Specific identification using COI sequence analysis of scombrid larvae collected off the Kona coast of Hawaii Island

Abstract

Physical condition and morphological similarity prohibit unambiguous specific identification in many studies of scombrid larvae,
often resulting in several larvae that are unidentified or identified only to genus. Recent molecular techniques allow for
the unambiguous identification of early life history stages, even of those specimens that may be damaged. Molecular and morphological
techniques were used to determine the species composition of scombrid larvae taken in 43 tows in a putative spawning area
off the Kona Coast of Hawaii Island, 19–26 September 2004. Most of these tows were taken at night, at depths of 10 or 14m,
for 1h each at 2.5 knots. All 872 scombrid larvae collected were identified to species, 29% from unambiguous morphological
criteria and 71% using a molecular marker [cytochrome c oxidase I (COI) gene sequence]. Yellowfin tuna (Thunnus albacares) and skipjack tuna (Katsuwonus pelamis) dominated the larval composition almost equally, with frequencies of 48 and 45%, respectively. Five percent of the collection
was identified as albacore T. alalunga, a higher frequency than reported in previous studies of scombrid larval assemblages around the Hawaiian Islands. This COI
molecular marker enabled complete description of species diversity in the assemblage of scombrid larvae collected.

Full text

Page 1

FULL PAPER
Specific identification using COI sequence analysis of scombrid
larvae collected off the Kona coast of Hawaii Island
Melissa A. Paine Æ Æ Jan R. McDowell Æ Æ
John E. Graves
Received: 2 August 2006/Revised: 9 July 2007/Accepted: 14 July 2007/Published online: 26 December 2007
? The Ichthyological Society of Japan 2007
Abstract
prohibitunambiguousspecificidentificationinmanystudies
of scombrid larvae, often resulting in several larvae that are
unidentified or identified only to genus. Recent molecular
techniques allow for the unambiguous identificationof early
life history stages, even of those specimens that may be
damaged. Molecular and morphological techniques were
used to determine the species composition of scombrid lar-
vaetakenin43towsinaputativespawningareaofftheKona
Coast of Hawaii Island, 19–26 September 2004. Most of
these tows were taken at night, at depths of 10 or 14 m, for
1 heachat2.5knots.All872scombridlarvaecollectedwere
identifiedtospecies,29%fromunambiguousmorphological
criteria and 71% using a molecular marker [cytochrome c
oxidase I (COI) gene sequence]. Yellowfin tuna (Thunnus
albacares) and skipjack tuna (Katsuwonus pelamis) domi-
natedthe larval composition
frequencies of 48 and 45%, respectively. Five percent of the
collection was identified as albacore T. alalunga, a higher
frequency than reported in previous studies of scombrid
larval assemblages around the Hawaiian Islands. This COI
molecular marker enabled complete description of species
diversity in the assemblage of scombrid larvae collected.
Physical condition and morphological similarity
almostequally, with
Keywords
COI
Larvae ? Scombrid ? Tuna ? Identification ?
Introduction
Scombrid fishes (e.g., tunas, mackerels, bonitos) are
important worldwide for their economic and ecological
value. Bigeye tuna (Thunnus obesus), yellowfin tuna
(T. albacares), albacore (T. alalunga), and skipjack tuna
(Katsuwonus pelamis) are central components of pelagic
fisheries that operate in Hawaii’s exclusive economic zone
(Boggs and Ito 1993; Xi and Boggs 1996). Little is known
about the distribution, abundance, ecology, and behavior
of early life history stages of these species around Hawaii,
but it is this early period that is crucial to understanding
survival and recruitment to fishable stocks (Sund et al.
1981).
The composition of scombrid larvae in the central
Pacific, particularly around Hawaii, has been described as
being dominated by yellowfin tuna, skipjack tuna, and
frigate or bullet tunas, Auxis spp. (Strasburg 1960; Miller
1979; Boehlert and Mundy 1994). The other scombrid
larvae that could be encountered around Hawaii are alba-
core, bigeye tuna, wahoo (Acanthocybium solandri),
kawakawa (Euthynnus affinis), striped bonito (Sarda
orientalis), and spotted chub mackerel (Scomber austral-
asicus) (see Collette and Nauen 1983). Boehlert and
Mundy (1994) identified only a few albacore, bigeye tuna,
wahoo, and kawakawa larvae among hundreds of scombrid
larvae collected in their surveys around the island of Oahu.
Many of the Thunnus larvae they collected could not be
identified to species, and so it is uncertain if only a few
bigeye tuna and albacore were found because their larvae
were present in the fraction that could not be identified.
Although the incidence of mature adult albacore, wahoo,
and bigeye tuna would indicate that spawning could occur
in this area, there have been few confirmed collections of
the larvae of these species. A macroscopic examination of
This is contribution number 2887 of the Virginia Institute of Marine
Science.
M. A. Paine ? J. R. McDowell ? J. E. Graves (&)
Virginia Institute of Marine Science,
School of Marine Science, College of William and Mary,
Rte 1208 Greate Rd, Gloucester Point, VA 23062, USA
e-mail: graves@vims.edu
123
Ichthyol Res (2008) 55:7–16
DOI 10.1007/s10228-007-0003-4

Page 2

albacore gonads collected within 20 miles of the Hawaiian
Islands suggested that some spawning was occurring
during the summer in the vicinity of the islands (Otsu and
Uchida 1959). An analysis of bigeye tuna gonads suggests
that spawning occurs well offshore, to the southwest of the
island of Hawaii (Nikaido et al. 1991), but the presence of
larvae, especially those of smaller size, would be a more
direct means to show that spawning occurs in the region
(Prince et al. 2005).
Larval tuna are found in abundance near landmasses,
especially tropical and subtropical islands (Boehlert and
Mundy 1994). Gilmartin and Revelante (1974) hypothe-
sized that nutrient-rich waters near the Hawaiian Islands
create favorable conditions for spawning and larval
survival. Additionally, physical oceanographic features
such as eddies may act to retain larvae near the islands in
waters that are favorable for growth and survival
(Boehlert and Mundy 1993; Seki et al. 2002). Most
studies of nearshore abundance of scombrid larvae around
Hawaii have taken place around the island of Oahu
(Higgins 1970; Miller 1979; Boehlert and Mundy 1994),
and few studies of larval scombrids have been conducted
specifically off the Kona coast of the Hawaii Island. The
studies off Oahu showed a high concentration of scombrid
larvae close to land on the leeward side of the island.
These observations and the finding that the Kona coast
may be a ‘‘hot spot’’ for billfish spawning (Hyde et al.
2005) suggest that, this area may also be a favorable
spawning area for scombrids.
Proper specific identification is essential for early life
history studies. Although identification of scombrid
adults is unambiguous (Collette and Nauen 1983), spe-
cificidentification of early
problematic as many morphological characters are diffi-
cult to interpret. Scombrid eggs are very similar in
appearance and can only be separated by pigment char-
acters that become lost after preservation (Richards
2006). Larvae of the genus Thunnus are especially
challenging to identify. Specific identification requires
clearing and staining to determine the position of the first
closed hemal arch for vertebral precaudal/caudal count,
but yellowfin tuna and bigeye tuna can only be separated
by the presence or absence of certain pigment characters
(Richards 2006). It is also not possible to separate larvae
of yellowfin tuna from albacore before the appearance of
black pigment cells at the tip of the lower jaw in
yellowfin tuna at 4.5 mm standard length (Matsumoto
et al. 1971). Juvenile stages (15–60 mm SL) of Thunnus
species are difficult to identify because the development
of body pigmentation obscures diagnostic larval charac-
teristics, and meristic counts are broadly overlapping
(Nishikawa and Rimmer 1987).
lifehistorystages is
Previous work on scombrid larval distributions has been
limited by dependence upon morphological identification.
These types of analyses typically have many larvae that
cannot be identified to species level because they are too
small to have developed distinguishing morphological
characteristics or are too disfigured (Strasburg 1960; Leis
et al. 1991; Beckley and Leis 2000). Only half the 227
Thunnus larvae that Boehlert and Mundy (1994) collected
in their September surveys were large enough to be iden-
tified to species, and the remainder had to be classified as
Thunnus spp. Also, bullet and frigate tuna larvae generally
are not distinguished and are grouped together as Auxis
spp. (Higgins 1970; Boehlert and Mundy 1994). To fully
utilize information from all scombrid larvae, a more reli-
able method of identification is necessary to accurately
describe the early life history characteristics of these
species.
Molecular markers can provide a means for positive
identification when morphological identification is uncer-
tain or impossible (Morgan 1975; Graves et al. 1988;
Bartlett and Davidson 1991; McDowell and Graves 2002;
Hyde et al. 2005; Perez et al. 2005). Techniques such as
polymerase chain reaction-restriction fragment length
polymorphism (PCR-RFLP) analysis have been used to
identify species of the scombrid tribes Thunnini and
Sardini (Chow et al. 2003) as well as the species of the
genus Thunnus (Chow and Inoue 1993; Takeyama et al.
2001). In addition, sequencing of a mitochondrial gene
region has been used to identify Thunnus species (Bartlett
and Davidson 1991; Ram et al. 1996; Quintero et al. 1998;
Terol et al. 2002). Previously, a method for the identifi-
cation of all scombrids occurring in the western Atlantic
Ocean was developed that utilizes sequence information
from the mitochondrial cytochrome c oxidase I (COI) gene
region (Paine et al. 2007). The high level of conservation of
COI allowed the design of a unique primer pair that
successfully amplifies the same fragment across the diverse
members of the Scombridae. Although the conservation of
COI results in fewer differences between species than a
more variable gene region, the differences detected in that
study were fixed and therefore sufficient for species iden-
tification. That study demonstrated that the COI marker
worked well in other ocean basins to identify those
scombridswith circumtropical
morphological identification is limited because of a dam-
aged sample or morphological characters are ambiguous,
this molecular marker provides a means for unambiguous
identification. In the present study, the COI molecular
marker was used in concert with morphological identifi-
cation to describe species composition of scombrid larvae
taken off the leeward coast of Hawaii Island in September
2004.
distributions.When
8 M. A. Paine et al.
123

Page 3

Materials and methods
Sample
conducted off the Kona coast of Hawaii Island (Fig. 1)
aboard NOAA R/V Oscar Elton Sette, 19–26 September
2004, using a 1.8-m Isaacs-Kidd Trawl with 0.5-mm mesh.
A total of 43 tows was taken at 2.5 knots for 1 h each, with
an estimated average 12,580 m3of water filtered in each
tow. Of these, 31 tows were performed at night, 27 of
which were taken at a depth of 10 m and the other 4 at
14 m. The nine preliminary daytime collections were
stepped oblique tows at various two-step intervals of
14/10 m, 8/14 m, 30/20 m, and 20/10 m. One tow was
taken at 20 m and another was just below the surface
(\1 m) in the early morning. One scombrid larva (21 mm)
was taken from a midwater trawl (50 m) performed at dusk
on September 20. Total collections spanned a distance of
80 km along the leeward coast over 2 days and subse-
quently five nights. Tows were carried out in all directions
in an area from 1 to 16 km offshore. Bottom depth aver-
aged 2,000 m. Temperature of the top 50 m layer was at
least 27.5?C, and the wind speed during sampling was very
light, never exceeding ten knots.
collection. Ichthyoplanktonsamplingwas
Plankton samples were preserved in 95% ethanol, and
putative scombrid larvae were removed. Each sorted larva
was given a unique identifying number, stored in 95%
ethanol, and analyzed individually. Individual larvae were
photographed using a digital camera attached to a stereo-
scope via a phototube capturing as much detail as possible
for future morphological and meristic analysis. Total length
was measured using a ruler placed below the larva and
approximated to the nearest 0.5 mm.
Application of marker to identify species.
phological criteria of Nishikawa and Rimmer (1987) and
Richards (2006) were used to identify as many of the
scombrid larvae as possible. The unique morphological
The mor-
characters used to positively identify larvae from these
guides were forebrain pigment and ventral pigment spot in
skipjack tuna, lower jaw pigmentation in kawakawa, lateral
tail pigmentation in frigate tuna, and snout length in
wahoo. Well-preserved larvae of skipjack tuna, kawakawa,
wahoo, and, at some sizes, frigate and bullet tunas were
distinguished following the aforementioned identification
criteria. Larval Thunnus are generally problematic and
were distinguished following the cytochrome c oxidase I
(COI) sequence analysis method developed previously
(Paine et al. 2007). This molecular identification method
was also used for larvae that could not be identified
because of physical damage or questionable morphological
characters.
DNA was isolated, amplified, and sequenced following
the short-fragment COI sequence analysis method (Paine
et al. 2007). In cases in which there was no amplicon, the
reaction was repeated with both the COI primers and with
universal COI primers (Paine et al. 2007). If there was no
amplification with the universal primers, then it was
inferred that the sample was too degraded for analysis. If
amplification resulted from the universal primers but not
from the scombrid specific primers, then the sample was
inferred to be a non-scombrid. All sequences were edited
using Sequencher version 4.2.2 (Gene Codes, Ann Harbor,
MI, USA). The species identity was inferred by noting
where the sample sequence clustered in an unweighted
pair-group method with arithmetic averaging (UPGMA)
tree constructed of reference sequences using absolute
number of differences in the program MacVector version
7.2 (Accelyrs, San Diego, CA, USA) (Paine et al. 2007). In
cases where an unknown sample clustered between two
species, informative base positions in the unknown
sequence were compared to reference sequences in a
molecular key. The positions at which a species has a
consistent, unique combination of nucleotide base pairs are
indicated in the molecular key shown in Fig. 2.
Preliminary genetic identification of a few larvae
showed them grouping between the yellowfin tuna and
bigeye tuna reference COI sequences, which only differ by
two base position differences. Upon referencing the
molecular key, it appeared that only one of the bases was
consistently discriminatory between the two species. To
refine the key and confirm species identity, part of the
mitochondrial cytochrome b (cyt b) gene was sequenced
for four known yellowfin tuna and four known bigeye tuna
samples to provide another set of mtDNA reference
sequences. The primers used were cytbL686 (50TCC TTG
GTT TCG TGA TCC30) and cytbH982 (50GGG TTC AGA
ATA GGA ATT GG30). All PCR and sequencing of cyt b
was carried out in the same manner as for COI, with a 53?C
annealing temperature.
Fig. 1 Map of Hawaiian Islands with scombrid larval collection area
shown by black oval
Genetic identification of Hawaiian scombrid larvae9
123

Page 4

Because introgression of albacore mtDNA into Pacific
bluefin tuna T. orientalis has been reported previously
(Chow and Kishino 1995), the possibility exists that bluefin
tuna may be misidentified based on mtDNA characters
alone. Preliminary mtDNA analyses identified 43 larvae as
albacore and, to address the issue of potential misidentifi-
cationof bluefin tunaas
introgression, the nuclear internal transcribed spacer (ITS)-
1 region of these larvae was sequenced using the primers F-
ITS-1 (50GAG GAA GTA AAA GTC GTA ACA AGG30)
and 5.8SR2 (50GTG CGT TCG AAR KGT CGA TGA
albacore resultingfrom
TCA AT30) (K. Johnson, Virginia Institute of Marine
Science, unpublished data). This fragment was amplified
and sequenced as previously described for the COI frag-
ment, except that a 45? annealing temperature and 5 ll Q
solution (Qiagen, Valencia, CA, USA) were used in the 25-
ll reaction. Using the program Sequencher version 4.2.2 to
find restriction sites, the restriction enzyme EagI (New
England BioLabs, Ipswich, MA, USA) was found to dis-
tinguish between bluefin tuna and albacore based on
sequence information. Samples were subsequently distin-
guished in a 15-ll restriction digestion reaction using 0.5 ll
Fig. 2 Molecular key of
interspecific differences in the
cytochrome c oxidase I (COI)
fragment between consensus
sequences of each scombrid
species. The sites that are useful
in distinguishing very closely
related species (i.e., between
Thunnus albacares and Thunnus
obesus; Thunnus thynnus from
other Thunnus) have an asterisk.
Species abbreviations with
number of reference samples
represented in each consensus
sequence are ASOL,
Acanthocybium solandri (21);
ALBC, Thunnus alalunga (17);
BLFT, T. thynnus (18); BET, T.
obesus (18); YFT, T. albacares
(18); EUTH, Euthynnus
alletteratus (10); SKJT,
Katsuwonus pelamis (19);
AUXR, Auxis rochei (16);
AUXT, A. thazard (10). Note
that Euthynnus alletteratus
occurs in the western Atlantic
and is the congener of E. affinis
that occurs in Hawaii. Because
E. affinis is the only Euthynnus
that occurs around Hawaii,
except for the rare waifs of E.
lineatus from the eastern
Pacific, the EUTH reference
consensus sequence can be used
to identify E. affinis
10 M. A. Paine et al.
123

Page 5

EagI, 1.5 ll 109 NEB43 109 reaction buffer, 11 ll
deionized sterile water, and 2 ll PCR product, digested at
37?C for 1 h, then at 65?C for 20 min.
Results
Forty-three ichthyoplankton tows yielded a total of 872
scombrid larvae. Scombrids were found in all collections
except two of the nine daytime tows. The daytime tows
averaged 2.3 ± 2.2 scombrids per tow and those taken at
night averaged 24.3 ± 21.2. Morphological characters
were used to positively identify 29% of the larvae, and the
remaining 71% were identified using the COI marker. All
DNA isolations of scombrid larvae were amplified
successfully on the first attempt, with the exception of two
specimens that amplified on the second effort. All these
isolations were amplified with the scombrid COI primers,
except four non-scombrids that amplified with the uni-
versal COI primers.
Four of the scombrid larvae grouped between the
reference bigeye tuna and yellowfin tuna sequences; two of
these individuals (OES18-36, OES22-2) are shown in
Fig. 3. Sequence information obtained from the mito-
chondrial cyt b gene demonstrated that three of these were
yellowfin tuna and one was a bigeye tuna.
Preliminary analyses identified 43 albacore larvae using
the COI marker. To exclude the possibility that some of
these putative albacore larvae were not bluefin with intro-
gressed ‘‘albacore-like’’ mtDNA, a portion of the nuclear
ITS-1 region was sequenced from these larvae and from
reference samples of bluefin tuna and albacore. The alba-
core larvae identified using the COI marker had ITS
sequences more similar to the known albacore samples and
consequently were inferred to be albacore. The nucleotide
base differences in the ITS-1 region for the reference
samples of these two species are shown in Fig. 4.
To provide a quick molecular diagnostic tool to dis-
criminate albacore and bluefin tuna in future studies, a
restriction fragment length polymorphism was found in the
nuclear ITS-1 region that distinguishes the two species. The
restriction enzyme EagI does not digest the *800 bp PCR
product in bluefin tuna but does create 350 and 450 bp
bands for albacore (Fig. 5). The northern bluefin tuna (six)
and albacore (nine) ITS-1 sequences generated in the study
by Chow et al. (2006) were also found to have the same
EagI restriction site pattern found in this study, and further
there were no intraspecific differences at this cut site.
Sample OES23-73 clustered nearest to EUTH (Euthyn-
nusalletteratus),theonlyEuthynnusincludedintheoriginal
key developed in our previous study of western North
Atlantic scombrids. We infer that this specimen is Euthyn-
nus affinis as this species is the only congener that is thought
to have a persistent population off Hawaii. Euthynnus
lineatus has been recorded from the Hawaiian Islands twice,
but in both cases the single specimens were thought to be
vagrantsfromthe easternPacific(Matsumoto 1976;Collette
and Nauen 1983). Because E. alletteratus is the only species
in this study without a circumtropical distribution, the
reference sequence information of E. alletteratus allows for
identification of E. affinis individuals.
The species composition of scombrids collected in all
tows is represented in Fig. 6. Yellowfin and skipjack tuna
larvae dominated the collections at frequencies of 48 and
45%, respectively. Albacore were found in 20 tows and
comprised 5% of the scombrid larvae. Information on all
collections as well as the range of lengths and mean lengths
(±SD) of the larvae collected of each species are given in
Table 1. Among the commonly encountered larvae, the
lengths of skipjack tuna were smaller on average than
albacore or yellowfin tuna (P\0.05). The length–
frequency distribution of the most common species
collected (yellowfin tuna, skipjack tuna, and albacore) at all
stations is presented in Fig. 7.
Discussion
Previous studies of scombrid larvae assemblages have been
limited by reliance on morphological identification and
typically had a fraction of the collection that could not be
assigned to a species. The combination of morphological
identification and the COI molecular marker used in this
study allowed for unambiguous species identification of all
scombrid larvae collected off the Kona coast of Hawaii
Island. For example, previous larval studies were not able
to distinguish past the generic level in Auxis, but using
sequence information from COI allowed for the distinction
between frigate tuna and bullet tuna. Also, previous studies
have had a Thunnus spp. group or have made tentative
specific identifications based on unreliable characteristics,
such as one or two very small and inconspicuous ventral
pigment spots that separate intact bigeye tuna larvae from
yellowfin tuna. In the present study, bigeye tuna and
yellowfin larvae were confidently identified using this
molecular marker. Additionally, previous studies could not
discriminate albacore from yellowfin tuna larvae below
4.5 mm SL. In this study, 46% of the Thunnus larvae
collected fell into this size range, and all were successfully
identified using the COI marker. The use of the COI marker
allowed for a complete description of species diversity of
the scombrid larval assemblage collected off the Kona
coast, which is in contrast to the surveys taken around
Oahu by Boehlert and Mundy (1994) in which almost half
of the more than 200 Thunnus larvae they collected in
September could not be identified to species.
Genetic identification of Hawaiian scombrid larvae11
123