Effect of disinfection with peracetic acid on the microbial community of a seawater aquaculture recirculation system for Pacific white shrimp (Litopenaeus vannamei)

During the past few years, the production of Pacific white shrimp (Litopenaeus vannamei) has increased in European countries due to the possibility of keeping these animals, independent from natural sea water, on inland farms in recirculating aquaculture systems (RAS) (Bauer et al., 2018). The primary advantages of keeping shrimps in RAS are a reduced consumption of water, a low environmental impact and the possibility of high stocking densities and therefore high productivity. Challenges that occur when tropical shrimp are kept in northern Europe are maintaining a high water temperature and an appropriate salinity. Pacific white shrimp are very tolerant against low and moderate salinity levels (Bray, Lawrence, & Leungtrujillo, 1994; Jayasankar et al., 2009) so that RAS can be operated at 10 to 13 ‰ salinity, which lowers the cost for artificial sea salt and reduces the pollution of the wastewater caused by high salinity. In northern European countries, waste heat from biogas plants can be used for heating the water in RAS so that the production of shrimp in these systems is a sustainable option for the local production of high-quality food. Due to the current development, fresh marine shrimp can be offered to customers in areas far away from the sea. Received: 17 April 2020 | Revised: 27 May 2020 | Accepted: 28 May 2020 DOI: 10.1111/jfd.13207


| INTRODUC TI ON
During the past few years, the production of Pacific white shrimp (Litopenaeus vannamei) has increased in European countries due to the possibility of keeping these animals, independent from natural sea water, on inland farms in recirculating aquaculture systems (RAS) (Bauer et al., 2018). The primary advantages of keeping shrimps in RAS are a reduced consumption of water, a low environmental impact and the possibility of high stocking densities and therefore high productivity. Challenges that occur when tropical shrimp are kept in northern Europe are maintaining a high water temperature and an appropriate salinity. Pacific white shrimp are very tolerant against low and moderate salinity levels (Bray, Lawrence, & Leungtrujillo, 1994;Jayasankar et al., 2009) so that RAS can be operated at 10 to 13 ‰ salinity, which lowers the cost for artificial sea salt and reduces the pollution of the wastewater caused by high salinity. In northern European countries, waste heat from biogas plants can be used for heating the water in RAS so that the production of shrimp in these systems is a sustainable option for the local production of high-quality food. Due to the current development, fresh marine shrimp can be offered to customers in areas far away from the sea.
Nevertheless, in intensive shrimp aquaculture, disease outbreaks might occur due to infections with viral, parasitic or bacterial pathogens (Austin & Zhang, 2006;Bauer et al., 2018;LeRoux et al., 2015;Lotz, 1997;Soto-Rodriguez, Gomez-Gil, & Lozano, 2010). By choosing specific pathogen-free post-larvae for stocking these systems, the infection risk for viral pathogens can be reduced. However, due to the high stocking densities and the large amount of organic material in recirculating water from non-utilized feed and faeces, very high numbers of bacteria in water and on all surfaces of the system might occur (Bauer et al., 2018). Additionally, the accumulation of micro-particles in the water leads to an increased growth of bacteria on these solids (Wold et al., 2014). Most bacteria present in RAS have only a low pathogenic potential for shrimp but might cause heavy losses under suboptimal farming conditions. Bacteria in the system attach to every possible surface, including the walls of the tanks, the filters and materials inside the filters and also to the surface of the animals. Bacteria from the surrounding water are colonizing the carapaces and the gills of shrimp. When these bacteria are taken up by shrimp, they may even colonize the intestine. For fish, it is known that a dense population of bacteria from the physiological microflora on mucosal surfaces acts as an important component of the external infection barrier (Balcazar et al., 2006) and the same might be true for the surface of shrimp. Also in shrimp, a diverse and stable microflora might therefore prevent bacterial infections. Nevertheless, within this physiological bacterial microflora, also potentially harmful organisms might occur. Under suboptimal conditions, the colonization of surfaces with potentially pathogenic bacteria can be the starting point of bacterial infections (Abraham, Sharon, & Ofek, 1999). In marine aquaculture facilities, especially high diversity of Vibrio spp. can be found. Among these bacteria, there are also potential pathogenic species that might induce disease (Bauer et al., 2018).
In RAS for fish production, different approaches are used to reduce the risk of bacterial infections. One approach relies on removing bacteria from the water to achieve a reduction in bacterial numbers on fish mucosa as well. With this aim in mind, physical and chemical methods are used to reduce the total amount of bacteria in the water.
One of the most frequently used chemical substances for reducing possible pathogens in aquaculture systems is peracetic acid (PAA), which is considered as highly efficient against pathogens and environmentally friendly (Kitis, 2004). Due to its low molecular weight, PAA is able to pass through outer membranes of bacteria and react with internal cell components leading to cell damage and release of cellular components (Finnegan et al., 2010). Thus, the substance is able to reduce all bacteria in the system in a very non-specific manner. PAA was used in experimental studies as a prophylactic measure over long-term periods to reduce the total amount of bacteria generally or as an alternative treatment for specific bacterial pathogens, like Flavobacterium spp., Aeromonas salmonicida or Yersinia ruckeri to reduce antibiotic use (Liu, Pedersen, Straus, Kloas, & Meinelt, 2017;Marchand et al., 2012;Meinelt et al., 2015). Most research on PAA was carried out in freshwater systems, and it is known that higher salinity levels accelerate the degradation of PAA. Obviously, especially due to the effect of ions like Ca 2+ or Mg 2+ , in sea water, the degradation of PAA is faster than in pure NaCl-solutions. Furthermore, it was assumed that a lower Mg 2+ and Ca 2+ ratio compared with the Na + and K + ratio forces a faster degradation (Liu, Steinberg, Straus, Pedersen, & Meinelt, 2014).
In aquaculture, there are two different strategies used to apply PAA products: pulse applications, short-term treatments using high concentrations of PAA, up to several times per day with concentrations of 1-2 mg PAA/L or continuous applications of concentrations below 0.2 mg PAA/L . In fish aquaculture systems, it could be shown that pulse application inhibits biofilm formation in the tanks, whereas continuous application of PAA promotes biofilm formation . By pulse application therefore, the amount of bacteria in the system could be reduced and stable biofilms were prevented. On the other hand, it is known that stable biofilms are important for the stability of the whole system and that chemical or physical treatments influence not only the amount but also the composition of the bacterial microflora in the system (Blancheton, Attramadal, Michaud, d'Orbcastel, & Vadstein, 2013;Gullian, Espinosa-Faller, Nunez, & Lopez-Barahona, 2012;Wietz, Hall, & Hoj, 2009). This might lead to a selection of bacteria that are more resistant to the used substance or method and therefore might destabilize the physiological microflora. Also, such unstable microflora favours colonization by fast-growing "r-strategist" populations, which comprises most pathogenic bacteria Skjermo, Salvesen, Oie, Olsen, & Vadstein, 1997;Skjermo & Vadstein, 1999). Another approach to prevent clinically relevant bacterial infections is therefore the stabilization of the physiological microflora by not disturbing the composition of the bacterial flora. This approach aims at a stable and diverse microflora where different bacterial species occupy practically all available ecological niches so that pathogenic bacteria are impeded from asserting themselves within the system. It could be shown that in a stable microbial environment characterized by high amounts of slow-growing K-strategists and low amounts of rapid growing opportunistic pathogenic r-strategists, fish survival rates were increased Skjermo et al., 1997). It can be assumed that this is transferable to shrimp aquaculture as well.
Most research activities were performed in aquaculture systems for fish. However, no systematic evaluation is available on the effect of PAA in seawater shrimp aquaculture. In the present study, therefore, the effect of different concentrations of PAA on the microflora in RAS for L. vannamei was tested. As the microflora should be kept as stable as possible despite the PAA treatment, a continuous application of PAA was performed.

| Examinations on toxic levels of PAA for L. vannamei
To determine the PAA concentrations for use in the main experiment, four pretests were performed to examine toxic levels of PAA for L. vannamei. In a first experiment, 59-day-old animals (n = 125) and, in a second experiment, 21-day-old animals (n = 155) were used.
Both experiments were performed in 12 plastic aquaria with a water volume of two litres each. In the first experiment, 10 shrimps and, in the second experiment, 12 shrimps were placed in each of the aquaria. In the first experiment, the water was adjusted to a salinity of 10‰ and a temperature of 30°C, and in the second experiment, to a salinity of 30‰ and a temperature of 30°C. Each of the aquaria was aerated and equipped with an identical amount of plastic fibre as holding material for the shrimp. In both experiments, PAA at concentrations of 1 mg/L, 0.1 mg/L or 0.01 mg/L was put in three of the aquaria in order to perform the test in triplicate. In every experiment, three control tanks remained untreated. After adding PAA to the water, the shrimps were kept in the aquaria for a period of 10 hr.

| Influence of different concentrations of PAA on bacterial growth
In order to test the effectiveness of PAA against bacteria, one isolate each of Aeromonas hydrophila, Pseudomonas fluorescens and Vibrio parilis was used. All isolates were incubated on three Columbia sheep blood agar plates (Oxoid) for 24 hr at 25°C. The bacteria were dissolved homogeneously in veal infusion broth, and the optical density of each suspension was adjusted to 0.2. The bacterial suspensions were aliquoted and mixed with PAA concentrations of 1 mg/L, 0.1 mg/L and 0.01 mg/L, respectively. PAA was diluted in a 0.9% saline solution (NaCl). As negative control, veal infusion broth mixed with the same concentrations of PAA was used. After a 24-hr incubation period at 25°C, a 10-step dilution series with dilution steps of 1:10 each was made with the incubated culture media. Each of these dilutions was spread out in triplicate on Columbia sheep blood agar plates. Bacterial colonies were counted after a 48-hr incubation period at 25°C.
The experiment was repeated with Vibrio parilis incubated with PAA concentrations of 4, 40, 400, 4,000 and 40,000 mg/L under the same conditions.

| Recirculating aquaculture systems
The main experiment was performed in four laboratory-scaled RAS.
Each RAS consisted of three keeping tanks with 100 L volume each and one reservoir tank for all technical devices with a volume of 150 L. In the reservoir tank with a pump that ensured a constant water circulation in the entire RAS, filter material, a skimmer and aerators were placed. All four RAS were maintained with the addition of food for six weeks before shrimp were introduced into the holding tanks. Water temperature was adjusted to 30°C and water salinity to 30‰.

| Shrimps
The RAS were stocked with post-larvae from L. vannamei (PL 12, approx. 12 days old), and 230 shrimp were added to each holding tank.
After a four-week acclimatization period, at the start of the experiment, the shrimp were approximately 39 days old with a mean body weight of 0.14 ± 0.18 g and a mean body length of 2.37 ± 1.37 cm.

Automatic feeders (Eheim GmbH & C. KG) installed in the individual
tanks allowed the shrimp to be fed on a regular basis. Up until day 9, the daily amount of food was 10.5 g per RAS. This was reduced after day 9, and for the rest of the duration of the experiment, only a total amount of 4.5 g food was dispensed daily per RAS.

| Application of PAA
The PAA product Wofasteril E400 (Kesla Pharma Wolfen GmbH) was used in the experiment. Three of the RAS were treated with different concentrations of PAA and the fourth RAS served as untreated control. According to the results from the pretest, final PAA concentrations of 0.1 mg/L, 1 mg/L and 10 mg/L should be achieved. However, to avoid strong concentration peaks, the cor-

| Experimental design
During the experiment, samples were taken six days before adding PAA and two, nine, 29 and 56 days after commencing PAA application. Chemical water parameters (pH, NH + 4 , NO − 2 , NO 3 ) were measured daily for the first 2 weeks of the experiment and at each of the additional sampling time-points. To assess the effect of PAA on the bacterial microflora in the RAS at each sampling time-point, water samples from each tank, swabs from the biofilm of the surface of each tank and swabs from the transition from the carapax to the abdominal segments of three shrimp specimens per tank were taken.
Additionally, swabs from the abdominal cavity of three shrimp specimens per tank were taken at the first and the last sampling timepoints. For this, the shrimp were killed individually in a 1 L plastic aquarium with iced water at a temperature of 0 ± 1°C. The ratio of ice to water was adjusted so that there was a clear excess of ice (approx. 3:1), but at the same time the individual shrimp were completely surrounded by iceless water. Shrimp were left in iced water for at least two minutes. The abdominal cavity was then opened with a sterile scalpel and sampled with a swab.
To investigate the total amount of bacteria in the water samples, dilution series with sterilized water of the same salinity were prepared from undiluted samples to a dilution level of 10 -5 and each dilution was spread on two sheep blood agar plates containing 30‰ artificial sea salt and incubated at 25°C for 48 hr. Colony-forming units (CFU) on the plates were counted after 12 and 48 hr, and the amount of CFU per ml of tank water was calculated. The amount of morphologically different CFUs was described semi-quantitatively (low: +; up to 10 colonies/plate, moderate: ++; 10-50 colonies/plate), high: +++; >50 colonies/plate), and all morphologically different colonies were subcultured on sheep blood agars containing 30‰ artificial sea salt. After a 48-hr incubation period at 25°C, subcultures were stored at −80°C in 2 ml of veal infusion broth until further analysis for identification of the bacterial species.
The swab samples from tanks surfaces, the carapaces and the abdominal cavities of shrimp specimens were plated on blood agar plates containing 30‰ artificial sea salt. The plates were cultivated at 25°C for a total of five days. Every day, the plates were checked for bacterial growth. The amount of bacterial colonies was assessed semi-quantitatively. On average, subcultures of bacteria were prepared after one day of incubation. Afterwards, from macroscopically different colonies, one colony was picked with a loop and plated and then fractionated on a separate blood agar plate. One day later, the subcultures were checked for purity and were stored at −80°C in 2 ml of veal infusion broth until further analysis for the bacterial species.

| Identification of bacteria
For species identification, pure cultures of the isolates were identified by 16S rRNA gene sequencing. For this, DNA was extracted by adding one colony per isolate to 500 µl of AF-buffer (Qiagen GmbH, Hilden, Germany), incubation at 92°C for 15 min while shaking and centrifugation at 13,000 g for 5 min. DNA concentrations were measured using spectrophotometry (NanoDrop ND-1000 Lab, Peqlab Biotechnologie GmbH) and adjusted to a concentration of 10 ng/μl with PCR grade water (Thermo Fisher Scientific Inc.). The V1-V9 region of the 16 S rRNA-encoding gene was amplified using forward and reverse primers designed by Jiang et al. (Jiang, Gao, Xu, Ye, & Zhou, 2011 for 30 s, 72°C for 60 s, and an extension step at 72°C for 7 min.
Sequencing of PCR products was performed by LCG Genomics GmbH, Berlin, Germany.

| Calculation of bacterial population diversity
The composition of the bacterial microflora was analysed using the following ecological terms: Prevalence-number of samples in which a particular bacterial species could be found divided by the number of samples examined expressed as a percentage (%).
Mean intensity-number of a particular bacterial species found in a sample divided by the number of samples in which this particular bacterial species could be found. An arbitrary scale was used for quantifying the bacterial species: 0 = absent, 1 = low amount, 2 = moderate amount, and 3 = high amount.
Mean abundance-total amount of a particular bacterial species in a particular sample divided by the number of samples examined; mean abundance is equivalent to mean intensity multiplied by prevalence.
The diversity of the bacterial community was evaluated by calculating the Shannon-Wiener index of diversity for individual samples (H′ = −∑(p i lnp i ) where p i is the relative intensity of bacterial amount i).

| Statistical analysis
The data were statistically analysed using the computer program SigmaPlot 12. When the data were normally distributed (tested with a Shapiro-Wilk test), an ANOVA on ranks was performed, followed by an all-pairwise multiple comparison procedure. The Tukey test was used for comparing groups with an equal number of data, and Dunn's method was used for comparing groups with an unequal number of data. When the test for normality failed, the Mann-Whitney ranksum test was used for comparing the data. Differences between tested data sets were considered significant at a probability of error of p < .05. Principal component analysis (PCA) was performed for the data on the microbial community in respect to their relationship between different treatment groups and different sample types.

| Toxicity of PAA for L. vannamei
None of the shrimp died in any of the four test groups treated with PAA concentrations of 0.1 up to 100 mg/L. In addition, no changes in the behaviour of the animals could be recognized. It could be concluded that for a 12-hr period, PAA concentrations of up to 100 mg/L are tolerable for L. vannamei.

| Influence of PAA on bacterial growth under laboratory conditions
In the first experiment, similar results were obtained for all three tested bacterial isolates. In the control group without PAA, 10 8 CFU/ ml were counted for all three bacterial isolates. No reduction in bacterial growth was observed when PAA at a concentration of 0.1 ml/L was added to the bacterial suspensions. A reduction to 10 7 or 10 4 CFU/ml was seen when the bacterial suspensions were treated with a PAA concentration of 1 mg/L or 10 mg/L.
In the second experiment with Vibrio parilis, the highest amount of bacteria could be detected in the untreated control group (1.05 × 10 9 CfU). Treatment with a PAA concentration of 4 mg/L only led to a very slight reduction in bacterial growth (1.02 × 10 9 CfU).
However, after a treatment with a PAA concentration of 40 mg/L a significant reduction in bacterial growth to 9.55 × 10 8 CfU was observed. No bacterial growth at all could be detected at all PAA concentrations above 40 mg/L.

| Animals
In all four RAS, the shrimp grew and gained weight over the experimental period. The animals treated with a PAA concentration of 0.1 mg/L were statistically significantly smaller at the end of the experiment compared with the animals in the control group and the group treated with a PAA concentration of 1 mg/L. Animals from the group treated with a PAA concentration of 0.1 mg/L were also lighter in weight compared with those from the control group, and also, their antenna length was shorter compared with both other groups ( Figure 1). The relation between antenna length and body length at day 56 was statistically significantly smaller for those animals treated with a PAA concentration of 1 mg/L compared with those animals in the control group.

| Water chemistry
The concentration of the nitrogen compounds, ammonia and nitrite in recirculating water remained relatively stable in the untreated control RAS and in the RAS treated with a PAA concentration of 0.1 mg/L.
In the RAS treated with PAA concentrations of 1 mg/L and 10 mg/L, two days after commencing PAA application, a statistically significant (p = <.05) increase in the concentration of ammonia and nitrite could be measured. To avoid negative effects on the shrimp, in all four RAS, the water was changed completely at this time-point and the amount of feed given was reduced for the rest of the experiment from a daily amount of 10.5 g per RAS to one of 4.5 g per RAS. At this time-point, the PAA addition was suspended to the RAS treated with a PAA concentration of 10 mg/L and the RAS was removed from the experiment in order to avoid shrimp losses. At day 29 post-start of PAA addition, statistically significantly higher concentrations of ammonia and nitrite were again measured in the RAS treated with 1 mg/L (p = <.05) compared to the control RAS and the RAS treated with a PAA concentration of 0.1 mg/L ( Figure 2). As expected in a running system without a denitrification unit, the nitrate levels increased during the experimental period in all RAS, whereas this increase was highest in the control group. As similarly described for other seawater RAS as well, the pH value decreased continuously during the course of the experiment and the most highly significant changes were detected in the RAS treated with PAA at a concentration of 1 mg/L ( Figure 2).

| Bacterial community
A taxonomic characterization of the bacterial community was performed by cultivating bacterial species. In this analysis, a total of 91 bacterial species were isolated from all examined samples (Table 1). The greatest number of bacterial species was isolated from the carapaces of the shrimp and from water samples (n = 61).
Almost the same number of bacterial species was found in the samples from the biofilms from the tanks (n = 56) while the lowest number of bacterial species was found in the samples from the abdominal cavity of the shrimp (n = 34) ( Table 1- At the beginning of the experiment, mainly V. alginolyticus could be detected, whereas at days 29 and 56 after commencing PAA application, higher amounts of V. harveyi were isolated (Table 1).

Bacterial community in biofilms on tank surfaces
Bacterial composition in the biofilms on tank surfaces also changed in a moderate way in all four RAS during the experimental period.
Like in the water, Vibrio sp. was the most abundant genus ( Figure 6).
The large water exchange before day nine and the reduced feeding F I G U R E 1 Body weight (g), body length (cm), antenna length (cm) and relation between antenna length and body length (%) of Litopenaeus vannamei before and 56 days after continuous application of different amounts of peracetic acid (PAA). The statistically significant differences between the treatments are marked by different letters of the shrimp changed the composition of the bacterial community in the biofilms in all three remaining RAS to a greater extent. This could be seen in particular at day 9 after commencing PAA application. At this time-point, higher amounts of bacterial species, like Virgibacillus spp., were present, which could not be isolated in such large amounts at previous sampling time-points. Photobacterium spp. could only be detected in the samples from RAS treated with PAA, especially at days 29 and 56 after application and at higher percentages in the RAS treated with a PAA concentration of 1 mg/L. The largest changes in the composition of the microflora in biofilms were found in the RAS treated with a PAA concentration of 10 mg/L two days after commencing the PAA application. Then, a significant reduction in Vibrio sp. from 58% before PAA application to 20% after two days of PAA application and higher percentages of other bacterial species were found ( Figure 6). Similar to the composition in the water, the amount of V. alginolyticus decreased over time, whereas the amount of V. harveyi increased in the control RAS and in the RAS treated with a PAA concentration of 0.1 and 1 mg/L, respectively (Table 2).

Bacterial community on the carapaces of shrimp
The composition of the bacterial community on the surfaces of the carapaces of the shrimp showed the largest differences in all four RAS during the experimental period. Vibrio sp. was the most abundant genus also on the carapaces, but to a lesser percentage than in the samples from the water and the biofilms of the tank surfaces.
A change in microbial composition was seen again after the large water exchange and the reduction in food at day 9 after commencing PAA application. Unlike in the community in water and biofilms, the number of different bacterial species did not decrease, but a higher diversity was seen. In the RAS treated with a PAA concentration of 10 mg/L two days after commencing the PAA application, there was not a large change in the bacteria composition as was seen in water and biofilm samples. On the carapaces of the shrimp from the RAS treated with a PAA concentration of 1 mg/L, the number of different bacteria that could be detected was reduced significantly at both later sampling time-points, days 29 and 56 ( Figure 7).

Bacterial community in the abdominal cavity of shrimp
In total, the number of different bacterial isolates was much lower in the microflora from the abdominal cavity compared with those from the water or the surfaces of tanks or carapaces. The composition of the microflora in the abdominal cavity from the examined shrimps did not alter greatly between the different treatment groups. Nevertheless, Photobacterium spp. could not be detected at the end of the experimental period at 56 days in all three remaining RAS (Figure 8).

| Relationship between the samples from different origins in the four treatment groups
In general, in the three RAS, which were sampled for the entire 56-day experimental period, the microflora in the abdominal cavity and pH values in four seawater recirculating aquaculture systems stocked with Litopenaeus vannamei and treated with different concentrations of peracetic acid (PAA). Shown are measurements before PAA application and 2, 9, 29 and 56 days after continuous application of PAA. The statistically significant differences between the treatments are marked by different letters TA B L E 1 Mean abundances of bacterial species isolated from tank water from four seawater RAS treated with different concentrations of PAA (0 mg/L = control; 0.1 mg/L; 1 mg/L; 10 mg/L) before start of the application and 2, 9, 29 and 56 days after start of continuous PAA application

| D ISCUSS I ON
In the present study, the effects of PAA at different concentrations on the bacterial microflora were analysed in seawater RAS for Pacific white shrimp, Litopenaeus vannamei. Therefore, the bacterial composition in the water, on the biofilms of tank surfaces and on the carapaces of shrimp and in the abdominal cavity was investigated.
Additionally, the performance of the shrimps and the chemical water quality were analysed.
PAA is widely used in aquaculture systems for fish production either prophylactically as a general water disinfectant with the aim of keeping the animals in the systems healthy  or, in the case of a disease outbreak due to bacterial or parasitic infections, to reduce these pathogens. As there are many regulations regarding the use of therapeutic agents for treating animals for human consumption and especially considering a reduction in the use of antibiotics, alternative strategies to maintain the health of aquatic animals in recirculation aquaculture systems are becoming increasingly important. Studies on the application of PAA documented that the substance is environmentally friendly because it degrades to biodegradable residues (Kitis, 2004). PAA can reduce the growth of fish-pathogenic bacteria in vitro at a dos- Nonetheless, PAA could also be an interesting substance for RAS for the cultivation of crustaceans like shrimp in order to maintain optimal water quality or to reduce the abundance of pathogens. For example, it could be shown that adding 10 mg PAA/L can effectively suppress the pathogenicity of the oomycete, Aphanomyces astaci, the pathogenic agent in crayfish plague (Jussila, Makkonen, & Kokko, 2011). For fish, PAA seems to be welfare-friendly, as a true habituation of fish to the substance associated with a decrease in the cortisol response after repeated exposure to PAA has been seen (Gesto et al., 2018;  after seven-day treatment with 2 mg PAA/L. These alterations were even more pronounced when the crayfish were treated with 10 mg PAA/L for an equal duration (Chupani, Zuskova, Stara, Velisek, & Kouba, 2016). However, none of the crayfish died, and after a recovery period of seven days, gill morphology returned to normal levels (Chupani et al., 2016). This is in accordance with the results of the preliminary experiment of the present study where all shrimp survived the exposure to PAA concentrations even up to 100 mg PAA/L. No losses were seen also in the main experiment; however, the RAS treated with a PAA concentration of 10 mg/L had to be removed from the experiment already two days after commencing In this study, PAA was applied to seawater RAS with a salinity of 30‰. It is known that the degradation of PAA is related to the salinity of water and that higher salinities lead to a faster degradation (Liu et al., 2014). The effect of a PAA prophylaxis or treatments in seawater RAS therefore might be reduced compared with freshwater RAS, and the maintenance of an effective concentration might be a challenge (Liu et al., 2014). For seawater RAS, multiple applications or additional PAA dosages are thus recommended to maintain effective concentrations (Liu et al., 2014). To avoid potential harm to the biofilter in RAS during PAA application, it is nevertheless suggested to apply PAA only at reduced flow rates .
Taking this into account, in the present study, PAA was applied to the RAS at normal flow rates, not as pulse application but continuously.
Thus, stable PAA concentrations in the RAS should be maintained, and especially, short-term high PAA concentrations in the biofilter should be avoided. The addition of PAA at the higher dosage of 10 mg/L in the present study interfered with the removal of nitrogenous compounds, which suggests that the bacteria of the biofilter were affected. Because this effect was not anticipated, the biofilter was not sampled and this assumption could not be confirmed. For freshwater RAS, it could be shown that a continuous application did not lead to PAA accumulation and it was assumed that a fast degradation of the substance was attenuated by microbial adaption ). An enhanced biofilm formation was seen visually, and it was suggested that acetic acid and acetate as active ingredients of PAA acted as an easy degradable dissolved organic substance, which could promote the growth of especially heterotrophic bacteria. Nevertheless, an examination of the composition of the microflora was not performed . Our former studies on the microbial community in freshwater RAS showed that  et al., 2016, 2018). The maintenance of stable chemical and microbiological water conditions in the RAS was therefore another reason for applying the PAA continuously. Regarding the chemical water parameters, in freshwater RAS, it could be seen that the pH value increased when PAA was continuously applied . In sea- bacteria. This is in contrast to previous findings in freshwater RAS treated continuously with PAA in which nitrite and nitrate levels were not higher than in a control RAS .
In the present study, a reduction in the abundance of bacteria in water, as was expected by applying PAA at higher concentrations to the RAS, was not achieved, neither with 1 mg/L nor with 10 mg/L.
The total number of bacteria in the water even increased significantly, especially two days after applying 1 mg/L. Additionally, a shift in the bacterial community seemed to be provoked by applying PAA at concentrations of 1 or 10 mg/L to the RAS. Nevertheless, the bacterial community was analysed by cultural techniques followed by molecular biological identification. By culturing, it is known that Columns depict the composition of the bacterial microflora before the start of the application and 2, 9, 29 and 56 days after commencing a continuous PAA application. Shown are the bacterial species that could be detected at least at one sampling time-point in one of the samples with an abundance of 10% or more. All other bacterial species are summarized under the heading "others." not all bacterial species might be detected as not all are growing on agar plates. This has to be considered when interpreting the data.
As mentioned previously, in freshwater RAS, an increased biofilm formation was described after continuous application of PAA . In the present study, the thickness of biofilms was not measured but an enhanced biofilm formation was visually not observed in the PAA treated RAS. The higher amounts of bacteria in tank water nevertheless indicated an enhanced reproduction of bacteria that were probably able to use PAA degradation products as a nutritional source. It could be seen that the composition of the bacteria within the RAS changed in relation to the amount of PAA applied to the tanks and that the composition of the biofilms at tank surfaces and carapaces of shrimp changed differently.
Especially in the RAS treated with a PAA concentration of 10 mg/L, a significant reduction in the amount of Vibrio sp. to 20% was recognized already two days after commencing the application. In contrast to this, the mean amount of Vibrio sp. was around 40%-60% of all detected bacteria and in all kinds of samples. This mean value is in accordance with findings in other seawater RAS and also in the oceans where Vibrio sp. represent the prevailing organism in the physiological microflora. The amount of Vibrio sp. within the physiological microflora in marine and brackish habitats can reach up to 40% (Bauer et al., 2018;Jeyasanta, Lilly, & Patterson, 2017;Urakawa & Rivera, 2006). A significant reduction in the abundance of these organisms in a seawater RAS running at 30‰ therefore can be regarded as a sign of destabilization of the system. By adding PAA at a concentration of 1 mg/L to the RAS, also changes in the bacterial composition occurred that also indicated a destabilization of the microflora in the RAS. Such unsteady environments which offer unexploited nutrient resources can favour fast-growing bacteria with an r-selection strategy. To avoid such unstable conditions in the water, the importance of microbial matured water in RAS was underlined by different authors Skjermo et al., 1997;Skjermo & Vadstein, 1999). Stable water conditions most likely favour the colonization of recirculating water and biofilms by bacteria with a K-selection strategy. K-selection mainly occurs in communities that are close to the carrying capacity (CC) of a system, where the CC is the number of bacteria that can be sustained in the system over a long period of time (Attramadal et al., 2016). According to this, the selection pressure of bacteria is used to prevent the proliferation of opportunistic bacteria with an r-selection strategy by filling all possible niches in the system with non-opportunistic bacteria. In this respect, matured water colonized by these non-opportunistic bacteria with a K-selection strategy would protect fish from bacterial diseases caused by opportunistic fast replicating bacteria with r-selection strategy (Skjermo et al., 1997).
It can be assumed that these findings are relevant for RAS stocked with shrimp as well. In the present study, the microbial com- Nevertheless, in the current study, differences in the microbial community were not only seen due to PAA application but also due to a large water exchange and reduced amount of food at day nine after commencing PAA application. This clearly indicates that management measures are of great importance for maintaining optimal keeping conditions in aquaculture systems and that these measures are more suitable for prophylaxis treatment of disease outbreaks than water disinfection.
In conclusion, using PAA in seawater aquaculture systems affected the microflora in the water and on biofilms depending on the used concentration. The highest continuously applied PAA concentration tested, at 10 mg/L, led to significant changes in chemical water parameters, in particular a drop in the pH value and to high levels of ammonia and nitrite, already after two days of application.
A continuous application of PAA at a concentration of 1 mg/L, which is often used in aquaculture systems for fish, also led to increased levels of ammonia and nitrite within two days of application and to a significant increase in the amount of bacteria in water. In the preliminary experiment, it could be demonstrated that PAA at a concentration of 1 mg/L was able to significantly reduce bacterial growth of different species in vitro. Therefore, it can be assumed that the bacterial number in the water increased in the aquaculture systems because heterotrophic bacteria were able to use degradation products of PAA as a nutritional source. There were indications that the welfare of the shrimps was affected because the relation between body length and antennae length was significantly smaller than in those shrimps in the control RAS. In the RAS treated with a PAA concentration of 0.1 mg/L, the chemical and microbiological parameters were comparable to those of the untreated control RAS or even more stable during the experimental period. In this RAS, a reduction in the total amount of bacteria was achieved by decreasing the amount of food given and by exchanging the water, but obviously not by the effect of PAA. Applying PAA continuously at concentrations of 0.1 mg/L therefore seems not to be effective in seawater aquaculture systems for shrimp, neither for prophylactic use nor as an alternative treatment for specific pathogens. Continuous application of 10 mg/L also is not recommendable for use in seawater shrimp aquaculture because this had a strong impact on chemical water parameters. Applying 1 mg/L PAA in seawater shrimp aquaculture might be an alternative in cases of disease outbreaks due to bacterial F I G U R E 1 0 Shannon-Wiener indices of diversity describing the composition of the bacterial microflora from four seawater recirculating aquaculture systems (RAS) treated with different concentrations of peracetic acid (PAA) (0 mg/L = control; 0.1 mg/L; 1 mg/L; 10 mg/L).
Depicted are index values from the recirculating water, the biofilms of tank surfaces and from shrimp carapaces, and from the abdominal cavity of shrimp before the start of the application and 2, 9, 29 and 56 days after commencing continuous PAA application. In the RAS treated with a PAA concentration of 10 mg/L, samples were collected only 2 days after commencing continuous PAA application and the abdominal cavity of shrimps was sampled before the application of PAA. The statistically significant differences between the treatments are marked by different letters or parasitic infections for short periods of time. Nonetheless, further research on the specific effect in a running seawater system is needed. A continuous prophylactic use of PAA in this concentration is not recommendable as no bacterial reduction was achieved and the chemical and microbiological parameters fluctuated strongly due to the application, leading to a destabilization of the system.
Whether repeated pulse applications of PAA are more effective in seawater shrimp aquaculture either for prophylactic or as an alternative treatment has to be investigated further.

ACK N OWLED G EM ENT
This work was supported by the Deutsche Bundesstiftung Umwelt (German Federal Environmental Foundation) under the grant number 30575-23.

CO N FLI C T O F I NTE R E S T S
All authors declare that they have no conflict of interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The authors confirm that the data supporting the findings of this study are available within the article. Additional data are available from the corresponding author upon reasonable request.