Patterns of morphological evolution of the cephalic region in

Patterns of morphological evolution of the cephalicregion in damselfishes (Perciformes: Pomacentridae)of the Eastern Pacific

ROSALÍA AGUILAR-MEDRANO1*, BRUNO FRÉDÉRICH2, EFRAÍN DE LUNA3 andEDUARDO F. BALART1

1Laboratorio de Necton y Ecología de Arrecifes, y Colección Ictiológica, Centro de InvestigacionesBiológicas del Noroeste, La Paz, B.C.S. 23090 México2Laboratoire de Morphologie fonctionnelle et évolutive, Institut de Chimie (B6c), Université de Liège,B-4000 Liège, Belgium3Departamento de Biodiversidad y Sistemática, Instituto de Ecología, AC, Xalapa, Veracruz 91000México

Received 20 May 2010; revised 21 September 2010; accepted for publication 22 September 2010bij_1586 593..613

Pomacentridae are one of the most abundant fish families inhabiting reefs of tropical and temperate regions. Thisfamily, comprising 29 genera, shows a remarkable diversity of habitat preferences, feeding, and behaviours.Twenty-four species belonging to seven genera have been reported in the Eastern Pacific region. The present studyfocuses on the relationship between the diet and the cephalic profile in the 24 endemic damselfishes of this region.Feeding habits were determined by means of underwater observations and the gathering of bibliographic data.Variations in cephalic profile were analyzed by means of geometric morphometrics and phylogenetic methods. Thepresent study shows that the 24 species can be grouped into three main trophic guilds: zooplanktivores, algivores,and an intermediate group feeding on small pelagic and benthic preys. Shape variations were low within eachgenus except for Abudefduf. Phylogenetically adjusted regression reveals that head shape can be explained bydifferences in feeding habits. The morphometric phylogeny recovered the subfamily Stegastinae and the relation-ship between Abudefduf troschelii and Chromis species. The cephalic profile of damselfishes contains a clear andstrong phylogenetic signal. © 2011 The Linnean Society of London, Biological Journal of the Linnean Society,2011, 102, 593–613.

ADDITIONAL KEYWORDS: Gulf of California – head shape – phylogenetic morphometrics – reef fishes –trophic niche.

INTRODUCTION

The damselfishes (Pomacentridae) are a species-rich(approximately 360), worldwide distributed family ofmarine fishes that inhabit tropical and temperatewaters (Allen, 1991). Most inhabit tropical coral reefs,although a number of species live in rocky reefs ofcooler temperate waters. They have been presentwithin these ecosystems for 50 million years (Bell-wood, 1996; Bellwood & Sorbini, 1996). In the EasternPacific, their range extends from Monterey Bay

(California, USA) in the north to the south of Chile,including all the oceanic islands of this region (Fig. 1).In this part of the Pacific, Pomacentridae is oneof the most abundant families, having radiated incoral reefs, rocky reefs, and kelp forests. Twenty-fourspecies belonging to seven genera (i.e. Stegastes,Microspathodon, Hypsypops, Nexilosus, Chromis,Azurina, and Abudefduf) have been reported in theEastern Pacific and all are endemic to this region(Robertson & Allen, 2008). The genera Azurina,Hypsypops, and Nexilosus are solely present in thispart of the Pacific, whereas the genus Microspath-odon contains two species in the Eastern Pacific and*Corresponding author. E-mail: [email protected]

Biological Journal of the Linnean Society, 2011, 102, 593–613. With 9 figures

© 2011 The Linnean Society of London, Biological Journal of the Linnean Society, 2011, 102, 593–613 593

two others in the Atlantic Ocean. On the other hand,the genera Abudefduf, Chromis, and Stegastes aredistributed worldwide (Allen, 1991). All genera rep-resented in this region belong to the basal groupsof the family Pomacentridae (i.e. the subfamilies Ste-gastinae, Chrominae and Abudefdufinae; as definedby Cooper, Smith & Westneat, 2009) (Fig. 2). Themembers of the most derived subfamily Pomacentri-nae have radiated in the Indo-West Pacific region only(Cooper et al., 2009). Despite the use of numerousmolecular data, the phylogenetic relationships withinthe family Pomacentridae are not fully resolved(Tang, 2001; Quenouille, Bermingham & Planes,2004; Cooper et al., 2009).

The damselfishes of the Eastern Pacific display aremarkable diversity with regards to habitat prefer-ences, feeding habits, and behaviours. Their colora-tion is highly variable, ranging from drab hues ofbrown, grey, and black to brilliant combinations oforange, yellow, and neon blue (Robertson & Allen,2008). Generally speaking, the damselfishes livingin coral reefs of the Indo-West Pacific region feedmainly on filamentous algae as well as small plank-tonic and benthic invertebrates (Allen, 1991; Kuo& Shao, 1991; Frédérich et al., 2009). The drab-coloured species feed mainly on algae, whereas most

of the brightly patterned species (e.g. membersof the genus Chromis) obtain their nourish-ment from the current-borne plankton (Allen, 1991).To our knowledge, the trophic ecology of only twodamselfish species from the Eastern Pacific has

Figure 1. Distribution of damselfish species in the coast and islands of the Eastern Pacific Ocean. Left: localities ofunderwater observations in the Gulf of California: National Marine Park of Loreto, La Paz Bay, and National Marine Parkof Cabo Pulmo.

Figure 2. Phylogeny of Pomacentridae family sensuCooper et al. (2009). Dotted lines refer to groups notrepresented in the Eastern Pacific region.

594 R. AGUILAR-MEDRANO ET AL.

© 2011 The Linnean Society of London, Biological Journal of the Linnean Society, 2011, 102, 593–613

been studied in detail: Stegastes rectifraenum andMicrospathodon dorsalis (Montgomery, 1980).

Ecomorphological studies attempt to understandthe relationships between the morphological varia-tion among species and their corresponding ecolo-gical variation (Norton, Luczkovich & Motta, 1995;Wainwright, 1996; Costa & Cataudella, 2007). Forexample, head morphology is subject to various con-straints dealing with the strategy of feeding and thetype of ingested food (Liem, 1979, 1993; Wainwright& Richard, 1995). The relationship between cephalicmorphology and diet has been broadly studied infreshwater fishes such as perches (Svanbäck & Eklöv,2002), centrarchids (Collar, Near & Wainwright,2005), and cichlids (Barel, 1983; Albertson, Streelman& Kocher, 2003), as well as marine fishes such assparids (Costa & Cataudella, 2007) and labrids (West-neat, 1995; Streelman & Karl, 1997; Wainwrightet al., 2004). The first ecomorphological studies inPomacentridae (Emery, 1973; Gluckmann & Vande-walle, 1998) suggested that a detailed study of cepha-lic morphology could reveal different trophic groups.Using geometric morphometrics (Rohlf & Marcus,1993; Lawing & Polly, 2009), Frédérich et al. (2008)highlighted strong size and shape variations in fourskeletal units of the damselfish skull among speciesliving in the Indo-West Pacific region and belonging todifferent trophic groups. Recently, Cooper & Westneat(2009) investigated the evolution of skull shapewithin Pomacentridae and showed that planktivoryhas involved important changes in damselfish headmorphology. In this latter study, the authors usedonly one species per genus to infer morphologicalevolution. Consequently, the inference of intra-genericvariation in trophic morphology was not possible.Recently, Barneche et al. (2009) confirmed the appli-cability of Bergmann’s rule to the family Pomacen-tridae, stating that the larger species are found athigher latitudes. The damselfishes of the EasternPacific are distributed along a wide range of latitudes(–33°N to 35°S; Fig. 1) and show a great variation ofbody size among species (standard length in the range11–36 cm; Table 1). Moreover, the genera Chromis,Hypsypops, Nexilosus and, Microspathodon containthe largest damselfishes (Allen, 1991). This axis ofbody size variation could have promoted morphologi-cal variation within this group of endemic species.

In the present study, we collect trophic data aboutall endemic damselfish species of the Eastern Pacificregion and apply geometric morphometric techniquesand phylogenetic methods to the cephalic region ofthese damselfishes. The main goal of the study is toprovide an overview of the trophic diversity of dam-selfishes in the Eastern Pacific and to determinewhether variations in cephalic shape can be explainedby size variation (i.e. allometry), feeding behaviour,

and/or phylogeny. In particular, we investigate theecomorphological hypothesis that cephalic shapevariation is associated with dietary differences indamselfishes of the Eastern Pacific. On the basisof previous analyses in damselfishes (Emery, 1973;Gluckmann & Vandewalle, 1998; Frédérich et al.,2008; Cooper & Westneat, 2009), we predict that dietis correlated to shape variations. We also test thepresence of phylogenetic signal (i.e. the expectationthat the phylogenetic relatedness is associated withshape similarity) (Cardini & Elton, 2008) in thecephalic region of damselfishes. We have no a priorihypothesis regarding this relationship because suchstudies are limited in fishes (Guill, Heins & Hood,2003).

MATERIAL AND METHODSTROPHIC DATA

Underwater observations (UO) were carried out inthree areas in the Gulf of California, Mexico (Fig. 1):(1) National Marine Park of Loreto Bay; (2) La PazBay; and (3) National Marine Park of Cabo Pulmo.The National Marine Park of Loreto Bay is located at26°07′ to 25°43′N and 111°21′ to 111°13′W. Coronado,Carmen and Danzante islands delimit this areawhere the substrate composition varies from fine sandto cobbles, boulders, and rocks (Campos-Dávila et al.,2005). La Paz Bay is located at 24°07′ to 24°21′N and110°17′ to 110°40′W, the reef presents fine sand, big(30 m) and small (12 m) boulders, big walls (15 m),and rocky reef (Aburto-Oropeza & Balart, 2001). TheNational Marine Park of Cabo Pulmo is the north-ernmost coral reef in the Eastern Pacific and islocated near the entrance of the Gulf of California ina transitional zone between the tropical and tem-perate Pacific at 23°50′N, 109°25′W (Alvarez-Filip,Reyes-Bonilla & Calderon-Aguilera, 2006). This parkpresents a barrier reef and a lagoon composed ofsmall boulders with sandy areas. The UO werecarried out during January, May, and August 2008,and January and March 2009, by scuba diving andsnorkeling. Observations of territorial species werecarried out for 2 min per organism; for fishes inconstant movement (e.g. planktivorous), a maximumof 10 min were devoted per organism. Observationswere conducted on 20–100 individuals per species,according to the availability. During these observa-tions, we mainly focused on two main components oftheir trophic ecology: (1) social behaviour: territorialand solitary species versus species forming groupsand (2) feeding habit: feeding areas (especially forterritorial, algivorous species), feeding strategy (i.e.biting, grazing or feeding in the water column), andtype of prey. Fourteen damselfish species were

HEAD SHAPE DIVERSITY IN DAMSELFISHES 595


Tab

le1.

Lis

tof

stu

died

dam

selfi

shsp

ecie

s

Spe

cies

Lor

eto

Bay

La

Paz

Bay

Cab

oP

ulm

oN

NM

NC

W

SL

A(c

m)

SL

B(c

m)

SL

C(c

m)

Abu

def

du

fco

nco

lor

(Gil

l,18

62)

88

013

1519

Abu

def

du

fd

ecli

vifr

ons

(Gil

l,18

62)

xx

x12

57

1316

18A

bud

efd

uf

tros

chel

ii(G

ill,

1862

)x

xx

3913

2613

1623

Azu

rin

aeu

pala

ma

Hel

ler

&S

nod

gras

s,19

038

80

1011

16A

zuri

na

hir

un

do

Jord

an&

McG

rego

r,18

98en

Jord

an&

Eve

rman

n,

1898

2525

013

1517

Ch

rom

isal

taG

reen

fiel

d&

Woo

ds,

1980

xx

x12

111

1015

15C

hro

mis

atri

loba

taG

ill,

1862

xx

x28

424

810

13C

hro

mis

cru

sma

(Val

enci

enn

es,

1833

enC

uvi

er&

Val

enci

enn

es,

1833

)16

160

1316

19C

hro

mis

inte

rcru

sma

Eve

rman

n&

Rad

clif

fe,

1917

1515

016

1728

Ch

rom

isli

mba

ugh

iG

reen

fiel

d&

Woo

ds,

1980

x25

1312

1114

12C

hro

mis

mer

idia

na

Gre

enfi

eld

&W

oods

,19

8025

250

1012

11C

hro

mis

pun

ctip

inn

is(C

oope

r,18

63)

x23

230

1925

25H

ypsy

pops

rubi

cun

du

s(G

irar

d,18

54)

x25

196

2023

36M

icro

spat

hod

onba

ird

ii(G

ill,

1862

)x

xx

205

1521

2531

Mic

rosp

ath

odon

dor

sali

s(G

ill,

1862

)x

xx

256

1918

2231

Nex

ilos

us

lati

fron

s(T

sch

udi

,18

46)

1616

022

2730

Ste

gast

esac

apu

lcoe

nsi

s(F

owle

r,19

44)

xx

2726

112

1818

Ste

gast

esar

cifr

ons

(Hel

ler

&S

nod

gras

s,19

03)

1515

011

1816

Ste

gast

esba

ldw

ini

All

en&

Woo

ds,

1980

1010

08

913

Ste

gast

esbe

ebei

(Nic

hol

s,19

24)

1010

011

1217

Ste

gast

esfl

avil

atu

s(G

ill,

1862

)x

x26

215

1114

15S

tega

stes

leu

coru

s(G

ilbe

rt,

1892

)x

xx

253

2212

1517

Ste

gast

esre

ctif

raen

um

(Gil

l,18

62)

xx

x24

024

1012

13S

tega

stes

red

empt

us

(Hel

ler

&S

nod

gras

s,19

03)

x23

230

1215

15Z

alem

biu

sro

sace

us

(Jor

dan

&G

ilbe

rt,

1880

)27

270

1417

20

Sit

esof

un

derw

ater

obse

rvat

ion

inth

eG

ulf

ofC

alif

orn

ia,

Mex

ico.

N,

nu

mbe

rof

spec

imen

su

sed

for

the

geom

etri

cm

orph

omet

ric

anal

ysis

;N

M,

nu

mbe

rof

spec

imen

sfr

omm

use

um

coll

ecti

ons;

NC

W,

nu

mbe

rof

spec

imen

sco

llec

ted

inth

ew

ild;

SL

,sta

nda

rdle

ngt

h;S

LA

,ave

rage

SL

ofth

efi

shes

use

din

the

pres

ent

stu

dy;S

LB

,max

imu

mS

Lof

the

fish

esu

sed

inpr

esen

tst

udy

;SL

C,m

axim

um

SL

repo

rted

byA

llen

(199

1).



directly observed and studied (Table 1). Consequently,the diet of all Eastern Pacific species was described bymeans of our UO and completed by a review of resultsfrom previous general studies. Using all these data,an alimentary item matrix of the 24 endemic dam-selfishes of the Eastern Pacific was built (Table 2). Itwas analyzed for the redundancy of the 16 alimentaryitems in all the species using the Kendall correlationindex for presence–absence data; this analysis wasconducted to group the alimentary items and reducethe main matrix. All alimentary items with a corre-lation higher than 0.80 were grouped (Table 2).

Additionally, the data of diet composition were alsoused for the estimation of the trophic level of eachdamselfish species. The TROPH index, initially usedfor Mediterranean fishes (Stergiou & Karpouzi, 2002),expresses the position of organisms within the foodwebs that largely define aquatic ecosystems. Realconsumers do not usually have TROPHs with integervalues and the definition of TROPH for any consumerspecies (1) is:

TROPH DC troph= + ⋅=

∑11

ij j

j

G

where G is the total number of prey species, DCij

represents the fraction of j in the diet of i and troph jis the fractional trophic level of prey j. The TROPHvalue was calculated from the dataset using TRO-PHLAB (Pauly et al., 2000), which is a stand-aloneapplication for estimating TROPH and its standarderror using qualitative information from the list ofitems known to occur in the diet of each species. If suchtrophic levels are missing, TROPHLAB uses defaulttroph values for various prey (based on data in FISH-BASE; Froese & Pauly, 2000). TROPH values wereused as feeding habit data, allowing the study of therelationships between head shape of damselfishes andtheir trophic ecology in a quantitative way, and givingan ecomorphological meaning to shape differences (seemethods below) (Costa & Cataudella, 2007).

GEOMETRIC MORPHOMETRICS AND

PHYLOGENETIC ANALYSIS

The sample comprised 509 specimens belonging to all24 damselfishes of the Eastern Pacific (Table 1). Addi-tionally, the embiotocid Zalembius rosaceus (Jordan &Gilbert, 1880) (family Embiotocidae) was used as an‘outgroup’. The Embiotocidae are now recognized asthe sister group of Pomacentridae (Streelman & Karl,1997; Mabuchi et al., 2007), Zalembius rosaceus is asmall benthic carnivorous fish distributed inside thesame area of the studied pomacentrids. Some speci-mens were speared during the field studies and theothers came from the museum collections of CIBNOR

(La Paz, BCS, México), CICIMAR (La Paz, BCS,México), SIO (San Diego, CA, USA), LACM (LosAngeles, CA, USA), and USNM (Washington, DC,USA) (Table 1). The list of museum specimens used inthis study is available upon request from the firstauthor (R.A.-M.).

The specimens were photographed in lateral viewwith a camera (Kodak ¥ 4 optical and 4 mega pixels)and the x- and y-coordinates of 18 homologous land-marks (Fig. 3) were digitized from the left side of eachindividual using TPSDIG, version 2.05. Superimposi-tion of landmark data was achieved using a general-ized procrustes analysis (Rohlf & Slice, 1990) whichaligned landmark configurations such that the sum ofsquared distances between corresponding landmarkswas minimized by scaling, translating, and rotatingspecimens with respect to a mean consensus con-figuration. The consensus configuration (‘the grandmean’) was obtained and used as the reference.Partial warp scores (PWs) including both uniform andnon-uniform components were calculated and used asdescriptors of shape variation (Bookstein, 1991; Rohlf,1993).

A principal components analyses (PCA) was usedto find hypothetical variables (components) thataccount for as much of the variance in the morpho-logical data (Davis, 1986). Subsequently, two kindsof discriminant analyses were carried out. Differ-ences among species were tested by means of analy-ses of variance (ANOVA), the Tukey–Kramer testand multivariate analyses of variance (MANOVA).Then, a canonical variance analysis (CVA) was per-formed to compare cephalic profile among groups.Indeed, CV axes allow us to maximize the differ-ences in shape among groups relative to withingroup variance. MANOVA and CVA were computedusing all shape variables (PWs). Deformation gridsusing the thin-plate spline (TPS) algorithm wereused to visualize the patterns of shape varia-tions along PC axes and CV axes (Thompson, 1917;Bookstein, 1991; Rohlf, 1993).

To determine whether shape data are hierarchicallyclustered, we applied phenetic and phylogeneticparsimony methods for grouping. Phenetic distancemethods are based upon a different conceptual frame-work of grouping than parsimony. We included aphenogram to determine whether there was evidencefor the assertion that similarity in morphometric datais not the result of a phylogenetic signal. Pheneticrelationships were summarized using a cluster analy-sis on the matrix of mean shape. Phenogram of the 25species was calculated using the unweighted pair-group method algorithm, and the Procrustes distanceas measure of the similarity. The goodness of fit of thecluster analysis was measured by the coefficient ofcophenetic correlation (Cardini & Elton, 2008). The



Tab

le2.

Die

tan

dtr

oph

icin

dex

ofal

lpo

mac

entr

ids

livi

ng

inth

eE

aste

rnP

acifi

c

Spe

cies

Die

t

TR

OP

Hin

dex

Ref

eren

ces

Ben

thic

Pel

agic

Oth

er

AB

CD

EF

GH

IJ

KL

MM

ean

±S

E

Abu

def

du

fco

nco

lor

01

10

00

00

00

00

03

±0.

38R

ober

tson

&A

llen

,20

08A

bud

efd

uf

dec

livi

fron

s0

11

00

00

00

00

00

3.6

±0.

52U

O;

Rob

erts

on&

All

en,

2008

Abu

def

du

ftr

osch

elii

01

01

10

01

10

10

13.

9±

0.61

UO

;H

obso

n,

1965

;P

eter

sen

&M

arch

etti

,19

89;

Gro

ve&

Lav

enbe

rg,

1997

Azu

rin

aeu

pala

ma

00

00

00

01

11

10

04

±0.

66R

ober

tson

&A

llen

,20

08A

zuri

na

hir

un

do

00

00

00

01

10

10

03.

8±

0.58

Rob

erts

on&

All

en,

2008

Ch

rom

isal

ta0

00

00

00

11

10

00

4±

0.65

Rob

erts

on&

All

en,

2008

Ch

rom

isat

rilo

bata

00

00

00

01

10

00

03.

9±

0.61

UO

;H

obso

n,

1965

;E

spin

oza

&S

alas

,20

05;

Rob

erts

on&

All

en,

2008

Ch

rom

iscr

usm

a0

00

00

00

11

11

00

4±

0.65

Nú

ñez

&V

ásqu

ez,

1987

;A

nge

l&

Oje

da,

2001

Ch

rom

isin

terc

rusm

a0

00

00

00

11

00

00

4±

0.65

Rob

erts

on&

All

en,

2008

Ch

rom

isli

mba

ugh

i0

00

00

00

11

00

00

3±

0.38

UO

;R

ober

tson

&A

llen

,20

08C

hro

mis

mer

idia

na

00

00

00

01

10

00

03

±0.

38G

reen

fiel

d&

Woo

ds,

(198

0)C

hro

mis

pun

ctip

inis

00

00

00

00

10

00

03.

4±

0.45

Bra

yet

al.,

1988

;R

oth

ans

&M

ille

r,19

91H

ypsy

pops

rubi

cun

du

s1

10

00

00

00

00

00

2.6

±0.

21H

ixon

,19

81;

Rob

erts

on&

All

en,

2008

Mic

rosp

ath

odon

bair

dii

01

01

10

00

00

00

02.

8±

0.36

UO

;R

ober

tson

&A

llen

,20

08M

icro

spat

hod

ond

orsa

lis

01

01

10

00

00

00

02.

8±

0.36

UO

;E

spin

oza

&S

alas

,20

05;

Rob

erts

on&

All

en,

2008

Nex

ilos

us

lati

fron

s0

11

10

00

00

00

00

2.9

±0.

32G

rove

&L

aven

berg

,19

97;

An

gel

&O

jeda

,20

01;

Rob

erts

on&

All

en,

2008

Ste

gast

esac

apu

lcoe

nsi

s0

10

10

00

00

00

00

3±

0.40

UO

;E

spin

oza

&S

alas

,20

05;

Rob

erts

on&

All

en,

2008

Ste

gast

esar

cifr

ons

01

01

11

00

00

00

03.

1±

0.38

Gro

ve&

Lav

enbe

rg,

1997

;R

ober

tson

&A

llen

,20

08S

tega

stes

bald

win

i0

10

10

00

00

00

00

3±

0.40

Rob

erts

on&

All

en,

2008

Ste

gast

esbe

ebei

01

01

01

00

00

00

03.

2±

0.38

Gro

ve&

Lav

enbe

rg,

1997

;R

ober

tson

&A

llen

,20

08S

tega

stes

flav

ilat

us

01

01

00

01

00

00

03

±0.

38U

O;

Pet

erse

n&

Mar

chet

ti,

1989

;E

spin

oza

&S

alas

,20

05;

Rob

erts

on&

All

en,

2008

Ste

gast

esle

uco

rus

01

01

00

00

00

00

03

±0.

40U

O;

Rob

erts

on&

All

en,

2008

Ste

gast

esre

ctif

raen

um

01

01

00

01

00

01

02.

9±

0.34

UO

;H

obso

n,

1965

;P

eter

sen

&M

arch

etti

,19

89;

Rob

erts

on&

All

en,

2008

Ste

gast

esre

dem

ptu

s0

10

10

00

00

00

00

3±

0.40

Rob

erts

on&

All

en,

2008

Zal

embi

us

rosa

ceu

s0

00

11

01

00

10

00

3.3

±0.

47R

ober

tson

&A

llen

,20

08

Ben

thic

diet

:A,

spon

ges;

B,

mac

roal

gae,

mic

roal

gae,

cru

stac

ean

s,an

dse

ssil

ew

orm

s;C

,se

ssil

em

ollu

sks;

D,

mob

ile

cru

stac

ean

s;E

,m

obil

ew

orm

s;F,

anem

ones

;G

,ga

stro

pods

;H

,fi

sheg

gs.

Pel

agic

diet

:I,

zoop

lan

kton

;J,

cru

stac

ean

s;K

,fi

shla

rvae

.O

ther

:L

,de

trit

us;

M,

ecto

para

site

s.U

O,

Un

derw

ater

obse

rvat

ion

s.



morphological phylogenetic analysis was executedusing the 32 shape variables (RWs) as characters forall studied species. For each species, we scored eitherRW intervals, RW means or both. Phylogenetic analy-ses were performed using new algorithms for thedirect optimization of continuous characters in TNT(Goloboff, Mattoni & Quinteros, 2006). We used acombination of ratchet, sectorial search, tree drifting,and tree fusing search algorithms for the selection ofoptimal trees and for the estimation of Jackknifevalues as a measure of support.

The molecular phylogenetic hypothesis by Cooperet al. (2009) and our morphological phylogeny wereused for phylogenetic adjusted regressions of theshape variables on the TROPH index, centroid size(CS), and standard length (SL) using the modulePDAP in MESQUITE (Midford, Garland & Maddison,2009). Our morphological and molecular phylogenieswere used to obtain an estimate of character distri-bution and correlation along a phylogeny (Garlandet al., 1993). Phylogenetic regression analyses totest interspecific allometry were performed usingthe mean values of CS and SL versus the 32shape variables. Similarly, we performed phylogeneticregression analysis using the mean values of the 32shape variables and the TROPH index to test therelationship between feeding habit and cephalicshape. Additionally, we compared these results withconventional nonphylogenetic regression analysesperformed in TPSREGRES, version 1.37.

Geometric morphometric analyses were performedusing computer programs from the TPS series(TpsDig, TpsRegres) written by F. J. Rohlf (http://life.bio.sunysb.edu/morph/), and the IMP series (PCAGen,CVAGen), created by H. D. Sheets (http://www2.canisius.edu/~sheets/morphsoft.html). Multivariateanalyses (correlation analyses, ANOVA, PCA, CVA,MANOVA) were computed with the statistical pack-ages: PAST, version 1.74 (Hammer, Harper & Ryan,2001; http://folk.uio.no/ahammer/past), Statistica,version 8.0 (http://www.StartSoft.com) and JMP,version 8.0 (SAS Institute Inc.). Phylogenetic analysiswas computed in TNT, version 1.1 (Goloboff, Farris &Nixon, 2008; http://www.zmuc.dk/public/phylogeny).Phylogenetic regressions were performed in thePDAP package (Midford, Garland & Maddison, 2010;http://mesquiteproject.org/pdap_mesquite/) for MES-QUITE, version 2.73 (Maddison & Maddison, 2010;http://mesquiteproject.org/mesquite/mesquite.html).

RESULTSTROPHIC AND BEHAVIOURAL DATA

Chromis limbaughi, Chromis punctipinnis, Hypsy-pops rubicundus, and Stegastes redemptus were onlyobserved at one locality (Table 1), whereas the trophicbehaviours were always consistent among sites in theother species. Unfortunately, Chromis alta was neverobserved at any location, maybe as a result of itsdepth preferences (down to 150 m; Allen, 1991). Gen-erally, the underwater observations corroborate dataobtained from the literature related to diet of eachstudied species. However, some new trophic behav-iours were also observed. For example, the mainlyalgivorous S. rectifraenum was observed recurrentlypicking feces from zooplankton feeders (e.g. Abu-defduf troschelii) directly in the water column.

Figure 3. Anatomical landmarks used in geometric mor-phometric analyses: (1) tip of the premaxilla; (2) nostril;(3–6) inferior, anterior, superior, and posterior margin ofthe eye; (7) center of the eye; (8) superior tip of thepreopercular; (9) superior tip of the operculum; (10) pos-terior tip of the operculum; (11, 12) superior and inferiorinsertion of the pectoral fin; (13) insertion of the pelvic fin;(14) insertion of the operculum on the body profile; (15)posterio-ventral corner of the preopercular; (16, 17) ante-rior and posterior extremity of the dentary; (18) posteriorextremity of the premaxilla.



The trophic behaviour appeared relatively constantwithin all genera, except within the genus Abudefduf(Table 2). Abudefduf troschelii mainly fed on plank-tonic preys in the water column, whereas Abudefdufconcolor and Abudefduf declivifrons mainly grazedalgae or fed on some small sessile animals attached tothe rocks.

The analysis of alimentary items shows a strongrelationship (r > 0.80) between four items: macroal-gae, microalgae, sessile crustaceans, and worms.Consequently, these prey items were grouped inTable 2. Thus, a mainly algivorous species such asall Stegastes species, Microspathodon species, Nex-ilosus latifrons, and H. rubicundus may be con-sidered as a benthic feeder grazing on algae andfeeding on small benthic invertebrates. On the basisof a review of the published results and our UO,four alimentary items are species-specific (Table 2):sponges (A): H. rubicundus; detritus (L): S. rectifrae-num; gastropods (G): Z. rosaceus; and ectoparasites(M): A. troschelii. Abudefduf troschelii showed thelargest diet of the damselfish investigated in thepresent study, feeding in benthic and pelagic areas.All seven species of Chromis and the two speciesof Azurina mainly feed on zooplankton. Zalembiusrosaceus is considered as a benthic carnivorousspecies, feeding only on mobile benthic crustaceans(shrimps, small crabs), worms, gastropods, andbivalves (Table 2).

As for the diet, the social behaviour was con-served within genera. Hypsypops rubicundus andStegastes species are solitary, protecting small ter-ritories, forming couples only during the repro-ductive season. Hypsypops rubicundus showed apreference for small caves around big rocks andall Stegastes species prefer small rocks. BothMicrospathodon species are solitary or live in pairsand protect big rocks with high vertical walls. Nex-ilosus latifrons was not observed in the wild,although some authors (Grove & Lavenberg, 1997;Angel & Ojeda, 2001; Robertson & Allen, 2008)reported that it is a grazing species living close tothe rocks in small groups (up to ten individuals). AllAbudefduf, Azurina, and Chromis species live ingroups (i.e. schooling species) and are territorialonly during the reproductive season. The main dif-ference between these species is the number of inte-grants in the groups (Abudefduf species: 5–30individuals versus Chromis species: 10–60 indivi-duals), although it proved complicated to achievea good description of these numbers because theychanged according to the study area and thehabitat. Neither Azurina species were observed inwild; thus, we do not have any estimation of thenumber of fish per group. Specimens of A. declivi-frons and A. concolor could also be found solitary.

GEOMETRIC MORPHOMETRICS

The main shape variations across species can beexamined by a distribution of specimens in the PCfigure defined by the axis PC1 and axis PC2 (Fig. 4).The first two PCs account for 70% of the total shapevariance (PC1 = 58.41% and PC2 = 10.32%). Shapevariation was relatively low within each genus exceptfor Abudefduf. This genus could be clearly dividedinto two groups, one comprising A. concolor and A.declivifrons and the other comprising A. troschelii.PC1 allowed a clear distinction between mainlybenthic feeders (low PC1 values) and mainly zoop-lankton feeders (high PC1 values). The main morpho-logical variation along axis PC1 is related to the snoutlength, the cephalic depth, and the eye size andposition. The benthic feeders, especially Microspath-odon species, were characterized by a very shortsnout and small eyes; the pelagic feeders (Azurinaspecies, Chromis species, and A. troschelii) were char-acterized by relatively larger eyes and longer cephalicprofile. Along axis PC2, the main shape variation wasrelated to the position of the mouth and the pectoralfin, and explains the variation within some generasuch as Stegastes and Chromis.

When species were used as grouping factor,the MANOVA revealed significant differences amongthem (lWILKS = 2.36 ¥ 10–5, F = 42.02, d.f.1 = 256,d.f.2 = 3665.2; P < 0.001) and pairwise comparisonssupported that all species showed significant dif-ferences in mean shape (P < 0.05). MANOVA wasrepeated using genus as grouping factor, and theresults also show significant differences among allgenera; however, Abudefduf genus show a strongpatter of segregation into two groups. According to thetrophic data, the feeding habit is consistant withingenera except for the genus Abudefduf. The segrega-tion pattern of Abudefduf was tested by exploratoryanalyses (ANOVA, PCA, and CVA). Because thepattern was repeated in all analyses, the finalMANOVA was performed using genera as groupingfactor, with the genus Abudefduf split into twogroups: (1) A. concolor and A. declivifrons and (2) A.troschelii. The results of MANOVA show significantdifferences among all groups (lWILKS = 0.0000263,F = 41.23, d.f. = 265, P < 0.001), and all pairwisecomparisons based on Mahalanobis distances showsignificant differences (P < 0.05).

Discrimination among groups can be also inter-preted by examining the ordination of specimensin the morphospace defined by the CV axes (Fig. 5).The first three CV axes accounted for 84% of thetotal shape variation in the dataset and allow thediscrimination of five main groups. The axis CV1distinguishes three groups (Table 3): (A) Micros-pathodon species; (B) A. concolor, A. declivifrons, H.



rubicundus, N. latifrons and Stegastes species; (C)Azurina species, Chromis species, A. troschelii and Z.rosaceus. This last group (group C) could be subdi-vided into three groups in the morphospace defined byCV1 and CV3: Z. rosaceus in the extreme high values,Azurina species in the middle values, and A. troscheliiand Chromis species in the extreme low values(Fig. 5, Table 3).

In general, the CVA strengthened the profile andthe position of the eye and the mouth. Having thelowest scores of CV1, the two species of Microspath-odon showed a higher and flatter cephalic profile;Chromis species, Azurina species, A. troschelii, and Z.rosaceus present a lengthened cephalic profile withbig eyes; H. rubicundus, N. latifrons, Stegastesspecies, A. concolor, and A. declivifrons showed anintermediate shape. Along CV2, the main shapevariation was related to the position of the mouth andof the pectoral fin. The two Microspathodon speciesshowed a more horizontally oriented pectoral fin, thecephalic profile is high and the mouth is small

(+CV2). The eye of the two Azurina species and Z.rosaceus was in a more forward and lower position,and the snout region is also longer than in allChromis species and A. troschelii (CV3; Fig. 5). Thephenogram shows a segregation of four morphologicalgroups (Fig. 6): (1) all species of the genus Stegastes,A. concolor, and A. declivifrons, which are mainlyalgal feeders; (2) all species of the genus Chromis andA. troschelii, which are mainly zooplankton feeders;(3) both Microspathodon species are algal feeders,with an extremely flat cephalic shape; and (4) bothAzurina species, which are zooplanktivorous, and Z.rosaceus, which is carnivore benthic feeder, form agroup having a highly sharp cephalic profile. Thecoefficient of cophenetic correlation is relatively high(r = 0.76).

PHYLOGENETIC ANALYSIS

The morphological phylogenetic hypothesis (Fig. 6)shows a grouping pattern with a high degree of

Figure 4. Scatterplot of principal components (PC) 1 and 2. Cross, Abudefduf; equis, Azurina; square, Chromis; triangle,Hypsypops; circle, Microspathodon; asterisk, Nexilosus; black circle, Stegastes; black square, Zalembius. Thin plate splinedeformation grids for the extreme points of each axis are shown; these are superimposed on the shapes predicted whenthe average landmark configuration of all specimens is deformed into that of a hypothetical specimen positioned at theextreme point of an ordination axis.



correlation between head morphology and feedinghabits. The morphological phylogeny is not totallycongruent with the molecular phylogeny, exceptin that the subfamily Stegastinae is recoveredwith the genus Stegastes, Nexilosus, Hypsypops,and Microspathodon. This clade presents mode-rate support values (Jackknife = 63). However, ourmorphometric phylogeny includes two species of

Abudefduf that belong to the Abudefdufinae sub-family according to the molecular phylogeny. Withinthis group, the clade of two species of Abudefduf ishighly supported (Jackknife = 96). The subfamiliesChrominae and Abudefdufinae are not recovered asmonophyletic groups in our morphometric phylogeny.Rather, the Chrominae is a partially resolved gradethat also includes A. troschelii.

Figure 5. Scatterplot of canonical variates (CV): CV1 versus CV2 and CV1 versus CV3. Black cross, Abudefduf concolorand A. declivifrons; grey cross, Abudefduf troschelii; equis, Azurina; white square, Chromis alta, C. crusma, C.intercrusma, C. limbaughi, C. meridian and C. punctipinnis; grey square, Chromis atrilobata; triangle, Hypsypops; circle,Microspathodon; asterisk, Nexilosus; black circle, Stegastes; black square, Zalembius. For both plots, thin plate splinedeformation grids for the extreme points of each axis are shown; these are superimposed on the shapes predicted whenthe average landmark configuration of all specimens is deformed into that of a hypothetical specimen positioned at theextreme point of an ordination axis.



RELATIONSHIP BETWEEN CEPHALIC SHAPE AND SIZE

Interspecific allometry was tested using linear regres-sions analyses. The relationship between shape andsize (CS and SL) is low (r2 � 0.4; Table 4). The lowestr2-values were found with the nonphylogeneticregression analysis and the highest ones with thephylogenetic regression analyses (Table 4). However,shape always showed a stronger relationship with CSthan with SL.

RELATIONSHIP BETWEEN CEPHALIC

SHAPE AND TROPHIC DATA

Both nonphylogenetic regression analyses and theequivalent with phylogenetically independent con-trasts show a significant positive relationshipbetween the TROPH index and shape variables(Table 4). Values of the coefficient of determination(r2) were lower in the phylogenetic regression analy-sis than in the nonphylogenetic analysis. These testsand mirror trees analyses clearly demonstrate that

the main shape variation among the studied dam-selfishes is related to their feeding habit (Fig. 7).Cephalic shapes are more related to trophic levelsthan to the phylogenetic relationships (Fig. 7,Table 4). Trophic data group the 24 damselfishes inthree main groups: mainly algal feeders, mainlyzooplankton feeders, and an intermediate groupfeeding on small pelagic and benthic preys. CVA andcluster analysis allow the discrimination of the threemain trophic groups and also show a clear degree ofvariation between two extremes of cephalic shape(i.e. zooplankton feeders, Azurina species, withhighly sharpness cephalic shape and algal feeders,Microspathodon species, with highly flattened cepha-lic shape). On the other hand, the phylogenetichypothesis constructed with shape data producestwo monophyletic groups: (1) subfamily Stegastinae,which group mainly algal feeders and (2) a groupcomposed of A. troschelii, Chromis meridiana, andC. limbaughi, which are mainly zooplanktivorous,feeding also on fish eggs.

Table 3. Analysis of variance and Tukey–Kramer test of the three first axes of canonical variance analysis

A B C D E F G H I

CV 1 A 5.694 7.324 7.605 0.743 8.480 0.143 0.118 4.959B < 0.0001 1.630 1.910 6.437 14.175 5.837 5.576 0.735C < 0.0001 < 0.0001 0.280 8.067 15.805 7.467 7.206 2.365D < 0.0001 < 0.0001 0.876 8.348 16.085 7.747 7.487 2.646E 0.246 < 0.0001 < 0.0001 < 0.0001 7.737 0.600 0.861 5.702F < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 8.338 8.598 13.439G 1.000 < 0.0001 < 0.0001 < 0.0001 0.631 < 0.0001 0.261 5.102H 1.000 < 0.0001 < 0.0001 < 0.0001 0.002 < 0.0001 0.986 4.841I < 0.0001 0.083 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001

CV 2 A 0.443 2.710 1.015 1.418 4.645 1.408 2.889 1.692B 0.799 3.153 1.458 0.975 5.088 0.965 2.446 1.249C < 0.0001 < 0.0001 1.695 4.127 1.935 4.118 5.599 4.402D 0.001 < 0.0001 < 0.0001 2.432 3.630 2.423 3.904 2.707E < 0.0001 0.005 < 0.0001 < 0.0001 6.063 0.010 1.471 0.275F < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 6.053 7.534 6.337G < 0.0001 0.033 < 0.0001 < 0.0001 1.000 < 0.0001 1.481 0.284H < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 1.197I < 0.0001 < 0.0001 < 0.0001 < 0.0001 0.987 < 0.0001 0.993 < 0.0001

CV 3 A 1.33 2.82 1.20 1.13 0.11 0.54 0.35 6.72B < 0.0001 4.15 0.13 0.20 1.44 1.87 0.98 8.05C < 0.0001 < 0.0001 4.02 3.95 2.71 2.28 3.17 3.90D < 0.0001 1.00 < 0.0001 0.07 1.31 1.74 0.85 7.92E 0.01 1.00 < 0.0001 1.00 1.24 1.67 0.78 7.85F 1.00 < 0.0001 < 0.0001 < 0.0001 < 0.0001 0.43 0.46 6.62G 0.80 < 0.0001 < 0.0001 < 0.0001 < 0.0001 0.86 0.89 6.18H 0.86 < 0.0001 < 0.0001 < 0.0001 < 0.0001 0.14 0.02 7.07I < 0.0001 < 0.0001 < 0.0001 < 0.0001 0.01 < 0.0001 < 0.0001 < 0.0001

A: Abudefduf concolor, Abudefduf declivifrons; B: Abudefduf troschelii; C: Azurina spp.; D: Chromis spp.; E: Hypsypopsrubicundus; F: Microspathodon spp.; G: Nexilosus latifrons; H: Stegastes spp.; I: Zalembius rosaceus. Above the diagonal:difference degree of Tukey–Kramer; below the diagonal: P-value.



The optimization of the TROPH index and morphol-ogy in the molecular phylogeny revealed that allspecies of the genus Stegastes and N. latifrons arehighly similar with respect to morphology and feedinghabits. The subfamily Stegastinae show a high cor-relation between morphology and feeding habits,

although some shape variation can be observedamong all Stegastes species, N. latifrons, and thetwo Microspathodon species. Morphological phylog-eny highlighted a pattern of convergence with twomembers of Abudefduf genus (A. concolor and A.declivifrons). Although A. concolor and A. declivifrons

Figure 6. Comparison of two hierarchical models. Phenogram (A) and morphometric phylogeny (B) compared to maintrophic groups of damselfish. Images of some species are added to help to visualize the pattern of cephalic shape variationin relation to each trophic group. In the phylogeny, the number on branches are support values (jackknife, 1000 replicates,cut = 50, jackknifing P = 36). Coph. Corr., coefficient of cophenetic correlation.

Table 4. Phylogenetic and nonphylogenetic regression analysis for testing interespecific allometry and the relationshipbetween morphology and trophic data

Least squareregression Variables r2 F d.f. P

Phylogenetic CS versus shape variables (molecular and morphological phylogeny) 0.41 16.32 23 0.0005SL versus shape variables (molecular and morphological phylogeny) 0.17 4.64 23 0.042TROPH index versus shape variables (molecular phylogeny) 0.51 23.16 23 < 0.0001TROPH index versus shape variables (morphological phylogeny) 0.51 23.16 23 < 0.0001

Nonphylogenetic CS versus shape variables 0.20 6.22 32–736 < 0.0001SL versus shape variables 0.09 2.44 32–736 < 0.0001TROPH index versus shape variables 0.36 13.17 32–736 < 0.0001

CS, centroid size; SL, Standard length.



Figure 7. Mirror tree optimization of the TROPH index (left) and morphology (right). A, morphological hypothesis;B, molecular hypothesis.



are close according to the morphological phylogeny,the feeding habit of A. declivifrons is different andmore similar to that of members of the Chrominaesubfamily. Both Microspathodon species and H. rubi-cundus are close according to their feeding habits.The cephalic shape is highly conserved in the sub-family Stegastinae through both phylogenies. Accord-ing to the categories of the TROPH index (Fig. 7), theChrominae subfamily shows five categories distrib-uted between medium and high values of the TROPHindex. On the other hand, the range of morphologicalvariation of this group is low (i.e. only two extrememorphological categories of colour code; Fig. 7).According to the morphological phylogeny, A. trosche-lii is closely related to the Chrominae. The con-vergence pattern of A. troschelii introduces a newmorphology within this group. The Abudefdufinaesubfamily is highly diverse. Two morphological cat-egories were found in this subfamily (Fig. 7); (1)A. troschelii is more similar in shape to the outgroup(Z. rosaceus) and (2) A. concolor and A. declivifronsshow similar values to the Stegastinae subfamilymembers. Each Abudefduf species has a differentcolour code for its trophic index (Fig. 7).

DISCUSSIONTROPHIC DIVERSITY OF DAMSELFISHES

The trophic data obtained in the present studyconfirm the division of damselfishes of the EasternPacific into three main trophic groups, as reported forspecies living in the Indo-West Pacific region (Allen,1991; Kuo & Shao, 1991; Frédérich et al., 2009): (1)the pelagic feeders mainly sucking zooplanktonicpreys (e.g. Chromis spp., Azurina spp.); (2) thebenthic feeders mainly grazing filamentous algae andpicking small invertebrates (e.g. Stegastes spp., Nex-ilosus latifrons, Hypsypops rubicundus, Microspath-odon spp.); and (3) an intermediate group gatheringspecies feeding both on small pelagic and benthicpreys (e.g. Abudefduf spp.).

The social behaviour of damselfishes is constantin the whole Indo-Pacific, Western Atlantic, andEastern Pacific region (Emery, 1973; Frédérich et al.,2009; present study): species mainly feeding onbenthic prey are solitary, whereas zooplanktivorousspecies live in groups. However, N. latifrons is rela-tively atypical because it is a grazing species livingin small groups (up to ten individuals) (Grove &Lavenberg, 1997; Angel & Ojeda, 2001; Robertson& Allen, 2008).

In general, the diet is consistent within damselfishgenera. However, our ecomorphological approach indi-cated that the three Abudefduf species should begrouped into two different trophic guilds. Consistent

with the trophic data (Table 2), the cephalic profile ofA. troschelii is more similar to that of Chromis andAzurina, suggesting that this species should be con-sidered as an omnivorous species mainly feeding onzooplankton. On the other hand, A. declivifrons andA. concolor are more similar to Stegastes, Hypsypops,and Nexilosus, showing that they should be consid-ered as omnivorous species and mainly benthicfeeders.

To our knowledge, H. rubicundus is the onlydamselfish feeding on sponges. No damselfish of theIndo-West Pacific is known to be a common con-sumer of such kind of prey (Allen, 1991; Frédérichet al., 2009). The angelfishes (Pomacanthidae) isspecialized in such prey catching in coral reef envi-ronments (Konow & Bellwood, 2005). Accordingto Robertson & Allen (2008), one Pomacanthidae,Pomacanthus zonipectus, is distributed in the coolertemperate regions and lives sympatrically withH. rubicundus. However, this angelfish achieves itsmaximum population density in the Tropical EasternPacific and is scarce in cooler temperate regions.Consequently, a lower competition level could permitan extension of the trophic width of damselfishessuch as H. rubicundus in temperate regions.

ECOMORPHOLOGY AND MORPHO-FUNCTIONAL

IMPLICATIONS

The bucco-pharyngeal cavity of a fish has been mod-elled as a truncate cone, whose small base comprisesthe circular opening of the mouth and whose largebase is located behind the branchial basket on thelevel of the opercles (Alexander, 1967; Lauder, 1980;Lauder & Lanyon, 1980; Liem, 1993). The efficiency ofthe cone depends on various factors such as themorphology of the skull and particularly the bucco-pharyngeal cavity (Liem, 1990). There are three basicmodes of feeding according to the degree of truncationof the cone (Liem, 1980, 1993): suction feeding, ramfeeding, and biting. However, a mode of prey captureis not exclusive; many teleosts are able to modulatetheir feeding mode and to move from one categoryto another (Liem, 1980, 1993; Ferry-Graham et al.,2002). If the head morphology of damselfishesprompts the consideration that they are good suctionfeeders (Emery, 1973; Frédérich et al., 2008; Cooper& Westneat, 2009; present study), geometric mor-phometric analyses allow a deeper understanding ofthe different ways of feeding and reveal functionaldifferences among species.

The main difference between morphological groupsis the degree of sharpness of the cephalic shape,which goes from a long angular cephalic profile as inZ. rosaceus, both Azurina species, A. troschelii, andall Chromis species; followed by angular but shorter



cephalic profiles as in A. concolor, A. declivifrons, H.rubicundus, N. latifrons, and all Stegastes species;and, finally, to an almost flat cephalic profile as inboth Microspathodon species (Fig. 6). The benthic car-nivorous Z. rosaceus feeds mainly on gastropods,mobile worms, and crustaceans (Table 2). As observedin some cichlids (Liem, 1993), a long angular cephalicprofile may facilitate the catching of these items.Despite a similar angular cephalic profile as in bothAzurina species and Chromis species, Z. rosaceusshows a great morpho-functional difference comparedto these zooplanktivorous species, which was not indi-cated by our geometric morphometric analyses.Indeed, this difference is solely observed when themouth is extended (i.e. during mouth protrusion)(Fig. 8). During feeding, the mouth is oriented moreventrally in Z. rosaceus, optimizing the capture ofbenthic animal preys, whereas the mouth is directedrostrally in Azurina and Chromis species, facilitatingprey capture in the water column. Further morpho-functional studies should precisely address the differ-ences in the degree of protrusion of the premaxillarybones during feeding among these species. Zooplank-tivorous damselfishes such as the Azurina andChromis species can be described as being particulate

feeding in that they attack the individual planktonicpreys they select visually. The possession of relativelylarge eyes (Fig. 5) should increase their ability to findand target planktonic preys, as exemplified in cichlids(Barel, 1983). Their elongated head profile facilitatesthe capture of these organisms using ram-suctionfeeding (Coughlin & Strickler, 1990), although furtherfunctional studies should aim to test whether thedifferences in cephalic profile between Chromis andAzurina species (Fig. 5) could be related to differencesin feeding strategy and performance. For example,the contribution of ram (i.e. the predator movementtowards the prey) and suction (i.e. the prey movementtowards the predator as a result of aspiration) duringfeeding may differ between both genera (Wainwrightet al., 2001).

Abudefduf troschelii is an omnivorous speciesfeeding mainly on zooplankton (Grove et al., 1986;present study). Robertson & Allen (2008) consideredthis species to comprise two feeding groups: omnivo-rous and planktivorous. Abudefduf troschelii shows avery similar cephalic profile to the almost exclusivezooplanktivorous Chromis and Azurina genera. Simi-larly, A. troschelii mainly occurs along rocky shores orcoral reefs, in shallow waters, foraging on zooplank-ton in aggregations. By contrast, A. concolor, A.declivifrons, H. rubicundus, N. latifrons, all Stegastesspecies and both Microspathodon species mainlygraze filamentous algae growing on rocks. Within thistrophic group, the two Microspathodon speciespresent a highly different morphology. Our underwa-ter observations revealed that the way of feeding inboth Microspathodon species differ from that of theothers. Indeed, all Stegastes species grazed algae orpicked up small invertebrates on small, mainly hori-zontal rocks or rubble, whereas the two Microspath-odon species scraped on big rocks with high verticalwalls. This type of feeding is probably facilitated byan almost flat cephalic profile in Microspathodonspecies. Moreover, the premaxillary bones on theiranterior region reveal a loose connective tissue whereteeth are continuously produced (Ciardelli, 1967).When the fish scrapes the rocky wall, the teeth arecontinuously eroded and, consequently, need to beproduced constantly (Trapani, 2001). Furthermore,this connective tissue that supports and nurtures theteeth could act as a buffer, supporting the movementof the teeth and the premaxillary bones on therock. Although, when the food items are similar, themorphological pattern could diverge if the methods ofprey catching differs.

The results of the present study indicate that dam-selfishes from the Eastern Pacific show strong differ-ences with respect to their cephalic profile. Thesedifferences are mainly related to the degree of sharp-ness and the position of the eye and the mouth. A

Figure 8. Difference in the mouth orientation when it isopen in (A) Zalembius rosaceus, (B) Azurina hirundo, and(C) Chromis atrilobata.



pattern of relationship between head morphology anddiet was found, as in previous analysis (Frédérich,Parmentier & Vandewalle, 2006; Frédérich et al.,2008; Cooper & Westneat, 2009). However, the rela-tionship between the cephalic profile and feedinghabit found in damselfishes extends beyond this. Inthe present study, head morphology was observed tobe related to the way that the food resource isextracted: two species can use the same food resource(Stegastes species and Microspathodon species) andpresent a different head morphology, and theserespond to the way that each species extracts theresources of the environment.

EVOLUTIONARY CONSIDERATIONS

The shape data of the present study confirm amorphological group composed by Z. rosaceus, ourchosen outgroup (Figs 4, 5). The family Embiotocidaeincludes mainly carnivorous (zoobenthos) species witha pointed cephalic profile. Interestingly, Z. rosaceusshows a cephalic profile more similar to the mostderived zooplanktivorous damselfishes (i.e. Chromisand Azurina) (Fig. 2). The similarity among thesethree genera is a convergence in the cephalic profile.The present study confirms the results reported byFrédérich et al. (2008), who stated that shape varia-tion is not correlated with size variation in damself-ishes. Interspecific allometry is low and size does notpredict convergent shapes. Consequently, variationsin size and shape may be viewed as two independentevolutionary factors explaining the diversification ofthe family Pomacentridae. When the relationshipsbetween CS, SL, and shape variables are analyzedthrough nonphylogenetic analyses, r2-values arelower than when they are analyzed through phyloge-netic analyses, indicating that size is related to thephylogeny.

The morphometric results of the present studyshow Azurina to be very similar in cephalic shape toChromis. In a Euclidean plane, the slender Azurinaspecies lie at the extreme point of the Chromis dis-tribution along an axis of cephalic length (Figs 4, 5).According to the morphological phylogeny, Azurinaspecies are closely related to Chromis species (Fig. 6);this relationship is also reported by Cooper et al.(2009), who proposed synonymizing the genusAzurina with Chromis and re-classifying the twospecies of Azurina. Their proposal was based on astrong phylogenetic relationship between Azurinahirundo from the Eastern Pacific and Chromis mul-tilineata from the Atlantic.

The genus Abudefduf had been traditionally segre-gated into two groups. According to our mophometricanalyses, the main difference between these groups isthe position of the mouth, which clearly corresponds

to two different feeding habits. In A. troschelii, themouth is superior (omnivorous, feeding mainly onzooplankton); this shape is very similar to that of C.meridiana and C. limbaughi (Fig. 6). By contrast, themouth of A. concolor and A. declivifrons is inferior(omnivorous, feeding mainly on benthic preys); thus,the shape is similar to Microspathodon bairdii andMicrospathodon dorsalis (Fig. 6). The molecular phy-logenetic analysis of Quenouille et al. (2004) consi-dered 16 Abudefduf species, and found three mainclades within this genus: A. declivifrons, A. concolor,and Abudefduf taurus are the sister group to thetwo principal groups. Abudefduf taurus is a mainlyherbivorous fish (Randall, 1967; Allen, 1991), withsimilar behaviour to A. concolor and A. declivifrons,showing strong preference for limestone shorelinesand tide pools in regions with surf (Allen, 1991;R. Aguilar-Medrano, pers. observ.). The second groupcomprises Abudefduf sordidus, Abudefduf septemfas-ciatus, and Abudefduf notatus, which are also mainlyalgal feeders (Allen, 1991). All Abudefduf species ofthese two groups possess a dark body with clearvertical bars and have a strong preference for veryshallow areas (0–4 m) with surf (Allen, 1991; R. A.-M.& B. F. pers. observ.). Their mouth is placed in a lowerposition than the others. The third cluster of Que-nouille et al. (2004) includes A. troschelii, these grouppresent species which show an overall lighter bodywith dark vertical bars and are laterally compressed;all are omnivorous, feeding mainly on zooplankton(Emery, 1973; Frédérich et al., 2009; present study).They are gregarious, their mouth is in a superiorposition, their habitat distributions are in a waterdepth in the range 1–15 m, and they are associatedwith rocky and coral reefs (Emery, 1973; Allen, 1991;Frédérich et al., 2009; present study).

Previous molecular phylogenetic analyses foundthat Microspathodon, Hypsypops, Nexilosus, and Ste-gastes are closely related genera belonging to thesubfamily Stegastinae (Tang, 2001; Quenouille et al.,2004; Cooper et al., 2009), although relationships arenot fully resolved (Fig. 2). The trophic data of thepresent study revealed that all these species arebenthic feeders, mainly grazing filamentous algae.Our geometric morphometric analyses and pheno-gram clearly distinguish the two Microspathodonspecies from the other species belonging to thistrophic group (i.e. H. rubicundus, N. latifrons, and allstudied Stegastes species) based on their highly flat-tened cephalic profile. Hypsypops rubicundus and N.latifrons are predominantly found in temperatewaters, in the extreme north and south, respectively,of the Eastern Pacific damselfish distribution. Theircephalic profiles are relatively similar to those ofStegastes species, although geometric data showsome differences among these three genera. In the



morphospace, H. rubicundus and N. latifrons speciesare placed peripheral of the Stegastes morphologicaldistribution (Fig. 5) and a schematic representation ofthe observed differences within this group is illus-trated in Figure 9. If we hypothesize that Nexilosus,Hypsypops, and Microspathodon are derived from thegenus Stegastes (Cooper et al., 2009), the results ofthe present study demonstrate that the main shapevariation during the evolutionary process was relatedto the form of the cephalic profile (pointed versusflattened), the size of the eye, and the position of theeye and mouth. From a Stegastes ancestor, a firstevolutionary step could produced two new forms(Fig. 9): (1) H. rubicundus with an pointed profile, along horizontal distance between the mouth and theeye, and a short vertical distance between the mouthand the eye, and (2) N. latifrons with a more roundedprofile, a short horizontal distance between the mouthand the eye, and a long vertical distance between themouth and the eye. The possible next evolutionarystep from the Nexilosus brand produced a stronglyspecialized cephalic profile: both Microspathodonspecies present a highly flattened profile, no horizon-tal space between the eye and the mouth and highvertical distance between the mouth and the eye(Fig. 9).

The molecular data reported by Tang (2001) showedthat Hypsypops is closer to Parma (Subfamily Stegas-tinae) than to Stegastes. Indeed, when the genusParma was first described, H. rubicundus wasincluded as one of its members (Tang, 2001). Thesetwo genera are predominantly distributed in temper-ate cool water and rocky reefs; furthermore, Parmaincludes some species that can be found in kelpforests, such as H. rubicundus (Allen, 1991; Buckle &Booth, 2009). Future geometric studies should includeParma species to obtain a good overview of the phe-notypic differences or similarities among Parma,Hypsypops, and Stegastes. The phylogenetic positionof N. latifrons is unknown (Cooper et al., 2009),although the data of the present study stronglysuggest that N. latifrons is closely related to Stegastesspecies in morphology and ecology, even though theformer lives in small groups. According to Robertson& Allen (2008), the main differences between thisgenus and Stegastes are that the margins of thepreopercle and infraorbital bones are smooth in Nex-ilosus and serrated in Stegastes. Both genera presenta single row of teeth and the number of dorsal finspines of Nexilosus (XIII) is included in the range ofStegastes (XII to XIV) (Allen, 1991). Similar to thequestion of synonymy between the genus Azurina andChromis, the close relationship between Nexilosus,Stegastes, and Hypsypops allows us to questionwhether Nexilosus and Hypsypops (both monospecificgenera: N. latifrons and H. rubicundus, respectively)

Figure 9. Schematization of the variation related tothe position of the eye and mouth in the cephalic profileof (A) Hypsypops rubicundus, (B) Stegastes flavilatus,(C) Nexilosus latifrons, and (D) Microspathodon dorsalis.



should be considered as valid genera. Further exhaus-tive phylogenetic studies are needed to better under-stand the phyletic relationships among Microspa-thodon, Hypsypops, Nexilosus, Stegastes, and Parma.

Our morphometric and phylogenetic analyses of thecephalic region show that the subfamilies Stegastinaeand Chrominae present a specific morphologicalpattern, but not Abudefdufinae (Fig. 7). The mor-phological average appeared twice throughout themolecular phylogeny (Fig. 7B): (1) Stegastinaesubfamily, in Stegastes genus, N. latifrons, and H.rubicundus, and (2) in Abudefdufinae subfamily, inA. concolor and A. declivifrons. In these groups, themean cephalic shape is related to medium-highvalues of the TROPH index. Nevertheless, species ofthe genus Chromis showed the same TROPH values.Consequently, A. concolor and A. declivifrons pre-sent convergent morphological adaptations to themembers of the subfamily Stegastinae in the familyin the Eastern Pacific.

In conclusion, the present study shows that varia-tion in the cephalic shape of Pomacentridae of theEastern Pacific can be clearly explained by differencesin diet and trophic behaviour. Shape and size may beviewed as two independent evolutionary factorsexplaining the diversification of Pomacentridae.Cepahlic shape is a significant predictor of trophichabit. Finally, the morphological groups discovered byour morphometric analyses partially agree with themain clades delimited by molecular phylogenetichypothesis. Consequently, the cephalic profile of dam-selfishes shows a clear and strong historical (phylo-genetic) signal only for the Stegastineae and partiallyfor Chrominae. Members of the subfamily Abudefdu-finae show convergences of cephalic shape. Themainly zooplanktivorous A. troschelii is very similarto Chromis species (Chrominae), and the mainly algalfeeders A. concolor and A. declivifrons are verysimilar to Microspathodon species (Stegastinae).

ACKNOWLEDGEMENTS

We thank the Consejo Nacional de Ciencia y Tec-nología (CONACYT, México) for a doctoral scholar-ship support to A.M.-R. to develop this work. We alsothank Lucia Campos Dávila (CIBNOR), Sandra J.Raredon (USNM), H. J. Walkers (SIO), Philip A.Hastings (SIO), Victor Cota Gómez (CICIMAR), JoséDe La Cruz Agüero (CICIMAR), and Rick Feeney(LACM) for their help with the museum collections.Thanks also to Fernando Aranceta Garza, IsmaelMascareñas Osorio, Juan J. Ramírez Rosas, MarioCota Castro, and Enrique Calvillo (all from CIBNOR)for their enthusiastic assistance on field trips. Thiswork was supported in part by the projects EP2 andEP3 of the Centro de Investigaciones Biológicas del

Noroeste, México (CIBNOR) and CT001 (CONABIO).Finally, we thank Mark Webster and two anonymousreviewers for their enlightening comments.

REFERENCES

Aburto-Oropeza O, Balart EF. 2001. Community structureof reef fish in several habitats of a rocky reef in the Gulf ofCalifornia. Marine Ecology 22: 283–305.

Albertson RC, Streelman JT, Kocher TD. 2003.Directional selection has shaped the oral jaws ofLake Malawi cichlid fishes. Proceedings of the NationalAcademy of Sciences of the United States of America 100:5252–5257.

Alexander RMcN. 1967. The functions and mechanisms ofthe protrusible upper jaws of some acanthopterygian fish.Journal of Zoology 151: 43–64.

Allen GR. 1991. Damselfishes of the world. Melle, Germany:Mergus Press, 271.

Allen GR, Woods LP. 1980. A review of the damselfish genusStegastes from the Eastern Pacific with description of a newspecies. Records of the Western Australia Museum 8: 171–198.

Alvarez-Filip L, Reyes-Bonilla H, Calderon-Aguilera LE.2006. Community structure of fishes in Cabo Pulmo Reef,Gulf of California. Marine Ecology 27: 253–262.

Angel A, Ojeda PF. 2001. Structure and trophic organizationof subtidal fish assemblages on the northern Chilean coast:the effect of habitat complexity. Marine Ecology ProgressSeries 217: 81–91.

Barel CDN. 1983. Towards a constructional morphology ofcichlid fishes (Teleostei, Perciformes). Netherlands Journalof Zoology 33: 357–424.

Barneche DR, Floeter SR, Ceccarelli DM, Frensel DMB,Dinslaken DF, Mário HFS. 2009. Feeding macroecology ofterritorial damselfishes (Perciformes: Pomacentridae).Marine Biology 156: 289–299.

Bellwood DR. 1996. The Eocene fishes of Monte Bolca:the earliest coral reef fish assemblage. Coral Reefs 15:11–19.

Bellwood DR, Sorbini L. 1996. A review of the fossil recordof the Pomacentridae (Teleostei: Labroidei) with a descrip-tion of a new genus and species from the Eocene of MonteBolca, Italy. Zoological Journal of the Linnean Society 117:159–174.

Bookstein FL. 1991. Morphometric tools for landmark data– geometry and biology. Cambridge: University Press.

Bray RN, Miller AC, Johnson S, Krause PR, RobertsonDL, Westcott AM. 1988. Ammonium excretion by macro-invertebrates and fishes on a subtidal rocky reef in southernCalifornia. Marine Biology 100: 21–30.

Buckle EC, Booth DJ. 2009. Ontogeny of space use and dietof two temperate damselfish species, Parma microlepis andParma unifasciata. Marine Biology 156: 1497–1505.

Campos-Dávila L, Cruz-Escalona VH, Galván-Magaña F,Abitia-Cárdenas F, Gutiérrez-Sánchez FJ, Balart EF.2005. Fish assemblages in a Gulf of California MarineReserve. Bulletin of Marine Sciences 77: 347–362.



Cardini A, Elton S. 2008. Does the skull carry a phyl-ogenetic signal? Evolution and modularity and in theguenons. Biological Journal of the Linnean Society 93: 813–834.

Ciardelli A. 1967. The anatomy of the feeding mechanismand the food habits of Microspathodon chrysurus (Pisces:Pomacentridae). Bulletin of Marine Sciences 17: 843–883.

Collar DC, Near TJ, Wainwright PC. 2005. Comparativeanalysis of morphological diversity: does disparity accumu-late at the same rate in two lineages of centrarchid fishes?Evolution 59: 1783–1794.

Cooper JG. 1863. On new genera and species of Californiafishes-Number I. Proceedings of the California Academy ofNatural Sciences, First Series 3: 70–77.

Cooper WJ, Smith LL, Westneat MW. 2009. Exploring theradiation of a diverse reef fish family: phylogenetics of thedamselfishes (Pomacentridae), with new classificationsbased on molecular analyses of all genera. Molecular Phy-logeny and Evolution 52: 1–16.

Cooper WJ, Westneat MW. 2009. Form and function ofdamselfish skull: rapid and repeated evolution into a limitednumber of trophic niches. BMC Evolutionary Biology 9:24.

Costa C, Cataudella S. 2007. Relationship between shapeand trophic ecology of selected species of Sparids ofthe Caprolace coastal lagoon (Central Tyrrhenian Sea).Environmental Biology of Fishes 78: 115–123.

Coughlin DJ, Strickler JR. 1990. Zooplankton capture by acoral reef fish: an adaptative response to evasive prey.Environmental Biology of Fishes 29: 35–42.

Cuvier G, Valenciennes A. 1833. Histoire naturelle despoisons. Paris: 9 FG Levrault.

Davis JC. 1986. Statistics and data analysis in geology.New York, NY: John Wiley & Sons.

Emery AR. 1973. Comparative ecology and functional oste-ology of fourteen species of damselfish (Pisces: Pomacen-tridae) at Alligator Reef, Florida Keys. Bulletin of MarineSciences 23: 649–770.

Espinoza M, Salas E. 2005. Estructura de las comunidadesde peces de arrecife en las Islas Catalinas y Playa Ocotal,Pacífico Norte de Costa Rica. Revista de Biología Tropical53: 523–536.

Evermann BW, Radcliffe L. 1917. The fishes of the westcoast of Peru and the Titicaca Basin. Bulletin United StatesNatural Museum 95: 1–166.

Ferry-Graham LA, Wainwright PC, Westneat MW, Bell-wood DR. 2002. Mechanisms of benthic prey capture inwrasses (Labridae). Marine Biology 141: 819–830.

Fowler HW. 1944. Results of the Fifth George VanderbiltExpedition (1941) (Bahamas, Caribbean sea, Panama, Gal-apagos Archipelago and Mexican Pacific Islands). TheFishes. Monographs of the Academy of Natural Sciences ofPhiladelphia 6: 57–529.

Frédérich B, Fabri G, Lepoint G, Vandewalle P, Par-mentier E. 2009. Trophic niches of thirteen damselfishes(Pomacentridae) at the Grand Récif of Toliara, Madagascar.Ichthyological Research 56: 10–17.

Frédérich B, Parmentier E, Vandewalle P. 2006. A pre-

liminary study of development of the buccal apparatus inPomacentridae (Teleostei, Perciformes). Animal Biology 56:351–372.

Frédérich B, Pilet A, Parmentier E, Vandewalle P.2008. Comparative trophic morphology in eight species ofdamselfishes (Pomacentridae). Journal of Morphology 269:175–188.

Froese R, Pauly D. 2000. FishBase 2000: concepts, designand data sources. ICLARM, Los Baños, Laguna. Availableat: http://www.fishbase.org

Garland T Jr, Dickerman AW, Janis CM, Jones JA. 1993.Phylogenetic analysis of covariance by computer simulation.Systematic Biology 42: 265–292.

Gilbert CH. 1892. Scientific results of explorations by the U.S. Fish Commission steamer ‘Albatross.’ 22. Descriptions ofthirty-four new species of fishes collected in 1888 and 1889,principally among the Santa Barbara Islands and in theGulf of California. Proceedings of the United States NaturalMuseum 14: 539–566.

Gill TN. 1862. Catalogue of the fishes of Lower Californiain the Smithsonian Institution, collected by Mr. J. Xantus.Proceedings of the Academy of Natural Sciences of Philadel-phia 14: 140–151.

Girard C. 1854. Observations upon a collection of Fishesmade on the Pacific coast of the United States, by Liut WP,Trowbridge USA, for the Museum of the Smithsonian Insti-tution. Proceedings of the Academy of Natural Sciences ofPhiladelphia 7: 142–156.

Gluckmann I, Vandewalle P. 1998. Morphofunctionalanalysis of the feeding apparatus in four Pomacentridaespecies: Dascyllus aruanus, Chromis retrofasciata, Chry-siptera biocellata and C. unimaculata. Italian Journal ofZoology 65: 421–424.

Goloboff PA, Farris JS, Nixon KC. 2008. T.N.T. Freeprogram for phylogenetic analysis. Cladistics 24: 774–786.

Goloboff PA, Mattoni CL, Quinteros AS. 2006. Continuouscharacters analyzed as such. Cladistics 22: 589–601.

Greenfield DW, Woods LP. 1980. Review of the deep-bodiedspecies of Chromis (Pisces: Pomacentridae) from theEastern Pacific, with descriptions of three new species.Copeia 1980: 626–641.

Grove JS, Gerzon D, Saa MD, Straing C. 1986. Dis-tribución y ecología de la familia Pomacentridae (Pisces)en las Islas Galápagos. Revista de Biología Tropical 34:127–140.

Grove JG, Lavenberg RJ. 1997. The fishes of the GalapagosIslands. Stanford, CA: Stanford University Press.

Guill JM, Heins DC, Hood CS. 2003. The effect of phylog-eny on interspecific body shape variation in darters (Pisces:Percidae). Systems Biology 52: 488–500.

Hammer Ø, Harper DAT, Ryan PD. 2001. PAST: Palento-logical Statistics software package for education and dataanalysis. Palentolologia Electronica 4(1): 9pp.

Heller E, Snodgrass RE. 1903. Papers from the HopkinsStanford Galapagos expedition, 1898–1899. XV. New fishes.Proceedings of the Washington Academy of Sciences 5: 189–229.

Hixon MA. 1981. An experimental analysis of territoriality in



the California reef fish Embiotoca jacksoni (Embiotocidae).Copeia 1981: 653–665.

Hobson ES. 1965. Diurnal-nocturnal activity of some inshorefishes in the Gulf of California. Copeia 1965: 291–302.

Jordan DS, Evermann BW. 1898. The fishes of North andMiddle America: a descriptive catalogue of the species offish-like vertebrates found in the waters of North America,north of the Isthmus of Panama. Part II. Bulletin of UnitedStates Natural Museum 47: 1241–2183.

Jordan DS, Gilbert CH. 1880. Description of a new flounder(Platysomatichthys stomias), from the coast of California.Proceedings of United States Natural Museum 3: 301–303.

Jordan DS, McGregor RC. 1898. List of fishes collected atthe Revillagigedo archipelago and neighboring islands In:Jordan DS, Evermann BW. 1898. The fishes of North andMiddle America: a descriptive catalogue of the species offish-like vertebrates found in the waters of North America,north of the Isthmus of Panama. Part II. Bulletin of UnitedStates Natural Museum 47: 1241–2183.

Konow N, Bellwood DR. 2005. Prey–capture in Pomacan-thus semicirculatus (Teleostei, Pomacanthidae): functionalimplications of intramandibular joints in marine angel-fishes. Journal of Experimental Biology 208: 1421–1433.

Kuo SR, Shao KT. 1991. Feeding habits of damselfish(Pomacentridae) from the southern part of Taiwan. Journalof the Fisheries Society of Taiwan 18: 165–176.

Lauder GV. 1980. The suction feeding mechanism insunfishes (Lepomis): an experimental analysis. Journal ofExperimental Biology 88: 49–72.

Lauder GV, Lanyon LE. 1980. Functional anatomy offeeding in the Bluegill sunfish Lepomis macrochirus: in vivomeasurement of bone strain. Journal of ExperimentalBiology 84: 33–55.

Lawing AM, Polly PD. 2009. Geometric morphometrics:recent applications to the study of evolution and develop-ment. Journal of Zoology 280: 1–7.

Liem KF. 1979. Modulatory multiplicity in the feedingmechanism in cichlid fishes, as exemplified by the inverte-brate pickers of Lake Tanganika. Journal of Zoology 189:93–125.

Liem KF. 1980. Adaptive significance of intra- and interspe-cific differences in the feeding repertoires of cichlid fishes.American Zoologist 20: 295–314.

Liem KF. 1990. Aquatic versus terrestrial feeding modes:possible impacts on the trophic ecology of vertebrates.American Zoologist 30: 209–221.

Liem KF. 1993. Ecomorphology of the Teleostean skull. In:Hanken J, Hall BK, eds. The skull: functional and evolu-tionary mechanisms. Chicago, IL: The University of ChicagoPress, 422–452.

Mabuchi K, Miya M, Azuma Y, Nishida M. 2007. Indepen-dent evolution of the specialized pharyngeal jaw apparatusin cichlid and labrid fishes. BMC Evolutionary Biology2007: 7–10.

Maddison WP, Maddison DR. 2010. Mesquite: a modularsystem for evolutionary analysis, Version 2.73. Available at:http://mesquiteproject.org

Midford PE, Garland JT, Maddison WP. 2010.

PDAP:PDTREE Package for mesquite, 2.73. Available at:http://mesquiteproject.org/pdap_mesquite/

Montgomery WL. 1980. Comparative feeding ecology of twoherbivorous damselfishes (Pomacentridae: Teleostei) fromthe Gulf of California, Mexico. Journal of ExperimentalMarine Biology and Ecology 47: 9–24.

Nichols JT. 1924. A contribution to the ichthyology of theGalapagos. Zoologica 5: 63–65.

Norton SF, Luczkovich JJ, Motta PJ. 1995. The role ofecomorphological studies in the comparative biology offishes. Environmental Biology of Fishes 44: 287–304.

Núñez L, Vásquez J. 1987. Observaciones tróficas y dedistribución espacial de peces asociados a un bosque sub-mareal de Lessonia trabeculata. Estudios en Oceanologia 6:79–85.

Pauly D, Froese R, Sa-a PS, Palomares ML, ChristensenV, Rius J. 2000. Trophlab manual. Manila: ICLARM.

Petersen CW, Marchetti K. 1989. Filial cannibalism in theCortez Damselfish Stegastes rectifraenum. Evolution 43:58–168.

Quenouille B, Bermingham E, Planes S. 2004. Molecularsystematics of the damselfishes (Teleostei: Pomacentridae):Bayesian phylogenetic analyses of mitochondrial andnuclear DNA sequences. Molecular Phylogenetics andEvolution 31: 62–68.

Randall JE. 1967. Food habits of reef fishes of the WestIndies. Studies in Tropical Oceanography 5: 665–847.

Robertson DR, Allen GR. 2008. Shorefishes of the TropicalEastern Pacific online information system, Version 1.0.Smithsonian Tropical Research Institute, Balboa, Panamá.Available at: http://www.neotropicalfishes.org/sftep, http://www.stri.org/sftep

Rohlf FJ. 1993. Relative warps analysis and an exampleof its application to mosquito wings. In: Marcus LF, Bello E,Garcia-Valdecasas A, eds. Contributions to morphometrics.Madrid: Monografias del Museo Nacional de CienciasNaturales, CSIC, 131–159.

Rohlf FJ, Marcus LF. 1993. A revolution in morphometrics.Trends in Ecology and Evolution 8: 129–132.

Rohlf FJ, Slice D. 1990. Extension of the Procrustes methodfor the optimal superposition of landmarks. Journal ofSystematic Zoology 39: 40–59.

Rothans TC, Miller AC. 1991. A link between biologicallyimported particulate organic nutrients and the detritus foodweb in reef communities. Marine Biology 110: 145–150.

Stergiou KI, Karpouzi VS. 2002. Feeding habits and trophiclevels of Mediterranean fish. Reviews in Fish Biology andFisheries 11: 217–254.

Streelman JT, Karl SA. 1997. Reconstructing labroid evo-lution using simple-copy nuclear DNA. Proceedings of theRoyal Society of London Series B, Biological Sciences 264:1011–1020.

Svanbäck R, Eklöv P. 2002. Effects of habitat andfood resources on morphology and ontogenetic growthtrajectories in perch. Oecologia 131: 61–70.

Tang KL. 2001. Phylogenetic relationships amongdamselfishes (Teleostei: Pomacentridae) as determined bymitochondrial DNA data. Copeia 2001: 591–601.



Thompson DA. 1917. On growth and form. Cambridge:Cambridge University Press.

Trapani J. 2001. Position of developing replacement teeth inteleost. Copeia 2001: 35–51.

Tschudi JJ. 1846. Ichthyologie. Pp. ii–xxx + 1–35 Pls 1–6In: Scheitlin Zolliskofer, ed. Untersuchungen über die FaunaPeruana. Scheitlin & Zollikofer, 1–693.

Valenciennes A. 1833. Poissons. In: Cuvier G, ValenciennesA, eds. 1833. Histoire naturelle des poisons. Paris: 9 FGLevrault.

Wainwright PC. 1996. Ecological explanation through func-tional morphology: the feeding biology of sunfishes. Ecology77: 1336–1343.

Wainwright PC, Bellwood DR, Westneat MW, GrubichJR, Hoey AS. 2004. A functional morphospace for the skull

of labrid fishes: patterns of diversity in a complex biome-chanical system. Biological Journal of the Linnean Society82: 1–25.

Wainwright PC, Ferry-Graham LA, Waltzek TB,Carroll AM, Hulsey CD, Grubich JR. 2001. Evaluatingthe use of ram and suction during prey capture bycichlid fishes. Journal of Experimental Biology 204: 3039–3051.

Wainwright PC, Richard BA. 1995. Predicting patterns ofprey use from morphology of fishes. Environmental Biologyof Fishes 44: 97–113.

Westneat MW. 1995. Feeding, function, and phylogeny:analysis of historical biomechanics and ecology in labridfishes using comparative methods. Systematic Biology 44:361–383.



Documents

Patterns of morphological evolution of the cephalic region in