Methods for Godwin Lab/ A. Voelker data:
Water oxygen and hydrogen isotopes were measured simultaneously by cavity ringdown laser spectroscopy (CRDS) using a L1102-i Picarro water isotope analyzer and A0211 high-precision vaporizer. Each sample was injected nine times into the vaporizer. Memory effects from previous samples were avoided by rejecting the first three analyses. Values for the final six injections were averaged with in-run precision of less than ±0.1 for d18O and ±1 for dD (1-standard deviation). Calibration of results to V-SMOW was achieved by analyzing internal standards before and after each set of 7 or 8 samples. Internal standards were calibrated against V-SMOW, GISP, and SLAP. All results are reported in parts per thousand (o/oo) relative to V-SMOW.
The instrument was calibrated at the start of each run using three working standards from the University of Cambridge (Delta, Botty, Spit) with known values calibrated against V-SMOW, GISP, and SLAP. An internal seawater standard (SPIT) was analysed after every three samples to correct for instrument drift.

Methods for d18O and dD analysis at the University of Chicago
All seawater samples were processed on a GasBench II (Thermo, Bremen, Germany) with a CTC GC Pal (CTC analytics) and interfaced to a Delta V Plus isotope ratio mass spectrometer (Thermo). A hydrogen equilibration method using Pt catalyst was used for dD analysis; a classical CO2-H2O equilibration was used for d18O analysis. In both cases, samples were entrained in a helium carrier gas with gas sample cleanup and removal of water vapor achieved on the GasBench II peripheral prior to isotopic analysis in continuous flow mode on the mass spectrometer.

For D/H analyses, Pt catalyst sticks (Thermo P/N 1091831) were loaded into Exetainer tubes (Labco, 12 mL, screw cap with butyl rubber septa inserts). Water (samples and standards) were added to each tube using fresh 1 mL syringes (BD syringe) and needles (20 or 22 Ga, 1 or 1.5 inch lengths, BD precision Glide), with 300 µL of seawater sample used for all D/H runs. Standards were either primary (VSMOW2, SLAP2, and/or GISP from the IAEA) or house standards calibrated to the primary standards. The exetainers with Pt catalyst and water were loaded into the GasBench II heater block at 26°C in a group of 5 rows of 10 tubes. The CTC GC PAL sampling and flushing needles were side by side and spaced by 5 positions in the heater block. The first set of 5 tubes were flushed (2% H2 in He (Airgas), 20 mL/min), then following a pause, sampling of the first set of tubes was completed coincident with flushing of the adjacent set of tubes. Total incubation time of H2 gas mix with water and Pt catalyst was 68 minutes for each tube (end of flushing to start of analysis). The dD analysis on headspace H2 takes 15 min: 60 sec transfer time, 3 reference gas pulses (20 sec each, 30 and 80 sec spacing, last used as the reference peak) and 8 sample pulses (100 µL sampling loop, 45 sec load time at 0.5 mL/min, 35 second injection time at 1 mL/min, through PoraPlot Q GC column, 25m x 0.32mm, Varian, at 50°C). Typically the middle six sample peaks are used for isotopic composition determination, and linear plot of the peak’s result vs. peak number determined and the value for the first peak in the series is calculated from the slope and intercept.
For d18O runs, capped Exetainer tubes were flushed with 0.3% CO2 in He (Airgas, ca. 20 mL/min) using the PAL autosampler. Then 200 to 300 microliters of seawater or standard was injected into the Exetainer tube as described for the D/H sample injection. The tubes were incubated in the Gasbench II heater block at 26°C for 24 hrs before the analysis was initiated. The CTC GC PAL sampling parameters, GasBench II injection loop program and pattern of reference and sample peaks, and peak processing was similar to that described for dD measurements.

Raw d18O analyses are corrected for the fractionation factor between water and CO2 as a function of temperature of equilibration, mass balance between water oxygen and CO2 oxygen present in the tube, and to the isotopic reference materials using standard practices (Friedman and O’Neil). The D/H analyses are corrected to the standards (Nelson, 2000), and then a uniform -2.0 o/oo salt correction (Martineau et al., 2012) is applied to compensate for isotopic enrichment of the headspace H2 for the seawater samples. Both oxygen and hydrogen isotope compositions are reported on the VSMOW scale