Identification of Shark Species Composition and
Proportion in the Hong Kong Shark Fin Market Based
on Molecular Genetics and Trade Records
SHELLEY C. CLARKE,∗ JENNIFER E. MAGNUSSEN,† DEBRA L. ABERCROMBIE,†
MURDOCH K. MCALLISTER,‡ AND MAHMOOD S. SHIVJI†∗∗
Joint Institute for Marine and Atmospheric Research, c/o National Research Institute of Far Seas Fisheries, 5-7-1 Orido-Shimizu,
Shizuoka 474-0022, Japan
†Guy Harvey Research Institute, Nova Southeastern University Oceanographic Center, 8000 North Ocean Drive, Dania Beach,
FL 33004, U.S.A.
‡Renewable Resources Assessment Group, Imperial College London, Royal School of Mines Building, Prince Consort Road,
London SW7 2AZ, United Kingdom
Abstract: The burgeoning and largely unregulated trade in shark fins represents one of the most serious
threats to shark populations worldwide. In Hong Kong, the world’s largest shark fin market, fins are classified
by traders into Chinese-name categories on the basis of market value, but the relationship between market
category and shark species is unclear, preventing identification of species that are the most heavily traded.
To delineate these relationships, we designed a sampling strategy for collecting statistically sufficient numbers
of fins from traders and categories under conditions of limited market access because of heightened trader
sensitivities. Based on information from traders and morphological inspection, we hypothesized matches
between market names and shark taxa for fins within 11 common trade categories. These hypotheses were tested
using DNA-based species identification techniques to determine the concordance between market category and
species. Only 14 species made up approximately 40% of the auctioned fin weight. The proportion of samples
confirming the hypothesized match, or concordance, varied from 0.64 to 1 across the market categories. We
incorporated the concordance information and available market auction records for these categories into
stochastic models to estimate the contribution of each taxon by weight to the fin trade. Auctioned fin weight
was dominated by the blue shark ( Prionace glauca), which was 17% of the overall market. Other taxa, including
the shortfin mako ( Isurus oxyrinchus), silky (Carcharhinus falciformis), sandbar (C. obscurus), bull (C. leucas),
hammerhead (Sphyrna spp.), and thresher (Alopias spp.), were at least 2–6% of the trade. Our approach to
marketplace monitoring of wildlife products is particularly applicable to situations in which quantitative data
at the source of resource extraction are sparse and large-scale genetic testing is limited by budgetary or other
market access constraints.
Key Words: DNA forensics, marketplace monitoring, marketplace sampling, seafood, wildlife
o o o o o
Identificaci´ n de la Composici´ n y Proporci´ n de Especies de Tibur´ n en el Mercado de Aletas de Tibur´ n en
Hong Kong Con Base en Gen´tica Molecular y Registros Comerciales
Resumen: El floreciente y en gran parte no regulado comercio de aletas de tibur´ n representa una de las
m´ s serias amenazas para las poblaciones de tiburones en todo el mundo. En Hong Kong, el mayor mercado
de aletas de tibur´ n en el mundo, los comerciantes clasifican a las aletas en categor´as con nombres en Chino
o ı o
con base en el valor de mercado, pero la relaci´ n entre la categor´a de mercado y las especies de tibur´ n no
es clara, lo que limita la identificaci´ n de especies que son sujetas de mayor comercio. Para delinear estas
correspondence to M. Shivji, email firstname.lastname@example.org
Paper submitted September 4, 2004; revised manuscript April 12, 2005.
Conservation Biology Volume 20, No. 1, 201–211
C 2006 Society for Conservation Biology
202 Species in Shark Fin Trade Clarke et al.
n u ı
relaciones, dise˜ amos una estrategia para recolectar con comerciantes un n´ mero estad´sticamente suficiente
de aletas bajo condiciones de mercado de acceso limitado debido a la intensificaci´ n de las sensibilidades de
o o o o
los comerciantes. Con base en informaci´ n de los comerciantes y en inspecci´ n morfol´ gica, hicimos hip´ tesis
sobre la correspondencia entre los nombres comerciales y los taxa de tiburones para aletas dentro de 11 cate-
ı o e o
gor´as comerciales comunes. Probamos las hip´ tesis utilizando t´cnicas de identificaci´ n de especies con base
en ADN para determinar la concordancia entre la categor´a de mercado y la especie. Solo 14 especies abar-
caron aproximadamente 40% del peso de las aletas a la venta. La proporci´ n de muestras que confirmaron la
o o ı
hip´ tesis de correspondencia, o concordancia, vari´ de 0.64 a 1 en las categor´as de mercado. Incorporamos
la informaci´ n de concordancia y los registros de venta disponibles en modelos estoc´ sticos para estimar la
o u o
contribuci´ n, seg´ n el peso, de cada tax´ n al comercio de aletas. El peso de aletas a la venta fue dominado
por Prionace glauca, que comprendi´ 17% del total del mercado. Otros taxa, incluyendo a Isurus oxyrinchus,
Carcharhinus falciformis, C. obscurus, C. leucas, Sphyrna spp. y Alopias spp., comprendieron por lo menos entre
2 y 6% del comercio. Nuestro m´todo para monitoreo del mercado de productos de vida silvestre es partic-
ularmente aplicable a situaciones de escasez de datos cuantitativos obtenidos en el sitio de extracci´ n del
a e a
recurso y cuando el an´ lisis gen´tico a gran escala est´ limitado por restricciones presupuestarias o de acceso
Palabras Clave: ADN forense, mariscos, monitor, muestreo, vida silvestre
Introduction fisheries (Camhi 1999) have contributed to growing con-
cerns that the fin trade may be driving shark catches to
Advances in molecular biology over the past decade have unsustainable levels. Policy responses to these concerns,
created a variety of powerful conservation tools, includ- including the Food and Agriculture Organization’s Inter-
ing the ability to identify the species or population origin national Plan of Action for Sharks (Food and Agriculture
of wildlife products, enabling new types of trade monito- Organization 1998) and the listing of three shark species
ring and analysis. The often secretive nature of the wildlife in the appendices of the Convention on International
trade and the prohibitively high cost of purchasing sam- Trade in Endangered Species of Wild Fauna and Flora
ples, however, may constrain sample access and acqui- (CITES), have been hampered by a lack of species-specific
sition and thus limit application of genetic technology. catch and trade data for most fisheries (Clarke 2004a).
Under these circumstances, the appropriate choice of Shark resource management based on an amalgam of
study objectives and the sampling design become criti- species is suboptimal because it is likely to exaggerate the
cal aspects of any research program aimed at surveying conservation concern for some species (e.g., sharks with
markets to monitor and quantify wildlife trade. high fecundities and low age of maturity) while poten-
Many researchers have applied genetic techniques to tially understating the vulnerability of less prolific shark
determine whether particular species, populations, or in- e
stocks (Smith et al. 1998; Cort´s 2002).
dividuals are present in wildlife trade (e.g., Malik et al. Hong Kong is the world’s largest shark fin market, rep-
1997; Birstein et al. 1998; Hoelzel 2001; Dalebout et al. resenting at least 50% of the global trade (Fong & Ander-
2002). These researchers, however, were concerned pri- son 2002; Clarke 2004b). Previous studies of the Hong
marily with demonstrating that the proposed technique Kong fin trade describe fin types with primarily English
is effective for forensic identification of individual mar- common names for sharks (Parry-Jones 1996; Vannuccini
ketplace products and did not attempt to characterize 1999; Fong & Anderson 2000) and rely solely on traders’
the composition and proportion of species in the trade knowledge of these English names, with no means of inde-
as a whole. Marketplace sampling and characterization pendently verifying the actual species origin. Traders use
of wildlife products are particularly valuable when the 30–45 market categories of fins ( Yeung et al. 2000), but
sources of extraction (e.g., landing ports for fish or hunt- the Chinese names of these categories do not correspond
ing camps for terrestrial mammals) are diffuse and lack to the Chinese taxonomic names of shark species (Huang
standardized, species-specific records. In such cases mon- 1994). Instead Chinese market categories for shark fins
itoring of species parts in trade, particularly in large ware- appear to be organized primarily by the quality of fin rays
houses at the center of global supply chains, may be the produced and secondarily by distinguishing features of
most effective means of obtaining data necessary for off- dried fins because often traders have never observed the
take assessment and management. whole animal.
Increasing awareness of the vulnerability of shark Our recent analysis of Hong Kong shark fin auction
species to exploitation (Camhi et al. 1998; Castro et al. data produced minimum estimates of quantities (metric
1999) and a documented proliferation of finning (i.e., re- tonnes [mt]) of shark fins traded in 11 market categories.
moval of fins and discarding of the carcass at sea) in some We used statistical models to account for missing records
Volume 20, No. 1, February 2006
Clarke et al. Species in Shark Fin Trade 203
and simulate a complete dataset (Clarke et al. 2004). The lieved to be common in the fin trade, to ensure species
11 studied market categories comprised 46% of the auc- specificity (i.e., absence of false positives).
tioned shark fins, with the remainder auctioned under Sequences and diagnostic validation data for the
other trade names or in unspecified categories. Over- species-specific primers used in this study are avail-
all, 18% of the auctioned fins were labeled as ya jian, able from the following sources: shortfin mako (Isu-
which on the basis of trader information and morpholog- rus oxyrinchus), longfin mako (I. paucus), silky (Car-
ical examination were hypothesized to originate from the charhinus falciformis), dusky (C. obscurus), and blue
blue shark (Prionace glauca). Several other categories sharks in Shivji et al. (2002); common (Alopias vulpi-
hypothesized to derive from other coastal and pelagic nus), bigeye (A. supercilious), and pelagic thresher (A.
sharks comprised 3–5% each of the estimated auctioned pelagicus) sharks in Abercrombie (2004); tiger (Galeo-
fin quantity (Clarke et al. 2004). Although this study pro- cerdo cuvier), bull (C. leucas), and spinner (C. brevip-
vides the first quantitative information on trade composi- inna) sharks in Nielsen (2004); and scalloped (Sphyrna
tion by market category, its usefulness is limited because lewini), smooth (S. zygaena), and great hammerhead (S.
Chinese market categories could not be related conclu- mokarran) sharks in Abercrombie et al. (2005). The sand-
sively to shark taxa and thus extraction rates for individual bar shark (C. plumbeus) primer sequence is in Pank et
species could not be assessed. al. (2001; additional validation data from M. Henning &
Here we investigated, using molecular genetic identi- M.S.S., unpublished data).
fication, concordances between the most common Chi- All primers are highly reliable in their diagnostic ac-
nese market categories and species in the Hong Kong curacy (above references), with the following caveats
shark fin market. These concordance results were de- for two primers. The dusky primer cross-amplifies two
signed to be modeled in conjunction with estimates of of its congeners, Galapagos shark (C. galapagensis) and
the traded weight of shark fins in each market category oceanic whitetip shark (C. longimanus). Because fins of
(Clarke et al. 2004) to better understand the composition the oceanic whitetip shark are unmistakably distinct in
and contribution by species to the fin trade. Another of appearance (Castro 1993), however, we used the dusky
our objectives was to devise effective market sampling primer to indirectly confirm morphological identification
strategies for animal parts under conditions fraught with of these fins. The uncertainty introduced when using the
sampling constraints and to guide other applications of dusky primer is thus limited to the inability to distinguish
molecular diagnostic methods to wildlife trade monitor- fins from dusky and Galapagos sharks only. The bull shark
ing faced with similar access limitations. primer faintly cross-amplifies its congener, the Caribbean
reef shark (C. perezi). Because cross-amplification of the
Caribbean reef shark by the bull shark primer is ineffi-
cient, however, consistently producing only a very faint
amplicon, discrimination between these two species has
not been problematic (Nielsen 2004). Nevertheless, a sec-
ondary testing of all fins identified as originating from
The recent development of diagnostic DNA sequence bull shark with an extensively validated Caribbean reef
tests with species-specific PCR primers for sharks be- shark primer (M. Henning & M.S.S., unpublished data)
lieved to be common in the shark fin trade has provided confirmed identification of bull shark fins.
a new tool for studying this trade. Detailed methodolo-
gies for these diagnostic tests are provided in Pank et
al. (2001), Shivji et al. (2002), Chapman et al. (2003),
Sampling Constraints in the Hong Kong Shark Fin Market
Abercrombie (2004), Nielsen (2004), and Abercrombie
et al. (2005), and are overviewed briefly here. The tech- Approximately 50 retail or wholesale establishments in
nique relies on DNA sequence differences among shark the dried seafood business district of Hong Kong (Sai Yin
species in the nuclear ribosomal DNA ITS2 locus for the Pun) deal in shark fin products. Of this number, about
development of species-specific primers. Each primer is 16 wholesalers take turns hosting daily shark fin auctions
a short, synthetic, single-stranded piece of DNA designed and at least an equal number import large quantities of
to recognize (anneal to) and PCR amplify only DNA from fins and then export them to mainland China for process-
a single shark species (e.g., a primer designed for short- ing without auctioning. Traders receive fins from at least
fin mako shark will amplify only DNA from fins of that 85 countries and territories (Clarke & Mosqueira 2002)
species), producing a species-specific DNA band (ampli- in poorly sorted shipments; because of on-site space lim-
con). Each primer we used here was tested extensively itations they immediately re-sort the fins into market cat-
against its target species from worldwide sources to en- egories and promptly auction or sell them. Our sampling
sure its global diagnostic utility (i.e., absence of false nega- opportunities were limited to these re-sorting periods be-
tives) and approximately 72 nontarget species, including cause traders refused us access to the fins at other times.
very closely related species (congeners) and those be- Limits on sampling during re-sorting periods arose from
Volume 20, No. 1, February 2006
204 Species in Shark Fin Trade Clarke et al.
the patchy nature of the shipments (i.e., during each sam- Sampling Program Design
pling episode fins tended to be clustered in a few market
categories from a particular source country) and the need We selected 11 market categories for study because they
to minimize sampling time to avoid delaying re-sorting op- were common in the trade and because we had diagnos-
erations. Furthermore, random sampling of the complete tic PCR primers for the species we hypothesized to be
range of fins on hand was not feasible because of the present in each category. Concordances between Chi-
logistics of moving large sacks in cramped warehouses. nese market names and taxa (Table 1) were hypothe-
Given these restrictions, and based on the results of initial sized based on information from traders and were com-
enquiries at more than 30 trading houses, instead of ran- pared with published information on morphologies of
dom sampling we tested samples of particular shark fin fins and whole sharks (Fisheries Agency of Japan 1999;
market categories across as wide a range of traders and Vannuccini 1999; Yeung et al. 2000; Froese & Pauly 2002).
source countries as possible. We avoided market categories representing mixed fin
Table 1. A priori hypothesized matches (x→x ) between traders’ market categories and shark taxa for fins targeted in the sampling.
Hypothesized samples required
Market predominant shark Assigned
category taxa within market Taxonomic a priori CV p = CV p =
(x)∗ category (x ) uncertainty value of p 0.1 0.05
Ya jian blue (Prionace visually distinct; low probability of 0.95 6 22
glauca) mixing with other species
Qing lian shortfin mako some traders mentioned infrequent 0.80 25 100
(Isurus mixing with the less abundant
oxyrinchus) longfin mako (I. paucus)
Wu yang silky (Carcharhinus observed presence of silvertip shark 0.70 43 172
falciformis) fins (C. albimarginatus); potential
for mixing with other carcharhinids
especially Galapagos shark (C.
Hai hu dusky (C. obscurus) visually similar to bai qing and ruan 0.90 12 45
sha but high value of these fins
causes traders to sort them
Bai qing sandbar (C. visually similar to hai hu and ruan sha 0.90 12 45
plumbeus) but high value of these fins causes
traders to sort them accurately
Ruan sha tiger (Galeocerdo visually similar to hai hu and bai qing 0.90 12 45
cuvier) but high value of these fins causes
traders to sort them accurately
Chun chi smooth and traders state this category contains all 0.60 67 267
scalloped hammerheads (Sphyrna and
hammerhead Eusphyra spp., 7 species in total)
(Sphyrna other than great (S. mokarran);
zygaena and S. hammerhead species not listed as
lewini) predominant taxa are believed to be
considerably less abundant
Gu pian great hammerhead high value suggests sorting is accurate 0.85 18 71
(S. mokarran) but category may inadvertently
contain other light-colored
hammerheads (Sphyrna spp.)
Wu gu thresher (Alopias some traders mentioned infrequent 0.90 12 45
spp., 3 species) mixing with longfin mako (I.
Sha qing bull (C. leucas) some traders mentioned infrequent 0.80 25 100
mixing with another unknown shark,
possibly pigeye (C. amboinensis)
Liu qiu oceanic whitetip (C. visually distinct; low probability of 0.95 6 22
longimanus) mixing with other species
∗ Trade names are romanized under the Pinyin (Putonghua) system. Chinese characters (complex form) are given to facilitate translation
between Pinyin and other forms of Chinese romanization.
Volume 20, No. 1, February 2006
Clarke et al. Species in Shark Fin Trade 205
lots such as those simply sorted by size because of low Individual warehouses were sampled between Novem-
value. ber 2000 and February 2002. At each establishment,
Given the highly sensitive nature of the fin trade and the traders were shown a list of desired fin types (in Chinese
accompanying logistical constraints to market sampling, characters) and asked to allow free-of-charge sampling of
and to reduce the uncertainty in downstream estimation small pieces of tissue from the edges of dried fins. Sam-
of the proportion of fins in the trade by individual species, ples were clipped with clean metal pincers or scissors,
we developed an explicit sampling strategy. By determin- and laboratory genetic analyses avoided using cut edges
ing the sampling effort required to estimate concordances when extracting DNA in case of cross-sample contamina-
within a given confidence interval for each market cate- tion introduced at the point of cutting. In the majority
gory, we minimized sampling bias toward the most eas- of visits, not all fin types could be sampled because of
ily sampled categories (i.e., most common or least valu- limited stocks. In addition, some traders were reluctant
able products). Simultaneously, we prioritized sampling to allow sampling of high-value fins for fear of product
of those categories hypothesized to contain a mixture damage. Many traders did not understand the rationale
of look-alike species’ fins over those expected to con- for replicate sampling within a category and this led to
tain distinct products from a single species. Determining sample sizes of n = 1 during many visits. We used the
sample sizes necessary per category in the absence of ex- information-theoretic evenness measure of Brillouin for
isting information requires setting an a priori probability, nonrandom samples (Zar 1999) to assess the diversity of
p, that the hypothesized match between market category samples collected in each market category by trader and
and taxonomic identity would hold for a given tested fin. geographic source region. Higher values for this statistic
In setting the a priori p we assumed that market cate- indicate that the samples were spread more evenly among
gories containing morphologically distinctive fins most the various trader or source regions.
likely derived from a single species and would thus have
high p values (i.e., near 1). We correspondingly reduced
the p values for other categories according to the de-
gree of taxonomic uncertainty as determined from trader
information and preliminary market reconnaissance
Implementation of the Sampling Design
The required number of samples, n, was determined We obtained records of 29% of all auctions held during an
based on the binomial model (sensu Zar 1999), with the 18-month period (October 1999–March 2001), including
formula numerous records from all major trading houses. Details
of the auction records and methodology for using existing
q records to simulate a complete set of records for the 18-
(CV p )2 p month period are described in Clarke et al. (2004).
A total of 596 fin samples from the 11 target market cat-
where p is the proportion correctly identified, q is the egories was collected. The number of collected samples
proportion incorrectly identified, and CV p is the coeffi- exceeded the target number for CVp = 0.10 for each mar-
cient of variation in p, as given by CVp = sp /p, where s p ket category, and for three categories ( ya jian, wu gu, and
is the standard deviation of p. Using this formula, and the liu qiu) the number of collected samples also exceeded
a priori p values, the requisite number of samples for two the target number for CVp = 0.05 (Tables 1 & 2).
CV levels for p (CVp = 0.05 and 0.10) was calculated for Our ability to assess (1) the fidelity of nomenclature
each market category (Table 1). across the trading community and (2) the traders’ ability
The sampling program aimed to estimate, based on ge- to consistently classify fins imported from different parts
netic analysis, the a posteriori probability p that a fin iden- of the world into coherent categories was determined
tified by a trader as market category x is taxon x with a by the distribution of samples among different trading
low CV for p (CVp < 0.10). The 11 market categories we houses and source ocean basins or regions. The distri-
analyzed genetically were modeled earlier with Markov bution by trader was skewed such that 372 of the 596
chain Monte Carlo methods to determine the percentage samples (62%) were obtained from just two traders. This
by weight of auctioned fins in each category on an annual is a reflection of the sharp decline in trade-community co-
basis (Clarke et al. 2004). We extended this model based operation following a shark conservation campaign con-
on the results of the concordance testing such that in each ducted in Hong Kong in March 2001. Because many of the
model iteration the stochastically determined weight of traders were unable to be precise about the geographic
fins in each market category was multiplied by a stochas- origin of the fins, we used general categories to assess
tic estimate of p, generated from data on the number of the distribution of samples across ocean basins or areas.
samples tested and the number of samples verifying the The two heavily sampled traders specialized in imports
hypothesized match for each category. This allowed us to from South America, and this is reflected in the fact that
determine the proportion of fins in the trade by species. 264 of 596 samples (44%) were said to have originated in
Volume 20, No. 1, February 2006
206 Species in Shark Fin Trade Clarke et al.
Table 2. Shark fin samples collected and their distribution within each market category by trader and geographic source (ocean basin/region)
Source ocean basins
Trader’s Traders Evenness or regions represented Evenness
market Hypothesized Samples represented for traders (of 8 total, excluding for sources
category species match collected (n) (of 21 total) (Brillouin J)a unidentified)b (Brillouin J)a
Ya jian P. glauca 37 11 0.81 5 0.77
Qing lian I. oxyrinchus 69 11 0.68 6 0.82
Wu yang C. falciformis 110 13 0.71 5 0.58
Hai hu C. obscurus 34 7 0.52 4 0.43
Bai qing C. plumbeus 40 9 0.64 3 0.69
Ruan sha G. cuvier 26 6 0.63 3 0.48
Chun chi S. zygaena or S. lewini 94 13 0.68 7 0.60
Gu pian S. mokarran 35 8 0.70 4 0.29
Wu gu Alopias spp. 75 9 0.69 6 0.81
Sha qing C. leucas 53 10 0.74 5 0.64
Liu qiu C. longimanus 23 9 0.76 3 0.76
a Theevenness statistics (Brillouin J) (Zar 1999) represent only the equity of the distribution of samples among traders or sources that were
sampled and are not a function of the total possible number of sampled traders or sources.
b Geographic sources of fins cited by the traders providing samples were grouped into eight categories as follows: South America (Brazil,
Ecuador); Eastern Atlantic/West Africa (Spain, Togo); Southeast Asia (Australia, Indonesia, Philippines); Indian Ocean (India, Sri Lanka,
Maldives, Bangladesh); Southern Africa (South Africa, Angola); Central America (Gulf of Mexico, Costa Rica); South Pacific (Fiji); Middle East
(may include fins from unspecified African countries).
South America. Nevertheless, one of the heavily sampled a posteriori values of p, the proportion of samples vali-
traders also received shipments of fins from other loca- dating the hypothesized match, ranged from 0.64 for sha
tions, and these other locations contributed 50% of the qing to 1.00 for liu qiu (Table 3). Calculation of a poste-
samples from this trader. riori estimates of the coefficient of variation of p (CV p )
Evenness of sampling diversity, based on the Brillouin suggested that for the majority of market categories the
J statistic, across traders ranged from 0.52 to 0.81 (Ta- sample size was sufficient to provide a robust estimate
ble 2). The lowest value resulted for the category hai hu of the a posteriori value of p because all values of CV p
(n = 34), for which 56% of the samples were collected were <0.11 (Table 3). The results from further testing of
from one trader, with another 29% from a second trader. samples that did not verify the hypothesized match are
The remaining samples were distributed among five other given in Table 4.
traders. Evenness of sampling diversity across geographic The results for some market categories, which were eas-
source regions ranged from 0.29 to 0.82 (Table 2), with ily distinguished from most other types of fins because
the lowest value for category gu pian (n = 35), for which of size, shape, or color, conformed closely to hypothe-
89% of the samples were collected from one source re- sized identities. For example, none of the liu qiu fins
gion, with the remaining samples distributed among three was from any shark other than oceanic whitetip, and the
other regions or an unidentified source. only ya jian that was not a blue shark fin was from a
bigeye thresher shark, which is similar in fin dimensions.
Gu pian fins, recognizable because of their light color
Species Identification of Samples
and large size, mainly derived from great hammerheads
All samples (n = 596) were initially tested with the as expected, and four of the five gu pian fins deriving
species-specific primer representing the hypothesized from other species were from the morphologically simi-
taxonomic match (Table 1). Complete sample amplifica- lar, though smaller, scalloped hammerhead.
tion failure (i.e., failure of the positive control amplicon Some market categories validated the hypothesized
to appear [Pank et al. 2001; Shivji et al. 2002]) occurred species matches at higher probabilities (i.e., were less
in <4% of the samples. If each amplifiable sample failed diverse) than expected. We expected that a posteriori val-
to amplify with its hypothesized species primer (a q fin), ues of p for chun chi, smooth or scalloped hammerhead
it was consecutively tested in a multiplex PCR format (de- (n = 94), concordance would be relatively low because
tailed in Shivji et al. ) against primers for each of of the potential presence of other hammerhead species
the 16 species we addressed (i.e., all the species listed (e.g., S. tudes, S. corona, S. media, and Eusphyra blochii)
in Table 1 plus longfin mako and spinner) until a match for which primers have not yet been developed, in the
was found (a q i sample) or until all the primers had been chun chi category. However, except for two samples iden-
tried without achieving identification (a q u sample). The tified as deriving from the great hammerhead, one sample
Volume 20, No. 1, February 2006
Clarke et al. Species in Shark Fin Trade 207
Table 3. Results of genetic analysis of shark fin samples by market category.
A posteriori estimateb
Trader’s Hypothesized genetic Samples Samples with as matching the
market category identification tested control failuresa hypothesized species p CV p sp
Ya jian P. glauca 37 1 35 0.97 0.028 0.027
Qing lian I. oxyrinchus 69 2 57 0.85 0.051 0.044
Wu yang C. falciformis 110 2 86 0.80 0.049 0.039
Hai hu C. obscurus 34 0 29 0.85 0.071 0.061
Bai qing C. plumbeus 40 6 25 0.74 0.103 0.076
Ruan sha G. cuvier 26 1 21 0.84 0.087 0.073
Chun chi S. zygaena or S. lewini 94 1 89 0.96 0.022 0.021
Gu pian S. mokarran 35 0 30 0.86 0.069 0.059
Wu gu Alopias spp. 75 7 50 0.74 0.073 0.054
Sha qing C. leucas 53 3 32 0.64 0.106 0.068
Liu qiu C. longimanus 23 0 23 1.00 0 0
Total 596 23 477
with control failures are those for which no positive control amplification could be obtained after several trials.
proportion of samples validating the hypothesized match is the a posteriori p value (equal to the number of matching samples per
number of samples tested minus control failures).
identified as pelagic thresher, and one sample that did not two sampling events from two different traders is less
amplify with any primer, only scalloped and smooth ham- easily explained.
merhead fins were present. Most of the remaining trade categories contained a
The other category that was less species-diverse than higher diversity of species than expected. The mixing
expected was wu yang (n = 110). Amplification with the of longfin mako fins with threshers in the wu gu cate-
silky shark species-specific primer was observed in 80% gory (22% of fins) and with shortfin makos in the qing
of the successfully tested samples. Most of the remain- lian category (9% of fins) was extensive. Of the q i sam-
ing amplifiable fins were identified as scalloped hammer- ples in the wu gu and qing lian categories, 88% and 75%,
heads (n = 6) or shortfin makos (n = 4). Only one each of respectively, were longfin mako fins. All the high-value
three other Carcharhinus congeners were present (Table carcharhinid fins (i.e., hai hu, bai qing, and ruan sha)
4), suggesting that these are most likely present in only contained other carcharhinid species and several uniden-
small numbers in the wu yang category. The presence tifiable fins that did not amplify with any existing primer.
of several shortfin makos in this category may be due to The market category with the lowest a posteriori value of
sampling error because some of the traders call shortfin p was sha qing, which was originally expected to show
makos wu yang ( , i.e., same romanization but differ- a strong concordance with the bull shark. Of the 50 sha
ent Chinese characters). Identification of six scalloped qing samples successfully tested, 17 could not be iden-
hammerheads in the wu yang samples collected during tified and may have derived from the morphologically
Table 4. Summary of shark fin samples not confirming the hypothesized match in each market category.
Market category q u samplesa q i samplesb Identity of q i samplesc
Ya jian 0 1 bigeye thresher (1)
Qing lian 2 8 longfin mako (6), bigeye thresher (1),
scalloped hammerhead (1)
Wu yang 9 13 scalloped hammerhead (6), shortfin mako (4),
sandbar (1), dusky or Galapagos (1), spinner (1)
Hai hu 4 1 silky (1)
Bai qing 8 1 dusky or Galapagos (1)
Ruan sha 3 1 bull (1)
Chun chi 1 3 great hammerhead (2), pelagic thresher (1)
Gu pian 1 4 scalloped hammerhead (4)
Wu gu 1 17 longfin mako (15), shortfin mako (1), blue (1)
Sha qing 17 1 tiger (1)
Liu qiu 0 0 not available
Totals 46 50
a Samples that did not amplify with any species-specific primer.
b Samples that were found to amplify with another species-specific primer.
c True species identity and number of samples corresponding to each species for q samples.
Volume 20, No. 1, February 2006
208 Species in Shark Fin Trade Clarke et al.
similar pigeye shark (C. amboinensis) for which no pri- in the sampling design whenever possible. For example,
mer is yet available. we acknowledge that traders’ market categories are based
on fin value, which may have little connection with scien-
Market Composition by Trade Category tific taxonomy. This issue was addressed for the categories
of interest through an a priori formulation of p (hypoth-
Hong Kong auction records typically contained 19 major
esized market category-taxon concordance), which was
market categories describing auction lots. These 19 cate-
used to calculate the necessary number of samples.
gories in conjunction with alternative names for the same
Our ability to obtain the requisite number and distri-
fin types, names for rare fin types, and combinations of
bution of samples across traders and geographic source
names applied to mixed lots resulted in nearly 100 total
regions was a key factor. Access constraints directed sam-
market categories. The stochastic model of auctioned fin
pling toward testing concordances between market cat-
quantities, extended to incorporate genetic information,
egories and shark taxa, rather than random sampling of
provided taxa-specific estimates of the composition, by
the trade as a whole. Target sample sizes were achieved
weight, of the shark fin trade and confirmed that blue
(Table 3) by following the concordance testing frame-
sharks comprised by far the largest distinct proportion
work developed, and reasonable sample distribution was
(17.3% by weight) of fins auctioned in the Hong Kong
obtained including 6–13 of the 21 traders and three to
market (Table 5).
seven of the eight geographic source regions, depending
on market category. Overall, given the covert nature of
the trade and the sensitivities generated by ongoing con-
Discussion servation campaigns, the implementation of the sampling
program was highly successful.
Evaluation of the Sampling Program
Our ability to draw sound conclusions regarding the Trader Sorting and Labeling
species composition of the shark fin trade based on our
results is a function of the robustness of the sampling de- A large volume of fins (54% by weight) traded in unstud-
sign, the extent of category-specific sorting and labeling ied, and often nonspecific, categories (Clarke et al. 2004)
by Hong Kong traders, and the species-diversity scope could not be characterized. This shortcoming could not
of the diagnostic primer library. Each factor is discussed be overcome but was accounted for by limiting the study
individually and then integrated in the context of each to common and relatively distinct types of fins. These un-
studied market category in the following sections. studied categories may contain additional fins of studied
taxa, particularly when fin size is small, and the low value
Sampling Design does not warrant careful sorting; therefore, taxa-specific
estimates from this study should be considered as mini-
Studies involving sampling of traded wildlife products can mum proportions in trade.
gain an even greater understanding of the trade if mer- Another difficulty presented by the trade labeling sys-
chant records can be accessed and understood. Shortcom- tem was the existence of cases where several taxa are con-
ings of such records should then be explicitly addressed tained in one market category (e.g., wu gu category used
Table 5. Means and 95% probability intervals for estimated proportion of the Hong Kong shark fin trade composed of each taxa.a
Proportion of trade by market category b Proportions of trade by taxa c
Market category mean lower 95% P.I. upper 95% P.I. Tested taxon/taxa mean lower 95% P.I. upper 95% P.I.
Ya jian 18.2 16.6 20.0 P. glauca 17.3 15.5 19.1
Qing lian 3.2 2.8 3.6 I. oxyrinchus 2.7 2.3 3.1
Wu yang 4.4 4.0 4.9 C. falciformis 3.5 3.1 4.0
Hai hu 1.7 1.5 2.0 C. obscurus 1.4 1.2 1.7
Bai qing 3.3 2.8 3.8 C. plumbeus 2.4 1.9 2.8
Ruan sha 0.16 0.10 0.23 G. cuvier 0.13 0.08 0.19
Chun chi 4.7 4.2 5.2 S. zygaena or S. lewini 4.4 3.9 4.9
Gu pian 1.8 1.5 2.1 S. mokarran 1.5 1.2 1.8
Wu gu 3.2 2.8 3.6 Alopias spp. 2.3 2.0 2.7
Sha qing 3.5 3.0 4.0 C. leucas 2.2 1.8 2.6
Liu qiu 1.9 1.6 2.1 C. longimanus 1.8 1.6 2.1
a The possible presence of additional fins of these taxa within the 54% of the trade that could not be characterized implies that the estimates
below are minimums.
b Proportions by market category (and probability intervals, P.I.) are taken from model results presented in Clarke et al. (2004).
c Proportions by taxon/taxa were derived for this study by extending the model to adjust the market category for taxonomy (see text).
Volume 20, No. 1, February 2006
Clarke et al. Species in Shark Fin Trade 209
for all three thresher species) and cases where many mar- a 100% concordance with their hypothesized market
ket categories may be used for one taxon (e.g., the longfin match. Given the ease of morphological identification of
mako found within several categories). Those cases that these fins by traders, the best estimate of oceanic whitetip
were detected early in the study were addressed through sharks’ contribution to the trade (1.8% [1.6–2.1%]) is
careful definition of target categories and sampling re- likely more accurate than other species because these
quests to traders (e.g., clarifying whether the fin was fins are less likely to be inadvertently sorted into other
sometimes called by another name). categories.
The three hammerhead species (scalloped, smooth,
Scope of the Primer Library and great) collectively formed a significant proportion of
the trade (approximately 5.9%). The occurrence of some
Although diagnostic primers for each shark species in the
scalloped hammerhead fins in the gu pian (great hammer-
shark fin trade (i.e., at least 50 species) are desirable, this
head) category most likely resulted from the fact that both
is a large and ongoing task. In the interim, especially in
scalloped and great hammerhead fins are similarly light
light of considerable concerns about the status of several
in color. Inadvertent introduction of large, light-colored
populations, shark resource management requires cur-
scalloped hammerhead fins into the gu pian category and
rent information on the fin trade to assess questions of
small, light-colored great hammerhead fins into the chun
sustainable use, and it is therefore necessary to pursue
chi category (Table 4) thus affects the traded weight es-
trade studies despite the current taxonomically incom-
timates for the various species of hammerheads in both
plete primer library.
The least reliable results were those for the dusky shark,
Estimates of Species Proportions in Trade
which was estimated to contribute approximately 1.4%
Combining results of the genetic analyses with the (1.2–1.7%) of the fins in the Hong Kong market. Although
stochastic modeling of trade records allowed the contri- already low, this figure most likely overestimates this
bution of individual species to the trade to be determined. species’ proportion in trade because the dusky primer
Shortfin mako fins showed an 85% concordance with the used does not distinguish between dusky and Galapagos
market category qing lian, which equates to a trade pro- sharks.
portion for that species of approximately 2.7% (probabil- Our results indicate that between 34% and 45% (95%
ity interval 2.3–3.1%). The actual percentage is likely to probability interval), and possibly more ( because of pres-
be somewhat higher, given the presence of shortfin mako ence in unstudied categories), of the Hong Kong shark fin
fins in the wu yang (silky shark) trade category. auction trade is composed of only 14 species (i.e., nine
The wu gu category contained mostly fins from the mostly single-species trade categories plus scalloped and
three thresher species (74%), resulting in these species smooth hammerheads and the three species of thresh-
collectively accounting for approximately 2.3% (2.0– ers). This finding may reflect the relative abundance of
2.7%) of the shark-fin trade. The actual proportion in trade these species in global fisheries, preferential demand for
is also likely to be slightly higher given the discovery of their fins in this market, or a combination of these and
thresher fins, albeit at low frequency, in other categories other factors. Our results indicate that blue sharks form a
(e.g., ya jian, qing lian, and chun chi) (Table 4). The particularly large component of the market, a finding con-
longfin mako presented a key nomenclatural issue that sistent with and potentially resulting from the fact that
affected both the wu gu (thresher) and qing lian (short- blue sharks form a large proportion of the shark bycatch
fin mako) categories. Discussions with traders suggested in pelagic longline fisheries targeting tuna and swordfish
that a minority of traders sort longfin mako fins into a (Nakano & Seki 2003). Although this species is one of
separate category (qing hua); most either combine them e
the most productive shark species (Cort´s 2002), further
with shortfin mako or thresher fins due to similarity of analyses are needed to determine whether the current
appearance and market value. demand exerted by the shark fin trade on this species is
Tiger sharks, assumed by traders and authors of pre- sustainable. Other species comprise approximately 0.1 to
vious studies to be hai hu (Chinese for ocean tiger), 4.4% each of the total trade volume, and, given the range
were instead highly concordant with the ruan sha cate- e
in biological productivity of these species (Cort´s 2002),
gory, which comprised only 0.08 to 0.19% of the trade. it is expected that some will be better able to withstand
Commonly traded carcharhinid sharks included the blue the growing demands of the fin trade than others.
(17.3% [15.5%–19.1%]), silky (3.5% [3.1–4.0%]), sandbar
(2.4% [1.9–2.8%]), and bull (2.2% [1.8–2.6%]) sharks. The
Present and Future Shark Fin Trade Characterization
presence of fins from these carcharhinid species in other
market categories was noted (Table 4) and would tend to Available information does not provide a definitive an-
increase these estimates. swer to the question of whether auction records are rep-
Oceanic whitetip sharks, although circuitously identi- resentative of the entire global trade. However, the Hong
fied in the genetic testing, are highly distinct and showed Kong market serves both Hong Kong and mainland China
Volume 20, No. 1, February 2006
210 Species in Shark Fin Trade Clarke et al.
(Clarke 2004a), and thus includes both high-end and S. J. Joung and Y. Y. Liao, generously contributed their ex-
budget-conscious consumers with a wide spectrum of pertise to the study. We thank E. J. Milner-Gulland, M. V.
tastes. Furthermore, Hong Kong auctions represent 20% Ashley, and two anonymous reviewers for comments on
of all Hong Kong imports (Clarke et al. 2004), and Hong previous versions of this manuscript.
Kong is thought to handle more than 50% of the global
shark fin trade (Clarke 2004b). Therefore, although fur- Literature Cited
ther studies are necessary to investigate other compo-
nents of the global shark fin market, our conclusions are Abercrombie, D. 2004. Efficient PCR-based identification of shark prod-
ucts in global trade: applications for the management and conserva-
based on what appears to be the largest trading channel tion of commercially important mackerel sharks (Family Lamnidae),
in the world. thresher sharks (Family Alopiidae) and hammerhead sharks (Fam-
Development of diagnostic primers for the CITES ily Sphyrnidae). M.S. thesis. Nova Southeastern University Oceano-
Appendix II-listed basking (Cetorhinus maximus) and graphic Center, Dania Beach, Florida.
whale (Rhincodon typus) sharks, as has been done for Abercrombie, D. L., S. C. Clarke, and M. S. Shivji. 2005. Global-scale
genetic identification of hammerhead sharks: application to assess-
great white sharks (Carcharodon carcharias) (Chapman ment of the international fin trade and law enforcement. Conserva-
et al. 2003), will allow products from these species to tion Genetics: in press.
be identified quickly, reliably, and at low cost. A larger Birstein, V. J., P. Doukakis, B. Sorkin, and R. DeSalle. 1998. Population
challenge for rare species in trade will arise from design- aggregation analysis of three caviar-producing species of sturgeons
ing a sampling program. In addition to the usual issues and implications for the species identification of black caviar. Con-
servation Biology 12:766–775.
associated with sampling coverage and statistical power, Camhi, M. 1999. Sharks on the line II: an analysis of Pacific state shark
awareness of conservation concerns and regulatory pro- fisheries. Living oceans program. National Audubon Society, Islip,
tections may lead traders to camouflage certain types New York.
of fins through cryptic labeling, as has been observed a
Camhi, M., S. Fowler, J. Musick, A. Br¨utigam, and S. Fordham. 1998.
for basking (S.C., personal observation) and great white Sharks and their relatives: ecology and conservation. Occasional pa-
per 20. IUCN Species Survival Commission, Gland, Switzerland.
(Shivji et al. 2005) shark fins. Even without deliberate sub- Castro, J. I. 1993. A field guide to the sharks commonly caught in com-
terfuge, rare fins lacking distinctive characters and market mercial fisheries of the southeastern United States. Technical mem-
values may be mixed within voluminous, undifferentiated orandum NMFS-SEFSC-338. National Oceanic and Atmospheric Ad-
market stocks and thus become nearly impossible to de- ministration, Miami.
tect. A more practical approach to expanding the list of Castro, J. I., C. M. Woodley, and R. L. Brudek. 1999. A preliminary eval-
uation of the status of shark species. Fisheries technical paper 380.
species we examined is to identify additional distinctive FAO, Rome.
market categories commonly used by traders and then Chapman, D., D. Abercrombie, C. Douady, E. Pikitch, M. Stanhope,
develop primers that can identify some of the species and M. Shivji. 2003. A streamlined, bi-organelle, multiplex PCR ap-
expected to be found in these categories. Molecular ge- proach to species identification: application to global conservation
netic testing of putative products derived from controlled and trade monitoring of the great white shark, Carcharodon car-
charias. Conservation Genetics 4:415–425.
species such as CITES-listed sharks is likely to remain lim- Clarke, S. 2004a. Shark product trade in mainland China and Hong Kong
ited to specific enforcement cases (Clarke 2004a). and implementation of the CITES shark listings. TRAFFIC East Asia,
Addressing the critical questions of shark resource sus- Hong Kong.
tainability will require further work to link species com- Clarke, S. 2004b. Understanding pressures on fishery resources through
position and proportion findings with estimates of the trade statistics: a pilot study of four products in the Chinese dried
seafood market. Fish and Fisheries 5:53–74.
catch weight or number of sharks represented by fins Clarke, S., and I. Mosqueira. 2002. A preliminary assessment of Euro-
in the market. Trade-based estimates of weight or num- pean participation in the shark fin trade. Pages 65–72 in M.Vacchi,
bers by taxon can then provide supporting information e
G. La Mesa, F. Serena, and B. S´ret, editors. Proceedings of the
to evaluate whether overexploitation is occurring and to ee
4th European elasmobranch association meeting. Soci´t´ Fran¸aise c
elucidate the role of the shark fin trade in driving shark d’Ichtyologie, Paris.
Clarke, S., M. McAllister, and C. Michielsens. 2004. Estimates of shark
catches and disposition. species composition and numbers associated with the shark fin trade
based on Hong Kong auction data. Journal of Northwest Atlantic
Fisheries Science 35:453–465.
Cort´s, E. 2002. Incorporating uncertainty into demographic modeling:
application to shark populations and their conservation. Conserva-
Acknowledgments tion Biology 16:1048–1062.
Dalebout, M. L., G. M. Lento, F. Cipriano, N. Funahashi, and C. S. Baker.
We are grateful to Imperial College London, the Wildlife 2002. How many protected minke whales are sold in Japan and Ko-
Conservation Society, the David and Lucile Packard Foun- rea? A census by microsatellite DNA profiling. Animal Conservation
dation, the Curtis and Edith Munson Foundation, the Ep- 5:143–152.
Fisheries Agency of Japan. 1999. Characterization of morphology of
pley Foundation, the Hai Stiftung Foundation, the Florida
shark fin products: a guide of the identification of shark fin caught
Sea Grant Program (R/LR-B-54), and the Japan Society for by tuna longline fishery. Global Guardian Trust, Tokyo.
Promotion of Science for their support of this research. Fong, Q. S. W., and J. L. Anderson. 2000. Assessment of the Hong Kong
Staff of the National Taiwan Ocean University, including shark fin trade. INFOFISH International 1/2000:28–32.
Volume 20, No. 1, February 2006
Clarke et al. Species in Shark Fin Trade 211
Fong, Q. S. W., and J. L. Anderson. 2002. International shark fin mar- logically similar sharks Carcharhinus obscurus and Carcharhinus
kets and shark management: an integrated market preference-cohort plumbeus (Carcharhinidae) using multiplex PCR. Marine Biotech-
analysis of the blacktip shark (Carcharhinus limbatus). Ecological nology 3:231–240.
Economics 40:117–130. Parry-Jones, R. 1996. TRAFFIC report on shark fisheries and trade in
Food and Agriculture Organization (FAO). 1998. International plan of Hong Kong. Pages 87–143 in The world trade in sharks: a com-
action for the conservation and management of sharks. Document pendium of TRAFFIC’s regional studies. TRAFFIC International, Cam-
FI:CSS/98/3. FAO, Rome. bridge, United Kingdom.
Froese, R., and D. Pauly. editors. 2002. FishBase database. Fishbase, Kiel, Shivji, M. S., D. D. Chapman, E. K. Pikitch, and P. W. Raymond. 2005.
Germany. Available from www.fishbase.org (accessed August 2004). Genetic profiling reveals illegal international trade in fins of the great
Hoelzel, A. R. 2001. Shark fishing in fin soup. Conservation Genetics white shark, Carcharodon carcharias. Conservation Genetics: in
Huang, Z. G. 1994. Zhongguo haiyang shengwu zhonglei xiefenbu Shivji, M., S. Clarke, M. Pank, L. Natanson, N. Kohler, and M. Stan-
(China marine organism categorization and ordering). China Ocean hope. 2002. Rapid molecular genetic identification of body-parts
Press, Beijing (in Chinese). from six pelagic shark species for conservation, management, and
Malik, S., P. J. Wilson, R. J. Smith, D. M. Lavigne, and B. N. White. 1997. trade-monitoring. Conservation Biology 16:1036–1047.
Pinniped penises in trade: a molecular-genetic investigation. Con- Smith, S. E., D. W. Au, and C. Snow. 1998. Intrinsic rebound potentials
servation Biology 11:1365–1374. of 26 species of Pacific sharks. Marine and Freshwater Research
Nakano, H., and M. P. Seki. 2003. Synopsis of biological data on the 49:663–678.
blue shark, Prionace glauca Linnaeus. Bulletin Fisheries Research Vannuccini, S. 1999. Shark utilization, marketing and trade. Fisheries
Agency (Japan) 6:18–55. Technical Paper 389. Food and Agriculture Organization, Rome.
Nielsen J. T. 2004. Molecular genetic approaches to species identifica- Yeung, W. S., C. C. Lam, and P. Y. Zhao. 2000. The complete book of
tion and delineation in elasmobranchs. M.S. thesis. Nova Southeast- dried seafood and foodstuffs. Wan Li Book Company Limited, Hong
ern University Oceanographic Center, Dania Beach, Florida. Kong (in Chinese).
Pank, M., M. Stanhope, L. Natanson, N. Kohler, and M. Shivji. 2001. Rapid Zar, J. H. 1999. Biostatistical analysis. 4th edition. Prentice-Hall Interna-
and simultaneous identification of body parts from the morpho- tional, Upper Saddle River, New Jersey.
Volume 20, No. 1, February 2006