207few drops of ultrapure HNO 3 to pH 1–2, for sample preserva-
tion (see Michard et al. 1994 ; Viollier et al. 1995 ; Busigny
et al. 2014 ). The filters were either discarded or preserved for
analysis of the suspended particulate matter (SPM).
Dissolved oxygen, temperature, and specific conductivity
were determined in situ using a Seabird SBE 19 Seacat pro-
filer (Sea-Bird Electronics Inc., Washington, DC) and turbid-
ity was measured using a NKE STBD 300 probe.
Settling particles were collected with sediment traps
(Uwitec, Austria) disposed at different depths along a verti-
cal profile in the water column from April, 1994, to May,
1995 (Viollier et al. 1997 ) and June to September, 2011
(Cosmidis et al. 2014 ).
A sediment core was collected in September 2009 at the
bottom of the lake (92 m) using a Uwitec gravity corer
(90 mm ext. diameter), allowing to sample a vertical profile
at regular intervals below the water-sediment interface
(Busigny et al. 2014 ). Sediment traps and core samples
were transferred into glove bags and placed under anoxic
conditions (N 2 atmosphere) immediately after collection.
Porewaters were separated from solid phase using Rhizons®
or centrifugation. During the separation process, oxygen
levels were monitored with an Oxi 340i WTW oxygen
meter and were always below the detection limit of 0.1
mg/L. Sediments were dehydrated by freeze-drying before
mineralogical, chemical and isotopic analyses. Finally, all
samples were stored at 4 °C until analyses were
conducted.
12.2.2 Chemical and Isotope Analyses
The dissolved Fe concentration and Fe(II)/Fe(III) speciation
were measured on board by spectrophotometry (Merck
SQ300 spectrophotometer) on filtered water samples follow-
ing the method by Viollier et al. ( 2000 ). Spectrophotometric
analyses (λ = 562 nm) of the iron-ferrozine complexes were
performed in a single aliquot before and after a reduction
step with hydroxylamine. The procedure was calibrated
using Fe(III) standards stable under normal conditions of
analysis. Manganese and phosphate concentrations were
determined in laboratory on filtered and acidified samples
from the water column using inductively coupled plasma
atomic emission spectroscopy (ICP-AES; Perkin Elmer
Optima 3000) or using inductively coupled plasma mass
spectrometry (ICP-MS; Thermo Element2). Sulfate (SO 4 2−)
concentrations were measured by ion chromatography
(Dionex DX-600 IC System) from samples filtered and poi-
soned with zinc acetate to avoid reoxidation of any sulfide
present. Total sulfide concentrations (ΣH 2 S) were deter-
mined by spectrophotometry using the methylene blue
method as described in Bura-Nakic et al. ( 2009 ).
Iron isotope compositions were measured on water sam-
ples collected from various depths in the lake. The method
was described in a previous contribution (Busigny et al.
2014 ) and can be summarized as follows. Samples were
acidified with HNO 3 to ensure that all Fe is in the ferric state.
Iron was then separated from matrix elements on anion
exchange chromatography in HCl medium (Strelow 1980 ).
Iron concentrations and isotopic compositions were mea-
sured using a Neptune ThermoFischer MC-ICP-MS
(Multiple Collector Inductively Coupled Plasma Mass
Spectrometer). We corrected for instrumental mass discrimi-
nation using the conventional sample-standard bracketing
(SSB) approach. The^56 Fe/^54 Fe ratio is reported in the usual δ
notation in per mil (‰) as:
δ^56 Fe=−[(^56 Fe/)^54 Fesample/(^56 Fe/)^54 Festandard 1 1000]×where the standard is IRMM-014 (Institute for Reference
Materials and Measurements; Taylor et al. 1992 ). Based on
replicate analyses of international rock standards, the exter-
nal precision and accuracy on δ^56 Fe values were always bet-
ter than 0.06 ‰ (2SD).12.2.3 Mineralogical AnalysesThe mineralogy of Lac Pavin particles from sediment traps
and sediment core samples has been characterized by
X-ray diffraction (XRD), Extended X-ray absorption fine
structure (EXAFS) spectroscopy, scanning electron
microscopy (SEM), scanning transmission X-ray micros-
copy (STXM) and transmission electron microscopy
(TEM) (Viollier et al. 1997 ; Cosmidis et al. 2014 ). The
coupling between these various methods provides infor-
mation on the identity of the mineral phases, their chemi-
cal composition (including redox state and speciation of
iron), size and shape. The detection limit for a specific
mineral phase in a complex natural sample is generally
considered to be ~1 %.12.2.4 Geochemical ModelingA geochemical modeling was developed herein to (i) con-
strain Fe and P reaction pathways and to (ii) obtain quantita-
tive constraints on physical, chemical and isotopic parameters
(such as Fe isotope fractionation associated with various
chemical reactions). Our model is a reactive-transport model
developed from AQUASIM software (Reichert 1998 ). It
describes spatial and temporal distribution of dissolved and
particulate chemical species in the lake water column. The
detailed methodology and formalism have been presented in12 Iron Wheel in Lac Pavin