Soil microbial diversity is still a difficult field to study, especially due to the several limitations of techniques. Since 95-99% of organisms cannot be cultivated by culture based — methodologies, the microbial diversity of soils shall be assessed by molecular biology techniques (Elsas & Boersama, 2011).

New DNA and RNA sequencing techniques provide high resolution information, especially using depth sequencing of metagenomic samples. Most of times a high amount of the obtained sequences are related with unknown genes or unknown organisms, involving a high cost per sample. Since soils imply in most of times in high spatial variability, which means high number of samples and replicates, fingerprinting techniques are recommended prior to sequencing in order to reduce costs for the high resolution techniques.

The first study of microbial diversity in Amazon soils using molecular techniques, by means of clone library, showed a high prokaryotic diversity (Borneman & Tripplett, 1997). Analyzing 100 sequences, differences between mature forest and pasture were detected, and about 18% of sequences were related to unknown Bacteria. A decade after, analyzing 654 clones similar results were detected in other study site, in which 7% of sequences could not be classified in any bacterial phyla (Jesus et al., 2009). In both studies land use changes was an important factor, and the unknown species were surveyed showing that depth sequencing should be used to better characterize the Amazon soils.

The most popular techniques for soil microbial communities fingerprinting are DGGE and the terminal restriction fragments length polymorphism (T-RFLP), which should be complemented by sequencing information to provide an overview of the study sites. Such techniques consist in extraction of nucleic acids from the soil samples; followed by amplification by PCR, aiming to target specific microbial groups according to the primers chosen (i. e. a universal primer for 16S rRNA gene will give a general prokaryotic overview of the samples). After PCR the amplicons should be analyzed by denaturizing gel separation (DGGE) or digestion with restriction enzymes and analysis of the dye labeled fragments (T — RFLP), or DNA sequencing. In turn metagenomics techniques allow sequencing without preview amplification by PCR and other techniques to be considered (Elsas & Boersama,

2011) .

T-RFLP consists in a PCR using dye labeled primers followed by a digestion with restriction enzymes, purification and reading in a DNA sequencer. The PCR amplifies a specific gene (mainly the 16S rRNA gene for prokaryotic diversity), and the restriction enzymes fragment the PCR products according to its polymorphism. The sequencer separates the fragments by length reading them in an electrophoresis run. So the presence of distinct fragment sizes found in different soil samples allows the diversity separation among them (Jesus et al., 2009). Clone libraries consist in cloning the PCR amplicons into bacterial vectors, followed by DNA sequencing. Since the PCR from environmental samples amplify different DNA sequences of different organisms at the same time, cloning technique allows the separation of amplicons and the sequencing of individual sequences (Borneman & Tripplett, 1997). Different studies using other molecular approaches to access the diversity of Amazon soils (Table 1) are described below.

In Western Amazon a T-RFLP analysis of the bacterial communities showed how it was influenced by soil attributes correlated to land use (Jesus et al., 2009). Community structure changed with pH and nutrient concentration. By DNA sequencing, bacterial communities presented clear differences among the different sites. Pasture and one of crops presented the highest diversity. Secondary forest presented similar diversity with the community structure of the primary forest, showing that bacterial community can be restored after agricultural use of the soils. Using the automated ribosomal intergenic spacer amplification (ARISA) technique distinct microbial structures were also observed between agricultural and forest soils (Navarrete et al., 2010). Seasonal changes in the two different years of sampling and distinct band patterns were observed for fungal, bacterial and archaeal richness.

Different patterns between Terra Preta soil (Dark Earth or Anthrosols) and an adjacent soil were observed in the Southwestern Amazon using 16S rRNA gene sequencing (Kim et al.,

2007) . Acidobacteria were predominant in both sites but 25% greater species richness was

Grossman et al. (2010) studying the three same Dark Earths sites, including one additional site, "Dona Stella", and using different molecular techniques also found difference among the samples.. T-RFLP of the 16S rRNA genes provided clear distinction between the two types of soils, and the same result was observed using DGGE and 16S rRNA sequencing. While T-RFLP provided a good fingerprinting between Anthrosols and Adjacent soils, 16S rRNA sequencing provided better resolution of the changes, indicating Verrucomicrobia as an important group to the Anthrosols, Proteobacteria and Cyanobacteria for Adjacent soils; while Pseudomonas, Acidobacteria and Flexibacter were found in both sites.

Studying the "Hatahara" site, differences in bacterial communities were also observed among Amazonian Dark Earth, black carbon and an adjacent oxisol by T-RFLP (Navarrete et al., 2010). By pyrosequencing it was shown that the most predominant phyla were Proteobacteria, Acidobacteria, Actinobacteria and Verrucomicrobia. About one-third of the sequences corresponded to unclassified Bacteria. For archaeal structure comparison by T — RFLP the soil attributes were more important than the type of soil, if it was Terra Preta or adjacent soils (Taketani, 2010). DNA sequencing showed that Candidatus spp. was the most abundant genera in both types of soils. An amoA clone library showed differences among the sampled sites, but also did not show differences between Terra Preta and the adjacent soil.

Using T-RFLP of bacterial 16S rRNA, distinct patterns were observed among biomes and land uses in the Southwestern Amazon (Lammel et al., 2010). Southwestern Amazon is divided in two mainly biomes, Tropical Forest and Cerrado (Brazilian Savanna). Over the last three decades these natural vegetations have been converted to pasture and agriculture. Land use was the most important factor to distinguish the bacterial communities, and it was correlated with the soil chemical changes: pH — due to liming and chemical fertility — due to fertilizers application. Pristine Tropical Forest and Cerrado formed distinct clusters, but they were more similar to each other than in relation to pasture or soybean field (Fig. 7).

In Eastern Amazon wetland soils Archaeal community was characterized by 16S rRNA gene libraries and by isolation of methanogenic Archaea (Pazinato et al., 2010). Archaeal diversity decreased with depth and the most of sequences belonging to Crenarchaeota, Methanosarcina and Metahnobacteriam genera were isolated from the sites.

These different techniques showed a high microbial diversity on Amazon soils. Fingerprinting techniques, such as T-RFLP and ARISA, were sensitive tools to detect difference in the microbial structure among the different sites and land uses. However only DNA sequencing provided a better resolution of the diversity, i. e. identify taxonomic groups and report unknown Bacteria that probably belong to new taxonomic groups. These pioneer studies showed, in general, that diversity does not decrease from pristine vegetation to agricultural uses, but the structure of microbial community as a whole is affected by land use changes. They can be restored after stopping the soil cultivation followed by secondary forest growth. The Amazon region is a "hot spot" regarding the soil microbial diversity.