Insights into Southern Ocean carbon export from the delta C-13 of particles and dissolved inorganic carbon during the SOIREE iron release experiment

C-13 contents of total dissolved inorganic carbon (delta C-13-DIC) and particulate organic matter (delta C-13(org)) were determined to examine the factors influencing phytoplankton C-13 contents and carbon export from the SOIREE iron-induced algal bloom. Suspended particles sieved into 200, 70, 20, 5, and 1 micrometer(mum) size classes displayed an extremely large range in delta C-13(org) Of 8 parts per thousand. delta C-13(org) values increased from - 28 parts per thousand for the 1-5 mum class to a maximum of - 20 parts per thousand for the 20-70 mum class, which was dominated by the large pennate diatom Fragilariopsis kerguelensis. Larger particles (70-200 and > 200 mum) had similar delta C-13(org) to the smaller (1-5 and 5-20 mum) particles, reflecting both the presence of long narrow Thalassiothrix antarctica diatoms and zooplankton that grazed on small phytoplankton. Comparison of results inside and outside the bloom identified cell surface/volume ratio (mainly reflecting cell size) as the dominant control of phytoplankton delta C-13(org), with subsidary roles for growth rate and seawater [CO2](aq). The SOIREE iron fertilization provoked an increase in the proportion of large ( > 20 mum) diatoms. This increased the delta C-13(org) of the bulk suspended particles within the mixed layer, but there was minimal increase in the delta C-13(org) of sub-surface suspended particles and negligible change in the delta C-13(org) of particles obtained with sediment traps suspended below the bloom. This suggests that there was no increase in carbon export over the similar to 13 day observation period. However, comparison to delta C-13(org) results from previous voyages, and to vertical changes in delta C-13-DIC, suggests that large diatoms control carbon export from the Antarctic Zone over the summer growth season. This result must be viewed with great caution as it is based on very sparse data and involves several assumptions. (C) 2001 Elsevier Science Ltd. All rights reserved.