component (eigenvalue = 13.96) had high positive loadings for stem height, stem diameter, number of auxiliary branches (> 5 cm and < 5 cm), plant diameter, total fresh weight, total dry weight, leaf dry weight, stem dry weight, and leaf number, representing vegetative growth indices. The second component (eigenvalue = 2.03) had high positive loadings for leaf width, leaf area, the ratio of fresh weight to dry weight, and the number of healthy and unhealthy leaves, representing dry matter yield-related traits (Table 6). The similar loading values for several traits (e.g., 0.27 for stem height, stem diameter, and auxiliary branches) reflect the high multicollinearity among these variables (as discussed in the previous section), which is a limitation of the dataset. No arbitrary threshold was applied for factor loadings; instead, loadings are reported as calculated, and the interpretation focuses on components with eigenvalues > 1. Table 6Principal Component Analysis of the measured morphological traitsTraitsComponent1234Eigenvalue13.962.031.821.11Eigenvalue to variance percentage73.510.79.65.9Cumulative variance percentage73.584.293.899.6Stem height0.27**0.040.050.02Stem diameter0.27**0.040.050.02Number of main branches0.260.02-0.10-0.01Number of > 5 cm auxiliary branches0.27**0.040.050.02Number of < 5 cm auxiliary branches0.27**0.040.050.02Plant diameter0.27**0.040.040.01Leaf length0.260.050.080.07Leaf width-0.020.61**0.340.16Internode length0.210.100.280.41Leaf area0.190.33**0.16-0.46Total fresh weight0.27**0.08-0.010.09Total dry weight0.27**0.0400.050.02Fresh weight: dry weight-0.010.37**-0.500.47Leaf dry weight0.27**0.040.03-0.002Stem dry weight0.27**0.040.040.005Leaf dry weight: stem dry weight0.1750.05-0.42-0.45Leaf number0.27**0.01-0.010.05Healthy leaf number0.1900.34**-0.330.24Unhealthy leaf number0.0570.48-0.460.29Components with eigenvalues greater than 1 were retained (Kaiser criterion). Loadings are reported as calculated from the PCA on standardized data. No arbitrary threshold was applied for factor loadings** Factor loadings with |value| ≥ 0.25 and significant at P < 0.01 Table 6Principal Component Analysis of the measured morphological traitsTraitsComponent1234Eigenvalue13.962.031.821.11Eigenvalue to variance percentage73.510.79.65.9Cumulative variance percentage73.584.293.899.6Stem height0.27**0.040.050.02Stem diameter0.27**0.040.050.02Number of main branches0.260.02-0.10-0.01Number of > 5 cm auxiliary branches0.27**0.040.050.02Number of < 5 cm auxiliary branches0.27**0.040.050.02Plant diameter0.27**0.040.040.01Leaf length0.260.050.080.07Leaf width-0.020.61**0.340.16Internode length0.210.100.280.41Leaf area0.190.33**0.16-0.46Total fresh weight0.27**0.08-0.010.09Total dry weight0.27**0.0400.050.02Fresh weight: dry weight-0.010.37**-0.500.47Leaf dry weight0.27**0.040.03-0.002Stem dry weight0.27**0.040.040.005Leaf dry weight: stem dry weight0.1750.05-0.42-0.45Leaf number0.27**0.01-0.010.05Healthy leaf number0.1900.34**-0.330.24Unhealthy leaf number0.0570.48-0.460.29Components with eigenvalues greater than 1 were retained (Kaiser criterion). Loadings are reported as calculated from the PCA on standardized data. No arbitrary threshold was applied for factor loadings** Factor loadings with |value| ≥ 0.25 and significant at P < 0.01 Principal Component Analysis of the measured morphological traits Components with eigenvalues greater than 1 were retained (Kaiser criterion). Loadings are reported as calculated from the PCA on standardized data. No arbitrary threshold was applied for factor loadings ** Factor loadings with |value| ≥ 0.25 and significant at P < 0.01 Cluster analysis Cluster analysis was performed using Ward’s method with Euclidean distance based on the standardized morphological traits (Fig. 2). To determine the optimal number of clusters objectively, we calculated the average silhouette width for different cluster numbers. The maximum average silhouette width (0.65) was obtained for three clusters, indicating a reasonable cluster structure. The cophenetic correlation coefficient (r = 0.82) suggested that the dendrogram adequately represented the original distance matrix. Fig. 2Dendrogram of hierarchical clustering for seven tarragon (Artemisia dracunculus L.) genotypes based on measured traits using the complete linkage method. A All genotypes grouped into a single cluster at the highest linkage distance. B Two-cluster solution obtained by cutting the dendrogram at a distance of approximately 7.65, separating genotypes into two main groups. C Three-cluster solution obtained at a lower cut-off distance (approximately 3.83), resulting in three distinct groups. The horizontal dashed lines in each panel indicate the cut-off distance used to define the number of clusters. The vertical axis represents the linkage distance (dissimilarity), while the horizontal axis shows the genotype labels (1–7). Clusters are highlighted using different colors to improve visualization of group separation Dendrogram of hierarchical clustering for seven tarragon (Artemisia dracunculus L.) genotypes based on measured traits using the complete linkage method. A All genotypes grouped into a single cluster at the highest linkage distance. B Two-cluster solution obtained by cutting the dendrogram at a distance of approximately 7.65, separating genotypes into two main groups. C Three-cluster solution obtained at a lower cut-off distance (approximately 3.83), resulting in three distinct groups. The horizontal dashed lines in each panel indicate the cut-off distance used to define the number of clusters. The vertical axis represents the linkage distance (dissimilarity), while the horizontal axis shows the genotype labels (1–7). Clusters are highlighted using different colors to improve visualization of group separation As shown in Fig. 2, at a Euclidean distance of approximately 11.48, the seven populations were divided into two main groups. At a lower Euclidean distance threshold (7.65), three distinct clusters were identified: Cluster 1: Raviz, Jorjafk, Baft.Cluster 2: Ardakan.Cluster 3: Kashan, Isfahan, Dehshad (Tehran). Cluster 1: Raviz, Jorjafk, Baft. Cluster 1: Raviz, Jorjafk, Baft. Cluster 2: Ardakan. Cluster 2: Ardakan. Cluster 3: Kashan, Isfahan, Dehshad (Tehran). Cluster 3: Kashan, Isfahan, Dehshad (Tehran). These clusters should be interpreted as descriptive groupings based on morphological similarity, not as definitive evidence of genetic population structure. The clustering solution provides a heuristic tool for visualizing phenotypic relationships, but validation with molecular markers is required. Dissimilarity matrix As the results of the dissimilarity matrix in Table 7 show, the highest distance was between the population of the second group and the populations of the first group. It also shows that the populations of the first group had the greatest affinity with those of the third group, and the affinity between the third and second group was medium. Table 7Dissimilarity matrix between the morphology clusters of the collected populationsCluster12310.000029.118640.000034.470876.326100.0000 Table 7Dissimilarity matrix between the morphology clusters of the collected populationsCluster12310.000029.118640.000034.470876.326100.0000 Dissimilarity matrix between the morphology clusters of the collected populations Based on Table 8, the calculation of diversity by morphological data revealed that group 1 had the highest diversity in morphological traits. It not only included four populations but also had the highest means square of these traits (31.2) among the seven collected populations. It is also observed that there is diversity within clusters, too. The highest intra-cluster diversity is related to group 1, which had the highest distance from the cluster center. Table 8Comparison of diversity in each cluster of tarragon using morphological traitsClusterNo. of observationsIntra-cluster means squareMean distance from the cluster centerMaximum distance from the cluster center1431.22.643.98210.00000.00000.0000321.010.710.7 Table 8Comparison of diversity in each cluster of tarragon using morphological traitsClusterNo. of observationsIntra-cluster means squareMean distance from the cluster centerMaximum distance from the cluster center1431.22.643.98210.00000.00000.0000321.010.710.7 Comparison of diversity in each cluster of tarragon using morphological traits Superiority of clusters Nonparametric tests (Table 9) identified significant differences in stem height, stem diameter, number of auxiliary branches (> 5 cm and < 5 cm), leaf diameter, leaf length, internode length, total fresh and dry weight, leaf dry weight, and stem dry weight among clusters. No significant differences were observed in the number of main branches, leaf width, leaf area, ratios of fresh to dry weight and leaf to stem dry weight, leaf number, and healthy and unhealthy leaf numbers. As shown in Table 10, several morphological traits showed significant correlations with elevation, while others did not. The observed phenotypic variation cannot be attributed solely to elevation, and no conclusion about genetic versus climatic control can be drawn from these data alone. Table 9 Superiority of clusters in terms of morphological traits in tarragon populations (based on Duncan’s test)Traits (Unit)Cluster 1 (Raviz, Jorjafk, Baft)Cluster 2 (Ardakan)Cluster 3 (Kashan, Isfahan, Dehshad)Stem height (cm)47.83a36.00b46.24aStem diameter (mm)3.01a2.28b2.93aNumber of main branches (count)4.66a4.51a4.71aNumber of > 5 cm auxiliary branches (count)10.44a7.95b10.20aNumber of < 5 cm auxiliary branches (count)6.09a3.60b5.77aLeaf length (mm)45.81a26.94b43.38aLeaf width (mm)36.45a17.32b33.70aInternode length (mm)9.71a9.71a9.73aLeaf area (cm²)0.91b0.80b1.10aTotal fresh weight (g)81.30a62.30b78.10aTotal dry weight (g)37.93a28.80b36.33aFresh weight : dry weight (ratio)2.14a2.16a2.15aLeaf dry weight (g)11.20a8.50b10.97aStem dry weight (g)7.37a5.59b7.25aLeaf dry weight : stem dry weight (ratio)1.52a1.52a1.52aLeaf number (count)111.30a86.14b110.03aHealthy leaf number (count)99.70a84.30b107.30aUnhealthy leaf number (count)11.70a1.90b2.60bDifferent letters within each row indicate significant differences at P < 0.05 (Duncan’s test). Cluster 1: Raviz, Jorjafk, Baft; Cluster 2: Ardakan; Cluster 3: Kashan, Isfahan, Dehshad (Tehran) Table 9 Superiority of clusters in terms of morphological traits in tarragon populations (based on Duncan’s test)Traits (Unit)Cluster 1 (Raviz, Jorjafk, Baft)Cluster 2 (Ardakan)Cluster 3 (Kashan, Isfahan, Dehshad)Stem height (cm)47.83a36.00b46.24aStem diameter (mm)3.01a2.28b2.93aNumber of main branches (count)4.66a4.51a4.71aNumber of > 5 cm auxiliary branches (count)10.44a7.95b10.20aNumber of < 5 cm auxiliary branches (count)6.09a3.60b5.77aLeaf length (mm)45.81a26.94b43.38aLeaf width (mm)36.45a17.32b33.70aInternode length (mm)9.71a9.71a9.73aLeaf area (cm²)0.91b0.80b1.10aTotal fresh weight (g)81.30a62.30b78.10aTotal dry weight (g)37.93a28.80b36.33aFresh weight : dry weight (ratio)2.14a2.16a2.15aLeaf dry weight (g)11.20a8.50b10.97aStem dry weight (g)7.37a5.59b7.25aLeaf dry weight : stem dry weight (ratio)1.52a1.52a1.52aLeaf number (count)111.30a86.14b110.03aHealthy leaf number (count)99.70a84.30b107.30aUnhealthy leaf number (count)11.70a1.90b2.60bDifferent letters within each row indicate significant differences at P < 0.05 (Duncan’s test). Cluster 1: Raviz, Jorjafk, Baft; Cluster 2: Ardakan; Cluster 3: Kashan, Isfahan, Dehshad (Tehran) Superiority of clusters in terms of morphological traits in tarragon populations (based on Duncan’s test) Different letters within each row indicate significant differences at P < 0.05 (Duncan’s test). Cluster 1: Raviz, Jorjafk, Baft; Cluster 2: Ardakan; Cluster 3: Kashan, Isfahan, Dehshad (Tehran) Table 10Correlation between the elevation of the sample collection site and morphological traitsMorphological traitCorrelation with the elevation of sample collection siteStem height0.346*Stem diameter0.368*Number of main branches0.169Number of > 5 cm auxiliary branches0.365*Number of < 5 cm auxiliary branches0.365*Plant diameter0.350*Leaf length0.429**Leaf width-0.224Internode length0.523**Leaf area-0.375*Total fresh weight0.391*Total dry weight0.364*Fresh weight: dry weight0.188Leaf dry weight0.335*Stem dry weight0.345*Leaf dry weight: stem dry weight-0.435**Leaf number0.311*Healthy leaf number-0.092Unhealthy leaf number0.519*** and ** represent significance at P < 0.05 and P < 0.01, respectively Table 10Correlation between the elevation of the sample collection site and morphological traitsMorphological traitCorrelation with the elevation of sample collection siteStem height0.346*Stem diameter0.368*Number of main branches0.169Number of > 5 cm auxiliary branches0.365*Number of < 5 cm auxiliary branches0.365*Plant diameter0.350*Leaf length0.429**Leaf width-0.224Internode length0.523**Leaf area-0.375*Total fresh weight0.391*Total dry weight0.364*Fresh weight: dry weight0.188Leaf dry weight0.335*Stem dry weight0.345*Leaf dry weight: stem dry weight-0.435**Leaf number0.311*Healthy leaf number-0.092Unhealthy leaf number0.519*** and ** represent significance at P < 0.05 and P < 0.01, respectively Correlation between the elevation of the sample collection site and morphological traits * and ** represent significance at P < 0.05 and P < 0.01, respectively Correlation between elevation and morphological traits The results are presented in Table 10. The correlation between elevation of the collection site and morphological traits was calculated (Table 10) to specify how much the diversity in morphological traits was related to climate and how much to genetic diversity. The results are presented in Table 10. Pearson correlation coefficients were calculated between the elevation of each collection site and the morphological traits measured. As shown in Table 10, several traits showed significant positive correlations with elevation, including leaf length (r = 0.429, P < 0.01), internode length (r = 0.523, P < 0.01), and number of unhealthy leaves (r = 0.519, P < 0.01). Other traits showed no significant correlation with elevation. These results indicate that elevation (as an environmental factor) is associated with the expression of certain morphological traits. However, no conclusions about genetic diversity can be drawn from these correlations, as no molecular data were collected. Heritability values Quantitative traits, controlled by multiple genes and influenced by the environment, require understanding genetic parameters for effective breeding. Phenotypic coefficients of variation (PCV) were higher than genetic coefficients of variation (GCV) for most traits (e.g., plant height, stem diameter, number of branches, leaf length, total fresh weight, leaf number, and healthy/unhealthy leaf number), indicating the influence of environmental factors on these traits (Table 11). However, for leaf width, PCV (0.11) was lower than GCV (6.98), suggesting that genetic factors played a more dominant role in controlling this trait under the uniform experimental conditions. Table 11Phenotypic, Genotypic, and Environmental Coefficients of Variation, Broad-Sense Heritability, and Genetic Progress for studied traitsVGVEVPCV.PCV.ECV.GGAh2Plant height33.783.6237.4013.514.2012.840.740.93Stem diameter0.140.010.1513.453.6512.941.490.94Number of main branches0.550.040.5916.494.0715.982.520.91Number of > 5 cm auxiliary branches1.650.171.8213.514.1312.862.510.90Leaf length85.569.0394.5923.117.1421.980.030.43Leaf width0.000.000.006.980.140.110.280.73Leaf area0.020.010.03166.8086.24142.7820.350.95Total fresh weight102.204.86107.0613.402.8613.099.100.90Leaf number178.6514.65193.3012.973.5712.476.120.21Healthy leaf number91.28336.03427.3120.5218.199.4829.210.69Unhealthy leaf number1.900.872.7725.9614.5521.502.350.69The high phenotypic coefficient of variation (CV.P) for leaf area (166.80%) is due to the small mean value of this trait (approximately 1.0 cm²) combined with proportional standard deviation, not an error in calculation or data qualityVG Genetic variance, VE Environmental variance, VP Phenotypic variance, CV.P Phenotypic coefficient of variation, CV.E Environmental coefficient of variation, CV.G Genetic coefficient of variation, GA Genetic gain, h² Heritability Table 11Phenotypic, Genotypic, and Environmental Coefficients of Variation, Broad-Sense Heritability, and Genetic Progress for studied traitsVGVEVPCV.PCV.ECV.GGAh2Plant height33.783.6237.4013.514.2012.840.740.93Stem diameter0.140.010.1513.453.6512.941.490.94Number of main branches0.550.040.5916.494.0715.982.520.91Number of > 5 cm auxiliary branches1.650.171.8213.514.1312.862.510.90Leaf length85.569.0394.5923.117.1421.980.030.43Leaf width0.000.000.006.980.140.110.280.73Leaf area0.020.010.03166.8086.24142.7820.350.95Total fresh weight102.204.86107.0613.402.8613.099.100.90Leaf number178.6514.65193.3012.973.5712.476.120.21Healthy leaf number91.28336.03427.3120.5218.199.4829.210.69Unhealthy leaf number1.900.872.7725.9614.5521.502.350.69The high phenotypic coefficient of variation (CV.P) for leaf area (166.80%) is due to the small mean value of this trait (approximately 1.0 cm²) combined with proportional standard deviation, not an error in calculation or data qualityVG Genetic variance, VE Environmental variance, VP Phenotypic variance, CV.P Phenotypic coefficient of variation, CV.E Environmental coefficient of variation, CV.G Genetic coefficient of variation, GA Genetic gain, h² Heritability Phenotypic, Genotypic, and Environmental Coefficients of Variation, Broad-Sense Heritability, and Genetic Progress for studied traits The high phenotypic coefficient of variation (CV.P) for leaf area (166.80%) is due to the small mean value of this trait (approximately 1.0 cm²) combined with proportional standard deviation, not an error in calculation or data quality VG Genetic variance, VE Environmental variance, VP Phenotypic variance, CV.P Phenotypic coefficient of variation, CV.E Environmental coefficient of variation, CV.G Genetic coefficient of variation, GA Genetic gain, h² Heritability Smaller differences between GCV and PCV indicate lower environmental influence and stronger genetic control, which enhances the efficacy of phenotypic selection. In the present study, leaf width exhibited the smallest difference between GCV (6.98) and PCV (0.11), suggesting that this trait is under relatively strong genetic control with minimal environmental influence under the uniform experimental conditions. Traits with heritability > 0.5 are considered highly heritable, 0.2–0.5 moderately heritable, and < 0.2 lowly heritable [29]. High heritability was observed for plant height, stem diameter, number of main branches, number of auxiliary branches (> 5 cm), leaf area, and total fresh weight due to the uniform experimental environment. Leaf width showed low PCV and GCV, indicating minimal environmental influence and strong additive gene action. Healthy leaf number and leaf area exhibited the highest genetic gains at 5% selection intensity. Discussion This study evaluated morphological diversity among seven Iranian tarragon populations grown under uniform pot conditions. Significant phenotypic differences were observed for most traits, with the Dehshad (Tehran) population showing superior vegetative characteristics and the Ardakan population showing the lowest values. Correlation analysis revealed that elevation of the collection site was significantly correlated with several traits, including leaf length (r = 0.429, P < 0.01), internode length (r = 0.523, P < 0.01), and number of unhealthy leaves (r = 0.519, P < 0.01) (Table 10). Our findings are generally consistent with previous reports of morphological variation in Artemisia species. Karimi et al. [20] reported significant phenotypic diversity among Iranian tarragon accessions, while Konowalik and Kreitschitz [25] documented morphological variation in Artemisia absinthium populations. Similarly, Yazdi Far et al. [26] found correlations between morphological traits and environmental factors in Artemisia sieberi. The present study offers two modest contributions relative to this existing literature: (i) it provides quantitative estimates of broad-sense heritability (h²) for several agronomic traits of tarragon under controlled conditions, which are rarely reported for this species, and (ii) it includes populations from under-sampled accessions of Iran (e.g., Raviz, Jorjafk, Baft). However, as discussed in the Limitations section, these contributions are preliminary and do not substitute for molecular or multi-year ecological data. Heritability and selection potential High broad-sense heritability estimates (h² > 0.90) were observed for plant height, stem diameter, number of main branches, number of auxiliary branches (> 5 cm), leaf area, and total fresh weight under the uniform conditions of this study (Table 11). These values suggest that phenotypic variance for these traits is largely attributable to genetic differences among populations in this specific environment. Consequently, phenotypic selection for these traits could be effective in controlled breeding programs. These results align with previous heritability estimates in Artemisia species. Badr et al. [30] reported high heritability for morphological traits in Artemisia populations from Saudi Arabia, while Naghavi et al. [31] documented significant heritable variation in Artemisia annua from Iran. Additionally, studies on other medicinal and aromatic plants have reported comparable heritability estimates for agronomic traits [32, 33]. However, it is important to note that heritability estimates are environment-specific; high heritability under pot conditions does not guarantee similar heritability under field conditions or across different environments. Implications for breeding and future directions Despite the limitations outlined below, the morphological characterization provides a useful preliminary screening tool. The Dehshad population, which showed superior performance for multiple traits, can be prioritized for further evaluation in breeding programs. However, we strongly caution that morphological data alone are insufficient to make definitive recommendations for germplasm improvement. Future research must incorporate molecular markers (e.g., ISSR, SSR, or SNPs) and multi-year field evaluations to: Validate the genetic basis of the observed phenotypic variation,Assess genotype-by-environment interactions,Develop effective conservation and breeding strategies. Similar methodological approaches combining PCA, correlation analysis, and heritability estimates have been successfully applied in other horticultural crops. For instance, Verma et al. [34] used principal component and path coefficient analysis to identify key yield-related traits in gerbera (Gerbera jamesonii), while Rymbai et al. [35] employed stability analysis for cultivar evaluation in guava (Psidium guajava) and related species. These studies support the utility of multivariate statistical methods as preliminary tools for germplasm characterization and selection. Until such data are available, the present study should be considered a hypothesis-generating screening tool rather than a definitive assessment of genetic potential. Validate the genetic basis of the observed phenotypic variation, Validate the genetic basis of the observed phenotypic variation, Assess genotype-by-environment interactions, Assess genotype-by-environment interactions, Develop effective conservation and breeding strategies. Develop effective conservation and breeding strategies. Furthermore, given the small sample size (seven populations) used in this study, future research should include a larger and more representative set of populations (at least 30–50) to enable robust multivariate analyses and to validate the preliminary PCA and clustering results reported here. Conclusions Significant phenotypic differences were observed in stem height, stem diameter, number of auxiliary branches, leaf length, internode length, total fresh and dry weight, leaf dry weight, and stem dry weight among tarragon populations from different regions of Iran. However, contrary to our initial statement, correlation analysis revealed that the elevation of sampling sites was significantly correlated with several morphological traits, including leaf length (r = 0.429, P < 0.01), internode length (r = 0.523, P < 0.01), and the number of unhealthy leaves (r = 0.519, P < 0.01) (Table 10). Therefore, environmental factors (elevation) do influence the expression of certain morphological traits, and we have corrected this statement throughout the manuscript. The moderate to high heritability estimates observed in this study (e.g., for plant height, stem diameter, leaf area, and total fresh weight) suggest that these traits are under relatively strong genetic control under uniform growing conditions. However, it is critical to emphasize that this study assessed morphological (phenotypic) diversity only. No molecular markers (e.g., ISSR, SSR, or SNPs) were used. Consequently, claims about “genetic diversity,” “genetic relationships,” or “population structure” are not supported by the data and have been removed from this revised manuscript. Instead, we refer to “phenotypic grouping” or “morphological clustering” based on the measured traits. Robust statistical methods, such as PCA and cluster analysis, aid in understanding phenotypic relationships and morphological grouping among the populations. Cluster analysis grouped the seven populations into three distinct clusters at a Euclidean distance of 7.65, while Principal Component Analysis identified two components explaining 84.2% of the total phenotypic variance. These groupings should not be interpreted as evidence of genetic population structure; validation using molecular markers is essential before any conclusions about genetic relationships or gene flow can be drawn. Despite these limitations, the present study provides a useful preliminary assessment of phenotypic variation among Iranian tarragon wild populations. The Dehshad (Tehran) population, which showed superior performance for multiple agronomic traits, can be prioritized as a candidate for further evaluation in breeding programs. However, we strongly recommend that future research incorporate molecular marker analysis (e.g., ISSR, SSR, or genotyping-by-sequencing) to determine the true genetic diversity and population structure of Iranian tarragon germplasm, along with multi-year field evaluations across multiple environments. Until such data are available, the morphological characterization presented here should be considered a preliminary screening tool for identifying promising accessions, rather than a definitive assessment of genetic diversity or population structure.