Microbial bioprospecting is the identification of microorganisms, genes, enzymes, and/or metabolic pathways with biotechnological applications in industry or in research itself. This thesis aimed at bioprospecting a range of unexplored, natural and artificial environments subjected to harsh physicochemical conditions by applying a multi-level strategy. The microbial communities were analyzed in a first step by means of metagenomic sequencing, which yielded an extensive catalog of the microbial species and/or genes occurring in the community, and allowed the ecological contextualization of new habitats. Culturing in artificial media was used as a complementary approach in order to isolate, select, and characterize microbial strains of particular relevance. Last, a bioinformatics algorithm was developed in order to visualize community assemblage based on the predicted associations and interactions shaping microbial consortia in environmental samples. As a result of our analysis, a range of individual microbial strains, genes and microbial consortia with potential biotechnological applications have been identified. Moreover, we propose a new, holistic approach for the efficient mining and analysis of environmental microbiomes.