Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL OF

MERCURY DYNAMICS IN LAKES

M. L. DIAMOND
Department of Geography, University of Toronto, 100 St. George Street, Toronto, Ontario, M5S 3G3
Canada
(e-mail: diamond@geog.utoronto.ca; fax: 416 978 6729)

(Received 22 January 1997; accepted in revised form 2 April 1998)

Abstract. A simple, mechanistic model of mercury (Hg) dynamics in a lake has been developed,
based on the fugacity/aquivalence approach of Mackay (1991) and Mackay and Diamond (1989) and
its extension to treat several interconverting chemical species (Diamond et al., 1992). The model
considers the distribution of inorganic (HgII), elemental (Hg◦ ) and methyl (MeHg) mercury species
between dissolved and particle-sorbed phases, and fate and transport in a system consisting of a well-
mixed water column and an active sediment layer. Hg can enter the lake from watershed runoff and
by atmospheric deposition directly to the lake surface. Once in the lake, Hg exchanges between water
and air, and water and sediments, and exits by sediment burial, advective flow and volatilization. The
model was applied to a hypothetical drainage lake on the Canadian Shield. Model estimates of water
and sediment concentrations compare well with measured values. The results suggest that the three
Hg species experience significantly different fates and persistence, with overall Hg dynamics domi-
nated by the fate of HgII (the predominant species). A sensitivity analysis illustrates the importance
of physical/chemical properties and lake characteristics on the total amount and behavior of Hg in
the lake.

Keywords: mass balance model, mercury, temperature lake

1. Introduction

Sport fish in a significant number of remote lakes have concentrations of mercury


(Hg) in excess of the 0.5 ppm Canadian guideline for human consumption (e.g.,
OMOEE and OMNR, 1995; McMurtry et al., 1989). Similarly high levels of Hg
in fish have been documented in remote lakes in Scandinavia (e.g., Meili, 1988;
Hakanson et al., 1990) and in several states in the United States (e.g., Grieb et al.,
1990; Sorensen et al., 1990; Driscoll et al., 1994). From scientific and regulatory
perspectives, it is puzzling and disturbing that many of these lakes have no known
direct emissions of mercury. Thus, the focus of Hg studies has shifted from local
situations with known point-sources (e.g., English-Wabigoon system, Parks et al.,
1986; Parks, 1988) to the regional scale and non-point sources of Hg (Lindqvist,
1991; Weiner and Stokes, 1990).
Several hypotheses have been proposed to account for high levels of Hg in fish
on regional scales. Depth profiles of mercury concentrations in lake sediments
and peat bogs suggest increases in Hg concentrations that are coincident with

Water, Air, and Soil Pollution 111: 337–357, 1999.


© 1999 Kluwer Academic Publishers. Printed in the Netherlands.
338 M. L. DIAMOND

industrialization (e.g., Evans, 1986; Rekolainen et al., 1986; Swain et al., 1992).
The source of Hg, according to Brosset (1987), is combustion facilities such as
coal-fired power plants, that release significant amounts of Hg that are subject to
long-range atmospheric transport. Other hypotheses point to the correlation be-
tween lake acidification and elevated Hg levels in fish (e.g., Meger, 1987; Winfrey
and Rudd, 1990). Much of the research has identified the importance of net methy-
lation rates, and factors that affect net methylation, as being critical to determining
fish Hg concentrations (e.g., Xun et al., 1987; Gilmour and Henry, 1991). Physical
lake characteristics have also been found to correlate with fish Hg concentrations,
however some studies have reported contradictory results. McMurtry et al. (1989)
found a weak positive correlation between lake area and fish Hg concentrations
but others have found an inverse relationship (Hakanson et al., 1988; Sorensen et
al., 1990; Bodaly et al., 1993). The inverse relationship is hypothesized to be due
to correlations among lake size, summer water and surface sediment temperature,
and methylation rates. Several studies have correlated fish Hg concentrations with
watershed area and the ratio of watershed to lake area, an indication of poten-
tial Hg loading (e.g., Hakanson et al., 1988; McMurtry et al., 1989; Suns and
Hitchin, 1990). Finally, Hakanson et al. (1988) and Sorensen et al. (1990) reported
that water retention time negatively correlated with fish Hg concentrations. Thus,
within-lake mechanisms leading to higher Hg concentrations in fish are unclear,
with two main arguments focusing on (a) biological and (b) geochemical and
physical processes (Watras et al., 1995a).
The empirical models relating fish Hg concentrations to various lake charac-
teristics have met with some success but, as with any empirical relationship, they
are restricted to the lakes and fish species from which they were developed. This
could, in part, explain some of the contradictory findings. Because they are empir-
ical, it is difficult to establish causal relationships (Hakanson, 1990). Alternatively,
mechanistic models should improve our understanding of Hg dynamics and the
processes critical to these dynamics, although our ability to compile such a model
is circumscribed by major uncertainties in various processes.
In this paper a model of Hg fate and transport, and species interconversion in
lakes is developed and applied. The objectives of the model are to provide a tool
for exploring Hg dynamics on a whole lake basis that relies on a minimum of input
data, and examine hypotheses concerning the role of physical-chemical properties
of Hg and physical processes on these dynamics. When applied to contaminated
systems, the model can be used to guide management decisions. The model is
an extension of the QWASI model (Quantitative Water Air Sediment Interaction)
model of Mackay et al. (1983), using the multi-species approach developed by
Diamond et al. (1992), that has been described by Diamond and Sang (1992)
and applied to several systems (e.g., Diamond et al., 1993; Mackay et al., 1995).
Similarly to other QWASI models, it is relatively simple in describing the ma-
jor chemical transport processes. The key to considering Hg, or other speciating
chemicals, is the model’s reliance on concentration fractions for each species in
DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL 339

each phase. In this application, the species proportions are assumed to be at steady
state (i.e. they do not change with time), but not at chemical equilibrium. This
assumption confers considerable simplicity to the model and reduces the amount
of data required. Thus, the model is widely applicable, but differs from the more
complex models of Hudson et al. (1994) and Harris (1991), that also treat Hg
movement in a lake, predicting Hg methylation based on lake chemistry.
In this paper, the model is first described and then applied to a hypothetical
drainage lake located on the Canadian Shield, such as those studied near Dorset,
Ontario. The model results are interpreted and a sensitivity analysis illustrates the
effect of lake characteristics on Hg behavior.

2. Model Development

The fugacity approach of Mackay and co-workers (1991) has been used to de-
scribe the multi-media behavior of various chemicals in a variety of environments
(e.g., Mackay and Paterson, 1991; DiGuardo et al., 1994). Fugacity, an equilib-
rium criterion, is applicable to chemicals with a measurable vapor pressure such as
most nonpolar organic chemicals, but for involatile chemicals, such as metals and
polymers, the equilibrium criterion aquivalence, is used (Mackay and Diamond,
1989).
Briefly, fugacity f (Pa) and aquivalence Q (mol m−3 ) are linearly related to
concentration C (mol m−3 ) through chemical capacity or Z values as C = fZ = QZ,
where Z (mol m−3 Pa−1 ) and Z (dimensionless) are the fugacity and aquivalence
capacities, respectively. The ratio of two Z values in phases 1 and 2 is the dimen-
sionless partition coefficient K12 . In the fugacity formalism the Z value for air ZA
is first established as 1 (RT)−1 , where RT is the gas constant-temperature group
(Pa m3 mol−1 ). With aquivalence, the Z value for water ZW is defined as 1.00. In
both formalisms, subsequent Z values are determined from partition coefficients,
e.g., ZP = ZW KP where ZP is for suspended particles and KP is the dimensionless
particle-to-dissolved chemical partition coefficient.
As described by Diamond et al. (1992), for multiple for interconverting species
j in phase i, concentration is
C = fij Zij = Qij Zij (1)
where Zij is related to the species-specific partition coefficient K12j . For each
phase there is a total concentration CT , total fugacity or aquivalence fT or QT ,
respectively, and a total Z value, ZT . CT is calculated as a weighted average
according to species concentration fractions Xij
Xij = Cij CT−1 and CT = 6Xij Cij (2)
ZT is calculated as an average weighted by aquivalence fractions Yij
ZT = 6Yij Zij where Yij = Qij Q−1
T and QT = 6Yij Qij (3)
340 M. L. DIAMOND

with 6Xij = 6Yij = 1. Species-specific Z values, e.g., Zij , can be calculated from
K12T by multiplying by the concentration fractions, X1j and X2j
−1
Z1j = Z2j K12j = Z2j (X1j X2j )K12T (4)

Z values for mixed phases consisting of, for example, dissolved and particulate
phases, can be defined for individual species, ZBij , and total chemical, ZBTj where
B refers to a bulk phase.
Under physical equilibrium conditions, equal aquivalence applies to the chem-
ical in all phases, e.g., QAT = QW T = QST , and each species in all phases, e.g.,
QAj = QWj = QSj where A, W and S refer to air, water and sediment com-
partments. However, since aquivalence pertains to physical, not chemical equi-
librium, species aquivalences within a phase will not necessarily be equal, e.g.,
QA1 6 =QA2 6 =QA3 .
Chemical movement, transformation and species interconversion are expressed
by D values (m3 h−1 ), calculated for advective flow as GZij where G (m3 h−1 )
refers to bulk material movement, transformation or species interconversion as
kj V Zij where kj (h−1 ) is a first order reaction rate constant and V (m3 ) is vol-
ume, or diffusion as Kj AZij or Bj AY −1 Zij where Kj (m h−1 ) is a mass transfer
coefficient, Bj is a diffusivity (m2 h−1 ), A (m2 ) is area, and Y (m) is path length.
Total chemical movement, DT , can be calculated as DT = 6Yij Dij . The rate of
movement, transformation or interconversion within or from phase i of species j,
Nij (mol m−3 ), is Nij = Qij Dij and NT = QT DT .
The mass balance equations for one of three species in a lake consisting of a
well-mixed water column underlain by a well-mixed sediment layer are

d(VW 1 ZW 1 QW 1 )/dt = EW 1 + QA1 (DA1 + DV 1 ) + QI 1 (DI 1 + DX1 ) +

QS1 (DT 1 + DR1 ) + QW 2 DW 21 + QW 3 DW 31 −

QW 1 (DV 1 + DT 1 + DD1 + DJ 1 + DY 1 + DW 12 + DW 13 ) (5)

for species 1 in water, and

d(VS ZS1 QS1 )/dt = QW 1 (DT 1 + DD1 ) + QS2 DS21 + QS3 DS31 −

QS1 (DT 1 + DD1 + DB1 + DS12 + DS13) (6)

in sediment, where E is direct emissions into the water (mol h−1 ), and the sub-
scripts A, I , W and S refer to air, water inflow, water and sediment, respectively
(see Table I for definitions of the D values). Air is treated as a semi-infinite com-
partment with a defined concentration. These equations differ from the single species
model by having terms for species interconversion, e.g., conversion of species 1
into 2 in water DW 12 , etc.
DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL 341
TABLE I
Definition of transport and species interconversion D values

Process D value Rate (mol h)

Sediment burial DB GB ·CS or (GB ·Zs ) ·QS or DB ·QS


Sediment resuspension DR GR ·CS or (GR ·ZS ) ·QS or DR ·QS
−1
Sediment to water diffusion DT KT ·AS ·CS ·KSW or (KT ·AS ·ZW ) ·QS or DT ·QS
Water to sediment diffusion DT KT ·AS ·CW or (KT ·AS ·ZW ) ·QW or DT ·QW
Sediment deposition DD GD ·CP or (GD ·ZP ) ·QW or DD ·QW
Water inflow DI GI ·CI or (GI ·ZW ) ·QI or DI ·QI
Water particle inflow DX GX ·CX or (GX ·ZP ) ·QI or DX ·QI
Water outflow DJ GJ ·CW or (GJ ·ZW ) ·QW or DJ ·QW
Water particle outflow DY GY ·CP or (GY ·ZP ) ·QW or DY ·QW
Atmospheric deposition DA DM + DC + DQ
−1
Rain dissolution DM GM ·CA ·KAW or (GM ·ZW ) ·QA or DM ·QA
Wet particle deposition DC GC ·CQ or (GC · ZQ ) ·QA or DC ·QA
Dry particle deposition DQ GQ ·CQ or (GQ ·ZQ ) ·QA or DQ ·QA
Volatilization DV KV ·AW ·CW or (KV ·AW ·ZW ) ·QW or DV ·QW
Species interconversion (water) DW VW ·CW ·k12 or (VW ·ZW ·k12 ) ·QW or DW ·QW
Species interconversion (sed) DS VS ·CS ·k12 or (VS ·ZS ·k12 ) ·QS or DS ·QS

Nomenclature and explanation: The groups in parentheses are the D values (m3 h−1 ), e.g., DB is
GB ·ZS ). The rate (mol h−1 ) is the product of D and aquivalence, Q (mol m−3 ), e.g., DB ·QS . G
values are flows (m3 h−1 ) of a phase, e.g., GB is m3 h−1 of sediment that is buried. C values are
concentrations (mol m−3 ), the subscripts being S sediment, W water, A air, Q aerosol, P water par-
ticles, I water inflow, X water particle inflow, I water outflow and Y particle outflow. QW , QS , QA ,
and QI are aquivalences of water, sediment, air, and water inflow. Z values are chemical capacities
−1
(dimensionless), the subscripts defined as for concentration. KSW , e.g., ZS ZW , is a sediment-water
−1
partition coefficient (dimensionless) and KAW , e.g., ZA ZW , is an air-water partition coefficient
(dimensionless). KT and KS are sediment-water and within sediment mass transfer coefficients
(mh−1 ). k12 is a first order rate constant for conversion of species 1 to 2 (h−1 ). AW and AS are
air-water and water-sediment areas (m2 ), and VW and VS are water and sediment volumes (m3 ).

A main advantage of this formalism is that the species-specific equations for


each phase can be summed to one equation per phase. The equations for water and
sediment are
d(VW ZBW T QW T )/dt = EW T + QAT (DAT + DV T ) + QI T (DI T + DXT )+
QST (DT T + DRT ) − QW T (DV T + DT T + DDT + DJ T + DY T ) (7)
d(VS ZBST QST )/dt = QW T (DT T + DDT ) − QST (DT T + DDT + DBT ). (8)
The terms for species interconversion cancel and the equations become analogous
to the single chemical case. The steady-state solution for (8) is
QST = QW T (DT T + DDT )/(DT T + DDT + DBT ) (9)
342 M. L. DIAMOND

and for (7) with substitution of (9) is

QW T = EW T + QAT (DAT + DV T ) + QI T (DI T + DXT )+

QST (DT T + DRT )/[(DV T + DT T + DDT + DJ T + DY T )−

(DT T + DDT )(DT T + DRT )/(DT T + DRT + DBT )] . (10)

Aquivalences in water and sediment can be solved for each species using the
species-specific Equations (5) and (6), however, this solution is mathematically
difficult since for N species in two phases there are 2 N equations and 2 N un-
knowns. In reality, the solution strategy is difficult because the rates of species
interconversion are frequently not known. Alternatively, (9) and (10) can be solved
for total chemical, and species-specific aquivalences can be obtained as the product
YI J QT . From species-specific and total aquivalences in all phases, all rates of
total chemical and species movement, concentrations, amounts and net rates of
species interconversion can be calculated. Absolute rates of interconversion could
be obtained if the 2 N rate constants for the N species in the two phases are known.
Model estimates can also be used to estimate the persistence (units of time)
of species and total chemical in water and sediments where, under steady-state
conditions, persistence is defined as the ratio of the amount of chemical in a phase
or the total system, to inputs.
The method of solving the equations in the absence of species interconversion
rates relies on the use of concentration fractions, Xij . As discussed above, using
these fractions assumes that the properties of all species in all phases are at steady
state but not necessarily chemical or physical equilibrium. This assumption may be
reasonable when considering ‘average’ conditions over time (Meili et al., 1991),
but may be inappropriate when considering seasonal variations (e.g., Bloom and
Effler, 1990) or systems that depart significantly from steady state. Under the lat-
ter conditions, the unsteady-state version of the model should be used with time
dependent concentration fractions. Although the effect of ambient chemistry on
Hg speciation is not considered explicitly; these effects are included implicitly;
through the use of emperically derived partition coefficients (Diamond, 1995) and
species fractions.

3. Model Application

Three species of Hg are considered in this application of the model: gaseous el-
emental (Hg◦ ), inorganic (forms of HgII) and monomethylmercury (CH3 HgX or
MeHg). Together they account for about 98% of total mercury in the water and sur-
ficial sediments of most lakes not receiving direct industrial inputs (Brosset, 1987).
Individually, the species have significantly different physical-chemical properties
DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL 343
TABLE II
Physical-chemical properties of Hg species

Hg◦ HgII MeHg

Molecular wt. (g mol−1 ) 200.59 271.5 215.73


Henry’s law constant (Pa m3 mol−1 ) 479 – 0.036
Vapour pressure (Pa) 0.246 – 1.76
Melting point (◦ C) –38.9 277 167

Source: Yarwood and Niki (1990).

(Table II). Dimethylmercury was not included because it is usually below analytical
detection limits in freshwater system (Fitzgerald et al., 1991; Vandal et al., 1991).
Z values for the model were developed as follows. For all species, ZW (water) is
defined as 1.00. Z values for air ZA , suspended particles ZP , and bottom sediment
ZS are calculated as the product of ZW and KAW , KP W or KSW , respectively. For
Hg◦ and MeHg, both with measurable vapor pressures, KAWj was estimated as
H (RT)−1 where H is the Henry’s law constant (Pa m3 mol−1 ). ZA for HgII was
assigned an arbitrarily low value of 10−6 that is consistent with its nonvolatility
(e.g., Diamond, 1995). KP Wj and KSWj were calculated from KP W T and KSW T
using (4) with the concentration fractions, Xij . KP W T and KSW T were calculated
as the products of a distribution coefficient KD (L kg−1 ) and particle density. A
sediment-to-water KD of 100 000 L kg−1 was selected from estimates for seven
Wisconsin seepage lakes (KD ranged from 63 000 to 1.5 × 106 with a mean of
300 000, Watras et al., 1991). A pore water-to-sediment KD of 10 000 L kg−1 was
selected from estimates for Little Rock and Pallette Lakes (ranging from 2500 to
24 000, Babiaz and Andren, 1991, Hurley et al., 1994a). Table III lists values of
Xij , which were taken from the literature, and calculated values of Zij and Yij .
Hg was assumed to enter the lake from:

1. Direct deposition to the lake at the rate of 10 µg m−2 y−1 (Mierle, 1990;
Fitzgerald et al., 1991).
2. Runoff from the catchment area estimated as the fraction of atmospherically
deposited Hg exported from the watershed. The proportion of Hg exported
was correlated with the export of DOC (e.g., Mierle and Ingram, 1991; Lee
and Iverfeldt, 1991), which is a function of the proportion of wetlands in the
watershed (Eckhardt and Moore, 1990)

% exported = 8.6 + 0.76 (% wetlands) (11)

3. Direct emissions from local anthropogenic activities.


4. Natural sources such as surrounding geology (Rasmussen, 1994).
344 M. L. DIAMOND

TABLE III
Concentration fractions, Xij , Z values, aquivalence fractions, Yij , and data sources

Hg◦ HgII MeHg HgT

Concentration fractions, Xij

Air (gaseous) 0.99 Fitzgerald et al., 1991 0.00 by difference 0.01 Lee and Hultberg, 1990
Water: dissolved 0.03 Watras et al., 1994 0.92 by difference 0.05 Watras and Bloom, 1994
Watras et al., 1995b Watras et al., 1995a,b
Lee and Iverfeldt, 1991
particles 0.01 assumed 0.94 by difference 0.05 Hurley et al., 1994b
inflow 0.01 assumed 0.94 by difference 0.05 Hurley et al., 1995
Sediment: solids 0.01 assumed 0.98 by difference 0.01 Watras et al., 1994
Watras et al., 1995a
pore water 0.16 Hurley et al., 1994a 0.79 by difference 0.05 Krabben’t and
Babiarz, 1992

Z values (dimensionless)

Air, ZA 0.204 10−6 1.2×10−5


Water, ZW 1 1 1
Suspended 67000 204000 200000 200000
particles, ZP
Bottom 1250 25000 4000 2000
sediment, ZS

Aquivalence fractions, Yij

Air 0.01 0 0.99


Water 0.03 0.92 0.05
Sediment 0.16 0.79 0.05

In this application, only direct deposition and runoff were considered since data
were unavailable for direct emissions and geological inputs.
It was assumed that Hg entering a compartment instananeously adopts the de-
fined particulate-to-dissolved proportions (Parks et al., 1986) and species propor-
tions, i.e., there is no kinetic limitation on species interconversion. Using this
assumption simplifies the model considerably, but likely contravenes reality. Again,
the chemical dynamics in the lake and the proportions of species are assumed to be
at steady state, but chemical equilibrium is not assumed among species.
The model is applied illustratively to a generic lake located on the Canadian
Shield using lake dimensions described by Harris (1991) (Table IV). The lake is
DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL 345
TABLE IV
Summary of lake characteristics

Dimensions Water flows

Watershed area (ha) 1000 Inflow = outflow (m3 yr −1 ) 5×106


% wetlands 16
Lake surface area (ha) 100 Particle concentrations and density
Water mean depth (m) 10
Water volume (m3 ) 106 Suspended particles (mg L−1 ) 2
Sediment depth (m) 0.05 Sediment porosity 0.10
Sediment volume (m3 ) 5×104 Density (kg m−3 ) 2000

Mass transfer coefficients (m h−1 ) Particle movement (g m−2 d−1 )

Volatilization: air side 2 Deposition 1.5


water side 0.02 Resuspension 0.5
Sediment-water 2×10−4 Burial 0.3

assumed to consist of a well mixed water column underlain by a layer of active


sediment 5 cm deep. Values for suspended particle concentration, and rates of sed-
iment deposition, resuspension and burial are typical for an oligotrophic lake. Mass
transfer coefficients for volatilization and diffusion were taken from Mackay et al.
(1994), Krabbenhoft and Babiarz (1992) and Diamond (1995). An atmospheric
concentration of total Hg of 1.5 ng m−3 , typical of remote temperate locations, was
specified (Shakelton, 1994; Fitzgerald et al., 1991).

4. Results and Discussions

4.1. C OMPARISON OF OBSERVED AND MEASURED CONCENTRATIONS

Table V summarizes the estimated and measured species and total mercury con-
centration in water and sediment. Estimated concentrations for HgT , MeHg and
Hg◦ are well within the range of values reported for temperate lakes (e.g., Evans,
1986; Meili et al., 1991; Watras and Bloom, 1994) and close to those predicted
by Harris (1991). The correspondence of model results with available data is en-
couraging considering the results rest on the physical-chemical properties of Hg
species and our understanding of lake processes, not withstanding the uncertainties
and assumptions contained in the model, and lack of calibration or testing.
346
TABLE V
Comparison of measured (M) and estimated (E) Hg concentrations, with data sources

Hg◦ HgII MeHg Total Hg

Water (ng L−1 )

Bulk M 0.020–0.2 Fitzgerald et al., 1991 – 0.05–0.3 Watras et al., 1991 2–10 Meili et al., 1991
0.017–0.157 Watras et al., 1994 – 0.05–0.5 Watras and Bloom, 1994 0.5–5 Watras and Bloom, 1994

M. L. DIAMOND
– – 0.04–2.2 Watras et al., 1995b 0.2–5 Watras et al., 1995b
– – – 0.6–2 Gill and Bruland, 1990
E 0.04 1.54 0.08 1.66
Dissolved M – – – 0.4–1.8 Gill and Bruland, 1990
E 0.04 1.28 0.07 1.40
Particle M – – 0.033–0.14 Hurley et al., 1994b 0.21–0.6 Hurley et al., 1994b
– – – 0.18–2 Gill and Bruland, 1990
E 0.003 0.26 0.01 0.28
Particle M – – 47–160 Hurley et al., 1994b 120–680 Hurley et al., 1994b
(ng g−1 ) E 1.4 130 7 139
TABLE V

DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL


(continued)

Hg◦ HgII MeHg Total Hg

Sediment

Solid M – – 1–5 Watras et al., 1995a 90–100 Water et al., 1995a


(ng g−1 ) – – – 120–700 Evans, 1986
– – – 50–550 Rekolainen et al., 1986
– – – 34–753 Sorensen et al., 1990
– – – 28–170 Babiarz and Andren, 1991
E 1.7 164 1.7 168
Pore water M – – 0.15 Krabbenhoft, 1991 5–70 Babiarz and Andren, 1991
(ng L−1 ) – – 0–0.3 Krabbenhoft and Babiarz, 1992 4–20 Krabbenhoft and Babiarz, 1992
– – – 10–30 Hurley et al., 1994a
E 2.4 12 0.7 15

347
348
M. L. DIAMOND
Figure 1. Mass balance of Hg species and total Hg in a generic shield lake.
DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL 349

4.2. M ASS BALANCES OF TOTAL AND HG SPECIES

Figure 1 illustrates Hg dynamics in a hypothetical Shield lake. Most HgT orig-


inates from watershed runoff, although this is entirely a function of the ratio of
watershed to lake area and the percentage of wetlands specified. Of the incoming
31 g yr−1 , about 18 g yr−1 or 60% is buried, 4 g yr−1 or 13% volatilizes, and
8 g yr−1 or 27% is exported. This results in water and sediment concentrations
of 1.7 ng L−1 and 168 ng g−1 , respectively. Whereas 60% of the incoming HgT
is buried, a much greater proportion exchanges between the water and sediments.
Sediment deposition delivers about 76 g yr−1 to the sediments of which 31 and
27 g yr−1 return to the water column by sediment resuspension and net sediment-
water diffusion, respectively. The high rate of diffusion is due to the relatively low
KSW , which results in a pore water concentration of 15 ng L−1 , which again is
well within the range of measured values. The net rate of sediment-water diffusion
estimated by the model (27 µg m−2 yr−1 ) is about one-third higher than the rate
of 17 µg m−2 yr−1 calculated by Krabbenhoft and Babiarz (1992) for the assumed
outflow area of Pallette Lake, a seepage lake in Wisconsin. Thus, the amount of
HgT that cycles and recycles between water and sediments is much greater than
the annual loading to the lake, similarly to other chemicals in relatively shallow
systems (Diamond et al., 1996). These estimates assume that the pE of the active
sediment layer is about 0 and that the lake is circumneutral since under reducing
conditions, particularly at low pH, Hg◦ should, theoretically, be the predominant
species in pore water (Hurley et al., 1994a). The diffusive flux would decrease with
a lower concentration fraction of Hg◦ in pore water.
Hg◦ , HgII, and MeHg experience significantly different fates (Figure 1). Hg◦
and MeHg comprise 1 and 5%, respectively, and HgII 94% of the 31 g yr−1 HgT
entering the lake (according to the specified inflow concentration fractions). Con-
sequently, HgII dominates the export, burial and sediment-water exchange fluxes
of HgT . About 13% of HgT inputs is lost through volatilization of Hg◦ , that is sup-
ported by the conversion of HgII (3.2 g yr−1 and MeHg (1.4 g yr−1 ). Thus, although
negligible Hg◦ is estimated to enter the lake, within lake conversion supplies Hg◦
that is lost to the atmosphere. Since most Hg◦ volatilizes, Hg◦ contributes only
about 1% of HgT in the lake’s water and sediments. The predicted volatilization
rate of 13% of annual inputs is within the range of 10 to 50% reported for temper-
ate Wisconsin lakes (Vandal et al., 1991; Watras et al., 1994). The rate, which is
equivalent to 4 µg m−2 yr−1 , is an order of magnitude lower than the equivalent
annual volatilization rate of 55 ± 31 µg m−2 yr−1 measured during summer by
Schroeder et al. (1989), but within the range of the equivalent annual rates of 1.7 to
91 µg m−2 yr−1 measured during days in winter and summer by Xiao et al. (1991)
over lakes in southern Sweden. Several mechanisms have been proposed for the
reduction of HgII to Hg◦ , such as reduction in the presence of humic substances
(Alberts et al., 1974; Allard and Arsenie, 1991), via photochemical reaction(s) that
could involve humic substances (Xiao et al., 1991), biological or photochemical
350 M. L. DIAMOND

TABLE VI
Estimated chemical persistence and half lives (yr)

Hg◦ HgII MeHg Total Hg


Persist. Half-life Persist. Half-life Persist. Half-life Persist. Half-life

Water 0.1 0.07 0.6 0.4 7.9 5.5 0.5 0.35


Sediment 92 64 92 64 92 64 92 64
Overall 3.5 2.4 65 45 168 116 55 38

reduction involving visible or UVA radiation (Amyot et al., 1994), or biologically


mediated reduction as a detoxifying mechanism (Silver, 1984).
The model suggests that 1 g yr−1 MeHg enters from watershed runoff and 0.5 g
yr−1 from direct atmospheric deposition (wet and dry deposition). Based on the
assumption that about 1% of gaseous HgT is MeHg, which is deduced from MeHg
contributions to wet and dry deposition (e.g., Lee and Hultberg, 1990; Watras et al.,
1995a), an estimated 0.5 g yr−1 is adsorbed from air to water. While MeHg inputs
from the watershed and from atmospheric deposition have been well documented
(e.g., Rudd, 1995; Bloom and Watras, 1989), air-water absorption has not been.
Within the lake, about two-thirds of inputs are lost by export and one third by
burial. These results, however, do not account for MeHg bioaccumulation which
is likely its dominant fate (Watras et al., 1994). Bioaccumulation would result
in maintaining MeHg in the lake, thereby decreasing MeHg export and burial.
As well, bioaccumulation or the sequestering of MeHg by biota, would decrease
the dissolved concentration in the water column which could, in turn, increase
absorption across the air-water interface.
Due to the differing fates of each species, their persistence or half-life in the
lake varies (Table VI). Persistence refers to the time required for concentrations
and amounts to fall to 37% of estimated values given no additional inputs and, as
discussed by Mackay (1991), is the same as that provided by a time dependent
model. Persistence is calculated as the ratio of the amount of chemical M (g) to in-
puts I that equal outputs O (g yr−1 ) (as the model is steady state). To accommodate
species interconversion, species-specific persistence is calculated as M O−1 . The
persistences of Hg◦ , HgII and HgT in the water column are significantly less than
that of the water itself (water residence time of 2 yr). This is due to rapid chem-
ical removal by sediment deposition, and for Hg◦ , removal by volatilization. The
equivalent half-life for HgT of 0.35 yr is within the range of 0.2 to 0.6 estimated
by Sorensen et al. (1990) for 34 drainage lakes in northern Minnesota.
Persistence of Hg in the sediments of 91 yr is equal to that of the sediment
layer itself. Thus, similarly to other conservative or slowly degrading chemicals,
Hg will remain in the sediments where it can act as a source to the water column
DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL 351
TABLE VII
Parameter values used in sensitivity analysis

Parameter Scenario

Trophic status Base Oligotrophic Eutrophic

Suspended solids (mg L−1 ) 2 0.5 5


Deposition (g m−2 yr) 1.5 0.3 6
Resuspension (g m−2 yr) 0.5 0.1 2
Burial (g m−2 yr) 0.3 0.06 0.5

Hydrology Base Long Short

Water residence time (yr) 2 8 0.5

Sorption Base Low High

KD (L kg−1 ) water column 100000 50000 200000


KD (L kg−1 ) sediments 10000 5000 20000

Speciation Base Low High

Hg◦ (%) 3 1 6

(e.g., Parks et al., 1986). The overall residence time of Hg in the system is between
50 and 60 yr reflecting the short residence time in the water retarded by the long
residence time in the sediments.

4.3. S ENSITIVITY ANALYSIS

Hg dynamics depend on lake characteristics, in addition to chemical properties


and species proportions. To illustrate this dependence, a sensitivity analysis was
conducted using the model to simulate various scenarios. The model was run for
six scenarios, the base case (as described above), oligotrophic conditions (low
suspended particle concentration, and low sediment deposition, resuspension and
burial rates), eutrophic conditions (the converse of the oligotrophic system), low
and high flow (water inflow and outflow decreased and increased four times, re-
spectively), and low and high sorptive capacity of suspended particles and bottom
sediments (halve and double ZP and ZS ). The latter scenarios would occur in lakes
with sediments having either a high sand (quartz and silica) content and minimal
organic matter or other ligands that complex Hg, or the converse (e.g., Lindberg
and Harris, 1974; Wang et al., 1989; Coquery and Welbourn, 1995). Table VII
352 M. L. DIAMOND

Figure 2. Sensitivity analysis: Hg concentrations in water and sediments.

summarizes the parameter values used in each scenario. Loadings are held constant
in all cases, including those in which water residence time (or flows) are changed.
Results of previous sensitivity analyses suggest that the QWASI model is sur-
prisingly robust to changes in the physical-chemical properties of chemicals, but
is sensitive to changes in the amount or sorptive capacity of suspended and bed
sediments (Diamond, 1995; Diamond et al., 1996).
The results of the analysis suggest that Hg water concentrations are highest
(>3 ng L−1 ) under oligotrophic conditions, in which minimal suspended parti-
cles are available to convey Hg to the sediments (Figure 2). Increased rates of
DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL
353
Figure 3. Sensitivity analysis: summary of rates of Hg movement.
354 M. L. DIAMOND

volatilization (due to higher dissolved concentrations) and export are insufficient


to reduce concentrations. This result is consistent with reports of greater Hg fish
concentrations in oligotrophic lakes (Lindqvist et al., 1991). Conversely, low water
concentrations (about 1 ng L−1 ) prevail under eutrophic and high sorption con-
ditions in which either the abundant suspended particles or their high sorptive
capacity for Hg result in high Hg sedimentation and burial rates, i.e., burial re-
moves about 80% of Hg inputs in eutrophic and high sorption scenarios (Figure 3).
Despite a high removal rate through burial, sediment concentrations are highest in
the high sorption scenario, explained by the greater mass of Hg that accumulates
under the high sorption versus eutrophic scenario (2.2 and 1.4 kg, respectively,
Figure 3).
Water and sediment concentrations (2 ng L−1 and about 200 ng g−1 , respec-
tively), and HgT mass in the lake (2 kg) are all elevated under low flow conditions
(high water retention time) indicating the importance of advective flows in remov-
ing Hg from the system. Conversely, Hg concentrations (<1 ng L−1 in water and
100 ng g−1 in sediment) and the total amount (1 kg) are lowest under the high
flow scenario (low water retention time). These results contradict the negative
correlation between water retention time and Hg fish concentrations reported by
Hakanson et al. (1988) and Sorensen et al. (1990). The apparent contradiction may
be due to Hg loadings being positively correlated with flows in the lakes examined
by these authors, whereas loadings were constant for all scenarios examined here.
Finally, a sensitivity analysis was used to determine the importance of species
fractions on estimated concentrations and chemical fate. The results indicate that
concentrations are most sensitive to the percent Hg◦ specified. With Hg◦ at 1% of
HgT , water concentrations increased to almost 2 ng L−1 and sediment concentra-
tions to 194 ng g−1 . The increase is due to negligible conversion of HgII to Hg◦ with
attendant volatilization. Thus, as Fitzgerald et al. (1994) suggested, volatilization
of Hg◦ may be an important ‘release’ mechanism that limits Hg concentrations in
water and, to a lesser extent, sediment.
Conversely, the release implies that the volatilized Hg is available for re-deposi-
tion into aquatic or terrestrial systems, rather than entering a more permanent sink
in the sediments. In this way, Hg is difficult to control as it cycles and re-cycles
between air and water or land, as is the case with persistent organics (e.g., PCBs).
However, unlike many persistent organics, Hg, with its relatively low sorptive affin-
ity for sediments, can also cycle and re-cycle between water and sediments, and be
available for biotic uptake, export or re-volatilization.

5. Conclusions

The simple model presented here provides estimates of HgT and species-specific
concentrations in water and sediments, while requiring minimal lake- and chemical-
specific data. Obtaining reliable results, including those for individual species, rests
DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL 355

on specified species concentration fractions derived from system-specific measure-


ments. Thus, the model does not require a priori knowledge of species interconver-
sion rates, which are poorly known for most systems. The model can be improved
by predicting species concentration fractions and accounting for the proportion of
HgII that is ‘reactive’ or available for interconversion (Fitzgerald et al., 1994).
The model was applied to a generic lake typical of the Canadian Shield. Esti-
mated water and sediment concentrations were within the range of values reported
for remote lakes: this correspondence was achieved using generic lake charac-
teristics and the physical-chemical properties of Hg species, but without model
calibration. Model results indicate that Hg◦ , HgII and MeHg experience different
fates and persistence that are consistent with field observations. The fate of HgT
is dominated by that of HgII, which constitutes most Hg in remote, ‘non-polluted’
systems. Most HgT is retained in the sediments through burial, with export being
of secondary importance. In-lake conversion of HgII to Hg◦ which volatilizes,
provides an important release mechanism that lowers HgT water concentrations.
These results are supported by a sensitivity analysis suggesting that Hg water
concentrations ought to be highest in oligotrophic lakes and lakes having high
water retention times. The greatest amount of Hg will accumulate in lakes with
sediments having a high sorptive capacity, i.e., rich in organic matter.
Perhaps the greatest value of the results presented lies in a comparative sense
rather than in assessing the absolute value of estimated Hg concentrations and
amounts. The conclusion suggested by comparing scenarios such as outlined in the
sensitivity analysis is that lake-specific factors, such as water retention time, can
play a role in controlling Hg concentrations and hence the amount of Hg available
for methylation.

Acknowledgements

Judy Josefowicz ably assisted with preparing the manuscript.

References

Alberts, J. J., Schindler, J. E., Miller, R. W. and Nutter, D. E.: 1974, Science 184, 895.
Allard, B. and Arsenie, I.: 1991, Water, Air, and Soil Pollut. 56, 457.
Amyot, M., Mierle, G., Lean, D. R. S. and McQueen, D. J.: 1994, Environ. Sci. Technol. 28, 2366.
Babiarz, C. L. and Andren, A. W.: 1991, in Mercury in Temperate Lakes 1990 Annual Report,
Prepared for Electric Power Research Institute, Palo Alto, CA.
Bloom, N. S. and Effler, S. W.: 1990, Water, Air, and Soil Pollut. 53, 251.
Bloom, N. S. and Watras, C. J.: 1989, Sci. Total Environ. 87/88, 199.
Bodaly, R. A., Rudd, J. W. M., Fudge, R. J. P . and Kelly, C. A.: 1993, Can. J. Fish. Aquat. Sci. 50,
980.
Brosset, C.: 1987, Water, Air, and Soil Pollut. 34, 145.
Coquery, M. and Welbourn, P. M.:1995, Water Res. 29, 2094.
356 M. L. DIAMOND

Diamond, M. L.: 1995, Environ. Sci. Technol. 28, 29.


Diamond, M. L., Mackay, D. and Welbourn, P. M.: 1992, Chemosphere 25, 1907.
Diamond, M. L. and Sang, S.: 1992, Abst., Soc. Environ. Toxicol. Chem. 13th Annual Meeting,
Cincinnati, OH.
Diamond, M. L., Mackay, D. and Sang, S.: 1993, Abst. 36th Conf. of Internat. Assoc. for Great Lakes
Research, Depere, WI.
Diamond, M. L., Mackay, D., Poulton, D. J. and Stride, F. A.: 1996, Water Res. 30, 405.
Di Guardo, A., Calamari, D., Zanin, G., Consalter, A. and Mackay, D.: 1994, Chemosphere 28, 511.
Driscoll, C. T., Yan, C., Schofield, C. L., Munson, R. and Holsapple, J.: 1994, Environ. Sci. Technol.
28, 136A.
Eckhardt, B. W. and Moore, T. R.: 1990, Can. J. Fish. Aquat. Sci. 47, 1537.
Evans, D. R.: 1986, Arch. Environ. Contam. Tox. 15, 505.
Fitzgerald, W. F., Mason, R. P. and Vandal, G. M.: 1991, Water, Air, and Soil Pollut. 56, 745.
Fitzgerald, W. F., Mason, R. P., Vandal, G. M. and Dulac, F.: 1994, in C. J. Watras and J. W. Huckabee
(eds.), Mercury Pollution: Integration and Synthesis, Lewis Publ., Boca Rotan, FL, pp. 203–220.
Gill, G. A. and Bruland, K. W.: 1990, Environ. Sci. Technol. 24, 1392.
Gilmour, C. C. and Henry, E. A.: 1991, Environ. Pollut. 71, 131.
Grieb, T. M., Driscoll, C. T., Gloss, S. P., Schofield, C. L., Bowie, G. L. and Porcella, D. B.: 1990,
Environ. Toxicol. Chem. 9, 919.
Hakanson, L.: 1990, Limnologica 20, 339.
Hakanson, L., Nilsson, A. and Andersson, T.: 1988, Environ. Pollut. 49, 145.
Hakanson, L., Andersson, T. and Nilsson, A.: 1990, Water, Air, and Soil Pollut. 50, 171.
Harris, R. C.: 1991, A Mechanistic Model to Examine Mercury in Aquatic Systems, M. Eng. Thesis,
McMaster Univ., Hamilton, Ontario.
Hudson, R. J. M., Gherini, S. A., Watras, C. J. and Porcella, D. B.: 1994, in C. J. Watras and J.
W. Huckabee (eds.), Mercury Pollution: Integration and Synthesis, Lewis Publ. Boca Raton, FL,
pp. 473–523.
Hurley, J. P., Krabbenhoft, D. B., Babiarz, C. L. and Andren, A. W.: 1994a, in L. A. Baker (ed.),
Environmental Chemistry of Lakes and Reservoirs, ACS Advances in Chemistry Series No. 237,
American Chemical Society, pp. 425–449.
Hurley, J. P., Watras, C. J. and Bloom, N. S.: 1994b, in C. J. Watras and J. W. Huckabee (eds.),
Mercury Pollution: Integration and Synthesis, Lewis Publ. Boca Raton, FL, pp. 69–82.
Hurley, J. P., Benoit, J. M., Babiarz, C. L., Shafer, M. M., Andren, A. W., Sullivan, J. R., Hammond,
R. and Webb, D. A.: 1995, Environ. Sci. Technol. 29, 1867.
Krabbenhoft, D. P.: 1991, in Mercury in Temperate Lakes 1990 Annual Report, Prepared for Electric
Power Research Institute, Palo Alto, CA.
Krabbenhoft, D. P. and Babairz, C. L.: 1992, Water Resource Res. 29, 3119.
Lee, Y. H. and Hultberg, H.: 1990, Environ. Toxicol. Chem. 9, 883.
Lee, Y. H. and Iverfeldt, A.: 1991, Water, Air, and Soil Pollut. 56, 309.
Lindberg, S. E. and Harriss, R. C.: 1974, Environ. Sci. Technol. 8, 459.
Lindqvist, O.: 1991, Water, Air, and Soil Pollut. 55, 1.
Lindqvist, O., Johansson, K., Aastrup, M., Andersson, A., Bringmark, L., Hovenius, G., Hakanson,
L., Iverfeldt, A., Meili, M. and Timm, B.: 1991, Water, Air, and Soil Pollut. 56, 333.
Mackay, D.: 1991, Multimedia Environmental Models, Lewis Publ., Chelsea, MI.
Mackay, D., Paterson, S. and Joy, M.: 1983, Chemosphere 12, 981.
Mackay, D. and Diamond, M. L.: 1989, Chemosphere 18, 1343.
Mackay, D. and Paterson, S.: 1991, Environ. Sci. Technol. 25, 427.
Mackay, D., Sang, S., Vlahos, P., Diamond, M., Gobas, F. and Dolan, D.: 1994, J. Great Lakes Res.
20, 625.
Mackay, D., Wania, F. and Schroeder, W. H.: 1995, Water, Air, and Soil Pollut. 80, 941.
DEVELOPMENT OF A FUGACITY/AQUIVALENCE MODEL 357

McMurtry, J. L., Wales, D. L., Scheider, W. A., Beggs, G. L. and Dimond, P. E.: 1989, Can. J. Fish.
Aquat. Sci. 46, 426.
Meger, S. A.: 1986, Water, Air, and Soil Pollut. 30, 411.
Meili, M.: 1988, in P. Mathy (ed.), Air Pollution and Ecosystems, Proc. Internat. Symp. Comm.
European Comm., D. Reidel Publ. Co., Dordrecht, Holland, pp. 940–947.
Meili, M., Iverfeldt, A. and Hakanson, L.: 1991, Water, Air, and Soil Pollut, 56, 439.
Mierle, G.: 1990, Environ. Toxicol. Chem. 9, 843.
Mierle, G. and Ingram, R.: 1991, Water, Air, and Soil Pollut. 56, 349.
OMOEE and OMNR or Ontario Ministry of the Environment and Energy and Ontario Ministry of
Natural Resources: 1995, Guide to Eating Ontario Sport Fish, Toronto, Ontario, 296 pp.
Parks, J. W., Sutton, J. A. and Lutz, A.: 1986, Can. J. Fish. Aquat. Sci. 43, 1426.
Parks, J. W.: 1988, Water, Air, and Soil Pollut. 42, 267.
Rasmussen, P. E.: 1994, Environ. Sci. Technol. 28, 2233.
Rekolainen, S., Verta, M. and Liehu: 1986, The Effect of Airborne Mercury and Peatland Drainage on
Sediment Mercury Contents in Some Finnish Forest Lakes, Publ. of the Water Res. Inst., National
Board of Waters, Finland. No. 65.
Rudd, J. W. M.: 1995, Water, Air, and Soil Pollut. 80, 697.
Schroder, W. H., Munthe, J. and Lindqvist, O.: 1989, Water, Air, and Soil Pollut. 48, 337.
Shakelton, M. N.: 1994, Abst, CIRAC/AWMA-OS Joint International Conference on Atmospheric
Chemistry, Atmospheric Pathways for Toxic Substances, Toronto, Ontario.
Silver, S.: 1984, in J. O. Nriagu (ed.), Changing Metals Cycles and Human Health, Springer-Verlag,
New York, pp. 199–223.
Sorensen, J. A., Glass, G. E., Schmidt, K. W., Huber, J. K. and Rapp, G. R., Jr.: 1990, Environ. Sci.
Techn. 24, 1716.
Swain, E. B., Engstrom, D. R., Brigham, M. E., Henning, T. A. and Brezonik, P. L.: 1992, Science
257, 784.
Suns, K. and Hitchin, G.: 1990, Water, Air, and Soil Pollut. 50, 255.
Vandal, G. M., Mason, R. P. and Fitzgerald, W. F.: 1991, Water, Air, and Soil Pollut. 56, 791.
Wang, J. S., Huang, P. M., Hammer, U. T. and Liaw, W. K.: 1989, in J. A. Nriagu (ed.), Aquatic
Toxicology and Water Quality Management, J. Wiley and Sons, pp. 153–159.
Watras, C. J., Hurley, J. P., Claas, S., Morrison, K., Bloom, N. and Hoffman, T. L.: 1991, in Mercury
in Temperate Lakes 1990 Annual Report, Prepared for Electric Power Research Institute, Palo
Alto, CA.
Watras, C. J. and Bloom, N. S.: 1994, in C. J. Watras and J. W. Huckabee (eds.), Mercury Pollution:
Integration and Synthesis, Lewis Publ. Boca Raton, FL, pp. 137–152.
Watras, C. J. et al.: 1994, in C. J. Watras and J. W. Huckabee (eds.), Mercury Pollution: Integration
and Synthesis, Lewis Publ. Boca Raton, FL, pp. 153–277.
Watras, C. J., Morrison, K. A. and Bloom, N. S.: 1995a, Can. J. Fish. Aquat. Sci. 52, 1220.
Watras, C. J., Morrison, K. A., Host, J. S. and Bloom, N. S.: 1995b, Limnol. Oceanogr. 40, 556.
Weiner, J. G. and Stokes, P. M.: 1990, Environ. Toxicol. Chem. 9, 821.
Winfrey, M. R. and Rudd, J. W. M.: 1990, Environ. Toxicol. Chem. 9, 853.
Xiao, Z. F., Munthe, J., Schroeder, W. H. and Lindqvist, O.: 1991, Tellus 43B, 267.
Xun, L., Campbell, N. E. R. and Rudd, J. W. M.: 1987, Can. J. Fish. Aquat. Sci. 44, 750.
Yarwood, G. and Niki, H.: 1990, A Critical Review of Available Information on Transformation
Pathways for Mercury Species in the Atmospheric Environment, Prepared for Atmospheric Envi-
ron. Service, Environ. Canada, Supply and Services Canada, Science Procurement Contract No.
KM171-9-0129.

You might also like