APPLIED AND ENVIRONMENTAL MICROBIOLOGY, JUlY 1994, p. 2259-2264

Vol. 60, No. 7

0099-2240/94/$04.00+0 Copyright © 1994, American Society for Microbiology

Rates of Benthic Protozoan Grazing on Free and Attached Sediment Bacteria Measured with Fluorescently Stained Sediment MATHIEU STARINK,l* IRINA N. KRYLOVA,2 MARIE-JOSE BAR-GILISSEN,1 ROLF P. M. BAK,3 AND THOMAS E. CAPPENBERG1 Centre for Limnology, Netherlands Institute of Ecology, 3631 AC Nieuwersluis,' and Netherlands Institute for Sea Research, 1790 AB Den Burg, Texel, The Netherlands, and Institute of Inland Waters Biology, Academy of Sciences of the Russia Microbiology Laboratory, Borok, Nekouz, Yaroslavl, 152742, Russia2 Received 16 August 1993/Accepted 24 April 1994

In order to determine the importance of benthic protozoa as consumers of bacteria, grazing rates have been measured by using monodispersed fluorescently labeled bacteria (FLB). However, high percentages of nongrazing benthic protists are reported in the literature. These are related to serious problems of the monodispersed FLB method. We describe a new method using 5-(4,6-dichlorotriazin-2-yl)-aminofluorescein (DTAF)-stained sediment to measure in situ bacterivory by benthic protists. This method is compared with the monodispersed FLB technique. Our estimates of benthic bacterivory range from 61 to 73 bacteria protist-' h-X and are about twofold higher than the results of the monodispersed FLB method. The number of nongrazing protists after incubation for 15 min with DTAF-stained sediment is in agreement with theoretical expectation. We also tested the relative aflinity for FLB of protists and discuss the results with respect to a grazing model.

In pelagic marine and freshwater systems, heterotrophic nanoflagellates are major consumers of bacteria (5, 13, 27). They are now widely recognized as important processors of energy and nutrients. When the nanoflagellates are preyed upon by zooplankton, the energy of the heterotrophic nanoflagellates is transported to higher trophic levels within the pelagic system. Although it is believed that microbial food webs also exist in benthic ecosystems (1, 15, 17), very little is known about the interactions between bacteria and protozoa in these food webs. A high level of bacterial production in sediments has been estimated (21, 33), but the fate of this production remains obscure. Benthic ciliates show high specific grazing rates but consume only a minor fraction of the benthic bacterial production, mainly because of the low level of ciliate standing stocks (21). Bak and Nieuwland (2, 3) measured densities of protists in marine sediment systems ranging from 5 x 103 to 350 x 103 cells cm-3. In freshwater systems, nanoflagellate densities in the top layer of the sediment were measured and found to range from 10 x 103 to 170 x 103 cells cm-3 in Lake Gooimeer and up to 220 x 103 cells cm-3 in Lake Vechten (30), The Netherlands. Considering these high densities, we may expect the heterotrophic nanoflagellates to play a significant role in benthic systems. Measuring grazing rates and bacterial production can explain the role of heterotrophic nanoflagellates in benthic microbial food webs. Several techniques developed to measure protist grazing rates are used in marine as well as freshwater ecosystems. Metabolic inhibitors (16), artificial fluorescent bacterium-size particles (7, 23), and fluorescently labeled bacteria (5, 26) have been applied to measure bacterivory in these systems. Bacterivory by benthic ciliates and heterotrophic nanoflagellates has been measured by the monodispersed fluorescently labeled bacteria (FLB) method (11, 20, 21). Although these methods are frequently used, they are

subject to criticism. Measuring grazing rates with eukaryotic inhibitors can cause considerable uncertainties because of inadequate specificity (32). It has also been shown that protists can select against fluorescent tracers such as latex microspheres or paint particles with the same size spectrum as bacteria (6, 26). More commonly, natural or cultured FLB are used to measure bacterivory. Hondeveld et al. (20) have adjusted the FLB method to estimate flagellate grazing directly in sediments, but they noticed a high percentage of benthic protists without ingested FLB, which may result in underestimation of grazing rates. This phenomenon, a limited quantity of bacterivores in the flagellate community, may be caused by differences in grazing preferences for attached and aggregate bacteria. Grazing preferences of surface-associated protozoa for attached bacteria have been demonstrated in batch cultures by Caron (9) and Sibbald and Albright (28). In view of the possible importance of flagellates grazing on attached bacteria, we developed a new method to measure grazing rates of benthic protists, using sediment stained with the fluorochrome 5-(4,6-dichlorotriazin-2-yl)-aminofluorescein (DTAF). Here we describe the procedure for preparing FLB of both types, free as well as attached to sediment particles. Estimates of benthic bacterivory with DTAF-stained sediment are compared with bacterivory on monodispersed FLB added to sediment. The effect of different relative FLB concentrations (from DTAF-stained sediment) on calculated ingestion rates is also examined and discussed with respect to a grazing model. MATERIALS AND METHODS

Preparation of stained sediment. Sediment collected from a littoral zone on the southern shore of Lake Gooimeer, near the town of Naarden in the central part of The Netherlands, was treated in the laboratory. Interstitial bacteria and bacteria attached to particles in the sediment were stained with DTAF (Sigma Chemical Co., St. Louis, Mo.) at a final concentration of 0.2 g liter-' and incubated at 60°C for 3 h. During the

* Corresponding author. Mailing address: Centre for Limnology, Netherlands Institute of Ecology, 3631 AC Nieuwersluis, The Netherlands. Phone: 31 2943 3599. Fax: 31 2943 2224.

2259

2260

APPL. ENVIRON. MICROBIOL.

STARINK ET AL.

incubation, the sediment was regularly stirred with a spoon. After incubation, eight 50-ml stainless steel centrifuge tubes were filled with 15 ml of stained sediment and washed with tap water in a continuous-flow angle rotor (Sorvall SS-34/KSB system) at 15,000 rpm (26,895 x g) at 5oC for 5 h in an SR-SC centrifuge (Sorvall Instruments, du Pont, Wilmington, Del.). The sediment was collected in a beaker and frozen at -20°C. A day before the actual grazing experiments, the sediment was thawed and put in a prerinsed wet dialyzing tube (Spectra/Por 4; molecular weight cutoff, 12,000 to 14,000). Then the sediment was dialyzed in a beaker overnight, with the diffusate flushed with a continuous flow of lake water. Preparation of FLB. A continuous culture inoculated with a mixed sample of sediment bacteria was set up to obtain sufficient bacteria for the preparation of FLB. Six liters of bacterial suspension from this continuous culture was concentrated by centrifugation with a continuous-flow angle rotor (Sorvall SS-34/KSB system) at 15,000 rpm (26,895 x g) and 50C for 2 h in a Sorvall SR-SC centrifuge. The bacterial pellets were resuspended in 10 ml of 0.05 M Na2HPO4 buffer (pH 9.0) (without NaCl) and stained with DTAF (Sigma) at a final concentration of 0.2 g liter-' for 2 h at 60°C as described by Sherr et al. (26). After three washes with the phosphate buffer, the cells were resuspended in 0.02 M Na4P207 (tetrasodium PPi) and frozen at -20°C. On the day of the experiment, the FLB were thawed and sonicated with a Soniprep 150 (Measuring & Scientific Equipment Ltd.), equipped with a miniprobe, at a low power level for a few 1-s bursts to disperse the bacteria. The remaining clumps were removed by filtration over a 3-,um-pore-size membrane filter (Nuclepore Corp., Pleasanton, Calif.). Comparison of methods. Intact littoral sediment cores were collected from Lake Gooimeer just outside the reed belt in June. After gently aspirating the overlying water, the upper 5 mm of sediment was collected in a beaker and gently homogenized with a spoon. Two subsamples of 15 ml were placed in a water bath and incubated at in situ lake temperature (18°C). One portion was gently mixed with 5 ml of DTAF-stained sediment. To the remaining portion, a 1.1-ml suspension containing 18 x 109 monodispersed FLB was added and the mixture was carefully mixed with a spoon. Both samples were incubated for 15 min. At regular intervals (starting at the beginning of the experiment) a subsample was taken and fixed with an equal volume of ice-cold 4% glutaraldehyde, resulting in a 2% final concentration, and put on ice (6). Two independent series were run in the field. A preliminary experiment was carried out in the laboratory, with the samples incubated for 30 min and subsampled every 5 min. The rate of ingestion of the total amount of bacteria was calculated by assuming that native bacteria were consumed in proportion to consumed FLB:

BACTOT protist-1 h- l = BACFLB

100 I= IFLB x% FLB

(1)

where IFLB is the uptake of FLB per protist per hour, BACNAT is the concentration of native bacteria per milliliter, BACFLB is

the concentration of FLB

per

milliliter, and BACTOT

is

the

total concentration of bacteria (BACNAT plus BACFLB). During the incubation period, the number of empty protists (protists not consuming FLB) was counted for both methods. The theoretically expected numbers of empty cells at a given

incubation period were calculated supposing a Poisson distribution of the ingested FLB per protist according to:

((x * t) + 1

y=

(3)

where a is the uptake of FLB per protist per min and t is the time in minutes. The regression parameter 13 represents the number of (seemingly) ingested FLB per protist at the onset of the incubation. This parameter may be >0 when some scored FLB were in fact not ingested by the protists and present in the food vacuole but on the outside of the cell. Combining equations 2 and 3 gives the expected percentage of empty cells for an uptake rate at as a function of time: Px=o

ln

t)

*

=

P=O =

+

13

(4)

- (t t) + a -

Grazing model. According to Fenchel (12), protozoan grazing rates can be described by Michaelis-Menten kinetics, which is functionally analogous to the predation model of Holling (18). To investigate the rate of uptake of FLB by protists for different bacterium/FLB ratios and affinities, we constructed a grazing model based on these Michaelis-Menten kinetics. The total concentration of bacteria (N = native bacteria + FLB) in our model was kept constant. If we assume that the maximum rate of uptake (Vmax) of bacteria and FLB by protists with the same size distribution is constant, then the rate of uptake of FLB (VFLB) as a function of the FLB concentration (x) is given by: Vmax *x

VFLB = K

K

F

.(N

-

x)

BAC

%FLB N 100 -

x

=

v VN-

VFLB Km

+ N*

%FLB

mFLB

100O

%FLB 100

Km FLB

+

(5)

/

Km BACmN-N

%FLB 100

When the affinities for FLB and native bacteria are equal, then = KmBAC and the equation can be simplified to:

KmFLB

VFLB =

[BACNAT + BACFLB IFLB L

(2) Px=o = e-Y where P,=O is the number of empty cells divided by the total number of cells at time t and y is the average uptake of FLB at time t. The mean uptake of FLB for an uptake rate a at time t is:

Vmax N * %FLB/100

Km FLB+N

(6)

Ratio series. Possible selectivity against FLB by protists was tested by comparing the rates of ingestion of FLB at different bacterium/FLB ratios and evaluated against the prediction by the grazing model. In August, a sample from the upper 5 mm of sediment was collected, from the same location as described above, in a beaker and gently homogenized. To obtain different bacterium/FLB ratios in the grazing assays, the freshly collected sediment was mixed in the field with different amounts of DTAF-stained sediment, resulting in concentration series of about 10, 25, 35, and 50%. The grazing assay samples were

VOL. 60, 1994

BENTHIC BACTERIVORY MEASURED WITH STAINED SEDIMENT

placed in a water bath at the in situ lake temperature (23°C). During the incubation, subsamples were taken at regular intervals and fixed with an equal volume of ice-cold 4% glutaraldehyde and put on ice. The experiment was carried out in triplicate in the field. The exact percentages of FLB in the DTAF-stained sediment grazing assays were determined for the subsamples collected at the start of the incubation period. Protozoan biovolumes and FLB uptake. From 0.5 ml of fixed sediment sample, the protists were extracted by an isopycnic centrifugation technique with nonlinear Percoll gradients (31). The extracted protists were stained with primulin (Janssen Chemica, Geel, Belgium) and examined on 1.0-,um-pore-size membrane filters (Nuclepore Corp., Pleasanton, Calif.) by epifluorescence microscopy (Zeiss Axiophot) at x 1,250 magnification (4). For biovolume estimates, at least 100 fixed organisms per sample were measured by eyepiece micrometer, and volumes were calculated from length and width, assuming a spherical or cylindrical shape. Equivalent spherical diameters were calculated from these volumes. The number of ingested FLB was determined at the same magnification with a Zeiss 487909 filter set (BP 450-490 excitation filter, FT 510 beam splitter, and LP 520 barrier filter). A total of 50 protists were examined per time point. Counting bacteria. The total amounts of bacteria and FLB in the remaining parts (3.5 ml) of the fixed sediment samples taken at the start (to) of the grazing experiments were determined. To disperse the attached bacteria from their attachment sites, the fixed sediment samples (3.5 ml) were diluted with 6 ml of Milli-Q water (deionized and filtered [0.2-,um pore size] tap water) and 0.5 ml of 0.2 M Na4P207 (tetrasodium PPi), resulting in a final concentration of 0.01 M Na4P207. The sediment suspensions were sonicated with a Soniprep 150 (Measuring & Scientific Equipment Ltd.) equipped with a miniprobe at a power level of 100 W for 2.5 min. Preliminary studies showed that this adjustment and sonication time resulted in maximum yields without destroying cells. After an appropriate number of decimal dilutions of the sonicated sediment suspension in a 0.05 M Trizma buffer (pH 4.0), the cells were stained with DAPI (4',6-diamidino-2-phenylindole) for at least 15 min in the dark (25). Diluting PPi-treated cells in Trizma instead of Milli-Q water resulted in bright blue cells. Yellow bacteria, difficult to distinguish from debris, were observed if cells were diluted only in Milli-Q water. The stained samples were collected on 0.2-,um-pore-size membrane filters (Nuclepore Corp.), and the bacteria were counted by epifluorescence microscopy at x 1,250 magnification on randomly selected fields until either 100 fields or 250 bacteria were observed. For biovolume estimates, 100 fixed organisms per sample were measured by eyepiece micrometer, and volumes were calculated from length and width, assuming a spherical or cylindrical shape. RESULTS Comparison of methods. An explorative experiment with benthic laboratory-cultured protists was carried out with DTAF-stained sediment and monodispersed FLB. In this experiment, for both methods the protists showed a linear uptake of ingested bacteria with time. The total ingestion rate measured with DTAF-stained sediment was twice as high as that calculated from the uptake of monodispersed FLB. The washing procedure turned out to be sufficient in removing all free DTAF, and all protists initially present in the stained sediment were destroyed. After DTAF-stained sediment was mixed with fresh sediment, only an insignificant amount (less than the counting error) of native bacteria became stained with

2261

15 0L

c1o z

--U-

;-

co w

5~

~

z

0

2

4

6

8

10

12

14 16 TIME [min]

FIG. 1. Total numbers of consumed bacteria per protist, calculated from the uptake of monodispersed FLB (open symbols) and from the uptake of FLB in stained sediment (closed symbols). Squares and circles represent duplicate experiments. Lines are regressions on the linear portion of the uptake curve. The uptake was corrected for consumption of FLB at to. DTAF, whereas 90.6% (±8.8%) of the total bacterial population in the DTAF-stained sediment was labeled with the

fluorochrome. After these promising results, a similar experiment was carried out (in duplicate) in the field and examined in more detail. The protist and bacterium concentrations in the field samples used for the grazing assays were 33.6 x 103 and 5.33 x 109 cells ml-' respectively. In the grazing assays with monodispersed FLB, 33.5% of the total bacterial population was labeled with DTAF. A comparable percentage of 30% labeled bacteria was found in the grazing assays with DTAFstained sediment. A linear uptake of bacteria by protists was observed for the first 10 min of the incubation with DTAF-stained sediment (r2 = 0.971) and for up to 15 min with monodispersed FLB (r2 = 0.930; Fig. 1). The rates of ingestion of bacteria calculated from the slopes of these linear regression lines were 61.3 ± 2.1 bacteria protist- h-1 for the DTAF-stained sediment method and a significantly (t test; P < 0.01) lower rate of 35.9 ± 1.9 bacteria protist-V h-' for the monodispersed FLB method. The equivalent spherical diameter size frequency distributions of the two protist populations involved in this field experiment 50 Sc. 40 0

ar. 30 0

oLL 20 z w

0 w

10

0-

0

r ^

7 =

44 =

-17

nn

n

7

rC

ESD [uml FIG. 2. Relative size distribution of protists in sediment mixed with DTAF-stained sediment (closed bars) and in sediment to which monodispersed FLB were added (open bars). ESD, equivalent spherical diameters (in micrometers).

2262

APPL. ENVIRON. MICROBIOL.

STARINK ET AL. r,

UEC-

93 4. w 0

-J

41.

-----

w

W 3.~

5 z

7

n(U a)z o

6

z

5

A 0 z

4

U

0

z

0 a:w

a-2.

z

.

L

*

U

3

8 0

2

4

6

8

10

12

14

0

16

10

20

30

40

50

60

PERCENTAGE FLB

TIME [min]

FIG. 3. Logarithmic (ln) decrease in empty cells during incubation with monodispersed FLB (open symbols) and with DTAF-stained sediment (closed symbols) and regression lines.

FIG. 5. Total bacterial densities plotted against FLB percentages measured in the grazing assays with different ratios of DTAF-stained and unstained sediment (-). The mean bacterial density of all series is indicated (- --).

(Fig. 2) were not significantly different at a 5% error level (Kolmogorov-Smirnov two-sample test; P = 0.07) and ranged from 2 to 29 p.m, with a mean of 10 pLm. The mean biovolumes for the FLB were 0.155 ± 0.165 pum3 for the monodispersed FLB and 0.151 ± 0.151 p.m3 for the FLB from the DTAFstained sediment; the two size frequency distributions were not significantly different (Kolmogorov-Smirnov two-sample test; P = 0.286). The decrease in empty protists during the incubation period was fitted with the nonlinear model equation 4 (see Materials and Methods). The calculated regression coefficients were as follows: ot = -0.106 + 0.009 and ,B = 3.906 ± 0.11 (r2 = 0.97) for protists incubated with monodispersed FLB and ao = -0.158 ± 0.007 and ,B = 4.495 + 0.056 (r2 = 0.99) for protists incubated with DTAF-stained sediment (Fig. 3). After incubation for 15 min with DTAF-stained sediment, the number of empty protists decreased by 81.2%, which is in agreement with the theoretically expected decrease of 81.5%. With the monodispersed FLB method, the decrease in empty cells after an incubation period of 15 min was 39.6%, which was significantly different (t test; P