Deep Genotypic Species Delimitation of Aspergillus Section Flavi Isolated from Brazilian Foodstuffs and the Description of Aspergillus annui sp. nov. and Aspergillus saccharicola sp. nov.

Aspergillus section Flavi is a fungal group that is important in food because it contains spoilage and potentially aflatoxigenic species. Aflatoxins are metabolites that are harmful to human and animal health and have been recognized as the primary natural contaminant in food. Therefore, recognizing the biodiversity of this group in food is necessary to reduce risks to public health. Our study aimed to investigate the diversity of Aspergillus section Flavi isolated from Brazilian foodstuffs such as cassava, sugarcane, black pepper, paprika, Brazil nuts, yerba-mate, peanuts, rice, and corn. A polyphasic approach integrating phenotypic data and multilocus genotypic analyses (CaM, BenA, and RPB2) was performed for 396 strains. Two new species in the Aspergillus subgenus Circumdati section Flavi are proposed using maximum-likelihood analysis, Bayesian inference, and coalescence-based methods: Aspergillus saccharicola sp. nov. and Aspergillus annui sp. nov. A. saccharicola sp. nov. belongs to the series Flavi, is a potentially aflatoxigenic species (B1, B2, G1, and G2), closely related to Aspergillus arachidicola, and was found mostly in sugarcane. A. annui sp. nov. was isolated from samples of sweet paprika. To accommodate A. annui sp. nov., a new series Annuorum was proposed.


Introduction
Mycotoxins are secondary metabolites produced by a variety of fungal species that colonize different crops around the world. More than 400 mycotoxins have been described, but only a few have relevance as food contaminants [1]. In addition to the risk to public health, mycotoxin contamination causes losses that can be estimated in billions of dollars [2]. According to the annual report of the Rapid Alert System for Food and Feed (RASFF), mycotoxins were the main hazard in notifications of border rejection in the European Union in 2020, with nuts, nut products, seeds, fruits, and vegetables accounting for more than 75% of notifications related to mycotoxins [3].
The strains were purified by monosporic culture and cultivated in yeast extract sucrose medium (liquid) at 25 • C for 3 days until the formation of a mycelial film was formed. Later, the mycelia were removed from the culture medium and macerated with liquid nitrogen. The NucleoSpin Microbial DNA kit (Macherey-Nagel, Düren, NRW, Germany) was used to extract genomic DNA according to the protocol recommended by the manufacturer. DNA was quantified spectrophotometrically using NanoDropTM 2000/2000c (Thermo Fisher Scientific, Pittsburgh, PA, USA).
Three-loci amplification was performed via PCR; part of the gene encoding calmodulin (CaM) was amplified using the CMD5 and CMD6 [32] or CF1 and CF4 [33] primer pairs; Bt2a and Bt2b primer pair [34] were used for amplification of part of the beta-tubulin gene (BenA), and finally, for amplification of the gene of the second largest subunit of RNA polymerase II (RPB2), the 5Feur and 7CReur primer pair [35] was used. Conditions for amplification were as described by Silva et al. [36].
The amplification products were separated using agarose gel (1% w/v) electrophoresis, stained with ethidium bromide, and photographed under ultraviolet light. The PCR products were purified with the ExoSAP-IT™ PCR Product Cleanup reagent (Thermo Fisher Scientific, Santa Clara, CA, USA) after amplification. The PCR fragments were subjected to direct sequencing using the method described by Sanger et al. [37]. Amplicons were sequenced in both directions (forward and reverse) using a BigDye ® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) in a SeqStudio Genetic Analyzer ® (Applied Biosystems, Waltham, MA, USA).

Phylogenetic Analysis
The partial sequences of the genes BenA, CaM, and RPB2 were used to analyze the diversity of Aspergillus section Flavi isolates. ClustalW was employed to align the sequences [38] in the BioEdit Sequence Alignment Editor v.7.1.3.0 software [39]. The three loci were concatenated for each of the 396 strains, and haplotype analysis was performed using the DnaSP v.6 DNA Sequence Polymorphism program [40].
Furthermore, maximum-likelihood (ML) trees were constructed using a representative of each of the characterized haplotypes to identify the strains. Phylogenetic trees were constructed for each of the loci separately, and for the combined dataset (CaM+BenA+RPB2). The sequences were aligned with those of type or neotype strains of species formally accepted in the Aspergillus section Flavi, using ClustalW.
For the construction of ML trees, the best nucleotide substitution model was calculated in the jModelTest2 [41], based on the Akaike Information Criterion (AIC). The ML trees were built in the MEGA 11 program [42], with 1000 bootstraps of replicates.
Additionally, descriptive DNA parameters such as nucleotide diversity (π), number of variables sites, and parsimony-informative sites were measured for each of the three loci individually and for the combined dataset using the DnaSP v.6 DNA Sequence Polymorphism program.
All intra-and interspecific variability (based on haplotypes) discovered in this study were deposited in GenBank, and accession numbers can be found in the Supplementary Material (Table S1).

Coalescence Analysis
For the coalescence analysis, in addition to the ML trees, we also performed Bayesian inference (BI) analysis for each of the three loci individually. The BI analysis was conducted in MrBayes 3.2.3. The most suitable nucleotide substitution model for each dataset was selected based on the lowest Bayesian Information Criterion (BIC) value in jModeltest2. For BI, the Markov Chain Monte Carlo (MCMC) algorithm was run for 5 × 10 6 generations with a sample frequency of 100 and with 25% of trees removed for burn-in. Convergence diagnostics were monitored based on standard deviations of frequencies below 0.01.
All of the ML and BI gene trees generated with their bootstraps and posterior probabilities scores were used as the input for the program Astral-II [43] in the CIPRES science gateway [44]. The Astral analysis was performed with the full annotation mode, greedy resolution, and 10 6 replicates for each locus. All other parameters were set to default values. The species tree was visualized using MEGA 11.

Morphological Analysis
Morphological analyses were performed according to the recommendations of Samson et al. [29]. For macromorphological observations, representatives of candidate species (groups 1 and 2) were inoculated at three points on Czapek Yeast Autolysate agar (CYA), malt extract agar (MEA), yeast extract agar Sucrose Agar (YESA), and incubated for 7 days at 25 • C in the dark. Furthermore, the strains were tested in CYA medium for 7 days at temperatures of 37 • C and 42 • C. The experiments were performed in triplicate.
Microscopic mounts were made in lactic acid from colonies of MEA (7 days at 25 • C) for micromorphological analysis (optical microscopy). The size of the microstructures: conidiophores, stipes, vesicles, conidia, metulae, and phialides were measured using the software Carl Zeiss™ AxioVision Release 4.8.2.

Secondary Metabolite and Mycotoxin Analysis
Secondary metabolites and mycotoxins were detected in three isolates from each of the new taxa. The six isolates were three points inoculated on CYA, YESA, WATM (Wickerham's Antibiotic Test Medium) [45], and Aspergillus flavus parasiticus agar (AFPA) [46] and incubated at 25 • C for 7 days in the dark. One extract was made from three agar plugs each of CYA and YESA, and another extract was made from three agar plugs of WATM and three agar plugs of AFPA. The extraction solvent used was ethyl acetate isopropanol (3:1, vol/vol) containing 1% formic acid. The extraction procedure and HPLC (high performance liquid chromatography) analysis was carried out as described by Nielsen et al. [47]. The extracts were subjected to HPLC-DAD and HPLC-DAD-MS-MS (high-performance liquid chromatography-diode array detection-mass spectrometry-mass spectrometry), as described by Nielsen et al. [47] and Wang et al. [48]. The identity was verified by comparing the results to authentic standards (retention time, retention index, UV spectra, and mass spectra).
Ultra-high-performance liquid chromatography-diode array detection-quadrupole time-of-flight mass spectrometry (UHPLC-DAD-QTOFMS) was performed on an Agilent Infinity 1290 UHPLC system (Agilent Technologies, Santa Clara, CA, USA) equipped with a diode array detector. Separation was achieved on a 150 mm × 2.1 mm i.d., 1.9 µm, Poroshell 120 Phenyl Hexyl column (Agilent Technologies, Santa Clara, CA, USA) held at 40 • C. The sample, 2 µL, was eluted at a flow rate of 0.35 mL min −1 using a linear gradient from 10% acetonitrile (LC-MS grade) in Milli-Q water buffered with 20 mM formic acid increasing to 100% in 15 min, staying there for 2 min before returning to 10% in 0.1 min. Starting conditions were held for 2 min before the following run.
Mass spectrometry (MS) detection was performed on an Agilent 6545 QTOF MS equipped with Agilent Dual Jet Stream electrospray ion source (ESI) with a drying gas temperature of 250 • C, a gas flow of 8 L min −1 , sheath gas temperature of 300 • C, and flow of 12 L min −1 . Capillary voltage was set to 4000 V and nozzle voltage to 500 V in positive mode. MS spectra were recorded as centroid data, at an m/z of 100-1700, and auto MS/HRMS fragmentation was performed at three collision energies (10, 20, and 40 eV) on the three most intense precursor peaks per cycle. The acquisition rate was 10 spectra s-1. Data were handled using Agilent MassHunter Qualitative Analysis software (Agilent Technologies, Santa Clara, CA, USA). Lock mass solution in 70% MeOH in water was infused in the second sprayer using an extra LC pump at a flow of 15 µL/min using a 1:100 splitter. The solution contained 1 µM tributylamine (Sigma-Aldrich) and 10 µM Hexakis (2, 2, 3, 3-tetrafluoropropoxy) phosphazene (Apollo Scientific Ltd., Cheshire, UK) as lock masses. The [M + H]+ ions (m/z 186.2216 and 922.0098, respectively) of both compounds were used.
An in-house fungal metabolite library search was carried out as described by Kildgaard et al. [49]. Data files were processed in MassHunter workstation B.07.00 with "Find by Auto MS/MS function" with a processing limit to 200 largest peaks and mass match tolerance m/z 0.05. The HRMS/MS library search was performed using a parent and fragment ion accuracy of 20 ppm + 2 mDa, with a minimal forward score of 50 and reverse score of 80.

Genotypic Analysis
Primarily, it is important to contextualize the taxonomic status of Aspergillus section Flavi; Frisvad et al. [9] carried out the most recent major review of the group, with the authors including seven new species: Aspergillus aflatoxiformans, Aspergillus aspearensis, Aspergillus austwickii, Aspergillus neoalliaceus, Aspergillus subflavus, Aspergillus pipericola, and Aspergillus vandermerwei. The authors also synonymized Aspergillus parvisclerotigenus and Aspergillus subolivaceus as A. flavus and Aspergillus albertensis as A. alliaceus. The other significant points were the renaming of Aspergillus korhogoensis to Aspergillus cerealis and Aspergillus bombycis to Aspergillus luteovirescens. Concluding the review, Aspergillus section Flavi harbored 33 species.
The taxonomy of this group has been dynamic, and the popularization of multiloci analysis associated with genealogical concordance has increased the number of descriptions of new species. After the review by Frisvad et al. [9], six new species of Aspergillus section Flavi have been proposed-Aspergillus texensis [50], Aspergillus agricola and Aspergillus toxicus [51], Aspergillus krugeri and Aspergillus magaliesburgensis [52], and more recently, Aspergillus burnetii [53].
The three loci analyzed here (CaM, BenA, and RPB2) are considered secondary barcodes of Aspergillus, as the official fungal barcode (ITS region) does not have discriminatory power for many Aspergillus species [29]. Based on the combined dataset (CaM+BenA+RPB2), 48 haplotypes were characterized; the distribution of haplotypes and the origin of the strains are provided in the Supplementary Material (Table S1). The best nucleotide substitution model was calculated based on jModelTest2 and is presented in Table 1. The RPB2 locus had the highest number of parsimony-informative sites; however, it also had the lowest nucleotide diversity ( Table 1). The CaM locus had the most polymorphic sites and the highest nucleotide diversity, whereas the BenA locus had the lowest values of polymorphic sites and parsimony-informative sites ( Table 1). The three loci appeared to have a high potential for resolution, but the use of one or the other should be avoided in some cases. For instance, A. parasiticus and A. novoparasiticus are not well discriminated by the BenA locus [9]. Based on the tree topologies of CaM (Figure 1), BenA (Figure 2), and RPB2 (Figure 3) loci, the CaM locus appears to be the best at resolving the Aspergillus section Flavi species.
Aime et al. [54] recently published a detailed guide on the requirement to describe new fungal species according to the International Code of Nomenclature for algae, fungi, and plants (ICNafp); one of the points highlighted by the authors is the sampling, both analyzed strains and loci. The use of a single locus for the description of a new species is strongly discouraged, but in exceptional cases, it may be acceptable.
The use of multiple unlinked loci increases informativeness and decreases bias caused by combined selection pressure on a single region of the genome, resulting in more reliable evolutionary hypotheses [55,56]. Aime et al. [54] suggest that working with multiple loci is ideal, and using genealogical concordance in the delimitation of cryptic species should be used whenever possible.
More broadly, the concept of genealogical concordance has been used to support the description of many fungal species [9,36,[57][58][59]. However, we emphasize that the delimitation of phylogenetic species provided by the genealogical concordance must serve as taxonomic hypotheses, and therefore, must be tested with additional data through integrative taxonomy.
Another fundamental point for our study was the understanding of the concept of phylogenetic species, which is generally supported by the monophyly criterion, which states that species is defined as the smallest group of organisms that includes the common ancestor and all its descendants [60].
In an attempt to improve the recognition of phylogenetic species' boundaries, Taylor et al. [61] and Dettman et al. [62] developed a theoretical-operational protocol, which they called Genealogical Concordance Phylogenetic Species Recognition (GCPSR). This method employs the use of multiple independent genes, for which gene trees are constructed and compared separately. When using the GCPSR, the subjectivity of recognizing the limits of a species can be reduced, since it is sufficient to look for the point at which there was a transition from concordance between gene trees (representing the divergent phylogenetic relationships of the species tree) to the incongruity between gene trees (representing the cross-linked associations between individuals within a species). In other words, the intersection of conflicting topologies indicates gene flow between individuals below the level of species (intraspecies).
Additionally, the analysis must be conducted with a good sampling of strains, as the description of a new species based on a single strain is not recommended. The evaluation of multiple strains allows for a more comprehensive view of the genetic composition of the group under analysis [63]. We used genealogical concordance analysis to 1188 nucleotide sequences obtained from 396 strains isolated from 9 foodstuffs to investigate the context of Aspergillus section Flavi species in Brazilian foodstuffs, According to this analysis, as can be seen in Figures 1-4, the most common species is undoubtedly A. flavus, with 194 strains distributed in 17 haplotypes (H1, H2, H4, H9,  H10, H25, H26, H32, H34, H36, H37, H38, H42, H43, H45, H47, H48), followed by A. parasiticus with 63 individuals, distributed in 9 haplotypes (H6, H7, H8, H24, H28,  H29, H30, H40, H41), and together these two species represented 64.2% of the total set of strains. This is not a surprise as A. flavus and A. parasiticus are considered the most frequent species of Aspergillus section Flavi and the most important from the perspective of aflatoxin production [64]. Other species found were: A. novoparasiticus, A. arachidicola, A. tamarii, A. caelatus, A. pseudocaelatus, A. nomiae, and A. pseudonomiae (Figure 4).      The CaM ML tree (Figure 1) allowed for the allocation of all haplotypes together with species formally accepted in Aspergillus section Flavi, except for haplotypes H12 and H16 (closer to A. arachidicola); H46 (intermediate to ser. Coremiiformes and ser. Alliacei); and H25 and H26 (closer to A. flavus/oryzae).
Moreover, the formation of subgroups within the A. flavus/oryzae clade is also visible ( Figure 1); A. flavus/oryzae exhibits high intraspecific diversity, and the formation of these subgroups is common. Some of these subgroups correspond to the type strains of A. kambarensis (CBS 542.69) and A. subolivaceus (CBS 501.65), both of which are considered synonymous with A. flavus sensu stricto.
The BenA locus exhibits less intraspecific variation for members of the A. flavus/oryzae group, resulting in a less branched topology; haplotypes H25 and H26 were grouped together with type strains of A. flavus and A. oryzae (Figure 2).
The BenA ML tree ( Figure 2) did not allow for a good resolution of the species of the A. parasiticus group (A. arachidicola, A. parasiticus and A. novoparasiticus). As already mentioned, the BenA locus does not lend itself to discrimination between A. parasiticus and A. novoparasiticus [9]. The H12 and H16 haplotypes, again, did not fully group the A. arachidicola-type strain (CBS 117610).
The BenA locus also did not allow for the grouping of Haplotype H46 to any species of Aspergillus section Flavi ( Figure 2); again, a distinct clade was formed between ser. Coremiiformes and ser. Alliacei, but phylogenetically closer to ser. Alliacei.
As in the CaM and BenA ML trees, for the RPB2 locus (Figure 3), the haplotype H46 also formed a distinct clade from the others, more phylogenetically related to being ser. Alliacei. The same occurred in the concatenated ML tree (Group II) (Figure 4). This topological concordance at different loci evidences an independent evolutionary lineage [61,62], thus, H46 shows reciprocal monophyly in relation to all species formally accepted in Aspergillus section Flavi. Therefore, based on the concept of phylogenetic/genealogical species, we started to treat it as a new species in Aspergillus section Flavi, entitled Aspergillus annui sp. nov.
Similarly to what occurs for A. parasiticus and A. novoparsiticus with the locus BenA, using the RPB2 locus for identification or discrimination between A. nomiae and A. pseudonomiae should be avoided, as the difference between the aforementioned species for the RPB2 locus is only one snip, which is in a marginal portion, and is easily lost in alignments of large datasets.
Interestingly, the RPB2 ML tree genealogy showed discordance with CaM and BenA for the H12 and H16 haplotypes that clustered together with A. parasiticus/A. sojae ( Figure 3) and did not form a distinct group (closer to A. arachidicola), as observed in the other two loci (Figures 1 and 2) and on the concatenated ML tree (group I) (Figure 4).
Similarly, in the ser. Kitamyces, the haplotypes H11, H13, and H17 had variable taxonomic positions (Figures 1-3), clustering together with A. caelatus or A. pseudocaelatus, depending on the locus analyzed. This incongruence between gene trees is frequently used to indicate species delimitation, and it is even one of the bases of the GCPSR mentioned above (when caused by recombination). However, it can also be an indication of incomplete lineage sorting (ILS) [65,66].
ILS is an evolutionary phenomenon that transcends speciation events and occurs when ancestral gene sequences fail to coalesce, i.e., the persistence of ancestral polymorphisms across speciation events results in incompletely classified loci, which consequently may cluster closely related non-conspecifics, rather than separate them [67][68][69][70], resulting in hemiplasy [71].
The best method for distinguishing between species boundary-breaking signatures from those associated with ILS is to use methods based on the coalescent theory, which relies on the use of multiple gene trees to examine coalescence, thus, producing species trees; such methods have been used to overcome this type of difficulty in genealogical approaches to species delimitation [72][73][74][75][76][77] and has been successfully applied to fungi [78][79][80][81][82].
Therefore, we constructed species trees based on the coalescent model to elucidate the taxonomic position of haplotypes H12 and H16 (clade A. parasiticus) and H11, H13, and H17 (in the ser. Kitamyces). In addition, we also implemented this approach for the haplotype H46 (A. annui sp. nov.) as additional evidence.
As seen in Figure 5, the set of 22 strains of haplotypes H12 and H16 formed a distinct group of A. parasiticus, A. arachidicola and A. novoparasiticus; GI is statistically well supported. This result shows that in fact GI represents an independent evolutionary lineage, and should be considered a distinct phylogenetic species in Aspergillus section Flavi, hereinafter referred to as Aspergillus saccharicola sp. nov. Furthermore, we constructed ML trees based on other loci (ITS and niaD), which also corroborated these results (Supplementary Material Figure S1).  Regarding the GII group (H46), the species tree based on coalescence ( Figure 5) corroborates the phylogenetic analyses (Figures 1-4) and solidly supports the description of A. annui sp. nov.
In the coalescence analysis of the ser. Kitamyces, we evaluated 12 strains from five different substrates (maize, black pepper, sugarcane, peanuts, and rice). The haplotypes H11, H13, and H17, that showed incongruences in gene trees, formed a statistically wellsupported group (bootstrap 74%) with the type strain of A. pseudocaelatus ( Figure 5); however, there was still the formation of subgroups, but with little statistical support. This confirms the incongruence between the gene and species trees. It was demonstrated that the ILS can be a substantial phenomenon in the ser. Kitamyces.
Additionally, we used a dataset of reference sequences (Table S2) to calculate a genetic distance matrix to analyze the divergence between the candidate species of the present study, A. annui and A. saccharicola, with the species related to them ( Figure 6). The genetic distance between A. annui sp. nov. and the species of the series Alliacei and Coremiiformes was greater than the genetic distance between the species currently accepted in these groups, such as A. alliaceus vs. A. neoalliaceus (0.015) and A. togoensis vs. A. coremiiformis (0.081) ( Figure 6). Furthermore, the genetic distance of A. annui sp. nov. in relation to the individuals of the ser. Alliacei and/or ser. Coremiiformes was equivalent or greater to that found between the individuals of the ser. Alliacei vs. ser. Coremiiformes (Figure 6), which denotes the need to establish a new series to accommodate this new species, which was also noticeable in all the genealogies presented (Figures 1-5).
The species with the smallest genetic distance in relation to A. annui sp. nov. was A. magaliesburgensis (0.105) (Figure 6), a species recently described by Visagie et al. [52].
Considering the members of the Alliacei and Coremiiformes series, the smallest genetic distance found between the members of these groups was 0.003 ( Figure 6), which occurred in the comparison of A. alliaceus vs. A. burnetti; similar values were obtained in the comparison of A. parasiticus and A. sojae (0.002).
It is important to note that A. sojae is a non-toxigenic domesticated variant of A. parasiticus; the same relationship exists between A. oryzae and A. flavus [83][84][85][86][87]. These species are still classified as separate taxa for economic and food safety reasons, but not for taxonomic reasons. A. oryzae and A. sojae are widely used in the food industry, and are generally recognized as safe by the FDA [4,86].
The genetic distance of A. saccharicola sp. nov. and the closest species (A. arachidicola), The genetic distance between A. annui sp. nov. and the species of the series Alliacei and Coremiiformes was greater than the genetic distance between the species currently accepted in these groups, such as A. alliaceus vs. A. neoalliaceus (0.015) and A. togoensis vs. A. coremiiformis (0.081) ( Figure 6). Furthermore, the genetic distance of A. annui sp. nov. in relation to the individuals of the ser. Alliacei and/or ser. Coremiiformes was equivalent or greater to that found between the individuals of the ser. Alliacei vs. ser. Coremiiformes (Figure 6), which denotes the need to establish a new series to accommodate this new species, which was also noticeable in all the genealogies presented (Figures 1-5).
The species with the smallest genetic distance in relation to A. annui sp. nov. was A. magaliesburgensis (0.105) (Figure 6), a species recently described by Visagie et al. [52].
Considering the members of the Alliacei and Coremiiformes series, the smallest genetic distance found between the members of these groups was 0.003 ( Figure 6), which occurred in the comparison of A. alliaceus vs. A. burnetti; similar values were obtained in the comparison of A. parasiticus and A. sojae (0.002).
It is important to note that A. sojae is a non-toxigenic domesticated variant of A. parasiticus; the same relationship exists between A. oryzae and A. flavus [83][84][85][86][87]. These species are still classified as separate taxa for economic and food safety reasons, but not for taxonomic reasons. A. oryzae and A. sojae are widely used in the food industry, and are generally recognized as safe by the FDA [4,86].
The genetic distance of A. saccharicola sp. nov. and the closest species (A. arachidicola), was also equivalent to that found among other species of the ser. Flavi, such as A. parasicitus vs. A. novoparasiticus (0.005) ( Figure 6).
All these results strongly support the description of A. annui sp. nov. and A. saccharicola sp. nov. based on the concept of phylogenetic/genealogical species; however, as recommended by the ICNafp, we also performed phenotypic analysis.

Secondary Metabolites Analysis
Aspergillus annui sp. nov. did not produce any mycotoxins on the four media CYA, YESA, WATM, and AFPA, but the metabolites ergokonin B, glycocholic acid, flavin, phytosphingosin, and sphinganin were detected ( Table 2). A. annui sp. nov. was found to produce the antifungal ergokonin B, which is also produced by Trichoderma sp. Xy24 and a Fusarium species [88,89]. Ochratoxin A, which is produced by most species in the Alliacei series, was not detected in A. annui sp. nov.
Moreover, even though sphinganine and phytospingosin are considered as primary metabolites, they may be involved in the biosynthesis of the secondary metabolites flavucides, which are antibacterial cerebrosides known from A. flavus [90]. Additionally, flavin is structurally related to the fluorescent secondary metabolites asperopterin A and B, which were isolated from a strain identified as A. oryzae [91,92].
The high-resolution MS/MS spectrum of dereplicated metabolites can be found in the Supplementary Material (Table S3).

Morphological Analysis
To support the description of these new taxa, detailed morphological descriptions and illustrations are provided below.  Figure 7.
Mycobank: MB845969 Etymology: The specific epithet refers to the substrate from which it was isolated, paprika pepper, Capsicum annuum.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/jof8121279/s1, Figure S1: Maximum-likelihood tree of Aspergillus parasiticus clade based on combined dataset sequences (ITS+niaD); Table S1: Description of the samples from this study. Strains, substrate of origin, haplotypes and GenBank accession numbers; Table S2: Dataset for genetic distance analysis; Table S3: The high-resolution MS/MS spectrum of dereplicated metabolites of Aspergillus annui sp. nov. and Aspergillus saccharicola sp. nov.

Institutional Review Board Statement: Not applicable.
Data Availability Statement: The sequences newly generated in this study have been submitted to the GenBank database.