a: Graduate College of Marine Studies, University of Delaware, Lewes, DE, 19958, USA
b: Department of Ecology, Evolution and Marine Biology, University of California, Santa Barbara, CA, 93106-9610, USA
c: Monterey Bay Aquarium Research Institute, 7700 Sandholdt Road, Moss Landing, CA, 95039-9644, USA.
Marine Chemistry (2002) 77: 239 253.
Received: 2001 July 19.
Revision received: 2001 December 3.
Accepted: 2001 December 3.
A broad community intercalibration exercise for accurate measurement of dissolved organic carbon (DOC) in seawater has been carried out over a period of 5 years. A set of 10 natural samples with DOC content from 40 to 200 µM C were accompanied by two glucose standards and a "zero C" blank; all sealed in glass ampoules. Samples were sent to all interested analysts for "blind" analysis; 62 laboratories in 17 countries participated. A total of 59 separate analyses were determined to be acceptable by screening criteria based on standards and blank; another nine sets of analyses did not pass the screening. The majority of the analyses, both those passing and those that did not, were performed with high temperature combustion (HTC) methods, six sets of analyses were done using wet chemical oxidation methods. From the 53 sets of acceptable HTC analyses, the coefficient of variation (%CV) for analytical comparability of the samples was 10% ("community precision"). It is estimated that the individual replicate injection precision for most instruments was approximately 2% and that no additional variability was caused by differences within the ampoules of individual samples. The additional variability over 2% was likely a result of both random and systematic differences in analytical capabilities from instrument to instrument and from day to day for individual instruments. With an arbitrary selection after the fact, smaller subsets of analysts can show comparability better than 10% and duplicate or triplicate runs on different days of the full sets of samples in several laboratories showed comparability in the 26.5% range. Experienced oceanic analysts, with internal or shared reference materials, can now show reproducibility and comparability at a level closer to 2%. Preliminary use of DOC reference materials by 14 participants showed day-to-day reproducibilities for their laboratories in the 26% range in most cases; several with poorer reproducibility do not normally perform DOC analyses on samples with concentrations as low as the deep ocean reference used here. Use of these reference materials can also give a demonstration of comparability between laboratories. For credibility of DOC analyses, it is necessary for analysts to use community reference materials and report results of their analytical performance with these references. This paper does not identify individual data nor should it be considered an evaluation of individual laboratories or analysts. The purpose is to show the summary picture of the international community of DOC analysts as it existed in the mid- to late 1990s.
© 2002 Published by Elsevier Science B.V.
The effort reported on here was possible only through the generous cooperation of many analysts over a period of about 5 years. The laboratories and analysts listed in Table 2 were willing to run the samples in a blind format and report the raw data to the first author. We thank them very much for their assistance. We thank Lenore Bennett for assistance with preparation of the samples and shipping of the early sets of ampoules sent out.
In addition to the contributions by those who ran the set of samples, several analysts reported data on their use of the DOC reference materials, the results of which are listed in Table 6. In addition to some who are co-authors of this paper (KBS, KRR), we acknowledge Georgina Spyres, Axel Miller, David Burdige, Rainer Amon, Nancy Kaumeyer, Penny Vlahos, Renee Styles, Xose Alvarez-Salgado, and Erik Lundberg (affiliations shown in Table 2) and also Javier Aristegui (of Universidad de Las Palmas de Gran Canaria) for contributing information on reference use.
This research was partially supported by NSF grants OCE 92-00970, OCE 94-02859, OCE 96-33926, and OCE 00-82238 to JHS, by OCE 96-17795, OCE 96-19222, and MCB 99-77918 to CAC and by OCE 91-15201, OCE 92-03953, OCE 93-10719, and OPP 95-30609 to ETP.