Thesis (Ph. D.)--Leland Stanford junior University, 1915.

The main goal of this thesis was to develop a multi-technique approach to characterise high molecular weight (colloidal) dissolved organic matter (DOM) occurring in estuarine and coastal marine environments and to investigate how these bulk properties may relate to their ability to influence the physicochemical speciation of metals such as iron and copper. This characterisation was undertaken systematically on the pre-filtered (0.4 μm) estuarine or marine sample, referred to here as the ‘bulk’, as well as on the ultrafiltered (< 5 kDa) and colloidal (> 5 kDa) fractions. Three successive studies were carried out. The first one took place in the estuarine mixing zone of a high pH, low turbidity black water river (River Thurso) and the other two in the receiving coastal waters (Thurso Bay). The optical and size distribution properties of these waters were examined in detail in the first study while their iron and copper-binding properties were examined in the second and third studies, respectively. Size fractionation results showed that the proportion of colloidal to soluble organic carbon (DOC) tended to decrease from the upper estuary (~ 60%) to coastal waters (~ 20%). With respect to trace metals, similar trends were observed as one progressed from the mouth of the river towards the open sea. In relation to their molecular absorption, fluorescence and size fractionation properties, both River Thurso DOM and Nordic Reservoir NOM Reference Material presented the same characteristics and mixing behaviour, indicating the dominance of humic and fulvic substances in the Thurso river-ocean system. The river-borne, humic colloids underwent two types of transformations upon mixing with the seawater end-member. The first one was the coiling or contraction of individual macromolecules ―monomers― with increasing salinity. The second one was the concurrent aggregation of these small monomer units (d = 2 – 4 nm) to form entities that were still colloidal, i.e. smaller than 0.4 μm. As a result of extensive association of iron and copper with the colloidal and soluble fractions respectively, not only organic carbon but also iron and copper behave conservatively in the River Thurso estuary. Throughout the coastal region of freshwater influence (S = 29 - 35) colloidal iron accounted for 30 - 80% of total dissolved iron and was present as iron-humic complexes supplied by the river and showing a uniform stability constant (log KFe’HS’ = 11.3  0.1, i.e. log KFe3+ HS’ = 21.3  0.1). Soluble iron was found to be largely complexed to ligands of marine origin with log K’Fe’HS’ = 11.9  0.1, thus revealing for the first time a difference between the iron-binding strengths of colloidal and soluble ligands. Terrestrial colloidal iron was found to be entirely, if slowly (~ 10 hours), accessible to the added competing ligand 2-(2-thiazolylazo)-pcresol (TAC) used for the determination of K’Fe’HS’. Furthermore, iron appeared to play a role in holding these terrestrial colloids together. Evidence for this came from variations in humic fluorescence intensity over time in response to dissociation of the ironhumic colloidal associations induced by a chelating resin. These results are consistent with the concept that iron derived from a peat-draining river is strongly but reversibly bound to humic substances and remains so under marine conditions. In the same coastal region, two types of high-affinity ligands binding over 99.99% of total copper were detected. The stronger ligand (L1, log KCu2+L1’ = 15.5 - 16.1), of riverine origin, was present in very low concentrations of 1 - 4 nM that correlated with ―but systematically fell short of― total copper concentrations. Its conditional binding constant tended to increase with salinity, with most of the increase taking place in the near-field portion of the river plume. The weaker, more abundant ligand (L2, log KCu2+L2’ = 11.8 - 12.8) was present in total concentrations of 60 – 170 nM and had a controlling influence on the value of the labile (i.e. inorganic) copper concentration which ranged from 0.001 to 0.0001 nM. Both organic ligands were fairly evenly partitioned between soluble and colloidal phases but their sources appeared to differ significantly. Ligand L1 appeared to be of riverine origin (although one cannot exclude active microbial production, as opposed to passive release from peat) while ligand L2 distributions suggested in situ production within Thurso Bay.