MATERIAL AND METHODS

Surface sediments were sampled off Namibia with multicorers during the RV Meteor cruises M48/2 in the year 2000 and M57/3 in 2003 (cruise reports in Emeis 2002, Brüchert 2005, respectively). The water depth at the stations ranges from 28 to 152 m (Table 1, Figure 1). Samples were taken in steps of 1 cm from one to four core tubes, depending on availability and recovery. Densely suspended or flocculent matter within the lowermost supernatant bottom water was collected with a spoon and added to the surface sample. Three multicorer tubes were sampled down core in larger sections, lumping together layers of homogenous sediment facies. All samples were stained onboard with a solution of 2 g Rose Bengal (C.I. 45440) l^-1 ethanol immediately after recovery. This solution does not affect subsequent isotopic measurements (Serrano et al. 2008). About 200 cm³ of the solution were added per 100 cm³ sample wet volume in a translucent plastic container (Kautex-bottle). The penetration of the preservative was enhanced by softly shaking the closed container, until no more unstained sediment lumps or bacterial mucus tubers attached to its wall. The best staining results with Rose Bengal are obtained if the sediment is exposed to the stain for three (sandy or silty sediment) or six weeks (sticky clay). Shorter exposure will reduce staining efficiency (Lutze and Altenbach 1991). This methodology was cross-checked by comparison of biomass measurements of stained foraminifera and living foraminifera picked from fresh sediments, both sampled from periodically dysoxic sediments of the Baltic Sea (Lutze and Altenbach 1991). These control measurements gave proof for a reliability of the detection of living, protoplasm filled tests above the 96 % level. The usage of lowered ethanol concentrations or water as a solvent for Rose Bengal, other preservatives, or short-term staining, will adversely affect staining efficiency (Lutze and Altenbach 1991). During cruise M48, the Tübingen working group isolated living foraminifers with protruding reticulopodia from our sampling sites for DNA analysis (Ertan et al., 2004).

Gas bubbles were observed in several samples during recovery, producing an intense smell of hydrogen sulfide, sometimes even foaming up within the multicorer tubes. In some instances, excess sediment falling from the multicorer tubes reduced the wooden ship planks within minutes (Appendix). A sediment volume correction was deduced in compensation of either the additional matter sampled from the nepheloid layer, or for the expanding gas volumes. After storage for several weeks, all particles had settled below the supernatant liquid, leaving a clear, translucent Rose Bengal solution. The sediment surface, clearly visible through the translucent sample container, was marked with a water- resistant felt tip. Subsequent to pouring the sample on the sieve, this mark was refilled with water for the detection of the settled and gas free sample volume. This procedure led to corrections ranging from +10 to -40 % of the theoretical volume calculated based on the diameter of the multicorer. The resulting sample volumes of approximately 100 to 300 cm³ are sufficient for the detection of a standing stock (stained tests) down to 0.1-0.03 specimens per 10 cm³. Samples were wet sieved for size fractions >63 µm and >250 µm, and dried at 40ºC. The fraction 150 to 250 µm was dry sieved from the fraction 63-250 µm.

Size fraction >250 µm was counted in all surface samples. This size fraction rules out, or at least reduces, increasing lateral advection for decreasing test sizes (Lutze 1980). The size fraction 150-250 µm was counted in all samples from cruise M57/3, in order to estimate the imprint of smaller size classes. Counting this size fraction was much more time-consuming in comparison to the fraction >250 µm. We estimate a factor of about two to 10 times, depending on the increasing number of diatom frustules. The size fraction 63 to 150 µm of the DMB was impossible to count quantitatively because of the absolute dominance of diatom frustules. Other particles were only occasionally found on picking trays. Organic debris and bacterial mucus caught in a fabric with diatoms resulted in large aggregates in the dried sample. Splitting this small fraction in a Micro-Splitter led to static adhesion of diatoms on the walls of the splitter and the containers, and clouds of diatom frustules floated in the air for seconds. Bremner (1983) was successful in point counting the >63 µm fraction of the DMB by treating the samples with fresh water onboard and storing them for longer periods under room temperature. We presume that the chemical breakdown of the organic matter before drying was crucial for his success. However, such treatment would operate to the detriment of protoplasm staining, an indispensable prerequisite for our study. Foraminifers were identified and quantified per sample volume on a picking tray under a Wild M3Z dissecting microscope. Selected specimen were investigated with a Wild M10 with a Planapo 1.6 objective and by scanning electron microscopy (SEM CARL ZEISS LEO 1430 VP). Tests were fixed on isolating wax, fluted with argon, and sputter-coated with gold in a Polaron E 5000.

Data on near bottom oxygen concentrations (Table 1) were obtained with a multi-sensor probe (CTD) at about 5 m above the ocean floor during cruise M48/2. Derived oxygen measures were calibrated by Winkler measurements from Niskin bottles for several stations (Emeis 2002). Mean bottom oxygen values for the stations of cruise M57/3 were calculated from a grid of 1709 measures provided by the database of the South African Oceanographic Center (Sadco 2007). The Standard deviation of Winkler calibrated CTD oxygen values from M48/2 were correlated with the Sadco (2007) derived oxygen values for the same stations; the resulting standard deviation is 0.3 ml O₂ l^-1. This range corresponds to the accuracy considered for standard CTD probes in general and is threefold below the mean seasonal variability of oxygen concentrations recorded on the Namibian shelf (Kristmannsson 1999). Given environmental terms for observed oxygen ranges follow Bernhard and Sen Gupta (1999).