The mean values of all the measured physicochemical parameters in the three lakes were within the standard limit for fish health [24,25,26,27,28,29]. In light of this, the lakes could be considered fitting for fish survival. Water quality factors such as DO, temperature, ammonia, phosphate, pH, alkalinity, hardness and clarity affect fish health in multiple ways. Each water quality factor interacts with and influences other parameters, sometimes in complex ways and what may be fatally toxic in one situation can be harmless in another. Nevertheless, we did not test many other parameters, which are important water quality factors for logistics reasons. The effect of sampling site, season and hour on the parameters is usually considerable. For instance, DO values significantly vary when the sampling site is surface water and along shorelines compared to deep water, or at the center. Our sampling was on surface water.
The oxygen requirements of fish also depend on a number of other factors including temperature, pH, and CO2 level of the water, as aforementioned, and the metabolic rate of the fish. Therefore, changes in these physicochemical parameters in the aquatic environment are primary causes of fish stress  although the health impact of such stress may depend not only on the severity of the stress, but on its duration and the fish's overall physiological status.
The mean DO values of 6.27 mg/l (Lake Hawassa), 5.10 mg/l (Lake Langano) and 3.93 mg/l (Lake Ziway) indicate relatively better aeration at lakes Hawassa and Langanoo than Ziway at least during the study period and sampling time. This might be attributable to variations in temperature and flow rate in the Lakes during the sampling season as lower temperature and good flow rate are associated with higher ODs [24, 30]. Most DO in ponds is produced during photosynthesis by aquatic plants and algae. For this reason, DO increases during daylight hours, declines during the night, and is lowest just before daybreak. DO concentrations below 5 mg/l may be harmful to fish and piping (gulping air at the surface) may be observed when DO falls below 2 mg/l. Low levels of DO are most frequently associated with hot, cloudy weather, algae die-offs, or heavy thunderstorms . Overall, poor water quality is a key factor for low fish yields. In a pond study, whereas increase in temperature and OD was correlated with tilapia growth rate, increase in conductivity and pH showed the opposite . Different fish species have different requirements for water DO concentration , and water temperature may influence fish feeding, growth and overall behavior.
Similarly, the data showed insignificant pollution by nitrogenous wastes at the sampling time although ammonia is a pollutant frequently found in aquatic ecosystems. In fish, ammonia can cause physical damage, alter its behavior such as lower swimming activity and feeding behavior, and oxidative stress response, and even cause death. Exposure to ammonia also increases fish physiological stress and recent evidence suggested that once exposed, fish suffer from reduced antioxidant defenses and thus increased oxidative tissue damage even if water quality was improved .
The only exception was phosphate level in Lake Langano which was relatively high (2.34–4.48 mg/l). The highest concentration of phosphate (4.48 mg/l) recorded for Lake Langano O’etu site (S2) was higher than the standard limit although the mean value was normal. The mean value of phosphate for Langanoo was substantially higher than that of Hawassa, and that of Ziway was much lower. In the early 1990s, it was reported that Lakes Ziway and Hawassa (then Hawassa) were phosphorus-limited, whereas Langanoo has surplus phosphorus . It appeared that Lake Langano remained phosphate surplus. In the 1960s, however, increases in phosphorus in the lower Lakes raised considerable public concern . The augment in phosphorus could be due to increased surface runoff from phosphate containing fertilizers and certain industrial wastes. Phosphate is an essential plant nutrient. However, high phosphate level can lead to eutrophication boosting plant/algae growth. These plants/algae eventually decay and cause DO depletion in the water threatening certain species of fish favoring phosphate-pollution-tolerant organisms.
Although it appeared that the physicochemical parameters of the Lakes were normal, the proportion (20%) of fish with visible health problems is not a good sign. Alternatively, it may be plausibly argued that encountering relatively lesser number of clinically diseased fish with visible pathological changes might be attributed to the good water quality of the Lakes during the study period notwithstanding the seasonal fluctuations in water quality [35, 36]. At least during the study season, majority of the fish caught were less stressed. The effect of stress on freshwater fish may be a factor of the severity of the stress, its duration and the overall physiological state of the fish .
The proportion of bacterial isolates from the intestine of live-caught fish (17.6%) was significantly higher than that of the liver (12.4%) or kidney (9.1%). But, there were slight variations with respect to the individual bacteria species. For instance, the number of isolates from live-caught fish intestine was equal to that of the liver concerning E. coli which was the most prevalent bacterium isolated. The second most prevalent in this study, E. tarda, was similarly isolated from the intestine and liver in equal proportions. E. tarda is a cause of rare but fatal food-/waterborne infection in man . Contrarily, E. aerogenes was most frequently isolated from the kidney. A study on marine fish found significantly higher potential pathogenic bacteria in kidneys than in liver samples and a variation was found between the fish species . The authors reported that significant differences were observed between fish species, organs and sites, indicating the importance of the environmental conditions on the fish microbiome. Although we did not investigate bacteria in different portions of the fish gut, it appears that microenvironment dynamics along the gut of various fish species may influence the composition and abundance of the bacterial flora .
In African catfish, higher bacterial load was recovered from the intestine than other organs like skin and gills . There is evidence that dense microbial populations occur within the intestinal contents, with numbers of bacteria much higher than those in the surrounding water, like the current finding, indicating that the intestines provide favorable ecological niches for these organisms . However, the method of intestinal bacteria sampling varied in different surveys. Some used anal swabs, others only the intestinal contents, while still in others intestinal tract and contents were homogenized and used for culturing . Austin and Al-Zahrani (1988)  distinguished between the flora of the gut contents and that intimately associated with the wall of the gastrointestinal tract, and noted that scanning electron microscopy showed only sparse microbial colonization of the wall. Although the genera present in the gut generally seem to be those from the environment or diet which can survive and multiply in the intestinal tract, there is evidence for a distinct intestinal microflora in some species . This same author reviewed the progressive decline in the numbers of aerobic heterotrophic bacteria along the digestive tract. Anaerobes were detected only in the upper intestine and in the intestinal contents.
However, other investigators  found that numbers of bacteria in freshwater salmonids increased between the stomach and the posterior portion of the intestine. The authors suggested that these numbers must represent active multiplication in the tract, as they could not be accounted for by ingestion. The numbers detected in this survey were probably an artificially low estimate since the methods used did not allow for the isolation and growth of strict anaerobes, species sensitive to oxygen, nutritionally fastidious species, or those requiring low growth temperatures (< 20 °C). In addition, the counts obtained were based on the total tissue weight in each sample, while the bacteria actually populate only the epithelial surface of the tract, and the rest of the tissue sterile. There was no significant difference in the bacterial flora of fish of different species, sex, breeding status, weight, or geographical source. But, microenvironment characteristics at various locations through the gastrointestinal tract of fish influence the composition and abundance of gut bacteria  and bacterial counts significantly differed between species, sources and feeding habits of examined fishes .
Fish internal organs such as the spleen, liver, and kidneys are expected to be sterile . Nonetheless, evidence accumulates for bacteria from internal organs of apparently healthy fish  in agreement with the current study. Possible fish immune compromise may explain such observations. Immune defect could happen due to stress. Fish stress could be associated with poor water quality, temperature changes, nutritional deficiencies, overcrowding, trauma, parasitism, primary viral infections [50,51,52], among others. For instance, fish immunity was found significantly affected by lower temperatures  making them susceptible to obligate or facultative pathogenic bacteria such as A. hydrophila.
The bacterial flora of the gut of two marine fish has been investigated in an attempt to clarify the relationship between these bacteria and the bacterial flora of their diets, and to determine the effect of the degree of specialization of the digestive tracts on their floras . Bacterial flora of fish with relatively undeveloped digestive tracts reflected that of the fishes' food, whereas fish with more specialized tracts have a distinctive gut microflora. In the red sea bream, the composition of the bacterial floras of the stomach and intestine changed with time after feeding. Half an hour after ingestion, most of the bacteria isolated from the stomachs resembled those of the fish meat diet, but after 6 h, vibrios resistant to bile and low pH predominated. Further work comparing representative strains of these indigenous vibrios from the red sea bream with other isolates from the stomach and intestine showed that the vibrios were able to survive in the presence of gastric juice at pH 4, and were able to grow, although at a reduced rate, at pH 5, while most of the other isolates were inhibited by these conditions .
In this study, the water samples were positive for A. sobria, Citrobacter spp., E. tarda, E. coli, K. pneumoniae, P. aeruginosa, S. typhi, S. dysenteriae and S. flexneri. Detection of these and other related bacteria both in marine and freshwater habitats and fish has been widely recorded [56,57,58,59]. Aeromonas is associated with a range of human opportunistic infections including enteritis and septicemia [60,61,62] and is one of the most common pathogens in tropical fish . E. coli was the most frequently detected bacterium in processed and live-caught fish samples in this study. Total and fecal coliforms such as E. coli are indicators of fecal contamination of aquatic environments and food. Detection of E. coli in processed fish samples could be due to unhygienic handling during processing.
Moreover, the fact that most of E. coli isolated during this study was from fish intestine reflecst warm-blooded animal pollution level of the water. E. coli can have a long-term survival and can multiply depending on fish and water temperatures [64,65,66,67]. It is known that fish possess distinct intestinal microbiota, but nutritional status or feeding habits, trophic level, species (attribute to complexity of the fish digestive system) and the environmental conditions (salinity of the habitats and the bacterial load in the water) are the most influential factors which change the intestinal microbiota composition and abundance . This may explain the observed differences in the distribution of bacteria from fish samples in the three Lakes. While bacteria in water can influence the microbial flora associated with fish, the reverse is also worth consideration.
Shigella spp. and Salmonella spp. are pathogenic bacteria found in animal, human or environmental reservoir. Although contamination of fish products with these bacteria is commonly from the environment, their incidence in ready-to-eat fish product due to unhygienic handling cannot be ruled out . Citrobacter, Enterobacter and Klebsiella are indigenous to general environment and frequently present in fish but most of these bacteria are considered non-pathogenic environmental strains. The bacterial species isolated from processed fish (E. coli, K. pneumonia and S. paraTyphi) were also recovered from the water samples.
The variation in the number of isolates and bacterial species between sampling sites of the study Lakes might be attributed to the relative distance and degree of exposure to the nearby point source pollution around the study area. Disruption of the environmental microbiome after an earthquake followed by seasonal variation in the water quality was noted although restoration of these microbial communities as a function of time and sanitation practices occurred in Nepal .
All positive live fish samples were positive for at least one isolate of all the 15 bacteria species recovered. This shows the higher bacteria species diversity in fish compared to the aquatic environment wherein only 4 species, E. aerogenes, E. cloacae, P. shigelloides and S. typhi, were characterized although the water bacteria prevalence was higher. The finding suggests that the bacteria detected in fish internal tissue might constitute the natural fish microbiota and/or the fish bacteria source might be their diet. Alternatively, the fish might have been exposed to seasonal or occasional biological pollutants which have been diluted or neutralized from the environment. Such dynamics in the aquatic ecosystem may explain particularly the absence of human urinary and respiratory tracts pathogens E. aerogenes and E. cloacae  in water samples and their detection in fish. Another surprising finding is the absence of P. shigelloides in the water samples. The common environmental reservoirs for this organism include freshwater ecosystems and estuaries and inhabitants of these aquatic environs, and a series of foodborne enteritis outbreaks have been solely or partially attributable to P. shigelloides . Another species that is persistently detected in freshwater environments and stands among major causes of food-/waterborne human illnesses , but which could not be detected in the water samples of the current Lakes was S. typhi.
Detection of the common fecal coliforms (E. coli, K. pneumonia, E. aerogenes) and Salmonella spp. in all of the sample types especially in processed fish, signals the danger of passage of these pathogens and their toxins to man via infected and contaminated fish products. Salmonella spp. and fecal coliforms were detected in 42% of water samples and 64% of processed fish samples in this study. Shigella spp. and Salmonella spp. are pathogenic bacteria found in animal or human reservoir and contamination of fish products by these bacteria is almost always due to poor hygiene.
Except A. sobria, all the others species were detected from tilapia in different frequencies. Similarly, all bacteria species were also detected from catfish with the exception of P. shigelloides. However, C. freundii, C, koseri and V. parahemolyticus were not isolated from carps. Even though it seemed that some bacteria species tended to be specifically associated with a particular fish species, the association was not statistically significant. Different studies reported the occurrence and antimicrobial resistance of A. sobria, E. tarda, P. shigelloides, P. aeruginosa, Citrobacter spp. and Klebsiella spp. from tilapia and catfish [73,74,75,76].
During the study period, differences were observed in the bacterial prevalence and frequency across the sampling sites of each Lake. This may be attributed to the relative distance and degree of exposure to a nearby pollution source around.
Although the report on fish bacteria and their occurrence in humans is limited in Ethiopia, there were some efforts. Among the bacteria found in this study, E. coli, Klebsiella spp., Enterobacter spp., Citrobacter spp., and Aeromonas spp. which are enterotoxin-producing were detected in stools of Ethiopian children with diarrhoeal disease in the late 1970s . E. tarda was isolated from the liver of a tilapia from Lake Ziway for the first time for the Lake . The other bacteria detected in this same study were E. coli, Kebsiella oxyloca, Citrobacter spp. and Yersinia enterocolitica. Another investigator  isolated Aeromonas spp. including A. sobria, E. tarda, Vibrio spp., E. aerogenes P. shigelloides, E. coli, K. pneumoniae, Shigella spp., Citrobacter spp. from fish of Lake Tana. The author also reported that all the bacterial species, which were isolated from the water samples, were also recovered from fish in the Lake. Moreover, an outbreak of A. hydrophila associated with a certain parasite in pond of African catfish fingerlings at Sebeta, central Ethiopia, was reported . Vibrio spp., Salmonella, Shigella and E. coli were detected from surface water and sediment samples of Lake Ziway and drinking water system of Batu (former Ziway) town, Ethiopia .
From the total 410 fish samples examined, six were found contaminated with Shiga toxin-producing E. coli strain in Ethiopia . The isolates were resistant to ampicillin and streptomycin disks. However, ciprofloxacin, gentamicin and nalidixic acid were found effective in inhibiting the growth of all of the isolates. Vibrio, Escherichia, Aeromonas, Pseudomonas, Salmonella and Streptococcus were detected from Nile tilapia in Hawassa with the bacterial population significantly higher in the intestine than in the liver . A short review on bacterial pathogens of fish presented pathogenic and zoonotic bacteria such as Edwardsiella, Salmonella, Escherichia, Staphylococcus, Vibrio and Aeromonas recovered from fish from various parts of Ethiopia . A more recent molecular study that analyzed the diversity of microbiota in different sections of tilapia gut found more diversity in Lake Chamo than Lake Hawassa .
In some countries like Poland  and Malaysia , the emergence of hitherto unreported pathogenic fish bacteria is becoming evident. Thus, fish bacteria detection methods in Ethiopia must take into account less known and unreported ones as well. Moreover, exploring possible reciprocal transmission of potential pathogenic bacteria from wild fish to aquaculture, and domestic animals or humans is essential. This will contribute towards microbial-source-detection investigations.
This work will serve as an initial step to establish a baseline dataset of microbial communities associated with wild freshwater fish in Ethiopia. But, it has certain notable limitations. It neither quantified the detected bacteria, nor molecularly identified them, and no antibiotic susceptibility test was done. The results would have been more robust if samples from fish skin and gills which are gateway routes of transient or resident microbiota and/or potential pathogenic bacteria have been included. Moreover, the study did not assess seasonal patterns of both water quality and fish microbiota.