behavioural and morphological differences between feral ......behavioural and morphological...

Journal of Fish Biology (2009) 75, 1206–1220

doi:10.1111/j.1095-8649.2009.02345.x, available online at www.interscience.wiley.com

Behavioural and morphological differences between feraland domesticated strains of common carp Cyprinus carpio

S. S. Matsuzaki*†, K. Mabuchi‡, N. Takamura*§, M. Nishida‡and I. Washitani*

*Department of Ecosystem Studies, Graduate School of Agricultural and Life Sciences,The University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan, ‡Ocean Research

Institute, The University of Tokyo, 1-15-1 Minamidai, Nakano-ku, Tokyo 164-8639, Japanand §Research Center for Environmental Risk, National Institute for Environmental

Studies, 16-2 Onogawa, Tsukuba-shi, Ibaraki 305-8506, Japan

(Received 12 January 2009, Accepted 8 May 2009)

Morphological and behavioural traits of a feral strain of the common carp Cyprinus carpio fromLake Biwa in Japan were compared with those of two domesticated strains reared in Japan (one com-mercial strain and one ornamental koi). To compare genetically inherited traits, all fish were rearedfrom eggs under similar environmental conditions. Using these fish, the following five traits werecompared among the three strains: body shape, consumption rate of two types of free-swimmingshrimp, medaka Oryzias latipes and bottom-dwelling chironomid larvae prey items, preference fora bottom habitat, feeding skills in detecting prey and escape response to predator attack. The feralstrain of fish had more streamlined bodies, higher consumption rates for free-swimming prey, agreater preference for a bottom habitat, possessed greater skill in detecting prey and were morecautious of predator attacks, compared with the fish of the two domesticated strains. These charac-teristics shown by the feral fish are probably adaptive to the natural environment. A genetic analysisbased on five nuclear single nucleotide polymorphism markers, however, suggested that the feralstrain was relatively recently derived from domesticated stocks. Considering this, the present resultsappear to indicate the possibility that domesticated C. carpio could re-adapt to the wild envi-ronment during a short evolutionary period, although further research using more feral strains isrequired. © 2009 The Authors

Journal compilation © 2009 The Fisheries Society of the British Isles

Key words: artificial selection; domestication; genetic markers; invasion; morphometrics.

INTRODUCTION

In many natural ecosystems, native biodiversity and ecosystem functions are threat-ened by intentional and unintentional releases of domesticated animals (IUCN, 1997;Barilani et al ., 2005; Bowman et al ., 2007; Randi, 2008). This is particularly the casewith fishes. Domesticated fishes, which have been subjected to artificial selection fordesirable production traits (e.g. rapid growth), are released into wild environmentsor escape from aquaculture (Rhymer & Simberloff, 1996; Hansen, 2002; Garant

†Author to whom correspondence should be addressed. Tel./fax: +81 29 850 2425; email: [email protected]

1206© 2009 The Authors

Journal compilation © 2009 The Fisheries Society of the British Isles

B E H AV I O U R A N D M O R P H O L O G Y O F F E R A L C Y P R I N U S C A R P I O 1207

et al ., 2003; Huntingford, 2004). Naturalization or feralization of domesticated ani-mals can have negative effects on the fitness of wild populations through ecologicaland genetic mechanisms (Manchester & Bullock, 2000; Volpe et al ., 2000).

Accumulating evidence suggests that domesticated fishes exhibit genetically inher-ited differences in morphological and behavioural traits from wild fishes, and havelower fitness in natural environments (Einum & Fleming, 1997; Fleming & Einum,1997; Garant et al ., 2003; McGinnity et al ., 2003; Metcalfe et al ., 2003). Impor-tantly, Araki et al . (2007) have demonstrated that even a few generations of domes-tication can cause a decline in fitness under natural conditions. These differences intraits and fitness are attributable to the adaptation to the culture environment (relax-ation of natural selection) or the genetic effects arising from artificially selectiveregimes in captivity (Brown & Day, 2002; Tymchuk et al ., 2006; Fraser, 2008). Theevidence, however, is mostly based on studies of salmonids and gadoids.

The common carp Cyprinus carpio L. is domestically cultured in vast numbersand has been introduced intentionally to wild environments throughout the world. Itcan cause serious environmental and economic problems in freshwater ecosystems(Millennium Ecosystem Assessment, 2005). This species is included in the list of100 world’s most invasive alien species (IUCN; www.iucn.org). A recent geneticsurvey has revealed that in Japan, although genetically pure native C. carpio strainsare endangered, non-native domesticated C. carpio, which have been introduced intowild environments, have established large feral populations (Mabuchi et al ., 2008).Based on previous studies on salmonids, domesticated fishes are expected to havelower fitness in natural environments than wild fishes, which have been selected fornatural conditions. The successful naturalization of domesticated C. carpio in Japan,therefore, poses an intriguing question. One possible mechanism for this success is‘rapid evolution’ suggested in the case of Atlantic salmon Salmo salar L., which isbelieved to have adapted rapidly to the wild environment over several generations,resulting in the establishment of vast feral populations (Volpe et al ., 2000; Jacobsen& Hansen, 2001). If this is the case with feral C. carpio in Japan, they should havebehavioural and morphological traits that are adaptive to the wild environment. Fewempirical studies have considered this.

In this study, the hypothesis that a feral strain of C. carpio from Lake Biwa inJapan has different morphological and behavioural traits, compared with those of tworepresentative domesticated (farmed) strains, that have been subjected to selectivebreeding over multiple generations in Japanese hatcheries was tested. In order tocompare genetically inherited traits, all three strains used in this study were rearedfrom eggs under identical hatchery environments. Their body shape, consumptionrates for free-swimming and bottom-dwelling prey, preference for a bottom habitat,feeding skill in terms of detecting prey and escape reaction to predator attack weretested. To confirm whether the feral C. carpio used are derived from domesticatedstocks and not from the Japanese native C. carpio, individuals were genotyped acrossfive nuclear markers, which discriminate between Japanese native and non-nativeorigins. Using the same five markers, the extent of hybridization between native andnon-native strains in the remaining two domesticated strains was assessed, whichenabled the genetic similarity among the strains used to be determined, based onselectively neutral markers.

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1206–1220

1208 S . S . M AT S U Z A K I E T A L .

MATERIALS AND METHODS

S T U DY S T R A I N S

According to Maruyama et al . (1987), Japanese institutes have imported several foreigndomesticated strains from some European and Asian countries since 1905. Since then, largenumbers of (non-native origin) domesticated C. carpio have been annually introduced intonatural water bodies throughout the country (MAFFJ, 2000).

The feral C. carpio (F strain) used in this study were the offspring of 11 wild matureparents that had been caught during the breeding season in Lake Biwa (35◦ 15′ N; 136◦05′ E), the largest lake in Japan (area 670 km2; mean depth 41 m; maximum depth 104 m).This strain was reared in the Shiga Prefectural Fisheries Experiment Station. The two repre-sentative domesticated (farmed) stocks used in this study were a commercial strain reared inthe National Institute for Environmental Studies (NIES) (D strain) and an ornamental strainknown as (Nishikigoi) reared in the Niigata Prefectural Inland Water Fisheries ExperimentalStation (O strain). The breeding histories and origins of the two domesticated strains werelargely uncertain, but both had been subjected to selective breeding and experienced captiveenvironments in aquaculture over multiple generations (>20 years).

All experimental fish were raised from the fertilized egg to the juvenile stage under similarconditions in hatcheries: in bare 500 l polyethylene tanks with well water (constant temper-ature) and supplemental aeration. The fish were fed commercial dry crumbles at a rate of2% body mass per day, and they were shifted among tanks approximately once a month. Byrearing the juveniles under similar environmental conditions, an attempt was made to reducepossible rearing tank effects (Fleming & Einum, 1997; Johnsson et al ., 2001; Metcalfe et al .,2003). All the juveniles used for the experiments were transferred to NIES and reared inidentical tanks until the start of experiments. All the experiments were performed in NIESwith size-matched fish (age 1+ years).

G E N E T I C A NA LY S I S

Because Japanese domesticated (and feral) strains of C. carpio are thought to be a hybridof Japanese native (originally wild) and non-native (domesticated) strains (K. Mabuchi,H. Takeshima, H. Senou, K. Nakai & M. Nishida, unpubl. data), the extent of hybridiza-tion of each strain was assessed using nuclear markers. A microsatellite-enriched genomiclibrary from the Lake Biwa wild strain of C. carpio (Mabuchi et al ., 2006) was used for theisolation of HapSTR markers and the development of primers (K. Mabuchi, H. Takeshima,H. Senou, K. Nakai & M. Nishida, unpubl. data). A HapSTR, a conjunction of haplotype andshort tandem repeat markers, consists of a microsatellite region and its flanking sequences(Hey et al ., 2004). In this study, single nucleotide polymorphisms (SNP) in the flankingsequences were used as markers, because significant levels of homoplasy have been reportedfor microsatellite allele length in various animals (van Oppen et al ., 2000). Five SNP withinfive different HapSTR loci were used as diagnostic markers for Japanese native v. non-nativepopulations. Primers for four of the five HapSTR loci (c20, 25, 37, 52) have been described(K. Mabuchi, H. Takeshima, H. Senou, K. Nakai & M. Nishida, unpubl. data). The fifth locus(Koi 57-58 ) used primers described by David et al . (2001).

Five SNP were genotyped by direct sequencing of the flanking regions for 31 F strain,29 D strain and 29 O strain fish. Total DNA was extracted from a fin clipping preservedin ethanol, using an AquaPure genomic DNA purification system according to the manufac-turer’s protocol (Bio-Rad; www.bio-rad.co.jp). Multiplex polymerase chain reaction (PCR)was performed using a Model 9700 thermal cycler (Perkin-Elmer; www.perkinelmer.com),and reactions were carried out using 40 cycles of a 10 μl reaction volume containing 4·5 μlof distilled water, 1·0 μl of 10 × PCR buffer (Takara; http://www.takara-bio.com/index.htm),1·0 μl of dNTP (4 mM), 0·3 μl of 0·5 unit Ex Taq (Takara), 1·0 μl of template and 0·25 μlof each primer (10 μM). The following 10 primers were used: c20-F, -R, c25-F, -R, c37-F,-R, c52-F, -R, Koi 57-58-F and -R. The thermal cycle profile was as follows: denatura-tion at 94◦ C for 15 s, annealing at 55◦ C for 15 s and extension at 72◦ C for 30 s. ThePCR products were electrophoresed on a 1·0% l 03 agarose gel column and stained with



ethidium bromide for band characterization via ultraviolet trans-illumination. Double-strandedmultiplex PCR products purified using a PreSequencing Kit (USB; www.usbweb.com) weresubsequently used for direct cycle sequencing using dye-labelled terminators (Applied Biosys-tems; www.appliedbiosystems.com). The primers used were five of the 10 primers used formultiplex PCR: c20-F, c25-R, c37-R, c52-R and Koi 57-58-F. All sequencing reactions wereperformed according to the manufacturer’s instructions. Labelled fragments were analysedusing a Model 3130XL DNA Analyzer (Applied Biosystems). Sequence editing and SNPgenotyping were performed using EditView ver. 1.01 (Applied Biosystems).

Arlequin ver. 3.11 (Excoffier et al ., 2005) was used to calculate pair-wise multilocusFST among the three strains (F, D and O) and test the significance of genetic differentiationamong the three strains by means of a Fisher exact test (Raymond & Rousset, 1995). Arlequinver. 3.11 (http://cmpg.unibe.ch/software/arlequin3/) was also used to test for deviations fromlinkage equilibrium as well as the Hardy–Weinberg equilibrium (Guo & Thompson, 1992).Hybrid indices for each individual were calculated by entering a score of 0 for each Japanesenon-native SNP and 1 for each Japanese native SNP. Scores for all five loci thus rangedfrom 0 (all Japanese non-native SNP) to 10 (all Japanese native SNP). SNP frequencies areavailable.

M O R P H O L O G I C A L M E A S U R E M E N T S A N D A NA LY S I S

Morphology was analysed using landmark-based geometric morphometrics by thin-platespline analysis (Svanback & Eklov, 2004; Langerhans et al ., 2007), which is a powerful,flexible and easily interpreted multivariate technique for analysing morphological variationamong objects.

Eighty-six size-matched individuals (F strain: n = 26, D strain: n = 30, O strain: n = 30,mean ± s.d. standard length, LS, 11·1 ± 1·1 cm) were analysed. The fish were anaesthetizedusing 2-phenoxyl-ethanol (0·3 ml l−1) and placed on a piece of polystyrene board. The finswere fixed using pins. A photograph was taken using an Olympus μ 710 digital camera(www.olympus.com). The camera was set at c. 30 cm from the object to minimize poten-tial wide-angle artefacts. Digital photographs were imported into TpsDig (Rohlf, 2006), andthereafter the x and y co-ordinates of 17 homogenous points on the left side of each fish weredigitized and captured (Fig. 1). The data were transferred to tpsRelw (Rohlf, 2007a), wherethe variation in shape was calculated by comparing the position of the landmarks between indi-viduals. TpsRelw then described the resulting shape variation by small-scale shape changes(partial warps or non-uniform scores) and shape changes along the whole-body axis (uniformscores); in the present case, 28 partial warps and two uniform components. The partial warpsand uniform scores of all individuals were analysed by multivariate discriminant functionanalysis (DFA). This technique combines all partial warps and uniform scores into a fewdiscriminant functions for each fish, which maximally discriminate among the three strains.Shape changes associated with the morphological variation obtained from the DFA werevisualized as deformations using tpsRegr (Rohlf, 2007b).

B E H AV I O U R E X P E R I M E N T S

Two experiments were performed to assess differences in behavioural traits between feraland domesticated strains.

Experiment 1: consumption rateAn outdoor predation experiment was conducted to assess the consumption rate for the

following two different types of prey among the three strains: live chironomid Propsilocerusakamusi larvae, as benthic prey, and shrimps Paratya compressa improvisa and Japanesemedaka Oryzias latipes (Temminck & Schlegel) as free-swimming prey (Weeks et al ., 1992).The experiment was performed in a randomized complete block design with eight replications.Aquaria (60 cm × 30 cm × 36 cm, 60 l) were placed in water baths (each 1 m × 2 m) tomaintain a constant temperature of 25◦ C. The aquaria were covered with black plastic sheetsand were filled with well water (c. 50 l) oxygenated with air bubbles. The bottom of each



Fig. 1. Photograph of the F strain of Cyprinus carpio used in the experiment showing the locations of the 17landmarks used to describe shape.

aquarium was covered with sand to a depth of 2 cm, and a polyvinyl chloride (PVC) grey pipe(5 cm diameter, 15 cm long) was placed at the centre of each aquarium to provide shelter.

Forty P. akamusi larvae (mean ± s.d. body length 17·0 ± 1·5 mm), 10 P. c. improvisa(mean ± s.d. body length 29·8 ± 1·2 mm) and five O. latipes (mean ± s.d. LS 28·5 ±1·6 mm) were added to each aquarium 1 day before the start of the experiment. The size-matched fish (mean ± s.d. LS 10·7 ± 0·8 cm) were not fed for 24 h before the experiment.Each fish was used only once during the experimental period. At the start of the exper-iment, P. akamusi larvae spread over the bottom and did not burrow into the sand. Oneindividual C. carpio was released into each aquarium after acclimation in a small cage(hand-net) held at the water surface for 2 h. The fish were removed 24 h after releaseand then each aquarium was searched thoroughly to count the number of remaining preyitems.

Experiment 2: habitat use, feeding skill and escape reactionTo evaluate preference for bottom habitat, feeding skill in terms of detecting prey and

escape reaction when subjected to predator attack, three sequential behavioural observationsusing digital video cameras were recorded.

An individual C. carpio was placed in each 60 l aquarium (visually isolated by blackplastic sheets covering three sides) under natural light conditions (36◦ N). Individual bodysize did not differ significantly among the strains (mean ± s.d. LS 10·3 ± 0·8 cm). Thefish were allowed to acclimate to the tank for one night before the start of observations. Forbehavioural observations, 12 aquaria were filmed simultaneously using three digital videocameras (DCR-SR300, SONY; www.sony.com) placed in front of the aquarium (>3 m).Three horizontal grid lines (every 10 cm) were drawn on the unscreened front side of theaquarium. Ground water was allowed to overflow into each aquarium through a soft vinyltube (6 mm diameter) to maintain sufficient oxygen concentrations and a water temperature of20◦ C. Each aquarium had one PVC grey pipe (5 cm diameter, 15 cm length), which servedas a refuge.

The first observation, termed habitat use was performed for 10 min. The time spent by thefish in the layer within 10 cm of the tank bottom (Hart, 2003) and the time spent hiding inthe shelter were recorded.

The second observation to assess feeding skills was termed feeding. One hour after the firstobservation, 15 P. akamusi larvae were carefully added into the back corner of the aquarium.The time taken for fish to first bite a larva was recorded over a period of 15 min (Sundstrom& Johnsson, 2001; Kelley et al ., 2005).

The third observation was escape reaction, in which each fish was subjected to a simulatedpredator attack using a wooden model of an egret Egretta garzetta (Iguchi et al ., 2001;Johnsson et al ., 2001). These are one of the most common piscivorous birds in Japan. Three



hours after the second observation, the head was suddenly presented, simulating a strike. Thetime elapsed from the start of the escape reaction (i.e. burst swimming) until the fish ceasedactive swimming was recorded to estimate the duration of the flight reaction to predatorattack (flight duration; Johnsson et al ., 2001). Whether fish escaped into the shelter after thestimulus was also recorded using a binominal code (0 or 1).

A series of these measurements were repeated daily for 5 consecutive days using each fishonly once; thus, the behaviours of 60 individuals (20 of each strain) were observed.

Statistical analysisTo examine differences in behavioural traits among the three strains, a generalized lin-

ear model (GLM) with a binomial or Poisson error distribution with a logit or log link,respectively, was applied. The information-theoretic model selection approach was used tocompare model fits of five possible groupings of the strains into sub-groups: (Null), [F v.(D, O)], [D v. (F, O)], [O v. (F, D)] and (F v. D v. O) (Table I). Different groupingscan be compared in terms of their Akaike information criterion (AIC). The best perform-ing model has the smallest AIC, and models within two AIC units of the best model areconsidered to be equally parsimonious (Burnham & Anderson, 2002). For example, on agrouping basis, [F v. (D, O)] indicates that the F strain almost certainly differs from the Dand O strains. All analyses were performed using the statistical software R version 2.7.1(www.r-project.org).

RESULTS

G E N E T I C C H A R AC T E R I Z AT I O N

The hybrid indices of the 31 feral individuals ranged from 0 to 6, indicating thatthe feral C. carpio used in this study were of non-native (domesticated strain) ori-gin. The values of pair-wise FST among the three strains were 0·003 for F v. D,0·195 for F v. O and 0·267 for D v. O. Exact tests of genetic differentiation showedsignificant differences between O and the remaining two strains (both P < 0·05),but showed no significant difference between F and D strains (P > 0·05). The dis-tribution of hybrid indices (Fig. 2), however, indicated that all three strains werecomposed of individuals with low hybrid indices (<5). Thus, the F strain is assumedto be relatively recently derived from domesticated stocks.

All three strains showed no significant departure from the Hardy–Weinberg equi-librium at the five SNP loci (for some loci in D and O strains, no test was performedbecause they were monomorphic in the strains). Thirty tests of linkage disequilib-rium for the loci detected two significant deviations in the F strain [c25 and c37(P < 0·05), c37 and c52 (P < 0·05)].

M O R P H O L O G I C A L T R A I T S

The DFA revealed highly significant differences in body shape among the strains,yielded two morphological functions and correctly classified 98·8% of the individualsto the three respective strains (Wilks’ λ = 0·016, P < 0·001). The results are givenfrom the first DFA, explaining the majority of the variation (88·7%) (Fig. 3). Thefish with scores distributed at the negative end of the scale (i.e. the F strain) had amore streamlined body, smaller head, narrower caudal peduncles, longer dorsal fins,a relatively more ventral placement of the eye, more downward bend and anteriorplacement of the pectoral fin with a shallower angle of attachment between the



Tab

leI.

Res

ults

ofm

odel

sele

ctio

nto

com

pare

mod

elfit

sbe

twee

nfiv

epo

ssib

legr

oupi

ngs

ofth

eC

ypri

nus

carp

iost

rain

s(F

,D

and

O)

into

sub-

grou

psin

the

beha

viou

rex

peri

men

ts.

Bol

dva

lues

are

the

best

grou

ping

with

smal

lest

Aka

ike

info

rmat

ion

crite

rion

valu

e.K

indi

cate

sth

enu

mbe

rof

para

met

er

Can

dida

tem

odel

Var

iabl

esN

ull

Fv

.(D

,O

)D

v.

(F,

O)

Ov

.(F

,D

)F

v.

Dv

.O

(K=

2)(K

=2)

(K=

2)(K

=2)

(K=

3)

Exp

erim

ent

1P

rops

iloc

erus

akam

usi

larv

ae13

9·214

0·314

0·114

1·114

1·9P

arat

yaco

mpr

essa

impr

ovis

a77

·858

·968

·576

·360

·0O

ryzi

asla

tipe

s73

·544

·863

·565

·446

·8E

xper

imen

t2

Hab

itat

use

obse

rvat

ion

Dep

thse

lect

ion

3624

3592

3626

3588

3581

Shel

ter

use

2301

322

890

2297

622

990

2289

2Fe

edin

gob

serv

atio

nFi

rst

bite

5113

546

586

5112

646

412

4487

9E

scap

ere

actio

nob

serv

atio

nFl

ight

dura

tion

41·0

38·1

42·0

41·3

40·4

Shel

ter

use

248·4

243·1

246·8

249·8

244·9



Hybrid index

0 2 4 6 8 100

2

4

6

8

10

12

14Fr

eque

ncy

Fig. 2. Frequency of individual hybrid indices showing genotypic distributions of the three strains of Cypriniuscarpio [F ( ) n = 31; D ( ) n = 29; O ( ) n = 29] used in this study. Values close to 0 indicate afish with all Japanese non-native single nucleotide polymorphism (SNP), while individuals with valuesclose to 10 indicate a fish with all Japanese native SNP.

pectoral fin (oblique) and the long axis of the body (horizontal). On the other hand,the fish with more positive scores (i.e. the O strain) had a deeper body, larger head,robust caudal peduncles, shorter dorsal fin, a more dorsal placement of the eye,more upward bend and posterior placement of the pectoral fin with a deeper angleof attachment. The D strain showed intermediate morphology, but was closer to theF strain.

B E H AV I O U R A L T R A I T S

Experiment 1: consumption rateThe number of P. akamusi larvae (benthic prey) consumed was best explained by

the null model, indicating that there was no difference between the strains [Fig. 4(a)and Table I]. The best model for explaining the consumption rates for P. c. improvisaand O. latipes (free-swimming prey) was [F v. (D, O)], indicating that the F strainconsumed more of both prey than the D and O strains [Fig. 4(b), (c) and Table I].For the consumption rate on P. c. improvisa, however, (F v. D v. O) within twoAIC units of the best model also had substantial support (Table I). This indicatesthat the D strain had a moderately higher consumption rate on P. c. improvisa thanthe O strain.

Experiment 2: habitat use, feeding skill and escape reactionIn the habitat use observations, the F strain was in the bottom 10 cm of the water

column more often than the D and O strains [Fig. 5(a) and Table I]. In addition,the O strain spent more time in the shallow habitat than the D strain. The F strainfish used the shelter more frequently and were more cautious than the other strains[Fig. 5(b) and Table I].

In the feeding observations, (F v. D v. O) was favoured in accounting for thevariation in timing of first bite, indicating that first bite differed among the threestrains [Fig. 5(c) and Table I]. The F strain began foraging on chironomid larvaemore rapidly than other strains, and the D strain bit sooner than the O strain.



Discriminant function 1

−6 −4 −2 0 2 4 6 8

Freq

uenc

y

0

2

4

6

8

10

Fig. 3. Distributions of the first discriminant function (from the discriminant function analysis) for each Cypri-nus carpio strain [F ( ), D ( ) and O ( )]. The grid plots below the graph show a visualization ofthe extreme morphologies obtained from tpsRegr. The 17 landmarks ( ; see Fig. 1) and the solid linesconnecting outer landmarks are drawn to aid interpretation. Deformations are 3×-scaled for visualizationpurposes.

In the escape reaction observation, [F v. (D, O)] best explained the observedvariations in flight duration and shelter use [Fig. 5(d), (e) and Table I]. Following thesimulated egret strike, the F strain had longer flight duration and a higher probabilityof escaping into the shelter than the D and O strains, indicating that the F strain wasmore cautious than the other two strains. For escape into the shelter, however, (F v. Dv. O) within two AIC units of the best model also had substantial support (Table I).Thus, the shelter use response to predator attack of the D strain was moderatelyhigher than the O strain.

DISCUSSION

There were striking differences in morphological and behavioural traits amongthe Lake Biwa feral (F) and domesticated strains (D and O). These phenotypicdifferences are considered to have a genetic basis because the fish were reared fromeggs under similar environmental conditions (Johnsson et al ., 2001; Metcalfe et al .,2003). Interestingly, the characteristics shown by the F strain (which is assumed tobe relatively recently derived from domesticated stocks) are likely to be adaptive tothe natural environment.

The F strain had a more streamlined, slender body with narrower caudal peduncles,a smaller head and shallower attachment angle of the pectoral fin, compared with thetwo domesticated strains (Fig. 3). These traits shown by the F strain are likely to bethe result of natural selection in the natural environments. In fact, wild C. carpio have



0

1

2

3

0

1

2

3

4

F D O

F D O

F D O

0

10

20

30

40

50

a

b

b

a

b

b

(a)

(b)

(c)

a a a

Strain

Num

bers

con

sum

ed (

indi

vidu

als

per

day)

Fig. 4. Behavioural traits related to predation for the F, D and O strains. Mean ± s.e. consumption rates (±s.e.) for (a) Propsilocerus akamusi, (b) Paratya compressa improvisa and (c) Oryzias latipes. The meanswith different italic lowercase letters are considered to differ from each other, based on the best groupingby model selection (Table I).

a more streamlined body than domesticated fish (Kafuku, 1966; Balon, 2004). In thewild, C. carpio with better swimming ability must have been selected. Interestingly,prolonged swimming speed performance is optimized with a relatively shallow caudalpeduncle and streamlined body shape in poeciliids (Langerhans et al ., 2007). Collaret al . (2008) have also reported that labrids with a shallower attachment angle ofthe pectoral fin achieve higher swimming speed.

The F strain showed higher consumption rates for free-swimming prey than theD and O strains [Fig. 4(b), (c)]. Although it is possible that the difference couldhave arisen due to a lack of experience of live prey (Olla et al ., 1998; Sundstrom& Johnsson, 2001), this seems unlikely. All fish, which were reared under similarenvironmental conditions, probably had similar feeding experience. Alternatively,



Mea

n tim

es w

hen

with

in 1

0cm

of ta

nk b

otto

m (

s pe

r 10

min

)

300

350

400

450

500

550

Flig

ht d

urat

ion

(s)

0

1

2

3

4

5Pr

obab

ility

to e

scap

e in

to th

e sh

elte

r

0·0

0·1

0·2

0·3

0·4

Tim

e to

fir

st b

ite (

s)

0

200

400

600

c

a

bb

ab

a

bb

(a) (b)

(d) (e)

StrainStrain

F D O F D O F D O

F D O F D O

Mea

n tim

es w

hen

with

in

shel

ter

(s p

er 1

0 m

in)

0

100

200

300

a

b

c(c)

a

b b

Strain

Fig. 5. Behavioural traits related to habitat use, feeding and escape reaction for the F, D and O strains. Times(a) spent when within 10 cm of the tank bottom and (b) when hiding in the shelter in the habitat usetest. (c) Time to first bite in the feeding test. (d) Flight duration and (e) probability of escape into theshelter in response to a simulated predation attack in the escape reaction test. Values are means ± s.e.The means with different italic lowercase letters are considered to differ from each other, based on thebest grouping by model selection (Table I).

difference in selective regimes between the captive and wild environment may beresponsible for this (Fleming & Einum, 1997; Fraser, 2008). While domesticated fishreceive a plentiful supply of nutritious pellets and are subjected to directed artificialselection for rapid growth, wild (including feral) fish can be exposed to naturalselection for efficient foraging so that the most skilled individuals can survive. Thus,the F strain’s higher consumption rates for free-swimming prey could be explainedby adaptation to the wild environment.

Although the F strain fish were more cautious [Fig. 5(b), (d), (e)], they werequicker to attack prey than the domesticated strains [Fig. 5(c)]. This indicates thatthe F strain has better feeding skills in detecting prey. Considering that wild-caughtbrown trout Salmo trutta L. attack prey more quickly than hatchery-reared fish (Sund-strom & Johnsson, 2001), the F strain’s faster attack may be explained by adaptationto the wild environment. Differences in prey-detecting skill may, however, haveinfluenced the consumption rates for free-swimming prey items (Fig. 4). The feed-ing skill of the D strain was higher than that of the O strain. Moreover, depthselection, consumption rate on P. c. improvisa and escape into shelter response topredator attack seemed to differ between the D and O strains [Table I and Figs 4(b)and 5(a), (e)]. This may be due to differences in breeding history (e.g. generationnumber) and the domestication process among culture facilities (Lucas et al ., 2004).



The F strain preferred a bottom habitat, while the domesticated strains (especiallythe O strain) were more often found in upper water layers [Fig. 5(a)]. Biro et al .(2004) reported that domesticated rainbow trout Oncorhynchus mykiss (Walbaum)were more willing to take risks than the wild strain by foraging more in food-rich, but risky pelagic habitats (risk of piscivorous fishes). For C. carpio shallowerlayers in natural waters are considered to be potentially risky due to an easierapproach for predatory birds. Cyprinus carpio with superior predator recognitionabilities (antipredator behaviour) must have been selected in the natural environ-ment. The more cautious response to predator attack of the F strain [Fig. 5(c), (d)]could be similarly explained. These characters are likely to be adaptive to the naturalenvironment.

Considering the results of the genetic analysis (Fig. 2), the present study indicatesthe possibility that domesticated C. carpio could re-adapt to the wild environmentduring a short evolutionary period. This should be interpreted, however, with cautionbecause of a lack of replication (only a single feral population was included in thepresent study). Thus, the study should be viewed as a first step in understandingthe re-adaptation of domesticated fish to wild environments. Further experiments arerequired to examine multiple feral and domesticated C. carpio populations, and testthe generality of the findings for other fish species.

Finally, the naturalization or feralization of domesticated C. carpio can have detri-mental influences on native freshwater ecosystems. This study demonstrated that feralC. carpio have some traits that would allow adaptation to the wild, but they maybehave differently from wild (native) C. carpio in some crucial aspects affectingnative ecosystems. If so, the presence of a feral population would threaten fresh-water ecosystem. For example, other experimental studies have demonstrated thatferal C. carpio can cause ecosystem changes by worsening water quality and reduc-ing submerged macrophytes (Matsuzaki et al ., 2007, 2009). From a conservationviewpoint, therefore, interactions, including hybridization and competition, betweenferal and wild strains (the Japanese native C. carpio) might be crucial and shouldbe investigated urgently.

We are sincerely grateful to K. Kawabe and A. Saji for tireless help in the field, and alsoT. Kubo, M. Akasaka and T. Kadoya for help with statistical analyses. We thank R. Svanback,B. Langerhans, K. Iguchi and J. Kelley for their valuable comments and suggestions on mor-phological or behavioural analysis. We also acknowledge C. Brown and two anonymousreviewers for critical review and improvement of earlier drafts of this study. Finally, wethank the following institutions for providing the test fish for the experiment: National Insti-tute for Environmental Studies, Shiga Prefectural Fisheries Experiment Station and NiigataPrefectural Inland Water Fisheries Experimental Station. In this study, all procedures complywith Japanese laws governing ethical conduct and the care and use of animals in research. Thisstudy was funded by grant-in-aid from the Ministry of Education, Culture, Sports, Scienceand Technology of Japan to S.M. (No.1811493) and K.M. (No.18780142).

References

Araki, H., Cooper, B. & Blouin, M. S. (2007). Genetic effects of captive breeding cause arapid, cumulative fitness decline in the wild. Science 318, 100–103.

Balon, E. K. (2004). About the oldest domesticates among fishes. Journal of Fish Biology65(Suppl. A), 1–27.

Barilani, M., Deregnaucourt, S., Gallego, S., Galli, L., Mucci, N., Piombo, R., Puigcerver,M., Rimondi, S., Rodriguez-Teijeiro, J. D., Spano, S. & Randi, E. (2005). Detecting



hybridization in wild (Coturnix c. coturnix ) and domesticated (Coturnix c. japonica)quail populations. Biological Conservation 126, 445–455.

Biro, P. A., Abrahams, M. V., Post, J. R. & Parkinson, E. A. (2004). Predators select againsthigh growth rates and risk-taking behaviour in domestic trout populations. Proceedingsof the Royal Society B 271, 2233–2237.

Bowman, J., Kidd, A. G., Gorman, R. M. & Schulte-Hostedde, A. I. (2007). Assessing thepotential for impacts by feral mink on wild mink in Canada. Biological Conservation139, 12–18.

Brown, C. & Day, R. L. (2002). The future of stock enhancements: lessons for hatcherypractice from conservation biology. Fish and Fisheries 3, 79–94.

Burnham, K. P. & Anderson, D. R. (2002). Model Selection and Multimodel Interference:A Practical Information-theoretic Approach . New York, NY: Springer-Verlag.

Collar, D. C., Wainwright, P. C. & Alfaro, M. E. (2008). Integrated diversification of loco-motion and feeding in labrid fishes. Biology Letters 4, 84–86.

David, L., Rajasekaran, P., Fang, J., Hillel, J. & Lavi, U. (2001). Polymorphism in ornamen-tal and common carp strains (Cyprinus carpio L.) as revealed by AFLP analysis and anew set of microsatellite markers. Molecular Genetics and Genomics 266, 353–362.

Einum, S. & Fleming, I. A. (1997). Genetic divergence and interactions in the wild amongnative, farmed and hybrid Atlantic salmon. Journal of Fish Biology 50, 634–651.

Excoffier, L., Laval, G., & Schneider, S. (2005). Arlequin ver. 3.0: an integrated softwarepackage for population genetics data analysis. Evolutionary Bioinformatics Online 1,47–50.

Fleming, I. A. & Einum, S. (1997). Experimental tests of genetic divergence of farmedfrom wild Atlantic salmon due to domestication. ICES Journal of Marine Science54, 1051–1063.

Fraser, D. J. (2008). How well can captive breeding programs conserve biodiversity? A reviewof salmonids. Evolutionary Applications 1, 535–586.

Garant, D., Fleming, I. A., Einum, S. & Bernatchez, L. (2003). Alternative male life-historytactics as potential vehicles for speeding introgression of farm salmon traits into wildpopulations. Ecology Letters 6, 541–549.

Guo, S. W. & Thompson, E. A. (1992). Performing the exact test of Hardy–Weinbergproportion for multiple alleles. Biometrics 48, 361–372.

Hansen, M. M. (2002). Estimating the long-term effects of stocking domesticated trout intowild brown trout (Salmo trutta) populations: an approach using microsatellite DNAanalysis of historical and contemporary samples. Molecular Ecology 11, 1003–1015.

Hart, P. J. B. (2003). Habitat use and feeding behaviour in two closely related fish species, thethree-spined and nine-spined stickleback: an experimental analysis. Journal of AnimalEcology 72, 777–783.

Hey, J., Won, Y. J., Sivasundar, A., Nielsen, R. & Markert, J. A. (2004). Using nuclear hap-lotypes with microsatellites to study gene flow between recently separated cichlidspecies. Molecular Ecology 13, 909–919.

Huntingford, F. A. (2004). Implications of domestication and rearing conditions for thebehaviour of cultivated fishes. Journal of Fish Biology 65, 122–142.

Iguchi, K., Matsubara, N. & Hakoyama, H. (2001). Behavioural individuality assessed fromtwo strains of cloned fish. Animal Behaviour 61, 351–356.

IUCN (1997). Conserving vitality and diversity. In Proceedings of the World ConservationCongress Workshop on Alien Invasive Species (Rubec, C. B. A. & Lee, G. O., eds),pp. 29–40. Ottawa: IUCN Species Survival Commission and North American WetlandsConservation Council.

Jacobsen, J. A. & Hansen, L. P. (2001). Feeding habits of wild and escaped farmed Atlanticsalmon, Salmo salar L., in the Northeast Atlantic. ICES Journal of Marine Science 58,916–933.

Johnsson, J. I., Hojesjo, J. & Fleming, I. A. (2001). Behavioural and heart rate responses topredation risk in wild and domesticated Atlantic salmon. Canadian Journal of Fisheriesand Aquatic Sciences 58, 788–794.

Kafuku, T. (1966). Morphological differences between domesticated common carp and wildone: speculation on the process of differentiation of the two carp races. Bulletin ofFreshwater Fisheries Research Laboratory 16, 71–82.



Kelley, J. L., Magurran, A. E. & Macias-Garcia, C. (2005). The influence of rearing experi-ence on the behaviour of an endangered Mexican fish, Skiffia multipunctata. BiologicalConservation 122, 223–230.

Langerhans, R. B., Gifford, M. E. & Joseph, E. O. (2007). Ecological speciation in Gambusiafishes. Evolution 61, 2056–2074.

Lucas, M. D., Drew, R. E., Wheeler, P. A., Verrell, P. A. & Thorgaard, G. H. (2004). Behav-ioral differences among rainbow trout clonal lines. Behavior Genetics 34, 355–365.

Mabuchi, K., Miya, M., Senou, H., Suzuki, T. & Nishida, M. (2006). Complete mitochondrialDNA sequence of the Lake Biwa wild strain of common carp (Cyprinus carpio L.):further evidence for an ancient origin. Aquaculture 257, 68–77.

Mabuchi, K., Senou, H. & Nishida, M. (2008). Mitochondrial DNA analysis reveals crypticlarge-scale invasion of non-native genotypes of common carp (Cyprinus carpio) inJapan. Molecular Ecology 17, 796–809.

MAFFJ (2000). The 10th Fisheries Census . Tokyo: Statistics Department, Ministry of Agri-culture, Forestry and Fisheries of Japan.

Manchester, S. J. & Bullock, J. M. (2000). The impacts of non-native species on UK biodi-versity and the effectiveness of control. Journal of Applied Ecology 37, 845–864.

Maruyama, T., Fujii, K., Kijima, T. & Maeda, H. (1987). Introductory Process of ForeignNew Fish Species . Tokyo: Fisheries Agency (in Japanese).

Matsuzaki, S. S., Usio, N., Takamura, N. & Washitani, I. (2007). Effects of common carp onnutrient dynamics and littoral community composition: roles of excretion and biotur-bation. Fundamental and Applied Limnology 168, 27–38.

Matsuzaki, S. S., Usio, N., Takamura, N. & Washitani, I. (2009). Contrasting impacts of inva-sive engineers on freshwater ecosystems: an experiment and meta-analysis. Oecologia158, 673–686.

McGinnity, P., Prodohl, P., Ferguson, K., Hynes, R., O’maoileidigh, N., Baker, N., Cotter, D.,O’hea, B., Cooke, D., Rogan, G., Taggart, J. & Cross, T. (2003). Fitness reduction andpotential extinction of wild populations of Atlantic salmon, Salmo salar, as a resultof interactions with escaped farm salmon. Proceedings of the Royal Society B 270,2443–2450.

Metcalfe, N. B., Valdimarsson, S. K. & Morgan, I. J. (2003). The relative roles of domestica-tion, rearing environment, prior residence and body size in deciding territorial contestsbetween hatchery and wild juvenile salmon. Journal of Applied Ecology 40, 535–544.

Millennium Ecosystem Assessment (2005). Ecosystems and Human Well-being: Synthesis .Washington, DC: Island Press.

Olla, B. L., Davis, M. W. & Ryer, C. H. (1998). Understanding how the hatchery environ-ment represses or promotes the development of behavioral survival skills. Bulletin ofMarine Science 62, 531–550.

van Oppen, M. J. H., Rico, C., Turner, G. F. & Hewitt, G. M. (2000). Extensive homoplasy,nonstepwise mutations, and shared ancestral polymorphism at a complex microsatellitelocus in Lake Malawi cichlids. Molecular Biology and Evolution 17, 489–498.

Randi, E. (2008). Detecting hybridization between wild species and their domesticated rela-tives. Molecular Ecology 17, 285–293.

Raymond, M. & Rousset, F. (1995). An exact test for population differentiation. Evolution49, 1280–1283.

Rhymer, J. M. & Simberloff, D. (1996). Extinction by hybridization and introgression. AnnualReview of Ecology and Systematics 27, 83–109.

Sundstrom, L. F. & Johnsson, J. I. (2001). Experience and social environment influence theability of young brown trout to forage on live novel prey. Animal Behaviour 61,249–255.

Svanback, R. & Eklov, P. (2004). Morphology in perch affects habitat specific feeding effi-ciency. Functional Ecology 18, 503–510.

Tymchuk, W. E., Biagi, C., Withler, R. & Devlin, R. H. (2006). Growth and behavioral con-sequences of introgression of a domesticated aquaculture genotype into a native strainof Coho salmon. Transactions of the American Fisheries Society 135, 442–455.

Volpe, J. P., Taylor, E. B., Rimmer, D. W. & Glickman, B. W. (2000). Evidence of naturalreproduction of aquaculture-escaped Atlantic salmon in a coastal British Columbiariver. Conservation Biology 14, 899–903.



Weeks, S. C., Gaggiotti, O. E., Schenck, R. A., Spindler, K. P. & Vrijenhoek, R. C. (1992).Feeding-behavior in sexual and clonal strains of Poeciliopsis. Behavioral Ecology andSociobiology 30, 1–6.

Electronic References

Rohlf, F. J. (2006). tpsDig, Version 2.10 . Stony Brook, NY: State University of New York.Available at http://life.bio.sunysb.edu/morph/

Rohlf, F. J. (2007a). tpsRelw, Version 1.45 . Stony Brook, NY: State University of New York.Available at http://life.bio.sunysb.edu/morph/

Rohlf, F. J. (2007b). tpsRegr, Version 1.35 . Stony Brook, NY: State University of New York.Available at http://life.bio.sunysb.edu/morph/


behavioural and morphological differences between feral ......behavioural and morphological...

Documents