Why do POC concentrations measured using Niskin bottle collections sometimes differ from those using in-situ pumps?

Systematic discrepancies between concentrations of particulate organic carbon (POC) measured using bottles (small volume) versus in-situ pumps (large volume) have long been of concern to oceanographers. Usually bottle samples yield higher values of POC than pump samples taken concurrently. Several factors, including pressure effects during filtration, particle formation during handling, contamination, and adsorption of dissolved organic carbon (DOC) onto filters, have been suggested as responsible for these differences. As part of a larger study of particles and particle fluxes in the Mediterranean Sea (MedFlux), we systematically and contemporaneously measured various factors that could allow comparison of particle types. The comparison of chlorophyll (Chl-a) and POC immediately suggested a disconnection between concentrations of carbon and Chl-a: bottle and pump estimates of Chl-a concentrations agreed at all depths, while POC and particulate organic nitrogen (PON) measurements did not, especially in the upper 50 m of the water column. These observations suggest that pumps and bottles are equally adept at capturing Chl-a, but that Chl-a:C and Chl-a:N are collected in different proportions by the two methods. Additional experiments using samples collected from the coastal waters of Long Island Sound and in the open Mediterranean, and using algal cultures of both non-fragile Thalassiosira pseudonana 3H and fragile Isocrysis galbana cells, showed no effects of filtration pressure on POC, PN, or Chl-a concentrations over the pressure range of 0.2-0.9 atm. In addition, shaking experiments using bottle samples from the Mediterranean did not show particle formation during handling. Corrections for DOC adsorption on the filters were made and cannot account for the discrepancy between pump and bottle POC. One possible explanation for the pump-bottle POC discrepancy is that bottle samples capture zooplankton that can evade pumps but not bottles. In support of this hypothesis, comparisons of zooplankton caught in Niskin bottles with those retained on >70 μm in-situ pump filters, showed greater abundances in bottle relative to pump samples. In addition, non-swimming zooplankton (radiolaria) were under-collected by pumps, suggesting that washout of particles off the 70 μm mesh of in-situ pumps during use or recovery may also cause lower POC values. This possibility needs to be further investigated. This study does not address whether zooplankton should or should not be included as POC, but only suggests constraints on the difference between the two methods.