Protozoa grazing cause changes in taxonomic structure of bacterial community

Detailed hypothesis: Protozoa could eliminate some competitive bacterial species and simultaneously help to grow up other species due to weaken competition between bacteria species.

Materials and Methods

Preliminary study - small scale experiment

At the beginning 150 ml of artificial sewage (medium) prepared according to the recipe showed in chapter II in Table 2.1 (page 30) was inoculated with bacteria community (filtration product of activated sludge from Niepołomice WWTP through 5µm nylon filter (Microporous Membrane)).

Bacteria were growing in flask mixed on mechanical agitator (Orbital Shaker PSU-10i) with constant 130 rpm speed for 2 days at 25^oC. After this period medium was ten times diluted with spring water

“Żywiec Zdrój” and spilled into six 100 ml volume Erlenmeyer flasks. Each flask was filled with 50 ml of medium with bacteria. Next, flasks were divided randomly into two experimental groups, three flasks per group. Into each flask of the treatment group dense culture of ciliates Colpidium colpoda, Spirostomum sp., Drepanomonas revoluta, Euplotes sp. and flagellates was introduced. Initial ciliates density in each flask was approximately 190 ind./ml. All flasks were constantly mixed with 130 rpm speed at laboratory temperature (approximately 21−23^oC). Every two or three days, 15 minutes after stopping the mixing, 5 ml of medium was replaced with new five times diluted artificial sewage.

Procedure of medium replacement was done under laminar air flow chamber (Telstar AH-100, Spain) to prevent contamination.

At the end of experiment, ciliates and flagellates density was estimated based on one subsample of 200 µL took from well mixed flasks diluted in 800 µL “Żywiec Zdrój” spring water and fixed with one drop

48 of Lugol with acetic acid solution (4%). Subsamples were put in 24 wells tissue culture plate (TPP, Switzerland) and counted under inverted Olympus IX 71 microscope. After 15 minutes of sedimentation approximately 40 ml of medium without bacterial aggregates from each flask was taken for chemical analysis of COD, NH4-N, NO3-N and PO43-. Additionally, MLSS from all flasks were measured. The experiment lasted 34 days.

Laboratory experiments in bioreactors – bench scale experiment

Experiment were conducted in six identical bioreactors under operational parameters detailed described in chapter II (see pages 30−31). In this experiment temperature was not strictly controlled and increased slowly from 18^oC at the beginning to 23^oC at the end of experiment. Each bioreactor was filled with 1300 ml of sterilized medium prepared according to Table 2.1 (page 31) and inoculated with bacterial community.

The bacterial community was obtained from activated sludge sample from Niepołomice WWTP (82 047 people equivalent, localized in Małopolska voivodship, Poland). The Niepołomice WWTP is effectively and stable working plant based on EBPR technology with up to 30% of industrial influent. Activated sludge sample was filtrated through 5 μm nylon filter (Microporous Membrane, UK). Filtrate with bacteria (5 ml) were growing on mixed artificial sewage for 2 days at 18°C and after that bioreactor cycle was switched on. Acclimatization of bacteria bioreactor conditions lasted a week. After acclimatization period into three bioreactors dense mix culture of ciliated protozoa: Drepanomonas revoluta, Colpidium colpoda, Euplotes sp., Blepharisma sp., Chilodonella sp. and flagellates were added. For the remaining three bioreactors the same volume of filtrate from the protozoa culture (obtained as a results of filtration through 5μm nylon filter) was added.

As the aim of this experiment was to culture activated sludge de novo. Mean MLSS value at the beginning of experiment was under detection limit. SRT was 69 days − during whole time of experiments excess sludge was not removed. The experiment was carried out 69 days. The protozoan population density, the size of the bacterial aggregates, the concentration of nitrogen and phosphorus compounds and the concentration of free-swimming bacteria in the effluent were analyzed once a week.

The experiment was conducting for 8 weeks after introduction of protozoa.

During the experiment, the biofilm began to grow on the surfaces of the bioreactor tank walls, so estimation of MLSS based on sub-samples taken only from reactors medium were unreliable. Firstly, sub-sample from reaction tank was collected and MLSS was measured. Next total dry mass value for each bioreactor was obtained by removing biofilm from reaction tank surface and mixing with suspended biomass from reactor. Subtract MLSS from total dry mass gave biofilm dry mass. As removing of biofilm required interruption of the process, biofilm was collected at the end of experiment.

Therefore, only values from the last day of experiment was taken to the statistical analysis. The MLSS and total dry mass value were obtained according to APHA (2005) standard methods guide. The MLSS

49 value in this experiment was approximately equal to total bacteria dry mass - mixed liquor volatile suspended solids (MLVSS) because non-biodegradable suspended matter was not added into bioreactors.

Sludge yield (Y) was calculated with following equation:

𝑌 =g MLSSend - g MLSSstart

g COD removed

where g MLSSstart was the total amount of MLSS present at the start of the experiment in our case = 0, g MLSSend represented the sum of MLSS and biofilm present at the end of the experiment (total dry mass). The term g CODremoved is the total amount of COD removed in the whole experiment period. To obtained CODremoved we calculated mean COD removed per day in each bioreactor based on values from 6 sampling days. Then mean COD removed per day per bioreactor was multiplied by number of experimental days.

The ciliates biomass for each species was taken from Foissner and Berger (1996) where wet mass of 1 µm³ = 1 pg with specific gravity of the protoplasm 1 (Finlay 1982). Biovolumes of rest community members were estimated by assuming geometric shapes and mean cell volume. The biomass has been transformed into a dry mass using mean biomass conversion ratio (dry weight/wet weight) 0.59 (Gates et al. 1982).

The coefficient of variation (CV) was chosen as measure of variability (stability) of the system performance because Cottingham et al. (2001) suggested that for untransformed data coefficient of variation is more preferable measure of variability than standard deviation.

The coefficient of variation for COD, Ntotal and Ptotal were calculated with following equation:

𝐶𝑉 = 𝑆𝐷 𝑋̅

where SD is standard deviation and 𝑋̅ is mean value for of all measured samples in one bioreactor.

The total mean of protozoa and metazoa dry mass, mean floc area, mean reduction rate, mean biodiversity index were calculated with following equation:

𝑚𝑒𝑎𝑛 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑝𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟 𝑋 = 𝑠𝑢𝑚 𝑜𝑓 𝑝𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟 𝑋 𝑣𝑎𝑙𝑢𝑒𝑠 𝑓𝑟𝑜𝑚 𝑎𝑙𝑙 𝑠𝑎𝑚𝑝𝑙𝑖𝑛𝑔 𝑑𝑎𝑦𝑠 𝑛𝑢𝑚𝑏𝑒𝑟𝑠 𝑜𝑓 𝑠𝑎𝑚𝑝𝑙𝑖𝑛𝑔 𝑑𝑎𝑦𝑠

Mean values were calculated for each bioreactor.

Amplification and Illumina sequencing

Total DNA was extracted from the activated sludge samples (1 ml volume) from six bioreactors in seven consecutive time points. For total DNA extraction Wizard genomic DNA purification kit (Promega) was used. Amplification and Illumina sequencing of 16S rRNA was done following

50 established protocols (Caporaso et al. 2011). The hypervariable regionV4 of bacterial and archeal 16S rRNA genes were PCR amplified in triplicates using primers 515f and 806r. The samples were indexed using a 12bp barcode added to the 5’end of the 806r primer. Two types of negative controls were included in each batch of PCR reactions: two extraction negative controls and two PCR negative controls (negative control gives information about contamination during procedure). Amplicon libraries were pooled at equimolar ratios and sequenced on an Illumina MiSeq sequencer (Illumina, USA). Raw sequences were processed using the QIIME2 software package (https://qiime2.org, Caporaso et al.

2010). The reads were demultiplexed, quality controlled and trimmed, retaining only reads with at least 75bp of consecutive high quality bases. To assign sequences to operational taxonomic units (OTUs), sequences that shared 97% or greater sequence identity were clustered and then aligned against the Greengenes database (version 13_8) using PyNAST (Caporaso et al. 2009) with default parameters set by QIIME2.

In our analysis we expressed bacteria diversity as number of Operational Taxonomic Units (OTUs) like other authors did (Hai et al. 2014, Delgado-Baquerizo et al. 2016, Fan et al. 2017). But in some analysis we also used well known in ecology Shannon-Wiener biodiversity index calculated by QIIME2 software.

51 Results

Preliminary experiment - small scale experiment

Results from pilot experiment were very promising. Effluent from Erlenmeyer’s flasks with ciliates had clearly lower concentration of N and P compounds and tend to had lower concentration of COD and higher total microorganisms dry mass (Table 4.1). Additionally, in all Erlenmeyer’s flasks flagellates were present. Density of free swimming ciliates Colpidium sp. decreased significantly with time from approximately 86 ind./ml at the beginning to 5 ind./ml at the end of experiment. Spirostomum and Drepanomonas individuals were eliminated from the community. Euplotes density was kept on stable level (approx. 20 ind./ml) during whole experimental 34 days’ period. At the end of experiment Chilodonella individuals were noticed, although it wasn’t observed in protozoa mixture added to experimental flasks. Most probably Chilodonella was present in ciliates inoculum in very low density, under detection limits.

Table 4.1. The effect of the ciliates presence on the quality of effluent from pilot scale experiment.

Mean ± SD, all values are in mg/l.

Group COD NH4-N NO3-N sum of N PO43- Dry mass Ciliates 24 ± 11 9.3 ± 1.1 7.3 ± 3.4 16.6 ± 2.7 2.5 ± 0.3 0.19 ± 0.08 Control 47 ± 11 12.1 ± 0.9 37.5 ± 12.3 49.6 ± 12.4 4.0 ± 0.2 0.05 ± 0.01

Laboratory experiments in bioreactors – bench scale experiment

In main experiment bioreactors were contaminated with small flagellates, naked amoebas and rotifers. In consequence we did not have real control group without protozoa. But bioreactors still differ in presence of ciliates and rotifers. Finally, we had three experimental groups:

1. Control (C) – 2 bioreactors only with flagellates and small naked amoebas 2. Ciliates (Ci) – 2 bioreactors with flagellates, small naked amoebas and ciliates

3. Rotifers (R) – 2 bioreactors with flagellates, small naked amoebas, ciliates and rotifers

Based on this observation we reformulated our scientific hypothesis: