Introduction

The Brazil nut (Bertholletia excelsa) is a monotypic South American tree species of the Amazon Rain forests, native to Brazil, Peru, Bolivia, Guianas, Venezuela, and Columbia. It occurs as scattered trees in large forests along the banks of large river systems. Tree exploitation is a source of socio-economic and environmental impacts for the region. Activity generates employment and income for local people; at the same time, it inhibits the deforestation (see review Freitas-Silva and Venâncio 2011).

The Brazil nut is a tall tree, reaching heights of up to 50 m, one of the largest of the Amazon forest. The lower limits documented for the species are 30–45 m. The fruit is a spherical capsule of ligneous mesocarp, and is extremely hard. Seeds are found in average numbers of 18, with a ligneous and rough shell. The nut (i.e., seed) is recognized by its high nutritional value. The fruits are collected after falling, and therefore are in contact with the ground where nuts are exposed to fungal infection (Souza et al. 2004).

Among the main problems identified in Brazil nut production is contamination by fungi that produce aflatoxins, primarily by species of Aspergillus section Flavi. Aflatoxins are toxic secondary metabolites produced by some fungal species when humidity and temperature are favorable. The occurrence of aflatoxins in Brazil nuts is a critical constraint for its commercialization, which has been cited in several reports (Pohland 1993; FSA 2004; Marklinder et al. 2005; Arrus et al. 2005; CAC 2005; Pacheco and Scussel 2007; Olsen et al. 2008; Pacheco and Scussel 2009; Freitas-Silva and Venâncio 2011).

Producing countries (Bolivia, Peru and Brazil) have implemented restrictions to avoid aflatoxin contamination in an attempt to meet international sanitary standards and strengthen export markets. For example, “Safenut”, a project coordinated by the European Union served to validate an aflatoxin level analysis method that is reliable and easy for industrialists to use for reducing and controlling nut aflatoxin contamination. The project completion in 2008 resulted in several recommendations; among them, the necessity for studies on the microbiota to better understand the infection process, and subsequent aflatoxin contamination (STDF 2008).

Aspergillus section Flavi taxonomy is complex and it is continually evolving. Usually, fungi of this section are subdivided into two classes based on their effects to food and human health (see Godet and Munaut 2010) i.e., aflatoxin producing and non-producing species. Aflatoxin producing species include A. flavus, A. parasiticus (Codner et al. 1963), A. nomius (Kurtzman et al. 1987), A. pseudotamarii (Ito et al. 2001), A. bombycis (Peterson et al. 2001), A. parvisclerotigenus (Saito and Tsurota 1993) A. arachidicola, and A. minisclerotigenes (Pildain et al. 2008). The first three species are the most damaging to foods such as rice, wheat, nuts, peanuts, and peppers (Geiser et al. 1998). The non-aflatoxin producing species are A. sojae, A. oryzae, A. tamarii, A. caelatus. A. oryzae and A. sojae are economically important species that are used in the preparation of fermented food products. At the molecular level, A. oryzae and A. sojae are closely related to A. flavus and A. parasiticus, respectively, but are maintained as separate species to avoid regulatory problems in the food industry (Geiser et al. 1998). Furthermore, recent studies have demonstrated that many previously described species in Aspergillus section Flavi are not supported by currently available data. Varga et al. (2011) used sequence data from three loci to clarify the taxonomy of this section. The phylogenetic analysis based on calmodulin and ITS sequence data encourage authors to included seven main clades with 20 or more taxa in the Aspergillus section Flavi. The main clades isolates form well-defined subclades on the trees, however, mostly are represented by a single isolate and further collections and studies are needed to clarify if they represent separated species.

Amongst the different types of aflatoxins (AF), the most important are aflatoxin B1, B2, G1, and G2. All these important aflatoxins have been detected in Brazil nuts (Olsen et al. 2008). A. flavus produces only AFB1 and AFB2, while the other aflatoxigenic species additionally produce AFG1 and AFG2. As reviewed by Freitas-Silva and Venancio (2011) there is a controversy concerning which species contribute greatest to aflatoxin production in Brazil nut. Aspergillus flavus has been described as dominant in Brazil nuts (Freire et al. 2000, Arrus et al. 2005), however this species are not able to produce AFG1 and AFG2. According to Olsen et al. (2008), Aspergillus flavus is not the main aflatoxin producer in contaminated Brazil nuts. Authors obtained evidences that A. nomius is an important producer, and recommended further studies to inspect for the presence of other B and G aflatoxin producers. The supposition that supports this controversy is the imprecise recognition of the fungal species (Freitas-Silva and Venancio 2011).

Since the identification of aflatoxin producer species may influence strategies for prevention and control of aflatoxins in Brazil nuts, the objectives of this study were to isolate and identify, by morphological and molecular data, fungal isolates from this important product extracted from the Amazonian region.

Materials and methods

Fungi isolation and identification by morphological analysis

A total of 20 samples from approximately 2 kg of Brazil nuts were collected in the Brazilian Amazon region. From each sample, approximately 100 g of shelled nuts (i.e., seeds) and 100 g of shells were disinfected separately by immersion in 0.4% sodium hypochlorite solution for 1 min. Fifty pieces of nuts or shells were plated separately onto Dichloran Glycerol (DG-18) agar containing chloramphenicol. Samples were incubated for 5 days at 25 °C. All Aspergillus section Flavi were isolated on Czapek yeast extract (CYA) agar, and incubated at 25 °C, 37 °C or 42 °C for 7 days. Identification was carried out according to Klich and Pitt (1988), and Pitt and Hocking (2009), based on macroscopic characteristics, such as diameter size, colony colour and texture, colour of colony reverse, presence of exudates and soluble pigment as well as microscopic characteristics, such as shape and size of vesicles and conidia, thickness of stipes and presence of phialides and metulas. For this evaluation, 40× and 100× microscopic magnifications were used.

Aflatoxin production

Isolates were evaluated for aflatoxin production following the protocol of Filtenborg et al. (1983). Briefly, the isolates were inoculated onto yeast extract agar containing 15% sucrose, and incubated at 25 °C for 7 days. A small piece of each colony was cut, and used for toxin extraction with chloroform:methanol (1:1). Plugs were placed on thin layer chromatography (TLC) plates (silica gel-G 500 mm thick), which were developed in toluene: ethyl acetate: 90% formic acid: chloroform (7:5:2:5), and visualized under UV light at 356 and 254 nm. Aflatoxin B1, B2, G1, and G2 standards (Sigma Chemical, St Louis, USA) were used for qualitative comparison of retention time and fluorescence.

DNA extraction

Isolates were cultivated in yeast extract and lactose (YEL) solid medium for 7 days. From this culture, we performed a suspension of approximately 107 conidia in 2.5 mL of Tween 80. The suspension was inoculated into bottles containing 50 mL of YEL liquid, and incubated in a shaker (180 rpm) at 28 °C for 16 to 24 h. After incubation, mycelia were collected by vacuum filtration and washed in sterile water. Nucleic acids were extracted according to Azevedo et al. (2000), and treated with ribonuclease A.

RAPD analysis

DNA amplification was performed using the arbitrary primers OPX 3, OPX 7, and OPX 11 (Invitrogen ®) in a PTC-100 thermocycler (MJ Research, Inc.) according to Fungaro et al. (1996). Isolate genotyping was conducted based on analysis of polymorphic loci generated by RAPD markers. A similarity matrix using Jaccard’s coefficient was constructed. Cluster analysis was performed using UPGMA (Unweighted Pair Group Method with Arithmetic Mean) with the SAHN module (Sequential, Agglomerative, Hierarchic, Nonoverlapping Clustering Methods) in NTSYS-PC (Numerical Taxonomy and Multivariate Analysis System) (Rohlf 2000). An Analysis of Molecular Variance (AMOVA) was conducted using the software Arlequin 3.0 (Excoffier et al. 2005) to examine divergence levels between groups evidenced by RAPD markers.

Partial amplification of the β-tubulin and calmodulin coding genes

Amplification of a β-tubulin gene region was performed using the following primer pair: Bt2a (5′ GGT AAC CAA ATC GGT GCT GCT TTC 3′) and Bt2b (5′ ACC CTC AGT GTA GTG ACC CTT GGC 3′), as described by Glass and Donaldson (1995). Similarly, a calmodulin gene region was amplified using the cmd5 (5′ CCG AGT ACA AGG AGG CCT TC 3′) and cmd6 (5′ CCG ATA GAG GTC ATA ACG TGG 3′) primers previously reported by Hong et al. (2006). Standard amplification reactions and cycling protocols were adopted, and amplicons submitted to direct sequencing in both directions (forward and reverse) in a MegaBaceTM 1,000 Molecular Dynamics system (Amersham, Pharmacia Biotech). The quality of the sequences was examined using the software package Phred/Phrap/Consed. The sequences obtained were aligned using Aspergillus section Flavi type species sequences deposited in the NCBI database (http://www.ncbi.nlm.nih.gov/) and MycoBank (http://www.mycobank.org). Editing and sequence alignment were performed using ClustalW (Thompson et al. 1994). The computer program package MEGA (Tamura et al. 2007) was used for building a phylogenetic tree based on the neighbor-joining method.

Results and discussion

Results from 241 isolates in Aspergillus section Flavi indicated 99 (41%) positive for aflatoxin production, with respective 72 (29.8%) and 27 (11.2%) B and G group aflatoxin production. Based on the differences found by means of macro- and microscopic characteristics, the 241 isolates were divided in 11 groups. Representatives of each morphological group (40 from nuts and 41 from shells) were selected to identify or confirm isolate taxonomy by means of molecular methods (Table 1). Figure 1 shows RAPD profiles for 41 isolates generated from nuts, and 40 from shells. One hundred and two RAPD loci were detected from nut isolates, which resulted in a dendrogram constructed with high confidence values, and isolates clustered into six groups. Among the groups, three (Groups II, IV, V) were aflatoxin non-producer isolates, two groups (Groups III, VI) were aflatoxin producers, and the major group (Group I) contained both types of isolates (aflatoxin non-producers and producers). Shell isolates generated 143 RAPD loci clustered into seven groups. Similarly, three groups (Groups II, IV, V) were comprised of aflatoxin non-producer isolates, three (Groups III, VI, VII) were aflatoxin producers, and the other group (Group I) had both types of isolates (aflatoxin non-producer and producer).

Table 1 Aspergillus section Flavi isolates obtained from nuts and shells of Brazil nuts collected from different States of the Amazonian region
Full size table
Fig. 1
figure 1

RAPD profiles from Aspergillus section Flavi isolates collected from Brazil nuts. a and b Isolates obtained from nuts and shells, respectively. MM molecular weight marker 1 kb plus; NC negative control. Codes followed by + and − correspond to aflatoxin producers and non-producers

Full size image

AMOVA performed with RAPD marker data obtained from nut samples indicated the main source of variation was between (FST = 0.81), rather than within groups (FST = 0 19). The pairwise FST values ranged from 0.77 to 1.0. The largest difference was detected between Group V (ITAL 116 isolate) and Group VI (ITAL 89) (Table 2a). In addition, the main source of variation based on shell sample isolates was between groups (FST = 0.83). Similarly, the pairwise FST values ranged from 0.79 to 1.0. The largest divergence was detected between Group VI (ITAL 245 isolate) and Group VII (ITAL 189), followed by Group V (ITAL 259 and ITAL 262) and Group VI (ITAL 246) (Table 2b).

Table 2 Pairwise FST values based on RAPD data among Aspergillus section Flavi isolate groups collected from Brazil nuts
Full size table

The high genetic distance values (not shown) and FST estimates suggest our isolate collection contains different species of Aspergillus section Flavi.

As the main source of genetic variation was between, rather than within RAPD groups, representatives from each group underwent nucleotide sequence analysis of β-tubulin and calmodulin coding genes. Compared to public databases using the BlastN algorithm, for most β-tubulin and calmodulin sequences we observed very low E-values and higher than 99% identity percentages with Aspergillus species described in the literature. However, for ITAL 116 (Group V collected from nuts), ITAL 262 and ITAL 259 (Group V collected from shells) isolates, the β-tubulin nucleotide sequences showed 89% maximum identity with those deposited in databases (A. tamarii and A. caelatus). Similarly, for the calmodulin gene, sequence analysis generated a maximum identity of 91% with A. caelatus sequences. The phylogenetic trees reconstructed from the comparisons of nucleotide sequences of our isolates and those of each type species of Aspergillus section Flavi are shown in Figs. 2 and 3.

Fig. 2
figure 2

Phylogenetic tree reconstructed from the β-tubulin gene (Bt2) sequences aligned with corresponding sequences of Aspergillus section Flavi type species deposited in public databases

Full size image
Fig. 3
figure 3

Phylogenetic tree reconstructed from the calmodulin gene sequences aligned with corresponding sequences Aspergillus section Flavi type species deposited in public databases

Full size image

Although RAPD data provided a random sample of the genetic variation dispersed throughout the genome, and gene sequence data offered information from genes recognized as suitable for distinguishing species in the Aspergillus section Flavi, the results here obtained from these two approaches were exactly in agreement. Isolates collected even from nuts or shells were clustered into approximately the same number of groups (6 or 7), independently if based on RAPD profiles or sequences data.

Among the 41 nut isolates, we identified the following species: A. flavus, A. caelatus, A. nomius, A. tamarii, and A. bombycis, corresponding to RAPD groups I, II, III, IV, and VI. Among the 40 shell isolates, we identified A. flavus, A. caelatus, A. nomius, A. tamarii, A. bombycis and A. arachidicola, corresponding to RAPD groups I, II, III, IV, VI, and VII (Table 1).

According to the literature, A. nomius and A. flavus were the only two aflatoxigenic species isolated from Brazil nuts. These species have been described as occurring in shell and shelled Brazil nuts. (De Mello and Scussel 2007, Olsen et al. 2008). In the present study, we identified in addition to A. nomius and A. flavus, two other aflatoxigenic species: A. bombycis and A. arachidicola. One of them, A. bombycis was found in both nuts and shells; however, A. arachidicola was found only in the shells. A. arachidicola was described in Agentinean peanuts and can produce both B and G-type aflatoxins (Pildan et al. 2008). A. bombycis was discovered by Peterson et al. (2001) during fungi isolation from insect frass collected in silkworm rearing houses in Japan, and they can also produce B and G-type aflatoxins. Among the four aflatoxin producing species identified in this study, A. nomius causes the greatest concern because of its frequency in Brazil nuts (23%), and all isolates produced aflatoxins B1 and G1.

The isolate denoted ITAL 116 collected from nuts (corresponding to Group V RAPD data), ITAL 262 and ITAL 259 collected from shells (corresponding to Group V RAPD data) formed a group phylogenetically distant from other Aspergillus species (Figs. 2, 3). Interestingly, a high number of nucleotide polymorphisms were detected on the β-tubulin gene Bt2 region of the three isolates relative to the type species presumed phylogenetically closest to the isolates i.e., A. tamarii and A. caelatus. A comparison with A. tamarii revealed 50 substitutions and three insertions, and 51 substitutions and two insertions were detected in A. caelatus (Fig. 4a). Sequence analysis of the calmodulin gene revealed 48 substitutions and one deletion compared to A. tamarii. Relative to A. caelatus, 38 substitutions were detected (Fig. 4b). Single nucleotide polymorphisms (SNP) were not found in β-tubulin and calmodulin genes (Fig. 4) of the ITAL 116, ITAL 262, and ITAL 259 isolates.

Fig. 4
figure 4

Alignment of nucleotide sequences of the β-tubulin gene (Bt2) (a) and calmodulin gene (b) of the ITAL 116, ITAL 259, and ITAL 262 isolates with corresponding Aspergillus tamarii and A. caelatus sequences

Full size image

The high level of nucleotide variation within the β-tubulin and calmodulin sequences in the ITAL 116, ITAL 262, and ITAL 259 isolates compared with the type species sequences deposited in public databases, and the high genetic divergence revealed through RAPD data suggests these three isolates from Brazil nuts represents a novel species not yet described in the literature.