To effectively characterise and distinguish between different organic matter samples, multiple chemical characterisation techniques
are often employed. Due to the structural complexity of organic matter and the unique information provided by different characterisation
techniques, it is often difficult to compare and combine data obtained from different analytical methods. In this study, we show
how non-parametric multivariate statistical approaches can be used to compare the relative pattern of similarity/dissimilarity between organic
samples characterised by two common solid-state analytical techniques: 13C nuclear magnetic resonance (NMR) spectroscopy and
flash pyrolysis-gas chromatography mass spectrometry (py-GCMS). These analytical methods were used to characterise a suite of plant
residues including the leaf, flower, bark and wood of several species. Using non-parametric multivariate statistical approaches we identified
similarities between the plant residue data using ordination plots, which enabled us to identify where NMR and py-GCMS distinguished
between residues differently. A mantel-type test called RELATE showed that there was significant (P<0.05) similarity between
the NMR and py-GCMS data in terms of their ability to differentiate between plant residues of different type; 61% of the sample discrimination
was common to both profiling techniques, while 39% of discrimination was method specific. Further multivariate comparisons
indicated that NMR was more sensitive to detecting differences in the organic composition of the plant residues.
Keywords: Multivariate statistics, NMR, Organic matter, Plant residue, Pyrolysis, RELATE, Resemblance.
Rights & PermissionsPrintExport