Particle-associated flagellates: swimming patterns

AQUATIC MICROBIAL ECOLOGYAquat Microb Ecol

Vol. 35: 141–152, 2004 Published April 28

INTRODUCTION

Marine snow aggregates in the ocean are importantas exporters of organic material, but are also sites ofelevated microbial activity. Thus, a significant part ofthe microbial activity in the pelagic environment mayoccur on or in the vicinity of aggregates (Azam & Long2001). Bacteria may cluster in the chemosphere sur-rounding particles (Jackson 1989, Grossart et al. 2001,Barbara & Mitchell 2003) and aggregates (Kiørboe &Jackson 2001) and they may attach to aggregates(Kiørboe et al. 2002), yielding local concentrations that

are orders of magnitude higher than bulk ambient con-centrations (Alldredge & Silver 1984, Simon et al.2002). The bacteria solubilize and remineralize theaggregate, thus reducing vertical fluxes (Smith et al.1992, Ploug et al. 1999). Heterotrophic flagellates, themost important grazers on pelagic bacteria, also occurat elevated concentrations in and around marineaggregates (Caron et al. 1986, Artolozaga et al. 2000),where they form communities that may be differentfrom those in the ambient water (Caron 1991, Patter-son et al. 1993). Many species directly attach to parti-cles, either to feed on the abundant attached bacteria

© Inter-Research 2004 · www.int-res.com*Email: [email protected]

Particle-associated flagellates: swimming patterns,colonization rates, and grazing on attached bacteria

Thomas Kiørboe1,*, Hans-Peter Grossart2, Helle Ploug3, Kam Tang4, Brigitte Auer5

1Danish Institute for Fisheries Research, Kavalergården 6, 2920 Charlottenlund, Denmark2Institute of Freshwater Ecology and Inland Fisheries, Alte Fischerhütte 2, 16775 Neuglobsow, Germany

3Max Planck Institute for Marine Microbiology, Celsiusstraße 1, 28359 Bremen, Germany4Virginia Institute of Marine Science, Gloucester Point, PO Box 1346, Virginia 23062, USA

5Alfred Wegener Institute for Polar and Marine Research, Am Handelshafen 12, 27570 Bremerhaven, Germany

ABSTRACT: Some pelagic flagellates colonize particles, such as marine snow, where they graze onbacteria and thus impact the dynamics of the attached microbial communities. Particle colonization isgoverned by motility. Swimming patterns of 2 particle-associated flagellates, Bodo designis andSpumella sp., are very different, the former swimming slowly in an erratic, random pattern, and thelatter faster and along smooth helixes of variable amplitude and frequency. At spatial scales exceed-ing ca. 50 µm, the motility of B. designis can be described as a random walk and modeled as diffu-sion. Spumella sp. shows directional persistence of the helical axes up to a scale of at least about0.5 mm, and its motility cannot, thus, be characterized as a random walk at such small scales. Motil-ity analyses predicted overall rates at which the 2 flagellates encountered and colonized model par-ticles (4 mm agar spheres) rather well. After initial colonization, the number of flagellates remainedapproximately constant for ~10 h or more. In B. designis this was due to a density-dependent attach-ment probability, while in Spumella sp. the declining accumulation rate was better explained by aconstant specific detachment rate and a constant (low) attachment probability (12%). The grazingimpact of B. designis on attached bacteria was estimated from short-term (4 to 8 h) differences indevelopment of attached bacterial populations in the presence and absence of the flagellate. B.designis ingested up to 120 bacteria ind.–1 h–1 (ingestion rate increasing with increasing density ofbacteria on the sphere) and had surface area clearance rates of up to 1.3 × 10–4 cm2 h–1. At flagellatedensities typical of marine snow, the implied bacterial grazing mortality exceeds bacterial growthand colonization, suggesting that flagellate grazing controls abundances of attached bacteria.

KEY WORDS: Marine snow · Flagellates · Bodo designis · Spumella sp. · Bacteria grazing

Resale or republication not permitted without written consent of the publisher

Aquat Microb Ecol 35: 141–152, 2004

(Caron 1987, Artolozaga et al. 2002, Grossart et al.2003a, Kiørboe et al. 2003), or to exploit the locallyenhanced concentrations of free bacteria that arestrained more efficiently by attached than by free fla-gellates (Christensen-Dalsgaard & Fenchel 2003). Col-onization of bacterivorous flagellates may control theabundance of attached bacteria and, hence, impact therate of aggregate degradation (Kiørboe 2003). Flagel-late grazing may also have a more indirect effect onaggregate degradation, since grazing may lead tohigher bacterial growth rates and better substrate uti-lization due to a more efficient nutrient regeneration(Güde 1986).

Despite the significance of the dynamics of attachedmicrobial communities for vertical material fluxes, andthe potential role of bacterivorous flagellates for thesedynamics, little is known about flagellate particlecolonization rates (Kiørboe et al. 2003) and grazing onattached bacteria (Caron 1987, Artolozaga et al. 2002,Kiørboe et al. 2003). For 2 flagellates that are com-monly found attached to aggregates, we hereinexamine motility patterns and how motility leads toparticle encounter; we quantify particle colonizationrates, and we provide estimates of the grazing onattached bacteria by 1 of the species.

MATERIALS AND METHODS

We used 2 flagellates for our experiments, Bododesignis (8 to 10 × 4 to 6 µm) and Spumella sp. (ca. 10µm), that were isolated from sediments in the WaddenSea. Both are known to attach to surfaces. The formerfeeds on attached bacteria (Caron 1987), while the lat-ter feeds only on bacteria in free suspension. Culturesof flagellates were maintained on bacteria that weregrown on wheat grains. All experiments were con-ducted at room temperature.

Motility patterns of Bodo designis and Spumella sp.Suspensions of either B. designis or Spumella sp. wereadded to a circular observation chamber (∅ 15.9 mm,height 2.45 mm) that was glued with stopcock greaseonto a microscopic glass slide and closed by a cover-slip. Swimming patterns were video-recorded underdark-field illumination in an inverted microscopeusing a 6.4× objective, yielding a viewing field of ca.600 × 700 µm2. The recordings were stored directly inthe computer and 2-dimension swimming tracks weredigitized using LabView Software (DiMedia) and atime resolution of 0.04 s. About 80 tracks were ana-lyzed for each species. To characterize the motility pat-tern, the RMS (root mean square) net distance covered(L) was plotted as a function of time (t). If the motilitycan be described as a diffusion process, then, in 2dimensions (Berg 1993):

L = 2√__D √

__t (1)

where D is the equivalent diffusion coefficient. D wasestimated from regression analysis.

In other experiments, observation chambers either2.45 or 4.85 mm high were filled with a dense suspen-sion of Bodo designis (1 to 10 × 104 cells ml–1), andevery 0.5 to 10 min the number of flagellates that hadcontacted or attached to the microscopic slide, N(t),were enumerated within a 1 × 1 mm2 square (Chris-tensen-Dalsgaard & Fenchel 2003). This providesanother way of estimating an equivalent diffusioncoefficient of the swimming flagellates, since (derivedfrom Carslaw & Jaeger 1959):

(2)

where l is half the chamber height and C0 is the initialconcentration of cells in the chamber. Again, D wasestimated by fitting Eq. (2) to the observations byregression analysis.

Colonization of agar spheres. We used previouslydescribed methods to quantify particle colonizationrates (Kiørboe et al. 2003). Briefly, 4 mm diameter agarspheres were used as model particles. These werefixed on drawn-out Pasteur pipettes and exposed todense suspensions of flagellates (1 to 10 × 103 cellsml–1). The number of flagellates attached to thespheres was then followed over time by samplingduplicate or triplicate spheres at time intervals rangingfrom 15 min to several hours. Most incubations lasted 6to 12 h. All incubations were conducted in 2 l beakerswith up to 36 spheres. In some cases the spheres hadbeen pre-colonized by bacteria, obtained either fromcultures (Marinobacter Strain PCOB-2) or from raw seawater (from Gullmar Fjord, Sweden) that had beenstrained through a 1 µm filter. Even in experimentsusing non-colonized spheres, bacteria introducedtogether with the flagellates (stemming from the cul-tures) rapidly colonized the spheres. Fourteen incuba-tions involved Bodo designis and 5 incubationsSpumella sp. Flagellates that have colonized a particlemay detach again, and 3 experiments were designedto quantify detachment rates. Flagellates were allowedto colonize agar spheres overnight; these were thenmoved to either sterile-filtered or to 1 µm-strained sea-water and the number of attached flagellates was mon-itored over time.

The accumulation of flagellates on agar spheres maybe predicted from motility analysis, since the rate atwhich flagellates encounter and attach to a sphere isgiven by:

(3)

ddFt

F pt A = β

N t lCn

D n t l

n( ) =

+( )

+

=

∞( )∑ – – /

0 2 21

8 1

2 1

2 1 4

0

2 2 2

ππe

142

Kiørboe et al.: Particle-associated flagellates

where F is the number of flagellates attached to thesphere, FA is the ambient concentration of flagellates(ind. cm–3), p the probability of attachment followingencounter with the sphere, and βt the encounter ratekernel (cm3 s–1). If the motility of the flagellates canbe described as a random walk, then the encounterrate kernel for a sphere of radius a is (Kiørboe et al.2002):

(4)

where D is the equivalent diffusivity of the flagellates.If the motility is rather characterized as linear swim-ming, then:

(5)

where vt is the propagation velocity of the flagellate. Bodo designis grazing on attached bacteria. Graz-

ing rates were quantified by comparing bacterialdevelopment on agar spheres or agar-coated glassslides in the absence and presence of Bodo designis.Grazing on the surface of agar spheres was quantifiedusing stained bacteria. Bacteria were first allowed tocolonize agar spheres; the bacteria were either pre-stained with Sybrgold for 2 h, or entire spheres withattached bacteria were submerged in 1:10 000 dilutedSybrgold solution for 2 h. Half the spheres were thenmoved to a dense suspension of B. designis and theother half to sterile filtered seawater and the abun-dance of attached stained and unstained bacteria(visualized by DAPI) as well as attached B. designiswere monitored for 3 to 8 h on both sets of spheres.Grazing of stained bacteria was computed as thedifference in density of attached (stained) bacteria inthe absence and presence of B. designis. We did 4incubations of this type.

In the other approach, agar-coated slides were firstincubated for 0 to 48 h with raw seawater collectedfrom the Øresund (Denmark); during this time, bacte-ria colonized the slides. We then transferred 10 slidesto each of two 100 ml beakers containing a dense cul-ture of Bodo designis (and bacteria). To one of thebeakers, eukaryotic inhibitors were added (200 mg l–1

each of colchicine and cycloheximide; Hahn & Höfle1999). In the non-inhibited beaker, the flagellatesrapidly colonized the slides, and densities of between4000 and 6000 cells cm–2 were typically reached within1 h, remaining at that density for the remainder of the4 to 8 h incubation period. Slides were sampled onceevery 30 min to 1 h, and bacterial and flagellate densi-ties were assessed by epifluorescence microscopy afterDAPI staining.

Assuming a constant colonization rate of bacteriaand constant mortality of attached bacteria due toflagellate grazing, the change in bacterial density (B,cells cm–2) on the slides due to colonization (K, cellscm–2 h–1), specific growth and detachment (µ and δ,both in units of h–1), and grazing mortality (g, h–1) canbe described as (cf. Kiørboe et al. 2003):

(6)

Replacing d = µ – δ and integrating yields:

(7)

where B0 is the initial bacterial density. Colonization,growth, and detachment of attached bacteria wereassumed to be identical between parallel grazed andnon-grazed beakers, and Eq. (7) was fitted to observedbacterial densities as a function of time, allowing esti-mation of grazing mortality (least-squares method).Grazing mortality divided by flagellate density yieldsan estimate of flagellate clearance rate (cm2 h–1), andclearance multiplied by bacterial density an estimateof ingestion rate (nos. of bacteria eaten flagellate–1 h–1).We conducted 14 such grazing experiments.

The impact of flagellate grazing on attached bacteriawas finally assessed by incubating agar spheres in1 µm-strained seawater with and without Bodo desig-nis. Spheres were sampled 3 times d–1 over 3 d forenumeration of attached bacteria and flagellates.

RESULTS

Motility patterns

Bodo designis and Spumella sp. have very differentswimming patterns (Fig. 1). B. designis swims with ananterior flagellum and ‘wriggles’ through the wateralong an erratic track (Fig. 1A) at an average velocityof about 35 µm s–1. The real average velocity in 3dimensions can be estimated as 1.22 times the average2-dimensional velocity (Kiørboe et al. 2002). Spumellasp. also swims with an anterior flagellum that pulls thecell through the water, but along a smooth helical trackat about 100 µm s–1 (Fig. 1B). The amplitude and thefrequency of the helix vary substantially and inverselywith each other, from about 20 to 120 µm and 0.1 to1.3 s–1, respectively (Fig. 2).

The swimming patterns can be further characterizedby examining the net distance covered as a function oftime (Fig. 3). In Bodo designis, the RMS net distancecovered (L) initially increases linearly with time (t)(slope equal to swimming speed = 35 µm s–1) but, atscales exceeding 2 s and ca. 70 µm, the distance in-

B B

Kd g

td g t d g t –– –= +

+( ) ( )+ +( ) ( )0 1e e

ddBt

K B B gB K g B – – – –= + = +( )µ δ δ µ

β π π = ⇒ =a v F a v pF tt t t A2 2

β ππ

ππ

t

t A

Daa

Dt

F DaF pta

Dt

.

.

= +( )

⇒

= +( )

4 1

4 12

0 5

0 5

143


creases as a power function of time, with a power ofabout 0.5 (Fig. 3A,B). Thus, at these larger scales themotility pattern can be described as random and char-acterized by an equivalent diffusion coefficient. Eq. (1)was fitted to the data (t > 2 s), and D was estimated as0.7 × 10–5 cm2 s–1.

In Spumella sp. there also appear to be 2 phases inthe RMS distance covered as a function of time, butthese are both linear (i.e. they have slopes of ~1 in logplots); the initial slope corresponds to the averageswimming velocity (vi), while the latter slope corre-sponds to the tangential swimming velocity (vt), i.e. thevelocity along the helical axis (Fig. 3C,D). For a perfecthelical pattern the 2 are related as (Svenssen & Kiør-boe 2000):

(8)

where f and A are the frequency and amplitude,respectively, of the helical pattern. fA can be estimatedfrom Fig. 2 as ca. 16.7 µm s–1, and νt as 16.5 µm s–1 in 2dimensions (20.1 µm s–1 in 3 dimensions) from Fig. 3C.Eq. (8) predicts a swimming velocity of 107 µm s–1 in 3dimensions, corresponding to 88 µm s–1 in 2 dimen-sions. Thus, at spatial scales up to at least 400 µm (dueto the limited field of view, we could not extend the

analysis to larger scales) Spumella sp. propagatesessentially linearly with an average propagation veloc-ity of 20 µm s–1.

The equivalent diffusivity of Bodo designis was alsoestimated from the rate at which the cells attach to amicroscope slide (Fig. 4). Depending on the size of thechamber most cells had attached to a surface within 15to 60 min. Not all flagellates attached immediatelyupon contact with the glass slide; initially, most movedaround on the surface of the slide until eventually at-taching with the posterior flagellum. In 10 trials the av-erage diffusivity estimated by fitting Eq. (2) to the datawas 0.8 ± 0.4 × 10–5 cm2 s–1 (range: 0.5 to 1.9 cm2 s–1).

Particle colonization

The Bodo designis colonization pattern was charac-terized in all cases by an initial increase in the numberof attached flagellates, followed by a period in whichthe number of attached flagellates remained more orless constant (Fig. 5A). There were no consistent dif-ferences in colonization pattern between spheres withand without bacteria, nor did the pattern vary consis-tently with the type of bacteria that had pre-colonized.Using the estimated diffusivity of 0.8 × 10–5 cm2 s–1,Eq. (4) predicts the initial accumulation of flagellateswell (assuming an attachment probability of p = 1). Theinsignificant detachment of B. designis cells from theagar spheres (Fig. 5B) and the independence of thedensity of attached flagellates from ambient concen-trations (Fig. 5C) suggest that the attachment probabil-ity declines with increasing density of attached flagel-lates. Apparently, B. designis’ attachment probabilityapproaches zero as its density approaches 2000 cellsper sphere (= 4000 cells cm–2). In 1 incubation lasting3 d, flagellate density was constant for 40 h subsequent

v v fAi t = +( )2 22π

144

Fig. 1. (A) Bodo designis and (B) Spumella sp. swimmingtracks. Positions are plotted at 0.04 s intervals

Fig. 2. Spumella sp. Relation between amplitude (A, µm) andfrequency (f, s–1) of swimming helixes. Hyperbolic relation is

fitted to the data: A = 17.0/(0.054 + f ), R2 = 0.99


to initial colonization, and then increased almost 10-fold during the subsequent 40 h.

The overall colonization pattern of Spumella sp.resembled that of Bodo designis (Fig. 6A). The initialincrease was linear over time as predicted by Eq. (5),and the slope is an estimate of the colonization rate.Initial colonization rates taken from all 5 experimentswere proportional to the ambient concentration offlagellates (Fig. 6B); the proportionality constant is anestimate of the encounter rate kernel × the attachmentprobability, β × p = 0.11 ml h–1 = 3 × 10–5 ml s–1. β canalso be estimated from motility analysis as β = πa2vt =2.5 × 10–4 ml s–1. Thus, attachment probability can beestimated as p = 3/25 = 0.12 = 12%. As for B. designis,cell accumulation leveled off subsequent to initial col-onization. However, in contrast to B. designis,Spumella sp. detaches rapidly from the particles(Fig. 6C, specific detachment rate = 0.4 h–1). Assuming

a constant detachment rate, δ, Eq. (5) modifies to(Kiørboe et al. 2002):

(9)

Eq. (9) provides a good fit to the data (Fig. 6A), and theestimated detachment rate 0.68 h–1 (range 0.66 to 0.70h–1) was within a factor of 2 of that estimated directly.

Bodo designis grazing rates on attached bacteria

After attachment to a surface by the tip of the poste-rior flagellum, Bodo designis stretches by up to 25 µmfrom its attachment point in all directions and collectsbacteria within this area by means of its anterior fla-gellum. Even when attached, however, the cell movesforward slowly — gliding at the tip of the flagellum —

F

pFt

A t – –= ( )βδ

δ1 e

145

Fig. 3. Bodo designis andSpumella sp. Motility analyses:RMS net distance covered as afunction of time. B. designis:data plotted on (A) linear and(B) log scales; Spumella sp.:data plotted on (C) linear and(D) log scales. Lines are re-gressions fitted to data shownby open symbols (dashed lines)or filled symbols (continuouslines). Grey symbols not

included in regressions

Fig. 4. Bodo designis. Examplesof flagellates colonizing glassslides, showing density of fla-gellates on glass bottom of (A)small and (B) large chambers—measuring 2.45 or 4.85 mm inheight—as a function of incu-bation time. Eq. (4) has been fit-ted to the data. Initial chamberconcentration of cells (C0) andestimated D-values are given


at an average velocity of 2.7 (±1.3) µm s–1, thus collect-ing bacteria within an approximate area per unit timeof 50 × 2.7 µm2 s–1 ≈ 5 × 10–3 cm2 h–1.

The presence of flagellates had in all cases a signifi-cant negative effect on the number of attached bacte-ria, both on the spheres and on the slides (Figs. 7 & 8).On the spheres, the cumulative number of bacteria

grazed (difference in number of attached stained bac-teria between spheres incubated in sterile filtered sea-water and in the presence of Bodo designis) wasapproximately linearly related to the cumulative num-ber of ‘flagellate-hours’ (= cumulated abundance of fla-gellates × time) (Fig. 7), and specific ingestion rateswere estimated as the slopes of these plots.

146

Fig. 5. Bodo designis. (A) Example of initial colonization ofagar sphere. The line is the predicted colonization (Eq. 4)assuming D = 0.8 × 10–5 cm2 s–1. (B) Detachment of B. designisfrom agar spheres. Data from 3 experiments have beennormalized by the initial abundance of cells per sphere andplotted together (each experiment has a different symbol).The line is the regression of the pooled data; the regression isnot significant. (C) Average abundance of attached flagellatesin 15 experiments following initial colonization as a functionof ambient concentration of flagellates. The line is the regres-

sion, which is not significant. Error bars are ±1 SE

Fig. 6. Spumella sp. (A) Example of colonization of agar sphere.Continuous line is a fit of Eq. (5) utilizing only the initial datapoints, while the dashed line is a fit of Eq. (9) utilizing all data;(B) colonization rate versus ambient concentration of flagellates.The fitted linear regression has an estimated slope of 0.11 ±0.03 ml h–1 (R2 = 0.81). (C) Detachment of Spumella sp. fromagar spheres. Data normalized by the initial abundance of cellsper sphere. The regression line (log cells per sphere vs time)is significant and yields an estimated detachment rate of

0.38 ± 0.12 h–1 (R2 = 0.73). Error bars are ±1 SE


In the experiments with unstained bacteria andinhibited flagellates, 3 different patterns were seen,depending on bacterial density and growth, and on fla-gellate grazing rates: (1) density of attached bacteriadecreased in grazing beakers and increased in non-grazed controls (Fig. 8A); (2) density of attached bacte-ria increased on both grazed and non-grazed slides butreached near steady-state, at least in the grazingbeaker (Fig. 8B); (3) bacterial density increased near-exponentially on both types of slides (Fig. 8C). WhileEq. (7) provided good fits to all 3 types of experimentaldata, parameter estimates were strongly intercorre-lated in the latter case, rendering such estimatesmeaningless; these estimates were disregarded.

Combining the various grazing estimates yields thefunctional response in ingestion and clearance rates tovariation in density of attached bacteria (Fig. 9). Inges-tion rates increased with bacterial density and rangedup to 120 bacteria flagellate–1 h–1. The functionalresponse was described by fitting Holling’s disk equa-

tion (Holling 1959) to the data (Fig. 9). From the fit, a‘handling time’ of ca. 30 s and a maximum surfaceclearance rate of 1.3 × 10–4 cm2 h–1 was estimated.

The impact of flagellate grazing on attached bacte-ria was finally evaluated by comparing the populationdevelopment of attached bacteria in the absence andpresence of Bodo designis (Fig. 10). Throughout (ca.80 h), bacterial density on the spheres was higher inthe absence than in the presence of flagellates.Bacteria also appeared to be much more aggregatedon the surface of spheres in the presence of flagellates(Fig. 11).

147

Fig. 7. Example of Bodo designis grazing on bacteria attachedto agar sphere. (A) Evolution of abundance of stained bacteriaon spheres in absence and presence of flagellates and ofnumber of ‘flagellate hours’. (B) Estimated number of grazedbactera versus ‘flagellate hours’; slope of the regression is anestimate of the grazing rate: 97 ± 10 bacteria flagellate–1 h–1

(R2 = 0.96). Error bars in (A) are ±1 SE

Fig. 8. Examples of Bodo designis grazing on bacteria attachedto agar-coated glass slides: evolution of number of attachedbacteria with (d) and without (s) addition of eukaryotic in-hibitors. Examples shown are 1 of each of the 3 patterns ob-served. Lines are fits of Eq. (7) to the data. Error bars are ±1 SE


DISCUSSION

Colonization

Marine snow aggregates are inhabited by densemicrobial communities. Flagellates, for example,occur in abundances of 10 to 10 000 per aggregate,increasing with increasing aggregate size (see datacompilation by Kiørboe 2001). Particles suspended innatural seawater in the laboratory will be colonizedby flagellates, but typically these occur in detectableabundances only after 1 to several days of incubation(e.g. Stuart et al. 1981, Pomeroy et al. 1984, Biddanda& Pomeroy 1988). In contrast, the flagellate coloniza-tion rates reported here suggest that high densitieswould be reached within a few hours. However, nei-ther of these laboratory rates are immediately applica-ble to field situations, owing to differences in ambientflagellate densities and in the flow environment.Microorganisms, including flagellates, colonize parti-cles mainly because the organisms are motile. Even arapidly sinking aggregate will scavenge only a fewsuspended immobile organisms, because the aggre-gate pushes the water away as it sinks. However,advection may facilitate encounters between sinkingaggregates and microorganisms that have motilitiesthat can be described as a diffusion process. Theenhancement can be estimated from the Sherwoodnumber and may be up to a factor of 20 for the largestand fastest sinking aggregates and for organisms withdiffusivities similar to those estimated here for Bododesignis (Kiørboe et al. 2001). Many flagellates havechemosensory capabilities that may allow them toaggregate at food patches (Fenchel & Blackburn1999) and further enhance the rate at which theycolonize marine snow aggregates. The effect ofchemosensory behavior is, however, somewhat

limited in a sinking aggregate because advectionwashes any signals away (Kiørboe & Jackson 2001).Characteristic concentrations of flagellates in theeuphotic zone are on the order of 103 cells ml–1. If theyhave motilities similar to those observed here for B.designis, then a 0.5 cm radius sinking aggregate will

148

Fig. 9. Bodo designis. Functional response in ingestion rate and surface area clearance rate to variation in density of attachedbacteria. Lines are fits of Holling’s disk equation to the data. For ingestion rate (I, bacteria flagellate–1 h–1) the relation is I = a ×B/(1 – a × τ × B), where a is the instantaneous rate of prey discovery (cm2 flagellate–1 h–1), τ is the prey handling time (h), and B isthe density of bacteria (bacteria cm–2). The fitted parameters are a = 1.3 ± 0.5 × 10-4 cm2 flagellate–1 h–1, and τ = 0.0081 ± 0.0017 h

(R2 = 0.74). Error bars are ±1 SE

Fig. 10. Long-term effect of Bodo designis grazing on bac-terial populations. (A) Number of attached bacteria in thepresence and absence of flagellates as a function of time.(B) Number of attached flagellates as a function of time. Error

bars are ±1 SE

Kiørboe et al.: Particle-associated flagellates 149

Without Bodo With Bodo

Fig. 11. Bacteria on the surface of spheres in the absence (left) and presence (right) of Bodo designis after 7.5, 37, 45 and 69 hincubation


encounter about 2000 flagellates h–1 (Eq. 5, assuminga Sherwood number of 10 estimated from Kiørboe etal. 2001). Thus, typical abundances of attached flagel-lates may be reached quickly, and the abundance ofattached flagellates is probably rarely limited byencounter rates. Rather, abundances of attachedflagellates may be governed by attachment anddetachment probabilities, e.g. mediated throughgrazing possibilities on the aggregate.

Grazing

The surface clearance rate is the imaginary surfacearea cleared of bacteria per flagellate per unit time.The estimated area potentially covered by a Bododesignis cell gliding on a surface is 10 to 100 times theestimated surface clearance rate, suggesting that B.designis collects or ingests only a small fraction of thebacteria encountered.

The bacterial ingestion rates estimated here, 5 to120 bacteria flagellate h–1, are similar to or somewhathigher than the few earlier reports on flagellategrazing rates on attached bacteria. The only previousestimates of flagellate grazing on bacteria attached tomarine snow are 1 to 10 bacteria flagellate h–1 for Bododesignis (Artolozaga et al. 2002) and 15 bacteria flagel-late–1 h–1 for an assemblage of small (~5 µm) flagellates(Kiørboe et al. 2003). Fenchel (1986) reported a func-tional response for Bodo sp. that was very similar tothat found here. Other reports on flagellate grazing onsediment-attached bacteria range up to 75 bacteriaflagellate–1 h–1 (Starink et al. 1994, 1996, Epstein 1997,Hammels et al. 2001). Because B. designis has a bio-volume about 200 times the biovolume of its bacterialprey (1 to 2 µm length), and assuming a growth yield of0.4, the present maximum grazing estimates wouldpredict a maximum flagellate growth rate of about0.2 h–1. This is similar to that found in free-living,bacterivorous pelagic flagellates (Fenchel 1982).

Dynamics of attached microbes

After initial colonization, the accumulation of bothBodo designis and Spumella sp. leveled off whenabundances had reached about 2000 cells per sphere(= 4000 cells cm–2), and remained at that density for upto 40 h (Fig. 10). A similar stationary density wasreached rapidly on the agar-coated glass slidesexposed to culture concentrations of B. designis. Suchflagellate densities are typical also of marine snowaggregates (see data compilation by Kiørboe 2001).The mechanisms controlling the abundance ofattached flagellates during this ‘stationary phase’

appeared to be different between the 2 species andto be caused by, respectively, a density-dependentattachment probability and a constant specific detach-ment rate. B. designis grazes on attached bacteria, anddensity-dependent attachment may be interpreted inthe context of possible intraspecific competition. Themaximum surface clearance rate of 10–4 cm2 flagel-late–1 h–1 at 4000 flagellates cm–2 implies a specificmortality rate of attached bacteria of 0.4 h–1 (= 10 d–1).This is higher than typical specific growth rates of bac-teria attached to natural marine snow aggregates,which are rather on the order of 2 d–1 (Grossart & Ploug2001, Grossart et al. 2003a). However, bacteria colo-nization also occurs from the ambient water. Withtypical bacterial motilities (D = 10–5 cm2 s–1, e.g. Berg1993) and ambient concentrations (106 ml–1), bacteriawould arrive at a rate of 105 cells h–1 at the surface of a4 mm diameter sphere, accounting for about half of thebacteria potentially ingested by the residing flagel-lates. Thus, bacterial grazing would approximately bebalanced by bacterial growth and colonization undernatural conditions. Higher grazing rates would lead todepletion of attached bacteria, or to a change in thebacterial community towards grazer-resistant forms, asseen in pelagic communities (Posch et al. 1999, Jür-gens & Matz 2002). Aggregation of bacterial cells onthe sphere surface (Fig. 11) might similarly serve asgrazer protection, because specific grazing rates wouldbecome saturated within patches. In the present exper-iments, protozoan grazing was unable to check thepopulation of attached bacteria (Fig. 10), probablyowing to their very high growth rate (in excess of10 d–1; authors’ own unpublished estimate based onthymidine incorporation). However, observed bacterialabundances on natural aggregates are on the order of106 bacteria per aggregate rather than the 108 bacteriafound here on the model aggregates (Fig. 10), and theabove considerations thus suggest that populations ofbacteria attached to natural aggregates would bechecked by flagellate grazing.

The implications of bacterial grazing go beyond pop-ulation dynamics, since it may also lead to changes inregeneration rates (Pomeroy et al. 1984, Rothhaupt &Güde 1992). For example, Sala & Güde (1999) foundthat heterotrophic nanoflagellates play an importantrole during the degradation of macrophytes becausethey grazed mainly on free bacteria and left theattached bacteria to grow. This promoted a shift fromthe hydrolysis of non-structural to structural polysac-charides. Hence, zooplankton and especially micro-zooplankton grazing may lead to pronounced changesin rates of bacterial uptake and remineralization oforganic matter.

The dynamics of the microbial communities thatdevelop on marine snow aggregates are governed by a

150


multitude of processes, including colonization, grazinginteractions, other inter- and intra-specific interactions,and growth (Azam & Long 2001, Long & Azam 2001,Gram et al. 2002, Grossart et al. 2003b, Kiørboe et al.2003). Attempts to model microbial dynamics with thelonger-term aim of describing material fluxes and ver-tical transport in the upper ocean are constrained byour limited knowledge of the component processes(Kiørboe 2003). The present study demonstrates thatflagellates may rapidly colonize sinking aggregates,and that their grazing on attached bacteria may havesignificant influence on the attached microbial com-munity and, therefore, on the fate of the aggregate.Grazer-induced changes in bacterial numbers andcommunity composition may cause changes in nutrientregeneration and organic matter remineralization.Thus, while the activity of the attached microbial pop-ulations is ultimately limited by their abundances anddynamics — which we can now describe — the exacteffects of bacterial grazing on aggregate degradationremain to be revealed.

Acknowledgements. We are grateful to Dr. Uffe H. Thygesenfor writing MatLab codes for the regression analyses. Thiswork was supported by grants from the Danish NaturalScience Research Council to T.K. (21-01-054), from theLeibniz Society to H.P.G., and by EU-funded Access to theResearch Infrastructure Kristineberg Marine Research Station(ARI P.45).

LITERATURE CITED

Alldredge AL, Silver MW (1984) Characteristics, dynamicsand significance of marine snow. Prog Oceanogr 20:41–82

Artolozaga I, Ayo B, Latatu A, Azíua I, Unanue M, Iberri J(2000) Spatial distribution of protists in the presence ofmacroaggregates in marine systems. FEMS Microbiol Ecol33:191–196

Artolozaga I, Valcárel M, Ayo B, Latatu A, Iberri (2002) Graz-ing rates of bacterivorous protists inhabiting diverseplanktonic environments. Limnol Oceanogr 47:142–150

Azam F, Long RA (2001) Oceanography — sea snow micro-cosms. Nature 414:495–496

Barbara GM, Mitchell JG (2003) Marine bacterial organisa-tion around point-like sources of amino acids. FEMSMicrobiol Ecol 43:99–109

Berg HC (1993) Random walks in biology. New, expandededition. Princeton University Press, Princeton, NJ

Biddanda BA, Pomeroy LR (1988) Microbial aggregation anddegradation of phytoplankton-derived detritus in sea-water. I. Microbial succession. Mar Ecol Prog Ser 42:79–88

Caron DA (1987) Grazing of attached bacteria by hetero-trophic microflagellates. Microb Ecol 13:203–218

Caron DA (1991) Heterotrophic flagellates associated withsedimenting detritus. In: Patterson DJ, Larsen J (eds) Thebiology of free-living heterotrophic flagellates, SpecVol 45. The Systematics Association, Clarendon Press,Oxford, p 77–92

Caron DA, Davis PG, Madin LP, Sieburth JMcN (1986)Enrichment of microbial populations in macroaggregates

(marine snow) from surface waters of the North Atlantic.J Mar Res 44:543–565

Carslaw HS, Jaeger JC (1959) Conduction of heat in solids,2nd edn. Clarendon Press, Oxford

Christensen-Dalsgaard KK, Fenchel T (2003) Increased filtra-tion efficiency of attached compared to free-swimmingflagellates. Aquat Microb Ecol 33:77–86

Epstein SS (1997) Microbial food webs in marine sediments. I.Trophic interactions and grazing rates in two tidal flatcommunities. Microb Ecol 34:188–198

Fenchel T (1982) Ecology of heterotrophic microflagellates. II.Bioenergetics and growth. Mar Ecol Prog Ser 8:225–231

Fenchel T (1986) The ecology of heterotrophic microflagel-lates. Adv Microb Ecol 9:57–97

Fenchel T, Blackburn N (1999) Motile chemosensory behav-iour of phagotrophic protests: mechanisms for and effi-ciency in congregating at food patches. Protist 150:325–336

Gram L, Grossart HP, Schlingloff A, Kiørboe T (2002) Possiblequorum sensing in marine snow bacteria: production ofacylated homoserine lactones by Roseobacter strainsisolated from marine snow. Appl Environ Microbiol 68:4111–4115

Grossart HP, Ploug H (2001) Microbial degradation of organiccarbon and nitrogen on diatom aggregates. LimnolOceanogr 46:267–277

Grossart HP, Riemann L, Azam F (2001) Bacterial motility inthe sea and its biogeochemical implications. Aquat MicrobEcol 25:247–258

Grossart HP, Hietanen S, Ploug H (2003a) Microbial dynamicson diatom aggregates in Øresund, Denmark. Mar EcolProg Ser 249:69–78

Grossart HP, Kiørboe T, Tang K, Ploug H (2003b) Bacterialcolonization of particles: growth and interactions. ApplEnviron Microbiol 69:3500–3509

Güde H (1986) Loss process influencing growth of planktonicbacterial populations in Lake Constance. J Plankton Res 8:795–810

Hahn MW, Höfle MG (1999) Flagellate predation on a bacte-rial model community: interplay of size-selective grazing,specific bacterial cell size, and bacterial community com-position. Appl Environ Microbiol 65:4863–4872

Hammels I, Muylaert K, Casteleyn G, Vyverman W (2001)Uncoupling of bacterial production and flagellate grazingon aquatic sediments: a case study from an intertidal flat.Aquat Microb Ecol 25:31–42

Holling CS (1959) Some characteristics of simple types ofpredation and parasitism. Can Entomol 91:385–398

Jackson GA (1989) Simulation of bacterial attraction andadhesion to falling particles in an aquatic environment.Limnol Oceanogr 34:514–530

Jürgens K, Matz C (2002) Predation as a shaping force for thephenotypic and genotypic composition of planktonic bac-teria. Antonie Leeuwenhoek 81:413–434

Kiørboe T (2001) Formation and fate of marine snow: small-scale processes with large-scale implications. Sci Mar65(Suppl):57–71

Kiørboe T (2003) Marine snow microbial communities: scalingof abundances with aggregate size. Aquat Microb Ecol 33:67–75

Kiørboe T, Jackson GA (2001) Marine snow, organic soluteplumes, and optimal sensory behaviour of bacteria. LimnolOceanogr 46:1309–1318

Kiørboe T, Ploug H, Thygesen UH (2001) Fluid motion andsolute distribution around sinking aggregates. I. Smallscale fluxes and heterogeneity of nutrients in the pelagicenvironment. Mar Ecol Prog Ser 211:1–13

151


Kiørboe T, Grossart HP, Ploug H, Tang K (2002) Mechanismsand rates of bacterial colonization of sinking aggregates.Appl Environ Microbiol 68:3996–4006

Kiørboe T, Tang K, Grossart HP, Ploug H (2003) Dynamics ofmicrobial communities on marine snow aggregates: colo-nization, growth, detachment, and grazing mortality ofattached bacteria. Appl Environ Microbiol 69:3036–3047

Long R, Azam F (2001) Antagonistic interactions amongmarine pelagic bacteria. Appl Environ Microbiol 67:4975–4983

Patterson DJ, Nygaard K, Steinberg G, Turley CM (1993) Het-erotrophic flagellates and other protists associated withoceanic detritus throughout the water column in the midNorth Atlantic. J Mar Biol Assoc UK 73:67–95

Ploug H, Grossart HP, Azam F, Jørgensen BB (1999) Photo-synthesis, respiration, and carbon turnover in sinkingmarine snow from surface waters of Southern CaliforniaBight: implications for the carbon cycle in the ocean. MarEcol Prog Ser 179:1–11

Pomeroy LR, Hanson RB, McGillivary PA, Sherr BF, KirchmanD, Deibel D (1984) Microbiology and chemistry of fecalproducts of pelagic tunicates: rates and fates. Bull Mar Sci35:426–439

Posch T, 2imek K, Vrba J, Pernthaler J, Nedoma J, Sattler B,Sonntag B, Psenner R (1999) Predator-induced changes ofbacterial size-structure and productivity studied on anexperimental microbial community. Aquat Microb Ecol 18:235–246

Rothhaupt KO, Güde H (1992) The influence of spatial andtemporal concentration gradients on phosphate partition-ing between different size fractions of plankton — furtherevidence and possible causes. Limnol Oceanogr 37:739–749

Sala MM, Güde H (1999) Role of protozoans on the microbialectoenzymatic activity during the degradation of macro-phytes. Aquat Microb Ecol 20:75–82

Simon M, Grossart HP, Schweitzer B, Ploug H (2002) Micro-bial ecology of organic aggregates in aquatic systems.Aquat Microb Ecol 28:175–211

Smith DC, Simon M, Alldredge AL, Azam F (1992) Intensivehydrolytic activity on marine aggregates and implicationsfor rapid particle dissolution. Nature 359:139–141

Starink M, Krylova IN, Bär-Gilisen MJ, Bak RP, CappenbergTE (1994) Rates of benthic protozoan grazing on free andattached sediment bacteria measured with fluorescentlystained sediment. Appl Environ Microbiol 60:2259–2264

Starink M, Bär-Gilisen MJ, Bak RPM, Cappenberg TE (1996)Bacterivory by heterotrophic nanoflagellates and bacterialproduction in sediments of a freshwater littoral system.Limnol Oceanogr 41:62–69

Stuart V, Lucas MI, Newell RC (1981) Heterotrophic utilisa-tion of particulate matter from the kelp Laminaria pallida.Mar Ecol Prog Ser 4:337–348

Svensen C, Kiørboe T (2000) Remote prey detection inOithona similis: hydromechanical versus chemical cues.J Plankton Res 22:1155–1166

152

Editorial responsibility: John Dolan,Villefranche-sur-Mer, France

Submitted: December 8, 2003; Accepted: January 30, 2004Proofs received from author(s): April 16, 2004

Documents

Particle-associated flagellates: swimming patterns