Variation in the seston C:N ratio of the Arctic Ocean and pan-Arctic shelves

Studying more than 3600 observations of particulate organic carbon (POC) and particulate organic nitrogen (PON), we evaluate the applicability of the classic Redfield C:N ratio (6.6) and the recently proposed Sterner ratio (8.3) for the Arctic Ocean and pan-Arctic shelves. The confidence intervals for C:N ranged from 6.43 to 8.82, while the average C:N ratio for all observations was 7.4. In general, neither the Redfield or Sterner ratios were applicable, with the Redfield ratio being too low and the Sterner ratio too high. On a regional basis, all northern high latitude regions had a C:N ratio significantly higher than the Redfield ratio, except the Arctic Ocean (6.6), Chukchi (6.4) and East Siberian (6.5) Seas. The latter two regions were influenced by nutrient-rich Pacific waters, and had a high fraction of autotrophic (i.e. algal-derived) material. The C:N ratios of the Laptev (7.9) and Kara (7.5) Seas were high, and had larger contributions of terrigenous material. The highest C:N ratios were in the North Water (8.7) and Northeast Water (8.0) polynyas, and these regions were more similar to the Sterner ratio. The C:N ratio varied between regions, and was significantly different between the Atlantic (6.7) and Arctic (7.9) influenced regions of the Barents Sea, while the Atlantic dominated regions (Norwegian, Greenland and Atlantic Barents Seas) were similar (6.7-7). All observations combined, and most individual regions, showed a pattern of decreasing C:N ratios with increasing seston concentrations. This meta-analysis has important implications for ecosystem modelling, as it demonstrated the striking temporal and spatial variability in C:N ratios and challenges the common assumption of a constant C:N ratio. The non-constant stoichiometry was believed to be caused by variable contributions of autotrophs, heterotrophs and detritus to seston, and a significant decrease in C:N ratios with increasing Chlorophyll a concentrations supports this view. This study adds support to the use of a power function model, where the exponent is system-specific, but we suggest a general Arctic relationship, where POC=7.4 PON^0.89.