Hydrogen isotope ratios of moss cellulose and source water in wetlands of Lake Superior, United States reveal their potential for quantitative paleoclimatic reconstructions

The hydrogen isotopic composition of organic substrates provides valuable information on past hydrological changes. To explore whether quantitative information can be obtained from this proxy, we measured D/H ratios of cellulose extracted from mosses inhabiting both hollow and hummock sites in three swales located near Lake Superior, United States. These isotope ratios were compared to those of potential water sources, namely swale, moss, and ground water, to evaluate whether cellulose D/H ratios reflect those of its water sources. For that evaluation, we relied on a better mechanistic understanding of the isotopic fractionation occurring during cellulose synthesis in plants as well as the relative simplicity of moss physiology in regard to water use strategies and a dataset that combines published D/H ratios for modern moss cellulose. We found that moss cellulose δD values for hollow species (− 118.1 ± 8.5‰) were significantly depleted (p = 0.016) in D by 8‰ in relation to those for hummock species (− 110.1 ± 8.1‰), possibly due to an averaging effect taking place in hollow species. In addition, we found that 40% of the hydrogen in cellulose from hollow species probably experienced isotopic exchange with the surrounding cell water, and this percentage is not statistically different from that (42%) obtained from an analysis of published data on modern moss cellulose. This analysis also reveals that the fraction of hydrogen in modern moss cellulose that is subjected to isotope exchange falls within a relatively narrow range (± 3%). Using this range and a mechanistic model, we were able to estimate source water δD values using those measured for moss cellulose, and these estimates were statistically indistinguishable from the measured δD values of moss and swale water. Further, the ratios of δD/δ18O of moss cellulose of hollow species yielded values that are similar to those reported for the local meteoric water line, thus adding a potential source of paleo-hydrological reconstructions from moss cellulose. Consequently, our combined results suggest not only that moss cellulose D/H ratios reflect those of source water, but that the latter values can be estimated fairly precisely from cellulose D/H ratios using a relatively simple mechanistic model. However, the accuracy of this model hinges on assumptions that need to be tested further to corroborate its widespread applicability.