211we assumed that their effect are negligible for the bulk Fe-P
cycles. In the reducing layer, Fe and P are supplied from the
sediment to the water column (benthic Fe and P flux from
Viollier et al. 1995 ; Table 12.3), and from a mineralized sub-
surface spring assumed to be located 8 m below the redox
transition zone (water flow estimate from Jézéquel et al. 2011 ).
The Fe isotope composition of this subsurface spring is pres-
ently unknown. However, we analyzed two different ferrugi-
nous springs in the Pavin area, supposed to be representative
of the purported subsurface spring feeding the monimolim-
nion (e.g. Assayag et al. 2008 ). Both springs, Fontaine Goyon
and Source Saint-Anne, have δ^56 Fe values of−0.80 ± 0.03 ‰
(n = 2) and −1.09 ± 0.02 ‰ (n = 2), respectively. In our model-
ing, Fe isotope composition of the subsurface ferruginous
spring was thus assumed to be −1 ‰ (Table 12.3). The vertical
eddy diffusion coefficients were calculated from temperature
and conductivity profiles (Lopes et al. 2011 ). Other parame-
ters such as chemical reaction rates, isotope fractionations
associated with chemical reactions and particle sedimentation
velocity were obtained from modeling by adjusting the param-
eters for a better fit of the experimental data. Table 12.3 pro-
vides a list of chemical reactions, kinetic constants and isotope
fractionations used in the present model. Figure 12.7 repre-
sents the results of the model obtained from values given in
Table 12.3. A first observation is that the model reproduces,
with good agreement, experimental data collected from the
field. Although a test of sensitivity was performed on the
model, we cannot rule out the possibility that another set of
parameters would also fit the data.
The sensitivity analysis shows that the concentrations of
particulate Fe along the water column profile are tightly con-
trolled by particle sedimentation velocity. On the contrary,
the concentrations of dissolved species are not significantly
affected by the sedimentation velocity. This explains the
relative constancy of dissolved Fe2+ and SRP (soluble reac-
tive phosphorus) concentrations over time in the water col-
umn, while turbidity profiles (tracking particles) in the
anoxic zone are variable (Fig. 12.3b). Although the model
does not perfectly match with the distribution of particulate
Fe, the overall shape is reproduced (Fig. 12.7d) and allows us
to estimate an average particle sedimentation velocity of
about 1 m.d−1. The variability of particles distribution along
the depth profile may reflect time dependent variations
related to the type of particles and/or fluxes. For instance, it
was demonstrated, from sediment trap monitoring, that the
main flux of particles to the sediment is related to diatoms
bloom in the spring, which brings about 60 % of the annual
particle deposit (Viollier et al. 1997 ).
Another important result from the model is that the profile
of Fe isotope composition (Fig. 12.7c) is mostly –if not
only– controlled by the values of Fe isotope fractionation
used for reactions of Fe oxidation and reduction, and vivian-
ite precipitation. Thus the model provides an estimate of
each of these isotope fractionation values (Table 12.3). The
deep ferruginous spring, located at the relative depth of 13 m
in the present model (i.e. ~65 m depth in Lac Pavin from
Jézéquel et al. 2011 ), is not able to produce any peak in the
Fe concentration at this depth, but can modify the whole pro-
file at the scale of the monimolimnion. These modifications
are observed only when the dissolved Fe concentration of the
ferruginous spring is largely increased (by a factor of 4 or
higher), which is not compatible with the depth profile of45556575859501234560400800120016000 100 200300 400July 2007
Sept. 1992SRP(μM)Fe2+(μM) Fe2+ = 3.45 × SRP
R^2 = 0.99Fe/P (molar)Depth(m)(A)(B)Fig. 12.2 (a) Relation between the concentrations of dissolved iron
and phosphate (SRP: soluble reactive phosphorus) in water samples
from Lac Pavin. Black squares and circles represent two different peri-
ods of sampling in July 2007 (MX15) and September 1992 (Data from
Michard et al. 1994 ) respectively. (b) Fe/P molar ratio in the dissolved
fraction of water samples collected at different depths in the water col-
umn. The symbols represent two periods of sampling like in (a)
12 Iron Wheel in Lac Pavin