Low dissolved DMSP concentrations in seawater revealed by small-volume gravity filtration and dialysis sampling

A variety of filtration and passive dialysis protocols were tested with seawater for determination of the dissolved concentrations of dimethylsulfoniopropionate ( DMSPd), a compound originating in the cytosol of many marine phytoplankton. Commonly used sampling procedures, such as in-line filtration, syringe pressure filtration, and gravity filtration of relatively large-volume samples ( 5-50 mL) through glass fiber filters, caused release of DMSPd from particulate material as evidenced by increasing DMSPd concentrations with volume filtered. Exposure of filters to air at the end of filtration caused particularly severe DMSPd release. Dialysis bags or Slide-A-Lyzer dialysis cassettes equilibrated with seawater DMSPd within 4 h, even in polar waters ( -1.8 degrees C), and appeared to give accurate DMSPd concentrations in some circumstances. However, incubation of seawater in laboratory containers ( e. g., glass jars) during dialysis sometimes caused artifactual release of DMSPd. We therefore adopted a small-volume gravity drip filtration ( SVDF) procedure, the essential elements of which were: ( i) collection of 20 to 50 mL seawater directly from the primary sample container ( e. g., Niskin bottle) into a dry, all-plastic filtration tower; ( ii) use of a 47-mm-diameter Whatman GF/F filter; ( iii) rapid (< 3 min) filtration by gravity ( hydrostatic) pressure; ( iv) collection of only the first 3.5 mL filtrate for DMSPd analysis; and ( v) never exposing the filtered plankton to air. The SVDF procedure appeared to yield the same low DMSPd concentrations as dialysis samplers incubated in the ocean water column. Using the SVDF procedure, DMSPd was found to be < 2.8 nM over a broad range of ocean water types and particulate DMSP concentrations. The maximum DMSPd concentration observed in our study ( 2.8 nM) was far lower than the reported worldwide average DMSPd concentration of 16.9 nM, raising the possibility that past data collections may have been influenced by filtration artifacts.