Research Dynamic Colloid-Water Partitioning of Pyrene through a Coastal Baltic Spring Bloom O ¨ RJAN GUSTAFSSON,* NINA NILSSON, AND THOMAS D. BUCHELI Institute of Applied Environmental Research (ITM), Stockholm University, 10691 Stockholm, Sweden
Colloidal organic particles constitute the dominant portion of particulate organic matter in surface seawater, but their influence on the phase speciation and bioavailability of hydrophobic organic compounds (HOCs) is sparsely evaluated. Studies on colloid-water partitioning have been focused on other regimes and have largely been performed on chemically defined subportions of total colloids such as the humic fraction. Available estimates of colloid-water partition coeffficients (Kcoc) are highly variable and not easily explained by regularly applied Kow-Koc relationships. Here, pyrene was partitioned to bulk natural colloids isolated using cross-flow ultrafiltration techniques from the surface water of a coastal bay. A key objective was to elucidate biogeochemical controls on the changing colloidsorbent qualities over the course of the dynamic allochtonous-autochtonous transition of a well-constrained boreal coastal spring bloom. The pyrene Kcoc was found to decrease from 12.9 ( 0.9 × 103 Lw/kgoc in the terrestrial runoff dominated regime to values around 2.9 ( 0.7 × 103 Lw/kgoc, once phytoplankton production became the governing source of organic matter to the surface waters. The changing Kcoc was well correlated with the molar extinction coefficient at 280 nm of the colloidal organic carbon. This study supports other reports of an improved prediction of HOC phase speciation through this simple molecular proxy of the “quality” of organic sorbents. While being poor sorbents on a carbon atom basis, relative to soils and sediments, coastal marine colloids, by their shear abundance, may significantly attenuate the truly dissolved exposures of HOCs with log Kow above 5.
Introduction Colloidal organic carbon (COC) is regularly around 10 times more abundant than larger “particulate” organic carbon (POC; filter-defined as > 0.7 µm) in surface seawater (e.g., refs 1 and 2). It is also well known from laboratory experiments that bioaccumulation of hydrophobic organic compounds (HOCs) in planktons and nectons is reduced in the presence of such filter-passing sorbents (e.g., refs 3 and 4). Yet, there is a deficiency of quantitative understanding of colloidal influence on the phase speciation and bioavailability of HOCs in the marine environment. Detailed evaluation of the phase distribution of HOCs in the portion of any aquatic system that pass through conventional (i.e., 0.2-0.7 µm) filters is experimentally * Corresponding address phone: +46-86747317; fax: +4686747638; e-mail:
[email protected]. 10.1021/es0003019 CCC: $20.00 Published on Web 09/19/2001
2001 American Chemical Society
challenging. Most available data on partition coefficients with submicron sorbents stem from HOC-spiked binding experiments. The resulting colloid-water distribution has then either been observed directly, by the proportional quenching of the probe’s fluoresence upon colloid binding (e.g., refs 5-8), or was obtained after separation of the phases on solidphase extraction columns (e.g., refs 9 and 10). Several studies have also made use of the apparent solubility enhancement (e.g., refs 11 and 12) and the reduced headspace activity (e.g., ref 13) resulting from HOC-colloid association. The majority of “colloid” partitioning studies to date have been performed on humic substances that have been either produced commercially under an undisclosed procedure (e.g., Aldrich humic acid) or represent an operationally defined and chemically isolated subportion of the total colloidal matter in natural waters. It has recently been demonstrated that obtained humic substance partitioning coefficients are highly dependent on the chosen isolation procedure (e.g., ref 14). For seawater, the conventional “humic” isolation procedure has been adsorption of acidified water onto nonionic XAD resins (e.g., ref 15). However, this method typically only recovers a minor fraction (5-15%) of the total filter-passing organic carbon in seawater, requires large manipulations of pH during isolation, and is selective for hydrophobic constituents. The current study is focused on HOC partitioning with natural organic colloids in surface seawater using less intrusive techniques. Largely as a result of the historical difficulty to physically separate the filter-passing sorbents from the aqueous solution, experimental distribution coefficients have frequently been normalized to the bulk filter-passing organic carbon, commonly referred to as “dissolved” organic carbon (“DOC”). As espoused below, this fraction does contain both organic matter that is truly dissolved in the aqueous phase (i.e., DOC) as well as COC (i.e., “DOC” ) DOC + COC). Bulk filter-passing “DOC”-normalized partition coefficients (i.e., K“DOC”) for HOCs in many natural water systems are seen to vary over orders-of-magnitude for any individual compound. They are generally also significantly lower than predictions from their octanol-water partition coefficients (Kow) and partitioning relationships developed primarily for natural organic matter of sediments and soils (reviewed in refs 16 and 17). There are several potential explanations to the more scattered and less energetic partitioning to suspended organic carbon compared to organic matter from sediments and soils. In contrast to the latter, which has undergone structural homogenization through diagenesis (18), it is conceivable that physical and chemical properties of, diagenetically younger, suspended organic macromolecules are more heterogeneous. As discussed in detail elsewhere (19), to qualify as a natural colloidal sorbent an organic macromolecule must be sufficiently large to physically accommodate a hydrophobic compound. With total surface areas of common HOCs such as polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs) of around 2 nm2, this criterion translates into macromolecular weights of 1000-10 000 D. Hence, smaller organic substances as well as macromolecules with extended tertiary configurations without an interior “core”, of at least such dimensions, may not absorb HOCs. Consequently, it is reasonable to employ 1000 D as a conservative point of reference to discriminate between DOC and COC, where COC obviously represents the sorbing portion of the “DOC”. Since COC normally is found to make up only 10-40% of the “DOC” (e.g., refs 2 and VOL. 35, NO. 20, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4001
20), this size constraint may in part explain the low K“DOC”. Solid-state NMR studies of seawater COC are suggesting that lipopolysaccharidic biopolymers are making up a large fraction of marine colloids (e.g., refs 2 and 21). Garbarini and Lion (22) have demonstrated that structurally related cellulose is rather inefficient as an HOC sorbent. Hence, on a carbon atom basis, it is reasonable to anticipate a less energetic interaction with seawater colloids than with organic matter in sediments and soils. The current study seeks to alleviate the dearth of information on the behavior and role of abundant seawater colloids as a partitioning medium for HOCs. To this end, we have probed the dynamic sorbing nature of surface water colloids in a coastal bay over the winter-to-summer transition, with the sources of organic matter shifting from terrestrial runoff to becoming dominated by autochtonous phytoplankton production.
Experimental Section Study Sites and Sampling Strategy. To perform the colloid binding experiments at different stages of the spring bloom, a suite of biogeochemical parameters were monitored approximately every third day between January 19 to May 19, 2000 in the surface waters of Ekhagen Bay. This semienclosed coastal system is located just north of Stockholm on the Swedish coast of the northwestern Baltic Sea (59°22.54N 18°03.58E) and has been described elsewhere (23). Briefly, samples were retrieved from 4-m depth in 14-m water column from a pier extending 40 m from the moderately inhabited and forested shore of the “Ekoparken” National Park. Water samples for ancillary parameters were obtained with an in situ prerinsed 5-L Ruttner collector. The 20-L samples for the colloid sorption experiments were pumped from 4-m depth using a PVC line as described earlier (23). All samples were immediately transported to the Stockholm University laboratory 200 m away and handled within 2 h. A Hydrolab Surveyor Probe was used to define depth profiles of oxygen saturation, conductivity, salinity, temperature, and pH. Between January 25 and March 7, the bay was frozen over with ice, and samples were collected through a 25-cm diameter drilled hole. To constrain end-member compositions and sorbent behavior, we also sampled the drainage water to the Ekhagen Bay, the underlying sediment porewater, and exudates from a common phytoplankton species grown in a laboratory culture. The Igelba¨cken creek, 20-L sampled October 31, 2000, drains a mixed forest and agricultural land before emptying into the Ekhagen Bay. Surficial bottom sediments of the Ekhagen bay was retrieved in November 1999 using an Ekman dredge and stored frozen (-20 °C) in amber glass jars. The thawed sediment (organic carbon content of 8.1%) was centrifuged in an MSE Mistral centrifuge at 2000 rpm for 30 min. The supernatant porewater was decanted and filtered through a precombusted 0.7 µm glass fiber filter (GF/F; Whatman Inc.). Finally, we studied the binding behavior of exudates from a laboratory culture of the diatom Thalassiosira weissflogii, ubiquitously present in the Baltic Sea (e.g., ref 24). Cultures were prepared in filtered, and twice autoclaved, 10‰ salinity Baltic seawater with the F/2 medium prepared according to Guillard and Ryther (25). Biogeochemical Parameters. Descriptive parameters of both the natural system and its sorbent characteristics were constrained over the course of the study. Twelve milliliter samples for total organic carbon (TOC) as well as for isolated DOC (
