41 - Association for the Sciences of Limnology and Oceanography

Transcription

41 - Association for the Sciences of Limnology and Oceanography
Limnol. Oceanogr., 41(S), 1996, 1790-1799
0 1996, by the American
Society of Limnology
and Oceanography,
Inc.
Separating biological and physical changes in dissolved oxygen
concentration in a coral reef
Steven Kraines’
Department
of Chemical
Systems Engineering,
University
of Tokyo, 7-3- 1 Hongo, Bunkyo-ku,
Tokyo
113, Japan
Yoshimi Suzuki
Department
of Earth Science, Shizuoka University,
Shizuoka, Japan
Koichi Yamada and Hiroshi Komiyama
Department
of Chemical
Systems Engineering,
University
of Tokyo
A hstract
Measured changes in dissolved oxygen (DO) concentrations in a coral I-ccf at Miyako Island are markedly
different from the diurnal sinusoidal pattern expected if biological processesare dominant. We were able to
resolve this paradox by constructing a simple model that combines the irflucnccs of organic production and
air-sea exchange with water flow and mixing through the reef system. The model successfully replicates the
observed DO behavior, suggesting that the simple flow system we propose adequately accounts for these
water advection effects. Our optimized estimates of gross production and respiration on the reef flat arc 13
m and 5.2 g C m-2 d-l, giving a P: R ratio of 2.5. We obtained best fits to the data in the lagoon with a
production rate of 4.9 g C m-2 d-l and a respiration rate of 5.2 g C m-: d-l for a P: R ratio of 0.9.
Coral reefs have received much attention recently because of their role in the global carbon budget and global
warming from COz accumulation
in the atmosphere
(Crossland et al. 199 1; Ware et al. 1992; Kinsey and
Hopley 199 1). CO2 fixation by net biological production
is one of the least understood parts of the global carbon
cycle, and coastal margins may play an important role in
this regard (Suzuki 1993; Walsh 199 1; Smith 198 1).
Among natural ecosystems, coral reefs have one of the
highest gross organic production
rates (Lewis 1977);
moreover, coral reef autotrophs may be increasing their
uptake of CO;! through both nutrient and CO2 fertilization
effects (Walsh 199 1; Broecker and Pcng 1993; Lewis 1985).
Also, coral reefs are far from closed systems in that seawater, the medium of the system, is exchanged relatively
rapidly. This advective transport of material to and from
the ocean is therefore a major element of the mass balance
in the reef (Smith 1984). The important role of advective
transport suggests significant transport of organic carbon
to the deep ocean by coral reefs (Walsh 199 1).
Kinsey (1985) reviewed metabolic rate studies in coral
reefs and concluded that the ratio of gross daily organic
production to community respiration in coral reef sysl Corresponding author.
Acknowledgments
We thank Kazuo Yamamoto for advice in designing the basic
model, Richard Zimmerman for extensive comments and advice on the manuscript, Steven V. Smith for careful review and
valuable comments, and the cooperative efforts of Tamotsu
Omori.
Financial support came from the Research Institute of Innovative Technology for the Earth (RITE).
terns, an indication of the dcgrec to which an ecosystem
is autotrophic, is approximately
one. Furthermore, the
other major billlogical process in coral reefs, the production of CaCO, skeletons, increases the concentration of
COz in the water (Ware et al. 1992). When this calcification process, which occurs during both day and night,
is taken into account, coral reefs seem to release COz.
Ware et al. (1992) quantified these results and concluded
that modern coral reefs arc a net source of 0.02-0.08 Gt
C as CO2 per ;/car. This is bctwecn 0.4 and 1.4% of anthropogenic CO, emissions from fossil fuel combustion.
Frankignoulle
et al. (1994) calculated that the ratio of
COz released per mole of CaCO, fixed by calcification
(0.6 under present conditions) will increase with increasing Pco2 in the ocean. Thus, coral reefs may be expected
to release even more CO2 with increased CO2 accumulation in the al.mosphere.
To more accurately determine the influence of coral
reefs on the global carbon budget and to quantify the
degree to which these ecosystems act as CO;! sources or
sinks, we mus; gain not only quantitative measures but
also a mechanistic understanding
of the carbon cycle
within the reel’. Gaining this mechanistic understanding
is not easy. Ccral reefs involve a complex interaction of
biological activity and water flow patterns, making it difficult to interpret data acquired from a limited number
of observation points in the field. Development of a comprehensive model that reflects the complex interactions
in the reef is an essential step toward understanding the
chemical cycles in the coral reef ecosystem as a whole
from limited samples. Such a model should incorporate
enough complexity to give meaningful results, yet it should
also be made ,IS simple as possible by focusing only on
the key factors. A simplified model is more understand-
1790
DO
changes in a coral reef
1791
able and also makes it easier to perform time-consuming
sensitivity analyses. Simpler models also tend to be more
robust because there are fewer parameters upon which
nonconservative
material fluxes are dependent.
WC can calculate rates of gross photosynthesis (P) and
respiration (R) from observed changes in dissolved oxygen (DO) concentration by means of the following equations:
ADO/At
= ADO/At,,-,,
- ADO/At,,,,
‘xxchgJ
and
ADO/&as
exchg)
= piston velocity
x ([DO] - [DO],,).
Here, exchange of oxygen across the air-sea interface is
quantified with the stagnant film model (Broecker and
Pcng 1974). The piston velocity is calculated from windspeed data taken every hour (Wanninkhof
1992), and
[DO],, is the concentration of DO in equilibrium
with
air calculated as a function of the water temperature and
salinity (Weiss 1970).
Photosynthesis is driven by light, so P is zero at night.
Therefore, the first equation can be used to calculate R
at night and net production (P - R) during the day. P
can be obtained by the differcncc. However, this equation
does not include the contribution
to mass transport by
the flow of water, which can bc extremely important in
coastal ecosystems, as pointed out by Imbcrger et al. (1983)
and Hamner and Wolanski (1988). DO concentration
profiles measured for 3 d at IO-min intervals in October
1993 at Bora Bay in Miyako Island (Fig. 1) clearly show
that concentration changes do not follow the simple diurnal pattern of rise during the day and fall at night prcdieted by these equations (Fig. 2a,b).
Previous research
Because of the large effects of hydrodynamicinfluences,
which we term here water advection, WC need to account
for advection in the above equations or remove the inllucnce of advection from the study area. Several field
experimental methods have been developed to remove
the hydrodynamic influences. These methods include the
basic flow respirometry concepts developed originally for
riverine systems (Lewis 1977) as well as other techniques
such as standing water rcspirometry in coral knolls with
plastic tents or domes (Smith and Harrison 1977) and
the Lagrangian transect technique (Barnes 1983). Although these techniques have advanced our understanding of metabolic processes in coral reefs, several researchers have noted that they arc limited to particular conditions and arcas of the reef system (Atkinson and Grigg
1984; Gattuso et al. 1993).
Furthermore, these methods depend on removing the
advection influence either by studying cases in which the
flow can be neglected or by measuring water moving with
the flow, and therefore they are not able to directly address
the role that water advection plays in the reef-lagoon
’ systems. To assess the contribution of water advection to
the biogeochemical cycle of oxygen in coral reefs, we de-
Fig. 1. Diagram of the coral reef and lagoon at Bora Bay,
Miyako Island, showing the basic flow pattern and the observation points Ml and M3. The Kuroshio flows from west to
east at about 100 cm s-l. Inflow at M3 averages 3.1 cm s-l;
outflow at M 1 averages 3.3 cm s- l (see Fig. 2).
veloped a DO mass flux model based on systems engineering chemical reactor modeling concepts (Levenspiel
1962).
Many DO models for aqueous systems exist. Van Duin
and Lijklema (1989) developed a one-dimensional model
in Lake Wolderwijd. They assumed the lake to be horizontally and vertically well mixed and achieved a good
fit to the data using metabolic parameters derived from
light and dark bottle experiments. Ginot and Herve (1994)
developed a short-term DO model to cstimatc gas exchange, respiration, and photosynthesis parameters. Stefan and Fang (1994) developed a one-dimensional
DO
model for regional lake analysis based on lake morphology and trophic status. These models consider small and
isolated lakes that can be assumed to be perfectly mixed.
Because water-flushing time is rapid in most coral reefs
(Parnell 1988), the assumption of perfect mixing is invalid
here. We therefore incorporate horizontal transport into
a simple model design.
Two models that combinc physical and biological processes should also be mentioned. Fasham et al.‘s (1990)
nitrogen-based model simulates plankton and bacterial
nitrogen cycles in open ocean waters. This model has been
incorporated into the Princeton general circulation model
for the North Atlantic Ocean. The model developed by
Kremer and Nixon (1978) for plankton-nitrogen
dynamics in Narragansett Bay combines a physical-biological
model with a flow system between eight regions that are
assumed to be vertically and horizontally homogeneous.
The flow rates were calculated by a separate flow simulation. These models are cxccutcd over timescales of
months to years, which is appropriate for large bays or
1792
Kraines
-%
E
0
12
8
4
0
et al.
NE
~
t
SE
t
12
13oct
24
12
14oct
24
, :',t
Time
Fig. 2. Graphs of modeled and observed DO concentrations in Bora Bay. The model
conditions are: reef flat (mmol mm2 min-‘)-P
= 1.6, R = ~0.3; lagoon (mmol mm2 min-‘),
P = 0.6, R = -0.3; Ax = 89 m; number of boxes in lagoon = 10. Shaded areas correspond
to the stagnant periods when the Row speed at M3, shown at the bottom, is < 1 cm s-‘. Dates
and times are shown at the bottom of the figure. [a, b.] Observed data are given by the dotted
lines; straight lines show modeled results; observed DO at M3 and model results for box1 (a)
and observed DO at Ml and model results for box 10 (b). [c.] Modeled concentrations at
selected boxes throughout the lagoon. [d, e, f.] Flow speeds at M3 and M 1 and depth in the
lagoon are shown at the bottom of the figure. The direction taken for the calculation of the
flow speeds is shown by the arrows to the right of panels d and e.
DO changes in a coral reef
oceanic regions with long residcncc times. In a study of
reefs in the Great Barrier Reef’, Parnell (1988) found that
small coral reef systems have residcncc times on the order
of days to hours. Also, flow patterns of reefs differ markedly from the circulation dynamics of large bodies of
water (Hamner and Wolanski 1988).
Smith and his colleagues developed a simple, highly
effective approach to account for advection in material
cycles in bodies of water with long rcsidcnce times by
examining the degree to which various elcmcnts that are
nonconservative
with respect to advection deviate from
spatial and temporal variations of a conservative tracer,
such as salinity (Smith and Atkinson 1983; Smith ct al.
unpubl. rep.; Smith and Vcch 1989). However, in a small
water system, such as Bora Bay, with no large freshwater
inflows and a very rapid water turnover time, salinity
variation is too small to cfXcctivcly trace water advection
and mixing.
Our model simulates the obscrvcd DO concentration
changes at Bora Bay over lo-min intervals and allows us
to separate these changes into the cffccts of physical and
biological processes. We use an optimization
fitting procedure described below to dcterminc values of gross photosynthesis (P) and community respiration (R) in the lagoon and the reef flat. The excellent match of the model’s
predictions with observed data indicates that the model
is able to account for the mixing and water flow problems
described above. To our knowledge, this is the first study
of coral reefs that models water mixing and advection
together with biological activity on a timcscale of 1 h by
means of systems engineering techniques.
Description of the site at Bora Bay
Bora Bay is located at the southeast tip of Miyako
Island at 24”45’N, 125’20’E. The Kuroshio (Japan Current) flows parallel to the reef flat at - 100 cm s-l and
may have a significant impact on the flow within the
lagoon (Ho and Marra pers. comm.). The lagoon is parabolic, with an area of -0.2 km2 (Fig. 1) and an average
depth of 3 m. The reef flat forms a nearly complete boundary between the ocean and the lagoon (area, -0.1 km2).
The depth range over the reef is -0.2 to 2 m during the
spring tides; thus, the lagoon becomes a closed system at
the low spring tide.
The land around the bay is heavily cultivated with sugar
cane and tobacco, and we observed large inorganic nutrient concentrations in agricultural runoff water into the
bay. Additionally,
coastal development projects, including the construction of several hotels, have increased sedimentation of the bay. According to local fishermen, there
has been a sharp decline in the fish population, which
may be due to overfishing or to the anthropogenic influence of eutrophication and sediment loading. Thus, Bora
Bay is best characterized as an anthropogenically
influenced reef system. However, in terms of the model we
present here, this characterization
of Bora Bay as nonpristine does not affect our conclusions regarding the ability of the model to reproduce the DO patterns observed
in coral reefs.
1793
Despite anthropogenic eutrophication,
the coral reef
and lagoon at Bora Bay contain a diverse marine flora
and fauna typical of coral reefs in the western Pacific.
Throughout the lagoon, many tropical algae species, both
calcareous and fleshy, were frcqucntly more abundant
than the corals (Zimmerman pcrs. comm.), which may
reflect the high level of eutrophication
in the bay.
The reef and lagoon system was divided into distinct
arcas for modeling. The ncarshorc region out to - 100 m
(a limestone bench substrate with red clay silt and calcareous sand) is rather stagnant and is dominated by siltcovered algal beds. Although biological activity is likely
to bc high in this warm, shallow area, water flow is very
slow, which indicates that this water mass is basically
separate from the bulk of the lagoon water. Because our
objective was to model the highly advective flow regime,
we excluded this region from the model. From 100 m to
the backreef, isolated coral knolls protrude from a floor
of coarse sediment composed of coral rubble and sand
> 1 mm in size; the coral knolls are more numerous toward the inner flat edge of the reef. The coral coverage
is O-20% here, but algae are dominant, with a coverage
of lo-40%. This region corresponds to the lagoon in the
model formulation.
The living biomass on the reef flat,
on the other hand, is dominated by hcrmatypic corals.
This community
is exposed to the air during the low
spring tides. The outer slope is also heavily covered with
coral and algae and drops rapidly to a depth of - 10 m,
forming a sharp boundary between the reef and the surrounding ocean waters (Zimmerman pers. comm.).
Materials and methods
Our model is based on observations WC made at two
points in the reef (Fig. 1). Ml is located on the east side
of the lagoon near the reef flat at an average depth of
-2.5 m. M3 is located on the west side of the lagoon near
the reef in an average depth of 2.3 m. Thcsc points are
in areas of high coral coverage and are surrounded by
several large coral knolls.
Water-flow measurements were carried out by Tokyo
Kyuuei Co. at IO-min intervals at sites M 1 and M3 with
a vane-type flow meter (precision, +2%; range, 2-250 cm
s-l) suspended 1 m from the sea bed.
’ Salinity, temperature, and DO were also measured continuously at Ml and M3 by Tokyo Kyuuci Co. at 1O-min
intervals with an electromagnetic induction conductivity
sensor for salinity, a thermistor sensor for temperature,
and a galvanic electrode sensor for DO measurements.
The sensor package was suspended 1 m from the seabed
(precision levels were +O.O5o/oofor salinity, +O.l”C for
temperature, and -I co.05 mg liter-l or 1.6 mmol m-3
for DO). The device executes a burst of three measuremcnts- 1 s before, during, and 1 s after each lo-min
interval. The three values are then averaged to give the
measured value.
Theflow system in Bora Bay-In rapid-water throughput systems such as coral reefs, concentration profiles are
expected to show a strong tide-induced pattern. To quan-
1794
Kraines et al.
Fig. 3. Diagram of the structure of the tanks-in-series flow
model used to simulate advection and mixing in the lagoon at
Bora Bay. Us are water flow velocities, [X’j,, is the concentration
of X at location y, and Ax is the width of the tanks in the lagoon.
Water enters the system across the M3 reef flat. Flux over the
reef flat, shown by the thick vertical arrows, begins mainly inward at box 0 and becomes increasingly outward with increasing
box number.
tify this tidal influence, we need to account for the water
flow throughout the entire tidal cycle. Here, we use the
flow meter measurements to estimate the flow system at
Bora Bay.
The results of our flow speed measurements at points
Ml and M3 are shown in Fig. 2. The flows are generally
unidirectional;
water travels from the ocean to the lagoon
at M3 and from the lagoon to the ocean at Ml (see Fig.
I). Thus, water seems to flow steadily from M3 to Ml,
parallel to the Kuroshio. If we assume that there is a linear
gradient of north-south unidirectional currents across the
reef flat from in at Ml to out at M3, we can calculate the
average flow volumes that enter on the west side and exit
on the east side.
If WC assume the linear gradient dcscribcd above, the
average instantaneous flow speed over the western half
of the reef flat is half the M3 flow speed and that over
the eastern half is half the M 1 flow speed. The flow volumes are simply the product of the average instantaneous
flow speed coming in or going out, the instantaneous
depth, and half of the reef flat length. Because the flow
speeds are about equal and opposite and the depths are
about the same, the constant component of the flow volume entering the lagoon across the western half of the
reef flat is about equal to the flow volume exiting across
the eastern half. For the first 2 d (13 and 14 October), we
obtained an average flow volume of -65,000 m3 h-l at
both M 1 and M3. The tidal component of water flow in
the reef is given by the area of the reef multiplied by the
tidal excursion and divided by one tidal period. With an
area of 365,000 m2 and the observed tidal excursion of
1.4 m, the tidal exchange is -40,000 m3 h-l.
Finally, we calculate the turnover time of water in the
lagoon as the volume divided by the sum of the two flow
volumes. If we assume an average depth in the lagoon of
3.5 m, as indicated by the depth profile taken by Ajia
Kousoku, the volume of the lagoon is 1,277,500 m3. The
combined flow volume is 105,000 m3 h-i. Therefore, the
turnover time Df water in the lagoon is - 12 h. The water
in the lagoon seems to be flushed twice per day by a
combination of the steady flow parallel to the Kuroshio
and the perpendicular ebb and flow tidal cycles (both of
about the same order).
Groundwater invasion is potentially a major factor for
systems with permeable floors such as coral reefs (Buddemeier and Oberdorfer 1988). However, the rate of
freshwater input, which we determined from changes in
salinity, is ~2,000 m3 h-l across the entire lagoon or 5%
of the combined oceanic flux, so WC have ignored it in
the simple water balance presented above. Freshwater
inflow is included in the model described below for the
calculation of isalinity and temperature changes.
The model-.Our model is based on the mass balance
of a finite volume of water, where the total change of DO
concentration is the sum of the changes due to water
advection, organic production, and gas exchange:
y
= y(biol)
+ dWl
--p
dKl
+ --&flux
+ y(gas
exchg) + y(vol
chg)
ux over reef flat)
between boxes).
We model the physical transport of water and materials
through Bora ISay by dividing the entire ocean and coral
reef-lagoon sy:;tem into three regions: the outer ocean
waters, the ree:Yflat, and the lagoon. The ocean surrounding Bora Bay has DO concentration between 170 and 2 10
mmol m-3 (2 SD) (Omori pers. comm.). This variation
is negligible compared to that of the lagoon. Therefore,
we assume tha; the oceanic water region has constant DO
concentrations and temperature. The Kuroshio seems to
flow strongly outside the reef flat and constantly sweep
fresh oceanic water into the outer ocean region.
We assume :glug flow transport over the reef flat (Levenspiel 1962) and metabolic rates indcpcndent of those
in the lagoon. Plug flow transport means that the material
transport due lo horizontal eddy diffusion on the reef flat
is assumed to be negligible in comparison to advective
transport.
The lagoon is modeled with the tanks-in-series approximation
fi)r imperfect mixing and is divided into
tanks in the e;lst-west direction (Fig. 3). The tanks-inseries method is one of two standard engineering approaches for modeling imperfectly mixed systems. A given volume is divided into parallel subsystems or tanks
of width Ax in the direction of the flow. With one tank
DO
changes in a coral reef
1795
F=[X m-2 min-l] (gas exchange, heat flux,
box
n=2
box n=l
Here: lJt&t) = measured flow speed at M3 in direction of
primary flow
U,,f(t) = flow speed on the reef flat calculated
uMa(t) as shown below
t = time
C = concentration of modeled substance at M3
C* = concentration exiting the reef flat
Cc = concentration of modeled substance in the ocean
Fig. 4. M3 reef flat plug flow model for simulation of concentration changes across the
reef outside M3. Water of a constant concentration flows over the reef flat at velocity UMf(t).
This velocity determines the time spent on the reef flat and the degree of biological and
physical modification. Mixing rcduccs these cffccts considerably as water flows to the moored
sensor at M3.
the system is perfectly mixed and diffusion is infinite. As
the number of tanks approaches infinity, the system reduces to plug flow and diffusion is zero. In general, increasing the number of tanks by decreasing Ax reduces
the rate of diffusion. This formulation
is equivalent to
the use of a dispersion factor or difhusion coefficient in
the mass transfer term, the other basic approach to modeling imperfectly mixed reactors (see Levenspiel 1962).
The sharp peaks and dips in the M3 DO profile (Fig.
2a) occur immediately after the lagoon becomes closed
during the low spring tide. During this period, water is
trapped at a very shallow depth over the reef flat and is
radically altered by biological activity. When the flow
cycle resumes, this highly modified water enters the lagoon followed by constant concentration oceanic water.
This advective transport pattern results in a peak in DO
at M3 but not at Ml because the water flow at Ml is from
the lagoon to the reef flat (Fig. 2b).
To simulate this DO concentration behavior, we use
the hyperbolic tangent function for light-dependency
of
photosynthesis (Kirk 1983; Barnes and Chalker 1990).
We use an I, of 200 pmol photons m-2 s-l as recommended by R. Zimmerman (pers. comm.) based on his
studies of the communities at Bora Bay. We calculated
noontime irradiance from the time and latitude of the
field study (Kirk 1983). Attenuation due to daily averaged
cloud coverage (obtained from the Japanese Metcorological Agency) is given by the equation proposed by Kremer
and Nixon (1978), and we use an average extinction coefficient of 0.1 m-l first approximation
for attenuation
due to water and detritus (R. Zimmerman pers. comm.).
WC model depth with a sinusoidal profile based on high
and low tide depths and times. Flow speed across the reef
flat owing to the tidal action is calculated from the water
balance in the lagoon, as described above.
The second stage of the model calculates the concentration profile of DO entering the lagoon from the reef
flat at M3. From the measured speeds at M3, we estimate
the velocity of water over the reef flat. By integrating these
time-dependent velocities, WC obtain the time that water
takes to travel over the reef flat. Calculation of this period
allows us to calculate the change in DO concentration
resulting from organic production and gas exchange as
the water travels across the reef flat (Fig. 4). We model
gas exchange as described above by using wind speeds
measured hourly by Tokyo Kyuuei. WC adjust for the fact
that M3 is located -20 m downstream from the edge of
the reef by using the concentration profile of the first tank
of the lagoon, as described below. The resulting profile is
shown in Fig. 2a.
The third stage of the model calculates the mass balance
for each of the tanks in the lagoon. The velocity of water
flowing across the reef flat edge is taken as a weighted
sum of the M3 velocity and the tidal velocity calculated
in stage one. We then calculate the DO concentration in
the water coming off the reef flat as in stage two. We
Kraines et al.
1796
Table 1. Results of DO optimization
of rates of gross photosynthesis (P), community
respiration (IX) and the P : R ratio. CO, fixation is the quantity of CO, absorbed by the reef
system from the water passing over. DO metabolic rates (mmol m-2 min-‘) and corresponding
CO2 fixation rates from organic production (g C m-2 d-l) are given (total in t C yr-I).
Metabolic
rate
Region
Reef flat
(area = 0.1 km2)
Lagoon
(area = 0.2 km2)
Total
(area = 0.4 km2)
P
R
P :R
P
R
P:R
P
R
P:R
1.6
-0.3
2.5
0.6
-0.3
0.9
0.9
-0.3
1.4
derive the output water velocity exiting from the east edge
from the water balance in the tank.
Running the simulation through the entire length of the
lagoon resulted in the concentration
profile sequence
shown in Fig. 2c. The sharp peaks observed at M3 gradually diminish and shift. A comparison of modeled and
observed concentration profiles at M 1 shows an excellent
lit (Fig. 2b).
We calculated temperature and salinity distributions
on the reef flat and in the lagoon as described for DOby means of empirical equations for heat transfer across
the water surface and an empirical formula for evaporation together with the estimates of freshwater inflow
derived above (Morikita et al. 1987; Brutsaert 1982). The
modeled profiles match continuous field measurements
very well. The modeled distributions
were used to calculate the temperature and salinity dependence of the DO
equilibrium
concentration in the water (Weiss 1970).
To implement our model, we constructed a program
that executes the following routines. First, the program
calculates lagoon depth from published values of the high
and low tide depths and times, which gives a smooth
curve for the change in depth with time. The calculated
and observed depth profiles are nearly indistinguishable.
The model then calculates the rate of photosynthesis in
the lagoon and on the reef flat as a function of time by
means of the hyperbolic tangent function of light dependence. The light profile is calculated by use of the cloud
coverage and extinction coefficient. Light-saturated photosynthesis rates are then entered. Observed hourly wind
speeds, observed water velocities at M3, and temperature
profiles calculated by the temperature model, which replicate observed tcmperaturc profiles well (data not shown),
are stored next. Because salinity is fairly constant at Bora
Bay, we use an average value of salinity.
Initial conditions in the lagoon are set at a uniform DO
concentration. The program then calculates the time it
takes for the water arriving at the current time step to
cross the reef flat and integrates the effect of biological
activity and gas exchange on the DO concentration across
the reef flat for this period. This reef-altered DO concentration is used as the boundary condition for box zero in
Organic production
co,
fix.
13.1
-5.2
4.9
-5.2
7.4
-5.2
-
Net
7.9
-0.3
2.2
Total
290
-20
270
the main computational routine. Changes in DO concentration from bic’logical activity, gas cxchangc, flow across
the central reef flat, and advective transport are calculated
through the lagoon. DO profiles in each box are stored
and then compared with the observed data. Values of P,
R, and Ax arc readjusted according to the criteria described below.
Most of the parameters used in the model arc based on
empirical, observational,
or published values. Ax is an
adjustable parameter related to the rate of mixing. Given
Ax, we can estimate P and R as the values that give the
best fit of the model to the observed data during the day
and night. We estimated the values of P, R, and Ax using
a best-fit algoril.hm. The concentration profile at M3 has
a gradual increase from dawn due to photosynthesis combined with flow through the reef, a sharp peak where gas
exchange is strongest, and a gradual decrease after dusk
mainly due to respiration. The width of the peak indicates
the degree of mixing in the lagoon. Changes in DO concentration during the ebb cycle (i.c. from high to low tide)
are not influenced by oceanic input over the reef flat.
Thus, we use tne peak starting and ending times to optimize Ax and the periods from 0600 to 1200 and 1800
to 2400 hours to calculate optimum values of P - R and
R, respectively,
Results
The results of the optimizations of P and R are shown
in Table 1. The optimum value for Ax was 89 m. The
corresponding best-fit values of reef flat rates to the M3
observed profile of Y and R are 1.6 and 0.3 mmol m-2
min-l. Conversion to g C m-2 d-l assumes that 1 mole
of carbon is metabolized for each mole of oxygen and
that the photom;ynthetic period is 11.4 h. The reef flat in
this system is a strong organic material producer, with a
P: R ratio of h 2.5 and a net production rate of 8 g C m-2
d- l. This rate is equivalent to 290 t C yr- l over the entire
reef flat.
The values IDf P and R in the lagoon that best fit the
M 1 observed :?rofile are 0.6 and 0.3 mmol m-2 min-‘,
DO changes in a coral reef
.-
5
2 -2
5*
Tizgam
LA!
iiF0
-EC
-ii
0
-10
t
a
-15
1797
-5
12
13 act
24
12
14 Ott
24
12
Time
15 act
Fig. 5. Separation of modeled changes of DO into biological and gas-cxchangc effects. [a.]
Separated contributions of biology (- - -) and gas exchange (-)
to DO concentration change
across the reef flat for water arriving at time t (model description given in Fig. 4). [b.] Same
as panel a but across the lagoon (model description given in Fig. 3). Note that the integration
time equals 1 min in panel b, but the integration time equals the time water takes to cross
the reef flat in panel a.
a lagoon P : R of 0.9 as shown in Table 1. The net
production in the lagoon is -0.3 g C me2 d-l, or -20 t
C yr-l. The total net production over the 0.3-km2 reef
and lagoon system is 270 t C yr- l and the P : R ratio is
- 1.4. This implies a net organic production in similar
coral reefs of -9 t C ha-l yr-l. This value might be an
overestimate because we use only data from Octoberone of the most productive times for a coral reef. A crucial
next step is to analyze data from other seasons at Bora
Bay. Note also that the figure for net organic production
is not directly translatable to total CO2 fixation in the
coral reef system because the process of calcification, which
has not been considered here, leads to a net release of
CO2 into the water (Ware et al. 1992).
giving
Discussion
Our model assumes pcrfcct vertical mixing in the lagoon, plug flow over the reef flat, constant oceanic con-
centrations, a stagnant-film gas-exchange model, and areaaveraged values for P and R. Dcspitc these assumptions
and a highly simplified flow system, the model yields an
excellent fit to the observed DO data as well as to salinity
and temperature over the entire diurnal concentration
profile. This excellent fit suggests that our flow system,
which combines a unidirectional
flow parallel to the Kuroshio with the tidal flow, adcquatcly accounts for the
effects of water advection and that WC have grasped the
essential elements that control the flux of DO through the
reef system. This flow system is based on readily obtainable data. Our model should therefore bc easily adaptable
to other coral reefs. Furthermore, because the flow system
was shown to be adequately modeled, WCcan extend the
model to assess the flux of materials to and from the outer
ocean. Finally, we can use the model to predict how reefs
may perform under diffcrcnt flow regimes.
Our value for the P: R ratio on the reef flat, 2.5, is at
the highest end of the values in Kinsey’s (1985) review
of the literature (0.7-2.5 on coral reef flats). Furthermore,
1798
Kraines et al.
the whole system P : R ratio, 1.4, implies that 40% of
gross community production is not oxidized and is either
stored somewhere in the reef or exported to the ocean.
Anaerobic metabolism in the sediments in this reef possibly contributes to the mineralization
of a significant
quantity of organic carbon; however, even anaerobic metabolism is usually followed by chemical oxidation of the
reduced product in lagoons, which would lead to the same
loss of DO that would be expected if all organic carbon
remineralization
were aerobic. Boucher et al. (1994) studied the community respiration quotient, an indicator of
how rapidly the reduced products of anaerobic metabolism are reoxidized, in coral reef lagoons and concluded
that the quotient is L 1. This conclusion supports the
chemical oxidation hypothesis and confirms that DO is
a reliable proxy for measuring CO2 changes owing to organic metabolism (Boucher et al. 1994; Clavier et al. 1994).
Recently, many researchers have focused on the probIcrn of partitioning
physical and biological components
of changes observed in various mcasurcd substances (Craig
and Hayward 1987; Emerson et al. 1991; Murray et al.
1994). Our model permits us to separate the influences
of gas exchange and biology on the observed DO concentration profiles. Gas exchange during the day accounts
for up to 50% of the biologically induced DO concentration change on&he reef flat and up to 20% of the change
in the lagoon. (Fig. 5). Therefore, it is important to accurately quantify gas exchange of DO across the air-sea
interface even in high-water-flow-through
systems such
as the coral reef at Bora Bay. During periods of slow water
flow on the reef flat, we observed formation of bubbles
on plant surfaces that seemed to be oxygen degassed directly from the plant resulting from the extreme supersaturation of dissolved oxygen in the water. This loss of
oxygen is not accounted for in the model. By ignoring
this oxygen loss, the model underestimates the rate of
organic carbon production during this supersaturated period. It is therefore possible that the P : R ratio, particularly on the reef flat, may be even higher than the value
we have reported.
Our method for estimating the P : I< ratio in coral reefs,
emphasizes the role of advection, which may partially
explain why our P: R results are higher than most published values. As discussed above, published P : R values
are generally based on methods that circumvent the problem of advection by removing the advective influence.
However, if not adequately accounted for, the mixing and
flow factors could act to reduce the apparent biological
effects on observed concentrations
through imperfect
mixing in the water column, which retards transport of
the biologically altered waters to the observation point
or through the advective supply of oceanic waters. Both
of these physical flow factors would reduce the magnitude
of instantaneous changes in DO due to biological activity.
The rates of respiration during the day and night are (to
a first approximation)
equal and would therefore have a
smaller instantaneous effect on DO concentration than
on gross production, which proceeds during the day only.
Therefore, a reduction in the magnitude of changes in
DO concentration
observed would result in a larger un-
derestimation of production than of respiration and thus
a lower P: R mtio.
Our calculations indicate that organic production may
be a significanly
greater sink for atmospheric carbon in
eutrophic coral reefs than in pristine reefs, which have
generally been the objects of coral reef studies. Furthermore, the relate vely healthy appearance of this reef despite
considerable eutrophic influence indicates that with a
strong unidirec:tional flow regime, which acts to reduce
the residence lime of water in the reef, coral reefs may
be able to flourish even under strongly eutrophic conditions. As noted by Ware et al. (1992), a considerable
amount of carbon fixed by this net organic production is
balanced by CO2 release during calcification. However,
their calculations were based on P : R ratios of - 1.1. Our
observations indicate that net production may be up to
six times larger in eutrophied reefs than in pristine reefs
with much less difference in respiration. This difference
in metabolic activity would result in a dramatic shift in
the net balance of CO2 absorption and release. The high
rate of CO2 fixation at Bora Bay seems to occur without
serious damage to the reef corals. By clarifying the mechanisms that cause this phcnomcnon, it may be possible
to discover how to modify other coral reefs to increase
fixation of CC), without adversely affecting other bcneficial aspects of the reef systems. Such biogeochemical
engineering cc,uld mitigate the global warming problem
as well as protect the reefs from eutrophication
damage.
References
ATKINSON,M. .I., AND R. W. GRIGG. 1984. Model of a coral
reef ecosysl.em. Part 2. Gross and net benthic primary production at French Frigate Shoals, Hawaii. Coral Reefs 3:
13-22.
BARNES,D. J. 1983. Profiling coral reef productivity and calcification using pH and oxygen electrodes. J. Exp. Mar. Biol.
Ecol. 66: 1449-l 6 1.
-,
AND B. E. CHALKER. 1990. Calcification and photosynthesis in reef-building corals and algae, p, 109-l 3 1. In
Z. Dubinsky [ed.], Coral reefs. Elsevier.
BOUCHER,G., J. CLAVIER,AND C. GARRIGUE. 1994. Oxygen
and carbon dioxide fluxes at the water-sediment
interface
of a tropical lagoon. Mar. Ecol. Prog. Ser. 107: 185-l 93.
BROECKER,
W. S., AND T. H. PENG. 1974. Gas exchange rates
between air and sea. Tellus 26: 2 l-35.
-,
AND T. H. PENG. 1993. Greenhouse puzzles, part 1:
Keeling’s world. Eldigio.
BRUTSAERT,W. 1982. Evaporation into the atmosphere. Reidel.
BUDDEMEIER,F:. W., AND J. A. OBERDORFER.1988. Hydrogeology and hydrodynamics
of coral reef pore waters, p.
485-490. ~rz Proc. 6th Int. Coral Reef Symp. V. 2.
CLAVIER,J., G. BOUCHER,AND C. GARRIGUE. 1994. Benthic
respiratory and photosynthetic
quotients in a tropical lagoon. C.R. Acad. Sci. Paris Life Sci. 317: 937-942.
CRAIG, H., ANI) T. HAYWARD. 1987. Oxygen supersaturation
in the ocean: Biological versus physical contributions. Science 235: 199-202.
CROSSLAND,C. J., B. G. HATCHER,AND S. V. SMITH. 1991.
Role of co.-al reefs in global ocean production. Coral Reefs
10: 55-64.
DO changes in a coral reef
EMERSON,S., P. QUAY, C. STUMP,D. WILBUR, AND M. KNOX.
199 1. 02, Ar, N,, and 222Rn in surface waters of the subarctic ocean: net biological 0, production. Global Biogeo- _
them. Cycles 5: 49-69.
FASHAM, M. J. R., H. W. DUCKLOW, AND S. M. MCKELVIE.
1990. A nitrogen-based model of plankton dynamics in
the ocean mixed layer. J. Mar. Res. 48: 591-639.
FRANKIGNOULLE,M., C. CANON, AND J.-P. GATTUSO. 1994.
Marine calcification as a source of carbon dioxide: Positive
feedback of increasing atmospheric CO,. Limnol. Occanogr. 39: 458-462.
GATTUSO, J.-P., M. PICHON, B. DELESALLE, AND M.
FRANKIGNOULLE.1993. Community metabolism and airsea CO2 fluxes in a coral reef ecosystem (Moorea, French
Polynesia). Mar. Ecol. Prog. Ser. 96: 259-267.
GINOT, V., AND J. C. HERVB. 1994. Estimating the parameters
of dissolved oxygen dynamics in shallow ponds. Ecol. Model. 73: 169-187.
HAMNER, W. M., AND E. WOLANSKI. 1988. Hydrodynamic
forcing functions and biological processes on coral reefs: A
status review, p. 103-l 13. In Proc. 6th Int. Coral Reef
Symp. V. 1.
IMBERGER,J., AND OTHERS. 1983. The influence of water motion on the distribution and transport of materials in a salt
marsh estuary. Limnol. Oceanogr. 28: 20 1-214.
KINSEY, D. W. 1985. Metabolism, calcification and carbon
production. 1: Systems level studies, p. 505-526. In Proc.
5th Int. Coral Reef Congr. V. 4.
-,
AND D. HOPLEY. 199 1. The significance of coral reefs
as global carbon sinks- response to global greenhouse. Palaeogeogr. Palaeoclimatol.
Palaeoecol. 89: 363-377.
KIRK, J. T. 0. 1983. Light and photosynthesis in aquatic ecosystems. Cambridge.
KREMER,J. N., AND S. W. NIXON. 1978. A coastal marine
ecosystem simulation and analysis. Springer.
LEVENSPIEL,
0. 1962. Chemical reaction engineering. Wiley.
LEWIS,J. B. 1977. Processes of organic production on coral
reefs. Biol. Rev. 52: 305-347.
. 1985. Groundwater discharge onto coral reefs, Barbados (West Indies), p. 477-48 1. Zn Proc. 5th Int. Coral
Reef Congr. V. 6.
MORIKITA, Y ., K. HATA, AND S. MIURA. 198 7. A numerical
model analysis of cold pollution and eutrophication
phe-
1799
nomena in reservoirs [in Japanese]. Public Works Res. Inst.
Min. Construction No. 2443.
MURRAY, J. W., AND OTHERS. 1994. Physical and biological
controls on carbon cycling in the equatorial pacific. Science
266: 58-65.
PARNELL,K. E. 1988. The hydrodynamics of fringing reef bays
in the great barrier reef marine park, p. 503-508. In Proc.
6th Int. Coral Reef Symp. V. 2.
SMITH, S. V. 198 1. Marine macrophytes as a global carbon
sink. Science 211: 838-840.
1984. Phosphorus versus nitrogen limitation in the
marine environment. Limnol. Oceanogr. 29: 1149-l 160.
-,AND
M.J. ATKINSON. 1983. Mass balance of carbon
and phosphorous in Shark Bay, Western Australia. Limnol.
Oceanogr. 28: 625-639.
-,
AND J. T. HARRISON. 1977. Calcium carbonate production of the Mare Incognitum, the upper windward reef
slope, at Enewetak atoll. Science 197: 556-559.
-,
AND H. H. VEEH. 1989. Mass balance of biogeochcmically active materials (C, N, P) in a hypersaline gulf. Estuarine Coastal Shelf Sci. 29: 195-2 15.
STEFAN,H. G., AND X. FANG. 1994. Dissolved oxygen model
for regional lake analysis. Ecol. Model. 71: 37-68.
SUZUKI, Y. 1993. Dynamic cycle of dissolved organic carbon
and marine productivity,;p.
53 l-549. In M. Heimann [ed.],
Global carbon cycles. Springer.
VAN DUIN, E. H. S., AND L: LIJKLEMA. 1989. Modelling photosynthesis and oxygen in a shallow, hypertrophic
lake.
Ecol. Model. 45: 243-260.
WALSH,J. J. 199 1. Importance of continental margins in the
cycling of carbon and nitrogen. Nature 350: 53-55.
WANNINKHOF,R. 1992. Relationship between wind speed and
gas exchange over the ocean. J. Geophys. Res. 97: 73737382.
WARE, J. R., S. V. SMITH, AND M. L. REAKA-KUDLA. 1992.
Coral reefs: Sources or sinks of atmospheric CO,? Coral
Reefs 11: 127-l 30.
WEISS,R. F. 1970. The solubility ofnitrogen, oxygen and argon
in water and seawater. Deep-Sea Res. 17: 721-735.
Submitted: 16 June 1995
Accepted: 20 March 1996
Amended: 24 July 1996