Content uploaded by Samuel Jackson Bentley
Author content
All content in this area was uploaded by Samuel Jackson Bentley on Nov 25, 2014
Content may be subject to copyright.
Bioirrigation modeling in experimental benthic mesocosms
by Yoko Furukawa
1
, Samuel J. Bentley
2
and Dawn L. Lavoie
1
ABSTRACT
Burrow irrigation by benthic infauna affects chemical mass transfer regimes in marine and
estuarine sediments. The bioirrigation facilitates rapid exchange of solutes between oxygenated
overlying water and anoxic pore water, and thus promotes biogeochemical reactions that include
degradation of sedimentary organic matter and reoxidation of reduced species. A comprehensive
understanding of chemical mass transfer processes in aquatic sediments thus requires a proper
treatment of bioirrigation. We investigated bioirrigation processes during early diagenesis using
laboratory benthic mesocosms. Bioirrigation was carried out in the mesocosms by Schizocardium
sp., a funnel-feeding enteropneust hemichordate that builds and ventilates a U-shaped burrow.
Interpr etation of the laborato ry results was aided by a two- dimensional multi component mo del for
transport and reactions that explicitly accounts for the depth-dependent distribution of burrows as
well as the chemical mass transfers in the immediate vicinity of burrow walls. Our study shows that
bioirrig ation sign i cant ly affec ts th e sp atial d istributio ns of pore wa ter solutes . Moreo ver, bio irriga-
tion promotes burrow walls to be the site of steep geochemical gradients and rapid chemical mass
transfer. Our results also indicate that the exchange function, a, widely used in one-dimensional
bioirrigation modeling, can accurately describe the bioirrigation regimes if its depth attenuation is
coupled to the depth-dependent distribution of burrows. In addition, this study shows that the
multicomponent 2D reaction-transportmodel is a useful research tool that can be used to critically
evaluatecommon biogeochemicalassumptionssuch as the prescribeddepth dependenciesof organic
matter degradation rate and C/N ratio, as well as the lack of macrofaunal contribution of metabolites
to the pore water.
1. Introduction
Many macroinvertebrates inhabit benthic boundary layers in marine and estuarine
environments. Some of these animals construct burrows as their habitats, which they
ventilate with O
2
-rich overlying water (i.e., bioirrigation). Consequently, O
2
and other
electron acceptors are introduced to sediments that are well away from water-sediment
interface (WSI), and metabolites such as dissolved inorganic carbon and ammonium are
removed to overlying water. Previous studies have recognized the signi cance o f this
bioirrigation process in sedimentary early diagenesis (e.g., Aller, 1982; Kristensen, 1988;
Marinelli, 1992; Martin and Banta, 1992; Emerson et al., 1984; Aller and Aller, 1998;
1. Naval Research Laboratory, S ea oor Sciences Branch, Stennis Space C enter, M ississippi, 39529, U.S.A.
email: yoko.furukawa@nrlssc.navy.mil
2. Louisiana State University , Coastal Studies Institute, Baton Roug e, Louisiana, 7080 3, U.S.A.
Journal of Marine Research, 59, 417–452, 2001
417
Furukawa et al., 2000). In these studies, bioirrigation is found to quantitatively affect the
microbial remineralization reactions and solute uxes. Comprehensive understanding of
the ch emical mass transfer in marine and estuarine en vironment s thus req uires a mec hanis-
tic and quantitativeunderstanding of the bioirrigation processes.
Today, the common quantitative treatment of bioirrigation involves a mathematical
expression of one-dimensional nonlocal exchange (e.g., Emerson et al., 1984; Boudreau,
1984; Martin and Banta, 1992). The measure for this exchange, nonlocal exchange
function a, is usually described as a simple function (e.g., constant, linear decrease,
exponential decay) of depth. Whereas this modeling strategy often produces agreement
between measured and modeled depth pro les of solutes (e.g., Matisof and Wang, 1998;
Kristensen and Hansen, 1999; Schlu¨ter et al., 2000), the mathematical formulae and
“best- t” parameter values for ado not allow deconvolution of actual mechanistic steps
involved in the bioirrigation processes (Meile et al., 2001). Moreover, this type of 1D
treatment gives little insights to the lateral spatial variability in chemical mass transfer
regimes associated with the burrows.
The most mechanistic model treatment of bioirrigation to date was established by Aller
(1980, 1982, 1984). In his so-called cylinder model, bioirrigated sediment is idealized as a
collection of laterally close-packed, identical cylinders, each with a cylindrical void space
(i.e., burrow) of an identical geometry in the center (Fig. 1). The outer surface of each
microenvironment (i.e., r5r
2
) represents the midpoint between two adjacent burrows,
and thus the point of zero radial ux. In this model, bioirrigation is mathematically
expressed as the radial diffusive exchange of solutes perpendicular to the burrow wall. The
concentration of a given solute at a given position at a given time, C
x
,r
,t
, can be determined
by the equation that describes vertical diffusion perpendicular to the water-sediment
Figure 1. The sch ematic model ge ometry of Alle r’s cylind er model (All er, 1980 ), with r15burrow
radius , and r25half-distancebetween burrows.
418 [59, 3
Journal of Marine Research
interface (WSI), radial diffusion perpendicular to the burrow wall, and net rate of
production or consumption of the solute due to biogeochemicalreactions:
w]Cx,r,t
]t5]
]x
X
wDsw
u2
]Cx,r,t
]x
D
11
r
]
]r
X
rwDsw
u2
]Cx,r,t
]r
D
1R(1)
(Aller, 1980; Boudreau, 1997, p. 63) where w [ porosity, t[time, D
sw
[diffusion
coef cient of the solute in seawater, u [ diffusive tortuosity, and R[overall rate of
production/consumption reactions. The diffusive tortuosity term in the equation can be
replaced by a porosity expression using the following empirical correlation:
u2512ln ~w2!(2)
(Boudreau, 1997, p. 132). The model originally presented by Aller (1980) assumes that the
burrow water composition is always the same as that of overlying water (i.e., 100%
irrigation). The original model geometry also assumes that all burrows have the identical
radius and depth extent, and are equally spaced. Boudreau and Marinelli (1994) extended
the model to allow periodic, discontinuous irrigation in which the burrow water composi-
tion starts to equilibrate with surrounding pore water while the burrows are not actively
irrigated. This model re ects the observations that many infauna species go through
alternating ventilation and rest cycles (e.g., Kristensen, 2000). Furukawa et al. (2001)
incorporated the metabolite contributionby burrowing macrofauna into the discontinuous
irrigation model.
The advantage to such a 2D modeling approach over the use of simple 1D nonlocal
exchange functions is that 2D models allow us to quantify the spatial and temporal
heterogeneity in sediment geochemistry. For example, Aller (1980) showed the wide range
of lateral variability in pore water compositions as a function of distance from the model
burrow wall using the original cylinder model. Marinelli and Boudreau (1996) used the
discontinuous irrigation model as well as an idealized laboratory microenvironment to
illustrate the steep gradients in redox and pH conditions in the immediate vicinity of
burrow walls. Furukawa et al. (2001) calculated the possible lateral variability in the
thermodynamic stabilities of calcium carbonate minerals due to bioirrigation and infauna
metabolism using the model that incorporated discontinuous irrigation and macrofaunal
metaboliteproduction.
Whereas the original cylinder model and its derivatives provide insights to the processes
that occur in association with burrows, their application to actual sediments is limited. This
is partly because of the assumption that all burrows are vertical and have the identical depth
extent. In reality, more than one species of burrowing infauna usually inhabit a given
sedimen tary area, cre ating burrows of v arious depth exte nts and ge ometries (e.g., D’Andrea
and Lopez, 1997; Levin et al., 1999). Many species of burrowing infauna are known to
create burrows with complex geometry that may not be adequately approximated by
vertica l cylinders (e.g ., Davey, 1994; Rowden and Jones, 1995; Ziebis et al., 19 96; Scaps et
2001] 419
Furukawa et al.: Bioirrigation modeling
al., 1998). A bioirrigation model would be more applicable to actual sediments if it were
able to account for the variable depth extents and tilt angles of burrows.
Our strat egy towa rd th e quan titative an d mech anistic u nderstandin g of b ioirrigatio n thus
involvesthe construction of a 2D bioirrigation model, with a simultaneous data collection
in lab oratory bent hic mesocosms. The mo del formulation takes into acco unt the para meters
that were previously considered (e.g., burrow radius, number of burrows per unit area of
seabed; Aller, 1980; Boudreau and Marinelli, 1994; Marinelli and Boudreau, 1996). In
addition, our model considers other parameters that are important in the adequate
description of eld and laboratory observations, such as the depth-dependent distribution
of burrows and burrow tilt angles. The laboratory experiments provide directly measured
data for constraining the model geometry and boundary chemical values. They also supply
depth pro les of solute species that are used to evaluate the model outputs. Consequently,
the simulation results demonstrate the quantitative signi cance of depth-dependentbioirri-
gation in terms of net chemical mass transfer. Moreover, the model results illustrate the
lateral variability in chemical mass transfer regimes especially in the vicinity of burrow
walls. This study also demonstrates the model as an evaluation tool for common
assumptions used in biogeochemical studies of early diagenesis, such as the prescribed
depth d ependenci es of organ ic matter (OM ) d egradation ra te and C /N ratio .
2. The model
a. Model geometry
We have extended Aller’s (1980) cylinder model (Fig. 1) to consider sediments with
variable burrow depths and tilt angles. Our model represents burrowed sediment using a
cylindrical microenvironment with a void space (i.e., “model burrow”) in center (Fig. 2a).
The cylinders are arranged in the lateral close packing as in Figure 1(a). Whereas the radius
of micro environmen t (r
2
) re mains co nstant th roughout th e dept h interval o f consi deration,
the radius of model burrow is a function of depth (i.e., r
1
(x)). The value for r
2
is
determi ned by the number of burrow openings at WSI (n(m
2
2
)) as follows:
r25
Î
1
2Î33n~m!. (3)
Each microenvironment has the depth of L
total
and the model burrow has the depth of
L
burrow
.
The function and parameters for r
1
(x) are selected such that the surface area of model
burrow wall at the depth interval x;x1d x (Fig. 2b and c),
E
x
x1dx
2pr1~x!dx (4)
420 [59, 3
Journal of Marine Research
describes the surface area of burrow wall interface at the depth interval of x;x1dx for
the sediment microenvironment that is being modeled. Thus, the value of r
1
at a given
depth xre ects three factors: (1) the ratio of burrow density at the given depth to burrow
density at WSI; (2) the burrow radii; and (3) the tilt angles of the burrows. Higher burrow
density yields more burrow wall surface area available for radial diffusive exchange, and
greater r
1
value. The l arger averag e burrow ra dius also tran slates to th e larger r
1
value. Th e
surface area of burrow wall interface for a vertical burrow within the nite depth interval
(dx ) is 2pr
burrow
dx. On the other hand, the burrow wall interface area of a burrow tilted by
angle uis greater, and given by 2pr
burrow
dx/cos u(Fig. 3). Hence, the greater the tilt angle
u, the greater the available burrow wall interface surface area for the given depth interval.
Figure 4 illustrates how this model geometry can be applied to real sediment. In this
example, all burrows have the identical radius (r
burrow
). First, the density of burrow
openings at WSI ( n(m
2
2
)) determines the outer radius of microenvironment, r
2
, to be
Î
1
2Î33n(m) (Eq. 3). The burrow wall area for this microenvironment within the top dx
(m) is given by 2pr
burrow
dx /cos u, when the average tilt angle of all burrows at WSI is u.
The surface area of this burrow can be represented by a vertical model burrow of the same
interface area whose radius is r
1
:
r1,SWI 5rburrow
cos u. (5)
Similarly, the value of r
1
at depth xcan be determined using the burrow density at depth x,
pn (m
2
2
) ( p,1 ), and average burrow tilt angle at depth x,u
x
, to be:
Figure 2. The schematic model geometry of the depth-dependentbioirrigation model. The number of
burrows decreases with depth in natural sediments, and it is represented by a microenvironment
whose bu rrow radi us decre ases with de pth.
2001] 421
Furukawa et al.: Bioirrigation modeling
r1~x!5prburrow
cos ux
. (6)
This model geometry requires that complex burrow geometry be represented by a single
function, r
1
(x), using the depth-dependent change in burrow wall interface area as a guide.
This simpli cation is based on the observations that the primary effect of burrows in early
diagenetic chemical mass transfer is the increased amount of diffusive interface between
oxygen ated water and a noxic se diments (Fors ter, 19 96; Krist ensen, 20 00).
b. Model equations
We consider burrowed sediment with no advection (i.e., no sediment accumulation).
The no advection assumption is justi ed when r
2
is relatively small compared to L
burrow
Figure 3. The wall surface area of a tilted burrow is greater for a unit depth interval than that of a
vertica l burrow.
Figure 4. A schematic diagram showing how the geometry parameters for the cylinder model ( r2,
r1(x)) are determined.
422 [59, 3
Journal of Marine Research
(Aller, 1980). We also assume that there is no compaction, and thus porosity (w) has a
constant value throughout the sediments. Consequently, we can write an equation similar
to Eq. (1) for the new geometry. The equation for the solute concentration within sediment
(i.e., 0 #x#L
total
and 0 #r#r
2
, excluding the burrow: 0 #x#L
burrow
and 0 #r#
r
1
(x)) is:
]Cx,r,t
]t5Dsw
12ln ~w2!
]2Cx,r,t
]x21Dsw
12ln ~w2!
1
r
]
]r
X
r]Cx,r,t
]r
D
1R. (7)
Within the burrow (i.e., 0 #x#L
burrow
and 0 #r#r
1
(x)), the solute concentration is
held at the value of overlying water (C
0
) to represent 100% (continuous) irrigation.
Parameter values and functions needed for the steady state solution of Eq. (7) are: model
burrow radius as a function of depth (r
1
(x) (m)), porosity (w), diffusion coef cient (D
sw
(m
2
s
2
1
)), and net production rate (R(M s
2
1
)). The outer rad ius of microen vironment, r
2
(m) is derived from Eq. (3) using the observed burrow opening density at WSI (n(m
2
2
)).
Other geometry-related parameters include L
burrow
and L
total
. The burrow parameters
(n(x), r
1
(x), r
2
,L
burrow
) and porosity (w) can be determined through direct analysis of
sediment samples. The net production/consumption reaction rate of a given solute species
within the sediments (R) may be found in literature (i.e., Van Cappellen and Wang, 1996),
estimated from O
2
micropro les (Marinelli and Boudreau, 1996), or determined by batch
incubation experiments (Aller and Yingst, 1980). The diffusion coef cient ( D
sw
) can be
found in literature, such as Boudreau (1997, p. 96).
c. Boundary conditions
Boundary conditions needed for the numerical solution of Eq. (7) are the same as the
ones introduced by Aller (1980), Boudreau and Marinelli (1994) and Marinelli and
Boudreau (1996), and include the following. Solute concentration at the WSI (i.e., at x5
0) is equal to the bulk solute concentration of the overlying water (C
bulk
):
Cx50,r,t5C05Cbulk (8)
At burrow wall (r5r
1
(x)), solute concentration is also equal to the bulk solute
conce ntration of t he overlyi ng water (i. e., 100% irrigation):
Cx,r5r1~x!,t5C0. (9)
The exception to the above (8) and (9) is oxygen: due to the diffusive boundary layer (e.g.,
Jørgensen and Revsbech, 1985), the concentration value at WSI and burrow wall (C
0
) is
typic ally less than the bulk value in the overlying water (i.e., C
0
,C
bulk
). The value of C
0
for oxygen may be directly determined by inspecting measured O
2
micropro les (Fu-
rukawa et al., 2000). The solute diffusive ux at the outer boundary of each sedimentary
microenvironment, r
2
, is zero in the radial direction, as in Aller’s cylinder model (Aller,
1980):
2001] 423
Furukawa et al.: Bioirrigation modeling
]C
]rUr5r2
50. (10)
Vertical solute ux at the bottom boundary (at x5L
total
) can be set to zero using a
suf c iently large v alue for L
total
. Al ternatively, wh en simul ating irrigat ion in an expe rimen-
tal tank with a closed bottom, L
total
can be de ned as the bottom of the tank. In either case,
the following equation represents this boundary condition:
]C
]xUx5L
50. (11)
The i nitial co ncentratio n distributi on can be set arbitra rily when solvin g for stea dy state.
d. Net production rates and numerical solutions
The net producti on rate, R(M s
2
1
), for each species is determined through the couplings
of geochemical reactions that are recognized as important to overall chemical mass
transfer. The reactions considered here include OM remineralization by O
2
, NO
3
2
, and
SO
4
2
2
, reoxidation of NH
4
1
and SS (5H
2
S1HS
2
) by O
2
, rapid NH
4
1
adsorption/
desorption at mineral surfaces, and rapid acid-base equilibration reactions among sul de
species (H
2
S and HS
2
) and dissolved inorganic carbon (CO
2
, HCO
3
2
, and CO
3
2
2
) (Table 1;
Van Cappellen and Wang, 1996; Boudreau, 1997; Furukawa et al., 2001). Particle-bound
Mn and Fe are important redox species when nonlocal particle mixing due to bioturbation
displaces them across the redox boundaries (Aller, 1994; Can eld et al., 1993). Head-down
deposit feeders are one example of bioturbators that displace sediment particles across the
redox boundaries. Our current model assumes contributions of metal oxides to OM
reminerali zation to be negligible. This is justi ed for the application we present in this
paper in which we mo del early d iagenesis i n laborat ory mesocosm s inhabi ted by Schizocar-
dium sp. (See Section 3a).
Eq. (7), the two-dimensional conservation equation, is written for each of the species
considered in the present study, O
2
, NO
3
2
, SO
4
2
2
, NH
4
1
,SS (5H
2
S1HS
2
), TCO
2
(5CO
2
1HCO
3
2
1CO
3
2
2
), and titration alkalinity (Alk
t
5HCO3
2
12CO
3
2
2
1HS
2
).
The rate expressi on for e ach of the con servation e quations i s sho wn in Table 1.
A FORTRAN code is used to numerically solve the above conservation equations and
bound ary conditi ons. The as sumption o f steady state is valid wh en r
2
is suf ciently small
compared to L
total
(Aller, 1980). Lo cally on e-dimension al method (LOD; Bou dreau, 1 997,
p. 350) with the Crank-Nicholson formula is employed as the numerical solution scheme.
A two -dimension al ( xand r) grid is applied to the cylindrical microenvironment,and C
x
,r
for ea ch no de is so lved a s a t ime-evoluti on problem (Press et al., 1992) until steady state is
reached. All conservation equations are coupled and solved simultaneously through the
coupli ng of reaction terms, R(i.e ., Van Capp ellen and Wang , 1996 ; Furukawa et al., 2001).
For example, the reaction term for the conservation equation of O
2
at Time Step Tis
424 [59, 3
Journal of Marine Research
determined by the concentration values of O
2
, NH
4
1
, and SS at Time Step T21 (see
Table 1, I-1).
In the end, the model determines the value of C
x
,r
for each grid point for all species, as
well as the radially averaged C
x
,r
for each of the depth intervals. The radially averaged
concentrations are used as the feedback tool because typical eld measurements of pore
water constituents yield depth pro les, in which lateral variability is averaged by the
typical sampling practices (i.e., horizontal slicing of core samples and long-term deploy-
ment of pore water peepers).
Table 1. Reactio ns and rat es within sedim ents (after Van Ca ppellen and Wa ng, 1996) .
Primary redox reactions:
(CH2O)x(NH3)y(H3PO4)z1(x12y)O21(y12z)HCO3
2
®(x1y12z)CO21yNO3
2
1
zHPO4
2
2
1(x12y12z)H2O
(CH2O)x(NH3)y(H3PO4)z1((4x13y)/ 5)NO3
2
®((2x14y)/ 5) N21((x23y1
10x)/ 5) CO21((4x13y210 z)/5)HCO3
2
1zHPO4
2
2
1((3x16y110z)/ 5) H2O
(CH2O)x(NH3)y(H3PO4)z1(x/2 )SO4
2
2
1(y22z)CO21(y22z)H2O®(x/2)H2S1(x1
y22z)HCO3
2
1yNH4
1
1zHPO4
2
2
Secondary redox reactions:
NH4
1
12O212HCO3
2
®NO3
2
12CO213H2O
H2S12O212HCO3
2
®SO4
2
2
12CO212H2O
Acid-base reactions (e quilibrium):
CO21H2O«HCO3
2
1H
1
HCO3
2
«CO3
2
2
1H
1
H2S«HS
2
1H
1
Adsorption reactions (equilibrium)
NH4
1
(aq) «NH4
1
(ads)
Reaction rates:
D[O2]
Dt5
x12y
xROC
[O2]
KO21[O2]22kNH4O2[O2][NH4
1]22kSSO2[O2]@SS#(I-1)
D[NO3
2]
Dt52y
xROC
[O2]
KO21[O2]14x13y
5xROC
[NO3
2]
KNO31[NO3
2]
K9
O2
K9
O21[O2]1kNH4O2[O2][NH4
1](I-2)
D[SO4
22]
Dt51
2ROC
[SO4
22]
KSO41[SO4
22]
K9
O2
K9
O21[O2]
K9
NO3
K9
NO31[NO3
2]1kSSO2[O2]@SS#(I-3)
D@SCO2#
Dt52ROC (I-4)
D[NH4
1]
Dt52y
xROC 2kNH4O2[O2][NH4
1](I-5)
D@SS#
Dt52D[SO4
22]
Dt(I-6)
D@Alkt#
Dt5
y12z
xROC
[O2]
KO21[O2]24x13y210z
5xROC
[NO3
2]
KNO3
21[NO3
2]
K9
O2
K9
O21[O2]
2x1y22z
xROC
[SO4
22]
KSO41[SO4
22]
K9
O2
K9
O21[O2]
K9
NO3
K9
NO31[NO3
2]
22kNH4O2[O2][NH4
1]22kSSO2[O2]@SS#
(I-7)
2001] 425
Furukawa et al.: Bioirrigation modeling
3. Experimental methods
a. Experimental macrofauna
This study used Schizocardium sp., a funnel-feeding enteropneust hemichordate, as the
bioirri gator of choice b ecause of its abundance at a n earby eld stat ion and adapta bility to
the laboratory environment. Each individual Schizocardium sp. creates and inhabits a
single U-shaped burrow, whose diameter is approximately 5 mm (Fig. 5; Bentley and
Richardson, 2001). It ingests sediment particles at the feeding pit and egests at the other
end of the burrow, creating a fecal mound. Thus, the ingested sediment particles remain
within the same redox environment in the close vicinity of WSI, justifying the omission of
particle-bound Mn and Fe in the reaction couplings for this particular application (see
Section 2d above). The experimental animals were collected during several cruises on
board R/V Kit Jones at a eld station just outside of St. Louis Bay in Mississippi Sound
(30° 14.09N, 89° 20.09W). Schizocardium sp. was the numerically predominant macro-
fauna caught in the 0.5 mm sieves at this site. The
210
Pb ge ochronolo gy data from t he same
station shows that the upper 5 ;10 cm of the sediments are well mixed (Bentley et al.,
2000), suggesting that the burrowing activities of Schizocardium sp. extend to these
depths. In the eld, active burrows are marked by a few mm-thick tan-colored (i.e.,
oxidized) halos along the walls, whereas the walls of abandoned burrows exhibit a dark
gray color (i.e., reduced) similar to the matrix sediments.
b. Laboratory mesocosms
Water-satu rated, n e-grained sedi ment from St. Lo uis Bay, Mis sissippi (30° 17. 09N, 89°
19.09W) was used as the experimental substrate. The sediment was rst passed through the
Figure 5. Body ( A) an d schema tic burrow morp hology (B) of Schizocardiumsp. (A) Eac h increment
on the scale is 1 mm. (B) Scale is approximate. This organism constructs a U-shaped burrow,
ingesting sediment from one burrow opening (creating a feeding pit), and defecating at the other
(creating a fecal mound). Typical burrow depths (in mesocosms and the eld) range from 7 to
.15 cm.
426 [59, 3
Journal of Marine Research
0.5 mm sieves in order to eliminate macrofauna. A small amount of water, whose salinity
was adj usted to match th at of th e Schizocardiumsp. co llection site (18 61 psu), was added
to the sediment in order to aid sieving. A portion of the sediment was frozen and then
thawed, which resulted in the elimination of meiofauna and sediment compaction due to
dewatering.
Two glass tanks (“Tank 3” and “Tank 6”), 60 cm 360 cm 345 cm high each, were
prepared for this study. Tank 3 was loaded with the sediment/water mixture that had not
been frozen, which was then allowed to sett le naturally. Most of sediment particles settled
within seven days creating a 13 cm-thick sediment column. Tank 6 was loaded with the
sediment/water mixture that had been frozen and thawed. The sediment was also allowed
to sett le natura lly, and cre ated a 1 6 cm-thi ck sedi ment colum n. After th e settli ng, Plexigl as
dividers (30 cm high) were inserted into each tank in order to separate the tank into four
equal quadrants (i.e., A, B, C and D) with the lateral dimension of 30 cm 330 cm. The
water levels were maintained in both tanks so that there was always approximately 30 cm
of water column above WSI. The overlying water in the tanks was circulated through
external aragonite-gravel lter chambers for the duration of experiments.
Prior to the Schizocardium sp. in troduction, T ank 6 was analyz ed usin g Clarke-t ype O
2
microelectrodes to obtain three O
2
micropro les with the 0.2 mm vertical spatial resolu-
tion. The micropro les were located near the center of Quadrant 6C, and were laterally
5 mm apart of each other.
After approximately 1.5 months of sediment introduction, all experimental quadrants
were populated with Schizocardium sp. An imal populations in th e experimental qua drants
are shown in Table 2. Only two of the quadrants from each tank were studied for the
purpose of this paper (Quadrants 3B, 3D, 6A and 6B) and no further replication was made.
Within a few days, most of the animals burrowed into the substrate (Fig. 6a). Within a
week, the worms visibly altered the seabed structure through their tube construction and
sediment in gestion/egestion . The pho tograph taken a fter four week s of ani mal introduc tion
in Tank 3D (Fig. 6b) shows the magnitude of structural modi cation by Schizocardiumsp.
bioturbation. For the duration of the experiments, the mesocosms were lit for 10 hours per
day, salinity and temperature were kept near constant (S518 61 , T
tank3
524 618C,
T
tank6
519 61°C), and overlying waters were circulated through the aragonite lters.
Throughout the duration of experiments, green algae were allowed to grow on WSI as well
as on the inside of tank walls. On average, 10–25% of the WSI was covered with the algae,
Table 2. Schizocardiumsp. populatio n in exper imental qu adrants .
Tank Quadrant
Number in 30 cm 3
30 cm quadrant
Population
densit y (m
2
2)
Burrow
openin gs (m
2
2)
Tank 3 3B 72 800 1600
3D 9 100 200
Tank 6 6A 32 356 712
6B 28 311 622
2001] 427
Furukawa et al.: Bioirrigation modeling
whereas the tank walls were always completely covered by the algae. The green algae on
walls were occasionally removed by wiping.
Fifty-six days after the animal introduction in Quadrant 3B, Clarke-type O
2
microelec-
trodes were used to measure ve O
2
micropro les at 0.2 mm vertical spatial resolution.
The microelectrode insertion was arranged so that all ve micropro les were on a lateral
straight line near the center of Quadrant 3B, with the lateral spacing between micropro les
to be 5 mm.
Fifty-seven days after introduction of the animals, the sediments in Quadrants 3B and
3D were sampled using two 15 cm-diameter Plexiglas tubes per each quadrant. Quadrant
6A was sampled 26 days after the animal introduction, and Quadrant 6B was sampled 60
days aft er the animal introd uction, u sing one 15 cm-di ameter t ube for e ach qu adrant. L arge
tubes were necessary in order to extract enough pore water from each 1-cm slices.
However, due to the tight tting of 1;2 15-cm tube(s) plus 17-cm 32-cm slab in each
15 315-cm quadrant, no sampling could be duplicated. One core from each quadrant was
then sliced into 1-cm sections, whose pH values were determined by the direct insertion of
a pH electrode calibrated with NBS-traceable buffers. The slices were subsequently
centrifuged and ltered to yield pore water samples. The pore water samples were analyzed
for depth-dependent values of titration alkalinity (by potentiometric titration), SO
4
2
2
(by
ion chromatography), and NH
4
1
(by spectrophotometry).
The remaining cores from 3B and 3D were used for physical property (porosity and
grain density) analysis. The method involves the initial determination of bulk density and
subsequent determination of dry density after drying the sediment samples at 105°C
overnight. The porosity data were used to estimate the diffusive tortuosity (Eq. 2). The
drying method used here eliminates water trapped within mineral aggregates that may not
contri bute to diffu sive tortuosity (Bourbie´ et al., 1987). However, we proceeded to use the
total porosity in Eq. (2), because the empirical equation correlating porosity to diffusive
tortuosity is based on porosity data obtained using a wide variety of drying methods for
both clayey and non-clayey sediments (see Boudreau, 1997, p. 130, and references
therein). The porosity gradients for 6A and 6B were estimated from a core taken in the
Figure 6. An experimental quadrant (3D) shortly after the Schizocardium introduction (A), and 4
weeks after the introduction (B).
428 [59, 3
Journal of Marine Research
quadrant not used in this study (Quadrant 6C). The Schizocardium sp. population in 6C
was 28 per quadrant (or 311 m
2
2
).
All experimental quadrants were also cored using 2.2 cm-thick and 17 cm-wide Plexi-
glas slab-shaped corers. The slab cores were X-rayed immediately after coring in order to
characterize the burrow distribution. The X-radiographs were then converted to binary
images of burrows vs. matrix sediments. These images were subsequently overlaid with
horizontal lines in 1-cm intervals in order to determine the depth-dependent burrow
distribution and burrow tilt angles with 1-cm depth resolution.
Transmission electron microscope (TEM) was used to visually inspect the microfabric
of sediments. Samples were taken from the immediate vicinities of WSI and burrow walls,
as well as from the matrix sediments well away from WSI and burrow walls using
mini-corers (Lavoie et al., 1996). They were subsequently treated with a xing agent
(Gluteraldehyde-buffer solution; Leppard et al., 1996), embedded in resin, ultramic-
rotomed, and imaged under TEM.
c. Batch incubation
The rates for microbial production of TCO
2
and NH
4
1
and consumption of SO
4
2
2
were
determined through batch incubation experiments in closed systems (Aller and Yingst,
1980). Afte r coring , approx imately the to p 5 cm of the rem aining se diments from Tank 3 D
were homogenized by hand mixing and placed into four centrifuge tubes. Visible fauna
were eliminated prior to placing in the tubes, thus the rates determined herein were
considered to be the microbial reaction rates. The tubes were completely lled with
sediment in order to minimize air space, and then placed in a glove bag lled with ultra
high purity N
2
. The centrifuge tubes were capped in the glove bag. All but one of the
centrifuge tubes were then placed in an airtight container with two small openings for N
2
circulation. The container was lled with a fast stream of N
2
, followed by a slow
continuous stream of N
2
in order to assure no atmospheric O
2
would be in contact with the
centrifuge bottles. The tubes were taken out of the container one at a time at 24 hours, 167
hours, and 384 hours after the initial loading, inserted with a pH electrode, and centrifuged.
The pore water samples were subsequently ltered and analyzed for titration alkalinity,
SO
4
2
2
and NH
4
1
concentrations using the same methods used above. The remaining
sediment - lled tube was analyz ed for pH an d centrifuged immedia tely after lling in order
to extract and analyze initial pore water. The time-dependent change in TCO
2
was
calculated using the measured time-dependent pH and alkalinity data and equilibrium
constants for the carbonate system found in Millero (1995). The pH measurement for
24-hour sample failed, thus TCO
2
value was not calculated for this sample.
4. Experimental results and determination of parameter values
a. Porosi ty, grain d ensity a nd di ffusion coe f cients
Porosity pro les of the experimental quadrants are shown in Figure 7. The mean
porosity for each quadrant is thus determined to be: w
3B
50.854 (range: 0.733 ;0.928),
2001] 429
Furukawa et al.: Bioirrigation modeling
w
3D
50.832 (range: 0.710 ;0.896), and w
6A
5 w
6B
50.709 (range: 0.660 ;0.791). The
porosity in Tank 6 is lower than that of Tank 3 because the Tank 6 sediment went through
freezing before loa ding, which caused dewatering.
The measured pro les (Fig. 7) indicate that the porosity is not constant within each
mesocosm tank. The assumption of constant porosity (Section 2b) with the use of mean
porosity values introduces uncertainty to the tortuosity-corrected diffusion coef cient,
DSW
12ln ~w2!. The estimated uncertainty in diffusion coef cients is up to 23% in Quadrant
3B, up to 23% and Quadrant 3D, and up to 13% in Quadrant 6A and 6C. The average grain
density of the mesocosm sediments was determined to be 2.55 g cm
2
3
.
b. O
2
micropro les and aerobic OC degradation rate
Figure 8 shows the O
2
micropro les taken in 6C prior to the Schizocardium sp.
introduction, and Figure 9 shows the O
2
micropro les taken in 3D prior to the core
sampling. The key parameter values for the Tank 3 pro les are also shown in Table 3. The
Figure 7. Depth pro les of measured porosity for Quadrants 3B, 3D, and Tank 6.
Figure 8. O2micropro les taken in Tank 6 prior to the Schizocardiumsp. introduction.
430 [59, 3
Journal of Marine Research
pro les from 3D were used to calculate the O
2
consumption rates in all tanks: the 6C
pro les ta ken be fore th e anim al intro duction wou ld yie ld an O
2
consumptionrate that does
not account for the possible change in the microbial activities due to the presence of
macrofauna. The procedure outlined below assumes that the observed pro les are free
from the effect of lateral O
2
diffusion along burrow walls. This is a reasonable assumption
because the radial O
2
penetra tion thickness alon g burrow wa lls is expected to be similar to
the vertical O
2
penetra tion depth at W SI, whic h is app roximately 2 –3 mm (Table 3 ). Thus,
if the number of burrow openings is 1,600 (m
2
2
) as in Quadrant 3B and burrow radius is
1.25 (mm), approximately 9% of the WSI surface is under the in uence of radial diffusion.
In other words, at WSI, a randomly inserted O
2
microelectrode has 91% chance of
recording a vertical O
2
pro le that is free from the effect of radial diffusion. One of the
pro les from 3D (Fig. 9a) exhibits the in uence of radial diffusion, and thus is excluded
from the following calculations.
At steady state, when vertical diffusion is the only transport mechanism, the mass
conservation equation describing the pore water pro le of O
2
is,
D9d2[O2]
dx21RO250 (12)
Table 3. Graphi cally estim ated value s of C0and LO2, and calculate d O2consumptionrates R0.
(b) (c) (d) (e)
T(°C) 24
DS W (m2s
2
1) 2.25 310
2
9
fW S I (%) 89.6
C0(M) .163 310
2
3.162 310
2
3.171 310
2
3.157 310
2
3
LO2(m) 2.8 310
2
32.6 310
2
32.4 310
2
32.6 310
2
3
R0(M/s) 27.7 310
2
828.8 310
2
8210.9 310
2
828.6 310
2
8
Average R0(M/s) 29.0 310
2
8
Figure 9. O2micropro les taken in Tank 3 prior to the core sampling. Pro le (a) exhibits the effect
of la teral O2diffusi on from a b urrow.
2001] 431
Furukawa et al.: Bioirrigation modeling
where D9is the molecular diffusion coef cient of O
2
after tortuosity correction, and R
O
2
is
the net production rate. For simplicity, we will assume that the rate of consumption of O
2
remains constant. The boundary conditionsare:
[O2]x505[O2]0(13)
[O2]x5LO250 (14)
where L
O
2
is the O
2
penetration depth. Consequently, the solution to Eq. (12) is
[O2]x5[O2]01
X
2[O2]0
LO2
1
RO2
0LO2
2D9
D
x2
RO2
0
2D9x2(15)
(Bouldin, 1968; Cai and Sayles, 1996). By imposing that the ux of dissolved O
2
must be
zero at x5L
O
2
, that is,
d[O2]
dx Ux5LO2
50 (16)
we obtain
RO2
0522D9[O2]0
LO2
2(17)
Eq. (15) predicts de pth pro les with a q uadratic curvat ure. An inspe ction of Figu res 8 and 9
indicates this to be true for the lower portions of the measured pro les. Typically, however,
a layer of nite thickness separates the upper boundary of the quadratic decay portion of
the pro le from the uniform bulk overlying water O
2
concentration(Fig. 10), indicating the
presence of diffusive boundary layer above the water-sediment interface (Jørgensen and
Revsb ech, 198 5). Thus , the co ncentratio n [O
2
]
0
in Eq. (17) is not the bulk water value, but
the value at the base of boundary layer (Fig. 10). Values of [O
2
]
0
and L
O
2
are estimated
graphically for the measured pro les by setting the water-sediment interface to be where
the quadratic decay of the pro le begins. These values are combined with an estimate of
the measured porosity values in the upper 1 cm of the sediments (i.e., 89.6% in Quadrant
3D) in Eq. (17) to obtain the values for R
O
2
0
. The D
O
2
SW
value was determined using the
formula given in Boudreau (1997, p. 109). The results are shown in Table 3.
The R
O
2
value determined here is related to the aerobic OC degradation rate by the
following reaction:
(CH2O)x(NH3)y(H3PO4)z1~x12y!O21~y12z!HCO3
2®
~x1y12z!CO21yNO3
21zHPO4
221~x12y12z!H2O.
Thus, the ratio of O
2
consu mption rate t o aerobi c OC d egradation rate is ( x12y)/ x. If we
assume the C/N ratio of labile OC in our experiments to be 106:16, the R
O
2
/R
OC
ox
ratio is
determined to be 1.3. Consequently, the average R
O
2
calculated here (9.0 310
2
8
M s
2
1
)
432 [59, 3
Journal of Marine Research
yields R
OC
ox
56.9 310
2
8
M s
2
1
. This R
OC
ox
value is actually the upper limit because part of
O
2
consumed within the upper 2–3 mm is utilized for SS and NH
4
1
reoxidation rather than
OM remineralization.
c. Batc h incuba tion, anaerobic OC deg radation ra te, and NH
4
1
adsorption constant
The batch incubation results, shown in Figure 11, were interpreted with the following
assumptions: (1) the SO
4
2
2
consumption and TCO
2
and NH
4
1
production rates remain
constant during the rst ;400 hours following the hand mixing; and (2) the changes in
concentration values re ects microbial remineralization reactions only and do not contain
effects from reoxidation by O
2
. Under these assumptions, the slopes of the incubation
results yield the microbial production rates for SO
4
2
2
and TCO
2
to be R
SO
4
5 21.29 3
10
2
9
(M s
2
1
) and R
S
CO
2
5 22.94 310
2
9
(M s
2
1
).
The ana erobic OC degradation rate, R
OC
an
, is related to the above R
SO
4
and R
S
CO
2
through
Eqs. (I-3) and (I-4) (Table 1). In the absence of O
2
and NO
3
2
, Equations (I-3) a nd (I-4) can
be rewritten:
D[SO4
22]
Dt51
2ROC
an (18)
D@SCO2#
Dt52ROC
an (19)
Figure 10. Schematic O2micropro le indicating the parameters necessary for the determination of
O2consu mption rate.
2001] 433
Furukawa et al.: Bioirrigation modeling
Eqs. (18) and (19) yield the R
OC
an
value to be 12.58 310
2
9
(M s
2
1
) and 12.94 310
2
9
(M s
2
1
),
respectively. We take the average valueto be the average anaerobic OC degradationrate within
the upper 5 cm of Tank 3 sediments, i.e., R
OC
an
5 12.8 310
2
9
(M s
2
1
).
Depth dependency of the anaerobic OC degradation rate in Tank 3 was prescribed as
follows. First we assumed that (1) the anaerobic OC degradation rate is a linear function of
depth , and (2) th e rate va lue reached zero at the bot tom of tan ks (i.e., R
OC
an
5a(L
total
2x)).
The value of parameter awas then determined by solving the following equation, because
the batch incubation experiment was carried out using the upper 0.05 m of Tank 3
sediment:
E
0
0.05
a~Ltotal 2x!dx
0.05 52.8 31029. (20)
This yi elds the value of afor Tank 3 to be 2.7 310
2
8
. Un fortunately , we d id not co nduct a
batch incubation experiment for Tank 6 sediments. The simulation results for 6A and 6B
using the Tank 3 R
OC
an
value, however, resu lted in u nderestimation o f anaerobi c productio n
of TCO
2
and consumption of SO
4
2
2
. This indicates that Tank 6 sediments contained more
labile OC than the Tank 3 sediments, possibly because the sediment freezing resulted in
death and decay of meiofauna. The value of afor Tank 6 was determined after treatment as
an adjustable parameter to be 5.0 310
2
8
.
NH
4
1
in the aqueous phase can be assumed to be in equilibrium with NH
4
1
adsorbedonto
clay mineral surfaces (Mackin and Aller, 1984):
Figure 11. The results of the batch incubation experimentsthat were used to determine the microbial
SO4
2
2
and TCO2consumption rates and NH4
1
adsorption coef cient of the experimental sedi-
ments.
434 [59, 3
Journal of Marine Research
{NH4
1(ads)} 5wKN
r~12w! [NH4
1(aq)] (21)
where K
N
is the dimensionless adsorption coef cient, {NH
4
1
(ads)} is the concentration of
adsorbed NH
4
1
in terms of moles per kg solids, [NH
4
1
(aq)] (M) is the concentration of
aqueous NH
4
1
,r(g cm
2
3
) is the grain density, and fis the porosity. The average grain
density of experimental sediment was 2.55 (g cm
2
3
), and the average porosity of sediment
in Qu adrant 3D was 0.832. Th e rate o f NH
4
1
produc tion in ana erobic part of the sed iment is
related to R
OC
an
through Eq. (I-5) (Table 1) by:
D([NH4
1(ads)] 1[NH4
1(aq)])
Dt52y
xROC
an (22)
where x:yis the C/N ratio of labile OM and [NH
4
1
(ads)] is the concentration of adsorbed
NH
4
1
in terms of moles per dm
3
pore water in contact. Manipulation of Eqs. (20)–(22)
yield:
~11KN!D[NH4
1(aq)]
Dt52y
xROC. (23)
The value of D@NH4
1(aq)]
Dtwas determined through the batch incubation experiment to be
11.18 310
2
10
(M s
2
1
). Assuming the C/N ratio of x:y5106: 16, and using the above w
and rvalues, Eq. (23) yields K
N
52.5.
d. Pore water chemistry
The results of bulk pore water analyses in the laboratory mesocosms (Figs. 12–15) as
well as the O
2
micropro les (Figs. 8 –9) exhibit depth pro les that are typical of
ne-grained siliciclastic sediments found in marine and estuarine environment: a rapid
consumption of O
2
within a few mm of WSI, gradual depth-dependent depletion of SO
4
2
2
,
Figure 12. Depth pro les of pH directly measured for the cores from experimental quadrants.
2001] 435
Furukawa et al.: Bioirrigation modeling
and depth-dep endent i ncreases in NH
4
1
and TCO
2
. These bulk pore water pro les are used
to evaluate the model results in the following sections.
e. X-radiography and model geometry
The X-radiography images from all quadrants were converted to binary images of
burrows vs. matrix sediment in order to characterize the burrow networks (Fig. 16). The
geometry of cylindricalmicroenvironment(r
2
,r
1
(x), L
burrow
,L
total
) was determined using
the binary images as follows.
First, the binary images were overlaid with horizontal lines in 1-cm intervals, and the
number of burrows intersected by each horizontal line was recorded (Fig. 17). The burrow
numbers captured in the slab cores were then normalized to the burrow density per square
meter by assuming that each indivi dual Schizocardium sp. create s two openings at WSI:
n~0!523P(24)
n~x!5n~0!3N~x!
N~0!(25)
Figure 14. Depth pro les of NH4
1
directlymeasured for the cores from experimentalquadrants.
Figure 13 . Dept h pro les of TCO2determined from directly measured pH and titration alkalinity
using th e formula in Mi llero (1995).
436 [59, 3
Journal of Marine Research
Figure 16. The x-r adiograph images o f the expe rimental quadran ts,an d binary imag es of the ori ginal
x-radi ographs that were us ed to determine the mod el geometry.
Figure 15. Depth pro les of SO4
2
2
directlymeasured for the cores from experimentalquadrants.
2001] 437
Furukawa et al.: Bioirrigation modeling
where n(0) is the burrow density at WSI (m
2
2
), Pis the Schizocardium populationdensity
for the given quadrant (m
2
2
), n(x) is the burrow density at depth x(m
2
2
), N(x) is the
number of burrows intersected by the horizontal line at depth xin the X-radiography
image, and N(0 ) is the num ber of burrows at WSI in t he X-radiography imag e. The bu rrow
density determined this way assumes that burrows that have been abandoned do not
contri bute to b ioirrigation (Al ler, 1 984).
The tilt angle expressed in the X-radiographs are the “apparent” tilt angles, u9, which are
2D projectio ns of the true tilt ang les, u(Fig. 18). Their relationship can be expressed in an
equation:
u5
E
0
p
2
arctan
X
tan u9
cos a
D
p
2
da(26)
where ais the angle between the vertical plane parallel to X-radiography slab and the
vertical plane that contains the burrow (Fig. 18). The angle ahas an equal probability to
take any value between 0 and 90 degrees. Subsequently, the arithmetic average of true tilt
angle (u) is determined for each 1-cm interval, and subsequently used in Eqs. (5) and (6) to
determine the model burrow radius (r
1
) for each 1-cm depth interval. The value of r
burrow
was assumed to be 1.25 310
2
3
(m). The 1-cm interval expression of r
1
for each quadrant
was then tted to a linear function of depth (i.e., r
1
(x)5r
1
(0 ) 1ax) using the standard
least square tting procedure.
L
burrow
for each quadrant was obtained as the value of xwhere r
1
(x) becomes equal to
zero. L
total
for each quadrant was determined as the total thickness of sediment substrate,
Figure 17. Parameters for model geometry (r1(x), r2,Lb u r ro w ) were determined by overlaying
1 cm-interval horizontal lines on the binary burrow image, and counting and examining the
burrows th at inte rsect with each horizo ntal line.
438 [59, 3
Journal of Marine Research
and r
2
was determined using Eq. (3), in which the value of nwas twice the value of
Schizocardium sp. population density (m
2
2
) in each quadrant. The geometry parameter
values are summarized in Table 4.
f. Electron micrographs
The TEM images of samples from WSI, burrow wall, and matrix sediment (Fig. 19)
show that the packing of clay mineral particles is different among these locations. Mineral
particles are organized into clusters with large inter-cluster pore spaces in sediments from
WSI and burrow wall. On the other hand, clay mineral particles in the matrix sediment are
distributed more evenly without large clusters or pore spaces. These images indicate that
the assumption of uniform diffusive tortuosity (Section 4a., above) may be an oversimpli-
cation of the diffusive transport processes in these sediments.
5. The hindcast and discussion
a. Summary of parameters and assumptions
Model calculations were executed using the parameters determined as described in
Section 4 (Table 4). The determination processes for all model parameters required
assumptions in order to keep the model system simple. The major assumptions include: (1)
Figure 18. A schematic diagram showing how apparent burrow tilt angles (u9) projected onto the
plane of X- radiograp hy are con verted to t he tr ue tilt a ngles (u). The vertical pl ane contai ning the
burrow (B) and the plan e of X-ra diograph y projectio n (X) interse ct at ang le a.
2001] 439
Furukawa et al.: Bioirrigation modeling
Table 4. Model parameters.
3B 3D 6 A 6B
Sediment column
height (1)
L
tot
(m) 0.13 0.13 0.16 0.16
Burrowed layer
thickness (1)
L
bur
(m) 0.09397 0 .06106 0. 07352 0.1016
Model cylinder
radius (1)
r
2
(m) 0.01343 0 .03877 0. 02014 0.02 154
Model burrow
radius (1)
r
1
(x) (m) 0.00313 7 2
0.03338x
0.0027 53 2
0.04508x
0.003844 2
0.05228x
0.00458 3 2
0.04512x
Model no de dimension dx and d r (m) 0.0004 0.0004 0.0004 0.0004
Porosity (2) f0.854 0.83 2 0.709 0.7 09
Bottom water
composition (2 )
[O
2
]
0
(M) 0.223 310
2
3
0.223 310
2
3
0.228 310
2
3
0.228 310
2
3
[NO
3
2
]
0
(M) {0.015 310
2
3
} {0 .015 310
2
3
} {0 .015 310
2
3
} {0.0 15 310
2
3
}
[SO
4
2
2
]
0
(M) 18.0 310
2
3
18.0 310
2
3
15.4 310
2
3
16.8 310
2
3
[NH
4
1
]
0
(M) 0 0 0 0
[SS]
0
(M) 0 0 0 0
[SCO
2
]
0
(M) 3.25 310
2
3
3.25 310
2
3
6.25 310
2
3
6.17 310
2
3
[H
1
]
0
(M) 6.59 310
2
9
6.59 310
2
9
3.46 310
2
9
5.74 310
2
9
Diffusion coef cien ts
(3)
D9
O
2
{1.71 310
2
9
} {1.64 310
2
9
} {1.18 310
2
9
} {1.18 310
2
9
}
D9
NO
3
2
{1.43 310
2
9
} {1.38 310
2
9
} {1.00 310
2
9
} {1.00 310
2
9
}
D9
SO
42
2
{7.94 310
2
10
} {7.64 310
2
10
} {5.50 310
2
10
} {5.50 310
2
10
}
D9
NH
4
1
{1.48 310
2
9
} {1.42 310
2
9
} {1.03 310
2
9
} {1.03 310
2
9
}
D9
H
2
S
{1.32 310
2
9
} {1.27 310
2
9
} {8.99 310
2
10
} {8.99 310
2
10
}
D9
HS
2
{1.29 310
2
9
} {1.24 310
2
9
} {9.24 310
2
10
} {9.24 310
2
10
}
D9
CO
2
{1.35 310
2
9
} {1.30 310
2
9
} {9.37 310
2
10
} {9.37 310
2
10
}
D9
HCO
3
2
{8.86 310
2
10
} {8.52 310
2
10
} {6.09 310
2
10
} {6.09 310
2
10
}
D9
CO
32
2
{6.92 310
2
10
} {6.66 310
2
10
} {4.81 310
2
10
} {4.81 310
2
10
}
C:N:P ratio (4) x:y:z{106:16:1} {106:16:1 } {106:16:1 } {106:16 :1}
Aerobic OC ox idation
rate (5)
R
OC
ox
6.9 310
2
8
6.9 310
2
8
6.9 310
2
8
6.9 310
2
8
Anaerobic OC
oxidation rate (6 )
R
OC
an
(x) (M s
2
1
)2.7 310
2
8
3
(L
tot
2x)
2.7 310
2
8
3
(L
tot
2x)
[5. 0 310
2
8
3
(L
tot
2x)]
[5 .0 310
2
8
3
(L
tot
2x)]
Rate cons tant for NH
4
1
reoxidation (7)
k
NH4
1
{0.16} {0.1 6} {0.16} {0. 16}
Rate cons tant for SS
reoxidation (8)
k
S
S
{5.1 310
2
3
} {5.1 310
2
3
} {5.1 310
2
3
} {5.1 310
2
3
}
NH
4
1
adsorption
coef cient (9)
k
N
2.5 2.5 2. 5 2.5
Monod saturation
constants (10)
K
O
2{0.02 310
2
3
} {0.02 310
2
3
} {0.02 310
2
3
} {0.02 310
2
3
}
K
NO
3{0.005 310
2
3
} {0 .005 310
2
3
} {0 .005 310
2
3
} {0.0 05 310
2
3
}
K
SO
4{1.6 310
2
3
} {1.6 310
2
3
} {1.6 310
2
3
} {1.6 310
2
3
}
Monod inhib ition
constants (10)
K9
O
2{0.02 310
2
3
} {0.02 310
2
3
} {0.02 310
2
3
} {0.02 310
2
3
}
K9
NO
3{0.005 310
2
3
} {0 .005 310
2
3
} {0 .005 310
2
3
} {0.0 05 310
2
3
}
Acidity cons tants for
SCO
2
(11)
K
1C
{1.14 310
2
6
} {1.14 310
2
6
} {1.02 310
2
6
} {1.02 310
2
6
}
K
2C
{7.17 310
2
10
} {7.17 310
2
10
} {5.89 310
2
10
} {5.89 310
2
10
}
Acidity cons tant for
H
2
S (12)
K
S
{3.44 310
2
7
} {3.44 310
2
7
} {3.44 310
2
7
} {3.44 310
2
7
}
Directly det ermined value s are in bo ld types, wherea s values d etermined afte r treate d as adjus table
parameters are in [] and values taken from references are in {}. Remarks: (1) Determined through
X-radiography data analysis (see Section 4e.); (2) Determined through direct measurements; (3)
Determined using directly measured porosity, salinity, and temperature with formulae given in
Boudreau (1997, Chapter 4.2); (4) Typical values for labile organic matter; (5) determined using the
measured O2micropro les (see Section 4b.); (6) the formula for Quadrants 3B and 3D were
determined by assuming that the incubation experiment results represent the average anaerobic OC
degradation rate of upper 5 cm of the sediment column, whereas the formula for Quadrants 6A and
6B were determined after the treatment as an adjustable parameter; (7) from Van Cappellen and
Wang (1996); (8) from Van Cappellen and Wang (1996); (9) determined through incubation
exper iments (see Se ction 4c.) ; (10) fro m Boudrea u (1996); (11 ) calculat ed from Mille ro (1995); ( 12)
calculatedfrom Millero (1995).
440 [59, 3Journal of Marine Research
the burrow water composition is equal to the overlying water composition (i.e., 100%
irrigation); (2) porosity and diffusive tortuosity are constant throughout the sediments; (3)
aerobic OC degradation rate is constant throughout the immediate vicinity of WSI and
burrow walls; (4) an aerobic OC degrada tion rate is a n impo sed line ar functio n of dept h and
becomes zero at the bottom of sediment column; (5) the system is at steady state; and (6)
the simple cylinder geometry (Fig. 2) reasonably describesthe actual burrow network.
The above assumptions may or may not be tolerated in the simulation of our laboratory
mesocosm systems. For example, previous studies on infaunal ventilation patterns show
that the burrows are usually intermittently ventilated and thus burrow water compositions
are not continuously equal to the overlying water compositions (Gust and Harrison, 1981;
Kristensen, 1989; Riisgård, 1991; Forster and Graf, 1995). The TEM images (Fig. 19)
indicate that diffusive tortuosity is variable throughout the sediments, with the values near
WSI and burrow walls expected to differ from the values in the matrix sediments. The
diffusive tortuosity is also affected by organic burrow linings (Aller, 1983). The distribu-
tions of OC and micro-organisms are expected to be spatially and temporally heteroge-
neous , and thu s the O C degradation rate is n ot a co nstant.
While the model is not an exact simulation of the experimental systems, it allows us to
(1) evaluate the adequacy of assumptions; and (2) quantitatively describe the processes
occurring at burrow walls, which cannot be accomplished with the traditional 1D models.
b. Calculation results and comparison
Model calculations were carried out using parameter values listed in Table 4. The
results are reported here in terms of ve rtical (1D) pro le s, which were generated by
Figure 19. Transmission electron microscopy (TEM) images of the samples from three distinct
regions of the benthic mesocosm sediments. The dark- and medium-grey features represent clay
mineral aggregates whereas the at, light-gray masses are the pore spaces, now lled with an
embedding medium. The total volume as well as the distribution and geometry of pore spaces are
signicantly different between images. This results in spatially heterogeneousdiffusive tortuosity
for solute species.
2001] 441
Furukawa et al.: Bioirrigation modeling
tak ing the radial averag e values of concentrations for each depth increment. The
radially averaged 1D pro les a llow the visual comparison between model re sul ts and
measured pro les.
The model-calculated 1D depth pro les of SO
4
2
2
are in general agreement with the
measured pro les (Fig. 20). For the simulations of 3B and 3D, no attempt was made to
adjust p arameter values in ord er to seek better t . The pa rameter value s determined a priori
(Table 4) and assumptions (1) through (6) in Section 5a. were the only constraints. The
agreement between calculated and measured SO
4
2
2
pro les indicate that the parameter
values and assumptions used here, including the model geometry, reasonably describe the
chemical mass transfer processes in these Quadrants. The simulations of 6A and 6B were
carried out similarly: the only difference was that, for 6A and 6B, the rate of anaerobic OC
degradation was treated as adjustable until a reasonable agreement between measured and
modeled pro les were met. An underestimation of SO
4
2
2
depletion occurred in the deeper
parts of 6A and 6B. This is probably attributed to errors in the prescription for depth-
dependency of R
OC
an
.
The m odel-calculat ed 1D depth pro les of TC O
2
are also in general agreement with the
measured pro les (Fig. 21). However, the overestimation of TCO
2
buildup, especially
visible in 3B and 3D, means that there may be additional mechanisms that remove TCO
2
from sediments at the rates greater than the rates of irrigative introduction of SO
4
2
2
into
sediments. The possible mechanisms include TCO
2
consumption due to primary produc-
tion, authigenic precipitation of carbonate minerals, and growth of individual Schizocar-
dium sp. An underestimation of TCO
2
buildup occurred in the deeper parts of 6A and 6B.
This phenomenon is coupled to the underestimationof SO
4
2
2
deplet ion in the same regions,
and probably also due to the errors in the assumption for depth-dependency of R
OC
an
.
The overestimation of TCO
2
buildup in model results can also be seen in the plots of
excess TCO
2
Vs. SO
4
2
2
removed (Fig. 2 2). Th e anaero bic OC di agenesis i n a clo sed syst em
is dominated by the reaction:
Figure 20. Model-simulated d epth pr o les of SO4
2
2
plotted together with actual me asured pro les
from the experimentalquadrants.
442 [59, 3
Journal of Marine Research
(CH2O)x(NH3)y(H3PO4)z1~x/2!SO4
221~y22z!CO21~y22z!H2O®
~x/2!H2S1~x1y22z!HCO3
21yNH4
11zHPO4
22
Consequently, the ratio between TCO
2
increase and SO
4
2
2
decrease is 2:1. The model
calculated this ratio for each depth increment of all model quadrants, and the resulting
slopes are very close to the ideal value (0.5). The shallow pore water (i.e., low DTCO
2
and
Figure 21. M odel-simula ted depth pro l es o f TCO2plotted tog ether with actu al measur ed pro les
from the experimentalquadrants.
Figure 22. For each depth increment, increase in TCO2(DTCO25[TCO2]2[TCO2]0) is plotted
against decrease in SO4
2
2
(DSO4
2
2
5[SO4
2
2
]02[SO4
2
2
]). The values from model results are
plotted in bold lines (top) whereas the values measured in the actual core slices are plotted with
lled circles (bottom). Lines for the stoichimetric closed system value (see Table 1) of DSO4
2
2
/
DTCO250.5 a re also shown for c omparison. Fo r the laborato ry systems, bes t- t lin ear lines and
the slop e values are al so shown.
2001] 443
Furukawa et al.: Bioirrigation modeling
DSO
4
2
2
values) exhibits much lower DSO
4
2
2
/DTCO
2
ratio because aerobic OC diagenesis
occurring in this region does not deplete SO
4
2
2
. The difference in diffusion coef cients
between TCO
2
and SO
4
2
2
and reoxidation of SS cause the DSO
4
2
2
/DTCO
2
ratio to deviate
from 0.5 slightly. The DSO
4
2
2
/DTCO
2
ratios obtained from the measured depth pro les of
experi mental quadra nts, however, d iffer substa ntially from the id eal closed syst em value of
0.5. In the actual experimental systems, diagenetic TCO
2
increase is not as rapid as the
SO
4
2
2
decrease. Additional TCO
2
sinks not implemented in the model, such as the primary
production, authigenic carbonate mineral formation, and growth of Schizocardium sp.
must b e present in these experimenta l systems.
The model-calculated 1D depth pro les of NH
4
1
signi ca ntly underesti mate the mea-
sured NH
4
1
buildup (Fig. 23). The K
N
value estimated from batch incubation experiments
may be too high for the actual NH
4
1
adsorption in the mesocosm tanks in which both
adsorption and desorption occur through displacement of clay particles between different
pore water environments due to bioturbation. However, simulations using the K
N
value of
zero (K
N
50. 0) still barely agree with the measured NH
4
1
buildup (Fig. 24). There must
be additional NH
4
1
sources that were not implemented in the model. They may include the
NH
4
1
excretion by macrofauna, which has been observed and quantitatively reported for
other macrofauna species (e.g., Kristensen, 1984).
Inspection of the measured DNH
4
1
/DTCO
2
ratio diagrams (Fig. 25) reveals that the ratio
is greater for mid-depth pore waters than for pore waters from deeper parts of the
sedime nts. T his is probably ca used by t he dept h-depend ent C/N rat io of sou rce OM. In the
mid-depth region, where freshly produced green algae with low C/N ratio is consumed by
both macrofauna and microorganisms, the resulting ratio of OM consumption products,
DNH
4
1
/DTCO
2
, is large. On the other hand, in deeper parts of the tanks where bioturbation
is infrequent and thus in ux of fresh OM is minimal, the OM degradation is dominated by
less labile source OM with higher C/N ratio that had been incorporated into the sediments
since the time of initial tank loading.
Figure 23. Model- simulated dep th pro les of NH4
1
plotted together with actual measured p ro les
from the experimentalquadrants.
444 [59, 3
Journal of Marine Research
c. Effect of bioirrigation on 1D solute distributions
The quantitative signi cance of bioirrigation was examined by comparing the above
hindcast results with model results obtained from a 1D diffusion-reaction model with no
radial diffusion term:
]Cx,r,t
]t5Dsw
12ln ~w2!
]2Cx,r,t
]x21R. (27)
The 1D simulations were carried out by using the same parameter values used in the above
hindcasts (see Table 4), except for the lack of burrow geometry terms r
1
and r
2
.
Figure 24. The dep th pro le o f NH4
1
was si mulated using KN50 for Quadrant 6A.
Figure 25. For each depth increment, increase in TCO2(DTCO25[TCO2]2[TCO2]0) is plotted
against increase in NH4
1
(DNH4
1
5[NH4
1
]2[NH4
1
]0). The values measured in the actual core
slices are plotted with lled circles. Best- t linear lines and their slope values are also shown.
2001] 445
Furukawa et al.: Bioirrigation modeling
Comparison of the results (Fig. 26) shows that bioirrigation quantitatively affects the
chemical mass transfer regimes in aquatic sediments. When all other parameters including
the depth distribution of anaerobic OM degradation rate are held the same, bioirrigation is
the determining factor for the depthdistributionof geochemically importantsolute species
such as SO
4
2
2
.
d. Chemical mass transfer in the vicinity of burrow walls
Bioirrig ation is qua ntitative lysig ni cant no t only bec ause it affec ts the ve rtica l distribu-
tion of solute species as discussed above, but also because it promotes burrow walls as the
interface between oxygenated burrow water and anoxic pore water. In sediment with no
burrows, WSI is the only interface between oxygenated overlying water and anoxic
sedime nts. The int erface acco mmodates ste ep geoc hemical gradi ents due to p rocesses suc h
as th e rapid c onsumption of OM a nd pore wate r O
2
, and rap id produc tion of TCO
2
and H
1
.
When the sediments are burrowed, burrow walls become the additional interface to
accommodate such rapid chemical mass transfer processes. The addition of burrow walls
as the oxic-anoxic interface is quantitatively signi cant. For example, the model burrow
geometry used for the simulation of Quadrant 3B represents 158% increase in the
oxic-anoxic interface area over the same model cylinder with no burrow, as calculated by
the following equation:
SA2
SA1 51
pr2
2
H
E
0
Lbur
2pr1~x!dx 1~pr2
22pr1~0!2!
J
(28)
where SA1 is the area of WSI for the model cylinder with no burrow, and SA2 is the
surface area of WSI plus burrow wall for the model cylinder. Similarly, the surface area
increase for the other model quadrants are determined to be 11% for 3D, 66% for 6A, and
96% for 6B.
Figure 26. M odel-simulate d depth p ro les o f SO4
2
2
are compared with the depth pr o les of SO4
2
2
calcul ated using the 1 D mod el with no irrigatio n (Eq. 2 7).
446 [59, 3
Journal of Marine Research
The model-generated gradients of O
2
concentrations and TCO
2
are illustrated in Figure
27. They show that the immediate vicinities of WSI and burrow wall interface are similar
in that they both are the sites of intense chemical mass transfer processes, as indicated by
the steep gradients of O
2
and TCO
2
concen trations. Howev er, the steady state assumption
of the model prevents us from using the model results to quantitatively discuss the
differences between these two types of interfaces. One important difference between the
two types of interfaces is the temporal dynamics. Chemical mass transfer in the vicinity of
burrow walls is temporally dynamic, and less likely to be at steady state than the chemical
mass transfer in the vicinity of WSI. Burrow walls are, in reality, ventilated periodically
rather than contin uously (Kristense n, 198 9, 200 0), and thus the bu rrow wal l in terfaces are
subject to oscillating geochemical parameters including dissolved O
2
concentrations and
pH (Furukawa et al., 200 1). A worm may rapidly bu ild a burrow in the sediment regio n that
has been previously fully reduced to create a non-steady state interface between oxic
burrow water and anoxic sediments. Another difference between the two types of
interfaces arises due to biochemical compounds that are excreted by burrowing macro-
fauna. For example, some burrowing enteropneusts are known to excrete brominated
compounds, which have been found to affect the microbial activities at burrow walls
(Jensen et al., 1992; Giray and King, 1997). Hansen et al. (1996) found the increased
microbia l activity in the vici nity of Mya arenaria burrow walls due to OM excretion by the
animals. Thus, microbial characteristics in the vicinity of burrow walls are different from
that in the vicinity of WSI. In addition, the diffusive tortuosity at the burrow walls is
expected to be different from that of bulk sediments or WSI, because burrowing animals
often strengthen their burrows by rearranging ne-grained particles (Fig. 19) as well as by
excretin g organic lining (Al ler, 1983). Thus, from a geochemical st andpoint, the chemical
Figure 27. Mesh diagrams showing the steep geochemical gradients in the immediate vicinities of
burrow walls as wel l as WSI. The rel ief in t hese mesh di agrams indi cates the con centration valu es:
where it is high, the concentration is high, and steep slopes indicate steep gradients in the
concen tration. Th ere is a reg ion o f u niform bott om-water O2and TCO2concentrationsalong the
burrow axi s, becau se the bu rrow is a ssumed to b e 100% irr igated.
2001] 447
Furukawa et al.: Bioirrigation modeling
mass transfer in the vicinity of burrow walls is not always analogous to that in the vicinity
of WSI. The microbial reaction rates, and all other parameters that are coupled to the
microbial reaction rates, are affected by the unique properties of burrow walls. Further
geoch emical, mi crobial and microfabric c haracterizati ons of bu rrow wa lls are n ecessary in
order to fully evaluate the quantitative effect of bioirrigation.
e. Comparison with 1D nonlocal treatment
Boudreau (1984) has shown that the radial diffusion term in Aller’s tube model (Aller,
1980) can be converted to a 1-D non-local transport term using the following relationship:
a~x!52D9r1
~r2
22r1
2!~r#~x!2r1!(29)
where ais the non-local exchange function, and r#(x) is the radial distance from the burrow
axis whose concentration value is equal to the radially averaged concentration for that
depth. By using the above term in 1D diagenetic equation:
D9d2C
dx22a~x!~ C
#
2C0!1R50. (30)
the 2D tube model becomes equivalent to the 1D nonlocal model. The 1D nonlocal
treatment of bioirrigation has been widely used: in most cases, a(x) is assigned a simple,
depth-dependenta priori function such as exponential decay and linear decay (Martin and
Banta, 1992; Matisof and Wang, 1998; Kristensen and Hansen, 1999; Schlu¨ter et al.,
2000).
The 1D nonlocal exchange function, a(x), can be determined using the hindcast results
for our laboratory mesocosms in the same manner shown by Budreau (1984):
a~ x!52D9r1~x!
~r2
22r1~x!2!~r#~x!2r1~x!! (31)
This is identical to Eq. (29), except for r
1
being a function of depth. Detailed derivation
steps can be found in Boudreau (1984; 1997), p. 75).
The concentration distributions and horizontally averaged concentrations of NH
4
1
and
SO
4
2
2
were used to evaluate the r#(x) as functions of depth for the model quadrants. The
depth-function for r#(x) can be determined by comparing the radial co ncentration di stribu-
tion for each de pth inte rval with the hori zontally ave raged conc entration for th e correspo nd-
ing depth interval. All other parameters can be found in Table 4. The values of r#(x) as a
function of depth are shown in Figure 28.
The r#(x) pro les derived from the concentration distributions if SO
4
2
2
and NH
4
1
are
nearly independent of depth (Fig. 28). In many cases, r
2
.. r
1
, thus (r
2
2
2r
1
(x)
2
) term in
Eq. (31) may be regarded as constant. Thus, using Eq. (31), we derive:
448 [59, 3
Journal of Marine Research
a~x!} r1~x!
r#2r1~x!. (32)
The above expression (32) indicates that, under the assumptions employed in this study,
the depth dependency of 1D nonlocal exchange coef cient a(x) is directly correlated to
the dep th depe ndent distri bution of bu rrows. Howev er, th e depe ndency is no t linear: rath er,
the depth attenuation of a(x) is greater than that of r
1
(x), as the denominator in the right
side of (32) increases with depth.
6. Conclusions
This study establishes that the two-dimensional bioirrigation model, in which depth
distrib ution of burrows is ex plicitly c onsidered , is an app ropriate tool for stu dying the early
diagenesis of sediments populated by Schizocardium sp. and other benthic infauna that
create and ush deep burrows. The model application to our benthic mesocosm environ-
ments reveals that the common assumptions used in early diagenetic studies can be
critically evaluated by examining the agreement between measured and model calculated
depth pro les of pore water species. Such assumptions include the prescribed depth
dependency of OM degradation rate, C/N ratio, as well as the lack of consideration for
macrofau nal met abolite co ntributio ns to th e po re water. The stea dy stat e model also s hows
the importance of burrow walls as the interfacebetween oxic and anoxicenvironmentsand
site of intense chemical mass transfer. It is evident that more studies are needed to fully
characterize the temporal dynamics of burrow wall environment and its effects on
microbial activities and subsequent chemical mass transfer. Finally, the 1D nonlocal
exchange coef cient for bioirrigation was found to be directly correlated with the
Figure 28. Dep th pro les of r#(x) calculated from comparing the model-calculated 2D distributions
of SO4
2
2
and NH4
1
concentrationswith the radially-averagedconcentrations.
2001] 449
Furukawa et al.: Bioirrigation modeling
depth-dependent distribution of burrows through the comparison between model-
calculated radial concentration distributions with the radially averaged concentration
values.
Acknowle dgments . We thank J. Wa tkins and C. Va ughan f or la boratory a ssistanc e and Rick Mang
and crew of R/V Kit Jones for eld assistance. Discussions with C. Meile, M. Richardson, P. Van
Cappellen and C. Koretsky helped the designs of laboratory experiments and calculation schemes.
Taxonomic assistance was provided by E. Ruppert and G. King. This study was funded by ONR
322GG (Dr. J. Kravitz, Program Manager, Program Element No. 0601153N). NRL contribution
number JA/7431-00-0014.
REFERENCES
Aller, R. C. 1980. Quantifying solute distribution in the bioturbated zones of marine sediments by
de n ing an ave rage microen vironment. Geochi m. Cosmochim. Acta , 44, 1955–1965.
—— 1982. The effects of macrobenthos on chemical properties of marine sediment and overlying
water, in Animal Sediment Relations, P. L. McCall and M. J. Tevesz, eds., Plenum Press, NY,
352 pp.
—— 1983. The importance of the diffusive permeability of animal burrow linings in determining
marine sed iment chemi stry. J. Mar. Res. , 41, 299 –322.
—— 1984. The importance of relict burrow structures and burrow irrigation in controlling
sedimen tary solu te dist ributions. Geoc him. Cos mochim. Ac ta, 48, 1929–1934.
—— 1994 . The sedi mentary Mn cy cle in Lo ng Is land Sou nd: Its r ole as in termediate o xidant a nd the
in uence of bioturbation, O2and Co r g ux on diagenetic reaction balances. J. Mar. Res., 52,
259–295.
Aller, R. C. and J. Y. Aller. 1998. The effect of biogenic irrigation intensity and solute exchange on
diagen etic reactio n rates in ma rine sed iments. J. Mar. Re s., 56, 905–936.
Aller, R. C. and J. Y. Yingst. 1980. Relationships between microbial distributions and the anaerobic
decomp osition of or ganic mat ter in su rface se diments o f Lon g Islan d Soun d, USA. Mar . Biol ., 56,
29 – 42.
Bentley, S. J., Y. Furukawa and W. Vaughan. 2000. Record of event sedimentation in Mississippi
Sound. Tran sactions —Gulf Coast Associ ation of Geolo gical Socie ties, 50, 715–723.
Bentley, S. J. an d M. D . Ri chardson. 2 001. Biogen ic in uences on p orosity a nd s hear str ength of th e
shallo w seabed: Results fr om long- term mesocosm stu dy. Mar. Geol ., (submitte d).
Boudreau, B. P. 1984. On the equivalence of nonlocal and radial-diffusion models for porewater
irrigation.J. Mar. Res., 42, 731–735.
—— 1996. A method-of-lines code for carbon and nutrient diagenesis in aquatic sediments.
Compute rs & Geo sciences , 22, 479– 496.
—— 1997 . Diagen etic Models an d Their Imp lementati ons. Springer-Ve rlag, Heidelb erg, 414 pp .
Boudreau, B. P. and R. L. Marinelli. 1994. A modeling study of discontinuous biological irrigation.
J. Mar. Res., 52, 947–968.
Bouldin, D. R. 1968. Models for describing the diffusion of oxygen and other mobile constituents
across th e mud-wa ter inter face. J. Ecol ., 56, 77– 87.
Bourbie´ , T., O. Coussy and B. Zinszner. 1987. Acoustics of Porous Media. Gulf Publishing
Company,Houston, 334 pp.
Cai, W. J. and F. L. Sayles. 1996. Oxygen penetration depths and uxes in marine sediments. Mar.
Chem., 52, 123–131.
Can eld , D. E., B. Th amdrup an d J. W . Hansen . 1993. The anaerobic degr adation of or ganic matter
in Danish coasta l sediment s: Iron re ductio n,ma nganes e reductio n,a nd su lfate red uction. Geo chim.
Cosmoch im. Acta, 57, 3867–3883.
450 [59, 3
Journal of Marine Research
D’Andre a, A. F. an d G. R. Lopez. 19 97. Benth ic macr ofauna in a s hallow wat er ca rbonate se diment:
major bio turbators at the Dry Tortugas . Geo-Mar. Lett., 17, 2 76 –282.
Davey, J. T. 1994. The architecture of the burrow of Nereis-diversicolor and its quan ti cation in
relatio n to sedi ment-wate r exchang e. J. Ex p. Mar. Biol. Ec ol., 179, 115–129.
Emerson, S., R. Jahnke and D. Heggie. 1984. Sediment-water exchange in shallow water estuarine
sedimen ts. J. Ma r. Res., 42, 709–730.
Forster, S. 1996. Spatial and temporal distribution of oxidation events occurring below the
sedimen t-water int erface. P.S.Z.N. I: Mar. Ecol., 17, 309–319.
Forster, S. and G. Graf. 1995. Impact of irrigation in oxygen ux into the sediment: intermittent
pumping by Callianassa subterranean and “piston pumping” by Lanice conchilega. Mar. Biol.,
123, 335–346.
Furukawa, Y., S. J. Bentley, A. L. Shiller, D. L. Lavoie and P. Van Cappellen. 2000. The role of
biologically-enhanced pore water transport in early diagenesis: An example from carbonate
sedimen ts in the vic inity of North Key Harbor, Dry To rtugas National Park, Flor ida. J. Mar. Res.,
58, 493–522.
Furukawa, Y., R. A. Socki, L. Elrod and K. K. Bissada. 2001. Two-dimensional numerical modeling
of geochemical interactions between infauna and sediments during early diagenesis of burrowed
shelf car bonate sedim ents in t he Dry Tortu gas, Florida . Geochim. Co smochim. Ac ta, (submi tted).
Giray, C. a nd G. M . Ki ng. 199 7. Effect o f natu rally occu rring bromop henols on s ulfate red uction and
ammonia ox idation in intertida l sediments . Aquat. Microb . Ecol., 13, 295–301.
Gust, G. and J. T. Harrison. 1981. Biological pumps at the sediment-water interface: mechanistic
evalua tion of th e alphei d shrimp Alpheus mackayi an d its irrig ation patt ern. Mar. Biol ., 64, 71–78.
Hansen, K., G. M. King and E. Kristensen. 1996. Impact of the soft shell clam Mya arenaria on
sulfate red uction in a n int ertidal sed iment. Aqua t. Micro b. Ecol., 10, 181–194.
Jensen, P., R. Emrich and K. Weber. 1992. Brominated metabolites and reduced numbers of
meiofauna organisms in the burrow wall lining of the deep-sea enteropneust stereobalanus-
canadensis.Deep-Sea Res. Part A., 37, 1247–1253.
Jørgensen, B. B. and N. P. Revsbech. 1985. Diffusive boundary layer and the oxygen uptake of
sedimen ts and detrit us. Limnol. Ocea nogr., 30, 111–122.
Kristense n, E. 19 84. Effect o f natura l concentr ations on exc hange of nu trients betwe en a po lychaete
burrow in e stuarine s ediments and ove rlying wate r. J. Exp . Mar. Bi ol. Ecol ., 75, 171–190.
—— 1988. Benthic fauna and biogeochemical processes in marine sediments: microbial activities
and uxes, in Nitrogen Cycling in Coastal Marine Environments, T. H. Blackburn and J. So-
rensen , eds., J ohn Wile y & Son s Ltd., Chi chester, 275 –299.
—— 1989. Oxygen and carbon dioxide exchange in the polychaete Nereis virens: in uence of
ventilation activity and starvation. Mar. Biol., 101, 381–388.
—— 20 00. Organ ic matter d iagenesis a t the oxic/anoxi c interfa ce in coast al marine s ediments, wi th
emphasison the role of burrowing animals. Hydrobiologia,426, 1–24.
Kristensen, E. and K. Hansen. 1999. Transport of carbon dioxide and ammonium in bioturbated
(Nereis diversicolor
) coa stal, marine sed iments. Biogeo chemistry ,45, 147–168.
Lavoie, D. M., R. J. Baerwald, M. H. Hulbert and R. H. Bennett. 1996. A drinking straw minicorer
for se diments. J. Sed. Res., 66, 1030.
Leppard, G. G., A. Heissenberger and G. J. Herndl. 1996. Ultrastructure of marine snow. 1.
Transmis sion el ectron microsc opy me thodolog y. Mar. Ecol. Pr og. Ser., 135, 289 –298.
Levin, L. A., N. E. Blair, C. M. Martin, D. J. DeMaster, G. Plaia and C. J. Thomas. 1999. Macrofau-
nal processing of phytodetritus at two sites on the Carolina margin: in situ experiments using
C-13-la beled diato ms. Mar. Ec ol.-Prog. Ser. , 182, 37–54.
Mackin, W. R. an d R. C. Aller. 19 84. Ammon ia ads orption in ma rine s ediments. Limnol . Ocean ogr.,
29, 250 –257.
2001] 451
Furukawa et al.: Bioirrigation modeling
Marinelli, R. L. 1992. Effects of polychaetes in silicate dynamics and uxes in sediments:
Importa nce of spe cies, a nimal activity and po lychaete e ffects o n b enthic d iatoms. J. Mar. Res., 50,
745–779.
Marinelli, R. L. and B. P. Boudreau. 1996. An experimental and modeling study of pH and related
solute s in an irri gated anoxic co astal sediment . J. Mar. Res. , 54, 939 –966.
Martin, W. R. and G. T. Banta. 1 992. Th e mea surement o f sed iment irri gation rat es: A compa rison of
the Br-tracer and 222Rn/226Ra di sequilibriu mte chniques. J. Mar. Res., 50, 125–154.
Matisof, G. and X. S. Wang. 1998 . Solute tran sport in sedi ments by fre shwater i nfauna l bioirrig ators.
Limnol. Oce anogr., 43, 1487–1499.
Meile, C., C. M. Koretsky and P. Van Cappellen. 2001. Quantifying bioirrigation in aquatic sedi-
ments: An i nverse mo deling approa ch. Limnol., Oc eanogr., 46, 1 64 –177.
Millero, F. J. 1995. Thermodynamics of the carbon dioxide system in the oceans. Geochim.
Cosmoch im. Acta, 59, 661–677.
Press, W. H., S. A. Teukolsky, W. T. Vetterling and B. P. Flannery. 1992. Numerical Recipes in
FORTRAN, Se cond Edit ion. Cambrid ge Univers ity Press, Ca mbridge, 9 63 pp.
Riisgård, H. U. 1991. Suspension feeding in the polychaete Nereis diversicolor. Mar. Ecol. Prog.
Ser., 70, 29 –37.
Rowden, A. A. and M. B. Jones. 1995. The burrow structure of the mud shrimp Callianassa-
subterranea (Decapoda,Thalassinidea) from the North-Sea. J. Na t. Hist., 29, 1155–1165.
Scaps, P., S. Brenot, C. Retiere and G. Desrosiers. 1998. Space occupation by the polychaetous
annelid Perinereis cultrifera: In uence of substratum heterogeneityand intraspecic interactions
on b urrow struc ture. J. Mar. Bio l. Assoc. UK., 8, 435– 449.
Schlu¨ ter, M., E. Sauter, H. P. Hansen and E. Suess. 2000. Seasonal variations of bioirrigation in
coasta lsediments :Modelling of eld d ata. Geochim. Cosmochim. Acta, 64, 821– 834.
Van Cappellen, P. and Y. Wang. 1996. Cycling of iron and manganese in surface sediments: A
generaltheory for the coupled transportand reaction of carbon, oxygen, nitrogen, sulfur, iron, and
mangan ese. Amer. J. Sci ., 296, 197–243.
Ziebis, W., S. Forster, M. Huttel and B. B. Jørgensen. 1996. Complex burrows of the mud shrimp
Callian assa truncata and th eir geo chemical i mpact i n the sea b ed. Nature, 382, 6 19 – 622.
Received : 17 April, 2000;re vised: 26 April, 2001.
452 [59, 3
Journal of Marine Research
