An advanced data evaluation strategy for assessing temporal changes in dissolved organic matter quality during flood events based on ultrahigh-resolution mass spectrometry

Water chemistry can change dramatically during a flood event. While variations in the concentration of inorganic ions, nutrients and bulk DOC as function of discharge have been  intensively investigated, changes in dissolved organic matter (DOM) quality were considered less detailed with respect to high resolution techniques. DOM is a highly complex mixture consisting of thousands of different elemental compositions. Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) is the analytical tool with the up to date highest DOM quality resolution. Here we present a novel data evaluation strategy for time series applied to a flood event in a small river catchment. As sampling area the Ammer River near Tübingen, southern Germany was selected. During a flood event with 900% MQ in 2021, six water samples were collected within five hours period at the Pfäffingen monitoring station while discharge changed by a factor of five. Water samples were filtered (Whatman GF/F), acidified and passed through PPL cartridges. Methanolic eluates were analyzed by FTICR-MS in negative ionization mode (ESI^-). Molecular formulas (MFs) were calculated for the mass range between 150–1000 Da using in-house software, considering the elements: carbon ¹²C_1–80, hydrogen ¹H_1–198, oxygen ¹⁶O_0–40, nitrogen¹⁴N_0–2, and sulphur ³²S_0–1. A total of 3500 formulas were shared among all six samples. Inter sample ranks were calculated for each molecular formula based on relative signal intensity, with rank 1 representing the highest and rank 6 the lowest abundance.  (1,2). From the inter sample ranks the rank sequences were derived (for example 3-6-1-5-4-2) and used as input tor hierarchical cluster analysis (HCH). Five superordinate clusters were selected for further evaluation. Rank distribution of formulas within each cluster were visualized via bar graph and molecular formulas were plotted in van Krevelen diagrams (H/C versus O/C). This visualization revealed flood-specific compositional dynamics in DOM. Sulfur-containing compounds (CHOS) exhibited their highest relative abundance at peak discharge (fourth sample), whereas aliphatic CHO compounds (H/C > 1.5) were most abundant at low discharge (first and last samples). In contrast, aliphatic CHO (H/C > 1.5) showed highest abundance at lowest discharge (first sample and last sample). Nitrogen-containing components (CHNO) showed different ranking distribution and revealed highest abundance in the second and third sample (before discharge peak).

In conclusion, DOM exhibits highly divers and dynamic behavior during flood events due to its complex composition and information received from bulk DOC concentrations alone seems to be insufficient to capture these compositional changes.

1) Herzsprung P. et al., Environ. Sci. Technol. (2012), 46, 5511-5518

2) Dadi. et al., Environ. Sci. Technol. (2017), 51, 13705-13713