Determination of particulate organic carbon (POC) in seawater: The relative methodological importance of artificial gains and losses in two glass-fiber-filter-based techniques

Abstract

Particulate matter in aquatic systems is an important vehicle for the transport of particulate organic carbon (POC). Its accurate measurement is of central importance for the understanding of marine carbon cycling. Previous work has shown that GF/F-filter-based bottle-sample-derived concentration estimates of POC are generally close to or higher than large-volume in-situ-pump-derived values (and in some rare cases in subzero waters are up to two orders of magnitude higher). To further investigate this phenomenon, water samples from the surface and mid-water Northeast Atlantic and the Baltic Sea were analyzed. Our data support a bias of POC concentration estimates caused by adsorption of nitrogen-rich dissolved organic material onto GF/F filters. For surface-ocean samples the mass per unit area of exposed filter and composition of adsorbed material depended on the filtered volume. Amounts of adsorbed OC were enhanced in the surface ocean (typically 0.5 μmol cm− 2 of exposed filter) as compared to the deep ocean (typically 0.2 μmol cm− 2 of exposed filter). These dependencies should be taken into account for future POC methodologies. Bottle/pump differences of samples that were not corrected for adsorption were higher in the deep ocean than in the surface ocean. This discrepancy increased in summer. It is shown that POC concentration estimates that were not corrected for adsorption depend not only on the filtered volume, true POC concentration and mass of adsorbed OC, but also on the filter area. However, in all cases we studied, correction for adsorption was important, but not sufficient, to explain bottle/pump differences. Artificial formation of filterable particles and/or processes leading to filterable material being lost from and/or missed by sample-processing procedures must be considered. It can be deduced that the maximum amounts of POC and particulate organic nitrogen (PON) that can be artificially formed per liter of filtered ocean water are ∼ 3–4 μM OC (5–10% of dissolved OC) and ∼ 0.2–0.5 μM ON (2–10% of dissolved ON), respectively. The relative sensitivities of bottle and pump procedures, and of surface- and deep-ocean material, to artificial particle formation and the missing/losing of material are evaluated. As present procedures do not exist to correct for all possible biasing effects due to artificial particle formation and/or miss/loss of filterable material, uncertainties of filtration-based estimates of POC concentrations need further testing. The challenge now is to further constrain the magnitude of the biasing effects that add to the adsorption effect to reduce the uncertainties of estimates of POC concentrations, inventories and fluxes in the ocean.