Fisheries science JSFS , tập 76, số 4, 2010 7

O R I G I N A L A R T I C L E FisheriesEffect of trawling with traditional and ‘T90’ trawl codends on fish size and on different quality parameters of cod Gadus morhua and haddock Melano

O R I G I N A L A R T I C L E Fisheries

Effect of trawling with traditional and ‘T90’ trawl codends on fish

size and on different quality parameters of cod Gadus morhua

and haddock Melanogrammus aeglefinus

Hanne Digre•Ulrik Jes Hansen•Ulf Erikson

Received: 22 July 2009 / Accepted: 21 April 2010 / Published online: 12 June 2010

Ó The Japanese Society of Fisheries Science 2010

Abstract The effect of trawling on fish size and on

dif-ferent quality parameters of cod (Gadus morhua) and

haddock (Melanogrammus aeglefinus) was evaluated after

conducting 16 valid hauls using two trawls in a double rig

fitted with a traditional and a novel ‘T90’ codend,

respectively The total catch volume during the fishing

period was 47.6 metric tons, with an average catch per

codend of 1.5 (range 0.5–2.9) tons The mean haul duration

was 5 h The catch was assessed according to fish size,

mortality, external damage, initial white muscle pH and

development of rigor mortis Fillet quality (colour, blood

spots, gaping) was assessed after 1 week of freeze-storage

Our results showed there was no difference between the

two types of nets in terms of catch volume, but

signifi-cantly slightly bigger fish were caught with T90 than with

the traditional trawl net (p \ 0.05) Haddock caught with

the traditional trawl net had more external injuries related

to the trawl gear than haddock caught with the T90 gear

(p \ 0.05) The gaping frequency for cod caught with the

traditional trawl net tended to be higher than cod caught

with the T90 gear, but the difference was not significant

(p = 0.07) No other differences in fish quality between

fish caught in the trawl nets were observed

Keywords Cod
Fish quality
Fish size
Haddock

Trawling
T90 codend

IntroductionSeafood products have often suffered from an inherent loss

of quality caused by the fishing gear, retrieval from thewater, and handling on deck Due to the nature of theindustry, it is rather difficult to industrialise the production

to provide for more gentle fish handling routines In tion, seafood is often stored onboard for a comparativelylong storage time until it can be landed and delivered to thefish processors or consumers In 2003, approximately8,000 metric tons (4% of the total catch volume) of Atlanticcod landed in Norway had suffered serious physical dam-age, with the result that 1,900 metric tons were down-gra-ded, losing economic value (F Gregersen, unpublisheddata 2005) Increasing the utilisation of each catch while alsoconcurrently raising the quality of the catch will result inhigher prices for the products and contribute to more sus-tainable fisheries in the present situation, whereas focus iscurrently being directed towards a better exploitation of theresources New technologies which improve quality mayfurther contribute to moving the industry towards quality,rather than quantity only

addi-Only a few attempts have been made to alter the trawlgear to improve the quality of the catch The focus has been

on the codend, with particular attention being directedtowards reducing the turbulence that is seen in full-scaleusage Recent investigations have demonstrated that adrastic reduction in the movements of the codend can beachieved just by turning the direction of the netting 90°(T90) in relation to how it is normally used in traditionaltrawl assembly (U J Hansen, unpublished data 2004;Fig.1) This effect is due to changes in the configuration ofthe knots in the netting—the knots are further apart withthe 90° shift in direction than in netting stretched in thenormal direction In effect, a T90 codend has a much larger

H Digre ( &)
U J Hansen
U Erikson

SINTEF Fisheries and Aquaculture, 7465 Trondheim, Norway

e-mail: Hanne.Digre@Sintef.no

H Digre

Department of Biotechnology, The Norwegian University

of Science and Technology, 7491 Trondheim, Norway

DOI 10.1007/s12562-010-0254-2

cross-sectional area, and it is practically free of all

move-ment Theoretically, these features increase the possibility

improving the catch quality

A typical processing line onboard a present-day trawler

comprises the following unit operations: (1) the trawl net is

pulled up a steep trawl slip, then over another upper edge

before the catch is gathered into the trawl deck; (2) the nets

are emptied in a tank; (3) the gills are cut, with subsequent

bleeding, gutting, deheading and packing Although not

widely studied, it has nevertheless been shown that

catching methods and subsequent onboard handling may

affect fish quality [1 8] The fish are often exhausted,

injured or killed as a result of inadequate catching methods,

transfer from sea-to-vessel methods, or on-deck handling

routines

The hauling process can damage a substantial part of the

catch because of the pressure on the codend when it is being

hauled towards the boat In many cases, the cause of death is

anoxia where fish are left in the open air on the deck The

catching process leads to elevated levels of plasma cortisol

[9,10], glucose [11], lactate, [11–14] haematocrit, Na?, K?,

and Cl-, whereas blood pH decreases [11] The more

strenuous the stress-related activity the fish are subjected to,

the more rapid the muscle ATP depletion [7,10] and severe

muscle activity causes early rigor mortis onset [10, 14]

Cole et al [7] showed that in terms of low initial flesh pH

(7.1), high blood lactate and depleted ATP stores,

hand-lining of blue cod Parapercis colias was the most fatiguing

capture method when compared with commercial potting,modified potting and rested harvested fish Chopin et al.[15] showed that stress during capture may vary betweendifferent gear types Auclair [2] evaluated the effects of thegillnet and trawl in the cod fishery and found that gillnettedfish were of a lower quality than trawl-caught fish Thisauthor also found that the bacterial contamination increasedwith increased fishing times, namely, 4, 12, 24 and 48 h,with significant differences after 24 h for both fishingmethods The use of the gillnet also caused flesh discolor-ation due to bleeding, and fish were lost to predators andparasites Chopin et al [15] showed that the severity anddegree of injuries increased with the time of entrapment fortrammel net-caught sea bream The quality of Atlantic codcaught by otter trawl was investigated by Botta and Bonnell[5], who concluded that the initial quality of the cod wasusually very good and that the reduction of the quality was aresult of catch volumes being too large ([5 tons during asingle tow), delayed bleeding ([1 h), storage method andtime ([6 days)

Adequate bleeding is considered to be necessary for agood product quality Botta et al [16] studied differentbleeding/gutting procedures on the sensory quality of freshraw Atlantic cod caught by a trawler (2–3 h tow lengths;catch amounts 2.3–13.6 tons) and showed that time beforegutting ([1 h) was more important than the bleeding/gut-ting methods Similar results were obtained by Kelly [17]and Valdimarsson et al [3] Wagner [18] evaluated theexternal appearance and consistency of cod caught usingtrawls and found that the quality was reduced withincreased hauling time and number of fish in the trawl net.Botta et al [4] compared the effect of season and catchingmethod (gillnet, handline, longline and trap) on the quality

of cod and showed that the catching method had an impact

on fillet colour, discoloration, bruising and overall qualitygrades of cod They concluded that the catching methodshad a greater impact than season on the quality of freshcod Furthermore, muscle pH is lower and the conditionfactor is generally higher in fish caught by gillnet comparedwith fish caught by longline [19] Hattula et al [6] studiedthe effect of gillnetting, poundnetting and trawling on themortality and quality of herring Mortality increased whenthe trawling time increased from 2 to 5 h Rigor mortisstarted earlier, and the nucleotide decomposition proceededfurther in gillnet-caught fish than in fish caught by the othermethods, indicating a loss of freshness due to stress in thecatching process O¨ zyurt et al [8] showed that quality andshelf life of pike perch (Sander lucioperca) were affected

by catching methods, with the acceptable shelf life being

7 days longer for pike perch caught by longline and poon than gillnetted fish

har-The objectives of this study were to compare a T90codend and a traditional trawl net in terms of fish injuries,

Fig 1 Trawl netting stretched in the direction of normal use (top)

and turned 90° (bottom)

fish size, handling stress and fillet quality and to obtain

more knowledge about how trawling impacts cod and

haddock quality

Materials and methods

Fishing gear

Cod and haddock were captured using a traditional and the

novel ‘T90’ codends by a typical North Atlantic factory

trawler (M/T J Bergvoll, length 57 m, BT 1499, HP 3900)

in November 2004 The vessel was selected due to its

ability to operate two trawls simultaneously This was ideal

for the aim of the study since the simultaneous use of the

two trawls would exclude a number of variables and make

the assessments more directly comparable

The trawls used were identical traditional cod trawls, but

fitted with different codends They were of the type Alfredo

no 5 (Refa-Frøystad Group, Tromsø, Norway) that is

standardly used by several trawlers in the North Atlantic

The codend on one side was a standard codend and on the

other side, it was of the new (T90) design (Figs.1,2) Both

trawls were fitted with the sorting grid which is specified

by the Norwegian authorities in the Technical Regulations

for the whitefish fishery in the Barents Sea (SDBS website:

http://www.lovdata.no/for/sf/fi/xi-20000310-0271.html)

The sorting grids (flexigrids) were mounted as a whole unitbetween the belly of the trawl and the extension piece infront of the codend The standard codend had an extensionpiece and codend made from standard diamond meshmaterials, while the T90 codend differed by having a largepart made from netting turned 90° (Fig.3) Normally, ajoining ratio of 2:3 is recommended to join the turnedmeshes from the T90 material to the standard diamondmesh Because the aforementioned regulation specifies theuse of selection grids in cod trawls and also specifies thenumber of meshes in the circumference, the joining ratio inthis case was a compromise—104 meshes were joined to

80 in the case of the T90 The extension piece and codendwere made from 6- and 8-mm double-braided polyamide,respectively The rear-most 4.8 m of the T90 codend wasmade from knotless netting in an attempt to create the bestconditions for the preservation of fish quality Knotlessnetting has a much smoother inside surface than double-braided knotted netting It should be mentioned that theT90 concept does not apply to knotless netting Both trawlswere fitted with the usual top and bottom side chafers Thenets were closed by weaving the top and bottom panelmeshes together with a cod line, which results in a longtransversal knot that has been seen in model tests to furtherreduce the movements of the catch The trawls were bothfitted with acoustic monitoring instruments to measure thedistance between the doors and the filling rate of thecodends

Fish captureSeventeen hauls were conducted during the period 20–24November 2004 in the Nordkapp Bank in the Barents Sea(71°N/24–27°E) The bank is located north of the coast ofFinnmark in northern Norway A trawl was badly damaged

in one haul, and that haul was not included in furtheranalyses The total catch for the 16 valid hauls was47.6 metric tons, with an average catch of 2.98 tons (range1.06–4.85 tons) per haul (= 2 codends) During the trials,effort was made to standardise the hauling conditions, butfactors such as the bottom conditions or the weather con-ditions occasionally prevented us from fulfilling this cri-terion The mean haul duration was 5 h (range 2.5–6.0 h),and the towing speed was kept close to 4.0 knots Thefishing was conducted at depths between 238 and 370 m.Data for each haul, including catch amount, are given inTable1

Processing lineThe processing line onboard consisted of the followingoperations: (1) the trawl net with the fish were hauled up ondeck; (2) the nets were emptied in a tank without water;

Fig 2 Two models of codends made from normal netting (top) and

T90 (bottom) demonstrate the difference due to the two

configura-tions Photographs were taken in the flume tanks of SINTEF Fisheries

and Aquaculture in Hirtshals, Denmark

(2) the throats of the fish were cut (within 2 h) and the fish

were placed in a tank without water prior to being gutted,

deheaded, frozen in blocks and stored in a freezing room

at -21°C Total processing time from the time the catch

was brought onboard until the fish were packed before

freezing was typically around 2.5 h

Fish sampling

Immediately after the catch, cod and haddock were

col-lected from the codends while still on deck Due to the tight

schedule, it was not always possible to make all

assess-ments of the fish from every haul A selection of the

assessments of the fish was conducted from a random

selection of the hauls Fish were selected at random, and

the number of mortalities, visual assessment of external

damage, and fish size were determined Mortality rateswere assessed for both species from Haul 8 and 11 (cod,

n = 83; haddock, n = 84) Visual assessments of externaldamage were done on cod (n = 520) and haddock(n = 481) from eight hauls (Hauls 1, 3, 4, 5, 7, 8, 11 and13) To evaluate the fish size, we measured the length ofindividual cod (n = 3,803) and haddock (n = 5,165) fromeight different hauls (Hauls 4, 5, 7, 8, 11, 12, 13 and 15).The fish were subsequently collected from the processingline after the throats were cut, and measurements of initialwhite muscle pH, body temperature, weight, fork length andrigor status were conducted on cod (weight 2.4 ± 0.1 kg,fork length 61 ± 1 cm; n = 102) and haddock (weight1.7 ± 0.1 kg, fork length 50 ± 1 cm; n = 97) from fivedifferent hauls (Hauls 1, 3, 7, 12 and 15), approximatelywithin 60 min after the fish were landed on deck

30.0#

Grid section

Intermediatesection

Cutting rate

knotless 9.4 mm

PA double

8 mm

Material

PA double

8 mm

PA double

6 mm

Fig 3 Specifications for the

two different codends, including

information on the material,

number of meshes and stretched

length, mesh size and cutting

rate

Cod (n = 40) and haddock (n = 40) from Haul 8 were

collected at random from the processing line before the

fish were frozen The fish were packed in Styrofoam

boxes, frozen onboard (-21°C), transported to our

labo-ratory and stored at -21°C for 8–9 days The fish were

then thawed at 2–4°C in a cold room for 2 days and

fil-leted (skin-on) by hand, following which muscle pH, fillet

colour and visual assessments of blood spots and gaping

were carried out (day 10 and 11 postmortem for haddock

and cod, respectively)

Analyses

The mortality rate was estimated by cessation of gill

movements and by gently touching the mid-line and the tail

immediately after the catch landed onboard to see if the

fish responded Visual assessments of injuries on whole fish

were conducted using a scoring system for different

parameters: gear injuries, scale loss, pressure injuries and

bruises (skin discolorations) A score was devised ranging

from 0 up to 2 based on descriptive terms for each

parameter (Table2)

The body temperature was measured in the white muscle

between the mid-line and the dorsal fin, and in filleted fish

the temperature was measured in the flesh A Testo 110

thermometer (Lenzkirch, Germany) was used The pH was

measured directly in the white muscle between the mid-line

and the dorsal fin using a shielded glass electrode (WTW

SenTix 41) connected to a portable pH meter (model WTW

315i; WTW, Weilheim, Germany) During the ments, the instrument was frequently calibrated using pH4.01 and pH 7.00 buffers Frequent cleaning of electrodeswas needed to obtain consistent results

measure-The rigor mortis progression during ice storage wasdetermined using the Rigor Status Method [0 = pre- orpostrigor; 1 = rigor onset (first sign of stiffness, for instance,

Table 1 Catching data for each haul, fishing conditions and catch amounts for traditional and T90 codend trawl nets

Haul no Date

November 2004

Haul duration (h)

Wind speed (m/s)

Fishing depth (m)

T90 (kg fish)

Traditional trawl net (kg fish)

Catch difference T90 vs trad (kg)

Total catch (kg)

of the fish (\2)

Bruises (discoloration

on the skin)

on parts of the fish

discoloration

in the neck or tail region); 2 = rigor (a larger area is clearly

in rigor); 3 = whole fish in rigor; 4 = stronger rigor;

5 = very strong rigor (the fish is extremely stiff, rod-like)]

[20] Evaluation was done by touching the fish to evaluate

muscle tension and by carefully lifting the fish a few

centi-metres above the ice to judge the degree of stiffness of the

fish

The gaping frequency was subjectively assessed using a

score of 0 to 5 according to the number of slits in the fillet

[21] Visual assessments of the number of fillet blood spots

according to a subjective scoring system from 0 to 2 was

used, where 0 = no blood spots, 1 = 1–4 small spots and

2 = large blood spots or several small ([4)

The colour (L*, a* and b*, International Commission of

Illumination (CIE) 1976 colour space; hue; chroma] of the

flesh was measured using the Minolta Chroma Meter

CR-200 (Minolta, Osaka, Japan) The instrument readings cover

an area of 8 mm in diameter The hue angle (0°/360° = red

hue, 240° = blue hue) and chroma (colour intensity or

saturation) were calculated as: hue = 360 - tan-1(b*/a*)

where a* \ 0 and b* \ 0, and as chroma = (a*2? b*2)‘

The measurements were carried out on the white muscle

between the mid-line and the second dorsal fin at three

different locations along the fillet The instrument was

calibrated using a standard white plate

Statistics

To test significance on the effect of trawl gear and haul

number on the different parameters (pH, temperature,

weight, length, L*, a*, b*, hue and chroma), a two-way

analysis of variance (ANOVA) was used A one-way

ANOVA was used to test significance between the two

trawl gears and between different catch amounts For the

non-parametric results (mortality, injuries, gaping and

blood spots) the Mann–Whitney test was used Significant

differences were defined as p \ 0.05 All values are

reported as mean values ± standard error of means

(SEM)

Results

Catch amount and fish size

During the trip a total of 47,600 kg of gutted fish was

landed from 16 different hauls (Table1) The two codends

contributed the same amount of total catch, namely,

24 tons fish from each trawl gear The average length of

cod and haddock from the various hauls is given in Fig.4

In terms of fish length between fish caught in the two trawl

nets, cod from Hauls 4, 7, 8 and 13 and haddock from

Hauls 5 and 7 were significantly different In total, cod

and haddock caught with the T90 codend were on average1.5 and 0.5 cm longer (fork length), respectively, thancod and haddock caught with the traditional codend(p \ 0.05)

Mortality and injuriesThe mortality of cod caught by the two trawl nets, evalu-ated immediately after the catches were brought onboard,was 2.4%, with no significant difference between the nets.For haddock caught by T90 and traditional trawl nets, themortality was 7.1 and 14.3%, respectively, but the differ-ence was not significant In both nets, haddock had a highermortality rate than cod (p = 0.03)

More than 94% of both species exhibited various extents

of scale loss, and 20–30% of the fish had some kind ofinjuries caused by the fishing gear (Table3) Bruises werefound on approximately 20% of both cod and haddock,while pressure injuries were found on 3–5 and 9% of the

Fig 4 Fork lengths of cod and haddock from the two trawl nets.

*Significant difference between the trawl nets Mean ± standard error of the mean (SEM) (n = 122–530)

total catch of cod and haddock, respectively Haddock were

found to have a significantly higher scale-loss scores and a

higher degree of pressure injuries than cod Moreover,

haddock caught with the traditional trawl net exhibited

more gear injuries than haddock caught with the T90 trawl

(p \ 0.05; Fig.5) There were no significant differences

between the two trawl nets in terms of external damages of

cod caught Catch amount had a significant impact on the

degree of injuries (Table4) When the catch amounts were

[2,000 kg, cod showed a higher degree of pressure

inju-ries and bruises, while gear injuinju-ries were lower (p \ 0.05)

The relationship between catch amount and injuries were

not so pronounced for haddock At catch amounts

[2,000 kg, haddock had more pressure injuries (p \ 0.05),

while gear injuries and scale-loss were higher when the

catch amount was \2,000 kg (p \ 0.05) No relationship

between catch amounts and bruises were observed for

haddock A general observation was that haddock was

more vulnerable to catch handling than cod, as both the

mortality and some of the injuries were higher for haddock

than cod (p \ 0.05)

Initial muscle-pH and rigor mortis

The mean white muscle pH of cod and haddock from all

the hauls and from the two trawl nets were 7.3 ± 0.2 and

6.8 ± 0.2, respectively There was no significant difference

between the trawl nets, but for cod the initial muscle pH

showed significant difference between some of the hauls

(Table5) Cod from Haul 3 had significantly lower initial

pH value compared with cod from Haul 12 and 15

How-ever, there were significant differences between the species

and initial muscle pH (p \ 0.001) The mean body

temperature ranged from 6.0 to 7.5°C according to sea

temperatures (Table5)

Rigor mortis assessments for both species are shown inFig.6 Rigor onset started within 5 h after catching Firmrigor (score [2) was evident in 63% of the fish after 10 hpostharvest; after 20 h postharvest, 100% of both specieswere in peak rigor No significant differences in the rate ofrigor development were observed between the two trawlnets Haddock seemed to enter rigor a little later than cod,but rigor peaked at the same time in both fish None of thedead fish showed visible signs of rigor when the catch wasbrought on board

Fillet qualityAfter 10 days (haddock) and 11 days (cod) postmortem,the ultimate pH values for cod (pH 6.9 ± 0.2) and haddock(pH 6.5 ± 0.1) were significantly different (p \ 0.05)(Table6) The mean gaping score for both species was low(Table6), and there was no difference between the twotrawl nets for both species However, the gaping score forcod fillets caught with the traditional trawl net tended to be

Table 3 Mean cod and haddock mortalities and injuries as

percent-ages of the whole catch with T90 and traditional trawl nets

* Significant differences between the species (p \ 0.05)

Fig 5 Effect of using T90 and traditional trawl nets to catch fish on external damage (gear injuries, scale-loss, pressure injuries and bruises) of cod and haddock *Significant difference between the two trawl nets (p \ 0.05) Mean ± SEM (n = 233–272)

higher than for cod caught with the T90 gear, but the

dif-ference were not significant (p = 0.07) Blood spots were

observed in 33 and 28% of cod and haddock fillets,

respectively, with a score \0.6 for both species (Table6)

Again, no differences between the two trawl nets for either

species nor between the species were observed (p [ 0.05)

The L*, a*, b*, hue and chroma values measured on

fillets of cod and haddock are shown in Table7 There

were no significant differences between the two trawl nets

for either species, and the data were therefore pooled

However, there was a difference between the species in the

lightness (L* value) of the fillets (p \ 0.05), with the

haddock being darker than the cod

Table 4 Mean cod and haddock injuries as percentages of catch

Values are given at the mean ± SEM

Different letters (A, B) denote significant differences between the catch amount for each species (p \ 0.05)

Table 5 Pooled initial cod and haddock white muscle pH and body

temperatures from five different hauls

Body temperature (°C)

Mean ± SEM Since there were no significant differences between

the two trawl nets, the data were pooled Different letters (A, B)

denote significant differences between the haul numbers

* Significant differences between the species (p \ 0.05)

Fig 6 Development of rigor mortis of cod (n = 102) and haddock (n = 97) caught with T90 and traditional trawl nets Data from five different hauls are pooled Mean ± SEM

Table 6 Gaping, blood spots and ultimate pH in thawed fillets of cod and haddock caught with T90 and traditional trawl nets

Values are given as the mean ± SEM, (n = 20)

* Significant differences between species (p \ 0.05) No significant differences between the trawl nets were found

Even if the size of the fish did not differ much between the

two trawl nets, the T90 caught significantly slightly larger

fish than the traditional trawl net for both species (Fig.4)

It was not possible to clarify whether this difference was

due to better selectivity, possibly as a result of more open

meshes and a larger cross-sectional area in the T90 codend

(Fig.1), which have been documented in the Baltic cod

fishery [22,23]

According to the impact of the trawl gears on injuries,

the T90 caused fewer gear injuries to haddock The

posi-tive properties of the T90, with its larger cross-section,

reduced flow and turbulence, appeared to reduce the

damage on the more delicate fish species—in this case,

haddock Similar effects were not seen for cod This is not

in accordance to unpublished results of U.J Hansen, L.H

Knudsen, P Nielsen and E.M Andersen (1996) who

reported that cod caught with a T90 codend had 28% fewer

injuries than cod caught with a traditional trawl net It can

be argued that the reduced movement of codend is not a

direct measurement of the ability to preserve fish quality

and reduce injuries However, there is strong evidence

from studies on the survival of fish after escape from the

trawl that the fish are more damaged from rubbing against

the netting in the trawl than from the act of penetrating the

netting and escapement [24, 25] The damage is seen as

pressure marks and other bruises to the flesh as well as

damage to the skin, especially in the head and tail regions

A possible explanation for these observations is that the

fish are thrown around in the incessant turbulent flow

inside the codend and are scraped against the knots in the

net

Nevertheless, regardless of codends, trawling had

neg-ative impacts on injuries on both fish species; in particular,

various degrees of scale loss were observed on almost all

fish The mortality rate was different between cod and

haddock, with haddock having the highest mortality rate

(Table3) This is in accordance to McCracken [26], who

found a greater mortality among haddock than Atlantic codwhen both were captured by otter trawl for tagging pur-poses and towed for approximately 30 min at speedsbetween 140 and 200 cm/s Beamish [13] found that themortality varied from 2 to 42% in haddock caught by ottertrawl during a 30-min tow and a recovery period for 12 h.The majority of deaths occurred during the first hour aftercatch

It is possible that a number of other factors affected thefish quality, positively or negatively, in our experiment andthereby diminished the effect of the different codends.Such factors could include the selectivity devises (flexigr-ids), the codend attachments (protection bags) and the thicktwines used in the trawl nets

The proportion of injured fish was in some cases higherwhen the catch size was larger (Table4) However, thecatches in this were rather small (Table1) Wagner [18]reported that in his study the quality of the cod, expressed

as external appearance and consistency, deteriorated astrawling time and catch size increased, while Hattula et al.[6] did not find any effect of catch size (100–3,500 kg) onthe quality of herring However, in both of these studies,the shorter the duration of the trawling, the larger theproportion of fish alive after being caught Botta andBonnell [5] showed that the reduction in the quality ofAtlantic cod caught by otter trawl was primarily due toexternal factors, such as delayed bleeding, storage methodand time as well as catch size Catching more than 5 tonsper haul decreased the overall grade of the cod

The initial pH of cod was comparatively high ering the presumed handling stress the fish were subjected

consid-to during the catching process (Table5) The differencesobserved between three of the hauls were probably due tocatch amount in the hauls (p = 0.03) The haul with thecod with the lowest initial muscle pH had the highest catchamount, 2.8 tons The initial pH of haddock was lower thanthat of cod, with no difference between trawl nets, hauls orcatch amounts Our results therefore indicate that haddock

is more susceptible to handling stress than cod Low pH atthe time of killing is widely recognised as an indicator ofhandling stress, as reported in salmon [27], eel [28] andturbot [29] Typical initial muscle pH values reported forexhausted cod is about 7.0 [30–32], which is lower thanthat reported for our cod, namely, pH 7.2–7.3 (Table5).Surprisingly, this means that the cod in our experimentwere just partially affected by capture stress In previousstudies, initial pH values for rested harvested farmed codwere reported to vary between 7.3 [30], 7.4 [32] and 7.9[31]

Rigor mortis is highly influenced by the ATP-content inthe muscle, and below a critical level, actin and myosinmake an irreversible bond and muscle enters rigor mortis[33] None of the fish showed visible signs of rigor when

Table 7 Cod and haddock fillet colour as L*, a*, b*, hue and chroma

values after 10 days (haddock) and 11 days (cod) postmortem

Values are given as the mean ± SEM (n = 40 As there were no

significant differences between the two trawl nets or both fish species,

the data were pooled

* Significant differences between species (p \ 0.05)

the catch was brought onboard, indicating that the fish had

not died early during the 5-h trawling event Rigor mortis

onset occurred 5 h after catching, with no significant

dif-ferences between trawl nets or fish species (Fig.6) Since

the initial pH value differed between the species, a

dif-ference in the development of rigor mortis could be

expected between the species, with cod entering rigor later

than haddock However, we have no plausible explanation

for this incidence The early rigor mortis onset indicated

that the trawled fish were exhausted, or nearly so, due to

struggling and handling during the dragging and emptying

of the trawl net Maximum rigor was attained after 21 h,

which is in accordance with Kristoffersen et al [31], who

found that cod exposed to preslaughter handling stress

reached maximum rigor after 20–24 h Hattula et al [6] did

not find any correlation between trawling time and rigor

index in herring trawled for 2 and 4–5 h, but they did find

that herring caught with gillnet developed rigor mortis

earlier than herring caught with trawling or poundnetting

Gillnetted pike perch (Sander lucioperca) also developed

rigor mortis earlier than fish caught by longline and

har-poon [8]

The ultimate pHs in cod and haddock were significantly

different (Table6) These results are in accordance with

those of Love [34], who found a higher ultimate pH in

wild-caught cod (pH 6.7) than in haddock (pH 6.4–6.6)

Hultmann and Rustad [35] reported a similar ultimate pH

value for cod caught in November in northern Norway

The mean gaping score for both species was low, and no

effects of the two trawl nets were found (Table6) These

results are in line with those of Fletcher et al [36] who

found no effects of exercise on the tendency of gaping of

king salmon However, Jerrett et al [37] doubted the

use-fulness of using gaping scores as an indicator of the

pro-pensity of the flesh to gape Love [34] found that the

connective tissue of cod is stronger at high postmortem pH

values (pH [ 6.6) than at lower pH values Below pH 6.6,

the muscle tends to gape more In comparison, the ultimate

pH values for our cod and haddock were 6.9 and 6.5,

respectively

There were some small blood spots in the cod and

haddock fillets (Table6), but no significant effects of the

two trawl nets were observed Our findings, with a low

score for blood spots, indicate that the fillets were only

slightly influenced by the catching and handling processes

Botta et al [4] compared the gillnet, handline, longline and

trap and reported that the method of catching significantly

(p B 0.001) affected the colour and discolouration/bruising

of Atlantic cod The colour of our fillets was not different

in fish caught in the different trawl nets, except that the cod

fillets were lighter than haddock fillets (Table7)

To conclude, the codend made from T90 netting caught

slightly—but significantly—larger cod and haddock than

the traditional codend Haddock caught with traditionaltrawl net had more gear injuries related to the trawl gearthan haddock caught with the T90 gear No other differ-ences in fish quality between the trawl nets were observed.Regardless of trawl type, most fish exhibited a high degree

of external damage, particularly in terms of scale loss.Haddock were more susceptible to suffering damage thancod (p \ 0.001) We recommend further research to con-firm the possibly positive properties of the T90 trawl net.Additional research is also recommended to provide morescientific knowledge about how the trawling processinfluences the fish quality of other fish species

Acknowledgments The financial support of The Research Council

of Norway (NFR project No 151831/120) and the Norwegian Seafood Federation are gratefully acknowledged The authors would also like to thank the crew of M/T J Bergvoll for excellent cooper- ation during the experiment.

References

1 Fraser DI, Weinstein HM, Dyer WJ (1965) Post-mortem lytic and associated changes in the muscle of trap- and trawl- caught cod J Fish Res Board Can 22:83–100

glyco-2 Auclair G (1984) Comparative-study of trawl and gillnet effects

on the quality of fish Canadian Institute of Food Science and Technology Can Inst Food Sci Technol J 17:R25

3 Valdimarsson G, Matthiasson A, Stefansson G (1984) The effect

of on board bleeding and gutting on the quality of fresh, quick frozen and salted products In: Møller A (ed) Fifty years of fisheries research in Iceland Icelandic Fisheries Laboratory, Reykjavik, pp 61–72

4 Botta JR, Bonnell G, Squires BE (1987) Effect of method of catching and time of season on sensory quality of fresh Atlantic cod (Gadus morhua) J Food Sci 52:928–931, 938

5 Botta JR, Bonnell G (1989) Causes of reduced quality of fresh Atlantic cod (Gadus morhua) caught by otter trawl In: Proc World Symp Fishing Gear and Fishing Vessel Design St John’s, Newfoundland, pp 340–344

6 Hattula T, Luoma T, Kostiainen R, Poutanen J, Kallio M, uronen P (1995) Effects of catching method on different quality parameters of Baltic herring (Clupea harengus L.) Fish Res 23:209–221

Su-7 Cole RG, Alcock NK, Handley SJ, Grange KR, Black S, Cairney

D, Day J, Ford S, Jerrett AR (2003) Selective capture of blue cod (Parapercis colias) by potting: behavioural observations and effects of capture method on peri-mortem fatigue Fish Res 60:381–392

8 O ¨ zyurt G, O¨zogul Y, O¨zyurt CE, Polat A, O¨zogul F, Go¨kbulut C, Ersoy B, Ku¨ley E (2007) Determination of the quality parameters

of pike perch Sander lucioperca caught by gillnet, longline and harpoon in Turkey Fish Sci 73:412–420

9 Pankhurst NW, Sharples DF (1992) Effects of capture and finement on plasma cortisol concentrations in the snapper, Pagrus auratus Aust J Mar Freshw Res 43:345–356

con-10 Lowe TE, Ryder JM, Carragher JF, Wells RMG (1993) Flesh quality in snapper, Pagrus auratus, affected by capture stress.

J Food Sci 58:770–773, 796

11 Hopkins TE, Cech J Jr (1992) Physiological effects of capturing striped bass in gillnets and fyke traps Trans Am Fish Soc 121:819–822

12 Parker RR, Black EC, Larkin PA (1959) Fatigue and mortality in

troll-caught Pacific Salmon (Oncorhynshus) J Fish Res Board

Can 16:429–448

13 Beamish FWH (1966) Muscular fatigue and mortality in

had-dock, Melanogrammus aeglefinus, caught by otter trawl J Fish

Res Board Can 23:1507–1521

14 Korhonen RW, Lanier TC, Giesbrecht F (1990) An evaluation of

simple methods for following rigor development in fish J Food

Sci 55:2

15 Chopin FS, Arimoto T, Inoue Y (1996) A comparison of the

stress response and mortality of sea bream Pagrus major captured

by hook and line and trammel net Fish Res 28:277–289

16 Botta JR, Squires BE, Johnson J (1986) Effect of bleeding/gutting

procedures on the sensory quality of fresh raw Atlantic cod

(Gadus morhua) Can Inst Food Sci Technol J 19:186–190

17 Kelly TR (1969) Discolouration in sea-frozen fish fillets In:

Kreuzer R (ed) Freezing and irradiation of fish Fishing News

(Books), London, pp 64–67

18 Wagner H (1978) Einfluss der Schleppzeiten und Steertfu¨llung

auf die Qualita¨t des Fisches (in German) Seewirtschaft 10:399–

400

19 Esaiassen M, Nilsen H, Joensen S, Skjerdal T, Carlehøg M,

Eilertsen G, Gundersen B, Elvevoll E (2004) Effects of catching

methods on quality changes during storage of cod (Gadus

mor-hua) Food Sci Technol/LWT 37:643–648

20 Erikson U (2000) Rigor measurements In: Kestin SC, Warris PD

(eds) Farmed fish quality Blackwell, Cornwall, pp 283–295

21 Andersen UB, Strømsnes AN, Steinsholt K, Thomassen MS

(1994) Fillet gaping in farmed Atlantic salmon (Salmo salar).

Norwegian J Agric Sci 8:165–179

22 Moderhak W (1997) Determination of selectivity of cod codends

made of netting turned through 90 degree Bull Sea Fish Inst

140:3–14

23 Moderhak W (1999) Investigations of the selectivity of cod

(Gadus morhua) codends with meshes turned through 90 degree.

Bull Sea Fish Inst 146:39–55

24 Main J, Sangster GI (1990) An assessment of the scale damage to

and survival rates of young gadoid fish escaping from the cod-end

of a demersal trawl Scottish Fish Res Rep No 46/90 Fisheries

Research Services, Aberdeen

25 Lehman KM, Sangster GI (1994) Assessment of the survival of

fish escaping from commercial fishing gears EC Contract Final

Report No TE 3.741

26 McCracken FD (1956) Cod and haddock tagging off Lockeport.

N S J Fish Res Board Can 64:10–15

27 Sigholt T, Erikson U, Rustad T, Johansen S, Nordtvedt TS, Seland A (1997) Handling stress and storage temperature affect meat quality of farmed-raised Atlantic salmon (Salmo salar).

J Food Sci 62:898–905

28 Morzel M, Van De Vis H (2003) Effect of the slaughter method

on the quality of raw and smoked eels (Anguilla anguilla L.) Aquac Res 34:1–11

29 Morzel M, Sohier D, Van De Vis H (2003) Evaluation of slaughtering methods for turbot with respect to animal welfare and flesh quality J Sci Food Agric 83:19–28

30 Stien LH, Hirmas E, Bjørnevik M, Karlsen Ø, Nordtvedt R, Røra AMB, Sunde J, Kiessling A (2005) The effects of stress and storage temperature on the colour and texture of pre-rigor filleted farmed cod (Gadus morhua L.) Aquac Res 36:1197–1206

31 Kristoffersen S, Tobiassen T, Steinsund V, Olsen RL (2006) Slaughter stress, postmortem muscle pH and rigor development

in farmed cod (Gadus morhua L.) J Food Sci Technol 41:861– 864

32 Misimi E, Erikson U, Digre H, Skavhaug A, Mathiassen JR (2008) Computer vision-based evaluation of pre- and post rigor changes in size and shape of Atlantic cod (Gadus morhua) and Atlantic salmon (Salmo salar) fillets during rigor mortis and ice storage: effects of perimortem handling stress J Food Sci 73:E57–E68

33 Iwamoto M, Yamanaka H, Watabe S, Hashimoto K (1987) Effect

of storage temperature on rigor-mortis and ATP degradation in plaice Paralichthys olivaceus muscle J Food Sci 52:1514–1517

34 Love RM (1979) The post mortem pH of cod and haddock muscle and its seasonal variation J Sci Food Agric 30:433–438

35 Hultmann L, Rustad T (2002) Textural changes during iced storage of salmon (Salmo salar) and cod (Gadus morhua).

J Aquat Food Prod Technol 11:105–123

36 Fletcher GC, Hallett IC, Jerret AR (1997) Changes in the fine structure of the myocommata-muscle fibre junction related to gaping in rested and exercised muscle from king salmon (On- corhynchus tshawytscha) Food Sci Technol/LWT 30:246–252

37 Jerrett AR, Stevens J, Holland AJ (1996) Tensile properties of white muscle in rested and exhausted Chinook salmon (On- corhynchus tshawytscha) J Food Sci 61:527–532

O R I G I N A L A R T I C L E Fisheries

Estimating larval supply of Ezo abalone Haliotis discus

hannai in a small bay using a coupled particle-tracking

and hydrodynamic model: insights into the establishment

of harvest refugia

Yoichi Miyake•Shingo Kimura•Tomohiko Kawamura•

Takashi Kitagawa•Motoyuki Hara•Hiroshi Hoshikawa

Received: 2 November 2009 / Accepted: 23 May 2010 / Published online: 17 June 2010

Ó The Japanese Society of Fisheries Science 2010

Abstract Most juveniles of Haliotis discus hannai have

been found to be descendants of wild individuals, although

most adults were artificially produced (released)

individu-als as a result of restocking inside the refugium located

near the head of Oshoro Bay, Hokkaido, Japan To estimate

the larval supply from released and wild individuals into

the refugium and to compare the suitability of locations as

larval sources, we simulated larval dispersal using a

cou-pled hydrodynamic and particle-tracking model The

sim-ulation results indicated that more larvae may be supplied

from the wild adults inside the bay to the refugium than

from the released adults These results are consistent with

the observed high abundance of wild juveniles in the

refugium Most larvae from the refugium were predicted to

disperse out of the bay We found that larval retention in

the bay was at least one order of magnitude higher than that

in the refugium Thus, it may be more effective in terms of

self-replenishment and reproduction if the refugium were

to be expanded to the bay scale There were only minor

differences among the compared sites at the head of the bay

in terms of their suitability as larval sources Consequently,

the establishment of new refugia in this area could beexpected to provide an effectiveness equal to that of thecurrent refugium

Keywords Abalone Delft3D  Larval dispersal Marine reserves  Harvest refugia  Hydrodynamics Numerical modeling

IntroductionEzo abalone Haliotis discus hannai is one of the mostvaluable fishery resources in Japan The overall catch ofthis species has declined drastically from 3,198 tonnes in

1969 to nearly one tenth of that in 1990 The release ofhatchery-reared juveniles of H discus hannai into areascontaining wild populations started in the late 1970s [1].However, the catch still remains low despite this long-termprogram of releasing artificially produced abalone seeds[2], necessitating new effective measures in fishery man-agement to facilitate the recovery of the resource in thenatural environment

Marine reserves can improve depleted populations ofharvested gastropod [3] and fish species [4] The reservesmay result in an increased abundance and size [5,6] of theprotected species or in increased egg production [6] Thereserves that could potentially enhance fisheries by acting

as centers of dispersal of propagules and adults into thesurrounding areas are often termed ‘‘harvest refugia’’ [7].Abalone species have planktonic larval stages beforetransitioning into crawling benthic stages [1] Although thedispersal of adult abalone is limited, abalone larvae candisperse on the scale of kilometers [8] However, the fer-tilization success lowers as the distance between spawningindividuals increases [9] Consequently, harvest refugia

Y Miyake ( &)  S Kimura  T Kawamura  T Kitagawa

Atmosphere and Ocean Research Institute,

The University of Tokyo, Kashiwa,

Chiba 277-8564, Japan

e-mail: miyakey@aori.u-tokyo.ac.jp

M Hara

National Research Institute of Aquaculture,

Fisheries Research Agency, 422-1 Nakatsuhamaura,

Minami-ise, Mie 516-0193, Japan

H Hoshikawa

Central Fisheries Research Institute,

Hokkaido Research Organization, Yoichi,

Hokkaido 046-8555, Japan

Fish Sci (2010) 76:561–570

DOI 10.1007/s12562-010-0260-4

which establish spawning populations can be considered to

be effective measures for the restoration of depleted

aba-lone resources

In Oshoro Bay, Hokkaido, a no-take zone (hereafter

referred to as the refugium) has been in place since 2001, and

hatchery-reared H discus hannai [approx 50 mm standard

length (SL)] have been released into the refugium with the

aim of investigating how the enhancement of adult abalone

density influences recruitment around this area [10] The

spawning of H discus hannai occurs briefly between July and

October in Oshoro Bay [10] Using a microsatellite DNA

marker, Hoshikawa and Hara [11] determined that only 12–

20% of the juveniles in the refugium originated from the

spawning group of released artificial seeds between 2003 and

2006 This percentage was low, considering the high

per-centage (90%) of released abalone [11] This low percentage

of offspring originating from the released seeds could occur if

the larval supply from the natural sources is much greater than

that from the artificial seeds To investigate this possibility,

we simulated the larval dispersal of H discus hannai using a

coupled particle-tracking and hydrodynamic model This

modeling approach can be used for studying the temporal and

spatial distribution of larvae and for quantitatively estimating

larval supply [12] For the simulations of hydrodynamics and

larval dispersal, we used the Delft3D-FLOW and -PART

simulation programs (WL|Delft Hydraulics, Delft, the

Neth-erlands) Simulated larval dispersal processes can provide

novel approaches for effective fishery management In

addition, information of quantitative estimates of larval

supply to and from the refugium in Oshoro Bay can provide

useful information on how the establishment of refugia in a

small bay can contribute to the management of depleted

resources In the study reported here, we focused on the local

larval dispersal in Oshoro Bay itself and in its vicinity

Small bays are common in coastal areas (i.e., abalone

fishery grounds), and an understanding of the suitability of

this geographical feature for the enhancement of marine

resources is important for fisheries management The

pri-mary objectives of our study were to (1) estimate larval

supply from wild and artificial seed adults into the

refu-gium, (2) estimate larval retention in the refugium and bay,

and (3) compare the suitability of the locations as larval

sources based on self-recruitment in that location and larval

retention in the bay by simulating the larval dispersal of

H discus hannai in Oshoro Bay and its vicinity

Materials and methods

Study area

The study area included the regions inside Oshoro Bay

(43°12.70N, 140°51.40E), which is located on the south

coast of Ishikari Bay, facing the Sea of Japan, and itsvicinity (Fig.1) Oshoro Bay is a small bay between twoheadlands, and its size is approximately 700 m long in bayaxis and 300 m at the bay mouth There is a breakwaternear the bay head that separates the bay into two regions.The refugium is situated at the shore opposite to thebreakwater The size of the refugium is 150 m long in thealong-shore direction and extends to where the area ofboulders and cobbles (2–4 m deep) end in the offshoredirection [10]

Hydrodynamic measurements in the fieldThe regional hydrodynamics were recorded by a total offour current meters during the spawning season of H dis-cus hannai (Fig.1) These current meters were deployedoutside of the bay (Stn A), at the bay mouth (Stn B and C)and the head of the bay (Stn D) in July–October 2007,where they recorded both the current speed and the watertemperature The measurement depths were 5 m (Stn A–C)and 3.5 m (Stn D) The current data were compared usingSpearman’s correlation coefficient The data collected fromconductivity (salinity), temperature (water), and depth(CTD) stations were recorded at depth increments of 0.1 moutside the bay on July 14 and at the bay mouth onSeptember 28, 2007 (Fig.1); these were used for thehydrodynamic model setup detailed below

Hydrodynamic modelThree-dimensional hydrodynamic model simulations werecarried out using the Delft3D-FLOW simulation program(WL|Delft Hydraulics) Delft3D-FLOW solves theunsteady shallow water equations in three dimensions inwhich the system of equations consists of the horizontalequations of motion, the continuity equation, and thetransport equations for conservative constituents [13].Figure1shows the model domain and its open boundaries(north, east, and west) The model grid and bathymetrywere prepared using Delft3D-RGFGRID and -QUICKIN,based on the bathymetry data provided by the JapanOceanographic Data Center and Hokkaido National Fish-eries Research Institute, Fisheries Research Agency Themodel domain extended 1,560 m longitudinally and3,000 m latitudinally The simulations were performed on

a Cartesian grid with a spatial resolution of 30 m in thehorizontal direction and 20 r-grid layers in the verticaldirection The time step was set to 3 s The eastern andwestern boundaries were forced with the logarithmic cur-rent profile converted from the east components of half-hourly current velocities observed outside the bay (Stn A).The model simulations were run in two periods, namely,summer (July 15–22) and fall (September 29–October 6),

applying a 1-day spin-up prior to these periods The CTD

measurement data on July 14 and September 28 were used

for the open boundary and the initial conditions The

northern boundary was forced with tidal changes based on

four tidal constituents (K1, O1, S2, and M2) calculated from

the tidal height data measured at the Oshoro tidal station

(Geographical Survey Institute Website:http://tide.gsi.go

jp/main.php?number=4) The free surface was forced by

the atmospheric pressure, wind stress, and heat exchange

Hourly data on atmospheric pressure at Otaru and wind at

Yoichi (Japan Meteorological Agency Website: http://

www.data.jma.go.jp/obd/stats/etrn/index.php) were used

The bay is located between two headlands that could cause

the changes in wind direction, and the wind direction

within the bay was thus changed to along the bay axis The

wind direction inside the bay was changed to northwest

(toward the bay head) when the north component was

prominent and to southeast (toward the bay mouth) when

the south component prevailed The heat exchange through

the free surface was modeled using a heat flux model for

which the relative humidity, air temperature, and the net

(short wave) solar radiation were prescribed Hourly data

of the relative humidity at Otaru and the air temperature at

Yoichi were prescribed The solar radiation (Sapporo), the

air temperature (Yoichi), the humidity (Otaru), the degree

of cloudiness (Sapporo), and the water temperature at Stn

A were used for calculating the net solar radiation (allmeteorological data except water temperature were fromJapan Meteorological Agency website) The backgroundhorizontal eddy viscosity and horizontal eddy diffusivitywere both set to 1 m2/s The viscosity and diffusivity,additional to the background values, were calculated withthe Horizontal Large Eddy Simulation (HLES) sub-gridmodel, which computes sub-grid scale turbulence [13] Thebackground vertical eddy viscosity and diffusivity wereboth set to 10-4 m2/s For computation of additionalvertical turbulent eddy viscosity and diffusivity, the k–eturbulence model was selected The model results werevalidated against observational data, including (1) tidalheight records at the tide station in Oshoro Bay (see above)and (2) the north and east components of currents at Stn A,

B, C, and D The model validations were performed usingthe model skill score [14] and root-mean-square-error(RMSE) The model skill scores are based on quantitativeagreement between model results and observations, andperfect agreement yields a skill score of one, while com-plete disagreement yields the score of zero [14] The modelskill scores and RMSE of the modeled hydrodynamics

5 5

5

5

5

10 10

20

25

25 25

30

500 m

Breakwater Oshoro Bay

N

Tide Station Refugium

Sapporo Otaru Yoichi

H1 H2

H3 IS1

IN1 IN2

ON1

OS1

R

Fig 1 Maps showing the

location of Oshoro Bay (open

circle; upper left), modeled

region (right), and particle

release sites (left bottom) Right

panel Five-meter bathymetric

contours, the location of the

refugium, positions of current

meters (filled triangles A–D),

and conductivity, temperature,

and depth (CTD) stations (filled

circles: July 14, open circles:

September 28) The open

boundaries were located at the

north, east, and west of the

modeled region Bottom left

panel Particle release sites

(gray-shaded areas R, IS1, IN1,

IN2, OS1, ON1, HR, H1–3)

were considered to be acceptable, with those for sea level

being 0.83 and 6.6 cm in the summer simulation and 0.93

and 4.6 cm in the fall simulation, respectively The mean

model scores and RMSE for currents in the summer

sim-ulation were 0.59 and 1.8 cm/s (east component) and 0.57

and 3.3 cm/s (north component), respectively; those in the

fall simulation were 0.49 and 1.5 cm/s (east component)

and 0.38 and 2.9 cm/s (north component), respectively

Particle-tracking model

The larval dispersal simulations were performed using

Delft3D-PART The PART computes the position of every

individual particle by advection and dispersion It simulates

transport processes by means of a particle-tracking method

using flow data from FLOW The random walk method

was used for the dispersion The horizontal dispersion

coefficient was set to 1 m2/s, and the vertical dispersion

coefficient was calculated as a function of the total depth

by the option ‘‘Depth averaged algebraic’’, which assumes

the vertical dispersion to be constant over depth [15]

Larval dispersal from the wild and artificial seed

spawning groups in the refugium (R, Fig.1) and the areas

inside the bay (IS1, IN1 and 2; Fig.1) were simulated by

releasing particles at these sites To investigate the larval

supply from the spawning groups outside the bay, we also

released the particles at OS1 and ON1 These particle

release sites consisted of a number of grid cells (Table1)

Two periods, summer (July 15–22) and fall (September

29–October 6), in the spawning season of H discus hannai

were chosen for larval dispersal simulations Larvae can

start to attach to the bottom at about 3 days of age [16], and

newly metamorphosed post-larvae were estimated to be

4–6 days old in the natural environment [17] Hence, the

competent period for settlement (hereafter referred to as

competent period) was assumed to be from 3 to 7 days

after particle release The influence of larval swimming

behavior on their dispersal remains largely unknown for

abalone species [12] Consequently, we employed passive

particles for the larval dispersal simulations Our focus was

on the dispersal of larvae that entered the water columnsince these particles cannot simulate the larvae that remain,for example, inside rock crevices As the larvae in thewater column are likely to disperse more widely than thosethat stay in the slower flow environment, the results in thisstudy are based on the maximum larval dispersal Thesimulations require a large number of particles to obtainaccurate results due to the stochastic nature of the disper-sion process [15] Thus, 1,000 particles were released ateach site To estimate the ratio between larval supply fromthe wild and artificial seed spawning groups to the refu-gium during the competent period, we weighted the num-ber of particles by multiplying the fecundity (Table1)because a consistent number of particles (i.e., 1,000) werereleased at each particle release site Due to the lack of fielddata (adult abundance and mean shell length) for spawninggroups outside the bay, the fecundities at IS1 and IN1 wereused for the particle release sites, OS1 and ON1, respec-tively The number of individuals in each particle releasesite was obtained by multiplying the abundance of adultindividuals (ind./m) by the number and size of grid cells.The abundance of adult individuals and mean shell lengthswere surveyed in 2006 (Table1; H Hoshikawa, unpub-lished data) Although these data were obtained 1 yearprior to the simulation periods, they were used togetherwith the particle-tracking model results so that larval sup-ply from different spawning groups could be taken intoaccount The number of spawned eggs per individual wasestimated using the following Eq.1 [18]

where Y is the number of spawned eggs For X [SL (mm)][18], the mean shell length of the spawning group (mm) inthe particle release site was used (Table 1, Hoshikawa H,unpublished data) The larval supply (Table2) from eachspawning group into the refugium was estimated Larvalretention inside the refugium and bay (Table2) wereestimated in the summer and fall simulations

Table 1 Details on the particle release sites and spawning groups

Particle release

site

Number of grid cells

Adult abundance a (ind./m) Mean shell length a (mm ± SE) Fecundity (910 6 eggs)

Self-recruitment and larval retention (Table2) were

used as the criteria to evaluate the suitability of the

loca-tions as larval sources The suitability was compared

among four sites, including HR (refugium) and H1–3(hypothetical larval sources) at the head of the bay (Fig.1)

A total of 1,000 particles were evenly released in five gridcells inside each site

ResultsObserved hydrodynamicsThe hydrodynamic measurements indicated that the along-shore current was predominant outside the bay (Stn A;Fig.2), with the prominent flow directions being north–northeast to the north of the bay mouth (Stn B) andnortheast and west south of the bay mouth (Stn C) Meancurrent speed was greater outside the bay than inside Themean current speeds at the current meter stations were 5.4,3.2, 2.4 and 1.3 cm/s at Stn A, B, C, and D, respectively,with flows generally weaker at sites closer to the bay head(Fig.2) The mean flow was weaker south of the baymouth (Stn C) than north of it (Stn B) In addition, thenorth component of the current to the north of the baymouth (Stn B) was significantly correlated with the north

Table 2 Definitions of terms

Number of females Half of the number of individuals in the

particle release sitea

multiplied by the number of spawned eggs per individual

Larval supply The number of particles that reached the

refugium during the competent period multiplied by the fecundity (Table 1 )bSelf-recruitment The number of particles returned to the

release site during the competent period against the number of particles releasedbLarval retention The proportion of particles inside the

subject area during the competent periodc

a The ratio between male and female was assumed to be 1:1

b Only particles introduced into the subject area for the first time

Fig 2 Fluctuations of low-passed (running mean 25 h) currents at the current measurement stations A (a), B (b), C (c), and D (d) from July 15 to October 14, 2007

(Spearman’s correlation, r = 0.51, P \ 0.01) and east

components of the current (r = 0.41, P \ 0.01) outside the

bay (Stn A) The east components of the currents outside

the bay (Stn A) and to the north of the bay mouth (Stn B)

were weakly correlated (r = 0.28, P \ 0.01) The

corre-lation between the current outside the bay (Stn A) and that

at the south of the bay mouth (Stn C) was weak The north

component of the current at the south of the bay mouth (Stn

C) was significantly correlated with that of the current at

the north of the bay mouth (Stn B, r = 0.44, P \ 0.01) and

weakly correlated with that outside the bay (Stn A,

r = 0.27, P \ 0.01) Observed hydrodynamics indicated

that the flows outside the bay and those at the bay mouth

were influenced by the same currents

Modeled hydrodynamics

The residual currents (vertically and temporally averaged)

during the summer and fall simulations showed relatively

stronger flows outside the bay and weaker flows inside of it

(Fig.3) These simulations were considered to be

consis-tent with the observed currents discussed above The flows

outside the bay mouth were predominantly northward and

westward in the summer and fall simulations, respectively

An eddy was generated at the bay mouth in both simulated

periods, but it was stronger in the summer simulation In

the middle portion of the bay, a weak counterclockwise

eddy formed in both periods, but it was relatively stronger

in the summer simulation The flows in the bay head area

were generally weaker than those in other parts of the bay,apparently due to currents being blocked by thebreakwater

Larval dispersal processesThe results of the larval dispersal simulation from therefugium are shown in Fig 4 The number of particles ineach grid cell was totaled for each day between Days 0–7,enabling larval dispersal and areas with high or low con-centration of larvae to be distinguished In the summersimulation, the particles released at the refugium (R,Fig.1) dispersed outside the bay from Day 0 onwards(Fig.4), moving predominantly toward the north andnortheast of the model domain During the competentperiod, the particles were concentrated inside the bay.Relative to the summer simulation, in the fall simulation,fewer particles remained inside the bay (Fig.4) The par-ticles outside the bay were dispersed toward the northernand western model boundaries In the early competentperiod, the particles inside the bay were more concentratedthan those outside the bay On the other hand, no particlesremained in the bay on the last day (Day 7) of the com-petent period

The particles were released at six sites (Fig 1; R, IS1,IN1, IN2, OS1 and ON1) We estimated the larval supplyfrom the wild and artificial seed spawning groups fromthese sites to the refugium during the competent period(Fig.5) In the summer simulation, the larval supply fromthe artificial seed spawning group into the refugium (R:12.43 9 108) was greater than that from the southern (IS1:9.18 9 108) and northwestern (IN1: 11.12 9 108) wildspawning groups The number of larvae supplied from thewild spawning groups outside the bay, OS1 and ON1, wassmaller (1.91 and 1.62 9 108, respectively) than that fromthe adjacent wild spawning groups in the bay (IS1 and IN1,Fig.5a) The total larval supply in the refugium consisted

of 60% wild spawning group descendants and 40% cial seed spawning group descendants (Fig.5c); thesepercentages were 62 and 38% when the larval supply fromthe spawning groups outside the bay was included(Fig.5d) In the fall simulation, the larval supply from theartificial seed spawning group (R: 1.71 9 108) into therefugium was greater than that from the southern (IS1:1.15 9 108) and northwestern (IN1: 1.01 9 108) wildspawning groups The number of larvae supplied from thenorthern wild spawning group outside the bay (ON1:0.81 9 108) was smaller than that from the inside the bay(IS1 and IN1), and there was no larval supply from thesouthern wild spawning group outside the bay (OS1;Fig.5b) The total larval supply in the refugium con-sisted of 54% wild spawning group descendants and 46%artificial seed spawning group descendants (Fig 5c); these

(b) Fall

(a) Summer

Fig 3 Modeled flow fields inside the bay and in its vicinity in the

summer (a) and fall (b) simulations The flow fields were averaged

vertically and temporally during the simulated periods, and the

vectors outside the bay were plotted at every two grid cells

percentages were 61 and 39% when the larval supply from

the spawning groups outside the bay was included

(Fig.5d)

Particles were released at R, and the proportions of

particles in the refugium and bay were compared during

the competent period (Fig.6) The proportion of particles

in the refugium declined from 0.30 to 0.02% in the

summer simulation and from 0.03 to 0% in the fall

simulation, while that in the bay decreased from 13.01 to

0.84% in the summer simulation and from 2.26 to 0% in

the fall simulation These results indicate that larval

retention in both the refugium and bay could be reduced

by at least one order during the competent period and that

larval retention in the bay was at least one order higher

than that in the refugium

To compare the suitability of the locations as larval

sources, we released the particles at four sites at the head of

the bay (HR and H1–3, Fig.1) The self-recruitment was in

a range of 3.3–5.3 and 0.4–1.0% in the summer and fall

simulations, respectively (Table3) There was little

dif-ference among the release sites in terms of self-recruitment

(summer: 0.1–2.0%, fall: 0.1–0.6%), although the

self-recruitment was slightly higher when the particles were

released at HR and H2 in the summer and fall simulation,

respectively Regardless of the particle release sites, more

than 85% of the released particles had dispersed out of the

bay by Day 3, which is the first day of the competent period

(Fig.7) Larval retention (i.e., proportion of particles) was

higher in the summer simulation There were only minor

differences among the particle release sites inside the bay

in terms of larval retention, although the highest retention

(total of proportion of particles during the competent

period) was obtained when the particles were released

at H2 and H1 in the summer and fall simulations,respectively

DiscussionThe observed current data and the modeled flow fieldsrevealed that the flows were weaker close to the bay head

An eddy at the bay mouth was generated by the namic model The larval dispersal simulations suggestedthat the seasonal difference in hydrodynamic conditionsinfluenced the distribution of larvae in the bay The eddies

hydrody-at the bay mouth and in the middle of the bay were moreevident in the summer simulation than the fall simulation

In the summer simulation, this hydrodynamic condition inthe bay apparently retained the particles in the bay, pre-venting dispersal toward the outside of the bay In contrast,fewer particles were retained inside the bay in the fallsimulation, apparently due to the weaker eddies in the bay.During the competent period in both simulations, the par-ticles were relatively concentrated inside the bay when theywere released at the refugium (R), indicating that it is morelikely that the larvae from the refugium will settle insidethe bay at a higher concentration than in other areas Thesimulation results suggest that the wild adults in the baysupplied larvae into the refugium, which is in agreementwith the observation that juveniles originating from thewild adults were found inside the refugium There was arelatively large larval supply from the northwestern andsouthern wild spawning groups inside the bay in both thesummer and fall simulations, and these adults could be

dominant larval sources for the wild individuals in the

refugium There was no larval supply from the southern

spawning group outside the bay in the fall simulation,

indicating that the larval supply from outside of the bay

could fluctuate depending on the hydrodynamic condition

or the season The estimation of larval supply in this study

does not reflect the seasonal difference in fecundity since

our focus was mainly on the ratio of larval supply from the

wild and artificial seed groups in different locations to the

refugium The gonad index can be expected to be higher in

the fall than the summer [10], and thus the number of

larvae supplied to the refugium could be relatively higher

than the estimated larval supply in the fall

The abundance and fecundity of artificial seeds in the

refugium was high (Table1); however, the simulation

results showed a lower percentage of larval supply from theartificial seed adults into this area compared to that fromthe wild spawning groups The percentage of the larvalsupply that originated from the spawning adults of artificialseeds, exclusive and inclusive of the spawning groupsoutside the bay, was 40–46 and 38–39% of the total larvalsupply into the refugium, respectively The percentage ofthe naturally occurring juveniles estimated using a DNAmarker was in the range of 12–20% of the juveniles in therefugium [11], but this range is only a rough indicator ofthe contribution of artificial seeds to reproduction [19].Thus, the simulation results were consistent with the ratio

of juveniles determined using the DNA marker However,there was a difference in the percentages of artificial seeddescendants between the simulation and the DNA markerresults This difference could have been caused by thepoorer environmental adaptability of artificial seeds com-pared to wild individuals [19] The simulation results showthat the larvae could be supplied from the spawning groupsoutside bay into the refugium This difference could alsooccur if the overall larval supply from the wild adultsoutside the bay (including the wild adults located in theareas that were not considered in this study) to the refu-gium is large To further clarify the influence of larval

(22) (28)

(5) (4) (7) (3)

Fig 5 Larval supply (a, b) and total larval supply (c, d) from the

wild and artificial seed spawning groups in the particle release sites

into the refugium during the competent period in the summer and fall

simulations Numbers in brackets in the upper two panels (a, b) are

the numbers of particles transported to the refugium Larval supply

from the spawning groups outside the bay are excluded in c and

included in d See text and Fig 1 for more details on release sites

(R, IS1, IN1, IN2, OS1, ON1)

Day

summer fall

summer fall

0 0.1 0.2 0.3 0.4 0.5

3 6 9 12 15

Fig 6 Proportion of particles, expressed as a percentage of total number of released particles in the refugium (left) and bay (right) during the competent period in the summer and fall simulations The particles were released at R The proportions are the daily average

Table 3 Self-recruitment (%) at particle release sites (HR, H1, H2 and H3)

HR Harvest refugium The percentages are expressed as the number of particles that returned

to the particle release sites during the competent period against the number of particles released

supply from outside of the bay, the ecological

investiga-tions of H discus hannai in this area and the connectivity

among spawning groups need to be investigated on a larger

scale Localized dispersal of abalone larvae has been

sug-gested in some studies [20–22], and the ratio of

descen-dants of artificial seeds in the refugium would be greater if

the larval dispersal was localized Thus, the relatively wide

larval dispersal seems to be the main factor causing the low

ratio of descendants of artificial seeds in the refugium

The larval dispersal simulations indicate that most of the

larvae from the refugium are likely to be transported

out-side the refugium and bay before the period of settlement

competency During the competent period, the proportions

of particles inside the bay were at least one order higher

than those inside the refugium In addition, the particles

tended to be more concentrated inside the bay than outside

the bay in both the summer and fall simulations (Fig.4)

Self-replenishment can be achieved when a reserve is of a

size that can contain a sufficient larval supply [23] Thus, it

may be more effective in terms of self-replenishment and

reproduction if the refugium were to be expanded to the

scale of the bay The areas inside the bay are not entirely

suitable for larval settlement since some areas have sandy

bottoms and, therefore, artificial substrata favorable for

larval settlement would need to be put in place to enhanceself-recruitment It would be expected that both adultabalone and larvae would be exported from the bay to itsvicinity following the establishment of a larger refugium.However, the environment for larval settlement should becarefully considered It has been suggested that depositedsediments over substrata have negative effects on larvalsettlement in H diversicolor [24] The environment forpost-larvae should also be assessed since mortality is highduring the first few weeks post-settlement [8] The survivaland growth during the postlarval period can be greatlyinfluenced by the food source [25] Food limitation wassuggested as the major factor controlling the postlarvalsurvival of H discus hannai in Miyagi prefecture [26].Caution is needed when weighing these factors since theycould influence the success of the refugium to a greatextent

The model results show that there is very little ence in terms of self-recruitment and larval retentionamong locations at the head of the bay As these sites didnot show any major great differences in their suitability as

differ-a ldiffer-arvdiffer-al source, the estdiffer-ablishment of new refugidiffer-a differ-at thehead of the bay could be expected to be as effective as thecurrent refugium

Quantitative estimates by models can be useful forfishery management, but their limitations have to be kept inmind Observational data are usually scarce in abalonefishery grounds, and Oshoro Bay is not an exception Onlytwo periods during the spawning season were covered inthis study due to the limited availability of observationaldata on such parameters as current, water temperature, andsalinity for boundary conditions in the hydrodynamicsimulations Thus, additional observations may be neces-sary to improve our hydrodynamic model Since larvalswimming behavior and mortality are still unclear forabalone species in the natural environment, these were notincorporated into the model, and the simulations were keptsimple Although the modeling approaches are subject tothese limitations, our models enabled us to estimate thelarval supply and to quantitatively assess harvest refugia in

a small bay

Acknowledgments We are grateful to Y Machiguchi for providing the bathymetry data We are also grateful to K Kawamoto for boat operation.

References

1 Kawamura T, Takami H, Saido T (2002) Early life ecology of abalone Haliotis discus hannai in relation to recruitment fluctu- ations Fish Sci 68[Suppl 1]:230–233

2 Seki T, Sano M (1998) An ecological basis for the restoration of abalone populations (in Japanese with English abstract) Bull Tohoku Natl Fish Res Inst 60:23–40

Fig 7 Proportion of particles, expressed as a percentage of total

number of released particles in the bay during the competent period in

the summer (left) and fall (right) simulations The particles were

released at HR, H1, H2, and H3 (see Fig 1 for details) The

proportions are the daily average

3 Duran LR, Castilla JC (1989) Variation and persistence of the

middle rocky intertidal community of central Chile, with and

without human harvesting Mar Biol 103:555–562

4 Russ GR, Alcala AC (2003) Marine reserves: rates and patterns

of recovery and decline of predatory fish, 1983–2000 Ecol Appl

13:1553–1565

5 Babcock RC, Kelly S, Shears NT, Walker JW, Willis TJ (1999)

Changes in community structure in temperate marine reserves.

Mar Ecol Prog Ser 189:125–134

6 Kelly S, Scott D, MacDiarmid AB, Babcock RC (2000) Spiny

lobster, Jasus edwardsii, recovery in New Zealand marine

reserves Biol Conserv 92:359–369

7 Allison GW, Lubchenco J, Carr MH (1998) Marine reserves are

necessary but not sufficient for marine conservation Ecol Appl

8:S79–S92

8 Sasaki R, Shepherd SA (1995) Larval dispersal and recruitment

of Haliotis discus hannai and Tegula spp on Miyagi coasts,

Japan Mar Freshw Res 46:519–529

9 Babcock R, Keesing J (1999) Fertilization biology of the abalone

Haliotis laevigata: laboratory and field studies Can J Fish Aquat

Sci 56:1668–1678

10 Hoshikawa H, Takahashi K, Tsuda F, Machiguchi Y (2006)

Effect of adult density on 0? juvenile density of the abalone

Haliotis discus hannai (in Japanese with English abstract) Bull

Fish Res Agency Suppl 5:119–126

11 Hoshikawa H, Hara M (2008) Contribution of the released

abalone, Haliotis discus hannai, to the reproduction of a field

population (in Japanese) Kaiyo Monthly 40:534–537

12 Miyake Y, Kimura S, Kawamura T, Horii T, Kurogi H, Kitagawa

T (2009) Simulating larval dispersal processes for abalone using a

coupled particle-tracking and hydrodynamic model: implications

for refugium design Mar Ecol Prog Ser 387:205–222

13 Hydraulics Delft (2008) Delft3D-FLOW user manual version

3.14 Deltares, Delft

14 Warner JC, Geyer WR, Lerczak JA (2005) Numerical modeling

of an estuary: a comprehensive skill assessment J Geophys Res

110:C05001

15 Delft Hydraulics (2007) Delft3D-PART user manual version

2.13 WL|Delft Hydraulics, Delft

16 Seki T, Kan-no H (1977) Synchronized control of early life in the abalone, Haliotis discus hannai INO, Haliotidae, Gastropoda (in Japanese with English abstract) Bull Tohoku Reg Fish Res Lab 38:143–153

17 Takami H, Oshino A, Sasaki R, Fukazawa H, Kawamura T (2006) Age determination and estimation of larval period in field caught abalone (Haliotis discus hannai Ino 1953) larvae and newly metamorphosed post-larvae by counts of radular teeth rows J Exp Mar Biol Ecol 328:289–301

18 Kobayashi T, Musashi T, Endo T, Hara M (2007) Estimation of number of spawned eggs in field condition of Pacific abalone Haliotis discus hannai (in Japanese with English abstract) Aquac Sci 55:285–286

19 Hara M, Hoshikawa H (2007) Reproductive impact of stocked abalone in Oshoro Bay (in Japanese) Kaiyo Monthly 39:274–279

20 Prince JD, Sellers TL, Ford WB, Talbot SR (1987) Experimental evidence for limited dispersal of haliotid larvae (genus Haliotis; Mollusca: Gastropoda) J Exp Mar Biol Ecol 106:243–263

21 McShane PE, Black KP, Smith MG (1988) Recruitment cesses in Haliotis rubra (Mollusca: Gastropoda) and regional hydrodynamics in southeastern Australia imply localized dis- persal of larvae J Exp Mar Biol Ecol 124:175–203

pro-22 Prince JD, Sellers TL, Ford WB, Talbot SR (1988) Confirmation

of a relationship between the localized abundance of breeding stock and recruitment for Haliotis rubra Leach (Mollusca: Gastropoda) J Exp Mar Biol Ecol 122:91–104

23 Shanks AL, Grantham BA, Carr MH (2003) Propagule dispersal distance and the size and spacing of marine reserves Ecol Appl 13:S159–S169

24 Onitsuka T, Kawamura T, Ohashi S, Iwanaga S, Horii T, Watanabe Y (2008) Effects of sediments on larval settlement of abalone Haliotis diversicolor J Exp Mar Biol Ecol 365:53–58

25 Kawamura T, Roberts RD, Takami H (1998) A review of the feeding and growth of postlarval abalone J Shellfish Res 17:615– 625

26 Sasaki R, Shepherd SA (2001) Ecology and post-settlement survival of the ezo abalone, Haliotis discus hannai, on Miyagi Coasts, Japan J Shellfish Res 20:619–626

O R I G I N A L A R T I C L E Biology

Phylogenetic analyses in cetacean species of the family

Delphinidae using a short wavelength sensitive opsin gene

sequence

Tomoko Koito•Kaoru Kubokawa•

Shinsuke Tanabe•Nobuyuki Miyazaki

Received: 6 October 2009 / Accepted: 12 March 2010 / Published online: 1 May 2010

Ó The Japanese Society of Fisheries Science 2010

Abstract The short wavelength sensitive (SWS) opsin

gene is expected to contain informative sites for

under-standing the speciation of the family Delphinidae, because

it is not functional in cetaceans We determined partial

SWS gene sequences from 15 delphinid species of 12

genera and from harbor porpoise for comparison We found

a 39-bp insertion that was shared by six species (the

insertion group: Delphinus delphis, Delphinus capensis,

Stenella longirostris, Stenella coeruleoalba, Lagenodelphis

hosei, and Sousa chinensis) and common base substitutions

shared by eight species (Stenella frontalis, Tursiops

truncatus, and six species of the insertion group) As these

insertions and substitutions are not found in the other seven

delphinids or in the cloven-hoofed mammals (which are

close to cetaceans), it is suggested that these eight species

are more closely related to each other than to the other

species This hypothesis is supported by phylogenetic

analyses The eight species with the substitutions formed a

clade containing two sister clades, one consisting of the

insertion group and the other consisting of the two otherspecies, in both neighbor-joining and Bayes analyses.Phylogenetic analyses also showed that Lissodelphisborealis and Lagenorhynchus obliquidens are closelyrelated and that their common ancestor diverged from theothers at an early stage of delphinid evolution

Keywords Delphinidae
Short wavelengthsensitive (SWS) gene
Insertion
Phylogenetic analysis

IntroductionCetaceans have a unique evolutional history They haveshifted their habitat from land to the sea [1,2], and havechanged their morphological and physiological systems,including five senses (touch, taste, hearing, eyesight andsmell), to adapt to the aquatic environment We are inter-ested in changes in the visual systems of cetaceans Mostland mammals have color vision, which is advantageous in

an environment where light intensity varies irregularlyspatially and temporally, such as in shallow water andforests However, in the ocean, visibility is less than onland because of the decrease in light intensity with depth.Moreover, toothed whales—major members of the ceta-ceans—have developed echolocation, which is a uniquefeature of bats and cetaceans, for foraging and communi-cation, in addition to their five other senses [3 5] Thus, thedependency of cetaceans on vision is likely to be lowerthan for land mammals, although some whales have beenreported to hunt their prey by sight alone [1]

All of the mammals studied to date have visual pigments

in rod cells, which are referred to as rhodopsin, as well aspigments in cone cells, which are often called cone pigments

or color visual pigments Each visual pigment consists of an

T Koito

Graduate School of Frontier Science, The University of Tokyo,

Kashiwa, Chiba 277-8561, Japan

T Koito
K Kubokawa
N Miyazaki

Ocean Research Institute, The University of Tokyo,

1-15-1 Minamidai, Nakano, Tokyo 164-8639, Japan

S Tanabe

Center for Marine Environmental Studies, Ehime University,

Matsuyama, Ehime 790-8577, Japan

Present Address:

T Koito ( &)

College of Bioresource Sciences, Nihon University,

1866 Kameino, Fujisawa, Kanagawa 252-0880, Japan

e-mail: koito.tomoko@nihon-u.ac.jp

Fish Sci (2010) 76:571–576

DOI 10.1007/s12562-010-0241-7

integral membrane protein, opsin, and a chromophore, either

11-cis-retinal or 11-cis-3,4-dehydroretinal [6,7] Most

ter-restrial mammals have dichromatic color vision based on

two types of spectrally different visual pigments: short

wavelength sensitive (SWS) cone and either long

wave-length sensitive (LWS) cone or middle wavewave-length sensitive

(MWS) cone visual pigments [8] It is also known that

humans and some primates are cone trichromats [9, 10]

Among these three pigments, SWS cone pigment has been

studied extensively, because it absorbs ultraviolet (UV) in

some animals and UV vision is known to be important for

foraging, communication and sexual selection [11] In

pre-vious studies, the opsin genes of the bottlenose dolphin and

common dolphin were cloned and characterized [12–15]

Genes of a rod class, a LWS cone class and a SWS cone class

were found, and the former two were demonstrated to be

functional However, as the SWS cone opsin gene is not

expressed in vivo and its sequence contained frame shift

mutation, it was concluded that the bottlenose dolphin SWS

gene is a pseudogene Moreover, Levenson and Dizon [16]

also showed that the SWS gene is a pseudogene in all

cetacean species analyzed, although hippopotamus, which

has a close relationship to the cetaceans, has a functional

SWS gene Thus, it was proposed that the SWS gene lost its

function through nucleotide substitution during the early

stages of the evolution of modern cetaceans

Although some molecular phylogenetic relationships

among cetaceans have been studied, mostly using

mitochondrial DNA (mtDNA) regions, species of the familyDelphinidae have not been analyzed yet, except for reports

by LeDuc et al [17] and May-Callado and Agnarsson [18]using the mtDNA cytochrome b gene Delphinidae speciesdiverged around 10 million years ago within a short timeperiod [1,19], and the degree of differentiation is relativelylow [20] SWS genes have also been used for phylogeneticanalysis of cetaceans [16] It was successfully used todetermine the general evolution of cetaceans, but the phy-logenetic relationships between species within the familyDelphinidae were not analyzed using SWS

In this study, we isolated partial SWS gene sequencesfrom 15 species in 12 genera of the family Delphinidae.Specific mutations and phylogenetic analyses were char-acterized in order to elucidate the relationship betweenSWS sequence differentiation and the speciation of thefamily Delphinidae

Materials and methodsSamples

Frozen muscle, liver and blubber tissues of 15 species fromthe family Delphinidae and the harbor porpoise, asdescribed in Table 1, were obtained from the Environ-mental Specimen Bank (es-BANK) at the Center forMarine Environmental Studies (CEMS), Ehime University

Table 1 List of samples used in this study

a Accession numbers of the sequences determined in this study

In addition, pig, cow and river hippopotamus sequences

were obtained in order to compare them with the

Del-phinidae sequences (Table1)

DNA amplification and sequencing

Genomic DNA was extracted using an auto nucleic acid

extraction machine (Quickgene 800) and a DNA extraction

kit (Fuji Film, Tokyo, Japan)

Oligonucleotide primers for PCR and sequencing were

designed in the regions of the SWS genes in which

sequences were conserved among the river hippopotamus,

common minke whale, blue whale, humpback whale,

bow-head whale, pygmy right whale, Indus river dolphin,

Sow-erby’s beaked whale, Amazon river dolphin, Franciscana,

bottlenose dolphin, beluga, Vaquita and harbor porpoise

(DDBJ/EMBL/GenBank accession numbers: AF54583–

AF54591, AF54594–AF54596 and U92557) One forward

primer, SWS-F2 (ATGAGCAAGATGTCAGAGGA),

cor-responds to the N-terminus of the coding region

Two reverse primers, SWS-R2 (CTCAAAGACCAAGA

AGGC) and SWS-R3 (CAGTGTAGCAGAAAATGAA

GA), amplify approximately 700 and 1300 bp fragments,

respectively, with SWS-F2

PCR was performed using ExTaq (TaKaRa Bio, Otsu,

Japan) under the following conditions: 1 ll of 10–100 lg

genomic DNA (approximately 10–100 lg) was amplified

in a 25 ll reaction mixture containing 0.08 lM dNTP,

0.02 lM forward and reverse primers, 2.5 ll 109 ExTaq

buffer and 0.3125 U of ExTaq DNA polymerase (TaKaRa

Bio, Otsu, Japan) Amplification was achieved through 40

cycles of denaturing at 95°C for 10 s, and annealing at

60°C for 15 s and 72°C for 1 min PCR products were

analyzed by size with ethidium bromide stained

agarose-gel electrophoresis Excess primers and nucleotides were

removed from the PCR products by treatment with

Exo-nuclease I (ExoI) and shrimp alkaline phosphatase (SAP)

After the addition of 5 U of ExoI and 0.5 U of SAP, the

PCR product was incubated at 37°C for 45 min and 85°C

for 15 min The 5 ll reaction mixture for direct sequencing

containing 0.4 ll BigDye ver 3.0 terminator premix, a

final concentration of 3.5 lM of sequence primer, and

1.0 ll of purified PCR products was subjected to 25 cycles

of denaturation at 95°C for 10 s, annealing at 50°C for 5 s,

and extension at 60°C for 4 min Sequences were

deter-mined with an automated sequencer (Applied Biosystems,

model 3130 xl)

Phylogenetic analysis

Sequences were aligned using the Clustal W program in

MEGA ver 3.1 [21] to detect insertion, deletion and base

substitution

The neighbor-joining tree was constructed using theNEIGHBOR program of PHYLIP 3.62 [22] after calculatingthe genetic distances by Kimura’s two-parameter model Thetransition/transversion ratio was set at 2.0 in the DNADISTprogram of PHYLIP Bootstrap replications were performed

1000 times by the SEQBOOT program Finally, a consensustree was constructed in the CONSENSE program

Bayesian analysis was performed using MrBayes ver.3.1.2 [23] with the following settings: a maximum likeli-hood model of six substitution types (‘‘nst = 6’’), with basefrequencies estimated from the data, and rate variationmodeling across sites using a c-distribution (rates =

‘‘invgamma’’) A Markov chain Monte Carlo search was runwith four chains for 310,000 generations (repeated threetimes), sampling the Markov chain every 100 generations.The sample points of the first 3100 generations, after whichthe chain reached stationarity, were discarded as ‘‘burn-in.’’

ResultsSWS gene sequences and number of substitution sitesPartial SWS gene sequences containing 1021–1167 bpwere obtained from all 16 species, and 1063 positions werealigned Nucleotide substitutions were found at 82 sites,including gaps, and these included 27 sites of transitions, 8sites of transversions, one site with a three-base change,and 46 sites of indels Interspecies substitutions withinDelphinidae included 96 sites, corresponding to 9.0% of allnucleotides

Specific insertion and substitutions

A specific 39-bp sequence (see Fig.1, nt 244–282) wasfound in six species, Lagenodelphis hosei (L hos), Sousachinensis (S chi), Stenella longirostris (S lon), Stenellacoeruleoalba (S coe), Delphinus capensis (D cap) andDelphinus delphis (D del), but was absent in the other ninedelphinid species examined in this study (Fig.1) Wehereafter refer to the six species with this insertion as theinsertion group, and the others as the non-insertion group.The insertion was also absent in cloven-hoofed mammals,pig (Sus scrofa domesticus, S scr), cow (Bos taurus, B.tau) and river hippopotamus (Hippopotamus amphibious,

H amp), which is considered to be the closest extant ative of cetaceans, and possesses functional SWS genes(Fig.1) The insertion sequence consists of 39 bp andcontains three repeats of the sequence CATGGATACTTTGT It is interesting that this sequence is alsofound at the position 287–300, which is conserved by allspecies sequenced in this study and river hippopotamus(Fig.1) In addition, two nucleotide mutations are common

to all species in the insertion group, and to two species in

the non-insertion group, Tursiops truncatus (T tru) and

Stenella frontalis (S fro) (Fig.2)

Phylogenetic analyses

Phylogenetic trees were constructed using the 1063 aligned

positions, with Phocoene phocoena (P pho) used as the out

group Neighbor-joining and Bayesian trees showed similar

topologies and contained two clades (Fig.3): Clade Iconsisted of L bor and L obl, and Clade II contained T.tru, S fro and the six species from the insertion group InClade II, the insertion group and T tru/S fro formed sistersubclades These clades are supported by a high bootstrapvalue or posterior probability The positions of species thatwere not included in the two clades differed between thetrees In addition, long-rostrum species like Tursiopstruncatus and short-rostrum species Pseudorca crassidens,

Fig 1 A comparison of partial

sequences of the short

wavelength sensitive (SWS)

opsin gene of cow (B tau), river

hippopotamus (H amp), pig (S.

scr), Phocoena phocoena (P.

pho) and 15 species of

Delphinidae Six species of

Delphinidae share the repetitive

insertion sequence The

repeated motif was found in the

SWS sequences of cetaceans

and river hippopotamus

Fig 2 Nucleotide sequences of

a portion of the short

wavelength (SWS) gene in 15

species of Delphinidae and

harbor porpoise Fixed mutation

sites are indicated by gray

boxes Fixed mutations were

seen in the insertion group,

Turiops truncatus (T tru) and

Stenella frontalis (S fro)

Globicephala macrorhynchus and Pepnocephala electra

did not separate out in the two trees

Discussion

The insertion group is a recently diverged group

In the present study, we isolated partial sequences of the SWS

gene from 15 delphinid species A specific insertion sequence

was found in Lagenodelphis hosei, Sousa chinensis, Stenella

longirostris, Stenella coeruleoalba, Delphinus capensis and

Delphinus delphis (Fig.1) As this insertion is absent in other

species of Delphinidae, P phocoena of Phocoenidae, and

Artiodactyla (pig, cow and river hippopotamus), it is

sug-gested that the insertion occurred after the divergence of the

common ancestor of the six species Moreover, two

nucleo-tide mutations are common to members of the insertion group

and two species in the non-insertion group (Fig.2) Thus, it is

suggested that the eight species have a common ancestor, in

which these mutations occurred

It is interesting that the repeat motif was found in

another part of the SWS sequence (nt 287–300, see Fig.1),

and that it was conserved among all of the delphinid

species examined and the river hippopotamus, which lacks

the insertion sequence A similar motif was found in pig

and cow, but the motif contained base substitutions

(YATGGGTACTTCGT) (Fig.1)

The evidence that the river hippopotamus shares the

same motif with Delphinidae supports the close

relation-ship between cetaceans and Hippopotamus, as proposed by

previous studies [24–26]

Stenella frontalis is closer to Tursiops truncatus than to

Stenella species in the insertion group

The result that Stenella frontalis is closer to Tursiops

truncatus than to the two Stenella species included in the

insertion group is not consistent with the taxonomy based

on morphological characteristics Although the reason forthis inconsistency is unknown, it is noteworthy that Perrin

et al [27] and Perrin [28] pointed out that Stenella frontalis

is similar to Tursiops truncatus based on eye stripe andblowhole stripe morphology In addition, the two generaStenella and Tursiops were not separated clearly by phy-logenetic analyses using the cytochrome b gene [17,18]

Speciation of DelphinidaeThe topology of the phylogenetic trees indicates that thecommon ancestor of Clade I diverged from the others at anearly stage of delphinid evolution, that the members ofClade II are more closely related to each other than to theother species in Delphinidae, and that within Clade II, thesix species of the insertion group diverged from T tru/S.fro (Fig.3) Clade II corresponds to the eight speciescontaining two base mutations described above, and the sixspecies of the insertion group contain a common insertionsequence; i.e., the positions of Clade II and its subcladesare supported by sequence mutations As gaps wereexcluded from phylogenetic analyses, the results of thephylogenetic analyses and the existence of the specificinsertion are independent

Although the delphinid SWS gene exhibited a lowsubstitution rate for a pseudogene, we found that it con-tained informative substitutions, including the 39-bpinsertion Thus, we propose that it is a potential target forelucidating phylogenetic relationships between delphinidspecies, top predators in marine ecosystems

Acknowledgments We would like to thank Noriko Tsunehiro and Masahiro Kunimoto of the Environmental Specimen Bank (es- BANK) at the Center for Marine Environmental Studies (CEMS), Ehime University, for helping with tissue handling We are indebted

to Drs Takanobu Mizuta and Yukiko Tando of the Ocean Research Institute, University of Tokyo for experimental support We also thank Dr Koji Inoue for comments on the manuscript This work was

1.00 1.00

1.00 1.00

Clade I

Clade I

Fig 3 Phylogenetic trees

calculated by a the

neighbor-joining method and b the

Bayesian method Bootstrap

values and posterior

probabilities are shown at nodes

unless the value is less than 60%

or 0.60, respectively The

insertion group was separated

from the non-insertion group,

and this separation was

supported by a high bootstrap

value or posterior probability

funded by a Sasakawa Scientific Research Grant from The Japan

Science Society.

References

1 Gaskin DE (1982) The ecology of whales and dolphins

Heine-mann, London

2 Berta A, Sumich JL (1999) Marine mammals: evolutionary

biology Academic Press, London

3 Aldridge HDJN, Rautenbach IL (1987) Morphology,

echoloca-tion and resource partiecholoca-tioning in insectivorous bats J Anim Ecol

56:763–778

4 Neuweiler G (1989) Foraging ecology and audition in

echolo-cating bats Trends Ecol Evol 4:160–166

5 Ketten DR (1997) Structure and function in whale ears

Bio-acoustics 8:103–135

6 Sakmar TP, Fahmy K (1996) Properties and photoactivity of

rhodopsin mutants Israel J Chem 35:325–337

7 Yokoyama S (2000) Molecular evolution of vertebrate visual

pigments Prog Retin Eye Res 19:385–419

8 Peichl L (2005) Diversity of mammalian photoreceptor

proper-ties: adaptations to habitat and lifestyle? Anat Rec Part A

287:1001–1012

9 Jacobs GH (1993) The distribution and nature of colour vision

among the mammals Biol Rev 68:413–471

10 Regan BC, Julliot C, Simmen B, Vienot F, Charles-Dominique P,

Mollon JD (2001) Fruits, foliage and the evolution of primate

colour vision Philos Trans R Soc Lond B 356:229–283

11 O ¨ deen A, Ha˚stad O (2003) Complex distribution of avian color

vision systems revealed by sequencing the SWS1 opsin from total

RNA Mol Biol Evol 20:855–861

12 Fasick JI, Cronin TW, Hunt DM, Robinson PR (1998) The visual

pigments of the bottlenose dolphin (Tursiops truncatus) Vis

Neurosci 15:643–651

13 Fasick JI, Robinson PR (1998) Mechanism of spectral tuning in

the dolphin visual pigments Biochemistry 37:433–438

14 Peichl L, Behrmann G, Kroger RHH (2001) For whales and seals

the ocean is not blue: a visual pigment loss in marine mammals.

17 LeDuc RG, Perrin WF, Dizon AE (1999) Phylogenetic ships among the delphinid cetaceans based on full cytochrome b sequences Mar Mamm Sci 15:619–648

relation-18 May-Callado L, Agnarsson I (2006) Cytochrome b and Bayesian inference of whale phylogeny Mol Phylogenet Evol 38:344–354

19 Fordyce RE (1994) The evolutionary history of whales and phins Annu Rev Earth Planet Sci 22:419–455

dol-20 Kingston SE, Rosel PE (2004) Genetic differentiation among recently diverged delphinid taxa determined using AFLP mark- ers J Hered 95:1–10

21 Kumar S, Tamura K, Nei M (2004) MEGA3: integrated software for molecular evolutionary genetics analysis and sequence alignment Brief Bioinform 5:150–163

22 Felsenstein J (1993) PHYLIP (Phylogeny Inference Package) 3.5c University of Washington, Seattle

23 Huelsenbeck JP, Ronquist F (2001) MRBAYES: Bayesian inference of phylogenetic trees Bioinfomatics 17:754–755

24 Irwin DM, Arnason U (1994) Cytochrome b gene of marine mammals: phylogeny and evolution J Mamm Evol 2:37–55

25 Nikaido M, Rooney AP, Okada N (1999) Phylogenetic ships among cetartiodactyls based on insertions of short and long interpersed elements: hippopotamuses are the closest extant rel- atives of whales Proc Natl Acad Sci USA 96:10261–10266

relation-26 Arnason U, Gullberg A, Janke A (2004) Mitogenomic analyses provide new insights into cetacean origin and evolution Gene 333:27–34

27 Perrin WF, Mitchell ED, Mead JG, Caldwell DK, Caldwell MC, Van Bree PJH, Dawbin WH (1987) Revision of the spotted dolphins, Stenella spp Mar Mamm Sci 3:99–170

28 Perrin WF (1997) Development and homologies of head stripes in the delphinoid cetaceans Mar Mamm Sci 13:1–43

O R I G I N A L A R T I C L E Biology

Growth, sex ratio, and maturation rate with age in the blackspot

tuskfish Choerodon schoenleinii in waters off Okinawa Island,

southwestern Japan

Akihiko Ebisawa•Kiyoaki Kanashiro•

Toshihiko Kiyan

Received: 21 December 2009 / Accepted: 1 April 2010 / Published online: 15 May 2010

Ó The Japanese Society of Fisheries Science 2010

Abstract The growth, sex ratio with age, and age at

sexual maturation were determined based on sectioned

otoliths in 257 specimens of the blackspot tuskfish

Choerodon schoenleinii collected in waters off Ryukyu

Island Opaque rings observed by reflected light in the

sectioned otoliths were found to form once a year from

January to July The three growth parameters of the von

Bertalanffy growth equation were L?= 68.1 (cm),

k = 0.263, and t0= -0.023 (year) The age at which the

sex ratio reached 50% by sexual transition was about

6.15 years, and the age at which 50% of females were

sexually mature was approximately 2 years The oldest

specimen among the samples was 17 years old

Keywords Choerodon schoenleinii Otolith  Growth 

Sex ratio with age Okinawa

Introduction

The blackspot tuskfish Choerodon schoenleinii is a large

labrid species distributed in the tropical western Pacific

region [1] Because the species is commercially important,many studies on its settlement area, growth, and feedingbehavior from the larval to juvenile stages [2, 3], repro-ductive cycle, sexual maturation, and sexual transitionbased on body size [4] have been carried out in Okinawa,southwestern Japan Because of the species’s importance,stock enhancement is being tried [5] Studies of homeranges and diel movement patterns [6], and otolith micro-structure [7] using hatchery-reared individuals have alsobeen carried out According to these studies, the speciesexhibits protogynous hermaphroditism in accordance withchanges in body color, like many other labrid species.Juveniles of the species inhabit sea-grass beds in theinnermost areas of relatively large gulfs The major habi-tats of the species in its adult stages are also restricted tothe adjacent regions [8] The major habitat of the Japanesespecies is very similar to that of the species in Australia,and extensive research into reproductive biology, sexualtransition, growth [9], and habitat portioning [10] have alsobeen carried out at Shark Bay, a large gulf on the westcoast of Australia, where only recreational fishing of thespeces has been conducted [9]

The species is caught primarily by night spire fishing,which captures the target after confirmation of the speciesand body size The characteristics of the fishing gear itselfallow only those fish within the size limit of the targetspecies to be captured A local rule restricting the capture

of this species [heavier than 1 kg body weight; mately 36 cm total length (LT)] has been established at thenorthern area of Okinawa Island The annual catch of thespecies in Okinawa prefecture in the past two decades hasfluctuated between 20 and 40 mt, however, no decreasingtrend has been observed (Ebisawa, unpublished data) Thepeaked body size in catch at about 28 cm LT, in areaswhere the body-size restrictions have not been introduced,

approxi-A Ebisawa ( &)  K Kanashiro  T Kiyan

Okinawa Prefectural Fisheries and Ocean Research Center,

1-3-1 Nishizaki, Itoman, Okinawa 901-0305, Japan

e-mail: ebisawaa@pref.okinawa.lg.jp

Present Address:

K Kanashiro

Okinawa Prefectural Sea Farming Center,

853-1 Oohama, Motobu, Okinawa 905-0212, Japan

seems to be too smaller for the species which attains at

about 70 cm LTat the maximum; thus, the stock is

deter-mined to be non-rationally used even decrease in catch is

not observed (Ebisawa, unpublished data)

For stock assessments or to determine the effect of stock

management, analyses based on the age composition in the

catch are necessary, and it is important to carry out growth

studies and to examine the relationship between sexual

maturation/transition and age The present study reveals the

growth of the species based on sectioned otoliths and

analyzes the relationship between age and sexual

matura-tion/transition, primarily using specimens from the same

source as a previous study [4] with additional specimens

collected thereafter The results are compared to those of

Shark Bay populations in order to elucidate the biological

profiles of the species under different environments

Materials and methods

The specimens for this study were 247 individuals among

289 specimens collected from 1986 to 1990 [4], and 35

newly collected specimens gathered from 2000 to 2006

The major fishing sites of the specimens were Nakagusuku

Bay, Kin Bay, and the Haneji area of Okinawa Island

(Fig.1) Most of the specimens were purchased from

commercial fishermen conducting night spire fishing A

pair of otoliths (sagittae) was removed from each

individ-ual after measurement of the total length (LT) and body

weight (WB), and determination of the sex by the external

appearance of the gonad The gonad was later prepared for

histological observation in order to determine both the sex

and the stage of ovarian maturation; these results have

already been reported by Ebisawa et al [4] Sex ratio (RS)

at each age was calculated based on the number of females

in the total, which included female, male, and

hermaphrodite individuals in each same-integer age group.The rate of ovarian maturity (ROM) in each age group wasdetermined based on the number of mature females amongwhole females obtained during the most active spawningperiod, which is from February to May [4] Maturity stagesfrom early peri-nucleolus to yolk vesicle were defined asimmature, and those from yolk globule to atretic weredefined as mature The smallest two specimens, otolithobservations of which appear in the ‘‘Results’’ section, areexcluded from these analyses owing to the lack of histo-logical observations of their gonads

Preparation and observation of otolithsEach otolith was embedded in epoxy resin, transverselysectioned about 450 lm thick with the core included,mounted on slide glass with a medium (Eukitt; O Kindler)and covered with a cover glass Reflected light observationwith a binocular microscope revealed translucent andopaque zones alternating around a central opaque area(Fig.2a) The opaque zones outside of the central opaquearea were counted as growth rings; the observer wasblinded to the details of the specimen (body size, monthcollected, sex) Whether the outermost edge was opaque ortranslucent was also determined Otolith pictures wererecorded at the first observation using a Polaroid PDMC-Iewith 160091200 pixels and 24-bit color A second obser-vation was carried out by the same reader based on theotolith picture without the reader being provided with theprevious data When the number of growth rings matchedbetween the first and second readings, no further readingswere carried out; when the numbers disagreed, two addi-tional readings from the otolith picture were performed.When a total of three readings agreed, the number ofgrowth rings was considered to be determined; if threeconcurring readings were not obtained, the otolith wasexcluded from later analysis Measurements of the otolith

Fig 1 Map of the Okinawa Islands with 200-m depth contour, and

the locations where the specimens were collected

Fig 2 Sectioned otolith of Choerodon schoenleinii from a a 46.2 cm

LTspecimen collected January 11, 2002; b the smallest specimen, 9.2 cm LT, collected January 1, 2005; and c a 10.6 cm LTspecimen collected May 4, 2005 All magnifications are equal, as shown by the 1-mm scale bar The innermost four growth rings in a are identified

by a line and a number e Outer edge of the otolith, D1 point at which the marginal growth index was obtained

growth ring for the analysis of the marginal growth index

(MGI) were conducted at the second reading from otolith

pictures MGI was calculated as X0/X1 where X0 is the

distance from the start of the outermost growth ring to the

outer edge of the otolith (distance between lines ‘‘4’’ and

‘‘e’’ in Fig.2a) and X1 is the distance from the start of the

inner growth ring to the outer edge of the previous

trans-lucent zone (distance between lines ‘‘3’’ and ‘‘4’’ in

Fig.2a) at area D1 of the otolith In case these

measure-ments at the area D1 were difficult, they were carried out at

area D2

The age of each specimen is given as follows Birth

month was defined as February, which is the start of the

most active spawning period of the species in the Okinawa

area [4] The decimal part of the age is the proportion of a

year from the birth month to the month in which the

specimen was collected Details of the growth ring

for-mation are explained further in the ‘‘Results’’ section, but it

should be noted that the birth month is included in the

period at which growth ring is formed Thus, the integer

part of the age is given as the number of growth rings

minus one for specimens that had a newly formed growth

ring but that were collected before the birth month (Fig.3b,

c; dots encircled by solid line) and is given as the number

of growth rings plus one for those that had not yet started to

form new growth rings but were collected after the birth

month (Fig.3d; dots encircled by broken line) In all other

cases, the integer part of the age is given as the number of

growth rings Growth parameters in the von Bertalanffy

growth equation were estimated by nonlinear regression

(SPSS for Windows, release 7.5.2, SPSS) in the data sets of

age and LTfor the specimens

Results

Period of growth ring formation

Monthly changes in both MGI and the condition of the

otolith edge were as follows Otoliths with translucent

edges in the single-growth-ring group were obtained

con-tinuously from July to January; their MGI values were the

lowest in July with continual increases thereafter (Fig.3a)

In the two-ring group, otoliths with a growth ring on their

edges were obtained from January to July, and those with

translucent edges from June to January (Fig.3b) Increases

in MGI values were continuous from the minimum values

of the otoliths with growth rings at their edges collected in

January to the maximum values of those with translucent

edges collected in December and January, in the two-ring

group Although monthly changes in groups with three and

more rings were almost identical, earlier emergences of

otoliths with growth rings at their edges collected in

November and in December in the three-ring group werethe exception Therefore, all otoliths with low MGI valuescollected from January to July had a growth ring at theiredge, while the edges of the otoliths collected from August

to October showed a translucent zone Thus, the growthring was determined to be formed annually from aboutJanuary to July

In one of the smallest otoliths (9.2 cm LT), which wasobtained in January 2005, about two-thirds of the centralarea was opaque (Fig.2b), and in another specimen

0 0.1 0.2 0.3 0.4 0.5

0 0.1 0.2 0.3 0.4 0.5

0 0.1 0.2 0.3 0.4 0.5

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

d

Fig 3 Monthly changes in marginal growth index (MGI) in a the single-growth-ring group; b the two-ring group; c the three-ring group; and d a mixed group with four and more growth rings A solid circle indicates an otolith with a growth ring at its edge; an open circle indicates one with a translucent edge The significance of the solid and broken encircling lines is explained in the text

(10.6 cm LT) obtained in May 2005, half of the central area

was opaque while the outer areas were translucent

(Fig.2c)

Determination of age and growth parameters

Specimens whose integer ages differ from the number of

growth rings are circled in Fig.3 Those whose integer age

is the same as the number of growth rings minus one are

circled by a solid line, including 1 specimen with two rings

collected in January (MGI = 0.15: Fig.3b), 11 specimens

with three rings gathered from November to January

(MGI \ 0.3: Fig.3c), and 6 specimens with four and more

rings collected in January (MGI \ 0.3: Fig.3d) Those

whose integer age is the same as the number of growth

rings plus one are circled by a broken line, including eight

specimens with four or more rings collected between

February and April (MGI [ 0.4: Fig.3d) The integer ages

of all other specimens were the same as the number of

growth rings The three parameters of the von Bertalanffy

growth equation were obtained as follows:

L1¼ 68:1 ðcmÞ; k¼ 0:263;

t0¼ 0:023 ðyearÞ ðr2¼ 0:86Þ

Increases in body size were obvious up to 6–7 years of age,

but ceased thereafter Body sizes of males were larger than

those of females in all same-age groups (Fig.4)

Sex ratio and ovarian maturity rate in each age group

Average LT± SD for each age group, sex ratio (RS), and

ovarian maturity rate (ROM) are shown in Table 1 All

individuals aged 1 and 2 years were female

Hermaphro-dite and male individuals appeared at 3 years of age and

older The average LT of males was considerably larger

than that of females RSdecreased to 50% at 6 years of age,

fluctuated from 0 to 75% between 7 and 11 years of age

probably due to the small sample sizes, and decreased to

0 between 12 and 17 years The age at which RSwas 50%was 6.15 years by fitting the logistic equation in RSof eachage class, omitting the extreme values of 0 and 75% of 7and 11 year olds, respectively ROM was 90% in 2 yearolds, reaching 100% in 3 year olds ROMin 1 year olds wasnot obtained because no specimens were obtained duringthe most active spawning period from February to May

DiscussionOpaque zones were found to be formed in otoliths duringthe winter period in Choerodon schoenleinii, which is incontrast to many other species, in which translucent zonesare formed during the winter period [11–19] In a previousstudy using oxytetracycline (OTC) marking, in the con-generic C rubescens in western Australia, opaque zoneformation was determined to occur during the spring andsummer, based on the location of opaque, translucent, andOTC marks in the otoliths, though the minute specificanalytical data for this determination were not provided[20] Another study found that, in the four species of thegenus Choerodon, including C schoenleinii in westernAustralia, MGI values, measured according to a baselinedefined as the outermost zones of successive opaque edges,reach minimum in November/December [9], indicating thattranslucent zone formation begins in the summer, although

Fig 4 The von Bertalanffy growth curve fitted to observed total

length (LT) and age in Choerodon schoenleinii

Table 1 Average LT(cm) ± SD for each sex, sex ratio (RS), rate of ovarian maturity (ROM), number of total specimens in each integer age group (n1), and number of female specimens obtained at the spawning period (n2)

the author states that opaque zone formation takes place in

the summer [9] Opaque zone formation has been

con-firmed to occur during the winter in some Labridae, such as

Thalassoma lunare, which is found in tropical eastern

Australia [21], and during the spring in three species of

Labridae in temperate eastern Australia [22] Thus, the

opposing periods of opaque zone formation as seen in C

schoenleinii are not specific to the labrid species

The first growth ring is determined as the innermost

opaque band outside of the central opaque area in the

present study In the otoliths of the two smallest specimens

obtained in the present study, a 9.2 cm LT specimen

obtained in January shows a translucent zone one-third of

the diameter of the otolith outside of the opaque central

area, and a 10.6 cm LTspecimen obtained in May has a

translucent zone about half of which is outside the central

area Accordingly, the central opaque area seems to be

formed until approximately December in specimens of

about 90 mm LT Settled juveniles in the sea-grass bed

grow from about 20 mm LTon average in May to about

70 mm LT in August, and the larger individuals attain

about 100 mm LTby August/September [2] Therefore, the

smallest specimen in the present study (9.2 cm LT,

obtained in January) clearly belongs to the 0 integer age

group, although it would have reached 1 year of age the

following February The 10.6 cm LTspecimen obtained in

May has to belong to the 1 integer age group even though

the first growth ring was not confirmed, since it was

col-lected after February, the designated birth month of the

species in the present study There were no otoliths with a

growth ring at their edges in the 1 integer age group

col-lected from July to January Thus, the first growth ring

seems to be formed during the relatively short period from

spring to early summer, shortly after the major spawning

period from February to May

The biological parameters of age in the species of the

Okinawa population (hereafter termed Opop) differed

greatly from those of the Shark Bay population [9]

(here-after termed SBpop) The growth of Opop was significantly

faster than that of SBpop, although the maximum size and

maximum ages in the two populations were approximately

equal (Fig.5; Table2) While the body size at which 50%

of females reached ovarian maturity (LOM50) was almost

equivalent between two populations, the age of 50%

ovarian maturity (AOM50) was younger in Opop due to

differences in growth rate (Table2) In addition, the great

difference in female sexual maturation between the two

populations has existed Rate of sexual maturation in

female ranged between 40–60% at 4–8 years of age and

attained 100% at 9 and 10 years of age, although the

sample sizes at the latter two age classes are very small in

the SBpop Therefore, AOM50 (3.45) indicated by

Fairc-lough is obtained along with the determination of 50% of

ovarian maturation at 4–8 years of age as fully mature;thus, about 25% of female sexually mature at 3.45 years ofage Therefore, the AOM50 in the SBpop become muchgreater if the same criteria of AOM50 employed in thepresent study applied In contrast, about 90 and 100% offemales at 2 and 3 years old, respectively, were sexuallymature in Opop The age at which the sex ratio reached50% (ARS50) was lower in Opop, and the number of malesout of the total number of specimens in SBpop was sig-nificantly smaller One of the plausible reasons for theextremely smaller number of males in the SBpop could bebiased sampling, because only a few specimens of largerbody size were collected However, the age of the youngestmale in the SBpop was 7 years old, whereas it was

LT

Age

0 10 20 30 40 50 60 70

0 2 4 6 8 10 12 14 16 18

Opop SBpop

Fig 5 Growth curve of Choerodon schoenleinii Open circles represent the Okinawa population, solid circles the Shark Bay population according to Fairclough [ 9 ]

Table 2 Comparison of biological parameters in Choerodon schoenleinii between the Okinawa population (Opop) and Shark Bay population (SBpop) by Fairclough [ 9 ]

AOM50and LOM50age and total length (LT) at which 50% of females are sexually mature; ARS50and LRS50, age and LTat which sex ratio reaches 50%

extremely younger at 3 years old in Opop (Table2) The

age of emergence of the youngest male was lower under

higher fishing pressure in comparison with the congeneric

C venustus experiencing lower fishing pressure in adjacent

regions [23] On the contrary, in a comparison of the body

size and age of the youngest male between two populations

of the congener C rubescens—one experiencing low

fishing pressure and higher growth due to high productivity

at the area and the other experiencing higher fishing

pres-sure—the body size of the smallest male was smaller at the

area of higher fishing pressure However, the age of the

smallest male was older there because of the slower growth

rate In addition, ARS50 was younger at the lower fishing

pressure area These contradictory findings indicate that the

reason for the sexual and maturational differences between

SBpop and Opop can not yet be determined The reason for

the difference seems to be linked not only to environmental

factors, such as biology, genetics, population density, food

availability, antibiotics, water temperature, and water flow,

but also artificial factors, such as fishing pressure and

methods If the effect of the artificial factors is not small,

management strategies for the species should be developed

by that take these factors into consideration

Assuming a normal distribution around the calculated

body size at age using the given mean square obtained

during the estimation of the growth parameters, the

per-centage falling below the size restriction (\36.0 cm LT) as

currently enforced at the northern area of the Okinawa

Island is 100% at 1 year, 78% at 2 years, and 22% at

3 years In hermaphroditic species, size-selective fishing

makes the proportion represented by the second sex

sig-nificantly smaller compared to dioecious species [24,25]

If the at-sex-change is flexible, the side effects of

size-selective fishing are small [26,27] In many labrid species,

a haremic social structure, in which the disappearance of

the male induces the sexual transition of the apex female

into male in the harem, has been reported [28–30]

Therefore, if this type of social structure exists in C

schoenleinii, size-selective fishing of larger individuals

does not necessarily lead to a shortage of males in the

population The specimens in the present study were

obtained before the introduction of size-selective fishing at

the study site This size restriction is currently conducted at

Haneji, but not yet at Kin and Nakagusuku It is therefore

necessary to obtain growth rate, age, and body size of both

sexual maturation and sexual transition, before and after

the introduction of the body-size-restrictive fishing at Kin

and Nakagusuku areas in order to fully elucidate the effects

of size-selective fishing

Acknowledgments This work was supported in part by the

‘‘Mor-phometric Survey of Important Fishery Resources in Seas Adjacent to

the Okinawa Islands’’ and ‘‘Research on Biological Characters of

Dominant Species of Coral Reefs around the Okinawa Islands,’’ both being conducted by the Japan Fisheries Agency We are grateful to

Mr W Noda of the University of the Ryukyu, who kindly provided small specimens of the species.

References

1 Westneat MW (2001) Labridae Wrasses (also hogfishes, zorfishes, corises and tuskfishes) In: Carpenter KE, Niem VH (eds) FAO species identification guide for fisheries purposes The living marine resources of the western central Pacific, vol 6 Bony fishes part 4 (Labridae to Latimeriidae) estuarine crocodiles sea turtles, sea snakes and marine mammals FAO, Rome, pp 3381– 4218

ra-2 Kanashiro K (1998) Morphology and changes of distribution and food habits with growth of late larvae and juveniles of black-spot tuskfish, Choerodon schoenleinii (Labridae), settled on seagrass beds of Okinawa Island, the Ryukyus (in Japanese with English abstract) Nippon Suisan Gakkaishi 64:427–434

3 Oota I (2007) A study of habitat selection and process of recruitment in Choerodon schoenleinii (in Japanese) Annu Rep Okinawa Fish Ocean Res Cent 68:249–250

4 Ebisawa A, Kanashiro K, Kiyan F, Motonaga F (1995) Aspects of reproduction and sexuality in the black-spot tuskfish Choerodon schoenleinii Jpn J Ichthyol 42:121–130

5 Yoseda K, Asami K, Yamamoto K, Dan S (2005) Current status

on broodstock management and seed production techniques in the black-spot tuskfish (Choerodon schoenleinii) In: Taiwan–Japan international symposium on marine biotechnology and its appli- cation (proceedings) Acta Sinica, Taipei, pp 104–106

6 Kawabata Y, Okuyama J, Asami K, Yoseda K, Arai N (2008) The post-release process of establishing stable home ranges and diel movement patterns of hatchery-reared black-spot tuskfish Choerodon schoneleinii J Fish Biol 73:1770–1782

7 Yamada H, Chimura M, Asami K, Sato T, Kobayashi M, Nanami

A (2009) Otolith development and daily increment formation in laboratory-reared larval and juvenile black-spot tuskfish Choer- odon schoenleinii Fish Sci 75:1141–1146

8 Kanashiro K, Motonaga F, Ebisawa A, Kiyan F (1990) The present condition of Choerodon scholenleinii fishery (in Japa- nese) Annu Rep Okinawa Fish Ocean Res Cent 52:40–48

9 Fairclough D (2005) The biology of four tuskfish species (Choerodon: Labridae) in Western Australia PhD Dissertation, Murdoch University, Perth

10 Fairclough DV, Clarke KR, Valesini FJ, Potter IC (2008) Habitat partitioning by five congeneric and abundant Choerodon species (Labridae) in a large subtropical marine embayment Estuarine Coastal Shelf Sci 7:446–456

11 Loubens G (1980) Biologie de quelques espe`ces de Poissons du lagon Ne´o-Cale´donien III Croissance Cah Indo-pac 2:101–153

12 Fowler AJ (1990) Validation of annual growth increments in the otoliths of a small, tropical coral reef fish Mar Ecol Prog Ser 64:25–35

13 Hart AM, Russ GR (1996) Response of herbivorous fishes to crown-of-thorns starfish Acanthaster planci outbreaks III Age, growth, mortality and maturity indices of Acanthurus nigrofus- cus Mar Ecol Prog Ser 136:25–35

14 Choat JH, Axe LM (1996) Growth and longevity in acanthurid fishes; an analysis of otolith increments Mar Ecol Porg Ser 134:15–26

15 Sadovy Y, Figuerola M, Roma´n A (1992) Age and growth of red hind Epinephelus guttatus in Puerto Rico and St Thomas Fish Bull 90:516–528

16 Shimose T, Tachihara K (2005) Age, growth and maturation of

the blackspot snapper Lutjanus fulviflammus around Okinawa

Island, Japan Fish Sci 71:48–55

17 Ebisawa A, Ozawa T (2009) Life-history traits of eight Lethrinus

species from two local populations in water off the Ryukyu

Islands Fish Sci 75:553–566

18 White BD, Palmer SM (2004) Age, growth, and reproduction of

the red snapper, Lutjanus campechanus, from the Atlantic waters

of the southeastern U.S Bull Mar Sci 75:335–360

19 Fowler AJ (1995) Annulus formation in otolith of coral reef

fish—a review In: Secor DH, Dean JM, Campana SE (eds)

Recent development in fish otolith research University of South

Carolina Press, Columbia, pp 45–63

20 Nardi K, Newman SJ, Moran MJ, Jones GP (2006) Vital

demo-graphic statistics and management of the baldchin groper

(Choerodon rubescens) from the Houtman Abrolhos Islands Mar

Freshwater Res 57:485–496

21 Ackerman J (2004) Geographic variation in size at age of the

coral reef fish, Thalassoma lunare (family: Labridae): a

contri-bution to life history theory PhD Dissertation, James Cook

University, Townsville

22 Morton JK (2007) The ecology of three species of wrass

(Pisces: Labridae) on temperate rocky reefs on New South

Wales, Australia PhD Dissertation, University of Newcastle,

Newcastle

23 Platten JR, Tibbetts IR, Sheaves MJ (2002) The influence of increased line-fishing mortality on the sex ratio and age of sex reversal of the venus tusk fish J Fish Biol 60:301–318

24 McGoven JC, Wyanski DM, Pashuk O, Manooch CS II, Sedberry

GR (1998) Change in the sex ratio and size at maturity of gag, Mycteroperca microlepis, from the Atlantic coast of the south- eastern United States during 1876–1995 Fish Bull 98:797–807

25 Coleman FC, Koenig CC, Collins LA (1996) Reproductive styles

of shallow-water groupers (Pisces: Serranidae) in the eastern Gulf

of Mexico and the consequences of fishing spawning tions Environ Biol Fish 47:129–141

aggrega-26 Allonzo SH, Mangel M (2004) The effects of size-selective fisheries on the stock dynamics of and sperm limitation in sex- changing fish Fish Bull 102:1–13

27 Molloy PP, Goodwin NB, Cote IM, Gage MJG, Reynolds JD (2007) Predicting the effects of exploitation on male-first sex- changing fish Anim Cons 10:30–38

28 Warner RR (1982) Mating systems, sex change and sexual demography in the rainbow wrasse, Thalassoma lucasanum Copeia 1982:653–661

29 Nemtzov SC (1985) Social control of sex change in the Red Sea razorfish Xyrichtys pentadactylus (Teleostei, Labridae) Environ Biol Fish 14:199–211

30 Robertson DR (1972) Social control of sex reversal in a coral-reef fish Science 177:1007–1009

O R I G I N A L A R T I C L E Biology

Natural growth of juveniles of the sea cucumber Apostichopus

japonicus: studying juveniles in the intertidal habitat in Hirao

Bay, eastern Yamaguchi Prefecture, Japan

Yusuke Yamana•Tatsuo Hamano•

Seiji Goshima

Received: 30 September 2009 / Accepted: 2 April 2010 / Published online: 15 May 2010

Ó The Japanese Society of Fisheries Science 2010

Abstract In order to determine the population dynamics

of juvenile Apostichopus japonicus, successive surveys

were conducted in the stony intertidal zone of Hirao Bay,

Seto Inland Sea The juveniles of the green and black types

of this species grew well from January to June Before this

high growing season, zero year old juveniles were too

small to be readily detected by visual observations during

field surveys During the high water temperature season

from August to November, juveniles over 1 year of age

estivated underneath rocks They awakened from their

estivation in November, and it took another month before

they could eat and discharge The juveniles mainly

migrated from the intertidal zone before they reached 3

years of age In the present work, three patterns of juvenile

growth were found It is suggested that these differences in

juvenile growth are mainly caused by differences in their

start times for growth initiation when they are zero yearsold

Keywords Apostichopus japonicus Ecology Growth Juvenile  Sea cucumber

IntroductionThe Japanese sea cucumber Apostichopus japonicus is acommercially important species in Japan, since it is eaten

as dried sea cucumber and as vinegared raw flesh Fisheryresources in Japan are currently overfished, and programsfor conserving and propagating these resources are activelyconducted However, because of a lack of basic knowledgeabout the ecology of these fishery resources, it is consid-ered questionable as to whether some programs will havetheir expected effects

In particular, knowledge about the growth of juveniles

in the field, which is essential when devising seedlingrelease programs, has been the focus of only limitedresearch One of the present authors, Hamano, conducted

an ecological survey on the population dynamics of

A japonicus in the intertidal zone of the western coast ofthe Seto Inland Sea, and reported that its juveniles reachedsizes of about 70 mm within 1 year [1] However, untilrecently there was no accurate method for measuring

A japonicus in the field because of the extreme flexibility

of its body [2], and so his population analysis was what inconclusive Such a problem is one of the reasonswhy there is a shortage of research on this topic

some-However, the main cause of this is the scarcity of suitablestudy sites Under ordinary circumstances, juveniles of

A japonicus are distributed with low density (\1.0 indiv.per m2) in rocky and boulder-strewn areas [3,4], and their

Y Yamana

Graduate School of Fisheries Science, Hokkaido University,

Hakodate, Hokkaido 041-8611, Japan

T Hamano

Department of Applied Aquabiology, National Fisheries

University, Shimonoseki, Yamaguchi 759-6595, Japan

S Goshima

Faculty of Fisheries Sciences, Hokkaido University,

Hakodate, Hokkaido 041-8611, Japan

Present Address:

Y Yamana ( &)

Wakayama Prefectural Museum of Natural History,

Kainan, Wakayama 642-0001, Japan

muted color is rarely noticed [5] Therefore, it has been

difficult to collect the number of individuals required for

population analysis In addition, it has not been easy to

research the growth of this species by tracking released

seedlings in the field, because the percentage of seedlings

found decreases significantly soon after release (see for

example [6])

Recently, however, a quite densely populated habitat

was detected in the intertidal zone of Hirao Bay in the Seto

Inland Sea, eastern Yamaguchi Prefecture, where the body

sizes of juveniles varied over a wide range [4], and so this

population was considered to be suitable for growth

anal-ysis In the present study, in order to elucidate the growth

of juvenile A japonicus in the field, we have conducted a

study of its population dynamics Here, for the first time, its

growth in the field was observed using a new measurement

criterion that allows higher precision and does not require

the death of the animal [2,7]

Materials and methods

The Japanese sea cucumber Apostichopus japonicus is

widely distributed throughout the Pacific Northwest and in

coastal areas throughout Japan [8] It has been

distin-guished into three body color types—green, black and

red—based on its ventral color [9], and recent studies have

revealed that the separate status of red and the other two

color types is valid based on genetic differences [10]

In Japan, except in the northern part, adult A japonicus

estivate; throughout the summer they stop moving to keep

out of the heat [9,11] During this estivation, all feeding and

growth ceases and the alimentary canal becomes atrophied;

however, juveniles under 5 g in body wall weight do not

show these characteristics of estivation [9] It is considered

that the activating water temperature of green adults is

generally below 17.5–19.0°C—growth is observed to

pro-gress below 16.0–17.0°C—and the critical water

tempera-ture at which they definitely enter estivation is 24.5°C [9]

The spawning season is spring to early summer around

the main island of Japan In the eastern part of Yamaguchi

Prefecture, preliminary research on the reproduction of

A japonicus has shown that the main spawning season lasts

from late April to early May [12], and planktonic larvae

require about 2 weeks prior to settlement [12]

In this study, we used the standard length Le [2] as a

measurement criterion for A japonicus This was

calcu-lated as an estimate of the anesthetized body length [13]

using measurements of body length (L) and body breadth

(B) taken from photographs L, which runs along the

midline from the tip of the snout to the end of the tail, and

B, which is measured across the center of the body except

for the papillae and the parapodia, were measured in mm

Le was calculated separately for each color type using thefollowing formulae from Yamana and Hamano [2]:Green color type:

LeðmmÞ ¼ 2:32 þ 2:02ðLBÞ1=2:Black color type:

LeðmmÞ ¼ 1:34 þ 2:12ðLBÞ1=2:The green and black color types of A japonicus [9]occur at the study site used in this work, and these colortypes usually differ in both body color but in geneticproperties [10] In this study, combined data on both colortypes were used for analysis, since there were no cleardifferences in body size or attaching position between thecolor types in the survey area, and intermediately coloredindividuals were often observed

Surveys were conducted during two periods, fromFebruary 2005 to March 2006 and from February 2008 toFebruary 2009 At low spring tide every month, individualswere collected from the fixed survey area in the intertidalzone of Hirao Bay However, the tide level in September

2008 was higher than it was during other months, so thesurvey could not be carried out during that month In thepresent study, we describe the monthly size distribution andconduct a cohort study using the method described below

Study site

In this study, we defined a 20 m2(2 9 10 m) fixed surveyarea in the intertidal zone of Hirao Bay, near the TanaMarine Biological Laboratory of the National FisheriesUniversity, eastern Yamaguchi Prefecture (Fig.1) There is

an artificial rocky shore inside this bay, which is situatedunder the coastal revetment for an arterial road and reported

to support higher densities of juvenile A japonicus than inother parts of the bay [4] It has a narrow intertidal zoneconstructed from ripraps (rubble rock) about 0.3 m3 involume, which pave the revetment to a width of 10 m(Figs 2a,3a) The half of each riprap closest to the land wasburied in the sediment, and they became a muddy sandbottom containing many dead oyster shells (Crassostreagigas), where a wide tide pool appeared at low spring tide.The survey area was located in the same place for everysurvey, along the boundary line between the riprap area andthe sediment on the line 5 m from the margin of the ripraps(Fig.3a), where the tide level was about ?0.4 to 0.2 mabove the datum line

The temperature of the seawater was recorded duringevery survey (Fig.4) The abundance of seaweed was alsorecorded, since this was considered to be an importantfactor that affects the life history of A japonicus [3,4], andthese data can be summarized as follows From February to

June, Sargassum thunbergii and Grateloupia filicina grew

on the ripraps and Ulva pertusa grew on the sediments

These seaweeds grew thickest and formed a dense cover

over the substrate in June (Fig.2b), and declined from July

to August (Fig.2c) The growth of seaweed was restarted

from January to February

The distribution pattern of A japonicus at the study site

can be roughly described as follows Especially the smaller

juveniles of A japonicus (smaller than about 30 mm in Le)

were mainly found around the rhizoids of S thunbergii on

the ripraps (Fig.3b), while some were found around

G filicina Relatively small individuals (smaller than about

100 mm in Le) were found on the ripraps and on U pertusa

in the tide pool (Fig 3c), while larger individuals wereseen on the sediment around the ripraps (Fig.3d) Duringthe estivating season, individuals were found estivating onthe undersides of the ripraps, regardless of their body size(Fig.3e)

In the first survey period in 2005, Typhoon Nabi(Typhoon No 200514, Japanese name) passed just abovewestern Yamaguchi Prefecture on September 6 Its forcewas ‘‘big-sized and extremely strong’’ [14] FromSeptember 3 to 8, the precipitation was 200–300 mm atHirao Bay, and 300–600 mm in the basin of the TabuseRiver that flows into the bay [14] There was severe damage

to fishery resources inside the bay caused by the suddendecrease in salinity (Hirao Branch Office of Japan FisheriesCooperatives of Yamaguchi Prefecture, private note)

Methods

At low spring tide every month, two highly skilledresearchers collected as many A japonicus in the fixedsurvey area as possible Such a method has often been used

to investigate the distributions of sessile and low-mobilityinvertebrates, and Larsson [15] reported that a perfectestimate of echinoderm density was possible if sufficienttime was spent carrying out the survey Individuals wereimmediately released after photographing them in theseawater-filled aquaria along with a scale using a digitalstill camera Body size measurements were obtained fromthe photographs in the laboratory

It had been reported that cultured seedlings of A nicus show pronounced differences in individual growth

japo-Fig 1 Location of Hirao Bay and the study site A star denotes the

study site TMBL, Tana Marine Biological Laboratory

Fig 2 a General view of the

study site in the intertidal zone

of Hirao Bay at the lowest

spring tide; b partial view of the

predetermined survey area in

June 2005; c the same view in

August 2005

rate (termed ‘‘uneven growth;’’ see for example [16]).

Since there was no information on whether this uneven

growth occurs in its natural habitat, it was difficult to

determine whether a normal cohort study could be

con-ducted or whether to analyze the dynamics of natural

populations However, in our previous study, it was

sug-gested that the increase in uneven growth was derived from

the unevenness of the available amount of food under the

rearing conditions used in the culture tank, which was

limited in capacity and food availability, and that such

conditions do not occur in nature [17] Therefore, in thepresent study, we considered that a normal cohort studycould be used for the growth analysis of A japonicus in thefield, and assumed that the age composition was expressed

by Gaussian distributions calculated from a size–frequencydistribution We also assumed that the single Gaussiandistribution consisted of a class cohort from the same year,which showed increases in size between monthly samplesduring the growth period

Based on the assumptions made above, the monthly sizedistribution of Le was described using 5 mm size-classes.From this, Gaussian distributions were estimated by the v2minimum method (Marquardt method) using the Solverfunction of MS Excel [18], and the calculated means of theGaussian distributions of year-class cohorts were related tothe passage of time in order to calculate the expandedBertalanffy growth curve using a periodic function [19] Inthis growth curve, according to preliminary research on thereproduction of this species [12], recruitment at the sitewas assumed to occur in June The means of year-classcohorts that contained a small number of individuals andcomprised \10% of the total number were not used tocalculate the growth formula The measurements of largeindividuals (over 200 mm) were also eliminated becausethey were too scarce for growth analysis

All of the parameters of the growth formula, except forthe maximum length, were calculated using the Solverfunction of MS Excel Since the juveniles emigrate fromthe intertidal zone as they grow [1], the calculation per-formed using the Solver function was not allowed for themaximum length in the present study It was postulated that

Fig 4 Monthly changes in water temperature at the study site Black

and white circles denote the 2005–2006 and 2008–2009 data,

respectively

Fig 3 a Cross-section of the

present study site C.D.L., the

chart datum level; M.S.L.,

the mean sea level b–d Figures

showing the distributions of

different body sizes of

Apostichopus japonicus during

the growing period, while

e shows the distribution in the

aestivating period

the maximum length for which the present formula was

applicable was the largest body size among all of the

individuals collected in the survey

Results

Activity

Individuals were actively moving from February to July

2005, but eating and discharging were not observed in July

(these activities are easily detected by noting the existence

of feces in the catch container) In August 2005, all of the

individuals were estivating and exhibited a stiffened dermis and a loss of flexibility From September toDecember 2005, no individuals were collected in the sur-vey area, or even from outside the area In January 2006, afew individuals were collected from the tide pool and wereactively moving From February to March 2006, activitywas almost the same as during the same season in theprevious year, but there were only a few individuals

epi-In 2008, individuals continued to actively move, eat anddischarge until July In August 2008, some individualswere estivating under the ripraps, while the others werestill in the tide pool and in an active state without eatingand discharging All individuals were estivating in October

Fig 5 Size–frequency distributions of individuals of the green and

black color types of the Japanese sea cucumber Apostichopus

japonicus at the study site The distributions were smoothed by using

moving means over three size-class intervals Triangles indicate the

means of the Gaussian distributions The distributions described by

dashed curves were not used for the present growth analysis due to sampling bias during the estivating period The cohort shown by white triangles had too few individuals (\10% of the total number of individuals)