Discriminatory classification of natural and anthropogenic waters in two U.K. estuaries

The ability to distinguish water inputs from both natural and anthropogenic sources was investigated in the complex environment of an urban estuary (Tyne) and a relatively pristine estuary (Tweed). We used a data set from a total of 11 estuarine transects, comprising measurements of bulk dissolved organic matter (dissolved organic carbon and nitrogen), dissolved nitrogen (total dissolved nitrogen, ammonium, nitrate+nitrite and dissolved organic nitrogen), optical absorbance measurements (a350, S290–350) and fluorescence excitation emission matrix measurements (fluorophores A, H, B and T intensity and A and H emission wavelength maxima). In order to investigate trends within the numerous parameters measured, multivariate statistics were employed. Principal components analyses showed 63.4% of the variability in the total data set can be explained by two sets of components and 74.9% of the variability by the spectrophotometric measurements alone. In both analyses the first component correlated to the mixing of terrestrial and marine waters and the second component was correlated to sources of pollution such as domestic sewage. Within the data set, river flow and terrestrially derived DOM were significantly correlated, and situations with high river input showed an increase in terrestrial signature in the estuary. Discriminant analyses were also carried out and indicated that 59.8% (total data set) and 53.3% (solely spectrophotometric data) of the samples can be correctly classified into their respective groups (water categories) assigned on the basis of salinity and sampling location. Overall the results clearly show the potential of spectrophotometric techniques to discriminate distinct water categories with different DOM characteristics. In particular, measurement of the fluorophore H emission maxima, the spectral slope parameter, S290–350, and fluorophores T and B intensity enabled discrimination of DOM from riverine, estuarine, marine, and sewage affected water categories. The results presented here indicate the ability of spectrophotometric data alone to distinguish between marine, anthropogenic and terrestrial DOM and distinguish terrestrial DOM from different catchments (Tyne vs. Tweed). With current advances in the in-situ deployment of absorbance and fluorescence spectroscopy it is anticipated that multivariate statistics will gain importance as a cost effective, powerful and diagnostic approach to assessing the distributions of water types and their associated DOM characteristics and fluxes at the land–ocean interface.