Microbial safety of commercial probiotics in aquaculture: A mixed methods approach applied to Bangladesh

Abstract

Aquaculture plays a vital role in Bangladesh's food security and economic growth, yet disease outbreaks continue to challenge productivity and drive antibiotic reliance. Probiotics are promoted as sustainable alternatives; however, concerns persist regarding their microbial integrity, safety, labelling accuracy, and regulatory oversight. This thesis systematically assesses the safety of commercial probiotics used in aquaculture in Bangladesh through an integrated approach combining stakeholder surveys, microbial characterisation, and metagenomic analyses. The study started with an analysis of probiotic purchasing behaviours among fish farmers (Chapter 2), employing Best-Worst Scaling (BWS) and multinomial logit modelling to elucidate the influence and relative importance of different label attributes—such as microbial composition and manufacturer claims—derived from 36 commercially available probiotics on purchasing decisions. The analysis revealed that microbial species composition and concentration were the most influential factors and significantly (p <0.05) guided purchasing decisions. In contrast, product name and the inclusion of storage advice exerted least influence. The findings also revealed a moderate level of brand loyalty among farmers, with 43.3% reporting recurring purchases of the same probiotic brand, indicating a preference for known or trusted products. Furthermore, aquaculture consultant suggestions influenced purchase decisions substantially, with 44.2% of respondents citing them as their primary source of advice. These findings underline the importance of certain label attributes along with external advice in influencing probiotic buying patterns among fish farmers. To assess the relative importance of potential safety hazards associated with probiotics applied in aquaculture, a modified e-Delphi study was conducted (Chapter 3) to capture expert consensus on key hazards. The expert panel identified contamination with viable pathogenic microorganisms, horizontal transfer of antimicrobial resistance (AMR) genes, and the transfer of virulence genes as the most important potential hazards, contributing 18.96%, 11.63%, and 9.1% respectively. Kendall's W coefficients increased in each Delphi round (Round 1: 0.224; Round 2: 0.434; Round 3: 0.448), indicating expert opinion convergence. The rising kendall's W coefficients in successive rounds indicate a growing consensus among experts about the relative importance of commercial probiotic hazards, emphasising the need for evidence-based prioritisation in microbial risk assessment and regulatory policy formulation. Twenty probiotic products were subsequently chosen for the phenotypic and genotypic characterisation of bacterial isolates, with the objective of identifying some of the hazards previously identified through expert consensus in Chapter 3. Direct antimicrobial activity was discovered in 10% (2/20) of the tested products, indicating possible contamination by chemicals such as heavy metals, disinfectants, or antibiotics. Studies using metagenomics and culture-dependent techniques found microbial species not listed on the product labels in 90% (18/20) of the products, including opportunistic pathogens like Listeria monocytogenes. From all products, 118 distinct bacterial isolates were obtained; 93 of them were examined for antibiotic production vs reference strains. Of them, 67% (62/93), showed inhibitory action suggesting possible antibiotic substances production. Phenotypic antimicrobial resistance was observed, notably against oxytetracycline (28%; 22/80), ampicillin (20%; 16/80), erythromycin (19%; 15/80), and chloramphenicol (9%; 7/80). The nature of this resistance is currently unknown, and it remains unclear whether these isolates exhibit intrinsic resistance or have acquired resistance mechanisms. Furthermore, the erythromycin resistance gene ermK was identified in 27% (3/11) of the isolates. A high prevalence of virulence-associated traits was also detected, including haemolysin (48%; 14/29), DNase (96%; 27/29), protease (65%; 19/29), and gelatinase (79%; 23/29) activity. In vivo virulence assessment using Galleria mellonella larvae demonstrated that 28% (8/28) of the probiotic isolates tested exhibited virulence greater or equal to Vibrio anguillarum Vib79, which is a fish pathogen known to exhbit virulence in this alternative host. These results show that some of the microbial hazards given priority in Chapter 3 are found in a subset of probiotic products currently on the market in Bangladesh, so stressing the vital need of thorough safety assessments before their commercial use in aquaculture. Finally, Chapter 5 examines whether labelling on commercial aquaculture probiotics in Bangladesh can reliably indicate microbial safety. Using the hazard prioritisation framework from Chapter 3 and laboratory data from Chapter 4, twenty products were evaluated for species-level accuracy, completeness of label attributes, and associations with expert-weighted hazard scores. All products showed discrepancies between declared and detected species, including undeclared and potentially pathogenic taxa. Correlation and regression analyses found no significant predictors of hazard scores, with overall label richness showing negligible explanatory value (regression coefficient β = -0.0714, indicating the estimated change in hazard score per unit increase in label richness; R² = 0.0004). Traceability-related attributes (e.g., country of manufacture, producer brand, certification) showed weak trends towards safer profiles, while missing core safety elements—such as microbial species declaration or concentration—were more often linked to higher hazard scores. Multiple correspondence analysis confirmed that absent safety information was a stronger differentiator than promotional claims. These findings indicate that labelling alone cannot provide a robust safety assessment, highlighting the need for independent testing and risk-based labelling standards in low-regulation markets. This thesis concludes that commercial probiotics utilised in aquaculture in Bangladesh may pose several hazards, including microbial contamination, mislabelling, and regulatory deficiencies, which are not identifiable just through label information. These findings highlight the importance of microbiological safety analysis as a fundamental measure in enhancing the implementation of current rules to guarantee the identification and removal of hazardous products from the market.