(Table S2) Relative abundance of archaeal 16S rRNA genes in sediment cores

Deep drilling into the marine sea floor has uncovered a vast sedimentary ecosystem of microbial cells (Parkes et al., 1994, doi:10.1038/371410a0; D'Hondt et al., 2004, doi:10.1126/science.1101155). Extrapolation of direct counts of stained microbial cells to the total volume of habitable marine subsurface sediments suggests that between 56 Pg (Parkes et al., 1994, doi:10.1038/371410a0) and 303 Pg (Whitman et al., 1998) of cellular carbon could be stored in this largely unexplored habitat. From recent studies using various culture-independent techniques, no clear picture has yet emerged as to whether Archaea or Bacteria are more abundant in this extensive ecosystem (Schippers et al., doi:10.1038/nature03302; Inagaki et al., doi:10.1073/pnas.0511033103 ; Mauclaire et al., doi:10.1111/j.1472-4677.2004.00035.x; Biddle et al., doi:10.1073/pnas.0600035103). Here we show that in subsurface sediments buried deeper than 1 m in a wide range of oceanographic settings at least 87% of intact polar membrane lipids, biomarkers for the presence of live cells (Biddle et al., doi:10.1073/pnas.0600035103; Sturt et al., 2004, doi:10.1002/rcm.1378), are attributable to archaeal membranes, suggesting that Archaea constitute a major fraction of the biomass. Results obtained from modified quantitative polymerase chain reaction and slot-blot hybridization protocols support the lipid-based evidence and indicate that these techniques have previously underestimated archaeal biomass. The lipid concentrations are proportional to those of total organic carbon. On the basis of this relationship, we derived an independent estimate of amounts of cellular carbon in the global marine subsurface biosphere. Our estimate of 90 Pg of cellular carbon is consistent, within an order of magnitude, with previous estimates, and underscores the importance of marine subsurface habitats for global biomass budgets.

Sea surface temperatures reconstructed from sediment core U1338

This study presents a new alkenone-derived Sea Surface Temperature (SST) record and d18ONoelaerhabdaceae data of the 2-5 µm carbonate fractions from the IODP site U1338 located in the Eastern Equatorial Pacific (EEP), over the Miocene-Pliocene. Our data and those available from other sites of the same area show the establishment of a cold tongue during the early Pliocene (4.4 - 3.6 Ma). SST and d18ONoelaerhabdaceae time-series indicate periods of significant salinity variations. Comparison with the d18Obenthic curve from sediment cores of the Equatorial Pacific Ocean allow us to distinguish between global changes and local surface salinity variations in the EEP. Ice sheet growth and evaporation-precipitation are then discussed as possible drivers of such changes, as well as the role of Central American and the Indonesian seaway restriction. Our data suggest a shallowing of the thermocline in the EEP, between 6.8 and 6 Ma, and its shoaling between 4.8 and 4.0 Ma, suggesting the appearance of the cold tongue (Steph et al., 2010). The Pliocene climate transition would therefore not be primarily driven by Northern hemisphere glaciation.

(Table 2) Carbonate chemistry and isotopes from northern Cascadia margin samples

Most authigenic carbonates previously recovered from the Cascadia slope have 87Sr/86Sr signatures that reflect shallow precipitation in equilibrium with coeval seawater. There is also evidence for carbonate formation supported by fluids that have been modified by reactions with the incoming Juan de Fuca plate (87Sr/86Sr = 0.7071; Teichert et al., 2005, doi:10.1016/j.epsl.2005.08.002) or with terrigenous turbidites (87Sr/86Sr = 0.70975 to 0.71279; Sample et al., 1993, doi:10.1130/0091-7613(1993)021<0507:CCICFF>2.3.CO;2). We report on the strontium isotopic composition of carbonates and fluids from IODP Site U1329 and nearby Barkley Canyon (offshore Vancouver Island), which have strontium isotope ratios as low as 0.70539. Whereas the strontium and oxygen isotopic compositions of carbonates from paleoseeps in the uplifted Coast Range forearc indicate formation in ambient bottom seawater, several samples from the Pysht/Sooke Fm. show a 87Sr-depleted signal (87Sr/86Sr = 0.70494 and 0.70511) similar to that of the anomalous Site U1329 and Barkley Canyon carbonates. Our data, when analyzed in the context of published elemental and isotopic composition of these carbonates (Joseph et al., 2012, doi:10.1016/j.palaeo.2013.01.012 ), point to two formation mechanisms: 1) shallow precipitation driven by the anaerobic oxidation of methane (AOM) with d13C values as low as -50 per mil and contemporaneous 87Sr/86Sr seawater ratios, and 2) carbonate precipitation driven by fluids that have circulated through the oceanic crust, which are depleted in 87Sr. Carbonates formed from the second mechanism precipitate both at depth and at sites of deep-sourced fluid seepage on the seafloor. The 87Sr-depleted carbonates and pore fluids found at Barkley Canyon represent migration of a deep, exotic fluid similar to that found in high permeability conglomerate layers at 188 mbsf of Site U1329, and which may have fed paleoseeps in the Pysht/Sooke Fm. These exotic fluids likely reflect interaction with the 52-57 Ma igneous Crescent Terrane, which supplies fluids with high calcium, manganese and strontium enriched in the non-radiogenic nucleide. Tectonic compression and dehydration reactions then force these fluids updip, where they pick up the thermogenic hydrocarbons and 13C-enriched dissolved inorganic carbon that are manifested in fluids and carbonates sampled at Barkley Canyon and at Site U1329. The Crescent Terrane may have sourced cold seeps in this margin since at least the late Oligocene.

Lithic grain and mineral composition of sand and sandstones from the North Pacific Ocean and the Bering Sea (Table 3)

Sand and sandstone compositions from different types of basins reflect provenance terranes governed by plate tectonics. One hundred and one thin sections of Upper Miocene to Holocene sand-sized material were examined from DSDP/IPOD Sites in the North Pacific Ocean and the Bering Sea. The Gazzi-Dickinson point-counting method was used to establish compositional characteristics of sands from different tectonic settings. Continental margin forearc sands from the western North America continental margin arc system are clearly different from backarc/marginal-sea sands from the Aleutian intraoceanic arc system. The forearc sands have average QFL percentages of 29-42-29, LmLvLst percentages of 32-34-34, 3 Fmwk%M and 0.82 P/F. Aleutian backarc sands have average QFL percentages of 8-22-69. LmLvLst percentages of 9-85-6, 0.5 Fmwk%M and 0.96 P/F. A trend of increasing QFL%Q and decreasing LmLvLst%Lv westward in the backarc region of the Aleutian Ridge reflects the influence of the Asiatic continental margin. Aleutian backarc sands without continental influence have average QFL percentages of 1-20-79, LmLvLst percentages of 1-98-1, 0 Fmwk%M and 0.99 P/F. Of the continental margin forearc samples, sands on the Astoria Fan (west of the Oregon-Washington trench) contain the highest LmLvLst%Lv and lowest P/F; sands from mixed transform-fault and trench settings (Delgada Fan and Gulf of Alaska samples) have slightly higher Qp/Q (0.03); and sands from the Pacific-Juan de Fuca-North America triple junction have the highest Fmwk%M. Delgada Fan and Gulf of Alaska sands have average QFL percentages of 27-38-35, LmLvLst percentages of 37-26-37, 2 Fmwk%M and 0.86 P/F. Astoria Fan sands have average QFL percentages of 35-41-24, LmLvLst percentages of 30-47-23, 3 Fmwk%M and 0.74 P/F. The triple-junction sands have average QFL percentages of 28-59-13, LmLvLst percentages of 25-26-49, 9 Fmwk%M and 0.87 P/F. The petrologic data from the modern ocean basins examined in this study can provide useful analogs for interpretation of ancient oceanic sequences. Our data suggest some refinements of, but generally substantiate, existing petrologic models relating sandstone composition to tectonic setting.