Microbial Community of Major Fresh Produce Producing Soils Revealed By 454-Pyrosequencing.

High-throughput sequencing is a promising method, as it provides enough sequencing depth to cover the complex microbial communities. In this study, microbial community structure and composition were studied in 32 soils from California (CA) and Arizona (AZ). Our goal was to correlate the composition of certain bacterial groups in soils as revealed by 16S rRNA pyrosequencing with some environmental parameters, and access how certain bacterial group may be affected by the environmental factors. Sampling location was a significant factor affecting microbial community, but management (organic or conventional) was not. Proteobacteria, Bacteroidetes, Actinobacteria, Acidobacteria, and Firmicutes dominated (>73.45%) bacterial communities from the three locations classified using RDP Classifier at a confidence threshold of 50%. Canonical correspondence analysis (CCA) of bacterial community structure and soil variables showed that EC, pH, WHC, sand, TN, and OC significantly (P =0.05) impacted microbial communities, whereas clay, WSOC, and MBC did not. CCA-based variation partitioning analysis (VPA) showed that soil physical properties explained 16.3% of the variation  in microbial communities, and soil chemical variable explained 12.5%, location explained 50.9%, and 13% unexplained. Kohonen self-organizing map of microbial community composition and associated soil chemical, physical and biological variables using artificial neural network analysis showed that community composition in soils was negatively correlated with OC, but positively correlated with TN, clay and WSOC. Both analytical techniques allow better fit of nonlinear data exhibiting complex or chaotic behavior for prediction of environmental variables affecting microbial community composition.