PROCESSES CONTROLLING THE ACCUMULATION OF DIATOMS IN SEDIMENTS: A MODEL DERIVED FROM BRITISH COLUMBIAN FJORDS

The transformation of biocoenosis to taphocoenosis of diatom assemblages was examined using monthly sediment traps over a 3-year period at four sites in the fjords of British Columbia. The fjords were chosen as a natural laboratory where productivity is high and seasonally variable and where geological rates of sediment accumulation approach the biological rates of production and flux. The composition of trap samples from different depths was quantified and compared with the surface sediment, recovered by box cores. Fragmentation caused by zooplankton grazing preferentially destroys 2050% of the early spring and 5075% of the late spring and early summer production. Grazing activity and incorporation into pellets does not otherwise affect preservation. Differential dissolution occurs during settling through the water column; taxa with a high surface-to-volume ratio are preferentially lost. Low-level lateral advection tends to concentrate heavy, robust taxa, while strong advection affects all particles equally and has no net biasing effect. The interaction of these processes results in a sediment assemblage in which winter and fall taxa are more common than they are in the plankton, while the early spring bloom is badly underrepresented. Multivariate analyses indicate that the sediment is dominated by the early summer bloom, but caution must be exercised in this interpretation, because the bloom is characterized by a species (Skeletonema costatum) which is common at all seasons of the year and therefore is not a reliable indicator of a single period of production. The end result is a decrease in seasonal contrast and at some sites a reordering of seasonal proportion in the sediment. While the seasonal composition can still be identified, there is also evidence of homogenization of sediment assemblages, with some of the original spatial variability obscured. Taken to an extreme, a unique fossil assemblage may be produced which cannot be statistically related to any combination of seasons of production. Three factors which might affect preservation of deep-sea assemblages (water depth, accumulation rate, and degree of bioturbation) do not have a measurable effect in fjord environments. Strong modification of assemblages can occur within the upper 500 m; deep-sea monitoring programs which do not include this interval may lead to erroneous conclusions concerning production and flux to the sediments.