*2.1. Validation*

Mycotoxin analysis in food and plant ingredients is always based on several factors, including the composition and nature of the matrix or ingredients to control, and the mycotoxin to investigate. In this study, the used modified dispersive liquid−liquid microextraction (DLLME) and LC-MS/MS methods were checked, and validation of the parameters of recoveries, linearity, LOD, and LOQ were carried out (Table 1 and Supplementary Table S1).

**Table 1.** Values of repeatability (mean recoveries of the triplicate and matrix effect ME) and sensitivity with a mix of blank AMP samples spiked at 10 times LOD.


a SD: Standard deviation.

Linearity was evaluated with calibration curves, which were constructed for each mycotoxin with methanol and a blank sample at concentration levels ranging from the LOQ to 1 μg/mL. All studied mycotoxins presented good linearity, with correlation coefficients (r2) higher than 0.9995. Substances present in the matrix that modified the instrumental response of the analyte were evaluated by matrix effects (MEs), resulting in enhancement or suppression of the signal, so that signal suppression/enhancement was evaluated comparing the slopes of calibration curves obtained with methanol and in the blank sample. ME values higher than 100% indicated enhancement of the signal, ME values lower than 100% indicated suppression of the signal, and ME values near 100% indicated no significant matrix effects. The accuracy has been evaluated with recoveries (R) of the analytical method, so that a mix of blank samples (negative, < LOD) were spiked at three levels (LOQ, 2 LOQ. and 10 LOQ). Recovery values ranged from 81 ± 4.26% (HT-2) to 125 ± 14.18% (EN A1) at 10 times LOQ (Table 2). In Supplementary Table S2, the R at 10 times LOQ (*n* = 3, intraday study) and the ME observed by each studied AMP sample (*n* = 3) are shown. High ion suppression in *Rosmarinus officinalis, Matricaria chamomilla,* and *Myrtus communis* samples were observed.

#### *2.2. Natural Occurrence of Mycotoxins*

Out of 40 total AMP samples, 36 samples (90%) presented at least one mycotoxin (Table 3). All analyzed samples of *Mentha spicata, Lavandula intermedia, Matricaria chamomilla,* and *Myrtus communis* were contaminated by at least one mycotoxin. The most frequent mycotoxins in AMP were AOH (85%), ZEN (27.5%), AFG1 (17.5%), TENT (17.5%), ENB (10%), AFG2 (7.5%), ENA1 (2.5%), and HT-2 (2.5%) (Table 3), while the mycotoxins AFB1, AFB2, OTA, BEA, ENA, ENB1, and T-2 were under the LOD. The highest mycotoxin value found in AMP samples

was registered in a sample of *Origanum vulgare,* with 309 ng/g of AOH (Table 3). Below, the occurrence in the analyzed samples by group of mycotoxins is detailed.


**Table 2.** MS/MS parameters for mycotoxin detection by multiple reaction monitoring (MRM).

a Rt: Retention time; b Q: quantification transition; q: qualification transition; c de-clustering potential (DP), collision energy (CE), collision cell entrance potential (CEP), and collision cell exit potential (CXP) are all expressed in voltage.

**Table 3.** Mycotoxin incidence, mean of positive (Mp) samples, and range levels distributed according to the eight studied species of AMP varieties.


Raw tea and herbal infusion materials were reported to contain up to 76 μg/kg of fumonisin B1, but no mycotoxins were detected in infusions [25]. In China, the presence of ZEN and its metabolite α-zearalenol ( α-ZEL) in 100 widely-consumed foods and medicinal plants was investigated. Authors reported that 12% of these tested samples were contaminated with ZEN at levels ranging from 5.3 to 295.8 μg/kg [26]. Another study from Spain reported the occurrence of T-2 and HT-2 in seeds of milk thistle (*Silybum marianum*) at levels ranging from 363 to 453.9 μg/kg and from 826.9 to 943.7 μg/kg, respectively [27]. A study from India reported that dried market samples of stem portions of *Tinospora cordifolia,* an important medicinal plant, were contaminated with AFB1, AFB2, OTA, patulin, and citrinine; however, fusarial species and their toxins were not detected in those samples [28]. In Latvia, the occurrence of 12 mycotoxins has been recently investigated in 60 herbal teas. Among the dry tea samples, 90% were positive from one to eight mycotoxins. ENB, DON, AFB1, and OTA were the most frequently detected mycotoxins in 55%, 45%, 20%, and 10% of samples, respectively. The authors reported that 32% and 100% of DON and ZEN, respectively, present in dry teas were extracted into the infusions ready for its consumption [29]. A study from Spain showed the presence of AFB2 (19.1–134.7 μg/L) and AFG2 (2.2 to 13.5 μg/L) in botanical dietary supplement infusion beverages, and ENB in two samples, although at low levels [30]. More recently, AFs were detected in green tea samples obtained from retail shops and supermarkets in three Moroccan areas; however, the rate transfer of AFs from herbal green tea to infusion was unavailable, as it was not investigated [31].

#### 2.2.1. Aflatoxins (AFG1 and AFG2)

AFG1 was detected in seven samples (17.5%), including two samples of *Lavandula intermedia* (5%) and five samples of *Myrtus communis* (12.5%) (Tables 3 and 4). Levels of AFG1 ranged from 4.9 to 8.6 ng/g, and the mean level of AFG1 in positive samples was 4.6 ± 1.4 ng/g (Table 4). Concerning the presence of AFG2, it was detected in three samples of *Rosmarinus officinalis* (7.5%). Levels of the AFG2 in this plant ranged from 26.2 to 41.1 ng/g, with a mean value of 27.7 ± 2.1 ng/g (Tables 3 and 4). It should be highlighted that three positive AMP samples (*Lavandula intermedia*, *Myrtus communis,* and *Rosmarinus officinalis)* exceeded the ML (10 ng/g) of Moroccan regulation set for the sum of AFs [32].


**Table 4.** Occurrence, mean levels, probable daily intake (PDI), and risk of dietary exposure of studied mycotoxins through analyzed Moroccan AMPs.

a Mp: mean of positive samples; Mt: mean of total analyzed samples; SD: Standard deviation; b LB: Lower bound; c UB: Upper bound; de

 Values close to LOD; Value between LOD and LOQ.

#### 2.2.2. *Fusarium* Toxins (ZEN and HT-2)

Determination of ZEN in AMP samples showed that 27.5% of samples were positive for this mycotoxin (Table 4) specifically: two samples of *Mentha spicata*, six samples of *Rosmarinus officinalis,* and three samples of *Origanum vulgare*. Levels of ZEN varied between 33.7 and 114.7 ng/g, and the mean ZEN level was 55.7 ± 26 ng/g. Recent studies have also detected ZEN in AMP as Duarte et al., who detected them in 19 herb samples with smaller ranged values (1.82–19.02 ng/g) than those detected in our analyzed samples [33]. Concerning the presence of HT-2, one AMP sample of *Verveine officinale* (2.5%) contained this mycotoxin, with levels up to 2.9 ng/g, and a mean level of 1.47 ± 2.6 ng/g.

#### 2.2.3. Emerging Mycotoxins (ENA1 and ENB)

In this survey, only ENA1 and ENB were detected among emerging mycotoxins in AMP samples. Regarding the occurrence of ENA1 in AMP samples, this toxin was detected only in one sample (2.5%) of *Verveine officinale,* with a contamination level up to 0.3 ng/g and a mean level of 0.16 ± 0.3 ng/g. For the presence of ENB in AMP samples, four samples (10%) were contaminated: one sample of *Verveine officinale* and three samples of *Lavandula intermedia* (Table 3). Levels of the ENB were detected in a range of 0.04–0.1 ng/g, and the mean level was 0.05 ± 0.1 ng/g.

#### 2.2.4. *Alternaria* Toxins (AOH and TENT)

*Alternaria* mycotoxins gain more and more interest due to their frequent contamination of food commodities. Indeed, these toxins are often detected in fruits, vegetables, and wines [34]. Besides the estrogenic activity demonstrated in vitro for certain *Alternaria* toxins, AOH causes DNA damage and cell cycle arrest [35]. In the present survey, 34 samples (85%) presented levels of AOH. Nine of them were *Origanum vulgare.* Levels of AOH ranged from 2.3 to 309 ng/g, and the mean level was 126.2 ± 40.4 ng/g. TENT was detected in seven samples (17.5%) as follows: five samples of *Myrtus communis* and two samples of *Lavandula intermedia*. Levels of TENT varied from 0.7 to 4.5 ng/g, and the mean level was 1.47 ± 0.85 ng/g.

#### *2.3. Co-Occurrence of Mycotoxins in AMP*

The co-presence of mycotoxins in a single sample could be a health concern due to the exposure of consumers to multiple fungal metabolites, which might exert greater toxicity than the exposure to a single one. The multi-mycotoxin occurrence in food and feed could be associated with health and reproductive disorders, lower performance in animals, and higher medical costs [36]. Concerning the mycotoxins' co-occurrence in AMP samples, this happened in 52% of samples. Figure 1 summarizes the data obtained on the multi-contamination of AMP samples, revealing that the ZEN + AOH combination was the most commonly present (20%).

Analytical results showed that four mycotoxins co-occurred in samples of *Verveine officinale* (AOH + HT-2 + ENA1 + ENB) and *Lavandula intermedia* (AFG1 + ENB + AOH + TENT), three mycotoxins were present in samples of *Rosmarinus officinalis* (AFG2 + AOH + ZEN), and *Myrtus communis* (AFG1 + AOH + TENT)*,* while two mycotoxins (AOH + ZEN) cooccurred in *Origanum vulgare* and *Mentha spicata* samples (Figure 1). Finally, positive samples of *Artemisia absinthium* and *Matricaria chamomilla* were contaminated individually by AOH.

To the best of our knowledge, limited data have been published on the multi-presence of mycotoxin in aromatic and medicinal herbs available worldwide. Indeed, a recent investigation from Spain was performed to screen the multi-contamination by mycotoxins (AFs, OTA, ZEN, T-2, DON, citrinin, and fumonisins) in 84 samples of aromatic and/or medicinal herbs, showing that 99% of the samples were contaminated with T-2 (99%), ZEN (98%), AFs (96%), OTA (63%), DON (62%), citrinin (61%), and fumonisins (13%) [37].

**Figure 1.** Co-occurrence mycotoxin distribution.

#### *2.4. Conjugated Mycotoxins in AMP*

Samples with contamination levels of target mycotoxins were injected into the LC– QTOF–MS system to confirm their presence and to study the possible co-occurrence of lesser known non-target mycotoxin metabolites formed during detoxification and glycosylated and sulfated conjugates, since the probability of identifying these compounds was reasonably greater when more mycotoxins were present.

In the present work, the exact mass and isotope pattern calculated from the molecular formula and plus/minus the expected adduct(s) of the suspected substance and experimental information (retention time behavior and presence of related substances) were used to screen that substance in the samples. Afterwards, non-target components of modified and conjugated compounds were studied to gain confidence through library match and/or diagnostic fragments.

All target mycotoxins previously quantified by LC–MS/MS were confirmed by LC– QTOF–MS (Table 4). However, the results of the study of non-target components derived from metabolism were only presented for ZEN. Although AOH was the most detected mycotoxin, ZEN was the second mycotoxin highly present in Moroccan AMP samples, and more susceptible to suffering from glycosylation, sulphuration, and hydroxylation reactions than AOH, as reported in the literature in the last decade [7,11,38]. After automatic acquisition mode MS/MS, α-ZEN, β-ZEL, ZEN-4-Glc, ZEN-14-Glc, ZEN-16-Glc, α-ZEL-14-Glc, β-ZEL-14-Glc, and ZEL-4-Sulf were selected to be identified by MassHunter Qualitative Analysis B.10.0. Mycotoxins or metabolites that have an available commercial standard (α-ZEL, β-ZEL, ZEN) were confirmed and quantified with the same program, and a "Find by Formula" data-mining algorithm with a mass error below ± 5 ppm with score values ≥ 70 (including isotope abundance and isotope spacing) was also used. A retention time window of ± 1 min was specified for peak detection to compensate for retention time shifts due to system-to-system variability. All relevant compound species including adducts: [M+H]+, [M+Na]+, [M−H]–, [M+HCOO]–, [M−OH]–, and [M+HCOOH−H]– were used as target masses. Mycotoxin metabolites' annotation was also supported by comparing the obtained MS/MS fragmentation spectra with the experimental spectra proposed in the databases of mycotoxins and related metabolites, Personal Compound Database and

Library (PCDL) Manager MassHunter. Metlin Metabolites PCDL MassHunter contains an accurate mass compound database, a collision cross section database, and an MS/MS accurate mass spectral library for mycotoxins.

In the present study, the ZEN conjugates and metabolites selected were α-ZEL, β-ZEL, ZEN-14-Glc, β-ZEL-14-Glc, and ZEN-4-Sulf. All were selected based on purity score values ≥ 70 and a mass error < ± 5 ppm. Tentative identification is listed in Table 5. β-ZEL was the most abundant metabolite found (22%). The glycosylated ZEN (ZEN-14-Glc) compound was detected in five samples (11%), with low intensity. Regarding ZEN sulfate conjugate (ZEN-4-Sulf), it was detected in four samples (9%). Previous studies have indicated the presence of ZEN-4-Sulf in fungal cell cultures in molar ratios from 1:12 to 1:1, compared with ZEN [14], and in wheat flour samples at levels of 9.7% ZEN-4-Sulf [38]. Here, in AMP samples, ZEN-4-Sulf was present at levels ranging from 5 to 2% of the ZEN level. As far as we know, this is the first time that these non-target metabolites have been reported in AMP samples, and the first time they have been identified by LC–QTOF–MS in AMP. QTOF-MS data cannot support quantification of compounds without the use of a reference compound so that the conjugated mycotoxins detected were not quantified, although their presence was positive. Supplementary Figure S1 shows a total ion current (TIC) chromatogram from a positive sample and its extracted ion chromatogram (EIC) of β-ZEL and ZEN-14-Glc detected. Their corresponding scan and product ion spectrum are included in Supplementary Figure S1.


**Table 5.** Identification of conjugated mycotoxins and ZEAs metabolites detected in AMP analyzed samples.

\* Tentative identification by Metlin Metabolites PCDL MassHunter (PCDL contains an accurate mass compound database, a collision cross section database, and an MS/MS accurate mass spectral library).

### *2.5. Dietary Exposure*

Several studies have reported the presence of mycotoxins in food and feed, while few data are still available of mycotoxins present in medicinal plants. The effect of mycotoxins found in some herbal plants on biochemical parameters of blood from mice were reported by Alwakeel (2009) [39]. The authors showed that analytical parameters, such as mean creatinine, urea, alanine aminotransferase, aspartate aminotransferase, and gamma-glutamyl

transpeptidase, were higher in the mice group fed or treated with herbal and fungal extracts than the control group, and the study confirms the implication of the AFs with induction of nephrotoxicity and hepatoxicity in animals.

No data is currently available on the annual consumption of AMPs in Morocco, and as for estimation, it is assumed that the annual AMP consumption is half of the annual raw tea consumption. According to FAO, the annual consumption per capita of green tea in Morocco averages 1.89 kg/year, so the annual consumption per capita of AMPs in Morocco is supposed to be 0.94 kg/year [40]. The risk of mycotoxins was assessed herein following both lower bound (LB) and upper bound (UB) approaches. For the LB approach, mean values were obtained by assigning a level of zero to free mycotoxin samples (where no mycotoxins were detected), or at levels below the LOQ (where mycotoxins were detected), whereas, in the UB approach, values equal to the LOD were assigned to samples where no mycotoxins were detected, and values equal to the LOQ were assigned to samples in which mycotoxin levels were below the LOQ.

In this study, mycotoxins AFG1, AFG2, ZEN, HT-2, AOH, TENT, ENA1, and ENB were detected in positive samples. For the AFs, these substances are confirmed as carcinogenic and classified by IARC in Group 1 (do not have an established TDI), so it is not possible to determine the threshold levels at which AFs have no effect [41]. It is recommended by JECFA, with regard to the safe level of AFs in foods, that AF levels must be reduced according to the "As Low As Reasonably Achievable" (ALARA) principle [42]. Furthermore, no TDI values have been established for emerging mycotoxins (BEA and ENs) and *Alternaria* toxins (AOH and TENT), so a risk assessment is not possible to calculate for these mycotoxins. For ZEN, the PDIs calculated were 0.67 ± 1.21 ng.kg−1.bw.day−<sup>1</sup> (LB approach) and 1.39 ± 1.20 ng.kg−1.bw.day−<sup>1</sup> (UB approach), and the TDIs (ZEN 250 ng.kg−1.bw.day−1) were 0.56% and 0.82% by the LB and UB approaches, respectively (Table 4).

Risk assessment shows that the intake of mycotoxins through the consumption of AMP beverages does not represent a risk for the population, except for AFs that are classified as carcinogenic compounds. Nevertheless, the presence of mycotoxins in AMPs could increase the exposure in large consumers. A focus on plants is relevant to gather more knowledge on larger spectra of mycotoxin contamination related with AMP handling conditions, their presence in extracted essential oils, the apparition of any toxic effect, or the effect on human health.

Another important point is that the conjugated metabolites, as well as their reductive forms, are not a part of ZEN´s regulations. In vitro analyses of the gastrointestinal digestive process showed no cleavage of ZEN´s conjugates, but in human microbiota fermentation, the conjugates were cleaved by the microbial enzymes [43,44]. Thus, ZEN uptake might be underestimated, due to the release of absorbable ZEN.

Recently, the EU-CONTAM Panel found it appropriate to set a group TDI for ZEN and its modified forms [45]. It must be considered that the estrogenic potency of ZEN derivatives differs. Potency factors assigned to these derivatives by the EFSA CONTAM Panel are 0.2 for β-ZEL and 60 for α-ZEL relative to ZEN. Moreover, for sulfate and glucoside conjugates, the same factors as the free forms are proposed. However, to obtain more data on the occurrence of ZEN metabolites in food and feed, standard compounds are needed.
