Polyphenolic and Biogenic Volatile Organic Compound Profiles of Endemic Metallophytes From the Thallium‐Rich Allchar Mining Area

ABSTRACT

The Allchar Mining Area (North Macedonia) is an extreme metalliferous environment enriched in Tl, As and Sb that supports diverse endemic metallophytes. This study investigates metabolic profiles in selected species through integrated analysis of polyphenols (HPLC/DAD/MSⁿ) and biogenic volatile organic compounds (BVOCs, headspace GC–MS). A total of 78 polyphenols and 123 volatile compounds were identified. Knautia caroli‐rechingeri and Centaurea leucomalla were dominated by caffeoylquinic acid derivatives, in which reached 405.40 mg/g dry herb in K. caroli‐rechingeri. Thymus thracicus var. alsarensis showed a distinct profile rich in rosmarinic and salvianolic acids, whereas Tl‐hyperaccumulating Viola species were characterized by flavone and flavonol glycosides together with a markedly reduced diversity of phenolic acids. BVOC profiles also differed markedly among species, including terpene‐rich profiles in Thymus, aldehyde‐dominated compositions in Knautia and Centaurea, and non‐terpenoid‐enriched profiles in Viola species. Multivariate analysis confirmed species‐specific metabolic patterns and significant concordance between polyphenolic and BVOC profiles (Mantel ρ = 0.72 and p = 0.008), while no significant association with reported Tl accumulation was observed. Metabolomic and metal accumulation patterns highlight species‐specific biochemical characteristics associated with life in metalliferous habitats.

Six plant taxa endemic to the thallium‐rich Allchar Mining Area (North Macedonia) were profiled by HPLC/DAD/MSn and headspace GC–MS. Seventy‐eight polyphenols and 123 biogenic volatile organic compounds were identified, and multivariate analysis showed congruent, species‐specific polyphenolic and volatile signatures with no significant association with previously reported thallium accumulation.

1. Introduction

The Allchar Mining Area, located on the slopes of Kozuf Mountain, represents one of the most geochemically distinctive environments worldwide. This ancient mining site has been exploited for centuries and is globally recognized as the only known deposit of the thallium‐bearing mineral lorandite, accompanied by elevated concentrations of arsenic, antimony, gold, nickel, lead and cadmium [1]. Owing to these unique geochemical characteristics, the locality has attracted considerable scientific interest, including its proposed relevance in solar neutrino detection and environmentally clean energy research.

Despite these extreme geochemical conditions, a notable diversity of plant species has been recorded in this area, including a significant number of endemic taxa. According to Micevski and Matevski [2], the site is of particular importance due to the presence of locally restricted endemic species such as Odontarrhena kavadarcensis, Centaurea kavadariensis, Centaurea leucomalla, Knautia caroli‐rechingeri, Viola allchariensis, Viola arsenica, and Viola × halacsyana. These taxa represent an important model system for studying plant survival and adaptation under extreme metalliferous conditions.

Soil contamination with heavy metals and other toxic elements is widely recognized as a major environmental issue. Elevated concentrations of potentially toxic elements such as Cd, Co, Cr, Hg, Mn, Pb, and Zn have been reported to exert pronounced phytotoxic effects, resulting in inhibition of key physiological and biochemical processes, reduced biomass, and, in severe cases, plant death [3]. Nevertheless, such general responses contrast with the persistence of metallophyte species observed in extreme environments such as the Allchar.

Certain plant species have evolved specialized mechanisms for metal and metalloid tolerance, including sequestration in metabolically active compartments or incorporation into stable chelate complexes [4, 5]. Plants adapted to metal‐rich substrates while maintaining normal growth are defined as metallophytes [6, 7]. These species are of particular importance for understanding ecological adaptation and evolutionary processes, as well as for applications in phytoremediation [8].

In metalliferous habitats, abiotic stress significantly alters plant chemical composition, often through modulation of secondary metabolism. A central mechanism involves oxidative stress induced by reactive oxygen species (ROS), both radical and non‐radical (O₂•⁻, •OH, H₂O₂, ¹O₂) that may cause membrane damage, reduced photosynthesis, and disruption of genetic processes [9]. Metals such as Cu, Pb, Cr, and As can directly induce ROS formation, whereas Cd acts mainly through enzyme inhibition [5, 10].

To mitigate oxidative stress, plants rely on antioxidant systems composed of enzymatic and non‐enzymatic components. Among the latter, polyphenolic compounds are of particular importance due to their redox activity, metal‐chelating properties, and regulatory roles in stress signaling [5, 11]. Flavonoids, especially aglycones, are highly effective in formation of stable complexes with metal ions and reduce oxidative damage [12, 13]. Recent reviews confirm that an increased flux through the phenylpropanoid pathway, and the consequent accumulation of phenolic acids and flavonoids, is among the most consistently reported biochemical responses of plants to abiotic stress, including exposure to potentially toxic elements [14, 15].

Heavy metals exposure has been shown to activate the phenylpropanoid pathway resulting in increased synthesis of flavonoids and related compounds. Such responses have been reported in several plant species under Cu^{2 +}, Cd^{2 +}, and combined metals [16, 17]. However, these effects are highly variable and depend on both plant species and metal type. For example, in Spinacia oleracea, mercury exposure increased flavonoid content but reduced total phenols, while arsenic had no significant effect [18]. In Larix sibirica, aluminum exposure increased multiple phenolic groups, whereas in polluted field conditions Spiraea media showed reduced flavonoid aglycones and increased phenylcarboxylic acids compared to control sites [19, 20]. Overall, these findings indicate highly complex, species‐specific, and pollutant‐dependent antioxidant responses.

In addition to polyphenols, biogenic volatile organic compounds (BVOCs), as major constituents of essential oils, represent another important class of secondary metabolites involved in plant stress responses. However, their role under heavy metal exposure remains poorly understood and is largely based on controlled experimental studies. In Ocimum basilicum, low arsenic concentrations have been reported to increase essential oil production, whereas higher concentrations (>150 mg/kg) reduce it and alter volatile composition, including increased linalool and decreased eucalyptol levels [21]. Due to the limited available data, no general conclusions can yet be drawn regarding BVOC responses under metal stress. Untargeted metabolomic approaches based on liquid chromatography coupled to multistage or high‐resolution mass spectrometry, increasingly supported by molecular networking, have proven effective for resolving such complex mixtures of specialized metabolites and were adopted here for that purpose [22].

Previous work on the Allchar flora has focused almost exclusively on the uptake, accumulation and tissue distribution of Tl and As [23, 24, 25]. The specialized‐metabolite chemistry of these taxa, in contrast, is largely unknown: apart from a single report on Thymus thracicus var. alsarensis [26], no polyphenolic profile has been published for any of the Allchar endemics, and no BVOC profile has been published for any of them.

Therefore, in the present study, for the first time, a comprehensive analysis of the polyphenolic profile and BVOCs of six selected endemic metallophytes from the Allchar Mining Area is performed using HPLC/DAD/MSⁿ and headspace GC–MS, respectively. Particular emphasis is placed on characterizing their specialized‐metabolite responses under long‐term natural exposure to the metalliferous conditions of the area, characterized by elevated concentrations of Tl, As, and Sb and on evaluating differences in comparison with botanically related species from non‐metalliferous environments. By combining two complementary analytical approaches, this study provides an integrated characterization of the major specialized metabolites associated with these endemic metallophytes and provides a chemical baseline for interpreting how these species persist under the metalliferous conditions of the area.

Because the studied taxa are endemic to the Allchar area and naturally occur under its metalliferous conditions, appropriate non‐metalliferous control populations of the same taxa are not available. Therefore, comparisons with botanically closely related species occurring on non‐metalliferous soils were used as a contextual reference, while acknowledging that such comparisons cannot unequivocally distinguish the effects of metal exposure from species‐specific, phylogenetic, phenological, and other environmental differences.

2. Results and Discussion

2.1. Analysis of BVOCs

Biogenic volatile organic compounds represent important products of both primary and secondary plant metabolism. Their qualitative and quantitative composition provides information on chemotaxonomic differentiation, as well as on specific environmental and growing conditions.

Qualitative analysis of volatile profiles revealed high chemical diversity among the investigated species. Representative chromatograms for all samples are provided in the (Figure S1a–f). A total of 123 compounds were identified, with terpenes representing the dominant group (74 compounds), including 23 polycyclic terpenes, 20 sesquiterpenes, 18 oxygenated monoterpenes, 11 monoterpenes, and 2 oxygenated sesquiterpenes. In addition, non‐terpenoid compounds were detected, including 11 aldehydes, 8 alcohols, 6 carboxylic acids, and smaller number of aliphatic hydrocarbons, esters, ketones, lactones, aromatic compounds, and nitrogen‐containing heterocyclic compounds. A representative from each group is shown in Figure 1, while Figure 2 presents a summary of the identified compounds; the complete detailed data are provided in Table S1. The dominance of terpenes is consistent with their well‐established contribution to the aroma and the bioactivity of plant essential oils [27].

FIGURE 1.

Chemical structures of detected BVOCs. The numbers in parentheses correspond to the compound numbers listed in Table S1.

FIGURE 2.

Heatmap of relative abundance (%) of BVOCs in the analyzed samples (green color) and number of different structures identified (blue color).

The highest number of different BVOCs have been identified in Thymus thracicus var. alsarensis (77 compounds) with monoterpenes being the predominant class (51.3% of total BVOCs, Figure 2). The most abundant monoterpenes were D‐limonene (29.7%) and γ‐terpinene (7.8%), both of which were not detected in any of the other investigated species. No previous data are available for the volatile profile of T. thracicus, however, marked differences were observed compared to T. vulgaris (common thyme), in which thymol (oxygenated monoterpene) is reported as dominant compound (72%), γ‐terpinene occurs only in trace amounts, and D‐limonene is absent [28]. In the present study, phenolic derivatives were also prominent in T. thracicus, including carvacrol (2‐methyl‐5‐(1‐methylethyl)phenol) (16.2%) and carvacryl acetate (0.6%), both uniquely detected in this species. Polycyclic terpenes also had relatively high abundance (13.1%), where β‐ionone was not detected, which is an exception only for this analyzed sample. Of the 20 sesquiterpenes detected across all samples, 19 were present in T. thracicus, indicating their potential chemotaxonomic relevance. Sesquiterpenes were not detected in previous studies on the most widely distributed species T. vulgaris [28].

In Centaurea leucomalla, 37 volatile compounds were identified, with aldehydes as the dominant group (44.3%), followed by aliphatic hydrocarbons (21.1%). Benzeneacetaldehyde (10.4%) was the most abundant aldehyde and has also been reported in C. scabiosa [29] and C. triumfettii [30]. Oxygenated monoterpenes accounted for 19.7%, with thymol as the major representative (19.3%), showing a markedly higher abundance compared to other studied species. Notably, thymol has not been previously reported in other Centaurea species [29, 30, 31, 32]. In contrast, sesquiterpenes, which are typically dominant in the genus [30, 32], were present only in low amounts in C. leucomalla (1.6% and 2.7% for oxygenated and non‐oxygenated sesquiterpenes, respectively).

The Knautia caorli‐rechingeri sample is characterized by 36 different BVOCs. Aldehydes were again predominant (41.6%) with 2‐methylbutanal (19.4%) and benzeneacetaldehyde (10.6%) as major constituents. Alcohols represented 16.6%, a markedly higher proportion compared to all other analyzed samples, where this group remained below 1.5%. The terpene profile is represented by oxygenated monoterpenes (7.8%) and polycyclic terpenes (2.2%), but no sesquiterpenes were detected. The highest content of oxygenated monoterpenes was observed for thymol (4.2%). Chrysanthenol and (−)‐myrtenol were uniquely detected in this sample. This research represents one of the first detailed BVOC characterizations of the genus Knautia.

In the three Viola species (V. macedonica, V. arsenica, and V. allchariensis), aldehydes were consistently the dominant class (33.6%, 30.6%, and 30.4%, respectively). 2‐Methylbutanal was the most abundant compound in all three species, as also observed in the samples of K. caroli‐rechingeri and C. leucomalla. A notable feature of these species was the relatively high proportion of furanones, lactones, and nitrogen‐containing heterocycles (10.2%–26.6%), which was considerably higher than in the other investigated taxa. Terpenes were present only in low amounts, with the exception of polycyclic terpenes (2.9%–11.3%). In this class, methyl salicylate was identified as a characteristic compound, a component that was dominant in V. dubyana and V. calcarata, according to previous research [33]. According to published data, the most widely distributed Viola species in Europe is V. odorata, whose composition is dominated by linalool and eugenol [34], two compounds that were not detected at all in the samples of these study.

Overall, the results indicate that endemic species from the Allchar area exhibit distinct and species‐specific BVOC profiles. Although these patterns may be associated with the specific environmental conditions of the Allchar habitat, the present data do not allow the contribution of individual metals to BVOC composition and abundance to be established. A potential role of metal‐related environmental conditions therefore remains a hypothesis that requires further investigation under controlled experimental conditions.

2.2. Polyphenolic Profile

2.2.1. Qualitative Analysis

A total of 78 compounds were detected and identified: phenolic acids (31), flavones (21), flavonols (20), iridoids and secoiridoids (4), flavan‐3‐ols (1) and fatty acids (1). Compound annotation was based on UV–vis spectra, MS and MS ⁿ fragmentation patterns, and comparison with literature data, following an untargeted annotation workflow analogous to that applied to other complex leaf extracts [22]. Full structural information, including retention times, UV maxima, and MS data, is provided in Table 1, while representative chemical structures are shown in Figure 3. Representative chromatograms for all samples are provided in the (Figure S2).

TABLE 1.

Retention times (t_R ), UV maxima, and MS data for identified compounds.

Nos.	Compound names	t R/min	λ max	[M‐H]−	MS2	Detected in sample a
Phenolic acid derivatives
Caffeoyl derivatives
Mono‐ and dicaffeoyl derivatives
2	3‐Caffeoylquinic acid	7.5	244, 300sh, and 326	353	191 b , 179, and 135	CL and KC
7	5‐Caffeoylquinic acid	13.3	240, 298sh, and 326	353	191	CL and KC
10	Caffeic acid	15.4	244, 294sh, and 326	179	161, 147, and 135	TA
12	4‐Caffeoylquinic acid	16.6	240, 298sh, and 326	353	191	CL and KC
35	1,3‐Dicaffeoylquinic acid	22.2	244, 300sh, and 326	515	353, 191, and 179	KC
50	1,4‐Dicaffeoylquinic acid	25.1	244, 298sh, and 328	515	353, 317, 299, 255, and 203	KC
54	3,4‐Dicaffeoylquinic acid	26.7	242, 296sh, and 330	515	353 and 299	CL
56	1,5‐Dicaffeoylquinic acid	27.5	242, 294sh, and 332	515	353 and 331	CL
62	3,5‐Dicaffeoylquinic acid	29.9	236, 296sh, and 326	515	353, 191, 179, and 173	KC
64	4,5‐Dicaffeoylquinic acid	30.4	248, 296sh, and 326	515	353 and 299	CL
	Salvianolic acids
46	Salvianolic acid A	24.5	234, 254, 288, 310, and 344sh	493	295	TA
53	Salvianolic acid K	26.6	236, 288sh, and 324	555	493 and 359	TA
73	Isosalvianolic acid A	34.3	234, 288, and 332	493	359, 331, 197, 179, and 161	TA
75	Salvianolic acid F	36.8	232, 272, and 310	313	269, 243, and 161	TA
	Glycosides
4	Caffeic acid 3‐O‐glucoside	9.6	252, 290, and 322	341	297, 179, and 135	KC
	Depsides (caffeic acid esters)
33	Rosmarinic acid glucoside	21.3	256 and 328	521	359 and 161	TA
47	Rosmarinic acid isomer 1	24.9	234, 284, and 316	359	223, 197, 179, and 161	TA
52	Rosmarinic acid isomer 2	25.9	236, 288sh, and 330	359	223, 197, 179, 161, and 133	TA
	Coumaroyl derivatives
13	Coumaroyl pentoside	16.9	232 and 314	295	163	KC
21	Coumaroylquinic acid	19.3	314	337	191 and 173	CL and KC
	Feruloyl derivatives
19	Feruloyl arabinoside	18.3	244, 296sh, and 328	325	193	KC
32	5‐Feruloylquinic acid	20.9	236, 294sh, and 326	367	191	CL
51	Caffeoyl feruloylquinic acid	29.9	244, 294sh, and 328	529	367, 353, and 191	KC
65	Feruloyl tartaric acid (fertaric acid)	30.7	236, 298sh, and 326	325	193, 178, 149, and 135	KC
	Hydroxybenzoic acid derivatives
1	Gentisoyl glucoside	6.7	316	315	153 and 108	CL
3	Protocatechuic acid glucoside	7.8	238 and 316	315	156	VAr
5	Protocatechuic acid glucoside pentoside	10.2	234 and 312	447	315, 271, 207, and 152	VAr
6	Hydroxybenzoic acid glucoside	13.1	298 and 326	299	137	VAr, VAl, and VM
	Other
8	Sinapic acid hexoside	14.1	238, 294sh, and 312	369	223, 205, and 129	VAr
14	Dimer of 3‐(3,4‐dihydroxyphenyl) lactic acid	16.9	278	377	359 and 161	TA
16	Hydroxyjasmonic acid‐O‐hexoside	17.9	272	387	369, 207, and 163	TA
	Flavones
	Apigenin derivatives
17	Vicenin‐2 (apigenin‐6,8‐di‐C‐glucoside)	18, 0	236, 272, and 338	593	575, 503, 473, 383, and 353	VAr, VAl, VM, and CL
22	Apigenin 6‐C‐glucosyl‐7‐O‐glucoside	19.3	236, 270, and 336	593	473, 431, 341, 311, and 283	KC
24	Apigenin 6‐C‐glucoside 8‐C‐arabinoside (schaftoside)	19.6	272 and 326	563	545, 519, 503, 473, 443, 383, and 353	VAl and VM
27	Apigenin‐di‐C‐glucoside	20.5	234, 272, and 334	593	503, 473, and 353	CL
37	Apigenin‐C‐rhamnoside‐C‐hexoside isomer	22.3	272 and 334	577	559, 503, 473, 383, and 353	VAr, VAl, and VM
38	Apigenin 8‐C‐pentoside 6‐C‐hexoside	22.6	232, 272, and 336	563	545, 503, 473, 443, 382, and 353	CL
40	Apigenin‐C‐rhamnoside‐C‐hexoside isomer	23.0	258, 272, and 336	577	559, 503, 473, 457, 383, and 353	VAr
42	Apigenin 8‐C‐hexoside 6‐C‐pentoside	23.3	234, 270, and 334	563	545, 503, 473, 443, 382, and 353	CL
43	Apigenin‐C‐rhamnoside‐C‐hexoside isomer	23.4	236, 272, and 336	577	473, 383, and 353	VAr and VM
44	Acacetin‐C‐glucoside	23.8	234, 270, and 336	445	325, 297, 282, and 231	KC
45	Apigenin‐6‐C‐glucoside‐8‐C‐rhamnoside	24.1	236, 270, and 340	577	559, 487, 457, 413, 383, and 353	VAr, VAl, and VM
	Luteolin derivatives
25	Luteolin 6‐C‐hexoside‐8‐C‐pentoside	20.0	274 and 330	579	489, 459, 399, and 369	VAr
26	Isoorientin (luteolin‐6‐C‐glucoside)	20.4	238, 258, 268, and 350	447	429, 357, and 327	KC
31	Chrysoeriol‐C‐glucoside	20.9	244, 258, 268, and 350	461	341, 313, and 299	KC
34	Luteolin‐8‐C‐(6”acetyl)‐β‐D‐glucopyranoside	21.6	272 and 350	489	447 and 327	VM
41	Luteolin‐7‐O‐glucoside	23.3	236, 284, and 330	447	285	TA
55	Luteolin‐4’‐O‐glucoside	27.4	236, 266, 298, and 334	447	327, 285, and 255	KC
61	Luteolin O‐hexoside	29.8	232, 286, and 332	447	285	TA
68	Luteolin O‐dihexoside	31.3	232, 284, and 332	609	447 and 285	ТА
71	Luteolin O‐acetyl‐dipentoside	32.4	230, 266, and 346	591	549, 531, and 285	ТА
	Isoflavone
39	Genistein‐C‐rhamnoside‐C‐hexoside	22.7	266 and 350	577	503, 473, 457, 413, 383, and 353	VAl
	Flavonols
	Quercetin derivatives
15	Quercetin di‐C‐hexoside	16.9	262, 290, and 356	609	619, 489, 399, and 369	CL
18	Quercetin di‐C‐hexoside	18.1	234, 252, 272, and 348	609	619, 489, 399, and 369	CL
29	Quercetin 7‐O‐glucoside	20.7	236, 254, 282, and 346	463	301	TA
30	Quercetin‐rhamnosyl‐rutinoside	20.7	256 and 354	755	609, 591, 489, 301, and 271	VAl
48	Isorhamnetin‐rhamnosyl‐rutinoside	24.9	256 and 354	769	623 and 315	VAr
49	Quercetin‐3‐О‐hexoside‐7‐О‐pentoside	24.9	276 and 344	595	301	CL
58	Isorhamnetin 3‐O‐hexoside 7‐O‐rhamnoside	28.5	254 and 354	623	315 and 300	VAr, VAl, and VM
59	Patuletin 3‐O‐hexoside	28.7	246, 292, 316, and 348	493	331	CL
63	Isorhamnetin (3‐O‐methylquercetin)	30.2	286 and 336	315	300	TA
67	Quercetin‐O‐acetylhexoside	31.1	274 and 346	505	301	CL
70	Patuletin acetyl hexoside	31.8	240, 258, and 354	535	331	CL
	Kaempferol derivatives
23	Kampferol‐O‐deoxyhexoside‐O‐feruloylrutinoside	19.4	258, 266, and 350	915	769, 593, 431, and 285	VAr
28	Kaempferol 3‐O‐glucoside	20.5	256, 268, 292, and 350	447	423, 357, 327, and 285	VAr and VM
57	Kaempferol‐3‐O‐rutinoside	27.9	238, 266, 304, and 352	593	285	VAr, VAl, and VM
60	Kaempferol dipentoside	28.9	254 and 350	549	417, 327, and 285	TA
66	Kaempferol acetyl pentosyl‐hexoside	30.7	232, 254, and 352	621	579, 561, and 285	TA
72	Kaempferol‐O‐acetylhexoside	33.0	234, 266, and 342	489	285	CL
76	Kaempferol 3‐(6''‐acetylglucoside)‐7‐rhamnoside	38.2	236, 268, and 314	635	575, 489, 349, and 285	KC
	Other
36	Myricetin‐hexoside	22.2	238, 260, and 358	479	317	CL
74	Flavanomarein	34.7	250, 298sh, and 330	449	287 and 173	KC
	Flavan‐3‐ol
11	Catechin	15.9	280	289	271 and 229	VM and VAr
	Iridoids and secoiridoids
9	Oleoside (secologanoside)	15.4	240, 296sh, and 324	389	345, 209, 165, and 121	KC
20	Loganic acid glucoside	18.9	266	537	375 and 179	VM and VAl
69	Unknown 1	31.5	292	429	385, 249, 205, 179, and 161	VM, VAl, and VAr
77	p‐Coumaroylgeniposidic acid	41.8	260 and 316	519	477, 459, 163, and 145	TA
	Fatty acids
78	Trihydroxy‐octadecenoic acid	46.7	228, 276, and 322	329	311, 293, 229, 211, and 183	TA

^a CL—C. leucomalla, KC—K. caroli‐rechingeri, TA—T. thracicus var. alsarensis, VAl—V. allcharinesis, VAr—V. arsenica, VM—V. macedonica.

^b Numbers in bold refer to base fragment ion in MS².

FIGURE 3.

Chemical structures of identified polyphenolic compounds. The numbers in parentheses correspond to the compound numbers listed in Table 1.

Phenolic acids (31 compounds) constituted the largest group, showing characteristic UV maxima at 240–250 nm and 320–330 nm, consistent with hydroxycinnamate‐type structures. Among them, caffeic acid derivatives (19 compounds) were dominant, including mono‐ and dicaffeoylquinic acids, salvianolic acids, glycosides, and depsides (caffeic acid esters). In general, Viola species exhibited lower diversity of phenolic acid derivatives compared to Knautia, Thymus, and Centaurea, in agreement with previous reports [35, 36]. Caffeic acid (10) was detected only in T. thracicus, whereas monocaffeoylquinic acids (2, 7, and 12) were present in both Centaurea and Knautia. In a previous study of this species, caffeic acid was detected only in glucoside form [26].

All six isomers of dicaffeoylquinic acids (35, 50, 54, 56, 62, and 64) were identified in these genera, showing characteristic MS fragmentation (m/z 515 → 353). All of the structures classified as salvianolic acids and depsides were detected in the T. thracicus sample, of which only compound 46 was previously reported [26].

Rosmarinic acid derivatives (33, 47, and 52) were mainly found in T. thracicus, consistent with a pronounced contribution of esterified caffeic acid conjugates to the phenolic pool of this taxon. Compound 14 was also detected in Thymus, supporting the diversity of caffeic‐acid‐derived structures in this genus previously reported in three other Thymus species from Croatia [37]. In contrast, coumaric acid derivatives (13 and 21) and feruloyl derivatives (19, 32, 51, and 65) were restricted to K. caroli‐rechingeri and C. leucomalla, suggesting species‐specific allocation of hydroxycinnamate metabolism. Sinapic acid hexoside (8) was detected exclusively in V. arsenica, while hydroxyjasmonic acid‐O‐hexoside (16) was present in Thymus species, in agreement with earlier reports (for T. thracicus) [26, 38]. Hydroxybenzoic acid derivatives (1, 3, 5, and 6) were characteristic for Viola species. Although, hydroxybenzoic acids, such as gentisic and protocatechuic acids have been reported in several Viola species [39, 40], but reports of their glycosylated derivatives remain scarce.

Flavones (21 compounds) included derivatives of apigenin, luteolin, chrysoeriol, and acacetin occurring mainly as C‐ and O‐glycosides. Additionally, C‐glycosylated derivative of genistein was detected, as the only representative of isoflavones. C‐Glycosylated apigenin derivatives were widely distributed across most species, reflecting a conserved biosynthetic pattern, whereas T. thracicus lacked apigenin derivatives but contained luteolin O‐glycosides, indicating a shift in flavonoid biosynthetic routing. Luteolin and chrysoeriol as aglycones were also detected. Previous profiling studies have demonstrated that Viola, Thymus, Centaurea, and, to a lesser extent, Knautia species are rich sources of flavone derivatives, particularly apigenin‐ and luteolin‐based glycosides [41, 42]. According to literature data, among the investigated taxa, Viola spp. showed the strongest evidence for the occurrence of apigenin, luteolin and chrysoeriol derivatives [39], including both glycosylated forms and corresponding aglycones, whereas reports of genistein derivatives remain unclear.

Flavonols (20 compounds) were dominated by quercetin and kaempferol derivatives. Quercetin glycosides (29, 48, 49, 58, 63, 67, and 70) were identified via the diagnostic ion at m/z 301, with characteristic neutral losses of hexose, pentose, and acetylhexose moieties. Kaempferol derivatives (23, 57, and 76) were exclusively O‐glycosides, yielding the aglycone ion at m/z 285. Notably, compound 76, identified as kaempferol 3‐(6''‐acetylglucoside)‐7‐rhamnoside, which to the best of our knowledge has not previously been reported in the genus Knautia and within Caprifoliaceae, indicating rare chemotaxonomic specialization under extreme environmental conditions. The compound has only been described once in the literature, in Delphinium formosum [43], indicating a very limited distribution. Its detection in this study therefore represents the first report for both the genus and the family, expanding the known chemotaxonomic occurrence of acetylated kaempferol glycosides.

Additionally, compound 74 was identified as flavanomarein. This compound is characteristic for the Asteraceae [44, 45] family, but it has not been reported for the Knautia genus, or within the family Caprifoliaceae.

Flavan‐3‐ols were represented by catechin (11), detected only in V. macedonica and V. arsenica, consistent with its known occurrence in the genus Viola. These compounds have been reported in V. odorata [46] and V. calcarata [47].

Iridoids and secoiridoids (9, 20, and 77) included oleoside derivatives, loganic acid, and coumaric acid conjugates, which were identified based on characteristic neutral losses (CO₂, hexose) and diagnostic fragment ions. Iridoids and secoiridoids are considerably less common than flavonoids in the Viola, Thymus, Knautia, and Centaurea genera. Their occurrence appears to be strongly genus‐ and species‐dependent. Among the investigated genera, Knautia species (Caprifoliaceae) are the most likely source of iridoid‐type metabolites, as iridoids are characteristic chemotaxonomic markers of the Caprifoliaceae group [48].

A single fatty acid, trihydroxyoctadecanoic acid (78), was identified based on typical oxylipin fragmentation and hydroxylation patterns [49].

2.2.2. Quantitative Analysis

The total content of the detected compounds varied considerably among the analyzed samples, following the order: K. caroli‐rechingeri (453.23 mg/g), V. macedonica (47.28 mg/g), C. leucomalla (38.12 mg/g), V. arsenica (28.23 mg/g), T. thracicus (20.86 mg/g) and V. allchariensis (18.04 mg/g), presented in Figure 4. The detailed distribution and content of individual compounds is provided in Table S2.

FIGURE 4.

Total content of detected compounds expressed as mg/g dry herb (A) and percentage distribution (B), of all samples analyzed in triplicates.

Pronounced qualitative and quantitative differences were observed in the phenolic acid derivatives. Monomeric caffeic acid derivatives, including mono‐ and dicaffeoylquinic acids, were only detected in K. caroli‐rechingeri and C. leucomalla, where they constituted the dominant portion of the total phenolic content. This agrees with previously reported dominance of caffeoylquinic acid derivatives in related Knautia and Centaurea taxa [45, 50].

The most abundant compound across all samples was 1,3‐dicaffeoylquinic acid, reaching 405.40 mg/g in K. caroli‐rechingeri, and also being present, though at lower levels, in the other studied species. To our best knowledge, this compound is common in artichoke and other spices [51], but it has not been reported in Knautia species and certainly not in such big amount. The second most abundant compound in K. caroli‐rechingeri was 5‐caffeoylquinic acid (28.93 mg/g), which was also detected in C. leucomalla (17.99 mg/g), and is considered a key biosynthetic precursor of dicaffeoylquinic acids [52].

The phytochemical profile of C. leucomalla was similar to that of K. caroli‐rechingeri, differing by the presence of 3,4‐dicaffeoyilquinic (4.81 mg/g) and 1,5‐dicaffeoylquinic acid (4.50 mg/g). The selective enrichment of dicaffeoylquinic acid derivatives is consistent with, but does not demonstrate, a contribution to antioxidative defence; a role in metal tolerance remains a hypothesis to be tested. Such enrichment has been reported by Moglia et al. [52], whose study reports elevated content of dicaffeoylquinic acids as a response to abiotic stress. It should be noted that both K. caroli‐rechingeri and C. leucomalla have been previously investigated for their capacity to accumulate toxic metalloids. It has been shown that these taxa tend to behave as excluders, retaining metals primarily in roots, as opposed to aerial parts [25, 53]. In contrast, caffeic acid was detected only in T. thracicus, with a low concentration of 0.16 mg/g.

In the same species, all salvianolic acids (46, 53, 73, and 75) and derivatives of rosmarinic acid (33, 47, and 52) were observed only in T. thracicus var. alsarensis. Rosmarinic acid represented more than 50% of the total phenolic acids (5.76 mg/g), which is in agreement with previously published data for T. vulgaris, T. serpyllum, T. trautvetteri and other Thymus species [54]. Salvianolic acid K (2.06 mg/g) and isosalvianolic acid A (2.00 mg/g) were the most abundant derivatives, representing almost 40% of all phenolic acids present. High content of salvianolic acids was previously reported for T. thracicus [26], as well as other closely related species [54]. In previous research, the T. thracicus var. alsarensis has exhibited a differentiated metal uptake strategy where Tl is translocated to above‐ground parts of the plant, whereas As is predominantly retained in the roots [55]. Taken together, the metal accumulation pattern and the enrichment in derivatives of salvianolic and rosmarinic acid appear to be functionally linked. The selective production of these highly active phenolics likely represents and adaptive biochemical strategy that complements metal uptake mechanisms, facilitating both detoxification and tolerance.

However, differences in total abundance were not the most distinctive feature across the samples. Instead, pronounced variation was observed in the qualitative composition and relative distribution of individual constituents.

In Viola species, flavones constituted the dominant phenolic class, particularly apigenin derivatives. The most abundant compound was apigenin‐6‐C‐glucoside‐8‐C‐rhamnoside, with concentrations of 25.62 mg/g (V. macedonica), 12.66 mg/g (V. allchariensis), and 15.74 mg/g (V. arsenica). All three species also contained vicenin‐2 (0.48–2.16 mg/g), previously reported in V. yedoensis [39].

Flavonols represented the second most abundant group, with total concentrations ranging from 0.19 to 6.16 mg/g across all Viola species. The most abundant compound was kaempferol‐3‐O‐glucoside (5.58 mg/g in V. macedonica). Kaempferol‐3‐O‐rutinoside (0.11–0.60 mg/g), previously reported in V. thianscianica [56] and V. odorata [57], was also detected in all analyzed samples.

The predominance of flavonoid derivatives, combined with the markedly reduced presence of phenolic acids, indicates a markedly different allocation within phenolic metabolism in the Viola species studied here. Notably, derivatives of coumaric and ferulic acids were not detected in any of the analyzed Viola samples, indicating possible suppression or redirection of hydroxycinnamate biosynthesis toward flavonoid formation, a pattern repeatedly described for plants exposed to abiotic stress [14, 15].

This observation is particularly notable given that these violets are known Tl and As hyperaccumulators, reaching concentrations of up to 58 900 and 381 mg/kg, respectively [24, 58]. In contrast to typical phytochemical profiles reported for species such as V. tricolor and V. odorata, where both flavonoids and phenolic acids (including coumaric and ferulic acids) contribute significantly to the phenolic pool [39], derivatives of these hydroxycinnamic acids were not detected in any of the analyzed samples. This absence suggests a possible redirection of hydroxycinnamate biosynthesis toward flavonoid formation or a stress‐induced suppression of phenolic acid accumulation.

This diversity of uptake and translocation mechanisms reflects species‐specific physiological controls over metal transport, sequestration, and detoxification. More broadly, hyperaccumulation is recognized as a specialized adaptation enabling plants to tolerate and sequester exceptionally high concentrations of toxic elements in their tissues without detrimental effects [6, 7]. The observed differences between analyzed taxa therefore align with the concept that even closely related or co‐occurring species can employ fundamentally different strategies for coping with metalliferous soils. These findings reinforce the importance of species‐specific investigations when interpreting both metal accumulation and associated metabolic profiles.

2.3. Multivariate Analysis of the Metabolite Profiles and Their Relation to Previously Reported Metal Accumulation Patterns

The variability observed in both polyphenolic and BVOC profiles can be interpreted in relation to the specific geochemical conditions of the Allchar locality, characterized by elevated concentrations of Tl, As, Sb, as well as Ni, Pb, and Cd [1]. Previously reported metal accumulation patterns provide a useful framework for interpreting the metabolic differentiation observed among the studied taxa.

Viola species, recognized as Tl hyperaccumulators with comparatively lower As uptake [23, 24], exhibit profiles dominated by flavonoids, and characterized by a lower diversity of phenolic acids. Such a profile suggests a preferential allocation toward flavonoid biosynthesis, potentially linked to their role in intracellular antioxidant protection and metal chelation. The absence of coumaric and ferulic acid derivatives may further indicate a possible redirection of hydroxycinnamate metabolism [59].

In contrast, Knautia caroli‐rechingeri, reported to accumulate higher levels of As, particularly in roots [25], is characterized by a strong enrichment in caffeoylquinic acid derivatives. Given their well‐documented antioxidant properties, these compounds may contribute to mitigating oxidative stress associated with arsenic exposure. A comparable, though less pronounced, pattern is observed in Centaurea leucomalla, suggesting that hydroxycinnamate derivatives may be a shared feature of these two taxa; a common adaptive significance is not demonstrated by the present data.

In Thymus thracicus var. alsarensis, rosmarinic and salvianolic acid derivatives predominated, consistent with its differentiated metal partitioning, with Tl mainly accumulated in aerial parts and As retained in roots [55]. These patterns suggest that the studied taxa employ distinct phenolic metabolic strategies that may contribute to protection against metal‐associated oxidative stress. The BVOC profiles showed similarly pronounced species‐specific differentiation, with terpene‐dominated profiles in Thymus, aldehyde‐rich profiles in Centaurea and Knautia, and a greater contribution of non‐terpenoid compounds in Viola.

Multivariate analysis confirmed that the observed differences represented consistent, species‐specific patterns (Figure 5). In the polyphenolic dataset, PC1 and PC2 accounted for 37.3% and 31.9% of the variance, respectively, with the three Viola species forming a distinct group associated mainly with apigenin C‐glycosides, while T. thracicus var. alsarensis and C. leucomalla were separated along PC2 by rosmarinic/salvianolic acid derivatives and 5‐caffeoylquinic acid, respectively (Figure 5A). For the BVOCs, PC1 explained 53.5% of the variance and separated T. thracicus var. alsarensis from the remaining taxa, primarily along a gradient involving D‐limonene, carvacrol and γ‐terpinene versus 2‐methylbutanal, acetic acid and benzeneacetaldehyde (Figure 5B). Hierarchical clustering supported these patterns, with the Viola group recovered in 100% and 89% of bootstrap replicates for the polyphenolic and BVOC datasets, respectively. The association of C. leucomalla and K. caroli‐rechingeri was supported in 95% of the polyphenolic replicates, while T. thracicus var. alsarensis was separated from the other taxa in 100% of the BVOC replicates (Figure 5D,E).

FIGURE 5.

Multivariate analysis of polyphenolic and BVOC profiles of the six Allchar endemic taxa and their relationship to previously reported metal accumulation. (A, B) Principal component analysis (PCA) of polyphenolic (A) and BVOC (B) profiles. (C) Mantel test showing the relationship between polyphenolic and BVOC dissimilarities. (D, E) Hierarchical clustering of polyphenolic (D) and BVOC (E) profiles; numbers indicate bootstrap support from 2000 resamplings. (F) Spearman correlations between metabolite classes and previously reported aerial‐part Tl concentrations. PP, polyphenolic class; BVOC, volatile class. Symbols and colors indicate taxa and previously documented metal‐handling behavior, respectively; sample codes are as listed in Table 2. Summary of all tests is given in Table S6

Importantly, the two chemically distinct metabolite datasets showed a high degree of congruence. Their dissimilarity matrices were strongly correlated (Mantel ρ = 0.72 and exact one‐sided p = 0.0083; Figure 5C), indicating that polyphenolic and volatile profiles independently captured a similar species‐specific organization. This concordance provides stronger evidence for taxon‐associated metabolic differentiation than would be obtained from either analytical platform alone.

In contrast, grouping taxa according to their previously documented metal‐handling behavior did not provide statistically significant separation. PERMANOVA comparing the three Tl‐hyperaccumulating Viola species with the three non‐hyperaccumulating taxa gave F = 3.67 for polyphenols and F = 2.06 for BVOCs, with an exact p = 0.100 in both datasets. Because only six taxa were available, the statistical resolution of this comparison was intrinsically limited; therefore, these results should not be interpreted as evidence either for or against an effect of Tl hyperaccumulation on overall metabolite composition.

A direct assessment of metabolite–metal relationships was likewise constrained by the absence of metal measurements in the present samples. Published aerial‐part concentrations were available for Tl in five taxa, As in three and Sb in only one, and originated from different studies, collection periods and, in some cases, plant organs (Table S3). Consequently, only exploratory Spearman correlations with reported aerial Tl concentrations could be evaluated. For this purpose, compound‐class composition of the six samples, expressed as the percentage of the total content of identified compounds is provided in Table S4. None of the tested metabolite classes remained significantly associated with Tl after correction for multiple comparisons (Figure 5F and Table S5). The only nominally significant association was observed for volatile alcohols (ρ = −1.00 and p = 0.017), but this value represents the minimum attainable p‐value for a perfect rank correlation with n = 5 and was not significant after Benjamini–Hochberg correction (adjusted p = 0.53). Similarly, total polyphenol content (ρ = −0.20 and p = 0.78) and the number of detected volatile compounds (ρ = −0.50 and p = 0.45) were not related to the reported Tl concentrations. These findings are consistent with our previous study of T. thracicus var. alsarensis [26], in which total polyphenol content was not associated with As or Tl concentrations measured in the same plants. Overall, the available evidence supports pronounced species‐specific metabolic differentiation, but does not provide sufficient evidence to attribute this variation directly to metal accumulation.

3. Conclusions

The present study demonstrates that endemic metallophytes from the Allchar Mining Area exhibit marked differences in both polyphenolic and BVOC composition, as revealed by complementary HPLC/DAD/MS ⁿ and headspace GC–MS analyses. Distinct metabolite patterns among the studied taxa highlight the biochemical diversity of plants inhabiting this geochemically extreme environment. These differences were reflected in the relative abundance and diversity of phenolic compounds and volatile metabolites, providing a broader biochemical perspective on the specialized metabolism of the studied endemic metallophytes.

For the first time, BVOC profiles were characterized for all six selected Allchar endemics, while polyphenolic profiles were reported for five of the six taxa. The work also represents one of the first detailed characterizations of volatile compounds within the genus Knautia. Several metabolites were newly reported for the corresponding taxa or taxonomic groups, including kaempferol 3‐(6′′‐acetylglucoside)‐7‐rhamnoside and flavanomarein in K. caroli‐rechingeri, thymol in C. leucomalla, and a high content of 1,3‐dicaffeoylquinic acid in K. caroli‐rechingeri. These findings extend the current knowledge of specialized‐metabolite diversity in the Allchar endemic flora and establish a useful chemical baseline for these poorly characterized taxa.

Multivariate statistical analysis confirmed clear species‐specific patterns in both polyphenolic and BVOC profiles. The significant concordance between the two metabolite datasets (Mantel ρ = 0.72 and p = 0.008) supports a consistent taxon‐specific organization across the two chemically distinct metabolite groups. In contrast, no significant associations were detected between metabolite classes and previously reported aerial‐part Tl concentrations after correction for multiple testing. Therefore, the observed metabolic differences cannot be directly attributed to metal accumulation based on the available data.

The observed metabolite signatures may provide hypotheses regarding potential roles of specialized metabolites in plant responses to metalliferous conditions and may contribute to understanding how these species persist in the Allchar environment. However, the present findings do not establish direct mechanisms of metal tolerance or adaptation. Further studies incorporating replicated sampling, concurrent determination of soil and plant Tl, As and Sb concentrations, comparisons of metalliferous and non‐metalliferous populations, and controlled exposure experiments are needed to clarify the functional relationship between metal exposure and species‐specific metabolic responses.

4. Experimental Section

4.1. Materials

Aerial parts of six endemic taxa were collected near the former As‐Sb‐Tl Allchar mine (Kožuf Mountain, North Macedonia), at the Crven Dol locality, during the full flowering period on 21 June 2024. The geographical coordinates of each sample are given in Table 2. Only material free of adhering soil and dust was taken. The plants were collected and taxonomically identified by K. Bačeva Andonovska (Research Centre for Environment and Materials, Macedonian Academy of Sciences and Arts, Skopje). Voucher specimens of all six taxa are deposited in the Macedonian National Herbarium (MKNH) at the Institute of Biology, Faculty of Natural Sciences and Mathematics, Ss. Cyril and Methodius University in Skopje, which is listed in Index Herbariorum under the acronym MKNH; the registration number of each voucher specimen is given in Table 2.

TABLE 2.

Collection data for analyzed samples.

Plant family	Plant species	Codes	Date of collection	Locations	Coordinates	Voucher nos.
Violaceae	Viola arsenica Beck	VAr	21 June 2024	Crven Dol	41.160; 21.951	242161 (MKNH)
Violaceae	Viola allchariensis Beck	VAl	21 June 2024	Crven Dol	41.162; 21.947	242162 (MKNH)
Violaceae	Viola tricolor subsp. macedonica Boiss. & Heldr	VM	21 June 2024	Crven Dol	41.164, 21.945	242163 (MKNH)
Lamiaceae	Thymus thracicus Velen. var. alsarensis	TT	21 June 2024	Crven Dol	41.159; 21.948	242164 (MKNH)
Asteraceae	Centaurea leucomalla Bornm.	CL	21 June 2024	Crven Dol	41.159; 21.952	242165 (MKNH)
Caprifoliaceae	Knautia caroli‐rechingeri Micevski	KC	21 June 2024	Crven Dol	41.162; 21.950	242166 (MKNH)

4.2. BVOCs Analysis

Plant material was ground using an electric mill. Approximately 50 mg of the powdered sample was weighed into a GC glass vial, using an analytical balance (±0.01 mg); each sample was prepared and analyzed in triplicate. The samples were analyzed using headspace‐gas‐chromatography‐mass‐spectrometry (HS‐GC–MS) method. The analysis was performed using an Agilent 8890N gas chromatograph coupled to a 5977B mass spectrometer in electron impact ionization mode at 70 eV. An Agilent HP‐5MS UI (30 m × 250 µm × 0.25 µm) column was used with helium as mobile phase with flow rate of 1 mL/min. Samples were incubated at 180°C for 20 min, then injected volume of 1 mL (pulsed split mode 1:10) and injection temperature of 250°C. The oven temperature gradient is given below. The oven temperature programme is given in Table 3:

TABLE 3.

Temperature programme of the oven used for the HS‐GC–MS analysis.

Rate/(°C/min)	Temperature/°C	Hold time/min
2	80	0
5	180	0
20	260	1

Identification of compounds was made by assignment of their mass spectra and comparing them with the NIST spectral libraries (NIST MS Search 2.3). Relative abundances are expressed as the percentage of the total peak area of the identified compounds, without correction for response factors.

4.3. Extraction of Non‐Volatile‐Compounds

The powdered material was accurately weighed (∼0.1 g) on an analytical balance (±0.01 mg). Each taxon was represented by one composite sample of aerial parts pooled from several individuals, and all extractions and instrumental analyses were performed in triplicate. The extraction of polyphenolic compounds was conducted using 15 mL natural deep eutectic solvent composed of choline chloride and acetic acid in molar ratio 1:2, with 40% (m/m) water content. Ultrasonic assisted extraction was conducted for 40 min, at 50°C. Extracts were filtered (RC 0.45 µm) and diluted 1:1 (V/V) prior to instrumental analysis. The procedure was previously optimized by our research group [60].

For HPLC/DAD/ESI‐MS ⁿ analysis Agilent 1100 Series instrument equipped with a diode‐array detector (DAD) and an ion‐trap mass spectrometer (Agilent Technologies, Waldbronn, Germany) was used. A XDB‐C18 Eclipse column (150 × 4.6 mm, 5 µm, Agilent, USA) was used for chromatographic separation. The mobile phase consisted of water with 1% formic acid (solvent A) and methanol (solvent B). Gradient elution was applied as follows: the run started with 20% B, which was maintained until 5 min, then gradually increased to 40% B at 15 min, 70% B at 45 min, and finally 90% B at 55 min, which was held until 60 min. After each run, a 10‐min equilibration period was applied. The flow rate was 0.5 mL min^{− 1}, the injection volume was 10 µL, and the column temperature was maintained at 38°C.

Spectral data from all peaks were accumulated in the 190–600 nm range, and chromatograms were recorded at 280, 300, 330 and 350 nm. MS and MS [2] spectra were acquired in the negative ionization mode across an m/z range of 100–1200 with an electrospray ionization (ESI) system. The nebulizing gas was nitrogen at 65 psi with a flow rate of 12 L/min.

Identification of compounds was based on UV and MS spectral characteristics and comparison with reference standards and literature data.

Hydroxycinnamic acid derivatives were quantified as caffeic acid equivalents at 330 nm. Flavones and flavonols were quantified as luteolin and isorhamnetin equivalents at 350 nm. Iridoids were quantified as catalpol equivalents at 330 nm. Calibration curves for all standards were constructed over the concentration range of 1–500 µmol L⁻¹.

4.4. Statistical Analysis

Principal component analysis (PCA) was performed on log(1 + x)‐transformed and Pareto‐scaled metabolite data to assess patterns in polyphenolic and BVOC composition. Hierarchical clustering was based on Bray–Curtis dissimilarities with average linkage, and cluster stability was evaluated by bootstrap resampling of compounds (2000 replicates). Concordance between polyphenolic and BVOC profiles was assessed using a Mantel test based on Spearman correlation of the corresponding dissimilarity matrices. Differences related to previously reported metal‐handling behavior were evaluated by PERMANOVA. Exploratory Spearman correlations were used to assess relationships between metabolite classes and previously reported aerial‐part Tl concentrations for the five taxa with available data, with Benjamini–Hochberg correction for multiple testing. Due to the limited number of taxa and available published metal data, As and Sb were not included in the correlation analysis. All permutation‐based tests were evaluated by complete enumeration to obtain exact p‐values. Statistical analyses were performed using Python 3.11 (NumPy 2.4, SciPy 1.17, scikit‐learn 1.8).

Author Contributions

M.C. and J.P.S. designed the study. K.B.A. collected and verified the plant samples. M.C. performed the experimental procedures. M.C. and J.P.S. performed the data analysis. M.C. and J.P.S. wrote the manuscript. M.S. reviewed the manuscript. All authors read and approved the final manuscript. All figures, schemes, and tables in this manuscript and in the Supporting Information were created originally by the authors for this work. No material has been reproduced or adapted from previously published sources, and no third‐party copyrighted images, logos or trademarks are included; no copyright permissions are therefore required.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: cbdv71761‐sup‐0001‐SuppMat.docx.

Data Availability Statement

Data available on request from the authors.