Separating individual contributions of major Siberian rivers in the Transpolar Drift of the Arctic Ocean

The Siberian rivers supply large amounts of freshwater and terrestrial derived material to the Arctic Ocean. Although riverine freshwater and constituents have been identified in the central Arctic Ocean, the individual contributions of the Siberian rivers to and their spatiotemporal distributions in the Transpolar Drift (TPD), the major wind-driven current in the Eurasian sector of the Arctic Ocean, are unknown. Determining the influence of individual Siberian rivers downstream the TPD, however, is critical to forecast responses in polar and sub-polar hydrography and biogeochemistry to the anticipated individual changes in river discharge and freshwater composition. Here, we identify the contributions from the largest Siberian river systems, the Lena and Yenisei/Ob, in the TPD using dissolved neodymium isotopes and rare earth element concentrations. We further demonstrate their vertical and lateral separation that is likely due to distinct temporal emplacements of Lena and Yenisei/Ob waters in the TPD as well as prior mixing of Yenisei/Ob water with ambient waters.

measured for 1 x 36 cycles at concentrations ranging between 5 and 20 ppb Nd in 250 µL solution with peak centering prior to the measurement session. The Nd standard JNdi-1 was measured after every 2-3 samples at the same concentration as the samples. The measured 143 Nd/ 144 Nd ratios were corrected for the instrumental mass fractionation using an exponential law and 146 Nd/ 144 Nd = 0.7219 6 . If possible (positive correlation between 143 Nd/ 144 Nd and 142 Nd/ 144 Nd), we applied a secondary mass bias correction using a linear correlation 7 of 143 Nd/ 144 Nd and 142 Nd/ 144 Nd. All data were normalized to the accepted value for the JNdi-1 standard 8 Table S1, the internal, external and propagated errors (calculated from internal and external error, 2SD) are shown. Blanks were processed identically to the samples and spiked with a 146 Nd spike for Nd quantification via MC-ICP-MS. Total procedural blanks (n = 3) from the shipboard MQ system were < 37 pg Nd, the procedural laboratory blanks (n = 16) were < 6 pg Nd, which represent < 3 % and < 1 % of the lowest sample concentration, respectively.

Rare earth element preconcentration and analysis
For the determination of [REE], isotope dilution ICP-MS analysis was applied following the method described in Behrens et al. 9 . Briefly, 10-20 mL seawater aliquots were spiked with a multi-element REE spike, processed with an automated seaFAST-pico system (ESI) in offline mode for REE preconcentration and matrix removal, and REEs were analyzed using a Thermo Finnigan Element II ICP-MS in combination with a Cetac Aridus II desolvating nebulizer system. Nitrogen supply allowed for reduction of oxide formation to < 0.04 % for Ce-oxide, corrections for oxide formation were therefore not necessary. For accuracy and external reproducibility, the seawater standard SAFe 3000 m (n = 13) was used. The values agreed well within 7 % of the published average [REE] of four different labs 9 and showed a reproducibility of < 10 % (2SD) and 35 % for Ce. Total procedural blanks from shipboard MQ water, procedural lab blanks (2 % HNO3, distilled, seaFAST preconcentration), and analytical blanks (2 % HNO3, distilled) were in general < 1.8 % (< 10 % for Ce) of the lowest sample concentration. The blanks were spiked for isotope dilution analysis after preconcentration and/or prior to the measurement. The ratios of HREE/LREE were calculated after Martin et al. 10 after REE normalization to Post Archean Australian Shale (PAAS 11 ). The standard deviation of the standard SAFe 3000 m (n = 12) for HREE/LREE was 4.6 % (2SD).

Intercalibration
The laboratory at the ICBM in Oldenburg is intercalibrated for [REE] through analysis of GEOTRACES standard SAFe 3000 m 9  Pacific water [16][17][18][19] . This approach, however, remains challenging due to shelf processes producing nutrient signatures in Siberian shelf derived waters similar to Pacific waters 19,20 and differences in estimates of Pacific water fractions based on the different methods are of up to 60 and 40 % in the central Arctic Ocean and Fram Strait, respectively 15,21 . Therefore, we refrain here from using Atlantic and Pacific fractions from the 4-component mass-balance but use the meteoric water fraction and SIM-corrected salinities, which are identical within error with those determined based on the 3-component analysis. The few values referred to for fractions of Pacific water based on the N/P method were calculated as described by Bauch et al. 19 .

Calculation of water mass fractions based on salinity, δ 18 O, [Nd], εNd and mass conservation (Nd-method)
Calculation of water masses is performed using a two-step method solving linear equations: In a first step, the fractions of marine water (mar), meteoric water (met) and sea-ice melt (SIM) are calculated using salinity and δ 18 O:  Supplementary Table S2.
To obtain a set of linear equations, equations (6) and (7) can be replaced by equations (8) and (9):