Methods

Study design and study area
The study adopted a cross-sectional design, systematically sampling market-ready fish from selected farms to capture representative data across varied aquaculture systems. Conducted in Kafue (15.7700° S, 28.1830° E) and Siavonga (16.5380° S, 28.7180° E)—two major aquaculture hubs in Zambia—the study compared fish from pond and cage culture systems to evaluate their influence on fish health, growth, and productivity.

Pond culture systems, predominant in Kafue, are typically earthen in type and offer greater control over environmental conditions such as water quality, feeding, and disease management. This controlled environment can enhance fish health and reduce stress, leading to consistent growth rates; however, ponds are limited by smaller production volumes and higher land and water management requirements. In contrast, cage culture systems, common in Siavonga on Lake Kariba, enable high-density production in open water bodies, benefiting from natural water exchange and lower infrastructure costs per unit biomass. Nonetheless, cage systems expose fish to fluctuating environmental conditions, higher pathogen transmission risks, and occasional stress from crowding, which can affect health and survival.

The study classified aquaculture farms following the Department of Fisheries (Ministry of Fisheries and Livestock, Zambia) criteria, based on production scale and system type, to evaluate capacity and market orientation. Large-scale farms comprised ponds exceeding 1 hectare or cage units with over 40 cages, producing more than 12 metric tonnes (MT) and 20 MT annually, respectively, mainly supplying commercial and export markets. Medium-scale farms operated ponds of 2,400 m²–1 hectare or 5–40 cages, yielding 3–12 MT and 12–20 MT per year, typically serving regional markets. Small-scale farms managed 600–2,400 m² ponds or 1–5 cages, producing below 3 MT and 12 MT annually, respectively, and primarily targeted local markets with modest infrastructure and limited management capacity.

Sample size estimation
This diverse representation enhanced the robustness of the findings and their applicability to the broader aquaculture industry. A two-stage sampling methodology was employed in Siavonga and Kafue districts to ensure a representative selection of farms and fish samples.

Stage 1: Selection of farms
The first stage involved developing a sampling frame in collaboration with district fisheries officers, identifying fish farms based on production capacity, geographical location, and culture system (cage, pond, or tank). A proportional allocation method was then applied to determine the number of farms to be sampled, ensuring representative coverage across farm sizes and production systems. The number of farms selected (nf) was calculated as:

$$
nf = \left(\frac{N_f}{N}\right) \times ns
$$

where:

Nf = total number of farms in the selected district [Siavonga (25) and Kafue (30)]

N = total number of registered farms in both districts (55)

ns = total number of farms required for the study (determined based on logistical feasibility and study objectives) (11)

Farms were then selected randomly within each district, ensuring proportional representation based on production capacity and system type. Five and six fish farms were selected in Siavonga and Kafue districts, respectively.

Stage 2: Collection of fish samples

Once the farms were selected, the second stage involved collecting fish samples from each farm. The number of fish sampled per farm was determined using Cochran’s formula [13] to ensure an adequate sample size:

$$
n = \frac{Z^{2} P (1 – P)}{d^{2}}
$$

where:

n = sample size,

Z is the Z statistic for a level of confidence (95%) = 1.96

P is the expected prevalence = 8.4% [14]

d is the precision =0.05

The sample size (n) was estimated at 118.

Sample collection
Apparently healthy Nile tilapia were sampled from selected farms for bacterial isolation, antimicrobial susceptibility testing, and antibiotic residue analysis. Target bacteria included Vibrio spp., Lactococcus garvieae, Streptococcus spp., Aeromonas spp., Escherichia coli, and Salmonella spp. Fish were humanely euthanised with clove oil (100–200 mg/L) following ethical guidelines, and bacteriological swabs from the liver and cloaca were aseptically collected. Samples were inoculated onto Blood agar (HiMedia, India) and MacConkey agar (HiMedia, India), with additional enrichment in alkaline peptone water (3% NaCl, pH 8) (HiMedia, India) for 24 hours. All samples were kept in cooler boxes at 4°C during transport to preserve viability and integrity for subsequent bacterial and antimicrobial residue analyses.

Isolation and idenfication of bacterial
Bacterial isolation involved incubating media plates (Blood and MacConkey agar) at 24–28°C for 24 hours upon arrival at the University of Zambia, School of Veterinary Medicine, Bacteriology Laboratory.

A loopfuls from each previously cultured APW tube (liver and cloacal tissue) were innoculated onto Thiosulfate-citrate-bile salts-sucrose (TCBS) agar (HiMedia, India) and incubated at 24–28°C for 24 hours. To purify the Vibrio suspect colonies, a single green and yellow colony from each grown type was streaked onto fresh TCBS agar plates and incubated overnight at 24–28°C. This procedure was repeated until pure, consistent colonies were obtained. Identification was accomplished based on the results of microscopic observation of stained smears using Gram staining and biochemical examination were carried by using the indole test, citrate utilization test, triple sugar iron test and salt tolerance test to detect growth of Vibrio species on 0%, 3% and 6% NaCl [15].

The isolation and identification of Aeromonas spp., Escherichia coli, and Salmonella spp. were preceded by inspecting and selecting colonies on previously inoculated MacConkey agar (HiMedia, India) to facilitate differentiation based on lactose fermentation. The selected isolates were subcultured on MacConkey agar (HiMedia, India) at 24–28°C for 24 hours. Following incubation, colony morphology was recorded, and suspected isolates were subjected to Gram staining for preliminary characterization. Biochemical identification was conducted using oxidase, triple sugar iron (TSI), citrate, and urease tests [15].

Isolation of Streptococcus spp. and L. garvieae was performed by selecting suspect colonies from previously inoculated Blood agar (HiMedia, India), which were subcultured on the same media and incubated at 24–28°C for 24–48 hours under facultative anaerobic conditions. Colony morphology and hemolytic patterns were recorded to differentiate species. Suspected isolates were subjected to Gram staining, catalase testing, and biochemical assays. Sugar fermentation tests, including lactose, mannitol, and trehalose utilization, were performed to further characterize isolates [16].

The isolates of Vibrio spp., Aeromonas spp., L. garvieae., Streptococcus spp., Salmonella spp., and Escherichia coli were stored in glycerol (Tryptone Soya Broth with 50% glycerol at −40 °C) for further antimicrobial susceptibility testing.

Determination of antibiotic susceptibility of selected bacteria
Antibiotic susceptibility testing (AST) was conducted using the disk diffusion method in accordance with the Clinical and Laboratory Standards Institute (CLSI) guidelines to evaluate bacterial resistance profiles [17]. Mueller-Hinton agar (HiMedia, India) plates were inoculated with bacterial suspensions standardized to 0.5 McFarland and incubated at 35–37°C for 16–18 hours. Zones of inhibition surrounding antibiotic disks were measured and interpreted based on CLSI breakpoints to classify isolates as susceptibile, intermediate or resistant. Target bacteria included Vibrio spp., L. garvieae, and Escherichia coli, tested against a panel of antimicrobials comprising Ampicillin (10 µg), Erythromycin (15 µg), Sulfamethoxazole (25 µg), Florfenicol (30 µg), Tetracycline (30 µg), Ceftazidime (30 µg), Ciprofloxacin (5 µg), Gentamicin (10 µg), Meropenem (10 µg), Imipenem (10 µg), Chloramphenicol (30 µg), Co-trimoxazole (25 µg), Amoxicillin-Clavulanic acid (20/10 µg), and Cefoxitin (30 µg).

Evaluation of antimicrobial residues in fish muscle
High-purity antibiotic standards (≥98%) for Sulfamethoxazole, Trimethoprim, Penicillin G, Ampicillin, Tetracycline, Oxytetracycline, Gentamicin, Tiamulin, Erythromycin and Tylosin were procured from Sigma-Aldrich (Germany), along with isotopically labeled internal standards for accurate quantification. High-Performance Liquid Chromatography (HPLC)-grade solvents were sourced from Merck (Germany), and ultrapure water was generated using a Pure Lab Ultra system (UK). Stock solutions (1 mg/mL) were prepared in methanol and stored at -20°C. Working standards (0.5–200 µg/L) were freshly prepared for calibration before each analytical batch.

Fish sample collection and processing
All collected fish samples were analyzed for antibiotic residues. Approximately 0.5 g of fish muscle was homogenized, and antibiotic residues were extracted using ethyl acetate as the extraction solvent, with EDTA added as a chelating agent to minimize interference in Fluoroquinolone detection. The extraction process involved vortexing, sonication, and centrifugation, after which the supernatant was subjected to solid-phase extraction (SPE) for purification using Oasis HLB cartridges (Waters, USA). The purified antibiotic extracts were then evaporated, reconstituted in 500 µL of mobile phase, filtered through a 0.2 µm nylon membrane, and analyzed using UHPLC–MS/MS.

UHPLC-MS/MS analysis and method validation
Samples were analyzed using a Shimadzu Nexera X2 UPLC system with a Hypersil Gold C18 column (30 mm, 1.9 µm) under gradient elution (ethanol and 0.1% formic acid in water) at a flow rate of 500 µL/min. The UPLC system was coupled to a triple quadrupole mass spectrometer (Shimadzu 8050), operated in positive ionization mode using multiple reaction monitoring (MRM) for precise quantification. Calibration curves (0.5–200 µg/L) were established using matrix-matched calibration, with internal standards for accuracy. Method validation included linearity (R² > 0.99), recovery (1, 10, 100 µg/kg), intra-/inter-day precision (RSD <15%), and limits of detection (LOD) and quantification (LOQ) based on ICH guidelines. This validated analytical method ensured high sensitivity, accuracy, and reproducibility for detecting antibiotic residues in fish, supporting reliable residue monitoring and food safety assessments.

Data analysis
Data were organized in Microsoft Excel 2013 and analyzed using DATAtab software [18]. Descriptive statistics (means, frequencies, and percentages) were applied to summarize bacterial prevalence and antibiotic resistance patterns. Associations between production systems and bacterial occurrence were evaluated using the Chi-square (χ²) test with a significance level of p < 0.05. The Multiple Antibiotic Resistance (MAR) index was calculated as a/b, where a represents the number of antibiotics to which an isolate was resistant and b represents the total number of antibiotics tested, to assess the intensity of antimicrobial resistance. Antibiotic residue levels were then compared with established Maximum Residue Limits (MRLs) to evaluate food safety.

Ethical consideration
Ethical approval was obtained from ERES Converge Institutional Review Board (IRB), reference number 2024-Oct-006, before commencement of the study. Written consent was also obtained from participating farmers through the Department of Fisheries before sampling.

Funding

The authors did not receive any specific funding for this work.