Content uploaded by Amanda Vincent
Author content
All content in this area was uploaded by Amanda Vincent on Jul 11, 2015
Content may be subject to copyright.
MARINE ECOLOGY PROGRESS SERIES
Mar Ecol Prog Ser
Vol. 291: 81–91, 2005
Published April 28
INTRODUCTION
Seagrasses are productive, biogenic habitats (Pollard
1984, Sheperd et al. 1989, Short & Wyllie-Escheverria
1996) that usually support a greater abundance and
diversity of fishes than surrounding unvegetated habi-
tats (Heck et al. 1989, Edgar & Shaw 1995, Jenkins et
al. 1997). This pattern has been attributed to lower pre-
dation risk (Choat 1982, Orth et al. 1984, Hindell et al.
2000), greater food availability (Edgar 1990), increased
sediment stability, and refuge from hydrodynamic
forces within seagrasses (Lewis 1984, Dean & Connell
1987). While hydrodynamic processes influence re-
cruitment patterns (Jenkins et al. 1997, Stoner 2003),
many marine species actively select seagrass beds over
unvegetated habitats, as well as microhabitats of dif-
ferent complexities within seagrasses (Bell & Westoby
1986b, Edgar 1990). Higher densities inside versus out-
side seagrass beds were noted for many species in
the family Syngnathidae (seahorses, pipefishes, pipe-
horses and seadragons) (Teixeira & Musick 1995, Diaz-
Ruiz et al. 2000, Kendrick & Hyndes 2003).
As heterogeneous habitats that vary in the degree of
structural complexity and exposure to tidal regimes
© Inter-Research 2005 · www.int-res.com*Email: janelle.curtis@elf.mcgill.ca
Distribution of sympatric seahorse species along
a gradient of habitat complexity in a seagrass-
dominated community
Janelle M. R. Curtis
1, 2,
*
, Amanda C. J. Vincent
1, 2
1
Department of Biology, McGill University, 1205 Dr. Penfield Avenue, Montréal, Québec H3A 1B1, Canada
2
Project Seahorse, Fisheries Centre, University of British Columbia, 2204 Main Mall, Vancouver,
British Columbia V6T 1Z4, Canada
ABSTRACT: We present estimates of local population abundance, distribution and habitat prefer-
ence for 2 European seahorse species, Hippocampus guttulatus and H. hippocampus. We predicted
that these sympatric species would partition their habitat into 2 broadly defined habitat types: com-
plex vegetated habitats and sparsely vegetated sand flats. We sampled populations using underwa-
ter visual census techniques over landscape (100s to 1000s m) and microhabitat scales (<1 m). Over
landscape scales, we estimated abundance and quantified habitat associations using generalized
linear models. Over microhabitat scales, we tested for holdfast (attachment site) preferences using
selection indices. Both species were patchy in distribution, but H. guttulatus mean density
(0.073 ind. m
–2
) was one order of magnitude greater than that of H. hippocampus (0.007 ind. m
–2
). At
a landscape scale, H. guttulatus abundance was positively correlated with an index of habitat com-
plexity, the percentage of substrate covered by flora and sessile fauna. Conversely, H. hippocampus
used more open and less speciose habitats that were subjected to greater oceanic influences. At
microhabitat scales, both species significantly preferred grasping holdfasts over barren surfaces, but
the species differed in holdfast preferences: H. guttulatus grasped all prospective holdfast types with
equal probability while H. hippocampus significantly avoided both fauna and flora that formed large
colonies or tracts of dense vegetation. Patterns in habitat use were consistent with differences in
morphology and foraging strategy. Despite similar life histories, these sympatric species may respond
differently to disturbances that modify habitat structure and complexity over landscape or micro-
habitat scales.
KEY WORDS: Habitat selection · Habitat complexity · Hippocampus · Syngnathidae · Habitat suit-
ability models · Underwater visual census · Foraging strategy
Resale or republication not permitted without written consent of the publisher
Mar Ecol Prog Ser 291: 81–91, 2005
(Bell & Westoby 1986a,b, Hovel et al. 2002), seagrasses
provide opportunities for sympatric species to partition
their habitat over multiple spatial scales. The effects of
habitat structure and complexity on the diversity and
abundance of seagrass-associated species are well
documented (e.g. Lewis 1984, Bell & Westoby 1986a,b,
Dean & Connell 1987, Heck et al. 1989, Edgar 1990,
Hovel et al. 2002, Hyndes et al. 2003). However, few
studies have contrasted the effects of seagrass bed
structural complexity on the habitat partitioning of
closely related species with similar life histories.
Within the Syngnathidae, sympatric species of
pipefish partition habitat both within and among sea-
grass beds according to their morphology, mobility,
foraging technique and prey use (Howard & Koehn
1985, Kendrick & Hyndes 2003). Seagrass-dwelling
syngnathids are generally cryptic and sedentary, and
either occupy the canopy or reside at the sediment-
water interface (Bell & Westoby 1986a, Teixeira &
Musick 1995). Howard & Koehn (1985) showed that
less mobile syngnathid species (inferred from prehen-
sile tails) consumed mainly planktonic prey and may
have relied more on dense macrophyte canopies, while
more mobile species (inferred from well developed
caudal fins) consumed both planktonic and epibenthic
prey and likely used a wider range of microhabitats in-
cluding bare substrate.
Seahorse species (genus Hippocampus) in temperate
regions are typically associated with seagrass habitats
(Foster & Vincent 2004). Seagrasses allow for crypsis,
provide holdfasts (attachment sites for prehensile tails)
and harbour abundant food for syngnathids, such as
small crustaceans (Howard & Koehn 1985, Tipton &
Bell 1988, Foster & Vincent 2004). However, anecdotal
observations suggest that some species also exploit
bare substrates: Hippocampus abdominalis, H. capen-
sis, H. guttulatus, H. hippocampus and H. kuda have
been encountered on open tracts of sand and/or distant
from potential holdfasts (Bell et al. 2003, Garrick-
Maidment & Jones 2004, S. Lourie pers. comm., K.
Martin-Smith pers. comm.). Laboratory experiments
also show that seahorses employ different foraging
strategies in vegetated and unvegetated habitats
(James & Heck 1994). Thus, some seahorse species
may use barren surfaces as well as more complex veg-
etation.
Hippocampus guttulatus (Cuvier 1829) and H. hip-
pocampus L. appear to partition their habitat over both
landscape and microhabitat scales (Boisseau 1967,
Lythgoe & Lythgoe 1971, Whitehead et al. 1986, Reina-
Hervás 1989). Both species occur in the northeastern
Atlantic Ocean and Mediterranean Sea (Boisseau
1967, Whitehead et al. 1986, Reina-Hervás 1989,
Lourie et al. 1999). While H. guttulatus is usually re-
ported from seagrass beds, H. hippocampus is re-
ported from soft bottoms among rocks and algae. Hip-
pocampus guttulatus tends to be dark green or brown
in colouration and commonly bears skin filaments,
suggesting that this species employs vegetated micro-
habitats. Conversely, H. hippocampus is variable in
colouration with fewer or no skin filaments, suggesting
that this species relies less on flora or filamentous
structures for crypsis. Recreational diver surveys sup-
port these inferences (Garrick-Maidment & Jones
2004). However, no previous studies have examined
the extent to which local abundances of these sym-
patric species can be predicted by habitat structure.
Our study examined the distribution and abundance
of Hippocampus guttulatus and H. hippocampus in a
coastal lagoon along a habitat gradient from sheltered
seagrass and macroalgae to largely unvegetated sand
flats exposed to increased tidal flow and water depth.
Despite conservation concern (Santos et al. 1995, Fos-
ter & Vincent 2004, IUCN 2004), this is one of few eco-
logical studies of either H. guttulatus or H. hippo-
campus. Our objectives were to (1) estimate local
population abundances, (2) quantify differences in
habitat preference over landscape and microhabitat
scales and (3) identify components of habitat structure
for predicting local abundances. Given the biological
characteristics of the study species, we predicted that
H. guttulatus uses vegetated areas, while H. hippo-
campus uses more open sandy habitats over landscape
scales. We further predicted that within these 2
broadly defined habitat types, both species would pre-
fer microhabitats with potential holdfasts. We expected
that habitat partitioning over both landscape and
microhabitat scales would reflect interspecific differ-
ences in morphology and foraging strategy. We evalu-
ated these predictions through a multi-scale sampling
program covering the lagoon system.
MATERIALS AND METHODS
Species descriptions. We adopted the taxonomy of
the genus Hippocampus as outlined in Lourie et al.
(1999), recognizing that H. guttulatus was historically
synonymized with H. ramulosus (Leach 1814). Further
research is required to clarify whether H. ramulosus
represents a different species (Lourie et al. 1999). Hip-
pocampus guttulatus and H. hippocampus are readily
distinguished in situ by differences in head, snout and
trunk shape (Lourie et al. 1999).
Measuring seahorses precisely is challenging because
of the curvature of the trunk and tail and because the
head is held at an angle to the trunk. We used the
measurement protocol outlined in Lourie et al. (1999)
with one important distinction: we measured lengths
as straight lines between the appropriate reference
82
Curtis & Vincent: Ecology of sympatric seahorses
points, with the head held at a right angle to the body.
All measurements reported in our paper are standard
lengths, except where stated otherwise.
Published details about the life histories of Hippo-
campus guttulatus and H. hippocampus are sparse
(Boisseau 1967, Foster & Vincent 2004). In our study,
adult H. guttulatus ranged in size from 108 to 210 mm
(n = 384). H. hippocampus was on average ~40%
smaller, with adults ranging from 87 to 146 mm
(n = 41). Juveniles of both species were planktonic
(Boisseau 1967, Pérez-Ruzafa et al. 2004). The smallest
settled H. guttulatus and H. hippocampus encountered
in this study were 65 mm and 62 mm, respectively.
Site description. We carried out our study in the
western and central parts of the Ria Formosa lagoon
(36°59’N, 7°51’W), in southern Portugal (Fig. 1). The Ria
Formosa is a productive coastal lagoon (Sprung 1994a)
characterized by high water turnover rates, sand flats,
salt marshes and a network of channels and tidal creeks
(Sprung 1994a, Machás & Santos 1999, Newton &
Mudge 2003). From July 2001 to September 2002, wa-
ter temperature varied seasonally from 10 to 28°C.
Salinity ranged very little (36 to 38 ppt) among our sam-
pling sites, but had declined occasionally to 33 ppt in
the Ria Formosa during previous winters (Monteiro
1989). Water is exchanged with the open ocean through
6 tidal inlets, 2 of which have been dredged. Tides are
semidiurnal and mean depth varies from 2 to 3 m, but
can be as great as 15 m in the main Faro-Olhão channel
(between Sites 11 and 22, Fig. 1). Total lagoon area is
~170 km
2
(Machás & Santos 1999) with an estimated
subtidal area of 26.7 km
2
(A. Rodrigues de Matos, Par-
que Natural da Ria Formosa, pers. comm.).
Underwater visual census. We surveyed 32 sites
(Fig. 1) using SCUBA: 16 sites were surveyed from 13
August to 19 September 2001 and 16 were surveyed
from 14 June to 22 August 2002. In 2001, coincident
with another study (Erzini et al. 2002), sites were
selected along a gradient of ocean influence. To verify
2001 results, sites were selected in 2002 from through-
out the subtidal area. We did not stratify random sam-
pling by habitat type (e.g. seagrass, macroalgae, sand
flat) in either year because the distribution of vegeta-
tion had not been mapped at the time of our study and
we could not discern habitat type from the surface
because of low water clarity (usually <1 m). Some parts
(~20%) of the subtidal area were not accessible
because of sand bars, strong currents (e.g. near inlets)
or high volumes of boat traffic, therefore sites were
surveyed as close to the pre-selected coordinates as
possible. Surveys were carried out using standard
underwater visual census (UVC) techniques (Samoilys
1997). Three randomly placed 2 m × 30 m belt tran-
sects, placed >5 m apart, were surveyed per site (total
surveyed area = 5760 m
2
). Data were averaged among
transects within sites. All sites were >100 m apart.
Fish counts and observations. The species, sex, trunk
length, life history stage (juvenile, adult), holdfast, ap-
pearance (colouration, presence of skin filaments) and
behaviour were recorded for all seahorses encountered
on transects. We used trunk lengths (cleithral ring to
the last trunk ring) to measure live seahorses in situ be-
cause the precision was greater than when measuring
standard length (tip of the snout to the cleithral ring,
and from the cleithral ring to the tip of the straightened
tail), as the tail did not need to be straightened. Trunk
lengths were converted to standard lengths using
regression equations developed for both species
(Table 1). Hippocampus guttulatus and H. hippocam-
pus were considered subadults if they were <50% size
at maturity, corresponding to 109 mm and 78 mm SL,
respectively (Curtis 2004). In order to characterize
interspecific differences in activity patterns associated
with habitat use, behaviour was noted upon detection
83
Fig. 1. Western and central
part of the Ria Formosa
lagoon in southern Portu-
gal (inset). Solid (2001) and
open (2002) circles corre-
spond to the location of 32
underwater visual census
(UVC) sites (site numbers
given)
Mar Ecol Prog Ser 291: 81–91, 2005
and was classified into 4 categories: (1) stationary, (2)
swimming, (3) ambush predation, when individuals
were observed rapidly flicking their heads and sucking
from the plankton while in a stationary position, and (4)
active foraging, when individuals were observed feed-
ing from the plankton, sediments or vegetation while
actively searching for prey.
Few individuals (<1%) swam away from observers
during UVCs; therefore surveys were not time
constrained. Using the same experienced observer
(J.M.R.C.) for all UVCs further reduced observer bias.
Population sizes were estimated by extrapolating
mean densities to the subtidal area. Confidence inter-
vals (95% CI) were estimated by bootstrapping the
density estimates (Efron & Tibshirani 1993).
Benthic community and habitat characteristics. The
abundance of conspicuous (>2.5 cm in height, length
or width) seagrasses, macroalgae, invertebrates (ses-
sile and relatively immobile) and artificial structures
was estimated as percent cover, defined here as the
percentage of the seafloor that was overlaid by a given
habitat component. Percent cover was quantified using
three 30 m line intercepts (line-intercept method,
Samoilys 1997) per site (averaged within sites). These
line intercepts corresponded to the same transects
used during fish counts. Microhabitat patch sizes (e.g.
bryozoan colonies, tufts of macroalgae) were estimated
by measuring the length of the transect line that cov-
ered patches. The mean density of seagrass short
shoots and mean leaf height were estimated using 5
randomly placed quadrats of 0.01 m
2
per transect
searched. The presence of mobile benthic inverte-
brates (e.g. Aplysia spp., Sepia oficinalis) and fishes
(e.g. Halobatrachus didactylus) was recorded to esti-
mate total species richness (flora and fauna) at each
site. Two composite variables were also considered:
(1) an index of habitat complexity, C
t
/A
t
(Bartholomew
et al. 2000), which corresponded to the area surveyed
within a site (A
t
) that was covered by vegetation or ses-
sile invertebrates (C
t
), and (2), M
t
/A
t
, which denotes
the proportion of A
t
that was covered by all species of
macroalgae (M
t
).
Depth was measured at the beginning and end of
each transect and averaged within sites (all UVCs
were carried out within 2 h of low
tide). Distances to the nearest tidal
inlet were estimated planimetrically.
Substrate particle size structure was
characterized by fractionating 3 ran-
domly placed sediment cores (0.02 m
diameter × 0.01 m depth) per site into
4 size classes (<110 µm, 110–1000 µm,
1000–1800 µm and >1800 µm), drying
them to constant weight, and estimat-
ing percent composition by weight.
Potential covariates were recorded and included date,
tidal and lunar phases, start time, horizontal visibility,
current strength and prevailing weather conditions
(wind strength and % cloud cover).
Landscape level habitat use. The abundance of Hip-
pocampus guttulatus approximated a Poisson distribu-
tion and we assumed the same for H. hippocampus.
The landscape level relationships between seahorse
counts and habitat variables were therefore character-
ized using generalized linear models (GLMs), assum-
ing a Poisson distribution (McCullagh & Nelder 1989)
with a log link function. Prior to model selection, the
data from both years were summarized in Spearman
rank correlation matrices to identify collinear variables
as well as suites of variables that were broadly corre-
lated with seahorse abundance. Poisson regressions
were carried out for both species in 2001, but only for
H. guttulatus in 2002, when few H. hippocampus were
observed. To evaluate the predictive accuracy of our
GLMs, we compared the densities of H. guttulatus
observed in 2002 to expected values based on the 2001
GLM for H. guttulatus. We used a jackknife resam-
pling procedure to test whether the correlation coeffi-
cient between observed and expected values differed
significantly from unity (Manly 1997). In all cases,
model selection was carried out using both forward
and backward selection, which produced the same
results. Variables were retained in the model if
p-values associated with type III Wald tests were
<0.05. Final regression models are reported. Data
were examined for sex- and stage-specific differences
in habitat associations.
Microhabitat and holdfast preferences. Habitat
preference is a measure of the strength of selection of
one habitat component over others (Johnson 1980).
The standardized selection index (Krebs 1999) was
used to test whether Hippocampus guttulatus and
H. hippocampus preferred using microhabitats that
offered potential holdfasts (covered substrate) to
using bare surfaces. The selection index was esti-
mated as U
t
/(C
t
/A
t
), where U
t
was the proportion of
individuals grasping a holdfast. A selection index >0.5
indicated relative preference for covered substrate,
while a selection index <0.5 indicated a relative pref-
84
Table 1. Hippocampus guttulatus and H. hippocampus. Regression equations for
predicting standard length (SL) from trunk length (TrL) for males (m) and females (f)
(Curtis 2004)
Species Sex Equation n r
2
p-value
H. guttulatus m logTrL = –0.325 + 0.872 × logSL 703 0.86 <0.0001
f logTrL = –0.474 + 0.950 × logSL 637 0.85 <0.0001
H. hippocampus m logTrL = –0.232 + 0.811 × logSL 56 0.90 <0.0001
f logTrL = –0.593 + 1.008 × logSL 47 0.92 <0.0001
Curtis & Vincent: Ecology of sympatric seahorses
erence for bare substrate (Krebs 1999). Rank prefer-
ence indices (Johnson 1980, Krebs 1999) for using dif-
ferent microhabitats as holdfasts were calculated
using PREFER 5.1 (Pankratz 1994). Data were exam-
ined for sex and stage-specific differences in micro-
habitat preferences.
RESULTS
Seahorse abundance and distribution
Pooling across years and sites, Hippocampus guttu-
latus was 10 times more abundant than H. hippo-
campus and occupied more than twice
the number of survey sites (Table 2).
Hippocampus guttulatus was en-
countered at all but 3 of the 32 UVC
sites. By contrast, H. hippocampus
occurred primarily in large channels
that were influenced by water flowing
through tidal inlets (Fig. 2a).
Site characteristics
Survey sites were located from 1.0 to
6.9 km from the open Atlantic Ocean
in 0.25 to 6.75 m of water depth. Hori-
zontal visibility ranged from 0.3 to
6.0 m. Approximately 70% of the total
surveyed area (all sites pooled) was
bare substrate (fine sand, coarse sand
or shell fragments). Of the total area
that was covered, 58.6% was sea-
grass, 20.8% was macroalgae and
17.8% was invertebrates (Table 3).
Fourteen sites were on sand flats, 7
sites were dominated by the seagrass
Cymodocea nodosa, 4 sites were dom-
inated by a mixture of C. nodosa and
macroalgae (mostly Ulva lactuca and
Codium spp.) and 7 sites were domi-
nated by a mixture of macroalgae
and colonies of benthic invertebrates
(mostly the bryozoan Zoobotryon ver-
ticillatum) interspersed with sand flats
(Fig. 2b). Where seagrass (C. nodosa)
was present, mean leaf height ranged
from 11.1 to 34.2 cm and mean short
shoot density ranged from 233.3 to
848.3 shoots m
–2
.
Landscape level habitat use
Hippocampus guttulatus and H.
hippocampus were broadly associated
with different suites of variables
(Table 4). After applying a sequential
Bonferroni correction to adjust p-
values for multiple correlations, the
only significant correlation was be-
85
Table 2. Hippocampus guttulatus and H. hippocampus. Indices of abundance.
Occupancy (%) refers to the percentage of sites where a species was detected
and n denotes number of individuals encountered during UVCs (CI = bootstrap-
estimated confidence limit)
Abundance index H. guttulatus H. hippocampus
2001 2002 Pooled 2001 2002 Pooled
years years
Total n 168 216 384 39 2 41
Juveniles n 026 022 048 01001
% Occupancy 87.5 93.8 90.6 62.5 12.5 37.5
Max. density (m
–2
) 0.27 0.51 0.51 0.072 0.008 0.072
Mean density (m
–2
) 0.060 0.085 0.073 0.013 0.001 0.007
Lower 95% CI 0.039 0.003
Upper 95% CI 0.110 0.014
Population size 1 895 700 186 900
Lower 95% CI 1 041 300 _80 100
Upper 95% CI 3 043 800 373 800
Fig. 2. (a) Relative frequencies of Hippocampus guttulatus (white) and H. hippo-
campus (black) at each site. The total number of seahorses encountered per site
is given. (b) The dominant habitat type at 32 UVC sites: sand flats (s), seagrass
beds (d), mixed seagrass and macroalgae ( ) and mixed macroalgae and
sessile invertebrates interspersed with bare sand ( )
a
b
Mar Ecol Prog Ser 291: 81–91, 2005
tween H. guttulatus density and the % cover of Ulva
lactuca. However, by examining the direction of the
correlations, the correlation matrix suggested that H.
guttulatus density was greater at sites with greater
habitat complexity (i.e. sites with greater seagrass
shoot density, vegetation cover and sessile inverte-
brates), and thus less abundant at sites with lower
habitat complexity (i.e. sites with a greater amount of
bare sand and shell fragments). Conversely, H. hippo-
campus was generally more abundant in sites that had
low habitat complexity (i.e. less speciose sites with a
greater amount of bare sand). H. hippocampus density
was also greater in sites with stronger oceanic influ-
ences, reflected in positive correlations with horizontal
visibility, depth and current strength, and a negative
correlation with water temperature.
In the GLMs, C
t
/A
t
was the most significant predic-
tor of Hippocampus guttulatus abundance in both
years, but M
t
/A
t
was also retained in the 2001 model
and % Ulva lactuca was retained in the 2002 model
(Fig. 3a, Table 5). There was a significant correlation
between the observed and expected H. guttulatus den-
sities for 2002 (r = 0.471, p = 0.033), but the correlation
was significantly less than unity (jackknifed r = 0.488,
SE = 0.203, t = 2.52, p < 0.02), meaning that the pre-
dicted values were significantly greater than densities
86
Table 4. Non-parametric Spearman rank correlations (r
s
) be-
tween habitat variables and seahorse density (data pooled
across all sites). Included are habitat variables that were most
highly correlated (r
s
> 0.3, p < 0.1) with at least one of the spe-
cies (other variables are not shown). *Significant after apply-
ing a sequential Bonferroni correction to adjust p-values for
the multiple tests in this table
Habitat variable Hippocampus Hippocampus
guttulatus hippocampus
r
s
pr
s
p
% Cymodocea nodosa 0.406 0.021 0.196 0.282
C. nodosa shoot density 0.408 0.100 –0.197– 0.305
% Ulva lactuca 0.557* 0.001 –0.059– 0.750
% Anemone –0.334– 0.064 –0.193– 0.289
M
t
/A
t
0.438 0.012 0.233 0.199
C
t
/A
t
0.512 0.003 –0.263– 0.145
Species richness 0.220 0.226 –0.386– 0.029
% Sand –0.249– 0.169 0.333 0.063
% Shell fragments –0.433– 0.013 0.142 0.437
Horizontal visibility –0.162– 0.376 0.373 0.035
Water temperature –0.264– 0.144 –0.372– 0.036
Depth 0.195 0.284 0.423 0.002
Current strength 0.164 0.379 0.542 0.002
Latitude –0.110– 0.556 –0.304– 0.096
Lunar phase –0.011– 0.952 0.321 0.074
Table 3. Primary producers and benthic invertebrates en-
countered during underwater visual censuses in the Ria
Formosa lagoon. Relative abundances (%) are a percentage of
the total surveyed area (32 sites pooled) that was covered by
benthic flora and fauna. Only taxonomic groups covering
>1% of the pooled transects are reported
Habitat component Dominant species %
Seagrasses Cymodocea nodosa 54.200
Zostera noltii 4.40
Macroalgae Chaetomorpha sp. 1.43
Chondrus crispus 1.09
Codium spp. 2.24
Dictyopteris sp. 1.29
Dictyota dichotoma 1.19
Ulva lactuca 9.03
Ulva rigida 2.43
Bryozoans Zoobotryon verticillatum 8.40
Bugula neritina 1.61
Tunicates (Ciona intestinalis, Clavelina lepadi- 3.77
formis, Phallusia mammillata, Styela plicata)
Tube-dwelling polychaetes Sabella spp. 2.90
Sea urchins Paracentrotus lividus 1.12
Total 95.100
Fig. 3. (a) Density of Hippocampus guttulatus (
D
) and H. hippocampus
(
s
) plotted against C
t
/A
t
(see text for details). (b) Hippocampus hippo-
campus density plotted against % sand
a
b
0.5
0.4
0.3
0.2
0.1
0.0
0.0 0.2 0.4 0.6 0.8 1.0
C+/A+
0 20 40 60 80 100
% Sand
0.08
0.06
0.04
0.02
0.00
Density (m
–2
) Density (m
–2
)
Curtis & Vincent: Ecology of sympatric seahorses
observed in 2002. The density of H. hippocampus was
significantly explained by (and positively associated
with) % bare sand in 2001 (Fig. 3b, Table 5). Within
years, qualitatively results were robust to the exclusion
of the 2 sites that supported the highest densities of H.
guttulatus and H. hippocampus, respectively, and
robust to the exclusion of sites with densities of zero.
There were no significant correlations or significant
interactions among variables retained in the regres-
sion models (even when no sequential Bonferroni
corrections were used to adjust critical p-values).
Microhabitat and holdfast preferences
Despite differences in habitat use over landscape
scales, both species significantly preferred covered
substrate to bare surfaces at microhabitat scales.
Approximately 89% of all seahorses employed hold-
fasts such as seagrass blades, macroalgae, tunicates,
bryozoans, polychaete tubes, sea urchins, and artificial
structures (e.g. ropes, nets, bricks). The selection index
for covered substrate was >0.5 for Hippocampus
guttulatus at all occupied sites (mean SI = 0.87 ± 0.25
SD, Fig. 4a) and >0.5 for H. hippocampus at 80% of
occupied sites (mean SI = 0.82 ± 0.36 SD, Fig. 4b). H.
guttulatus did not exhibit a preference for any holdfast
type (F
19,18
= 1.16, W = 3.13, K = 100, all pairwise com-
parisons <3.13). Conversely, H. hippocampus signifi-
cantly preferred grasping artificial structures, the small
tuft-forming bryozoan Bugula neritina, sea urchins, and
small or tuft-forming macroalgae (e.g. Colmpomenia
sinuosa, Dictyota dichotoma, Padina pavonia and Ulva
rigida) to the abundant bryozoan Zoobotryon verticil-
latum, which generally formed large colonies (mean ±
SD patch length = 1.25 ± 1.32 m, n = 50). H. hippocam-
pus also significantly preferred using small or tuft-
forming macroalgae to colonial ascidians, and to sea-
grasses (patch length = 5.03 ± 4.09 m, n = 22) and Ulva
lactuca (patch length = 1.12 ± 1.28 m, n = 47), which
generally formed large tracts of densely vegetated
substrates.
Morphology and foraging strategies
The 2 seahorse species differed sig-
nificantly in their appearance and for-
aging strategies. Colouration varied
both within and between species
(Fisher Exact test, p < 0.0001): Hippo-
campus guttulatus primarily occurred
in shades of brown (69.0%) and green
(28.7%) and H. hippocampus occurred
in shades of mottled brown (61.6%),
yellow (26.3%), maroon (7.9%) and rust
(4%). A greater percentage (70.3%) of
H. guttulatus bore skin filaments on
their heads and trunks than H. hippo-
campus (36.0%; χ
2
= 48.28, p < 0.001).
The 2 species also differed in mobility
and predation technique (Fig. 5). H.
87
Fig. 4. Proportion of available habitat
covered by benthic flora and fauna
(C
t
/A
t
, black bars) and proportion of
individuals using covered habitat (U
t
,
grey bars) for (a) Hippocampus gut-
tulatus and (b) H. hippocampus. Only
occupied sites are shown
a
b
Fig. 5. Relative frequency of Hippocampus guttulatus
(black bars, n = 382) and H. hippocampus (white bars,
n = 41) individuals observed stationary, swimming,
using ambush predation techniques or actively foraging
Table 5. Significant explanatory variables retained in generalized
linear models for Hippocampus guttulatus and H. hippocampus
density. Regression coefficients (Coeff), corresponding standard
errors (SE) and p-values (p) are given
Species (year) Variable(s) Coeff SE p
H. guttulatus % covered substrate 3.10 1.00 0.002
(2001) % macroalgae 3.37 1.66 0.042
H. guttulatus % covered substrate 2.17 0.96 0.023
(2002) % Ulva lactuca 7.69 2.94 0.009
H. hippocampus % sand 25.220 0.00 <0.0001
(2001)
Mar Ecol Prog Ser 291: 81–91, 2005
hippocampus swam significantly more frequently than
H. guttulatus, which tended to sway passively with the
currents while grasping holdfasts (χ
2
= 33.24, p < 0.001).
H. hippocampus was also more active when feeding; a
greater proportion was observed actively foraging for
planktonic and epibenthic prey than H. guttulatus
(χ
2
= 3.24, p = 0.072). We found no evidence of sex- or
stage-specific differences in habitat associations, hold-
fast preferences, appearance or foraging technique
within species.
DISCUSSION
Temperate seagrass communities support diverse
and abundant populations of syngnathids (Pollard
1984, Howard & Koehn 1985, Hindell et al. 2000,
Hyndes et al. 2003), which generally prefer complex
vegetated habitats to unvegetated habitats (Howard &
Koehn 1985, Flynn & Ritz 1999, Kendrick & Hyndes
2003). Our study revealed that 2 sympatric seahorse
species, Hippocampus guttulatus and H. hippocampus,
with similar life histories (reviewed in Foster & Vincent
2004) differed markedly in their habitat use over
multiple spatial scales and along a gradient of habitat
complexity: one species was positively associated with
habitat cover at both landscape and microhabitat
scales, whereas the other species used more open and
less speciose habitats at the landscape scale despite
preferring covered microhabitats. As predicted, habitat
associations over both spatial scales were linked to
differences in morphology (size, colouration, presence
of skin filaments), activity and foraging strategy.
Abundance
The density of Hippocampus guttulatus was one
order of magnitude greater than mean densities
observed in similar transect surveys of H. abdominalis
in Australia (K. Martin-Smith pers. comm.), H. capen-
sis in South Africa (Bell et al. 2003) and H. comes and
H. barbouri in the Philippines (A. Maypa pers. comm.).
High seahorse abundance in the Ria Formosa may be
attributable to high local productivity (Sprung 1994a)
and plankton biomass (Sprung 1994b). In some aquatic
ecosystems, fish biomass responds positively to in-
creases in primary production (Garg & Bhatnagar
2000). Seahorses, which prey primarily on small vagile
crustaceans (Howard & Koehn 1985, James & Heck
1994), may be less prone to food limitation in growth
and reproduction in productive habitats such as the Ria
Formosa than in less productive coastal habitats. How-
ever, mechanisms underlying differences in abun-
dance between H. guttulatus and H. hippocampus
within the Ria Formosa require further study. Our sam-
pling constraints may have caused us to miss areas of
high H. hippocampus abundance, which appeared to
be linked with greater depths and stronger currents
(see below).
Landscape level habitat associations
The amount of total habitat covered by vegetation and
sessile invertebrates, C
t
/A
t
, appeared to influence the
distribution and abundance of Hippocampus guttulatus.
This is consistent with the distributions of many marine
organisms, which are positively correlated with the
amount of habitat structure (Heck et al. 1989, Kendrick &
Hyndes 2003), drift algae (Kulczycki et al. 1981) or sea-
grass (Kupschus 2003). Although the predation effi-
ciency of predators typically decreases in more complex
habitats (Choat 1982), increased habitat complexity ei-
ther had no impact on (James & Heck 1994) or increased
(Flynn & Ritz 1999) the predation efficiency of syng-
nathids practicing ambush predation. By contrast, H.
hippocampus exploited sparsely vegetated sand flats,
which were less speciose and subject to stronger oceanic
influences. The 3 sparsely covered sites where H. hip-
pocampus outnumbered its congener (5, 17 and 20) were
located close to tidal inlets where horizontal visibility,
current strength and depth—all positively correlated
with H. hippocampus abundance—were the greatest. H.
hippocampus also outnumbered H. guttulatus in more
open and exposed coastal habitats near Barcelona (sand
flat) and Malaga (seagrass bed), in Spain (J.M.R.C. pers.
obs.), but were not reported from seahorse samples col-
lected from a coastal lagoon in the Mediterranean Sea
(Pérez-Ruzafa et al. 2004).
Fish distribution and abundance are often predicted
by multiple habitat variables including salinity and
temperature (e.g. Brown et al. 2000, Kupschus 2003).
Despite having quantified 60 habitat variables, the
GLMs retained only 1 or 2 explanatory variables that
were related to the extent of local habitat complexity.
This may reflect (1) low variance in other important
ecological variables within the Ria Formosa lagoon (e.g.
salinity; Monteiro 1989), (2) the influence of additional
ecological processes (e.g. predation, hydrodynamic
processes, recruitment dynamics) on local patterns of
abundance (Choat 1982, Orth et al. 1984, Hindell et al.
2000, Hovel et al. 2002, Stoner 2003), (3) low power to
detect weakly correlated variables, or (4) the potential
influence of unmeasured habitat variables (e.g. size of
seagrass beds, distance between suitable habitat
patches). Examining patterns of abundance and habitat
use over regional spatial scales may reveal other vari-
ables for predicting seahorse distribution and abun-
dance, including salinity and water temperature.
88
Curtis & Vincent: Ecology of sympatric seahorses
Microhabitat and holdfast preference
Both European seahorse species preferred using
covered microhabitats as holdfasts to using bare sub-
strate, although Hippocampus hippocampus was ob-
served more frequently on bare surfaces than H. guttu-
latus. Our results were consistent with an experimental
study showing that captive H. abdominalis preferred
using artificial seagrass microhabitats to bare substrate
(Flynn & Ritz 1999). The use of holdfasts with prehen-
sile tails likely helps these relatively poor swimmers to
stabilize themselves in strong currents. Seahorses en-
countered on bare substrate (e.g. H. capensis, H. gut-
tulatus, H. hippocampus, H. kuda and H. abdominalis)
may be temporarily exploiting open habitats, be in
transit between covered microhabitats or have been
displaced from their home ranges by strong wave
action during storms (as inferred for H. breviceps in
Moreau & Vincent 2004).
Although both Hippocampus guttulatus and H. hip-
pocampus in this study used prehensile tails to grasp
structures, they differed significantly in the types of
holdfasts they preferred, as reflected by interspecific
differences in appearance. Syngnathids mimic vegeta-
tion in color, shape and behaviour (Howard & Koehn
1985, Kendrick & Hyndes 2003), which likely reduces
their visibility to both predators and prey. The skin fil-
aments and predominantly brown and green coloura-
tion of H. guttulatus, a holdfast generalist, was consis-
tent with its use of seagrasses, macroalgae and colonial
invertebrates (e.g. Zoobotryon verticillatum) for cam-
ouflage. The colour patterning of H. hippocampus, a
holdfast specialist, was similar in colour to sand,
shell fragments and some sessile invertebrates. This
colouration was consistent with H. hippocampus being
less reliant on vegetation for camouflage than H. gut-
tulatus.
Interspecific differences in habitat preferences be-
tween Hippocampus guttulatus and H. hippocampus
were also linked to differences in foraging strategy
and prey use. H. guttulatus was relatively sedentary
and ambushed planktonic prey more frequently than
its more active congener. H. hippocampus swam more
often and actively foraged for both planktonic and
epibenthic prey. Differences in activity and foraging
strategy between these 2 species were consistent with
a study in which H. erectus adopted different prey cap-
ture techniques in complex and bare habitats: H. erec-
tus tended to ambush prey in artificial seagrass habi-
tats and actively forage for prey while on bare
substrates (James & Heck 1994). These results were
also consistent with observations of 4 sympatric pipe-
fishes within a seagrass bed: less mobile species con-
sumed mainly planktonic prey and were associated
with dense macrophytes, while more mobile species
consumed both planktonic and epibenthic prey and
probably used a variety of habitats, including bare
substrate (Howard & Koehn 1985). Differences in
microhabitat use and behaviour coupled with size-
selective prey use resulted in little diet overlap among
pipefishes during most of the year.
Implications for conservation
While our results suggest that the 2 European sea-
horse species may respond differently to an overall
reduction in seagrass cover, both species would likely
be affected by a loss of holdfasts, and reductions in
exported drift (Melville & Connolly 2003, Vanderklift &
Jacoby 2003). Habitat loss has been linked to the extir-
pation of several marine fish populations (Wolff 2000,
Dulvy et al. 2003) and is assumed to threaten popula-
tions of temperate seahorses including Hippocampus
hippocampus (Santos et al. 1995, Musick et al. 2000,
Pogonoski et al. 2002, IUCN 2003). Moreover, syn-
gnathids are incidentally captured in non-selective,
bottom-dragged fishing gears, especially trawls (Vin-
cent 1996), which may also reduce habitat structure
(Watling & Norse 1998, Turner et al. 1999) and holdfast
availability.
Seagrasses are threatened marine habitats (Sheperd
et al. 1989, Short & Wyllie-Escheverria 1996) and have
declined along some European coasts (Pasqualini et al.
1999, Wolff 2000). Excessive seagrass damage within
the Ria Formosa and in other parts of the North
Atlantic Ocean or Mediterranean Sea could result in a
reduction of Hippocampus guttulatus habitat, and
hence local population sizes. Although H. hippocam-
pus does not rely on seagrasses per se for refuge or
camouflage, its response is difficult to predict. This is
because the implications of seagrass loss reach beyond
the physical provisioning of refuge and camouflage for
both predators and prey: the production of many
marine fishes stems from detritus exported from sea-
grass beds (Melville & Connolly 2003).
Because fisheries-independent data are often scant
and costly to obtain, the development of habitat suit-
ability models for predicting species distributions and
responses to environmental change is appealing
(Brown et al. 2000, Kupschus 2003). Our preliminary
models were strengthened by qualitative concordance
between GLMs for Hippocampus guttulatus from 2
sequential years: C
t
/A
t
was included in GLMs from
both years and densities observed in 2002 were signif-
icantly correlated with predicted values. The index of
habitat complexity, C
t
/A
t
, may thus be useful in a habi-
tat suitability model for predicting the relative abun-
dance of H. guttulatus in similar areas that have not
yet been surveyed. Recently, habitat use models extra-
89
Mar Ecol Prog Ser 291: 81–91, 2005
polated to broader spatial and temporal scales success-
fully predicted the relative abundances of several
marine fishes (e.g. Brown et al. 2000, Kupschus 2003).
However, our results also underscore the importance
of examining habitat associations over multiple spatial
scales as habitat preferences are not necessarily con-
sistent over a range of spatial scales.
Acknowledgements. This is a contribution from Project Sea-
horse. We gratefully acknowledge the following for excellent
field and lab assistance: H. Balasubramanian, K. Bigney, B.
Gunn, S. Lemieux, C.-M. Lesage, E. Murray, J. Nadeau, Nor-
berto, S. Overington, M. Veillette, K. Wieckowski, and espe-
cially S. V. Santos. We thank D. Marsden and D. Réale for sta-
tistical advice. We also thank members of the Universidade do
Algarve and the Parque Natural da Ria Formosa, particularly
K. Erzini, N. Grade, E. Marques, A. L. Quaresma, J. Ribeiro,
Sr. Silverio, and M. Sprung, for useful discussions and kind
logistical support. A. Gonzalez, S. Lourie, J. Marcus, K. Mar-
tin-Smith, I. Morgan, S. Morgan, D. O’Brien and 4 anonymous
reviewers provided constructive comments that greatly
improved the manuscript. Guylian Chocolates, Belgium, gen-
erously funded our study. J.M.R.C. was supported by NSERC
PGS-B (Natural Sciences and Engineering Research Council
of Canada), FCAR (Fonds de Recherche et Technologies
Québec), and a McConnell McGill Major Fellowship. The
John G. Shedd Aquarium in Chicago generously supported
A.C.J.V.’s operating costs, through its partnership with Pro-
ject Seahorse.
LITERATURE CITED
Bartholomew A, Diaz RJ, Cicchetti G (2000) New dimension-
less indices of structural habitat complexity: predicted and
actual effects on a predator's foraging success. Mar Ecol
Prog Ser 206:45–58
Bell EM, Lockyear JF, McPherson JM, Marsden AD, Vincent
ACJ (2003) First field studies of an endangered South
African seahorse, Hippocampus capensis. Environ Biol
Fish 67:35–46
Bell JD, Westoby M (1986a) Importance of local changes in
leaf height and density to fish and decapods associated
with seagrasses. J Exp Mar Biol Ecol 104:249–274
Bell JD, Westoby M (1986b) Abundance of macrofauna in
dense seagrass is due to habitat preference, not predation.
Oecologia 68:205–209
Boisseau J (1967) Régulations hormonales de l'incubation
chez un vertébré mâle: recherches sur la reproduction de
l'Hippocampe. PhD dissertation, Université de Bordeaux
Brown SK, Buja KR, Jury SH, Monaco ME (2000) Habitat suit-
ability index models for eight fish and invertebrate species
in Casco and Sheepscot Bays, Maine. North Am J Fish
Manage 20:408–435
Choat JH (1982) Fish feeding and the structure of benthic
communities in temperate waters. Annu Rev Ecol Syst 13:
423–449
Curtis JMR (2004) Life history, ecology and conservation
of European seahorses. PhD thesis, McGill University,
Montreal
Dean RL, Connell JH (1987) Marine invertebrates in an algal
succession. III. Mechanisms linking habitat complexity
with diversity. J Exp Mar Biol Ecol 109:249–273
Diaz-Ruiz S, Aguirre-Leon A, Perez-Solis O (2000) Distribu-
tion and abundance of Syngnathus louisianae and Syng-
nathus scovelli (Syngnathidae) in Tamiahua Lagoon, Gulf
of Mexico. Cienc Mar 26:125–143
Dulvy NK, Sadovy Y, Reynolds JD (2003) Extinction vulnera-
bility in marine populations. Fish Fish 4:25–64
Edgar GJ (1990) The influence of plant structure on the
species richness, biomass and secondary production of
macrofaunal assemblages associated with western Aus-
tralian seagrass beds. J Exp Mar Biol Ecol 137:215–240
Edgar GJ, Shaw C (1995) The production and trophic ecology
of shallow-water fish assemblages in southern Australia
III. General relationships between sediments, seagrasses,
invertebrates and fishes. J Exp Mar Biol Ecol 194:107–131
Efron B, Tibshirani RJ (1993) An introduction to the bootstrap.
Chapman & Hall, New York
Erzini K, Bentes L, Coelho R, Correia C, Lino PG, Monteiro P,
Ribeiro J, Goncalves JMS (2002) Recruitment of sea
breams (Sparidae) and other commercially important spe-
cies in the Algarve (Southern Portugal). Final Report, DG
XIV/C/1, Ref 99/061, University of the Algave, Faro
Flynn AJ, Ritz, DA (1999) Effect of the habitat complexity and
predatory style on the capture success of fish feeding on
aggregated prey. J Mar Biol Assoc UK 79:487–494
Foster S, Vincent ACJ (2004) The life history and ecology of
seahorses, Hippocampus spp.: implications for conserva-
tion and management. J Fish Biol 65:1–61
Garg SK, Bhatnagar A (2000) Effect of fertilization frequency
on pond productivity and fish biomass in still water ponds
stocked with Cirrhinus mrigala (Ham.). Aquac Res 31:
409–414
Garrick-Maidment N, Jones L (2004) British Seahorse Survey
Report. The Seahorse Trust, Topsham
Heck KL Jr, Able KW, Fahay MP, Roman CT (1989) Fishes and
decapod crustaceans of Cape Cod eelgrass meadows—
species composition, seasonal abundance patterns and
comparison with unvegetated substrates. Estuaries 12:
59–65
Hindell JS, Jenkins GP, Keough MJ (2000) Evaluating the
impact of predation by fish on the assemblage structure of
fishes associated with seagrass (Heterozostera tasmanica)
(Martens ex Ascherson) den Hartog, and unvegetated
sand habitats. J Exp Mar Biol Ecol 255:153–174
Hovel KA, Fonseca MS, Myer DL, Kenworthy WJ, Whitfield
PE (2002) Effects of seagrass landscape structure, struc-
tural complexity and hydrodynamic regime on macrofau-
nal densities in North Carolina seagrass beds. Mar Ecol
Prog Ser 243:11–24
Howard RK, Koehn JD (1985) Population dynamics and feed-
ing ecology of pipefish (Syngnathidae) associated with
eelgrass beds of Western Port, Victoria. Aust J Mar Freshw
Res 36:361–370
Hyndes GA, Kendrick AJ, MacArthur LD, Stewart E (2003)
Differences in the species- and size-composition of fish
assemblages in three distinct seagrass habitats with differ-
ing plant and meadow structure. Mar Biol 142:1195–1206
IUCN (2004) 2004 IUCN Red list of threatened species (avail-
able at: www.redlist.org)
James PL, Heck KL Jr (1994) The effects of habitat complexity
and light intensity on ambush predation within a simu-
lated seagrass habitat. J Exp Mar Biol Ecol 176:187–200
Jenkins GP, Black KP, Wheatley MJ, Hatton DN (1997) Tem-
poral and spatial variability in recruitment of a temperate,
seagrass-associated fish is largely determined by physical
processes in the pre- and post-settlement phases. Mar
Ecol Prog Ser 148:23–35
Johnson DH (1980) The comparison of usage and availability
90
Curtis & Vincent: Ecology of sympatric seahorses
measurements for evaluating resource preference.
Ecology 61:65–71
Kendrick AJ, Hyndes GA (2003) Patterns in the abundance
and size-distribution of syngnathid fishes among habitats
in a seagrass-dominated marine environment. Estuar Coast
Shelf Sci 57:631–640
Krebs CJ (1999) Ecological methodology. Addison-Wesley,
Menlo Park, CA
Kulczycki GR, Virnstein RW, Nelson WG (1981) The relation-
ship between fish abundance and algal biomass in a sea-
grass drift algae community. Estuar Coast Shelf Sci 12:
341–348
Kupschus S (2003) Development and evaluation of statistical
habitat suitability models: an example based on juvenile
spotted seatrout Cynoscion nebulosus. Mar Ecol Prog Ser
265:197–212
Lewis FG (1984) Distribution of macrobenthic crustaceans
associated with Thalassia, Halodule and bare sand sub-
strata. Mar Ecol Prog Ser 19:101–113
Lourie SA, Vincent ACJ, Hall HJ (1999) Seahorses: an identi-
fication guide to the world’s species and their conserva-
tion. Project Seahorse, London
Lythgoe J, Lythgoe G (1971) Fishes of the sea. Blandford
Press, London
Machás R, Santos R (1999) Sources of organic matter in Ria
Formosa revealed by stable isotope analysis. Acta Oecol
20:463–469
Manly BFJ (1997) Randomization and Monte Carlo methods
in biology, 2nd edn. Chapman & Hall, London
McCullagh P, Nelder JA (1989) Generalized linear models.
Chapman & Hall, London
Melville AJ, Connolly RM. (2003) Spatial analysis of stable
isotope data to determine primary sources of nutrition for
fish. Oecologia 136:499–507
Monteiro CC (1989) La faune ichthyologique de la Ria For-
mosa: répartition et organisation spatio-temporelle des
communautés. PhD dissertation, Université des Sciences
et Techniques du Languedoc, Montpellier
Moreau MA, Vincent ACJ (2004) Social structure and space
use in a wild population of the Australian short-headed
seahorse, Hippocampus breviceps Peters, 1869. Mar Freshw
Res 55:231–239
Musick JA, Harbin MM, Berkeley SA, Burgess GH and 6 oth-
ers (2000) Marine, estuarine, and diadromous fish stocks
at risk of extinction in North America (exclusive of Pacific
salmonids). Fisheries 25:6–30
Newton A, Mudge SM (2003) Temperature and salinity
regimes in a shallow, mesotidal lagoon, the Ria Formosa,
Portugal. Estuar Coast Shelf Sci 57:73–85
Orth RJ, Heck KL Jr, Van Montfrans J (1984) Faunal commu-
nities in seagrass beds—a review of the influence of plant
structure and prey characteristics on predator prey rela-
tionships. Estuaries 7:339–350
Pankratz C (1994) PREFER v5.1 Preference assessment.
Northern Prairie Science Center, Jamestown, ND
Pasqualini V, Pergent-Martini C, Pergent G (1999) Environ-
mental impact identification along the Corsican coast
(Mediterranean Sea) using image processing. Aquat Bot
65:311–320
Pérez-Ruzafa A, Quispe-Becerra JI, García-Charton JA,
Marcos C (2004) Composition, structure and distribution
of the ichthyoplankton in a Mediterranean coastal lagoon.
J Fish Biol 64:202–218
Pogonoski J, Pollard DA, Paxton JR (2002) Conservation
overview and action plan for Australian threatened and
potentially threatened marine and estuarine fishes. Envi-
ronment Australia, Canberra
Pollard DA (1984) A review of ecological studies on seagrass
fish communities, with particular reference to recent stud-
ies in Australia. Aquat Bot 18:3–42
Reina-Hervás JA (1989) Contribucion al estudio de la F. Syn-
gnathidae (Pisces) en las costas del sureste de España.
Arquiv Museu Bocage 1:325–334
Samoilys MA (1997) Underwater visual census surveys. In:
Samoilys MA (ed) Manual for assessing fish stocks on
Pacific coral reefs. Department of Primary Industries,
Townsville, p 16–29
Santos RS, Hawkins S, Monteiro LR, Alves M, Isidro EJ (1995)
Marine research, resources and conservation in the
Azores. Aquat Conserv 5:311–354
Sheperd SA, McComb AJ, Bulthuis DA, Neverauskus V,
Steffensen DA, West R (1989) Decline of seagrasses. In:
Larkum AWD, McComb AJ, Sheperd SA (eds) Biology of
seagrasses. Elsevier, Amsterdam, p 346–393
Short FT, Wyllie-Escheverria S (1996) Natural and human-
induced disturbances of seagrasses. Environ Conserv 23:
17–27
Sprung M (1994a) Macrobenthic secondary production in the
intertidal zone of the Ria Formosa—a lagoon in Southern
Portugal. Estuar Coast Shelf Sci 38:539–558
Sprung M (1994b) High larval abundances in the Ria Formosa
(Southern Portugal)—methodological or local effect?
J Plankton Res 16:151–160
Stoner AW (2003) What constitutes essential nursery habitat
for a marine species? A case study of habitat form and
function for queen conch. Mar Ecol Prog Ser 257:275–289
Teixeira RL, Musick JA (1995) Trophic ecology of two con-
generic pipefishes (Syngnathidae) of the lower York River,
Virginia. Environ Biol Fish 43:295–309
Tipton K, Bell S (1988) Foraging patterns of two syngnathid
fishes: importance of harpacticoid copepods. Mar Ecol
Prog Ser 47:31–43
Turner S, Thrush S, Hewitt J, Cummings V, Funnell G (1999)
Fishing impacts and the degradation or loss of habitat
structure. Fish Manage Ecol 6:401–420
Vanderklift MA, Jacoby CA (2003) Patterns in fish assem-
blages 25 years after major seagrass loss. Mar Ecol Prog
Ser 247:225–235
Vincent ACJ (1996) The international trade in seahorses.
TRAFFIC International, Cambridge
Watling L, Norse EA (1998) Disturbance of the seabed by
mobile fishing gear: a comparison to forest clearcutting.
Conserv Biol 12:1180–1997
Whitehead PJP, Bauchot ML, Hureau JC, Nielsen J, Tor-
tonese E (1986) Fishes of the north-eastern Atlantic and
the Mediterranean. Vol 2, UNESCO, Paris
Wolff WJ (2000) Causes of extirpations in the Wadden Sea, an
estuarine area in The Netherlands. Conserv Biol 14:876–885
91
Editorial responsibility: Otto Kinne (Editor-in-Chief),
Oldendorf/Luhe, Germany
Submitted: January 30, 2004; Accepted: November 9, 2004
Proofs received from author(s): April 11, 2005