Limnol. Oceanogr., 32(2), 1987, 436441 0 1987, by the American Society of Limnology

and Oceanography,

Inc.

Slow uptake of 32P over a barrier reef flat M. J. Atkinson Department of Zoology, University of Western Australia, Nedlands, Western Australia 6009

D. F. Smith CSIRO Division of Fisheries Research, P.O. Box 20, North Beach, Western Australia 6020 Abstract A double-label experiment with 32Pand 3H was conducted over a reef flat dominated by algae in the Indian Ocean. Very little 32P was removed relative to 3H, demonstrating relatively slow exchange of P between the benthos and the water column. The best estimate of the constant of proportionality (k) between the rate of respiration-normalized P uptake (P*/R*) and concentration of P ([PI) was 9.2 liters (mmol 0,)--l, about the same value reported for two other reef flats.

The concentration of phosphate (P) in the water column has been shown to change little over several reef flats (Odum and Odum 1955; Pilson and Betzer 1973; Atkinson 198 1; Johannes et al. 1984). Atkinson (1987) showed that the maximal exchange rate of P between the water column and the reef benthos is surprisingly slow compared to the flux of P across some reef flats. The maximal exchange rate of P, as determined in laboratory experiments, was about equal to the net rate of P uptake observed over two reef flats. Thus it was concluded that a typical reef flat community removes only a small percentage (l-5%) of the P crossing the reef, and most, if not all, of the P that is taken up by the community is used for net community production. Atkinson (1987) presented a relationship between rate of P uptake, the ambient concentration of P in the water, and the respiration rate of the community. To verify this relationship between the metabolic variables and to demonstrate that reef flats take up P relatively slowly, we performed the following large-scale experiment on 32P uptake over a barrier reef flat. We used the C.S.I.R.O., Division of Fisheries Research, Field Station on Rat Island, Houtman Abrolhos Islands. Hatcher Research Associates provided both the logistic and administrative support. We especially thank B. Hatcher for assistance in the field.

Background The rate of P uptake (P* = dP/dt) normalized to the rate of community respira-

tion (R* = dO,/dt) has been shown to be proportional to the concentration of P in the ambient water ([PI), by a constant (k) that is about 10 liters (mmol O,)- * (Atkinson 1987). P*/R* = k[P].

(1)

P* and R* can be expressed as a rate per unit volume or a rate per unit area, as long as the units are consistent. Thus as concentration of P or rate of community respiration increases, so does the rate of P uptake. We can rewrite Eq. 1 to calculate the time scale for depletion of P in the water above a square meter of reef, assuming the reef water remains stationary. Dividing Eq. 1 by P* and k and multiplying by the depth of the water (D) results in an equation for the time of P depletion (T).

T = Dl(kR*) = [P]D/P*.

(2)

T can be estimated for a hypothetical reef flat by assuming the following values: D = lm,k= 10 liters (mmol 0,)) *, and R* = 500 mmol 0, m-2 d-l; T is therefore about 5 h. The residence time of water over many reef flats is from 0.2 h to at most 0.5. Therefore we might expect only 4-10% (0.2/5 and 0.5/5) of the P to be removed from the water in the time it takes a patch of water to cross the reef flat. In general, this change in concentration is barely detectable (10% X 0.2 PM = 0.02 PM). Thus, instead of trying to measure small, nearly undetectable changes in concentration of P to calculate the rate constant for

436

437

Slow uptake of j2P field uptake (k), we used the following relationship between k and R* (Atkinson 1987):

k = b/R”,

(3)

where b is the first-order rate constant of In 32P vs. time, d(ln 32P)/dt. P* and [P] cancel out of Eq. 1 because the uptake of 32P is directly proportional to the specific activity. In this field experiment, the concentration of 32P in each water sample was corrected for dilution by horizontal diffusion by normalizing the 32P activity of the sample to its 3HOH activity. The 3HOH will show no biologically mediated changes as its specific activity approaches zero, due to the high concentration of HOH in seawater (i.e. 55 M). As the water mixes, both 32P and 3H should be diluted at equal rates. If 32P is removed from the water relative to 3H, then 32P should decrease faster than 3H; if only a small percentage of 32Pis removed, as we predict, then there should be little change in 32P with respect to 3H. We can quantify the relationship between 32P and 3H with Eq. 3. The 32P/3H experiment was conducted at a different time than the determination of R*. Since respiration rate was measured per area of reef flat, the difference in depth between the determination of R* and the 32P/ 3H experiment must be included in Eq. 3, which becomes

k = b(Dp)lR* = {d[ln(32P/3H)]ldt}Dp/R* (4) where Dp is the average depth (m) of the water during the 32P experiment. Solving Eq. 4 over a finite interval and simplifying results in the following equation for k:

k = ln(32P,3H0/32P03H,)Dp/(AtR*).

28”40’s

(5)

To evaluate k, we need the best estimate of the initial 32P and 3H activities (32P0 and 3H0) and the final 32P and 3H activities (“P, and 3H,). The best statistical estimates of 32P,-,and 32Pt will be the estimates derived from a linear regression of log 32P vs. log 3H. The duration of the experiment in days is At; k is in units of liter (mmol O,)- l; k will be calculated with Eq. 5 and our data.

iz+f

INDIAN OCEAN

DISAPPEARINQ SL. REEF FLAT

Fig. 1. Location map of study site. the 32P/3H experiment- x . A solution of 32P,3H, and flourescein dye was mixed into the upstream water and then followed across the reef flat. The dye patch was followed visually and samples of the solution were taken about once per minute. 3H was used as a conservative tracer relative to 32P.

Methods The experiment was conducted on Disappearing Island reef flat, a 600-m-wide barrier reef flat in the Houtman Abrolhos Islands, Indian Ocean (Fig. 1). R* was estimated by measuring changes in the concentration of O2 between upstream and downstream water samples collected at night. An initial water sample was taken in a BOD bottle at the algal crest of the reef flat, the water was then followed across the reef with chemical luminescent drogues (Cyalumes, American Cyanamid Co.), and finally a downstream water sample was collected. The change in time between upstream and downstream samples was noted (At). Upon collection, the water samples were immediately fixed for Winkler

438

Atkinson and Smith

titrations; these samples were titrated the next day. Water depth (D) was measured with a leadline every 10 s while our boat drifted across the reef flat. The areal rate of respiration (R*) of the reef flat was calculated with the equation (Atkinson and Grigg 1984)

R” = (0, - O,)D/At.

(6)

Five transects were performed between 2000 and 2200 hours on 8 October 1984. The tracer bolus consisted of a solution of [32P]orthophosphate (1.5 GBq ml-l) and 3H-labeled water (37 GBq ml- I) supplied by Amersham (Australia) Pty Ltd. Both solutions of radioisotopes were mixed with 1 liter of membrane-filtered (pore size, 0.4 pm) seawater containing fluorescein. The tracer bolus was introduced into the water column at the upstream end of the reef flat at 1503 local time by pouring it into water only 20 cm deep which was flowing unidirectionally over the reef crest at a speed of about 0.5 m s- l. The water depth gradually increased downstream from 20 to 100 cm over the length of the transect. The combined effect of rough bottom topography and addition of tracer to the rapidly moving shallow water produced a uniform front of water downstream with fluorescein and label reaching from the bottom to the surface of the water column. An estimate of the time to mix the water vertically over the mean depth of the transect (0.668 m) can be made with a knowledge of the bottom roughness and the water velocity (White 1974). If we assume a roughness of 10 cm and a flow of 0.5 m s-l, the time scale to mix water 0.7 m deep is 1 min. The radioactive patch was sampled at intervals of once per minute as it crossed the reef flat. Samples of the labeled water were collected by dipping a 50-ml syringe half way into the water column and filling the syringe. The full syringe was then placed in a shoulder bag. Simultaneous measurements of water depth were taken with a leadline (Dp). The transect was completed in 24 min. Within 30 min after the last sample was collected, the radioactive samples were filtered through an in-line, 0.2pm pore-size membrane filter (Millipex, Millipore Corp.) attached to

a syringe and stored in clean scintillation vials. At the laboratory in Perth, 10 ml of sample were transferred from the field scintillation vials to clean scintillation vials containing 10 ml of InstaGel (Packard Instr. Inc.), a scintillant for aqueous samples. After dark adaptation, all samples were counted to the same number in a Packard liquid scintillation counter (model 3360). 32P was counted in the 32P window nine times in 22 d, and then a first-order decay curve was fitted to these data. Initial activity of 32P was calculated from the decay curve. After 6 months, when the 32P had decayed below background, “H was counted in the 3H window. All samples were counted to < 1% error in net cpm. This counting method was used to obtain the best possible values for 32P and 3H. The activities of 32P and 3H were transformed to log 32P and log 3H so that the low values of 32P and 3H would have statistical weight equal to the high values in the linear regression. A geometric mean regression was calculated (Ricker 1973) because both 32P and 3H had measurement error.

Results The concentration of 32P and 3H in the radioactive patch decreased over 300-fold during the 24-min transit of the water across the reef flat (Table 1, Fig. 2). Relative to the conservative tracer, 3H, very little 32P was removed from the water, demonstrating that the reef did not take up 32P rapidly from the ambient water. According to Eq. 5, we needed to know the rate of community respiration (R*), the mean depth of the water (Dp), and the linear regression between 32P and 3H to calculate the uptake rate constant (k) for this 32P/3H experiment. R* was estimated to be 526 mmol O2 m- * d- I, t60 (Table 2). The concentration of O2 was about 2% above saturation at the upstream end of the transect and about 2% below it at the downstream end. Using a diffusion coefficient of O2 for a 5 m s- Lwind speed from Marsh and Smith (1978), we estimated the gas flux in or out of the water to be only 3% of the respiration rate; therefore it was not included in our estimate of R*. The mean depth of the water

439

Slow uptake of 32P Table 1. j2P and 3H activities (cpm 10 ml-’ of sample) for each water sample across the reef flat. Time since the experiment began-l. The linear regression between log 32Pand log 3H was used to calculate k, the first-order rate constant. Statistics for the regression are also shown. I (min)

=P

‘H

0.43 5,621.g 37,372 13,999 1.03 3,045.o 14,214 2.00 2,802.2 3.10 1,613.5 10,691 4.10 7 14.69 4,050 5.27 909.53 4,123 6.70 287.95 2,028 2,134 8.15 355.32 9.73 307.27 1,681 11.48 100.85 660.9 14.95 139.27 738.3 16.77 104.80 648.3 18.20 91.06 481.9 19.13 81.43 482.9 20.32 72.25 491.9 21.37 71.70 427.2 22.65 36.78 233.7 23.87 17.48 102.6 log 32P= -0.83024 + l.O2009(log 3H) r-2= 0.994 Explained MS = S2y = 8.6552 Unexplained MS = S2yx = 0.03 19 x2 = 8.3670 SE regression coefficient = 0.019528 95% C.L. of the slope = kO.041399 95% C.L. of the 32Pestimate = + 1.07 cpm Mean 32P= 270 cpm

during the 32P/3H experiment was 0.668 m. The linear regression between log 32P and log 3H was log 32P = l.O2009(log 3H) 0.83024, r2 = 0.994 (Table 1). To estimate 32P0and 32PI for use in Eq. 5, we substituted log 3H at t = 0.43 in the above regression (Table l), resulting in 3H, = 37,372 and 3*po = 6,819.8; next, we substituted log 3H at t = 23.87 in the regression, resulting in 3H, = 102.6 and 32P1= 16.63. Thus Eq. 5 is k = ln(16.63 x 37,372/6,819.8 x 102.6)

1

The uptake rate constant of 9.2 liters (mmol 0*)-l falls within the range of 7-l 0 calculated from laboratory experiments and net rates of P uptake over two reef flats (Table

3).

I 4 (LOG

I 5

cpm)

Fig. 2. Log 32Pvs. log 3H (cpm 10 ml-’ of sample) during the course of the experiment (data from Table 1). Note that in 24 min the 32Pand 3H were diluted 300-fold. The line is the regression fit to the data: log 32P = l.O2009(log 3H) - 0.83024. The regression is used to calculate the uptake rate constant (see Table 3).

The regression slope between log 32P and log 3H, 1.0201, had a 95% C.I. of kO.0414. Therefore, the resultant range of k, calculated by using the slopes 1.0 and 1.06 15 in the above analysis, was between zero (in the case when the slope equals 1.0) and 28 (in the case when the slope equals 1.06 15). Thus there was a small chance that there was no uptake of 32P and also a small chance that the rate of 32P uptake could have been up to threefold faster than our best estimate. The fact that the regression slope was close to 1.O demonstrates the slow uptake of 32P by the reef. Table 2. Summary of the calculations to estimate the areal rate of respiration (mmol 0, m-2 d-l) (R*). 0, is initial 0, concentration; 0, is final 0, concentration. Owe (PM) 0,

k = 9.2 liters (mmol 0,)-l.

I 3 %I

.(0.668)/- 526(23.44/60 x 24) or

V 2

00

(0, - 0”)

Time (min)

Depth (m)

225 234 -9 23 0.752 224 237 -13 20 0.656 228 238 -10 15 0.591 226 234 -8 21 0.642 225 235 -10 15 0.702 Mean f SE Net 0, diffusion is