Received: 12 November 2021 | Accepted: 17 February 2022

DOI: 10.1111/2041-210X.13842

RESEARCH ARTICLE

Development of an eDNA-­based survey method for urban

fish markets

John L. Richards1 | Victoria Sheng1 | Haze Wing Yi Chung1 | Min Liu2 |

Rainbow Hin Hung Tsang1 | Shelby E. McIlroy1 | David Baker1

1
Swire Institute of Marine Science, School
of Biological Sciences, The University of Abstract
Hong Kong, Pokfulam Road, Hong Kong
1. Fish and fishery products are among the most highly traded commodities in the
SAR
2
State Key Laboratory of Marine
world, and over-­exploitation continues to threaten the biodiversity and sustain-
Environmental Science and College ability of global stocks. The state of knowledge for many fish species is limited
of Ocean and Earth Sciences, Xiamen
University, Xiamen, Fujian, China
by current monitoring techniques, which rely on labour-­intensive visual or ge-
netic surveys of individual specimens (often at inconsistent or coarse taxonomic
Correspondence
Shelby E. McIlroy
resolution).
Email: [email protected] 2. To address the need for more efficient methods that effectively monitor trade,
David Baker we developed a novel application of eDNA-­based metabarcoding that can iden-
Email: [email protected]
tify a broad range of fish taxa from effluent water draining from urban fish
Funding information markets. Using samples collected at three Hong Kong fish markets over a 5-­
Research Grants Council, University
Grants Committee, Grant/Award Number: day period, we tested two DNA capture protocols (filtration vs. precipitation)
CRF, C7013-­19G with the goal of ensuring that methods were relatively easy to implement with-
Handling Editor: Lynsey Harper out sacrificing species coverage relative to each other and to a standard visual
survey.
3. eDNA-­based survey methods were able to overcome many of the challenges
associated with visual identification, such as resolving morphologically simi-
lar species and accounting for butchered fish products. However, rare and/or
quickly transiting species occasionally went undetected. The two variations of
eDNA isolation methods performed almost identically, emphasizing the relative
robustness of eDNA metabarcoding, and allowing for user flexibility depending
on the turbidity of collected waters. Our results demonstrate the ability of a few
spatial and temporal replicates to generate an efficient and accurate market-­
wide species inventory that can complement visual inspections and guide tar-
geted investigations of illegal trade.
4. This research provides important baseline information on the capabilities of eDNA
metabarcoding as a non-­invasive survey tool, with the potential to complement
conventional surveys. To our knowledge, this study is the first proof of concept
using eDNA metabarcoding to monitor wildlife trade and also highlights the poten-
tial for eDNA metabarcoding to be applied to investigations of urban ecology.

John L. Richards and Victoria Sheng contributed equally to this work.

© 2022 The Authors. Methods in Ecology and Evolution © 2022 British Ecological Society.

1568 | 
wileyonlinelibrary.com/journal/mee3 Methods Ecol Evol. 2022;13:1568–1580.
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RICHARDS et al. Methods in Ecology and Evolu on 1569

KEYWORDS
conservation, eDNA, fish market, laboratory methods, metabarcoding, species identification,
wildlife trade

1 | I NTRO D U C TI O N the number of individuals and species observed in each stall. Keeping
pace with trade volume in a highly diverse region requires extensive
1.1 | Over-­exploitation of fish stocks and working hours and tedious inspection of each individual sample. In
unmonitored trade addition to being time-­consuming, visual surveys specifically are
often limited by the requisite taxonomic expertise to identify po-
Over three billion people world-­wide rely on fish for at least 20% of tentially hundreds of fish species rapidly and accurately in an active
their annual protein intake, with a 122% increase in fish consumption market. Concerns of manpower, time and efficacy of visual surveys,
from 1990 to 2018 (FAO, 2018). However, a lack of both research and particularly in the context of large urban markets with 10 –­100 s of
regulation has allowed this US$401 billion industry to threaten the stalls (such as those in Hong Kong) have thus far held back the roll
sustainability of vulnerable stocks, many of which are already listed as out of extensive monitoring programs.
threatened categories in the IUCN Red List (FAO, 2018). At particular Similar physical appearances of related fish species can lead
risk are species with late sexual maturation and reproduction which are to critical gaps in trade enforcement. For example, the family
subsequently unable to recover from high fishing pressures (Sadovy Epinephelidae (groupers) is a common and high-­value commodity,
de Mitcheson et al., 2017). Frequently, fish are sourced from develop- and while the majority of species within this family are presently
ing countries, where unmonitored/unmanaged fishing can have wide-­ not threatened, there are several traded species which are listed as
reaching ecological, economic and criminal consequences (Sumaila & Endangered, Vulnerable or Near Threatened (IUCN) (To & Sadovy de
Bawumia, 2014). Many areas of high biodiversity are also prime targets Mitcheson, 2009). Due to their similarity in appearance but differ-
for illegal, unreported and unregulated (IUU) fishing, with regions of the ences in price and conservation assessment, such species can be de-
South China Sea/Indo-­Pacific serving as fishing grounds for the poach- liberately or accidently misidentified by vendors. Within wholesale
ing that fuels the trade networks supplying Hong Kong, Taiwan and markets serving hundreds of vendors, the pace of product turnover
mainland China's demand for fresh seafood (FAO, 2018; Lam & Sadovy and sheer physical size may hinder attempts to record every species
de Mitcheson, 2011; Sadovy de Mitcheson et al., 2017). present, especially when illegally traded endangered species are de-
In Asia, global supply chains are consolidated and culminate in liberately hidden by vendors (Baker, 2008; Rees et al., 2014). Even
Hong Kong, an epicentre of both local consumption (To & Sadovy de the most well-­developed visual surveys lack the ability to identify
Mitcheson, 2009) and re-­exportation of fisheries products (Sadovy de products once fish are processed or filleted on ice. Given the con-
Mitcheson et al., 2017). Here, modes of fresh/live seafood distribution straints of visual surveys, particularly within a large-­scale setting in
also involve consolidation of goods procured from aquaculture and a place like Hong Kong where deliberate misreporting (Sadovy de
fishing in countries such as China, Indonesia and the Philippines (Fong Mitcheson et al., 2017) and reticence of traders and market manag-
& Zheng, 2016). Imports enter through a few centralized distribution ers to permit inspection are common obstructions to surveillance,
centres where they are offloaded and partitioned by land-­based whole- a method that can non-­invasively capture a wide range of target
salers to retail and restaurant establishments (Fong & Zheng, 2016). species over greater spatial and temporal scales could prove invalu-
While the framework for fisheries monitoring is well-­developed within able for monitoring (Baker, 2008). Further, supporting and improving
Hong Kong, deliberate misidentification and misreporting have been government surveying and monitoring methods has been noted as a
highlighted as key obstacles of monitoring the live seafood trade, with key suggestion in many assessments of Hong Kong (indeed, global)
import volumes being underestimated by as much as 50% (Sadovy de fish market management (FAO, 2018; Sadovy et al., 2003; Sadovy
Mitcheson, 2017). Landed fish are also rarely identified to the species de Mitcheson et al., 2017). The simplicity of water sample collec-
level, and confusion between local common names, English common tion and ease with which laboratory processing can be greatly scaled
names and scientific names make it difficult to reliably compile the fish- up means that large areas can potentially be covered with relatively
ers' self-­reported catches (Lee & Sadovy, 1998; Situ & Sadovy, 2004; To minimal increases in time or personnel and no requisite training be-
& Sadovy de Mitcheson, 2009). Despite the integral need for gathering yond basic laboratory technique.
trade information and identifying loopholes in trade regulations and
their enforcement, current methods lack efficiency and efficacy.
1.3 | Application of eDNA for trade monitoring

1.2 | Conventional survey challenges Developments in next generation sequencing continue to ad-
vance the efficiency and accuracy of biodiversity assessments.
Currently, markets are surveyed mainly by visual taxonomic identi- Metabarcoding, the ability to identify many taxa within a sample
fication. In such surveys trained individuals visit markets and record simultaneously, has reshaped our understanding of biodiversity
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1570 Methods in Ecology and Evolu on RICHARDS et al.

from deep sea microbes (Sogin et al., 2006) to coral reef ecosystems Three retail wet markets were sampled in this study: Kennedy
(Leray & Knowlton, 2015). Recent developments apply metabarcod- Town (KT), Shek Tong Tsui (STT) and Sai Ying Pun (SYP). Permits
ing techniques to environmental DNA (eDNA), that is, genetic ma- for fieldwork were unnecessary as all the drains sampled were pub-
terial present in environmental sources (e.g. soils, sediment, water) licly accessible. Each are located <2 km apart and are close to—­and
without obvious biological source material. eDNA-­based metabar- served by—­the same wholesale seafood market, the largest in Hong
coding methods allow for non-­invasive biodiversity assessments Kong. Unlike supermarkets, wet markets are composed of multiple
(Deiner et al., 2015; Ficetola et al., 2008) as well as time-­series moni- individual vendors selling live or frozen fish products displayed on
toring (Thomsen & Willerslev, 2015). Through primer design, me- tables or in aquaria (Figure 1b). The maintenance and cleaning of
tabarcoding techniques can be applied to broad taxonomic groups the public areas includes daily bleaching and washing by the market
such as eukaryotes, or targeted to lower levels of taxonomy such management. While cleaning within each stall varies by vendor, the
Actinopterygii, a class that includes 99% of living fishes. same bleaching and washing is usually performed at the end of each
The advantages of metabarcoding (mass-­identification, ‘snap- day (personal communication with market management). Live fish
shots’ of community compositions, a range of potential eDNA accounted for <25% of the fish at markets we surveyed. Instead,
sources like water, air or soil) make it well-­suited to the needs of wild- whole fresh fish and filleted fish are commonly displayed on top of
life trade surveys (Deiner et al., 2016; Karlsson et al., 2020; Wineland ice. The drainage systems of each wet market varied slightly from
et al., 2019). As a complementary methodology, metabarcoding can one facility to the next, but all consist of a ‘moat’ running under the
strengthen traditional visual surveys and biodiversity assessments display tables along the perimeter of the market area, which feeds
and improve confidence in taxonomic assignment or enable survey- run-­off from both live tanks and iced displays into a single drainage
ing in environments that would otherwise be challenging. channel shared by several stalls (Figures 1 and 2).
Samples were collected from each of the three wet markets over
5 days, at the same time each morning (11:00 a.m., after the vendors
1.4 | Research overview and objectives had received their daily supply of fish). Sterilized hand pumps were
used to draw 1.05 L water from several points along the wet market
In this study, our goal was to test the applicability of eDNA methods drains (Figure 2).
as a surveillance tool, which could potentially minimize expertise and
labour requirements and enhance the detection of species that are
not visible at the time of inspection. To test this, we extracted eDNA 2.2 | DNA capture, extraction and sequencing
from wet market drain-­water samples. Drainage systems collect
run-­off from vendor stalls and holding tanks throughout the mar- Samples were immediately brought back to the laboratory for fur-
ket, hypothetically providing a rough snapshot or en masse ‘species ther processing. Large particulates (e.g. scales, tissue, blood clots,
footprint’ of fish present in the market for a given time window. We inorganic sediment) were filtered from the samples using 1-­μm and
employed the two leading DNA capture techniques for eDNA stud- 0.5-­μm sterilized sieves. Triplicate subsamples of 15 ml (total vol-
ies, filtration and precipitation, and compared their outcomes. While ume = 45 ml) were taken from each sample for precipitation-­based
filtration is more common in the capture of eDNA and subsequent DNA capture and extraction, and 1 L was filtered. Reverse osmosis
extraction, the turbid nature of the water samples presents a chal- (RO)-­filtered water was employed as a negative extraction control to
lenge. Precipitation has also been used successfully to capture DNA ensure there was no cross-­contamination among water samples dur-
from water samples and is reportedly less inhibited by turbidity and ing DNA capture and extraction. All equipment (i.e. pumps, sieves,
the presence of particulate organic and inorganic contamination in filter funnels, scissors and forceps) were sterilized by washing with
the water sample (Ficetola et al., 2008). We then compared the spe- 10% bleach and 15 min exposure to UV immediately before use.
cies identified by eDNA metabarcoding with a photographic visual For precipitation, each 15 ml sample was transferred into a
survey to better understand relative performance. 50-­ml tube with 1.5 ml of 3 M sodium acetate and 33 ml of 100%
ethanol. The tubes were then spun at 3,100 g for 1 hr at 4°C. The
supernatant was poured off and 360 μl of ATL extraction buffer and
2 | M ATE R I A L S A N D M E TH O DS 40 μl of proteinase K were added (DNeasy Blood and Tissue Kit,
Qiagen). For filtration, 1 L of sample was vacuum-­filtered through a
2.1 | Study site and sample collection 47-­mm diameter glass microfiber filter (pore size, 0.22 μm; Millipore
Express). Where filters became clogged, they were exchanged with
For the purpose of establishing a proof of concept, this study fo- new ones. Those filters (3 per litre of filtered water) were then cut
cused on retail markets which are smaller in scale and allow for eas- using sterile scissors and placed into three 1.5 ml tubes, each con-
ier visual comparison. Retail markets also have closing hours during taining 360 μl of ATL extraction buffer and 40 μl of proteinase K.
which floors and surfaces are rinsed with bleach, potentially creating Sample tubes from the precipitation and filtration methods were in-
a reset point for eDNA source material. Although we are not able to cubated at 56°C overnight in a shaking water bath. DNA extractions
test this explicitly, we collected samples at a set time across 5 days. were then completed according to the manufacturer's protocol, with
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RICHARDS et al. Methods in Ecology and Evolu on 1571

F I G U R E 1 Photographs from the wet markets surveyed. (a) Fresh and butchered fish are displayed and sold on ice which melts into
common drainage areas. (b) Plectropomus leopardus (circled in red) among closely related species with similar in appearance, for example,
other red, spotted Perciformes such as Plectropomus areolatus and P. maculatus. (c) Live fish tanks with commonly found live reef fish. (d and
e) Whole fish on sale next to some butchered fish parts lacking characteristics for visual IDs

50 μl of elution buffer preheated to 56°C used for two rounds of Nextera adapter sequences were used to generate libraries for val-
elution. Samples were stored at −20°C until amplification. idation of fragment size by Bioanalyzer and quality control using
DNA extracts were quantified by Nanodrop (ThermoFisher), and Qubit and qPCR.
the subdivided extracts recombined to form a 30 μl representative of
each sample (market/day) (Figure 2). All extracts for use in the PCR
protocol optimization were of similar concentrations. The mitochon- 2.3 | Photographic surveys
drial 12S gene was amplified using the MiFish primers developed by
Miya et al. (2015), with the addition of Nextera adapters. PCR condi- While detailed inspection including specimen manipulation is often
tions were optimized across a range of annealing temperatures (62°–­ necessary for exact species identification, this was not possible
66.5°). DNA extract concentrations were similar across all samples given the nature of the present study. Instead, digital photographs
and so PCR template concentrations were optimized across a range were taken at each market of all operating fish vendor stalls con-
of dilutions: neat, ½, ⅕ and 1/10. The size, quality and intensity of the currently with water sample collection. A total of 372 photos were
resulting products were assessed with gel electrophoresis. Finally, for taken using an iPhone X in .jpg format with each about 2–­3 MB in
all samples, PCR was performed with a 20 μl mixture containing 10 μl size. The number of photos taken averaged around 20–­30 photos
Master Mix (Taq, Qiagen), 2 μl of each forward and reverse 10 μM per market/day. The number of specimens per photo varied on the
primer, 4 μl water and 2 μl of DNA extract at a 1/10 dilution. Each PCR stall display. Some photos captured up to ~50 fish laid out on a dis-
run included a no-­template control where 2 μl of RO was substituted play table (see, e.g. Figure 1d) where other photos captured two or
for the DNA extract. The thermal cycling profile included 94°C initial three live fish in a tub. Later, fish were identified from photos by a
denaturing for 1 min; 35 cycles of 94°C for 45 s, 65°C for 1 min and professional fish taxonomist who has worked in the region for about
72°C for 1 min; and a final extension step of 72°C for 10 min. 20 years. Regional fish taxonomic books were used for identification
All post-­amplification procedures were performed in a separate (Liu et al., 2013; Sadovy & Cornish, 2000; Shao et al., 2015; www.
workspace under a fume hood using fresh reagents and filter tips. fishdb.sinica.edu.tw; www.fishb​ase.org). Species-­level identification
Amplification products were checked by a 1.5% agarose gel and was given where possible, while some individuals could only be re-
quantified using the Qubit dsDNA HS assay. The same PCR mix and solved to the genus or family level.
thermal cycling program as above were employed. Successful am-
plifications were submitted to the University of Hong Kong Centre
for PanorOrmic Sciences, Genomics Core, for 251-­bp paired-­end 2.4 | Analysis
sequencing on the Illumina MiSeq platform. Library preparation
was performed based on the 16S Metagenomic Sequencing Library Paired, raw read files were uploaded to the MiFish pipeline (http://
Preparation Protocol (Illumina). Purified amplicons with partial mitof​ish.aori.u-­tokyo.ac.jp/mifis​h/) for data processing. In brief,
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1572 Methods in Ecology and Evolu on RICHARDS et al.

F I G U R E 2 Study site, sampling and workflow schematic. (a) Water samples were collected at several sampling points along moat-­like
underground drainage systems (red bottles). (b) Water samples from each market were subdivided for either filtration or precipitation DNA
capture protocols. DNA extracts were quantified and pooled for amplification, library prep and sequencing

reads were filtered for quality by removing sequences of atypical (Iwasaki et al., 2013; Sato et al., 2018). A similarity threshold of 97%
lengths. Primer and adapter sequences were trimmed and reads were and an e cut-­off of 10−5 were used for taxonomic assignment based
clustered; chimeras were removed using the uclust and uchime algo- on BLAST.
rithms respectively (Edgar, 2010; Edgar, 2016). Sequences with ≥10 Fish species detected were compared across methods and with
reads were clustered, while those with <10 were subjected to pair- the negative extraction controls to account for possible contam-
wise alignment and considered identical when matched with >99% ination. While the extraction controls did not reach sequencing
similarity to a well-­represented sequence (≥10 reads). Processed minimum thresholds, we took a conservative approach and se-
reads were then subjected to BLASTn against the MitoFish database quenced them anyway, removing any species present for a given
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RICHARDS et al. Methods in Ecology and Evolu on 1573

method/day if found in the corresponding method/day negative

extraction control.
Sampling effort and survey methods were evaluated with species
accumulation curves plotted with the r package BiodiversityR (Kindt
& Coe, 2005) using the ‘exact’ method. In this case, each unique
identification was included, even where the taxonomic resolution
remained above the species level. The number of discrete and over-
lapping taxa identified using the three methods (eDNA by filtration,
eDNA by precipitation and photographic survey) were compared
according to the taxonomic resolution (at genus and species levels).
Venn diagrams were generated using the r package VennDiagram
(Chen & Boutros, 2011).
The taxonomic composition of each survey method and market
was compared using principal coordinates analysis (PCoA) ordina-
tion based on the Jaccard distance matrix of identified taxa for F I G U R E 3 Species accumulation curves. Data from each
each day, market and method. This was done using the r package methodology, and both eDNA methods combined, are shown for
samples taken across 5 days from three Hong Kong wet markets.
phyloseq (McMurdie & Holmes, 2013). Due to variation in taxo-
Shaded envelope shows 95% confidence intervals. Each unique
nomic resolution among methods, taxa were first collapsed at the identification was included, even where the taxonomic resolution
family level. remained above the species level. One sample from each eDNA
method, filtration and precipitation, failed to amplify

3 | R E S U LT S to the family level. This disparity between the taxonomic resolution

of visual-­based and DNA-­based methodologies, as well as the open
All negative controls for extractions and PCR no-­template controls nature of the wet markets which can potentially receive new species
failed to amplify as determined by visualization with gel electro- each day limited our ability to estimate a regional or market-­wide
phoresis. Qubit assays were used on extraction controls which total species number.
further confirmed that these were well below the threshold for Overlap between eDNA and photographic survey methods was
sequencing (average concentration: 0.3 ng/μl). Of the 30 water moderate with 40 species detected by all three methods, 39 species
samples taken from the drainage of retail markets, 28 successfully (including the Sabah grouper) detected only by the photographic
amplified in PCR (one amplification failure each for filtration and survey and 74 species detected only by eDNA methods (Figures 4
precipitation). PCR products were quantified with Qubit assays and 5). Among those 17 taxa limited to the genus level in photo-
and all water samples passed QC controls (minimum concentra- graphic IDs, 14 belonged to a genus detected by eDNA, while 17
tion: 3 ng/μl, average concentration: 7.5 ng/μl). Sequencing efforts belonged to a family also detected by eDNA.
returned an average of 41,000 reads per sample with each sample PCoA indicated that the first and second axes together explain
having an average throughput of 8.9 Mb and 86% of bases had a ~40% of the variation among family-­level observations. The photo-
quality score ≥Q30. graphic surveys could be clearly distinguished from eDNA methods
The near plateauing species accumulation curves indicate that along the first axis which explained 29% of the variation. Within the
sampling effort was sufficient in recording most species for each photographic survey, separation of market clusters could be ob-
methodology, and that taxa were relatively stable across the days served along the second axis (12% of the variation), though markets
and markets (Figure 3). Across all surveys and methods, a total of were less distinguishable by the eDNA methods. There was not a
157 species (including the hybrid Sabah grouper Epinephelus fus- clear effect of sampling day on family richness. Plotting fish families
coguttatus x Epinephelus lanceolatus) and 106 genera were detected along each principal component revealed that, while most families
within the wet markets (Figures 4a and 5). There was variation fell near the centre of the ordination, a few were separated along
among the methodologies in both the overall number and identity PCoA-­1, suggesting a methodological bias for or against identifying
of taxa detected. eDNA capture by filtration had the highest species specific families did occur (Figure 6).
detection (111 species), followed by precipitation (109 species). Of
the 118 species detected by either eDNA method, 90% were de-
tected by both filtration and precipitation. Within the eDNA sur- 4 | DISCUSSION
vey methods, three sequences were not able to be assigned beyond
the genus level, though for two of these, a congeneric species was In this study, we aimed to develop a high throughput, DNA se-
also detected. In the photographic survey, 83 (including the Sabah quencing method to complement and strengthen the visual
grouper) species were identified. An additional 18 taxa could only surveys which are most often used to monitor trade in urban
be identified to the genus level and 14 more could only be identified wet markets. We hypothesized that eDNA metabarcoding of
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1574 Methods in Ecology and Evolu on RICHARDS et al.

F I G U R E 4 Venn diagrams of number

of taxa recovered across methods.
A: taxa identified to species level B:
taxa identified to genus level. Taxa
only identified to the family level
were not included. The hybrid fish
species Sabah grouper Epinephelus
fuscoguttatus × Epinephelus lanceolatus)
was classified as a distinct species

wastewater collected from the drainage systems would be well-­ 4.1 | Taxa recovery
suited for auditing the presence of wet market species given that
it accumulates fish scales, blood and tissues shed during trans- Our results suggest that the sampling design was broadly com-
port, washing and cutting processes (Jamal & Pugazhendi, 2021). prehensive (Figure 3), and market species were relatively stable
Overall, we found that 40 species were consistently detected throughout the study period (Figure 5). Differences in the taxonomic
by both visual and eDNA-­b ased methods, though, surprisingly, resolution between visual and DNA-­based methods limited our abil-
many more were restricted to either visual or eDNA only (39 ity to estimate a total species number within the markets, while the
and 74 species respectively; Figure 4a). A large portion of these rates of identification over time (i.e. shape of the curve) highlighted
discrepancies could be attributed to difficulties in resolving important methodological features. For example, the number of spe-
closely related taxa with morphological or sequence similarities cies recorded in the photographic surveys increased relatively rap-
with implications for the conservation of species complexes. In idly and then nearly plateaued suggesting efficiency but also bias for
the case of the hybrid Sabah grouper, while the eDNA may have a standard set of clearly identifiable and commonly displayed fish
been picked up by the survey, it is not possible to identify the species (Figure 3). Conversely, the cumulative number of species
hybrid from barcoding results, and its presence was noted only identified by eDNA methods matched our expectations for an open
in the photographic survey. However, other differences pointed wet market system with a core set of taxa accounted for within a few
directly to the unique advantages of each method, for example, samples/days, and regular addition of novel products in the market.
the occurrence of likely fillet products (such as Anguillid eels) The species detected across retail markets were largely expected;
in eDNA assays but not in visual surveys. Given the trade-­offs around 90% were common commercial fishes recorded in the data-
in the accuracy and efficiency of each method, we suggest that base of the Fish Marketing Organization (FMO: https://www.fmo.
the ‘snapshot’ of species present generated with eDNA methods org.hk/) and previous market surveys in Hong Kong (Sadovy de
provide an excellent complement to visual identifications which Mitcheson, 2017). The two eDNA-­based methods yielded similar
collectively can increase the scope of monitoring and regulating results to each other (Figure 4), and each was able to detect and/or
trade in urban wet markets. resolve an overall higher number of species than the photographic
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RICHARDS et al. Methods in Ecology and Evolu on 1575

F I G U R E 5 Cladogram was generated using PhyloT (Letunic & Bork, 2007 http://phylot.bioby​te.de/) using NCBI taxonomy. Only species
IDs from the photographic survey made to at least the genus level were used (asterisk denotes identifications made to the genus level).
Colours represent survey methods with results for each market/day
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1576 Methods in Ecology and Evolu on RICHARDS et al.

F I G U R E 6 Biplot of principal
coordinates analysis of market taxa.
Jaccard distance metric was calculated for
the difference in fish families observed
in market surveys from three Hong Kong
wet markets using three different survey
methodologies. Fish families were then
plotted along each principal component

surveys. The eight and four taxa unique to filtration and precipita- IDs were based on the analysis of photos. While photo and video
tions methods, respectively, did not stand out taxonomically from recordings allow for more time and consultation with multiple taxon-
the 118 species identified across both eDNA methods (Figure 3 –­ omists, if necessary, they may not capture all angles of the specimen
eDNA combined). This, and the fact that combining the results of sometimes required for accurate identification. Just as with visual
the two methods caused little change in how species were detected surveys, sequence-­based identifications can also be confounded
(Figure 3), suggests that these were the result of random variation in by similarity among closely related taxa and/or a complete lack of
the detection of rare eDNA from various samples and may indicate reference sequences for understudied groups. In this study, the
that increased water volumes could lead to a more robust species MitoFish primers, bioinformatic pipeline and database were used,
audit. Whereas the clear taxonomic differences in species detected which currently has sequences for more than 6600 fish species.
by eDNA (collectively) and photographic methods (Figures 4 and 6) Despite setting a high minimum similarity threshold (97%) for tax-
are more likely due, at least in part, to the challenging nature of ef- onomic assignments (see Sato et al., 2018), poorly resolved species
ficient and accurate visual surveys. can be misidentified to close matches. Similarity in marker regions
Misidentifications are most likely among closely related taxa due may also lead to inaccurate taxonomic classification by barcoding
to morphological and/or sequence similarities. In this study, visual but the use of multiple markers in the future may help to improve
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RICHARDS et al. Methods in Ecology and Evolu on 1577

the accuracy of identification and would not represent a significant Anguillid species were detected in the eDNA-­based surveys while
increase in workload. Indeed, comparing the survey results of the no eel species were reported in the photographic surveys, demon-
different methods at higher taxonomic groups, for example, genus strating the utility of eDNA-­based surveys to incorporate fish prod-
level (Figure 4a) or family level (Figure 6) resolved some, but cer- ucts as well as live fish. Many other fish products (fillets, butchered
tainly not all, of these discrepancies. whole fish) were displayed that, while lacking features to make visual
identification possible, could still act as a source of eDNA.
Further, considering the high value of rare, often endangered
4.2 | eDNA methods improve monitoring of cryptic species, fish market vendors are resistant to inspection and may
species, processed meats and covert trade attempt to conceal these frequently traded species. Hidden tanks
were noticed during our visual market surveys, and neither close in-
The use of genetic identification carries some advantages over vis- spection nor photography of these areas were permitted by vendors.
ual inspection, particularly with regard to morphologically similar This issue has also been noted in previous studies (Baker, 2008).
species. While both eDNA methods detected many species in the While we are unable to highlight any conspicuous illegal trade can-
family Cyprinidae (Carp), only three species were detected by the didates, given the common drainage system of the wet market as
photographic survey (Figure 5). It is likely that morphological simi- a whole, it is likely that taxa from concealed tanks, whatever their
larity among species within this family presents challenges for ac- purpose, are included in eDNA but not visual surveys.
curate species-­level identification, especially where multiple species
co-­mingle in a single large display tank. Consumer preferences to-
wards certain economically important groups of fish (e.g. groupers, 4.3 | Visual surveys provide fine spatial–­temporal
or carps and perches) can make it more likely that these species are resolution of both common and quickly traded species
displayed in large numbers with a mix of species marketed under
common name. While visual surveys can be limited by the availability of a labour
Incorrect species identifications that are masked by morpholog- force with taxonomic expertise, the ability to provide instant, on
ical similarities can also result in distortions of other factors such as sight, vendor-­specific identifications is a critical part of any trade
market prices or conservation assessments, as are often reported in regulation program. eDNA methods may be better able to resolve
fish market surveys (Sadovy de Mitcheson, 2017). For example, three cryptic taxa, but the samples take time to process and sequence,
grouper species, Plectropomus leopardus, P. maculatus and P. areola- after which the products can be sold, thus precluding collection
tus are very similar in appearance (see WWF, 2016; Figure 1a,b), but of direct evidence. Furthermore, there were several incidences of
only P. areolatus is listed as ‘vulnerable’ by the IUCN (Rhodes, 2018). species visually observed but not detected with eDNA methods.
In addition, the average retail prices for all three of these species Though some of these taxonomic groups were noted for their cryp-
vary, and there have been reports of fish vendors mislabelling spe- tic morphologies, this was not always the case. There were several
cies on sale and using the difficulty in differentiating among these cases in which a species was observed one to two times within a
species to their advantage (WWF, 2016). Indeed, in our survey re- market but never detected by eDNA methods (e.g. Pseudorasbora
sults, all three species were detected across all three markets by the parva, Oncorhynchus mykiss, Drepane punctata; Figure 3). This
photographic survey, while only P. maculatus was detected by eDNA suggests that less common/low biomass species or those which
(Figure 5). Decreasing the likelihood of misidentification of a conser- have a short life span within the market, for example, are rare
vation target among a species complex including those of low pri- or very quickly sold, may not shed enough eDNA for detection.
ority is an important advantage of eDNA applications in improving Indeed, low biomass has been shown to result in higher variation
trade regulation and enforcement. in sequence abundance (Elbrecht & Leese, 2015)—­variation which
We were also able to infer that eDNA methods are more likely to could be further exacerbated by water flow patterns, especially
detect fish species frequently sold butchered, descaled, filleted or as in our survey we were sampling from an intermittently flowing
otherwise modified for consumption in a way that obscures or elim- drainage channel rather than a static pool. The limits of detec-
inates distinctive morphological identifiers by comparing detection tion that could result in false negatives for low biomass or quickly
of species commonly sold butchered in the markets. Regular visits to transiting species with eDNA methods are unclear and will likely
Hong Kong wet markets and past surveys have shown the presence vary depending on specific market conditions, for example, water
of Anguillid eels as common market fishes in Hong Kong (Richards circulation, UV exposure and cleaning practices. eDNA from com-
et al., 2020). Within the live trade, Anguillid eels contain morpholog- mon (high biomass) species may be more resistant to sequence
ically similar species of conservation priority; the international trade abundance variation due to a higher concentration of eDNA pre-
in European eel Anguilla anguilla being CITES-­listed and of particular sent in drain effluent. Replication has been shown to be a safe-
conservation concern (Musing et al., 2018). However, eels are often guard against the false negatives caused by these concentration
sold butchered in Hong Kong markets (as was advertised on dis- differences in simulations and experiments (Ficetola et al., 2015),
plays in the three markets we surveyed). While eel products can be and further highlights the need for site-­specific study planning
identified generally, taxa are impossible to distinguish visually. Two and sampling regimes.
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1578 Methods in Ecology and Evolu on RICHARDS et al.

One interesting comparison is that of a common and well-­ to monitor fish markets and landing points discreetly and using those
described eel species, Muraenesox bagio which was recorded by data to conduct targeted investigations and collect traditional DNA
visual ID every day of the survey across all three markets. In con- sampling as evidence of illegal activity.
trast, eDNA methods detected this species on a subset (10 of 15
accounting for both capture methods and each market) of those days AC K N OW L E D G E M E N T S
(Figure 3). Thus, even common species may be missed in a single Funding was provided by the Collaborative Research Fund (CRF),
eDNA sample, emphasizing the importance of sampling design and Research Grants Council (CRF, C7013-­19G) awarded to DMB. The
replication in eDNA studies. authors thank the University of Hong Kong Centre for PanorOmic
Sciences (CPOS) for sequencing facilities, service and support.

4.4 | Recommendations for monitoring trade in C O N FL I C T O F I N T E R E S T

urban wet markets The authors declare they have no conflict of interest.

As fisheries stocks continue to decline world-­wide, and particu- AU T H O R S ' C O N T R I B U T I O N S

larly in the South China Sea and Indo-­P acific, increased aware- J.L.R., V.S. and D.B. conceived and designed the study; J.L.R. and
ness, understanding and characterization of the trade has become H.W.Y.C. performed sample collection and laboratory work; J.L.R.,
vital to maintaining and implementing responsible and sustain- V.S., S.E.M., R.H.H.T. and M.L. analysed the data. All authors contrib-
able fishing practices (FAO, 2018). The use of eDNA in biodiver- uted to the drafting and editing of the manuscript. All authors read
sity monitoring is widely recognized (Karlsson et al., 2020; Shaw and approved the final manuscript.
et al., 2016; Thomsen & Willerslev, 2015), and such methods could
also have important applications for monitoring wildlife trade in PEER REVIEW
urban contexts. The peer review history for this article is available at https://publo​ns.​
Our approach provides a baseline framework to compare ad- com/publo​n/10.1111/2041-­210X.13842.
vantages and limitations of eDNA mass-­survey methods as a com-
plement to traditional visual surveys. Both eDNA capture methods DATA AVA I L A B I L I T Y S TAT E M E N T
(filtration and precipitation) were able to provide similarly consis- Raw sequence data have been uploaded to the NCBI Short Read
tent species recovery, so that the preferable method can be chosen Archive (SRA) under BioProject accession number PRJNA796006.
based on technical constraints (e.g. turbidity or the amount of water OTU (presence/absence), taxa and sample metadata are available
that can be collected). While our study focused on the use of eDNA at https://github.com/shelb​y26/eDNA-­wetma​rket and archived on
for monitoring a market area, the success we had in retrieval, ampli- Zenodo 10.5281/zenodo.6340046 (Chue Hong, 2019).
fication and sequencing of eDNA from the urban environment sup-
ports the use of this technique in monitoring species/taxa in other ORCID
urban contexts, such as invasive fish or amphibian species in river- John L. Richards https://orcid.org/0000-0003-4428-1580
ine port areas and urban water bodies (Charvoz et al., 2021; Clusa Shelby E. McIlroy https://orcid.org/0000-0003-2482-8817
et al., 2021). David Baker https://orcid.org/0000-0002-0308-4954
In conclusion, this study demonstrated a novel approach to mon-
itor wildlife trade with the use of eDNA surveys in urban wet market REFERENCES
conditions that can better inform regulatory and enforcement ef- Baker, C. S. (2008). A truer measure of the market: The molecular ecol-
forts. Many IUCN and CITES-­listed fish and shark species are known ogy of fisheries and wildlife trade. Molecular Ecology, 17(18), 3985–­
3998. https://doi.org/10.1111/j.1365-­294X.2008.03867.x
to be traded in Hong Kong (Lam & Sadovy de Mitcheson, 2011; Lee
Charvoz, L., Apothéloz-­Perret-­Gentil, L., Reo, E., Thiébaud, J., &
& Sadovy, 1998; Sadovy et al., 2003), and species-­level market in- Pawlowski, J. (2021). Monitoring newt communities in urban area
formation provides improved intelligence on the scope, scale and using eDNA metabarcoding. PeerJ, 9, 1–­15. https://doi.org/10.7717/
seasonality/trends of traded species. To our knowledge, this study peerj.12357
Chen, H., & Boutros, P. C. (2011). Venn diagram for five sets. BMC
demonstrates the first instance of an eDNA survey for use in mon-
Bioinformatics, 12(35), 297. https://doi.org/10.2307/2689606
itoring wildlife trade. Our data show that eDNA metabarcoding of Chue Hong, N. (2019). How to cite software: current best practice.
drainage water has the capacity to generate market-­wide audits of Zenodo, https://doi.org/10.5281/ZENODO.2842910
species presence which can prompt further visual inspection and Clusa, L., Garcia-­Vazquez, E., Fernández, S., Meyer, A., & Machado-­
Schiaffino, G. (2021). Nuisance species in lake constance revealed
enforcement of illegal trade restrictions. In this way, the regular ap-
through eDNA. Biological Invasions, 23(5), 1619–­1636. https://doi.
plication of eDNA monitoring has the potential to provide valuable org/10.1007/s1053​0 -­021-­02462​-­2
intelligence for monitoring seasonal market patterns and identifying Deiner, K., Fronhofer, E. A., Mächler, E., Walser, J. C., & Altermatt, F.
markets with repeated instances of protected fish trading. While not (2016). Environmental DNA reveals that rivers are conveyer belts
intended nor suitable as court evidence, eDNA metabarcoding has of biodiversity information. Nature Communications, 7, 1–­9. https://
doi.org/10.1038/ncomm​s12544
great potential as a surveillance tool, enabling enforcement officials
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RICHARDS et al. Methods in Ecology and Evolu on 1579

Deiner, K., Walser, J. C., Mächler, E., & Altermatt, F. (2015). Choice of Miya, M., Sato, Y., Fukunaga, T., Sado, T., Poulsen, J. Y., Sato, K.,
capture and extraction methods affect detection of freshwater Minamoto, T., Yamamoto, S., Yamanaka, H., Araki, H., Kondoh, M.,
biodiversity from environmental DNA. Biological Conservation, 183, & Iwasaki, W. (2015). MiFish, a set of universal PCR primers for
53–­63. https://doi.org/10.1016/j.biocon.2014.11.018 metabarcoding environmental DNA from fishes: Detection of more
Edgar, R. (2016). UNOISE2: Improved error-­correction for Illumina than 230 subtropical marine species. Royal Society Open Science,
16S and ITS amplicon sequencing. BioRxiv, 081257. https://doi. 2(7), 150088. https://doi.org/10.1098/rsos.150088
org/10.1101/081257 Musing, L., Shiraishi, H., Crook, V., Gollock, M., Levy, E., & Kecse-­nagy,
Edgar, R. C. (2010). Search and clustering orders of magnitude faster K. (2018). Implementation of the CITES Appendix II listing of European
than BLAST. Bioinformatics, 26(19), 2460–­2461. https://doi. Eel Anguilla anguilla 1, 82.
org/10.1093/bioin​forma​tics/btq461 Rees, H. C., Maddison, B. C., Middleditch, D. J., Patmore, J. R. M., &
Elbrecht, V., & Leese, F. (2015). Can DNA-­based ecosystem assess- Gough, K. C. (2014). REVIEW: The detection of aquatic animal spe-
ments quantify species abundance? Testing primer bias and cies using environmental DNA –­A review of eDNA as a survey tool
biomass-­sequence relationships with an innovative metabarcod- in ecology. Journal of Applied Ecology, 51(5), 1450–­1459. https://doi.
ing protocol. PLoS ONE, 10(7), 1–­16. https://doi.org/10.1371/journ​ org/10.1111/1365-­2664.12306
al.pone.0130324 Rhodes, K. (2018). Plectropomus areolatus (errata version published in
FAO. (2018). The state of world fisheries and aquaculture 2018 –­ 2021). The IUCN Red List of Threatened Species, 8235. Retrieved
Meeting the sustainable development goals. In Fao. Retrieved from from https://doi.org/10.2305/IUCN.UK.2018-­2.RLTS.T6441​1A192​
http://www.fao.org/state​-­of-­fishe​ries-­aquac​ulture 479585.en.
Ficetola, G. F., Pansu, J., Bonin, A., Coissac, E., Giguet-­Covex, C., De Richards, J. L., Sheng, V., Yi, C. W., Ying, C. L., Ting, N. S., Sadovy, Y., &
Barba, M., Gielly, L., Lopes, C. M., Boyer, F., Pompanon, F., Rayé, Baker, D. (2020). Prevalence of critically endangered European eel
G., & Taberlet, P. (2015). Replication levels, false presences and (Anguilla anguilla) in Hong Kong supermarkets. Science Advances,
the estimation of the presence/absence from eDNA metabarcod- 6(10), 2–­7. https://doi.org/10.1126/sciadv.aay0317
ing data. Molecular Ecology Resources, 15(3), 543–­556. https://doi. Sadovy de Mitcheson, Y. (2017). Live reef food fish wet market survey
org/10.1111/1755-­0998.12338 (September, pp. 1–­21). ADM Capital Foundation.
Ficetola, G. F., Miaud, C., Pompanon, F., & Taberlet, P. (2008). Species Sadovy de Mitcheson, Y., Tam, I., Muldoon, G., le Clue S., Botsford, E. &
detection using environmental DNA from water samples. Biology Shea, S. (2017). The trade in live reef food fish –­Going, going, gone.
Letters, 4(4), 423–­425. https://doi.org/10.1098/rsbl.2008.0118 Volume 1: Main report. Retrieved from https://www.hku.hk/f/uploa​
Fong, Q., & Zheng, Q. (2016). Synopsis of the Hong Kong seafood market. d/17452/​LRFFE​xSumm​aryEng.pdf
Synopsis of the Hong Kong seafood market. https://doi.org/10.4027/ Sadovy, Y., Kulbicki, M., Labrosse, P., Letourneur, Y., Lokani, P., &
shksm.2016 Donaldson, T. J. (2003). The humphead wrasse, Cheilinus undulatus:
Iwasaki, W., Fukunaga, T., Isagozawa, R., Yamada, K., Maeda, Y., Satoh, Synopsis of a threatened and poorly known Giant coral reef fish.
T. P., Sado, T., Mabuchi, K., Takeshima, H., Miya, M., & Nishida, M. Reviews in Fish Biology and Fisheries, 13(3), 327–­364. https://doi.
(2013). MitoFish and MitoAnnotator: A mitochondrial genome da- org/10.1023/B:RFBF.00000​33122.90679.97
tabase of fish with an accurate and automatic annotation pipeline. Sadovy, Y., & Cornish, A. S. (2000). Reef fishes of Hong Kong. Hong Kong
Molecular Biology and Evolution, 30(11), 2531–­2540. https://doi. University Press.
org/10.1093/molbe​v/mst141 Sato, Y., Miya, M., Fukunaga, T., Sado, T., & Iwasaki, W. (2018). MitoFish
Jamal, M. T., & Pugazhendi, A. (2021). Treatment of fish market waste- and MiFish pipeline: A mitochondrial genome database of fish
water and energy production using halophiles in air cathode mi- with an analysis pipeline for environmental DNA metabarcoding.
crobial fuel cell. Journal of Environmental Management, 292, 112752. Molecular Biology and Evolution, 35(6), 1553–­1555. https://doi.
https://doi.org/10.1016/J.JENVM​AN.2021.112752 org/10.1093/molbe​v/msy074
Karlsson, E., Johansson, A.-­M., Ahlinder, J., Lundkvist, M. J., Singh, N. Shao, K. T., Chang, J. S., Jeng, M. S., Tu, T. H., Ho, C. W., Chiu, Y. W.,
J., Brodin, T., Forsman, M., & Stenberg, P. (2020). Airborne micro- Chan, T. Y., Ho, P. H., Chao, S. M., Joung, S. J. & Lin, P. L. (2015). A
bial biodiversity and seasonality in northern and southern Sweden. guide book of common economic aquatic animals and plants in Taiwan
PeerJ, 8, e8424. https://doi.org/10.7717/peerj.8424 (pp. 1–­499). Fisheries Department, Agriculture Committee.
Kindt, R., & Coe, R. (2005). Tree diversity analysis. (January 2017). World Shaw, J. L. A., Clarke, L. J., Wedderburn, S. D., Barnes, T. C., Weyrich,
Agroforestry Centre (ICRAF). L. S., & Cooper, A. (2016). Comparison of environmental DNA me-
Lam, V. Y. Y., & Sadovy de Mitcheson, Y. (2011). The sharks of South East tabarcoding and conventional fish survey methods in a river system.
Asia -­ Unknown, unmonitored and unmanaged. Fish and Fisheries, Biological Conservation, 197, 131–­138. https://doi.org/10.1016/j.
12(1), 51–­74. https://doi.org/10.1111/j.1467-­2979.2010.00383.x biocon.2016.03.010
Lee, C., & Sadovy, Y. (1998). A Taste for live fish: Hong Kong's live reef Situ, Y., & Sadovy, Y. (2004). A preliminary study on local species diver-
fish market. Retrieved from http://pubs.iclarm.net/Naga/na_​ sity and seasonal composition in a Hong Kong wet market. Asian
2301.pdf Fisheries Science, 17(3), 235–­248.
Leray, M., & Knowlton, N. (2015). DNA barcoding and metabarcoding Sogin, M. L., Morrison, H. G., Huber, J. A., Welch, D. M., Huse, S. M., Neal,
of standardized samples reveal patterns of marine benthic diver- P. R., Arrieta, J. M., & Herndl, G. J. (2006). Microbial diversity in the
sity. Proceedings of the National Academy of Sciences of the United deep sea and the underexplored ‘rare biosphere’. Proceedings of the
States of America, 112(7), 2076–­2081. https://doi.org/10.1073/ National Academy of Sciences of the United States of America, 103(32),
PNAS.14249​97112 12115–­12120. https://doi.org/10.1073/PNAS.06051​27103
Letunic, I., & Bork, P. (2007). Interactive Tree Of Life (iTOL): An online Sumaila, U. R., & Bawumia, M. (2014). Fisheries, ecosystem justice and
tool for phylogenetic tree display and annotation. Bioinformatics, piracy: A case study of Somalia. Fisheries Research, 157, 154–­163.
23(1), 127–­128. https://doi.org/10.1093/bioin​forma​tics/btl529 https://doi.org/10.1016/j.fishr​es.2014.04.009
Liu, M., Chen, X., & Yang, S. Y. (2013). Marine fishes of southern Fujian, Thomsen, P. F., & Willerslev, E. (2015). Environmental DNA –­An emerg-
China (p. 62). Ocean Press. ing tool in conservation for monitoring past and present biodiver-
McMurdie, P. J., & Holmes, S. (2013). Phyloseq: An R package for re- sity. Biological Conservation, 183, 4–­18. https://doi.org/10.1016/J.
producible interactive analysis and graphics of microbiome cen- BIOCON.2014.11.019
sus data. PLoS ONE, 8(4), e61217. https://doi.org/10.1371/journ​ To, A. W. L., & Sadovy de Mitcheson, Y. (2009). Shrinking baseline: The
al.pone.0061217 growth in juvenile fisheries, with the Hong Kong grouper fishery
 |

2041210x, 2022, 7, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13842 by Nat Prov Indonesia, Wiley Online Library on [09/04/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1580 Methods in Ecology and Evolu on RICHARDS et al.

as a case study. Fish and Fisheries, 10(4), 396–­4 07. https://doi. marke​t-­Group​s-­Fail-­to-­Provi​de-­Suffi​cient​-­and-­Accur​ate-­Infor​matio​n-­​on-­
org/10.1111/j.1467-­2979.2009.00326.x Seafo​od-­Produ​c ts-­Consu​mers-­are-­being​-­misle​d-­and-­overc​harge​d-­​
Wineland, S. M., Arrick, R. F., Welch, S. M., Pauley, T. K., Mosher, J. J., for-­misla​belle​d-­seafood
Apodaca, J. J., Olszack, M., Holmes, J. N., & Waldron, J. L. (2019).
Environmental DNA improves eastern hellbender (Cryptobranchus
alleganiensis alleganiensis) detection over conventional sampling
How to cite this article: Richards, J. L., Sheng, V., Chung, H.
methods. Environmental DNA, 1(1), 86–­96. https://doi.org/10.1002/
edn3.9 W. Y., Liu, M., Tsang, R. H. H., McIlroy, S. E. & Baker, D.
WWF. (2016). Major Hong Kong Supermarket Groups Fail to Provide (2022). Development of an eDNA-­based survey method for
Sufficient and Accurate Information on Seafood Products Consumers urban fish markets. Methods in Ecology and Evolution, 13,
are being misled and overcharged for mislabelled seafood|WWF
1568–1580. https://doi.org/10.1111/2041-210X.13842
Hong Kong. Retrieved from https://www.wwf.org.hk/en/news/
press_relea​s e/?17260/​P ress​- ­R elea​s e-­M ajor​- ­H ong-­Kong-­S uper​