Beyond the Obvious: Exploring Peat Vibrational Spectroscopy (FTIR-ATR) Data Using Principal Components Analysis on Transposed Data Matrix (tPCA), Store Mosse Bog (Sweden)

Peatlands are a major soil carbon reservoir, despite only covering about 3% of the land surface. Waterlogged peatlands act as sinks of atmospheric carbon, through accumulation of partially decomposed plant remains. But there are concerns that boreal peatlands may shift from sinks to sources due to enhanced peat mineralization with climate change. Characterizing the molecular composition of the peat is key to understanding the responses of peatlands to climate change, as it is expected that not all peat constituents will be equally affected. Mid-infrared (MIR) vibrational spectroscopy is a fast, cost-efficient technique that provides information on the molecular composition of many different materials, although the interpretation of MIR spectra may be hampered by the compositional complexity of the material analyzed. Chemometric methods, such as the application of principal components analysis (PCA), can help to untangle the MIR spectrum but, to date, most previous studies using this approach only consider a few main vibrations representative of the main peat components (i.e., cellulose, lignin, etc.), losing part of the information contained in the spectra. In this chapter, we explore the application of PCA on transposed MIR data (tPCA), using a previously well studied peat sequence from Store Mosse (SM-S2008), one of the largest peatlands in southern Sweden. Here, we exemplify how using the whole spectrum of each sample in the statistical analysis enables a proper characterization of the constituents of the peat. We describe both the scores’ spectra of the extracted components and identify the peat constituents they represent. We also describe the records of samples’ loadings, which show the variation of the constituents along the peat sequence. To support our interpretations, we compare the tPCA components with those extracted in a previous investigation using direct PCA (dPCA), correlate the tPCA components with elemental (C, N, and C/N ratio) and isotopic (δ13C and δ15N) properties of the peat, and model these same properties using the principal components. The results show that the major, and more obvious, constituents of the peat are easily identified here, but other relevant signals, more difficult to detect, such as that of microbial biomass are also extracted. Most of the tPCA components were highly correlated with the peat properties and were significant in the statistical models. These findings show that tPCA allows extracting detailed information about peat constituents, which is essential for understanding how peatlands may respond to ongoing and future climate change.