Ethical approval (FVM BAUEC/2019/193) and a research permit (NACOSTI/P/18/64308/21246) were obtained from the Faculty of Veterinary Medicine Biosafety, Animal Use and Ethical Committee, University of Nairobi, and from the National Commission for Science, Technology and Innovations (NACOSTI), respectively, prior to study commencement. Informed verbal consent to conduct the research was obtained from the Regional Director of Fisheries. All experimental procedures involving O. niloticus were performed in accordance with internationally accepted guidelines for the care and use of animals, and the Animal Research: Reporting of In Vivo Experiments 2.0 guidelines were strictly followed for study design, fish handling, and reporting [32].

The study was conducted between January 2023 and November 2024 along the shorelines of Lake Naivasha, specifically at the Karagita Landing Beach. This landing beach was deliberately selected because of declining fish stocks reported at other locations within the lake.

The lake supports high aquatic biodiversity and hosts several fish species, including blue-spotted tilapia (Oreochromis leucostictus), red-bellied tilapia (Coptodon zillii), largemouth bass (Micropterus salmoides), Louisiana red swamp crayfish (Procambarus clarkii), river cyprinid (Barbus paludinosus), common carp (Cyprinus carpio), O. niloticus, and African sharptooth catfish (Clarias gariepinus) [33]. Lake Naivasha is the second-largest freshwater lake in Kenya after the Kenyan portion of Lake Victoria [33]. It has a surface area of 139 km² and a mean depth of 3.35 m, with a maximum depth of 7 m; however, these parameters fluctuate under extreme hydrological conditions [34]. Naivasha town is situated approximately 80 km northwest of Nairobi, and the lake basin covers an area of ~3400 km². Lake Naivasha is located in the Eastern Rift Valley at a latitude of 0°46′10″ (0.7694), a longitude of 36°20′25″ (36.3403), and an altitude of 1890 m above sea level (Figure 1) [35].

Fish included in this study were apparently healthy at the time of capture and free from visible external lesions or parasitic infestations. Individuals exhibiting skin lesions, fin rot, or external parasite infestations were excluded. Any exclusions made after capture were documented, and such fish were excluded post-capture.

A total of 38 table-sized O. niloticus (18 males and 20 females) meeting the inclusion criteria were collected. The fish had a mean body weight of 418.6 ± 39.0 g, standard length of 20.9 ± 0.8 cm, and total length of 25.8 ± 0.9 cm. Fish were randomly captured using seine nets attached to motorized boats at Karagita Landing Beach between 0700 and 0800 h in January 2023. Water quality parameters were not assessed at capture because sampling covered large portions of the lake; however, January typically corresponds to the hot–dry season.

Following capture, fish were placed in two separate 100-L plastic tanks containing source water and transported alive to the Bacteriology Laboratory, Department of Veterinary Pathology, Microbiology, and Parasitology, University of Nairobi. Laboratory analyses commenced within 2 h of arrival.

Prior to necropsy, fish were humanely anesthetized using tricaine methane sulfonate (Syncaine®, Abbott Laboratories, Chicago, IL, USA) and euthanized in accordance with institutional animal-care guidelines. Postmortem procedures were conducted under aseptic conditions following standardized protocols described by Noga [36] and Roberts [37]. Dissecting instruments and bench surfaces were sterilized between samples using flaming and 70% ethanol, respectively, and gloves were changed between handling individual fish to minimize contamination. Separate cutting sets were used for each fish, and the necropsy sequence was standardized from external surfaces to internal organs to prevent microbial carryover.

Before opening the body cavity, fish skin surfaces were swabbed with 70% ethanol. Each fish underwent external examination, and gross lesions and biodata were recorded. A midline incision was made from the vent to the operculum, followed by a lateral incision along the abdominal wall to expose the viscera. The esophagus and rectum were severed, and the entire gut was removed. The hepatopancreas and mesentery were bluntly dissected and discarded. The gut was collected in sterile Petri dishes for bacterial isolation.

Field and laboratory blanks were included by processing phosphate-buffered saline (PBS; pH 7.2) alongside gut samples as negative controls to monitor environmental and procedural contamination.

Up to 25 g of gut tissue and contents were aseptically weighed and homogenized with 225 mL buffered peptone water to obtain an initial 1:10 dilution using a stomacher blender. The homogenate was serially diluted, and 0.1 mL aliquots of selected dilutions were inoculated onto tryptone soya agar (TSA; HiMedia Laboratories Pvt. Ltd., Mumbai, India) in duplicate and incubated aerobically at 24°C–25°C. After 24 h, plates were examined for growth and colony morphology. Single colonies were randomly selected, subcultured on TSA, and purified by repeated streaking. Pure isolates were transferred to tryptone soya broth (TSB; HiMedia) supplemented with 20% glycerol and stored at −80°C. Recovery was confirmed by thawing selected isolates and assessing growth on TSA.

Following the isolation of 50 bacterial strains, preliminary screening was performed. Hemolytic activity was assessed by streaking isolates onto 5% sheep blood agar and incubating aerobically at 24°C–25°C for 24 h. Hemolysis was classified as α, β, δ, or γ. Isolates exhibiting γ or α hemolysis were selected for further analyses.

Antibiotic susceptibility testing was performed using the Kirby–Bauer disk diffusion method in accordance with Clinical and Laboratory Standards Institute guidelines [38]. Bacterial suspensions were adjusted to a 0.5 McFarland standard (~1.5 × 10⁸ CFU/mL) and spread onto Mueller–Hinton agar (Oxoid Ltd., Basingstoke, UK). Antibiotic disks (HiMedia) included ampicillin, tetracycline, streptomycin, sulfonamides, nalidixic acid, trimethoprim–sulfamethoxazole, gentamicin, nitrofurantoin, chloramphenicol, and kanamycin. Inhibition zones were measured after 24 h and interpreted as sensitive or resistant as described by Patel et al. [39]. Escherichia coli American Type Culture Collection [ATCC]® 25922 and Staphylococcus aureus ATCC® 25923 were used as quality control strains.

Pathogenicity was evaluated based on cumulative mortality, clinical signs, and bacterial re-isolation. Sample size estimation was conducted using G*Power version 3.1.9.6 [44]. A total of 150 healthy O. niloticus were acclimatized and randomly assigned to treatment and control groups. Fish received intraperitoneal injections of bacterial suspensions or PBS (control). Fish were monitored for 10 days, and bacteria were re-isolated from spleen and kidney of moribund or dead fish. All challenge experiments were conducted under biosafety level 2 conditions.

Candidate isolates were identified based on colony morphology, Gram staining, biochemical tests, and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (Bruker Daltonics GmbH, Bremen, Germany), following manufacturer criteria [45, 46].

Enzymatic activity indices were calculated as described previously [41]. Bacterial counts were log_10- transformed, and data were expressed as mean ± SD. Normality and homogeneity were assessed using Shapiro–Wilk and Levene’s tests. Differences among groups were analyzed using chi-square tests and one-way analysis of variance, followed by Tukey’s post hoc test, with statistical significance set at p < 0.05. Analyses were performed using IBM SPSS Statistics (IBM Corp., Armonk, NY, USA), version 31.

Proteolytic and amylolytic activities of the selected isolates are presented in Figure 2. Among the 10 isolates screened, 40% (4/10) demonstrated protease activity, while 70% (7/10) exhibited amylase activity. Only four isolates (C, E, H, and I) expressed both enzymes.

Isolate H exhibited the highest enzymatic indices, with proteolytic and amylolytic values of 0.5 and 0.6, respectively (Figure 3). Isolates A, F, and G showed amylolytic activity only, with isolate G presenting the highest amylase index. In contrast, isolates B, D, and J lacked both enzymatic activities and were excluded from further evaluation. Consequently, seven isolates (A, C, E, F, G, H, and I) were retained for subsequent assays.

Survival of the selected isolates under bile salt stress is illustrated in Figures 4 and 5. All isolates (7/7) survived in 0.3% bile salts, whereas only 57% (4/7) remained viable in 2% bile salts.

Isolates A, C, and I showed complete growth inhibition at 2% bile concentration, indicating bile intolerance. In contrast, isolates G and H exhibited minimal reductions in viable counts at both bile concentrations relative to the control, demonstrating strong bile salt tolerance (Figure 5).

The acid tolerance of four selected probiotic candidates (E, F, G, and H) was evaluated under simulated gastrointestinal conditions (Table 2). At baseline (0 h), viable counts differed significantly among isolates across all pH levels (p < 0.05).

Extreme acidity (pH 1.5) resulted in immediate reductions in viability compared with pH 3.0 and 7.2. Isolate G consistently exhibited the highest survival at all pH values, whereas isolate F showed the lowest tolerance. After 1.5 h of exposure, viable counts declined significantly at both acidic conditions (p < 0.05), with isolate G remaining the most acid-tolerant.

Following 3 h of exposure, further viability losses were observed. Isolate G retained the highest survival at pH 1.5 and 3.0, while isolate F exhibited complete loss of viability at both acidic levels. At pH 7.2, isolates G and H had significantly higher viable counts than E and F (p < 0.05).

All four tested isolates demonstrated the ability to adhere to stainless steel surfaces (Figure 6). Isolate H exhibited the highest adhesion capacity (approximately 4.7 × 10⁴ CFU/mL), indicating strong colonization potential.

Isolate G also showed substantial adhesion, while isolate F exhibited moderate adherence. In contrast, isolate E demonstrated the lowest adhesion among the four candidates.

Growth patterns of all candidate isolates followed typical bacterial population dynamics, comprising lag, exponential, and stationary phases (Figure 7). The exponential growth phase commenced at approximately 3 h for all isolates, although growth intensities varied.

Isolate G demonstrated the most vigorous and prolonged exponential growth, achieving the highest optical density values. Conversely, isolate E showed the least robust growth, reaching the stationary phase earlier and at lower optical density. By 8–9 h, most isolates transitioned into the stationary phase, likely due to nutrient depletion or accumulation of metabolic by-products.

Mortality was observed among fish inoculated with isolate G (80%), isolate E (10%), and PBS control (20%) (Table 3). No abnormal clinical signs were noted in fish challenged with isolate E or in the control group.

In contrast, fish exposed to isolate G exhibited pronounced clinical manifestations, including lethargy, weakness, stagnation near the aquarium surface, scale desquamation, hydronephrosis, and congestion, particularly at the fin bases (Figure 8).

Mixed bacterial cultures resembling isolate G were consistently re-isolated from the kidneys and spleens of moribund and dead fish. No pathogenic bacteria were recovered from fish injected with isolate E or PBS, confirming that mortality was attributable to isolate G. These results indicated that isolates E, F, and H were non-pathogenic and suitable for further probiotic evaluation, whereas isolate G was pathogenic and excluded.

Morphological, biochemical, and MALDI-TOF MS analyses were used to identify the probiotic candidates and the pathogenic isolate (Table 4; Figure 9). The three non-pathogenic probiotic isolates were identified as R. marisflavi (E), M. luteus (F), and P. vaccinostercus (H). The pathogenic isolate was identified as A. ichthiosmia (G).

Among the 50 gut bacterial isolates recovered from wild O. niloticus, only 10 (20%) fulfilled the preliminary probiotic screening criteria of γ-hemolysis and broad antibiotic susceptibility. Of these, seven progressed to advanced functional assays, and only three isolates (E, F, and H) successfully passed all enzymatic activity, acid–bile tolerance, adhesion, growth kinetics, and pathogenicity evaluations.

This progressive reduction from 50 initial candidates to three validated strains highlights the stringent nature of the screening protocol and underscores the scarcity of probiotic-grade autochthonous bacteria within the gut microbiota of wild tilapia. Notably, this study represents the first comprehensive isolation and probiotic screening of indigenous gut bacteria from wild O. niloticus inhabiting Lake Naivasha, thereby filling a critical knowledge gap left by earlier Kenyan studies that primarily focused on pond-reared fish or commercial probiotic products.

By targeting wild fish populations, the study provides novel insights into naturally adapted microbial communities that may exhibit enhanced colonization efficiency, competitive exclusion, and physiological compatibility within local aquaculture environments.

The present study identified R. marisflavi, M. luteus, and P. vaccinostercus as potential probiotic candidates from the gut of wild tilapia. These bacterial species are rarely reported in fish gastrointestinal ecosystems.

Notably, P. vaccinostercus has not previously been described as a probiotic organism from any aquatic species. Its recent taxonomic reclassification within the genus Paucilactobacillus further emphasizes the limited genomic and functional information currently available for this species. The high-confidence species-level identification achieved by MALDI-TOF MS therefore constitutes the first report of P. vaccinostercus as a non-pathogenic, functionally promising probiotic strain in tilapia.

Similarly, R. marisflavi represents an underexplored bacterial taxon with considerable probiotic potential. Together, these isolates constitute valuable microbial resources for future probiotic development. However, molecular verification remains essential to confirm their safety profiles, particularly given reports of opportunistic pathogenicity among closely related strains. Depositing these isolates in accredited microbial repositories is strongly recommended to facilitate downstream genomic analyses and formulation research.

Although no prior studies have specifically reported R. marisflavi, M. luteus, or P. vaccinostercus as probiotics in O. niloticus, the functional traits observed in the present isolates align closely with those documented for other probiotic genera isolated from fish gastrointestinal tracts.

Previous investigations have demonstrated that probiotic bacteria from O. niloticus commonly exhibit enzymatic activity, bile and acid tolerance, and adhesion capacity—features that collectively enhance nutrient digestion, gut health, and disease resistance [21]. Reda et al. [40] and Athulya et al. [47] similarly reported substantial enzymatic and antimicrobial activities among gut-derived isolates, including Lactococcus lactis, Enterococcus faecalis, Lysinibacillus spp., Citrobacter freundii, and Bacillus spp., thereby reinforcing their probiotic suitability.

Adhesion to intestinal surfaces plays a crucial role in microbial persistence, pathogen exclusion, and immune stimulation [48]. According to Torres-Maravilla et al. [49], adhesive probiotics enhance mucosal immunity and epithelial barrier integrity, enabling sustained probiotic colonization and protective host responses.

Effective probiotic candidates must also withstand gastrointestinal stresses, including acidic pH and elevated bile concentrations. The physiological bile concentration in fish intestines ranges from 0.4% to 1.3%, while experimental screening commonly employs concentrations between 2.5% and 10% [50]. In the present study, isolate H exhibited superior tolerance to both acidic and bile environments over prolonged exposure periods, suggesting high probiotic suitability.

Comparable findings were reported by Balcázar et al. [50] and Coulibaly et al. [51], who isolated LAB from O. niloticus that demonstrated strong acid–bile tolerance, adhesion capacity, and antagonistic activity against fish pathogens. Iorizzo et al. [52] further confirmed these probiotic traits in Lactiplantibacillus plantarum derived from trout intestines.

Unlike many earlier screening studies that relied solely on in vitro assays, the current investigation uniquely incorporated in vivo pathogenicity testing to ensure host safety, consistent with the Food and Agriculture Organization (FAO) probiotic evaluation guidelines. Collectively, the results affirm the gut microbiome of O. niloticus as a rich reservoir of safe indigenous bacteria with multifunctional probiotic potential.

The observed antagonistic activity also aligns with global evidence that probiotics produce antimicrobial compounds, such as organic acids, bacteriocins, siderophores, and lipopeptides, that inhibit common aquaculture pathogens, including Aeromonas, Vibrio, and Streptococcus species [53].

Probiotic strains exhibiting characteristics similar to those observed in this study have previously been isolated from the gastrointestinal tract of O. niloticus [54]. Meidong et al. [55] further demonstrated that L. plantarum showed strong acid–bile tolerance, adhesion capacity, and non-hemolytic behavior.

In contrast to commonly used probiotic genera such as Bacillus and Lactobacillus, P. vaccinostercus displayed comparable acid and bile resistance, along with superior adhesion potential, suggesting high adaptation to the tilapia gut environment.

P. vaccinostercus, recently reclassified within the Paucilactobacillus clade, is recognized for its metabolic versatility and aerotolerance—traits that support probiotic functionality. Genomic studies have identified genes associated with vitamin biosynthesis, stress tolerance, and antimicrobial activity in Paucilactobacillus species, suggesting roles in nutrient assimilation, pathogen exclusion, and immune modulation [53, 56].

Despite the established probiotic use of LAB in aquaculture, strain-level characterization of P. vaccinostercus remains limited, and rigorous safety assessment, including screening for transferable AMR genes, is essential before application [49].

When benchmarked against commercial probiotics such as Bacillus spp. and L. plantarum, the isolates identified in this study, particularly P. vaccinostercus, demonstrated comparable or superior performance across acid tolerance, bile resistance, enzymatic activity, and adhesion. This suggests that locally adapted strains may offer enhanced ecological compatibility, cost efficiency, and sustainability for Kenyan aquaculture systems.

R. marisflavi, formerly classified as Bacillus marisflavi, belongs to the family Bacillaceae within the phylum Firmicutes and was reassigned to the genus Rossellomorea based on phylogenomic evidence [57, 58].

Recent experimental studies, including in vivo trials in O. niloticus, have shown that dietary supplementation with R. marisflavi, often in probiotic consortia, can improve growth performance, feed utilization efficiency, survival under bacterial challenge, and immune parameters [59].

The probiotic effects of R. marisflavi are attributed to multiple mechanisms, including competitive exclusion of pathogens, secretion of digestive enzymes (amylases, proteases, lipases), and production of antimicrobial metabolites [57]. The present study confirmed several of these probiotic attributes under in vitro conditions.

The current findings demonstrated that M. luteus exhibits multiple probiotic-associated traits, including antagonism against fish pathogens, extracellular enzyme production, acid–bile tolerance, adhesion capacity, and favorable antibiotic susceptibility profiles.

Previous experimental studies have further reported growth-promoting and protective effects of M. luteus in O. niloticus under pathogenic challenge, supporting its probiotic candidacy [59].

However, recent evidence also highlights the strain-specific nature of M. luteus. A 2025 study from India identified pathogenic M. luteus strains causing severe disease and high mortality in farmed tilapia [60]. These contrasting outcomes emphasize the importance of comprehensive safety evaluation, including virulence assessment and AMR profiling, before probiotic application.

The present study also isolated A. ichthiosmia from the gut of O. niloticus. Although this isolate exhibited several probiotic-like traits, including enzymatic activity, acid–bile tolerance, and adhesion capacity, it was ultimately confirmed as pathogenic.

The experimental challenge fulfilled Koch’s postulates, with affected fish exhibiting lethargy, congestion at the fin bases, mortality, and re-isolation of the same bacterial strain from internal organs.

Given the limited prior documentation of A. ichthiosmia pathogenicity in fish, these findings represent one of the first well-characterized reports identifying this species as a fish pathogen. This underscores the critical necessity of in vivo safety testing alongside functional probiotic screening.

The development of indigenous probiotic strains offers significant potential for sustainable aquaculture within Kenya’s Blue Economy framework by reducing reliance on antibiotics and mitigating AMR risks.

Utilization of native, non-pathogenic strains aligns with FAO–WOAH–WHO One Health strategies by enhancing fish health through biological mechanisms while preserving ecological balance.

The application of MALDI-TOF MS in this study demonstrated its utility for rapid, accurate bacterial identification, strengthening local diagnostic capacity. However, limitations include the absence of whole-genome sequencing, strain-level genomics, and metabolomic profiling.

Future research will integrate genomic, metabolomic, and in vivo performance evaluations to confirm safety, elucidate probiotic mechanisms, and support regulatory approval for commercial application.