IMIS

The majority of the sequences (> 99%) formed three clusters within the Marine Group I Crenarchaeota. The most important cluster, with 78.8% of the clones, included Candidatus Nitrosopumilus maritimus, a mesophilic archaeon able to oxidize ammonia. The most important subgroup in that cluster was the APA4-0cm subgroup (with 62.3% of the clones). This subgroup might represent important Crenarchaeota in the functioning of the bathypelagic sedimentary ecosystems of the Weddell Sea because it dominated the clone libraries in all sampling stations, and was found in sediments separated by very large geographic distances. Only one clone grouped within the Euryarchaeota. This euryarchaeal clone could not be affiliated with any of the previously defined clusters and might represent a novel Euryarchaeal lineage.
Geographic coverage: ANDEEP 3 cruise track, from Cape Town (SA) to Punta Arenas (CH). Four study regions were selected, but the main focus was on the Powell Basin and the Weddell Basin of the Weddell Sea, and their slopes. Two comparative samples were taken further north in the adjacent Agulhas and southern Cape Basins, which are separated from each other by the Agulhas Ridge. Four study regions were selected, but the main focus was on the Powell Basin and the Weddell Basin of the Weddell Sea, and their slopes. The major South Atlantic deep-sea basins started forming during Jurassic and Cretaceous times in connection with the Gondwana break-up and seafloor spreading (Brandt et al., 2004a, 2007; Lawver and Gahagan, 2003). The Weddell Basin is separated from the northerly basins by the South- west India Ridge (LaBrecque, 1986). The Powell Basin on the western side of the Weddell Sea was formed in the Tertiary by geological processes opening the Drake Passage and tectonic movements in the Scotia Sea (Lawver and Gahagan, 2003; Mitchell et al., 2000). The oceanography of the deep South Atlantic seafloor is defined by its prominent water mass, the Antarctic Bottom Water (Tomczak and Godfrey, 2001). The Antarctic Bottom Water expands north- wards into the Atlantic basins east and west of the Mid-Atlantic Ridge, like the Agulhas Basin, but can only enter the basins north of the Walvis Ridge (e.g., Cape Basin) via the northerly Romanche Fracture Zone. The Weddell Sea Bottom Water (WSBW), defined by a temperature of 0.7 1C and a salinity of 34.64 ppt (Orsi et al., 1993), is the main water mass above the Weddell Sea benthos (Fahrbach et al., 2001). The WSBW flows from the western Weddell Sea into the Scotia Sea and South Sandwich Forearc, and its circulation is driven by the Weddell Sea gyre. The sediments in the bathyal and abyssal Weddell and Powell Basins are dominated by silt and clay.
Sampling methods: Archaeal communities were analyzed in sediments in three bathypelagic stations (2200-3400m) sampled during the ANTXXII-3 cruise of RV Polarstern (ANDEEP3). Sediments were sampled using a box-corer. The sediments were immediately and aseptically sampled using 3 ml polyethylene (PET) cryovials (top 2 cm of the sediments). The cryovials were immediately frozen in liquid nitrogen and the samples were then stored at -20°C. Six replicate samples of the same core were collected at each site.
Quality control: PCR-generated chimeric sequences were first detected and eliminated using three different softwares : CHIMERA_CHECK version 2.7 of the Ribosomal Database Project II (Cole et al., 2003), Bellerophon (Huber et al., 2004), and Pintail version 0.33 (Ashelford et al., 2005). Sequences were then submitted to BLAST version 2.2.12 to identify the closest relatives and download their 16S rDNA sequence (Madden et al., 1996). All sequences were manually aligned and analyzed using Se-Al version 2.0a11 (Rambaut, 1996). Alignments and similarity calculations were also performed using the EMBOSS-Align tool of the EMBL-EBI (http://www.ebi.ac.uk). Distance and maximum-likelihood trees were generated with the Phylip program package, version 3.6 (Felsenstein, 2002). Distance trees were generated with "Dnadist" using Jukes-Cantor distances and Neighbour-Joining. Maximum-likelihood trees were generated with "Dnaml" (Ti/Tv=2.0; empirical base frequencies, one category of sites with a constant rate of variation). The statistical significance of the phylogenetic groups within the trees was tested by using bootstrap analysis with the Phylip programs "Seqboot" and "Consense" (100 bootstrap replicates, using using Jukes-Cantor distances and Neighbour-Joining). Trees were created using the program Treeview (version 1.6.6.).