Exploratory Analysis of Geochemical Data and Inference of Soil Minerals at Sites Across Canada

Abstract

Soil plays a crucial role in ecosystem functioning, e.g., soil minerals provide important provisioning and regulating ecosystem services. The determination of soil mineral composition can help to link geochemical processes to underlying bedrock and surficial geology, however analysing quantitative soil mineralogy by X-ray diffraction can be expensive. This study used data from the North American Soil Geochemical Landscapes Project sampled at sites (n = 560) across Canada; exploratory analysis of major elements from the C-horizon, < 2 mm size fraction, was carried out to determine whether geochemical data can infer site-specific qualitative soil minerals. Results for the raw geochemical data indicated relative variability of major elements across Canadian provinces with noticeable differences for silica and calcium. Geochemical data are compositional, and as such their statistical assessment is subject to the problem of closure. In the current study, all raw geochemical data were centred log-ratio-transformed prior to statistical analysis to overcome closure. Graphical measures indicated skewed element data prior to centred log-ratio transformation, which produced a more symmetric distribution. Correlations between elements suggested tentative soil mineral composition, such as silica and aluminum from aluminosilicates minerals. Principal component analysis of transformed geochemical data revealed three distinct groups of calcium, magnesium; iron, titanium, manganese; and aluminum, potassium, silica, sodium, while phosphorus had smaller relative variability independent of these groups. The interpretation of these groups was based on soil minerals and identified as carbonates, silicates, and weathered secondary oxides. These minerals corresponded geospatially