Supplementary material: XRF core scanning data, CaCO3 values, foraminiferal carbon and oxygen isotope data, Mg/Ca data, clay mineralogy and helium isotope results

The Paleocene-Eocene Thermal Maximum (PETM; ~56 Ma) is associated with abrupt climate change, carbon cycle perturbation, ocean acidification, as well as biogeographic shifts in marine and terrestrial biota that were largely reversed as the climatic transient waned. We report a clear exception to the behavior of the PETM as a reversing climatic transient in the eastern North Atlantic (Deep-Sea Drilling Project Site 401, Bay of Biscay) where the PETM initiates a greatly prolonged environmental change compared to other places on Earth where records exist. The observed environmental perturbation extended well past the d13C recovery phase and up to 650 kyr after the PETM onset according to our extraterrestrial 3He-based age-model. We observe a strong decoupling of planktic foraminiferal d18O and Mg/Ca values during the PETM d13C recovery phase, which in combination with results from helium isotopes and clay mineralogy, suggests that the PETM triggered a hydrologic change in western Europe that increased freshwater flux and the delivery of weathering products to the eastern North Atlantic. This state change persisted long after the carbon-cycle perturbation had stopped. We hypothesize that either long-lived continental drainage patterns were altered by enhanced hydrological cycling induced by the PETM, or alternatively that the climate system in the hinterland area of Site 401 was forced into a new climate state that was not easily reversed in the aftermath of the PETM.

Geochemistry of basalts from ODP Leg 49, North Atlantic Ocean

The geochemistry of basalts recovered from seven sites in the North Atlantic is described with particular reference to minor elements. Three sites (407, 408, and 409) along the same mantle flow line, transverse to the Reykjanes Ridge at about 63°N, provide information on the composition of basalts erupted over a 34-m.y. interval between 2.3 and 36 m.y. ago. At Site 410, at 45°N, penetration into 10 m.y.-old crust west of the ridge axis permits comparisons with young basalts dredged from the median valley at 45°N. Three sites in the FAMOUS area at about 36°N provided material from very young (1 m.y.) basaltic crust (Site 411), and material to test the geochemical coherence of basalts of different ages (1.5 and 3.5 m.y.) on either side of a fracture zone (Sites 412 and 413). These sites complement earlier data from dredged and drilled sites (Leg 37) in the FAMOUS area. At Site 407, four geochemically distinct basalt units occur, with different normative and rare-earth element (REE) characteristics, and there is a clear correlation with magnetic stratigraphy. Yet there is a remarkable consistency in incompatible element ratios between these units, indicating derivation from an essentially similar mantle source. The basalts from the younger sites, 408 and 409, show a similar range of normative and REE variation, but incompatible element ratios are identical to those at Site 407, indicating that basalts at all three sites were produced from a mantle source which was geochemically relatively uniform. Rare-earth differences between the basalts can be interpreted in terms of variations in the degree and depth of partial melting causing HREE (+Y) retention in the source, although there may be some inter-site differences with respect to REE. A similar picture is presented at 45°N. Apparently a range of tholeiitic, transitional, and alkalic basalts were being erupted 10 m.y. ago, which have almost identical geochemical characteristics to those recently erupted in the median valley at 45°N. Incompatible element ratios are markedly different from those recorded at the Reykjanes Ridge. Basalts recovered from the FAMOUS sites are geochemically similar to previous samples recovered from the FAMOUS area, and their incompatible element ratios are similar, but not identical, to those at 45°N. However, total trace element levels are consistently lower than in 45°N basalts, which might imply smaller degrees of partial melting and/or greater depths of magma generation at 45°N, or higher trace element levels in the mantle source at 45°N. Few of the basalts recovered on Leg 49 have the geochemical characteristics of typical "MORB" (e.g., Nazca Plate, Leg 34). The data strongly support models invoking geochemical inhomogeneity in the source regions of basalts produced at the Mid-Atlantic Ridge. However, the data also introduce an additional time factor into such models and demonstrate the uniformity of the mantle source at a particular ridge sector (over periods in excess of 30 m.y.), while emphasizing the marked differences along the ridge. Mixing models invoking "depleted" and "enriched" mantle sources would seem to be inadequate to account for the observed variations.

(Table T1) Mineral composition of oriented clay aggregates from ODP Leg 190 sites

Three sites were cored on the landward slope of the Nankai margin of southwest Japan during Leg 190 of the Ocean Drilling Program. Sites 1175 and 1176 are located in a trench-slope basin that was constructed during the early Pleistocene (~1 Ma) by frontal offscraping of coarse-grained trench-wedge deposits. Rapid uplift elevated the substrate above the calcite compensation depth and rerouted a transverse canyon-channel system that had delivered most of the trench sediment during the late Pliocene (1.06-1.95 Ma). The basin's depth is now ~3000 to 3020 m below sea level. Clay-sized detritus (<2 µm) did not change significantly in composition during the transition from trench-floor to slope-basin environment. Relative mineral abundances for the two slope-basin sites average 36-37 wt% illite, 25 wt% smectite, 22-24 wt% chlorite, and 15-16 wt% quartz. Site 1178 is located higher up the landward slope at a water depth of 1741 m, ~70 km from the present-day deformation front. There is a pronounced discontinuity ~200 m below seafloor between muddy slope-apron deposits (Quaternary-late Miocene) and sandier trench-wedge deposits (late Miocene; 6.8-9.63 Ma). Clay minerals change downsection from an illite-chlorite assemblage (similar to Sites 1175 and 1176) to one that contains substantial amounts of smectite (average = 45 wt% of the clay-sized fraction; maximum = 76 wt%). Mixing in the water column homogenizes fine-grained suspended sediment eroded from the Izu-Bonin volcanic arc, the Izu-Honshu collision zone, and the Outer Zone of Kyushu and Shikoku, but the spatial balance among those contributors has shifted through time. Closure of the Central America Seaway at ~3 Ma was particularly important because it triggered intensification of the Kuroshio Current. With stronger and deeper flow of surface water toward the northeast, the flux of smectite from the Izu-Bonin volcanic arc was dampened and more detrital illite and chlorite were transported into the Shikoku-Nankai system from the Outer Zone of Japan.

Sediment properties of IODP Holes 303-U1302A and 303-U1305C

Episodes of ice-sheet disintegration and meltwater release over glacial-interglacial cycles are recorded by discrete layers of detrital sediment in the Labrador Sea. The most prominent layers reflect the release of iceberg armadas associated with cold Heinrich events, but the detrital sediment carried by glacial outburst floods from the melting Laurentide Ice Sheet is also preserved. Here we report an extensive layer of red detrital material in the Labrador Sea that was deposited during the early last interglacial period. We trace the layer through sediment cores collected along the Labrador and Greenland margins of the Labrador Sea. Biomarker data, Ca/Sr ratios and d18O measurements link the carbonate contained in the red layer to the Palaeozoic bedrock of the Hudson Bay. We conclude that the debris was carried to the Labrador Sea during a glacial outburst flood through the Hudson Strait, analogous to the final Lake Agassiz outburst flood about 8,400 years ago, probably around the time of a last interglacial cold event in the North Atlantic. We suggest that outburst floods associated with the final collapse of the Laurentide Ice Sheet may have been pervasive features during the early stages of Late Quaternary interglacial periods.

Heat-flow studies in the Peru Trench subduction zone

New heat-flow values were obtained in the central Peru Trench area during site surveys and drilling of Ocean Drilling Program (ODP) Leg 112 by measuring temperatures with ordinary surface heat-flow probes and in the drill holes and by estimating from bottom-simulating reflectors resulting from gas hydrates. The values determined by these methods are consistent with each other within the limits of error. When combined with existing data, heat-flow distribution from the trench to the coast was delineated. Heat flow is lower than 40 mW/m**2 at the bottom of the trench and 40 to 50 mW/m**2 on the landward slope. The low heat flow at the trench bottom can be explained partly by a high sedimentation rate. Heat flow is variable about where the Mendana Fracture Zone meets the trench. This low heat flow might result from hydrothermal circulation in the fracture zone, which some scientists believe is a new propagating rift. On the landward slope, no significant difference in heat flow is recognized between the northern side and the southern side of the fracture zone, in spite of differences in the age of the subducting plate and the tectonic history. Heat flow on the landward slope may be slightly higher than that in most other subduction zones.

XRF-scanned elemental concentrations from sediment cores off Peru

We present a high-resolution marine record of sediment input from the Guayas River, Ecuador, that reflects changes in precipitation along western equatorial South America during the last 18ka. We use log (Ti/Ca) derived from X-ray Fluorescence (XRF) to document terrigenous input from riverine runoff that integrates rainfall from the Guayas River catchment. We find that rainfall-induced riverine runoff has increased during the Holocene and decreased during the last deglaciation. Superimposed on those long-term trends, we find that rainfall was probably slightly increased during the Younger Dryas, while the Heinrich event 1 was marked by an extreme load of terrigenous input, probably reflecting one of the wettest period over the time interval studied. When we compare our results to other Deglacial to Holocene rainfall records located across the tropical South American continent, different modes of variability become apparent. The records of rainfall variability imply that changes in the hydrological cycle at orbital and sub-orbital timescales were different from western to eastern South America. Orbital forcing caused an antiphase behavior in rainfall trends between eastern and western equatorial South America. In contrast, millennial-scale rainfall changes, remotely connected to the North Atlantic climate variability, led to homogenously wetter conditions over eastern and western equatorial South America during North Atlantic cold spells. These results may provide helpful diagnostics for testing the regional rainfall sensitivity in climate models and help to refine rainfall projections in South America for the next century.

(Table T1) Composition of clay minerals from ODP Leg 204 sediments

This report describes the results of semiquantitative analysis of clay mineral composition by X-ray diffraction. The samples consist of hemipelagic mud and mudstone cored from Hydrate Ridge during Leg 204 of the Ocean Drilling Program. We analyzed oriented aggregates of the clay-sized fractions (<2 µm) to estimate relative percentages of smectite, illite, and chlorite (+ kaolinite). For the most part, stratigraphic variations in clay mineral composition are modest and there are no significant differences among the seven sites that were included in the study. On average, early Pleistocene to Holocene trench slope and slope basin deposits contain 29% smectite, 31% illite, and 40% chlorite (+ kaolinite). Late Pliocene to early Pleistocene strata from the underlying accretionary prism contain moderately larger proportions of smectite with average values of 38% smectite, 27% illite, and 35% chlorite (+ kaolinite). There is no evidence of clay mineral diagenesis at the depths sampled. The expandability of smectite is, on average, equal to 64%, and there are no systematic variations in expandability as a function of burial depth or depositional age. The absence of clay mineral diagenesis is consistent with the relatively shallow sample depths and corresponding maximum temperatures of only 24°-33°C.