The study site, Hassel and Rappbode pre-dams, are located in the eastern Harz Mountains (Fig. 1), as part of the largest drinking water reservoir (the Rappbode Reservoir System) in Germany. The Hassel and Rappbode pre-dams are impoundments reconstructed from two tributaries in front of the main reservoir in the 1960s. They serve as sedimentation basins for the reduction of particle and nutrient loads (Friese et al. 2014). The Hassel pre-dam has a surface area of 0.26 km2 and an impounding capacity of 1.64 Mio m3 with a residence time around 65 days (Dadi et al. 2016). The deepest site of the pre-dam is 14 m, while the average depth is 5.0 m. The sub-catchment of Hassel pre-dam covers 44.6 km2 with almost equal share of forest, cropped land and grassland area. The Rappbode pre-dam is adjacent to Hassel pre-dam and morphologically similar to Hassel pre-dam in terms of surface area (0.26 km2), water impounding capacity (1.66 Mio m3), mean and maximum depths (5.7 m and 17.0 m, respectively) and catchment area (47.6 km2). The catchment of the Rappbode pre-dam consists of 72 % of forest, 22 % grassland and almost no agricultural land (Wendt-Potthoff et al. 2014). Differences in the coverage of agriculture lands between the Hassel and Rappbode pre-dam catchments lead to contrasting trophic states in these two pre-dams and can be preserved in sediments accumulated over past few decades.
Sediment cores were collected from four sites, two in mesotrophic Rappbode pre-dam and two in the eutrophic Hassel pre-dam. In the Rappbode pre-dam, a shallow sampling site (51.70°N, 10.79°E, 4 m in water depth) was located near the inflow, and a deep sampling site (51.71°N, 10.80°E, 14 m in water depth) was close to the main dam (Fig. 1). The shallow (51.70°N, 10.83°E) and deep sampling sites (51.71°N, 10.83°E) in the Hassel pre-dam have the same water depths (4 m and 14 m, respectively) and offer good systematic comparisons with the Rappbode pre-dam (Fig. 1).
Sediments and suspended particulate organic matter sampling
Sediments were sampled with a modified Kajak gravity corer (UWITEC, Austria) fixed with a polyvinyl chloride (PVC) cylindrical tube (9 cm in diameter, 60 cm in length) in April, 2011. Four replicate sediment cores were taken from each sampling site. Sediment cores were protected with a tight cover and kept secured upright position during transportation. In the laboratory, sediment columns were then cut into 2 cm slices and transferred into pre-weighted polypropene plastic vials for freeze-drying. Dried sediments were ground with agate mortar. Before this, large pieces of plant detritus were picked out manually. Well-ground sediments were weighted and kept for subsequent measurements.
Suspended particulate organic matter (POM) in the water column was sampled to obtain general information about algae in the pre-dams. Duplicate surface water samples were collected (20 L for each sample) at the deep sites of Hassel and Rappbode pre-dams. In the laboratory, the suspended material was separated from the bulk water samples by ultrafiltration. Briefly, a volume of 19 L water sample was passed through a 0.45 µm ultrafiltration cassette (C613, Pellicon), and then the residual 1 L of concentrated water sample was centrifuged (10,000 rpm for 15 min, Beckman Coulter J2-MC High Speed Centrifuge, Minnesota, United States of America). The centrifugal precipitate was dried at 45 ˚С, and the dry particulate material was ground for POM characterization. We assumed that the material was mostly made up of algae as supported by findings of Friese et al. (2014) and Barth et al. (2017), who worked at the same sampling sites.
Chemical fractionation of OM in sediments
A newly developed four-step chemical fractionation, which treated sediments subsequently with hot water, hydrochloric acid (HCl), hydrogen peroxide (H2O2) and disodium peroxodisulfate (Na2S2O8), was applied to group the bulk sedimentary OM into four portions (Liu et al. 2020). In brief, the powdered sediments (3 g) were first treated with hot water (80 ℃) for 18 to 24 hours as a mild extraction. The obtained hot water extracted residues (2 g) were successively hydrolyzed with 6 M HCl at 80 ℃ for about 20 h. One gram of the remaining solids was further oxidized repeatedly by H2O2 (10 %) at a temperature of 50 ℃ until effervescence ceased. In the end, half a gram of the recalcitrant residue was exposed to Na2S2O8 which was buffered with sodium bicarbonate (NaHCO3) overnight (80 ℃). The retained bicarbonate in suspension was neutralized by 1 M HCl. Solid residues from each chemical extraction step were collected and dried for further characterization. The chemical fractionation procedure proceeded in triplicate for one or two depths as representative shallow and deep sediments of each sediment core (Fig. 2), to evaluate the inherent variability of ground sediments.
To obtain a full depth profile of OM composition, sediments from each 2 cm slice were subsampled and treated using the above sequential fractionation method, and the sedimentary OM fractions were determined for elemental and stable isotope compositions. Our previous work demonstrated that hot water extracts only a small fraction of microbial and soluble proteins and polysaccharides. Furthermore, HCl extracts the remaining majority of proteins, carbohydrates and carboxylic lipids; H2O2 oxidizes lignins, lipids and some complex proteins; and Na2S2O8 removes almost the remaining lipids (Liu et al. 2020). Down-core variations in the composition of sedimentary OM fractions thus can be ascribed to the diagenesis of specific organic components.
Stable isotope ratio analyses
Total organic carbon (TOC), total nitrogen (TN), stable carbon and nitrogen isotope compositions of untreated bulk sediments and chemical extraction residues were measured by isotope ratio mass spectrometry (IRMS, DELTA V advantage, Thermo Fisher Scientific, Bremen, Germany), which was coupled with an elemental analyzer (EA, Flash 2000, Thermo Fisher Scientific, Bremen, Germany). Five milligrams of untreated bulk sediments as well as hot water- and HCl-treated sediment residues were selected for stable isotope ratio measurement. Sample weights were increased to 20 mg for H2O2- and Na2S2O8-treated sediment residues in order to increase peak signals for isotope analyses, especially for 15N. Homogenized sediment residues were combusted into purified gases like CO2, N2, H2, SO2, after which CO2 and N2 were determined by EA and IRMS. The elemental compositions were measured as mass fraction (wt. %) and converted to absolute C and N weights in one gram of dry sediments (mg g-1). Carbon to nitrogen ratios (C/N) are given in atomic ratios in this work. Stable carbon and nitrogen isotope ratios (δ13C, δ15N) were presented in parts per mil (‰) with reference to Vienna PeeDee Belemnite (V-PDB) and atmosphere, respectively (Coplen et al. 2006; Kayler et al. 2011). The 1-s measurement precision was 0.1 ‰ for δ13C and 0.2 ‰ for δ15N based on repeat analyses (n=6) of well mixed bulk field samples (see Materials and methods in Liu et al. (2020)).
Two-source mixing model for sedimentary organic matter
Atomic C/N ratios of sediments is the proxy for the relative contribution of aquatic and terrestrial OM to sediments (Das et al. 2013). Full depth profiles of atomic C/N ratios were obtained from the analytical results of sedimentary C and N contents. To evaluate the relative composition of land- and algal-derived OM in sediments, a two-source (aquatic and terrestrial) mixing model based on the following equation (Guillemette et al. 2017) was employed,
where f is the relative proportion of terrestrial OM in sediments; C/N sediment, C/N terrestrial, and C/N aquatic are the atomic C/N of the bulk sediments, the soil from the catchment and algae in the euphotic zone of the reservoir, respectively. The mean soil C/N in Rappbode and Hassel catchments are 15.4 and 13.1, respectively (Schönfeldt 2013), and the algal C/N value in this study is derived from that of the suspended POM in reservoirs.
Chemical composition analysis using 13C nuclear magnetic resonance
Preliminary isotopic measurements indicated the discrepancy of δ13C and δ15N shifts between the treatments with HCl and H2O2 among different sampling sites. This discrepancy was pronounced for deep sediments below 10 cm. Therefore, both the HCl and H2O2 resistant residues of relatively deep sediments (14-16 cm) were analyzed by 13C nuclear magnetic resonance (NMR) for chemical structure (Fig. 2) in order to interpret the site-specific shifts in δ13C and δ15N after H2O2 oxidation. The analyses were performed on a Bruker Avance 300 NMR spectrometer (Bruker Biospin, Bremen, Germany), coupled with Cross Polarization/Magic Angle Spinning (CP/MAS). The spectrometer was operated at a 13C frequency of 75.47 MHz. The proton 90° pulse was set to 3.3 µs, and the decoupling strength during acquisition was 69 kHz. The 13C NMR spectra were acquired applying a spinning rate of 5 kHz with a 4 mm diameter cylindrical zirconia rotor, a recycle delay of 2 s and a contact time of 2 ms. All spectra were processed with a Lorentzian line broadening of 1 Hz.
Spectra of 13C NMR analyses were analyzed using TopSpin 3.5 (Bruker, Biospin). Chemical shifts in the spectra were externally referred to the glycine resonance at 176.03 ppm and recorded from -100 to 300 ppm after linear baseline correction. Integrations of 13C NMR spectra were accomplished by grouping the chemical shift into regions as alkyl C (0-45 ppm), N-alkyl-methoxy C (45-60 ppm), O/O2-alkyl C and acetals (60-100 ppm), aromatic C (100-145 ppm), O-aromatic C with phenols (145-165 ppm), esters, carboxyl or carbonyl C (165-215 ppm) (Rodríguez-Murillo et al. 2011). Aliphaticity and aromaticity of the sedimentary OM were taken as the areas of 13C NMR spectroscopic integration between 0 and 100 ppm and 100 to 165 ppm, respectively.
Mean values and standard deviations of elemental and stable isotope compositions were calculated for sediment samples with replicates (see Fig. 2). Magnitudes of organic components removed progressively by chemical fractionation were quantified from differences of TOC or TN contents of sediment residues between two successive fractionation steps. Down-core variations in TOC, TN, δ13C and δ15N values of untreated bulk sediments were simulated by generalized additive models (GAM) with the R package “mgcv”. Geochemical proxies (elemental compositions and stable isotope ratios) were compared between deep and shallow sites within the same pre-dam and between two pre-dams by unpaired Student’s t-test. All statistical analyses were conducted under R computational environment in R version 3.4.3 (R Core 2013). Paired Student’s t-test was applied for the comparisons of the stable isotope ratios between two successive fractionation steps. The significance level (p) of all the statistical analyses was set at 0.05.