Section 7 update: Cluster analysis and statistical comparison of molecular community profile data

Characterization of the diversity and structure of microbial communities in the environment has been made possible by technical developments in molecular biology during the last two decades. Since Muyzer et al. [18] in 1993 first reported the use of denaturing gradient gel electrophoresis (DGGE) for the analysis of whole bacterial communities, this fingerprinting technique has become the most popular tool to characterize microbial communities [Muyzer et al. 1998 and other chapters in this manual], despite the development of many other community profiling techniques such as Amplified Ribosomal DNA Restriction Analysis (ARDRA), Length Heterogeneity-PCR (LH-PCR), Ribosomal Intergenic Spacer Analysis (RISA), Single-Strand Conformation Polymorphism (SSCP) and Terminal-Restriction Fragment Length Polymorphism (T-RFLP). Cultureindependent community profiles are made for defined habitats by using phylogenetically relevant macromolecules, particularly 16S rRNA gene fragments directly amplified from the environment [see e.g. 20, 24, 26]. This has revealed an unsuspected complexity of whole bacterial communities as is exemplified by Krave et al. [13], who showed stratification of bacterial communities in a pine forest soil in Central Java, Indonesia (see Figure 1). Potentially it is possible to give a complete description of the microbial communities by classical cloning and sequencing techniques, but this is usually not feasible in practice [13, 20]. The fingerprinting of microbial communities by electrophoretic separation of PCR-amplified fragment pools was developed as an alternative solution to labour-intensive and expensive cloning and sequencing analysis. Usually fingerprinting is used in combination with cloning and selective sequencing (see e.g. [7, 13, 22, 24]). Often community fingerprints are only visually compared, however far more information can be obtained by rigorous analysis of the community profiles, such as the calculation of diversity statistics and numerical comparison of the profiles, using clustering or ordination techniques.