Lakes play an important role in the global carbon (C) cycle as they process carbon from terrestrial (allochthonous) and within lake (autochthonous) sources and may store C over long periods of time. Some arctic lakes contain high concentrations of dissolved organic carbon (DOC) that does not absorb light and thus remains colourless. The origin of this DOC remains unknown, but the sediment of these lakes have been suggested to accumulate primarily autochthonous (algal) C. I developed an experimental chamber for hydrogen (H) isotope pre-treatments and applied a novel H isotope tracing approach to determine the origin of the DOC and sediment C of a lake on SW Greenland known to contain colourless DOC. I hypothesized that autochthonous C was the prime source of DOC and sediment C, in line with previous theories. Analyses of algae and soil samples from the catchment revealed that local allochthonous and autochthonous C sources had a δ2H composition of -139 ‰ and -209 ‰, respectively. In contrast to my hypothesis, the analysed DOC had a mean δ2H isotopic composition of -147 ‰ indicating a dominance (ca 80-90 %) of allochthonous C. Similarly, the sediment had a mean δ2H isotopic composition of -155 ‰, suggesting that about 84 % of the C accumulating in the sediment was derived from terrestrial sources. The terrestrial origin was supported by field observations of high DOC seepage water (up to 70 mg L-1) with uncharacteristically low light absorption values entering the lake during high precipitation events. My results indicate that terrestrial processes are fundamental C sources for arctic lakes, even in regions with very low precipitation.

The stable isotope ratio of non-exchangeable hydrogen (δ2Hn) emerged as a promising new tool for source attribution in aquatic ecology, which can better discriminate between the organic matter sources. However, determining the absolute isotopic values is problematic due to the functional groups with hydrogen (H) that can easily exchange with the ambient water. This can lead to significant uncertainties in the absolute isotopic values, which eventually translates into errors in source attribution. However, controlled H exchange experiments with dual water equilibrations can alleviate this problem. However, current methods report significant variation in the absolute values, likely originating in partial H exchange, uncertainty in fractionation factors between exchangeable H and ambient water, and residual moisture. Here I used two methods for H exchange based on steam and liquid water using novel equipment for sample preparation called the Isobox. I evaluated the exchangeable H fraction in materials commonly analysed in ecology. Furthermore, I have investigated how these methods and associated analytical uncertainties can influence the mathematical mixing models used to resolve the source attribution using known sample mixtures made with soils and algae. The results show that 1) H exchange experiments using liquid water equilibration provide a higher exchangeable H fraction than the steam-based method, which was approximately three times higher. 2) The two-source mixing model proved to effectively determine the source attribution with known soil and algae mixtures, as evidenced by both water and steam equilibration. However, prior exposure to isotopically divergent waters can lead to source attribution errors, particularly with steam-based methods that provide low exchangeable H fractions. 3) When labile H is fully exchanged, source attribution does not depend on absolute δ2Hn determination, and simple one-water equilibration is sufficient. 4) Additional uncertainties in source attribution could originate from fractionation factors and δ2H measurement variability. The findings of this study conclude that the variations in fractionation factors did not significantly alter the mixing model as the error was below 5%. 5) Based on these experiments, I recommend keeping the analytical uncertainty of δ2H below ±5‰, which amounts to about a 6% error factor in source attribution. Finally, these experiments and analyses show how the methods can generate reliable data, depending on the research questions and whether absolute or relative isotopic values are required. This study provides different analytical pathways.