Evaluation of the Farming Potential of Echinacea Angustifolia DC. Accessions Grown in Italy by Root-Marker Compound Content and Morphological Trait Analyses

The Echinacea genus includes a number of species that are commercially employed for the preparation of herbal products. Echinacea angustifolia DC. is one of these and is widely used, mainly for its immunomodulating properties, as it contains a wide range of compounds that belong to different chemical classes. In particular, echinacoside, cynarin and lipophylic alkylamides are the main specialized metabolites of the roots and can be considered to be marker compounds. In this work, 65 E. angustifolia accessions have been compared in a field trial in Italy, with the aim of investigating the variability/stability of the weight and chemical composition of their roots in order to identify the accessions that are most promising for future genetic-improvement programs. The morphological characteristics of the aerial parts have also been investigated. Seventeen samples were discarded due to germination or plantlet-development issues. Seven of the remaining accessions were identified as being different Echinacea species after a combined phytochemical and morphological evaluation. The morphological traits of the epigeal part, the root weight and the chemical composition data of the 41 confirmed E. angustifolia accessions were submitted to multivariate statistical analysis and a moderately homogenous sample distribution, with low selected-marker variability, was observed. Good echinacoside content was detected in almost all roots (>0.5%). However, two groups of accessions stood out because of their interesting features: One group possessed small roots, but had a high concentration of marker compounds, while another had highly developed roots and a good amount of marker compounds. These accessions can therefore be exploited for future selection work.


Introduction
The Echinacea genus (Asteraceae family) is indigenous to the Great Plains of North America and encompasses a small number of herbaceous grassland perennial taxa. Increasing interest in Echinacea as a commercial herbal product, including in herbal teas, capsules, tablets, powders, tinctures and beverages, has promoted its field cultivation, which has also been extended to Europe. E. angustifolia DC. (narrow-leaved coneflower), E. pallida (Nutt.) Nutt. (pale coneflower) and

Marker-Compound Quantification and Morphological Confirmation of the Accessions
Three E. angustifolia marker compounds, namely cynarin, echinacoside and dodeca-2E,4E,8Z,10E/ Z-tetraenoic isobutylamide, were identified and quantified in the roots of 3 or 4 individuals of each accession (Table S1). A representative chromatographic profile, obtained through the HPLC-PDA analysis, and the mass spectra of the commercial standards of the marker compounds are shown in Figure S1. The concentration of the three marker compounds ranged from 0 to 0.51% for cynarin, 0.39-2.86% for echinacoside and 0-0.9% for dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide (Table S2). The root weight was also recorded (from 1.1 to 53.3 g) (Table S2). Principal component analysis (PCA) was performed to better understand the variability/stability of these markers in the samples. The percentage contents of cynarin, echinacoside and dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide were considered as variables and, initially, all individuals were considered separately. In the score plot obtained for the first two components (85.8% of the total variation explained), the samples were homogenously dispersed with no positive or negative correlation with PC1 and PC2. A small cluster that negatively correlated with PC1 (accessions n • 50,54,55,57,58) was present ( Figure S2). In order to simplify the data under investigation, the same statistical analysis was applied using the mean values for each accession ( Figure 1). In this case, the analysis was expanded to the first three components. The combination of PC1 (59.6% of the variation explained) with both PC2 (30.7% of the variation explained) and PC3 (9.8% of the variation explained) gave results that were comparable to those reported in Figure S2, which had all the samples grouped together and a small separated cluster (accessions n • 50,54,55,57,58). The distribution of the variables suggests that there is negative correlation with the small cluster and cynarine and dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide content ( Figure 1B,D).
The concentration of the marker compounds in each accession was thoroughly evaluated. Neither cynarin nor dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide were detected in the five accessions that were grouped together (Table S2). This difference is confirmed by the box plot analysis for cynarin and dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide ( Figure 2). For cynarin, the box plot analysis highlights more outliers that are characterized by both lower (accession n • 51) and higher (accessions n • 16,60) quantities of this compound (Figure 2A). The concentration of the marker compounds in each accession was thoroughly evaluated. Neither cynarin nor dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide were detected in the five accessions that were grouped together (Table S2). This difference is confirmed by the box plot analysis for cynarin and dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide ( Figure 2). For cynarin, the box plot analysis highlights more outliers that are characterized by both lower (accession n° 51) and higher (accessions n° 16,60) quantities of this compound (Figure 2A). In order to better understand whether the phytochemical differences can be attributed to misleading species identification in the specific accessions, the main morphological traits of all the plants were considered. McGregor's taxonomy reports nine species and four varieties for Echinacea genus [21]. Most of the considered accessions exhibited morphological characters that are typical of E. angustifolia; dark green leaves with a rough surface, green stem with short and dense hairs (a reddish green stem was observed for accessions n° 6,17,18,19,30,31,43 and long and dense hairs were recorded for accessions n° 1,2,3,7,8). The color of the ray flowers was deep pink in most of the accessions, followed by pink (1, 2, 3, 6, 7, 9, 13, and 16) and deep pink-fuchsia (46, 47, 48, 52, 59, 60, 63, 64, and 65). The teeth of the ray flowers were >2 mm in 27 accessions, between 1 and 2 mm in 12 accessions (3,5,7,8,13,17,19,20,24,27, 33 and 34) and <1 mm in 2 accessions (9 and 16). At the same time, untypical morphological traits were recorded for accessions 49, 50, 51, 54, 55, 57, and 58. In particular, n° 49, 54, 55, 57, and 58 showed morphological characteristics that are typical of E. simulata or hybrids between that and E. angustifolia (tall plant, lanceolate leaves with long petiole, yellow pollen, long and dark pink-fuchsia ray flowers). The recorded characters of accession n° 51 were typical of E. pallida (tall plant, lanceolate leaves with long petiole, long and light pink ray flowers, whitish pollen). The morphological traits of n° 50 were typical of E. paradoxa var. neglecta (tall plants, leaves with a smooth and light green, pink ray flowers, yellow pollen) ( Figure 3). The combination of the quantification results and this first morphological evaluation led to accessions n° 49,50,51,54,55,57 and 58 being excluded in order to evaluate the variability/stability of those considered to be authentic In order to better understand whether the phytochemical differences can be attributed to misleading species identification in the specific accessions, the main morphological traits of all the plants were considered. McGregor's taxonomy reports nine species and four varieties for Echinacea genus [21]. Most of the considered accessions exhibited morphological characters that are typical of E. angustifolia; dark green leaves with a rough surface, green stem with short and dense hairs (a reddish green stem was observed for accessions n • 6,17,18,19,30,31,43 and long and dense hairs were recorded for accessions n • 1,2,3,7,8). The color of the ray flowers was deep pink in most of the accessions, followed by pink (1, 2, 3, 6, 7, 9, 13, and 16) and deep pink-fuchsia (46, 47, 48, 52, 59, 60, 63, 64, and 65). The teeth of the ray flowers were >2 mm in 27 accessions, between 1 and 2 mm in 12 accessions (3,5,7,8,13,17,19,20,24,27, 33 and 34) and <1 mm in 2 accessions (9 and 16). At the same time, untypical morphological traits were recorded for accessions 49, 50, 51, 54, 55, 57, and 58. In particular, n • 49, 54, 55, 57, and 58 showed morphological characteristics that are typical of E. simulata or hybrids between that and E. angustifolia (tall plant, lanceolate leaves with long petiole, yellow pollen, long and dark pink-fuchsia ray flowers). The recorded characters of accession n • 51 were typical of E. pallida (tall plant, lanceolate leaves with long petiole, long and light pink ray flowers, whitish pollen). The morphological traits of n • 50 were typical of E. paradoxa var. neglecta (tall plants, leaves with a smooth and light green, pink ray flowers, yellow pollen) ( Figure 3). The combination of the quantification results and this first morphological evaluation led to accessions n • 49,50,51,54,55,57 and 58 being excluded in order to evaluate the variability/stability of those considered to be authentic E. angustifolia accessions.

Phytochemical Evaluation and Morpho-Quantitative Characteristics of E. Angustifolia Accessions
The variability of cynarin, echinacoside and dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide concentrations in the remaining 41 E. angustifolia accessions was evaluated using PCA. Three principal components were selected and a homogenous sample distribution was observed, without the formation of distinctive clusters ( Figure S3), as reported in the score plot. Some accessions that were further separated from the main central group were either characterized by higher amounts of cynarin (accessions n° 16,17,60), or lower percentages of dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide (accession n° 25), as confirmed by the box plots ( Figure S4). A further investigation was performed by combining phytochemical data with agronomic features (i.e., number of germinated/developed plants) and a deeper morpho-quantitative analysis. The eight considered morphological traits varied in the accessions (Table S3). Accessions n° 6,5,18,25,17, and 45 were higher than accessions n° 2,8,21,27,34,37,38,39,40,43,44,46,47, and 48. The length and width of basal rosette leaves were variable: total length (petiole included) varied between 19.3 cm (accession 12) and 9.1 (accession 37), and width between 3.3 cm (accession 17) and 1.4 (accession 37). The shape of the leaf blade was also very variable as its "leaf length/width ratio" varied from 5.2 to 11.8. The leaf shape of accessions 25, 22, 9, and 24 was linear-lanceolate, while that of the remaining accessions was more or less lanceolate. The stems per plant varied from 1 (accession 64) to 12 Further elaborations (PCA and HCA, hierarchical cluster analysis) were therefore carried out, and the agronomic and morphological features of the plants were considered together with the phytochemical profile of the roots (in terms of concentration and absolute yield of the marker compounds). The PCA loadings plot ( Figure 4B) of the first two components (45.8% of the total

Phytochemical Evaluation and Morpho-Quantitative Characteristics of E. Angustifolia Accessions
The variability of cynarin, echinacoside and dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide concentrations in the remaining 41 E. angustifolia accessions was evaluated using PCA. Three principal components were selected and a homogenous sample distribution was observed, without the formation of distinctive clusters ( Figure S3), as reported in the score plot. Some accessions that were further separated from the main central group were either characterized by higher amounts of cynarin (accessions n • 16,17,60), or lower percentages of dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide (accession n • 25), as confirmed by the box plots ( Figure S4). A further investigation was performed by combining phytochemical data with agronomic features (i.e., number of germinated/developed plants) and a deeper morpho-quantitative analysis. The eight considered morphological traits varied in the accessions (Table S3). Accessions n • 6,5,18,25,17, and 45 were higher than accessions n • 2,8,21,27,34,37,38,39,40,43,44,46,47, and 48. The length and width of basal rosette leaves were variable: total length (petiole included) varied between 19.3 cm (accession 12) and 9.1 (accession 37), and width between 3.3 cm (accession 17) and 1.4 (accession 37). The shape of the leaf blade was also very variable as its "leaf length/width ratio" varied from 5.2 to 11.8. The leaf shape of accessions 25, 22, 9, and 24 was linear-lanceolate, while that of the remaining accessions was more or less lanceolate. Further elaborations (PCA and HCA, hierarchical cluster analysis) were therefore carried out, and the agronomic and morphological features of the plants were considered together with the phytochemical profile of the roots (in terms of concentration and absolute yield of the marker compounds). The PCA loadings plot ( Figure 4B) of the first two components (45.8% of the total variation explained) showed a positive correlation between PC1 and plant height, leaf length, root weight, and absolute amount of the marker compounds in the roots. A positive correlation was also observed with the number of germinated and developed plants. On the other hand, a negative correlation between PC1 and the concentrations of echinacoside and dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide is noticeable. The Pearson correlation coefficient calculation (Table S4)  variation explained) showed a positive correlation between PC1 and plant height, leaf length, root weight, and absolute amount of the marker compounds in the roots. A positive correlation was also observed with the number of germinated and developed plants. On the other hand, a negative correlation between PC1 and the concentrations of echinacoside and dodeca-2E,4E,8Z,10E/Ztetraenoic isobutylamide is noticeable. The Pearson correlation coefficient calculation (Table S4) confirmed these results. In particular, the root weight is statistically (p < 0.05) positively correlated with plant height, leaf length, N° of stems and flower/plants, and absolute amount of all three marker compounds and number of transplanted plants, while it is statistically negatively correlated to the concentration of echinacoside. The PCA scores plot ( Figure 4A) showed a moderately homogenous sample distribution. However, three clusters can be identified: The first cluster includes accessions 3, 5, 12, and 18 and is characterized by tall plants with a high root weight and a high absolute amount of the marker compounds; the second cluster includes accessions 6, 13, 16, 17, and 25 and is characterized by tall plants, but with a low root weight and marker-compound content; finally, a third group of samples (accessions 37, 38, 39, 43, 59, and 60) is characterized by plants with a low plant height and root weight, but a high concentration of marker compounds. These groups are confirmed by the hierarchical cluster analysis reported in Figure 5.  The PCA scores plot ( Figure 4A) showed a moderately homogenous sample distribution. However, three clusters can be identified: The first cluster includes accessions 3, 5, 12, and 18 and is characterized by tall plants with a high root weight and a high absolute amount of the marker compounds; the second cluster includes accessions 6, 13, 16, 17, and 25 and is characterized by tall plants, but with a low root weight and marker-compound content; finally, a third group of samples (accessions 37, 38, 39, 43, 59, and 60) is characterized by plants with a low plant height and root weight, but a high concentration of marker compounds. These groups are confirmed by the hierarchical cluster analysis reported in Figure 5.

Discussion
In this work, sixty-five E. angustifolia accessions were gathered from a germplasm bank, botanical gardens and seed companies and cultivated in a controlled experimental field. The first issue that was encountered was seed germination and the development of the plantlets. In fact, E. angustifolia is one of the most difficult echinacea species to grow successfully and is characterized by either poor or erratic germination, caused mainly by seed dormancy. Nevertheless, in line with previous works and thanks to treatment with gibberellic acid, 73.8% of accessions resulted in successful plantlet development [5,15,19,22]. The phytochemical profile of about 3/4 individuals from the resulting 48 accessions was evaluated first. Although few individuals were considered, the high number of the investigated accessions resulted in a high amount of data that were subjected to a multivariate data analysis (i.e. PCA, HCA) to obtain robust and evident results. Three marker compounds were selected. Echinacoside is the main caffeic acid derivative of E. angustifolia roots, but is also present in E. pallida roots. Cynarin, on the other hand, is typical of E. angustifolia roots. The roots of E. angustifolia contain large quantities of alkylamides, which are less abundant in E. purpurea aerial parts and mostly absent from the roots of E. pallida. Of the alkylamides, dodeca-2E,4E,8Z,10E/Ztetraenoic acid isobutylamide is considered to be the main constituent [3,13,20,23]. The principal component analysis, based on the percentage content of the compounds as its variables, clearly identified five accessions (n° 50,54,55,57 and 58) as being classified as other Echinacea species. These results were confirmed by the morphological analysis. In addition, untypical morphological traits were observed for two other accessions (n° 49 and 51), which were not clearly chemically discriminated. E. angustifolia plants are 30-60 cm in height with one or more stems being mostly unbranched. The leaves are lanceolate to linear-lanceolate and the inflorescences are characterized by a cone-shaped disk and by purple, pale pink or barely white spreading ray flowers [24]. The misidentification and confusion of Echinacea species is common since they are morphologically similar and their systematics are still unclear [2,4,21,25]. The combination of the morphological and chemical characteristics of these accessions allowed us to assume that accessions n° 54,55,57, and 58, which are characterized, besides their morphological traits, by an absence of cynarin and dodeca-2E,4E,8Z,10E/Z-tetraenoic acid isobutylamide, can be identified as E. simulata. On the basis of its chemical profile, n° 49 is probably a hybrid and n° 50 and 51 can be identified as E. paradoxa var neglecta and E. pallida, respectively. [1,20,23]. For the accession 51, it would have been useful to make the analysis of the chromosome number, but being the only doubtful accession, this analysis was not carried out. We based the identification, as for the others accessions, on the morphological traits and

Discussion
In this work, sixty-five E. angustifolia accessions were gathered from a germplasm bank, botanical gardens and seed companies and cultivated in a controlled experimental field. The first issue that was encountered was seed germination and the development of the plantlets. In fact, E. angustifolia is one of the most difficult echinacea species to grow successfully and is characterized by either poor or erratic germination, caused mainly by seed dormancy. Nevertheless, in line with previous works and thanks to treatment with gibberellic acid, 73.8% of accessions resulted in successful plantlet development [5,15,19,22]. The phytochemical profile of about 3/4 individuals from the resulting 48 accessions was evaluated first. Although few individuals were considered, the high number of the investigated accessions resulted in a high amount of data that were subjected to a multivariate data analysis (i.e. PCA, HCA) to obtain robust and evident results. Three marker compounds were selected. Echinacoside is the main caffeic acid derivative of E. angustifolia roots, but is also present in E. pallida roots. Cynarin, on the other hand, is typical of E. angustifolia roots. The roots of E. angustifolia contain large quantities of alkylamides, which are less abundant in E. purpurea aerial parts and mostly absent from the roots of E. pallida. Of the alkylamides, dodeca-2E,4E,8Z,10E/Z-tetraenoic acid isobutylamide is considered to be the main constituent [3,13,20,23]. The principal component analysis, based on the percentage content of the compounds as its variables, clearly identified five accessions (n • 50,54,55,57 and 58) as being classified as other Echinacea species. These results were confirmed by the morphological analysis. In addition, untypical morphological traits were observed for two other accessions (n • 49 and 51), which were not clearly chemically discriminated. E. angustifolia plants are 30-60 cm in height with one or more stems being mostly un-branched. The leaves are lanceolate to linear-lanceolate and the inflorescences are characterized by a cone-shaped disk and by purple, pale pink or barely white spreading ray flowers [24]. The misidentification and confusion of Echinacea species is common since they are morphologically similar and their systematics are still unclear [2,4,21,25]. The combination of the morphological and chemical characteristics of these accessions allowed us to assume that accessions n • 54,55,57, and 58, which are characterized, besides their morphological traits, by an absence of cynarin and dodeca-2E,4E,8Z,10E/Z-tetraenoic acid isobutylamide, can be identified as E. simulata. On the basis of its chemical profile, n • 49 is probably a hybrid and n • 50 and 51 can be identified as E. paradoxa var neglecta and E. pallida, respectively. [1,20,23]. For the accession 51, it would have been useful to make the analysis of the chromosome number, but being the only doubtful accession, this analysis was not carried out. We based the identification, as for the others accessions, on the morphological traits and the chemical profile that, in this case, revealed the absence of cynarin and presence of dodeca-2E,4E,8Z,10E/Z-tetraenoic acid isobutylamide in the roots. As reported by Jedrzejczyk (2020), the identification of Echinacea species is difficult, even from the biomolecular point of view; E. angustifolia is really very similar to other Echinacea species, such as E. paradoxa var. neglecta and E. pallida.
The multivariate statistical analysis of the marker-compound content in the considered authentic E. angustifolia accessions confirmed that the sample distribution was quite homogenous. However, when considering the root weight, the morphological traits of the aerial part and the root phytochemical features together, the correlations between some parameters can be noticed and some groups of accessions can be discriminated. In particular, the plants with well-developed aerial parts also present a good root-weight yield; these plants are also characterized by high vegetative success. On the other hand, the concentration of the marker compounds in these plants is lower. This can be attributed to the fact that the highest content of metabolites is found in the smaller and thinner roots, due to an accumulation of such constituents in the root cortex of the taproot [8].
These considerations and the results of the multivariate statistical elaborations allowed us to identify two groups of accessions as having interesting features. In particular, one group of accessions (i.e., accessions 37, 38, 39, 43, 59, and 60) is characterized by plants with small roots, but a considerably high concentration of marker compounds: For accessions 37 and 38, the concentration of echinacoside and dodeca-2E,4E,8Z,10E/Z-tetraenoic acid isobutylamide is higher than the mean values obtained for these compounds; for accession 39, the cynarin and dodeca-2E,4E,8Z,10E/Z-tetraenoic acid isobutylamide concentrations are higher than the mean; for accession 43, the echinacoside and cynarin concentrations are higher than the mean; while for accessions 59 and 60, the concentrations of all three of the marker compounds are higher than the mean. These accessions can therefore become a reference quality standard. On the other hand, a second group of accessions (i.e., accessions 3, 5, 12 and 18) is characterized by highly-developed roots, and a, nevertheless, good amount of the marker compounds, not only in terms of absolute amount, but also in terms of their concentration: For accession 3, the dodeca-2E,4E,8Z,10E/Z-tetraenoic acid isobutylamide concentration is higher than the mean value for this compound; for accession 5, the cynarin concentration is higher than the mean; for accession 12, the cynarin and echinacoside concentrations are higher than the mean; while for accession 18, the concentrations of all three of the marker compounds are higher than the mean. This group of accessions shows good vegetative development and promising marker-compound content, at the same time, and can therefore be exploited for future selection works.

Plant Material and Morpho-Quantitative Evaluation
Seeds belonging to 65 E. angustifolia accessions were obtained from different sources, as listed in Table 1. Seeds were treated with gibberellic acid (0.2 g L −1 ) and sowed, at the end of March of the first year, into plastic boxes containing a mixture of a substrate for horticulture (Manna Flor PT, Manna, Bolzano, Italy) and sand (about 2/3 and 1/3 in volume, respectively) and kept in a greenhouse at approximately 8 • C. Manna Flor PT composition was as follows: white peat (H2-H4) 25%, black peat (H6-H9) 75% with pH (CaCl 2 ) 5.  Table S1). The single roots were weighed, cut and dried (40 • C for 48 h), and subjected to solvent extraction for subsequent phytochemical analysis.

Extraction of Plant Material
One hundred milligrams of each dried and ground root were submitted to ultrasonic extraction with 10 mL of methanol/water (70:30, v/v), twice, for 10 min each. The resulting two extracts were combined and centrifuged at 4000 rpm for 10 min. The supernatant was brought to a volume of 30 mL and filtered through a 13 mm diameter, 0.22 µm pore diameter, and hydrophilic PTFE syringe filter, prior to HPLC-PDA-MS/MS analysis. The repeatability of extract composition was evaluated on 35 accessions with the RSD% between three extracts < 5%. The efficiency of the extraction process was determined by calculating the recovery of the three markers, which were 119%, 112% and 101% for cynarin, echinacoside and dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide, respectively.

Qualitative Analysis
Each extract (5 µL) was analyzed using a Shimadzu Nexera × 2 system coupled with a photodiode array detector SPD-M20A in series to a Shimadzu LCMS-8040 triple quadrupole system equipped with an electrospray ionization (ESI) source (Shimadzu, Dusseldorf, Germany) for the identification of the three markers.
Mass spectrometer operative conditions were: heat block temperature, 200 • C; desolvation line (DL) temperature, 230 • C; nebulizer gas (N 2 ) flow rate, 3 L/min; and drying gas (N 2 ) flow rate: 15 L/min. Mass spectra were acquired both in positive and in negative full-scan mode, over the range 50-1000 m/z, event time 0.5 s.
The three markers were identified in the extracts by comparing their retention times, UV and MS spectra with those of authentic standards.

Quantitative Analysis
Each extract (5 µL) was analyzed in triplicate using a Shimadzu UFLC XR (Shimadzu, Dusseldorf, Germany) equipped with a photodiode array detector SPD-M20A with the same column, mobile phases, flow rate, and gradient program as in the qualitative analysis (see Section 4.4.1).
UV spectra were acquired in the 220-450 nm wavelength range, and the resulting chromatograms were integrated at 325 nm for the quantification of echinacoside and cynarin, and at 254 nm for dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide.
Quantitation was performed using an external standard calibration method, and the results expressed as % weight/weight. The calibration curve of cynarin was built by analyzing it at five concentrations in the range 1-50 mg/L. For echinacoside, five concentrations in the range 1-100 mg/L were used, and for dodeca-2E,4E,8Z,10E/Z-tetraenoic isobutylamide, six concentrations in the range 0.5-100 mg/L were used. The determination coefficient (R 2 ) was higher than 0.998 in all cases. Analytical performance was measured in terms of repeatability and intermediate precision (RSD% never exceeding 3%).
All data were processed using Lab Solution software, version 5.92 (Shimadzu, Dusseldorf, Germany).

Statistical Analysis
Multivariate analysis (PCA, principal component analysis and HCA, hierarchical cluster analysis) as well as Pearson correlation was carried out on the raw data; the clustering was based on Euclidean distance with the group average method used as fusion criterion. All the computations were evaluated using SPSS 15.0 (IBM Corporation, Armonk, NY, USA) software.

Conclusions
The results obtained for several accessions that were considered as E. angustifolia confirm, from one side, the difficulties in the cultivation and the correct identification of this species. The combination of the main morphological traits and the quantification of cynarin and dodeca-2E,4E,8Z,10E/Z-tetraenoic acid isobutylamide marker compounds can help in the recognition of other possible Echinacea species, although the taxonomy of this genus is still unclear. The E. angustifolia samples that are considered authentic have a similar chemical profile with acceptable echinacoside content, in agreement with the European Pharmacopoeia (2020). However, when combining root weight, the morphological traits of the epigeal part and phytochemical data, some accessions stand out thanks to their interesting features. In particular, a set of accessions with small roots, but a high concentration of marker compounds, can be considered a quality reference standard, while another group with highly-developed roots and a good amount of the marker compounds is a good compromise between the search for the optimal agronomic and phytochemical features. These accessions could therefore be exploited as starting materials for future work selection.