We used stable carbon (l3C/I2C) and nitrogen ( f S~/ ' " ) isotope analysis to investigate linkages between sources of primary production and the pelagic and benthic components of the Northeast Water (NEW) Polynya off northeastern Greenland. Ice algae was enriched in 13C (mean 6I3C = -18.6 vs -27.9%) and I5N [mean 6 1 5~ = 8.3 vs 4.9%0) over particulate organic matter [POM) suggesting that the relative importance of these sources might be traced isotopically. Most grazing crustaceans and filter-feeding bivalves had 6I3C and 6 1 5~ values in the range of -21 to -23% and 7 to 9x0, respectively, indicating a direct pathway from POM. Close benthic-pelagic coupling was also confirmed for other benthic organisms examined with the exception of the predatory or deposit feeding echinoderms Ophioctin, Ophiacantha and Pontaster. Compared with other Arctic and temperate marine food webs, stable-carbon isotope values for the NEW Polynya were depleted in I3C. A S1'N trophic model that incorporated taxon-specific isotopic fractionation factors indicated that the NEW Polynya consisted of 4.5 to 5 trophic levels. Stable-isotope analysis Inay be well suited to establ~shing the importance of polynyas as sites of high primary productivity and tight benthic-pelagic coupling relative to regions of more permanent ice cover.

Wllson (1986) has criticized the application of a 3-level interactive model, epibenthic predators -predatory infauna -other infauna, to marine soft-bottom communities. His model, though inexact, shows that preferential predation on predatory infauna by epibenthic predators is a requirement of the 3-level model. His reanalysis of the caging studies used to test the 3-level model, however, is unacceptable as it analyzes only 32 % of the original data set and ignores predat~on by omnivores which is known to be important in structuring some soft-bottom communities. Despite its shortcomings. Wilson's reanalysis still supports a 3-level model. Experiments have demonstrated that many softbottom communities are more, not less, complex than the 3-level model indicates. Strict application of a 2-level model risks missing important interactions among the infauna.