The community composition of arbuscular mycorrhizal fungi (AMF) was investigated in roots of four different plant species (Inula salicina, Medicago sativa, Origanum vulgare, and Bromus erectus) sampled in (1) a plant species-rich calcareous grassland, (2) a bait plant bioassay conducted directly in that grassland, and (3) a greenhouse trap experiment using soil and a transplanted whole plant from that grassland as inoculum. Roots were analyzed by AMF-specific nested polymerase chain reaction, restriction fragment length polymorphism screening, and sequence analyses of rDNA small subunit and internal transcribed spacer regions. The AMF sequences were analyzed phylogenetically and used to define monophyletic phylotypes. Overall, 16 phylotypes from several lineages of AMF were detected. The community composition was strongly influenced by the experimental approach, with additional influence of cultivation duration, substrate, and host plant species in some experiments. Some fungal phylotypes, e.g., GLOM-A3 (Glomus mosseae) and several members of Glomus group B, appeared predominantly in the greenhouse experiment or in bait plants. Thus, these phylotypes can be considered r strategists, rapidly colonizing uncolonized ruderal habitats in early successional stages of the fungal community. In the greenhouse experiment, for instance, G. mosseae was abundant after 3 months, but could not be detected anymore after 10 months. In contrast, other phylotypes as GLOM-A17 (G. badium) and GLOM-A16 were detected almost exclusively in roots sampled from plants naturally growing in the grassland or from bait plants exposed in the field, indicating that they preferentially occur in late successional stages of fungal communities and thus represent the K strategy. The only phylotype found with high frequency in all three experimental approaches was GLOM A-1 (G. intraradices), which is known to be a generalist. These results indicate that, in greenhouse trap experiments, it is difficult to establish a root-colonizing AMF community reflecting the diversity of these fungi in the field roots because fungal succession in such artificial systems may bias the results. However, the field bait plant approach might be a convenient way to study the influence of different environmental factors on AMF community composition directly under the field conditions. For a better understanding of the dynamics of AMF communities, it will be necessary to classify AMF phylotypes and species according to their life history strategies.