Development of an eDNA‐based survey method for urban fish markets

Over three billion people world-wide rely on fish for at least 20% of their annual protein intake, with a 122% increase in consumption from 1990 to 2018 (FAO, 2018). However, lack both research and regulation has allowed this US$401 industry threaten the sustainability vulnerable stocks, many which are already listed as threatened categories IUCN Red List At particular risk species late sexual maturation reproduction subsequently unable recover high fishing pressures (Sadovy de Mitcheson et al., 2017). Frequently, sourced developing countries, where unmonitored/unmanaged can have wide-reaching ecological, economic criminal consequences (Sumaila & Bawumia, 2014). Many areas biodiversity also prime targets illegal, unreported unregulated (IUU) fishing, regions South China Sea/Indo-Pacific serving grounds poaching that fuels trade networks supplying Hong Kong, Taiwan mainland China's demand fresh seafood 2018; Lam Sadovy Mitcheson, 2011; In Asia, global supply chains consolidated culminate an epicentre local (To 2009) re-exportation fisheries products Here, modes fresh/live distribution involve consolidation goods procured aquaculture countries such China, Indonesia Philippines (Fong Zheng, 2016). Imports enter through few centralized centres they offloaded partitioned by land-based wholesalers retail restaurant establishments While framework monitoring is well-developed within deliberate misidentification misreporting been highlighted key obstacles live trade, import volumes being underestimated much 50% Landed rarely identified level, confusion between common names, English names scientific make it difficult reliably compile fishers' self-reported catches (Lee Sadovy, 1998; Situ 2004; To 2009). Despite integral need gathering information identifying loopholes regulations enforcement, current methods efficiency efficacy. Currently, markets surveyed mainly visual taxonomic identification. surveys trained individuals visit record number observed each stall. Keeping pace volume highly diverse region requires extensive working hours tedious inspection individual sample. addition time-consuming, specifically often limited requisite expertise identify potentially hundreds rapidly accurately active market. Concerns manpower, time efficacy surveys, particularly context large urban 10 –100 s stalls (such those Kong) thus far held back roll out programs. Similar physical appearances related lead critical gaps enforcement. For example, family Epinephelidae (groupers) high-value commodity, while majority presently not threatened, there several traded Endangered, Vulnerable or Near Threatened (IUCN) Due similarity appearance but differences price conservation assessment, be deliberately accidently misidentified vendors. Within wholesale vendors, product turnover sheer size may hinder attempts every present, especially when illegally endangered hidden vendors (Baker, 2008; Rees Even most ability once processed filleted ice. Given constraints large-scale setting place like Kong 2017) reticence traders market managers permit obstructions surveillance, method non-invasively capture wide range target over greater spatial temporal scales could prove invaluable 2008). Further, supporting improving government surveying noted suggestion assessments (indeed, global) management 2003; The simplicity water sample collection ease laboratory processing greatly scaled up means covered relatively minimal increases personnel no training beyond basic technique. Developments next generation sequencing continue advance accuracy assessments. Metabarcoding, taxa simultaneously, reshaped our understanding deep sea microbes (Sogin 2006) coral reef ecosystems (Leray Knowlton, 2015). Recent developments apply metabarcoding techniques environmental DNA (eDNA), is, genetic material present sources (e.g. soils, sediment, water) without obvious biological source material. eDNA-based allow non-invasive (Deiner 2015; Ficetola 2008) well time-series (Thomsen Willerslev, Through primer design, applied broad groups eukaryotes, targeted lower levels taxonomy Actinopterygii, class includes 99% living fishes. advantages (mass-identification, ‘snapshots’ community compositions, potential eDNA water, air soil) well-suited needs wildlife 2016; Karlsson 2020; Wineland 2019). As complementary methodology, strengthen traditional improve confidence assignment enable environments would otherwise challenging. study, goal was test applicability surveillance tool, minimize labour requirements enhance detection visible inspection. this, we extracted wet drain-water samples. Drainage systems collect run-off vendor holding tanks throughout market, hypothetically providing rough snapshot en masse ‘species footprint’ given window. We employed two leading studies, filtration precipitation, compared outcomes. more subsequent extraction, turbid nature samples presents challenge. Precipitation used successfully reportedly less inhibited turbidity presence particulate organic inorganic contamination (Ficetola then photographic survey better understand relative performance. purpose establishing proof concept, study focused smaller scale easier comparison. Retail closing during floors surfaces rinsed bleach, creating reset point Although able explicitly, collected set across 5 days. Three were sampled study: Kennedy Town (KT), Shek Tong Tsui (STT) Sai Ying Pun (SYP). Permits fieldwork unnecessary all drains publicly accessible. Each located <2 km apart close to— served by—the same largest Kong. Unlike supermarkets, composed multiple selling frozen displayed tables aquaria (Figure 1b). maintenance cleaning public daily bleaching washing management. stall varies vendor, usually performed end day (personal communication management). Live accounted <25% surveyed. Instead, whole commonly top drainage varied slightly one facility next, consist ‘moat’ running under display along perimeter area, feeds iced displays into single channel shared (Figures 1 2). Samples days, morning (11:00 a.m., after had received fish). Sterilized hand pumps draw 1.05 L points immediately brought further processing. Large particulates scales, tissue, blood clots, sediment) filtered using 1-μm 0.5-μm sterilized sieves. Triplicate subsamples 15 ml (total = 45 ml) taken precipitation-based filtered. Reverse osmosis (RO)-filtered negative extraction control ensure cross-contamination among extraction. All equipment (i.e. pumps, sieves, filter funnels, scissors forceps) 10% bleach min exposure UV before use. transferred 50-ml tube 1.5 3 M sodium acetate 33 100% ethanol. tubes spun 3,100 g hr 4°C. supernatant poured off 360 μl ATL buffer 40 proteinase K added (DNeasy Blood Tissue Kit, Qiagen). filtration, vacuum-filtered 47-mm diameter glass microfiber (pore size, 0.22 μm; Millipore Express). Where filters became clogged, exchanged new ones. Those (3 per litre cut sterile placed tubes, containing K. Sample precipitation incubated 56°C overnight shaking bath. extractions completed according manufacturer's protocol, 50 elution preheated rounds elution. stored −20°C until amplification. extracts quantified Nanodrop (ThermoFisher), subdivided recombined form 30 representative (market/day) use PCR protocol optimization similar concentrations. mitochondrial 12S gene amplified MiFish primers developed Miya al. (2015), Nextera adapters. conditions optimized annealing temperatures (62°–66.5°). extract concentrations so template dilutions: neat, ½, ⅕ 1/10. quality intensity resulting assessed gel electrophoresis. Finally, samples, 20 mixture Master Mix (Taq, Qiagen), 2 forward reverse μM primer, 4 1/10 dilution. run included no-template RO substituted extract. thermal cycling profile 94°C initial denaturing min; 35 cycles s, 65°C 72°C final extension step min. post-amplification procedures separate workspace fume hood reagents tips. Amplification checked 1.5% agarose Qubit dsDNA HS assay. mix program above employed. Successful amplifications submitted University Centre PanorOrmic Sciences, Genomics Core, 251-bp paired-end Illumina MiSeq platform. Library preparation based 16S Metagenomic Sequencing Preparation Protocol (Illumina). Purified amplicons partial adapter sequences generate libraries validation fragment Bioanalyzer qPCR. detailed including specimen manipulation necessary exact identification, possible study. digital photographs operating concurrently collection. A total 372 photos iPhone X .jpg format about 2–3 MB size. averaged around 20–30 market/day. specimens photo display. Some captured ~50 laid table (see, e.g. Figure 1d) other tub. Later, professional taxonomist who worked years. Regional books identification (Liu 2013; Cornish, 2000; Shao www.fishdb.sinica.edu.tw; www.fishbase.org). Species-level possible, some only resolved genus level. Paired, raw read files uploaded pipeline (http://mitofish.aori.u-tokyo.ac.jp/mifish/) data brief, reads removing atypical lengths. Primer trimmed clustered; chimeras removed uclust uchime algorithms respectively (Edgar, 2010; Edgar, Sequences ≥10 clustered, <10 subjected pairwise alignment considered identical matched >99% well-represented sequence (≥10 reads). Processed BLASTn against MitoFish database (Iwasaki Sato threshold 97% e cut-off 10−5 BLAST. Fish detected controls account contamination. did reach minimum thresholds, took conservative approach sequenced them anyway, any method/day if found corresponding control. Sampling effort evaluated accumulation curves plotted r package BiodiversityR (Kindt Coe, 2005) ‘exact’ method. case, unique included, even resolution remained discrete overlapping (eDNA survey) (at levels). Venn diagrams generated VennDiagram (Chen Boutros, 2011). composition principal coordinates analysis (PCoA) ordination Jaccard distance matrix day, This done phyloseq (McMurdie Holmes, 2013). variation methods, first collapsed failed amplify determined visualization assays confirmed these below (average concentration: 0.3 ng/μl). Of markets, 28 (one amplification failure precipitation). passed QC (minimum ng/μl, average 7.5 efforts returned 41,000 having throughput 8.9 Mb 86% bases score ≥Q30. near plateauing indicate sampling sufficient recording stable days 3). Across 157 (including hybrid Sabah grouper Epinephelus fuscoguttatus x lanceolatus) 106 genera 4a 5). There methodologies overall identity detected. highest (111 species), followed (109 species). 118 either method, 90% precipitation. assigned though these, congeneric survey, 83 grouper) identified. An additional 18 level 14 disparity visual-based DNA-based methodologies, open receive estimate regional market-wide number. Overlap moderate 39 74 Among 17 IDs, belonged eDNA, eDNA. PCoA indicated second axes together explain ~40% family-level observations. clearly distinguished axis explained 29% variation. separation clusters (12% variation), distinguishable methods. clear effect richness. Plotting families component revealed that, fell centre ordination, separated PCoA-1, suggesting methodological bias specific occur 6). aimed develop throughput, complement monitor markets. hypothesized wastewater auditing accumulates tissues shed transport, cutting processes (Jamal Pugazhendi, 2021). Overall, consistently though, surprisingly, restricted (39 respectively; 4a). portion discrepancies attributed difficulties resolving closely morphological similarities implications complexes. case grouper, picked barcoding results, its survey. pointed directly occurrence likely fillet Anguillid eels) surveys. trade-offs suggest ‘snapshot’ provide excellent identifications collectively scope regulating Our results design broadly comprehensive 3), period Differences rates shape curve) important features. recorded increased nearly plateaued standard identifiable Conversely, cumulative expectations system core samples/days, regular novel largely expected; commercial fishes Marketing Organization (FMO: https://www.fmo.org.hk/) previous yielded 4), detect and/or resolve higher than eight four precipitations respectively, stand taxonomically – combined). This, fact combining caused little change how suggests result random rare various robust audit. Whereas (collectively) 6) due, part, challenging efficient accurate Misidentifications due similarities. IDs photos. video recordings consultation taxonomists, necessary, angles sometimes required Just sequence-based confounded complete reference understudied groups. primers, bioinformatic used, currently 6600 species. (97%) assignments (see 2018), poorly matches. Similarity marker inaccurate classification markers future help represent significant workload. Indeed, comparing different groups, 4a) some, certainly all, discrepancies. carries inspection, regard morphologically Cyprinidae (Carp), It challenges species-level co-mingle tank. Consumer preferences towards certain economically groupers, carps perches) numbers marketed name. Incorrect masked distortions factors prices assessments, reported species, Plectropomus leopardus, P. maculatus areolatus very WWF, 1a,b), ‘vulnerable’ (Rhodes, addition, vary, reports mislabelling sale difficulty differentiating advantage (WWF, Decreasing likelihood complex low priority applications infer frequently sold butchered, descaled, modified way obscures eliminates distinctive identifiers butchered Regular visits past shown eels (Richards 2020). contain priority; international European eel Anguilla anguilla CITES-listed concern (Musing (as advertised surveyed). generally, impossible distinguish visually. Two demonstrating utility incorporate fish. (fillets, fish) lacking features still act considering value rare, resistant attempt conceal Hidden noticed neither nor photography permitted issue studies highlight conspicuous illegal candidates, whole, concealed tanks, whatever purpose, availability force expertise, instant, sight, vendor-specific part program. cryptic taxa, take process sequence, sold, precluding direct evidence. Furthermore, incidences visually Though morphologies, always case. cases times never Pseudorasbora parva, Oncorhynchus mykiss, Drepane punctata; common/low biomass short life span quickly enough detection. abundance (Elbrecht Leese, 2015)—variation exacerbated flow patterns, intermittently flowing rather static pool. limits false negatives transiting unclear will vary depending conditions, circulation, practices. (high biomass) concentration drain effluent. Replication safeguard simulations experiments 2015), highlights site-specific planning regimes. One interesting comparison well-described Muraenesox bagio ID contrast, subset (10 accounting market) Thus, missed sample, emphasizing importance replication studies. stocks decline world-wide, Sea Indo-Pacific, awareness, characterization become vital maintaining implementing responsible sustainable practices widely recognized (Karlsson Shaw Thomsen contexts. provides baseline compare limitations mass-survey Both (filtration precipitation) similarly consistent recovery, preferable chosen technical amount collected). success retrieval, environment supports technique species/taxa contexts, invasive amphibian riverine port bodies (Charvoz 2021; Clusa conclusion, demonstrated inform regulatory enforcement efforts. shark known (Lam Lee 2003), improved intelligence scope, seasonality/trends knowledge, demonstrates instance trade. show capacity audits prompt restrictions. way, application valuable seasonal patterns repeated instances protected trading. intended suitable court evidence, great enabling officials landing discreetly conduct investigations evidence activity. Funding provided Collaborative Research Fund (CRF), Grants Council (CRF, C7013-19G) awarded DMB. authors thank PanorOmic Sciences (CPOS) facilities, service support. declare conflict interest. J.L.R., V.S. D.B. conceived designed study; J.L.R. H.W.Y.C. work; V.S., S.E.M., R.H.H.T. M.L. analysed data. contributed drafting editing manuscript. approved peer review history article available https://publons.com/publon/10.1111/2041-210X.13842. Raw NCBI Short Read Archive (SRA) BioProject accession PRJNA796006. OTU (presence/absence), metadata https://github.com/shelby26/eDNA-wetmarket archived Zenodo 10.5281/zenodo.6340046 (Chue Hong,