Particulate organic carbon mass distribution at the Bermuda Atlantic Time-series Study (BATS) site

Abstract

Abstract Errors in total particulate organic carbon (total POC) measurements caused by particles settling in Niskin water samplers, loss of bacterial cells during filtration and undersampling of rare particles such as the diazotrophic cyanobacterium Trichodesmium spp. were investigated at the Bermuda Atlantic Time-series Study (BATS) site. Regular core samples of temperature, primary production, bacterial abundance, chlorophyll-a (Chl-a) and POC were collected at monthly intervals from 1991 to 1996. During this period of time, shorter investigations of particles settling in water samples (1991–1992), bacterial cells lost during filtration (1992–1993), and Trichodesmium abundance (1995–1996) were performed at the BATS site. The BATS site shows striking seasonal patterns in hydrography and phytoplankton primary productivity, with a strong maximum immediately following the deep winter mixing of the water column. Following the peak in primary production, bacterial abundance showed only slightly elevated levels in spring. Maxima of Chl-a and POC also were associated with the primary production peaks, but these particle concentrations became less pronounced through summer and fall. An average of 26% of total POC collected in Niskin water bottles settled below the spigot before it could be sampled. An average of 47% of all bacterial cells passed the nominal pore size of a Whatman GF/F filter, and total POC measurements generated from GF/F filtered seawater samples had to be corrected for this loss. The average integrated stocks of total POC in the upper 65 m of the water column was 32% pigmented phytoplankton, 15% microheterotrophs, 54% other detrital matter (32 : 15 : 54). Phytoplankton C equaled bacterial C in the 65–135 m depth range (16 : 19 : 65), but phytoplankton C was virtually non-existent deeper than 135 m (2 : 14 : 74). Bacterial C biomass was higher than phytoplankton in surface waters outside the spring bloom period, but carbon not accounted for by phytoplankton and bacteria (other C) showed an overall dominance throughout the year. Uncorrected, suspended POC collected on GF/F filters (POCSW) was nearly equal to the sum of phytoplankton C and bacterial C alone, and hence, the other C fraction of total POC was largely generated by the addition of settling particles (POCDreg). Seasonal occurrences of rare particles such as Trichodesmium colonies in surface waters in late summer may account for as much as 17–56% of total POC. Settling particles and Trichodesmium colonies, seldom included in POC estimates from temperate and tropical regions, constituted more than half of total POC measured in surface waters at BATS.