Philosophiae Doctor - PhD / This thesis presents significant advances into the microbial ecology of hypolithic communities in two hyperarid deserts. Deserts account for one fifth of the Earths total surface area. These zones differ substantially in terms of climate, geomorphology, hydrology and vegetation. Desert biomes are, however, generally depauperate with particularly with respect to macroorganisms. Hypoliths, photosynthetic microbial assemblages associated with quartz rocks, are widely distributed in hot and cold desert environs where they may represent a large fraction of the standing biomass and mediate key ecosystem processes, including nutrient cycling. However, important questions regarding their (i) development (ii) community structure and assembly patterns and (iii) functional structure remain unaddressed. Here, molecular tools (T-RFLP, clone libraries and pyrosequencing) and multivariate data analyses were used to address these questions. This study presents evidence of species recruitment in the development of hypolithic communities in the Namib Desert. Hypolithic bacterial communities were compared at a fine scale (10 m radius). Multivariate analysis of T-RFLP-derived data showed that hypolithic and open soil communities were structurally distinct. Applying the ecological concept of ‘indicator species’, 6 and 9 indicator lineages were identified for hypoliths and soil, respectively. Hypolithic communities were dominated by cyanobacteria affiliated to Pleurocapsales, whereas actinobacteria were prevalent in the open soil. These results are consistent with the concept of species sorting and suggest that the underside of the quartz rocks provide conditions suitable for the development of discrete and demonstrably different microbial assemblages.However, strong evidence for neutral assembly processes was found, as almost 90% of the taxa present in the hypoliths were also detected in the open soil. All together, these results suggest that hypolithons do not develop independently from microbial communities found in the surrounding soil, but selectively recruit from local populations.The bacterial community structure and assembly patterns in hypolithons from Miers Valley (Antarctica) were investigated. Previous studies in this valley have identified three morphologically distinct hypolithic community types: cyanobacteria dominated(Type I), fungus dominated (Type II) and moss dominated (Type III). The bacterial composition of surface soils and hypolithic communities were shown to be clearly and robustly distinct, using T-RFLP analysis. Moreover, the bacterial assemblages were similar in Type II and Type III hypolithons and clearly distinct from those foundin Type I. Using16S ribosomal RNA gene (rRNA) 454 pyrosequencing,Proteobacteria were shown to be the most important bacterial component of all three types of hypolithic communities. As expected, Cyanobacteria dominated Type I hypolithons, whereas Actinobacteria dominated Types II and III hypolithons. Using a probabilistic dissimilarity metric and random sampling, deterministic processes were demonstrated to be relatively more important in shaping the structure of the bacterial community found in Type II and Type III hypolithons. Taken together, these results suggest that hypolithic development favors a sequential pathway with Type II hypolithons serving as an intermediate development state between Type I and Type In a more in depth analysis of the diversity patterns of key nutrient cycling genes in Antarctic Miers Valley edaphic communities, genes coding for carbon fixation (greenand red-like cbbL), nitrogen fixation (nifH), nitrification (amoA) and denitrification(nirK and nirS), were targeted. Multivariate analysis (PERMANOVA) showed that hypolithic and open soil communities were functionally distinct. Type I hypoliths were functionally more diverse than soils, suggesting higher potential for enzymatic activities. Taxonomic structure (derived from 16S rRNA data) showed congruence with functional traits (genes involved in C and N cycling). Redundancy analysis suggested that chemical variables (S, F, and NO3) were important structuring forces in the different communities. Taken together, the results suggest that stochastic processes such as dispersion cannot override the influence of environmental factors on functional diversity patterns.