In this study, the novel data mining technique Market Basket Analysis (MBA) was applied for the first time in biogeochemical and ecological investigations. The method was tested on the fern Athyrium distentifolium, in which we measured concentrations of the elements Ca, Cd, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, Pb, and Zn. Plants were sampled from sites with different types of bedrock in the Tatra National Park in Poland. MBA was used to investigate whether specimens of Athyrium distentifolium that contain elevated levels of certain elements occur more frequently on a specific type of bedrock and to identify relationships between the type of bedrock and the concentrations of the elements in this fern. The results were compared with those of the commonly used principal component and classification analysis (PCCA) technique. MBA and PCCA ordination both yielded distinct groups of ferns growing on different types of bedrock. Although the results of MBA and PCCA were similar, MBA has the advantage of being independent of the size of the data set. In addition, MBA revealed not only dominant elements but, in the case of limestone bedrock, also showed very low concentrations of Cd, Fe, Mn, and Pb in ferns growing on this type of parent material. MBA, thus, appeared to be a promising data mining method to reveal chemical relations in the environment as well as the accumulation of chemical elements in bioindicators. This technique can be used to reveal associations and correlations among items in large data sets collected on a national or even larger scale.
