Use of Factor Analysis in Interpreting Genetic Processes in Lacustrine Soils of Ontario, Canada

Abstract

A multivariate statistical technique, Q-mode factor analysis, was applied to 120 samples collected from five lacustrine soils of southwestern Ontario. In each sample, 23 variables were measured: 11 chemical, and 12 particle size classes. Three factors were found to account adequately for the variation in the original data, i.e. 120 samples × 23 variables. The factors can be interpreted as indicators of depositional and weathering environments. To simplify such an interpretation and to try and separate the effects of the depositional environment from the pedogenic processes, the chemical data was treated separately from the grain size data. Two factors were found to describe adequately the variation in chemical data and three factors accounted for the variation in the grain size data. The chemical factors were interpreted to represent the processes of weathering occurring in the soil profile. The grain size factors were postulated as representing energy levels in the environment of deposition of the parent materials, i.e. a low energy environment where the sediments were formed predominantly through settling from suspension, a relatively higher energy environment where a significant portion of the sediments were deposited from traction currents, and an intermediate energy level where both settling and traction were important processes. These interpretations were arrived at considering only laboratory data and they agree with morphological information obtained from the field.In the exploratory or preliminary stages of research, factor analysis is a useful tool for manipulating a large amount of data. This type of analysis can complement morphological studies that are still needed for a full understanding of the genesis and classification of soils.