Environmental biology of fishes, tập 94, số 4, 2012

This study documents the habitat associations of adult and juvenile butterflyfishes on an inshore reef of the Great Barrier Reef GBR to investigate if ontogenetic shifts in habitat use o

Ontogenetic shifts in the habitat associations

of butterflyfishes (F Chaetodontidae)

Nicholas J Clark&Garry R Russ

Received: 20 April 2011 / Accepted: 4 November 2011 / Published online: 18 January 2012

# Springer Science+Business Media B.V 2012

Abstract The habitat associations of species are vital

in determining an organism’s vulnerability to

envi-ronmental and anthropogenic stress In the marine

environment, post-settlement processes such as

onto-genetic shifts in habitat use can affect this

vulnerabil-ity by subjecting a species to differing biological and

environmental conditions at various life stages This

study documents the habitat associations of adult and

juvenile butterflyfishes on an inshore reef of the Great

Barrier Reef (GBR) to investigate if ontogenetic shifts

in habitat use occur, and if such shifts relate to the

trophic ecologies of species Coral-feeding species

displayed highly concordant distributions among

adults and juveniles In contrast, adults and juveniles

of species with wider dietary selectivities (generalists)

displayed significantly different distributions across

reef zones Juvenile generalist feeders were limited to

the shallow, patchy areas of the reef flat whilst adult

conspecifics displayed comparatively wide

distribu-tions Butterflyfishes with a heavy reliance on corals

for food appear to settle preferentially in areas with

high abundances of adult conspecifics, which may

partially explain why coral specialists are more

vulnerable to localized depletion events In contrast,

generalist species utilize distinct habitats as adults and

juveniles, suggesting that generalist butterflyfishesexpand their ranges and are therefore subjected tochanging environmental conditions as they reachadulthood

Keywords Habitat association Ontogeny Abundance Coral reef Butterflyfish Chaetodontidae

Introduction

The larvae of coral reef fishes utilize a suite ofsensory cues to settle into specific regions of the reef(Stobutzki1998; Leis and Carson-Ewart2003), areasthat may differ from habitats occupied by adultconspecifics (Mumby 2006) Recruits and juveniles

of reef fish often utilise distinct habitats compared toadults (Jones et al.2010) and thus each life-stage may

be subject to different ecological and environmentalinfluences Moreover, alterations to benthic composi-tion and complexity, for example coral bleachingevents (Graham et al.2009), may modify recruitmentpatterns (Moore and Elmendorf 2006; Feary et al

2007) Thus, post-settlement processes, such asontogenetic shifts in habitat use, may influence aspecies’ vulnerability to stress Therefore, we canbetter understand how disturbances may affect reeffish population dynamics by determining the habitatassociations of vulnerable species

Butterflyfishes (Family Chaetodontidae) are aconspicuous, diverse group of fishes that exhibit

DOI 10.1007/s10641-011-9964-2

School of Marine & Tropical Biology and the ARC Centre

for Coral Reef Studies, James Cook University,

Townsville QLD 4811, Australia

e-mail: nicholas.clark@my.jcu.edu.au

strong associations with coral reefs (Pratchett et al.

2008b) The heavy reliance of these fish on coral reefs

is due to an abundance of coral-feeding species, with

butterflyfishes making up 61% of all corallivorous

fishes (Bellwood et al 2010) Moreover, several

species exhibit considerable feeding selectivity,

pref-erentially feeding on only a few species of coral

(Pratchett 2007) despite relatively wide geographic

distributions (Allen et al 2003; Froese and Pauly

2008) Consequently, butterflyfishes often show

marked decreases in abundance following disturbance

to coral reefs (Jones et al.2004; Pratchett et al.2006)

Lower abundances could in turn lead to reduced larval

production and decreased recruitment (Donelson et al

2008), exacerbating population declines Given their

strong ties to reefs and high vulnerability to

distur-bance (Wilson et al 2006; Pratchett et al 2008a),

butterflyfishes are a model study group to explore

ontogenetic variation in habitat association on coral reefs

The habitat associations of butterflyfishes indicate

a heavy reliance on living corals as adults and variable

associations with coral as juveniles (Harmelin-Vivien

1989; Pratchett et al 2008b) This suggests that

some species of butterflyfish may display ontogenetic

shifts in habitat In addition to ontogenetic shifts in

habitat association, the distributions of juvenile reef

fishes may also be strongly regulated by both

inter-and intra-specific competition (e.g., Munday 2001)

For instance, in a study conducted at Lizard Island on

the mid shelf of the Great Barrier Reef (GBR),

Berumen and Pratchett (2006) found that dominant

butterflyfish competitors such as Chaetodon

baron-essa aggressively defended territories against both

conspecifics and subordinate species and increased

their territory sizes as abundance of coral prey

declined This propensity for aggression, particularly

against conspecifics, may directly influence the

settle-ment preferences and spatial distributions of juveniles

Therefore, some recruits may be forced to settle into

sub-optimal habitats due to antagonistic interactions

with adults (e.g., Munday2001) In particular,

compe-tition among coral feeders may result in intraspecific

spatial partitioning in areas with relatively low

cover of commonly preferred Acropora and

Pocil-lopora corals, such as the inshore reefs of the GBR

(Emslie et al 2010) Therefore, the objective of this

study was to document the distributions and habitat

associations of adult and juvenile butterflyfishes on

inshore fringing reefs of the GBR It was predicted

that coral-feeding butterflyfishes would display genetic variation in habitat association in response tolow prey availability For generalist feeders, whichexhibit higher dietary versatilities than coral feeders(Pratchett 2005), we also predicted an ontogeneticshift in habitat association Generalist feeders expandtheir range of prey items as they reach adulthood(Harmelin-Vivien1989), which has been suggested toresult in an expansion in habitat use upon maturity(Pratchett et al.2008b)

onto-Methods

Study sites

This study was conducted in September and October

2010 in the Palm Islands (18˚34′S, 146˚29′E) on theinshore GBR (Fig.1) The Palm Island group consists

of nine islands located approximately 15 km off themainland Surveys were made at three levels ofexposure to winds and currents (sheltered, obliquelyexposed, and exposed; Fig 1) to investigate ifexposure affected coral and butterflyfish communities(following Pratchett et al 2008b) The degree ofexposure was determined according to multipleobservations of surge, wind speed, and the overalldirection of currents during the study Sites with fullprotection from prevailing winds were consideredsheltered, sites in which the reef faced parallel tooncoming surge and currents were consideredobliquely exposed, and sites that faced directly intosurge, currents, and prevailing winds were consideredexposed The fringing reefs of the Palm Islands have

an extensive shallow reef flat and mild reef slopegradient at sheltered sites, and a relatively narrow reefflat and steeply sloping reef wall at exposed sites.However, despite the variation in gradients of the reefslope between exposed and sheltered sites, thephysiognomic reef zones of flat, crest, and slope aretypically distinguishable at each location At alllocations, the reef flat has patchy coral cover andextensive areas of rubble, sand, and algae-coveredrock In contrast, the reef crest and reef slope arerelatively rugose areas consisting of large bommies(coral heads) and crevasses In general, the study sitesare characterized by high abundances of Alcyoniidsoft corals and scleractinian corals belonging to thefamily Poritidae (Emslie et al 2010)

Twenty-seven sites were surveyed for butterflyfish

abundance and benthic cover (Fig 1) Butterflyfish

abundance was estimated by underwater visual census

(UVC) along 50 m transects using a combination of

SCUBA and snorkel Surveys were made in the three

reef zones (flat, crest, and slope), as coral reef fishes

often occur in assemblages that are characteristic of

each physiognomic zone (Russ1984a,b) Study sites

were chosen to be relatively well-spread around the

coasts of Pelorus, Curacao and northern Orpheus Islands

(Fig 1) To maintain safe dive practice, study sites

were selected daily at sea based on weather

con-ditions Transects were delineated by a 50 m

fibre-glass tape laid parallel to the reef crest in each of the

three reef zones, with depths ranging from 1–2 m

along the flat and crest and from 4–10 m along the

slope The first observer counted butterflyfish within

1 m either side of their body while simultaneously

laying the transect tape Individual butterflyfish were

visually placed into one of three size classes [<3 cm, 3–

10 cm, and >10 cm total length (TL)] corresponding to

the life-stage categories of new recruit, juvenile, and adult

(following Fowler 1990 and Pratchett et al 2008b)

Size estimations were checked periodically

underwa-ter using a ruler printed on the back of the datasheets

Benthic composition was measured by a second

observer This permitted investigation of butterflyfish

habitat associations Live coral cover was estimated

using a variation of the line point intercept technique

(Lam et al 2006) A total of 100 randomly placed

points on the 50 m transect tape were sampled and

recorded into one of six categories (table Acropora,staghorn Acropora, Pocillopora, Porites, other hardcoral, and soft coral) The sample points were placed

on the tape using a randomly generated number tableprior to the survey period Three replicate transectswere surveyed in each zone at each site, yielding atotal of 243 transects across the 27 sites and a surveyarea of 24 300 m2

Analysis

Coral cover was analysed with an analysis of variance(ANOVA) to investigate if variation in butterflyfishabundance was related to spatial variation in coral cover

To test for variations in the distributions of adult andjuvenile butterflyfish conspecifics, a series of 3-factormultivariate analyses of variance (MANOVA) wereimplemented Prior to analysis, coral cover data andbutterflyfish abundance data were pooled at the site leveland standardised by area surveyed Butterflyfish abun-dances were also pooled by trophic guild (i.e hard coral-feeders, soft coral-feeders, or generalists) to investigatethe role of trophic guild in determining ontogeneticvariation in habitat association All analyses of variancewere tested using a 3-factor orthogonal model, with island(random), exposure (fixed), and reef zone (fixed) included

as the test variables Butterflyfish abundance data weresquare-root transformed and benthic cover data werearcsine-root transformed to meet (M)ANOVA assump-tions of normality, sphericity and homoscedasticity ofvariance Tukey’s HSD post-hoc tests were used toidentify homogenous groups following all analyses ofvariance Rather than performing correlations between

Australia

sites exposed sites Pelorus

Fig 1 Map of sampling

sites within the Palm

Islands on the inshore Great

Barrier Reef

butterflyfish and coral taxa, a principal components

analysis (PCA) was used to display variation in the

community structure of both butterflyfish and corals The

analysis was performed using the correlation matrix

generated from the transformed butterflyfish and benthic

cover data Statistical analyses were performed using

Statistica 10.0 (www.statsoft.com)

Results

The butterflyfish assemblage

A total of 1409 individual butterflyfish were recorded

on the 243 transects The majority of individuals

recorded were adults or juveniles (Table 1) Recent

recruits were rare (except for C aureofasciatus,

accounting for 48 of 50 total recruits) Thus, this

study only quantifies habitat associations of adult and

juvenile butterflyfish Obligate hard-coral feederswere the most abundant trophic guild, accountingfor 60.7% of individuals Generalist and soft-coralfeeders accounted for the remaining 39.3% Due tothe relatively low numbers of juveniles recorded forseveral species, only the six most abundant species(Chaetodon aureofasciatus, C rainfordi, C melanno-tus, C lunulatus, C vagabundus, and Chelmonrostratus; 91.5% and 87.9% of juvenile and adultrecordings, respectively) were included in the PCA

Benthic composition

Among the 27 study sites, mean scleractinian (hard) coralcover was 17.4±1.3% In addition, hard coral cover wasvariable among reef zones (Fig 2) However, thisvariation was not statistically significant (ANOVA,Table2) Most hard corals belonged to the‘other hardcorals’ category (mean 8.5±1.6%; Fig 2) Corals

Table 1 Total abundances of butterflyfish (displayed by size class) within the Palm Islands Totals are derived from 243 transects (HC), hard-coral feeder; (SC), soft-coral feeder; (G), generalist feeder

Mean abundance per site

belonging to the genus Porites were relatively

abundant (mean 6.7±1.2%), while both Acropora

(staghorn and table morphologies) and Pocillopora

were in much lower abundance (mean cover of 1.6±

0.04% and 0.6±0.01%, respectively; Fig 2) Table

Acropora was rare in all locations However, overall

cover of table Acropora was strongly related to

exposure, peaking along the exposed reef flat of

Orpheus and Pelorus Islands (Table 2) where waveaction and surge were strongest Pocillopora and tableAcropora were in relatively low abundance onCuracoa Island, where soft corals were abundant(Table 3) Porites species showed no patterns ofabundance among islands (Table 2), having peakcover in sheltered locations along the crest and slope(Fig 2) Soft coral was the dominant benthiccomponent surveyed, with a mean cover of 29.3±2.2% among sites (Fig 2) Nevertheless, the totalcover of both hard and soft corals was variable acrossstudy sites Total hard coral cover was lower inobliquely exposed sites than either sheltered orexposed sites (ANOVA, Table 2) In contrast, softcoral cover was higher in obliquely exposed sites(ANOVA, Table3), where extensive beds of soft coralwere frequently observed Additionally, soft coral wasmore abundant on Curacoa Island than either Orpheus

or Pelorus Islands (ANOVA, Tables2and3) No otherbenthic components varied across islands The remain-ing benthic categories surveyed (Pocillopora, staghornAcropora, and other hard corals) exhibited no variationamong zones or exposures (ANOVA, Table2)

Habitat associations of adult vs juvenile butterflyfish

Butterflyfish abundances varied considerably amonglocations and benthic compositions Comparisonsamong adults and juveniles revealed significantintraspecific variation in the distributions of generalistfeeders and the soft coral feeder C melannotus Forthe generalist feeders, ontogenetic variation in the use

of reef zones occurred (MANOVA, Table 4) Adultgeneralists were abundant in all reef zones, whilejuveniles were typically recorded only in the shallows

of the reef flat [Fig.3(c), (d)] In addition, abundances

of adult and juvenile generalists did not varysignificantly according to either island or level ofexposure (MANOVA, Table 4) For C melannotus,adult and juvenile distributions varied across expo-sures Adults of C melannotus were significantlyassociated with exposed and obliquely exposedlocations (MANOVA, Tukey’s Homogenous Groups;Table 4) while juveniles exhibited no clear patternamong locations (Fig 4) However, adults andjuveniles of C melannotus were both significantlymore abundant on Curacoa Island than on eitherOrpheus or Pelorus Islands (MANOVA, Tukey’sHomogenous Groups; Tables 1 and 4), showing a

Fig 2 Mean cover of benthic components within the Palm

Islands according to reef zone and exposure a flat; b crest; c

slope TA, table Acropora; SA, staghorn Acropora; POC,

Pocillopora; POR, Porites; OHC, other hard coral; THC, total

similar abundance pattern across islands to that of soft

coral cover In contrast to generalists and soft coral

feeders, hard coral feeding butterflyfishes showed no

significant variation in the distributions of adults and

juveniles across islands, exposures, or reef zones

(MANOVA, Table 4) Both adult and juvenile coral

feeders were significantly more abundant in sheltered

sites than either obliquely exposed or exposed sites

(MANOVA, Tukey’s Homogenous Groups: Table4),

showing a similar pattern to that of hard coral cover

Both hard- and soft-coral feeders showed no clear

intraspecific pattern in abundance across reef zones

[Fig.3(a), (b)]

The PCA revealed some intraspecific variation in

benthic association for the six most abundant

butterfly-fishes (Fig.5) For two of the coral-feeding species (C

aureofasciatus and C lunulatus), adults and juveniles

displayed comparable intraspecific distributionsamong reef locations (Fig 4) However, juveniles ofboth C aureofasciatus and C lunulatus differedsubstantially in their associations with benthic com-ponents compared to adult conspecifics (Fig.5) For C.aureofasciatus, adult abundance was positively corre-lated with several benthic categories (namely Porites,other hard corals, and soft corals), while juvenileswere primarily associated with staghorn Acropora(Fig 5) In contrast, adults of C lunulatus werepredominately associated with staghorn Acropora andPocillopora, while juveniles were more common inareas with high coverage of Porites (Fig.5)

In contrast to C aureofasciatus and C lunulatus,the remaining four species analysed displayed similarcorrelations among age groups and benthic categories(C rostratus, C vagabundus, C melannotus, and C

Table 2 3-factor ANOVA results comparing benthic

compo-nents among locations Island (random), exposure (fixed), and

reef zone (fixed) were included as test variables Results are

based on mean cover of benthic components, pooled across

sites Benthic cover values were arcsine-root transformed to meet ANOVA assumptions Numerical values are F-statistics.

Island (random:

Orpheus, Pelorus, Curacoa)

Exposure (fixed:

sheltered, obliquely exposed, exposed)

Island X Exposure

Zone (fixed:

flat, crest, slope)

Island X Zone

Exposure X Zone

Island X Exposure X Zone

Table 3 Mean cover of benthic

components at each study

island within the Palm Islands

rainfordi; Fig 5.) For the two generalists (C

vaga-bundus and C rostratus), adults and juveniles

displayed similar associations with the benthos

despite contrasting distributions across zones

(Fig 4), though juveniles of both species were more

strongly associated with the ‘other hard corals’

category than adults (Fig 5) Benthic associations

among adult and juvenile C melannotus (soft coral

feeder) were also similar, correlating primarily withsoft coral cover, though once again juveniles weremore strongly associated with‘other hard corals’ thanadults (Fig.5) For C rainfordi, an obligate hard coralfeeder, both adults and juveniles associated primarilywith table Acropora and Pocillopora species (Fig.5)

No species analysed was strongly associated withnon-coral benthos (Fig 5)

Table 4 3-factor MANOVA results from intraspecific

compar-isons of adult and juvenile butterflyfish abundances, pooled by

trophic guild Island (random), exposure (fixed), and reef zone

(fixed) were included as test variables Abundance and coral

cover data were pooled across sites and standardised by survey area Butterflyfish abundance data were square-root trans- formed to meet MANOVA assumptions Numerical values are

Island (random:

Orpheus, Pelorus, Curacoa)

Exposure (fixed:

sheltered, obliquely exposed, exposed)

Island X Exposure

Zone (fixed:

flat, crest, slope)

Island X Zone

Exposure X Zone

Island X Exposure X Zone

Fig 3 Mean abundances of adult and juvenile butterflyfishes,

pooled by trophic guild: (a) adult hard coral feeders; (b)

juvenile hard coral feeders; (c) adult generalist feeders; (d)

Fig 4 Mean abundances

(± SE) of adults (top graphs)

and juveniles (bottom

graphs) of the six most

abundant species of

butterflyfish in the Palm

Islands (F) flat; (C) crest;

(S) slope

Ontogenetic variation in the distributions of coral reef

fishes has been observed in a variety of species, with

evidence suggesting that dietary preference, the

availability of shelter, and the presence of adult

conspecifics may all contribute to this pattern of habitat

association (Lecchini and Galzin2005; Mumby2006;Jones et al 2010) The results of this study agreestrongly with a previous study of butterflyfishdistributions on coral reefs (Pratchett et al 2008b),indicating that some species of butterflyfish alsodisplay variations in their distributions as they mature.Furthermore, the patterns revealed in both studiesappear to relate strongly to trophic guild However, incontrast to the study by Pratchett et al.2008b, whichwas carried out on mid-shelf reefs of the Great BarrierReef (GBR) with relatively high abundance ofAcropora and Pocillopora corals, this study wasperformed on an inner-shelf GBR reef with signifi-cantly lower prey abundance for hard coral feeders(Emslie et al 2010) Thus, our results furtherhighlight the importance of trophic guild in determin-ing ontogenetic differences in butterflyfish distribu-tions For example, although Berumen and Pratchett(2006) noted an increase in territory size among adultcoral-feeders as abundance of coral prey declined, ourresults found that juvenile hard-coral feeders dis-played comparable distributions to adult conspecifics

in habitats dominated by soft coral This indicates thateither recruits of hard-coral feeders settle preferentiallyinto habitats occupied by adult conspecifics, or alterna-tively, that juveniles exhibit higher survival rates inthese areas (e.g., see Harmelin-Vivien1989; Pratchett

et al.2008b) However, hard-coral feeders were not theonly species to exhibit such limited ontogenetic varia-tion in abundance Adult and juvenile abundances of thesoft-coral feeder C melannotus were similar acrosslocations and ontogenetic variation was displayedonly according to degree of exposure In contrast tocoral feeders, generalist feeders displayed substantialontogenetic variation in distribution Juvenile gener-alist feeders were limited almost exclusively to theshallow, patchy areas along the reef flat, whilst adultsdisplayed comparatively wide distributions

Our original hypothesis of spatial variation inabundance of adult and juvenile generalists wassupported by our results (see Table 4; Fig.3) Thisontogenetic variation in distribution suggests a corre-lation between distribution and dietary versatility Theexpansion in dietary flexibility often shown bygeneralists may partially explain the relatively limiteddistributions of juveniles compared to adults Juve-niles typically feed only on benthic invertebrates,whilst adults have been shown to feed on a variety ofitems including scleractinian coral polyps and

Fig 5 Principal components analysis (PCA) showing

intra-specific variation in the associations of the six most abundant

butterflyfishes (a) with individual benthic components (b) in

the Palm Islands Capital letters indicate adults while lower

case letters indicate juveniles Species are as follows:

Chaetodon aureofasciatus, Chaetodon lunulatus, Chaetodon

rainfordi, Chaetodon vagabundus, Chaetodon melannotus, and

Chelmon rostratus

gastropods (Harmelin-Vivien 1989) As generalists

were restricted almost entirely to the reef flat as

juveniles, this study supports previous evidence that

immature generalist butterflyfishes often feed in

rubble or sandy areas of the shallows while adults

display comparatively wider distributions (Pratchett

et al 2008b) However, despite being limited to

relatively patchy habitats with low coral cover, the

abundances of C vagabundus and C rostratus

juveniles correlated strongly with the ‘other hard

corals’ category This suggests that the low abundance

of living corals was nevertheless important in

attract-ing immature individuals (e.g., see Feary et al.2007)

Moreover, areas devoid of living coral were typically

unoccupied by butterflyfish, regardless of trophic

guild Juvenile reef fishes often exhibit strong habitat

selection upon settlement (Williams and Sale 1981;

Doherty et al 1996; Feary et al 2007), and habitat

structure has been shown to be an important factor in

recruitment and post-settlement success among reef

fishes (Almany 2004) This study supports the

conclusion that living coral may be essential to the

recruitment of butterflyfish regardless of trophic guild

(Pratchett et al.2006)

For C melannotus, only exposure was a significant

factor in explaining ontogenetic variation in habitat

association Whilst generally labelled as a soft-coral

feeder, C melannotus does take a considerable

portion of bites from scleractinian corals (Pratchett

2005), a feeding preference that may or may not vary

with the onset of maturity For instance, the adult

distribution observed in this study strongly reflected

soft coral abundance, peaking at exposed and

semi-exposed locations In contrast, juveniles displayed a

stronger correlation with‘other hard corals’ than did

adults, and showed no variation in abundance

according to exposure This finding supports the

results of Pratchett et al 2008b, who found that C

melannotus juveniles displayed substantially different

distributions to adults at Lizard Island Whilst both

adults and juveniles of C melannotus displayed

similar patterns across islands to that of soft coral

cover in this study, it may be that juveniles rely more

heavily on hard corals than adults, either for food or

shelter Indeed, many reef fish species depend greatly

on live coral during settlement and early juvenile life

(Syms and Jones2000; Jones et al.2004) However, it

must be noted that this study did not determine the

feeding behaviours of butterflyfish, and thus we can

only speculate if C melannotus demonstrated a shift

in feeding preference with growth We thereforesuggest that future studies investigate ontogeneticdietary shifts in soft-coral feeding butterflyfish

As coral-feeding butterflyfish begin feeding onscleractinian corals almost immediately upon settlement(Harmelin-Vivien 1989), concordant patterns in thedistributions of adult and juvenile hard-coral feedersare to be expected Indeed, an abundance of residentadults may indicate high resource availability, andselection may favour individuals that settle in areaswith high abundances of preferred prey (e.g., seeStrathmann et al 2002) However, despite thesimilarities in juvenile and adult abundance displayed

by hard-coral feeders, some ontogenetic variation inbenthic association was displayed by the two mostabundant obligate corallivores, C aureofasciatus and

C lunulatus C aureofasciatus juveniles were ciated strongly with staghorn Acropora, while adultsdisplayed moderate correlations with a variety ofbenthic components On the other hand, juveniles of

asso-C lunulatus were mostly associated with Poritesspecies; however, adults were seen in abundance nearPocillopora and staghorn Acropora The differences

in benthic associations exhibited by C aureofasciatusand C lunulatus suggest that post-settlement forcesmay have affected the within-site distributions ofjuveniles Possible factors that may have contributed

to these differences are the availability of shelterwithin microhabitats (e.g., Almany 2004) and antag-onistic interactions between adults and juveniles (e.g.,Jones1987; Webster and Hixon2000; Ben-Tzvi et al

2008) Juvenile reef fishes often associate morestrongly with high-shelter areas than do their adultcounterparts to increase post-settlement survival(Mumby2006; Jones et al.2010) This may partiallyexplain why C aureofasciatus juveniles associatedstrongly with staghorn Acropora corals that provide ahigh degree of protection In addition, Berumen andPratchett (2006) suggest that both inter- and intra-specific competition may be important in determiningthe distributions of butterflyfish However, the lack ofsub-transect observations of fish and relatively lowsample sizes for individual species in this study preventsuch conclusions Although butterflyfish typically recruityear-round on coral reefs (Abesamis and Russ 2010),recruits are seen in highest abundance in the summermonths from January to April on the GBR (Leis

1989) As this study was performed during September

and October, we may have missed the peak of

recruitment This could explain why sample sizes

were biased towards adults and late-stage

juve-niles Therefore, we recommend future observations

of individual butterflyfish be carried out at different time

periods to elaborate on the trends shown in this study

Several conclusions from our results may prove

beneficial to the study of butterflyfish population

dynamics Firstly, the marked ontogenetic variations

in habitat association displayed by generalist feeders

indicate a distinction between habitats utilized by

adult and juvenile conspecifics This pattern of habitat

use suggests an ontogenetic shift between new recruit/

juvenile and adult habitats, providing valuable

eco-logical data to supplement genetic and demographic

analyses of population connectivity Such elucidations

can also be incorporated into refined larval dispersal

models, making future predictions regarding the

ability of populations to survive disturbances more

tangible (i.e., see Werner et al 2007; Jones et al

2009) Additionally, the concordant patterns of habitat

use displayed by adult and juvenile hard-coral feeders

suggest that the abundance of juveniles may be partly

related to the presence of conspecific adults

(Harmelin-Vivien 1989) rather than to habitat alone Such

dependence may partially explain the common lag in

recovery of butterflyfish populations following

declines in abundance after bleaching events (Jones

et al.2004; Pratchett et al.2008a) In fact, a range of

coral reef fishes has been shown to exhibit a delayed

response to recovery of reefs following disturbance

(Williams1986; Mclanahan et al.2002; Halford et al

2004; Graham et al 2007) This indicates that a

partial reliance of recruits on the presence of adults

may be a common phenomenon among reef fishes

In conclusion, this study indicates that the varying

degrees of ontogenetic shifts in the habitat

associa-tions of butterflyfish may be influenced by a number

of factors These may include trophic guild,

compe-tition, and the presence of adult conspecifics Our

results may prove useful in developing an

under-standing of the mechanisms contributing to

demo-graphic and genetic connectivity in butterflyfish

populations Ultimately, this work emphasizes the

need for detailed ecological observations in the study

of reef fish population dynamics

Simonson, J Kerry, J Hopf, D Buchler, N Summers, and

T Heintz for invaluable assistance in the field and five anonymous reviewers for comments that vastly improved the manuscript We are also thankful to H Burgess and the staff at OIRS for ongoing logistical support This research was supported by a grant to G.R.R from the Australian Research Council (ARC) Centre of Excellence for Coral Reef Studies.

References

Abesamis RA, Russ GR (2010) Patterns of recruitment of coral reef fishes in a monsoonal environment Coral Reefs doi: 10.1007/s00338-010-0653-y

Allen G, Steene R, Humann P, Deloach N (2003) Reef fish identification: tropical Pacific New World Publications, Jacksonville

Almany GR (2004) Differential effects on habitat complexity, predators and competitors on abundance of juvenile and

Bellwood DR, Klanten S, Cowman PF, Pratchett MS, Konow

N, van Herwerden L (2010) Evolutionary history of the butterflyfishes (f: Chaetodontidae) and the rise of coral

Ben-Tzvi O, Abelson A, Polak O, Kiflawi M (2008) Habitat selection and the colonization of new territories by

Berumen, ML & Pratchett, MS (2006) Effects of resource availability on the competitive behaviour of butterflyfishes (Chaetodontidae) Proceedings of the 10th International Coral Reef Symposium: 644–650

Doherty P, Kingsford M, Booth D, Carleton J (1996) Habitat selection before settlement by Pomacentrus coelestis Mar

Donelson JM, McCormick MI, Munday PL (2008) Parental condition affects early life-history of a coral reef fish J

Emslie MJ, Pratchett MS, Cheal AJ, Osborne K (2010) Great Barrier Reef butterflyfish community structure: the role of shelf position and benthic community type Coral Reefs

Froese R, Pauly D (2008) FishBase, World Wide Web

August 21, 2010 Graham NAJ, Wilson SK, Jennings S, Polunin NVC, Robinson

I, Bijoux JP, Daw TM (2007) Lag effects in the impacts of mass coral bleaching on coral reef fish, fisheries, and

Graham NAJ, Wilson SK, Pratchett MS, Polunin NVC, Spalding MD (2009) Coral mortality versus structural collapse as drivers of corallivorous butterflyfish decline.

Halford A, Cheal AJ, Ryan D, Williams DMcB (2004) Resilience

to large-scale disturbance in coral and fish assemblages on

the Great Barrier Reef Ecol 85:1892–1905

Harmelin-Vivien ML (1989) Implications of feeding

speciali-zation on the recruitment processes and community

Jones GP (1987) Competitive interactions among adults and

Jones GP, McCormick MI, Srinivasan M, Eagle JV (2004)

Coral decline threatens fish biodiversity in marine

Jones GP, Almany GR, Russ GR, Sale PF, Steneck RS, van

Oppen MJH, Willis BL (2009) Larval retention and

connectivity among populations of corals and reef fishes:

Jones DL, Walter JF, Brooks EN, Serafy JE (2010)

Connectiv-ity through ontogeny: fish population linkages among

mangrove and coral reef habitats Mar Ecol Prog Ser

401:245–258

Lam K, Shin PKS, Bradbeer R, Randall D, Ku KKK, Hodgson

P, Cheung SG (2006) A comparison of video and point

intercept transect methods for monitoring subtropical coral

Lecchini D, Galzin R (2005) Spatial repartition and ontogenetic

shifts in habitat use by coral reef fishes (Moorea, French

Leis JM (1989) Larval biology of butterflyfishes (Pisces,

Chaetodontidae): what do we really know? Environ Biol

Leis JM, Carson-Ewart BM (2003) Orientation of pelagic

larvae of coral-reef fishes in the ocean Mar Ecol Prog Ser

Mclanahan TR, Maina J, Pet-Soede L (2002) Effects of the

1998 coral mortality event on Kenyan coral reefs and

Moore KA, Elmendorf SC (2006) Propagule vs niche

limitation: untangling the mechanisms behind plant

spe-cies’ distributions Ecol Lett 9:797–804

Mumby PJ (2006) Connectivity of reef fish between mangroves

and coral reefs: Algorithms for the design of marine

Munday PL (2001) Fitness consequences of habitat use and

competition among coral-dwelling fishes Oecologia

Pratchett MS (2005) Dietary overlap among coral-feeding

butterflyfishes (Chaetodontidae) at Lizard Island, northern

Pratchett MS (2007) Dietary selection by coral-feeding

butter-flyfishes (Chaetodontidae) on the Great Barrier Reef.

Pratchett MS, Wilson SK, Baird AH (2006) Declines in the abundance of Chaetodon butterflyfishes following exten- sive coral depletion J Fish Biol 69:1269–1280

Pratchett, MS, Baird, AH, McCowan, DM, Coker, DJ, Cole,

AJ, and Wilson, SK (2008a) Protracted declines in coral cover and fish abundance following climate-induced coral bleaching on the Great Barrier Reef Proceedings of the 11th International Coral Reef Symposium, Ft Lauderdale,

Pratchett MS, Berumen ML, Marnane MJ, Eagle JV, Pratchett

DJ (2008b) Habitat associations of juvenile versus adult

Russ GR (1984a) Distribution and abundance of herbivorous grazing fishes in the central Great Barrier Reef I Levels

of variability across the entire continental shelf Mar Ecol

Strathmann RR, Hughes TP, Kuris AM, Lindeman KC, Morgan

SG, Pandolfi JM (2002) Evolution of local recruitment and its consequences for marine populations Bull Mar Sci

Werner FE, Cowen RK, Paris CB (2007) Coupled biological and physical models: present capabilities and necessary developments for future studies of population connectivity Oceanography 20:54–69

Williams DM (1986) Temporal variation in the structure of reef slope fish communities (central Great Barrier Reef): short- term effects of Acanthaster planci infestation Mar Ecol

Homatula laxiclathra (Teleostei: Balitoridae), a new species

of nemacheiline loach from the Yellow River drainage

in Shaanxi Province, northern china

Jin-Hui Gu&E Zhang

Received: 19 May 2011 / Accepted: 4 November 2011 / Published online: 16 November 2011

# Springer Science+Business Media B.V 2011

Abstract Homatula laxiclathra, new species, is here

described from the Wei-He of the Yellow River

drainage in Shaanxi Province, northern China It is

similar to H berezowskii, H longidorsalis, and H

variegata in the shared possession of a shallower

body with a uniform depth, a character distinguishing

all of them from all other congeners, but differs from

these three species in the width of vertical brown bars

on the caudal peduncle This new species, along with

H berezowskii, differs from H longidorsalis, and H

variegata in head length, caudal-peduncle depth,

length of the dorsal adipose creast of the caudal

peduncle, body squamation, and intestinal coiling

Homatula laxiclathra and H berezowskii are further

distinct in the caudal-fin shape and interorbital width

Keywords Homatula laxiclathra New species

Yellow River drainage Shaanxi Province

Introduction

Hu and Zhang (2010), in their describing Homatulapycnolepis from the Yangbi-Jiang of the upperMekong River drainage of Yunnan Province, southernChina, provided a re-definition of Homatula, andplaced in the genus all species previously included inParacobitis in the Chinese literature with the excep-tion of three hypogean species: P longibarbata, P.maolanensis and P posterodorsalus Nearly at thesame time, Min et al (2010) preferred to use Para-cobitis as the generic name for the species here placed

in Homatula, and described P nanpanjiangensis as anew loach species from the Nanpan-Jiang of the PearlRiver drainage in Yunnan Province, South China Min

et al.’s generic classification, though, is not inagreement with the consensus of workers outsideChina; e.g., Kottelat (1990), Bǎnǎrescu and Nalbant(1995) and Nalbant and Bianco (1998) Paracobitis,

as defined by these workers, is known only fromWestern Asia Chinese species previously recognized

in Paracobitis have a great geographical gap with thewestern Asian species of this genus, and thus belong

to a distinct genus Homatula (Bǎnǎrescu and Nalbant

1995) For this reason, we insist that Homatula is theavailable generic name for the epigean species thatwere referred to Paracobitis in the Chinese literature

So far ten valid species are designated to Homatula:

H acuticephala, H anguillioides, H berezowskii, H.erhaiensis, H nanpanjiangenis, H oligolepis, H

DOI 10.1007/s10641-011-9965-1

Chinese Academy of Sciences, Institute of Hydrobiology,

Wuhan 430072 Hubei Province,

e-mail: zhange@ihb.ac.cn

J.-H Gu

e-mail: swxgjh@126.com

J.-H Gu

Graduate School of Chinese Academy of Sciences,

pycnolepis, H potanini, H variegata, and H.

wujiangensis

Hu and Zhang (2010) cleared up some taxonomic

problems surrounding H variegata They concluded

that the type locality of H variegata is in the upper

Yangtze River drainage; that H berezowskii, which

was formerly considered as a junior synonym of H

variegata, is a valid species, and that the specimens

previously recognized as H variegata from the

Wei-He of the Huang-Wei-He (= Yellow River) drainage in

Shaanxi Province represent an unnamed species of

Homatula The purpose of the present investigation is

to provide a detailed description of this unnamed

species

Material and methods

Measurements were taken using digital calipers

connected to a computer to the nearest 0.1 mm The

last two branched rays in dorsal and anal fins are

closely approximated at the base, we thereby count

them as a single ray each Measurements and counts

were made on the left side of specimens whenever

possible, following the methods of Kottelat (1990)

Predorsal, prepectoral, prepelvic and preanal lengths

were taken, respectively from the anteriormost tip of

the snout to the dorsal-, pectoral-, pelvic- and anal-fin

origins, Measurements of parts of the head are given

as proportions of head length Head length and

measurements of other parts of the body are given

as proportions of standard length Statistics 5.0

(Wilkinson et al.1992) was used for the basic statistic

analysis on morphometric data and also for the

principal component analysis on the

variance-covariance matrix of the log–transformed

measure-ments Abbreviations for collections are: IHB,

Insti-tute of Hydrobiology, Chinese Academy of Sciences,

Wuhan; YU, Yunnan University, Kunming, and KIZ,

Kunming Institute of Zoology, Chinese Academy of

Sciences, Kunming Abbreviations: SL, standard

length, and HL, head length

Results

Homatula laxiclathra, new species (Fig.1a)

Paracobitis variegatus: Gao 1992: 86 (Zhouzhi

and Huxian)

Holotype

IHB 73V10738, 136.7 mm SL, Wei River, a tributary

of Yellow River drainage, at Zhouzhi County,Shaanxi Province, North China; collected in 1973;

no information about collectors

Paratypes

IHB 80VI0956-7, 80VI0959, 80VI0961, 80VI0964-8,80VI0971-3, 80VI0976, 80VI1185, 82VI0103,82VI0106-8, 82VI2279, 82VI2283-4, 21, 67.6–121.9 mm SL, same locality as holotype

Diagnosis

Homatula laxiclathra is similar to H variegata, H.berezowskii, and H longidorsalis in the sharedpresence of slender bodies with uniform depth [9.3–15.8% (mean 12.5) SL; see Figs 1 and 2a, andTable 1], a character distinguishing them from allremaining species of this genus where they havedeeper bodies [13.6–23.9% (mean 16.6) SL] andtheir body depth evenly decreases posterior to theposterior end of the dorsal-fin base The newspecies differs from H variegata, H berezowskii,and H longidorsalis in having vertical brown bars onthe caudal peduncle twice as wide as (vs muchnarrower or slightly wider than) their interspaces(Fig.1) This new species, along with H berezowskii,

is further distinguished from H variegata and H.longidorsalis by the presence of a shorter head (length15.6–21.1% SL vs 18.3–25.3; Fig 3a, Table 1), ashallower caudal peduncle (depth 8.4–11.4% SL vs.9.6–13.6; Fig 3b, Table 1), a shorter adipose crestalong the dorsal midline of the caudal peduncleanteriorly not extending to the midway of the anal-fin base (vs extending beyond the vertical throughthe anal-fin origin) (Fig 1a–d), a scaleless (vs.partially scaled) predorsal body, and an intestine with(vs without) a loop or bend behind the stomach(Fig 4) It is further distinct from H berezowskii inhaving an oblique (vs truncate) posterior margin ofthe caudal fin, and a narrower interorbital space[width 19.9–24.9 (average 22.5)% HL vs 24.1–29.7(average 27.0); see Fig 2b] The new species has anintestine with a loop anteriorly reaching the posteriorsurface of the U-shaped stomach, whereas the other

related species do not have a loop at this portion of

the intestine (Fig.4a, c)

Description

Morphometric data for type specimens are given in

Table 1 Body elongate and cylindrical, anteriorly a

little depressed and posteriorly compressed laterally.Head, thorax, abdomen and anterior half of predorsalbody scalesless; scales only present on back and sides

of posterior half of predorsal body as well aspostdorsal body Caudal peduncle compressed laterally.Lateral line complete, extending along midline of bodydirectly

5 10 15 20 25

SL (mm)

a

2.5 3.5 4.5 5.5 6.5

HL (mm)

b

Fig 2 a relation between body depth and SL for H laxiclathra

(black up-pointing triangle), H variegata (white circle), H.

berezowskii (white diamond), H longidorsalis (white square),

and all remaining species (black circle); and b relation between interorbital width and HL for H laxiclathra (black up-pointing triangle) and H berezowskii (white diamond)

Fig 1 Lateral view of four

species of Homatula: a H.

laxiclathra, IHB 73VI0738,

136.7 mm SL; China:

Yellow River drainage:

Wei He; b H variegata,

Head short, and depressed in frontal view, wider

than high, roughly triangular in dorsal view Snout

blunt, slightly shorter than postorbital length of

head Eye small, close to dorsal profile of head,

invisible from ventral view Interorbital space wide

and flat Nostrils closely set, nearer to anterior

margin of eye than to snout tip; anterior nostrils

situated at a nostril valve

Mouth inferior and arched Lips thick, slightly

furrowed, but not papillated; upper lip with a small

median incision, and lower lip with a marked median

incision Jaws covered by lips; upper jaw with a

well-developed process dentiformis corresponding with a

marked notch on lower jaw Three pairs of barbels;

two rostral pairs, inner one reaching corners of mouth

and outer one not reaching a vertical line of anterior

nostril, and one maxillary pair short

Fins flexible; dorsal fin with 3 simple and

8 branched rays; distal margin convex as a whole;

origin nearer to snout tip than to caudal-fin base

Pectoral fin with 1 simple and 9–10 branched rays,inserted slightly posterior to vertical through posteri-ormost point of operculum, tip of the longest fin raynot extending beyond halfway to insertion of pelvicfin Pelvic fin with 1 simple and 6–7 branched rays,inserted below second or third branched rays of dorsalfin, tip of the longest fin ray extending not beyondhalfway of distance between pelvic-fin insertion andanal-fin origin Anal fin with 3 simple and 5 branchedrays, with a convex distal edge; origin closer topelvic-fin insertion than to caudal-fin base Posteriormargin of caudal fin oblique Caudal peduncleuniformly deep; with adipose crests along its dorsaland ventral midlines Adipose crest along dorsalmidline of caudal peduncle anteriorly not extendingthrough anteriorly the position of the anal-fin origin.Intestine forming a loop anteriorly reaching poste-rior surface of U-shaped stomach (Fig 4a) Gasbladder osseous, anterior chamber invisible, fullyenclosed in capsule; posterior chamber degenerative

11 14 17 20 23 26

(a) HL and SL and (b)

caudal peduncle depth

and SL, for H laxiclathra

(black up-pointing triangle),

H variegate (white circle),

H berezowskii (white

dia-mond) and H longidorsalis

(white square)

Fig 4 Ventral view of

intestine coiling pattern

In formalin-stored specimens, head yellowish brown

Ground color of body pale Four or five brown blotches

on median predorsal region of body 15 to 17 rectangular

yellowish brown vertical bars along each side of body (5

or 6 predorsal, 3 subdorsal, and 7 to 8 postdorsal) twice

as wide as interspace Dorsal and caudal fins with dark

brown spots A dark brown vertical bars on caudal-fin

base Dorsal surface of pectoral, pelvic and anal fins

grayish, and caudal fin gray

Distribution

Known only from the Wei He of the Yellow River

drainage in Shaanxi Province, North China (Fig.5)

Etymology

The specific epithet, used as an adjective, is madefrom the combination of Latin words “laxus”(wide)and clathrus (barred), alluding the presence of widervertical bars on each side of body

Discussion

The description of a new species brings the total number

of species of Homatula to 12 Zhu (1989), in hismonograph of Chinese species of the Nemacheilinae,recorded five species of the genus under the genericname Paracobitis: P anguillioides, P erhaiensis, P.oligolepis, P potanini and P variegatus His generic

Fig 5 Map showing distribution of all Homatula species in China

classification of Paracobitis was subsequently

accept-ed widely, and four new Chinese species or

subspe-cies were added to the genus Ding and Deng (1990)

described P wujiangensis from the Wu-Jiang, a

tributary flowing to the upper Yangtze River in

Sichuan Province; Zhou and He (1993) described P

acuticephalus from the Erhai Lake system in Yunnan

Province; Yang et al (1994) described P variegatus

longidorsalis from the Nanpang-Jiang of the Pearl

River drainage in Yunnan Province; and Min et al

(2010) described P nanpangjiangensis from the

Nanpang-Jiang of the upper Pearl River drainage in

Yunnan Province However, all above-mentioned

species or subspecies of Paracobitis exclusive of P

nanpangjiangensis were transferred to Homatula by

Hu and Zhang (2010) when they described H

pycnolepis from the Yangbi-Jiang, a tributary of the

upper Mekong River drainage in Yunnan Province;

meanwhile, they validated Nemachilus berezowskii

Günther1896 from the synonym of H variegata P

nanpangjiangensis was not included in Homatula by

Hu and Zhang (2010), because Min et al’s (2010)

paper was published only a little earlier and Hu and

Zhang could not include it in their revision This

species is here transferred to Homatula Given that the

subspecies P variegatus longidoralis is here

consid-ered as a full species of Homatula, the here described

species H laxiclathra is the twelfth species of this

genus

Prior to the present investigation, the status of P

variegatus longidoralis remains contentious Min et

al (2010) considered it as valid, but Hu and Zhang

(2010) commented that it was not quite distinct from

H variegata The ongoing molecular phylogenetic

analysis, however, indicates that samples of P

variegatus longidoralis form a monophyletic lineage

(Chen X Y., KIZ; pers comm.) Based on this,

combined with no subspecies status under

phyloge-netic species concept, this subspecies is here rendered

as a full species In terms of Min et al (2010), H

longidorsalis differs from H variegata in having an

anterior nostril placed at a short tube (vs a valve), 9

(vs 8; rarely 9) branched dorsal-fin rays, maxillary

barbels extending to the vertical of the anterior or

middle (vs middle or posterior) margin of eye; and

numerous vermiform markings on the top of head (vs 1–

4 vermiform markings on the parietal area or obscure)

Twelve species here recognized in Homatula can

be tentatively subdivided into four groups on the basis

of squamation, completeness in the lateral line, andbody shape The first group comprises four specieshaving a densely scaled body except for head: H.acuticephala, H anguillioides, H erhaiensis and H.pycnolepis They occur close to each other in a smallarea in the upper Mekong River drainage (theYangbi Jiang and the Erhai Lake system) inYunnan Province, South China The second group

is composed of two species with an incompletelateral line, H potanini and H wujiangensis, andboth are known only from the upper Yangtze Riverdrainage The third group includes four species,namely H laxiclathra described here, H variegata,

H berezowskii and H longidorsalis, currently knownfrom the Yellow River, Yangtze River and Pearl Riverdrainages; these four species have slender bodies withuniform depth (9.3–15.8% SL; see Fig 2a) Thefourth group includes two species, P nanpanjiangen-sis, and H oligolepis, both confined only to the upperPearl River drainage in Yunnan and Guizhou prov-inces The two species have a scaleless body, or withrudimentary scales present only on the caudal-finbase

The results of the principle component analysisperformed on the variance-covariance matrix oflog-transformed measurements for the examinedspecimens of four species of the third group(Table 2, Fig 6) revealed that the combination ofPC2 against PC3 enabled the separation of H.laxiclathra and H berezowskii from H longidorsalisand H variegata These two pairs were distinguish-able by PC2, the main shape axis, on which the mainloadings are head length and caudal peduncle depth.Homatula laxiclathra and H berezowskii are distin-guished from H longidorsalis and H variegata inhaving a shorter head (length 15.6–21.1% SL vs.18.3–25.3; see Fig 3a), and a shallower caudalpeduncle (depth 8.4–11.4% SL vs 9.6–13.6) Bothare also distinct from H longidorsalis and H.variegata in having a shorter adipose crest along thedorsal midline of the caudal peduncle anteriorly notextending beyond (vs extending beyond) thevertical through the anal-fin origin (Fig 1a–d), ascaleless (vs partially scaled) predorsal body, and anintestine with (vs without) a bend or loop behind thethickened portion of intestine Apparently, our studyfurther confirms Hu and Zhang (2010) conclusion that

H berezowskii is a valid species distinct from H.variegata

6, 114.1–126.0 mm SL; China: Yunnan Province:Eryuan County: upper Mekong River drainage.Homatula berezowskii, IHB 73VI1191-3, 73VI1194,

4, 89.0–125.4 mm SL; China: Shanxi Province:Lueyang County: Jialing River of Yangtze Riverdrainage -IHB 73VI1044, 1, 120.7 mm SL; China:Shanxi Province: Fengxian County: Jialing River ofYangtze River drainage.-IHB 82VI2489, 1, 122.6 mmSL; China: Gansu Province: Chengxian County: JialingRiver of Yangtze River drainage -IHB 82 V2386, 1,85.8 mm SL; China: Gansu Province: Huixian County:Jialing River of Yangtze River drainage -IHB82VI2753, 82VI2755, 82VI2757, 3, 97.3–117.0 mmSL; China: Gansu Province: Wudu County: JialingRiver of Yangtze River drainage -IHB, uncat., 3, 74.89–113.35 mm SL; China: Hubei Province: ZhushanCounty: Du He, a tributary to Han Jiang of middleYangtze River drainage

Homatula erhaiensis, IHB 64VI0012, holotype, 1,68.8 mm SL; -IHB 64VI0001-11, 64VI0013-5,646775–7, 646779, paratypes, 18, 64.4–86.8 mmSL; China: Yunnan Province: Eryuan County: upperMekong River drainage -IHB 1270142–8, 1270150–

4, 12, 49.1–79.9 mm SL; China: Yunnan Province:Eryuan County: upper Mekong River

Homatula longidorslis, KIZ 874042–3, 874045–6,

874050, 1987005739, 1987005748, 1987005752, 8,50.26–80.92 mm SL; China: Yunnan Province: YiliangCounty: Nanpan Jiang of Pearl River drainage.Homatula nanpanjiangensis, KIZ 1994000021,

1 9 9 4 0 0 0 0 2 5 , 1 9 9 4 0 0 0 0 2 8–9, 1994000031,1994000033–5, 1994000037, 9, 65–90.55 mm SL;China: Yunnan Province: Qujing County, Nanpan Jiang,Pearl River drainage

Homatula oligolepis, KIZ 774557–9, 774560,

856145, 652099, 6, 83.51–171.05 mm SL; China:Yunnan Province: Zhanyi County: Pearl River drainage.Homatula potanini, IHB 42IX0661-2, 42IX0664,42IX0666-7, 79IV0597-8, 79IV0600, 79IV0605,79IV0609-10, 82 V0301-4, 15, 68.6–83.3 mm SL;China: Sichuan Province: Emei County: Yangtze Riverdrainage -IHB 78IV0175, 78IV0228-9, 78IV0233-34,

Table 2 Loadings on the first three principal components

extracted from morphometric data for: H laxiclathra, H.

variegata, H berezowskii, and H longidorsalis

Fig 6 Scatter plot on the 2nd and 3th principle components

extracted from morphometric data for H laxiclathra (black

up-pointing triangle), H variegata (white circle), H berezowskii

(white diamond) and H longidorsalis (white square)

78IV0239, 78IV0243, 79IV0401-2, 79IV0483-6,

820004–5, 15, 62.3–82.4 mm SL; China: Sichuan

Province: Leshan City: Yangtze River drainage

Homatula pycnolepis, IHB 814042–3, 814045–51,

9, 90.53–118.83 mm SL; China: Yunnan Province:

Jianchuan County: Yangbi Jiang of Mekong drainage

Homatula variegata, IHB 42VI0726, 1, 90.79 mm

SL; China: Sichuan Province: Xikang County: Jinsha

Jiang of Yangtze River drainage -IHB 82VI0517, 1,

119.32 mm SL; China: Sichuan Province: Dechang

County: Jinsha Jiang of Yangtze River drainage -IHB

82VI0461, 1, 89.87 mm SL; China: Sichuan Province:

Huili County: Jinsha Jiang of Yangtze River drainage

-IHB 64VI0600, 1, 103.44 mm SL; China: Gansu

Province: Wen County: Jialing Jiang of Yangtze River

drainage -IHB 572090, 1, 83.39 mm SL; China:

Chongqing City: Wuxi County: Daning He of Yangtze

River drainage -IHB, uncat., 7, 66.08–114 mm SL;

China: Sichuan Province: Yalong Jiang of Yangtze

River drainage -IHB uncat, 1, 120.97 mm SL; China:

Wu Jiang of upper Yangtze River drainage -IHB,

uncat., 7, 71–129.6 mm SL; China: Gesala: Yalong

River of Yangtze River drainage -KIZ 2004051170,

2004051172–7, 2004051181–2, 2004051179, 10,

74.37–96.77 mm SL; China: Yunnan Province: Yanjin

County: Baishui He of Yangtze River drainage -KIZ

L.N Du (KIZ) for permitting us to take digital photographs and

measurements of the specimens of Homatula nanpanjiangensis,

H longidorsalis, and H oligolepis under their care Our thanks

are also given to Z W Sun for her useful advices on this

manuscript This research was supported by a grant

(KSCXZ-YW-0934) from the Innovation Program of the Chinese Academy of Sciences.

References

Nemacheilinae with description of two new genera ostei: Cypriniformes: Cobitidae) Trav Mus Hist Nat Gr

Günther A (1896) Report on the collections of reptiles, batrachians and fishes made by Messrs Potanin and Berezowski in the Chinese provinces Kansu and Sze- chuan Ann Mus Zool Acad Sci St Pétersb 1:199–219

Hu YT, Zhang E (2010) Homatula macrolepis, a new species of nemacheilinae loach from the upper Mekong drainage, South China (Teleostei: Balitoridae) Ichthyol Explor

Kottelat M (1990) Indochinese nemacheilines: a revision of nemacheiline loaches (Pisces: Cypriniformes) of Thailand, Burma, Laos, Cambodia and southern Viet Nam Verlag

Min R, Chen XY, Yang JX (2010) Paracobitis sis, a new loach (Balitoridae: Nemacheilinae) from

Nalbant TT, Bianco PG (1998) The loaches of Iran and adjacent regions with description of six new species (Cobitoidea) Ital J Zool 65:109–125

Wilkinson L, Hill M, Welna JP, Birkenbeuel GK (1992) Systat for windows: statistics, version 5 edition Systat Inc, Evanston, pp 1–750

Yang JX, Chen YR, Kottelat M (1994) Subspecific ation of Paracobitis variegatus with comments on its

Zhou W, He JC (1993) The Parocobitis fishes distributed in Erhai area, Yunnan, China (Pisces: Cobitidae) Zool Res

Zhu SQ (1989) The loaches of the subfamily Nemacheilinae in China (Cypriniformes: Cobitidae) Jiangsu Science and

Ecomorphological analysis as a complementary tool to detect

changes in fish communities following major perturbations

in two South African estuarine systems

Antoni Lombarte&Ana Gordoa&

Alan K Whitfield&Nicola C James&

Víctor M Tuset

Received: 5 November 2010 / Accepted: 6 November 2011 / Published online: 7 January 2012

# Springer Science+Business Media B.V 2011

Abstract Ecomorphological changes as a result of

natural perturbations in estuarine fish communities

were investigated in two South African estuaries

(Swartvlei and East Kleinemonde), both before and

after the loss of aquatic macrophyte beds in these

systems The fish communities were analysed using

an ecomorphological diversity index (EMI) and the

results compared to a traditional index, the

Shannon-Wiener diversity index The EMI revealed that the

major changes in fish community composition

recorded in both estuaries were associated with

quantitative variations at the species level Both

estuaries essentially lost their macrophyte beds and

ended up with the same type of bottom habitat (bare

sediment) In both cases the fish morphological

variability decreased immediately after aquatic

macro-phyte loss and then increased to end above the initial

value The ecomorphological analysis appeared to be

sensitive to major ecological disturbances that occurred

during the study period and this was confirmed by themorphospace configuration The results indicate that theecomorphology of the fish community responds tohabitat changes and that this change corresponds toalterations in the representation of the different feedingtypes These findings therefore contribute to themeasurement of morphological changes in estuarinefish assemblages as a result of habitat changes within theecosystem and we propose that ecomorphologicalanalyses add another dimension to the informationprovided by existing diversity indices in studyingchanging fish communities

Keywords Ecomorpholgy Index of diversity SouthAfrican estuarine systems Fish communities

Introduction

Animal and plant communities are subjected tonumerous biotic and abiotic disturbances, caused bydifferent factors (including human activities) whichoften modify the composition and structure of theseassemblages (Begon et al 1986) Major disturbancewithin an aquatic ecosystem can result in a shift infish community structure (Olden et al.2008; Villéger

et al 2010) The morphological differences betweenspecies are indicative of differences in ecologicalstrategies (Norton et al.1995) in relation to locomo-tion (Fulton et al 2001; Wainwright et al 2002),habitat use (Gatz1979; Winemiller1991; Lombarte et

Blanes, Catalonia, Spain

South African Institute for Aquatic Biodiversity,

Grahamstown 6140, South Africa

al.2003; Lombarte et al.2010) and feeding strategies

(Motta et al 1995; Wainwright and Richard 1995;

Pouilly et al.2003; Barnett et al.2006; Wagner et al

2009)

Changes in fish community structure after a

disturbance could be functional and are likely to

be related to the loss of habitats and changes in

food resources Loss of biodiversity is likely to

affect or even threaten critical functional roles

within the ecosystem (Paddack et al 2006;

Villéger et al 2010) However, biodiversity species

or eveness richness indices may not detect functional

changes because these indices ignore the actual

species and consequently cannot identify their role

within the community (Green and Vascotto 1978,

Ernst et al.2006; Petchey and Gaston2006; Violle et

al.2007) It is therefore important to select appropriate

criteria for assessing the impact of perturbations and

also the degree of resilience of the community

(Bellwood et al 2006)

The most popular diversity indices used to document

community changes cover two different aspects, namely

the number of species (richness) and the proportional

abundances of species (heterogeneous diversity,

even-ness) (Whittaker 1960; Gray 2000) Most classical

biodiversity indices do not take into account factors

such as the phylogeny or changes in morphological

composition of species that constitute the community

(Magurran 1988) However, the extent or magnitude

of changes within a community as a result of natural

or anthropogenic perturbations can be measured using

morphological and functional traits (Ernst et al.2006;

Mason et al 2007; Olden et al 2008; Flynn et al

2009; Villéger et al.2010)

Biodiversity assessments based on the taxonomic

relatedness of the species have been undertaken by

Warwick and Clarke (1995, 1998) and Clarke and

Warwick (1998, 1999, 2001) These indices of

taxonomic diversity take into account the“weighted”

taxonomic differences between species (Mouillot et

al.2005) Somerfield et al (1997) and Mouillot et al

(2005) found no consistent pattern of decreasing

taxonomic diversity with increasing environmental

impact At the turn of the century a taxonomic index,

based on the nearest mean nodal taxonomic distance

(MND), was created by Webb (2000) More recently,

Bohannan and Hughes (2003) introduced modern

molecular techniques for determining microbial

bio-diversity based largely on Webb’s taxonomic index

The morphological characters of the species within

a community are considered essential to determine thefunctional structure of that community (Schoener

1974) This has led to the concept of ecomorphology,which defines a community or species assemblagethrough a morphospace determined by the morpho-logical data of the species that comprise that assem-blage (Karr and James1975; Gatz1979; Bock1990;Lombarte et al.2003) Morphospaces can be analysedusing geometric morphological methods (Neige2003;Antonucci et al 2009), from which an ecomorpho-logical diversity index (EMI) can be generated(Recasens et al 2006) to provide additional informa-tion on the morpho-functional structure of thecommunity (Winemiller 1991; Fulton et al 2001;Mason et al 2007)

The aim of this study was to increase the range

of methods available for detecting the impact ofdisturbances on aquatic systems Based on thepremise that functional loss may alter the morpho-logical characteristics of a fish community, themorphotypes of two South African estuarine fishcommunities, before and after major disturbances,were examined We also explore the sensitivity of

a traditional diversity index (Shannon-Wienerdiversity index) and recent Taxonomic DiversityIndex for detecting changes in the fish communityand compare these indices with the sensitivity of

an ecomorphological index based on multivariateanalyses of quantitative morphological data (Reca-sens et al 2006)

Material and methods

Study areas and biotic characteristics

The changes in two fish assemblages over time wereanalysed in two warm-temperate South Africanestuaries, the Swartvlei (34°01′51′′S; 22o

47′49′′E)and East Kleinemonde (33°32′21′′S; 27°02′55′′E).The data analysed in this study are derived fromdifferent sampling programmes, details of which areavailable from Whitfield (1986) for Swartvlei andJames et al (2008) for the East Kleinemonde Bothsystems experienced river inflow changes, whichaltered their habitats Prior to the loss of aquaticmacrophytes, the estuaries analysed in this studypresented differences in habitat complexity The

Swartvlei littoral zone was characterised by a more

complex habitat structure than the East Kleinemonde,

with the former system having an outer Potamogeton

zone, a Chara zone and an inner Potamogeton zone,

whereas the East Kleinemonde had either a Ruppia or

Potamogeton zone only In addition, the Swartvlei

system is classified as an estuarine lake whereas the

East Kleinemonde is classified as a temporarily open/

closed estuary (Whitfield1992)

The fish community associated with the

Swartvlei littoral zone was monitored between

1979 and 1982 (Whitfield 1986) During this period

the littoral zone was covered by extensive

Potamo-geton pectinatus beds in 1979, epipsammic

filamen-tous algal mats in 1980 and bare sand in 1982 The

fish community of the East Kleinemonde Estuary was

monitored between 1999 and 2007 (James et al

2008) Similar major changes occurred in the littoralzone of this estuary where extensive Potamogetonpectinatus and Ruppia cirrhosa beds were recordedbetween 1999 and 2002 but were mostly absentbetween 2003 and 2007 (Sheppard et al 2011)

Analyzed biodiversity indexes

The Ecomorphological Diversity Index (EMI) wascalculated for the fish assemblages from standardisedimages of the left side of each species A total of

27 metrics (Fig 1) were selected for each specimen(Recasens et al 2006) After digitalising the metricmaps of each species, they were rotated, scaled (tounit centroid size) and translated using a generalisedleast-square superimposition procedure (generalisedProcrustes) to remove scale and orientation distortions

Fig 1 Examples of the

fea-tures used in the geometric

morphological analysis of

selected fish species from the

East Kleinemonde Estuary: a

Liza dumerili, b Liza

falciformis Images were

obtained from the SAIAB

image collection

( http://saiab.ac.za:8080/

saiab/index.html )

(using tpsRel 1.24; Rohlf 2001) A thin-plate spline

representation was used to fit an interpolated function

to an average map (consensus configuration) of the

profile shape and derive the uniform (relative warp)

components of the shape variation Changes in shape

were visualised using relative warp analysis (Adams

et al 2003; Rohlf and Marcus2003; Zelditch et al

2003) The correlation between shape vectors of each

fish community were computed for a percentage of

covariance up to 99% using the program tpsPLS

(Rohlf 2003) in order to determine the similarity

between studied fish assemblages

In the morphological analysis the relative

abun-dance of each fish species in the assemblage was

considered, with the number of images analysed per

species corresponding to the percentage of individuals

of that species within the community For example,

Monodactylus falciformis accounted for 42% of the

initial abundance within the Swartvlei fish community

(Table1); thus 42 standardised images of this species

were used in the relative warp analysis Any species

whose abundance was equal to or less than 1% were

analysed as one image

The first eight relative warp scores (>98.5% of the

total morphological variability) were selected as the

variables for describing each species For each

assemblage, a cluster analysis was carried out based

on Euclidean distances (Sokal and Rohlf 1962) Thecomplete linkage method and the aggregation algo-rithm (Sokal and Michener 1958) were employed asthey both yield a high cophenetic correlation coeffi-cient To estimate the EMI for each estuary and phase,the mean nodal distance (Webb2000; Bohannan andHughes 2003) of the dendrograms, produced by thecluster analyses, was estimated (Fig.2)

To determine how the Ecomorphological DiversityIndex (EMI) behaves in relation to more conventionaldiversity and trait diversity indices, it was firstlycompared with the Shannon-Wiener Diversity Index(Shannon and Weaver1949): H ¼ SUMðpi»lnðpiÞÞ,where pi = the proportion of individuals of eachspecies This index is widely used and has evenbeen incorporated into the environmental legisla-tion of some countries to determine the impact ofdisturbances on the marine environment (Gray

2000) A Taxonomic Diversity Index (Δ) was alsoused and is defined as the average path lengthbetween every pair of individual organisms in asample, with individuals in the same genus beingcloser than individuals in the same family (Warwickand Clarke1995,1998)

Finally, the consensus configurations (i.e thehypothetical fish shape) of each morphospacewere compared The consensus fish community

Table 1 Characteristic fish species of the Swartvlei littoral area during the presence and absence of aquatic macrophyte beds Fish relative abundance is expressed as a percentage of individuals in relation to the total number captured

Swartvlei

East Kleinemonde mavrophyte 1999

East Kleinemonde sand 2007 East Kleinemonde sand 2004

Fig 2 Dendrogram based on Euclidean distances and complete

linkage of the first eight scores of the morphological analysis of

the fish assemblages Black dots correspond to the nodal

distance between species or groups of species 1a Swartvlei,

aquatic macrophytes; 1b Swartvlei, filamentous algae; 1c

Swartvlei, sand substrate; 2a East Kleinemonde, aquatic macrophytes; 2b East Kleinemonde, sand/mud substrate 1 year after the disturbance; (2c) East Kleinemonde, sand/mud substrate 3 years after disturbance

shapes over the study period were obtained by

means of another warp analysis (using tpsRel

1.24; Rohlf 2001), in order to ascertain the evolution

of the standard body shape of the ichthyofaunal

community over time and to explore how these

changes were associated with alterations in aquatic

habitats

Results

The fish species composition of each estuary

under different habitat phases is shown in Tables 1

and2 The different morphospaces (Figs.3and4) are

represented by their first three dimensions as these

explain between 74% and 95% of the morphological

variability In both the Swartvlei and East Kleinemonde

estuaries, the main changes were not at the qualitative

level (species presence/absence) but rather at the

quantitative level As the habitat changed over time the

evenness between the teleostean species was altered,

rather than any major changes in overall species

diversity (richness)

Swartvlei

The morphospaces of the Swartvlei littoral fishassemblage were similar throughout the study period(Fig 3a, b, c) Basically four groups could berecognised: midwater foraging species with a discoi-dal shape (e.g M falciformis), horizontally movingfusiform species that swim close to the bottom whenfeeding (e.g mugilids), species with both character-istics, some of which are more demersal (e.g.Rhabdosargus holubi) and others that forage through-out the water column (e.g Lichia amia) The firstcomponent, which explained between 56 and 61% ofthe total morphological variability, represented thegeneral body shape of the fishes, which changed fromdiscoidal in the case of M falciformis to fusiform forElops machnata and the Mugilidae The secondcomponent explained 24–32% of the total variabil-ity and was indicative of the level of development

of the caudal fin, which is most developed in E.machnata and least developed in R holubi

The correlations between shape vectors indicatedhigher differences between the second and last period(0.45), while shape differences between the initial

Table 2 Characteristic fish species of the East Kleinemonde Estuary during the presence and absence of aquatic macrophyte beds Fish relative abundance is expressed as a percentage of individuals in relation to the total number captured

East Kleinemonde

Fig 3 Morphospace

repre-sentation (three first

dimen-sions) of the ichthyofaunal

communities of Swartvlei

over time: a aquatic

macro-phyte phase; b epipsammic

filamentous algal phase;

c sandy bottom phase

Fig 4 Morphospace

repre-sentation (three first

dimen-sions) of the ichthyofaunal

communities of the East

Kleinemonde over time:

a aquatic macrophyte phase

(1999); b sandy/muddy

bottom phase immediately

after the disturbance (2004);

c final sandy/muddy bottom

phase (2007)

period and remaining periods were lesser (0.62 1a

versus 1b; 0.70 1a v 1c), indicating some degree of

recuperation in species composition

East Kleinemonde

The East Kleinemonde littoral zone was a less

complex habitat than Swartvlei but also had three

morphological groups dominating the three

morpho-spaces (Fig 4) The morphospace was mostly

determined by the general body shape, where the

first component explained between 61% and 79% of

the total morphological variability, and fluctuated

between the archetypal sparid shape of R holubi

and the fusiform shape of the mugilids The second

component, which explained between 11% and 29%

of the total variability, was associated with the

position of the dorsal and ventral fins, which are

posterior in M falciformis and anterior in

Oreochro-mis mossambicus

In the initial phase (macrophyte) the first axis,

which explained 60% of the morphological

variabil-ity, did not show any vertical segregation, which is

determined by the second axis (Fig 4a) The

macrophyte phase had a consensus configuration

skewed towards the sparid shape because R holubi

clearly dominated the community The following two

stages both had bare substrates and the warp analyses

showed that the first axis increased in variability

while the second decreased (Fig 4b, c) The

consen-sus configuration shifted gradually from a sparid

shape to an elongated mugilid shape and the vertical

segregation (2nd axis) declined in influence, which is

indicative of losing vertical structure with the

corresponding decrease in the abundance of the

monodactylids

The correlations between shape vectors revealed

considerable differences between the initial and final

period (0.34), thus indicating a dramatic change in the

fish assemblage Shape differences between the

intermediate period and remaining two were less

significant (0.56 2a versus 2b; 0.59 2b v 2c)

Consensus configuration analyses of the fish

assemblages

A warp analysis was carried out based on the

consensus configurations of the ichthyofaunal

com-munities of each estuary at the different stages In

Swartvlei (Fig 5a) a clear temporal change wasobserved in the predominant average fish Initially,the consensus configuration was a discoidal shape(similar to M falciformis) appropriate for swimmingvertically through tall submerged macrophyte beds.Following senescence of the aquatic macrophytes,epipsammic filamentous algal mats developed and thefish fauna consensus configuration was then replacedwith a sparid type silhouette (similar to R holubi).Finally, during the sandy bottom phase the consensusconfiguration evolved into a fusiform shape similar tothe mugilid silhouette which was the predominantgroup of fishes during this period This shape isappropriate for moving horizontally across an un-structured bottom topography without obstacles

In the East Kleinemonde Estuary fish community(Fig 5b) temporal changes before and after thedisappearance of the aquatic macrophytes were alsoclear During the initial phase (aquatic macrophytes)the average fish shape was a sparid type close to thedominant species R holubi, while in the years afterthe aquatic macrophytes had disappeared, the consen-sus configuration changed to the fusiform mugilidsilhouette

Comparison of biodiversity indexes

In general, for the three indices the East Kleinemondefish community remained less diverse than theSwartvlei fish community (Fig 6a, b) However, theresults indicate that the ecomorphological index(EMI) is sensitive to the observed changes in theSwartvlei and East Kleinemonde estuarine fishassemblages In both cases the morphological vari-ability decreased immediately after aquatic macro-phyte loss and then increased later to end above theinitial value This could be attributed to the widervariety of fish body shapes present during the sandyand sandy/mud phase of the Swartvlei and EastKleinemonde systems respectively

The Shannon-Wiener Diversity Index behaveddifferently for each time series analysed TheShannon-Wiener Index was weakly related to habitatchange in the Swartvlei system, while in the EastKleinemonde Estuary it responded positively tomacrophyte disappearance (Fig 6a) The TaxonomicDiveristy Index (TDI) showed differences in bothtime series analyzed In Swartvlei, important increaseswere only detected in the last phase of the study

period, i.e 2 years after the initial perturbation

(Fig 6b) In the East Kleinemonde, TDI was only

sensitive to change between the original phase and the

period just after perturbation (Fig 6b) So, the

behaviour of the TDI is not homogeneous in relation

with both perturbation processes

Discussion

The results from this study confirmed that higher

plant species diversity and morphological variability

within the Swartvlei littoral zone was more likely to

support a more diverse fish community than the East

Kleinemonde Estuary (Whitfield1998) The response

or sensitivity of the different indices to changes in the

fish communities after disturbances differed slightly

In the two study systems, certain diversity indicesactually increased to maximum values in the latterstages of the unvegetated phase, while speciesrichness did not change substantially The formerresult is due to the reduced abundance of a smallnumber of dominant species, thus‘smoothing out’ thebalance between the species and consequently theShannon-Wiener index increased Therefore, the use

of diversity or evenness indices to monitor changes infish communities may yield confusing results andshould be interpreted with caution from an ecological

or ecosystem perspective (Fulton et al 2001; wright et al 2002; Mason et al.2007)

Wain-The results presented here show that morphometricanalyses are sensitive to changes in fish communities,with fish diversity and morphological variability reach-ing a maximum value on bare substrata due to thepersistence of macrophyte associated species (even in

Algae(1980)

Macrophyte (1979)

No vegetation (1982)

a 1a

1b

1c

No vegetation (2004)

Macrophyte (1999)

No vegetation (2007)

b

2a

2b 2c

Fig 5 Consensus

configu-rations during the different

phases: a Swartvlei

(1a aquatic macrophytes,

1b filmentous algae, 1c

sandy bottom); b East

Kleinemonde (2a aquatic

macrophytes, 2b

sand/mud-dy substrate after 1 year, 2c

sand/muddy substrate after

3 years)

the absence of aquatic plants) and the influx of

previously scarce groups such as Mugilidae Although

it is generally accepted that differences in habitat

structure between sites are reflected by similar

differ-ences in their biotic diversity, this generalisation may

deviate when the differences in habitat structure occur at

the same site (Gray2000; Villéger et al 2010)

Both estuaries in this study ended up with the same

type of bottom habitat (bare sediment) and the fish

communities evolved towards higher diversity (H´)

and morphological variability (EMI) during this

phase This result is almost certainly due to the

occupation of the habitat by various mullet species

that were mostly absent during the macrophytedominant phases and would have introduced greatermorphological variation to the fish communities Themorphological diversity also expresses the level ofgenetic diversity of a community, and a largermorphological variation is also associated with alarger genetic variation (Macleod and Forey2002).Swartvlei had the highest phylogenetic diversity

of the two estuaries, as reflected by a highernumber of orders (Elopiformes, Siluriformes,Mugiliformes and Perciformes) and families (Elopidae,Aridae, Mugilidae, Carangidae, Haemulidae, Monodac-tylidae, Pomatomidae, Sciaenidae and Sparidae) TheEast Kleinemonde had only two orders (Mugiliformesand Perciformes) and eight families (Mugilidae, Car-angidae, Cichlidae, Haemulidae, Monodactylidae,Pomatomidae, Sciaenidae and Sparidae) Furthermore,species with extreme morphological characters were notvery abundant and the Mugilidae were predominant overthe other groups in terms of species diversity

The results from the three warp analyses ofSwartvlei showed that, in spite of the large changes

in the littoral habitat, there were no substantialchanges in the morphospace but large changes in theconsensus configuration At each phase (macrophyte,algae, sand) the first three axes (main components ofvariability), expressed the same type of morphologicalgradient However, the consensus configuration whichrepresented the morphospace centroid underwent asequential change from a discoidal to an elongatedshape as the habitat changed from predominantlyaquatic macrophyte to a bare sand substratum Thischange was driven by the decreased representation ofthe discoidal Monodactylus falciformis and increasedabundance of the fusiform Mugilidae

The changes in the proportion of species over thestudy period are clearly reflected in the altered centroids

of the morphospaces (as expressed by the consensusconfiguration) In the Swartvlei fish community, thedifferent shapes correspond mainly to differencesbetween the plant associated M falciformis and R.holubi and non-plant associated detritivore feeders.During the aquatic macrophyte phase the dominantspecies was M falciformis, which moves throughoutthe water column and preys predominantly onmacrophyte epifauna and zooplankton located on orbetween the plants (which reach up to 2 m in height).With the loss of aquatic macrophytes and proliferation

of epipsammic filamentous algae (situated on or close

Fig 6 Diversity index changes over time a comparison

between Ecomorphological Diversity Index (EMI) indicated

by solid lines and Shannon-Wiener Diversity Index (SDI)

indicated by dashed lines b comparison between

Ecomorpho-logical Diversity Index indicated by solid lines) and Taxonomic

Diversity Index (TDI) indicated by dashed lines 1a Swartvlei,

aquatic macrophytes; 1b Swartvlei, filamentous algae; 1c

Swartvlei, sandy bottom 2 years after disturbance; 2a East

Kleinemonde, aquatic macrophytes; 2b East Kleinemonde,

sandy/muddy bottom 1 year after the disturbance; 2c East

Kleinemonde, sandy/muddy bottom 3 years after disturbance

to the bottom) the sparid R holubi, which favours a

filamentous algal diet, was dominant Finally, when

the filamentous algal mat disappeared to create a bare

sandy bottom, there was an increase in the percentage

of Mugilidae (e.g Liza richardsonii, Liza dumerili

and Mugil cephalus) which are all horizontally

moving detritivorous species with fusiform body

shapes In East Kleinemonde, the initial aquatic

macrophyte-associated fish community was

dominat-ed by R holubi, which fedominat-eds on plants and benthic

invertebrates Following the disturbance which resulted

in the disappearance of aquatic macrophytes from the

East Kleinemonde, the Mugilidae (horizontally moving

detritivorous fishes associated with open waters) became

the predominant group The freshwater detritivore O

mossambicus declined in abundance and piscivores,

such as Argyrosomus japonicus, entered the fish

assemblage Three years later the bottom remained

without vegetation and the resulting morphospace

was very similar to the previous one, with Mugilidae

more abundant but represented by fewer species

Fish body shapes are directly associated with the

locomotion of the different species (Hertel 1966;

Lindsey 1978; Fulton et al 2001; Wainwright et al

2002; Costa and Cautadella 2007) For example,

Mugilidae are characterised by locomotion involving

the entire body moving horizontally, whereas

Mono-dactylidae tend not to use their entire body and are

more specialised for vertical and turning movements

The high abundance of Monodactylus falciformis

meant that the centroid of the morphological space

was skewed towards a discoidal contour The second

axis still described a high percentage of variability

(27%) and clearly separated Rhabdosargus holubi

from Lichia amia, which represent extreme types of

diet (omnivore versus piscivore respectively) They

differ in the position and size of the mouth and in the

length of the pectoral fin: Lichia amia has the biggest

mouth (associated with the piscivorous diet) and the

shortest pectoral fin

During the Swartvlei algal phase, the centroid

shifts from the discoidal shape to a typical sparid

shape (ventral depression) This is explained by the

reduced abundance of M falciformis and increased

abundance of R holubi In the final stage of the sandy

bottom phase the loss of fish height is evident and the

centre shifted towards fusiform shapes characteristic

of the Mugilidae due to increased abundance of this

family The above results show that the morphology

of the fish community corresponds to habitat changesand represents the dominant feeding type under eachregime These findings contribute to an under-standing of how morphological variation corre-sponds to a variation in the ecology of fishcommunities (Recasens et al 2006; Pulcini et al

2007; Costa and Cautadella2007)

Ecological morphology or ecomorphology is thestudy of the relationships between environmental factorsand body shape at the species level (Motta and Kotrschal

1992), and functional morphology is the study of therelationship between shape and the role of the species(Motta et al 1995; Barnett et al 2006; Petchey andGaston 2006; Violle et al 2007) This study hasconfirmed the validity of this relationship at a fishcommunity level, with the Ecomorphological Indexhaving been shown to reflect the changes in fishcommunity structure and is indicative of resourceutilisation changes with alterations in its ecological niche(Schoener 1974; Cornwell et al 2006; McGill et al

2006; Petchey and Gaston 2006; Mason et al 2007;Flynn et al.2009; Villéger et al.2010) The evolution

of a consensus body form in a particular ecosystemreflects environmental changes and particularly alter-ations in trophic structure The present study hashighlighted the need to use various analyses thatcomplement one another, thus allowing fish communi-ties to be compared in space and time as has been shownfor marine benthic faunal communities by Gray (2000).The ecomorphological analyses shown here are also auseful‘tool’ in biodiversity studies and can be used toexamine taxonomic or genetic relatedness of bioticassemblages (Warwick and Clarke1998; Webb2000;Bohannan and Hughes2003; Ernst et al.2006; Olden

et al 2008; Villéger et al 2010), especially whenothers indices are weakly influenced by ecologicaldisturbances and fail to assess environmental impact(Somerfield et al.1997; Lydy et al 2000; Mouillot et

al 2005)

MEC CGL2004-0384-E project and National Research tion (NRF) of South Africa.

Founda-References

Adams DC, Rohlf FJ, Slice D (2003) Geometric morphometrics: ten

Antonucci F, Costa C, Aguzzi J, Cautadella S (2009)

Ecomorphology of morpho-functional relationships in the

family of Sparidae: a quantitative statistic approach J

Morphol 270:843–855

Barnett A, Bellwood DR, Hoey AS (2006) Trophic

Begon M, Harper JL, Townsend CR (1986) Ecology: individuals,

populations and communities Blackwell, Oxford

Bellwood DR, Wainwright PC, Fulton CJ, Hoey AS (2006)

Functional versatility supports coral reef biodiversity Proc

Bock WJ (1990) From biologische anatomie to ecomorphology.

Bohannan BJM, Hughes J (2003) New aproaches to analyzing

287

Clarke KR, Warwick RM (1998) A taxonomic distinctness

index and its statistical properties J Appl Ecol 35:523–

531

Clarke KR, Warwick RM (1999) The taxonomic distinctness

measure of biodiversity, weighting of step lengths between

hierarchical levels Mar Ecol Prog Ser 184:21–29

Clarke KR, Warwick RM (2001) A further biodiversity index

applicable to species lists, variation in taxonomic

Cornwell WK, Schwilk DW, Ackerly DD (2006) A trait-based

test for habitat filtering: convex hull volume Ecology

Costa C, Cautadella S (2007) Relationship between shape and

trophic ecology of selected species of Sparids of the

Caprolace coastal lagoon (Central Tyrrhenian Sea) Env

Ernst R, Linsenmair KE, Rodel MO (2006) Diversity erosion

beyond the species level: dramatic loss of functional

diversity after selective logging in two tropical amphibian

communities Biol Conserv 133:143–155

Flynn DFB, Gogol-Prokurat M, Nogeire T, Molinari N, Richers

BT, Lin BB, Simpson N, Mayfield MM, DeClerck F

(2009) Loss of functional diversity under land use

Fulton CJ, Bellwood DR, Wainwright PC (2001) The

relation-ship between swimming ability and habitat use in wrasses

Gatz AJ (1979) Community organization in fishes as indicated

Gray JS (2000) The measurement of marine species

diversity, with an application to the benthic fauna of

the Norwegian continental shelf J Exp Mar Biol Ecol

Green RH, Vascotto GL (1978) A method for the analysis of

environmental factors controlling patterns of species

590

Hertel H (1966) Structure, form and movement Reinhold, New

York

James NC, Whitfield AK, Cowley PD (2008) Long-term

stability of the fish assemblages in a warm-temperate

Karr JR, James FC (1975) Eco-morphological configurations

and convergent evolution of species and communities In:

Cody ML, Diamond JM (eds) Ecology and evolution of

communities Harvard University Press, Cambridge, pp 258–291

Lindsey CC (1978) Form, function, and locomotory habits in fish In: Hoar WS, Randall DJ (eds) Fish physiology:

Lombarte A, Olaso I, Bozzano A (2003) Ecomorphological trends

in Artedidraconidae (Pisces: Perciformes: Notothenioidei) of

Lombarte A, Palmer M, Matallanas J, Gómez-Zurita J, Morales-Nin B (2010) Ecomorphological trends and phylogenetic inertia of otolith sagittae in Nototheniidae.

Lydy MJ, Crawford CG, Frey JW (2000) A comparison of selected diversity, similarity, and biotic indices for detecting changes in bentic-invertebrate community structure and

Macleod N, Forey PL (2002) Introduction: morphology, shape, and phylogenetics In: Macleod N, Forey PL (eds) Morphol- ogy, shape and phylogeny CRC, Boca Raton, pp 1–7 Magurran AE (1988) Ecological diversity and its measurement Princenton University Press, New Jersey, p 179

Mason NWH, Lanoiselee C, Mouillot D, Irz P, Argillier C (2007) Functional characters combined with null models reveal inconsistency in mechanisms of species turnover in

McGill BJ, Enquist BJ, Weiher E, Westoby M (2006) Rebuilding community ecology from functional traits.

Motta PJ, Kotrschal KM (1992) Correlative, experimental, and comparative evolutionary approaches in ecomorphology.

Motta PJ, Clifton K, Hernandez P, Eggold BT (1995) Ecomorphological correlates in ten species of subtropical seagrass fishes: diet and microhabitat utilization Env Biol Fish 44:37–60

Mouillot D, Gaillard S, Aliaume C, Verlaque M, Belscher T, Troussellier M, Chi TD (2005) Ability of taxonomic indices to determinate coastal lagoon environments based

on macrophyte communities Ecol Indic 5:1–17 Neige P (2003) Spatial patterns of disparity and diversity of the recent cuttlefishes (Cephalopoda) across the Old World J

Norton SF, Luczkovich JJ, Motta PJ (1995) The role of ecomorphological studies in the comparative biology of

Olden JD, Poff NL, Bestgen KR (2008) Trait synergisms and the rarity, extirpation, and extinction risk of desert fishes.

Paddack MJ, Cowen RK, Sponaugle S (2006) Grazing pressure

of herbivorous coral reef fishes on low coral-cover reefs.

Petchey OL, Gaston KJ (2006) Functional diversity: back to basics and looking forward Ecol Lett 9:741–758 Pouilly M, Lino F, Bretenoux JG, Rosales C (2003) Dietary- morphological relationships in a fish assemblage of the Bolivian Amazonian floodplain J Fish Biol 62:1137–1158 Pulcini D, Costa C, Aguzzi J, Cataudella S (2007) Light and shape:

a contribution to demonstrate morphological differences in

Recasens L, Lombarte A, Sánchez P (2006) Teleostean fish composition and structure of an artificial reef and a natural

rocky area in Catalonia (North Western Mediterranean).

Bull Mar Sci 78:71–82

Rohlf FJ (2001) TPS Dig 1.31 and TPS relative wards software.

Stony Brook, New York

Rohlf FJ (2003) tpsPLS, partial least-squares, version 1.12.

Stony Brook, New York

Rohlf FJ, Marcus LF (2003) A revolution in morphometrics.

Schoener TW (1974) Resource partitioning in ecological

Shannon CE, Weaver W (1949) The mathematical theory of

communication University of Illinois Press, Urbana

Sheppard JN, James NC, Whitfield AK, Cowley PD (2011)

What role do beds of submerged macrophytes play in

structuring estuarine fish assemblages? Lessons from a

warm-temperate South African estuary Estuarine Coastal

Sokal RR, Michener CD (1958) A statistical method for

evaluating systematic relationships Univ Kans Sci Bull

38:1409–1438

Sokal RR, Rohlf FJ (1962) The comparison of dendrograms by

objective methods Taxon 11:33–40

Somerfield PJ, Olsgard F, Carr MR (1997) A further

examina-tion of two new taxonomic distincness measures Mar Ecol

Villéger S, Ramos-Miranda J, Flores-Hernández D, Mouillot D

(2010) Contrasting changes in taxonomic vs functional

diversity of tropical fish communities after habitat

Violle C, Navas ML, Vile D, Kazakou E, Fortunel C, Hummel

I, Garnier E (2007) Let the concept of trait be functional!

Wagner CE, McIntyre PB, Buels KS, Gilbert DM, Michel E (2009) Diet predicts intestine length in Lake Tanganyika’s cichlid fishes Funct Ecol 23:1122–1131

Wainwright PC, Richard BA (1995) Predicting patterns of prey use from morphology of fishes Env Biol Fish

Warwick RM, Clarke KR (1998) A taxonomic distinctness index

Webb CO (2000) Exploring the phylogenetic structure of ecological communities: an example for rain forest trees.

Whitfield AK (1986) Fish community structure response to major habitat changes within the littoral zone of an estuarine coastal lake Env Biol Fish 17:41–51

Whitfield AK (1992) A characterization of southern African

Whitfield AK (1998) Biology and ecology of fishes in southern African estuaries Ichthyol Monogr JLB Smith Inst

Whittaker RH (1960) Vegetation of the siskiyou mountains.

Winemiller KO (1991) Ecomorphological diversification in lowland fresh-water fish assemblages from 5 biotic

Zelditch ML, Sheets HD, Fink WL (2003) The ontogenic

Body shape variation and colour change during growth

in a protogynous fish

Carmelo Fruciano&Concetta Tigano&

Venera Ferrito

Received: 27 May 2011 / Accepted: 14 November 2011 / Published online: 1 December 2011

# Springer Science+Business Media B.V 2011

Abstract Protogynous sequential hermaphroditism is

very common in marine fish Despite a large number

of studies on various aspects of sequential

hermaph-roditism in fish, the relationship between body shape

and colour during growth in dichromatic species has

not been assessed Using geometric morphometrics,

the present study explores the relationship between

growth, body shape and colouration in Coris julis (L

1758), a small protogynous labrid species with

distinct colour phases Results show that body shape

change during growth is independent of change in

colour phase, a result which can be explained by the

biology of the species and by the social control of sex

change Also, during growth the body grows deeper

and the head has a steeper profile It is hypothesized

that a deeper body and a steeper profile might have afunction in agonistic interactions between terminalphase males and that the marked chromatic differencebetween colour phases allows the lack of strictinterdependence of body shape and colour duringgrowth

Keywords Geometric morphometrics Labridae Labrids Colour change Protogyny Parallelcoordinates

Introduction

Hermaphroditism is widespread in marine fish where

it is present both as simultaneous and sequentialhermaphroditism (Shapiro1979; Warner1984) Whilemany aspects of sequential hermaphroditism—e.g.gonadal tissutal and morphological variation, bodycolour variation and size at sex change—in fish havebeen extensively studied, to the best of the authors’knowledge the relationship between body shape andcolour has not been assessed in fish species in whichdifferent sexual phases show different colouration.This has happened in spite of the fact that the size-advantage hypothesis of sex change and its variationshave been extensively explored (Warner 1984,1988;Muñoz and Warner 2003, 2004; Warner and Muñoz

2008) and in spite of the fact that both body shapefeature and colouration are correlated to reproductive

or territorial success in a few fish species (Warner and

DOI 10.1007/s10641-011-9968-y

Electronic supplementary material The online version of this

article (doi:10.1007/s10641-011-9968-y) contains

supplemen-tary material, which is available to authorized users.

Dipartimento di Scienze Biologiche,

Geologiche e Ambientali, Sezione di Biologia animale

“Marcello La Greca”—University of Catania,

Laboratory for Zoology and Evolutionary Biology,

Universitätsstrasse 10,

78464 Konstanz, Germany

Schultz 1992; Kuwamura et al 2000) The family

Labridae comprises species in which the sexes differ

in colour The first colouration, usually associated to

females, is often called “primary livery” while the

second colouration, usually associated with males, is

called “secondary livery” (Tortonese 1970) Various

labrid species are also diandric—that is, there are

individuals, called “primary males”, which are males

but present the primary livery (Reinboth 1967;

Warner and Robertson 1978) Coris julis (Linnaeus

1758) is a small protogynous diandric labrid species,

which is common along most of the Mediterranean

Sea coasts but also inhabits Eastern Atlantic coasts

Recent studies have shown that C julis lacks genetic

structuring at the Mediterranean scale (Fruciano et al

2011a) but nonetheless shows regional morphometric

variation (Fruciano et al 2011b; Fruciano et al

2011c) and that patterns of morphological variation

in geographic space can be different between colour

phases Coris julis also shows a certain degree of

variability in colour pattern In fact, primary

individ-uals are known to vary in colour from a brown-based

pattern to a reddish pattern as water depth increases

and have been found to be greener in Caulerpa

taxifolia meadows (Michel et al.1987; Arigoni et al

2002) The primary and secondary liveries of C julis

are so different that they were originally described as

two different species: Labrus giofredi (Risso, 1810)

and C julis (L 1758) Behavioural observations

(Lejeune 1982; Bentivegna and Cirino 1984) have

shown that individuals with primary liveries are

sedentary, while individuals with secondary liveries

are territorial and engage in agonistic behaviours

when they meet each other, especially during the

reproductive season Social factors (male/female

ratio) have been shown to induce sexual inversion in

the species (Bentivegna and Cirino 1984)

Histolog-ical aspects of sex inversion in C julis have been

described by Bruslé (1987) and Bentivegna et al

(1985), who also noticed a correlation between colour

phase and gonadic state The variation in colouration

during sex change, which creates a transitional form

with intermediate colouration traits called “transition

livery”, has been described by Bentivegna and Cirino

(1984) who also commented that the changes in

colour marks during transition do not always follow

the same order It is unclear, however, if the hormonal

changes which are believed to result in colour (and

sex) change in this and other labrid species (Reinboth

1975,1988; Reinboth and Brusle-Sicard1997; Frisch

2004; Ohta et al.2008) at the same time also cause avariation in body morphology or if the two processesare at least partially independent In fact, given thatcertain features of body shape have been shown to beimportant in territorial interactions among secondarymales (Warner and Schultz 1992), it could beexpected that a change in sex and colouration wouldresult at the same time in a change in body shape Forthis reason, analysing the relationships between size,shape and colouration—as opposed to studying onlyone trait at a time—can help understanding therelative importance of each morphological feature inthe biology of hermaphrodite fish species Therefore,the aim of this study was to determine if colourationand body shape changes happen simultaneously byusing geometric morphometrics coupled with bothexploratory and hypothesis-testing statistical tools

Materials and methods

Dataset preparation

For the present study, a total of 263 Coris julisspecimens, sampled with fish traps, nets, fishing rodsand hand lines at 9 different Mediterranean sites(Fig 1; Table 1), was used Fish were preserved in95% ethanol and brought to the laboratory for theacquisition of morphometrical data The colour phase

of each specimen was determined by visual inspection

of the colour marks, identifying as transitional theindividuals with colour patterns intermediate betweenprimary and secondary individuals, as described inBentivegna and Cirino (1984)

Pictures of the left side of each specimen weretaken using an Olympus C-3030 digital cameramounted on a copy stand Each specimen wasphotographed two times and points were digitizedtwo times for each pictures, obtaining a total of 4 sets

of coordinates per specimen (such a design wasdeemed appropriate following the results of a prelim-inary study on measurement error on a subset ofspecimens) In the course of the data gathering phase,several measures have been taken to minimize biasand error: the digital camera was relatively distant(495 mm) from the specimens to reduce the effect ofparallax (Mullin and Taylor 2002), fish were keptstraight by running a long needle of appropriate

length through the right side of the body (Windsor

Aguirre, pers comm.) as to limit dorso-ventral

arching, all the steps of the analysis were performed

by the same operator, individuals of one population

were not all photographed and digitized within the

same session but in different ones to avoid any bias in

the way the operator performed his tasks (Windsor

Aguirre, pers comm.); further details on

methodo-logical steps are provided by Fruciano (2009)

Twenty points (Fig.2), comprising both landmarks

(i.e homologous points) and semilandmarks (i.e

points which are not homologous but retain positional

correspondence), were digitized using the software

tpsDig (Rohlf 2006) The landmark/semilandmark

configurations were then subjected to a generalized

Procustes analysis with sliding semilandmarks

(Bookstein 1997) using the software tpsRelw (Rohlf

2007a), setting ten iterations and the minimization of

the squared Procrustes distance as sliding criterion

because this criterion removes all the tangential

varia-tion along outlines (Perez et al.2006)

To reduce both directional and non-directional

measurement error, the full dataset comprising four

landmark configurations for each specimen was

subjected to the procedure described in Valentin et

al (2008), then the resulting coordinates of each

specimen (now adjusted for body arching) were

averaged in order to obtain a single landmark

configuration for each specimen Centroid size (thesquare root of the summed squared distances of eachlandmark from the center of the form; Bookstein

1989) was also computed for each of the fourlandmark configuration per specimen and then aver-aged to obtain an average centroid size per specimen

Statistical analyses

Body shape variation during growth was studied withtwo approaches: regression on a size measure (cen-troid size) and exploratory plots of both relative warpsand relative warps in size-shape space (Mitteroecker

et al.2004)

Regression of shape variables on centroid sizewere performed with tpsRegr (Rohlf 2007b), visual-izing shape variation with a “wireframe graph”produced by the software MorphoJ (Klingenberg

2011)

Relative warps in size-shape space were computed

as explained by Mitteroecker et al (2004), adding tothe usual shape variables the natural logarithm ofcentroid size and then performing a principal compo-nent analysis (PCA) Plots of individual scores on thefirst three PC axes were finally obtained with thesoftware STATISTICA (StatSoft, Inc.)

To further explore the relationships among size,shape and colour phase in more than three dimensions

Fig 1 Geographic

locations for the samples

used in the study.

AU = Augusta, SR, Italy;

LE = Porto Cesareo, LE,

Italy; ML = Badia de Palma,

Mallorca, Spain; MA =

Mazara del Vallo, TP, Italy;

NA = Capo Posillipo,

Naples, Italy; OR =

Oris-tano, Italy; PN = Pantelleria,

Italy; RI = Riposto, CT,

Italy; SP = Split, Croatia

of multivariate space, we also obtained parallelcoordinate plots of relative warps Using parallelcoordinates (Inselberg 1985; Wegman 1990) is anapproach that allows a visualization of data with morethan three dimensions by avoiding the use oforthogonal axes and substituting them with parallelaxes For each observation (individual) the value of acertain variable is represented by a point on thecorresponding vertical axis (each axis represents avariable) The points for each observation on eachaxis are then usually linked by segments so that eachobservation (individual) in a sample is represented by

a poly-line with vertices on the axes with the position

of the i-th vertice corresponding to the value ofthe i-th variable While the parallel coordinatestechnique has been employed in various fields,especially for a visual exploration of data, its use

in biology has been rare (Shapley 2004) and, to thebest of the authors’ knowledge, it has never been usedwith geometric morphometric data

To check for possible bias in the analysis due tounequal sample sizes or to geographic variation, wealso carried out the above-mentioned analyses on asingle population (Mallorca) and on a subset ofgeographically close sampling sites (Augusta,Riposto, Mazara del Vallo, Pantelleria)

To quantify the degree of overlap between theportions of morphospace occupied by primary andsecondary individuals, we computed the ratio of theconvex hull volume (the volume of the n-dimensionalminimal convex set enclosing a certain set ofobservation; see Cornwell et al.2006for an example

in ecology) shared by both primary and secondaryindividuals on the total convex hull volume forprimary and secondary primary and secondary indi-viduals Convex hull computations were performed

on the first ten relative warps using the Quickhullalgorithm (Barber et al.1996) implemented in Octave(http://www.octave.org/)

Results

The regression of shape on centroid size was highlysignificant (Wilk’s Lambda 0.27; p<0.001) andexplained 23.5% of total variance Figure 3 depictsbody shape changes associated with changes in sizeand shows a deepening of body during growth andthat the relative positions of the eye and the forehead