(Appendix) Characteristic directions at DSDP Hole 76-534A

Kimmeridgian-Tithonian red marly limestones and Berriasian white limestones were recovered at Site 534 of DSDP Leg 76 in the western North Atlantic. These yielded a well-defined magnetostratigraphy with the characteristic magnetization carried by hematite in red sediments and magnetite in white sediments. The polarity sequence is correlated to the magnetostratigraphy of Kimmeridgian-Tithonian-Berriasian pelagic carbonates of northern Italy and southern Spain, allowing precise biostratigraphic age correlations. The Berriasian/Tithonian boundary occurs within the upper half of Core 90, the late Tithonian/early Tithonian boundary at the base of Core 96, and the Tithonian/Kimmeridgian boundary at the top of Core 102. Correlations are also made to M-16 through M-22 of the marine magnetic anomaly M-sequence. Poor recovery and irregular magnetic properties of the underlying Kimmeridgian-Oxfordian-Callovian marls and claystones prevented determination of a polarity sequence, but the entire interval has mixed polarity. Valanginian gray marly limestones have very weak magnetizations, and preliminary results are inadequate to determine the polarity pattern.

(Appendix) Petrographic information, trace element and Sr isotope ratios of Pito Deep samples

Fluid flow through the axial hydrothermal system at fast spreading ridges is investigated using the Sr-isotopic composition of upper crustal samples recovered from a tectonic window at Pito Deep (NE Easter microplate). Samples from the sheeted dike complex collected away from macroscopic evidence of channelized fluid flow, such as faults and centimeter-scale hydrothermal veins, show a range of 87Sr/86Sr from 0.7025 to 0.7030 averaging 0.70276 relative to a protolith with 87Sr/86Sr of ~0.7024. There is no systematic variation in 87Sr/86Sr with depth in the sheeted dike complex. Comparison of these new data with the two other localities that similar data sets exist for (ODP Hole 504B and the Hess Deep tectonic window) reveals that the extent of Sr-isotope exchange is similar in all of these locations. Models that assume that fluid-rock reaction occurs during one-dimensional (recharge) flow lead to significant decreases in the predicted extent of isotopic modification of the rock with depth in the crust. These model results show systematic misfits when compared with the data that can only be avoided if the fluid flow is assumed to be focused in isolated channels with very slow fluid-rock exchange. In this scenario the fluid at the base of the crust is little modified in 87Sr/86Sr from seawater and thus unlike vent fluids. Additionally, this model predicts that some rocks should show no change from the fresh-rock 87Sr/86Sr, but this is not observed. Alternatively, models in which fluid-rock reaction occurs during upflow (discharge) as well as downflow, or in which fluids are recirculated within the hydrothermal system, can reproduce the observed lack of variation in 87Sr/86Sr with depth in the crust. Minimum time-integrated fluid fluxes, calculated from mass balance, are between 1.5 and 2.6 * 10**6 kg/m**2 for all areas studied to date. However, new evidence from both the rocks and a compilation of vent fluid compositions demonstrates that some Sr is leached from the crust. Because this leaching lowers the fluid 87Sr/86Sr without changing the rock 87Sr/86Sr, these mass balance models must underestimate the time-integrated fluid flux. Additionally, these values do not account for fluid flow that is channelized within the crust.

(Appendix) Recent surface areas and historical lengths of glaciers from Patagonia and Tierra del Fuego, South America

Fluctuations in the length of 72 glaciers in the Northern and Southern Patagonia Icefield (NPI and SPI, respectively) and the Cordillera Darwin Icefield (CDI) were estimated between 1945 and 2005. The information obtained from historical maps based on 1945 aerial photographs was compared to ASTER and Landsat satellite images and to information found in the literature. The majority of glaciers have retreated considerably, with maximum values of 12.2 km for Marinelli Glacier in the CDI, 11.6 km for O'Higgins Glacier in the SPI and 5.7 km for San Rafael Glacier in the NPI. Among the 20 glaciers that have retreated the most relative to their size, small (less than 50 km**2) and medium (between 50 and 200 km**2) glaciers are the most affected. However, no direct relation between glacier retreat and size was found for the 72 glaciers studied. The highest percentage retreat in the CDI was by the CDI-03 Glacier (37.9%) and Marinelli Glacier (37.6%). In the SPI, relative retreats were heterogeneous and fluctuated between 27.2% (Amelia Glacier) and 0.4% (Viedma Glacier). In the NPI, relative retreat was very high for Strindberg and Cachet glaciers (35.9% and 27.6%, respectively) but for the remaining glaciers in this icefield it ranged between 11.8% (Piscis Glacier) and 3.6% (San Quintin Glacier). In addition to surface area, the surface slope (calculated on the basis of the DEM SRTM) was also related to the relative retreat and no straightforward relation was found. From a global point of view, we suggest that glacier retreat in the region is controlled firstly by atmospheric warming, as it has been reported in this area. Besides the general increase in temperature observed, no signal of a geographical pattern for the fluctuations in glacier length was found. Consequently, glaciers appear to initially react to local conditions most probably induced by their exposition, geometry and hypsometry. The heterogeneity of rates of retreat suggests that differences in basin geometry, glacier dynamics and response time are key features to explain fluctuations of each glacier.