Skip to main content
Dryad logo

Aquaculture at the crossroads of global warming and antimicrobial resistance

Citation

Reverter, Miriam et al. (2020), Aquaculture at the crossroads of global warming and antimicrobial resistance, Dryad, Dataset, https://doi.org/10.5061/dryad.dv41ns1tr

Abstract

In many developing countries, aquaculture is key to ensuring food security for millions of people. It is thus important to measure the full implications of environmental changes on the sustainability of aquaculture. We conducted a double meta-analysis (460 articles) to explore how global warming and antimicrobial resistance (AMR) impacts aquaculture. We calculated a Multi-Antibiotic Resistance index (MAR) of aquaculture-related bacteria (11,274 strains) for 40 countries, of which mostly low- and middle-income countries present high AMR levels. Here we show that aquaculture MAR indices correlate with MAR indices from clinical bacteria, temperature and countries’ climate vulnerability. We also found that infected aquatic animals present higher mortalities at warmer temperatures. Countries most vulnerable to climate change will probably face the highest AMR risks, impacting human health beyond the aquaculture sector, highlighting the need for urgent action. Sustainable solutions to minimize antibiotic use and increase system resilience are therefore urgently needed.

Methods

Data collection

Literature research strategy

We systematically searched all peer-reviewed journal articles and theses using Web of Science and Google scholar up to March 1st, 2019 that investigated 1) mortalities from cultured aquatic animals due bacterial infections (dataset 1) and 2) AMR from aquaculture environments (dataset 2). Since AMR changes over time, we only retained articles on this subject published within the last 10 years. We performed two independent literature searches for each of the subjects following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Supplementary Figure 2, 3). The following keyword combinations were used: 1) (aquaculture* OR farm* OR rear*) AND (fish OR shrimp OR shellfish) AND (mortality OR outbreak OR infection) AND (Aeromonas OR Edwardsiella OR Flavobacterium OR Streptococc* OR Vibrio OR Yersinia) and 2) (antimicrobial or antibiotic) AND (resistance OR susceptibil*) AND (aquaculture OR fish OR shrimp OR shellfish).

These searches produced a total of 3,526 records for the dataset 1 and 4,512 records for the dataset 2 that were filtered in a three-stage process (Supplementary Figure 1, 2). After removal of duplicates, issued from combining several database searches, title and abstract of the remaining records (2,458 for dataset 1 and 2,556 for dataset 2) were scanned for relevance in the studied topics. Then, the full-text of the retained articles (837 for dataset 1 and 697 for dataset 2) were assessed.

 

Inclusion criteria and data extraction

Dataset 1: Only research articles where an experimental infection was performed with a clear identified protocol were considered. Natural outbreaks were not considered due to the difficulty of determining 1) whether a previous treatment (e.g. vaccine or antibiotic) was applied, 2) exact temperature during the duration of the outbreak and 3) whether the outbreak was uniquely caused by one clearly identified bacterial pathogen. All selected studies met the following criteria: 1) experimental infections were performed with pure bacterial cultures previously characterized, 2) dose of infection and mode of infection were clearly identified, 3) the life stage of the organism infected was reported, 4) temperature during the duration of the outbreak was clearly reported and constant (± 1°C), 5) the animal mortality was reported as % and 6) aquatic infected animals were not exposed to any substance or stress that might have interfered with the mortality outcome When a study included several experiments under different temperatures, host species or pathogen species, we considered them distinct observations. Following, all the aforementioned criteria we obtained a dataset containing 582 observations extracted from 273 studies (Supplementary Figure 2, 4, Supplementary data 1). For each of the observations we extracted the following data: pathogen and host taxonomy (species, family and phylum) host developmental stage (larvae, juvenile, adult), country, temperature of the infection, cumulative mortality, mode of infection (injection or immersion) and infective dose.

Dataset 2: Only research articles reporting antimicrobial resistances of bacteria isolated directly from the aquaculture environment (cultured animals recovered at the farmed site, water or sediment) were considered. Articles reporting antimicrobial resistances of isolated bacteria from cultured animals recovered in any other site than the farming environment, such as retail markets or imported products, were not included to avoid a bias introduced by potential contamination during transport. All selected studies met the following criteria: 1) antimicrobial activity of bacterial strains was reported for at least 3 antibiotics, 2) at least the bacterial genus was identified in order to be able to disregard susceptibilities to antibiotics for which they are naturally resistant (Supplementary Table 20) and 3) bacteria studied were known as pathogenic for aquatic cultured animals. Since Pseudomonas species are known to present numerous intrinsic AMR, they were excluded from the analysis in order to avoid biased results. For the calculation of the countries’ MAR indices, we established a minimum requirement of 30 bacterial strains. This led to a dataset that contained antimicrobial resistances of 11,274 strains extracted from 187 studies (Supplementary Figure 5). For each of these studies the following information was extracted: country of the study, bacterial species or genus, source of isolation (host species or type of farm), number of antibiotics tested and number of resistant strains.

Funding

French National Research Institute for Development (IRD)