Published works

Soil organic matter has strong effects on many soil properties such as water holding capacity, soil structure and stability, nutrient availability and cation exchange capacity.  Therefore, characterising soil organic matter is necessary to improve soil management.  Pyrolysis coupled to gas chromatography-mass spectrometry (pyr-GC-MS) is one of many techniques that have been successfully used in this characterisation.  However, a major limitation of pyr-GC-MS is that generates large amounts of mass-spectrometry data preventing fast, high throughput data analysis.  This hinders our ability to identify compounds in complex matrices such as SOM that could be useful for predicting their characteristics.  In this study, we aimed to investigate whether it was possible to rapidly identify significant differences among pyr-GC-MS data from soil from burnt and unburnt areas using an unsupervised statistical approach and identify the specific features that cause them.  Of nearly 400 useful compounds extracted from the pyr-GC-MS data, only 15 were found to be necessary to classify between burnt and unburnt soil.  We discuss how these features could be useful in the classification of soil disturbance such as fire or, potentially, as a quantitative measure of fire impact (intensity or severity).