Zoonotic public health risks in farmed versus wild-caught wildlife for traditional markets for food: A systematic review and meta-analysis

Abstract

Intensive farming, wildlife trade, and close human–animal contact in traditional markets for food (TMFs) are widely considered drivers of zoonotic disease emergence. However, comparative evidence assessing differences between farmed and wild-caught wildlife at the market interface remains limited. This study evaluated whether animals sourced from farmed versus wild-caught systems differ in pathogen detection as a proxy for zoonotic hazard and does not measure transmission or human infection outcomes.

PubMed, Scopus, Web of Science, and MEDLINE were systematically searched up to August 11, 2025, without language restrictions. Comparative studies reporting zoonotic pathogens in farmed versus wild-caught animals were included. Odds ratios (ORs) were pooled using random-effects models, with heterogeneity assessed using χ² and I² statistics.

Twenty studies were included from 23,840 identified titles. Farmed aquatic species had lower odds of parasitic infection than wild-caught species (OR 0.11, 95% CI 0.03–0.44). For bacterial zoonoses, the pooled OR was 1.21 (95% CI 0.44–3.34). Subgroup analyses showed higher odds in farmed frogs (OR 13.54) and freshwater turtles (OR 4.24). Data on terrestrial animals were limited to snakes, water monitor lizards, and rats. Farmed reptiles had lower odds of parasitic infection than wild-caught reptiles (OR 0.26, 95% CI 0.12–0.60). Farmed rats had lower odds of SARS-CoV-2 detection (OR 0.13, 95% CI 0.06–0·30). Overall, most studies had a high risk of bias, and heterogeneity was substantial (I² up to 98%).

Parasitic zoonoses with indirect life cycles were consistently less prevalent in farmed aquatic and reptile animals. In contrast, bacterial findings were inconsistent, and viral data was scarce. Despite widespread concern regarding zoonotic risk in wildlife trade, direct comparative evidence between production systems remains limited and highly heterogeneous. Overall, zoonotic hazard is likely influenced by environmental and management conditions, although the current evidence base is insufficient to support generalised conclusions. High heterogeneity and risk of bias warrant cautious interpretation and further standardised research.

Highlights

• •Research is needed on zoonotic spillover risks in traditional food markets selling wildlife and wildlife products.
• •Zoonotic hazards and control strategies vary by animal species, source, purpose, and market environment.
• •Farmed wildlife have lower odds of selected parasitic zoonoses compared with wild-caught animals.
• •Evidence for bacterial zoonoses was inconsistent, and major gaps exist for viruses, hosts, and defined market conditions.
• •Zoonotic hazard varies by pathogen and host species and farming wildlife alone does not uniformly reduce zoonotic hazards.

1. Introduction

Wildlife is sold worldwide in traditional markets for food (TMFs) and is often of financial, cultural, and nutritional importance [1], [2], [3]. Since the COVID-19 pandemic, these markets have received increased attention due to their potential role in zoonotic spillover. Prior to COVID-19, contact with wild animals and wildlife trade had already been associated with zoonotic pathogens, including Ebola virus, Marburg virus, Nipah virus, and Hendra virus, among others [3], [4]. A landmark 2005 study estimated that 73% of emerging and re-emerging infectious diseases were zoonotic [5]. TMFs may also include animals transferred across international borders, which can facilitate broader geographic dissemination of pathogens. For example, a study in Benin found that only 25% of sampled carcasses originated locally, with the remainder sourced from neighbouring countries and traded across regions including India and China [6]. TMFs sell diverse wildlife species and products originating from backyard farming operations, illegal trade, wild capture, and formal livestock systems. A scoping review identified passerine birds, turtles, lizards, and snakes as the most commonly reported wildlife species sold at TMFs [3]. Markets with a greater diversity of species have been hypothesised to increase opportunities for cross-species pathogen exchange. According to Scopes et al., China and Southeast Asia have the greatest diversity of wildlife species in these markets, particularly reptiles and birds [3]. Although on-site slaughter has been reported in many markets, data on water, sanitation, and hygiene (WASH) conditions remain limited [3].

Growing demand for wildlife products has led to the expansion of wildlife farming to increase supply. Much of the literature on wildlife farming has focused on China, where production has expanded to meet demand for fur, traditional medicine, and other products [7]. Concerns have been raised that wildlife farming may increase consumer demand, fail to reduce pressure on wild populations, and allow illegally sourced animals to enter legal supply chains [7], [8]. Additional concerns relate to the welfare and treatment of both farmed and wild-caught animals [8]. However, comparatively little attention has been given to the public health implications of wildlife farming, particularly in relation to zoonotic pathogens. Following the COVID-19 pandemic, emphasis has largely been placed on market-level interventions. In 2021, WHO interim guidance recommended suspending trade in live wild mammals, alongside strengthening hygiene, sanitation, and surveillance systems [9]. In 2023, the World Health Assembly requested updates to this guidance, including clarification on species scope and the role of farmed versus wild-caught animals [10]. Whether farmed and wild-caught wildlife differ in zoonotic hazard has therefore emerged as a key unresolved question within broader efforts to reduce spillover risk in TMFs.

Intensive farming, wildlife capture and trade, and close human–animal contact in TMFs have been proposed as factors influencing zoonotic disease emergence. However, direct comparative evidence evaluating differences between farmed and wild-caught wildlife remains limited. Understanding whether zoonotic hazard differs by species, production system, and environmental conditions is therefore critical for informing public health interventions. This systematic review and meta-analysis aimed to compare zoonotic pathogen presence in farmed versus wild-caught wildlife and wildlife products sold in TMFs.

2. Methods

2.1. Study design

A systematic review with meta-analysis was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines (see S1) [11]. The review followed the PICOS (Population, Intervention, Comparison, Outcome, Setting)-question format: Would the sale of farmed wild animals (I) or captured wild animals (C) reduce the risk of pathogen detection (O) at the animal-human-environment interface (P) in traditional markets for food (S)? The analysis does not measure human infection outcomes or transmission events but the relative pathogen presence as a proxy for zoonotic hazard at the animal-human-environment interface. Zoonotic risk is operationalized as comparative pathogen detection in animals sourced from farmed versus wild captured systems, and we aimed to determine the difference in associated risk between farmed and captured wild animals for sale or trade in traditional markets for food. This review was not prospectively registered due to the exploratory nature of the research question.

2.2. Search strategy

PubMed, Scopus, Web of Science, and MEDLINE were searched from inception to August 11th, 2025 without language restrictions (see S1 for detailed search strategy). Grey literature sources were consulted, and reference lists of included studies were screened to identify further eligible studies.

Search results were imported into the Rayyan software for screening and selection [12]. Titles and abstracts were independently assessed by reviewer pairs (AB, AM, AS, CZ) using the predefined eligibility criteria described below. Members were conservative with their initial screening to ensure inclusion of all possible titles for further assessment. Full texts of potentially eligible studies were retrieved and assessed in duplicate. Discrepancies were resolved through discussion or, if needed, by a third reviewer serving as the tie-breaker (FS). Non-English articles were translated prior to assessment (CZ, FS, WS). Multiple published reports of the same study were consolidated, with the study considered the unit of analysis. Identified reviews were included for the purpose of screening their reference lists to identify additional individual studies eligible for inclusion.

2.3. Eligibility criteria

Only primary studies with a comparative design were eligible, comparing zoonotic pathogen risks, defined as pathogen presence in animals or risk of transmission at the human–animal–environment interface, in farmed and captured wild animals (live, dead, or processed) in the context of TFMs traded for any purpose. Any wild animal species was eligible; domesticated animals were excluded as defined by the World Organization for Animal Health (WOAH) [13].

2.4. Risk of bias assessment

The risk of bias in prevalence studies was assessed using the JBI Critical Appraisal Checklist for Prevalence Studies [14]. All domains were assessed as having a ‘low’, ‘unclear’, ‘high’, or ‘not applicable’ risk of bias, and the reasons for each bias classification were documented. The ROBVIS tool was used to illustrate the risk of bias judgement in traffic light plots [15].

2.5. Data extraction and statistical analyses

The following individual study characteristics were extracted: study design, study site, sampling period, species specification, sample specifics, zoonotic pathogens, number of samples for the farmed and wild animal groups, and the absolute numbers. The odds ratios (ORs) with corresponding 95% confidence intervals (95%CI) were calculated for each comparison. The data extraction was performed by one person (AM, CZ) with verification by a second (AS, FS) for each reviewer pair. Any disagreements were resolved through consensus and the involvement of a co-author with a veterinary and zoonotic background (AS, FS).

For the meta-analysis, study animal populations were first grouped into terrestrial, arboreal, avian, and aquatic categories according to the WOAH criteria [16]. Within each category, further subgrouping was conducted based on the zoonotic pathogen type: bacterial, viral, or parasitic. Pooled estimates were calculated using random-effects models when two or more studies reported comparable outcomes. Where fewer than two studies were available, forest plots were generated to visually present individual study estimates. The meta-analyses were conducted with Review Manager 5 (RevMan 5) version (5.4.1), The Cochrane Collaboration, (2020), available at revman.cochrane.org. The calculated Chi-squared (χ²) test and the I² statistic were interpreted to quantify heterogeneity. Sensitivity analysis was considered in case of K > 5 studies and conducted by removing studies one by one to identify the most influential study. Publication bias was not assessed, as the small number of studies per outcome would yield unreliable and low-powered estimates [17].

3. Results

Overall, evidence was limited, highly heterogeneous, and predominantly derived from aquatic species, with minimal data on viral pathogens or mammals. The search yielded 23,840 titles wherefrom 20 studies were included for final analyses (Fig. 1). These were conducted in local markets, fish markets, reptile markets and TMFs in Viet Nam (n = 4), Thailand (n = 3), Bangladesh (n = 3), Indonesia (n = 3), China (n = 1), Egypt (n = 1), Hong Kong (n = 1), India (n = 1), Spain (n = 1), Sri Lanka (n = 1), and Taiwan (n = 1). Seventeen studies included aquatic species, including fish (n = 12), shellfish (n = 2), frogs (n = 2), and freshwater turtles (n = 1) [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33]. Three included terrestrial species, rats (n = 1), snakes (n = 1), and monitor lizards (n = 1) [34], [35], [36]. No avian or arboreal species, and besides the rats, no other mammals were identified. Parasitic zoonoses were reported most often followed by bacterial zoonoses. Only one paper reported viral zoonoses. All included studies traded animals for food consumption purposes. A complete overview of included studies and their characteristics is reported in Table 1.

Fig. 1.

PRISMA flowchart.

Table 1.

Study characteristics of included studies.

Study information (author, year, country, region)	Study design	Title	Study site (market, study period)	Species specifics (ecosystem and host class)	Sample specifics (type of sample, specifics, pathogen, pathogen detail)	Results (comparison groups, effect, SE)
						Farmed	Wild	Effect
Ahmed et al 2023. Egypt	Cross-sectional	Parasites of public health importance in Nile and cultured fish in El-Minya Governorate	Market (wild) and fish farms (farmed). Mar 2021 - Feb 2022	Tilapia nilotica (Oreochromis niloticus), catfish (Clarias gariepinus), bajad (Bagrus bajad), carp fish (Cyprinus carpio). Aquatic, fish	Encysted metacercaria (EMC): Cynodiplostomumsp., Prohemistomum sp., Clinostomum phalacrocoracis in muscle and gills, Ichanthochephala sp., Capillaria sp. In stomach and intestines, and Cryptospordiu sp. in mucosal scrapings of the intestine	87/100 (87·00%) positive samples	100/100 (100·00%) positive samples	OR: 0·03 (95%CI 0·00 to 0·55)
Baten et al. 2021. Sylhet, North-Eastern Bangladesh	Cross-sectional	Status of Pathogenic Bacteria with their Antibiotic Susceptibility in Indigenous and Exotic Climbing Perch, Anabas testudineus (Bloch, 1972) in North-Eastern Bangladesh	Fish markets selling indigenous (wild) and exotic (farmed) climbing perch. Dec 2014 - Nov 2015	Aquatic, fish	Eshcerichia coli (E. coli) taken from muscle, gill and intestines mixed	20/30 (66·67%) positive samples	17/30 (56·57%) positive samples	OR: 1.48 (95%CI 0·49 to 4·44)
Colon et al., 2022. Hong Kong	Cross-sectional	Serotype Diversity and Antimicrobial Resistance Profile of Salmonella enterica Isolates From Freshwater Turtles Sold for Human Consumption in Wet Markets in Hong Kong	Wet markets where live turtles are available, wild-caught or farmed recorded. Jan - March 2021	Chinese softshell turtles (Pelodiscus sinensis), Red-eared sliders (Trachemys scripta elegans), Chinese striped neck turtles (Mauremys sinensis). Aquatic, reptile	Salmonella enterica in fecal or colon sample of freshwater turtles	18/35 (51·43%) positive samples	3/15 (20·00%) positive samples	OR: 4.24 (95%CI 1·02 to 17·67)
Chen et al. 2021. Taiwan	Cross-sectional	Prevalence, virulence-gene profiles, antimicrobial resistance, and genetic diversity of human pathogenic Aeromonas spp. from shellfish and aquatic environments	Shellfish farming areas (farmed) and fishing markets (wild). Feb - Aug 2017.	No species info, shellfish. Aquatic, crustaceans/molluscs unspecified	Aeromonas spp. From shellfish meat	8/60 (13·33%) positive samples	22/63 (34·92%) positive samples	OR: 0·29 (95%CI 0·12 to 0·71)
Hong et al. 2016. Guangzhou, China	Cross-sectional	Prevalence of Spirometra mansoni in dogs, cats, and frogs and its medical relevance in Guangzhou, China	Agricultural product markets and aquatic product wholesale market. May - Dec 2013	Frogs (Rana tigrina rugulosa (farmed), Rana tigrina rugulosa (wild), Rana guentheri (wild), Rana catesbeiana(farmed), Rana limnocharis (wild). Aquatic, amphibians	Spirometra mansoni in frog muscles	0/1382 (0·00%) positive samples	229/567 (40·39%) positive samples	OR: 0 (95%CI 0·00 to 0·01)
Huong et al, 2020. Southern provinces in Vietnam	Cross-sectional	Coronavirus testing indicates transmission risk increases along wildlife supply chains for human consumption in Viet Nam, 2013–2014	Live field rat trade supply chain (traders, markets, restaurant site) (wild) and wildlife farms specialising in raising rodents (farmed). Jan 2013 - March 2014	Field Rats (Rattus argentiventer, R. tanezumi, R. norvegicus, R. exulans, R.losea, and Bandicota indica) v. Farmed rats (Rhizomys sp.,). Terrestrial, mammal	Coronaviruses in rats (oral and tissue for wild and feces, urine, and swabs of the pen floors for farms)	6/96 (6·25%) positive samples	239/702 (34·0%) positive samples	OR: 0·13 (95% CI 0·06 to 0·30)
Januarista et al. 2021. Banyuwangi District, Indonesia	Cross-sectional	Prevalence and intensity of endoparasitic helminths on swamp eel (Monopterus albus) from natural caught and cultivation	Cultivators and catchers of swamp eels. Jan - March 2020	Swamp eel (Monopterus albus). Aquatic, fish	Endoparasitic helminths (Eustrongylydes ignotus andPingus sinesis) in skin, fins, body cavity, kidneys, gonads, and digestive tract.	3/60 (5%) positive samples	14/60 (23·3%) positive samples	OR: 0·17 (95%CI 0·05 to 0·64)
Kamalika et al. 2008. Puttalam and Kurunegala District, Sri Lanka	Cross-sectional	Prevalence of Salmonella in marketed Penaeus monodon shrimps in North Western Province, Sri Lanka	Wild captured in lagoons and cultured in farms. Nov 2006 - Mar 2007	Shrimp (Penaeus monodon) Aquatic, crustaceans	Salmonella sp. in shrimps	10/90 (11·11%) positive samples	13/90 (14·44%) positive samples	OR: 0·74 (95%CI 0·31 to 1·79)
Labony et al., 2024. Mymensingh, Kishoreganj, Bangladesh	Cross-sectional	Zoonotic human liver flukes, a type 1 biocarcinogen, in freshwater fishes: genetic analysis and confirmation of molluscan vectors and reservoir hosts in Bangladesh	Wild and farmed fish collected from local markets, sources of fish verified by fishermen. Jul 2018 - Jun 2022	Rohu (Labeo rohita), mrigal (Cirrhinus cirrhosus), grass carp (Ctenopharyngodon idella), olive barb (Puntis sarana), silver carp (Hypophthalmichthys molitrix), orange fin labeo (Labeo calbasu), bata (Labeo bata), tilapia (Oreochromis mossambicus), spotted snakehead (Channa punctata), striped snakehead (Channa striata), yellowtail catfish (Pangasianodon hypophthalmus), walking catfish (Clarias batrachus), and stinging catfish (Heteropneustes fossilis) Aquatic, fish	Fish-borne human liver flukes metacercariae (Clonorchis sinensis, Opisthorchis spp., and Metorchis spp.) in fish flesh	39/199 (19·59%) positive samples	22/126 (17·46%) positive samples	OR: 1·15 (95%CI 0·65 to 2·05)
Moratal et al. 2022. Comunidad Valenciania, Spain	Cross-sectional	Molecular Characterization of Cryptosporidium spp. in Cultivated and Wild Marine Fishes from Western Mediterranean with the First Detection of Zoonotic Cryptosporidium ubiquitum	Off-shore aquaculture farms and synanthropic fish that aggregate around floating pens (farmed) and wild fish from extractive fisheries obtained at fish market (wild). Jul 2020 – Oct 2021, Mar – June 2021.	Meagre (Argyrosomus regius), European seabass (Dicentrarchus labrax), Gilthead seabream (Sparus aurata), European conger (Conger conger), Annular seabream (Diplodus annularis), Common two-banded seabream (Diplodus vulgaris), Blackbelly rosefish (Helicolenus dactylopterus), Brown wrasse (Labrus merula), Large-scaled gurnard (Lepidotrigla cavillone), Blackbellied angler (Lophius budegassa), European hake (Merluccius merluccius), Blue whiting (Micromesistius poutassou), Red mullet (Mullus barbatus), Surmullet (Mullus surmuletus), Axillary seabream (Pagellus acarne), Common pandora (Pagellus erythrinus), African armoured searobin (Peristedion cataphractum), Greater forkbeard (Phycis blennoides), Forkbeard (Phycis phycis), Salema (Sarpa salpa), Atlantic mackerel (Scomber scombrus), Lesser spotted dogfish (Scyliorhinus canicula), Comber (Serranus cabrilla), Brown comber (Serranus hepatus), Greater weever (Trachinus draco), Mediterranean horse mackerel (Trachurus mediterraneus), Pouting (Trisopterus luscus), Stargazer (Uranoscopus scaber), Garfish (Belone belone), Bogue (Boops boops), Thicklip grey mullet (Chelon labrosus), Sharpsnout seabream (Diplodus puntazzo), White seabream (Diplodus sargus), Round sardinella (Sardinella aurita), Chub mackerel (Scomber japonicus), Blotched picarel (Spicara maena), Black seabream (Spondyliosoma cantharus), Pompano (Trachinotus ovatus), Argentine (Argentina sphyraena), Imperial scaldfish (Arnoglossus imperialis), Red gurnard (Chelidonichthys cuculus), and Spotted flounder (Citharus linguatula). Aquatic, fish	Cryptosporidium spp. in gastrointestinal tissue scrapings mixed with intestinal contents	7/147 (4·80%) positive samples	10/257 (3·89%) positive samples	OR: 1·24 (95%CI 0·46 to 3·32)
Nilavan et al 2021. Kerala, India	Cross-sectional	Prevalence of Vibrio mimicus in Fish, Fishery Products, and Environment of South West Coast of Kerala, India.	Seafood obtained from fish markets, fish farms, and fish landing centres. May 2017 - Mar 2020	Brackish water fish: Piaractus brachypomusand Oreochromis mossambicus among freshwater fish; Villorita cyprinoides, Liza tade, Etroplus suratensis, Arius maculates, Penaeus monodon, Mugil cephalus, Scylla serrata, Perna viridis, and Paphia malabarica Freshwater fish: Piaractus brachypomus, Oreochromis mossambicus Marine fish: Sardinella longiceps, Stolephorus indicus, Metapenaeus dobsonii. Aquatic, fish	Vibrio mimicus from muscles or the whole prawn.	Fish market: 9/169	Fish landing centre (wild): 4/67	OR: 0·89 (95%CI 0·26 to 2·98)
Phan et al., 2016. Vietnam	Cross-sectional	Comparative Risk of Liver and Intestinal Fluke Infection from Either Wild-Caught or Cultured Fish in Vietnam	Wild-caught fish collected from local markets, farmed fish directly from farmers. May - Aug 2014	Acheilognathus macropterus, Anabas testudineus, Beaufortia leveretti, Carassius auratus, Channa maculata, Cirrhinus cirrhosus, Clarias fuscus, Ctenopharyngodon idellus, Cultrichthys erythropterus, Cyprinus carpio, Elopichthys bambusa, Garra sp., Glyptothorax honghensis, Hemibarbus macracanthus, Hemibarbus medius, Hemiculter leucisculus, Hypophthalmichthys molitrix, Mastacembelus armatus, Microphysogobio sp., Onychostoma laticeps, Opsariichthys bidens, Oreochromis niloticus, Osteochilus salisbury, Prochilodus lineatus, Rhinogobius sp., Sarocheilichthys nigripinnis, Schistura spp., Squaliobalbus curriculus, Toxabramis houdermeri, Traccatichthys pulcher, Xenocypris argentea Aquatic, fish	Fish-borne zoonotic trematode metacercariae (Clonorchis sinensis, Haplorchis pumilio, Haplorchis taichui, Haplorchis yokogawai, Centrocestus formosanus, Procerovum varium) in fish (whole or pooled subsample including head, gills, muscles of the main body, and the caudal and anal fins)	87/186 (46·77%) positive samples	255/450 (56·67%) positive samples	OR: 0·67 (95%CI 0·48 to 0·95)
Rana et al., 2023. Bangladesh	Cross-sectional	Antimicrobial Resistance, Biofilm Formation, and Virulence Determinants in Enterococcus faecalisisolated from Cultured and Wild Fish	Wild and farmed fish collected from local markets, wild-caught or farmed recorded. Oct 2021 - Dec 2022	Koi fish (Anabas scandens), stinging catfish (Heteropneustes fossilis) Aquatic, fish	Enterococcus faecalis in fish intestines	40/72 (55·56%) positive samples	47/60 (78·33%) positive samples	OR: 0·35 (95%CI 0·16 to 0·75)
Ribas and Poonlaphdecha., 2017. Thailand	Cross-sectional	Wild-Caught and Farm-Reared Amphibians are Important Reservoirs of Salmonella, A Study in North-East Thailand	Studied in the context of market sale, not sampled in markets (farm, natural park, wildlife sanctuary), Mar - Jun 2014	Duttaphrynus melanostictus, Fejervarya limnocharis, Hoplobatrachus rugulosus, Kaloula mediolineata, Kaloula pulchra, Polypedates megacephalus, Hylarana erythraea, Hylarana lateralis Aquatic, amphibians	Salmonella spp. in frog intestines	55/60 (91·67%) positive samples	13/29 (44·83%) positive samples	OR: 13·54 (95%CI 4·19 to 43·71)
Saksirisampant and Thanomsub. 2012. Thailand	Cross-sectional	Positivity and intensity of Gnathostoma spingeruminfective larvae in farmed and wild-caught swamp eels in Thailand.	Farm (farmed) and local market (wild). Jul 2008 - Jun 2009	Swamp eel (Monopterus albus). Aquatic, fish	Gnathostoma spingerum in swamp eel liver.	106/1037 (10·22%) positive samples	78/383 (20·4%) positive samples	OR = 0·45 (95%CI 0·32 to 0·61)
Sieu et al., 2009. Mekong Delta provinces, Vietnam	Cross-sectional	Prevalence of Gnathostoma spinigerum Infection in Wild and Cultured Swamp Eels in Vietnam	Fish markets (farmed and wild), and local fishermen (wild) Nov 2005 - Aug 2007	Asian swamp eels (Monopterus albus) Aquatic, fish	Gnathostoma spinigerum in eel body (muscle, liver, visceral organs)	0/1020 (0·00%) positive samples	125/3851 (3·25%) positive samples	OR: 0·01 (95%CI 0·00 to 0·23)
Thu et al., 2007. Mekong Delta provinces, Vietnam	Cross-sectional	Survey for zoonotic liver and intestinal trematode metacercariae in cultured and wild fish in An Giang Province, Vietnam	Fish farms and wild-caught fish sampled from markets, Jun 2005 - Mar 2006	Tra catfish (Pangasius hypophthalmus and Pangasius bocourti), Anabas testudineus, Botia helodes, Channa striata, Cirrhinus jullieni, Macrognathus siamensis, Mastacembelus armatus, Mystus albolineatus, Oreochromis mossambicus, Osteochilus hasselti, Pristolepis fasciatus, Rasbora trilineata Aquatic, fish	Fish-borne zoonotic trematode metacercariae (Haplorchis pumilio, Opisthorchis viverrini, Procerovum sp.) in fish (whole or pooled subsample including head, gills, muscles of the main body, and the caudal and anal fins)	3/459 (0·65%) positive samples	14/108 (12·96%) positive samples	OR: 0·04 (95%CI 0·01 to 0·16)
Wiriya et al., 2013. Suphan Buri, Nakhon Pathom and Chachoengsao, Thailand	Cross-sectional	Fish-borne trematodes in cultured Nile tilapia (Oreochromis niloticus) and wild-caught fish from Thailand	Fish farms and wild-caught fish sampled from markets, Sep - Oct 2011, Apr - May 2012	Long-fatty finned mystus (Mystus singaringan), Siamese mud carp (Henicorhynchus siamensis), Greenback mullet (Liza subviridis), Snakeskin gourami (Trichogaster pectoralis), Silver barb (Barbodes gonionotus),Moonlight gourami (Trichogaster microlepis), Climbing perch (Anabas testudineus), Hard-lipped barb (Osteochilus hasselti), Catfish (Clariidae spp.), Nile tilapia (Oreochromis niloticus), Bagrid catfishes (Bagridae spp.) Aquatic, fish	Fish-borne zoonotic trematode metacercariae (Stellantchasmus falcatus, Haplorchis pumilio and Procerovum varium) in fish (gills, body muscle with dorsal fin, pelvic fin and pectoral fin, a piece of body muscle with anal fin, caudal fin)	6/230 (2·61%) positive samples	80/150 (53·33%) positive samples	OR: 0·02 (95%CI 0·01 to 0·06)
Yudhana et al., 2019. Indonesia	Cross-sectional	The medical relevance of Spirometra tapeworm infection in Indonesian Bronzeback snakes (Dendrelaphis pictus): A neglected zoonotic disease.	Wild and farmed from reptile market, status recorded. Sampling time unspecified.	Bronzeback snakes (Dendrelaphis pictus) Terrestrial, reptiles	Spirometra spp. in snakes (muscle tissue, subcutaneous tissue, and coelom)	96/197 (48·73%) positive samples	128/181 (70·72%) positive samples	OR: 0·39 (95%CI 0·26 to 0·60)
Yudhana et al., 2021. Indonesia	Cross-sectional	Sparganosis (Spirometra spp.) in Asian Water Monitor (Varanus salvator): A medical implications for veterinarians, breeders, and consumers.	Wild and farmed from reptile market and breeders, status recorded. Sampling time unspecified.	Asian water monitor lizard (Varanus salvator) Terrestrial, reptiles	Spirometra spp. in Asian water monitor lizards (muscles and subcutaneous tissues)	35/85 (41·18%) positive samples	183/228 (80·26%) positive samples	OR: 0·17 (95% CI 0·10 to 0·30)

A meta-analysis including ten studies assessed parasitic zoonoses in aquatic species, with subgroups for fish (K = 9) and frogs (K = 1) (Fig. 2). Farmed fish had lower odds of zoonotic parasites compared to wild-captured fish, OR = 0.19 (95%CI 0.06 to 0.57), including endoparasites, fish-borne trematodes and Gnathostoma spp. The single study on frogs reported an OR of 0.00 (95%CI 0.00 to 0.01) for Spirometra spp. Across all studies, the pooled OR for parasitic zoonoses for any farmed aquatic species was 0.11 (95%CI 0.03 to 0.44). Heterogeneity was high (I² up to 95%) due to the underlying differences between studies, limiting comparability and interpretation of pooled estimates. A sensitivity analysis for the robustness of the study was conducted for the studies involving fish species by removing the studies with the broadest 95%CI. Upon removal, the subgroup point estimate increased to OR = 0.64 (95%CI 0.39 to 1.07) and reduced the heterogeneity from 95% to 76% (Fig. 3). (See Fig. 4.)

Fig. 2.

Forest plot for aquatic species with parasitic zoonoses.

Fig. 3.

Sensitivity analysis for fish species with parasitic zoonoses.

Fig. 4.

Forest plot of zoonotic bacterial infections in aquatic species.

Six studies assessed the presence of zoonotic bacterial pathogens (Aeromonas spp., Enterococcus spp., Escherichia coli, Salmonella spp., Vibrio mimicus) in farmed and wild-captured aquatic species. We conducted 1 meta-analysis (K = 7) with subgroups for fish (K = 2), shellfish (K = 3), frogs (K = 1) and freshwater turtles (K = 1). The overall odds for bacterial zoonotic pathogens was OR = 1.21 (95%CI 0.44 to 3.34) for farmed aquatic species. There was no significant difference in bacterial zoonoses in farmed versus captured fish, OR = 0.70 (95%CI 0.16 to 2.98), or in shellfish, OR = 0.55 (95%CI 0.27 to 1.10). One study reported that farmed frogs had an OR of 13.54 (95%CI 4.19 to 43.71) for bacterial zoonoses compared to wild captured frogs. The study on freshwater turtles with a relatively small number of tested species reported that farmed freshwater turtles were associated with higher odds for bacterial zoonoses, OR = 4.24 (95%CI 1.02 to 17.67). Both the frogs and water turtles were tested for Salmonella spp. The limited number of studies precluded further sensitivity analyses.

Three studies assessed terrestrial species, reporting only on parasites and one virus. Of these, two studies investigated Spirometra spp. in bronzeback snakes and Asian water monitor lizards and found that farmed reptiles had a pooled OR of 0.26 (95% CI 0.12 to 0.60) (Fig. 5). One study in farmed and wild-captured rats reported that farmed rats had OR = 0.13 (95%CI 0.06 to 0.30) for SARS-CoV-2 (Fig. 6).

Fig. 5.

Forest plot of zoonotic parasitic infections in terrestrial species.

Fig. 6.

Forest plot of zoonotic viral infection in terrestrial species.

Risk of bias was assessed as high in 16 of 20 studies, primarily due to unspecified sampling strategies, use of convenience samples, and small sample sizes (Fig. 7). Overall, the evidence base was characterized by substantial risk of bias. Many studies were not primarily designed to assess pathogen prevalence. However, studies generally scored well on clarity of study subjects and setting (D4), data analysis (D5), and methods of condition measurement (D6–D8), indicating adequate internal validity. The main sources of bias were related to differences in sampling approaches rather than analytical methods.

Fig. 7.

Risk of bias in the included studies.

4. Discussion

This study examined comparative zoonotic hazard associated with the trade of farmed or captured wild live animals in TMFs. Parasitic infections in farmed aquatic and terrestrial species were lower than wild captured animals intended for TMFs. Data on bacterial infections were inconsistent, with higher odds in farmed aquatic turtles and frogs, but no statistically significant difference between farmed and wild-captured fish and shellfish. Finally, the only viral pathogen and only mammal host were in the same study, limiting comparison with other sites and settings. Despite global concern regarding viral spillover from wildlife markets, only one comparative study assessing viral pathogens met the inclusion criteria [37], highlighting a substantial evidence gap. These findings indicate that zoonotic hazard varies by pathogen and host, and that comparative evidence remains insufficient to support generalised assumptions across production systems.

The lower odds of parasitic infection observed in farmed aquatic and terrestrial species likely reflect reduced exposure to intermediate hosts in managed production systems, supporting the role of environmental control as a key determinant of zoonotic risk. The pooled point estimate remained suggestive of a protective association with farming after excluding studies with very wide confidence intervals. However, the association was no longer statistically significant (OR 0.64, 95% CI 0.39–1.07), indicating that although the direction of association was unchanged, the strength of the evidence was sensitive to study inclusion. Parasites such as Gnathostoma spingerum, Spirometra spp., and fish-borne trematodes all have complex, indirect lifecycles involving intermediate hosts such as copepods and snails [38], [39], [40], [41], [42]. Interrupting these transmission pathways through environmental management may reduce parasite presence, for example by limiting access to intermediate hosts or contamination in aquaculture systems [39], [40], [43]. It is possible that farmed wildlife may also reduce other parasites with indirect life cycles [44]. These findings are consistent with environmentally mediated parasite ecology and suggest that production conditions can influence pathogen presence for parasites with indirect life cycles.

The higher odds of bacterial pathogens observed in farmed frogs and freshwater turtles indicate that farming does not uniformly reduce zoonotic hazard and may introduce different exposure pathways. Farmed turtles have been reported to carry a high prevalence of Salmonella spp., with some studies reporting rates up to 90%, contributing to regulatory restrictions in certain settings [20], [45]. In addition, the use of antibiotics in aquaculture may contribute to antimicrobial resistance, which should be considered when evaluating potential benefits and unintended consequences of farming systems [46].

Only one study assessed a viral pathogen, focusing on SARS-CoV-2 in farmed and wild-caught rats. The limited scope and sample size preclude broader inference. Comparative evidence for viral zoonotic hazards in wildlife trade contexts is therefore extremely limited. Outside of a market context, surveillance for SARS-CoV-2 in wild rats has been mixed: rats in both Belgium and Canada were negative in one study each and positive by serology and PCR in the United States in another study [47], [48], [49]. While no paper compared wild caught and farmed Asian palm civets, in one study, SARS was found in civets in markets (unknown source) but not in civets at any farm tested. The authors believed the animals got infected in the market but were unable to determine the original source. This further shows the difficulty in tracing wildlife sources as well as the risks that occur from the multiple species and sources found in TMFs [50]. More research would be needed to make any broader claims on SARS-CoV-2, much less other viral diseases. This includes major viral zoonoses of public health importance spread by non-human primates, such as the haemorrhagic fever viruses (Filoviridae), and pox viruses (Poxviridae) [51]. While no study compared farmed and wild caught wild birds, multiple reports of avian influenza in wild birds from TMFs that also sell domestic birds have been reported, with potential for interspecies transmission at the market interface [52], [53]. The co-occurrence of farmed, wild-caught, and domestic animals within TMFs creates complex ecological interfaces that may facilitate cross-species pathogen exposure. Prevention and control measures therefore need to be host- and context-specific, rather than based solely on animal source.

For groups dependent on wildlife capture for livelihoods or as a primary protein source, the perceived benefits may outweigh concerns about zoonotic spillover [54]. Social values further shape the acceptability of consuming wildlife. In parts of China and Southeast Asia, wildlife consumption is associated with perceived health benefits, medicinal value, or social status, and some consumers may distrust farmed alternatives, viewing wild-caught animals as more natural or higher quality [54], [55], [56]. Vendors may also view wildlife farming primarily as an economic opportunity rather than a disease risk-reduction strategy [8]. Shifting from wild-caught to farmed wildlife may reduce zoonotic hazard for specific pathogen–host combinations, particularly environmentally mediated parasitic infections. However, outcomes are likely to depend on biosecurity, regulatory oversight, and pathogen-specific dynamics [57].

TMFs represent complex ecological interfaces characterized by high animal density, multi-species co-housing, on-site slaughtering, and variable hygiene conditions. These contextual factors may modify zoonotic hazard independently of animal source and warrant consideration in risk-reduction strategies [58]. The importance of sanitation and hygiene as well as awareness of risk, is a critical part of risk reduction from wild or domestic animals in TMFs. A study in a bushmeat market in Cameroon found most workers thought the risk was low despite working alongside inadequate sanitation and hygiene measures [59]. Context-specific policies that integrate regulatory oversight with culturally sensitive risk communication and stakeholder engagement are, therefore, essential to prevent unintended consequences, including increasing food insecurity [60].

Despite screening over 70 full text studies, only 20 were included in the final meta-analysis. Most studies were excluded because of lack of a comparator group; furthermore, 19 titles were unclear on the source for the wildlife, either farmed or wild-caught. This highlights both the scarcity of comparative data and challenges in accurately determining wildlife source within market systems. This difficulty is also one of the concerns with farming wildlife: illegal, wild-caught wildlife can be passed off as farmed [8]. The geographic distribution was also limited primarily to Asia with one study each from Spain and Egypt. Half of all studies came from lower-middle income countries with the rest coming from high or upper-middle income countries. Sex was also not considered as a variable by all but one paper on freshwater turtles (Colon et al., 2022). This type of host demographic information if included could provide more nuanced measurements of risk. All studies included aquatic or terrestrial species which led to the current division because of insufficient numbers to further divide by species. The absence of avians and mammals other than rats limit the scope of the suggestions that can be drawn from this review. The review shows that most current evidence comes from aquatic food sources. More comparative studies on other non-aquatic species are necessary to understand the risk difference between farmed and wild-caught wildlife animals.

The high heterogeneity observed across pooled analyses likely reflects differences in host species, pathogens, sampling methods, and production systems. Pooled estimates should therefore be interpreted as directional rather than as precise effect sizes. Quantifying this variation remains valuable for understanding source-associated risks and informing context-specific combinations of risk mitigation measures. The predominance of cross-sectional designs and non-random samplinglimits generalizability and may overestimate or underestimate true comparative prevalence. In general, the primary concerns stem from the representativeness of the sampling methods, high difference between studies, rather than the quality of the analyses themselves. Many studies relied on convenience sampling, and verification of animal source (farmed versus wild-captured) was often based on vendor reporting rather than traceable supply-chain documentation, introducing potential misclassification. While trying to adjust for this variety in risk of bias assessment, we have also experienced limitations in using the JBI checklist. This list is developed for human studies and does not allow for comparative designs. The response-rate item is less applicable, as most animals were sampled post-mortem and animals cannot consent for the procedure. The checklist also lacks a standardised rule for deriving an overall appraisal, leaving room for differences in how assessors weigh individual methodological concerns.

These limitations highlight the need for more systematic sampling and comparative studies across species, settings, and pathogen groups that directly compare farmed and wild-caught wildlife.

5. Conclusion

The most consistent finding across studies was a potentially lower prevalence of parasitic zoonoses with indirect life cycles in farmed aquatic and terrestrial species. In contrast, findings for bacterial pathogens were inconsistent, and comparative data on viral pathogens were scarce. Differences were not observed for bacterial pathogens in fish and shellfish, while higher odds were reported in farmed frogs and freshwater turtles. Only one study assessed a mammalian host. Overall, these findings suggest that pathogen presence is influenced by environmental and management conditions rather than animal source alone. Given the high heterogeneity and risk of bias across studies, results should be interpreted with caution, and further standardised comparative research is needed.

Data sharing

Data presented in this study is available in supplementary documents.

CRediT authorship contribution statement

Alexandra Sack: Writing – review & editing, Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Data curation. Angellica Marta: Writing – review & editing, Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Data curation. Frida Sparaciari: Writing – review & editing, Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Data curation, Conceptualization. Amber Barnes: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Will Sack: Writing – review & editing, Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Data curation. Erik A. Karlsson: Writing – review & editing, Writing – original draft, Visualization, Supervision, Resources, Project administration, Investigation, Funding acquisition, Conceptualization. Charifa Zemouri: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.

Funding

Funding for studies on traditional markets for food at Institut Pasteur du Cambodge (IPC) comes from the Food and Agriculture Organization of the United Nations (FAO) and the UK Academy of Medical Sciences. C.Z. and A. M. were funded, in part, by the World Health Organization (WHO). Funders had no role in the study design; data collection, analysis, or interpretation; writing of the manuscript; or the decision to submit for publication.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Erik Karlsson reports financial support was provided by Food and Agriculture Organization of the United Nations. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

All data included in the manuscript as described