Tocotrienol-Dominated Berberidaceae Species’ Seed Tocochromanols: Screening via Ultrasound-Assisted Extraction in Ethanol

Abstract

Inspired by the lack of wide-scale family level screenings, the profile of tocochromanols in Berberidaceae family species belonging to the Berberis, Mahonia, Caulophyllum, Jeffersonia and Podophyllum genera was studied. Seeds were acquired from botanical gardens around the world and tocopherol and tocotrienol content was tested using ultrasound-assisted extraction in ethanol (UAEE) and compared to saponification protocol and analyzed by an RP-HPLC-FLD system. The UAEE protocol produced 93% average tocochromanol recovery compared to the saponification protocol. All investigated samples were tocotrienol-dominated, the lowest proportions being in B. regeliana, B. thunbergii and B. aristata at means of 55%, 56% and 58%, respectively. The main tocochromanol constituents were α-tocotrienol and γ-tocotrienol. The highest α-tocotrienol content was observed in B. tchonskyana at 9.14 mg 100 g⁻¹ dw, and the highest γ-tocotrienol and sum of free tocochromanol content was observed in J. diphylla at 18.00 and 23.76 mg 100 g⁻¹ dw, respectively. Principal component analysis and k-means cluster analysis based on a free tocochromanol profile indicated γ-tocotrienol and α-tocotrienol content as the main differentiators. However, a comprehensive sample set could only be collected for the Berberis genus, warranting further research into Berberidaceae seed tocochromanols.

1. Introduction

The barberry (Berberidaceae) family is a flowering eudicot plant family in the Ranunculales order. It encompasses 13 genera, Achlys, Berberis, Bongardia, Caulophyllum, Epimedium, Gymnospermium, Jeffersonia, Loentice, Nandina, Plagiorhegma, Podophyllum, Ranzania and Vancouveria, and about 770 species, depending on the source. It is found worldwide in temperate regions and on tropical mountains. Additionally, the Mahonia genus used to be separate from Berberis but is now accepted as part of Berberis, and the once separate Diphylla genus is now accepted under the Podophyllum genus, which is subdivided into Podophyllum, Dysosma and Sinopodophyllum subgenera. Members of the family are common across Eurasia and the Americas, although they are only cultivated in Europe, parts of Asia and North America [1,2]. Berberis may be considered the most common of the genera and the most industrially significant.

Although several Berberis species are generally used as decorative shrubs, several are used in herbal medicine—B. vulgaris (European barberry) berries are high in vitamin C and therapeutic alkaloids, such as berberine, which is a promising agent in cancer [3,4]—and other disease therapies, owing to its antioxidant, neurotransmitter, enzyme and immunomodulation capabilities [5]. In addition to berberine, other alkaloids have also been observed in different Berberis species, including oxyberberine, palmatine, isocoridine, lambertine, magniflorine, and oxycanthine in B. vulgaris, berbamine in B. aristate, (+)-N-methylcoclaurine, (−)-pronuciferine, (+)-9-hydroxynuciferine, and (+)-orientine in B. montana, berbaumine, aromaline, and isotentrandine in B. stolonifera and jatrorrhizine in B. umbellata [1]. Like berberine, they are largely discussed as anticancer agents rather than toxins. The main polyphenolic compounds in barberries are rutin and apigenin [6]. In cuisine, barberries (B. vulgaris) are sometimes used in seasonings to add acidity to a dish, while B. microphylla (calafate) and B. darwinii (michay) are used in Argentinean and Chilean cuisines, made into jams and other desserts.

Tocotrienols (T3s) and tocopherols (Ts) are the most common tocochromanols. Both are prenyllipid antioxidants consisting of a chromane ring, a lipophilic branched aliphatic chain, saturated in tocopherols, and unsaturated with three double carbon bonds in tocotrienol molecules. Tocochromanols occur in virtually all plant organs and across most plant lineages. Yet, in the majority of species, tocopherols prevail—particularly α-tocopherol, which dominates in green tissues, and γ-tocopherol, which is typically the main form in seeds—whereas tocotrienols are frequently absent or present only at low levels, largely confined to seeds. These lipophilic metabolites play a central antioxidative role in plants by limiting oxygen toxicity. Specifically, tocopherols and tocotrienols intercept lipid peroxyl radicals and thus inhibit the chain propagation of membrane lipid peroxidation. Generated tocopheroxyl and tocotrienoxyl radicals can subsequently be reduced back to their native forms through the action of other antioxidants; conversely, tocochromanols may also contribute to preserving other antioxidant pools from oxidation. Inevitably, this protective function can translate into a gradual depletion of tocochromanols in plant material as oxidative processes proceed [7]. Despite their chemical lability, the diagnostic value of tocochromanols has been widely exploited in authenticity assessments as markers, for instance, to detect butter [8] and Arabica coffee [9] adulteration with cheaper substitutes. This is made possible by characteristic tocochromanol composition in certain food matrices. One such distinctive pattern is tocotrienol-dominated tocochromanol profiles. A number of articles have documented tocotrienol scarcity in dicotyledonous plants [10,11], but recent broad investigations of dicot families indicate that several dicot families’ seed tocochromanols are tocotrienol-dominated [12,13,14,15]. This finding offers an opportunity to identify new potential sources of tocotrienols.

Phylogeny is related to the appearance of several compounds within and between plant families, including different alkaloid classes and non-protein amino acids in the legume (Fabaceae) family, different alkaloids in the nightshade (Solanaceae) family, iridoids and sesquiterpenes in the Lamiaceae family [16], and specific pyrrolizidine alkaploids and phenolid compounds in Boraginaceae species [17,18]. Apiaceae species’ fruits and seeds are chemically linked due to the presence of petroselinic acid and tocotrienol-dominated tocochromanol profiles [13,14,15,19]. Other studies report that several classes of rare bioactive secondary metabolites, including naphthodianthrones, phloroglucinol derivatives, xanthones, and others, have been identified in members of the Hypericum genus. Recent evidence suggests that the Hypericum genus may also be characterized by the occurrence of appreciable levels of tocotrienols in most parts of the plant [20,21,22]—compounds that are highly scarce in photosynthetic tissues. However, certain secondary metabolites may only appear in certain branches of a family, like tocotrienols and tocochromanols do in the Rutaceae family [23], and factors affecting their production, such as ontogenetic, diurnal, and seasonal variation, should be considered. Therefore, despite preliminary findings that indicate that certain classes of secondary metabolites are chemotaxonomically linked at lower taxonomic levels, a cautionary note must be added [24]. Although phytochemical traits are highly unlikely to be grounds for reconsidering phylogenetic structures, they are highly informative when considering inherent and adaptive metabolic processes in plants.

Chemotaxonomy generally refers to the classification of a species based on molecular structure instead of or in combination with genetic information, but is used as a term to describe a chemical similarity of a genetically similar plant in the present study due to lack of an appropriate synonym. It has proven a useful hypothesis to assist in finding numerous plant families and species whose seeds primarily contain tocotrienols in Vitaceae [25], Celastraceae [26], Cornus (Cornaceae) [27], Ilex (the only remaining genus in Aquifoliaceae) [28], and Rutaceae [23]. It is based on the following formula. If tocotrienols have been reported in plant species “species 1” and “species 2” belonging to the same family “family 1”, then it is highly probable to find the presence of uninvestigated plant species “species 3”, “species 4”, and “species n” in “family 1”. Using this approach, a dominance and new sources of tocotrienol have been found in the seeds of the species belonging to the following families: Aquifoliaceae [28], Cornaceae [27], Celastraceae [26], Rutaceae [23], and Vitaceae [25], and the leaves of the Hypericum and Clusia genus [20,21]. There is some existing evidence of tocotrienols in Berberis species—the presence of γ-tocotrienol and α-tocotrienol in B. microphylla [29]. Another report indicates not only the presence of γ-tocotrienol and α-tocotrienol, but also β-tocotrienol. However, based on the chromatograms, this is probably a case of misidentification [30]. In B. integerrima seed oil, only the presence of α-tocopherol and γ-tocopherol has been reported because tocochromanol standards were limited to tocopherols [31,32]. Information on tocotrienol contents in other Berberidaceae genera is not available. The lack of tocotrienol standard use is probably one of the main reasons for the underestimation of tocotrienols in nature [33]. Therefore, in the present study, with the cooperation with the botanical gardens around the world, 44 Berberidaceae species seeds were investigated to verify the potential of chemotaxonomical tools for searching for new species rich in tocotrienols.

2. Materials and Methods

2.1. Reagents

Potassium hydroxide, pyrogallol, sodium chloride (reagent grade), n-hexane, methanol, ethyl acetate, and ethanol (HPLC grade) were obtained from Sigma-Aldrich (Steinheim, Germany). The 96.2% (v/v) ethanol was received from SIA Kalsnavas Elevators (Jaunkalsnava, Latvia). The eight tocochromanol standards, α, β, γ, and δ homologs of tocopherols and tocotrienols (>95%, HPLC), were purchased from LGC Standards (Teddington, Middlesex, UK) and Merck (Darmstadt, Germany).

2.2. Plant Material

Plant material of the Berberidaceae family was obtained from botanical gardens across the world, mainly Eurasia (Austria, Belgium, Czech Republic, Germany, Estonia, France, Hungary, Iceland, Kyrgyz Republic, Latvia, Lithuania, Poland, Portugal, Russia, Slovakia, and Ukraine) and North America (Canada), sent via mail, as part of seed exchange programs between botanical gardens. The Berberidaceae family was one of over a hundred families investigated as part of this project. The full list of botanical gardens that supported this project can be found in the Supplementary Materials. The species verification of the provided plant material (seeds) was performed by the staff of the donor botanical garden using their genotype resources. To reduce the impact of factors such as risk of misidentified species, crossbreeding, environmental factors, and others, origin (botanic garden) and diversification were prioritized alongside species diversity. Synonymic species were checked using online databases, such as wikispecies.com (Accessed on 10 January 2024) (classification into subfamilies) and worldfloraonline.com (Accessed on 10 January 2024) (synonymic species), using the consensus or most recent reference available. Species in the Mahonia genus were treated as part of the Berberis genus, except for Mahonia × wagneri, which is listed as a synonym of Berberis aquifolium in some sources, and as a separate nothospecies in others. Sinopodophyllum was treated as part of Podophyllum. Original species names, and the number of replications for each species, provided by the botanical garden are provided in the Supplementary Materials. Seeds were obtained and analyzed between 2019 and 2024. Altogether, 44 species’ seeds from 4 genera were analyzed, as well as one Mahonia hybrid. A small number of seeds were collected and air-dried at ambient temperature to retain viability before this research period, but this did not affect the results, as the seeds provided by the botanical gardens were from the preceding vegetative season. The obtained seeds were cataloged as they were received and cleaned from other plant part residues, e.g., fruit flesh, if required. Upon receipt, samples were analyzed within ~6–41 months, with an average 17 months after the collection of seeds. For more details about each analyzed sample, please see the Supplementary Materials. The seeds were stored in paper and plastic bags, away from direct light, under room-temperature conditions (21 ± 3 °C). Seed moisture of the received samples can be estimated at 10 ± 3%. Seeds were frozen at −80 °C for 1–3 h, freeze-dried using a FreeZone freeze-dryer system (Labconco, Kansas City, MO, USA) at a temperature of −51 ± 1 °C and < 0.01 mbar for 24–48 h, depending on the size and number of seeds. Lyophilized seeds contained 3–7% moisture. Due to generally limited obtained seed mass, an average moisture content of value 5% was used as the default/constant for tocochromanol calculation for all samples (0–2% systematic bias). Dry seeds (0.1–1 g) were powdered using an MM 400 mixer mill (Retsch, Haan, Germany) and tocochromanols were extracted within the same day using ultrasound-assisted extraction and 96.2% (v/v) ethanol (UAEE), as described below in Section 2.3.2 (all samples), and Section 2.3.1 for seven randomly selected samples (recovery study).

2.3. Tocochromanol Extraction

2.3.1. Saponification

The saponification protocol of powdered seed samples (0.05–0.10 g) was performed in the presence of 2% (w/v) pyrogallol (an antioxidant) in ethanol and 60% (w/v) aqueous potassium hydroxide, and then the samples were incubated in a water bath at 80 °C for 25 min. After saponification, tocochromanols were extracted three times using n-hexane:ethyl acetate solution (9:1, v/v). The details of the saponification step and subsequent tocochromanol extraction were reported earlier [34]. The organic solvent was evaporated, and the remaining sample was re-dissolved in 1 mL of ethanol, transferred to 2 mL glass vials and analyzed immediately by a reversed-phase liquid chromatography system with fluorescence detection (RPLC-FLD).

2.3.2. Ultrasound-Assisted Extraction in Ethanol (UAEE)

The greener method was adopted from a protocol developed for the extraction of tocochromanols from cranberry seeds [34]. Briefly, the powdered seeds (0.05–0.10 g) were placed in 15 mL tubes and supplemented with 96.2% (v/v) ethanol (5 mL), mixed (1 min) at 3500 rpm using vortex and treated by ultrasound of nominal ultrasonic power 160 W and ultrasound frequency 35 kHz using a Sonorex RK 510 H ultrasonic bath (Bandelin Electronic, Berlin, Germany) at 60 °C for 15 min. Extraction was carried out in sealed tubes and within a covered ultrasonic bath to restrict oxygen and light exposure. Immediately after completion of the ultrasonic step, the samples were mixed (1 min) as before, centrifuged at 11,000× g at 21 °C for 5 min, and transferred directly to a 2 mL glass vial and analyzed in an RPLC-FLD system.

To examine whether antioxidant supplementation contributed to the preservation of tocochromanols during UAEE, we performed an additional experiment on Berberis anhweiensis and Berberis tischleri. Seeds were collected in September–October 2023 and stored in paper bags at room temperature (21 ± 3 °C) until analysis in January 2026. UAEE was performed as described above, with the sole modification that pure ethanol was replaced by an ethanolic solution of pyrogallol (2%, w/v).

2.3.3. Method Validation

Since the extraction of tocopherols and tocotrienols from all tested seeds using UAEE differs from most studies investigating tocochromanol content in plant material, the results were compared with the standard saponification protocol. Recovery (%) tests of tocopherols and tocotrienols from the seeds of seven randomly selected species (Berberis genus: B. aquifolium, B. sieboldii, B. thunbergii, B. anhweiensis, and B. tischleri, and Podophyllum genus: P. hexandrum and P. peltatum) (7 × 3 UAEE vs. 7 × 3 saponification) were performed. Measurement repeatability (%) for both extraction protocols was evaluated. Repeatability (coefficient of variation) was calculated based on the independent determinations of a sample by analyzing three replicates on the same day. Error of measurement (standard deviation) was calculated based on the independent determinations of a sample by analyzing three replicates on the same day.

2.4. Tocochromanol Determination by Reversed-Phase Liquid Chromatography with Fluorescent Detection (RPLC-FLD)

The determination of four tocopherols and four tocotrienols was done according to a reported method using authentic standards and calculated using calibration curves produced earlier [8]. Separation was performed on a Luna PFP column (3 µm, 150 × 4.6 mm) (Phenomenex, Torrance, CA, USA) using 93% (v/v) methanol as mobile phase with 1 mL/min flow rate and a column oven at 40 °C temperature. Measurements were done on a LC 10 series (Shimadzu, Kyoto, Japan) system equipped with an RF-10AXL fluorescence detector using the following detection parameters of excitation and emission, λ_ex = 295 nm and λ_em = 330 nm, respectively. Details of method validation are provided in Table S1 (Supplementary Materials).

2.5. Statistical Analysis

The results of all the performed experiments of different species seed samples are presented as means ± standard deviation (n = 2–11). Since β-T and δ-T were absent from the samples, they were not included in statistical analyses. Base R and opensource R package dplyr 1.2.0 were used for multivariate analysis of variance (MANOVA). The dataset was not normally distributed, which is typical for similar datasets and screenings, so the Kruskal–Wallis test was employed as well using the agricolae package, and factoextra and ggpubr were used for principal component analysis (PCA) and k-means cluster analysis, and ggplot2, ggthemes, gghighlights, ggextra, ggrepel, scales and forcats were used for data visualization. Data was analyzed without transformation, differences were considered significant at p < 0.05, and outliers were included in the analysis to retain natural variability in the dataset.

3. Results and Discussion

3.1. Differences in Tocochromanol Recovery by Two Protocols: Saponification and UAEE

Saponification is the most common sample preparation protocol for tocopherol and tocotrienol analysis. It has the highest tocochromanol recovery [8], but requires long preparation time, uses nauseating solvents such as hexane, and does not distinguish between free and esterified tocochromanols [33]. The present study used a simplified UAEE of free tocochromanols, which is a less time-consuming, more cost-efficient, and more environmentally friendly method for testing a large number of samples. The method has proven suitable for tocochromanol analysis of the seeds of cranberry (Vaccinium macrocarpon) [34], grape (Vitis spp.) [35], other species belonging to the Vitaceae family—such as Ampelopsis japonica, Cyphostemma juttae, Parthenocissus quinquefolia, and three species of the Vitis genus, V. aestivalis, V. coignetiae, and V. riparia [25]—as well as Hypericum perforatum inflorescences [36], providing recovery similar to a saponification protocol. However, the results should not be compared with saponified sample scores directly and a pilot study with saponification is advisable when investigating new plant material. Esterified tocochromanols can be a minor or major fraction in plant material, their proportion ranging from almost none to almost all tocochromanols. However, that evidence is restricted to a single study, and it considers only bell pepper, chili pepper, cucumber, and walnut [37]. Additionally, tocochromanols can be physically bound in the plant material [33]. Therefore, seven species’ seeds—five from the Berberis genus, B. aquifolium, B. anhweiensi, B. sieboldii, B. tischleri and B. thunbergii, and two from the Podophyllum genus, P. hexandrum, and P. peltatum—were prepared using of both UAEE and a saponification protocol. Recovery differed between tocochromanols and species. Recovery (relative to the saponification protocol) of individual tocochromanols is as follows: 75–94% for α-tocopherol, 90–98% for α-tocotrienol, and 72–99% for γ-tocotrienol—detected in all seven samples—and 51–90% for γ-tocopherol, 90–94% for β-tocotrienol and 86–97% for δ-tocotrienol—detected in all three samples. β-tocopherol and δ-tocopherol were not detected in any of the seven tested samples. This outcome is largely as anticipated. β- and δ-tocochromanol homologues are rare and, when present, typically accumulate at much lower concentrations than the α- and γ-homologues—with some notable exceptions. β-Tocopherol dominance can be found in wheat germ (Triticum aestivum), robusta and arabica coffee (Coffea canephora and Coffea arabica) beans, oak (Quercus rubra) acorns, apple (Malus spp.) and Guelder-rose (Viburnum opulus) seeds. Meanwhile, predominance of δ-tocopherol has only been found in the seeds of Borago genus (Boraginaceae family) and Echium gentianoides. High concentrations of β-tocotrienol can be found in bran oils of spelt (Triticum spelta) and wheat (Triticum aestivum), and black caraway (Nigella sativa) seed oil. High content of δ-tocotrienol can be found in annatto (Bixa orellana) seed oil and rubber tree (Hevea brasiliensis) latex [33].

The sum of tocopherol and tocotrienol recovery from seeds of Berberidaceae species ranged between 75 and 94% and 87–98%, with average values of 85% and 94%, respectively (Figure 1, Table S2 in Supplementary Materials). Supplementing the extraction solvent with pyrogallol (2%, w/v in ethanol) did not alter free tocochromanol recoveries relative to UAEE performed without antioxidant addition. This result suggests that, under the conditions employed (15 min at 60 °C), UAEE does not induce measurable losses of free tocopherols and tocotrienols (Table S3 in Supplementary Materials). The slightly higher results obtained for saponified samples in comparison to UAEE can be explained by releasing the free tocochromanols from ester or glucoside derivatives, or non-extractable, physically bonded tocochromanols [33]. Conversely, in a few instances, the standard deviation bars imply recoveries above 100%, primarily as a consequence of the analytical variability inherent to extraction and quantification workflows. However, saponification is not necessarily a loss-free reference procedure. Degradation of tocochromanols may occur during alkaline hydrolysis, even when pyrogallol is included as a protective antioxidant. Indeed, in grape seeds, saponification has been associated with tocotrienol losses, consistent with incomplete protection [35]. Such observations suggest that pyrogallol-mediated protection may be insufficient under certain conditions, or that the pyrogallol concentration is not optimal. Pyrogallol, by far the most frequently used additive, has been applied across a broad concentration range in ethanol (2% (w/v), present work; 2.5% (w/v) [8]; up to 6% (w/v) [38,39]. Crucially, pyrogallol is not a simple antioxidant in chemical terms: it has been reported to both generate and scavenge H₂O₂ [40], a reactive species that can promote oxidation and compromise apparent recoveries. Taken together, these observations suggest that optimization of both the antioxidant type and its dosage may be warranted, and should be addressed through targeted methodological studies in the future. However, a further complication lies in the matrix itself. Plant tissues contain diverse endogenous antioxidants, and their interactions may be synergistic or antagonistic, ultimately modulating oxidative stability during saponification and extraction. Consequently, this is a multifaceted issue that cannot be resolved without a dedicated, matrix-aware optimization strategy.

Lower extractability of tocopherols than tocotrienols can be a result of not only a higher proportion of bound tocopherols, but also their physicochemical properties, especially α-tocopherol, resulting in lower extractability [35]. Due to the low tocopherol content, challenges in identifying bound tocochromanols, and measurement errors, the current study did not attempt to characterize these structures. This knowledge gap about bound tocochromanols and their extractability is largely attributable to the fact that comparative extraction studies using both direct and saponification-based protocols are rarely performed in screening studies. Additionally, detecting bound tocochromanols is inherently demanding: it is time-consuming, hindered by the limited availability of appropriate analytical standards, advanced equipment (GC-MS), and further constrained by the generally low concentrations of these compounds in plant tissues [37].

The two analytical methods exhibited similarly good repeatability. Generally, the lowest repeatability was observed for tocochromanols present at the lowest concentrations. Since free tocotrienols demonstrated higher concentrations across the Berberidaceae family, their repeatability was on average three times greater (coefficient of variation) than that of free tocopherols (Supplementary Materials). The recovery and repeatability of the UAEE protocol is suitable for application in comparative research of Berberidaceae family samples with 88–96%, an average of 93%, recovery of total tocochromanols compared to a saponification protocol, demonstrating its suitability for daily sample screening.

3.2. Free Tocochromanol Profile

Analyzed genera, species, and their free tocochromanol contents are presented in

Table 1. Free tocochromanol content in analyzed Berberidaceae family species, mg 100 g⁻¹ dw.

Genus	Species	γ-T3	α-T3	γ-T	α-T	Sum	Ratio γ/α-T3	Ratio T3s/Ts
Berberis	aemulans (n = 2)	3.56 ± 0.48	3.48 ± 0.49	0.29 ± 0.18	2.54 ± 0.81	9.88 ± 1.96	1.0	2.5
	aetnensis (n = 4)	3.50 ± 1.37	6.79 ± 1.87	ND	3.07 ± 2.42	13.36 ± 1.28	0.5	3.4
	aggregata (n = 3)	7.50 ± 1.25	2.20 ± 1.02	ND	4.17 ± 0.9	13.87 ± 2.99	3.4	2.3
	amurensis (n = 3)	3.68 ± 1.27	2.01 ± 0.94	0.34 ± 0.14	2.92 ± 1.3	8.96 ± 3.32	1.8	1.7
	angulosa (n = 2)	0.81 ± 0.13	5.18 ± 0.79	ND	2.43 ± 1.29	8.43 ± 0.37	0.2	2.5
	anhweiensis (n = 2)	1.33 ± 0.30	2.72 ± 0.96	ND	ND	4.05 ± 0.66	0.5	NA
	aquifolium (n = 8)	7.04 ± 2.00	1.42 ± 0.66	0.42 ± 0.27	1.28 ± 0.58	10.17 ± 3.24	5.0	5.0
	aristata (n = 2)	3.67 ± 1.37	2.77 ± 1.01	ND	4.64 ± 1.12	11.09 ± 3.50	1.3	1.4
	bealei (n = 2)	0.70 ± 0.13	1.33 ± 0.23	ND	0.28 ± 0.05	2.31 ± 0.42	0.5	7.3
	brachypoda (n = 3)	0.88 ± 0.36	6.47 ± 1.08	ND	1.63 ± 0.89	8.98 ± 1.69	0.1	4.5
	bretschneideri (n = 2)	2.95 ± 0.81	1.75 ± 0.30	ND	ND	4.71 ± 0.52	1.7	NA
	canadensis (n = 2)	1.04 ± 0.14	5.90 ± 0.38	ND	3.21 ± 0.60	10.16 ± 0.08	0.2	2.2
	chinensis (n = 2)	2.70 ± 0.11	3.11 ± 0.11	0.41 ± 0.13	3.48 ± 0.44	9.69 ± 0.79	0.9	1.5
	diaphana (n = 2)	4.22 ± 0.66	0.70 ± 0.12	0.32 ± 0.11	3.02 ± 0.82	8.26 ± 0.16	6.0	1.5
	dielsiana (n = 2)	1.74 ± 0.56	6.11 ± 0.35	ND	ND	7.85 ± 0.91	0.3	NA
	gagnepainii (n = 2)	2.96 ± 0.25	2.68 ± 1.08	ND	1.51 ± 1.24	7.16 ± 0.09	1.1	3.7
	giraldii (n = 2)	4.81 ± 0.63	1.87 ± 0.47	ND	1.34 ± 0.63	8.03 ± 0.47	2.6	5.0
	heteropoda (n = 3)	1.71 ± 1.23	8.95 ± 1.79	ND	1.66 ± 1.32	12.33 ± 1.70	0.2	6.4
	integerrima (n = 2)	5.88 ± 0.67	1.10 ± 0.33	0.85 ± 0.15	1.48 ± 0.19	9.31 ± 0.01	5.3	3.0
	johannis (n = 2)	10.20 ± 1.58	2.92 ± 1.50	ND	3.39 ± 1.02	16.52 ± 0.94	3.5	3.9
	julianae (n = 3)	3.80 ± 1.03	1.60 ± 1.01	ND	0.86 ± 0.37	6.26 ± 2.06	2.4	6.3
	karkaralensis (n = 2)	4.08 ± 0.79	5.21 ± 0.83	ND	1.15 ± 0.81	10.44 ± 0.86	0.8	8.1
	koreana (n = 5)	4.89 ± 1.74	2.66 ± 0.67	0.43 ± 0.28	3.47 ± 1.17	11.44 ± 2.57	1.8	1.9
	nervosa (n = 2)	1.45 ± 0.58	0.78 ± 0.21	ND	0.86 ± 0.41	3.09 ± 0.78	1.9	2.6
	orientalis (n = 2)	0.69 ± 0.37	5.66 ± 0.88	ND	1.58 ± 0.37	7.94 ± 0.88	0.1	4.0
	ottawensis (n = 3)	1.09 ± 0.38	5.25 ± 1.07	ND	1.62 ± 0.71	7.96 ± 0.82	0.2	3.9
	regeliana (n = 2)	1.19 ± 0.61	4.18 ± 0.78	0.74 ± 0.13	3.63 ± 1.16	9.74 ± 2.41	0.3	1.2
	repens (n = 3)	5.70 ± 2.60	1.68 ± 0.31	ND	0.22 ± 0.09	7.59 ± 2.57	3.4	33.5
	sanguinea (n = 2)	4.33 ± 0.62	1.59 ± 0.51	ND	0.63 ± 0.13	6.55 ± 0.02	2.7	9.4
	sargentiana (n = 2)	1.59 ± 0.25	6.48 ± 2.27	ND	1.98 ± 0.16	10.05 ± 2.36	0.2	4.1
	sieboldii (n = 3)	4.13 ± 0.86	5.35 ± 0.79	0.22 ± 0.07	5.48 ± 0.50	15.17 ± 2.11	0.8	1.7
	thibetica (n = 4)	1.62 ± 0.47	5.24 ± 1.13	ND	2.62 ± 0.89	9.47 ± 1.98	0.3	2.6
	thunbergii (n = 3)	1.28 ± 0.25	4.99 ± 0.56	0.35 ± 0.33	4.62 ± 0.51	11.25 ± 1.18	0.3	1.3
	tischleri (n = 2)	3.20 ± 0.49	5.44 ± 1.69	ND	4.27 ± 1.39	12.90 ± 0.18	0.6	2.0
	tschonoskyana (n = 2)	4.22 ± 1.60	9.14 ± 2.58	0.13 ± 0.05	4.89 ± 0.26	18.39 ± 4.40	0.5	2.7
	vernae (n = 3)	2.24 ± 0.75	5.50 ± 0.81	ND	3.57 ± 0.66	11.32 ± 0.32	0.4	2.2
	vulgaris (n = 7)	1.26 ± 0.55	4.97 ± 1.08	ND	1.06 ± 0.73	7.29 ± 1.42	0.3	5.9
	wilsoniae (n = 3)	11.32 ± 3.29	3.31 ± 0.95	ND	5.01 ± 0.96	19.64 ± 3.46	3.4	2.9
Mahonia	× wagneri (n = 2)	4.88 ± 1.02	1.40 ± 0.65	ND	0.43 ± 0.15	6.73 ± 0.21	3.5	14.6
Caulophyllum	robustum (n = 2)	0.23 ± 0.11	1.26 ± 0.37	ND	0.25 ± 0.12	1.75 ± 0.36	0.2	6.0
	thalictroides (n = 3)	4.35 ± 2.59	6.55 ± 3.78	1.40 ± 1.42	1.01 ± 1.28	13.31 ± 8.92	0.7	4.5
Jeffersonia	diphylla (n = 2)	18.00 ± 1.74	0.23 ± 0.04	ND	3.91 ± 3.63	23.76 ± 5.20	78.3	4.7
Podophyllum	hexandrum (n = 9)	3.88 ± 1.35	6.21 ± 1.76	ND	0.71 ± 0.38	11.37 ± 2.96	0.6	14.2
	peltatum (n = 3)	5.01 ± 1.05	7.48 ± 0.40	ND	0.62 ± 0.11	13.69 ± 1.11	0.7	20.1

Data are presented as means ± standard deviation. The number of analyzed samples is provided after the species (n = number of samples in species). δ-Tocotrienol was detected only in Jeffersonia diphylla, P. hexandrum and P. peltatum (1.53–1.71; 0.15–0.73; and 0.32–0.53 mg 100 g⁻¹ dw, respectively). β-Tocotrienol was detected only in P. hexandrum and P. peltatum (0.15–0.52 and 0.12–0.20 mg 100 g⁻¹ dw, respectively). T, tocopherol; T3, tocotrienol; dw, dry weight calculated based on estimation of 5% moisture in all freeze-dried seeds (see Section 2.2); ND, not detected; NA, not analyzed.

. In total, 44 species from four genera (Berberis, Caulophyllum, Jeffersonia, and Podophyllum) were analyzed for free tocochromanol (hereafter referred to as tocochromanol) content, including one unplaced species, Mahonia × wagneri, which would be categorized under the Berberis genus by some sources. Berberis was the most widely represented genus in the sample set (38 species and one Mahonia hybrid). Two species were tested from Caulophyllum and Podophyllum each, and one biological replicate could be sourced for Jeffersonia diphylla, the only accepted member of its genus. However, relative to the total number of species in the genera, Berberis is the least represented, with 39 (including Mahonia × wagneri/Berberis aquifolium) out of around 615 accepted species (6.34%), whereas the representation of the other genera is more complete: the sole member of the Jeffersonia genus, 2 out of 17 Podophyllum species, and two out of three Caulophyllum species.

The primary free tocochromanols present in the Berberidaceae family were α-tocotrienol, γ-tocotrienol, and α-tocopherol (0.23–9.14, 0.23–18.00, and 0–5.48 mg 100 g⁻¹ dw, respectively), with tocotrienols predominating over tocopherols. γ-Tocopherol was detected in approximately 25% of the species studied in the concentration 0–1.40 mg 100 g⁻¹ dw, while other tocotrienols and tocopherols were observed even less frequently or were entirely absent. None of the analyzed Berberis, Caulophyllum, or Mahonia samples contained δ-tocotrienol or β-tocotrienol; they were only detected in Jeffersonia diphylla, which contained 1.62 ± 0.13 mg 100 g⁻¹ dw δ-tocotrienol, while Podophyllum hexandrum contained 0.34 ± 0.26 and 0.23 ± 0.17 mg 100 g⁻¹ dw δ-tocotrienol and β-tocotrienol, respectively, and P. peltatum contained 0.42 ± 0.1 and 0.15 ± 0.04 mg 100 g⁻¹ dw δ-tocotrienol and β-tocotrienol, respectively.

There was a statistically significant difference between barberry family species (p < 0.001) and genera (p < 0.001). As presented in Figure 2, there is some distinction between tocochromanol contents and profiles in different genera. While γ-tocotrienol and α-tocotrienol contents are similar in Berberis, Caulophyllum and Podophyllum, α-tocopherol content tended to be much lower in Caulophyllum and Podophyllum. Meanwhile, the obtained Jeffersonia samples contained very little α-tocotrienol, and had much higher γ-tocotrienol content. However, conclusions about the true upper and lower limits of individual tocochromanols in the Berberidaceae family cannot be drawn due to the limited sample sets for Caulophyllum, Jeffersonia and Podopyllum.

Variation within genera and species can be caused by a variety of factors, including genetic and physiological differences (natural variability), fruit ripeness, climatic conditions and plant stress. There is limited research on the phytochemical composition of seeds in the Berberidaceae family, as research has primarily focused on the edible berries of two species—B. microphylla [29,30] and B. integerrima [31,32]—of which only the latter was analyzed in the present study. In all four available reports, tocochromanols have been identified in seed oil. In B. microphylla, α-tocotrienol and γ-tocotrienol had the highest tocochromanol proportions [32], which is consistent with the tocochromanol profile of the Berberis genus (Table 1).

However, another report likely misidentified compounds, as suggested by inconsistencies in the retention times across chromatograms for the same tocochromanol homologues [30]. The two available reports on B. integerrima only used tocopherol standards [31,32], and so only report the presence of tocopherols, whereas the use of tocotrienol standards in the present study has confirmed both their presence in the seed as well as preferential accumulation.

The present work clearly demonstrates a predominance of free tocotrienols in the seeds of the Berberidaceae family, while their comparison with the literature highlights the necessity of including tocotrienol standards to achieve a comprehensive characterization of plant material. The frequent omission of tocotrienol standards in combination with tocopherol standards is likely one of the main reasons for the underestimation of tocotrienols in plant-based matrices [33]. In all 44 Berberidaceae species examined, tocotrienols predominated over tocopherols, with wide tocotrienol-to-tocopherol ratio range across genotypes (1.2–33.5). The balance between the two main tocotrienols was more nuanced: α-tocotrienol exceeded γ-tocotrienol in the majority of genotypes (n = 25), whereas γ-tocotrienol dominance was observed in fewer cases (n = 18). Only a single genotype exhibited an α-tocotrienol/γ-tocotrienol ratio of one (Table 1). The free tocochromanol profile and concentrations observed in Berberidaceae seeds closely resemble those reported in grape seeds (Vitis spp., Vitaceae family). In grape seeds, γ-tocotrienol is typically the dominant homolog, followed by α-tocotrienol [35,41]. In contrast, seeds from the Berberidaceae family show genotype-dependent alternation in free tocotrienol dominance. While grape seeds have found industrial applications due to the large volume of by-products from juice and wine production, such usage is unlikely for Berberidaceae seeds, which are primarily derived from ornamental rather than commercially cultivated berries. However, the above reports about B. microphylla and B. integerrima highlight some of their potential applications in seed oil production, albeit in lower quantities than grapeseed oil.

3.3. Free Tocochromanol Composition as Shaped by Phylogeny

The Kruskal–Wallis test identified no statistically significant differences between the main tocochromanol (γ-tocotrienol, γ-tocopherol and α-tocopherol) contents, except for slightly higher γ-tocopherol content in Berberis and Caulophyllum. Of minor tocochromanols, Podophyllum had slightly higher β-tocotrienol and δ-tocotrienol contents, and Jeffersonia had distinctly elevated δ-tocotrienol and γ-tocotrienol contents (Figure 3), while α-tocochromanol contents were low. This may be explained by metabolic differences in the species, like lower γ-tocopheryl methyltransferase activity in the genus. The enzyme adds a methyl group to γ- and δ-tocochromanols, producing α- and β-tocochromanols, respectively. If tocochromanol proportion is used for analysis, similar trends are evident.

To determine the main differentiating tocochromanols, PCA was performed on the dataset. PCA revealed δ-tocotrienol, β-tocotrienol and γ-tocotrienol as the main differentiating molecules (Figure 4A). The Berberis genus was split into α-tocotrienol and γ-tocotrienol-associated branches. High γ-tocotrienol content appears to be the main delineator between Jeffersonia diphylla and γ-tocotrienol-rich Berberis species (Figure 4B). According to the PCA results, the Caulophyllum and Podophyllum genera are similar to α-tocotrienol-rich Berberis species. Eigen values and explained variance are provided in Table S4 (Supplementary Materials), and variable loadings are provided in Table S5 (Supplementary Materials).

PC1 accounted for 58.32% of the variation, while PC2 explained 27.87% of the variation, for a total of 86.19% explained variance. PC1 had negative loadings with β-tocotrienol (−9.14), α-tocotrienol (−3.88) and β-tocopherol (−1.81), and positive loadings with γ-tocotrienol (9.11), δ-tocopherol (8.00), δ-tocotrienol (3.21), α-tocopherol (1.39) and γ-tocopherol (1.34). PC2 had negative loadings with α-tocotrienol (−8.96), β-tocotrienol (−7.23), β-tocopherol (−4.45), γ-tocotrienol (−3.37), δ-tocotrienol (−2.32), α-tocopherol (−2.87), and γ-tocopherol (−1.53), and only had a positive loading with δ-tocopherol (7.38).

Using the silhouette method, two was determined to be the optimal number of clusters for k-means cluster analysis. Representatives within cluster 1 and 2 by genus and species are presented in Figure 5, and cluster mean tocochromanol contents are provided in Table S6 (Supplementary Materials).

Cluster 2 had higher γ-tocotrienol content and lower α-tocotrienol content, while other free tocochromanol contents were similar. Species in Berberis were separated between the two clusters, as was Mahonia × wagneri, while all Caulophyllum and Podophyllum species were categorized under cluster 1. Jeffersonia diphylla was categorized under cluster 2. Mahonia × wagneri is considered a synonym of Berberis aquifolium by some sources, and was categorized under the same cluster.

Certain tocochromanols (β-tocotrienol and δ-tocotrienol) were only present in a few species. While they had no significant impact on PCA variance or loadings, they significantly affected cluster assignment—44% of samples were reassigned from cluster 1, and 52% were reassigned from cluster 2 upon their removal from analysis.

PCA based on tocochromanol proportion (% of the sum of free tocochromanol content) results in PC1 and PC2, explaining 76% and 22% of variance, for a total of 99% explained variance. PC1 had high loadings with γ-tocotrienol (0.70) and α-tocotrienol (−0.71), and PC2 with γ-tocotrienol (−0.41), α-tocotrienol (−0.41), and α-tocopherol (0.81), as provided in Table S5 (Supplementary Materials). Scores for individual points are depicted in Figure 6. Eigen values and explained variance are provided in Table S4 (Supplementary Materials), and variable loadings are provided in Table S5 (Supplementary Materials). Using tocochromanol proportion for PCA resulted in higher α-tocotrienol and α-tocopherol contribution to scores and less separation between genera on the plot.

K-means cluster analysis based on tocochromanol proportion assigns the samples to two clusters of sizes 78 and 48; cluster mean tocochromanol proportions are provided in Table S6 (Supplementary Materials). The clusters are differentiated by higher α-tocotrienol proportion in cluster 1 and higher γ-tocotrienol proportion in cluster 2, while other tocochromanol proportions are similar between the clusters (Figure 7). Mean tocochromanol contents within cluster are provided in Table S6 (Supplementary Materials). Sensitivity to minor tocochromanol presence is significantly lower—82 and 98% of samples remain in assigned cluster after exclusions of minor tocochromanols. Assignment between clusters is similar to assignment based on free tocochromanol contents (82% and 100% of samples remained in the same cluster).

The apparent preference for γ-tocotrienol or α-tocotrienol accumulation is not unique to the Berberis genus. While legume seeds predominantly contain tocopherols, especially α-tocopherol and γ-tocopherol, their proportion can differ even between species within the same genus [42,43], as it is in the leaves of the Hypericum genus, which have a significant tocotrienol contents in most above-ground plant parts [20]. While grapevine seeds have relatively consistent α-tocotrienol content, γ-tocotrienol content is highly variable [44]. Free tocochromanol contents had high proportions between high and low values in Caulophyllum thalictroides and Podophyllum hexandrum seeds (max/min proportion was up to 11.1 for α-tocopherol and up to 8.3 for γ-tocopherol, up to 4 for α-tocotrienol, and up to 5.6 for γ-tocotrienol), but total free tocochromanol content max/min proportions were not as extreme—up to 3.5 (Table S8 in Supplementary Materials).

Disagreement between genetic and tocochromanol content similarity is to be expected, as tocochromanol and metabolically related compound biosynthesis are regulated by only a fraction of the whole genome. However, metabolic differences can be observed between branches of plant families. For example, certain branches of the Poaceae family exclusively use the C3 photosynthetic pathway, others use the C4 pathway, and other are mixed, with species using either C3 or C4 [45]. The C4 photosynthetic pathway likely evolved as an adaptation to warm and arid climates; C4 plants include a variety of monocotyledonous and eudicot plants, and the pathway causes chemical differences between plants [46]. Though there is no known link between tocochromanol composition and the photosynthetic pathway, it is a good example of metabolic differences between related plants, and tocochromanols are related to heat, salinity and drought stress response in plants [47], and their content may be affected by the native climate of the plant more than taxonomic classification [42]. There are a few similarities between phylogenetic branching and tocochromanol profiles in the present study. Jeffersonia has a distinctly different tocochromanol content and composition from the rest of the family, according to PCA and cluster analysis. It is estimated to have an earlier divergence time than the rest of the family. Jeffersonia is placed closest to Podophyllum on phylograms [48]; both contained relatively minor δ-tocotrienol (Jeffersonia and Podophyllum) and β-tocotrienol (Podophyllum) contents. Meanwhile, Caulophyllum, the next closest genus, did not contain either minor tocochromanol, and Podophyllum and Caulophyllum were not assigned to the same cluster as Jeffersonia in the present study. Divergence between Berberis species’ tocochromanol contents is difficult to compare to low overlap between species in the present study and existing phylogenetic studies on the family [48,49].

As mentioned in the Introduction, differences and similarities between different taxa have been observed in other plant families as well. In the Rutaceae family, tocotrienols predominated in only certain branches (Cneoroideae and most of Zanthoxyloideae subfamily among the investigated), and were completely or virtually absent in others (Ruta genus and Aurantioideae subfamily among the investigated) [23]. Most of the investigated Celastraceae species’ seeds preferentially accumulated γ-tocotrienol, with Euonymus species accumulating similar contents of α-tocotrienol and highly variable γ-tocotrienol contents, and only Celastrus species accumulating notable δ-tocotrienol and δ-tocopherol contents [26]. In the Vitaceae family, certain genera (Ampelocissus, Ampelopsis, and Parthenocissus) accumulate mostly α-tocotrienol, with a few exceptions (P. henryana) [25], while the Vitis species and interspecific crosses accumulate γ-tocotrienol and α-tocotrienol in a relatively similar proportion, and may have significant α-tocopherol content, depending on the variety [44]. In Ilex, the only remaining genus in the Aquifoliaceae family, preference for tocotrienol accumulation is consistent among across the 29 investigated species [28], while in the Cornus genus, species’ seed tocochromanol contents and proportions were only similar in the Kraniopsis subgenus [27]. A family wide tocochromanol screening of the Arecaceae (palm) family (84 species) observed consistent tocotrienol domination and highly variably tocochromanol composition [50]. Like the present study, it observed a family wide trend, but little relation to the phylogenetic structure of the family. Smaller screenings of the Apiaceae family have also observed tocotrienol-dominated seed tocochromanol content profiles (67–99% of free tocochromanols) [13]. Several previous studies have explored the chemotaxonomic relevance of tocochromanol and fatty acid profiles in families including Brassicaceae [51], Boraginaceae [52], and Apiaceae [14] as well. In each of these studies, however, none of them addressed the distinction between free and esterified tocochromanols, even though only free tocochromanols were quantified. Clear declarations regarding the analysis of free tocochromanols remain uncommon in the chemotaxonomic literature [25].

A number of factors can influence tocochromanol composition in the seeds, including natural variability [53,54] and fruit ripeness [55,56]. While the dominance of γ-tocotrienol and α-tocotrienol in Berberis species is apparent, this study cannot provide affecting factors for their proportion. Further studies are needed to understand the prevalence of individual tocochromanols in other plant parts as well as genera other than Berberis, since only a limited number of species could be collected for the present study.

Here, we explicitly acknowledge and discuss this and other methodological issues relevant to this and other investigations.

3.4. Uncertainties and Systematic Bias in the Measurements

3.4.1. Accumulation of Tocochromanols During Seed Development and Seed Maturity

Tocochromanol content increases during seed development, but content tends to stabilize at approximately the mid-point from flowering to full ripeness, and the qualitative profile or the absolute concentration of tocochromanols changes very little in the final 30–40 days of maturation in Japanese quince (Chaenomeles japonica) and grape (Vitis vinifera) seeds [56,57]. Grape seeds show a particularly illustrative dynamic: tocopherols prevail during early development, whereas tocotrienols accumulate following a logarithmic trend after mid-development [56]. In botanical garden practice, seeds are harvested at full maturity to preserve viability and germination capacity for the following season. Given typical tocochromanol accumulation patterns, small differences of a few days or even weeks near maturity are unlikely to meaningfully affect seed tocochromanol content; we consider seed maturity to be a minor factor and thus reasonably negligible in the context of the present study.

3.4.2. Abiotic Factors

Disentangling the effects of abiotic drivers—temperature, soil water status, irradiance, soil type and composition, and related factors—on seed tocochromanol accumulation is inherently challenging and rarely yields unequivocal outcomes. For this reason, controlled greenhouse experiments are frequently used; yet even well-designed greenhouse trials cannot fully reproduce the complexity of open field conditions. In soybean, experiments comparing three species across two temperature regimes and contrasting water availability showed that both drought and higher temperature increased α-tocopherol, while concomitantly decreasing γ- and δ-tocopherol, with only modest changes in total tocochromanol content [58]. Moreover, the effect of high-temperature stress on tocopherol levels in soybean differs depending on the developmental stage at which the stress occurs [59]. These results imply a stress-driven rebalancing of the tocochromanol pool, favoring α-tocopherol at the expense of other homologues—consistent with its prominent role in protecting the embryo against reactive oxygen species [60]. It is also plausible that the environmental determinants of seed tocochromanol content differ between crops cultivated primarily for seeds versus those cultivated for fruit. The effect of genotype and growing conditions can have similar effects on the tocochromanol composition and content, as they have done in carrot roots [53], but the impact of variety has been more significant than growing conditions in rapeseed, lupin, corn, oat, and durum wheat tocochromanols [61,62,63,64,65]. Geographic origin can have a stronger effect on tocochromanol content and composition than oil palm population [54]. Under controlled conditions, temperature and soil moisture may have a stronger effect on tocochromanol composition or total content, depending on the cultivar; heat and drought stress generally result in increased α-tocopherol content [58], but exposure to heat stress affects seed tocochromanol content most significantly in the late productive stage in soy [59], an annual plant. There are very few reports on perennial plant tocochromanol content changes in response to stress. A decline in α-tocopherol production has been observed in olive fruits over successive years, and the authors hypothesized blockage of the metabolic pathway in response to stress [55], but the study did not consider plant maturity as a factor. Chicory plants appear to upregulate α-tocopherol and γ-tocopherol production in response to drought stress [66], and different soil types have a moderate effect on tocochromanol accumulation in aerial parts of Hypericum plants [67].

Accordingly, variation among seed accessions of the same species sourced from different botanical gardens may reflect site-specific differences in soil type, precipitation patterns, temperature, and light environment. The systematic bias associated with abiotic factors is difficult to estimate due to the complexity of multiple interacting variables. Nevertheless, it is generally accepted that genetic determinants exert a greater influence than abiotic factors [61,62,68]. The biome (dry/wet and temperate/tropical) of the natural aerial of the species may have a greater impact on lipid content than plant subfamily, as has been observed in Fabaceae [42], but can be considered a separate issue, since it involves species’ individual genetic predisposition and adaptation to that environment.

3.4.3. Biological Replications

Evidence from seed matrices in which tocopherols and tocotrienols co-occur—such as cranberry (Vaccinium macrocarpon) [34] and grape (Vitis spp.) [35] seeds—suggests typical variability on the order of ~3–15%. Accordingly, we consider this range to be a realistic approximation of the uncertainty inherent in the present dataset.

3.4.4. Genetic Background (Genotypes/Variety/Cultivar)

Studies on soybean show that tocopherol profiles and concentrations are influenced by genotype, environmental conditions, and their interactions. Of these, genetic factors account for the greatest share of variability [61,62,68], although genetic variance can still be high even if secondary lipid production is primarily determined by variety [63]. Drawing on previous analyses of seed matrices containing tocopherols and tocotrienols—cranberry (V. macrocarpon, eight cultivars) [34] and interspecific grape crosses (V. vinifera × V. amurensis × V. riparia, 11 genotypes) [35]—the difference between genotypes with the lowest and highest content may reach 250%. Published data suggest that, in some crops, these differences can be even more pronounced. Additionally, there is significant imbalance between the number of species analyzed within the included genera in the sample pool. The coefficient of variation differed between different species and tocochromanols, tocopherol and minor tocotrienol content being more variable on average. All individual tocochromanol coefficients of variation were much higher than the sum of free tocochromanol content, while the proportion between highest and lowest values was similar between individual and the sum of free tocochromanol contents (Tables S7 and S8 in Supplementary Materials).

3.4.5. Free and Bound Tocochromanols

Bound tocochromanols may be present as esters, glycosides, or through physical association with the plant matrix. Yet, quantitative research on these forms is very limited [37], and available data do not indicate that seeds contain substantial quantities of bound tocochromanols. While a higher proportion of tocochromanol esters cannot be completely excluded—much like the natural occurrence of β-tocotrienol—the current state of knowledge suggests that the probability of bound tocochromanols exceeding 10% of the total is extremely low (<0.1%). Therefore, the potential systematic bias attributable to bound forms in this plant material can be conservatively estimated at 0–10%. Beyond the proportion of free and derivatized tocochromanols in plants, differences between their functionality are not well-understood. Current research reports poor digestibility in artificial digestion juices [37], and much of the research focuses on tocopheryl acetate, a food additive, while in nature tocochromanol esters are mostly formed with fatty acids [69,70,71,72,73]. Moreover, while saponification is necessary to release these bound forms and determine total tocochromanol content in a matrix, many, if not most, screening studies omit the step, and extract in hexane directly, but express results as total, not free tocochromanols [14,51,52].

3.4.6. Seed Moisture

Because plant material was limited, seed moisture could not be measured for every sample. Therefore, we normalized all seed data using a fixed moisture content of 5%, derived from the observed range across the tested samples (3–7%). We acknowledge that this assumption represents a strict simplification and may introduce a systematic bias on the order of 0–2%.

3.4.7. Seed Storage

Evidence from soybean storage experiments conducted for 12 months across different temperature and humidity regimes points to a generally high stability of tocopherols. Under moderate conditions (25 °C; 12% moisture), tocopherol losses were limited to 6–7% after 12 months [74]. Similarly, winter oilseed rape (Brassica napus) seeds—obtained during germplasm screening for genetic variation—can be stored under suitable conditions for at least 6 months without detectable tocochromanol degradation [75], and it can take several years for noticeable tocochromanol degradation to occur in a variety of species, with α-tocopherol and γ-tocopherol being the most sensitive to oxidation [76].

However, longer storage may result in substantial declines. Berberis anhweiensis and Berberis tischleri seeds stored for approximately 27 months in paper bags at room temperature (21 ± 3 °C) showed 10–40% lower tocotrienol contents when compared to seeds collected several years earlier from the same site (botanical garden). On average, losses were ~30% for tocotrienols and ~40% for α-tocopherol. Integrating these observations with published evidence, we conservatively estimate the systematic bias associated with seed storage (time from seed collection to analysis) to be 5–25%.

3.4.8. Other Unknown Systematic Bias, e.g., Cross-Pollination

Undoubtedly, additional variables not captured in our study may contribute to variation in seed tocochromanol composition and abundance, with cross-pollination being one plausible example. In the absence of published evidence on this specific aspect, we cannot reliably quantify its potential contribution, nor that of other unmeasured factors that may modulate seed tocochromanol profiles.

At the same time, species in the investigated genera appear to preferentially accumulate either α- or γ-tocotrienol; a similar free tocochromanol composition can be hypothesized for other genera in the family. Taxonomic assignment and tocochromanol biosynthesis are determined by and regulated by species’ genetic make-up, and, while there were some differences and similarities in tocochromanol contents attributable to taxonomy, the present study examined only one group of metabolically related compounds, and did not control for environmental factors, accession, or post-harvest factor effects. The methyl erythritol phosphate and shikimate pathways are involved in the biosynthesis of a large variety of compounds, including carotenoids, chlorophyll (the former), and aromatic amino acids (the latter), and the regulation of biosynthetic processes differs between plant organs [77]. The present study did not examine the fruit flesh, leaves, or other organs besides the seeds. Future research is recommended on tocochromanol biosynthesis and metabolic regulation, as well as other metabolically related compound groups. To determine direct links between phylogeny and tocochromanol or other compound contents, controlled, multi-site research with known accessions is required. Moreover, it may be recommended to analyze tocochromanol content when investigating Berberidaceae extracts with antioxidant functions.

4. Conclusions

UAEE had good repeatability and slightly lower recovery than the saponification protocol. UAEE can be considered a reliable free tocochromanol extraction method for Berberidaceae seeds. It can be recommended for routine monitoring, but should be validated for any new plant material, since it only allows the determination of free tocochromanol content, and information on the proportion of esterified and otherwise bound tocochromanols is very limited at this time.

Surveyed Berberidaceae species’ seed free tocochromanols were consistently tocotrienol-dominated and contained α-tocotrienol and γ-tocotrienol in highest proportion. Species in the Berberis genus may preferentially accumulate either, whereas the analyzed Mahonia, Caulophyllum and Podophyllum species mostly contained α-tocotrienol. Only Jeffersonia and Podophyllum samples contained small amounts of δ-tocotrienol and β-tocotrienol, which were differentiated from each other by their main tocochromanols (α-tocotrienol or γ-tocotrienol). These findings largely support the proposed chemotaxonomy-based hypothesis, but additional screening is necessary. Since free tocotrienols were the predominant tocochromanols in the seeds of all 44 Berberidaceae species (128 samples) examined here, it remains highly possible that other species within this 769-member family may also have tocotrienol-rich profiles. However, we cannot exclude the possibility that certain species—or even specific clades within the family—may instead be tocopherol-dominant, as we have previously observed in other tocotrienol-dominant families such as Rutaceae [23], Cornaceae [27], or Celastraceae [26]. Such deviations could plausibly reflect evolutionary legacies/phylogeny and/or environmental modulation, including stress-related drivers such as elevated temperature, drought, or other abiotic constraints. Conclusions on tocochromanol content, proportion and biosynthetic regulation in the family are limited by the number of tested species and genera, and analyzed compounds. However, the data show that tocotrienol absence should not be assumed in plant matrices.

It should be acknowledged that this study has some inherent limitations, including uncontrolled growing conditions (biotic and abiotic factors) and variable time between post-harvest and seed analysis, both of which may introduce systematic bias. The potential impact of these and other confounding factors on tocochromanol profiles is addressed in detail within the manuscript.