Growth rates of common pelagic ciliates in a highly eutrophic lake measured with a modified dilution method

Before each experiment, all the equipment was acid-washed and thoroughly rinsed with deionized water. Lake water for the experiments was prescreened through 100-μm mesh nylon gauze to remove larger zooplankton that could not be distributed representatively in the treatments. This water was treated as whole water and was carefully mixed. To produce the 10-μm filtrate, part of the water was sequentially gravity-filtered through 25-μm mesh gauze, and then twice through 10μm mesh gauze. To prevent clogging, the first filtration with 10-μm mesh gauze was carried out through three parallel filtration sets. Next, filtrates were combined and filtered through the fourth filtration set. Samples were taken from the 10-μm filtrate to assess the impact of filtration on bacterial abundance, flagellate abundance, and chlorophyll concentration. Subsequently, whole water and 10-μm filtrate were used to prepare the dilutions, in which fractions of whole water were 20% (in triplicate), 40%, 60%, 80%, and 100% (undiluted whole water, in triplicate). All dilutions were prepared in 120-ml bottles (9 in total), and each bottle was gently mixed by rotating it 50 times. Two bottles, one each for 20% and 100% dilutions, were used to assess changes in bacterial and flagellate abundances as well as chlorophyll a concentration during incubation. Next, initial samples were collected for ciliate counts (7 samples), bacterial and flagellate counts (one each for 20% and 100% dilutions), and water was sampled for chlorophyll a concentration measurements. Experimental incubation was conducted for 3.5–4.8 h under natural conditions (temperature, wave motion, irradiance), using an anchored experimental set-up. The exact duration time was used to calculate growth rates. After incubation, final samples were taken for the determination of ciliate, bacterial and flagellate abundances, and chlorophyll a concentration measurements (similarly to the initial samples). Finally, I took an additional sample from the lake to compare the ciliate dynamics in the whole experimental water with that in the environment.

The ciliates were analyzed in samples fixed with acid Lugol’s solution (final concentration 0.5%). The samples were stored in the dark at 4°C until they were analyzed using the Utermöhl (1958) method. The observations were conducted using an Olympus CKX41 inverted microscope. The ciliates were identified according to Foissner & Berger (1996). It was crucial to analyze the entire bottom of the Utermöhl chamber and not half of it to minimize the error caused by the uneven distribution of specimens on the bottom. Samples were dense with algae and detritus, so small volumes of water were taken for analyses: 3 ml or 10 ml for the more diluted treatments. If necessary, additional counts were performed to gather a sufficient number of specimens. In each experiment, the entire ciliate community was analyzed in one initial sample collected from whole water (see abundance values in Table 1). In the remaining samples, only two ciliate species and their growth rates were analyzed. The ciliate species were selected according to (i) high abundance, which allowed for an accurate assessment of changes in the abundance during incubation, and (ii) distinct morphology, which facilitated the identification of all specimens present in samples fixed with Lugol’s solution. Because fixation with Lugol’s solution obscures taxonomic features, the ciliates were identified to the genus level. During incubation, no evident changes in ciliate cell volumes were observed. For each dilution, the apparent growth rate (k, h^-1) was calculated assuming the exponential growth during incubation:

where N₁ was the final abundance (cells ml^-1), N₀ was the initial abundance (cells ml^-1), and t was the duration of incubation (h).

The experiments were analyzed according to Landry & Hassett (1982). The encounter rates between predators and prey were gradually reduced along the dilution gradient from undiluted to the most diluted water. Thus, the apparent growth rate (k, h^-1) in each dilution depended on the specific growth rate (μ, h^-1), i.e. the intrinsic rate of increase, grazing pressure (g), and the fraction of whole water in the dilution (D, %):

Plotting the apparent growth rates against fractions of whole water in dilutions enabled the estimation of specific growth rates (μ, h^-1) for the theoretically complete relaxation of grazing pressure (D = 0%, y-intercept, Fig. 1). A single measurement comprised 7 experimental treatments and the analysis of 15 samples (7 initial and 7 final samples, and also one additional environmental sample); thus, the growth rate estimates were statistically significant. Growth rates were measured during short incubations (3.5–4.8 h), therefore they were expressed per hour.

Figure 1

Measurements of ciliate growth rates were compared with potential (maximum) growth rates estimated from an allometric equation formulated on the basis of ciliate growth in cultures. The model published by Müller & Geller (1993) was chosen because it is based on the largest data set and is considered to be the best (Macek et al. 1996, Montagnes 1996). In this model, potential specific growth rates (μ, d^-1) were estimated from the mean cell volume of the organism studied (V, μm³) and the ambient temperature (T, °C):

To calculate the volume of organisms studied, they were measured in samples of the whole water collected before incubation. Measurements were carried out with an ocular micrometer or image analysis software. The volume (V, pm³) of Rimostrombidium sp., Cyclidium sp., and Urotricha sp. were calculated from mean values of their length (L, pm) and width (W, pm) using standard formulae for the volume of a rotational ellipsoid:

The volume of ciliates can shrink after fixation with Lugol’s solution. This shrinkage is species-specific and also depends on the physiological state of the cells (Choi & Stoecker 1989; Wiackowski et al. 1994). Thus, no correction was applied to avoid introducing additional error. The potential specific growth rates computed (μ, d^-1) were divided by 24 to calculate values per hour. Growth rate estimation for tintinnid ciliates (Tintinnidium sp.) was more complicated, because tintinnid loricae are built by protoplasts and the energy allocated for lorica construction has to be taken into account (Rychert 2011; Agatha et al. 2013). First, the carbon content of a tintinnid ciliate (CC, pgC), both lorica and protoplast, was calculated from the volume of lorica using the coefficient 0.053 pgC μm^-3 (Verity & Langdon 1984; Seuthe et al. 2011). The volume of lorica (V, μm³) was calculated from the mean width (W, μm) and length (L, μm) of lorica. Lorica shape was described as a cylinder and hemisphere:

Next, the carbon content was calculated back to the volume using the coefficient 0.11 pgC μm^-3 (Turley et al. 1986) to calculate the surrogate cell volume, i.e. volume of hypothetical protoplast containing the organic carbon allocated for production of both protoplast and lorica. This surrogate cell volume was used to calculate the potential growth rate according to the formula by Müller & Geller (1993), similarly as with aloricate ciliates.

Additionally, the tintinnid ciliate growth (μ, d^-1) was calculated based on the lorica oral diameter (LOD, μm) according to the formula by Dolan (2010), which was further developed by Montagnes (2013):

The estimated growth rates (d^-1) were recalculated per hour.

During all six experiments, I assessed the impact of dilution on food resources of the organisms studied in the whole water and in 20% dilution. Three parameters were studied: bacterial abundance, flagellate (ANF and HNF, size: 2–10 μm) abundance, and chlorophyll a concentration. Their changes during incubation were small to moderate, so the mean values (N) were calculated from initial (N₀) and final (N₁) values according to Frost (1972):

The quantity of food resources was analyzed in each experiment separately according to the food demands of the organism in question. When necessary, they were recalculated into carbon units. The nutritional value of algae was calculated from chlorophyll a concentration (Table 1) assuming the C:Chl a conversion factor of 40.7 (Montagnes et al. 1994). The bacterial biomass was calculated from the abundance (Table 1) using a coefficient of 17 fgC bacterium^-1 (Cleven & Weisse 2001), which is a mean value among those reported in the literature. The biomass of nanoflagellates was also calculated from their abundance (Table 1) assuming a conservative mean flagellate size (3 μm) and a conversion factor of 0.22 pgC μm^-3 (Børsheim & Bratbak 1987).

Rimostrombidium sp. grew at rates of 0.16 h^-1 at 15.4°C and 0.12 h^-1 at 15.6°C (Fig. 1A–B, Table 1), which corresponded to the highest estimates of the maximum growth rates reported in the literature for small, satiated oligotrich ciliates, e.g. 0.15 h^-1 at 15°C (Rivier et al. 1985) or 0.11 h^-1 at 20°C (Ohman & Snyder 1991). The field research conducted in hypereutrophic waters by Lavrentyev et al. (2004) also revealed similar growth rates of Rimostrombidium humile: 0.11 h^-1; however, measurements were taken at a higher temperature (22–25°C).

Tintinnidium sp. demonstrated an exceptionally high growth rate of 0.19 h^-1 around midnight at 15.6°C (Table 1), whereas around noon the specific growth rate was close to zero (Fig. 1C–D). This indicated that Tintinnidium sp. demonstrated a diel rhythm in growth with cell divisions occurring during the night. Some researchers studying the growth of tintinnids reported this cycle (Campbell 1926; Biernacka 1952), while others did not (e.g. Heinbokel & Coats 1986). Assuming that estimates for the night were 0.19 h^-1 and for the day - zero (no growth), the mean diel growth rate would be 0.10 h^-1. This value is twice as high as the maximum growth rate reported for Tintinnidium fluviatile (0.05 h^-1, Verity 1986) and several small tintinnid ciliates studied in cultures (0.05-0.06 h^-1, 18°C, Heinbokel 1978). A similar low value of 0.05 h^-1 was also calculated using a formula relating the growth rate to the lorica oral diameter (Dolan 2010; Montagnes 2013, see Materials and Methods). On the other hand, Lavrentyev et al. (2004) reported an even higher growth rate of Tintinnidium fluviatile: 0.12 h^-1 during the field research at higher temperatures of 22–25°C.

Cyclidium sp. grew very fast in June 2013 (0.12–0.13 h^-1 at 18.2–18.4°C, Fig. 1G–H, Table 1) and around noon in June 2012 (0.08 h^-1 at 20.0°C). However, in the latter experiment, Cyclidium sp. seemed to be relieved from grazing pressure only in the most diluted water (Fig. 1E), consequently the specific growth rate was underestimated. Assuming that the so-called “saturated feeding effect”, in which grazing pressure is very strong and relaxes only in more diluted water (Gallegos 1989; Calbet & Saiz 2013), additional growth rate calculations were carried out only for the most diluted treatments (20–60%). This growth estimate (0.21 h^-1) was higher, as was the statistical significance (Fig. 1E). However, a 2.5 times higher growth rate than in the 20% dilution (Fig. 1E) is very unlikely, and it seems to be overestimated. Thus, the actual specific growth rate was somewhere between the low, underestimated value (0.08 h^-1) and the high, overestimated value (0.21 h^-1). Generally, the estimates of the growth rates of Cyclidium sp. are similar to the maximum estimate for Cyclidium sp. of a similar size (887 μm³) at 0.16 h^-1 at 20°C (Boenigk & Novarino 2004) and higher than estimates of the maximum growth rate reported in the literature for Cyclidium glaucoma, e.g. 0.079 h^-1 at 20°C (Taylor 1978) or 0.10 h^-1 at 20°C (Finlay 1977).

In conclusion, the growth rates measured in this study corresponded well with the maximum values reported in the literature at similar temperatures or were even higher. Therefore, the hypothesis posed in the Introduction was confirmed. As mentioned in the Introduction, ciliate growth could be approximated with the allometric equation by Müller & Geller (1993). However, it does not necessarily estimate the maximum growth rates of particular species, but approximate the “general ciliate growth” and are especially useful for estimating the growth rates of multi-species ciliate communities (Montagnes 1996). To perform a general comparison (Fig. 2), the mean diel growth rates from measurements were compared with growth rates estimated with the allometric formula according to Müller & Geller (1993). This comparison also indicated that the growth rates measured are at the same level as rates expected from the growth observed in cultures.

Figure 2

Unsuccessful experiments with Urotricha sp. were unexpected, because organisms from this genus were introduced into cultures and were thoroughly studied under laboratory conditions (Weisse et al. 2001; Weisse et al. 2002). Thus they could be considered as robust. I cannot offer any explanation why Urotricha sp. was sensitive to the experimental procedure. Studies were performed in the highly eutrophic and brackish lake with unstable environmental conditions (Wielgat-Rychert et al. 2015). Therefore, it is possible that the sensitivity of Urotricha sp. was induced by an unidentified, perhaps abiotic, factor.

The dilution method is based on the two main assumptions: (i) the growth of the organisms studied is not food-limited in subsequent dilutions and (ii) grazing pressure depends linearly on the dilution factor, which is proportional to the concentration predators (Landry & Hassett 1982; Calbet & Saiz 2013). The first assumption holds if the experimental procedure does not cause a reduction in the food resources in dilutions or, alternatively, organisms are food-satiated, both in the whole water and in the dilutions, despite the reduced food resources in dilutions. Filtration excluded bacteria attached to detritus particles. In the 20% dilution, bacterial abundance was 55–94% (mean: 81%, Table 1) of that observed in whole water. However, neither attached bacteria nor detritus particles larger than 10 μm are edible for the organisms studied. Similarly, the abundance of flagellates was also reduced. Their abundance in 20% dilution was 60–93% (mean: 80%, Table 1) of that observed in whole water. Filtration had a pronounced effect on chlorophyll a concentration: the 10-μm filtrate contained only 10–20% (mean: 16%, Table 1) of the chlorophyll a concentration recorded in whole water. However, filtration excluded large, abundant colonial green algae, like Scenedesmus quadricauda and Pediastrum spp., colonial diatoms (genera: Asterionella, Nitzschia, and Fragilaria), and other large algae, which are not food resources for the organisms studied. Therefore, only the chlorophyll a concentration present in the 10-um-filtrate was treated as a food resource available to the organisms studied. The food resources were analyzed according to the food demands of the organisms in question in each experiment. However, food resources were assessed on the basis of three basic parameters (chlorophyll concentration, abundance of nanoflagellates, and abundance of bacteria; Table 1), consequently I estimated the potential food and not the most optimal prey grazed preferentially by organisms studied.

Rimostrombidium sp. and Tintinnidium sp. were studied in September 2011 (Table 1). On this occasion, the biomass of algae smaller than 10 μm was 286 μg C l^-1 (night) and 256 μg C l^-1 (day). The biomass of flagellates in the 20% dilution was 35 μg C l^-1 (night) and 53 μg C l^-1 (day), and the bacterial biomass was 89 μg C l^-1 (night) and 55 μg C l^-1 (day). Coccoid cyanobacteria were not studied separately, but their biomass was included in the group of small algae because they also contain chlorophyll a. These values were high, even though the biomass contributed by algae (smaller than 10 μm) and nanoflagellates partially overlapped because both autotrophic and heterotrophic nanoflagellates were counted. Rimostrombidium sp. feeds primarily on small algae (including coccoid cyanobacteria), but it can also ingest bacteria (Cleven 2004; Sonntag et al. 2006; Peštová et al. 2008). When only algae were considered, the food resources considerably exceeded the satiating food level for oligotrichs and aloricate choreotrichs at 100 μg C l^-1 (Montagnes 1996) or 150 μg C l^-1 (Gismervik 2005). This was also true if only part of the small algae (e.g. only cyanobacteria) was available as food for Rimostrombidium sp., which is likely because the organism studied was small (mean diameter of fixed protoplasts was about 12 μm). In addition, Rimostrombidium sp. could also feed on bacteria. The second organism studied in September 2011 was Tintinnidium sp. Tintinnids can ingest particles up to one third of their lorica oral diameter, but they prefer food particles sized about one fifth of the lorica oral diameter (reviewed by Montagnes 2013). In this study, the lorica oral diameter of Tintinnidium sp. was 12.5±1.4 μm (mean±SD), thus its food resource were nanoflagellates (the mean diameter of which was about 3 μm) and the smallest algae. According to Heinbokel (1978), Verity (1986), and Montagnes (2013), tintinnids require 50–80 μg C l^-1 for the saturated growth. Thus, the food demands of Tintinnidium sp. were satisfied by nanoflagellates and the small share of algae smaller than 10 μm. In conclusion, both Rimostrombidium sp. and Tintinnidium sp. were food-satiated, both in the 20% dilution and in the whole water, where food resources were more abundant. The excess of available food allowed them to achieve very high growth rates.

Cyclidium sp. was studied twice, in June 2012 and in June 2013. According to Fenchel (1986), Dolan & Coats (1991), Posch et al. (2001), Jezbera et al. (2005), and Bautista-Reyes & Macek (2012), organisms from the genus Cyclidium feed on bacteria and cyanobacteria. According to the functional response of Cyclidium glaucoma, bacterial abundance over 3.0 x 10⁷ cells ml^-1 (Posch et al. 2001) or 8.6 x 10⁷ cells ml^-1 (Taylor 1978; Fenchel 1980) saturates its food demands. During the experiments performed in June 2012, bacterial abundances in the most diluted treatment (20%) were 0.79 x 10⁷ cells ml^-1 (day) and 0.91 x 10⁷ cells ml^-1 (night, Table 1), thus below the saturated food level. The bacterial abundance in whole water was higher by 6–10% (depending on experiment, Table 1), but it was still below the satiation level. It is difficult to verify whether Cyclidium sp. was food-limited on that occasion or whether it was satiated due to the presence of coccoid cyanobacteria (additional food resource that was not counted, however, their abundance was few times lower than that of bacteria) or due to the exploitation of food patches, because the experiment performed around midnight was unsuccessful and that performed around noon resulted only in a rough estimation of the growth rate (0.08–0.21 h^-1, Fig. 1E). It should be admitted that data on coccoid cyanobacteria would not be conclusive, because not all species or even ecotypes from the genus Cyclidium feed efficiently on the former (Šimek et al. 1995; Peštová et al. 2008). During the experiments performed in June 2013, bacterial abundances were higher and exceeded 10⁷ cells ml^-1, both in the 20% dilution and in the whole water (1.07–1.64 x 10⁷ cells ml^-1, Table 1). However, they were slightly lower than the satiation level (3.0 x 10⁷ cells ml^-1, Posch et al. 2001). In June 2013, the experimental procedure caused a decrease in the bacterial abundance between the whole water and the 20% dilution (9–16%, Table 1). Theoretically, Cyclidium sp. growth could be limited in the whole water and in the dilutions (more likely). However, the high growth rates observed on this occasion (discussed above) preclude this possibility. Cyclidium sp. could be satiated due to the presence of coccoid cyanobacteria (not counted) and could exploit food patches and maintain the maximum growth rates despite the fact that the mean abundance of food particles calculated for the larger volume was not satiating. Unfortunately, coccoid cyanobacteria were not counted separately. It is also possible that the organism studied requires less food for saturated growth. The Cyclidium sp. studied by Boenigk & Novarino (2004) grew very fast (0.16 h^-1) when cultured with a nominally food-limited abundance of bacteria (2 x 10⁷ cells ml^-1).

The second assumption of the dilution method is that grazing pressure depends linearly on the dilution factor. Predatory ciliates, which could prey on the organisms studied, were observed during all the experiments. Depending on the occasion, their abundance in whole water ranged from 2.7 to 13.0 cells ml^-1. During all the experiments, I observed many specimens from the genera Askenasia and Monodinium. Ciliates from the genera Lacrymaria, Didinium, and Mesodinium were observed on two of the three dates. Litonotus sp. and others were less frequent. These ciliates are predatory (Gast 1985; Pratt & Cairns 1985; Dolan & Coats 1991; Nielsen & Kiørboe 1994; Mironova et al. 2012). Dilution of predatory ciliates in subsequent dilutions was accompanied by a linear increase in the apparent growth rates of the organisms studied (Fig. 1). This was directly demonstrated in the experiments with Rimostrombidium sp. (Fig. 1A–B), Cyclidium sp. (June 2013, Fig. 1G–H), and in one experiment with Tintinnidium sp. (Fig. 1D). As described above, in the experiment performed with Cyclidium sp. around noon in June 2012 (Fig. 1E), the relaxation of grazing pressure was incomplete, which resulted in the underestimated growth rate, whereas in the corresponding experiment performed around midnight (Fig. 1F), the relaxation of grazing pressure was unsuccessful. Higher grazing pressure at night could be explained by the abundance of predatory ciliates in both experiments. In the experiment performed around noon, their abundance was lower (3.7 cells ml^-1 in whole water) than around midnight (7.0 cells ml^-1 in whole water). Unreleased grazing pressure around midnight suggested that the abundance of predatory ciliates in the 20% dilution was still high enough to exert top-down control on Cyclidium sp. In conclusion, the second assumption of the dilution method was positively verified in the experiments that were carried out in September 2011 and in June 2013. This assumption was violated in June 2012 and on that occasion the growth rate was analyzed with the additional method assuming the so-called “saturated feeding effect” (see Results and Fig. 1E) and is reported as a range.

As illustrated in Fig. 1, the apparent growth rates observed in whole water (dilution 100%) were typically negative. One could expect values close to zero, that is, balanced growth and mortality rates. At the end of each experiment, an additional environmental sample was taken to compare changes in the whole experimental water with that at the study site. In 6 out of 8 cases, apparent growth rates in whole water in bottles were lower than in the environment; however, the difference was not statistically significant (Wilcoxon’s signed rank test). Theoretically, negative growth rates in less diluted water could be caused by trophic cascade effects, because whole water was pre-screened through 100-μm mesh nylon gauze and during incubation metazooplankton smaller than 100 μm was released from grazing pressure and could graze more strongly on ciliates (e.g. Klaas et al. 2008; First et al. 2009; Calbet & Saiz 2013). This would result in the overestimated slope of the grazing curve and grazing rates. However, it is of lesser importance in this study, because only growth rates and not grazing rates were taken into account. In most cases, the specific growth rates estimated (Fig. 1, y-intercept) were comparable to apparent growth rates observed in the 20% dilution.

The dilution method relaxes grazing pressure, but does not remove other mortality agents such as (i) viral lysis (McManus 1993), (ii) toxins excreted by other protists (Verity 1986, Leakey et al. 1994), or (iii) parasites (Verity 1986; Coats & Bachvaroff 2013; Coats et al. 2014). However, these mortality agents seem to be far less important than grazing, because successful elimination of grazing pressure generally results in very high growth rates close to the intrinsic rate of increase (e.g. Nielsen & Kiørboe 1994; Franzé & Lavrentyev 2014; this study).