Antibiotic residues from aquaculture farms and their ecological risks in Southeast Asia: a case study from Malaysia

ABSTRACT

Background and Objectives: One major source of antibiotic contamination in the sea is from aquaculture. We monitored the concentration of commonly used antibiotic classes and antibiotic resistance genes (tet(M), sul1, sul2 and sul3) in aquaculture farms in Peninsular Malaysia.

Methods: Antibiotic residues and resistance genes were quantified using high-performance liquid chromatography and real-time PCR respectively. Risk quotients in European technical guidance document on risk assessment was used to assess the potential environmental risk.

Results: We detected 23 antibiotics with tetracyclines, sulfonamides and quinolones were the most frequently detected classes, indicating a wide distribution of antibiotics in Malaysian aquaculture farms. The dendrogram and heatmap revealed three groups of antibiotic concentration patterns but with no differences in the types of antibiotics usage among aquaculture farms. The ARGs (10⁻³ copies/16S) were detected in >90% of the sites except for sul3. Ciprofloxacin, enrofloxacin, norfloxacin and lincomycin posed risks to cyanobacteria and algae in Kelantan, Perak and Pahang.

Conclusion: Relative to Asian aquaculture farms, the residues detected here were at low or moderate levels except for quinolones. This study will be useful to develop effective management of aquaculture wastewater in order to mitigate antibiotic pollution and transmission of ARGs to humans through the food chain.

KEYWORDS:

• Antibiotic residues
• resistance genes
• risk assessment
• aquaculture

Introduction

Aquaculture plays an important role as a main source of animal protein in global diets (FAO 2016; Mohd et al. 2017; Department of Fisheries (DOF) 2019). In order to meet the demand of the world’s growing population and to achieve sustainable food production and security, aquaculture production will have to be increased by an additional 46.4 million metric tons by 2030 (World Bank 2013). Asia is referred to as “home of aquaculture,” accounting for about 89% of global production in 2016 (FAO 2020). In terms of aquaculture production, Malaysia is ranked 15th in the world and 6th in Asia with an estimated production of 427,022.66 metric tonnes worth USD 731.81 million (FAO 2016; Dermawan 2019). In many Southeast Asia countries, aquaculture production has contributed significantly to their national economies [Gross Domestic Product (GDP), 0.2˗5% increase]. Therefore, aquaculture industry has been recognized as a potential pillar to strengthen economic growth (Lundgren et al. 2006; SEAFDEC 2017).

However, one of the major threats to the aquaculture industry worldwide is bacterial infection. More than USD 6 billion per annum is lost from the aquaculture industry due to disease (Stentiford et al. 2017). Both extensive and intensive aquaculture farming have greatly enhanced the transmission opportunities for waterborne pathogens that can spread at faster rates compared with terrestrial systems (McCallum, Harvell, and Dobson 2003). For example, Vibrio parahaemolyticus, the causative agent of acute hepatopancreatic necrosis disease (AHPND) and formerly known as early mortality syndrome (EMS) causes devastating losses that reached billions of dollars annually since its first outbreak in Southern China in 2009 (Lightner et al. 2012a, 2012b; Tran et al. 2013). This disease is rapidly spreading and has affected several countries in Southeast Asia consecutively, e.g., Vietnam in the year 2010, Malaysia (2011), Thailand (2012), Philippines (2013), and has even spread to the Americas e.g., Mexico in 2013 (Tran et al. 2013; Nunan et al. 2014; De La Peña et al. 2015).

In order to treat and prevent bacterial diseases in aquaculture, antibiotics are commonly used as therapeutic and/or prophylactic agents. Tetracyclines, sulfonamides, oxolinic acid and erythromycin are commonly used antibiotics in aquaculture farming (ASEAN, 2013). These antibiotics are permitted for use in food producing animal based on the recommended Maximum Residue Level (MRL) set by joint Food and Agriculture Organization of the United Nations and the World Health Organization (FAO/WHO), Codex Alimentarius Commission, and European Union legislation (FAO and WHO 2020). However, MRLs differ between geographic regions depending on the antibiotic usage profiles and local food safety regulatory agencies. Moreover, most Southeast Asian countries lack legislation, regulatory surveillance and monitoring systems on the use of antibiotics (Chuah et al. 2016; FAO 2016). Antibiotic contamination continues to be found in the environment and aquaculture products (Le and Munekage 2004; Lin, Yu, and Lin 2008; Oliveira et al. 2014; Xiong et al. 2015; Lai et al. 2018). Although Malaysia has banned the use of nitrofurans and chloramphenicol in aquaculture farming, the United States Food and Drug Administration (FDA) continues to detect both these residues in seafood from Malaysia, in which 44 cases were reported between 2009 and 2018 (Food and Drug Administration 2018).

Aquaculture waste has been identified as one of the main contributors of antibiotic pollution in the environment (De Jesus Gaffney et al. 2016) as the infrastructure for proper aquaculture waste management is critically lacking. Furthermore, many countries, particularly those in the developing countries have yet to develop standards on concentrations of antibiotics discharge from wastewater effluents (Sasikaladevi, Kiruthika Eswari, and Nambi 2020). Therefore antibiotics, antibiotic resistance bacteria (ARB) and antibiotic resistance genes (ARGs) from aquaculture are released directly to the environment. These chemical and biological pollutants can impact public health and marine ecosystems (World Health Organization (WHO) 2018). However, quantitative data on the residual levels of antibiotic and antimicrobial resistance in water samples from aquaculture remain scarce (Managaki et al. 2007; Suzuki and Hoa 2012; Shimizu et al. 2013; Yan et al. 2018). Current levels of antibiotic use in aquaculture worldwide are difficult to determine as different countries have different distribution and standards to assess the pollution levels (Romero, Feijoó, and Navarrete 2012). With limited information on the level of contamination of antibiotics in Malaysian aquaculture farms, the potential risk of residual antibiotic toward the ecosystem remains unclear. Hence, the aim of this study is to examine the distribution and composition of antibiotics in aquaculture farms, and their ecological risk, as well as determine the prevalence of ARGs in bacteria from aquaculture farms.

Materials and methods

Sampling

For this study, we obtained permission from 29 aquaculture farms located at the seven main aquaculture production states (Perak, Selangor, Pahang, Kelantan, Penang Island, Malacca and Johor) of Peninsular Malaysia (Figure 1, Department of Fisheries (DOF) 2019). During sampling, we were only allowed to collect the surface water using a stainless-steel bucket. The water samples were then passed through a 20 μm mesh net and stored into a clean 2 L amber glass bottle. Samples were kept cold during transportation before further processing in the laboratory.

Chemical and standards

Twenty-five commonly detected antibiotics [trimethoprim (TMP), ciprofloxacin (CIP), enrofloxacin (ENRO), ofloxacin (OFX), norfloxacin (NOR), nalidixic acid (NAL), carbadox (CAR), lincomycin (LIN), azithromycin (AZM), clarithromycin (CTM), erythromycin-H₂O (ETM), roxitromycin (RTM), tylosin (TYL), sulfadimethoxine (SMA), sulfapyridine (SPD), sulfathiazole (STZ), sulfamethoxazole (SMX), sulfamethazine (SMT), sulfamerazine (SMR), sulfamethizole (SMZ), doxycycline (DOX), minocycline (MNC), chlortetracycline (CTC), oxytetracycline (OTC) and tetracycline (TC)] that belong to six antibiotics classes were selected: diaminopyrimidine, quinolones, macrolides, sulfonamides, tetracyclines and others (lincomycin, carbadox), and analyzed according to Shimizu et al. (2013). Standards for SPD, RTM, TC, MNC and NOR were obtained from Sigma-Aldrich Co. (St. Louis, Mo, USA), SMX, SMR, SMA, SMT, TMP, CTM, OTC, CTC, NAL, and CAR were from Wako Pure chemicals Co. (Osaka, Japan), STZ, SMZ, ETM, TYL, and LIN from Honeywell Riedel-de Haen Co. (Seelze, Germany), AZM from LKT laboratories Co. (St Paul, USA), DOX from ICN Biomedicals Co. (Santa Ana, USA), OFX and ENRO from Hayashi pure chemicals Co. (Osaka, Japan). Oxytetracycline-¹³C,d₃, sulfamethoxazole-d₄, clarithromycin-d₃, roxithromycin-d₉ and norfloxacin-d₅ were used as surrogate standards and were purchased from Hayashi pure chemicals Co. (Osaka, Japan). All the above antibiotic standards were prepared by dissolution in methanol and kept at −20°C, and all solvents were of HPLC grade. Methanol, acetonitrile, formic acid (>99.5%) and ethylenediamine tetraacetic acid disodium (Na₂EDTA) were obtained from Wako Pure Chemicals (Osaka, Japan). Ultra-pure water was used in this study (Milli-Q ultrapure water system, Millipore, Bedford, MA, USA).

Extraction and quantification of antibiotic residues in water

One to two Liter water samples were measured precisely and filtered through glass fiber filters (GF/F, Sartorius, Gӧttingen Germany) with a nominal pore size of 0.7 µm. The pH of the filtrate was then adjusted to pH 3 with 3 M sulfuric acid, followed by the addition of 0.2 g Na₂EDTA as chelating agent. Targeted antibiotic was extracted by Solid Phase Extraction (SPE) using Oasis® Hydrophilic-lipophilic balance (HLB) extraction cartridges (500 mg, Waters, USA) and the Visiprep^TM SPE vacuum manifold (Supelco, USA). Prior to extraction, the cartridge was pre-treated with 6 mL each of methanol, ultra-pure water and 10 mM acidified Na₂EDTA buffer. Filtrate was then loaded and passed through the SPE cartridge at a flow rate of 10 min/mL. After loading the filtrate, 10 mL of acidified ultra-pure water was used to wash the cartridge followed by drying the cartridge under nitrogen flow for 30 minutes. Then, the analyte was eluted with methanol (2 mL) containing 0.1% (v/v) formic acid four times. A 50 µL of surrogate mixture (clarithrobycin-d₃, norfloxacin-d₅ (500 ng/mL each, in methanol), oxytetracycline-¹³C, -d₃, roxithromycin-d₉ and sulfamethoxazole-d₄,) was then spiked in the eluent. A rotary evaporator was used to concentrate the eluent to an approximate volume of 0.5 mL and kept in amber vial. The concentrated eluent was then dried completely under gentle nitrogen gas at 35°C and reconstituted with H₂O/acetonitrile (94:6 v/v) containing 0.1% formic acid to a final volume approximately 20 mL – 100 mL, providing a pre-concentration factor of 10 to 100.

A liquid chromatograph (LC) (Accela, Thermo Fisher Scientific, Yokohama, Japan) was used to determine and quantify TCs. The LC system was equipped with a tandem mass spectrometer (LC-MS/MS) (Quantum Access, Thermo Fisher Scientific, Yokohama, Japan) using a positive electrospray ionization (ESI) operating in positive mode with selected reaction monitoring (SRM) mode. Separation of TCs was achieved on a Waters Xterra MS C18 column (2.1 mm ID x 50 mm with particle size 2.5 µm; Waters, USA) in combination with a Waters Xterra MS C18 guard column (2.1 mm ID x 20 mm with particle size 3.5 µm, Waters, USA). An injection of 20 µL and a binary solvent gradient system with a flow rate of 0.2 mL/min were used. Solvent A was 0.1% formic acid in water and solvent (B) was acetonitrile. The separation of TCs were carried out using the following gradient program: initiated with 5% B for 5 min, followed by increasing solvent B in a linear pattern to 95% over 11 min. A 17 min post time allowed re-equilibration of column, before the initial gradient composition was reestablished for the next analysis. The collision energy and isolation width (m/z) for precursor ion and two product ions are summarized in Table S1.

The identification and quantification of antibiotics were determined by comparing the retention times and the area ratios of the two product ions in each sample with the standard. The accepted margin of error for the retention time and the area ratio of the two product ions between the sample and standard was within 0.3 min and 20% area ratio, respectively. A seven point (1, 3, 5, 10, 30, 50 and 100 ng/mL) external calibration curves was generated for quantification and used at regular interval. The linearity correlation index (R²) was above 0.99. The final concentration for the majority of samples were within the calibration lines. If the final concentration was below the lowest standard concentration (1 ng/mL), the sample’s concentration was calculated by extrapolation. The concentration of the selected antibiotics was adjusted against the recovery of the surrogates as indicated in Table S2.

Analytical performance

Based on successive dilution of the standard mixture solution, 0.03 ng/mL was determined as the lowest concentration of reliable detection for all the target antibiotics except for TYL and tetracyclines where 0.3 ng/mL was the lowest concentration of reliable detection. Considering the highest preconcentration factor (i.e., 100), the limit of detection (LOD) was determined at 0.3 ng/L of sample water for all the target antibiotics except for TYL and tetracyclines that have a LOD of 3 ng/L. A blank was run together with each batch of sample analysis and the limit of quantification (LOQ) was established based on 10 times blank value. LOQ were normally 20 ng/L for the target antibiotics except for TYL and tetracyclines with 200 ng/L of LOQ.

Effluent from a sewage treatment plant (STP) was analyzed in triplicate to determine the reproducibility. Relative standard deviations (RSD) of concentrations of the selected antibiotics were < 30% except for tetracycline whose concentrations were below LOQ (Table S2). Higher deviations for tetracyclines were ascribed to their lower sensitivity and low concentrations in sewage effluents. For the aquaculture samples, SPE was done on-site and spiked with surrogate after elution of target compounds from SPE cartridge. Therefore, extraction efficiency on SPE was not corrected, though loss during evaporation and transfer and matrix effects on LC-MS/MS analysis were corrected by using surrogates. Thus, we checked the extraction efficiency via the analysis of sewage effluents spiked with or without native standards before SPE. Recoveries i.e., extraction efficiencies, ranged from 78% to 132% (Table S2) and our reported concentrations were reliable with this range of accuracy.

Quantitative PCR of antibiotic resistance genes (ARGs)

In this study, we assessed the ARGs for sulfonamide and tetracycline resistance genes due to the long use of these antibiotics in human and veterinary clinics. Among the tetracycline resistance genes, tet(M) was chosen as it is suspected to have the widest host range (Roberts, Schwarz, and Aarts 2012) and its origin date back to ancient times (Kobayashi et al. 2007). tet(M) also has a high genetic diversity (Rizzotti et al. 2009) and is highly distributed in the environment (D’Costa et al. 2011). For sulfonamide resistance genes, sul1, sul2 and sul3 were selected.

For ARGs quantification, the total DNA of natural assemblages of bacteria were captured on 0.2 µm polycarbonate membrane filter (Merck Millipore, Germany). The targeted ARG was extracted according to Suzuki et al. (2013). Quantitative PCR (qPCR) was conducted using CFX 96^TM Real-Time system (Bio-Rad, Laboratories, Hercules, CA, USA) for four selected ARGs: tet(M), sul1, sul2 and sul3. These four genes were prevalent ARGs in aquatic environments (Suzuki et al. 2013, 2015), and thus appropriate representatives of ARGs. The 16S ribosomal RNA (16S rRNA) gene was also analyzed to quantify the total bacteria in the collected samples. qPCR amplification was performed in a reaction mixture containing of 1X SsoFast^TM EvaGreen® Supermix (Biorad, Laboratories, Hercules, CA, USA), 500 nM of each primer, 1 μL template DNA and sterile Milli-Q water (Merck Millipore, Germany) in a total volume of 20 μL. Each sample was measured in triplicates. The amplification condition and primer sequences are listed in Table 1. A known copy number of plasmid DNA that carried the cloned target genes was used to generate the standard curve. Ten times serial dilution was performed to generate a five-point standard curve in triplicate for each qPCR analysis. The ARGs were normalized to 16S rRNA (Copies/16S) and used to report and discuss the results.

Table 1. qPCR primer sequences, target and conditions of reactions

Target gene	Primers	Sequences	Conditions	References
16S rRNA	Bact1369F Bact1492R	cggtgaatacgttcycgg ggwtaccttgttacgactt	95°C for 30 sec (1 cycle); 95°C for 5 sec and 50°C for 10 sec (40 cycles)	Suzuki, Taylor, and DeLong (2000)
tet(M)	tet(M) F tet(M) R	gcaattctactgatttctgc ctgtttgattacaatttccgc	95°C for 30 sec (1 cycle); 95°C for 10 sec and 57°C for 20 sec (40 cycles)	Tamminen et al. (2011)
sul1	qsul 1 F qsul 1 R	ccgttggccttcctgtaaag ttgccgatcgcgtgaagt	95°C for 30 sec (1 cycle); 95°C for 5 sec and 51°C for 10 sec (40 cycles)	Heuer and Smalla (2007)
sul2	qsul 2 F qsul 2 R	cggctgcgcttcgatt cgcgcgcagaaaggatt		Heuer et al. (2008)
sul3	qsul 3 F qsul 3 R	tccgttcagcgaattggtgcag ttcgttcacgccttacaccagc	95°C for 30 sec (1 cycle); 95°C for 5 sec and 60°C for 20 sec (40 cycles)	Pei et al. (2006)

For all genes, melting curves was at 60°C – 95°C for 5 sec/step

Ecological risk assessment

For the potential ecological effect brought by the detected antibiotic toward the environment, a Risk Quotient (RQ) calculation was formulated following the European technical guidance document on risk assessment (European Commission 2003). The risk quotient (RQ) was calculated with the formula shown below:

$RQs=\frac{Measured\hspace{0.17em}environmental\hspace{0.17em}concentration\hspace{0.17em}\left(MEC\right)}{Predicted\hspace{0.17em}no-effect\hspace{0.17em}concentration\hspace{0.17em}\left(PNEC\right)}$

The value of assessment factor was decided based on the type of toxicity data EC50/LC50 (European Commission 2003). In this study, the toxicity data of each selected antibiotic on four types of aquatic live (algae, bacteria, fish and invertebrate) were used from the literature review of toxicological studies (Table S3). The predicted no-effect concentration (PNEC) is the division of EC50/LC50 and assessment factor. The RQ were classified into three levels of risk according to Hernando et al. (2006) with RQ value more than 1 categorized as high risk. RQ value in the range of 0.1 and 1 is categorized as medium risk, while RQ value less than 0.1 is under the level of low risk.

Statistical analysis

Correlation and linear regression analysis were conducted to analyze the effect between antibiotic residue concentration and antibiotic resistance genes detected in aquaculture farm. Data of antibiotic residue concentration were subjected to logarithmic transformation. Data were assessed by one-way ANOVA and Tukey test using PAST version 4.03 (Hammer, Harper, and Ryan 2001). Significant differences between antibiotic residue concentration and ARG were evaluated, and p < 0.05 was considered statistically significant. RStudio version 1.4.1106 was used to generate dendrogram and heatmap to conduct cluster analysis and identify the distribution, the pattern of usage and concentration of antibiotic residues in aquaculture farm.

Results

Antibiotic residues

Twenty-three antibiotics belonging to six classes were detected in Malaysian aquaculture farms, including seven sulfonamides (SPD, STZ, SMR, SMT, SMZ, SMX, SMA), five quinolones (CIP, ENRO, OFX, NOR, NAL), four tetracyclines (MNC, OTC, TC, DOX), five macrolides (AZM, TYL, ETM, CTM, RTM), TMP and LIN. The concentration of detected antibiotics ranged from < LOQ to 9.58 × 10⁵ ng/L (Table 2). Tetracyclines had the highest detection frequencies (83%) followed by sulfonamides (72%) and quinolones (69%).

Table 2. The concentration of tetracyclines, sulfonamides, quinolones, macrolides, trimethoprim, lincomycin and carbadox in surface water of aquaculture farm

Location		Antibiotic ng/L
		Tetracyclines (TCs, 83%)						Sulfonamides (SAs, 72%)
		MNC (24%)	OTC (41%)	TC (31%)	CTC	DOX (17%)	Mean	SPD (17%)	STZ (20%)	SMR (34%)	SMT (17%)	SMZ (13%)	SMX (17%)	SMA (10%)	Mean
Johor	J1	<LOQ	n.d.	n.d.	n.d.	n.d.	6.91	n.d.	2.55	<LOQ	n.d.	n.d.	n.d.	1.40	1.32
	J2	n.d.	<LOQ	n.d.	n.d.	n.d.	7.10	n.d.	<LOQ	n.d.	n.d.	n.d.	n.d.	n.d.	0.49
	J3	n.d.	n.d.	<LOQ	n.d.	n.d.	4.70	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	-
	J4	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	n.d.	n.d.	3.35	n.d.	3.35
	J5	5.05 x 10^1	<LOQ	n.d.	n.d.	n.d.	30.41	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	-
	J6	n.d.	n.d.	2.31 x 10^1	n.d.	n.d.	23.04	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	-
Malacca	M1	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	-
	M2	n.d.	<LOQ	n.d.	n.d.	<LOQ	5.09	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	-
	M3	n.d.	<LOQ	n.d.	n.d.	<LOQ	4.23	n.d.	n.d.	1.57	n.d.	n.d.	n.d.	n.d.	1.57
	M4	2.43 x 10^1	n.d.	n.d.	n.d.	n.d.	24.26	n.d.	<LOQ	1.23	n.d.	n.d.	3.78	n.d.	1.87
Selangor	S1	n.d.	<LOQ	<LOQ	n.d.	n.d.	4.61	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	-
	S2	n.d.	<LOQ	n.d.	n.d.	n.d.	6.30	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	-
Kelantan	K1	n.d.	<LOQ	n.d.	n.d.	n.d.	3.88	n.d.	n.d.	n.d.	n.d.	n.d.	1.94	n.d.	1.94
	K2	n.d.	<LOQ	n.d.	n.d.	n.d.	111.60	<LOQ	n.d.	<LOQ	n.d.	n.d.	n.d.	n.d.	0.14
	K3	n.d.	<LOQ	n.d.	n.d.	n.d.	26.39	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	-
Perak	P1	n.d.	n.d.	2.00 x 10^1	n.d.	<LOQ	12.87	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	4.03
	P2	n.d.	<LOQ	n.d.	n.d.	n.d.	168.82	1.67 x 10^−1	5.20 x 10^−1	4.71 x 10^−1	n.d.	n.d.	n.d.	n.d.	0.39
	P3	1.22 x 10^1	n.d.	n.d.	n.d.	n.d.	12.18	n.d.	n.d.	3.33 x 10^−1	n.d.	n.d.	n.d.	n.d.	0.33
	P4	n.d.	n.d.	<LOQ	n.d.	n.d.	0.99	n.d.	n.d.	1.18	n.d.	n.d.	n.d.	1.57	1.37
	P5	n.d.	n.d.	n.d.	n.d.	2.34 x 10^2	234.23	1.20	n.d.	n.d.	n.d.	4.79	n.d.	n.d.	2.99
	P6	2.45 x 10^2	<LOQ	7.30 x 10^1	n.d.	n.d.	180.62	2.91 x 10^1	n.d.	4.20	n.d.	n.d.	n.d.	n.d.	16.66
	P7	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	3.93 x 10^−1	n.d.	8.36 x 10^−1	n.d.	n.d.	0.61
Pahang	PA1	n.d.	n.d.	n.d.	n.d.	<LOQ	0.55	n.d.	n.d.	<LOQ	n.d.	n.d.	n.d.	n.d.	0.05
	PA2	<LOQ	n.d.	1.35 x 10^1	n.d.	n.d.	20.42	n.d.	2.82 x 10^2	3.83	5.68	n.d.	n.d.	n.d.	97.30
	PA3	n.d.	n.d.	<LOQ	n.d.	n.d.	0.49	n.d.	n.d.	n.d.	n.d.	1.11	3.21	n.d.	2.16
	PA4	n.d.	n.d.	n.d.	n.d.	n.d.	-	<LOQ	n.d.	n.d.	2.21	3.63	n.d.	n.d.	2.00
Penang Island	PI1	n.d.	n.d.	<LOQ	n.d.	n.d.	4.65	n.d.	<LOQ	<LOQ	7.21 x 10^−1	n.d.	1.97	n.d.	0.70
	PI2	<LOQ	<LOQ	n.d.	n.d.	n.d.	6.50	n.d.	n.d.	n.d.	1.14	n.d.	n.d.	8.54 x 10^−1	1.00
	PI3	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	2.98	n.d.	n.d.	n.d.	2.98
Location		Antibiotic ng/L
		Quinolones (FQs, 69%)						Macrolides (MLs, 55%)						Others			∑Antibiotic
		CIP (14%)	ENRO (52%)	OFX (21%)	NOR (14%)	NAL (28%)	Mean	AZM (17%)	TYL (7%)	ETM (44%)	CTM (13%)	RTM (7%)	Mean	TMP (10%)	LIN (55%)	CAR
Johor	J1	n.d.	3.08	2.75 x 10^−1	2.51 x 10^1	n.d.	9.47	n.d.	n.d.	<LOQ	<LOQ	n.d.	0.15	n.d.	n.d.	n.d.	3.96 x10^1
	J2	n.d.	1.81	n.d.	n.d.	n.d.	1.81	6.88 x 10^−1	n.d.	n.d.	n.d.	n.d.	0.69	n.d.	<LOQ	n.d.	1.04 x 10^1
	J3	n.d.	4.56	n.d.	n.d.	n.d.	4.56	1.54	n.d.	<LOQ	<LOQ	n.d.	0.63	n.d.	n.d.	n.d.	1.11 x 10^1
	J4	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	<LOQ	n.d.	3.77
	J5	n.d.	n.d.	n.d.	n.d.	2.21 x 10^1	22.06	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	<LOQ	n.d.	8.31 x 10^1
	J6	n.d.	n.d.	n.d.	n.d.	1.84 x 10^1	18.42	n.d.	n.d.	<LOQ	2.7 x 10^−1	n.d.	0.36	n.d.	7.47 x 10^1	n.d.	1.17 x 10^2
Malacca	M1	n.d.	n.d.	n.d.	n.d.	3.12 x 10^1	31.20	n.d.	n.d.	<LOQ	n.d.	n.d.	0.86	n.d.	n.d.	n.d.	3.21 x 10^1
	M2	n.d.	n.d.	n.d.	n.d.	n.d.	-	<LOQ	n.d.	<LOQ	n.d.	<LOQ	0.33	n.d.	<LOQ	n.d.	1.12 x 10^1
	M3	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	1.00 x 10^1
	M4	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	2.99 x 10^1
Selangor	S1	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	<LOQ	n.d.	<LOQ	0.48	n.d.	<LOQ	n.d.	1.04 x 10^1
	S2	n.d.	<LOQ	n.d.	n.d.	n.d.	-	n.d.	n.d.	<LOQ	<LOQ	n.d.	0.85	n.d.	<LOQ	n.d.	8.28
Kelantan	K1	1.46 x 10^1	3.73	<LOQ	<LOQ	n.d.	0.23	n.d.	n.d.	<LOQ	n.d.	n.d.	0.28	n.d.	n.d.	n.d.	2.88 x 10^1
	K2	n.d.	4.93	n.d.	3.56 x 10^1	n.d.	5.67	n.d.	2.28 x 10^−1	<LOQ	n.d.	n.d.	0.21	n.d.	3.77 x 10^1	n.d.	1.91 x 10^2
	K3	n.d.	n.d.	n.d.	n.d.	n.d.	20.25	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	2.64 x 10^1
Perak	P1	n.d.	5.55 x 10^1	5.45 x 10^−1	n.d.	n.d.	-	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	8.58 x 10^1
	P2	n.d.	1.64 x 10^1	n.d.	n.d.	n.d.	28.03	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	1.86 x 10^2
	P3	n.d.	n.d.	n.d.	n.d.	n.d.	16.43	<LOQ	n.d.	n.d.	n.d.	n.d.	0.34	n.d.	1.80 x 10^−1	n.d.	1.30 x 10^1
	P4	n.d.	n.d.	n.d.	n.d.	1.62 x 10^1	-	n.d.	n.d.	<LOQ	n.d.	n.d.	1.14	5.24 x 10^−1	1.60 x 10^−1	n.d.	2.17 x 10^1
	P5	n.d.	8.27 x 10^−1	n.d.	n.d.	1.03 x 10^2	16.17	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	3.44 x 10^2
	P6	1.31 x 10^5	9.58 x 10^5	n.d.	6.67 x 10^3	9.46 x 10^2	51.72	6.95	n.d.	n.d.	n.d.	n.d.	6.95	n.d.	9.05 x 10^−1	n.d.	1.09 x 10^6
	P7	8.61 x 10^1	2.13 x 10^2	n.d.	n.d.	<LOQ	274 x 10^3	n.d.	n.d.	<LOQ	n.d.	n.d.	1.11	n.d.	1.84	n.d.	3.03 x 10^2
Pahang	PA1	n.d.	n.d.	5.33 x 10^−1	n.d.	n.d.	0.53	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	LOQ	n.d.	1.16
	PA2	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	5.40 x 10^−1	n.d.	3.33 x 10^2
	PA3	n.d.	1.08	n.d.	n.d.	n.d.	1.08	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	5.89
	PA4	1.58 x 10^1	3.55	<LOQ	n.d.	1.93 x 10^1	9.72	n.d.	n.d.	n.d.	n.d.	n.d.	-	4.68	n.d.	n.d.	4.95 x 10^1
Penang Island	PI1	n.d.	<LOQ	2.94 x 10^−1	n.d.	n.d.	0.33	n.d.	n.d.	<LOQ	n.d.	n.d.	0.53	n.d.	<LOQ	n.d.	8.76
	PI2	n.d.	7.46	n.d.	n.d.	n.d.	7.46	n.d.	<LOQ	<LOQ	n.d.	n.d.	0.46	3.47 x 10^−1	<LOQ	n.d.	2.42 x 10^1
	PI3	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	n.d.	n.d.	n.d.	n.d.	-	n.d.	<LOQ	n.d.	3.23

LOQ = Limit of quantification, n.d. = non- detected, Mean and sum were calculated using the measured values if above the LOQ, the 1/2 MQL if < LOQ and 0 if not detected

^aJ = Johor, M = Malacca, S = Selangor, K = Kelantan, P = Perak, PA = Pahang, PI = Penang Island

For the tetracycline compounds tested, OTC was the most frequently detected (41%) but the concentrations were less than LOQ. TC and MNC were detected in the range <LOQ – 73 ng/L and <LOQ – 245 ng/L, respectively. TC was detected in the farms from Johor (J6: 2.3 ng/L), Perak (P1: 2.0 ng/L and P6: 7.3 ng/L) and Pahang (PA2: 1.4 ng/L). MNC was detected in Johor (J1: < LOQ, J5: 5.1 ng/L), Malacca (M4: 2.4 ng/L), Pahang (PA2:< LOQ), Penang Island (P12: < LOQ), where the highest concentration was recorded in Perak (P6: 245 ng/L,). DOX was only detected in one farm located in Perak (P5: 234 ng/L) whereas CTC was not detected in any of the farms.

All the sulfonamide compound tested was detected (< LOQ to 282.4 ng/L) in all the states except for Selangor. SMR (41%) and STZ (21%) were the most frequently detected. The highest concentration of STZ (282.4 ng/L) and SPD (29 ng/L) were found in Pahang (PA2) and Perak (P6), respectively whereas the concentration of other sulfonamides compounds in most of the farm (95%) were less than 6 ng/L. SMT was only detected in farms located in Pahang (PA2: 5.68 ng/L and PA4: 2.21 ng/L) and Penang Island (PI1, PI2, PI3: 0.72 ng/L, 1.14 ng/L and 2.98 ng/L) whereas SMZ was present in Pahang (PA3: 1.11 ng/L and 3.63 ng/L) and Perak (P5:4.79 ng/L and P7:0. 84 ng/L).

The five tested quinolone antibiotics were detected with concentrations ranging from <LOQ – 9.58 × 10⁵ ng/L. ENRO was the most frequently detected (52%) followed by NAL (28%), OFX, (21%), CIP (14%) and NOR (14%). CIP, ENRO and NAL were found dominant in Perak while NOR and OFX were dominant in Kelantan and Pahang, respectively. The highest concentration of ENRO, CIP, NOR and NAL were detected in P6 in Perak with concentrations at 9.58 × 10⁵ ng/L, 1.31 × 10⁵ ng/L, 6.7 × 10³ ng/L and 946 ng/L, respectively.

Macrolides (AZM, ETM, CTM, TYL and RTM) were found in notably low concentrations ranging from <LOQ – 6.9 ng/L and accounted for the lowest total concentration (20 ng/L). No macrolides were detected in Pahang. TMP was only detected in Pahang (PA4), Perak (P4) and Penang Island (PI2) at concentrations of 4.7 ng/L, 0.5 ng/L and 0.4 ng/L, respectively. LIN was found in all states (<LOQ – 74.7 ng/L) with the highest concentration detected in Pahang (PA6: 74.7 ng/L). CAR was not detected in all the water samples at all sites.

Antibiotic resistance genes

The sul genes detected among the aquaculture farms ranged from 10⁻⁷– 10°copies/16S (Figure 2). Among the targeted sul genes, sul2 was the most abundant (10⁻⁶– 10°copies/16S) followed by sul1 (10⁻⁵ – 10⁻¹copies/16S). The sul3 was the least abundant (10⁻⁷–10⁻⁴ copies/16S) or not detectable at most sites. The co-existence of sul genes were observed in which both sul1 and sul2 were predominantly found in 93% of the aquaculture farms with the exception of P1, Perak where only sul1 was present and PA1, Pahang where only sul2 was present. In contrast, the abundance of tet(M) (2.36 × 10⁻⁵ – 3.12 × 10⁰ copies/16S) was higher than sul genes. The highest abundance of tet(M) were detected in two farms located in Perak (P6: 2.42 × 10⁰ copies/16S and P7: 3.12 × 10⁰ copies/16S) (Figure 2).

Environmental ecological risk

In this study, the detected antibiotics posed negligible risk to fish. However, for quinolones, in which CIP detected in P6 and P7, Perak, K1 in Kelantan and PA4 in Pahang, exhibited high risk to Microcystis aeruginosa. ENRO detected in Johor (J1, J2, J3), Kelantan (K1, K2), Penang Island (PI1, PI2), Perak (P2) and Pahang (PA4) posed medium risk to Vibrio fischeri whereas in Perak (P1, P6 and P7), ENRO posed high risk. NOR posed medium risk to Vibrio fischeri at Kelantan (K1) and Johor (J1) whereas at P6, Perak, high risk was detected (Table 3).

Table 3. The calculated risk quotients (RQs) for the selected antibiotic detected in aquaculture farm

Red = high risk, Yellow = medium risk, Green = low risk, Grey = no antibiotic was detected.

* no toxicology data available

In contrast, TC only posed medium risk to Microcystis aeruginosa at Perak (P1, P6), Johor (J6) and Pahang (PA2) whereas DOX posed medium risk to Synechococcus leopoliensis at P5, Perak. Among the sulfonamides, only SMX detected at Johor (J4), Malacca, (M4), Kelantan (K1), Pahang (PA3) and Penang Island (PI1) posed medium risk to Synechococcus leopoliensis. LIN and CTM posed medium to high risk to Pseudokirchneriella subcapitata at Johor (J6) and Kelantan (K2), respectively. RQ was not calculated for NAL, MNC and SMZ due to the lack of toxicology data.

Discussion

Antibiotics in aquaculture waters

In our study, 23 antibiotic residues belonging to six classes were identified (total concentration: 1.099 × 10⁶ ng/L) in which tetracyclines, sulfonamides and quinolones were the most prevalent antibiotics detected suggesting the wide usage of these antibiotics in aquaculture farms in Malaysia (Table 2). Other studies in Asia have reported the frequent use of similar antibiotic compounds in aquaculture production (Rico et al. 2012; Lulijwa, Rupia, and Alfaro 2019). The dendrogram and heatmap revealed three distinct clusters for the 23 antibiotics with different concentrations patterns (Figure 3). Cluster I contained two antibiotics CIP and ENRO with higher concentrations in P6. The second cluster OTC, NAL, MNC and NOR with moderate and high concentrations while third cluster contains 17 antibiotics with the lowest concentrations. However, farms that used specific combination of various antibiotics were not specific within and across the regions. When we analyzed among the farms sampled in this study, no difference in the types of antibiotics usage among aquaculture farms (p > 0.05) were observed except for P6 farm in Perak which was notably different from the other farms with the highest concentration of antibiotics detected.

The variation in distribution, composition, detection frequency and concentration levels among the aquaculture farms may be attributed to the usage practices in aquaculture (e.g., disease, type of feed and feed additive containing antibiotic used by farmer), physicochemical reaction of antibiotic toward environmental parameters, and microbiological degradation of antibiotic by the aquatic or sediment bacteria (Hektoen et al. 1995; Le, Munekage, and Kato 2005). In this study, low or negligible concentrations of macrolides were observed as most bacterial pathogens of fish are Gram-negative bacteria (Haenen 2017). Macrolides are broad spectrum antibiotics that are effective against most Gram-positive bacteria, and ETM is the only macrolide used in fish farming and shrimp hatcheries in Southeast Asia (Lavilla-Pitogo 2017). Also, LIN was found in all the farms as it is commonly used in livestock and aquaculture infections (FAO 2005).

Most of the farms (n = 25, 86.2%) sampled in this study, used more than two antibiotics (average of five antibiotics). The highest total number of antibiotic compounds used in fish farms was nine, detected in four farms located at Kelantan (K2), Johor (J1) and Penang Island (PI1 and PI2), respectively. For shrimp farms, only one farm located in Perak (P6) used 11 antibiotic compounds. It is well known that there exists an extensive use of antibiotics in aquaculture around the globe (Tuševljak et al. 2013) with Asian countries more notable for their wider range of approved antibiotics (Serrano 2005). The sampled farms in Malaysia used a relatively lower and less diverse number of antibiotic compounds than the top producing countries in Asia [Thailand, (13 antibiotics used), China (33) and Vietnam (39)] (Rico et al. 2012; Phu et al. 2016; Lulijwa, Rupia, and Alfaro 2019). In contrast, Japan has significantly reduced antibiotic usage, and no antibiotic is reported in their aquaculture industry (Lulijwa, Rupia, and Alfaro 2019).

Tetracyclines

Tetracyclines were the most prevalent antibiotic detected in this study. Tetracyclines are widely used in the aquaculture industry, animal husbandry and human therapy due to their low cost and high efficacy against a broad spectrum of bacteria, parasites, and fungi (Mo et al. 2017). In Malaysia, tetracyclines are the second highest antibiotic used and is recently reported to reach 73,910 kg per year (Zakaria 2017). These antibiotics are commonly administered in feeds or dissolved in water to be absorbed by the gills. Among tetracyclines, our results showed that OTC, TC, MNC were the most commonly used tetracycline compounds among the farms but no CTC was detected. Our results contrasted with previous findings that reported OTC, TC, CTC and DOX as the most used tetracycline for treatment and prevention of diseases in aquaculture (Shamsuzzaman and Biswas 2012; Hazrat et al. 2016). However, this difference could be attributed to the different farms and the year of sampling (Table S4). In Malaysia, OTC, TC and CTC are permitted antibiotics used in aquaculture, and are registered under the National Pharmaceutical Regulatory Agency (NPRA), Ministry of Health. OTC, CTC and DOX also fall under WHO’s criteria of critically important antibiotics for human health and their usage are being restricted in veterinary and aquaculture sectors (Hassali et al. 2018).

Although the use of OTC as a growth promoter was banned by the EU in 2006 (Castanon 2007), OTC remains the most commonly used antibiotic in animal production and aquaculture and are often detected in aquaculture water in different countries (Nonaka, Ikeno, and Suzuki 2007; Suzuki and Hoa 2012). In Malaysia, pricing is one of the main reasons why OTC is more popular. In contrast, the usage of CTC and DOX is limited as they are more expensive (Treves-Brown 2000). As OTC is poorly absorbed in fish, and a high dosage of OTC is often required, excess OTC from aquaculture is eventually discharged to the environment (Lunestad and Goksøyr 1990).

When OTC concentrations were not detected in this study (<LOQ), this might be attributed to its physiochemical properties, biodegradation and photodegradation. OTC half-life ranges from 21–25 mins in aquaculture water, 2 days in freshwater, 12 days in seawater to as high as 150 days in marine sediment depending on the environmental conditions (e.g., pH, temperature, salinity, light) (Choo 1994; Brooks, Maul, and Belden 2008; Leal, Esteves, and Santos 2016). The degradation rates of OTC in river sediment and wastewater sludge have been reported to be higher than TC (Chang and Ren 2015; Yang et al. 2020).

Although tetracyclines remain one of the top three antibiotics used in the top producing countries in Asia, Lulijwa, Rupia, and Alfaro (2019) reported a reduction from 92% to 73% in the usage of tetracyclines (Sapkota et al. 2008). In order to tackle the indiscriminate use of antibiotics, some Asian countries have banned the use of selected tetracyclines in aquaculture. TC was recently banned in Malaysia (The Sun Daily 2020) and is also not currently used in Vietnam and Singapore whereas CTC is not used in Indonesia, Singapore and Vietnam (ASEAN, 2013; Whitehead 2016). Singapore is the only country that do not use OTC (ASEAN, 2013) whereas CTC and OTC are prohibited in China (Liu, Steele, and Meng 2017). In Thailand, OTC and TC are still authorized for use in food animal (Lulijwa, Rupia, and Alfaro 2019).

MNC was also found in this study. MNC is a semisynthetic, second‐generation tetracycline derivative which is typically used in humans to treat acne (Garrido‐Mesa, Zarzuelo, and Gálvez 2013). To the best of our knowledge, no study has reported the presence of MNC residues in aquaculture water environments in worldwide. Moreover, this antibiotic is not authorized for use in aquaculture farms in Malaysia. Thus, the presence of MNC residues in seven farms with the highest concentration detected in P6, Perak (245 ng/L), suggested a misuse of antibiotic.

Sulfonamides

After tetracyclines, sulfonamides were the second most prevalent antibiotic found in aquaculture farms in this study. Sulfonamides were found in all the farms with the exception of two farms located in Selangor. The presence of sulfonamides in farm waters concurred with other studies (Jayachandran, Lleras-Muney, and Smith 2010). Sulfonamides are ubiquitous in the developing Asia aquatic ecosystem due to their low cost and more importantly, sulfonamides can be absorbed through gills. In addition, sulfonamides are highly soluble in water and highly mobile thus they can be easily transported and distributed in aquatic environments (Shi et al. 2012; Liu, Steele, and Meng 2017).

All the selected commonly used SAs antibiotics in animal treatment were detected in this study. SMR and STZ were the most commonly detected in Malaysia aquaculture farm. This was in contrast with other studies where SMT and SMX were commonly found in aquaculture or its adjacent environment (Giang et al. 2015; Hossain et al. 2017; Lai et al. 2018). Sulfonamides are commonly used alone or in combination with TMP or ormetoprim for better efficacy to combat bacterial infection. However only three farms (Pahang:PA4, Perak: P4 and Penang Island: PI2) were detected with low concentrations of TMP, suggesting that the usage of combination sulfonamides and TMP was less common in Malaysian aquaculture.

Similar to Vietnam, SMX, SMT, STZ and SMR were also detected in this study (Hoa et al. 2011; Giang et al. 2015; Harada 2018; Thai et al. 2018). The sulfonamide composition detected here is less diverse relative to China [SMX, SMT, SPD, sulfadiazine, sulfametoxydiazine, sulfomonothoxine, sulfameter, sulfaquinoxaline, sulfachloropyridazine] and Taiwan [SMX, STZ, sulfadiazine sulfaguanidine, sulfathazine, sulfamonomethoxine and sulfadimethoxine] (Lin, Yu, and Lin 2008; Chen et al. 2017; Lai et al. 2018; Wang et al. 2018a; Zhong et al. 2018; Yuan et al. 2019).

Quinolones

For quinolones, ENRO, NAL and OFX were the most commonly used. The selected quinolones were detected among the farms with the highest total concentration of 1.097 × 10⁶ ng/L. These antibiotics (ENRO, NAL and OFX) are widely administered in Asia aquaculture, and have become more popular than oxytetracycline over the last two decades (Hektoen et al. 1995). They are stable in water and sediment (Kümmerer 2004; Le and Munekage 2004). Lulijwa, Rupia, and Alfaro (2019) revealed that about 55% of the global major aquaculture producing countries applied ENRO whereas the usage of CIP and NOR were at a lower frequency. This is distinctly different from Thailand and Vietnam (Suzuki and Hoa 2012) where 74% of Thailand shrimp farms primarily used NOR (Holmström et al. 2003). In shrimp pond areas in the mangroves of Vietnam, Le and Munekage (2004) reported that NOR is detected in all shrimp ponds and surrounding canals whereas in Taiwan aquaculture, OFX, CIP and flumequine are present (Lin, Yu, and Lin 2008; Lai et al. 2018) (Table S4). In recent years, ENRO has been banned in Taiwan, Vietnam, Thailand but the ban in Malaysia only began from the year 2020 (MARD 2014; Tsai et al. 2019; The Sun Daily 2020). This could explain why we were still able to detect ENRO in the farms in Malaysia.

Regional comparison of antibiotic use

The composition of antibiotics varies between different countries showing the different practice of antibiotic administration in aquaculture. From the comparison of the published antibiotic concentrations available in aquaculture (Figure 4(a)), the concentration of tetracyclines detected in this study were still lower than Thailand (2 -500 ng/L) and China (0.32 – 15 × 10³ ng/L) but higher than Taiwan (11 – 75 ng/L) and Korea (7.1 – 95.4 ng/L). The concentrations of sulfonamides were comparable to aquaculture water in Taiwan but relatively lower compared to mariculture and aquaculture farms in China (0.4 –12429 ng/L), and Vietnam (0.08 – 2,390,000 ng/L).

For quinolones, the concentrations detected in Malaysia were higher than aquaculture farms in Thailand (13.2 – 490 ng/L), Taiwan (0.2 – 331 ng/L), Korea (0.88 - 54.5 ng/L) and Vietnam (0.2 - 60000 ng/L) (Figure 4(a)). NOR and CIP are currently not used in Indonesia, Singapore, and Thailand (ASEAN, 2013). Although CIP has been banned in China and Vietnam, illegal use of this banned antibiotic is still being reported (Mo et al. 2017; MARD 2016; Chi et al. 2017; Tran et al. 2018). On the other hand, the LIN detected in this study was comparable to Vietnam (8–10 ng/L, Shimizu et al. 2013) and Korea (<LOQ – 14.8 ng/L, Kim, Lee, and Oh 2017) but lower than China and Taiwan (<LOQ – 1643 ng/L, Zhong et al. 2018).

A study by Segura et al. (2015), suggested that the level of a country’s income has an effect on the occurrence of antibiotic in environment. However, our results revealed that the type of antibiotic measured and used will also influence the occurrence of antibiotic in environment. Our results (after excluding P6 result) showed that quinolone was two-fold higher compared to the low-income group (Ghana, India, Indonesia, Kenya, Philippines, Vietnam, Mozambique, Pakistan) whereas tetracyclines and sulfonamides measured were in the category comparable with low income group (sulfonamides: 15.5–112 ng/L tetracyclines: 29.3–289 ng/L). Our observations are in contrast to the status of Malaysia as an upper middle-income country (World Bank 2020).

Quantitative detection of antibiotic resistant genes

The prevalence of sul genes in Malaysian aquaculture farm was in the following frequency: sul2> sul1> sul3. Our results suggested that these genes were ubiquitous in aquaculture farms in Malaysia and was similar with marine mariculture in Japan (Suzuki et al. 2019) and marine mariculture in China (Chen et al. 2017). However, for pond aquaculture in China (Xiong et al. 2015; Su et al. 2017; Yuan et al. 2019) and effluent in Korea aquaculture farm (Jang et al. 2018), a different order of sul1> sul2> sul3 has been reported. The variation in gene distribution observed could be attributed to the difference in farming practices, bacterial population composition, types and antibiotic dosages used (Shimizu et al. 2013; Muñoz-Atienza et al. 2013; Rico et al. 2013). For instance, integrated fish farming practised throughout Asia is often with a closed system aquaculture where pond water does not frequently exchange. This can result in antibiotic resistance genes (ARGs) accumulating in pond water and sediment through horizontal gene transfer (Watts et al. 2017).

The sul1 and sul2 values detected in this study were comparable with net-pen aquaculture in Japan (Suzuki et al. 2019), Taiwan (Suzuki et al. 2019) and higher than aquaculture farm in Tianji, China (Gao et al. 2012), coastal aquaculture farm in South Korea (Jang et al. 2018) and aquaculture farm sediment (Gao et al. 2018). In comparison with the floating open cage aquaculture farm in Singapore, our sul2 abundance was lower (Ng et al. 2018). For sul3, the abundance was generally lower than China (Xiong et al. 2015) and Japan (Suzuki et al. 2019) (Figure 4(b)).

Other than sul genes, we also measured tet(M) in this study. Our results are consistent with other findings that tet(M) was prevalent in aquaculture farms. The concentrations detected in this study were comparable with aquaculture farms in Taiwan (Suzuki et al. 2019), China (Gao et al. 2012; Xiong et al. 2015; Niu, Zhang, and Zhang 2016; Yan et al. 2018), South Korea (Jang et al. 2018) and Japan (Suzuki et al. 2019) (Figure 4(c)).

In this study, no statistically significant (p > 0.05) correlation was found between concentrations of antibiotic and resistance genes. The targeted genes are historically “older” ones, which are already widely dispersed even as antibiotic contamination is reduced or non-prevalent. Similar uncoupling of ARGs and antibiotics have been reported in other areas (Takasu et al. 2011; Suzuki et al. 2015). Studies have shown that exposure to low concentrations of antibiotics for a long period can exert selective pressure and their transformation products also contribute in the development and dissemination of resistant bacteria and ARGs (Gullberg et al. 2011).

The prevalence of ARGs in a farm may be due to long term application of antibiotics in feed and water which may result in the accumulation of antibiotic residues at aquaculture farm. The farm then becomes a resistance hotspot to promote the growth of antibiotic resistance bacteria by exchanging resistance genes and thus altering the microbial community in water and sediment (Mohamed et al. 2000). The leaching of free-antibiotic, unconsumed antibiotic-feed or undegraded antibiotic-feces from aquaculture may also reach the groundwater and ocean through water circulation. This eventually end up in humans that consume antibiotic-contaminated drinks and food which pose a risk to public health.

Environmental ecological risk

For the ecological risk analysis, farms located in Kelantan, Perak and Pahang had the highest RQs. Three quinolones (ENRO, NOR, CIP) and LIN were found to have the potential to pose high risk to M. aeruginosa, S. leopoliensis, and P. subcapitata in aquaculture farms in Malaysia. Our results concurred with the findings from South Yellow Sea and aquaculture pond water around Lake Honghu in China where these antibiotics could pose high risk to cyanobacteria and algae (Du et al. 2017; Wang et al. 2017). In this study, SMX, CTM and TC were found to pose medium risk to cyanobacteria and algae. Several studies have also reported that SMX and CTM posed medium risk to various primary producers in rivers and pond waters where there are aquaculture activities (Zheng et al. 2012; Wang et al. 2017). In contrast, recent reports revealed that SMX and TC posed high risk to algae in Pearl River and Yellow Sea, China (Xu et al. 2013; Du et al. 2017; Wang et al. 2017).

Studies have shown that the coexistence of mixed antibiotics would pose a direct or indirect threat to the environment (González-Pleiter et al. 2013; Liu et al. 2014; Wang et al. 2018b). However, the risk caused by mixed antibiotics to the environment was not evaluated in this study as we employed a single-compound approach. In the future, environmental toxicity risk employing multi antibiotic approach is needed as the coexistence of mixed antibiotics can cause more significant adverse impact.

Conclusions

The present study reported on the antibiotic residues, ARGs and its associated potential ecology in the seven-main aquaculture production state in Peninsular Malaysia. Our study detected 23 antibiotics with the total concentration 1.099 × 10⁶ ng/L in which tetracyclines (83%), sulfonamides (72%) and quinolones (69%) had the highest detection frequency, indicating a wide distribution of antibiotics in aquaculture farms in Malaysia. Oxytetracycline, tetracycline, minocycline, sulfamerazine, sulfathiazole, enrofloxacin, nalidixic acid, and ofloxacin were the most abundant antibiotics. The minocycline was detected for the first time in aquaculture farms. The antibiotic residues detected were at a low or moderate level compared with Asian aquaculture farms except for quinolones. The dendrogram and heatmap showed that three antibiotic concentration patterns and no differences in the types of antibiotics usage among aquaculture farms were observed except for P6 farm in Perak. Overall, the relative abundance of resistance gene decreased according to the following frequencies: sul2> tet(M)>sul1> sul3 and no significant correlation was observed between antibiotic residue and resistance genes. Ciprofloxacin, enrofloxacin, norfloxacin and lincomycin were found to have high ecological risk to cyanobacteria and algae in Kelantan, Perak and Pahang. Overall, this study intensifies our understanding on antibiotic profile and bacterial resistome in aquaculture wastewater, as well as their potential impacts to organisms in environment. Prevention with proper environmental management should be conducted for aquaculture wastewater to mitigate the risks of antibiotic resistance to environment and public health through food chain.

Figure 1. Map of sampling sites. Red label indicates fish farm, blue label indicates prawn/shrimp farm and green label indicates prawn and fish farm

Figure 2. Relative abundance of selected antibiotic resistance genes, tet(M), sul1, sul2 and sul3 in seven main aquaculture production states. Vertical axis is copy number with normalized by 16S rRNA gene

Figure 3. Heatmap of 23 detected antibiotics clustered by concentration profiles in 29 aquaculture farms. Each cell represents the detection concentrations of antibiotic in each farm after logarithmic transformation (log ng/L). The dendrogram presented the sample site cluster analysis on the vertical axis and three antibiotic clusters on the horizontal axis (I, II and III)

Figure 4. Comparison of antibiotic concentration (a), abundance of sul genes (b) and tet(M) (c) in aquaculture farm in East and Southeast Asia. Antibiotic concentration: China: .He et al. (2012), Chen et al. (2015), Xiong et al. (2015), Song et al. (2016), Chen et al. (2017), Wang et al. (2018a), Yuan et al. (2019), Han et al. (2020); Vietnam: Le and Munekage (2004); Takasu et al. (2011), Hoa et al. (2011), Shimizu et al. (2013), Andrieu et al. (2015), Giang et al. (2015), Harada (2018), Thai et al. (2018); Thailand: Takasu et al. (2011); Shimizu et al. (2013), Oliveira et al. (2014); Taiwan: Lin, Yu, and Lin (2008); Lai et al. (2018); Philippines: Suzuki et al. (2013); Korea: Kim, Lee, and Oh (2017). Antibiotic resistance genes: China: Gao et al. (2012), Xiong et al. (2015), Gao et al. (2018); Korea: Jang et al. (2018); Japan and Taiwan: Suzuki et al. (2019); Singapore: Ng et al. (2018)

Disclosure of potential conflicts of interest

The author(s) declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

This research was supported by Malaysian Ministry of Higher Education (HiCoE Phase II (IOES-2014D), FP048-2013A, SF022-2013), University Malaya (PG309-2016A, IF030A-2017) and KAKENHI, JSPS (25257402, 16H01782, 17H04476). A part of this work was performed as collaboration project of Leading Academia in Marine and Environment Pollution Research (LaMer), Ehime University.

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

This work was supported by a research grant from the Malaysian Ministry of Higher Education [HiCoE Phase II (IOES-2014D), FP048-2013A, SF022-2013], University Malaya [PG309-2016A, IF030A-2017] and KAKENHI, JSPS [25257402, 16H01782, 17H04476]