Relative strength of resource and predation limitation of heterotrophic nanoflagellates in a low-productive sea area

Abstract

The magnitude of resource and predation limitation of heterotrophic nanoflagellates (HNF) was studied in two short-term enclosure experiments performed in a low-productive sea area in the northern Baltic Sea in 2001. A cross-factorial design was used to simultaneously assess the relative importance of the two factors. Resource limitation was removed by adding bacteria, and predation limitation was eliminated by selective filtration. The first experiment was performed in June just after the spring bloom decline and the second in September at the end of the productive season. Resource limitation prevailed during both experiments, contributing to 60% of the net growth increase in June and 74% in September. Removal of predators had a significant effect only in June. Evidence for simultaneous resource and predation limitation was thus found only during the post-bloom situation. The results were applied to a model on resource and predation control of HNF abundances. To evaluate seasonal differences, field data on HNF and bacteria from a whole year study were applied to the model. Except for a few occasions during spring, the model indicated prevailing resource control of HNF at two locations with slightly different productivity.

Received July 10, 2005; accepted in principle August 25, 2005; accepted for publication September 5, 2005; published online September 8, 2005
 Communicating editor: K.J. Flynn

INTRODUCTION

Two of the most important factors shaping biological communities are resources and predation. Limitation, regulation and control of populations or trophic levels apply to different successive processes. Limitation refers to the initial response of the net growth rate after release of a limiting factor, regulation to the prolonged density-dependent effect on abundances and control to the equilibrium levels of organisms (Osenberg and Mittelbach, 1996). Both resources and predators can affect the three successive stages. Short-time experiments can be used for studies of limitation, while long-time experiments are needed for studies of regulation and control. Furthermore, controlling factors may be inferred from relations between prey and predator in natural ecosystems (Gasol, 1994). It is likely that the limiting and controlling factors are coupled. Samuelsson and Andersson (Samuelsson and Andersson, 2003) found that release of predation limitation of heterotrophic nanoflagellates (HNF) resulted not only in higher initial growth rates of the HNF but also in higher final biovolumes. The relative importance of resource and predation for the abundance of a population is believed to depend on the position in the trophic hierarchy and the productivity of the system. According to linear food chain theory, the controlling factor should alternate between adjacent trophic levels (Hairston et al., 1960). However, empirical data of plankton systems imply that the importance of predation increases with increasing productivity (Sarnelle, 1992; Mazumder, 1994; Steiner, 2001).

In pelagic environments, microorganisms constitute the major producers and consumers. Among these, heterotrophic microorganisms form the main link between dissolved organic material (DOM) and higher trophic levels, such as metazooplankton and fish (Fenchel, 1988; Porter, 1996). Osmotrophic bacteria utilize DOM as a carbon and nutrient resource. The dominating predators on bacteria are HNF (Sanders et al., 1992; Sherr and Sherr, 2002; Samuelsson and Andersson, 2003). Apart from being a grazer, HNF have a positive effect on bacterial growth through nutrient remineralization (Selph et al., 2003). HNF are also important grazers on picoplanktonic cyanobacteria and algae (Sanders et al., 2000; Sherr and Sherr, 2002). The relative importance of heterotrophic bacteria and phototrophic picoplankton as resource for HNF may depend on their biomass distribution. Dinoflagellates, ciliates, rotifers and metazooplankton are recognized predators on HNF (Dolan and Gallegos, 1991; Weisse, 1991; Gasol et al., 1995). Previous studies on HNF limitation have mostly focused on single factors (Jürgens et al., 1996; Calbet et al., 2001; Samuelsson and Andersson, 2003), although some experiments indicate the occurrence of simultaneous resource and predation limitation of HNF (Pace and Funke, 1991; Weisse, 1991; Kivi et al., 1993). To our knowledge, the relative importance of these factors has not been well quantified, neither in oligotrophic nor in eutrophic systems. To be able to measure the relative effect of resource and predation limitation, the limiting factors must be eliminated experimentally. Field data indicate different modes of control of HNF. In some studies, HNF abundances were found to increase with increasing bacterial abundance, indicating prevailing resource control (Sanders et al., 1992; Gasol et al., 1995, 2002), while in other a weak coupling between bacterial and flagellate numbers indicates predation control (Wieltschnig et al., 2001). Weak couplings between prey and predators may also occur because of other mechanisms, e.g. species interactions within the predators (interference or omnivory) or a high proportion of inedible prey (Micheli, 1999; Gasol et al., 2002). Correlative data can give an indication of governing factors, but not full understanding of complex mechanisms.

Gasol (Gasol, 1994) developed a qualitative model to assess bottom-up versus top-down control of HNF abundances. On the basis of empirical observations and energetic calculations, he calculated a maximum attainable abundance (MAA) of HNF that could be maintained by a certain abundance of bacteria. The model assumed that the flagellates only graze bacteria and that all bacteria were edible. A large data set including bacteria and HNF abundances from both marine and freshwater systems from different seasons showed that MAA seldom was reached. Gasol calculated an average line from all the relationships seen in nature and referred it to a mean realized abundance (MRA) line. Since both controlling factors were expected to be included in the large data set, the MRA line was assumed to average the simultaneous effect of resource and predation on the HNF community. Above the MRA line, HNF were suggested to be bottom-up controlled and below top-down controlled. Later it was found that the HNF to bacteria ratio were also dependent on the type of HNF predators in the system (Gasol et al., 1995). Especially cladocerans had a strong negative impact on HNF abundances. The Gasol model was used to assess the seasonal change in bottom-up versus top-down control of HNF in the Adriatic Sea (Solic et al., 1998). More recently, the Gasol model (Gasol, 1994) was modified for the determination of factors controlling the bacterial abundance (Gasol et al., 2002).

The aim of this study was to empirically determine the relative importance of resource and predation limitation of HNF in an oligotrophic sea area, the northern Baltic Sea. Two short-term studies were performed during summer and autumn in the low-productive northern Baltic Sea. Since predation only influences the net growth rate and not the specific growth rates of the prey, this study was only focused on the change in cell abundance. As heterotrophic bacteria are the main food resource for HNF in the area (Wikner et al., 1990; Samuelsson and Andersson, 2003), the results were compared with predictions from the Gasol model. In addition, bacteria and flagellate abundances in two sea areas with different productivity were applied to the Gasol model in order to identify a possible shift in the relative importance of resources and predation. Annual primary production differs with a factor of 4 in the two sea areas in the northern Baltic Sea (Andersson et al., 1996). We expected an increase in the relative importance of predation control, since heterotrophic flagellates are thought to be more predation controlled in systems with higher productivity (Sanders et al., 1992).

METHOD

Limitation experiments

Experiments were performed in the northern Baltic Sea in late June 2001, after the spring-bloom decline, and in late September, at the end of the productive season. Seawater was collected using a Ruttner sampler from 4 m (June) and 2 m (September) depths at a coastal station (63°30.5′ N, 19°47.8′ E). The water depth at the station was 16 m. The water was immediately prefiltered through a 90-μm nylon net to exclude metazooplankton and stored in acid-rinsed 10-L polycarbonate bottles. In the laboratory, two additional filtrations were made through 5- and 10-μm pore size polycarbonate filters (47 mm Ø Poretics) with a maximum filtration pressure of 4.0 kPa. The smallest pore size (5 μm) was used to isolate bacteria and small flagellates, while a 10-μm filter was used to include large flagellates and small ciliates. The 10 μm fraction was included to elucidate possible effects of mid-sized predators on bacterivorous nanoflagellates. The largest pore size (90 μm) included the whole microbial community, except cells living in large colonies or attached to large particles. To eliminate resource limitation, Pseudomonas fluorescens, an isolated bacterium from the study area with a size of 0.6 μm³ cell⁻¹ (Pinhassi and Hagström, 2000; strain BAL 18), was added in a concentration of ∼1 × 10⁷ cells mL⁻¹ to half of the treatments. The experimental design included six treatments: <5, <10 and <90 μm fractions with and without addition of bacteria. All treatments were incubated in situ at 4 m (June) and 2 m (September) depths in 250-mL dialysis bags (Spectra/Por 1, 12–14 kDa cutoff, 75 mm flat width). Bags were sealed with a double knot and clamped to seven steel frames. All treatments were performed in 16 replicates, and sampling was done daily during the 3-day experiment by removing the whole bag, i.e. four bags of each treatment were returned to the laboratory on each sampling occasion starting from day 0. Sub-samples were taken for analysis of inorganic nutrients, dissolved organic carbon (DOC), chlorophyll a (Chl a) and for the enumeration of bacteria, heterotrophic flagellates and ciliates. Simultaneously, samples from the surrounding seawater were analysed for the same parameters together with temperature and salinity. Cell growth on the walls of the dialysis bags was assumed to be negligible owing to the short incubation time.

Full-factorial analysis of variance (ANOVA) was used to test for differences in HNF numbers (log values) attributable to resource addition or predator removal. Since the experimental design made the samples independent in time, we analysed time as an independent factor. The interaction between time and the other factors reflected differences in community growth between treatments. Other independent factors were bacterial enrichment (+ and −, n = 2) and predator removal (<90 μm and <5 μm or <10 μm and <5 μm, n = 2).

Analysis of samples

Salinity and temperature were measured in situ with a conductivity meter (WTW LF 196). Photosynthetic active radiation (PAR) was determined at the incubation depths using a Li-COR underwater sensor Li 193 SA (Li-COR bioscience in Lincoln, IL, USA). Inorganic nutrients (PO₄³⁻, NO₂⁻, NO₃²⁻ and NH₄⁺) and DOC were measured daily in the surrounding water and in one of the treatment replicates. Water was filtered through 0.2-μm cellulose-acetate filters (Gelman Supor, GelmanScience, Ann Arbor, MI, USA), and samples for nutrient analysis were stored at −20°C. Samples for DOC analysis were acidified with 100 μL 2 M HCl and stored at +4°C. All filtration equipment and filters were rinsed in 1.2 M HCl. The inorganic nutrient samples were analysed using a TRAACS auto-analyser (Bran & Luebbe GmbH, Norderstedt, Germany), according to standard analytical methods (Grasshoff et al., 1983). DOC was measured with a high temperature carbon analyser (Shimadzu TOC-5000). Concentration of Chl a was measured with an Elmer LS 30 fluorometer (MPS-2000, Shimadzu Corporation, Kyoto, Japan) and calculated according to HELCOM (HELCOM, 1998). Samples of 50 mL (June) or 100 mL (September) were filtered through 25 mm Ø GF/C filters (Millipore) at a maximum pressure of 13.3 kPa, and Chl a was extracted in 5 mL (June) or 10 mL (September) of 95% ethanol for 24 h at room temperature in darkness.

Samples for bacterial and flagellate enumeration were preserved in 0.2-μm sterile-filtered glutaraldehyde (1% final concentration). For counting of heterotrophic bacteria, 1 or 3 mL of the conserved samples were filtered onto 0.2-μm black polycarbonate filters (Poretics, Osmonics, Livermore, CA, USA) and stained with acridine orange (Hobbie et al., 1977). Estimates of cell numbers and volumes were acquired by image analysis (Blackburn et al., 1998) in a Zeiss Axiovert 100 epifluorescence microscope (450–490 nm, Zeiss Filter Set nr 9). Carbon biomass of bacteria was calculated according to; m_b = v_b^0.7 × 0.12 pg μm⁻³, where m_b is the carbon content in pg C cell⁻¹ and v_b is cell volume in μm³ cell⁻¹ (Norland, 1993). Flagellates were filtered onto 0.6-μm black polycarbonate filters and stained with proflavine (Sherr et al., 1993). At least 30 fields (100 × 100 μm), or >100 cells, were counted at 1000× magnification in a Nikon TE 300 or Zeiss microscope equipped with an epifluorescence light source (103 W mercury lamp). Pigmented cells were distinguished from non-pigmented by their auto fluorescence. Mixotrophic flagellates were estimated by identification, according to their morphology and coloration. Ciliates were enumerated at the start and at the end of the experiment, while dinoflagellates were only estimated at the start. Ciliate/dinoflagellate samples were fixed with 0.2% alkaline Lugol’s solution. A volume of 50 mL was settled in a sedimentation chamber for 20 h and counted using the Utermöhl technique at 400× magnifications. Half of the sedimentation chamber was scanned during counting.

Net growth rate (μ) of the HNF community was calculated from the slope of a linear plot of ln HNF versus time. The change in net growth rate owing to resource addition or predator removal was used as a measure of the strength of limitation (Osenberg and Mittelbach, 1996). Small-sized flagellates, ≤5 μm, numerically dominated the flagellate community, constituting >80% of the cells. The calculation of growth rates and limiting factors, therefore, regarded small-sized flagellates ≤5 μm.

Resource limitation (RL) was calculated using three different equations, reflecting variation in the presence of predators and predation limitation (PL) using two equations reflecting the presence or absence of added resource (Table I).

Removal of limiting factors

An important requirement for this study was the ability to eliminate resource and predation limitation among HNF. To exclude resource limitation, we chose P. fluorescens, (Pinhassi and Hagström, 2000; strain BAL 18), which has been used in many functional response studies with protists (Fenchel, 1982; Eccleston-Parry and Leadbeater, 1994). The edibility of P. fluorescens was tested by comparing its consumption with that of an edible Escherichia coli wild-type strain NC3 (Andersson et al., 1985) in a HNF growth response experiment. The E. coli strain had never been in contact with seawater HNF and was therefore assumed to lack predation defence mechanisms. The bacteria were pregrown in ZoBell growth medium (ZoBell, 1946), concentrated by centrifugation, washed and resuspended in 0.2-μm filtered seawater. Bacteria were added to a final concentration of 10⁸ cells mL⁻¹ to Falcon test tubes containing 50 mL of 5-μm filtered seawater. Samples without any addition of bacteria were prepared as controls. The samples were incubated in a climate room at 15°C, and the HNF community was monitored over 3 days. To test the concentrations of bacteria that were sufficient to release the resource limitation, an in situ food response experiment was performed in dialysis bags with different concentrations of P. fluorescens, from 1 × 10⁶ to 1 × 10⁸ cells mL⁻¹. This experiment was performed in early September, and the samples were treated as in the other in situ experiments. The temperature at the incubation depth was 13.4°C.

To test whether predation limitation was fully removed in the 5-μm treatment, a dilution experiment was performed using 5-μm filtered seawater. Among protozoa, mainly small HNF pass through this filtration. However, theoretically some of the predators on HNF could still be present in the 5 μm filtration. The food resource (bacterial concentration) was not changed since the experimental water was diluted with 0.8-μm filtered seawater. Dilution treatments performed this way will thus remove predation limitation (Landry et al., 1995).

Seasonal study

Seasonal data from 2000 on bacteria and HNF concentrations from an offshore station in the Bothnian Bay (BB: 65°10′ N, 23°50′ E) and an offshore station in the Bothnian Sea (BS: 62°05′ N, 18°32′ E) were collected and applied to a qualitative model by Gasol (Gasol, 1994). Ten samplings per station were performed during the annual cycle. Integrated seawater samples were collected from 0 to 20 m depth using a plastic hose (Ø 2.5 cm) with vacuum pressure. A sub-sample of 30 mL was taken to estimate flagellate abundance, and a 3-mL sample was taken for enumeration of bacteria. The samples were treated as described above.

RESULTS

Efficiency of limitation removal

The chosen food resource, P. fluorescens, was found to be edible by flagellates. The HNF growth rate with ambient bacterial concentrations was 0.34 day⁻¹, while the growth rate with addition of P. fluorescens and E. coli was 0.69 and 0.83 day⁻¹, respectively. The food response experiment performed in early September showed that the maximal growth rate of flagellates was 0.36 ± 0.03 day⁻¹ and that this was reached at a bacterial concentration of 1 × 10⁷ cells mL⁻¹ (data not shown), indicating that resource limitation was eliminated at this concentration. The test of removal of predation limitation resulted in similar HNF growth rates in the <5 μm and <5 μm-diluted treatments (0.50 ± 0.04 and 0.47 ± 0.05 day⁻¹, respectively), indicating that predation was fully eliminated in the <5 μm filtration.

Experimental conditions

The physicochemical environment was partly similar in the two experiments. PAR was on average 30 μmol quanta m⁻² s⁻¹ at the incubation depth during both experiments, and the salinity varied between 3 and 4. However, a thermocline at 5 m depth was observed in June, while in September there was no stratification of the water mass. The mean water temperature was 14.5 ± 1.7°C in June and 11.9 ± 0.3°C in September. During both experiments inorganic nutrients showed no significant difference (t test, P > 0.05) between the surrounding water and the dialysis bags, except for ammonium, which was 2–6 times higher in the bags (t test, P < 0.05). In June, the DIN/DIP ratio in the free water mass was close to Redfield ratio (DIN/DIP = 0.43/0.02 μmol L⁻¹ = 21), while in the dialysis bags phosphorus limitation prevailed (DIN/DIP = 1.65/0.02 μmol L⁻¹ = 82). In September, both the free water mass and the dialysis bags exhibited a shortage of phosphorus (DIN/DIP = 3.85/0.05 μmol L⁻¹ = 77). The concentration of DOC in the open water was ∼350 μmol L⁻¹ in both experiments.

The initial abundance of heterotrophic bacteria was 4.9 ± 0.2 × 10⁶ cells mL⁻¹ in June and 2.7 ± 0.4 × 10⁶ cells mL⁻¹ in September. The corresponding biomass was 115 ± 7 μg C L⁻¹ in June and 48 ± 5 μg C L⁻¹ in September. The addition of bacteria in June increased the abundance to 9.7 × 10⁶ cells mL⁻¹ and in September to 11.7 × 10⁶ cells mL⁻¹ (Fig. 1). These values were equivalent to 304 and 404 μg C L⁻¹. In the non-enriched treatments, there was no significant change in bacterial numbers during the incubations (Fig. 1). However, in the treatments with addition of bacteria, the biomass did significantly decrease from day 2 in June (Sheffe’s post hoc; P < 0.05; cf. Fig. 1).

Potential predators on HNF were ciliates and heterotrophic/mixotrophic dinoflagellates. The initial abundance of ciliates in the <90 μm treatment in June was 6300 ± 1700 cells L⁻¹ and in September 1400 ± 200 cells L⁻¹. All of the ciliates were removed in the <5 μm treatments, while in the <10 μm treatment about 50–70% of the ciliates were removed. Among heterotrophic ciliates Balanion sp., Lohmaniella/Strombilidium spp. Strombidium sp. and Urothricha sp. were common both in June and September. The abundance of ciliates, especially Balanion sp. and a colorless Mesodinium sp. increased in the treatments with addition of bacteria in June. In all other treatments, including the September experiment, the ciliate abundance remained constant or decreased (data not shown). The initial abundance of dinoflagellates in the incubation water was 400 cells L⁻¹ in June and 11 200 cells L⁻¹ in September. The dinoflagellates belonged to the genera Amphidinium, Dinophysis, Gymnodinium and Katodinium. The development of dinoflagellate populations was not monitored during the experiment.

The concentration of Chl a in the open water was 2.36 ± 0.16 μg L⁻¹ in June and 0.61 ± 0.06 μg L⁻¹ in September. Chl a increased in all treatments with a rate of 0.44 ± 0.18 μg day⁻¹ in June and 0.19 ± 0.18 μg day⁻¹ in September. Mixotrophic nanoflagellates varied between 700 and 1700 cells mL⁻¹ in June and between 20 and 90 cells mL⁻¹ in September. Common mixotrophic nanoflagellates were Chrysochromulina spp., Pseudopedinella tricostata and Pseudopedinella sp. (with six chloroplasts). There was no significant difference in abundance between treatments or between day 0 and day 3.

Response by heterotrophic flagellates

The abundance of HNF in the surrounding water during the experimental incubation varied in time between 2800 and 8400 cells mL⁻¹ in June and between 900 and 1400 cells mL⁻¹ in September. The initial abundance of HNF in the experiments was on average 3200 ± 800 cells mL⁻¹ in June and 1100 ± 200 cells mL⁻¹ in September. In June, the abundances of HNF increased across all size fractions (<5, <10, <90) on average from 3200 to 12 000 cells mL⁻¹ in treatments with ambient bacterial concentrations and to 22 000 cells mL⁻¹ in treatments with addition of bacteria (Fig. 1). In September, the abundance increased across all size fractions from on average 1100 to 1400 cells mL⁻¹ in treatments without addition of bacteria and to 1800 cells mL⁻¹ in treatments with addition. The HNF community growth rate in June was thus generally higher than in September. In June, growth rates varied between 0.26 and 0.81 day⁻¹ and in September between 0.02 and 0.45 day⁻¹ (Fig. 2). There was on average a 55% increase in the HNF growth rates because of resource addition in June (P < 0.001, Table II) and a 188% increase in September (P = 0.005). Removal of predators caused a 33% increase in growth rate in June (P = 0.022), while no significant difference was detected in September (P > 0.05). However, there was an indication of increased growth rates because of removal of predation also in September (Fig. 2). Mid-sized predators (<10 μm) did not have any significant effect on the growth rate of HNF. The relative importance of resource and predation limitation was 60 and 40%, respectively, in June, and 74 and 26%, respectively, in September (Table III). These figures were calculated using the formulas in Osenberg and Mittelbach (Osenberg and Mittelbach, 1996). However, these are conservative measures, where only one of the limiting factors should be altered in each treatment [equations (1) and (4)]. If there was no interaction between predation and resource limitation, it would in principle be possible to use treatments where both predation and resource availability had been modified [equations (2), (3) and (5)]. In the June experiment, we detected no interaction between the limiting factors, indicating that we could utilize all equations, 1–5, to estimate the relative importance of different limiting factors. The coefficient of variation of the change in growth rate owing to resource or predation limitation was 10–14%, indicating that the different measures could be considered as replicates. However, there was a very high coefficient of variation in September, 119%, in the analysis of predation limitation. Consequently, we only used equations (1) and (4) to estimate the relative importance of resource and predation limitation in both the June and the September experiment.

Comparison with a qualitative model

We compared our data (Fig. 3) on HNF limitation using a theoretical framework presented by Gasol (Gasol, 1994). In June, the relationship between flagellates and bacteria in open water was above the MRA line (Fig. 3a), which indicated that the flagellates were mainly resource controlled. This was also seen in our experiment, since addition of bacteria led to increased HNF growth rates, while excluding predators had a weaker effect on HNF growth rates. The HNF–bacteria relationship in September was below the MRA line, implying that the flagellates were more predation controlled than resource controlled. This was not in agreement with our experimental results. The discrepancy between the model and the experimental results might be because of the model assumption that all bacteria are edible, regardless of bacterial size or behavior, e.g. colony forming and motility (Gasol, 1994). It has been shown that HNF graze on bacteria in a size range of 0.07−1 μm³ (Andersson et al., 1986; González et al., 1990; Hahn and Höfle, 2001; Wu et al., 2004). We therefore analysed the size distribution of the bacterial community in the open water in June and September. The proportion of bacteria >0.07 μm³ was 54 ± 3% in June and 41 ± 4% in September. The abundance of bacteria in the theoretical model was corrected with the percentage of bacteria >0.07 μm³. No upper limit were set since there were almost no (<0.5 %) bacteria >1 μm³ occurring naturally in our system. Furthermore, the model assumes that the flagellates only feed on bacteria. The dominating bacterivorous flagellates are ≤5 μm in size (Fuhrman and McManus, 1984; Rassoulzadegan and Sheldon, 1986; Wikner and Hagström, 1988), thus the flagellate numbers were also corrected to only include cells ≤5 μm. The relationship between bacteria and HNF was consequently moved proportionally to the x axis and y axis (Fig. 3). The corrected relationship appeared above the MRA line, which indicated a more resource limited community, in accordance with our experiment.

Seasonal data from 2000 on bacteria (>0.07 μm³) and HNF (≤5 μm) concentrations showed that throughout the year the more productive BS maintained a higher abundance of bacteria than the BB (Fig. 4, paired t test, P < 0.001). There was generally also a higher abundance of small flagellates in the BS than in the BB, although this was not significant (paired t test, P = 0.09). The abundances of bacteria and flagellates at the nearshore experimental station were higher than what is normally found at the offshore stations (Figs 1 and 4). Among HNF ≤5 μm, chrysophytes were commonly occurring throughout the year. Important species in the chrysophyte community belonged to the genus Paraphysomonas (Berglund et al., unpublished data). Other small flagellates were Monosiga spp. and unidentified species possessing one flagellum.

There was a high positive correlation between bacteria and flagellate concentrations in the BB (r = 0.92, P = 0.001), while the bacteria–flagellate relationship in the BS showed higher variation (Fig. 4). The relationship between size-corrected bacteria and HNF numbers were plotted in the theoretical model (Fig. 3b). At both stations, the data points ended up above the MRA line at 17 out of 20 occasions, indicating that the HNF were mainly resource limited. The relationship was below the MRA line in the BB at two occasions in May and in the BS at one occasion in January.

DISCUSSION

Relative importance of resource and predation limitation

In this study, two short-time experiments were performed to empirically determine the relative importance of resource and predation limitation for the net growth rates of bacterivorous HNF in the northern Baltic Sea. The results showed that HNF primarily were limited by resource, although a simultaneous predation limitation was measured. The abundance of the main resource (bacteria) and the predators (protozoa and zooplankton) show large seasonal variations in the area (Fig. 4; Samuelsson et al., Linköping University, unpublished data). There is therefore reason to believe that also the relative importance of resource and predation as limiting factors show seasonal variations. For example, during and after the spring bloom, a period characterized by high concentrations of large flagellates and ciliates, the importance of predation on small flagellates may be more pronounced. The results of our in situ experiments were in agreement with this. The experiment performed during the decline of the spring bloom (June) showed that the importance of predation was larger than at the end of the productive season (September). Furthermore, when field data of bacteria and HNF abundances were applied to the model by Gasol (Gasol, 1994), May—values from the BB indicated predation control of the flagellates. However, in general the field data indicated that the flagellates in the two basins were mainly resource controlled throughout the year. No differences between the basins were observed. The productivity differs approximately four times between the two basins. According to earlier studies, predation control should increase with productivity (Sanders et al., 1992; Sarnelle, 1992; Gasol, 1994). The difference in productivity between the basins thus seemed to be too small to generate such a shift in the importance of different controlling factors in the microbial communities.

The Gasol model is based on the relationship between bacterial and HNF abundance (Gasol, 1994). However, the abundance of bacteria is not the best measure of prey availability, since the size and nutritional quality of bacteria show large variations. Furthermore, bacteria may be inedible because of characteristics, such as size or behavior (Andersson et al., 1986; Jürgens and Matz, 2002). It is also well known that HNF may use other food resources than heterotrophic bacteria, e.g. picocyanobacteria, picoeukaryotes or detritus (Kuosa, 1991; Sanders et al., 2000; Sherr and Sherr, 2002). Therefore, care should be taken when interpreting the results of the model. However, in our study area the main food resource for HNF seems to be heterotrophic bacteria (Wikner et al., 1990; Samuelsson and Andersson, 2003). Pico-algae are certainly eaten, but they are not quantitatively an important food resource. This indicates that one of the assumptions of the model, that HNF only eat bacteria, is convincingly met. By size correcting the bacterial abundance, the second assumption was probably fulfilled, i.e. only the edible bacteria were included into the model. The relationship between bacterial and HNF abundance in the September experiment was very close to the MRA line. The MRA lines should not be considered as a sharp border between resource or predation control of HNF. There is certainly some variability inherent in the line.

This work and an earlier study in the same sea area (Samuelsson and Andersson, 2003) indicated that also heterotrophic ciliates and bacteria were predominantly resource limited. Bacterial growth rates increased more when resource limitation was decreased than when predation limitation was removed (Samuelsson and Andersson, 2003). Similarly, ciliates only showed an increase in abundance because of resource enrichment. This is not in agreement with the prediction that the main controlling/limiting factor should alternate between adjacent trophic levels (Hairston et al., 1960; Oksanen et al., 1981). One explanation to this deviation might be that the productivity of the system is too low for significant predation limitation. Further, several factors violate the predictions of the linear food web theory, such as prey refuges, heterogeneous trophic levels and omnivory (Leibold, 1989; Diehl and Feissel, 2000; Steiner, 2001). It is known that phytoplankton coexisting with potential predators could become relatively invulnerable to predation (Leibold, 1989). Increase of inedible prey may also cause resource limitation of edible prey and could therefore cause resource limitation of both the prey and predator level (Osenberg and Mittelbach, 1996). Similarly bacteria are known to have predation defence mechanisms (Hahn and Höfle, 2001; Šimek et al., 2001; Jürgens and Matz, 2002), which might explain resource limitation of HNF. In the studied sea area, omnivory is significant in the microbial food web, even though the main transport of carbon is through a linear food chain (Samuelsson and Andersson, 2003). For example, many ciliates are known to graze both flagellates and bacteria (Sherr and Sherr, 2002; Samuelsson and Andersson, 2003). This may cause both HNF and ciliates to become mainly resource limited. Nano-sized flagellates and ciliates may also eat particles of their own size making it difficult to classify them into trophic levels (Havskum and Hansen, 1997). Thus, it is evident that the microbial food web has properties that could weaken predator–prey interactions.

The species composition of the communities might have influenced the response to the resource and predation manipulations. For example, it has been shown that the response of phytoplankton and zooplankton to nutrient enrichment is dependent on the species composition of the herbivore community (Leibold and Wilbur, 1992). It is well known that the bacterivorous HNF includes species with considerable differences in their functional biology (Eccleston-Parry and Leadbeater, 1994; Boenigk and Arndt, 2002). Variations in feeding parameters such as handling time, clearance rate and optimal prey size in addition to a probable difference in predation vulnerability of HNF (e.g. because of size, motility or attachment to particles) imply that all species may not be limited by the same factor. The results from this study should therefore be applied at the community level and not interpreted as if all HNF populations were primarily resource limited.

It is known that HNF have species-specific predator–prey relationships with bacteria (Matz et al., 2002). In this study, a single species of bacteria was added. The pregrown bacterium had a relatively large cell volume (∼0.6 μm³) compared with the volumes of the natural bacterial community. This might have promoted HNF species preferring larger prey. HNF also show a large variation in size, the same species can vary, e.g. from 150 μm³ to 500 μm³, depending on the physiological status of the cells (Fenchel, 1982). Concentration and addition of the natural bacterial community would have been difficult, since bacteria and HNF are in the same size range. Furthermore, the natural community may include different proportions of inedible bacteria that would have made it problematic to determine whether resource limitation really was excluded. Addition of a more diverse bacterial community might have increased the number of flagellate species responding to the higher resource level. If this holds true, we probably underestimated the importance of resource limitation. However, this would not change the major conclusion of this study, i.e. that HNF primarily were resource limited.

We did not analyse the effect of viruses (or parasites) on the HNF community. Viruses may act as predators on eukaryotic populations and thereby decrease the net growth rate (Brussaard, 2004). Virioplankton may also be a food resource for HNF, even though their quantitative importance may be low (Bettarel et al., 2005). However, the additional source of death or nutrition for HNF owing to viruses should have been relatively equal in all treatments since viruses may pass all the filter cutoffs.

In this study, we assumed that the three days following limitation removal represented the initial response. In June, the net growth rate of the flagellates during these three days was constant, indicating that our assumption was correct. In September, however, the HNF showed a 1-day lag in five out of six treatments (Fig. 1). This was probably because of starvation (cf. discussion about HNF growth rates), which strengthens our conclusion that the flagellates were resource limited during the period. To mimic natural conditions and minimize experimental artifacts, all incubations in this study were performed in dialysis bags incubated in situ. The bags should allow an exchange of nutrients and dissolved organic matter with the surrounding water. However, this was not totally fulfilled, since increased concentrations of ammonium were observed within the dialysis bags compared with the open water. Both heterotrophic bacteria and protozoa probably produced inorganic nitrogen and phosphorus during the incubations. In particular, ammonium and phosphate is released as a consequence of protistan grazing (Dolan, 1997). Since phosphorus generally is a limiting nutrient in the studied sea area (Zweifel et al., 1993; Andersson et al., 1996), phosphate might have been extensively taken up by bacteria and phytoplankton, while the ammonium was not depleted. The concomitant increase in resource (bacterial production) for the HNF could have led to an overestimation of their growth rates and an underestimation of the resource limitation. An increase of HNF in the control, without addition of bacteria, indicates that this was the case. This would, however, not change the conclusion that HNF were mainly resource limited. The relatively stable bacterial number despite increased HNF concentrations also indicated increased bacterial growth. We find no reason to believe that the predation limitation was underestimated in the study and, therefore, the major conclusions would be correct despite eventual experimental drawbacks.

HNF growth rates

The HNF growth rates in the non-enriched treatments were in general in the same range as observed in previous studies in the Baltic Sea (Bjørnsen et al., 1988; Kuuppo-Leinikki et al., 1994; Samuelsson and Andersson, 2003). However, the obtained maximum growth rates in the bacterium-enriched and predator-free treatments were lower than reported for isolated flagellates (Eccleston-Parry and Leadbeater, 1994). We estimated growth rates corresponding to ∼20–25% of the maximum values for chrysophytes, which is a common group in the HNF community (Berglund et al., unpublished data). A possible explanation may have been that only a certain fraction of the natural HNF community was active and able to respond to the added bacteria. The reason for a high proportion of inactive cells could be starvation or viral infection. We find it likely that starvation is an important factor in resource-limited systems like the northern Baltic Sea. Viruses are presumably of less importance in oligotrophic systems with low cell concentrations (Brussaard, 2004).

It is known from other areas that there are seasonal differences in growth rates of the HNF community (Bjørnsen et al., 1988; Bloem and Bär-Gilissen, 1989; Weisse, 1997). In agreement, the HNF community showed three times higher growth rates in June than in the September experiment. Differences in the functional biology of the HNF communities may have contributed to the varying growth rates (Eccleston-Parry and Leadbeater, 1994; Boenigk and Arndt, 2002). The species composition of the HNF assemblage was not analysed in the experiments. However, from epifluorescence microscopy and molecular identification (sequencing), we know that most of the small flagellates (≤5 μm) in the studied area are chrysophytes (Berglund et al., unpublished data), and ∼50% of the occurring species are similar in June and in September. It is therefore likely that other factors caused the large differences in growth rate. Weisse (Weisse, 1997) showed that seasonal differences in growth of specific HNF species could be explained by variations in bacterial abundance and temperature. It was found that the growth of Spumella sp. was highly related to food supply and temperature, while growth of Kathablepharis sp. was dependent on temperature only. In our study, the discrepancy in growth rate remained after removing all predators and after the addition of bacteria (0.70 ± 0.10 day⁻¹ in June and 0.26 ± 0.06 day⁻¹ in September, Fig. 2). Differences in resource could thus not explain the variation in growth rates. Lower resource levels in September could, however, have caused starvation of the flagellates. The recovery time before exponential growth is positively correlated to the proceeding starvation period (Fenchel, 1982). In agreement, a growth lag phase was observed in September but not in June causing lower net growth rates (Fig. 1). Growth rates of protists respond linearly to temperature, with a slope of 0.07 day⁻¹ °C⁻¹ (Montagnes et al., 2003). The mean temperature was 2.5°C colder in September than in June. This would give a maximum growth rate of 0.48 day⁻¹ in September if the water had been 2.5°C warmer. Temperature thus could explain about 70% of the difference in growth rates between the experiments. The remaining difference was probably caused by a combination of differences in species composition and proceeding growth conditions.

CONCLUSIONS

Bacterivorous HNF in the oligotrophic northern Baltic Sea were found to be mainly resource limited. This was concluded from short-term experiments and by applying bacteria and HNF abundances to the Gasol model (Gasol, 1994). Furthermore, the adjacent trophic levels, bacteria and ciliates, also seemed to be limited by resource. Our results confirm the general occurrence of resource limitation in low productive areas. It would be interesting to perform similar studies in productivity gradients, since it is likely that there will be a continuous change in the relative strength of the limiting factors.

This study was supported by grants from the EU (BASYS MAS3-CT96-0058) and Umeå Marine Sciences Centre. We thank E. Lundberg and C.-H. Stangenberg for chemical analyses, B. McKie for improving the English and G. Englund for valuable comments on an earlier version of this article.

REFERENCES

Author notes

¹Department of Ecology and Environmental Science, Marine Ecology, Umeå University, SE-901 87 Umeå, Sweden, ²Umeå Marine Science Centre, SE-910 20 Hörnefors, Sweden and ³Department of Biology, Åbo Akademi University, Tykistökatu 6 A, FIN-20520 Turku, Finland