Arbuscular mycorrhizal (AM) fungi have gained immense research interest due to their symbiotic relation with roots of most terrestrial land plants. Diverse positive effects on plant health and fitness, as well as a high ecological relevance for many other below-and aboveground processes have been recognized. Despite growing research ambitions, sufficient knowledge on drivers and their relative impacts on AM fungal communities at varying scales and ecosystems is lacking. Very little attention has been paid to the interrelation of the spatial structure of AM host plants (here meaning the small- scale distribution of host plant individuals or populations) and AM fungal community dynamics, although deeper insights could move forward the development of theoretical community frameworks applicable to these important endosymbionts. The work presented in this thesis aimed to study the effects of host plant spatial structure for the establishment of mycorrhizal symbiosis as well as for AM fungal diversity and community composition. To do so, controlled experiments mimicking micro- landscapes of AM fungal habitats were conducted and compared with observational data from temperate forests in Germany. In the experiments, we used meta-community designs to test the responses of root colonization, AM fungal diversity and community composition to levels of habitat connectance or habitat heterogeneity. We assayed rates of root colonization across forests sites differing in AM-associating plant cover. For assessing the importance of further spatio-temporal parameters and host-specificity we surveyed AM fungal communities within one of the forest sites over two years. As expected, we observed a consistent pattern of higher rates of root colonization at a higher proximity of host plants, which in the case of the field study was a higher cover of AM- associating woody plants and in the case of the experiment treatments a higher connectance of host plant patches. There were generally no clear effects of habitat structure on AM fungal diversity and community composition in both the field and the experimental study. In the field, the main finding was that spatial parameters, followed by host identity, more strongly affected AM fungal assemblages than temporal parameters. In the experiments, we observed within-experimental-unit differences of beta-diversity between pairs of long- and short-distance patches as well as between pairs of different habitat quality, which rendered those communities less predictable. In conclusion, we present empirical evidence for a co-variation of AM plant host structure and AM fungal abundance. These findings are of importance for the restoration or management of natural forest habitats, as well as for sustainable agriculture. Furthermore, this work supports previous findings of high stochasticity in AM fungal communities, highlighting the need for further research into spatial scale-specific drivers of this important organism group.