Environmental biology of fishes, tập 93, số 4, 2012

Male body size further contributed to zebrafish reproductive success: offspring sired bylarge males exhibited higher hatching probability andthese offspring also hatched earlier and larg

Early development and larval behaviour of two clingfishes,

Lepadogaster purpurea and Lepadogaster lepadogaster

(Pisces: Gobiesocidae)

I Tojeira&A M Faria&S Henriques&C Faria&

E J Gonçalves

Received: 26 January 2011 / Accepted: 19 September 2011 / Published online: 8 October 2011

# Springer Science+Business Media B.V 2011

Abstract The recent revision on the taxonomic status

of Lepadogaster lepadogaster resulted in the division

of this species into L lepadogaster and L purpurea,

the clarification of each species’ distribution ranges

and the elimination of L zebrina (now in synonymy

with L lepadogaster) This new taxonomic status led

to the need of clarifying the early development of the

two species Embryonic development lasted 21 days

in L purpurea at a mean temperature of 14.2°C, and

16 days in L lepadogaster at a mean temperature of

16.5°C Newly hatched larvae of both species

measured 5.2 mm, had the mouth and anus opened,

pigmented eyes and almost no yolk At hatching andthroughout development the two species can bedistinguished by the ventral pigmentation which isabsent in L purpurea The change to a benthic mode oflife was gradual in both species, with larvae increasinglyspending more time close to the bottom until definitelysettling Larval development lasted 33 days in L.purpurea at a mean temperature of 14.6°C and 18 days

in L lepadogaster at a mean temperature of 16.5°C.Locomotion and foraging behaviours are describedfor both species L lepadogaster showed a higherfrequency of swimming and foraging behaviour whencompared with L purpurea

Keywords Lepadogaster lepadogaster Lepadogaster purpurea Gobiesocidae Clingfishes Early development Larval behaviour

Introduction

Clingfishes are distributed worldwide throughoutmany different habitats in tropical and temperate seas(Briggs 1955, 1986, 1990) However, knowledge oftheir behaviour (Gonçalves et al 1996, 1998) andecology (Henriques et al 2002) is extremely poor.This is related to their small size, which enables them

to occupy very cryptic microhabitats (Thresher1984).These species have a ventral sucking disk whichprovides an extra adaptation to explore crevices, holesand narrow spaces between rocks, as well as to resist

DOI 10.1007/s10641-011-9935-7

I Tojeira:A M Faria:C Faria:E J Gonçalves ( *)

Eco-Ethology Research Unit,

Instituto Superior de Psicologia Aplicada,

Centro de Investigação em Educação,

Faculdade de Ciências da Universidade de Lisboa,

Lisboa, Portugal

strong water movements (which are prevalent in the

intertidal and shallow subtidal habitats where they

occur) Demersal eggs are deposited on the underside

of stones, with the male guarding the egg mass which

may contain multiple batches at different stages of

development (Breining and Britz2000)

Lepadogaster purpurea (Bonnaterre 1788) and L

lepadogaster (Bonnaterre 1788) are two abundant

species of clingfishes (Briggs 1955, 1986)

Lepa-dogaster purpurea ranges from Scotland to Senegal,

the Canary and Madeira Islands and the

Mediterra-nean, while L lepadogaster occurs from as far north

as the extreme west of Galicia (Spain) to

north-west Africa, the Canary and Madeira Islands and the

Mediterranean (Henriques et al.2002) They are very

closely-related species, quite similar in their

morphol-ogy Adults can be distinguished by the different head

marks (or ocelli), as well as the number of the papillae

of the sucking disc regions The different body

colouration patterns, the length of the nostrils and

the distance between the eyes are other distinctive

characters used to identify each species (Henriques et

al 2002) They differ in microhabitat preferences;

both species occur in rocky boulder fields of the

intertidal and subtidal zones down to about 7 m of

depth, but L purpurea shelters in larger boulders and

can be found in greater depth than L lepadogaster

(Henriques et al 2002).The most striking difference

between these species is however the breeding period

Lepadogaster purpurea breeds mainly during the

winter until the beginning of the spring (October to

April) and L lepadogaster breeds mainly during the

spring until the beginning of the summer (March to

July) (Henriques et al.2002)

Due to this close resemblance, Lepadogaster

lepadogaster was considered until recently one single

species with two subspecies: L lepadogaster

lepa-dogaster and L lepalepa-dogaster purpurea (Henriques et

al.2002) These authors revised the taxonomic status

of L lepadogaster and divided this species into two

different ones: L lepadogaster and L purpurea This

taxonomic confusion and geographic overlap renders

the previously scattered descriptions of the early

stages of Lepadogaster (Guitel 1888) useless and

clarification is needed in order to correctly ascribe the

right larvae to the right species

The objective of this study is therefore to clarify

the early development of L lepadogaster and L

purpurea providing a detailed description of the

embryonic and larval stages The early ontogeny oflocomotor and foraging behaviours is also described

Materials and methods

Fourteen specimens of L purpurea were captured inJanuary and February 2006, and fourteen specimens

of L lepadogaster were captured in April 2006 atAlpertuche beach (38°28′ N; 8°59′ W) located at theArrábida Marine Park (Portugal), during the breedingseason of each species Individuals were kept in a

250 l tank illuminated with fluorescent light (60 W)

12 h per day and were fed twice a day with a varieddiet (shrimps, clams, cockles and mussels) Watertemperature was kept at 13°C for L purpurea and 15°Cfor L lepadogaster, according to the sea temperature

at the sampling site The substratum included severallayers of sand, with small (5–10 cm) and large (20–

30 cm) stones Shelter was formed by flat rocks thatwere also used as breeding sites by the males.For each species, eight batches were obtained Thecomplete embryonic development sequence for L.purpurea was based on a batch laid on 17 March

2006 (mean±S.D water temperature = 14.2±0.7°C,range =13–15°C, n=20) and for L lepadogaster ontwo batches laid on 1 June 2006 (mean±S.D watertemperature=16.50±0.46°C, range=16–17°C, n=16)and 11 June 2006 (mean±S.D water temperature=16.50±0.43°C, range =16–17°C, n=16) The remain-ing batches were used to complete descriptions,confirm specific developmental features, and validatethe sequence and timing of events and were notsampled on a daily basis No differences in thesequence and timing of events between these batchesand the ones used for descriptions were found Eggswere collected daily, the egg capsules were openedand the embryos distended to allow detailed observa-tions Mean number of eggs per batch, egg densityand egg mass area was calculated for each species.The larval development sequence was describedbased on two batches for each species For L purpurea,batches hatched on 6 February 2006 (mean±S.D watertemperature=14.60±0.54°C, range=13–15°C, n=39)and 3 March 2006 (mean±S.D water temperature=14.80±1.68°C, range =13–23°C, n=32), and for L.lepadogaster batches hatched on 13 May 2006(mean ± S.D water temperature = 17.80 ± 0.32°C,range =17–18°C, n=34) and on 1 June 2006 (mean±

S.D water temperature=17.10±0.43°C, range=16–

18°C, n=33) Upon hatching, larvae were collected

by aspiration from the parental aquarium and reared

in 25 l tanks illuminated with fluorescent light (15 W)

24 h per day A constant flow of seawater was

maintained Larvae were fed twice a day with a

mixture of Brachionus sp and Artemia sp nauplii

(2,040 individuals per 600 ml) and microalgae

During the first three days after hatching,

decapsu-lated eggs of Artemia sp were added to the mixture

Larvae were collected daily until metamorphosis

Both eggs and larvae (after having been anesthetized

with MS-222) were observed under a Nikon SMZ-800

stereomicroscope, photographed with a Nikon Coolpix

5400 camera and preserved in 4% saline formalin

buffered with sodium borate All larval measurements

correspond to standard length (LS)

In addition to embryonic and larval descriptions,

behavioural observations were conducted on a daily

basis from day 1 to day 17 post hatching for L

purpurea, and from day 1 to day 20 post hatching for

L lepadogaster Larval behaviours were categorized

into modal action patterns (MAPs) (Table1) A modal

action pattern is defined as a spatiotemporal pattern of

coordinated movement in which the pattern clusters

around some mode making the behaviour

recogniz-able (Barlow1968) During observations aeration was

stopped in order to avoid the influence of turbulence

on larval behaviour The focal animal technique

(Martin and Bateson 1993) was used to observe a

randomly selected larva for a 1-min interval This was

done for a total of ten larvae per day During each

observation period, the occurrence of seven modal

action patterns, grouped into three classes wasrecorded (Table 1) Locomotory and non-directedbehaviours were recorded as time variables, whereasforaging behaviours were recorded as frequencyvariables

A one way analysis of variance (ANOVA) wasused to test for differences between species in each

of the three classes of MAPs (locomotory, directed, and foraging) Prior to statistical analysis,the data were checked for homogeneity of variances(F-max test) and distribution normality (Shapiro-Wilk’s W test)

non-Results

Embryonic development

Both species laid the egg masses in a single layerunderneath the rocks, with L purpurea preferringlarger stones Generally, the male provided allparental care, fanning and rubbing the eggs untilhatching Newly laid eggs of both species were brightyellow; however eggs turned orange towards the end

of development They were oval in shape, with alower flattened surface containing fine filaments forattachment to the rocks (Fig.1(a), (b)) Egg diameterwas significantly different between the two species(mean±S.D.=1.80±0.04 mm, range =1.70–1.90, n=

53 for L purpurea and mean ±S.D.=1.90±0.03 mm,range =1.80–1.90, n=46 for L lepadogaster; t-test: t=

−2.35, df=34, p<0.05) For L purpurea, mean (± S.D.) batch area was 4.19±1.36 cm2(range =1.94–5.18,

Table 1 Definition of modal action patterns [MAPs —after Barlow ( 1968 )] observed in developing Lepadogaster larvae

Locomotory

Swim Forward movement of the larva through the water column using tail beats

Pause-Travel Larvae scans for prey; if prey is not located it moves a short distance, stops, and scans again Non-directed

Pause Larva is motionless and stationary in the water column

Sink Larva is motionless and descends through the water column, usually head first

Foraging

Orientation The head is directed towards a prey item

Fixate The larva is stationary and bends its body into an “S” shape position; typically follows orient Lunge The larva moves towards the prey item from the fixate position in an attempt to capture it

n=6), mean (± S.D.) number of eggs per batch was

144.40±46.46 eggs (range =60.10–177.10, n=6) and

mean (± S.D.) egg density was 34.50±3.39 eggs.cm−2

(range = 31.00–39.00, n=6) For L lepadogaster,

mean (± S.D.) batch area was 4.61 ± 2.46 cm2

(range =1.84–8.60, n=7), mean (± S.D.) number of

eggs per batch was 123.10±53.74 eggs (range=44.16–

176.70, n=7) and mean (± S.D.) egg density was

27.71±4.27 eggs.cm−2(range=22.00–32.00, n=7)

Embryonic development lasted 21 days in L

purpurea (mean ± S.D water temperature = 14.20 ±

0.67°C, range=13–15°C, n=20) and 16 days in L

lepadogaster (mean±S.D water temperature=16.50±

0.46°C, range=16–17°C, n=16) The main

ontoge-netic events of embryonic development for each

species are shown in Table 2 Embryos of both

species had a non-segmented yolk, consisting of a

large oil droplet surrounded by several small ones

Circulation of the blood fluid was first registered

on day 9 in L purpurea, and on day 6 in L

lepadogaster Pigmentation first appeared in the eyes,

followed by the lateral side of the body In L

purpurea, round melanophores started to cover the

middle region, then spread to the anterior and

posterior body areas (cephalic and caudal areas)

These round melanophores changed into star shaped

melanophores throughout development In L

lepa-dogaster, star shaped melanophores started to cover

both the lateral side of the body and the caudal region

until the end of the myomeres, spreading to the

cephalic region in the following days

The major difference between the embryos of thesetwo species was the absence or residual pigmentation

in the ventral fin fold region of L purpurea (Fig 1

(a)), which clearly contrasted with the strong tation due to the star shaped melanophores in the gutventral region in L lepadogaster (Fig 1 (b)) Threestar shaped melanophores, which became ramifieduntil hatching, were also present in the lower jaw of

pigmen-L lepadogaster (Fig.1(b)) but absent in L purpurea.Near hatching, movements of the embryos increased,

in particular eye movements Hatching of the entirebatches occurred throughout a 2-day period Larvae ofboth species hatched with the head first and immedi-ately swam to the surface where they seemed to gulpair, probably to fill the gas bladder Figure 2 showslarvae collected at different developmental stages andTable3presents the main ontogenetic events of larvaldevelopment for both species

Larval development

Newly hatched larvae of both species measured onaverage 5.20±0.08 mm (mean±S.D.) (range=5.00–5.30 mm, n = 5 for L purpurea; range = 5.20–5.30 mm, n=5 for L lepadogaster) and hatched withthe mouth and anus opened, lips and jaws differen-tiated, eyes completely formed and fully pigmented,the nostrils opened and the yolk almost fully absorbed(Fig 2 (a-I), (b-I)) The opercula were open, withthree branchial arches present; the characteristicnostril tentacles of adult fishes were not yetFig 1 (a) Dorsal view of the embryo of Lepadogaster purpurea and (b) Lepadogaster lepadogaster

developed; the liver and the heart were completelyformed and the blood circulation was noticeable.Larvae hatched with both pectoral fins differentiatedbut without any rays and with the median fin foldranging from the cephalic area to the anus.

Pigmentation patterns were quite similar (with fewexceptions) and changed throughout development.Both larvae were strongly pigmented at the time ofhatching, with two parallel rows in the dorsal areacomposed of ca 30 ramified melanophores in L.purpurea and 26 in L lepadogaster, which covered

Lepa-the pre- and post-anal area until Lepa-the last 4–6

myomeres In the lateral trunk, ramified

melano-phores were distributed from the post-opercular area,

forming a V-pattern coincident with the myomeres,

until the last 4–6 myomeres In the ventral trunk,

there were ca 12 ramified melanophores in L

purpurea and eight in L lepadogaster Close to the

urostyle, two round melanophores were present in

both species On the ventral area of the median fin

fold there were ca 30 star-shaped melanophores in L

purpurea and 21 in L lepadogaster The gut region

was heavily pigmented on both the dorsal and lateral

sides, with ramified melanophores In L purpurea,

the anus was fully surrounded by melanophores while

in L lepadogaster, melanophores agglomerated near

the anus but did not surround it The gas bladder had

a distinct dorsal pigmentation in both species

Similarly to what was observed for the embryos, the

most noticeable difference between species was the

absence or residual ventral pigmentation in the gut

region of L purpurea, while in L lepadogaster more

than 30 ramified melanophores were present in the

gut ventral region On the base of the pectoral fin

there were two melanophores in L purpurea and

three in L lepadogaster On the dorsal cephalic area,

larvae of both species had three major sets of ramified

melanophores organised in the following pattern

(counting from the tip of the nose to the post-ocular

area): 4+11+13 in L purpurea and 4+9+7 in L

lepadogaster On the median head, behind theopercula, there were ca 10 ramified melanophores

in both species Inside the internal otic vesicles, from

a dorsal view, there were two melanophores visible in

L purpurea and four in L lepadogaster A singlepunctiform melanophore on the gular region, as well

as two ramified melanophores in the throat region(four in the inferior lip and one in the opercula), werepresent in L lepadogaster but absent in L purpurea

In L purpurea, three ramified melanophoresappeared in a row in the post-anal lateral side abovethe notochord on day 4 (6.7–7.3 mm) These pig-ments disappeared later in development at day 10(7.8–8.3 mm) At day 12 (8.0–8.3 mm), the notochordflexion started, the ventral disc (modification of thepelvic fins) started to differentiate and the larval bodybecame less pigmented, with the exception of the star-shaped melanophores in the anal median fold and inthe cephalic region [Fig 2 (a-II)] At day 33 (9.4–9.5 mm), all individuals were settled and acquired abenthic life style, and fin rays were formed: D=17(17–21); C=11 (11–14); A=11 (10–12); P=21 (20–23) Larvae begun slowly to metamorphose andpigmentation started to become similar to the adultfish By day 39 (9.5–10 mm) the nostril tentacleswere already formed

In L lepadogaster, a regression on the expansion

of the melanophores in the lateral region wasregistered at day 6 (7.1–7.3 mm), coinciding with

Table 3 Ontogenetic events of larval development of L.

purpurea and L lepadogaster in order of first appearance

(days after hatching): [1] filled gas bladder; [2] yolk absorption;

[3] exogenous feeding; [4] caudal fin rays; [5] pectoral fin rays;

[6] notochord starts to flex; [7] ventral disk differentiation; [8]

larvae started to settle; [9] dorsal fin rays; [10] anal fin rays;

[11] ossified vertebra; [12] teeth; [13] notochord flexion completed; [14] all larvae settled; [15] median fin fold reabsorption; [16] tentacles differentiation; [17] juvenile typical pigmentation Size ranges are shown for the main ontogenetic stages

the notochord flexion, and the ventral disc started to

develop [Fig 2 (b-II)] At day 13 (8.0–8.3 mm),

seven ramified melanophores were noticeable in the

anterior, medium and posterior area of the ventral disc

region Larva started to make contact with the bottom

of the aquarium at day 7 and by day 18 (8.4–9.3 mm)

all larvae were settled At this time, all fin rays were

formed: D=17 (17–21); C=12 (11–14); A=11 (10–

12); P=21 (20–23), and the nostril tentacles begun to

differentiate

Juvenile pigmentation started to appear after

settlement (Fig 2 (a-III), (b-III)) In both species,

the eyes were silver with some pale red and orange

colours, the head was carmine with some whitish

spots and the body was heavily pigmented with

carmine-orange pigments In L purpurea, the

unpig-mented ventral region started to acquire a pinkish

shade Both species presented three sets of

melano-phores in the cephalic region (from the tip of the nose

to the post-ocular area: 2+11+13) and three ramifiedmelanophores could be distinguished on the base ofthe pectoral fin In L purpurea, three melanophoreswere present at the base of the dorsal fin between the1st, 3rd and 5th rays; the caudal fin had two ramifiedmelanophores between the 3rd and 5th rays and 1ramified melanophore at the end of the notochord; asingle melanophore in the anterior area of the suckingdisc, two in the median area and one in the posteriorarea were characteristic at this stage In L lepa-dogaster, three star shaped melanophores were pres-ent in the caudal fin region at day 21 (8.3–8.5 mm).These were the same melanophores that were visible

at the end on the notochord before flexion started Thedorsal region of the gut was heavily pigmented and inthe anus opening there was a set of melanophores Inthe ventral area of the gut, two rows of ramifiedmelanophores were also distinguished Only the base

of the dorsal and anal fins was pigmented with the

Table 4 Comparison of the main developmental aspects between the two species during the embryonic, larval and juvenile stages

L purpurea Absence or residual

pigmentation in the ventral fin fold region

Melanophores surrounding the anus

The unpigmented ventral region start to acquire a pinkish shade Absence of pigmentation

in the lower jaw

Absence or residual pigmentation in the gut region

3 melanophores at the base of the dorsal fin (between the 1st,3rd and 5th rays)

Absence of pigmentation in the lower jaw

2 ramified melanophores in the caudal fin (between the 3rd and 5th rays)

1 ramified melanophore at the end

of the notochord

1 melanophore in the anterior area

of the sucking disc, 2 in the median area and 1 in the posterior area

L lepadogaster Strong pigmentation of star

shaped melanophores in the gut ventral region

Melanophores agglomerate near the anus but do not surround it

3 star melanophores in the caudal fin region

3 star shaped melanophores (became ramified in the end of development) present in the lower jaw

More than 30 ramified melanophores in the gut region

A set of melanophores at the anus opening

1 punctiform melanophore on the gular region

2 rows of ramified melanophores

in the ventral area of the gut

2 ramified melanophores in the throat region (4 in the inferior lip and 1 in the opercula)

Base of dorsal and anal fins heavily pigmented Anal fin with star shaped melanophores

anal fin presenting star shaped melanophores A

single ramified melanophore in the inferior jaw was

observed A comparison of the ontogenetic features

between species is shown in Table4

Larval behaviour

Larval behaviour in both species (see Table 1) was

characterized by a constant locomotory activity

(Fig 3) Nevertheless, L lepadogaster was more

active (ANOVA, d.f.=34, F=6.03, P=0.02), spending

80 to 100% of the time in locomotion activities

Locomotion in both species was characterized by a

shift from pause-travel behaviour to swimming

around day 5, which is clearer in L lepadogaster

(Fig 4) Although very variable, non-directed modal

action patterns (MAPs) tended to decrease with age

L purpurea spent more time with pause and sink

MAPs when compared to L lepadogaster (ANOVA,

d.f.=34, F=10.36, P=0.003) (Fig 5) Foraging

be-haviour frequencies were highly variable, especially

for L purpurea, which also exhibited the lowest

foraging MAPs (ANOVA, d.f.=34, F=4.48, P=0.04)(Fig 6) In both species, orientation to the prey wasalways followed by fixation; however, the frequency

of attacks (lunge MAP) was lower since attack did notalways followed orientation and fixation behaviours

Discussion

Close resemblances between the gobiesocids dogaster lepadogaster and Lepadogaster purpurealead to taxonomic confusion that needed a clarifica-tion, especially at the embryonic and larval stages In

Lepa-L purpurea the mean length of the long axis of theeggs was 1.8 mm and 1.9 mm in L lepadogaster,which agrees with the available descriptions on othergobiesocids (Allen1984; Hefford1910; Padoa 1956;Russel 1976) However, Breining and Britz (2000)report smaller lengths for L lepadogaster, rangingfrom 1.5 to 1.8 mm According to several authors, thenumber of eggs per clutch can vary in the wild from

200 to 250 (Allen 1984; Russel1976), or up to 300

0 10 20 30 40 50 60 70 80 90 100

Days after hatching

Fig 3 Mean proportion of

time Lepadogaster

purpurea (closed symbols)

and Lepadogaster

lepadogaster (open

symbols) larvae spent

performing total locomotory

(triangles), total

non-directed (circles) and

total foraging (squares)

MAPs throughout

development

0 10 20 30 40 50 60 70 80 90 100

Days after hatching

Fig 4 Mean proportion of

time Lepadogaster

pur-purea (closed symbols) and

(Padoa 1956) Nevertheless, in this study, the larger

clutch (obtained in captivity) contained 177 eggs and

three different developmental stages were recognized,

which agrees with previous descriptions (Breining

and Britz 2000) Eggs of the two species differed in

the amount of pigmentation in the embryo, with L

lepadogaster presenting significantly more pigments,

especially in the mouth and ventral region of the gut

The advanced developmental stage of larvae at the

time of hatching is typical of marine fishes with male

parental care that spawn demersal eggs (Thresher

1984; Sponaugle et al 2002; Hickford and Schiel

2003) After hatching, larvae use successive

swim-ming impulses to immediately swim to the surface

where they seem to gulp air, probably filling their

swim bladders This behaviour has also been reported

in other demersal spawners, such as Gobius paganellus

(Pisces: Gobiidae) (Borges et al.2003)

At hatching and throughout development the two

species can be distinguished by the ventral

pigmen-tation which is absent in L purpurea Larvae of both

species can be easily distinguished from those of L.candolii (the only other species of the genus) whichare considerable less pigmented (Guitel1888; Russel

1976; Allen1984)

The change to a benthic mode of life wasgradual in both species, with larvae increasinglyspending more time close to the bottom untildefinitely settling Larval development lasted

18 days in L lepadogaster and approximately

30 days in L purpurea Different results couldhowever be expected if larvae were reared alternatinglight and dark conditions Nevertheless, the light:darkregime used did not seem to have induce relevantchanges in the duration of the larval period of L.lepadogaster, since the pelagic larval duration ofreared larvae was well within the values reported forwild larvae (range = 11–18 days; Beldade et al.2007)and other spring-spawners temperate gobiesocids (e.g.Apletodon dentatus = 15 days, Gouania wildenowi =

17 days, Lepadogaster candollei = 13 days; Raventósand Macpherson 2001) There are however no data

0 5 10 15 20 25 30 35 40

Days after hatching

Fig 5 Mean proportion of

MAPs (pause —triangles;

sink —circles) throughout

development

0 1 2 3

Days after hatching

Fig 6 Frequency of orient

(triangles) and lunge

(circles) MAPs of

Lepa-dogaster purpurea (closed

symbols) and Lepadogaster

lepadogaster (open

sym-bols) larvae throughout

development

available for wild L purpurea larvae or other

winter-spawner gobiesocids, and therefore a word of caution

is needed since photoperiod may affect growth rates

and pelagic larval durations These differences in

larval development are relevant for species that hatch

with the same size and similar morphologies and may

be explained by the striking difference in the breeding

periods of the two species The lower water

temper-ature in the winter probably increases developmental

time of both eggs and larvae since it is well know that

developmental time decreases with increasing

tem-perature in many fish species (Blaxter1969)

Behavioural observations showed that L purpurea

was less active when compared to L lepadogaster,

spending more time in resting activities (pause and

sink) This again may be a consequence of the longer

developmental period of L purpurea, which implies

spending significantly longer times in the water

column We hypothesise that during this extended

larval period in the winter and early spring, larvae

will likely encounter periods of low plankton

avail-ability Spending more time in pause and sink

behaviours is likely to be a better strategy for saving

energy since locomotion and foraging activities are

energetically costly to larvae (Kiørboe and Munk

1986)

Both species changed from a more ‘saltatory’

strategy to a ‘cruise’ strategy during development,

approximately coincident with the beginning of

notochord flexion The pause-travel behaviour is

probably associated to a saltatory strategy of

search-ing for prey where the search for prey occurs only

while pausing between swimming events (Browman

and O’Brien 1992a, b) As larvae became more

developed, most of their time was spent swimming,

and they adopted a cruise strategy: the search for prey

occurs while swimming (Munk and Kiørboe 1985)

Additionally, foraging behaviour increased with

de-velopment, which can be explained by an enhanced

swimming capacity and visual acuity, which in turn

will improve encounter rates and feeding success

(Miller et al 1993) The nature of the non-directed

activities, such as sink and pause, is not

straightfor-ward (Rabe and Brown 2001) Sinking has been

reported in other species, such as the snapper Pagrus

auratus (Pankhurst et al 1991) and the black sea

bream Acanthopagrus schlegeli (Fukuhara 1987),

and, like the pause behaviour, it has been interpreted

as a resting behaviour

The available descriptions for northeast Atlanticand Mediterranean larvae of Gobiesocidae are clearlyincomplete (see Guitel 1888; Padoa 1956; Russel

1976) In particular, the two species described in thisstudy have been mistakenly identified until veryrecently (Henriques et al 2002) Consequently, thefew larval developmental studies available for Lep-adogaster (Padoa 1956; Russel 1976) need to bereassessed since developmental stages were all mixedinto a single species and several misidentifications arelikely to have occurred For example, the descriptionsmade by Guitel (1888) for L purpurea (previouslyconsidered as L gouanii—see Henriques et al 2002)were in fact larvae of L lepadogaster The availabledrawings clearly show the presence of ventralpigmentation and pigments on the lower jaw whichare absent from L purpurea, unmistakably ascribingthese larvae to L lepadogaster

The correct identification of fish larvae is the basisfor ecological and taxonomic studies of the pelagicstage of fishes (Leis and McCormick2002) Errors inidentification can lead to misinterpretations of eco-logical processes (Powles and Markle 1984) andstudies like the present one are essential to improveour knowledge on the early life stages of marine fish

Acknowledgements We would like to thank R Lourenço for the illustrations and F Gil for help during larval rearing This work was supported by a PhD grant to AMF (SFRH/BD/21742/ 2005) and through the Pluriannual Program (R & D Unit 331/ 94), financed by Fundação para a Ciência e a Tecnologia (FCT).

References

Allen LG (1984) Gobiesociformes: development and ships In: Moser HG, Richards WJ, Cohen DM, Fahay MP, Kendall AW Jr, Richardson SL (eds) Ontogeny and systematics of fishes The American Society of Ichthyol- ogists and Herpetologists, pp 629–636

relation-Barlow GW (1968) Ethological units of behavior In: Ingle DJ (ed) The central nervous system and fish behavior University of Chicago Press, Chicago, pp 217–232 Beldade R, Pedro T, Gonçalves EJ (2007) Pelagic larval duration of 10 temperate cryptobenthic fishes J Fish Biol 71:376 –382 doi: 10.1111/j.1095-8649.2007.01491.x

Blaxter JHS (1969) Development: eggs and larvae In: Hoar

WS, Randall DJ (eds) Fish physiology, vol III Academic, New York, pp 178 –252

Borges R, Faria C, Gil F, Gonçalves EJ, Almada VC (2003) Embryonic and larval development of Gobius paganellus

(Pisces: Gobiidae) J Mar Biol Assoc UK 83:1151–1156.

doi: 10.1017/S0025315403008415h

Breining T, Britz R (2000) Egg surface of three clingfish

species, using scanning electron microscopy J Fish Biol

56:1129 –1137 doi: 10.1111/j.1095-8649.2000.tb02128.x

Briggs JC (1955) A monograph of the clingfishes (order

Xenopterygii) Stanf Ichthyol Bull 6:1 –224

Briggs JC (1986) Gobiesocidae In: Whitehead PJP, Bauchot

M-L, Hureau J-C, Nielsen J, Tortonese E (eds) Fishes of

the North-Eastern Atlantic and the Mediterranean.

UNESCO, Paris, pp 1351 –1359

Briggs JC (1990) Gobiesocidae In: Quéro JC, Hureau JC,

Karrer C, Post A, Saldanha L (eds) Check-list of the fishes

of the eastern tropical Atlantic UNESCO, Societas

Euro-paea Ichthyologorum & JNICT, Lisboa, pp 474 –480

Browman HI, O ’Brien WJ (1992a) The ontogeny of search

behaviour in the white crappie Pomoxis annularis Env

Biol Fish 34:181 –195 doi: 10.1007/BF00002393

Browman HI, O’Brien WJ (1992b) Foraging and prey-search

behaviour of golden shiner (Notemigonus crysoleucas)

larvae Can J Fish Aquat Sci 49:813–819 doi: 10.1139/

f92-092

Fukuhara O (1987) Larval development and behavior in early

life stages of black sea bream reared in the laboratory Bull

Jpn Soc Sci Fish 53:371 –379

Gonçalves EJ, Almada VC, Almeida SP, Gonçalves DM, Repas

M, Simões N (1996) Observations on the agonistic

behaviour of Lepadogaster lepadogaster purpurea (Pisces:

Gobiesocidae) J Fish Biol 49:367 –369 doi: 10.1111/

j.1095-8649.1996.tb00032.x

Gonçalves D, Gonçalves EJ, Almada VC, Almeida SP (1998)

Comparative behaviour of two species of Lepadogaster

(Pisces: Gobiesocidae) living at different depths J Fish Biol

53:447 –450 doi: 10.1111/j.1095-8649.1998.tb00992.x

Guitel F (1888) Recherches sur les Lepadogasters Arch

Zoologie Exp Gen, 2ème série, 6:423–647

Hefford AE (1910) Notes on teleostean ova and larvae

observed at Plymouth in spring and summer, 1909 J

M a r B i o l A s s o c U K 9 : 4 7–48 doi: 1 0 1 0 1 7 /

S0025315400073227

Henriques M, Lourenço R, Almada F, Calado G, Gonçalves

DM, Guillemaud T, Cancela ML, Almada VC (2002) A

revision of the status of Lepadogaster lepadogaster

(Tele-ostei: Gobiesocidae): sympatric subspecies or a long

misunderstood blend of species? Biol J Linn Soc

76:327 –338 doi: 10.1046/j.1095-8312.2002.00067.x

Hickford MJ, Schiel DR (2003) Comparative dispersal of larvae from demersal versus pelagic spawning fishes Mar Ecol Prog Ser 252:255–271 doi: 10.3354/meps252255

Kiørboe T, Munk P (1986) Feeding and growth of larval herring, Clupea harengus, in relation to density of copepod nauplii Environ Biol Fish 17:133 –139

Leis JM, McCormick MI (2002) The biology, behavior, and ecology of the pelagic, larval stage of coral reef fishes In: Sale

PF (ed) Coral reef fishes: dynamics and diversity in a complex ecosystem Academic Press, San Diego, pp 171 –199 Martin P, Bateson P (1993) Measuring behavior: an introduc- tory guide Cambridge University Press, Cambridge Miller TJ, Crowder LB, Rice JA (1993) Ontogenetic changes in behavioural and histological measures of visual acuity in three species of fish Environ Biol Fish 37:1 –8

Munk P, Kiørboe T (1985) Feeding behaviour and swimming activity of larval herring (Clupea harengus) in relation to density of copepod nauplii Mar Ecol Prog Ser 24:15–21 doi: 10.3354/meps024015

Padoa E (1956) Uova, larva e stadi giovanili de Teleostei Fauna e Flora del Golfo di Napoli Monogr 38:783–877 Pankhurst PM, Montgomery JC, Pankhurst NW (1991) Growth, development and behaviour of artificially reared larval Pagrus auratus (Bloch & Schneider, 1801) (Sparidae) Aust

J Mar Freshw Res 42:391 –398 doi: 10.1071/MF9910391

Powles H, Markle D (1984) Identification of fish larvae In: Moser HG, Richards WJ, Cohen DM, Fahay MP, Kendall

AW Jr, Richardson SL (eds) Ontogeny and systematics of fishes The American Society of Ichthyologists and Herpetologists, pp 31 –33

Rabe J, Brown JA (2001) The behavior, growth, and survival of witch flounder (Glyptocephalus cynoglossus) larvae in relation to prey availability: adaptations to an extended larval period Fish Bull 99:465 –474

Raventós N, Macpherson E (2001) Planktonic larval duration and settlement marks on the otoliths of Mediterranean littoral fishes Mar Biol 138:1115–1120

Russel FS (1976) The eggs and planktonic stages of British marine fishes Academic, London

Sponaugle S, Cowen R, Shanks A, Morgan S, Leis J, Pineda J, Boehlert G, Kingsford M, Lindeman K, Grimes C, Munro

J (2002) Predicting self-recruitment in marine populations: biophysical correlates and mechanisms Bull Mar Sci 70:341 –375

Thresher RE (1984) Reproduction in Reef Fishes T.F.H Publications, Neptune City

Paternal body size affects reproductive success

S Uusi-Heikkilä&A Kuparinen&C Wolter&

T Meinelt&R Arlinghaus

Received: 18 February 2011 / Accepted: 10 October 2011 / Published online: 11 November 2011

# Springer Science+Business Media B.V 2011

Abstract Across many fish species, large females

tend to exhibit higher individual reproductive success due

to elevated fecundity and the provisioning of better

conditioned eggs and offspring compared to small

females By contrast, effects of paternal body size on

reproductive success are less well understood We

disentangled the maternal- and paternal-size dependent

effects on reproductive output and early life history inzebrafish (Danio rerio) In the laboratory, females andmales from four size categories (small, medium-sized,large and very large) were allowed to spawn freely in

a full factorial design with 10 replicates per sizecombination As expected, larger females producedmore eggs and better conditioned offspring compared

to smaller females Male body size further contributed

to zebrafish reproductive success: offspring sired bylarge males exhibited higher hatching probability andthese offspring also hatched earlier and larger thanoffspring fertilized by small males However, thelargest males experienced lower mating success andreceived fewer eggs than males of the smaller sizeclasses While male body size substantially affectedreproductive success in zebrafish, it remained unclearwhether and to what degree direct paternal effects(e.g., related to sperm quality) or indirect paternaleffects stemming from differential allocation pat-terns by females were the mechanism behind ourfindings Answering this question constitutes animportant future research topic

Keywords Maternal effect Paternal effect Reproductive fitness Reproductive success

Introduction

Females are known to invest more resources into anembryo relative to males whose contribution if often

DOI 10.1007/s10641-011-9937-5

S Uusi-Heikkilä (*):C Wolter:R Arlinghaus

Department of Biology and Ecology of Fishes,

Leibniz-Institute of Freshwater Ecology

and Inland Fisheries,

Müggelseedamm 310,

12587 Berlin, Germany

e-mail: silva.uusi-heikkila@igb-berlin.de

A Kuparinen

Ecological Genetics Research Unit,

Department of Biological and Environmental Sciences,

University of Helsinki,

P.O Box 65, 00014 Helsinki, Finland

T Meinelt

Department of Ecophysiology and Aquaculture,

Leibniz-Institute of Freshwater Ecology

and Inland Fisheries,

Müggelseedamm 301,

12587 Berlin, Germany

R Arlinghaus

Inland Fisheries Management Laboratory,

Department for Crop and Animal Sciences,

Faculty of Agriculture and Horticulture,

Humboldt-University of Berlin,

Invalidenstrasse 42,

10115 Berlin, Germany

confined to sperm only, thus it is commonly believed

that a progeny’s phenotype is more strongly

influ-enced by the female’s phenotype than by the

phenotype of the male (Chambers and Leggett1996;

Heath et al.1999) Any effect of maternal phenotype

on the offspring’s phenotype is referred to as maternal

effect (Bernardo1996; Mousseau and Fox1998) The

positive relationship between female body size and

offspring performance is supported by findings

according to which large females of many marine

and freshwater fish species spawn greater numbers

and often larger eggs and larvae compared to their

smaller-sized conspecifics (reviewed in Wootton

1998; Green 2008; Marshall et al 2008) However,

such results should not be prematurely generalized

across fish species and ecological contexts (McLean

et al 2004; Kamler 2005; Marshall et al 2010) In

fact, larvae hatching from eggs produced by large

females may also be smaller than larvae hatching

from eggs produced by small females due to the

size-dependent variance in hatching time and differences

in larval resource intake (Heath et al.1999)

Accord-ing to the fundamental life-history trade-offs (e.g.,

trade-off between egg number and size; Stearns1989;

Roff 2002) it is unlikely that large females can

maximize all reproductive traits simultaneously Thus,

the often-cited positive influence of maternal size on

offspring performance and reproductive success may

not always apply in nature (Marshall et al.2010)

Relative to maternal-size effects, the effects of

paternal body size on reproductive success may be

less pronounced and it has also been studied less

intensively (Chambers and Leggett 1996) Males’

contribution to offspring development has not been

assumed to be as distinct as that of females’ because

males do not provide any extra-nuclear material or

nutrition to the developing offspring (Marteinsdottir

and Steinarsson1998; Kennedy et al.2007) However,

paternal-size effects can operate directly, either

through genetic contribution to the developing

off-spring (e.g ‘good genes’ –hypothesis; Andersson

1994) or via physiological and energetic pathways

For example, larger males may have larger testis and

longer spermatozoa with higher motility, which may

elevate fertilization rates (Gage et al.2004), compared

to small males (e.g., Howard et al.1998; Skinner and

Watt 2007a) Furthermore, the effect of male body

size on reproductive success may be indirectly

expressed by increased female reproductive

investment when mating with a high quality (e.g.,large) male (Howard et al 1998; Kolm 2001) Theadvantage female gains by investing more reproduc-tive resources towards high quality male (known asdifferential allocation; Burley 1988) is thought to beassociated with the prospects these males offer to thefemale’s fitness, such as better oviposition sites ormore intensive parental care in nest-guarding species(e.g., Sabat1994)

Much of the previous research focused on detectingparental-size dependent effects on reproductive success

in fish has been conducted by using artificial fertilizationexperiments in the laboratory (e.g., Chambers et al

1989; Marteinsdottir and Steinarsson 1998) Thisinhibits sexual selection, mate choice and differentialresource allocation patterns to be expressed, poten-tially biasing study findings regarding to maternal andpaternal-size effects on reproductive traits (Thériault

et al 2011) To address this issue, a model species,which allows individuals to spawn freely and expressmate choice may be useful We used zebrafish (Daniorerio, Hamilton) to investigate maternal and paternal-size dependent effects on reproductive success usingnatural spawning events without artificial insemina-tion in a full factorial design We defined reproductivesuccess as a combination of important reproductivetraits, such as spawning probability, clutch size, eggand larval size, embryo survival and hatchingprobability

Zebrafish is a small-bodied, batch-spawning cyprinidspecies, which spawn all year round under laboratoryconditions (Spence and Smith 2006) Domesticatedstrains spawn at intervals of 1 to 6 days, and clutchsize is known to correlate positively with inter-spawning interval (Spence and Smith 2006), femaleage (Eaton and Farley 1974) and body size (Spenceand Smith 2006; Uusi-Heikkilä et al 2010) Inaddition, female reproductive success may correlatewith male body size as females have been shown toprefer (Pyron 2003) and differentially allocate eggstowards larger males if exposed to large and smallmales in a short sequence (Skinner and Watt2007b).However, other zebrafish studies have not reportedfemale mating preference towards large males(Spence and Smith 2006; Hutter et al 2010), andultimate female mate choice may be related to othervisual (Hutter et al.2010) or olfactory cues (Gerlachand Lysiak 2006) In addition to mate preferences,sex-ratio and population density have been shown to

affect zebrafish mating behavior and reproductive

success (Spence and Smith2005; Spence et al.2006)

High density and biased sex-ratio may lead to

increased levels of aggressive interactions among

males, which can have a negative effect on female

egg production (Pritchard 2001; Spence and Smith

2005; Paull et al 2010) Although parental-size

dependent effects on zebrafish mating success,

repro-ductive output and early life-history traits have been

demonstrated earlier in trials comparing spawners

composed of similarly-sized individuals

(Uusi-Heikkilä et al 2010), the contribution of either

maternal or paternal-size effects and their interaction

on reproductive success remains obscure The

objec-tives of the present study were to investigate whether

the higher reproductive success of large zebrafish

spawners relative to small spawners (Uusi-Heikkilä

et al 2010) is determined mostly by female body size

or whether the variation in reproductive success is

also related to male body size or the interaction

between female and male body size We hypothesized

that both female and male size contribute to the

reproductive success in zebrafish but the effect of

female body size on early life-history traits was

expected to be greater than the effect of male body

size

Materials and methods

Fish holding conditions

Our experimental fish were third generation offspring

from a wild zebrafish population captured from a

river system 70 km west of Coochbihar (West-Bengal,

India, 22.56°N, 87.67°E) Fish were raised in six

glass fiber - polyester tanks (diameter: 79 cm, height:

135 cm, volume: 320 l) in a light (14 h light: 10 h

dark) and temperature controlled (26.8±0.79°C, mean ±

S.D.) recirculation facility with an inflow rate of

0.25 l s−1 The recirculation system was run with

insipid tap water, and the water quality was controlled

weekly for pH (8.4 ± 0.1), nitrite (N-NO2-; <

0.3 mg l−1), ammonium (N-NH4 ; < 0.05 mg l−1),

and daily for oxygen levels (7.9±0.6 mg l−1) The

stocking density per holding tank was 0.9 ± 0.2

individuals l−1 We fed fish with freshly hatched

Artemia-nauplii (Inve Aquaculture NV, www.inve

com) and commercial flake food (TetraMin, Tetra

GmbH, www.tetra.net; 47% protein, 10% fat) adlibitum Fish were fed five times per day with smallamounts of food as it has been shown to result in anefficient feed utilization and to maximize growth andreproductive output (Priestley et al.2006)

At age 250 days post fertilization (dpf), femalesand males were caught using a dip net Zebrafish startmaturing at age 90 dpf (Schilling 2002) and atstandard length of about 19 mm (Uusi-Heikkilä et

al 2011) so by the time our experiment was initiatedall fish were mature We measured females and malesfor standard length (SL) to the nearest mm and thenassigned them into four different size categories:small (24–25 mm), medium (26–27 mm), large (28–

29 mm) and very large (30–31 mm) The size rangeswere based on a preliminary experiment, wherefemales below 24 mm were found having anextremely low reproductive success, and fish above

31 mm were rare in our experimental populations Wecoupled females and males from different sizecategories with each other employing a full factorialdesign and consequently produced 16 different sizecombinations of females and males, each replicated

10 times (altogether 160 couples) This full factorialdesign allowed us to disentangle the size-dependentfemale and male contributions to reproductivesuccess

By the time the experiment was initiated,spawners (i.e., mature females and males) weretransferred into a standalone spawning facility(Aquarien-Bau Schwarz, 37081 Göttingen, Germany,

by separating adults from eggs as previously applied

by Uusi-Heikkilä et al (2010) A grid of a mesh size

of 2×2 mm was inserted inside of each spawning box(volume 3 l, length: 21 cm, width: 11 cm, height:

13 cm) Each box was additionally equipped withgreen plastic filter material serving as a spawningsubstrate Spawning boxes were stocked with onefemale and one male Reproductive success of the fishfrom the four different size categories was assessedfor four consecutive days Zebrafish are known tospawn every 1–6 days (Spence and Smith2006), thus

it was likely that each couple willing to spawnreproduced at least once during the 4 days spawningperiod

Reproductive traits

Reproductive output

Zebrafish usually spawn within the first few hours

after sunrise (Hisaoka and Firlit 1962), thus the

assessment of reproductive output took place between

0800 and 1000 h During the 4 days spawning period,

we cleaned the spawning boxes each morning,

assessed the occurrence of a spawning event and

counted the number of eggs spawned per female per

one spawning event For assessing the egg fertilization

probability, we distinguished fertilized eggs from

unfertilized eggs Zebrafish eggs are translucent, and

fertilized eggs can be easily identified by the presence of

a multi-cellular blastodisc, which is not present in

unfertilized eggs (Kimmel et al.1995) Only clutches

larger than 30 eggs were used in the egg fertilization

probability estimation to avoid inflated egg fertilization

probability estimates due to random egg mortalities in

very small clutch sizes

Egg traits

Egg trait measurements included the assessment of

egg size and egg mortality rate We measured egg size

as egg yolk diameter Yolk size can be a better

indicator of the egg quality than egg size (Kamler

2005) because perivitelline space is not contributing

substantially to the egg quality (Alderdice1988) The

eggs were photographed and the yolk diameter was

measured from the photographs under a profile projector

(Quick Scope; AT112-220 F; Mitutoyo;www.mitutoyo

co.jp) with an accuracy of 0.1 μm Eggs for size

measurements were selected from the first clutches

spawned and these eggs were only used for size

measurements, not for the subsequent analyses

Post-fertilization egg survival was estimated from

the first clutch females spawned Egg quality can

decrease in the course of spawning duration (Paull et

al 2008; Uusi-Heikkilä et al 2010) and therefore

eggs only from the first, and presumably highest

quality, clutch were collected From each female’s

first clutch, we transferred 15–48 fertilized eggs

(depending on the total amount of fertilized eggs

produced per couple) into a 24-well Multiwell Plates

(BD Falcon; nontreated polystyrene; Jacob et al

2007) so that one egg per well was incubated in

2 ml of tap water Consequently, eggs were not

influenced by other eggs or their contaminants andcould be treated as independent data points in thestatistical analyses After adding eggs to the plates wetransferred the plates into a rearing incubator (Tin-tometer GmbH, 44287 Dortmund, Germany, www.tintometer.de) at 27°C Water in the wells was notchanged during the incubation (Jacob et al 2007).The cell well plates were controlled during the next

48 h for the egg mortalities, which were estimated bycounting the number of dead eggs from each plate

Larval traits

Larval traits for the different sized parents wereassessed as larval age-at-hatch, larval length-at-hatch, and larval yolk-sac volume Larval traits,similarly as egg traits, were assessed from the firstclutches spawned Larvae were hatching in the cellwell plates and the number of larvae hatched wasrecorded each day during 7 days The standard length

of each newly hatched larva was measured under thedissection scope Larvae of age 4 dpf were used tocompare the larval length-at-hatch between the dif-ferent sized couples Measures based on larval length,however, may not be a reliable indicator of the quality

of the larva (Kamler 2008) Therefore, we graphed individual larva to measure larval yolk-sacvolume as an indicator of larval energy resources(Kamler 2008) We used the digitizing softwareImage Tool for Windows (version 3.0) to measurethe height and width of the yolk-sac from thephotographs The yolk-sac volume was estimatedusing the following formula (Chambers et al 1989):

photo-V¼ p=6LH2;where L represents the length (horizontal measure-ment; mm) and H the height (vertical measurement;mm) of the yolk-sac

Statistical analyses

Due to non-normally distributed and heteroscedasticdata, we used generalized linear models (GLM;Crawley 2007) to disentangle maternal and paternal-size effects on reproductive output and early life-history traits In all the analyses, female and male sizecategories and their interactions were treated as fixedeffects Due to the fact that zebrafish establish

potentially body-size based dominance hierarchies

(Pritchard 2001; Paull et al 2010) and the effect of

these hierarchies on zebrafish mating success and

reproductive output is largely unknown, we

addition-ally tested the effect of the relative size difference

between female and male on all the reproductive

traits The individual couple was set as a random

variable to account for the fact that other parental

traits than body size could contribute to the

differ-ences in reproductive success among couples (Spence

and Smith 2006; Hutter et al 2010) Spawning day

was also treated as a random variable when estimating

the effects of parental body size on variables

measured over the whole experimental period (i.e.,

spawning probability, clutch size and fertilization

probability) The amounts of variance associated to

the random variables were estimated through variance

components Couples which did not produce any eggs

during the four days spawning period were excluded

from the clutch-size analysis Count data, such as

clutch size and age-at-hatch, were modeled using

Poisson regression All probability data (i.e., spawning

probability, egg fertilization probability, egg survival

probability and hatching probability), were modeled

using binomial regressions In the analyses of larval

age-at-hatch, larval length-at-hatch and larval yolk-sac

volume, egg size could not be treated as a covariate

because the eggs measured were not the same eggs from

which the larvae hatched Using an average value of egg

size per couple as a covariate in these analyses was not

feasible due to the low number of observations per

couple for which both egg and larval traits were

measured Instead, we did a simple correlation analysis

(Pearson’s correlation) between the average egg size and

the average larval age-at-hatch, length-at-hatch and

yolk-sac volume If data was over-dispersed, the

quasi-Poisson or quasi-binomial distributions were used to

account for the overdispersion To estimate differences

among size categories we first fitted the full model and

then used the stepwise model reduction that in our case

referred to aggregating size categories, which had most

similar response variable values with each other

To summarize the effects of individual traits on

overall reproductive success we used spawning

probability, clutch size, egg fertilization probability,

egg survival probability and larval hatching probability

as components to estimate an integrative measure of

reproductive fitness (e.g., Mousseau and Roff 1987;

Danzmann et al 1989) The components (i.e., the

coefficients for each size combination predicted bythe model) were multiplied to obtain the expectednumber of hatched larvae, i.e., our fitness measurewas obtained by multiplying the model-based proba-bilities that an egg survives and hatches (as a product

of spawning probability, egg fertilization probabilityand hatching probability) further multiplied with thepredicted number of eggs for each size category ofeither males or females This measure describes theeffective offspring production as predicted by thestatistical models and is not to be confused with life-time fitness The final fitness values are given asrelative values where the values of different sizecombinations are standardized by the average valuefor the small female : small male size combination Inother words, this chosen reference value is used as avalue of 1 and all other values are relative to thisreference Our final integrative measure of reproduc-tive success described the expected number ofhatching larvae, which was considered a proxy offitness, as a function of female and male size,expressed relative to the small female : small malereproductive fitness

All data were considered statistically significant atP<0.05 All statistical analyses were performed with

R 2.11.1 with packages MASS and lme4 (R opment Core Team2009) Data are presented as meanvalues with standard errors (SE)

Devel-Results

Reproductive output

The spawning probability was not affected by femalebody size whereas male body size had a significanteffect on the female’s probability to spawn (Table1).Females from all size categories had a significantlylower probability to spawn with very large males(0.17±0.03) compared to mating with large (0.50±0.04), medium-sized (0.38 ±0.04) or small males(0.42±0.04; Table1) The interaction between femaleand male size and the relative size of males andfemales did not affect the probability to spawn(Table 1) Spawning day captured a relatively smallamount of variance (5.5%) not explained by theparental body size, while the individual couple wasresponsible for relatively large amount of variance(72.9%) in terms of spawning probability

Table 1 The effects of female, male, female × male body size and relative size difference between female and male on zebrafish reproductive traits Estimated values are given for the significant covariates, which are indicated in bold

Very large 1.378 (1.046) Male

Small (Intercept) 2.227 (0.685)

Very large 2.145 (1.138)

The number of eggs produced by zebrafish

correlated with female’s body size (Fig 1a,

Table 1) Very large females released significantly

more eggs compared to large, medium-sized and

small females (Table 1) Furthermore, females

re-leased on average significantly smaller clutches

(number of eggs produced over four spawning days)

when mated with very large males (49.2±13.1 eggsover four spawning days) compared to matings withlarge (72.5 ± 7.98), medium-sized (62.2 ± 7.88) orsmall (62.2±7.33) males (Fig 1a, Table 1) Neitherthe interaction nor the size difference between femaleand male body size did affect the number of eggsproduced (Table1) In terms of clutch size, 36.9% of

-value from the deletion of the variable from the full model

b P-values derived from the χ 2 –statistics

the variance not captured by parental body size was

associated to the individual couples and 7.8% to the

spawning day

The probability of egg fertilization was not

influenced by the parental body size or the female ×

male interaction (Table 1) After controlling for the

effect of body size, only 1.2% of the variance was

associated to the spawning days, whereas 51.8% of

the variance was associated to the individual couples

Egg traits

We found no difference in egg size (measured as egg

yolk diameter) among female size categories and no

significant female × male interaction (Table 1)

However, females released significantly smaller eggs

when crossed with medium-sized males (0.507 ±

0.002 mm) compared to the eggs released when

mated with small (0.519 ±0.002 mm), large (0.523 ±

0.001 mm) or very large males (0.515±0.002 mm;

Fig 1b, Table1) A large proportion of the variance

not explained by the parental body size (51.3%) wasassociated to the individual couples

In terms of egg survival probability there was asignificant interaction between female and male bodysize, but the pattern was not straightforward (Table1).Certain combinations, for instance small and largefemales mated with small males, large males matedwith either small or medium-sized females and verylarge males mated with very large females exhibitedlower egg survival probabilities (<90%) compared toother size combinations where egg survival probabilitiesexceeded 90% (Table2) 59.8% of the variance in eggsurvival probability was explained by characteristicsother than body size of the individual couple.Hatching probability was unaffected by femalebody size, but was affected by male body size.The average hatching probability of embryosfertilized by very large males (0.70 ± 0.02) wassignificantly higher than embryos fertilized bylarge (0.61 ±0.02), medium-sized (0.60 ± 0.02) orsmall males (0.49 ±0.02; Table 1) There was nointeraction effect between male and female size

20 40 60 80 100 120

0.49 0.50 0.51 0.52 0.53

3.5 4.0 4.5 5.0

3.0 3.1 3.2 3.3 3.4

0.010 0.015 0.020 0.025

Fig 1 a average clutch size

(number of eggs over four

spawning days), b average

egg size, c average larval

age-at-hatch, d average

larval length-at-hatch and e

average yolk-sac volume

produced by each male size

category across all female

size categories (filled

circles) The average value

of crosses between each

male and female size

cate-gory is indicated by the

open symbols Error bars

indicate standard errors

62.1% of the variance in hatching probability was

associated to the individual couples

Larval traits

Regardless of female body size, offspring sired by

very large (4.25±0.08 d) and large males (4.42±

0.06 d) hatched significantly earlier than offspring of

medium-sized (4.74±0.01 d) and small males (4.96±

0.01 d; Fig.1c, Table1) No variance in hatching time

was associated to the individual couples There was no

correlation between the average egg size and the average

larval age-at-hatch (df=31, r=−0.127, P=0.483)

Female body size was not a significant variable in

determining larval length-at-hatch (Table 1) Instead,

larvae which hatched from eggs fertilized by very

large (3.32 ± 0.02 mm) and large males (3.36 ±

0.01 mm) exhibited greater standard length than

larvae which hatched from eggs fertilized by

medium-sized (3.27±0.03 mm) and small males (3.26±0.03;

Fig.1d) When aggregating size categories, very large

and large males sired significantly larger offspring

compared to offspring of medium-sized and small

males (Table 1) In the larval length analysis, 21.5%

of the variance, not explained by parental body size,

was associated to the individual couples The average

larval length-at-hatch did not correlate with the

average egg size (df=31, r=0.180, P=0.316)

Female body size, but not male body size, had a

significant effect on larval yolk-sac volume Very large

females produced larvae with significantly greater

yolk-sac volume (0.021 ±0.001 mm3) relative to large

(0.016 ± 0.001 mm3), medium-sized (0.015 ±

0.001 mm3) and small females (0.013±0.001 mm3;

Fig 1e, Table 1) 25.1% of the variance in yolk-sac

volume was associated to the individual couples The

average yolk-sac volume did not correlate with theaverage egg size (df=25, r=−0.031, P=0.880).Integrative measure of reproductive fitness

The integrative measure of reproductive success (i.e.,reproductive fitness) varied with both female and malebody size from 0.2 to 4.0 relative to the reproductivevalue exhibited by small female: small male crossingschosen as the reference category (Fig.2) Large males,independent of the crossing, had the highest predictedfitness values, which was almost three times highercompared to the reference value of a small female :small male size combination Particularly the combi-nation of large female and large male yielded thehighest reproductive fitness value (4.0), which wasfour times higher than the reference value Interest-ingly, the average absolute reproductive fitness value

of very large females was somewhat lower (0.71)compared to the value of large females (1.13), but itwas still larger than the average absolute fitness value

of medium (0.61) and small females (0.46) across allmale sizes The very large males exhibited the lowestfitness values compared to all other male sizes

Discussion

Our study is the first to disentangle maternal and size effects on reproductive success in zebrafish Asexpected, female size contributed to reproductive outputand larval quality, and more unexpectedly male sizecontributed to a wide variety of reproductive parametersinvolving spawning frequency, clutch size, egg size,embryo development rate and larval size-at-hatch Theintegrated reproductive fitness measure showed that

paternal-Table 2 The average egg

sur-vival probabilities and their

standard errors for different

female and male size

combina-tions in zebrafish

N refers to the number of

indi-vidual eggs used in the egg

survival probability estimation

Small male Medium-sized

male

Large male Very large

male Small female 0.827 (±0.03) 0.989 (±0.01) 0.889 (±0.03) 0.979 (±0.01)

(N=150) (N=94) (N=162) (N=146) Medium-sized female 0.952 (±0.02) 0.945 (±0.02) 0.880 (±0.02) 0.990 (±0.01)

(N=188) (N=201) (N=251) (N=105) Large female 0.760 (±0.04) 0.956 (±0.02) 0.955 (±0.01) 0.963 (±0.03)

(N=104) (N=189) (N=294) (N=54) Very large female 0.944 (±0.02) 0.919 (±0.02) 0.946 (±0.01) 0.855 (±0.03)

(N=124) (N=136) (N=350) (N=173)

large, but not very large, fish exhibited the highest

reproductive success among both males and females, and

while large females were reproductively superior to

medium and small females, the very large males were

the least reproductively fit of all male sizes (Fig.2) Our

results altogether showed that male body size

contri-butes substantially to variation in several early

life-history traits in zebrafish, and, therefore, size-dependent

paternal effects might be more important for

reproduc-tive success in this species than previously believed

The positive relationship between female body size

and fecundity has been shown in several fish species

(Wootton 1998) and was also evident in our study,

similar to earlier reports in zebrafish (Spence and

Smith2006; Uusi-Heikkilä et al.2010) In addition to

egg number, egg size often correlates positively with

female body size across a range of fish species (e.g.,

Green2008; Marshall et al.2008) This is in contrast

with our results, which revealed that zebrafish egg

size varied independently of female body size

However, it has been previously indicated that egg

size measured as egg diameter may not be a

biologically relevant parameter for determining

zebra-fish reproductive success (Uusi-Heikkilä et al 2010)

and this may explain the lack of correlation between

egg size and female body size in the present study

The assumption is further supported by the lack of

correlation between egg size and a range of larvaltraits (e.g., length-at-hatch) in our study In fact, infish egg quality might be better reflected in embryodevelopmental rates or larval parameters than in eggsize (Kamler 2005, 2008) In our study, very largefemales did not produce larger eggs, but theyproduced higher quality larvae, in terms of yolk-sacvolume, compared to large, medium-sized or smallfemales This is consistent with the previous findings

of greater egg and larval qualities produced by largerfemales of many fish species (Marteinsdottir andSteinarsson 1998; Kennedy et al 2007), as it isknown that larvae with larger yolk-sacs may showincreased survival in the wild by being more resistant

to starvation under food-limited conditions (Miller et

al 1988; Kjørsvik et al.1990; Marshall et al 2010).Yet, despite the greater egg numbers and larger larvalqualities exhibited by the very large females in ourstudy, they showed a consistently lower integratedreproductive fitness value compared to large females,while still maintaining higher relative fitness com-pared to medium and small females This apparentlyinconsistent finding can be explained by the slightlylower model-predicted spawning probabilities, eggfertilization probabilities and hatching probabilities byeggs produced by very large females compared tolarge females The multiplied effects, although

Male size category

Fig 2 The estimated

integrated reproductive

fitness (i.e., the expected

number of hatched larvae)

for different female and

male size combinations in

zebrafish Lighter colour

corresponds to lower

esti-mated fitness value The

values are expressed as

relative to the reference size

category of small female :

small male=1 Colors in the

figure change smoothly and

the example colors in the

legend correspond to the

relative fitness values

individually not statistically significant (Table 1),

surmounted the significantly higher egg number

produced by the very large females, resulting in

slightly lower reproductive fitness values for very

large females Given the lack of trait-dependent

significant differences for female size for many traits

such as spawning and fertilization probabilities, one

should cautiously interpret this finding and not

prematurely discard the reproductive value of very

large females

We identified a range of pronounced paternal-size

effects on several reproductive traits such as spawning

probability, egg size, clutch size, embryo

develop-mental rate (i.e., hatching time), larval hatching

probability, and egg survival probability in zebrafish

Early hatching larvae have higher muscular activity

during the embryogenesis compared to late-hatching

larvae (Kimmel et al 1995) Thus, the higher

hatching probability and early hatching time of the

larvae produced by the large and very large males

could be an indicator of faster developmental rate and

better larval condition Previous studies have

demon-strated maternal-size effects on embryo

developmen-tal rate (Marteinsdottir and Steinarsson 1998;

Kennedy et al 2007), but the evidence for paternal

contributions to offspring development is limited

(Saillant et al 2001; Bang et al.2006) In our study,

larvae sired by very large and large males also

hatched significantly larger, in terms of standard

length, compared to larvae sired by medium-sized

and small males Larger body size at hatch may

increase larval fitness in the wild due to the greater

mouth gape and higher swimming activity, which

allows the larva to predate more efficiently and use

wider variety of prey (Miller et al.1988)

Considering that the contribution of sperm to

offspring development is mostly genetic, hypothesizing

on the nature of direct, non-genetic paternal effects on

early-life history traits is challenging It has been shown

that the amount and quality of sperm varies among

males and this variation can be size-dependent (Howard

et al 1998) In zebrafish, however, sperm quality,

quantity and motility have been shown to vary with

fish age and swimming activity level (Kemadjou

Njiwa et al 2004) but not with body size (Skinner

2004) This is partly supported by our results, which

showed no differences in egg fertilization probability

explained by male body size Therefore, we are not

convinced that size-dependent sperm quality is a

sufficient explanation for the pronounced dependent paternal effects we identified

size-Male size may also affect reproductive success in aless obvious indirect way through female mate choiceand differential allocation patterns (Skinner and Watt

2007b) Differential allocation patterns are

particular-ly likeparticular-ly when the experimental fish are allowed tospawn freely, as in our experiment, and eggs are notstriped and fertilized artificially Females benefit fromthe allocation of reproductive resources to betterquality mates as these partners may provide bettergenes or more resources to the offspring (Andersson

1994) For example, in zebrafish territorial males areknown to be larger (Spence and Smith 2005) andfemales allocating reproductive resources towardslarger, territorial males may benefit from betteroviposition sites Zebrafish females have indeed beenshown to prefer (Pyron 2003, but see Hutter et al

2010) large males, and they have also been found todifferentially allocate eggs towards larger males in asecond spawning when mated in short sequences witheither a large or a small male (Skinner and Watt

2007b) In our study, female zebrafish thus mighthave allocated higher quality eggs to larger malesbecause the more territorial (i.e., larger) males mayexhibit higher reproductive success, as is empiricallyshown to be true under low density conditions(Spence and Smith 2005; Spence et al 2006) Suchfemale preferences for large male body size would berevealed as a significant male-size effect in ourstatistical analysis In earlier studies the higherreproductive success by larger males has not beenconsistently evident (Spence and Smith 2005,2006;Spence et al 2006) However, differences in studyfindings on the importance of male size for reproduc-tive success should be viewed in terms of the malesize gradient used in the experiments For example,Spence and Smith (2006) did not find male size to berelated to reproductive success in zebrafish whileusing males ranging between 33.8 and 37.4 mm Inour study, we used males ranging from 24 to 31 mm.Potentially, the larger size gradient of males in ourstudy facilitated the emergence of clear paternal-sizeeffects on reproductive fitness, which may haveinvolved both direct (e.g., genetic quality, spermquality) and indirect (differential allocation byfemales) male-size effects

The relationship between male size and tive fitness in our zebrafish study was nonlinear In

reproduc-fact, we found that the very large males exhibited

consistently lower reproductive fitness compared to

large males (Fig.2) Unlike among females, the very

large males exhibited the lowest integrative

reproduc-tive fitness value of all male sizes Interestingly, very

large males sired high quality offspring once spawning

occurred, but they had substantially lower spawning

probabilities and they received significantly smaller

clutches compared to the other-sized males Because

this effect was not caused by the relative size difference

between females and males, it appears that the

advan-tages of very large body size are traded off against

unknown fitness costs of being too large We can only

speculate about the likely mechanisms, but

mating-related physiological or behavioral factors (e.g.,

court-ship behavior and sexual harassment; Partridge and

Fowler 1990) may play a paramount role In

Drosophila melanogaster male body size have been

shown to enhance male’s mating success but

simul-taneously to have a detrimental effect on female’s

fitness leading to a lower egg number received by the

large male (Fowler and Partridge 1989; Pitnick and

García-González2002) So far similar mechanisms of

sexual conflict have not been demonstrated in fish,

however it is possible that our experimental design

facilitated continuous sexual harassment of females

by very large males, which may have induced

substantial stress on females resulting in reduced

matings (Morgan et al.1999; Small2004) Thus, the

persistent and partly aggressive spawning behavior of

very large males, which could lead to a high mating

success first, may not be advantageous in repeated

spawnings and may introduce fitness costs for both

females and males Investigating the potential costs of

mating with very large males in zebrafish constitutes

an important avenue for further studies on

size-dependent sexual conflict in this species

Our experimental study controlled for density and

sex ratio, which both can affect zebrafish reproductive

success (Spence and Smith2005; Spence et al.2006)

and additionally allowed the fish spawn naturally

instead of using artificial fertilization Zebrafish has

been suggested to spawn in groups (Spence et al

2008) but a recent behavioural study showed that wild

zebrafish spawn in pairs rather than in groups (Hutter

et al 2010) Therefore, we believe that our

experi-ment allowed us to determine reliably the effect of

body size on reproductive success in zebrafish despite

the unnatural spawning conditions the fish were

exposed to Our experimental design allowed us tounravel some additional aspects related to zebrafishreproduction For example, stocking two fish in aspawning box helped us to reveal the relatively highamount of variation in reproductive output and early life-history traits that was associated to individual couplesindependent of male or female body size (sensu Paull et

al 2008) This additional variation in reproductivesuccess could be related to hormonal factors (van denHurk and Lambert 1983; van den Hurk et al 1987),genetic incompatibility (Gerlach and Lysiak2006), ordominance hierarchies (Pritchard 2001) Our studythus suggests that one should expect a strong effect ofbody size on reproductive performance but additionalfactors, potentially related to mate choice, are para-mount in explaining reproductive success in zebra-fish The great individual variance in reproductiveoutput has implications for experimental design ofstudies that investigate reproductive success in zebra-fish or use reproductive parameters in ecotoxico-logical studies as large sample sizes are needed toaccount for the large variability in individualreproductive performance (Paull et al 2008)

To conclude, our study is the first to unambiguouslyidentify the maternal and paternal-size dependent effects

on zebrafish reproductive success We revealed aninsofar overlooked importance of male body size forreproduction in this species and our findings also suggestthe importance of body size-dependent sexual conflictand female differential allocation as potential mechanismexplaining the pronounced male-size effects Implicatingbeyond our laboratory approach and assuming thatsimilar effects exist in other fish species, our findings

of paternal-size effects in addition to maternal-sizeeffects are worth being considered when deriving harvestregulations designed to protect exploited stocks Inparticular, our results suggest that ignoring the impor-tance of male body size for recruitment of fish mightconstitute a shortcoming when assessing the impact ofsize-selective fishing and skewed sex ratio on recruit-ment dynamics (Langangen et al.2011)

Acknowledgments Funding for this study was through the Adaptfish Project grant to RA and CW by the Gottfried- Wilhelm-Leibniz-Community ( www.adaptfish.igb-berlin.de ) and through the Academy of Finland for AK We thank Sarah Becker and Amanda O ’Toole for help in collecting the data and

in husbandry of zebrafish Additionally, we thank Thomas Mehner and the participants of the course on Scientific Writing

at IGB for helpful discussion on an earlier draft of this article,

and four reviewers for feedback The finalization of this study

was financially supported by the project Besatzfisch ( www.

besatz-fisch.de ) by the German Ministry for Education and

Research in the Program on Social-Ecological Research (grant

# 01UU0907, www.besatz-fisch.de ).

References

Alderdice DF (1988) Osmotic and ionic regulation in teleost

eggs and larvae In: Hoar WS, Randall DJ (eds) Fish

physiology Academic, London

Andersson M (1994) Sexual selection Princeton University Press

Bang A, Grønkjær P, Clemmsen C, Høie H (2006) Parental

effects on early life history traits of Atlantic herring (Clupea

harengus L.) larvae J Exp Mar Biol Ecol 334:51–63

Bernardo J (1996) Maternal effects in animal ecology Amer

Zool 36:83 –105

Burley N (1988) The differential-allocation hypothesis: an

experimental test Am Nat 132:611 –628

Chambers RC, Leggett WC (1996) Maternal influences on

variation in egg sizes in temperate marine fishes Amer

Zool 63:180 –196

Chambers RC, Leggett WC, Brown JA (1989) Egg size, female

effects, and the correlation between early life history traits

in capelin, Mallotus villosus: an appraisal at the individual

level Fish Bull 87:515 –523

Crawley RJ (2007) The R book Wiley, West Sussex

Danzmann RG, Ferguson MM, Allendorf FW (1989) Genetic

variability and components of fitness in hatchery strains of

rainbow trout J Fish Biol 35(supplA):313–319

Eaton RC, Farley RD (1974) Spawning cycle and egg

production of zebrafish, Brachydanio rerio, in the

laboratory Copeia 1974:195–204

Fowler K, Partridge L (1989) A cost of mating in female

fruitflies Nature 338:760 –761

Gage MJG, Macfarlane CP, Yeates S, Ward RG, Searle JP, Parker

GA (2004) Spermatozoa traits and sperm competition in

Atlantic salmon: relative sperm velocity is the primary

determinant of fertilization success Curr Biol 14:44 –47

Gerlach G, Lysiak N (2006) Kin recognition and inbreeding

avoidance in zebrafish, Danio rerio, is based on phenotype

matching Anim Behav 71:1371 –1377

Green BS (2008) Maternal effects in fish populations Adv Mar

Biol 54:1 –105

Heath DD, Fow CW, Heath JW (1999) Maternal effects on

offspring size: variation through early development of

Chinook salmon Evolution 53:1605–1611

Hisaoka KK, Firlit CF (1962) Ovarian cycle and egg

production in the zebrafish, Brachydanio rerio Copeia

1962:788–792

Howard RD, Martens RS, Innis SA, Drnevich JM, Hale J

(1998) Mate choice and mate competition influence male

body size in Japanese medaka Anim Behav 55:1151 –1163

Hutter S, Penn DJ, Magee S, Zala SM (2010) Reproductive

behaviour of wild zebrafish (Danio rerio) in large tanks.

Behaviour 147:641 –660

Jacob A, Nússle S, Britschgi A, Evanno G, Müller R,

Wedekind C (2007) Male dominance linked to size and

age but not to ‚good genes’ in brown trout (Salmo trutta) BMC Evol Biol 7:207

Kamler E (2005) Parent-egg-progeny relationships in teleost fishes:

an energetics perspective Rev Fish Biol Fish 15:399–421 Kamler E (2008) Resource allocation in yolk-feeding fish Rev Fish Biol Fish 57:789 –796

Kemadjou Njiwa JR, Müller P, Klein R (2004) Variations of sperm release in three batches of zebrafish J Fish Biol 64:475 –482

Kennedy J, Geffen AJ, Nash RDM (2007) Maternal influences

on egg and larval characteristics of plaice (Pleuronectes platessa) J Sea Res 58:65 –77

Kimmel CB, Ballard WW, Kimmel SR, Ullmann B, Schilling

TF (1995) Stages of embryonic development of the zebrafish Dev Dynam 18:143 –200

Kjørsvik E, Mangor-Jensen A, Holmefjord I (1990) Egg quality

in fishes Adv Mar Biol 26:71 –113 Kolm N (2001) Females produce larger eggs for large males in

a paternal mouthbrooding fish Proc R Soc B 268:2229– 2234

Langangen Ø, Edeline E, Ohlberger J, Winfield IJ, Fletcher JM, James JB, Stenseth NC, Vøllestad LA (2011) Six decades

of pike and perch population dynamics in Windermere Fish Res 109:131 –139

Marshall DJ, Allen RM, Crean AJ (2008) The ecological and evolutionary importance of maternal effects in the sea Ocean Mar Biol Ann Rev 46:203 –250

Marshall DJ, Heppell SS, Munch SB, Warner RR (2010) The relationship between maternal phenotype and offspring quality: do older mothers really produce the best off- spring? Ecology 91:2862 –2873

Marteinsdottir G, Steinarsson A (1998) Maternal influence on the size and viability of Icelandic cod Gadus morhua eggs and larvae J Fish Biol 52:1241 –1258

McLean JE, Bentzen P, Quinn TP (2004) Does size matter? Fitness-related factors in steelhead trout determined by genetic parentage assignment Ecology 85:2979–2985 Miller TJ, Crowder LB, Rice JA, Marshall EA (1988) Larval size and recruitment mechanisms in fishes: toward a conceptual framework Can J Fish Aquat Sci 45:1657–1670 Morgan MJ, Wilson CE, Crim LW (1999) The effect of stress

in reproduction in Atlantic cod J F Biol 54:477 –488 Mousseau TA, Roff DA (1987) Natural selection and the heritability of fitness components Heredity 59:181 –197 Mousseau TA, Fox CW (1998) The adaptive significance of maternal effects Trends Ecol Evolut 13:403 –407 Partridge L, Fowler K (1990) Non-mating costs of exposure to males in female Drosophila melanogaster J Insect Physiol 36:419 –425

Paull GC, Van Look KJW, Santos EM, Filby AL, Gray DM, Nash

JP, Tyler CR (2008) Variability in measures of reproductive success in laboratory-kept colonies of zebrafish and implica- tions for studies addressing population-level effects of environmental chemicals Aquatic Toxicology 87:115–126 Paull GC, Filby AL, Giddins HG, Coe TS, Hamilton PB, Tyler

CR (2010) Dominance hierarchies in zebrafish (Danio rerio) and their relationship with reproductive success Zebrafish 7:109 –117

Pitnick S, García-González F (2002) Harm to females increased with male body size in Drosophila melanogaster Proc R Soc B 269:1821 –1828

Priestley SM, Stevenson AE, Alexander LG (2006) The

influence of feeding frequency on growth and body

condition of the common goldfish (Carassius auratus) J

Nutr 136:1979S–1981S

Pritchard VL (2001) Behaviour and morphology of the

zebra-fish (Danio rerio) PhD Thesis, University of Leeds,

Leeds, UK

Pyron M (2003) Female preferences and male-male interactions

in zebrafish (Danio rerio) Can J Zool 81:122 –125

R Development Core Team (2009) R: A language and

environ-ment for statistical computing Vienna, Austria: R Foundation

for Statistical Computing See http://www.R-project.org

Roff DA (2002) Life-history evolution Sinauer Associates,

Sunderland

Sabat AM (1994) Mating success in brood-guarding male rock

bass, Ambloplites rupestris: the effect of body size Env

Biol Fish 39:411 –415

Saillant E, Chatain B, Fostier A, Przybyla C, Fauvel C (2001)

Parental influence on early development in the European

sea bass J Fish Biol 58:1585–1600

Schilling TF (2002) The morphology of larval and adult

zebrafish In: Nüsslein-Vollhard C, Dahm R (eds)

Zebrafish —A practical approach Oxford University

Press, Oxford

Skinner AMJ (2004) Sexual selection in the zebra fish (Danio

rerio) and the guppy (Poecilia reticulata) PhD Thesis,

University of Sheffield, Sheffield, UK

Skinner AMJ, Watt PJ (2007a) Phenotypic correlates of

spermatozoon quality in the guppy, Poecilia reticulata.

Behav Ecol 18:47 –52

Skinner AMJ, Watt PJ (2007b) Strategic egg allocation in the

zebra fish, Danio rerio Behav Ecol 18:905 –909

Small BC (2004) Effect of dietary cortisol administration on

growth and reproductive success on channel catfish J Fish

Biol 64:589 –596

Spence R, Smith C (2005) Male territoriality mediates density and sex ratio effects on oviposition in the zebrafish (Danio rerio) Anim Behav 69:1317–1323

Spence R, Smith C (2006) Mating preference of female zebrafish, Danio rerio, in relation to male dominance Behav Ecol 17:779 –783

Spence R, Jordan WC, Smith C (2006) Genetic analysis of male reproductive success in relation to density in the zebrafish, Danio rerio Front Zool 3:5

Spence R, Gerlach G, Lawrence C, Smith C (2008) The behaviour and ecology of the zebrafish, Danio rerio Biol Rev 83:13 –34

Stearns SC (1989) Trade-offs in life-history evolution Funct Ecol 3:259 –268

Thériault V, Moyer GR, Jackson LS, Blouin MS, Banks MA (2011) Reduced reproductive success of hatchery coho salmon in the wild: insights into most likely mechanisms Mol Ecol 20:1860 –1869

Uusi-Heikkilä S, Wolter C, Meinelt T, Arlinghaus R (2010) Size-dependent reproductive success of wild zebrafish Danio rerio in the laboratory J Fish Biol 77:552–569 Uusi-Heikkilä S, Kuparinen A, Wolter C, Meinelt T, O’Toole

AC, Arlinghaus R (2011) Experimental assessment of the probabilistic maturation reaction norm: condition matters Proc R Soc B 278:709 –717

van den Hurk R, Lambert JGD (1983) Ovarian steroid glucuronides function as sex pheromones for male zebrafish, Brachydanio rerio Can J Zool 61:2381 –2387

van den Hurk R, Schoonen WGEJ, van Zoelen GA, Lambert JGD (1987) The biosynthesis of steroid glucuronides in the testis of the zebrafish, Brachydanio rerio, and their pheromonal function as ovulation inducers Gen Comp Endocrinol 68:179 –188

Wootton RJ (1998) Ecology of teleost fishes, 2nd edn Chapman & Hall, London

Movements of lumpsucker females in a northern Norwegian

fjord during the spawning season

Hiromichi Mitamura&Eva B Thorstad&

Ingebrigt Uglem&Pål Arne Bjørn&Finn Økland&

Tor F Næsje&Tim Dempster&Nobuaki Arai

Received: 17 February 2011 / Accepted: 13 October 2011 / Published online: 30 October 2011

# Springer Science+Business Media B.V 2011

Abstract The lumpsucker Cyclopterus lumpus is

distributed throughout the North Atlantic Ocean and

migrates considerable distances between offshore

feeding areas and shallow inshore spawning grounds

The number of the lumpsucker has declined since the

mid 1980s, probably as a result of overexploitation

The lumpsucker is the preferred host of the sea louse

Caligus elongates, which is a problem for marine

aquaculture However, little is known about the

biology of the lumpsucker The aims of the study

were to 1) examine the movements of female

lumpsucker during the spawning migration, and 2)

assess the potential for lumpsucker to act as a vectorfor transmission of parasites and diseases betweenaquaculture farms and wild fish Twenty femalelumpsuckers tagged with acoustic transmitters werereleased during the spawning season in the inner part

of Øksfjord, northern Norway and their distributionwas recorded by 22 automatic acoustic receivers Theaverage time until departure from the fjord was 3 days,and within 1 week all fish had left the fjord Timing

of departure from the fjord was unrelated with eithertidal current patterns or the time of the day A highproportion of the fish (75%) were recorded within

200 m of fish farms, but they did not stay forextended periods at these farms Our results suggestthat mature female lumpsucker exhibit a movementpattern characterized by rapid fjord-scale migrationsduring the spawning season, and that they are notattracted to salmon farms in the same way as a range

of other fish species

Keywords Cyclopterus lumpus Lumpfish Management Øksfjord Spawning migration Telemetry

Introduction

The lumpsucker Cyclopterus lumpus is distributedthroughout the North Atlantic Ocean, and migratesconsiderable distances between offshore feeding areasand shallow inshore spawning areas (Davenport

Norwegian Institute of Fisheries and Aquaculture Research,

NO-9192 Tromsø, Norway

T Dempster

SINTEF Fisheries and Aquaculture,

NO-7465 Trondheim, Norway

T Dempster

Department of Zoology, University of Melbourne,

Melbourne, Victoria 3010, Australia

1985) Female lumpsucker are fished commercially

for their roe, which is used for caviar substitute

Spawning occurs in spring and early summer

(Davenport 1985), and commercial fishing takes

place in inshore waters during the spawning

migra-tion The number of the lumpsucker has declined

from the mid 1980s, probably as a result of

over-exploitation (Davenport 1985; Goulet et al 1986;

Sunnanå 2007) Despite the economic importance of

this species, parts of their life cycle are not well

known, and little is known about their spawning

migrations Knowledge of migration patterns during

the reproductive season is important for sustainable

management of the lumpsucker fishery

Aquaculture is one of the most important human

impact factors in Norwegian marine ecosystem

Community compositions of wild fish around coastal

salmon farms have been described and large-scale

attraction of wild fish to farms demonstrated (Bjordal

and Skar 1992; Dempster et al 2009) Salmon

aquaculture may affect the organisms living around

the farms, for example by increasing the risk of

parasite and disease transfer between cultured

organ-isms and the wild fauna An ectoparasitic copepod,

the sea louse Caligus elongatus, is found on more

than 80 different fish species (Kabata 1979), and

lumpsuckers are among the preferred hosts of C

elongatus (Øines et al.2006; Heuch et al.2007; Øines

and Heuch 2007) C elongatus is recognized as a

problem for the Atlantic salmon Salmo salar farming

industry (Pike and Wadsworth1999; McKenzie et al

2004) Thus, information on the movements of

lumpsucker in relation to fish farms is crucial for

assessing their potential for transferring C elongatus

between fish farms or between wild and farmed fish

The aims of this study were to 1) examine the

movement patterns of female lumpsucker during the

spawning period, and 2) assess the potential for

lumpsucker to act as a vector for transmission of

parasites and diseases between aquaculture farms and

wild fish

Materials and methods

The study was carried out in Øksfjord, Norway

(Fig 1) Four salmon farm sites were located in the

fjord, but only three contained fish during the study

In addition, the fjord contained an Atlantic cod farm

Twenty female lumpsuckers (body mass: 3.2 kg±0.9)were captured at four different sites in Øksfjordduring 7–10 July, 2006 with benthic gill nets(Fig 1, Table 1) Immediately after capture, the fishwere transported in an 800 l tank with circulatingwater onboard of a boat to a recovery pen at therelease site (Fig 1) The fish were kept in therecovery pen (circumference: 20 m, depth 5 m) untiltagging on 10 July, 2006 The fish were tagged withcoded acoustic transmitters (VEMCO Ltd., Canada,model V13-1H-69KHz-S256, Min/max off times 40/

120 s, battery life 244 days, diameter of 13 mm,length of 36 mm, mass in water of 6 g) Thetransmitter mass in water ranged between 0.1 and0.4% of the fish body mass Surgery was carried outunder anesthesia induced with 2-phenoxy-ethanol(Sigma Chemical Co., St Louis, MO, U.S.A., 1 mlper 1 l seawater, immersion period 1 min 39 s±14 s)

An incision (~ 1.5 cm) was made on the ventralsurface posterior to the pelvic girdle of the fish,and the transmitter was inserted into the peritonealcavity The incision was closed using two or threeindependent silk sutures (1/0 Ethicon) The im-plantation procedure lasted approximately 2 min.After tagging, the fish were allowed to recover for

3 days in the recovery pen before they werereleased in the inner part of the fjord on 13 July

2006 (Fig 1) Three days is sufficient for ical recovery and to resume normal activity levelsafter handling and tagging in a range of other fishes(e.g Peake et al 1997; Thoreau and Baras 1997).Studies of pike Esox lucius, pikeperch Stizostedionlucioperca and barbel Barbus barbus have alsodemonstrated that transmitters can be implantedshortly before spawning (~ 48 h), without affectingmigrations and spawning behavior significantly(Jepsen et al 2002) The survival after tagging was100% We are confident that any distress to the fishwas minimal and that all of the tagged fish survivedfor throughout the study period

physiolog-To minimize stress during surgery, the tive status of the females were not examined per se,but ripe eggs were observed during the surgicalprocedure for all of the females, although eggs can bedifficult to see through the small incision whenimplanting transmitters Furthermore, dissection of

reproduc-19 other females within the same size range captured

at the same time and locations confirmed that 18 ofthese had a large mass of ripe eggs, and that the ripe

egg of the remaining female was considered to be

spent The mass/length relationship indicated that all

of the 20 tagged females were considered to have

ripe eggs All of the 39 females did not have sea

louse

The movements and distribution of the fish were

recorded by eight arrays composed of 22 automatic

receivers (receiver model VR2) deployed in the fjord

system (Fig 1) The receivers recorded the unique

identification code of each tag, and date and time of

detection The depth at the sites where the receivers

were deployed varied between 23 and 300 m

Typically, the receivers were attached to the ropes at

approximately half the distance to the bottom, apart

from the receivers deployed at the fish farms (Fig.1),

which were suspended 20 m below the water surface

Range tests showed that the minimum detection range

varied between 500–700 m in radius Modified

receivers with shorter detection ranges (150–200 m

radius; Uglem et al.2008) were deployed at the fish

farms to record possible farm attraction of the tagged

fish Residence periods around fish farms are

calculated If the time lapse between consecutive datarecorded in the vicinity of each fish farm was lessthan 60 min, we estimated that the fish continuouslyspent its time around the fish farm Distances betweenreceiver arrays were measured along a mid-fjordtransect The distances moved should thus beregarded as minimum distances, because fish areunlikely to have taken the shortest route between thereceiver arrays Net movement speed throughout thefjord or in the inner part of the fjord was calculated.The speed in the inner part of fjord from the releasesite to the array 4 was calculated for fish which weredetected by the array 1–3 The distance covered foreach fish was divided by time between release andfirst detection of the outermost array 8 for or the array

4 The data were downloaded from the receivers on 3and 4 October 2006 Tidal data from the NorwegianHydrographic Service (www.vannstand.statkart.no/main.php) were used All values are given as mean ±standard deviation

Results

All lumpsuckers departed from the study area within

7 days after release (mean: 3 days, range: 1–7, Figs.2,

3, Table1), and they did not return to Øksfjord withinthe nearly 3-month study period Altogether ninetagged fish were detected in the inner part of the fjord(arrays 1 to 3), before they moved out of the fjord(Table1, Fig.3) The net movement speed throughoutthe fjord for fish detected in the inner part of the fjord(arrays 1 to 3) was not significant different from thatfor fish which was not detected (T-test, t=0.26, df=

18, P= 0.80) The minimum movement distance,which was based on cumulative distance movedbetween arrays after release until departure from thefjord, was 22.9±6.8 km (range: 17–39 km) Thiscorresponds to an average movement speed relative tothe ground of 0.72±0.42 km hour−1(Table 1) Therewas no significant association between net movementspeed from the release site to receiver array 8 furthestout in the fjord (Fig 1) and body length (linearregression analysis, t = 0.76, P= 0.46) or capturelocation (Kruskal Wallis test, X2=0.98, df=3, P=0.81) All 20 fish showed directional and rapidmovement to the outermost receiver array, oncethey had left array 6 (Fig 3) The fish did not driftpassively out of the fjord with the tidal current, as 12

Fig 1 Øksfjord in Finnmark County, northern Norway, with

the four locations where lumpsucker were captured (Δ A, B, C

and D), and the release site of all tagged fish (■) Automatic

receivers were arranged in arrays across the fjord, which are

labeled 1 to 8 ( ● indicate receivers with a normal range,

whereas o indicate receivers with reduced range deployed at

fish farms) In addition one normal range receiver was located

at the release site There was one Atlantic cod farm in the fjord,

close to capture site C, and four Atlantic salmon farm sites, but

only three of these were in use during the study as the farm

close to the release site was not in use

tagged fish (60%) showed directional movements out

of the fjord during periods with incoming tide The

proportion of fish departing array 6 did not differ

between the period of outgoing (8 fish, 40%) and

incoming tide (12 fish, 60%) (Goodness of fit test,

χ2=0.8, p=0.37), or between day and night (day:

06:00–18:00, n=9, night: 18:00–06:00, n=11,

Good-ness of fit test,χ2=0.2, p=0.65)

During the outward migration, 15 tagged fish

(75%) were detected in the close vicinity of fish

farms (within the 150–200 m detection distance of the

range-restricted receivers), but all left the fjord

without staying for long periods around farms

Residence periods in the vicinity of fish farms were

short, averaging 64±51 min (range: 11–182 min)

Fourteen fish were detected at the Atlantic salmon

farms (61±51 min, range: 5–346 min), and 3 fish

were detected at the cod farm (56±54 min, range: 23–

118 min) On average, each fish was recorded at a fishfarm 2.3 times, giving a total of 35 recordings atfarms Nine fish were detected at more than one fishfarm Fifteen tagged fish (75%) were detected in theclose vicinity of the fish farm which did not containfish Residence period was on average 42±33 min(range: 2–168 min) On average, each fish wasrecorded at a fish farm 1.5 times, giving a total of

22 recordings The residence period at the fish farmwhich did not contain fish was not significantdifferent from that at fish farm containing fish(Mann-Whitney U test, U=81.5, p=0.20)

Speed (km/h)

in the fjord

Speed (km/h) in the inner fjord

Duration monitored (h)

A, B, C and D refer to sites shown in Fig 1

In or Out means that individual was or was not detected in the inner part of the fjord (array 1 to 3).

Speed in the fjord or in the inner fjord means that net movement speed of an individual between the release site and the array 8 or an individual which showed the movement pattern “In” between the release site and the array 4

spawning season Observations of spent females

disappearing from fisheries catches in the Gulf of

White Sea in Russia (Mochek 1973) are consistent

with our findings that females departed from the fjordimmediately or within 7 days after release Mochek(1973) noted that spent females disappeared from thecatches after 2–3 days Moreover, an Icelandictagging project showed that lumpsucker tagged in aspawning area and subsequently recaptured in thesame locality were either recaptured within the first

3 weeks after tagging, or the next spawning season 1year later (Schopka 1974) Hence, spawning ofindividual lumpsucker females in local spawningareas appears to occur within one to 3 weeks, afterwhich they leave the spawning ground Our resultssuggest that the tagged females either completed theirspawning within Øksfjord and then migrated back tothe ocean, or that they visited other spawning areas ofadjacent fjords There are several neighbouring fjords

to the study area, where we had no acoustic receivers,which the tagged females may have visited Lump-sucker are highly migratory, moving between theocean feeding grounds and inshore spawning areas,and females may easily move between neighbouringfjord systems in search of males A reproductivestrategy involving that females spawn at severallocations scattered over a large geographical rangemay be adaptive if the environmental conditions varyconsiderably among various spawning locations andthat this also affects the fitness of individual broods.This supposition is supported by the fact thatlumpsucker males guard a nest with eggs, which can

be the result of single or multiple female spawnings,while females leave the nests immediately afterspawning (Goulet et al.1986)

The female lumpsucker in this study swam activelyduring both day and night, and often did so againstthe tidal current For many species, such as plaicePleuronectes platessa, sole Solea solea and Atlanticcod, selective tidal stream transport is used as amigratory mechanism (McCleave and Arnold 1999;Forward and Tankersley2001) Selective tidal streamtransport can lead to considerable reduction in theenergy necessary for horizontal movement How-ever, there was no indication that female lump-sucker passively drifted with the currents, or thatthey used the tidal current actively to move out ofthe fjord As there is likely to be local circularcurrents in a fjord system, the females might makefurther effort to get away from such retentioncurrents in Øksfjord The fish were, on average,moving a minimum of 0.72 km hour−1 relative to

Fig 3 Typical movement patterns of tagged lumpsucker

females a An individual with a directional movement to the

outermost array b An individual that moved into the inner part

of the fjord after release “R” represents the release location,

with receiver arrays 1 –3 located further in the fjord than the

release site and receiver arrays no 4 –8 located outside

Fig 2 Proportions of tagged lumpsucker detected in the

different receiver arrays after release in Øksfjord in 2006 “R”

represents the release location, with receiver arrays 1 –3 located

further in the fjord than the release site and receiver arrays no.

4 –8 located outside The smallest bubble represents one fish

(5%), while the largest represent 20 fish (100%) The bubbles

within each day do not sum up to 100% as one fish could have

been detected in several zones within the same day

the ground in this study Such extensive movements

suggest that they are particularly vulnerable to capture

by the passive gears (benthic gill nets) used in the

inshore commercial fishery during the reproductive

season

A basic assumption for horizontal transfer of

pathogens from one population of fish to another is

that they interact A high proportion (75%) of the

lumpsucker females were detected within 150–200 m

of fish farms and they could therefore theoretically be

regarded as potential vectors of diseases or parasites

originating from the farms If and to what extent

lumpsuckers transfer diseases or parasites depend on

that they actually share pathogens with the farmed

salmon, which in turn also are transferred among the

different species The female lumpsuckers were

present at the farms only for hours and it is therefore

unlikely that they on an individual level act as major

vectors of pathogens between fish farms and wild

fish This assumption does, however, depend on the

number of lumpsuckers present in a fjord with fish

farms If this number is high there might still be an

increased risk of disease transfer despite short

residence times at the farms Moreover, we

hypothe-size that the periods the female lumpsucker were

observed close to fish farms does not represent a

specific attraction to farms, as it thus would be

reasonable to expect longer and repeated visits to

farms Other fish species that are attracted to fish

farms, such as Atlantic cod and saithe Pollachius

virens, may be resident for many months (Uglem et

al 2008,2009), probably due to the constant supply

of waste fish food (Dempster et al 2009) Only two

out of 24 saithe (size: 40–65 cm) tagged in the same

way as the lumpsucker and released simultaneously at

nearby locations, left the fjord during the 3 month

study period, while the remaining 22 saithe were

observed regularly in the vicinity of salmon farms

(Uglem et al 2009) Although it is unknown how

long it takes for pathogens to transfer from fish to

another around a fish farm, the pathogen-transfer

possibility would be much higher for

farm-associated fish (e.g saithe) than unfarm-associated fish

(e.g female lumpsucker) Since lumpsuckers do

not feed during spawning migrations (Collins

1976; Davenport1985; Davenport and Thorsteinsson

1989), it is unlikely that they are attracted by the

uneaten food and concentration of other prey items

around the farms

Acknowledgments The study benefited from using the infrastructure, including automatic listening stations, of a concurrent study of Atlantic cod in Øksfjord, Finnmark, funded

by The Norwegian Seafood Federation Research Foundation This study was funded by the Norwegian Institute for Nature Research (NINA), the Norwegian Research Council through the strategic research project EcoMA, and by Grant-in-Aid for JSPS Fellows (18 –2409) Japan We would like to thank Rune Nilsen, Tom Andreassen, Pablo Sanchez-Jerez, Anders Økland, Arvid Fredriksen and Trygve Solvar Johansen for assistance during the field work.

References

Bjordal Å, Skar AB (1992) Tagging of saithe (Pollachius virens L.) at a Norwegian fish farm: preliminary results on migration ICES Council Meeting Papers 1992/G:35 Collins MAJ (1976) The lumpfish (Cyclopterus lumpus L.) in Newfoundland waters Can Field-Nat 90:64 –67

Davenport J (1985) Synopsis of biological data on the lumpsucker Cyclopterus lumpus (Linnaeus, 1958) FAO Fisheries Synopsis no 147, 31 pp

Davenport J, Thorsteinsson V (1989) Observations on the colours of lumpsuckers, Cyclopterus lumpus L J Fish Biol 35:829 –838

Dempster T, Uglem I, Sanchez-Jerez P, Fernandez-Jover D, Bayle-Sempere J, Nilsen R, Bjørn PA (2009) Coastal salmon farms attract large and persistent aggregations of wild fish: an ecosystem effect Mar Ecol Prog Ser 385:1– 14

Forward RB, Tankersley RA (2001) Selective tidal-stream transport of marine animals Oceanogr Mar Biol Annual Rev 39:305 –353

Goulet D, Green JM, Shears TH (1986) Courtship, spawning, and parental care behaviour of the lumpfish, Cyclopterus lumpus L., in Newfoundland Can J Zool 64:1320 –1325 Heuch PA, Øines Ø, Knutsen JA, Schram TA (2007) Infection

of wild fishes by the parasitic copepod Caligus elongates

on the south east coast of Norway Dis Aquat Org 77:149 – 158

Jepsen N, Thorstad EB, Baras E, Koed A (2002) Surgical implantation of telemetry transmitters in fish: how much have we learned? Hydrobiologia 483:239 –248

Kabata Z (1979) Parasitic Copepoda of British fishes The Ray Society, London

McCleave JD, Arnold GP (1999) Movements of yellow- and silver-phase European eels (Anguilla anguilla L.) tracked

in the western North Sea ICES J Mar Sci 56:510–536 McKenzie E, Gettinby G, McCart K, Revie CW (2004) Time- series models of sea lice Caligus elongatus (Nordmann) abundance on Atlantic salmon Salmo salar L in Loch Sunart, Scotland Aquacul Res 35:764 –772

Mochek AD (1973) Spawning behaviour of the lumpsucker (Cyclopterus lumpus L.) J Ichtyol 13:615 –619

Øines Ø, Heuch PA (2007) Caligus elongatus Nordmann genotypes on wild and farmed fish J Fish Dis 30:81 –91 Øines Ø, Simonsen JH, Knutsen JA, Heuch PA (2006) Host preference of adult Caligus elongatus Nordmann in the

laboratory and its implications for Atlantic cod

aquacul-ture J Fish Dis 29:167–174

Peake S, McKinley RS, Scruton DA, Moccia R (1997)

Influence of transmitter attachment procedures on

swim-ming performance of wild and hatchery-reared Atlantic

salmon smolts Trans Am Fish Soc 126:707 –714

Pike AW, Wadsworth SL (1999) Sealice on salmonids; their

biology and control Adv Parasitol 44:233 –337

Schopka SA (1974) Preliminary results from tagging of

lumpsucker (Cyclopterus lumpus) in Icelandic waters

Uglem I, Dempster T, Bjørn P-A, Sanchez-Jerez P, Økland F (2009) High connectivity of salmon farms revealed by aggregation, residence and repeated migrations of wild fish among farms Mar Ecol Prog Ser 384:251 –260

A new blind loach, Oreonectes elongatus sp nov.

(Cypriniformes: Balitoridae) from Guangxi, China

Li Tang&Yahui Zhao&Chunguang Zhang

Received: 5 May 2011 / Accepted: 13 October 2011 / Published online: 4 November 2011

# Springer Science+Business Media B.V 2011

Abstract A new loach, Oreonectes elongatus sp

nov is described based on collections from Mulun

Township, Huanjiang County, Guangxi in China It is

distinguished from its congeners by the combination

of the following characters: most elongate body (body

depth/SL 8.62–10.68%), blind, a forked caudal fin,

obvious adipose dorsal crest and ventral crest; entire

body naked and de-pigmented Although the new

species has a similar distribution with O macrolepis,

it can be distinguished by scales (absent in O

elongatus vs present in O macrolepis), shape of

snout (elongate vs round), the opposite position of

the dorsal and pelvic fins origins (behind vs front)

The new species shares the same possession of dorsal

and ventral crests, a forked caudal fin, eyeless, naked

body and incomplete lateral line with O translucens,

but can be distinguished from the latter by caudal fin

crest (more developed and translucent in O

trans-lucens), longer anterior nostril tube and barbel,extreme of pectoral fin reaching 2/3 of the distancebetween origin of pectoral and pelvic fins, morevertebrae (4+38–39 vs 4+32)

Keywords Oreonectes elongatus sp.nov .Balitoridae New species Cavefish China

Introduction

There are about a hundred hypogean fishesdistributed in China, which is nearly one-third ofall the described hypogean fish species in theworld (Romero et al 2009) with new speciesdescribed each year The genus Oreonectes is adistinct group in hypogean fishes with almost allspecies in this genus having cave-dwelling behavior.Oreonectes are small loaches distributed in SouthChina and North Vietnam (Zhu1989; Kottelat2001;

Du et al 2008).The species of Oreonectes share thefollowing characters: head depressed, anterior andposterior nostrils separated, anterior nostril in anelongate tube with a barbel at the end, and air bladderenclosed in a paired bony capsule (Zhu 1989).Günther first described the genus in 1868 with O.platycephalus as the type species found in HongKong, China In 1981, the second species O.anophthalmus was described by Zheng (1981) based

DOI 10.1007/s10641-011-9943-7

L Tang:Y Zhao ( *):C Zhang ( *)

Key Laboratory of the Zoological Systematics and Evolution,

Institute of Zoology, Chinese Academy of Sciences,

1 Beichen West Road, Chaoyang District,

Beijing 100101, China

e-mail: zhaoyh@ioz.ac.cn

e-mail: fish@ioz.ac.cn

L Tang

Graduate School, Chinese Academy of Sciences,

Yuquan Road, Shijingshan District, Beijing 100049, China

on some eyeless specimens collected from a cave in the

Qifeng Mountain, Wuming County, Guangxi Zhuang

Autonomous Region of China Since then, several

species have been discovered and reported

intermit-tently, including: O furcocaudalis Zhu and Cao1987,

O retrodorsalis Lan et al 1995, O translucens

Zhang et al 2006, O microphthalmus Du et al

2008, O polystigmus Du et al 2008, O macrolepis

Huang et al 2009, O luochengensis Yang et al

2011a, and O guananensis Yang et al.2011b Among

the ten species, O platycephalus is the only one

having a broad distribution including parts of

Guangxi and Guangdong in China (Zhu1989; Du et

al.2008; Huang et al.2009) as well as Quang Ninh

in North Vietnam (Kottelat 2001), and is often

found in open streams Interestingly, the rest species

are all definitely cave dwellers and endemic to

Guangxi with narrow distributions; they mainly occur

in the Xijiang River, the largest tributary of the Pearl

River in Southwest China (Günther1868; Zhu1989;

Du et al.2008)

In 2009, three specimens belonging to a blind

Oreonectes species were collected from the

under-ground waters of Mulun Township, Huanjiang County,

Guangxi in China The specimens showed several

distinguishing characters compared to congeners, and

subsequent examination enables us to recognize them as

a new species

Materials and methods

Type specimens of the new species, Oreonectes

translucens, O microphthalmus and O macrolepis

as well as the specimens of O furcocaudalis used to

compare in the paper belong to the collection of the

Institute of Zoology, Chinese Academy of Sciences,

Beijing (ASIZB) and Kunming Institute of Zoology,

Chinese Academy of Sciences, Kunming (KIZ)

(institutional abbreviation according to Leviton et al

1985) All these specimens were preserved in 75%

alcohol Detailed information on the specimens is

given in the“Comparative materials” below

Counts and measurements were taken on the left

side of the fish body as shown in Fig 1 All

measurements were taken point to point with digital

calipers to 0.1 mm Vertebrae were counted from

radiographs which were made with a Kodak

DXS-4000 X-ray unit

Results

Oreonectes elongatus sp nov (Figs.2 and3)

Holotype ASIZB 189288, 78.2 mm standard length(SL), from Donglei Cave in Mulun Township,Huanjiang County, Guangxi Zhuang AutonomousRegion, China, collected in November 2009 by Mr.Wei Longtao et al

Paratypes ASIZB 189289, 78.1 mm SL, deposited inASIZB SCAU20100106, 76.4 mm SL, deposited inSouth China Agricultural University, collected inanother cave—Shangzhaida Cave in Mulun township.Other data for the two paratypes are the same as forthe holotype

Diagnosis Oreonectes elongatus can be distinguishedfrom its congeners by the combination of thefollowing characters: the most elongate body (bodydepth/SL 8.62–10.68% vs 11.70–16.55% of itscongeners), eyeless, caudal fin forked, well-developed adipose crests on caudal peduncle; bodynaked and de-pigmented The new species is similar

to O translucens by lacking eyes and having a forkedcaudal fin, but differs from the latter by theappearance of anterior nostril (long nostril tube andbarbel of the anterior nostril, 18.6–26.3% HL) andcaudal crests (adipose and not transparent) as well asmore counts of vertebrae (4+38–39 vs 4+32).Description General body features are shown inFig 2 The meristic and morphometric data for typespecimens of the new species are given in Table1 Bodyvery elongate and cylindrical, posterior portion slowlycompressed from the end of the dorsal fin to the base

of the caudal fin Dorsal profile slightly convex fromsnout to dorsal fin insertion point, ventral profilenearly straight

Head slightly depressed and flattened, widthgreater than depth, gradually narrow from opercleforward to the tip of the snout, Snout long andelongate, anterior edge obtuse Eyes absent Anteriorand posterior nostrils are well separated with a shortdistance Anterior nostril in a short tube, whichextends into a relatively long barbel Mouth inferiorand curved Upper and lower lips smooth, a markedmedian groove in the middle of the lower lip Rostraland maxillary barbels very long, all extending over

posterior nostrils, outer rostral barbels’ length over 1/2

of head length (average 57.9±2.9% HL)

Dorsal fin serrated along posterior edge, withthe length nearly a full head length; dorsal-finorigin nearer to caudal-fin base than to snout tip,and posterior to pelvic-fin origin Anal fin originnext to anus, tip nearly reaching middle of caudalpeduncle Pectoral fin long and narrow, greaterthan 1/2 the distance between origins of pectoraland pelvic fins Pelvic fin relatively slender,extending slightly over anus Caudal fin forked.Caudal peduncle long and compressed laterally,

Fig 1 Principal measurements taken on Oreonectes species.

Oreonectes furcocaudalis is treated as a typical example All

measurements are taken on the left side of the fish specimens.

Standard length (A –K), from the tip of the upper jaw to the

position of the last half-centrum; body height at dorsal fin

origin (a) and maximum body height (b); body width at dorsal

fin origin (c) and at anus (x), maximum body width (d);

predorsal length (A –B), prepelvic length (A–F), preanal length

(A –I), prepectoral length (A–D), and pre-anus length (A–H),

from the tip of the upper jaw to the insertion of the dorsal fin,

pelvic fin, anal fin, pectoral fin and anus respectively; length of

dorsal fin (e), length of pelvic fin (f), length of anal fin (g),

length of pectoral fin (h), and length of caudal fin (i), from the

insertion of each fin to the tip of the longest ray; length of

dorsal-fin base (B –C), length of pelvic-fin base(F–G), length of

anal-fin base (I –J), length of pectoral-fin base (D–E), from the

anterior end to the posterior end of each fin base; length of

caudal peduncle (J–K), from the posterior end of anal-fin base

to last half-centrum; depth of caudal peduncle (j), the depth at

the most slender point of caudal peduncle; head length (k),

from the tip of the upper jaw to the posterior margin of operculum; head height (l), from the point between head and body vertically to the ventral midline; head width (m), the width at the nape; length of snout (n), from the tip of the upper jaw to the anterior margin of the eyes; eye diameter (o), the diameter of the circumorbital series; interorbital width (p), the shortest distance between the orbits; length of the anterior nostril tube and barbel (q), from the base of the anterior nostril tube to the tip of the nostril barbel; preanterior nostril length (r), the distance between the tip of the upper jaw and the base of anterior nostrils; distance between anterior and posterior nostrils (s), the shortest length from the posterior margin of the anterior nostril to the anterior margin of the posterior nostril; distance between posterior nostril and eye (t), the shortest distance between the posterior margin of posterior nostril and the anterior margin of the eye; distance between posterior nostrils (u), the shortest distance between posterior nostrils; length of outer rostral barbel (v) and length of maxillary barbel (w), the longest length of outer rostral barbels and maxillary barbels

Fig 2 Lateral view of holotype of Oreonectes elongatus sp.

nov (ASIZB 189288, 78.2 mm SL)

with adipose crests along both dorsal and ventral

sides Dorsal crest origin posterior to anal-fin

insertion, ventral crest origin a short distance from

the end of anal fin base; both crests connecting to

caudal-fin rays Body naked Lateral line

incom-plete, with 4 pores behind opercle, connecting to

the cephalic lateral-line system

Coloration Overall body pale, yellowish without any

special marks in alcohol, all fins including caudal

crests pale

Distribution The new species is only distributed in a

few caves from Mulun Township, Huanjiang County,

Guangxi Zhuang Autonomous Region, China (Fig.4)

The subterranean waters belong to Dahuangjiang

River drainage, Pearl River system

Etymology The species name derives from the Latin

elongatus meaning elongated

Discussion

Remarks

The first comprehensive review of Oreonectes was

conducted by Du et al (2008) In that paper, they

described two new species and divided the genus into

two species groups, i.e platycephalus and

furcocau-dalis groups, mainly based on caudal fin shape

(rounded vs forked), snout shape (round vs elongate)

as well as the appearance and development of the

caudal crests In 2009, Huang et al described another

species—O macrolepis and sorted it to the

furcocaudalis group Oreonectes elongatus sp nov.has a forked caudal fin, which indicates it belongs tothe group furcocaudalis Besides the new species,there are another four valid species in the group: O.furcocaudalis, O translucens, O microphthalmus,and O macrolepis

Oreonectes elongatus sp nov and O lis have a forked caudal fin with crests on both sides

furcocauda-of the caudal peduncle and an incomplete lateral line

in common Lacking eyes and scales, origin of dorsalfin slightly behind pelvic-fin base origin can distin-guish the new species from O furcocaudalis.Oreonectes elongatus sp nov and O microphthal-mus share the following common characters: a forkedcaudal fin, adipose crests present on both sides of thecaudal peduncle, elongate snout, incomplete lateralline, body de-pigmented However, O elongatus can

be distinguished from O microphthalmus by eyesbeing absent (vs vestigial with only black pigment in

O microphthalmus), the number of dorsal fin rays 8–

9 (vs 9–10), the number of anal fin rays 6 (vs.7),body naked (vs whole body covered with degen-erated scales or naked), and dorsal fin origin behindthe origin of anal fin (vs front)

Oreonectes elongatus sp nov and O macrolepisare both distributed in Huanjiang County, but indifferent townships The two species have thefollowing characteristics in common: a forked caudalfin, absent eyes, possessing dorsal and ventral crest,body skin with no markings or spots, lateral lineincomplete The new species differs from O macro-lepis by naked body (vs scaled body), disappearedeye (vs present but degenerated eye) and elongatesnout (vs round snout)

The closest known relative to the new speciesappears to be Oreonectes translucens (Fig 5), forboth of them possess caudal crests, a forked caudalfin, vestigial eyes, naked body and incomplete lateralline Du et al (2008) synonymized O translucenswith Triplophysa longibarbatus without checking thetype specimens O translucens has a clear shortdistance between anterior and posterior nostrils, andwell-developed nostril tubes (Zhang et al 2006),which apparently should not belong to Triplophysa.Therefore, O translucens is still a valid species Thenew species can be distinguished from O translucens

by having adipose caudal crests (vs transparent andwell-developed crests), slender anterior nostril tubewith long barbel (18.6–26.3% HL) (vs short nostril

Fig 3 Head (dorsal view) of holotype of Oreonectes elongatus

sp nov., showing anterior nostril with tube and barbel (a) and

posterior nostril (b)

tube with valve-like barbel, 7.6–9.4% HL), pectoral

fins end about 2/3 between the origin of pectoral fin

and pelvic fin (vs extended to the origin of pelvic fin)

as well as vertebrae 4+38–39 (vs 4+32) They are

also different in head length (20.2–22.8% vs 27.8–

32.3% of SL) and outer rostral-barbel length (55.0–

2 Eyes vestigial, only with black pigment on

location of the eyes; whole body covered with

scales except the abdomen……….…….3

Totally eyeless, whole body naked………….4

3 Snout round……….…O macrolepis

Snout elongate………O microphthalmus

4 Long anterior nostril tube and barbel, caudal

crests adipose and less than half of the caudal

peduncle depth…………O elongatus sp nov

Relatively short anterior nostril with

valve-like barbel, caudal crests transparent and

developed over 1/2 of the caudal peduncle

in Nov 1999

Oreonectes macrolepis (13 specimens, all typespecimens), KIZ 2008008131 (holotype), 55.0 mmSL; KIZ2008008130, 64.1 mm SL; KIZ 2008008294,61.8 mm SL; KIZ 2008008295, 44.9 mm SL; KIZ

2008008296, 41.9 mm SL; KIZ 2008008297,45.9 mm SL; KIZ 2008008298, 44.6 mm SL; KIZ

2008008299, 39.5 mm SL; KIZ 2008008300,47.4 mm SL; KIZ 2008008301, 40.1 mm SL; KIZ

2008008132, 48.4 mm SL; KIZ 2008008133,36.7 mm SL; KIZ 2008008134, 37.6 mm SL Allspecimens are from Dacai township, HuanjiangCounty, Guangxi, China in June 2008

Oreonectes microphthalmus (10 specimens, alltype specimens), KIZ 2004009395 (holotype),39.1 mm SL; KIZ 2004009394, 33.7 mm SL; KIZ

2004009396, 35.9 mm SL; KIZ 2004009397,48.7 mm SL; KIZ 2004009398, 31.0 mm SL; KIZ

2004009399, 32.4 mm SL; KIZ 2003007094,37.6 mm SL; KIZ 2003007095, 42.7 mm SL; KIZ

2003007096, 31.4 mm SL; KIZ 2003007097,31.2 mm SL All specimens checked here are fromLuocheng County in Guangxi

Fig 4 Distribution of Oreonectes elongatus sp nov and

relative furcocaudalis species ( ★ O elongatus sp.nov., ● O.

furcocaudalis, ▲ O translucens, ■ O microphthalmus, ◆ O.

macrolepis)

Fig 5 (a) Lateral view and (b) nostrils of holotype of Oreonectes translucens B-a, anterior nostril, B-b, posterior nostril