Mar Biol (2010) 157:2397–2406

DOI 10.1007/s00227-010-1504-7

ORIGINAL PAPER

Patterns in Wsh response to seagrass bed loss at the southern Ryukyu Islands, Japan

Yohei Nakamura

Received: 2 March 2010 / Accepted: 22 June 2010 / Published online: 9 July 2010© Springer-Verlag 2010

Abstract An extensive seagrass bed on a fringing coralreef at Amitori Bay (southern Ryukyu Islands) disappearedcompletely in 2009 after a typhoon. Seagrass bed loss had asigniWcant negative inXuence on not only seagrass bed resi-dents but also commercially important coral reef Wshes thatutilize seagrass beds as nurseries or feeding grounds. Withseagrass bed loss, mean species’ richness and densities ofoverall seagrass bed Wshes per transect decreased by morethan 75 and 85%, respectively. Most of the aVected Wsheswere benthivores, piscivores, detritivores, and herbivores.Of 21 dominant species, 13 disappeared completely and 4showed severe reductions in densities following seagrassbed loss, whereas the densities of 4 bottom-dwelling gobiesdid not change signiWcantly. Thus, this study demonstratedthat most seagrass bed Wshes lack the ability to adapt toseagrass habitat loss, suggesting that increasing globalseagrass loss will cause serious reductions in seagrass-associated Wshes and Wshery resources.

Introduction

Coral reefs, mangroves, and seagrass beds are majorfeatures of tropical coastal habitats, and their contributionto ecological and economic values of such habitats havereceived much attention over the last few decades(Birkeland 1997; Nagelkerken 2009). However, these habi-

tats are now rapidly collapsing worldwide, due as much tohuman-induced disturbances corresponding to the growthof coastal human populations (e.g., eutrophication andover-harvesting) as to recent global climate changesand natural disturbances (e.g., outbreaks of consumers andtyphoons) (Hughes et al. 2003; Orth et al. 2006). Estimateshave suggested mean annual rates of recent global loss of1–2% of coral reefs (Gardner et al. 2003; Bruno and Selig2008), 2.1% of mangrove forests (Valiela et al. 2001) and5% of seagrass beds (Waycott et al. 2009).

Several studies of coral reefs and mangroves have dem-onstrated that habitat loss has signiWcant negative eVects onthe productivity and biodiversity of associated Wshes (Joneset al. 2004; Wilson et al. 2006; Mumby et al. 2004). Thedramatic changes in the abundance of several Wsh speciesand trophic guilds in response to habitat degradation oftenlead to a phase shift in Wsh assemblage structure. For exam-ple, Wilson et al. (2006) found an increase in herbivorousWsh abundance in bleached coral areas owing to enhancedalgal growth on dead corals. Shinnaka et al. (2007) sug-gested that mangrove deforestation has led to a decrease inthe numbers of benthic crustacean feeders, whereas anincrease in those of zooplankton feeders. These Wndingssuggest that the loss of coral and mangrove habitats islikely to shift the local ecosystem to an alternative state andnegatively aVect local biodiversity and Wshery resources.

Despite the rapid global decline in seagrass beds over thelast few decades, little is known about the eVects of suchlosses on seagrass-dependent Wshes and Wsheries (Gillanders2006); this is largely due to a low level of public awarenessand the lack of suYcient scientiWc personnel necessary forlong-term monitoring of faunal trends in changing seagrasshabitats (Orth et al. 2006). Seagrass bed degradation mayinvolve several transitional phases. Partial disturbances of alarge seagrass bed may reduce seagrass height/density

Communicated by D. Goulet.

Y. Nakamura (&)Graduate School of Kuroshio Science, Kochi University, 200 Monobe, Nankoku, Kochi 783-8502, Japane-mail: ynakamura@kochi-u.ac.jp

123

2398 Mar Biol (2010) 157:2397–2406

or remove seagrass leaves from some portions of the bed,leading to fragmentation of the habitat into several smallerpatches. When disturbances occur continually, the frag-ments may continue to diminish and eventually vanish(Horinouchi et al. 2009). To date, some studies have inves-tigated the inXuence of seagrass bed fragmentation onWsh populations (Bell et al. 2001; Fernandez et al. 2005;Horinouchi et al. 2009; Macreadie et al. 2009); however,the eVects of complete seagrass bed loss on Wshes have notyet been clariWed in detail (but see Hughes et al. 2002). Themost common approach for understanding these eVects wasa comparison of the faunal composition between areas withseagrass beds and those in which seagrass has been lost ordid not occur (un-vegetated sandy bottom) (Nagelkerkenet al. 2001; Vanderklift and Jacoby 2003; Pihl et al. 2006),although Wndings must be interpreted with caution owing tonatural spatial variability of ecological systems and the biasthat may result.

Amitori Bay is located on northwestern Iriomote Island(southern Japan); fringing reefs are well developed alongthe bay. A seagrass bed (ca. 7 ha) extending onto the reefXats support a high biomass of benthic and epibenthicinvertebrates (Nakamura and Sano 2005) and Wshes(Nakamura and Sano 2004a) and serve as nurseries andfeeding grounds for coral reef Wshes (Nakamura andTsuchiya 2008). In 2009, however, the seagrass bed disap-peared completely and was largely replaced by a bare sandybottom. The reason for such a loss is unknown, although atyphoon has been suggested as a plausible cause. Sinceseagrasses provide living space and food resources for avariety of Wshes and invertebrates, seagrass bed loss islikely to start to shift the local ecosystem function into analternative state. Accordingly, this study examines patternsof Wsh response to seagrass bed loss at the community level(species’ richness and density) and based on functionalgroups (trophic and habitat use types) and dominant spe-cies, including historical data.

Materials and methods

Study site

The study was conducted on the fringing reef at AmitoriBay (24°20�N, 123°42�E), situated on the western side ofIriomote Island, Ryukyu Islands, Japan (Fig. 1). The reefedge along the coast was approximately 270 m from theshore. The coral area, which was 160 m oVshore to the reefedge, primarily comprised live and dead Acropora spp.,Porites spp., and coral rubble. No macroalgal (e.g., Sargas-sum spp.) bed existed in the coral area. The single seagrassbed (7 ha) extended along the coastline between 40and160 m from the shore (ca., 0.5–1 m at low tide; 2–3 m at

high tide) and was dominated by Enhalus acoroides, andother plants including Thalassia hemprichii, Syringodiumisoetifolium, and Cymodocea rotundata. The mean shootdensity of E. acoroides did not change seasonally(117.1 § 30.2 shoots/m2 [mean § standard deviation] inNovember 1999, February, May, and August 2000),although a signiWcant diVerence was observed in seagrassleaf height (23.9 § 6.5 cm and 66.9 § 13.5 cm in Februaryand August 2000, respectively; Nakamura and Sano2004a). The sand area was situated adjacent to the shorelineand was characterized by a sandy bottom lacking vegeta-tion. Water temperature varied from 22 to 29°C seasonally,reaching a maximum in July and August and a minimum inFebruary during the study period.

By mid-August 2009, an extensive seagrass bed in thebay had completely disappeared (Fig. 2), although someseagrass shoots remained. The extensive seagrass bed hadpreviously been observed in 2005 (Nakamura and Tsuchiya2008), but seagrass bed fragmentation occurred in 2007 and2008 (Sato M, personal observation). Although the primaryfactor responsible for the loss remains uncertain, typhoonsShanshan, Wipha, and Morakot, which hit in middle Sep-tember 2006 and 2007 and early August 2009, respectively,remain a plausible cause because strong currents and wavesdue to southerly winds occurred during these typhoons andXoating seagrasses were observed at that time. SigniW-cantly, Amitori Bay is particularly susceptible to thesesoutherly winds due to its topography. Moreover, the slow-moving typhoon Morakot hit during the spring tide periodwhen seagrass beds well exposed during low tides would bemore susceptible to the damage by storm. Seagrass beds oVthe northwestern coasts of Iriomote Island, which were pro-tected from southerly winds by the surrounding mountains,were not lost during this typhoon. In addition, human-induced disturbances (e.g., water pollution and land

Fig. 1 Map of the study areas at Iriomote Island, southern RyukyuIslands, Japan. Dotted line indicates reef margin. (Wlled circle) studysite; (Wlled square) control site

IriomoteIsland

East China Sea

N

24 20‘N

123 42‘E

Study site

0 2.5 5 km

Sonai

Amitori

123

Mar Biol (2010) 157:2397–2406 2399

reclamation) were negligible at the bay as it was uninhabi-tated, except for a research station (Okinawa regionalresearch center, Tokai University).

Sampling design

Fish assemblages in the study area were assessed using anunderwater visual belt transect survey method in late Julyand August 2000, 2001, 2005, and 2009, because this is theseason of high recruitment for several Wshes in the area(Nakamura and Sano 2004a). This nondestructive methodallowed reasonable estimation of the densities of bottom-dwellers and fast-swimmers compared to seine net sam-pling in seagrass beds (Horinouchi et al. 2005) and hasbeen used in several ecological studies of seagrass Wshassemblages (Nagelkerken et al. 2001; Guidetti and Buss-otti 2002; Nakamura and Tsuchiya 2008). In each year,seven 20 £ 1 m belt transects, which were parallel to theshore and separated from each other by at least 10 m, wereestablished randomly in the area (area with seagrass bed inthe former 3 years and seagrass loss in 2009) using a scaledrope. Accordingly, the locations of the transects withinthe area varied from year to year. Each transect wasapproached slowly by a single diver using SCUBA, anddensity and total length (TL) of all Wshes within the transectarea were recorded for 30 min. Each census was conductedat high tide between 1,000 and 1,600 h wherein, each tran-sect was surveyed once. Species’ richness and individualdensity of Wshes in the study area were expressed as the

mean numbers of species and individuals per transect(20 m2, seven transects per year). At some transects, coral-associated juveniles (e.g., pomacentridae and chaetodonti-dae) were observed around small coral patches (size range10–30 cm in diameter) in the seagrass bed. These juvenileswere excluded from all analyses because they were rare (<5individuals during the 4 census years) and their occurrencewas not relevant to this study. In each Wsh census year, Wve50 £ 50 cm quadrats were established randomly in the sea-grass bed and shoot densities of E. acoroides within thequadrats were recorded. Shoot density was expressed as themean density of Wve quadrats, and leaf height wasexpressed as the mean length of 15 leaves chosen randomlywithin the Wve quadrats.

The Sonai coast, which is located 6 km east of AmitoriBay, was selected as an undisturbed reference site, becauseit had an E. acoroides-dominated bed (ca. 5 ha) and showedno apparent seagrass loss. The site was similar to AmitoriBay in some environmental factor, such as water depth,water temperature, salinity (ca. 34.5–35.5‰), underwatervisibility (>15 m), and height and shoot density of seagrassleaves (Table 1). In fact, some diVerences in Wsh assem-blage structure existed between Amitori and Sonai beforethe disturbance (see Results); however, even when the envi-ronmental factors of the sites are similar, Wsh assemblagestructures rarely match completely. Nevertheless, Sonaiwas regarded as a “control” site because the seagrass bedwas quite healthy (undisturbed) and supported severalWshes, and thus, comparison with this site could provideuseful information. Fish assemblage and seagrass charac-teristics (shoot density and leaf height) surveys at the con-trol site were conducted in a similar manner (20 m2, N = 7)and over a similar period as those at Amitori Bay, i.e., in2005 and 2009.

To ascertain changes in the trophic structure of the Wshassemblage following seagrass loss, Wshes surveyed wereassigned to one of six trophic categories based on publisheddietary data of such Wshes at Amitori Bay (Nakamura et al.2003a). These categories included benthivores (which con-sumed benthic invertebrates), herbivores (which fed onseagrasses or algae), planktivores (which fed on planktoniccopepods), omnivores (which consumed benthic inverte-brates and plants), piscivores (which fed on Wshes) anddetritivores (which fed on detritus). Of these, planktivoresand omnivores were excluded from the analysis becausesuch Wshes were rare (<1% for total individual numbers inthe 4 census years at Amitori Bay).

To clarify the manner in which seagrass habitat lossaVected habitat use patterns of seagrass bed Wshes, Wshessurveyed were categorized (Nakamura and Tsuchiya 2008)as follows: (1) permanent residents A, juveniles and adultsliving in seagrass beds as well as other habitats (e.g., coraland sandy bottom habitats); (2) permanent residents B,

Fig. 2 a An Enhalus acoroides-dominated bed at Amitori Bay inAugust 2002. b Seagrass lost area at Amitori Bay on 16 August 2009

b

a

123

2400 Mar Biol (2010) 157:2397–2406

juveniles and adults living only or mainly in seagrass beds;(3) seasonal residents A, juveniles living in seagrass bedsas well as other habitats; (4) seasonal residents B, juvenilesliving only or mainly in seagrass beds; (5) transients, Wshesoccurring in seagrass beds in the course of foraging over avariety of habitats; and (6) casual species, Wshes appearingonly occasionally in seagrass beds.

Data analysis

Seagrass shoot density and leaf height in the 4 census yearsand the eVects of the seagrass loss episode on Wsh assem-blage structures (e.g., species’ richness and density in theoverall assemblage and separate trophic and habitat usegroups) were tested using one-way analysis of variance(ANOVA). If the ANOVA results indicated signiWcantdiVerences during the 4 years (probability level 0.05), theTukey multiple comparison test was used to determinewhich of the means diVered signiWcantly. A two sample ttest was used to examine diVerences in seagrass character-istics (shoot density and leaf height) and Wsh assemblages(species’ richness and density) between Amitori Bay andthe control site in both 2005 and 2009. Prior to the analyses,data were transformed to log (x + 1) and variances weretested for homogeneity using the Bartlett’s test. Kruskal–Wallis, Mann–Whitney U, and Steel–Dwass tests wereperformed instead of parametric tests for heterogeneousvariances.

Results

Comparison of seagrass shoot density and height betweenseagrass replete years (2000, 2001, and 2005) and the sea-grass lost year (2009) at Amitori Bay demonstrated that leafheights were signiWcantly lower in 2009 than in 2000,2001, and 2005, whereas no diVerences were observed inshoot density among the 4 census years (Table 1). Seagrassshoot density and leaf height did not diVer between AmitoriBay and the control site in 2005. In 2009, on the other hand,

although seagrass shoot density did not diVer between thetwo sites, leaf height at Amitori Bay was signiWcantlylower than that at the control site.

The mean species’ and individual numbers of Wshes pertransect were signiWcantly lower in the seagrass lost areathan in the seagrass bed in 2000, 2001, and 2005 at AmitoriBay, these variables decreased by 75–80% and 85–90%,respectively, with seagrass loss, respectively. On the otherhand, species’ and individual numbers of Wshes were highat the control site in 2009 (Fig. 3). A total of 48 specieswere observed during the 4 census years at Amitori Baywith the total number of species observed in the seagrassbed in 2000 and 2001 being ca. 40 compared with 30 in2005 and 12 in the seagrass lost area in 2009.

Species’ and individual numbers of benthivores, pisci-vores, and herbivores did not diVer signiWcantly between

Table 1 Mean shoot density and leaf height of Enhalus acoroides per1 m2 at Amitori Bay, Iriomote Island, before (2000, 2001, and 2005)and after seagrass bed loss (2009), with results of a one-way ANOVA

and Kruskal–Wallis test examining diVerences in the mean shootdensity and leaf height of Enhalus acoroides among 4 years9

9 SigniWcantly diVerent data coded by a and b (P < 0.01 by Steel–Dwass test)99 SigniWcantly diVerence in the mean leaf height of Enhalus acoroides between Amitori Bay and control site in 2009; *P < 0.01

N 2000 2001 2005 2009 P values

Amitori Amitori Amitori Control Amitori Control99

Seagrass shoot density 5 114 § 21 111 § 10 117 § 8 107 § 8 98 § 30 138 § 53 0.4

Seagrass height 15 64 § 14a 66 § 7.5a 70 § 7a 67 § 13 5 § 1b 67 § 11* <0.001

Fig. 3 Mean numbers (+SD) of species and individuals per transect(20 m2, N = 7) at Amitori Bay, Iriomote Island, before (2000, 2001 and2005) and after seagrass bed loss (2009). Numbers above bars indicatetotal species’ numbers. SigniWcant data coded by a and b (P < 0.05).*SigniWcant diVerences in mean species’ number and density of Wshesbetween Amitori Bay and control site in 2005 and 2009: *P < 0.05,**P < 0.01

0

10

20

30

0

50

100

150

AmitoriControl

**

*

2000 2001 2005 2009

**

aa

a

b

aa

a

b

Mea

n no

. of s

peci

es

Mea

n no

. of i

ndiv

idua

ls

3941

30

26

1227

123

Mar Biol (2010) 157:2397–2406 2401

2000, 2001, and 2005, whereas these variables signiWcantlyor marginally decreased in 2009 at Amitori Bay (abundantat control site) (Fig. 4). Species’ and individual numbers ofdetritivores decreased both at Amitori and the control sitein 2005, and few detritivores were observed at Amitori Bayin 2009, but they were abundant at the control site in thesame year. In 2005, no large schools of detritivorous juve-nile scarids were observed at Amitori Bay or the controlsite, and in 2009, detritivorous Calotomus spinidens andLeptoscarus vaigiensis juveniles were completely absent atAmitori Bay (Table 2).

Species’ and individual numbers in all habitat groups,except for the seasonal resident A group, did not diVer sig-niWcantly between seagrass replete years (2000, 2001, and2005), although these variables did decrease signiWcantly in2009 at Amitori Bay. However, these habitat groups wereabundant at the control site in 2005 and 2009 (Fig. 5). Spe-cies’ and individual numbers of the seasonal resident Agroup showed a gradual decrease in 2005 and 2009, par-tially due to the absence of schools of scarid juveniles(Table 2), whereas Wshes of this group were abundant at thecontrol site in 2009.

Comparisons of Wsh abundance for 21 dominant speciesbetween the seagrass replete years and the seagrass lossyear showed that 13 species were absent and four species(Lethrinus harak, Parupeneus barberinus, Stethojulis stri-giventer, and scarid juveniles) marginally or signiWcantly

decreased in the seagrass loss year, whereas the densities offour gobiid species remained stable (Table 2). Species witha signiWcant density decrease in 2009 included a benthivor-ous seasonal resident B (L. atkinsoni), benthivoroustransients (P. ciliatus, P. barberinoides, and P. barberinus),benthivorous permanent resident A (S. strigiventer), piscivo-rous permanent resident A (Cheilio inermis), and herbivo-rous permanent resident B (C. spinidens). All of theseWshes were abundant at the control site.

Cymolutes torquatus was the only species that occurredonly after seagrass bed loss. Although not observed in theseagrass beds (including the control site), two individualsof this species occurred in the transects in the seagrass lostarea.

Discussion

The present study showed that several seagrass-associatedWshes, including both residents and nursery/foraging spe-cies, but not bottom-dwelling gobies, are vulnerable to sea-grass bed loss. Moreover, C. torquatus, a common speciesover sandy areas on reef Xats (Myers 1999), had begun tooccur in the seagrass lost area (i.e., two individualsoccurred in and another two just outside of the transects),suggesting that Wsh assemblage structures would eventuallybecome similar to those of sandy bottom habitats. In an

Fig. 4 Mean numbers (+SD) of species (left side) and individu-als (right side) per transect (20 m2, N = 7) for each trophic group at Amitori Bay, Iriomote Island, before (2000, 2001 and 2005) and after seagrass bed loss (2009). SigniWcant data coded by a, b and c (P < 0.05). *Sig-niWcant diVerences in mean species’ number and density of Wshes between Amitori Bay and control site in 2005 and 2009: *P < 0.05, **P < 0.01

0

5

10

15

20

25

0

1

2

3

4

5

0

1

2

3

4

5

0

2

4

6

0

20

40

60

80

100

0

2

4

6

8

10

0

10

20

30

40

0

15

30

45

60

Benthivores

Piscivores

Herbivores

Detritivores

Mea

n no

. of s

peci

es

Mea

n no

. of i

ndiv

idua

ls

aa

a

b

a a

ab

b

a

ab

ab

b

aa

bc

a

a

a

b

a

aa

b

a

aba

bc

a

a

bc

*

**

**

**

**

**

**

**

*

**

**

*

AmitoriControl

2000 2001 2005 2009 2000 2001 2005 2009

123

2402 Mar Biol (2010) 157:2397–2406

Tab

le2

Mea

n nu

mbe

rs (§

SD

) of

indi

vidu

als

in th

e 21

dom

inan

t spe

cies

(de

term

ined

fro

m 2

000,

200

1, a

nd 2

005

cens

uses

) pe

r tr

anse

ct9 a

t Am

itor

i Bay

, Iri

omot

e Is

land

, bef

ore

(200

0, 2

001,

and

2005

) an

d af

ter

seag

rass

bed

loss

(20

09),

wit

h re

sults

of

a on

e-w

ay A

NO

VA

and

Kru

skal

–Wal

lis te

st e

xam

inin

g diV

eren

ces

in th

e m

ean

num

bers

of

indi

vidu

als

per

tran

sect

for

eac

h sp

ecie

sam

ong

4ye

ars99

9T

rans

ect:

20m

2 , N=

799

Sig

niW

cant

ly d

iVer

ent d

ata

code

d by

a, b

, or

c (P

<0.

05 b

y T

ukey

and

Ste

el–D

was

s te

sts)

999

B, b

enth

ivor

es; P

, pis

civo

res;

D, d

etri

tivo

res;

H, h

erbi

vore

s; D

/H, o

ntog

enet

ic s

hift

in f

ood

habi

ts f

rom

det

ritiv

ores

to h

erbi

vore

s99

99Pr

A, p

erm

anen

t res

iden

t A; P

rB, p

erm

anen

t res

iden

t B; S

rA, s

easo

nal r

esid

ent A

; SrB

, sea

sona

l res

iden

t B; T

r, tr

ansi

ents

§S

igniW

cant

ly d

iVer

ence

s in

the

mea

n in

divi

dual

num

bers

bet

wee

n A

mito

ri B

ay a

nd c

ontr

ol s

ite

in 2

005

and

2009

; *P

<0.

05, *

*P<

0.01

Fam

ilySp

ecie

sT

ropi

c ca

tego

ry99

9H

abita

t ca

tego

ry99

9920

0020

0120

0520

09P

val

ues

Am

itori

Am

itor

iA

mito

riC

ontr

ol§

Am

itori

Con

trol

§

Apo

goni

dae

Apo

gon

ishi

gaki

ensi

sB

PrA

4.9§

8.0

5.6§

13.0

9.0§

15.5

0.6§

1.5

01.

4§

3.8

0.6

Che

ilod

ipte

rus

quin

quel

inea

tus

PSr

A2.

1§

2.5

0.7§

1.3

0.1§

0.4

0.3§

0.6

02.

3§

5.6

0.06

Lut

jani

dae

Lut

janu

s gi

bbus

BSr

A0.

7§

1.9

1.6§

4.2

00

00.

1§

0.4

0.6

Let

hrin

idae

Let

hrin

us o

bsol

etus

BSr

B1.

9§

2.3a

1.0§

1.0ab

0.3§

0.5ab

0.3§

0.5

0b0.

1§

0.4

0.02

Let

hrin

us a

tkin

soni

BSr

B1.

7§

2.0ab

1.9§

1.8ab

1.6§

1.0a

0.9§

0.9

0b0.

6§

0.8

0.03

Let

hrin

us h

arak

BSr

B1.

4§

1.1

1.7§

4.6

2.7§

4.6

1.7§

1.7

0.4§

0.8

0.4§

0.5

0.29

Mul

lida

eP

arup

eneu

s ci

liat

usB

Tr

1.4§

1.0a

0.6§

1.1ab

1.1§

1.2ab

0*0b

0.1§

0.4

0.02

Par

upen

eus

barb

erin

oide

sB

Tr

0.4§

0.8ab

0.9§

1.3ab

2.1§

2.0a

0.1§

0.4*

*0b

0.3§

0.8

<0.

01

Par

upen

eus

mul

tifa

scia

tus

BT

r1.

7§

2.2

1.3§

1.9

0.9§

1.5

0.6§

1.1

00.

9§

0.9*

0.16

Par

upen

eus

barb

erin

usB

Tr

1.4§

1.6ab

2.4§

2.1ab

2.7§

1.3a

0.7§

0.8

0.1§

0.4b

2.1§

3.8

0.02

Lab

rida

eSt

etho

juli

s st

rigi

vent

erB

PrA

17.6§

7.1a

17.3

§11

.7a

16.3§

11.3

a3.

6§

4.2

1.0§

1.5b

16.0

§11

.4**

<0.

01

Che

ilio

iner

mis

PPr

A1.

6§

1.0a

1.9§

1.6a

1.1§

1.1a

0.6§

0.8

0b1.

1§

1.1*

0.01

Scar

idae

Cal

otom

us s

pini

dens

D/H

PrB

7.4§

8.3a

6.6§

2.7a

3.3§

2.5a

1.3§

2.1

0b1.

3§

1.3*

<0.

01

Lep

tosc

arus

vai

gien

sis

D/H

PrB

0.6§

1.0

1.9§

2.4

1.0§

1.2

1.0§

1.2

00.

4§

0.5

0.1

Scar

idae

spp

. (ju

veni

les)

DSr

A17

.1§

18.1

a7.

9§

6.3ab

0.1§

0.4b

0.1§

0.4

0.6§

1.5b

19.0

§16

.8*

<0.

01

Gob

iidae

Cte

nogo

biop

s po

mas

tict

usB

PrA

0.7§

0.8

0.3§

0.7

0.4§

1.1

0.3§

0.8

0.6§

1.0

0.3§

0.5

0.83

Ast

erro

pter

yx s

emip

unct

ata

DPr

A6.

3§

6.2a

5.6§

6.2ac

0.3§

0.8b

00.

1§

0.4bc

0.4§

0.8

<0.

01

Van

derh

orst

ia o

rnat

issi

ma

BPr

A4.

0§

4.7

2.3§

1.9

1.4§

1.6

0.4§

0.8

0.6§

0.8

0.7§

0.8

0.24

Cry

ptoc

entr

us c

aeru

lem

acil

atus

BPr

A9.

9§

6.8

10.9

§6.

03.

7§

3.6

2.1§

1.9

5.3§

4.2

0.7§

1.9*

0.05

Siga

nida

eSi

ganu

s fu

sces

cens

HPr

A1.

0§

1.4

0.4§

2.9

8.3§

10.9

14.4

§10

.50

6.1§

12.0

0.06

Siga

nus

spin

usH

PrA

0.9§

1.2

6.6§

7.7

2.0§

3.4

2.3§

4.1

07.

3§

7.6*

0.04

123

Mar Biol (2010) 157:2397–2406 2403

eelgrass bed in Massachusetts, USA, Hughes et al. (2002)found that 11 of the 13 most common species collecteddeclined in abundance and biomass with the loss of entireseagrass bed, whereas striped bass Morone saxatilis andscup Stenotous chrysops showed a higher mean abundanceafter the loss, the latter two species being common in thebay environment. Although Hughes et al. (2002) did notevaluate demersal Wshes (e.g., gobies), possibly due to thelimited sampling eYciency of the otter trawl used, suchWndings are supported by the present study, demonstratingthat most seagrass bed Wshes cannot adapt well to seagrassloss.

As to the question of the primary reason for Wsh disap-pearance at Amitori Bay in 2009—is this due to seagrassbed loss or other factors?—A possible explanation is natu-ral Wsh extinction independent of seagrass loss. SeveralWshes occurred in the control seagrass bed in 2009, making

this possibility unlikely. Another possible factor is thephysical removal of Wshes together with seagrass by strongtyphoon-generated currents and wave action. However,even if small epiphytic Wshes had been washed away, otherWshes would have been able to escape to adjacent reef slopehabitats from the shallow seagrass bed. Some C. spinidenswere observed in a coral area adjacent to the seagrass lostarea a week after the typhoon. None had returned to theseagrass lost area after the typhoon had passed; however,indicating that this area was no longer a suitable habitat formost seagrass bed Wshes. A subsequent visit to the seagrasslost area 3 months later (November 2009) conWrmed thatthe Wshes had not returned.

The results of the study demonstrated that benthivores,piscivores, herbivores, and detritivores were negativelyaVected by seagrass loss. Nakamura et al. (2003a) reportedthat benthivores at Amitori Bay were the most abundant

Fig. 5 Mean numbers (+SD) of species (left side) and individu-als (right side) per transect (20 m2, N = 7) for each habitat type group at Amitori Bay, Irio-mote Island, before (2000, 2001 and 2005) and after seagrass bed loss (2009). SigniWcant data coded by a, b, and c (P < 0.05). *SigniWcant diVerences in mean species’ number and density of Wshes between Amitori Bay and control site in 2005 and 2009: *P < 0.05, **P < 0.01

0

2

4

6

0

2

4

6

0

4

8

12

16

20

0

2

4

6

0

2

4

6

0

30

60

90

120

0

5

10

15

20

25

0

20

40

60

0

3

6

9

12

15

0

3

6

9

12

15

**

Mea

n no

. of s

peci

es

Mea

n no

. of i

ndiv

idua

ls

2000 2001 2005 2009

Amitori

Control

aa

a

b

aa

a

b

aa

aa

b

a

b

a

a a

bb b

b

a

a

bb

bb

a aa

aa

a

a

aa a

aa

*

**

**

**

*

**

**

**

2000 2001 2005 2009

Permanent resident A

Permanent resident B

Seasonal resident A

Seasonal resident B

Transients

123

2404 Mar Biol (2010) 157:2397–2406

Wshes in number, and that the most important food items forthe seagrass bed Wshes are small benthic and/or epiphyticcrustaceans, such as harpacticoid copepods and gammari-dean amphipods, the densities of which were greater in theseagrass bed than in the adjacent coral and sandy bottomhabitats (Nakamura and Sano 2005). Herbivores and detriti-vores feed predominantly on seagrasses and seagrass/epiphyte-derived detritus, whereas the most dominantpiscivorous Wsh C. inermis feeds on the epiphytic gobyPleurosicya bilobata, along with shrimps and gastropods(Nakamura et al. 2003a). Such food items that are stronglyassociated with seagrass have disappeared or decreasedalong with seagrass habitat loss, which may partly explainthe decrease in some Wshes.

In addition to food resources, seagrasses provide livingspace and protection from predators (Heck et al. 2003; Hor-inouchi 2007). Accordingly, loss of seagrass canopieswould lead to a shortage of sheltered living spaces for per-manent and seasonal residents. In a tethering experiment atAmitori Bay, Nakamura and Sano (2004b) found that thepresence of seagrass cover as shelter had a positive eVecton the survival of juvenile S. strigiventer and Apogonishigakiensis. These species decreased markedly with sea-grass loss in 2009. On the other hand, seagrass loss did notaVect the densities of the bottom-dwelling gobies, namelyCryptocentrus caerlemacilatus, Vanderhorstia ornatissima,Ctenogobiops pomastictus, and Asterropteryx semipunc-tata, suggesting that the sheltered living space function ofseagrasses for these gobies was poor or absent. Indeed, theformer three species often inhabit burrows of alpheidshrimps and were distributed in a sandy bottom habitatadjacent to the seagrass bed, whereas A. semipunctata pre-fers to live in gaps among coral rubble at Amitori Bay(Nakamura and Sano 2004a).

Although August is the season of high Wsh recruitmentat Amitori Bay (Nakamura and Sano 2004a; Nakamuraand Tsuchiya 2008), recently settled juveniles were notobserved in the seagrass lost area, suggesting that Wshlarvae seldom settle or persist in this habitat. In a Weldexperiment using artiWcial seagrass units at Amitori Bay,Nakamura et al. (2003b) found that units with long denseseagrass leaves supported a number of Cheilodipterusquinquelineatus recruits, whereas few or no recruitsappeared on other unit types with shorter or sparser leaves.Moreover, lethrinids use chemical cues for orientation andsettling into seagrass beds (Arvedlund and Takemura2006). Accordingly, they fail to settle on seagrass-freereefs (Nakamura et al. 2009). Thus, seagrass loss haslikely resulted in recruitment failure due to the lack of pre-ferred settlement structure and seagrass-derived olfactorycues.

Mangrove/seagrass habitat loss may exert signiWcantnegative impacts on coral reef Wsh stocks; the reduced

density of some species on such reefs is possibly related tothe absence of former nurseries (Nagelkerken et al. 2002;Dorenbosch et al. 2006). The present study conWrmed thatseagrass loss caused a marked decrease in lethrinidjuveniles that use seagrass bed nurseries (Beck et al. 2001).Generally, several coral reef Wshes similarly characterizedare important Wsheries’ species (Jackson et al. 2001).Lethrinidae and Scaridae, for example, are the most oftenlanded commercial families around the Ryukyu Islands,suggesting that seagrass bed loss may negatively impactWshery stocks in this region.

Although this study was conducted at a single location inthe southern Ryukyu Islands, patterns of Wsh response toseagrass bed loss observed in this study can be predicted tofollow the same general pattern in other instances of sea-grass loss in other locations in the Ryukyu Islands and else-where in the Western PaciWc, because the Wsh fauna andhabitat use patterns of Wshes are similar in this huge area(Jones and Chase 1975; Kochzius 1999; Nakamura andTsuchiya 2008). Moreover, the present study demonstratedthat most seagrass bed Wshes lack the ability to adapt toseagrass loss. An important consideration is whether sea-grass-associated Wshes (especially for Wshes of permanentresident B and seasonal resident B) can persist in alterna-tive habitats if they emigrate from the seagrass lost area.Even if some individuals survive in alternative habitats,these individuals are believed to represent a minority. Onlya few individuals of seagrass-associated Wshes wereobserved in these alternative habitats.

The results of this study were based on only 3-monthobservation after the disturbance; therefore, further long-term observation is needed to conWrm whether thepresent study result reXected the stable condition (i.e.,alternative state) or whether it is just a matter of timebefore the seagrass ecosystem recovers. Nevertheless,seagrass beds have drastically declined throughout theworld in recent years, mainly due to anthropogenicimpacts (e.g., marine pollution) resulting from increasedcoastal populations (Orth et al. 2006; Waycott et al.2009). This suggests that increasing global seagrass losswill pose an increased risk of extinction of seagrass-associated Wshes and Wshery resources and highlights thehigh vulnerability of the seagrass ecosystem and theimportance of its conservation.

Acknowledgments I am grateful to H. Kohno, K. Sakihara,A. Mizutani and the Okinawa Regional Research Center (ORRC),Tokai University, for assistance in the Weldwork. Constructivecomments on the manuscript from G. Hardy and three anonymousreviewers were much appreciated. This study was supported by thegrants from ORRC (No. 01-001, 02-002), the twenty-Wrst century COEProgram of the University of the Ryukyus, and a Grant-in-Aid forYoung Scientists (B) from the Ministry of Education, Culture, Sports,Science and Technology of Japan (No. 21780178).

123

Mar Biol (2010) 157:2397–2406 2405

References

Arvedlund M, Takemura A (2006) The importance of chemical environ-mental cues for juvenile Lethrinus nebulosus Forsskål (Lethrinidae,Teleostei) when settling into their Wrst benthic habitat. J Exp MarBiol Ecol 338:112–122. doi:10.1016/j.jembe.2006.07.001

Beck MW, Heck KL, Able KW, Childers DL et al (2001) The identiW-cation, conservation, and management of estuarine and marinenurseries for Wsh and invertebrates. Bioscience 51:633–641.doi:10.1641/0006-3568(2001)051

Bell SS, Brooks RA, Robbins BD, Fonseca MS, Hall MO (2001)Faunal response to fragmentation in seagrass habitats: implica-tions for seagrass conservation. Biol Conserv 100:115–123.doi:10.1016/S0006-3207(00)00212-3

Birkeland C (1997) Life and death of coral reefs. Chapman and Hall,New York

Bruno JF, Selig ER (2008) Regional decline of coral cover in the Indo-PaciWc: timing, extent, and subregional comparisons. PLos ONE2(8):e711. doi:10.1371/journal.pone.0000711

Dorenbosch M, Grol MGG, Nagelkerken I, van der Velde G (2006)Seagrass beds and mangroves as potential nurseries for the threat-ened Indo-PaciWc humphead wrasse, Cheilinus undulatus andCaribbean rainbow parrotWsh, Scarus guacamaia. Biol Conserv129:277–282. doi:10.1016/j.biocon.2005.10.032

Fernandez TV, Milazzo M, Badalamenti F, D’Anna G (2005) Compar-ison of the Wsh assemblages associated with Posidonia oceanicaafter the partial loss and consequent fragmentation of the meadow.Estuar Coast Shelf Sci 65:645–653. doi:10.1016/j.ecss.2005.07.010

Gardner TA, Cote IM, Gill JA, Grant A, Watkinson AR (2003) Long-term region-wide declines in Caribbean corals. Science 301:958–960. doi:10.1126/science.1086050

Gillanders BM (2006) Seagrasses, Wsh, and Wsheries. In: Larkum WD,Orth RJ, Duarte CM (eds) Seagrasses: biology, ecology and con-servation. Springer, The Netherlands, pp 503–536

Guidetti P, Bussotti S (2002) EVects of seagrass canopy removal onWsh in shallow Mediterranian seagrass (Cymodocea nodosa andZostera noltii) meadows: a local-scale approach. Mar Biol140:445–453. doi:10.1007/s00227-001-0725-1

Heck KL Jr, Hays G, Orth RJ (2003) Critical evaluation of the nurseryrole hypothesis for seagrass meadows. Mar Ecol Prog Ser253:123–136. doi:10.3354/meps253123

Horinouchi M (2007) Review of the eVects of within-patch scale struc-tural complexity on seagrass Wshes. J Exp Mar Biol Ecol350:111–129. doi:10.1016/j.jembe.2007.06.015

Horinouchi M, Nakamura Y, Sano M (2005) Comparative analysis ofvisual censuses using diVerent width strip-transects for a Wshassemblage in a seagrass bed. Estuar Coast Shelf Sci 65:53–60.doi:10.1016/j.ecss.2005.05.003

Horinouchi M, Tongunui P, Nanjo K, Nakamura Y, Sano M, Ogawa H(2009) DiVerences in Wsh assemblages between fragmented andcontinuous seagrass beds in Trang, southern Thailand. Fish Sci75:1409–1416. doi:10.1007/s12562-009-0166-1

Hughes JE, Deegan LA, Wyda JC, Weaver MJ, Wright A (2002) TheeVects of eelgrass habitat loss on estuarine Wsh communities of south-ern New England. Estuaries 25:235–249. doi:10.1007/BF02691311

Hughes TP, Baird AH, Bellwood DR, Card M et al (2003) Climatechange, human impacts, and the resilience of coral reefs. Science301:929–933. doi: 10.1126/science.1085046

Jackson EL, Rowden AA, Attrill MJ, Bossey SJ, Jones MB (2001)The importance of seagrass beds as a habitat for Wsheries species.Oceanogr Mar Biol Annu Rev 39:269–303

Jones RS, Chase JA (1975) Community structure and distribution ofWshes in an enclosed high island lagoon in Guam. Micronesica11:127–148

Jones GP, McCormick MI, Srinivasan M, Eagle JV (2004) Coraldecline threatens Wsh biodiversity in marine reserves. Proc NatlAcad Sci USA 101:8251–8253. doi:10.1073/pnas.0401277101

Kochzius M (1999) Interrelation of ichthyofauna from a seagrassmeadow and coral reef in the Philippines. In: Séret B, Sire JY(eds) Proc 5th Indo-PaciWc Fish Confer, Paris, pp 517–535

Macreadie PI, Hindell JS, Jenkins GP, Connolly RM, Keough MJ(2009) Fish response to experimental fragmentation of seagrasshabitat. Conserv Biol 23:644–652. doi:10.1111/j.1523-1739.2008.01130x

Mumby PJ, Edwards AJ, Arias-Gonzalez JE, Lindeman KC et al(2004) Mangrove enhance the biomass of coral reef Wsh commu-nities in the Caribbean. Nature 427:533–536. doi:10.1038/nature02286

Myers RF (1999) Micronesian reef Wshes: a comprehensive guide tothe coral reef Wshes of Micronesia. Coral Graphics, Guam

Nagelkerken I (2009) Ecological connectivity among tropical coastalecosystems. Springer, New York. doi:10.1007/978-90-481-2406-0

Nagelkerken I, Kleijnen S, Klop T, van den Brand RACJ, Cocheret dela Morinière E, van der Velde G (2001) Dependence of Caribbeanreef Wshes on mangroves and seagrass beds as nursery habitats: acomparison of Wsh faunas between bays with and without man-grove/seagrass beds. Mar Ecol Prog Ser 214:225–235.doi:10.3354/meps214225

Nagelkerken I, Roberts CM, van der Velde G, Dorenbosch M, van RielMC, Cocheret de la Morinière E, Nienhuis PH (2002) Howimportant are mangroves and seagrass beds for coral-reef Wsh?The nursery hypothesis tested on an island scale. Mar Ecol ProgSer 244:299–305. doi:10.3354/meps244299

Nakamura Y, Sano M (2004a) Overlaps in habitat use of Wshesbetween a seagrass bed and adjacent coral and sand areas atAmitori Bay, Iriomote Island, Japan: importance of the seagrassbed as juvenile habitat. Fish Sci 70:788–803

Nakamura Y, Sano M (2004b) Is there really lower predation risk forjuvenile Wshes in a seagrass bed compared with an adjacent coralarea? Bull Mar Sci 74:477–482

Nakamura Y, Sano M (2005) Comparison of invertebrate abundance ina seagrass bed and adjacent coral and sand areas at Amitori Bay,Iriomote Island, Japan. Fish Sci 71:543–550. doi:10.1111/j.1444-2906.2005.00998.x

Nakamura Y, Tsuchiya M (2008) Spatial and temporal patterns of sea-grass habitat use by Wshes at the Ryukyu Islands, Japan. EstuarCoast Shelf Sci 76:345–356. doi:10.1016/j.ecss.2007.07.014

Nakamura Y, Horinouchi M, Nakai T, Sano M (2003a) Food habits ofWshes in a seagrass bed on a fringing coral reef at Iriomote Island,southern Japan. Ichthyol Res 50:15–22. doi:10.1007/s102280300002

Nakamura Y, Horinouchi M, Sano M (2003b) InXuence of seagrassleaf density and height on recruitment of the cardinalWshCheilodipterus quinquelineatus in tropical seagrass beds: anexperiment study using artiWcial seagrass units. La Mer 41:192–198

Nakamura Y, Shibuno T, Lecchini D, Watanabe Y (2009) Habitatselection by emperor Wsh larvae. Aquat Biol 6:61–65.doi:10.3354/ab00169

Orth RJ, Carruthers TJB, Dennison WC, Duarte CM et al (2006) A glo-bal crisis for seagrass ecosystems. Bioscience 56:987–996.doi:10.1641/0006-3568(2006)56

Pihl L, Baden S, Kautsky N, Ronnback P, Soderqvist T, Troell M,Wennhage H (2006) Shift in Wsh assemblage structure due to lossof seagrass Zostera marina habitats in Sweden. Estuar CoastShelf Sci 67:123–132. doi:10.1016/j.ecss.2005.10.016

Shinnaka T, Sano M, Ikejima K, Tongnunui P, Horinouchi M,Kurokura H (2007) EVects of mangrove deforestation on Wsh

123

2406 Mar Biol (2010) 157:2397–2406

assemblage at Pak Phanang Bay, southern Thailand. Fish Sci73:862–870. doi:10.1111/j.1444-2906.2007.01407.x

Valiela I, Bowen JL, York JK (2001) Mangrove forests: one of theWorld’s threatened major tropical environments. Bioscience51:807–815. doi:10.1641/0006-3568(2001)051

Vanderklift MA, Jacoby CA (2003) Patterns in Wsh assemblages25 years after major seagrass loss. Mar Ecol Prog Ser 247:225–235. doi:10.3354/meps247225

Waycott M, Duarte CM, Carruthers TJB, Orth RJ et al (2009) Acceler-ating loss of seagrasses across the globe threatens coastal ecosys-tems. Proc Nat Acad Sci USA 106:12377–12381. doi:10.1073/pnas.090560106

Wilson SK, Graham NAJ, Pratchett MS, Jones GP, Polunin NVC(2006) Multiple disturbances and the global degradation of coralreefs: are reef Wshes at risk or resilient? Global Change Biol12:2220–2234. doi:10.1111/j.1365-2486.2006.01252.x

123