Introduction

When the temperature of water drops from 22 to 5°C its viscosity increases by 70%, thus the change in temperature alters not only the metabolic rate of organisms but also changes the physical characteristics of the ambient fluid. Such effects of temperature are potentially confounding at hydrodynamic scales where viscous forces dominate motion. It is the dimensionless Reynolds number (Re) that quantifies the relative importance of the two forces that matter during swimming: inertial forces and viscous forces. The Re is “the nearest thing we have to a completely general guide to what’s likely to happen when solid and fluid move with respect to each other” (Vogel 1994). When Re <1, viscous forces dominate the flow. Re is the product of three variables: the velocity of flow (U), the kinematic viscosity of the fluid (v) and the so-called characteristic length (l), which, in the case of swimming objects usually means the length measured in the direction of movement. Numerically Re is expressed as:

$$ \operatorname{Re} = lU\nu ^{{ - 1}} $$

In this study, the feeding of the major freshwater herbivores, cladoceran zooplankton, was investigated from a biomechanical viewpoint. Suspension feeding of cladocerans is characterised by Re~10−1–10−4 (Porter et al. 1983). Therefore, it was expected that the efficiency of food collection (and in consequence, growth) of freshwater cladocerans (Daphnia sp.) would be strongly influenced by viscosity changes of the water.

The aim of this study is to test two hypotheses:

Hypothesis 1

First it was tested if, and if so to what extent water viscosity is responsible for the decline in the growth of Daphnia species observed with a decrease in temperature.

It is not just the viscosity that influences Re, but also the size of the animals (that is proportional to l). The size of herbivorous cladocerans has a tremendous impact on the effectiveness of filtration and their vulnerability to the interference of bloom-forming filamentous blue-greens. Filamentous algae, which often dominate in meso- and eutrophic lakes, can cause a considerable reduction in growth but they primarily inhibit large species. To this day the physiological causes of the influence of body size on the fitness of cladocerans are not known.

In laboratory experiments (Hawkins and Lampert 1989; Gliwicz and Lampert 1990; DeMott et al. 2001) interference was demonstrated many times; the long filaments are hypothesised to interfere with the feeding process, causing dramatic reductions in growth rate. The inhibition of growth is particularly strong in large daphnids (D. pulicaria), while small species (D. galeata, D. cucullata) are not, or only weakly influenced. At low food concentrations the addition of the non-toxic filamentous Aphanizomenon flexosum even improves growth of D. galeata (Kurmayer 2001). Von Elert et al. (2003) observed that the growth rate of D. galeata on the filamentous blue-green Anabaena variabilis was nearly equal to its growth rate on Scenedesmus obliquus, when A. variabilis was enriched with sterols. Field studies (Threlkeld 1986) have shown that at low temperatures the inhibitory effect of blue-green algae on growth and fecundity of D. pulicaria decreases dramatically; however, in these studies the direct cause of inhibition was not determined. Other studies show much less clear pattern. Filamentous species are often associated with large-bodied Daphnia (Lynch 1980), and there are well-documented cases of D. pulicaria feeding on filamentous cyanobacteria (Epp 1996; Paterson et al. 2002). The feeding experiments of Lampert (1981) and Kirk and Gilbert (1992) have shown that it is mainly the toxicity of blue-greens that is responsible for the reduction of fitness in Daphnia.

Hypothesis 2

The second hypothesis that was tested is that the body size and temperature dependence on cyanobacterial interference is caused basically by the same physical mechanism, the change in the Reynolds number (Re) of the feeding apparatus of Daphnia.

If one considers two objects with identical shape, then equality of Res means that the physical character of the flow (streamlines) will be the same, regardless of the actual values of l, U or v. This law of dynamic scaling was used in this study, to “virtually” reduce the size of large daphniids by increasing the viscosity of water. This reduces the Re of the filter screens just as the change in body size would. It was expected that large Daphnia species in water of higher viscosity (especially D. pulicaria, see Materials and methods and Fig. 1) will behave qualitatively like smaller Daphnia, and will not be sensitive to the interference of filaments of cyanobacteria (Fig. 1).

Fig. 1
figure 1

a The change in the viscosity of water with temperature change, and the modification of viscosity with dextran. b Approximate Reynolds numbers (Re) of the setulae of the Daphnia species used. Re were calculated using morphological data from Brendelberger (1988). For all species the diameters of setulae of D. pulicaria were used as characteristic length. Flow velocity (U) was estimated as clearance rate/filtering area, thus the calculation ignores the fact that the flow is not steady but periodic. For D. pulicaria the Re for 12°C was also calculated. It was assumed that the 30% increase in viscosity that occurs when the temperature changes from 22°C to 12°C reduces U by 25%. c The mesh sizes of the filtering combs of the four species used (calculated after Brendelberger 1988)

Full size image

Materials and methods

Daphnia

Clones of four Daphnia species were used in the experiments: D. magna, D. pulicaria, (obtained from the Max-Planck Institute for Limnology, Plön), D. hyalina and D. galeata, (obtained from the Limnological Institute of the University of Konstanz, Konstanz). Prior to the experiments, the clones had been maintained in the laboratory for several months, fed with the green alga S. obliquus, and kept in constant, dim light at room temperature (20±2°C). The culture medium was membrane-filtered (0.45 μm) lake water from a hypertrophic lake in Warsaw.

Algae

The green alga S. obliquus (clone SAG 276-3a) and the filamentous blue-green Cylindrospermopsis raciborskii (originating from Lake Balaton, Hungary) were used in the experiments. Both species were grown in xenic (containing bacteria) batch cultures on Z/4 medium, with continuous illumination from white fluorescent lamps. Scenedesmus was grown at room temperature (18–22°C), while Cylindrospermopsis at 30±2°C. Most Scenedesmus in the culture was present in unicellular form, while Cylindrospermopsis forms short, straight trichomes of approximately 100 μm length, that are not coiled and do not form flakes. Cylindrospermopsis is a bloom-forming subtropical species, also found in southern Europe, which can produce the toxin cylindrospermopsin (Hawkins et al. 2001). There were two reasons for choosing this cyanobacterium as the interfering species: due to its straight shape it does not accumulate in the flow-through systems used in the experiments (see below), and secondly, previous studies have demonstrated that filtering rates (Hawkins and Lampert 1989), growth and life history parameters of Daphnia (Gliwicz and Lampert 1990) are strongly decreased in the presence of this cyanobacterium. The amount of algae and blue-greens used in the experiments reported here was determined spectrophotometrically (800 nm). The organic carbon content of Scenedesmus and Cylindrospermopsis suspensions was determined with a Shimadzu TOC analyser.

Viscosity modification

The viscosity of water was modified in two ways: by decreasing water temperature and by dissolving highly branched polymers (Fig. 1). Two polymers were used: dextran 500, a sucrose polymer produced by Leuconostoc mesenteroides, (molecular weight 500,000; SigmaAldrich), and ficoll, a synthetic polymer of sucrose and epichlorohydrin (molecular weight 400,000; SigmaAldrich). Both polymers have commonly been used in high concentrations to increase viscosity in studies of bacterial movement (Berg and Turner 1979). The small concentrations used in this study have a negligible effect on water density and osmotic pressure (Podolsky 1994).

The concentrations used were 0.54% (w/v) for dextran and 1.6% (w/v) for ficoll, and both increased the viscosity by 30%, from 1.0 cP (the viscosity of water at 22°C) to 1.3 cP (the viscosity of water at 12°C). Kinematic viscosity was determined with a Cannon-Ubbelohde viscometer. The reasons for choosing these polymers were the following: both are highly branched, their molecules are roughly spherical, and their solutions are Newtonian. The solutions of unbranched polymers (e.g. methylcellulose) are Newtonian only at very low concentrations; at higher concentrations the long branches form a “net” and the fluid becomes “elastic”: non-Newtonian (Berg and Turner 1979). Furthermore, two previous studies on the effects of viscosity on suspension feeding (Podolsky 1994; Bolton and Havenhand 1998) found these polymers to be non-toxic.

Experimental design

Two series of experiments were carried out, at constant, dim light, all in a flow-through system similar to the one described in Stich and Lampert (1984), which minimises the effect of sedimentation and makes it possible to maintain a constant food concentration. The general design of all experiments was similar: vessels of 250 ml with food suspension were kept in temperature-controlled water baths. Filtered lake water (0.45 μm) containing algal suspension was pumped into the vessels at a constant rate of 1.5 l day−1 vessel−1 with a peristaltic pump, from stirred reservoirs. The water in the controls and the polymer solutions used in the experiments were reused: they were refiltered on a 0.45-μm-membrane filter every 24 h and fresh food suspensions were added. The reasons for this practice were economic, due to the high cost of the polymers used. Although the secondary metabolites of Daphnia, Scenedesmus, Cylindrospermopsis and bacteria accumulate in this way, no signs of toxicity were found. All experiments were started with fresh water. To minimise the load of secondary metabolites, Scenedesmus was centrifuged at low speed (700 r.p.m.), and resuspended in membrane-filtered lake water prior to its addition.

At the start of each experiment Daphnia neonates younger than 24 h were pipetted into the vessels. The number of neonates in the vessels depended on the species used (eight D. magna, ten D. pulicaria, 15–12 D. galeata and D. hyalina). At the beginning of the experiments the dry weights of five subsamples of neonates were measured on a Cahn 25 microbalance. The experiments were ended after 5 days, the body size and clutch size of ovigerous animals were determined and all animals were dried and weighed on a Cahn 25 microbalance. Growth rates were calculated as g=[ln (W t /W0)]/t, where W0 and W t are the weights of the animals at the beginning and the end of the experiments, respectively, and t is the duration of the experiment. In all experiments all treatments were done in three replicates.

In the first experimental series there were three treatments: high temperature (22±0.1°C, viscosity 1.0 cP), low temperature (12±0.1°C, viscosity 1.3 cP), and high temperature with artificially increased viscosity (22±0.1°C, viscosity 1.3 cP, only Dextran was used). In the first series D. magna and D. pulicaria were used. Ten experiments were performed, differing only in food (Scenedesmus) concentration. The food concentrations were 1.4, 1.0, 0.5, 0.35, 0.2, 0.125, 0.1, 0.0 (no food, experiment without the low-temperature treatment) mg C/l for D. magna and 0.4 and 1.4 mg C/l for D. pulicaria. The experiments were finished when the first clutches of daphnia appeared (after 5 days).

In the second experimental series all four Daphnia taxa were used. Similar temperatures and viscosities were used as in the first series, but with three more treatments, in which a mixture of S. obliquus and C. raciborskii was used as food. The concentration of S. obliquus was 0.4 mg C/l, the concentration of C. raciborskii was 0.25 mg C/l, (~1,600–1,800 filaments/ml). The concentration of blue-greens was kept moderate in order to minimise food quality (biochemical limitation) effects. The cold (12°C) treatments were done separately. Since at low temperatures the size of Daphnia at first reproduction increases significantly (at 22°C it was 1.8–1.9 mm, while at 12°C it was 2.6–2.8 mm for D. pulicaria) the cold treatments were ended after 10–14 days, when the animals were approximately 2-mm-long juveniles.

A Becton Dickinson flow cytometer (FACScalibur) was used in three flow-through experiments (with D. magna and D. pulicaria) to check whether dextran causes any deterioration of food particles (Scenedesmus and Cylindrospermopsis), or enhances bacterial growth (see supplementary figures). The FACS recorded four parameters: side scatter (SSC), forward scatter (FSC), fluorescence of fluorescein, (emission at 520 nm) and fluorescence of phycoerythrin (emission at 575 nm). The number of particles in the samples were counted with CellQuest software.

Results and discussion

The three marine studies that have separated the effect of a change in viscosity from a change in metabolism (Podolsky 1994; Bolton and Havenhand 1998; Hagiwara et al. 1998), found a strong effect of viscosity on the amount of collected particles in all cases. This also seems to be the case in large Daphnia, since at limiting food concentrations a 30% increase in viscosity decreases growth by 15–30% for D. magna and D. pulicaria (Fig. 2). The fact that at high food concentrations (1.0, 1.4 mg C/l) growth is reduced only marginally means that any toxicity of dextran or increased respiration rate is not the cause. This pattern of growth rates was expected if viscosity acts by reducing the flow through the filtering apparatus and in consequence reduces the amount of collected particles.

Fig. 2
figure 2

a Growth rates of D. magna (diamonds) and D. pulicaria (squares) in different food concentrations. Black symbols show controls at the high temperature (22°C), grey symbols show controls at the low temperature (12°C), open symbols show growth rates at 22°C with viscosity adjusted to its level at 12°C. Significant (P<0.05; t-tests) differences between the high-temperature treatments are marked by an asterisk. b The life history parameters of D. magna in the experiments shown in (a). The regression lines of control and treatments with dextran do not differ significantly (analysis of covariance df=1, 169; F=0.652; P=0.42), thus the life history response of D. magna to dextran (elevated viscosity) cannot be distinguished from that to hunger

Full size image

The growth when no food was added was probably caused by microbial growth in the experimental vessels due to dextran, but no decrease in viscosity was observed during the experiments. Dextran supported moderate microbial growth, with high bacterial concentrations appearing characteristically on the second day of the growth experiments (see Supplementary Figs. 1, 2 and 3). Prior to and subsequent to this, the amount of bacteria was higher with dextran than in the control, but the amounts in both were comparable (see supplementary figures). The positive growth of D. magna at 0.0 mg C/l with dextran suggests that the effect of viscosity on growth is underestimated in growth experiments.

The slopes and intercepts of regressions between the size at first reproduction and clutch size of animals cultured with and without dextran do not differ from what would be expected in the case of hunger (Fig. 2b). Regardless of the efforts made to separate toxicity from food limitation, some sort of biochemical interaction of Daphnia metabolism and the polymers is possible (although theoretically dextran 500 does not penetrate cell membranes; supplier information).

No significant change of Cylindrospermopsis filament length (as reported in Gliwicz and Lampert 1990) was observed in any of the experiments. Overall, the negative effect of C. raciborskii on growth rates was much smaller than reported in previous experiments with the same species (Gliwicz and Lampert 1990), despite the comparable concentrations of algae used. In this study the only species for which C. raciborskii strongly inhibited growth was D. pulicaria, the species that has much larger intersetular distances (Brendelberger and Geller 1985) than the other species used. Contrary to this, the growth of the other “large” species, D. magna was weakly affected by filaments, what is in agreement with expectation based on its very dense filters (Brendelberger and Geller 1985; Geller and Müller 1981). D. galeata could utilize C. raciborskii under all experimental conditions used (Fig. 3).

Fig. 3
figure 3

Growth rates and life history parameters of D. pulicaria, D. hyalina and D. galeata and two clones of D. magna. The strongest interaction of blue-green filaments and increased viscosity is visible in D. pulicaria, where the presence of dextran, ficoll or low temperature made Cylindrospermopsis similarly edible. Growth rates and life history patterns of D. hyalina and D. magna clone 1 show qualitatively similar reactions to D. pulicaria, while D. galeata and D. magna clone 2 could feed on the filaments at both viscosities. Note the differences in scales

Full size image

Although filaments had a much smaller influence on growth than expected, there was a significant interaction between elevated viscosity/low temperature and the presence of filaments (Table 1). D. pulicaria and D. hyalina had significantly higher growth rates in the presence of both filaments and dextran, than in the presence of dextran alone; the filaments caused no growth reduction at a low temperature (D. pulicaria); and D. magna clone 1 is significantly less sensitive to filaments when cultured at an elevated viscosity (Table 1, Fig. 3.). Body size is clearly not the only factor influencing sensitivity to filaments: D. hyalina and D. galeata are nearly identical in size, whereas D. magna is slightly larger than D. pulicaria. Overall, the response of different Daphnia species was consistent with the Res of their filters, and not their body size.

Table 1 ANOVA table for the experiments shown in Fig. 3
Full size table

The result that higher water viscosity makes D. pulicaria and D. hyalina less sensitive to filament interference helps to explain several observations: the coexistence of D. pulicaria with filamentous species in the field (Lynch 1980; Epp 1996; Paterson et al. 2002) and the observed temperature dependence of growth inhibition due to filaments (Threlkeld 1986). The observation that during seasonal succession large-bodied Daphnia can delay cyanobacterial blooms but cannot prevent them (Kasprzak et al. 1999) is also consistent with the finding that interference by filaments depends on water viscosity (temperature).

At present there is no consensus on the particle-collection mechanisms of water fleas. Calculations and visual observations (Gerritsen et al. 1988, see also J. R. Strickler’s website) suggested that only a minor portion of water goes through the filters, as the filtering appendages work like paddles rather than like rakes, and the main filtration mechanism is direct interception. Other studies (Brendelberger 1991; Gophen and Geller 1984) found strong correlations between filter mesh size and the minimum size of the collected particles that supports the sieve theory. The results of this study do not provide direct evidence for any of the collection mechanisms discussed above, but can be explained if both direct interception and sieving operate simultaneously on the filters (as discussed for copepods in Koehl 1996). The intersetular distances of D. pulicaria are much larger than those of D. magna (Brendelberger and Geller 1985), thus a possible cause for its filament sensitivity is that filaments clog the intersetular gaps. The strong positive effect of increased viscosity on growth of D. pulicaria in the presence of filaments may paradoxically be explained by decreased flow through the screens—which switches off sieving (see also Koehl 1996). This would reduce the amount of water filtered, but make possible the utilization of blue-greens, assuming that direct interception is not inhibited by the filaments. The correlation between the growth reduction in the presence of dextran (without filaments) and the growth reduction in the presence of cylindrospermopsis (without dextran) supports the hypothesis that the responses to viscosity and blue-green filaments have similar causes (Fig. 4.). Although these hypotheses are speculative at the present stage, it seems to be very difficult to explain the growth patterns of D. pulicaria and D. hyalina with reference to only one mechanism of particle retention.

Fig. 4
figure 4

Growth changes of the four Daphnia species compared to the high temperature (22°C)-low viscosity (no polymer) control. The correlation (P=0.043) between the response for elevated viscosity and the presence of blue-greens suggests that these two growth responses may have the same underlying cause

Full size image

This study represents the first quantitative examination of the importance of physical properties of water in determining the competitive abilities of a freshwater suspension-feeding organism. The non-intuitive way in which temperature influences filtration through viscosity may be very important when the effects of global warming are considered on aquatic ecosystems, especially in freshwater, since they experience much wider temperature fluctuations than marine environments. (For the effect of viscosity on cladoceran filtration rate see also the paper of Loiterton et al. 2004, which was published after the submission of this manuscript).