Description of the sampled locations

Zempoala, Morelos (fresh water, 19°01′20″ N, 99°16′20″ W). The Zempoala Lagoons comprises 7 pristine water bodies located in a protected park in the state of Morelos at 2670-3686 masl. The area is surrounded by a temperate forest of pines, firs and oaks. Samples were obtained from one of the permanent lagoons.

Carboneras, Tamaulipas (brackish water, 24°37′41.88″ N, 97°42′59.19″ W). Fishery and leisure town located at 58 km from San Fernando in the state of Tamaulipas. It belongs to the central section of Laguna Madre.

Mezquital, Tamaulipas (sea water, 25°14′55.70″ N, 97°31′05.54″ W). Located in the eastern side of the wider part of Laguna Madre, it is connected to the Gulf of Mexico through an artificial navigation channel.

Media Luna, Tamaulipas (brackish water, 25°09′47.64″ N, 97°40′16.35″ W). Located at the western side of the wider part of Laguna Madre, and due to poor road conditions and to the absence of large settlements, this is one of the less spoiled areas. The distance between the Mezquital and Media Luna sampling places is approximately 12 km.

El Rabón, Tamaulipas (hypersaline sea water, 25°26′23.68″ N, 97°24′34.79″ W). At the northern end of Laguna Madre, this area has suffered serious transformations due to human activity and had become dry. Recently, the wetlands have been restored by pumping in sea water.

Carpintero. Tamaulipas (fresh water, 22°14′01.12″ N, 97°51′20.67″ W). Belonging to the Pánuco River basin and located in a protected natural park covering 7 ha, Carpintero Lagoon is currently used for fish and crocodile breeding grounds. In spite of the urban location of the site in the City of Tampico it is relatively unspoiled.

Vicente Guerrero, Tamaulipas (fresh water, 24°03′43.70″ N, 98°44′13.44″ W). This water body is a dam built in 1971 in the Soto La Marina River basin. It has a surface of 22.1 km² at 134 m above sea level. With an average volume of 500 million m³ it is used for aquatic sports and sport fishing.

Bahía de Banderas, Jalisco (sea water, 20°38′58.93″ N, 105°24′51.94″ W). Located in the border between the states of Nayarit and Jalisco in the Pacific Coast, it is the largest bay in Mexico with a surface of 773 km². The Ameca River mouth divides the bay, which has a high population density based on Puerto Vallarta.

Cruz de Huanacaxtle, Nayarit (sea water, 20°44′12.96″ N, 105°23′17.53″ W). A coastal site with low anthropogenic impact located in the north end of Bahía de Banderas in the state of Nayarit. The distance between the Bahía de Banderas and Cruz de Huanacaxtle sampled sites is approximately 15 km.

Zacapulco, Chiapas (mangrove swamp water, 15°04′07″ N, 92°45′20″ W). Unspoiled mangrove area located 200 km northwest of the border of the state of Chiapas with Guatemala on the Pacific Ocean coastline. At 14 masl, the estuarine salinity oscillates with the tides. The perennial vegetation provides coverage for the habitat of many aquatic birds. There is no seasonal daytime change and temperatures vary between 25 and 35 °C.

Santa Catarina, Querétaro (fresh water, 20°47′30.6″ N, 100°27′01.66″ W). An artificial water reservoir located 25 km northwest of the city of Querétaro in the state of Querétaro at 2035 masl.

Sample collection

We decided to collect samples from water instead from organic matter to recover a wider selection of the fungal population in each site, not only of those directly involved in decay. Samples were collected from 0.5 m below the water surface using a clean and surface sterilized container. Typical sample volumes were 20 L. The whole sample was passed through 3 layers of sterile cheesecloth and afterwards filtered through 5 μm PVDF membranes (Millipore). The biomass layer was scrapped with a spatula from the membranes, washed in 3-5 ml of the same water and centrifuged in 3 different Eppendorf tubes for replica analysis. Final biomass volumes were of 0.1-0.3 ml. Pellets were frozen at −20 °C until extraction.

DNA extraction

The whole pellet was processed and total DNA extracted using the Ultra Clean Soil DNA Kit (MoBio, Carlsbad, USA) in accordance to the manufacturer's instructions. Total DNA was analyzed for integrity by agarose gel electrophoresis.

Ribosomal DNA amplification

Total DNA aliquots were amplified using primers nu-SSU-0817 and nu-SSU-1536 (^{Borneman & Hartin, 2000}), yielding a primary product of approximately 750 bp. Reaction mixtures contained 2.5 mM MgCl₂, 1 mM dNTPs mix, 5 pM of each primer, 50 ng of total DNA as template, 1X reaction buffer and 5 U of Taq polymerase (Altaenzymes, Alberta, Canada). Reaction mixtures were subjected to an initial denaturation step of 5 min at 95 °C followed by 30 amplification cycles (30 s at 95 °C, 30 s at 55 °C, 45 s at 72 °C) and a final extension step of 2 min at 72 °C. Amplifications were performed in a PCR sprint thermal cycler (Thermo Electron Corporation, USA). The resultant fragments were separated by agarose gel electrophoresis, the amplified fragment was visualized by ethidium bromide staining, excised and purified with the QIAquick gel extraction kit (QIAGEN, Hilden, Germany).

Library construction

Amplified fragments were directly cloned in the pGemT-easy vector (Promega, Madison, USA) and transformed by electroporation into Escherichia coli strain DH5α. Insert-containing clones were detected by the lack of coloration in the presence of X-gal and IPTG and picked in fresh plates. Plasmids from 6 randomly selected clones from each library were extracted by alkaline lysis and sequenced for insert verification. In those cases where the amplification product could not be detected in at least 4 out of the 6 clones or when the sequenced inserts were not of fungal ribosomal genes, the procedure was repeated to obtain a better yield or quality.

DNA sequencing

A group of 384 randomly selected clones from each library was sequenced in sets of 4 96-well plates. Colonies were picked and plasmid DNA extracted by alkaline lysis. The concentration and integrity of isolated DNA were verified by agarose gel electrophoresis and sequenced with the T7 primer in the Sequencing Unit of the Centro de Ciencias Genómicas (UNAM).

Control experiments

Genomic DNA of different organisms was used as control before sample amplification. From prokaryotic sources: E. coli, Rhizobium meliloti and Spirulina maxima. From fungal sources: Aspergillus nidulans, Debaryomyces hansenii, Yarrowia lipolytica, Saccharomyces cerevisiae, Schizophyllum commune and Bjerkandera adusta. For an arthropod source we used DNA from Centruroides limpidus and from plant sources, Arabidopsis thaliana and Phaseolus vulgaris. In all cases the samples were kindly provided by colleagues at the Instituto de Biotecnología (UNAM) and the Centro de Investigación en Biotecnología (UAEM).

Sequence analysis

Sequenced DNA files were provided in FASTA format and the name of each file was manually edited in order to allow clear identification of each sequence in the future. An initial depuration was performed for each set of 384 sequences and very short or ambiguous sequences were removed. The rest of the sequences were submitted to individual identification against GenBank using the BLAST tool (^{Altschul et al., 1997}). While most of the sequences matched entries from fungal origins those that clearly corresponded to other phyla were removed. In samples from marine sources, the cutoff was less clear based on the poor number of well-characterized marine fungi so we decided to leave all ambiguous sequences in the set for further analysis. Approved sets were automatically aligned using the Clustal W algorithm (^{Larkin et al., 2007}) contained in Mega version 4 (^{Tamura, Dudley, Nei, & Kumar, 2007}). Most of the sequence aligned in the first round, although optimization of the surroundings of variable regions required manual alignment. Vector borne fragments were removed after alignment. A set of ribosomal sequences from well-identified organisms (Table 1) was added to each alignment for internal reference and the group was re-aligned and manually optimized with Clustal W. We sometimes detected groups of experimental sequences that did not cluster with reference sequences but within themselves. In these cases we identified each one of the sequences by looking for the closest match in the databases using the BLAST tool. For some sequences, the closest match resulted to be an entry from a characterized species but in other cases we recovered entries from environmental surveys. Phylogenies were reinforced by including sequences from characterized species to the reference sequence list and, sometimes, sequences from uncultured sources (Table 2). Sequence clustering was performed using the Neighbor Joining algorithm (^{Saitou & Nei, 1987}) contained in Mega version 4.

Amplification of fungal 18S ribosomal DNA from reference isolates

Control genomic DNA from different sources was tested for amplification with primers nu-SSU-0817 and nu-SSU-1536. Templates from non-fungal sources were unable to support amplification while fungal sources specifically amplified the internal fragment of 18S rDNA (data not shown). In 5 out of 6 fungal samples (A. nidulans, D. hansenii, Y. lipolytica, S. cerevisiae, and B. adusta) the size of the amplified product matched the expected 762 bp (data not shown). Sequence data quality was assessed by sequencing a control library generated by amplification of 18S rDNA from S. cerevisiae.

Identification and phylogenetic analysis of nu-SSU-0817 and nu-SSU-1536 amplification libraries

Total DNA from water samples from 11 different sites (see Section ‘Materials and methods’) was isolated, a fragment of the 18S rDNA amplified and cloned in genetic libraries for individual clone sequencing. In those rare cases where rDNA sequences from phylogenetic groups other than fungi were recovered, these were removed from the collection before clustering. The only exception to this behavior was the sample from Carpintero lagoon, where the 146 sequences recovered were more similar to reference sequences from dinoflagellates than from fungi. This set of sequences was removed from our analysis.

Clustering analysis of the libraries

We tested the ability of our molecular marker, a fragment of the 18 rDNA gene containing regions V4-V8, to reconstruct a clustering profile consistent with the taxonomic classification using our ensemble of reference sequences (Fig. 1). Once the robustness of the clustering method was demonstrated, the environmental sequences from 10 of our 11 libraries (after removing the samples from Carpintero Lagoon for the reasons described above) were organized by site along with the fungal reference sequences as described in Section ‘Materials and methods’. Serial reconstructions of the 10 libraries were performed using the manually optimized multiple alignment from each site, which included the recovered reference sequences, until a robust and informative phylogram was obtained. Based on these phylograms, we identified the most probable taxonomic classification of the environmental sequences by their clustering with reference sequences.

Figure 1 Reconstructed phylogram of reference sequences including supporting environmental sequences and other metazoa (Alveolata, Eccrinales, Ichtryosporeae and Choanoflagellida) indicating the taxonomic classification of the organisms. Sequences from Arthropoda and Cnidaria were added to root the phylogram. 

By the methodology described above, we were able to identify 529 of the environmental sequences at the family level, 1077 at the order level and 288 at the Class/Subphylum level. Only 326 sequences, all from the same site, did not group with any reference sequence but to uncultivated clones from marine sources (Table 3). The most abundant phylum found was Ascomycota (1458 sequences) (Fig. 2), followed by Chytridiomycota (133 sequences) (Fig. 3). Sequences belonging to phylum Basidiomycota (45 sequences) (Fig. 4) and to subphylum Mucoromycotina (21 sequences) were seldom recovered as well as non-fungal sequences belonging to Choanoflagellidae and Eccrinales (29 sequences) (Fig. 5).

Figure 2 Ascomycota diversity in the sampled sites identified by Class (sub-phyllum)/order/family. 

Figure 3 Chytridiomycota diversity in the sampled sites identified by Class (sub-phyllum)/order/family. 

Figure 4 Basidiomycota diversity in the sampled sites identified by Class (sub-phyllum)/order/family. 

Figure 5 Non fungal microbial diversity in the sampled sites identified by Class (sub-phyllum)/order/family. 

Diversity estimation

Molecular Operational Taxonomical Units (MOTUS) diversity was estimated using the Shannon Index, which gives a measure of both species numbers and the evenness of their abundance as presented in Table 4 (^{Shannon, 1948}). The value of the index ranges from low values (reduced species richness and evenness) to high values (extended species evenness and richness). In this work, the lowest diversity value was obtained for Media Luna and Bahía de Banderas, samples with a single MOTU identified in each (H′ = 0). In contrast, the highest diversity value was obtained for the Zempoala lagoon (H′ = 2.285), slightly higher than the total diversity by family (H′ = 2.130).

To determine if the different groups were randomly present in the various sample sites or if there was a predominance of certain groups in a given site, Friedman's test for the 2-way classification was performed. The entire population was distributed using the 10 different sites (blocks) and 18 taxonomic groups classified at the order level (treatments). Application of Friedman's procedure to the data resulted in a X² value of 35.71 and 17 degrees of freedom.