Subterranean ecological research and multivariate statistics: a review (1945-2006)

Subterranean ecosystem studies using multivariate ordination and/or agglomerative classification statistical methods were reviewed in the Science Citation Index (SCI) between 1945 and 2006. Nearly 57,000 publications cited subterranean habitats or their associated biota in the SCI abstracts, however, multivariate statistics applied to strictly hypogean taxa occurred only in only 65 papers from 1990 onwards. Over 90% of the multivariate applications were devoted to morphometric or genetic studies of single species and to relationships between the environment and species assemblages. In terms of taxa and ecosystem types, stygobite and waterless cave studies featuring multivariate applications predominated, respectively. Only six different methods (Agglomerative Clustering, Canonical Correspondence Analysis, Correspondence Analysis, Discriminant Analysis, non-metric Multidimensional Scaling, Principal Component Analysis) were used among the .30 multivariate techniques available within the biostatistical toolbox. The retrieved set of publications (n5 65) was sorted in a simple table by keyword according to type of biota, habitat, research topic and multivariate method, while online biostatistical resources are appended. Further comments are made on the use of statistics in the biological sciences in general.