Fluxes of organic carbon normalised to a depth of 1000 m from 18 sites in the Atlantic and the Southern Ocean are presented, comprising nine biogeochemical provinces as defined by Longhurst et al. (1995. Journal of Plankton Research 17, 1245-1271). For comparison with primary production, we used a recent compilation of primary production values derived from CZCS data (Antoine et al., 1996. Global Biogeochemical Cycles 10, 57-69). In most cases, the seasonal patterns stood reasonably well in accordance with the carbon fluxes. Particularly, organic carbon flux records from two coastal sites off northwest and southwest Africa displayed a more distinct correlation to the primary production in sectors (1 x 1°) which are situated closer to the coastal environments. This was primarily caused by large upwelling filaments streaming far offshore, resulting in a cross-shelf carbon transport. With respect to primary production, organic carbon export to a water depth of 1000 m, and the fraction of primary production exported to a depth of 1000 m (export fraction=EF1000), we were able to distinguish between: (1) the coastal environments with highest values (EF1000=1.75-2.0%), (2) the eastern equatorial upwelling area with moderately high values (EF1000=0.8-1.1%), (3) and the subtropical oligotrophic gyres that yielded lowest values (EF1000=0.6%). Carbon export in the Southern Ocean was low to moderate, and the EF1000 value seems to be quite low in general. Annual organic carbon fluxes were proportional to primary production, and the export fraction EF1000 increased with primary production up to 350 gCm-2 yr-1. Latitudinal variations in primary production were reflected in the carbon flux pattern. A high temporal variability of primary production rates and a pronounced seasonality of carbon export were observed in the polar environments, in particular in coastal domains, although primary production (according to Antoine et al., 1996. Global Biogeochemical Cycles 10, 57-69), carbon fluxes, and the export fraction remained at low.
