Using oxygen isotopes to establish freshwater sources in Bedford Basin, Nova Scotia, a Northwestern Atlantic fjord

Abstract

A weekly time-series of oxygen isotope (δ18O) measurements was collected over a 16-month period from near-surface (1 m) and near-bottom (60 m) waters of Bedford Basin, a coastal fjord adjacent to the Scotian Shelf, off eastern Canada. The time-series was complemented with δ18O measurements of local precipitation (rain and snow), river, and wastewater runoff. The isotopic composition of precipitation displayed strong seasonality with an average (volume-weighted) δ18O value of −5.39‰ (±0.96) for summer and a depleted value of −10.37‰ (±2.96) over winter. Winter precipitation exhibited more depleted and variable δ18O of solid precipitation relative to rainfall. The annual, amount-weighted average δ18O of Sackville River discharge (−6.49‰ ± 0.82) was not statistically different from precipitation (−7.24‰ ± 0.92), but exhibited less seasonal variation. Freshwater end-members (zero-salinity intercepts) estimated from annual and seasonal regressions of δ18O versus salinity (S) for Bedford Basin near-surface samples were consistent with the δ18O of summer precipitation and the annual, amount-weighted average for the Sackville River. However, the isotopically depleted signature of winter precipitation was not observed clearly in near-surface waters of Bedford Basin, which might reflect isotope enrichment during sublimation from accumulated snowfall prior to melting and discharge, or retention and mixing within the drainage basin. In near bottom waters, most of the δ18O-S variation (average freshwater end-member: 7.47‰ ± 2.17) could be explained by vertical mixing with near-surface waters (average freshwater end-member: −6.23‰ ± 0.34) and hence with locally-derived freshwater. However the near-bottom δ18O-S variation suggested an additional contribution of a freshwater end-member with a δ18O of −15.55‰ ± 2.3, consistent with a remotely-derived freshwater end-member identified previously for the Scotian Shelf. Residuals from a long-term regression of δ18O-S were generally within the range expected due to analytical uncertainty (±0.05); however near-surface waters exhibited seasonal variability of small amplitude, which was consistent with the timing and δ18O variability of local freshwater inputs.