Environmental biology of fishes, tập 93, số 3, 2012

Fish origin, size, and release location all had a significant effect on whether an individual demonstrated downstream movement.. Specifically, we analyzed four parameters of migration: d

Migration characteristics of hatchery and natural-origin

Oncorhynchus mykiss from the lower Mokelumne

River, California

S Casey Del Real&Michelle Workman&

Joseph Merz

Received: 28 January 2011 / Accepted: 12 October 2011 / Published online: 22 November 2011

# Springer Science+Business Media B.V 2011

Abstract The lower Mokelumne River (LMR), located

in the California Central Valley, supports a

population of natural-origin Oncorhynchus mykiss

In addition, the Mokelumne River Fish Hatchery

(Hatchery) contributes hatchery produced O mykiss

to the system annually We conducted a 3 year

acoustic tagging study to evaluate the migratory

characteristics of LMR hatchery and natural-origin

O mykiss to the Pacific Ocean Specifically, we

analyzed downstream movement and migration rates,

routes, and success of acoustically tagged O mykiss

of hatchery and natural origin under variable release

locations in non-tidal and tidal habitats Results from

our study suggest there are significant differences in

the proportion of hatchery and natural O mykiss that

demonstrate downstream movement Fish origin, size,

and release location all had a significant effect on

whether an individual demonstrated downstream

movement Mokelumne origin O mykiss that initiated

downstream movement utilized numerous migrationroutes throughout the Delta during their migrationtowards the Pacific Ocean We identified four primarymigration pathways from the lower MokelumneRiver through the Sacramento-San Joaquin Deltawhile the Delta Cross Channel was closed.However, several other pathways were utilized.Origin had a significant effect on O mykiss success

in reaching key points in the Delta and through theEstuary Fish size had a significant effect on whether

an individual reached the marine environment Of the

467 O mykiss tagged, 34 successfully reached thePacific Ocean (Golden Gate Bridge), and of these, 33were hatchery-origin and 1 was natural-origin Ahigher proportion of hatchery-origin fish (10% oftagged) migrated to the ocean compared to natural-origin fish (<1%) Our study provides valuableinformation on the differences between hatchery andnatural-origin O mykiss migration characteristics aswell as unique insight into the migratory behavior oflittle studied non-Sacramento River origin salmonids

Introduction

Steelhead rainbow trout (Oncorhynchus mykiss) hibit one of the most complex life histories of thePacific salmonids (Oncorhynchus spp.) including theability to utilize a variety of diverse habitats and

ex-DOI 10.1007/s10641-011-9967-z

East Bay Municipal Utility District,

One Winemasters Way, STE K-2,

Cramer Fish Sciences,

13300 New Airport Road, Suite 102,

Auburn, CA 95602, USA

Keywords Mokelumne River Oncorhynchusmykiss Acoustic telemetry Migratory behavior Hatchery release strategies

flexible life history traits ranging from resident (rainbow

trout) to anadromous (steelhead) forms (Behnke

2002; Good et al 2005; Zimmerman et al 2008)

Populations of O mykiss once extended throughout

many of the tributaries and headwaters of California’s

Central Valley (CV) (Busby et al 1996; McEwan

2001) Due to the popularity of O mykiss

propaga-tion, they were widely stocked throughout the state

dating back to the 1870s (Behnke1992; Moyle2002)

Today, the majority of CV O mykiss are restricted to

nonhistorical or remnant spawning and rearing habitat

below nonpassable dams and these populations are

heavily subsidized by hatchery production to

mitigate habitat loss and support a large sport

fishery (Yoshiyama et al 1996; Lindley et al.2006)

Even so, numerous stressors continue to impact CV O

mykiss including water diversions and withdrawals,

dams and in-stream structures, conversion of riparian

areas, species introductions, water pollution, and

disruption of coarse sediment supplies (McEwan

2001)

The steelhead component of CV O mykiss is

difficult to monitor because they often migrate and

spawn during periods of high, turbid waters and may

survive spawning or die away from spawning grounds

(McEwan 2001) Furthermore, O mykiss juveniles

often emigrate at larger sizes than CV Chinook salmon

(O tshawytscha) making them less susceptible to

the most common migrant monitoring techniques

used for CV salmonids (DuBois et al 1991;

McEwan2001) In addition, data on the relationship,

interaction, and contrasting dispersal patterns of

steelhead and resident rainbow trout are limited

(Busby et al 1996; NMFS 2003) Recent advances

in acoustic telemetry technology have allowed for the

tracking of movement and migration of individual

fish providing essential information in developing

resource management objectives and recovery goals

for CV O mykiss (Welch et al.2004; Hall et al.2009)

In this study we employed acoustic telemetry

technology to characterize migration patterns of

hatchery and natural-origin O mykiss in the lower

Mokelumne River, California (LMR), a system with

both hatchery and natural production Specifically, we

analyzed four parameters of migration: downstream

movement, migration rates, migration routes, and

migratory success to the Pacific Ocean (as defined

by reaching the Golden Gate Bridge) against three

variables: fish origin, size, and release location

Our objectives were to assess the differences inmigration characteristics using the biologicalparameters identified above

Study siteThe Mokelumne River is a snow-fed system thatdrains approximately 1624 km2of the central SierraNevada The river presently has 16 major waterimpoundments, including Salt Springs (0.175 km3;completed 1931), Pardee (0.244 km3; completed 1929)and Camanche (0.515 km3; completed 1963) reser-voirs The LMR stretches 103 river kilometers (rkm)from Camanche Dam, the lowest nonpassable dam,

to its confluence with the San Joaquin River within thecentral Sacramento-San Joaquin Delta (Delta) (Fig.1).The river is considered part of the North Valley FloorCritical Habitat for CV O mykiss (NMFS 2005).Between New Hope Landing and the San JoaquinRiver confluence, the Mokelumne River is connected

to the Sacramento River via the Delta Cross Channeland Georgiana Slough and to the Central Delta viaLittle Potato and Little Connection sloughs (Fig 2).The LMR currently supports two anadromous salmo-nids which are supported by hatchery production, fall-run Chinook salmon and O mykiss The MokelumneRiver Fish Hatchery (Hatchery) produces O mykiss tocompensate for the decrease in natural fish productionand habitat loss due to the construction of CamancheDam During years when the projected O mykiss eggtake did not meet the Hatchery’s production goals,Mokelumne River stock was augmented withimported eggs and/or fry from the Nimbus Hatchery(American River), the Feather River Hatchery, and theColeman National Fish Hatchery (Sacramento River)(Fig.1) Anadromous, natural-origin O mykiss in theLMR are listed as threatened under the EndangeredSpecies Act (ESA) (NMFS1998) However, the non-anadromous forms (rainbow trout) and hatchery-produced O mykiss are not ESA listed Both residentand anadromous forms of O mykiss are present in theLMR (Satterthwaite et al.2009)

Salmon and steelhead that emigrate out of theLMR must negotiate a maze of natural and man-madetributaries, sloughs, and river channels as they migratethrough the interior Delta to reach the Pacific Ocean

As salmonids navigate the complex network ofchannels that have been significantly altered by waterresource project operations, they are influenced by

both anthropogenic impacts and environmental

pro-cesses that affect migration rates, straying, predation,

and survival (Perry et al.2010) Migration through the

highly modified Delta system may be significantly

more risky than it historically was (Baker and

Morhardt 2001; Brandes and McLain 2001) and the

greatest management concern with respect to preserving

anadromy in CV O mykiss may be reduced survival of

emigrating smolts (Satterthwaite et al.2009)

Materials and methods

Fish collection

O mykiss were collected from four sources within the

LMR: (1) Hatchery-origin O mykiss directly from the

Hatchery, consisting of either Mokelumne River or

Feather River broodstock (1 and 2 year-old fish); (2)Reconditioned kelts obtained from the Hatchery; (3)Natural-origin O mykiss of various life stagescollected using standard boat electrofishing techni-ques (Meador et al 1993) at several locationsthroughout the non-tidal river (within 20 km ofCamanche Dam); and (4) Actively outmigratingnatural-origin O mykiss captured at two rotary screwtraps (RST) (downstream migrant traps used tosample emigrating anadromous salmonids) (Volkhardt

et al 2007) (Table1) The downstream RST (LowerRST) is located near the Mokelumne River tidewaterdownstream of Woodbridge Irrigation District Dam(WIDD) below the Lower Sacramento Road Bridge,61.8 rkm upstream of the confluence with the SanJoaquin River The upstream RST (Upper RST) is inthe non-tidal portion of the LMR above the ElliottRoad Bridge at rkm 87.4 (Fig.1)

Fig 1 Lower Mokelumne River in relationship to Sacramento, San Joaquin, Feather, and American rivers, Sacramento-San Joaquin Delta, and San Francisco Estuary

Surgical implantation of tags

We surgically implanted acoustic transmitters and

passive integrated transponder (PIT) tags in 467

hatchery and natural-origin O mykiss between 2007

and 2009 (Table 1) The tag types included Vemco

V9-2L-69 kHz R64K coded transmitters (implanted in

442 hatchery and natural-origin O mykiss of various

life stages) and Vemco V13-1L-69 kHz R64k coded

transmitters (implanted in 25 reconditioned hatchery

kelts) The V9-2L coded transmitters were 29 mm

long, weighed 4.7 g in air, and had an estimated

battery life of 292 days The corresponding values for

V13-1L coded transmitters were 36 mm long,

weighted 11 g in air, and had an estimated battery

life of 616 days The PIT tags (manufactured by

Destron Fearing) were 12.5 mm long and 2.0 mm

wide and weighed 0.11 g in air The minimum fork

length (FL) of tagged fish was 180 mm to obtain anoptimal transmitter-to-body-weight ratio that did notexceed 5% (Adams et al 1998) Tag burden for allweighed fish was (mean±SE) 2.8±1.4%

Surgical tagging occurred in the field at variouslocations along the LMR and in the Hatchery.Standardized tagging procedures were used at eachlocation O mykiss were anesthetized with tricainemethanesulfonate (natural-origin) or carbon dioxide(hatchery-origin) in aerated water until reactivity andresponses to handling were minimal, but operculummovement was still present Fish fork length and weightwere measured and fish were placed ventral side up in aV-shaped wooden platform with a foam rubber saddlesecured to a transportable open tank Water within thetank was maintained at a level sufficient to keep the gillswetted and was changed every seven to ten surgeries

An acoustic transmitter and a PIT tag were inserted

through a 2.54 cm incision into the peritoneal cavity of

each fish just off the midline and anterior to the pelvic

fins The incision was made using a number 12 surgical

scalpel blade and closed with 2–3 interrupted stitches

Tagged hatchery fish were held in raceways for 24 h

following surgery to allow for recovery and assessed for

abnormal behavior, tag shedding, or mortality before

release Fish tagged in the field were allowed to recover

in aerated holding tanks prior to release the same day

Fish release

In winter 2007, we initiated the first phase of the

3 year study by tagging and tracking three release

groups consisting of hatchery yearling smolts,

reconditioned hatchery kelts, and natural-origin O

mykiss Between January and May of 2008, we

implemented the second year of this study In year

two, we released eight tag groups, incorporated new

release locations, and included hatchery-reared

2-year-old fish and actively-outmigrating natural-origin

O mykiss by focusing on RST captures In 2009, year

three of the study, hatchery, post-spawn kelts, and

natural-origin O mykiss of various life stages were

tagged and released between January and May

(Table 1)

O mykiss releases at Antioch, Selby, New HopeLanding, and in the LMR at Elliott Road (Fig.1) werepumped into a Freightliner transport truck, driven totheir respective release location, and gravity fed intothe receiving waters On Site hatchery yearling smoltreleases were pumped directly from the raceways via15.24 cm diameter aluminum irrigation pipe into theLMR adjacent to the Hatchery Kelts were placed inhauling tanks, transported to the river below theHatchery, and released by using handheld dip nets.Tagged hatchery-origin fish were released either withother hatchery fish or independently O mykiss taggedduring electrofishing surveys were released upstream

of their collection site while fish tagged during RSToperations were released downstream of the traps Allreleases occurred during daylight hours

Data collection

We used stationary Vemco monitoring receivers todetect our Vemco coded transmitters We deployed 10acoustic receivers (Vemco VR2W-69 kHz) in theLMR from the base of Camanche Dam to theconfluence with the San Joaquin River Each receiverrecorded the identification number and time stampfrom the coded acoustic transmitters as tagged fish

RST Rotary Screw Trap; Ave FL Average Fork Length

traveled within the detection range, conservatively

estimated to be 250 m (Espinoza et al 2011) Data

were downloaded quarterly in the field using a

wireless personal computer interface Members of

the California Fish Tracking Consortium downloaded

data from over 300 receivers deployed throughout the

Sacramento-San Joaquin River System, Delta, and

San Francisco Estuary Data from downloaded

receivers were submitted to the California Fish

Tracking Consortium database which provided access

to data from the full array of receivers Following

each release of tagged O mykiss, the Consortium

database was monitored for a minimum of 1 year to

track fish movement

Data analysis

Acoustic tag detection data were processed to

eliminate false-positive detections following methods

of Pincock (2008) and Skalski et al (2002)

False-positive detections typically occur when more than

one tag is simultaneously present within the range of

a given monitor, and simultaneous tag transmissions

“collide” to produce a valid tag code that is not

actually present at the monitor (Pincock 2008; Perry

et al 2010) We considered detections valid if a

minimum of two consecutive detections occurred

within a 30-min period at a given telemetry station

and the detections were consistent with the

spatio-temporal history of a tagged fish moving through the

system of telemetry stations (Skalski et al.2002)

Statistical analysis of movement, migrations rates,

migration pathway selection, and migration success

was based on fish detected by the array of receivers

Release groups that resulted in an expected frequency

of less than five fish in more than 20% of the

analyzed categories or an expected frequency of less

than one in any category being analyzed were not

included in statistical analyses (Zar 1984), but

qualitative assessments were reported All statistical

tests were performed using JMP version 8.0.1

Downstream movement

We compared movement by fish origin and release

location across years using contingency table analysis

(Chi square) (Table 1 for categories) We compared

movement by size using ANOVA Fish were classified

into two main movement groups: downstream (towards

the Golden Gate Bridge) or no downstream movement.The no downstream movement group is made up ofthose fish detected by the array of receivers thatdemonstrated no migration (no net directionalmovement) or upstream movement (movement awayfrom the Golden Gate Bridge)

Migration rates

We estimated migration rates for fish that exhibiteddownstream movement as passage times of individualfish between receivers The migration rate of a fishthrough each reach was calculated as the distancebetween receivers divided by the time Time wasdefined as time of last detection at the previousreceiver to time of first detection at next receiver Weanalyzed migration rates (mean km/h) for each releasegroup using ANOVA

Migration routes

We compared migration pathways used by O mykissreleased in the Mokelumne River at New Hope orupstream that demonstrated downstream movementthrough the interior Delta to Chipps Island (Fig 2).Four pathways were identified: 1) Pathway 1 downthe North Fork of the Mokelumne River to the SanJoaquin River; 2) Pathway 2 down the South Fork ofthe Mokelumne River to the North Fork and SanJoaquin River; 3) Pathway 3 down the South Fork ofthe Mokelumne River into Little Potato Slough andthrough Potato Slough into the San Joaquin River;and 4) Pathway 4 down the South Fork of theMokelumne River into Little Potato Slough, LittleConnection Slough, and into the San Joaquin River.Other important pathways through the Delta includedFranks Tract, Three Mile Slough, and GeorgianaSlough A fish was categorized as using a specificpathway if it was detected moving downstream througheach primary section of a pathway (represented bydetection stations) that led towards Chipps Island Fishthat used a combination of pathways or used sections ofthe interior Delta outside of these four pathways weredescribed by the alternative migration corridor that wasutilized Statistical tests using contingency table analysis(Chi square) were performed on migration routeselection of designated pathways through the interiorDelta based on origin and release location Route

selection analysis based on size was performed using

ANOVA

Migration success

Key reference locations were established to assess

migration success of each release group These

locations include WIDD, New Hope, Chipps Island,

Richmond Bridge, and the Golden Gate Bridge

(Fig 1) The proportions of fish in each tagged

release group detected at each reference location were

based on release group totals Each reference location

site immediately downstream of release locations

accounted for 100% of the upstream release group

Release groups located immediately upstream of a

reference location were excluded from the analyses of

migration success to the first downstream site

Migration success of all release groups were compared

by origin using contingency table analysis (Chi square)

and by size using ANOVA Migration success of

hatchery-origin yearling release groups were compared

by release location using contingency table analysis

(Chi square)

Results

In this study we tagged 330 hatchery-origin and 137

natural-origin O mykiss of various life stages

Ninety-one percent (n= 301) of all acoustically tagged

hatchery releases and 37% (n=51) of natural-origin

releases were detected by the array of receivers

Downstream movement

Of the 404 acoustically tagged hatchery yearling smolts

and natural-origin O mykiss released, 169

demonstrat-ed downstream movement, 124 demonstratdemonstrat-ed no

downstream movement, and 111 were not detected

by the array of receivers Fish origin, size, and release

location revealed differences between migration and

residualization (no movement)

Fish origin had a significant effect on downstream

movement of all O mykiss release groups independent

of release location between 2007 and 2009 (Chi

square=25.26; P<0.001; Table 2) Comparing all

hatchery yearling smolt and natural-origin O mykiss

release groups, a significantly higher proportion of

hatchery-origin fish moved downstream (65%), than

natural-origin fish (22%), independent of releaselocation (Chi square =33.58; P<0.001) Of natural-origin fish that moved downstream, 64% wereconsidered ‘active migrants’, based on the collection

at RSTs Of the natural-origin O mykiss releases thatshowed no downstream movement, 95% (n=38)exhibited resident characteristics via non-directionalmovements detected by the receivers in the non-tidalLMR Of the hatchery yearling releases that had nodownstream movement, 95% (n = 80) strayedupstream

Fish size had a significant effect on downstreammovement (F = 11.29; df = 1; P=0.001) across allrelease groups (Table 2) The average fork length of

O mykiss that demonstrated downstream movementwas 262 mm with a standard deviation of 82 mm Theaverage fork length of O mykiss that demonstrated nodownstream movement was 295 mm with a standarddeviation of 100 mm

Movement of hatchery-origin O mykiss yearlingsmolts differed significantly (Chi square =8.52; P=0.036) based on release locations Downstreammovement was observed from all release locations.The Antioch release had the highest downstreammovement with 83% towards the Pacific Ocean(Fig 3) The On Site release in the non-tidal LMRhad the second highest downstream migration (81%).The proportion of fish that exhibited no downstreammovement from Antioch, San Pablo, and New Hopereleases varied from 17% to 39%

During the 2007 to 2009 study period, there wasalso a significant difference between the movement ofnatural-origin O mykiss release groups (Chi square=17.23; P<0.001) Of the fish that exhibited nodownstream movement, 95% were part of the InRiver release groups collected during electrofishingsurveys Of the In River release group, 90% exhibited

no downstream movement In comparison, a higherproportion of the natural-origin fish tagged at the RSTsites demonstrated downstream movement Six out ofeight tagged and released at the Lower RST and one

of one at the Upper RST exhibited downstreammovement (Fig.3)

Due to the small sample size for release groups ofkelts and 2-year-olds, they were not included in thestatistical analysis of downstream movement byrelease location However, detected movement ofthese life stages is noteworthy Of the reconditionedkelt releases, 54% (n=13) demonstrated downstream

movement New Hope (Kelt) releases and On Site

(Kelt) releases demonstrated 75% (n=6) and 44% (n=

7) downstream movement, respectively Of the

2-year-old releases, 17% (n=6) demonstrated

down-stream movement Of the San Pablo (2-year-old) and

Moke River (2-year-old) release groups, 14% (n=4)

and 33% (n=2) demonstrated downstream movement,

respectively

Migration rates

Between 2007 and 2009, there was no significant

difference between the migration rates of O mykiss

from different release groups (F=1.80; df=9; P=

0.072) The Antioch hatchery release of yearling

smolts showed the greatest sustained migration rates

with an average of 1.86 km/h Kelt migration ratesranged from 1.58 km/h (On Site) to 1.64 km/h (NewHope) while 2-year-old O mykiss migration ratesranged from 1.29 km/h (Moke River) to 1.61 km/

h (San Pablo) The natural-origin In River releasegroup had the lowest average migration rate of0.72 km/h (Table 3)

We recovered ocean travel time data on fivehatchery O mykiss (two yearlings released at NewHope; one yearling released at San Pablo; one MokeRiver 2-year-old released at Elliott Rd.; and one keltreleased at New Hope) Travel rates were calculatedover approximate straight-line distances between theGolden Gate Bridge and the acoustic receiver arraylocated off of Point Reyes (~54 km north of theGolden Gate) The New Hope kelt showed the greatest

Values within the figure represent number of fish

location on downstream movement, emigration pathway, and

success to key landmarks within the lower Mokelumne River,

Delta, and San Francisco Estuary Values represent all release

groups, except analyses of movement and migration success by release location which analyze hatchery yearling release groups A P-value≤0.05 is considered significant (Bold)

Migration Success by Location

WIDD Woodbridge Irrigation District Dam; ISS Insufficient sample size

sustained migration rate of 1.33 km/h and reached Point

Reyes in 1.7 days Hatchery yearling migration rates

ranged from 0.02 km/h (San Pablo) to 0.17 km/h (New

Hope) while the Moke River 2-year-old O mykiss

migration rate was 0.20 km/h A New Hope yearling

last detected at the Golden Gate Bridge 2 h before the

New Hope kelt took 16 days to reach Point Reyes A

yearling released in San Pablo Bay spent just over

145 days traveling between the Golden Gate Bridge

and Point Reyes

Migration routes

Between 2007 and 2009, 67 acoustically tagged

hatchery and natural-origin O mykiss of various life

stages released at or above New Hope Landing

demonstrated downstream movement via the designated

migration pathways Migration route selection, based on

all release groups, was not significantly related to fish

size (F=0.88; df=2; P=0.420) or release above or

within tidal influence (Chi square =0.96; P=0.618)

(Table 2) Of the hatchery yearling smolts, 43% used

Pathway 1, 23% used Pathway 2, 4% used Pathway 3,

and 2% used Pathway 4 In addition, 28% used other

pathways including Franks Tract (13%), Three Mile

Slough (11%), and Georgiana Slough (4%)

Fifty-seven percent of the reconditioned kelts migrated

through Pathway 1 while 29% utilized Franks Tract

and 14% migrated through Pathway 4 All of the fish

from the Moke River (2-year-old) release group

migrated through Pathway 1 Of the natural-origin

O mykiss, 60% used Pathway 1, 20% used Pathway

2, and 20% used Georgiana Slough

Migration success

While migration proportions reflect low overall

downstream success based on release totals, fish that

reached the first reference location subsequently had

relatively high migration success On Site releases of

hatchery yearling smolts had the highest overall

success to the first downstream reference point with

57% detected This was followed by 44% of On Site

kelts reaching the first reference location downstream

Twenty-five percent of the Moke River release group

successfully migrated to the first downstream reference

point In River releases of natural-origin O mykiss had

the lowest overall downstream detection at the first

reference point (New Hope) with only 0.8% detected

There was a significant difference in the size of fishthat successfully migrated to WIDD (F=5.32; df=1;P=0.027; Table 2) The average fork length of fishthat reached WIDD was 332 mm while the averagefork length of fish that did not was 433 mm Inaddition, fish origin had a significant effect onmigration success to New Hope (Chi square =11.39;P=0.001; Table2)

Migration success between New Hope and ChippsIsland ranged from 100% for Moke River 2-year-oldfish to 50% of the natural-origin fish Of the NewHope hatchery yearling and kelt releases, migrationsuccess to Chipps Island, the first downstreamreference location, was 17% and 22%, respectively(Fig.4) Fish origin (Chi square =7.29; P=0.007) had

a significant effect on migration success to ChippsIsland while size and release location did notsignificantly influence migration success through theDelta (Table 2)

Between Chipps and Richmond Bridge, migrationsuccess between reference locations ranged from100% of On Site kelts, Moke River 2-year-old fish,and New Hope kelts to 36% of New Hope yearlings.Seventeen percent of the Antioch yearling releasewere detected at the first downstream referencelocation Fish origin (Chi square =4.02; P=0.045)and size (F = 8.09; df = 1; P=0.005) significantlyinfluenced success to Richmond Bridge (Table 2).Migration success from Richmond Bridge toGolden Gate Bridge was relatively high in com-parison to total release group success Larger fishhad a better chance of reaching both the Richmondand Golden Gate bridges (locations with highersalinity) Fish origin (Chi square = 5.55; P=0.018)and size (F=11.18; df=1; P=0.001) had a significanteffect on migration success to the Golden Gate Bridge(Table 2) Of the hatchery yearling smolt releases,20% of On Site, 14% of San Pablo, and 9% ofAntioch releases reached the Golden Gate Bridge.The New Hope release group had the lowestpercentages of success to the Golden Gate Bridgewith 4% in 2007 and 5% in 2009 On Site releases ofreconditioned kelts had the highest proportion reachthe Golden Gate Bridge with 33% in 2007 However,

in 2008, only 10% of On Site kelts reached theGolden Gate Bridge Twenty-two percent of the NewHope reconditioned kelts and 25% of the 2-year-oldhatchery Moke River release group reached theGolden Gate Bridge One natural-origin fish was

recorded successfully migrating to the Golden Gate

(Fig.4)

Discussion

Acoustic technology has provided a method to better

compare hatchery-origin and natural-origin O mykiss

In a state-dependent life history model, Satterthwaite

et al (2009) predicted a mixture of anadromous and

resident O mykiss in the Mokelumne River, but with

anadromous fish dominating given baseline survival

assumptions Our results demonstrate the Mokelumne

River O mykiss population is a mixture of resident

and anadromous fish and that origin (hatchery vs

natural) has a significant effect on whether an

individual fish demonstrates migration tendencies

We showed that hatchery fish had a significantly

higher propensity to migrate, while the natural

population demonstrates very little anadromy

Downstream movement

In an effort to increase survival and promote returns,

the Hatchery has utilized numerous release locations

for hatchery-reared O mykiss However, returns have

remained low We found that release location can

significantly influence downstream migration trends

in hatchery yearling smolt O mykiss even though all

hatchery release groups demonstrated relatively high

downstream movement (59%)

The natural-origin O mykiss population in the LMRexhibits both anadromous and non-anadromous lifehistories Of the acoustically tagged natural-originfish detected by the array of stationary receivers, 78%demonstrated no downstream movement Conversely,once a natural-origin fish began downstream migration(for instance O mykiss captured at a RST) theycontinued in a downstream direction at a relativelyhigh proportion

Migration routes

Steelhead emigrating from the Mokelumne River have

numerous migration pathway options when traversing

the complex network of natural and man-made channels

of the interior Delta Each migration route poses

different benefits and risks associated with migration

rates, energy costs, predation, and entrainment that

ultimately affect migration success Due to the small

number of fish migrating through the Delta and the

utilization of diverse migration routes, current and

future management actions in the Delta may tionately affect Mokelumne River O mykiss

dispropor-Migration success

In seawater challenges, Beakes et al (2010) foundthat CV O mykiss survival off the California centralcoast varied significantly with fish size (with largerfish being more likely to survive than smaller fish).Similarly, we observed that success to key referencelocations within the saline environment of the San

WIDD (Rkm 173)

New Hope(Rkm 142)

Chipps Island(Rkm 70)

Richmond(Rkm 15)

Golden Gate(0)

Hatchery yearlings (Antioch release)Hatchery yearlings (San Pablo release)Hatchery 2-year-olds (San Pablo release)

5 5 3

30

35

6 35

(d)

within the figure represent release group totals followed by fish detection totals at each downstream reference location

Francisco Estuary was significantly related to fish

size

If we gauge‘successful’ migration as migration to the

Golden Gate Bridge, a majority of our successes have

been of hatchery-origin Reconditioned kelts released

On Site in 2007 had the highest proportion reach the

Golden Gate Bridge, thus active reconditioning of

hatchery spawned kelts may be a viable option for

increasing anadromy On Site releases of both hatchery

yearling and reconditioned kelts performed well during

the study period, but continued releases adjacent to the

hatchery will need to be weighed against potential

negative impacts to natural-origin salmonids rearing in

the LMR

Management implications

The diversity of O mykiss life history forms

demon-strates the relative phenotypic plasticity of the species

(McEwan2001) The year round presence of Age 1+

O mykiss of various life stage categories sampled

during fish community surveys on the LMR (Merz

2002) reflects the flexible life history patterns of O

mykiss within the Mokelumne River Zimmerman et al

(2008) revealed that the Central Valley O mykiss

population is skewed towards the non-anadromous

resident form as 77% of the analyzed O mykiss in his

study were progeny of resident rainbow trout

Similarly, results from our study suggest a large

proportion of natural-origin O mykiss in LMR

demonstrates a resident life history

Due to the precipitous declines of O mykiss in the

Central Valley and an apparent shift towards the

non-anadromous life history forms, the connection between

anadromous and non-anadromous O mykiss and their

management as a single or separate population has

profound implications for conservation and recovery

(Busby et al 1996; Zimmerman and Reeves 2000;

McEwan 2001) Since anadromous and

non-anadromous trout may form an interbreeding population

(Seamons et al.2004; Araki et al.2007) with females

producing progeny with opposite life history traits

(Viola and Schuck 1995; Riva-Rossi et al 2007;

Zimmerman et al.2008), steelhead management may

need to include protection of non-anadromous forms

and the connectivity between the resident and

anadromous fish (McEwan2001)

The largest population declines of natural-origin O

mykiss in California were a consequence of the dam

building era prior to the 1960s as spawning andrearing habitats became isolated (McEwan 2001).However, continued declines of O mykiss numbersimply additional threats and stressors still need to beaddressed For anadromous species migrating out ofthe Mokelumne River, Delta management remains acritical issue influencing migration success While theDelta Cross Channel remained closed throughoutthe study period, its management is presumed tosubstantially influence Mokelumne River salmonids.Further investigation is needed to assess its effects

on salmonid migration, straying, and survival Inaddition to Delta management, suppression ofanadromous life history traits, loss of geneticdiversity, and introgression of hatchery rainbowtrout into natural-origin populations continue to beserious concerns for steelhead conservation andmanagement

provided by East Bay Municipal Utility District, the California Urban Water Agencies, and the Mokelumne River Partnership.

We gratefully acknowledge J Miyamoto, J Smith, J Setka, E Rible, C Hunter, M Saldate, J Shillam, P Sandstrom, E Chapman, W Heady, all field staff who helped develop and collect data for this study, and the collaborative support of the Mokelumne River Fish Hatchery and the California Fish Tracking Consortium.

References

Adams NS, Rondorf D, Evans S, Kelly J, Perry R (1998) Effects of surgically and gastrically implanted radio transmitters on swimming performance and predator avoidance of juvenile Chinook salmon (Oncorhynchus tshawytscha) Can J Fish Aquat Sci 55:781–787 Araki H, Ardren W, Olsen E, Cooper B, Blouin M (2007) Reproductive success of captive-bred steelhead trout in the wild: evaluation of three hatchery programs in the Hood

Baker P, Morhardt E (2001) Survival of Chinook salmon smolts

in the Sacramento-San Joaquin Delta and Pacific Ocean In: Brown R (ed) Fish Bulletin 179: contributions to the biology of Central Valley salmonids, vol 2, 2 California Department of Fish and Game, Sacramento, pp 163–182 Beakes MP, Satterthwaite W, Collins E, Swank D, Merz J, Titus

R, Sogard S, Mangel M (2010) Smolt transformation in two California steelhead populations: effects of temporal

Behnke R (1992) Native trout of western North America Am Fish Soc Monog 6 Am Fish Soc, Bethesda, Maryland Behnke R (2002) Trout and salmon of North America Free Press, New York

Brandes P, McLain J (2001) Juvenile Chinook salmon abundance,

distribution, and survival in the Sacramento-San Joaquin

Estuary In: Brown R (ed) Fish Bulletin 179: contributions to

the biology of Central Valley salmonids, vol 2 California

Burgner RL, Light J, Margolis L, Okazaki T, Tautz A, Ito S (1992)

Distribution and origins of steelhead trout (Oncorhynchus

mykiss) in offshore waters of the North Pacific Ocean Int

Busby PJ, Wainwright T, Bryant G, Lierheimer L, Waples R,

Waknitz F, Lagomarsino I (1996) Status review of west

coast steelhead from Washington, Idaho, Oregon, and

California U.S Dept of Commer., NOAA Tech Memo.

NMFS-NWFSC-27 Seattle, WA

DuBois RB, Miller J, Plaster S (1991) An inclined-screen smolt

trap with adjustable screen for highly variable flows N

Espinoza M, Farrugia T, Webber D, Smith F, Lowe C (2011)

Testing a new acoustic telemetry technique to quantify

long-term, fine-scale movements of aquatic animals Fish

Res 108:364–371

Good TP, Waples R, Adams P (2005) Updated status of federally

listed ESUs of west coast salmon and steelhead U.S Dept.

Commer., NOAA Tech Memo NMFS-NWFSC- 66 Seattle,

WA

Hall JE, Chamberlin J, Kagley N, Greene C, Fresh K (2009)

Effects of gastric and surgical insertions of dummy

ultrasonic transmitters on juvenile Chinook salmon in

T07-267.1

Lindley ST, Schick R, Agrawal A, Goslin M, Pearson T, Mora

E, Anderson J, May B, Greene S, Hanson C, Low A,

McEwan D, MacFarlane R, Swanson C, Williams J (2006)

Historical population structure of Central Valley steelhead

and its alteration by dams San Franc Estuary Watershed

Sci 4, Issue 1, Artic 2

McEwan DR (2001) Central Valley steelhead In: Brown R (ed)

Fish Bulletin 179: Contributions to the biology of Central

Valley salmonids, vol 1, 1 California Department of Fish

McKinnell S, Pella J, Dahlberg M (1997) Population-specific

aggregations of steelhead trout (Oncorhynchus mykiss) in

Meador MR, Cuffney T, Gurtz M (1993) Methods for sampling

fish communities as part of the National Water-Quality

Assessment Program U.S Geological Survey Open-File

Merz JE (2002) Seasonal feeding habits, growth, and movement of

steelhead trout in the lower Mokelumne River, California.

Moyle PB (2002) Inland fishes of California University of

California Press, Berkeley

NMFS (National Marine Fisheries Service) (1998) Endangered

and threatened species: threatened status for two ESUs of

steelhead in Washington, Oregon, and California Fed

Regist 63(53):13347–13371

NMFS (National Marine Fisheries Service) (2003) Updated

status of federally listed ESUs of West Coast salmon and

steelhead West Coast Salmon Biological Review Team.

NOAA Fisheries Northwest and Southwest Fisheries

Science Centers

NMFS (National Marine Fisheries Service) (2005) Designation

of critical habitat for seven evolutionarily significant units

of pacific salmon and steelhead in California Fed Regist 70(170):52488–52627, 2 Sept 2005

Perry RW, Skalski J, Brandes P, Sandstrom P, Klimley A, Ammann A, MacFarlane B (2010) Estimating survival and migration route probabilities of juvenile Chinook salmon

Pincock DG (2008) False detections: what they are and how to remove them from detection data Vemco, Amirix System Inc Doc-004691-08 Halifax, Nova Scotia, Canada Riva-Rossi C, Pascual M, Babaluk J, Garcia-Asorey M, Haldenk

N (2007) Intra-population variation in anadromy and reproductive life span in rainbow trout introduced in the

Satterthwaite WH, Beakes M, Collins E, Swank D, Merz J, Titus

R, Sogard S, Mangel M (2009) State-dependent life history models in a changing (and regulated) environment: steelhead

in the California Central Valley Evol Appl 3:221–243 doi: 10.1111/j.1752-4571.2009.00103.x

Seamons TR, Bentzen P, Quinn T (2004) The mating system of steelhead, Oncorhynchus mykiss, inferred by molecular analysis of parents and progeny Environ Biol Fish

Skalski JR, Townsend R, Lady J, Giorgi A, Stevenson J, McDonald R (2002) Estimating route specific passage and survival probabilities at a hydroelectric project from smolt

Viola AE, Schuck M (1995) A method to reduce the abundance

of residual hatchery steelhead in rivers N Am J Fish

Volkhardt, GC, Johnson S, Miller B, Nickelson T, Seiler D (2007) Rotary screw traps and inclined plane screen traps.

O’Neil A, Pearsons T (eds) Salmonid field protocols handbook: Techniques for assessing status and trends in salmon and trout populations Am Fish Soc, Bethesda, Maryland, pp 235–266

Welch DW, Ward B, Batten S (2004) Early ocean survival and marine movements of hatchery and wild steelhead trout (Oncorhynchus mykiss) determined by an acoustic array: Queen Charlotte Strait, British Columbia Deep-Sea Res II

Yoshiyama RM, Gerstung E, Fisher F, Moyle P (1996) Historical and present distribution of Chinook salmon in the Central Valley drainage of California Sierra Nevada Ecosystem Project, Final Report to Congress, Vol 3 Centers for Water and Wildlife Resources, University of California, Davis

Zar JH (1984) Biostatistical analysis, 2nd edn Prentice-Hall, New Jersey

Zimmerman CE, Reeves G (2000) Population structure of sympatric anadromous and nonanadromous Oncorhynchus mykiss: evidence from spawning surveys and otolith microchemistry Can J Fish Aquat Sci 57:2152–2162 Zimmerman CE, Edwards G, and Perry K (2008) Maternal origin and migratory history of Oncorhynchus mykiss captured in rivers of the Central Valley, California Final Report prepared for the California Department of Fish and Game Sacramento, California

Migration patterns of juvenile Lutjanus argentimaculatus

in a mangrove estuary in Trang province, Thailand,

as revealed by ultrasonic telemetry

Matiss Zagars&Kou Ikejima&Nobuaki Arai&

Hiromichi Mitamura&Kotaro Ichikawa&

Takashi Yokota&Prasert Tongnunui

Received: 31 January 2011 / Accepted: 27 October 2011 / Published online: 12 November 2011

# Springer Science+Business Media B.V 2011

Abstract Migrational patterns of mangrove jack

Lutjanus argentimaculatus were studied in a

man-grove estuary in Trang province, Thailand, using

ultrasonic telemetry Ultrasonic coded transmitters

were surgically implanted in 18 fish and 16 of them

were subsequently monitored by nine fixed receiversinstalled along Sikao Creek estuary in June andNovember 2006 Due to technical limitations all ofthe individuals were released in the middle of thecreek Their movements were monitored for a period

of up to 1 month, the data being used to describe shortterm migration of juvenile Lutjanus argentimaculatus

in the creek and to find possible environmental cuesfor the observed movements All of the individualsshowed a tide related movement pattern, suggestingforaging in the small mangrove channels and/ormangrove forest during high tides 50% of the fishleft the study area for the open coast area within ashort time following release, indicating that a part ofjuvenile L argentimaculatus may move in betweenestuarine habitats instead of being site attached Asthe fish were reared in fish cages for a certain period

of time before the study this behavior could partly beexplained by the time spent in captivity It was foundthat L argentimaculatus showed higher movementactivity during night high tides possibly explained by

an increased availability of the sough after food items

Keywords Mangrove estuary Lutjanusargentimaculatus Migration Ultrasonic telemetry

Introduction

Many studies over the last few decades have focused

on evaluating the importance of mangrove habitats for

School of Environment, Resources and Development,

Asian Institute of Technology,

P.O Box4, Kulong Luang, Pathumthani 12120, Thailand

e-mail: ikejima@kochi-u.ac.jp

P Tongnunui

Rajamangala University of Technology Srivijaya,

Amphur Sikao, Trang 92150, Thailand

different fish species (e.g Robertson and Duke

1987; Laegdsgaard and Johnson 1995; reviewed in

Robertson and Blaber1992; Blaber2000) Robertson

and Duke (1987) followed by Thayer et al (1987)

compared fish assemblages in mangroves and

proxi-mal habitats Both studies showed that mangroves

contained a considerably greater abundance and

species’ richness of fish than adjacent,

non-mangrove habitats, such as seagrass beds and

mud-flats In the years following, many studies have

confirmed the above findings throughout the tropical

seas (Sheaves1992; Kimani et al.1996; Nagelkerken

et al.2000; Ikejima et al.2003)

Inter-annual, seasonal, lunar and diel changes have

been recorded for mangrove ichthyofauna (Robertson

and Duke 1990; Laegdsgaard and Johnson 1995;

Ikejima et al 2003; Mumby et al 2004), suggesting

that the majority of species migrate in and out, or

within the habitat For example, Robertson and Duke

(1990) showed a clear difference in fish assemblages

between high and low tides, implying a regular

pattern of tide-related fish movements in mangroves

Sheaves (2005) pointed out the importance of

migration as an essential part of the life cycle of

fishes in mangrove habitats, indicating that

man-groves are a part of an “interconnected habitat

mosaic” and should be studied within the context of

connectivity with other habitats In many mangrove

systems, migration to alternative habitats is

unavoid-able because the habitat is exposed during low tide

Even in areas where mangroves remain inundated

throughout the year, fish may shift habitats for

feeding, reproduction and life stage-specific habitat

use (Nagelkerken et al.2002; Sheaves2005)

Tidal fish migration has been studied by Krumme

and Saint-Paul (2003) and Krumme (2004) using

hydroacoustic equipment Hydroacoustic techniques

are effective for showing overall patterns of fish

migration in mangrove creeks, although, as was noted

by Krumme (2004), they do not show behavioral

variations between species and individuals

Ontogenetic migration of several reef fish species

has been successfully investigated using visual

cen-suses E.g Cocheret de la Morinière et al (2002)

observed relative density distributions of different

size-classes of selected species and demonstrated three

modes of post-settlement migration among mangroves,

seagrass beds and coral reefs Furthermore, Nakamura et

al (2008) described the ontogenetic migration of coral

inhabiting black tail snapper from mangrove habitat

to coral reefs, using stable isotope approach.However, these methods do not provide evidence ofshort term movements, such as diel or tidalmovements and visual censuses are not applicable inhighly turbid environments, such as mangrove creeks

in Thailand

Ultrasonic telemetry allows direct monitoring ofthe movement patterns of individual fish, giving aninsight into their behavioral biology The method hasbeen widely applied in a variety of habitats, givingnew insights into fish migratory behavior, e.g homingand site fidelity of greasy grouper in coral reefs(Kaunda-Arara and Rose 2004), homing behavior ofblack rockfish in coastal waters (Mitamura et al

2005), and the diurnal and tidal movements ofsnapper Pagrus auratus in a river estuary (Hartill et

al.2003) Recent studies have shown that this methodcan also be successfully applied to study movements

of groupers (serranidae) and snappers (lutjanidae) inmangroves and associated habitats Frias-Torres et al.(2007) used ultrasonic telemetry in a mangrovehabitat for monitoring individual fish movements.Despite the limited number of individuals monitored,they showed that the fish movements were correlatedmainly with tidal cycle Luo et al (2009) usedultrasonic telemetry combined with tagging and videorecording to observe daily as well as seasonalmovement patterns of grey snapper in Florida Keysand showing that the fish moved between the inshorehabitats for taking shelter and foraging, and per-formed longer movements to offshore reefs during thereproductive season Ultrasonic telemetry was alsosuccessfully used in a Bahamian tidal creek to showthat individuals of two lutjanid species show intra-population variation in movement patterns contra-dicting the existing assumption that individuals of agiven population possess the same behavioral charac-teristics (Hammerschlag-Peyer and Layman2010).Mangrove jack Lutjanus argentimaculatus is arelatively large fish, reaching up to 120 cm forklength and 8.5 kg in weight, inhabiting the Indo-westPacific Ocean (Russell et al.2003) It is commerciallyimportant for fishery and aquaculture industries inSouth East Asia, and sports fishing in Australia (Doiand Singhagraiwan 1994; Russell and McDougall

2005) Spawning occurs in offshore habitats, postlarvae and juveniles then moving to coastal nurseryhabitats, such as river estuaries and mangroves

(Sheaves 1995; Russell and McDougall2005) After

reaching reproductive maturity (at around 450, and

500 mm fork length for male and female,

respective-ly; Russell et al.2003) L argentimaculatus returns to

offshore habitats A study of movements of L

argentimaculatus in Australia using conventional

tagging suggested that during their period of

resi-dence in inshore areas, the species underwent local

migration within the river systems (Russell et al

2003) It has also been noted that during high tides L

argentimaculatus enters mangrove forests in order to

feed, indicating that the local tidal cycle was an

environmental cue for short term migration (Russell et

al 2003; Sheaves 2005) Nevertheless, since

move-ments have not been followed directly, uncertainties

remain concerning the movement pattern of L

argentimaculatus within estuarine habitats

In the present study we described short term

migration of L argentimaculatus in the Sikao

mangrove creek using ultrasonic telemetry In contrast

to previous studies of snappers in mangrove systems

we examined short term movements of juvenile fish

within their nursery habitat We specifically examined

to what extent cyclic changes in tidal height and time

of day are shaping the short term movement pattern of

L argentimaculatus The availability of food items

for predatory fish in mangrove habitats varies with the

tidal pattern due to the migration of smaller fish and

changing accessibility of the forest (Sheaves 2005)

The daily cycle is also known to influence migration

patterns in many lutjanids (e.g Nagelkerken et al

2000; Luo et al.2009) Thus we hypothesized that the

frequency of movements of L argentimaculatus in

Sikao creek change with tidal and diel cycles

Materials and methods

Study area

The study was conducted in Sikao Creek, a mangrove

estuary located in Trang province, west coast of

Thailand (Fig 1) The particular study site was

chosen due to its relatively small size, the presence

of undisturbed mangrove forest, and the availability

of periodically inundated side creeks making it

suitable for the examination of movement patterns

of L argentimaculatus and its relation to tidal

changes The forest, dominated by Rhizophora

apiculata, has been designated as a protected area

A relatively short dry season (January to April) isfollowed by a long wet season (May to December).The estuary is subject to semi-diurnal tides with atidal range of 1.0 (neap tide)–2.5 (spring tide) m(Ikejima et al 2003) One kilometer from an openbay, Sikao Creek splits into two main creeks (A andB; Fig.1), there being numerous side creeks branch-ing from the main creeks The main creek and thefirst level side creeks remain inundated through-out the diel and lunar tidal cycles, the mangroveforest and smaller side creeks are inundated onlyduring high tides Sikao Creek is connected to ashallow bay lined with large rocks The closestneighboring mangrove system is located ~4 kmaway in the NW direction, the closest seagrassbed ~5 km away in NW direction and the closestcoral reef ~18 km away in SW direction There is

no major river input L argentimaculatus iscommonly found in the area and an active fisheryfor the juveniles and subadults has been observedwithin the creek, individuals often being caught usingcage traps for culture in fish cages

The first study was conducted in the middle ofthe rainy season from 6 June 2006 to 22 July 2006during 39 semi-diurnal tides The second wasconducted at the end of the wet season, from

1 November 2006 to 2 December 2006 during 55semi-diurnal tides

Fish tagging

In total 18 fish were used in the two study periods(Table 1), selection for tagging being based on thecriteria of size suitable for tagging and good health(indicated by active movement, and no externalsymptoms of injuries or diseases) TL of fish rangedfrom 199 to 303 mm, with the mean (± SD) of 242±

37 mm (excluding 2 individuals discarded from theanalysis) All the individuals were within the sizerange of juveniles of mangrove jack reported fromAustralia (75–541 mm TL, Sheaves 1995;

<450~500 mm FL, Russell et al 2003) It has beenestimated that transmitter mass should not exceed 2%

of the fish’s mass (Jepsen et al.2002) For all the fishtransmitter weight in water never exceeded 1.4% ofthe body weight Due to logistical reasons it wasimpossible to attain wild individuals directly from thestudy area Thus seven of the individuals tracked in

the first study period were wild L argentimaculatus,

caught by fishermen in creek A and reared in fish

cages for 3 months in a near vicinity to the study site

(Fig.1) One fish was caught by a sports fisherman at

the river mouth of the Creek All ten fish used in the

second study period were wild L argentimaculatus,

caught by fishermen within creek A and reared in fish

cages for 1 week In order to evaluate if different

rearing periods in the fish cages had influence on fish

health condition allometric condition factor (K = W

L−b) was calculated for experimental fish (Godinho

1997) The value of b was obtained from the weight–

length relationship using linear regression: LogeW¼

Logeaþ bLogeL, where W = weight, and L = total

length The parameters (a, b) were obtained from

pooled data of all experimental fish (n=18), then K

was calculated for fishes of each study period There

was no significant difference in mean K values

between the two periods (t-test, t=−0.13, p=0.86,

df=16), suggesting that the rearing conditions were

close to natural and longer rearing period did not

influence fish health condition As the fish were

taken, reared for a short time period and released in

the Sikao Creek (Fig 1) we suggest that there was a

minimal effect of rearing on their natural behavior,which is further supported by the observed movementpatterns which were similar to those of other lutjanids

in coastal habitats (seeDiscussion)

Ultrasonic coded transmitters (V9-1L-R256, VemcoLtd, Canada), 8.5 mm in diameter, 25 mm long andweighing 2.2 g in water were used The min/max period

of pulse transmissions was 10–30 s maximizing thepossibility of tag transmission when the fish were withinthe detection range of the receiver array The trans-mitters were surgically implanted into the abdominalcavity of anesthetized fish (induced by 0.1% 2-phenoxyethanol) During the operation, fish were fixedbetween rubber sponges in a bath of aerated seawater

An incision of about 10 mm length was made in theinferior abdominal wall of the fish in order to insert thetransmitter An operating needle and sutures were usedfor closing the wound The antibiotics oxytetracyclinehydrochloride and polymixin B sulfate were applied.After the surgery, the fork length and weight of eachindividual were measured The fish were held in anaerated plastic experimental tank (500 l in volume) for afurther 1 day, allowing recovery before release Allindividuals were released at receiver 7 (Fig.1)

Fig 1 Map of Sikao Creek showing mangrove areas and the

positions of the receivers during both study periods as well as

the detection range of the receiver array Numbers indicate

receiver positions A and B indicate the main creeks, and C, D

and E some of the larger side creeks, FC indicates fish cages were the fish were reared location Tagged fish were released at Receiver 7

Tracking and monitoring systems

Nine fixed hydrophone receivers (VR-2, Vemco Ltd.)

were used in the experiment The receivers were

60 mm in diameter and 340 mm long and logged data

on the presence of tagged fish Receivers were

powered by lithium dry cell that lasted for up to

180 days and had a flash memory for data recording

ID number, date and time were recorded when a

tagged fish was within the detection range of the

receiver Creek B (Fig 1) was chosen for the

experiment due to its smaller size allowing more

efficient monitoring of the area with the available

equipment Nine fixed monitoring receivers were

installed within 5 km of the creek’s lower reaches

(Fig.1) in order to monitor the movements of tracked

fish in the creek and to detect possible exits to coastal

waters or creek A Due to the regular boat traffic the

receivers were positioned at the sides of the creek

Detection range experiment

A boat was used to tow an activated transmitterthrough the study area in a direction from receiver 1

to receiver 9 (Fig 1) during spring high ebbing tide.This was assumed to be the period with the highestbackground noise level thus the lowest detectionefficiency due to the very high current speed Thetransmitter was towed through the main channel alongthe opposite coast of the one where receivers werepositioned and into all the side channels In additionthe transmitter was fixed for 15 min in the mangrovefringe of the main channel in distances of 50, 150 and

250 m from the receiver 7 in order to assess if apossible entrance of fish to the mangrove forest wouldblock the acoustic signal Weights were used in order

to keep the transmitter 30 cm above the bottom of thecreek Location of the boat was recorded continuouslyusing eTrex Vista HcX global positioning device

Table 1 Summary of the

phys-ical parameters, tracking data of

individual fish, and the number

of movements each of day/night

time, and the tidal stage

a

Non consecutive days

b

Two individuals were excluded

from the analysis due to the very

high frequency and fixed

loca-tion of detecloca-tions, indicative of

dead fish or shed transmitter

ID code

TL(mm) BW(g)

Period I, released on 9~13 June 2006

(Garmin Ltd, USA) To match the time of each

detected signal with GPS locations of the transmitter

the receivers and GPS clock times were synchronized

prior to the experiment

In addition a fixed transmitter hanging 30 cm from

the creek bottom was set 80 m from the receiver 1 for

24 h in order to assess the influence of changes in

tidal height on the detection of the signals in the

present acoustic environment During the first hour of

this experiment all the events of passing boats were

recorded in order to assess the influence of boat

noises on detection of the signals

Data analysis

In this study detected signal patterns were interpreted

as fish movements in two ways: 1) the shift of signal

reception from one receiver to another, which

reflected movement from the detection range of one

receiver to that of the other, 2) the gaps in signal

reception by one receiver during an otherwise

continuous detection period All changes in signal

reception from one receiver to another were defined

as movements, regardless of the length of signal gap

between the two receivers If the fish was detected in

succession by two receivers with overlapping

detec-tion ranges, it was not defined as a movement To

quantify movement patterns of fish based on recorded

signals, three levels of gaps in signal detection from

the same individual by the same receiver were tested

for definition as movements at intervals of−>10, >20

and >30 min

In order to estimate if the appearance of the gaps in

signal reception had periodical nature Fourier

Analysis using Igor-Pro software was performed on

pooled data from all fish and all receivers separately

for the two study periods

The chi-square goodness of fit test (Zar1999) was

used to determine if the observed movements from

one receiver to another and appearance of gaps during

high and low tide periods differed significantly from

the expected frequency if they occurred randomly,

following ratio of high: low tide period High tide

period was defined as that during which the predicted

water height (Hydrographic Department, Royal Thai

Navy) was 2 m or more above the lowest low water

The rest was defined as a low tide period This

resulted in 14 and 10 h of high and low tide periods a

day (i.e a ratio of 58:42) Thus the null hypothesis

was that the fish movement frequency had a 58:42ratio during high and low tide periods respectively.The chi-square goodness of fit test (Zar1999) wasalso applied to determine if the observed frequency ofmovements from one receiver to another and appear-ance of gaps during daylight and night-time differedsignificantly from the expected frequency, if theyoccurred randomly Daylight was defined as theperiod from 6 am to 6 pm, the reminder being nighttime This resulted in 12 and 12 h of day and nightperiods (i.e a ratio of 50:50) Thus the null hypothesiswas that the fish movement frequency had a 50:50 ratioduring day and night periods respectively

At first, heterogeneity chi-square analysis (Zar1999)was performed to test the “interaction” between theabove two factors The null hypothesis was that bothday and night samples has a 58:42 ratio of high to lowtide movements As it was rejected the chi-squaregoodness of fit test (Zar1999) was performed for theday and night samples separately The sequentialBonferroni test (Rice1989) was applied to adjust thesignificance level of the multiple tests The sameprocedure was applied to test if the movement ratio oflow and high tide periods had a 50:50 ratio of nightand daytime period These analyses were performedonly for data pooled in categories of total movementsfrom all individuals during high/low tide and day/night periods because 1) the objective of the test was

to examine the movement patterns of a group of

L argentimaculatus, 2) there was a limited number ofobservations for each individual

In order to examine the dependence of movementpattern on fish size, the mean TL of the group ofindividuals which left the creek during the first 7 days

of the study and did not return during the study periodand those which stayed in the study area wascompared using T test In addition, a linear regressionanalysis was applied to determine if within the formergroup there was a significant correlation between fishsize and days spent in the study area

Probability level atα=0.05 was considered icant in all tests Analyses were performed usingSPSS 17.0 software

signif-Results

Individual data on fish sizes, number of days trackedare shown in Table 1 Eighteen data sets were

obtained during the two study periods However, two

individuals (ID 113 and 263) were discarded from

further analysis due to the very high frequency and

fixed location of detections, indicative of dead fish or

shed transmitter The hydrophone receiver array

recorded 127 497 signal transmissions from the 16

analyzed individuals over 169 non-consecutive days,

the longest tracking period being 28 days and the

shortest 1 day (Table1)

Detection range of the receiver array

As a result of towing the activated transmitter along

the main channel detection ranges of each individual

receiver were obtained, and the resulting detection

area is shown in Fig 1 It was determined that the

signal was lost (no signal was detected) when entering

any of the side channels as well as mangrove fringe in

distances of 50, 150 and 250 m from a receiver The

results from the experiment to determine the influence

of tidal height on signal detection showed that the

signals were received continuously throughout two

full tidal cycles It was also shown that the noise from

passing boats (approx 10–30 m from the transmitter)

did not interrupt with signal detection

Definition of movements

The frequency of gaps in signal reception of

>10 min did not differ significantly from the

expected frequency of random appearance of gaps

(x2=2.23, p>0.05, df=1) during high and low tide

periods However, the frequency of gaps in signal

reception of >20 and >30 min differed significantly

from the expected frequency of random appearance of

gaps (20 min, x2=11.8, p<0.01, df=1; 30 min, x2=

40.4, p<0.001, df=1) Because it was believed that

the appearances of shorter gaps were partially

associated with, for example, local topography or

background noises blocking signal reception, they

were excluded from further quantitative analysis

Accordingly, for the final quantitative analysis, a

movement was defined as a gap in signal reception

for more than 30 min or the changing of detection

from one receiver to the other The gaps in signal

reception defined as movements were present in all

fish (from 80% to 95% of movements of individual

fish)

Fourier analysis revealed notable 12 h periodicity

in the appearance of movements from one receiver toanother and appearance of gaps in signal reception forfish from both study periods which is consistent withthe points of maximum tidal heights during thediurnal tidal cycle (Fig 2)

Movement patterns of L argentimaculatus

Most fish could be categorized according to theobserved movement pattern For 8 of 16 tagged fish(ID’s 261, 264, 266, 108, 110, 114, 115 and ID 269)the migrational pattern was characterized by a shorttime period spent in the study area (1–10 days, exceptID110) with a following exit towards the open coast(e.g ID 264, Fig.3a) For the first 3 days of tracking

ID 261 and ID 264 were found in the area aroundreceivers’ No 7 and 8 and 6, 7 and 8 respectively.Two days after release, ID 261 left the creek towardsthe open coast and did not return On the contrary ID

264 left and re-entered the study area several times,swimming as far upstream as station 7 and returningafter a short time period Nevertheless after 7 daysspent in the Sikao creek, ID 264 left the creek towardsthe open coast and did not return ID 266 stayed in thestudy area only for 1 day, using two following hightides to swim downstream and exit the creek All thefish in this group from the second study period spentfirst 3–8 days of tracking in the area between thestations 7 and 9 followed by exit towards the opencoast However during all the study period ID 110was irregularly detected by the receiver 1 indicatingthat it spent time in the deep trench which connectsthe Sikao creek to the open coast or in the coastal areaclose to the Sikao creek

For 5 of 16 tagged fish (ID’s 267, 268, 109,

112 and ID 262) the movement pattern waslocalized The fish spent the whole period oftracking in the same area and no exit towards theopen coast was detected (e.g ID268, Fig 3b) ID

267, ID 268, ID 109, and ID 262 were found in thearea between stations 6 and 9 On the contrary ID 112swam downstream and spent the tracking period inthe creek mouth All the individuals regularly left thearea of detection ranges of the respective receiversduring high tides

Three fish could not be included in the abovegroups (ID’s 265, 270, 111) ID 265 spent 3 daysaround receivers 6, 7 and 8 Only four signals from

the receivers’ 3 and 1 were detected during the rest of

the study period indicating that the fish moved to the

creek A In the first 5 days of tracking ID 270 stayed

in the area of receiver 5 After moving to the area

around receiver 1 the fish was caught by a sports

fisherman ID 111 spent 4 days around receiver 9 and

left the study area towards the open coast without

detection by the receivers 8 and 7 but with signals

detected by all other receivers This is unlikely to

happen in natural conditions and it may imply that ID

111 was caught by a fisherman and transported

downstream

Comparisons of movement patterns with biotic

parameters

The movements of individual fish and the fraction of

movements during high and low tides are summarized

in Table 1 Of the total movements, ten were

directional the rest were indicated by gaps in signal

reception On average 72.0% and 84.8% of the

movements during both day and night periods were

detected during high tide Heterogeneity chi-square

test showed that the frequency ratio of movements

during high and low tide was not homogeneous in

day and night periods (x2=4.2, p<0.05, df=1),

having greater skew toward high tide at night Thefollowing goodness of fit test detected significantdifference from the frequency of random movementsduring the high and low tide periods for both day andnight (day x2=9.5, p<0.01, df=1; night x2=39.1, p<0.001, df=1)

43.1% and 62.3% of high and low tide, respectively,were detected in daytime, having significant heteroge-neity in frequency ratio of day and night movementsbetween tidal periods (x2= 6.1, p < 0.05, df = 1).However, following goodness of fit test for each ofhigh and low tide period detected no significantdifference in the frequency of movements duringday and night from the expected frequency (i.e 50 :50; high x2=3.18, p>0.05, df=1; low x2=3.7, p>0.05, df=1)

The mean TL of fish that left the creek for theopen coast area during the study period wasdetermined to be 251 mm (211–303 mm) com-pared with 223 mm (192–252 mm) TL for thosethat did not leave the creek; the differencebetween the mean values was not significantlydifferent (t=1.23, p=0.24, df=11) Within the group

of individuals which left the creek for the open coast,there was no significant correlation between fish sizeand days spent in the creek (t=−.007, p=0.95, df=7)

Fig 2 Fourier analysis of

movement periodicity for:

first study period, and

second study period A 12

h periodicity was evident

for both groups

Influence of tidal cycle on the observed movement

pattern of L argentimaculatus

We found that majority of the fish movements were

associated with high tides and a signal from a

transmitter was shown to be lost when entering

mangrove fringe or side channels

Our observations showed that L argentimaculatus

used the mangrove side channels and/or forest for

feeding during high tides and returned to the main

channels during low tide periods High tides provide

access to increased food availability for fish in

mangrove systems (e.g Robertson and Duke 1987;

Salini et al.1990), triggering immigration of fish into

the mangrove habitats (Krumme and Saint-Paul2003;

Sheaves 2005; Meynecke et al 2008) Based on

observed changes in consumed food items, Sheaves

(2005) noted that L argentimaculatus entered a

mangrove forest in Australia to feed at high tide andretreated to subtidal areas during low tide.Furthermore, Sheaves and Molony (2000) showedthat crabs inhabiting mangrove forest floor (FamilySesaermidae) are a major food item of L argentima-culatus in Australia Our own observations show thatsesarmid crabs are an abundant group of benthicanimals in Sikao creek’s mangrove forests leading to

a suggestion that they are a sought-after food item of

L argentimaculatus in the side creeks and/or grove forest during high tide periods High tides arealso characterized by higher densities of small andjuvenile fishes in mangrove creek and forest habitats(Thayer et al.1987; Ikejima et al.2003) Piscine preyhas been repeatedly shown to be of importance forseveral lutjanid species (e.g Rooker1995; Kulbicki et

man-al 2005) Additionally it has been found thatjuveniles of grey snapper were hunting for smallfishes in mangrove prop root habitat in South Florida(Thayer et al.1987)

Fig 3 Time series of signal transmissions from ID 264 (a) and

ID 268 (b) detected at individual receivers plotted against

corresponding variations in tidal height 1, 3, 6, 7, 8 indicate

receiver numbers Each black dot represents a separate signal reception Water depths are given as predicted water heights above extreme low water

Sheaves (2005) implied that predation is another

factor modifying the tide-related migrational patterns

of fish in mangrove habitats, suggesting that during

the periods when the forest is exposed, fish move to

alternative microhabitats, such as fallen trees, to avoid

predators Considering the relatively large size of the

individual fish studied here and the declining number

of large carnivorous fish, such as barracuda

(Sphyraena spp.) and barramundi (Lates calcarifer),

in the catches of local fishermen during the last

10 years (Tongnunui, personal observation) a strong

predation pressure on the studied fish is unlikely

within the study area This is further supported by

Meynecke et al (2008) who showed that during the

period a mangrove creek in Australia was accessible

larger Lutjanus russellii entered later and exited

earlier than smaller individuals suggesting that they

use the creek mostly for foraging in contrast to

smaller fish which spend more time in the creeks

due to the shelter provided by mangrove roots We

found no correlation between fish size and days spent

in the creek Thus we propose that in the current study

all L argentimaculatus used mangrove side channels

and/or forest mostly for feeding

Site fidelity of L argentimaculatus

Six fish spent up to 28 days in the study area within

creek B (2 km long) revealing a considerable degree

of site fidelity (Fig.3b) Previous studies have shown

that other species of Lutjanidae stay site attached in

estuarine habitats for a long time before reaching

certain size ca 400 mm in mangrove jack (Sheaves

1995) and moving to permanent adult habitats (e.g

Russell et al.2003; Nanami and Yamada2008) It has

also been observed that L argentimaculatus are

associated with structural habitat such as mangrove

prop roots (Russell et al 2003) The observed site

fidelity may be due to that the study area provided

suitable permanent habitat for a part of juvenile

mangrove jack population

However, most of the studied fish moved towards

the open coast The range of TL (192–303 mm) of the

fish in the present study was within the size range

(71–541 mm TL) of juvenile L argentimaculatus

found in the mangrove estuaries of northeast Australia

(Sheaves 1995), so we assume that this movement

was not explained by permanent migration to adult

habitats Our view is supported by Russell and

McDougall (2005) findings that juvenile mangrovejack travel up to 130 km within coastal habitats inAustralia without moving to permanent offshorehabitats Further proof comes from Hammerschlag-Peyer and Layman (2010) who showed that twospecies of lutjanids exhibited considerable intrapopu-lation variability in movement patterns instead ofshowing a constant pattern across the respectivepopulations Thereby our observations could beexplained by a part of mangrove jack migrating toother coastal habitats (i.e nearby mangrove systems,rocky and seagrass habitats) possibly in order tooptimize foraging or reduce intraspecific competition(e.g Hammerschlag-Peyer and Layman 2010)

Influence of diurnal cycle on the movement patterns

of L argentimaculatus

Previous studies have found contradicting resultsconcerning changing activity patterns of snappers inrelation to diurnal cycle Starck (1971) andNagelkerken et al (2000) found that lutjanids areinactive during day time but shift to active feedingbehavior during night time Duarte and Garcia (1999)showed that mutton snapper (Lutjanus analis) isactively feeding throughout the day L argentimacu-latus in this study were moving more actively duringnight high tides Krumme and Saint-Paul (2003)observed a significantly higher flux of migrating fish

in the Brazilian mangrove creeks during night timeand our unpublished data showed a higher abundance

of fish in night time samples (Ikejima et al.2003) Inaddition it is known that sesarmid crabs are moreactive during night time (e.g Seiple and Salmon

1982; Moser and Macintosh 2001) Thus the nal activity peak could be explained by an increasedabundance of mangrove jack’s foods in the mangroveforest and/or side channels during night time

noctur-Limitations of the study

It has to be noted that due to methodologicallimitations some interpretations of the observedmigrational patterns of L argentimaculatus have to

be viewed with caution Due to logistical limitationsthe studied fish were taken from aquaculture cagesand all the individuals were released at the samelocation in the middle of the study area which couldinfluence their behavior The fact that several

individuals left the creek towards open coast after a

relatively short time period could partially be explained

by unnatural behavior caused by this limitation Another

methodological shortcoming of the study was the

limited coverage of the receiver array which did not

extend to creek A and nearby coastal habitats

Increasing the coverage of the receiver array and the

length of experiment may reveal variation of individual

migrational patterns, and possible movements between

coastal nursery habitats

Conclusions

Ultrasonic telemetry successfully depicted the

migra-tional pattern of L argentimaculatus in the mangrove

creek We concluded that the tidal cycle is an

important environmental cue, determining the short

term migrational pattern of juvenile mangrove jack in

Sikao Creek, and suggest that individual fish utilize

the side creeks and/or mangrove forests for foraging

during high tide periods, retreating to the main creek

during low tides Evidence was also found for part of

the juvenile L argentimaculatus population leaving

Sikao Creek for other coastal habitats instead of

staying site attached This supports Sheaves (2005),

who indicated that mangroves are part of an

interconnected habitat mosaic and should be studied

in the context of connectivity with other habitats The

present data, together with the observed fishing

pressure on mangrove jack, revealed a pressing need

to develop a scheme for sustainable management of

the mangrove habitat in Thailand, so as to provide

protection for the habitat and associated fish stocks

from growing human interference

University for their help during the fieldwork, and M Holmer

for provided helpful discussion and advice We are grateful to

the National Research Council of Thailand for the granting of

permits to conduct this research in Thailand We also thank G.

Hardy for English correction, and anonymous reviews for

invaluable comments on previous drafts.

References

Blaber SJM (2000) Tropical estuarine fishes, ecology,

exploi-tation and conservation Blackwell Science, Cleveland, p

372

Cocheret de la Morinière E, Pollux BJA, Nagelkerken I, van der Velde G (2002) Post-settlement life cycle migration patterns and habitat preference of coral reef fish that use seagrass and mangrove habitats as nurseries Estuar Coast

Doi M, Singhagraiwan TSS (1994) Juvenile red snapper, Lutjanus argentimaculatus, occurring along the Eastern

Duarte LO, Garcia CB (1999) Diet of the mutton snapper Lutjanus analis (Cuvier) from the Gulf of Salamanca,

Frias-Torres S, Barroso P, Eklun AM, Schull J, Serafy JE (2007) Activity patterns of three juvenile goliath grouper (Epinephelus itajara) in a mangrove nursery Bull Mar Sci

of the characiform Triportheus guentheri Environ Biol Fishes 50:319–330

Hammerschlag-Peyer CM, Layman CA (2010) Intrapopulation variation in habitat use by two abundant coastal fish species Mar Ecol Prog Ser 415:211–220

Hartill BW, Morrison MA, Smith MD, Boubee J, Parsons DM (2003) Diurnal and tidal movements of snapper (Pagrus auratus, Sparidae) in an estuarine environment Mar

Ikejima K, Tongnunui P, Medej T, Taniuchi T (2003) Juvenile and small fishes in a mangrove estuary in Trang Province, Thailand: seasonal and habitat differences Estuar Coast

Jepsen N, Koed A, Thorstad EB, Baras E (2002) Surgical implantation of telemetry transmitters in fish: how much

Kaunda-Arara B, Rose GA (2004) Homing and site fidelity in the greasy grouper Epinephelus tauvina (Serranidae) within a marine protected area in coastal Kenya Mar Ecol Prog Ser 277:245–251

Kimani EN, Mwatha GK, Wakwabi EO, Ntiba JM, Okoth BK (1996) Fishes of a shallow tropical mangrove estuary, Gazi, Kenya Mar Freshw Res 47:857–868

Krumme U (2004) Patterns in tidal migration of fish in a Brazilian mangrove channel as revealed by a split-beam

Krumme U, Saint-Paul U (2003) Observations of fish migration in

a macrotidal mangrove channel in Northern Brazil using a

Kulbicki M, Bozec YM, Labrosse P, Letourneur Y, Mou-Tham

G, Wantiez L (2005) Diet composition of carnivorous fishes from coral reef lagoons of New Caledonia Aquat

Laegdsgaard P, Johnson CR (1995) Mangrove habitats as nurseries: unique assemblages of juvenile fish in subtrop- ical mangroves in eastern Australia Mar Ecol Prog Ser 126:67–81

Luo JG, Serafy JE, Sponaugle S, Teare PB, Kieckbusch D (2009) Movement of gray snapper Lutjanus griseus among subtropical seagrass, mangrove, and coral reef habitats.

Meynecke JO, Poole GC, Werry J, Lee SY (2008) Use of PIT tag and underwater video recording in assessing estuarine fish movement in a high intertidal mangrove and salt

Mitamura H, Arai N, Sakamoto W, Mitsunaga Y, Tanaka H,

Mukai Y, Nakamura K, Sasaki M, Yoneda Y (2005) Role

of olfaction and vision in homing behavior of black

rockfish Sebastes inermis J Exp Mar Biol Ecol 322:123–

134

Moser SM, Macintosh DJ (2001) Diurnal and lunar patterns of

larval recruitment of brachyura into a mangrove estuary

Mumby PJ, Edwards AJ, Arias-Gonzalez JE, Lindeman KC,

Blackwell PG, Gall A, Gorczynska MI, Harborne AR,

Pescod CL, Renken H, Wabnitz CCC, Llewellyn G (2004)

Mangroves enhance the biomass of coral reef fish

Nagelkerken I, Dorenbosch M, Verberk WCEP, Cocheret de la

Morinière E, van der Velde G (2000) Importance of

shallow water biotopes of a Caribbean bay for juvenile

coral reef fishes: pattern in biotope association,

communi-ty structure and spatial distribution Mar Ecol Prog Ser

202:175–192

Nagelkerken I, Roberts CM, van der Velde G, Dorenbosch M,

van Riel MC, Cocheret de la Morinière E, Nienhuis PH

(2002) How important are mangroves and seagrass beds

for coral-reef fish? The nursery hypothesis tested on an

Nakamura Y, Horinouchi M, Shibuno T, Tanaka Y, Miyajima T,

Koike I, Kurokura H, Sano M (2008) Evidence of

ontogenetic migration from mangroves to coral reefs by

black-tail snapper Lutjanus fulvus: stable isotope

Nanami A, Yamada H (2008) Size and spatial arrangement of

home range of checkered snapper Lutjanus decussatus

(Lutjanidae) in an Okinawan coral reef determined using a

Rice WR (1989) Analyzing tables of statistical tests Evolution

Robertson AI, Blaber SJM (1992) Plankton, epibenthos and

fish communities In: Robertson AI, Alongi DM (eds)

Tropical mangrove ecosystems American Geophysical

Union, Washington, pp 173–224

Robertson AI, Duke NC (1987) Mangroves as nursery sites:

comparisons of the abundance and species composition of

fish and crustaceans in mangroves and other nearshore

Robertson AI, Duke NC (1990) Mangrove fish-communities in tropical Queensland, Australia: spatial and temporal patterns in densities, biomass and community structure Mar Biol 104:369–379

Rooker JR (1995) Feeding ecology of the Schoolmaster snapper Lutjanus apodus (Walbaum), from Southwestern

Russell DJ, McDougall AJ (2005) Movement and juvenile recruitment of mangrove jack, Lutjanus argentimaculatus (Forsskål), in northern Australia Mar Freshwr Res

Aus-to predation on penaeid prawns Mar Biol 105:363–374 Seiple WH, Salmon M (1982) Comparative social behavior of two grapsid crabs, Sesarma reticulatum (Say) and S cinereum (Bosc) J Exp Mar Biol Ecol 62:1–24

Sheaves MJ (1992) Patterns of distribution and abundance of fishes in different habitats of a mangrove-lined tropical estuary, as determined by fish trapping Aust J Mar Freshw

Sheaves M, Molony B (2000) Short-circuit in the mangrove

Starck W (1971) Biology of the Gray snapper Lutjanus griseus (Linnaeus), in the Florida Keys In: Starck WI, Schroeder

R (eds) Investigations on the gray snapper, Lutjanus griseus Studies in Tropical Oceanography No 10 University of Miami Press, Coral Gables, pp 13–150 Thayer GW, Colby DR, Hettler WF Jr (1987) Utilization of the red mangrove prop root habitat by fishes in south Florida.

Zar JH (1999) Biostatistical analysis, 4th edn Prentice-Hall, New Jersey, p 663

Winter energy allocation and deficit of juvenile walleye

pollock Theragra chalcogramma in the Doto area,

northern Japan

Kouji Kooka&Orio Yamamura

Received: 31 March 2011 / Accepted: 27 October 2011 / Published online: 19 November 2011

# Springer Science+Business Media B.V 2011

Abstract Seasonal energy allocation and deficits of

marine juvenile fishes have considerable effects on

their survival To explore the winter survival mechanism

of marine fishes with low lipid reserves in their early

life, juvenile walleye pollock Theragra chalcogramma

were collected along the continental shelf of northern

Japan over a 2-year period, and energy allocation and

deficit patterns were compared between wild and

laboratory-starved fish Contrary to expectations, wild

fish generally continued to accumulate protein mass

and concurrently tended to reduce lipid mass from

late autumn through winter The most plausible

explanation for the continuous structural growth is

that juvenile pollock give priority to reducing

mor-tality risk from size-selective predators under

quasi-prey-limited conditions Exceptionally, inshore small

fish reduced both constituents during a winter The

inshore fish consumed 2.5 times more lipid energy

than protein energy in November–December, but

protein was more important than lipids as a source

of energy in December–January and in February–

March However, dependence upon protein reserves

was lower for the wild fish than for the

laboratory-starved fish, suggesting milder nutritional stress of the

wild fish than that observed in the starvationexperiment Moreover, the lipid contents of mortal-ities in the starvation experiment were mostly <1%,whereas few wild fish had such lipid contents in thefield These results suggest that juvenile pollock areable to avoid both starvation and predation byaccumulating protein reserves

Keywords Energy reserves Nutritional condition Metabolic fuel Overwintering Recruitment Survival

Introduction

Several groups of animals have the ability to withstandlong-term starvation by using stored energy reserves.These organisms use three types of body constituents(carbohydrate, lipid, and protein) to meet metabolicdemands under severe conditions, although carbohy-drate content makes up only a small fraction in teleostfishes (Weatherley and Gill 1987) The relativepredominance of these constituents depends uponthe duration of nutritional stress (Molony 1993;Hervant et al 2001) Méndez and Wieser (1993)defined four sequential phases in the metabolicresponse to starvation: 1) stress: animals use carbo-hydrate as the major metabolic substrate during earlyperiods of starvation, 2) transition: the major meta-bolic substrate switches from carbohydrate to lipid, 3)adaptation: the major metabolic substrate switchesfrom lipid to protein during extended periods of

DOI 10.1007/s10641-011-9957-1

Hokkaido National Fisheries Research Institute,

Fisheries Research Agency,

Kushiro,

Hokkaido 085-0802, Japan

e-mail: kkooka@fra.affrc.go.jp

starvation, and 4) recovery: a period of rapid growth

and initiation of energy storage if food once again

becomes available Therefore, temporal changes in

these constituents of wild animals may provide useful

information for exploring levels of nutritional stress

Among temperate juvenile fishes, winter mortality

is a common phenomenon that affects spatial

distri-bution and recruitment variability (Johnson and Evans

1990; Hurst and Conover1998), whereas the fishes in

the tropics may not suffer from the mortality due to

the low fluctuations of water temperature and prey

availability Several causal mechanisms underlie

winter mortality in fishes: thermal stress, starvation,

predation, parasites, and pathogens (Hurst 2007)

Because fishes with low lipid reserves more often

suffer from mortality caused by these mechanisms

(Hurst2007), it is important that juvenile fishes store

lipids to survive the winter However, some juvenile

fishes in sub-arctic marine systems, such as gadoid

fishes, exhibit low lipid reserves compared to other

sympatric juvenile fishes (Van Pelt et al 1997;

Anthony et al.2000) The winter survival mechanism

of these exceptional fishes remains uncertain

Walleye pollock, Theragra chalcogramma, is a

semi-demersal gadoid fish and one of the most important

fishery resources in the world (FAO 2009) Juvenile

walleye pollock are zooplanktivorous (Ciannelli et al

2004) and are preyed upon by bottom-dwelling fishes

and seabirds (Hatch and Sanger 1992; Yamamura

2004) Therefore, this species is a main component of

the sub-arctic marine systems of the North Pacific

(Springer 1992; Yamamura2004)

The Doto area is the most important nursery

ground for walleye pollock around the southern part

of Hokkaido Island (Fig.1) Juvenile walleye pollock

migrate from their main spawning ground to the

continental shelf of this area during September and

reside there until maturity (Shida 2002; Honda2004;

Honda et al 2004) Bottom water temperatures are

warm (up to 12°C) in September–October and

extremely cold (0–1°C) in March–April (Yamamura

2004) The mean abundance of mesozooplankton also

fluctuates from a maximum (730 mg m−3) in May–

July to a minimum (20 mg m−3) in January–March on

a wet-mass basis (Saito et al.1998; Kasai et al.2001)

The objective of the present study was to explore

the winter survival mechanism of juvenile walleye

pollock First, we collected wild fish in the Doto area

over a 2-year period and examined seasonal energy

allocation and deficit patterns during periods of preyscarcity Second, we conducted a laboratory experimentand compared energy deficit patterns between wild andlaboratory-starved fish

Materials and methods

Fig 1 Study area in the North Pacific, the southern part of Hokkaido Island, and the Doto area with the location of the main spawning ground (hatched area), juvenile sampling site for the laboratory experiment (star), juvenile summer migration route (arrows), and the location of monthly sampling (rectangle)

station, up to 100 fish were randomly selected and

placed in plastic bags on ice in a cooler box The

water temperature proximate to the sea floor was

measured at each station using a salinity/temperature/

depth sensor In addition, the water temperature was

measured at a 180 m depth station from August to

November 2004

In the laboratory, fish were measured to the nearest

1 mm (fork length) and weighed to the nearest 0.1 g

For the analysis of seasonal variation in biochemical

composition, we randomly subsampled 5–10 fish

(generally 10) from each seine sample To compare

lipid contents between direct and indirect methods,

we selected 30 samples to cover a wide range of lipid

contents and subsampled two or three fish from each

sample These fish were subsequently placed in

separate plastic bags and stored at −30°C for later

biochemical analyses Abundance (the number of fish

per seine) was calculated by dividing the catch by the

mean individual body mass for each seining

Starvation experiment

Juvenile walleye pollock were obtained from the main

nursery area for early juveniles in June 2004 (Fig.1)

These fish were transported to the Hokkaido National

Fisheries Research Institute, Kushiro, where they

were then transferred to two, 800-L circular tanks

with flow-through seawater filtered by sand Fish

were maintained at 8±1°C (mean ± SD) under a

12-hour light:12-12-hour dark photoperiod and were fed

thawed frozen euphausiid Euphausia pacifica to

satiation daily

To explore the effects of initial lipid reserves on

changes in lipid and protein contents of food-deprived

fish over time, we manipulated the lipid reserves of

fish in the two holding tanks in early January 2005

For the high-lipid treatment, after the water

temper-ature was lowered to 5.5°C, the fish were fed 3%

body mass d−1until the onset of the experimental set

up For the low-lipid treatment, after the water

temperature was raised to 10°C, the fish were

deprived of food for 2 weeks Thereafter, the

temperature was lowered to 5.5°C, and the fish were

fed 1% body mass d−1 until the onset of the

experimental set up In all cases, water temperatures

were changed at a rate of 1°C d−1

The experimental design consisted of two

treat-ments with two replicates each (four tanks in total)

Fish from each holding tank were randomly selectedand transferred to two of the four 275-L experimentaltanks on 8 February 2005 Prior to the transfer, thesefish were deprived of food for 48 h and measured tothe nearest 1 mm after anesthesia with methyl p-aminobenzoate; however, they were not taggedbecause of their sensitivity to handling (Smith et al

1986) Each experimental tank contained either 31fish (123±11mm and 121±9mm) with high lipidreserves or 33 fish (122±9mm and 123±9mm) withlow lipid reserves At the same time, we randomlyremoved an additional 15 fish from each holdingtank to determine their initial lipid reserves Thesefish were killed immediately with an overdose.After the temperature of the tanks was loweredfrom 5.5°C to 0.5°C (1°C d−1), with a ration of 1%body mass d−1, the experiment was initiated immedi-ately This additional manipulation likely had littleeffect on the estimation of initial energy reserves.According to a bioenergetics model developed byCiannelli et al (1998), the expected body mass gainwas only 0.3% after the 5-day period of manipulation.The experiment ran for 56 days, during which eachtank was checked at least twice a day All dead fishwere removed from the tanks On days 35 and 56, 15surviving fish were randomly removed from eachtreatment (7 or 8 fish per tank), except for the low-lipid treatment on day 56, when only 13 fish survived.The removed fish were killed immediately All dead

or randomly removed fish were measured and frozenfor future biochemical analysis

Measurement of energy density and proximatecomposition

The whole body of individual fish was freeze-driedand ground with a mortar Energy density wasmeasured using an adiabatic bomb calorimeter.Protein content was calculated by multiplying thenitrogen content by a conversion factor of 6.25 Thenitrogen content of dry tissue was measured using anelemental analyzer calibrated with acetanilide Lipidcontent was estimated indirectly, because the drymass of an individual fish was usually insufficient forthe direct measurement of both energy density andlipid content The calculation followed Harris et al.(1986): Lipid mass = (Body dry mass × Energydensity—Protein mass×23.5) / 39.5 To correct anybias in the calculated values, we extracted total lipids

directly for 30 subsamples (2–3 fish each) using a

Soxhlet apparatus and a solvent system of 7:2 (v/v)

hexane/isopropyl alcohol (Anthony et al 2000), and

the remaining tissue was used for the indirect

measurements Water content was determined as the

difference between wet mass and dry mass Ash

content was measured by a muffle furnace at 550°C

for 12 h In the present study, energy density (kJ g−1)

and lipid, protein, and ash contents (%) were

expressed on a wet-mass basis

Analysis of seasonal body condition and energy

allocation

Seasonal body condition and energy allocation of

juvenile pollock were assessed by monthly changes in

the absolute values of whole-body energy (kJ fish−1) and

the mass of lipid and protein (g fish−1) The use of

relative values could result in erroneous assessments

(Caulton and Bursell1977) Monthly changes in the

whole-body energy and the mass of lipid and protein

were calculated using the expected mean body mass

(wet mass) and the measured means of energy density

and lipid and protein contents for each month The

expected body mass was estimated from a growth

curve to minimize any sampling bias A third-order

polynomial line (Brodeur and Wilson 1996), which

provides a simple and appropriate representation of

reduced growth during winter, was fitted to the

monthly measured means of body mass The analysis

was conducted separately for the fish collected in the

inshore (30–90 m) and offshore (120 and 150 m)

habitats because growth patterns differed between

habitats during the 2003–2004 season

To examine the extent of nutritional stress in the

fish during the period, the ratio of consumed lipid and

protein (L/P) was calculated for the field and

laboratory-starved fish The L/P was calculated on

an energy basis because lipid energy equivalent was

1.7 times higher than protein energy equivalent in the

present study Consumed lipid and protein energy

were calculated on a daily basis (J d−1fish−1) because

sampling intervals were not equal

Statistics

For the field samples, we performed multivariate

analyses of variance (MANOVAs) to confirm seasonal

variation in energy density and proximate composition

Data from individual wild fish were adjusted to those of125-mm fish (approximate grand mean) from theallometric relationships in each month Abundance data

in the present sampling design is not appropriate for thestatistical comparison of abundance between the inshoreand offshore habitats because bathymetric distributions

of juvenile walleye pollock are highly skewed (Honda et

al 2004) Thus, in each month, the abundance ofsampling stations other than the station showing peakabundance tended to be much lower For thelaboratory experiment, we performed repeated meas-ures analyses of variance (RM-ANOVAs) to explorethe effects of treatment and time on lipid and proteincontents In the analysis, each tank mean was treated

as an experimental unit The 15 fish removed fromeach of the treatment holding tanks were randomlyassigned as the initial data of the replicated experi-mental tanks of each treatment (7 or 8 fish per tank).Prior to analysis, energy density data were ln-transformed, and percentage data were arcsinesquare-root transformed to meet the assumptions ofthese statistical models (Scheiner and Gurevitch

2001) To test whether remarkable size-selectivewinter mortality occurs or not, mean length andcoefficient of variation (CV) were regressed againstsampling date according to Hurst and Conover(1998) This analysis is based on the assumption that

if size-selective mortality is the causal mechanism in afish population with no growth, mean length increasesand CV decreases in the length-frequency distribu-tions (Hurst and Conover 1998) In the presentanalysis, data from late October through April wereused as a conservative approach because juvenilepollock may grow in September from the analysis ofotolith microstructure (Nishimura et al 2007) andbecause an opaque zone was formed at the otolithedge of juvenile pollock in May (Kooka andYamamura, unpubl data), suggesting new growth inthe early spring Significance levels were set at 0.05.All statistical analyses were performed using JMPversion 6.0 (SAS 2005)

Results

Field sampling

In the 2003–2004 season, juvenile walleye pollocktended to be more abundant in the offshore habitats

from late summer through winter, but few offshore

fish were collected in the spring (Fig 2a) The

offshore fish were much larger than the inshore fish

during the winter (Fig.2c) In the 2004–2005 season,

fish abundance was much higher in the inshore

habitats than in the offshore habitats throughout the

sampling period (Fig.2b) Offshore abundance was at

most 26 ind seine−1in winter, and then 3 ind seine−1

in spring No clear size separation was observed

between the inshore and offshore habitats, although

there appears to be a difference in size between 60 m

and 150 m depths from late summer through winter

(Fig 2d) In both seasons, abundance in the 30 m

depth was <100 ind seine−1

Bottom water temperature exhibited pronounced

seasonality (Fig 3) The mean bottom temperatures

from late summer through autumn were higher than

6.0°C in 2003 and 5.5°C in 2004 The temperature in

December 2003 and 2004 was 7.4°C and 5.5°C,

respectively The temperature dropped abruptly by

January of both years, decreased further from 2.5°C to

0.3°C by March 2004 and from 2.2°C to 0.4°C by

April 2005, and then began to rise in May

The lipid contents of juvenile pollock measuredusing the direct and indirect methods correlated wellwith each other: Y=0.99X−0.53 (0.9≤X≤5.7, r2

=0.95,P<0.0001), where Y and X are lipid contents obtained

by the indirect and direct methods, respectively Theslope of the regression did not differ from 1 (t28=−0.97,P=0.34), and the intercept differed significantly from

0 (t28=−3.61, P<0.01) Thus, we added 0.53 to thecalculated lipid content value to correct for this bias.The proximate composition and energy densities ofjuvenile pollock showed clear seasonality, with asignificant effect of month in both the 2003–2004 season(MANOVA, Wilks’s lambda=0.0028, F32,75=9.2, P<0.0001) and the 2004–2005 season (Wilks’s lambda=0.0012, F40,93=11.5, P<0.0001) Both energy densityand lipid content peaked by early autumn, decreased

by early winter, and remained low in winter beforeincreasing again by the end of spring (Fig 4a, b).Protein content tended to be high until the end ofautumn and low thereafter (Fig 4c) Water contentshowed the opposite pattern from that of lipid content(Fig 4d) Ash content did not exhibit a clearconsistent trend between sampling seasons (Fig.4e)

a

c

b

d

Fig 2 Seasonal bathymetric

distribution in the abundance

and mean fork length of

juvenile walleye pollock

from September 2003

through June 2004 (a, c) and

from August 2004 through

Starvation experiment

The initial lipid contents of juvenile pollock in the

high- and low-lipid treatments were 1.83±0.08 and

0.84±0.06% (mean±SE), respectively Lipid content

decreased more rapidly by day 35 in the high-lipid

treatment (0.02% d−1) than in the low-lipid treatment(0.0014% d−1; Fig 5) RM-ANOVA reflected thesevarying patterns, with a significant interaction betweentreatment and time (treatment: F1,2=116.4, P<0.01;time: F2,4=73.2, P<0.001; interaction: F2,4=44.2, P<0.01) The effects of treatment and time on proteincontent were not significant (10.1±0.4−11.8±0.3%,treatment: F1,2=6.0, P=0.13; time: F2,4=3.6, P=0.13;interaction: F2,4=0.4, P=0.68) The differences in therates of decrease of lipid content between treatmentsresulted in different utilization patterns of lipid andprotein energy (Table 1) In the high-lipid treatment,lipid and protein were consumed evenly during thefirst 35-day period (L/P was 0.93), and mainly proteinwas consumed during the second 21-day period (L/Pwas 0.23) In the low-lipid treatment, protein was themain energy source throughout the experiment(Table 1) The individual lipid contents of fishmortalities averaged 0.74±0.03%, and 91% of mortal-ities had lipid contents below 1% (Fig.6)

Fig 3 Seasonal variation in bottom-water temperature Error

Fig 4 Seasonal variation in

(a) energy density, (b) lipid

content, (c) protein content,

(d) water content, and (e)

ash content of juvenile

walleye pollock These data

were not adjusted for size.

Error bars represent ±1 SE

Seasonal energy allocation and deficit

Regression analysis showed that mean fork length of

juvenile pollock increased from mid autumn through

late winter, whereas CV did not decrease during the

period (Fig 7) The slope of the regression was

significant for both mean length (t4=3.00, P=0.04)

and CV (t4=6.39, P=0.003) in the 2003–2004 season

Both minimum and maximum lengths increased from

83 mm to 116 mm and from 157 mm to 208 mm,

respectively, during the period In the 2004–2005

season, the slope was significant for mean length (t4=

7.05, P=0.002), but was not for CV (t4= −0.05, P=

0.96) Also, both minimum and maximum lengthsincreased from 78 mm to 129 mm and from 157 mm

to 188 mm, respectively, during the period

The seasonal body condition of juvenile pollockdiffered between habitats and years In the 2003–2004season, the whole-body energy of inshore fish tended

to decrease from October through March, whereas thebody energy of offshore fish remained relativelyconstant from October through January and thenincreased (Fig 8c) In the 2004–2005 season, thebody energy generally increased throughout theseason (Fig.8d)

Juvenile pollock tended to allocate acquired tein energy to structural growth at the expense of lipidaccumulation from late autumn through winter In the2003–2004 season, both inshore and offshore fishaccumulated lipid mass by October that made up 20.0and 20.4% of the increased body organic massincrements for inshore and offshore fish, respectively(Fig 8e, g) Subsequently, the inshore fish tended tolose both lipid and protein mass until March, whereasthe offshore fish showed gains in protein massaccompanied by a loss of lipid mass until February(Fig 8e, g) In the 2004–2005 season, both inshoreand offshore fish accumulated lipids by September,comprising 40.6 and 58.0% of the increased bodyorganic mass increments for inshore and offshore fish,respectively (Fig 8f, h) Subsequently, inshore andoffshore fish exhibited gains in protein until April andFebruary, respectively, whereas they tended to lose orretain lipids; however, they accumulated considerableamounts of lipids in January and March, making up

pro-Fig 5 Changes in lipid content during 56 days of the

starvation experiment for juvenile walleye pollock in the

high- and low-lipid treatments Error bars represent ±1 SE

Table 1 Mean length and daily

changes in lipid and protein

energies of juvenile walleye

pollock in the starvation

experi-ment and in the field (inshore

habitat from late autumn

13.6–21.9% of their increased body organic mass

increments (Fig.8f, h)

In the 2003–2004 season, the inshore fish switched

their main source of energy from lipid to protein from

late autumn through the first half of winter Themonthly changes in lipid and protein energy indicatedthat the inshore fish consume both constituents asmetabolic fuel after summer (Table 1) The inshorefish consumed 2.5 times more lipid energy thanprotein energy in November–December, but proteinwas more important than lipids as a source of energy

in December–January and in February–March The L/Pduring these periods were higher than those oflaboratory-starved fish during the second 21-day period

of the high-lipid treatment During this particular winter,95% of the field-collected fish had lipid contents above1% (Fig.6)

Discussion

Bathymetric distribution

There were some variations in bathymetric tion inter-annually and seasonally The relationshipsbetween sampling depth and abundance showed thatoffshore fish tended to be more abundant than theinshore fish from late summer through winter in the2003–2004 season and vise versa in the 2004–2005season The difference in the inshore/offshore distri-bution pattern seems to reflect initial distributionpattern in late summer rather than the presence orabsence of the ontogenetic habitat shift to offshorehabitats The maximum offshore abundance (7,817ind seine−1) was 6-fold higher than the maximuminshore abundance in September 2003 Also, themaximum inshore abundance (1,501 ind seine−1)was 6-fold higher than the maximum offshoreabundance in the August 2004 Large-scale habitatshift within the shelf may not occur during theseasons

distribu-Juvenile pollock were apparently absent from theoffshore habitats in the spring of the 2003–2004season and in the winter and spring of the 2004–2005season First, in the spring of the 2003–2004 season,

it is probable that smaller offshore fish moved toinshore habitats and larger offshore fish moved tomid-water layers Length-frequency distributions inMarch and May 2004 suggest that smaller offshorefish (130–160 mm) in March were collected in theinshore habitats in May, whereas larger offshore fish(160–210 mm) were not collected in the habitats(Kooka et al 2009) Moreover, in the Doto area,

Fig 7 Overall mean fork length and coefficient of variation of

juvenile walleye pollock from (a) late October 2003 through

early March 2004 and (b) from early November 2004 through

mid April 2005 Solid and broken lines represent regression

lines Error bars represent ±1 SE

Fig 6 Lipid content of fish mortalities in the laboratory

experiment and inshore fish from January through March 2004

juvenile pollock occur both near the bottom and in the

mid-water layers in spring (Shida2002; Honda2004)

Second, in the winter of the 2004–2005 season, there

were fewer offshore fish although the offshore fish

seemed to perform better (Fig 8) One possible

explanation is that the offshore fish were cannibalized

by the winter In autumn, adult walleye pollock aregenerally abundant in the slope region (>150 mdepth)(Shida 2002), and warmer water temperatures

in the shelf region prevent adult pollock fromcannibalism (Yamamura et al 2001) In August

2004, bottom water temperatures in the offshore

Fig 8 Seasonal variation in

(a, b) body mass, (c, d)

whole-body energy, (e, f)

absolute mass of lipids, and

(g, h) protein of juvenile

walleye pollock in the

offshore (120 and 150 m)

habitats In seasonal

varia-tion in body mass (a), error

bars represent ±1 SE, and

growth curves were fitted to

the monthly mean body

mass, where Y is the

expected body mass and X

is number of days since 1

August In the offshore

habitats, the analysis was

restricted to mid-winter

because no or insufficient

fish were sampled to obtain

reliable mean body mass

data after mid-winter Mean

body mass data for offshore

fish in December 2003 were

excluded from the curve

fitting because of large

deviations from the

expected line

habitats ranged from 5.8°C to 8.0°C, whereas the

temperature at 180 m depth was 3.5°C Low water

temperatures of 2.6–3.4°C were observed in the

offshore habitats in October and November 2004,

but not in the autumn 2003 These results suggest that

the shoreward development of the slope water would

allow adult pollock to intrude the offshore habitats

Sequential phases in metabolism during starvation

In teleost fishes, proteins are initially conserved at the

expense of lipids during starvation (Weatherley and

Gill 1987) However, in the present study, the

transition phase (lipids as the main source of energy)

was not observed in the laboratory, whereas wild

juvenile pollock mainly used lipids in November–

December 2003 These results are probably due to the

difference in initial lipid reserves between wild and

laboratory fish A lean fish species, the largemouth

bass Micropterus salmoides, concurrently used lipids

and proteins in similar proportions, whereas a fatty

fish, the herring Clupea harengus, initially used

lipids, suggesting that the difference is related to

initial lipid reserves (Niimi1972)

Under the high-lipid treatment, juvenile pollock

consumed lipid and protein energy evenly during the

first period, whereas >80% of lost energy was

obtained from protein during the second period

These results suggest that the fish were in the early

adaptation phase at the start of the experiment

Juvenile pollock in the low-lipid treatment consumed

mainly protein throughout the experiment In the

high-lipid treatment, cumulative starvation mortality

was 1.6% after the first period and 19.1% after the

second period, whereas in the low-lipid treatment,

mortality rates were 28.8% and 74.5% after the first

and second periods, respectively (Kooka et al.2007a)

The low-lipid fish likely entered the terminal adaptation

phase during the second period, as this phase ultimately

leads to mass mortality (Castellini and Rea 1992;

Gibney et al.2003)

High survival in the high-lipid treatment after the

first period suggest that a low water temperature of

0.5°C has no detrimental effect on survival if they

have sufficient energy reserves This is consistent

with a laboratory experiment of juvenile pollock from

Puget Sound, northwestern United States (Sogard and

Olla 2000) However, the fish from Puget Sound

survived much longer time (at least 105 days) than the

fish in the present study The difference is probablyattributed to nutritional condition at the onset of theexperiment The lipid content of pre-experimental fish

in Sogard and Olla (2000) and the present study was27% and 10%, respectively, on a dry mass basis

Seasonal energy allocation and deficit

Body mass of juvenile walleye pollock tended toincrease from mid autumn through late winter exceptfor inshore fish during the winter of 2004 Somaticgrowth, rather than size-selective mortality, is likely

to be responsible for the increases in body massduring the periods, because the evidence of size-selective mortality was not found in the regressionanalysis Increases in both minimum and maximumlength also support the somatic growth throughout theperiods The increase of CV in the 2003–2004 season

is due to no winter growth in length for inshore fish(125–126 mm)

Juvenile pollock were likely to be exposed to preylimitation from autumn through winter In the 2003–

2004 season, the continuous decrease in whole-bodyenergy indicated prey limitation of inshore fish fromOctober through March and of offshore fish fromOctober through January The offshore fish subse-quently increased whole-body energy and body mass(Fig 8a, c), but their energy density remained low(3.4 kJ g−1) by March In the 2004–2005 season,juvenile pollock constantly increased their bodyenergy and body mass (Fig 8b, d), but their energydensity decreased from October through April(Fig 4a) Under a sufficient food supply, juvenilepollock increased both body mass and energy density(Kooka et al 2007b) Thus, prey were likely limitedbut not completely depleted during the first half ofwinter for offshore fish during the 2003–2004 seasonand from autumn through winter for both inshore andoffshore fish in the 2004–2005 season

We observed an unexpected energy allocationpattern from late autumn through winter In general,juvenile fish in seasonal environments allocate con-siderable energy to lipid reserves at the expense ofstructural growth as winter approaches and then uselipids or both lipids and proteins during winter (Hurstand Conover 2003; Biro et al 2004) In our study,juvenile pollock also gave priority to lipid storageover structural growth by early autumn, but theytended to allocate substantial energy to structural

growth at the expense of lipid accumulation from later

autumn through winter, with the exception of inshore

fish during the 2003–2004 season This result

suggests an energy allocation strategy for successful

overwintering in this juvenile gadoid fish under

quasi-prey-limited conditions The most plausible

explana-tion for the continuous structural growth is that

juvenile pollock give priority to reducing mortality

risk from size-selective predators by increasing body

size, as in Atlantic cod (Copeman et al.2008) In the

Doto area, smaller juvenile pollock are exposed to

strong predation pressure from predatory demersal

fishes from late summer through winter (Yamamura

2004) In addition, they are potential prey for larger

conspecifics (>300 mm) in the spring (Yamamura et

al 2001), indicating that winter growth is also

beneficial for avoiding cannibalism Moreover, larger

juvenile pollock have lower mass-specific metabolic

rates (Paul1986) and take more feeding opportunities

by capturing and handling larger prey (Brodeur1998;

Ciannelli et al 2004) These traits would increase

winter survivorship because the lower metabolic rates

and effective feeding would reduce both time spent

feeding and resultant encounter opportunities for

predators (Garvey et al.2004)

Juvenile walleye pollock in the Doto area

accumulated lipids from late summer to early

autumn, but exhausted the lipids by the onset of

winter, suggesting that they relied on feeding tomeet their metabolic demands during winter Thereliance on feeding is consistent with field obser-vations in Lynn Canal and Frederick Sound,southeastern Alaska (Heintz and Vollenweider

2010) The Alaskan fish gave priority to structuralgrowth over lipid storage in autumn but did not inwinter They concluded that increased average length

in winter is due to size dependent mortality because

of decreases in lipid and protein energy, RNA/DNAratios, and the coefficient of variation in length-frequency distributions Much higher winter temper-atures (about 5.5°C) in the southeastern Alaska(Heintz and Vollenweider 2010) may deprive them

of surplus energy for somatic growth throughtemperature-dependent metabolic rates

Although the inshore fish in the 2003–2004 seasonhad the lowest level of winter energy reserves, energylevels nonetheless appeared to be far from causingmass-starvation mortality First, the lowest L/P in thefield was higher than those in the laboratory, yieldinglow starvation mortality The inshore fish exhibitedhigher whole-body lipid and protein energy compared

to laboratory-reared survivors on day 35 (Table 2).Thus, the fish likely did not enter the terminaladaptation phase, which ultimately leads to mass-starvation mortality Second, few survivors had lipidcontents <1% in the field The lipid contents of

Table 2 Mean length, wet mass and whole body lipid and protein energy of juvenile walleye pollock in the field (inshore habitats from December 2003 through March 2004) and mortalities in the starvation experiment

Mean length

(mm)

Mean wet mass (g)

b

c

Wet mass and lipid and protein energy are adjusted to a 126-mm fish

mortalities were mostly <1% in the starvation

experiment, but in the low-lipid treatment, >70% of

starved fish were capable of surviving by day 35

(Kooka et al.2007a) despite the lower lipid level The

mobilizable energy (i.e the difference in whole-body

energy between field-collected fish and mortalities in

the laboratory) and daily energy expenditure levels of

the average-sized inshore fish would allow them to

survive more than 1 month until January and 2 months

until March, if they were starved (Table 2) Thus, if

starvation is a common mechanism of winter mortality,

severely lipid-depleted survivors should have been

observed in the field, as in the laboratory experiment

Unfortunately, evaluating the levels of energy

deficit during the second half of the winter of 2004

was impossible because no field samplings were

conducted in April 2004 However, starvation was

unlikely to have been the causal mechanism of

mortality during that period The average-sized

inshore fish in March 2004 would have survived for

a 105-day period without food before starving to

death, which is longer than the period until the onset

of spring (Table2)

One question arises as to why severely

lipid-depleted (<1%) survivors were virtually absent in

inshore habitats One possible explanation is that

juvenile pollock can maintain their lipid levels to

above starvation mortality through auxiliary feeding

In juvenile salmon, appetite is regulated to maintain

lipid reserves, and previously lipid-depleted fish are

able to quickly restore lipid reserves when food is

available (Metcalfe and Thorpe 1992; Álvarez and

Nicieza 2005) A more plausible explanation is that

the severely lipid-depleted fish have been culled by

predation Juvenile fishes in marine systems tend to

be targeted by more various predators than those in

freshwater systems (e.g Brodeur and Wilson 1996)

In the systems, inshore fish with lower lipid levels

also exhibited reduced protein levels in the winter

of 2004 and possibly suffered from predation

mortality because of impaired swimming

perfor-mance (Lankford et al 2001) However, the impact

of predation on overall winter mortality was likely to

be less extensive, because juvenile pollock in the

offshore area were more abundant and were much larger

and better conditioned than those in the inshore fish

In conclusion, mild nutritional stress and consistent

structural growth suggest that the majority of juvenile

pollock did not die because of the exhaustion of

energy reserves during winter but smaller fish mayhave suffered from predation In juvenile gadoidfishes, no evidence has been reported for winterstarvation mortality in the field Similarly, juvenileAtlantic cod in cold-water eelgrass habitats exhibitedconsistent growth with reduced lipid reserves until theonset of winter, suggesting strong predation pressure(Copeman et al 2008) In the laboratory, juvenilegadoid fishes are generally capable of feeding andgrowth in cold waters (Brown et al 1989; Kooka et

al 2007b) Thus, they are able to avoid bothstarvation and predation by accumulating proteinreserves under cold conditions with low prey avail-ability and in the presence of predators

the crew members of the fishing vessel Yutaka, for their assistance at sea; K Morita and K Hattori for their help with sample collection; H Kasai, N Hasegawa, and H Kunou for their help with stoichiometric analysis; and Y Kawaharada and

M Ishiguro for their laboratory assistance Earlier draft of this manuscript was improved through critical comments and useful suggestions of two anonymous reviewers This work was funded by research fellowships from the Japan Society for the Promotion of Science for Young Scientists (to K K.) and by the Dynamics of Commercial Fish Stocks (DoCoFis) program from the Fisheries Agency of Japan (to O Y.) The experiment complies with current laws of Japan.

References

Álvarez D, Nicieza AG (2005) Compensatory response

‘defends’ energy levels but not growth trajectories in

doi: 10.1098/rspb.2004.2291 Anthony JA, Roby DD, Turco KR (2000) Lipid content and energy density of forage fishes from the northern Gulf of

S0022-0981(00)00159-3 Biro PA, Morton AE, Post JR, Parkinson EA (2004) Over- winter lipid depletion and mortality of age-0 rainbow trout

Brodeur RD (1998) Prey selection by age-0 walleye pollock, Theragra chalcogramma, in nearshore waters of Gulf of Alaska Environ Biol Fish 51:175–186

Brodeur RD, Wilson MT (1996) A review of the tion, ecology and population dynamics of age-0 walleye pollock in the Gulf of Alaska Fish Oceanogr

Brown JA, Pepin P, Methven DA, Somerton DC (1989) The feeding, growth and behaviour of juvenile cod, Gadus

Castellini MA, Rea LD (1992) The biochemistry of natural

Caulton MS, Bursell E (1977) The relationship between

changes in condition and body composition in young

Tilapia rendalli Boulenger J Fish Biol 11:143–150

Ciannelli L, Brodeur RD, Buckley TW (1998) Development

and application of a bioenergetics model for juvenile

Ciannelli L, Brodeur RD, Napp JM (2004) Foraging impact on

zooplankton by age-0 walleye pollock (Theragra

chalcog-ramma) around a front in the southeast Bering Sea Mar

Copeman LA, Parrish CC, Gregory RS, Wells JS (2008)

Decreased lipid storage in juvenile Atlantic cod (Gadus

morhua) during settlement in cold-water eelgrass habitat.

FAO (2009) The state of world fisheries and aquaculture 2008.

Electronic Publishing Policy and Support Branch

i0250e/i0250e00.htm Accessed 17 September 2009

Garvey JE, Ostrand KG, Wahl DH (2004) Energetics,

preda-tion, and ration affect size-dependent growth and mortality

of fish during winter Ecology 85:2860–2871

Gibney MJ, Macdonald IA, Roche HM (2003) Nutrition and

metabolism Blackwell, Oxford

Harris RK, Nishiyama T, Paul AJ (1986) Carbon, nitrogen and

caloric content of eggs, larvae, and juveniles of the walleye

Hatch SA, Sanger GA (1992) Puffins as samplers of juvenile

pollock and other forage fish in the Gulf of Alaska Mar

Heintz RA, Vollenweider JJ (2010) Influence of size on the

sources of energy consumed by overwintering walleye

pollock (Theragra chalcogramma) J Exp Mar Biol Ecol

Hervant F, Mathieu J, Durand J (2001) Behavioural,

physio-logical and metabolic responses to long-term starvation

and refeeding in a blind cave-dwelling (Proteus anguinus)

and a surface-dwelling (Euproctus asper) salamander J

Exp Biol 204:269–281

Honda S (2004) Abundance estimation of the young cohorts of

the Japanese Pacific population of walleye pollock

(Theragra chalcogramma) by acoustic surveys Bull Fish

Honda S, Oshima T, Nishimura A, Hattori T (2004) Movement

of juvenile walleye pollock, Theragra chalcogramma,

from a spawning ground to a nursery ground along the

Pacific coast of Hokkaido, Japan Fish Oceanogr 13

Hurst TP (2007) Causes and consequences of winter mortality

10.1111/j.1095-8649.2007.01596.x

Hurst TP, Conover DO (1998) Winter mortality of

young-of-the-year Hudson River striped bass (Morone saxatilis):

size-dependent patterns and effects on recruitment Can J

Hurst TP, Conover DO (2003) Seasonal and interannual

variation in the allometry of energy allocation in juvenile

striped bass Ecology 84:3360–3369

Johnson TB, Evans DO (1990) Size-dependent winter mortality

of young-of-the-year white perch: climate warming and

invasion of the Laurentian Great Lakes Trans Am Fish

Kasai H, Saito H, Kashiwai M, Taneda T, Kusaka A, Kawasaki

Y, Kono T, Taguchi S, Tsuda A (2001) Seasonal and interannual variations in nutrients and plankton in the Oyashio region: a summary of a 10-years observation along the A-line Bull Hokkaido Natl Fish Res Inst

jembe.2007.03.003 Kooka K, Yamamura O, Ohkubo N, Honda S (2009) Winter lipid depletion of juvenile walleye pollock Theragra chalcogramma in the Doto area, northern

10.1111/j.1095-8649.2009.02279.x Lankford TE Jr, Billerbeck JM, Conover DO (2001) Evolution

of intrinsic growth and energy acquisition rates II offs with vulnerability to predation in Menidia menidia.

Méndez G, Wieser W (1993) Metabolic responses to food deprivation and refeeding in juveniles of Rutilus rutilus

Metcalfe NB, Thorpe JE (1992) Anorexia and defended energy levels in over-wintering juvenile salmon J Anim Ecol

Miyake H, Yoshida H, Ueda Y (1996) Distribution and abundance of age-0 juvenile walleye pollock, Theragra chalcogramma, along the Pacific coast of southeastern

Molony BW (1993) Effects of feeding history on mobilisation and deposition of body constituents and on growth in juvenile Ambassis vachelli (Pisces: Chandidae) Mar Biol 116:389–397

Niimi AJ (1972) Changes in the proximate body composition of largemouth bass (Micropterus salmoides) with starvation Can J Zool 50:815–819

Nishimura A, Hamatsu T, Shida O, Mihara I, Mutoh T (2007) Interannual variability in hatching period and early growth

of juvenile walleye pollock, Theragra chalcogramma, in the Pacific coastal area of Hoddaido Fish Oceanogr

Paul AJ (1986) Respiration of juvenile pollock, Theragra chalcogramma (Pallas), relative to body size and temper-

Saito H, Kasai H, Kashiwai M, Kawasaki Y, Kono T, Taguchi

S, Tsuda A (1998) General description of seasonal variations in nutrients, chlorophyll a, and netplankton biomass along the A-line transect, western subarctic Pacific, from 1990 to 1994 Bull Hokkaido Natl Fish Res Inst 62:1–62

SAS (2005) JMP user guide, release 6 SAS Institute Inc, Cary Scheiner SM, Gurevitch J (2001) Design and analysis of ecological experiments, 2nd edn Oxford University Press, New York

Shida O (2002) Age-dependent bathymetric pattern of walleye pollock, Theragra chalcogramma, off the Pacific coast of

eastern Hokkaido Sci Rep Hokkaido Fish Exp Stn 63:

9–19

Smith RL, Paul AJ, Paul JM (1986) Effect of food intake and

temperature on growth and conversion efficiency of

juvenile walleye pollock (Theragra chalcogramma

(Pallas)): a laboratory study J Cons Int Explor Mer

Sogard SM, Olla BL (2000) Endurance of simulated winter

conditions by age-0 walleye pollock: effects of body

size, water temperature and energy stores J Fish Biol

Springer AM (1992) A review: walleye pollock in the North

Van Pelt TI, Piatt JF, Lance BK, Roby DD (1997) Proximate composition and energy density of some North Pacific forage fishes Comp Biochem Physiol A 118:1393–1398

Weatherley AH, Gill HS (1987) The biology of fish growth Academic, London

Yamamura O (2004) Trophodynamic modeling of walleye pollock (Theragra chalcogramma) in the Doto area, northern Japan: model description and baseline simula-

Yamamura O, Yabuki K, Shida O, Watanabe K, Honda S (2001) Spring cannibalism on 1 year walleye pollock in the Doto area, northern Japan: is it density dependent? J