Methodology:

Silicon and nitrate concentrations were measured with spectrophotometry. Dissolved iron concentration, manganese concentration, iron isotopes and silicon isotopes were measured with inductively coupled plasma-mass spectrometry.

The reaction transport model employed is the Biogeochemical Reaction Network Simulator. The model was run under steady state assumption.

Pore water data depth resolution: 5 cm - 15 cm

Sediment data depth resolution: 10 cm - 50 cm

Model data depth resolution: 0.1 cm - 1 cm

Data collection:

(1) Thermo Scientific X-Series 2 quadrupole inductively coupled plasma-mass spectrometer (dissolved iron and manganese concentration measurements)

(2) Thermo-Finnigan Neptune multicollector-inductively coupled plasma-mass spectrometer (silicon and iron isotope measurements)

(3) Maple computing environment version 2016.0 (compile inputs for reaction transport modeling)

Data quality:

The accuracies of the pore water elemental concentration and isotope analyses were assessed with measurements of standard solutions. Errors associated with sediment and pore water analyses were assessed with replicate measurements of samples and standards. Analytical errors for: (i) NO3- concentration is about 0.5%; (ii) dMn concentration and dFe concentration are 0.7% - 8.9%; (iii) DSi concentration is about 2.5%; (iv) Si-Alk content is 25%; (v) Si-HCl content is 15%; (vi) delta30 Si is 0.08 per-mille - 0.34 per-mille, and (vii) delta56 Fe is 0.04 per-mille - 0.08 per-mille. Model's best fit with observational data was determined by minimizing the root-mean-square error.