Controls of δ34S and δ18O in dissolved sulphate: Learning from a detailed survey in the Llobregat River (Spain)

The S and O isotopic composition of dissolved SO4, used as a tracer for SO4 sources, was applied to the water of the Llobregat River system (NE Spain). The survey was carried out at 30 sites where surface water was sampled on a monthly basis over a period of 2a. The concentration of dissolved SO4 varied from 20 to 1575 mg L−1. Sulphur isotopic compositions clustered in two populations: one – 93% of the samples – had positive values with a mode of +9‰; the other had negative values and a mode of −5‰. Data for δ18OSO4 showed a mean value of +11‰, with no bi-modal distribution, though lower values of δ18O corresponded to samples with negative δ34S. These values can not be explained solely by the contribution of bedrock SO4 sources: that is, sulphide oxidation and the weathering of outcrops of sulphates, though numerous chemical sediments exist in the basin. Even in a river with a high concentration of natural sources of dissolved SO4, such as the Llobregat River, the δ34S values suggest that dissolved SO4 is controlled by a complex mix of both natural and anthropogenic sources. The main anthropogenic sources in this basin are fertilizers, sewage, potash mine effluent and power plant emissions. Detailed river water sampling, together with the chemical and isotopic characterisation of the main anthropogenic inputs, allowed determination of the influence of redox processes, as well as identification of the contribution of natural and anthropogenic SO4 sources and detection of spatial variations and seasonal changes among these sources. For instance, in the Llobregat River the input of fertilisers is well marked seasonally. Minimum values of δ34S are reported during fertilization periods – from January to March – indicating a higher contribution of this source. The dual isotope approach, δ34S and δ18O, is useful to better constrain the sources of SO4. Moreover, in small-scale studies, where the inputs are well known and limited, the mixing models can be enhanced and the contribution of the different sources can be quantified to some extent.