Content uploaded by Simon James Pittman
Author content
All content in this area was uploaded by Simon James Pittman
Content may be subject to copyright.
MARINE ECOLOGY PROGRESS SERIES
Mar Ecol Prog Ser
Vol. 348: 273–284, 2007
doi: 10.3354/meps07052
Published October 25
INTRODUCTION
Mangroves are generally considered to offer an
important nursery function for many fish species and
have been linked to highly productive coastal fisheries
worldwide (Blaber 2000, Manson et al. 2005, Faunce &
Serafy 2006). It is now increasingly acknowledged,
however, that mangroves may not always function in
isolation to support fish populations (Sheaves 2005,
Adams et al. 2006). In many regions, mangroves are
connected to a mosaic of habitat types (e.g. seagrasses,
coral reefs) through the movements of fish at a range of
scales in time and space (Parrish 1989, Mumby et al.
2004, Pittman et al. 2004). In Caribbean mangroves,
many of the most frequently observed fish species use
multiple habitat types through daily home range
movements (Rooker & Dennis 1991, Nagelkerken et
al. 2000, Verweij et al. 2006) and broader scale
onshore–offshore ontogenetic habitat shifts (Cocheret
de la Morinière 2002, Christensen et al. 2003, Mumby
et al. 2004). As such, the nursery function of Caribbean
mangroves may depend on spatial connectivity to
other habitat types (Sheaves 2005); an attribute of the
species–environment relationship, which is deter-
mined primarily by the interaction between animal
mobility and the composition and spatial arrangement
of resources across the seascape.
Nevertheless, a spatially explicit and quantitative
consideration of the surrounding seascape structure is
still a relatively novel approach in marine ecology.
Concepts and analytical techniques developed in
landscape ecology, however, provide an existing
© Inter-Research 2007 · www.int-res.com*Email: simon.pittman@noaa.gov
Using seascape types to explain the spatial patterns
of fish in the mangroves of SW Puerto Rico
Simon J. Pittman*, Chris Caldow, Sarah D. Hile, Mark E. Monaco
NOAA/NOS/CCMA Biogeography Team, 1305 East-West Highway, Silver Spring, Maryland 20910, USA
ABSTRACT: Many of the most abundant fish species using mangroves in the Caribbean also use
other habitat types through daily home range movements and ontogenetic habitat shifts. Few stud-
ies, however, have considered the structure of the surrounding seascape when explaining the spatial
distribution of fish within mangroves. This study develops an exploratory seascape approach using
the geographical location of mangroves and the structure of the surrounding seascape at multiple
spatial scales to explain the spatial patterns in fish density and number of species observed within
mangroves of SW Puerto Rico. Seascape structure immediately surrounding mangroves was most
influential in determining assemblage attributes and the density of juvenile Haemulon flavolineatum,
which were significantly higher in mangroves with high seagrass cover (>40%) in close proximity
(<100 m) than mangroves with low (<40%) or no adjacent seagrasses. Highest mean density of juve-
nile Ocyurus chrysurus was found in offshore mangroves, with high seagrass and coral reef cover
(>40 and >15%, respectively) in close proximity (<100 m). In contrast, juvenile Lutjanus griseus
responded at much broader spatial scales, and with highest density found in extensive onshore man-
groves with a large proportion (>40%) of seagrass within 600 m of the mangrove edge. We argue that
there is an urgent need to incorporate information on the influence of seascape structure into a wide
range of marine resource management activities, such as the identification and evaluation of critical
or essential fish habitat, the placement of marine protected areas and the design of habitat restora-
tion projects.
KEY WORDS: Seascape structure · Mangroves · Seagrasses · Coral reef ecosystems · Spatial scale ·
Fish · Puerto Rico · Caribbean
Resale or republication not permitted without written consent of the publisher
Mar Ecol Prog Ser 348: 273–284, 2007
framework that can be readily applied to shallow-
water marine systems (Robbins & Bell 1994). Land-
scape ecology has been defined as the study of pro-
cesses occurring across spatially heterogeneous
mosaics and the biotic responses to the resultant pat-
terns (Turner 1989). The underlying premise is that the
composition and spatial arrangement of mosaic ele-
ments affect ecological processes in ways that would
be different if the composition and arrangement were
different (Dunning et al. 1992, Wiens et al. 1993). From
an operational perspective, an exploratory seascape
approach must quantify environmental variability at
multiple spatial scales (Pittman & McAlpine 2003,
Pittman et al. 2004). A single-scale approach cannot
incorporate important interactions occurring at other
scales and is particularly inappropriate for studies of
multi-species assemblages. Species, and even differ-
ent life-stages of the same species, vary in their
response to the environment due to functional differ-
ences in dietary requirements, habitat specialisations,
body size and movements (Addicott et al. 1987, Wiens
1989, Wiens et al. 1993). Furthermore, comparing the
strength of the species–environment association
across scales allows determination of the scale exhibit-
ing the strongest influence, typically referred to as the
‘characteristic scale’ (Holland et al. 2004).
The present study links daytime underwater fish
census data collected within mangroves of the La
Parguera Natural Wildlife Reserve (SW Puerto Rico)
with geographical and seascape variables quantified
within circles of different radii surrounding fish sur-
vey sites. We focus on the spatial patterns of density
for 2 species of Haemulidae (Haemulon flavolineatum
and H. sciurus) and 3 species of Lutjanidae (Lutjanus
apodus, L. griseus and Ocyurus chrysurus), as well as
the density and number of species of the fish assem-
blage observed within mangroves. We develop an
exploratory approach (1) to identify the characteristic
scale(s) of response to habitat by examining the
strength of the correlation between the fish response
variable and individual seascape variables across the
scale range and (2) at these characteristic scales, we
then examine the influence of multivariate seascape
structure to identify the mangroves and associated
seascape types that support the highest and lowest
density of fish and number of fish species across the
region.
MATERIALS AND METHODS
Study area. Fish censuses were conducted between
2001 and 2006 within mangroves of the La Parguera
region of SW Puerto Rico (Fig. 1). The data was col-
lected as part of a larger project designed to monitor
and characterise fish–habitat relationships across
Caribbean coral reef ecosystems (National Oceanic
and Atmospheric Administration’s Caribbean Coral
Reef Ecosystem Monitoring Program).
The shoreline and islands of the area are lined with
mangrove communities dominated by the red man-
grove Rhizophora mangle. To the west, the onshore
mangroves are extensive and extend to approximately
800 m from the land. To the east, there are less exten-
sive fringing mangroves and many small mangrove
islands. Adjacent sediments support seagrasses (domi-
nated by Thalassia testudinum), macroalgae and
unvegetated sand and sandy mud interspersed with
coral reefs and patch reefs, which vary in patch size
and benthic community composition. The tidal range is
relatively small (<0.5 m), and mangrove prop roots at
the seaward edge of the mangrove stands are continu-
ally immersed throughout the tidal cycle. For all man-
groves sampled, mean water depth was 66 cm (range
30 to 152 cm).
Sampling fish. A total of 145 spatially random man-
grove survey sites were selected using a geographical
information system (GIS) and a nearshore benthic
habitat map for Puerto Rico (Kendall et al. 2002).
Sampling was conducted in winter (January/February/
March) and fall (August/September/October) seasons
each year, with approximately 10 to 12 spatially ran-
dom sites selected across the study area for each field
season. To quantify fish abundance, observers identi-
fied, counted and estimated the body lengths (fork
length, FL) of all fish sighted within a 25 m long by 4 m
wide belt transect (100 m
2
) using a standard daytime
underwater visual census technique (adapted from
Brock 1954). Transects were placed to directly follow
the edge of the mangrove tree-line. Lengths were esti-
mated in size-class ranges with 5 cm increments (i.e.
<5, 5 to 10, 10 to 15 cm). For individual species, only
juveniles and sub-adults were included in the analysis
and were distinguished based on lengths less than the
mean length at maturity reported by García-Cagide
et al. (1994).
Environmental data. Two types of spatial data were
acquired: (1) geographical variables and (2) spatially
explicit seascape variables. Geographical variables
included distance to shore and the geographical loca-
tion (latitude and longitude) of each fish transect
extracted from the spatial data using a GIS. Seascape
variables included the relative proportion of major
habitat types (percent cover of mangroves, seagrasses,
coral reefs and unvegetated sediments), the number of
habitat types (habitat richness) and mean water depth
quantified within circles of different radii (50, 100, 200,
300, 400, 500 and 600 m) surrounding each fish tran-
sect. Each circle surrounding a fish transect is referred
to here as a seascape sample unit. The proportion and
274
Pittman et al.: Seascape types and spatial patterns of fish in mangroves
richness of habitat types were extracted from the
nearshore benthic habitat map for Puerto Rico. To cap-
ture the full range of heterogeneity represented in the
map, habitat richness was calculated as the total num-
ber of habitat classes within a seascape sample unit
from all available classes at the level of ‘habitat type’
(maximum 13 habitat classes). Mean water depth in
the surrounding seascapes was quantified from a
bathymetric model developed using a triangulated
interpolated network (TIN) on depth soundings data
(National Geophysical Data Center) and supple-
mented by data from SCUBA divers. Vertical accuracy
of the TIN model was validated (mean vertical error
<0.3 m) using an independent set of bathymetric data.
The TIN model was then converted into a raster map
(5 × 5 m spatial resolution).
Scale selection. The range of radii for seascape sam-
ple units selected includes both sub-home range and
supra-home range spatial scales for juvenile Haemuli-
dae and Lutjanidae. Information on the scale of move-
ment was based on the average maximum distance
travelled for multiple individuals of each species using
existing data from acoustic tracking and mark/recap-
ture studies for durations ranging from hours to months
(Table 1). The maximum distance travelled during rou-
tine daily movements is considered an appropriate
measure for scaling ecologically relevant environmen-
tal patterns (Addicott et al. 1987). In the present study,
a maximum radius of 600 m was selected to minimise
overlap among sample units.
Statistical analyses. To examine the strength and
directionality of the fish response to the individual
seascape variables, and to determine the characteristic
scale, we calculated the Spearman rank correlation
coefficient (rho = r) between all response variables and
all seascape variables at each spatial scale. We tested
for significance of the observed r by comparing it with
a simulated distribution of r from re-sampled data
275
Fig. 1. Location of the study area in SW Puerto Rico. Map showing the 4 major habitat types included in this study as delineated
in the NOAA’s benthic habitat map
Mar Ecol Prog Ser 348: 273–284, 2007
using 1000 random permutations. The spatial scale
with the greatest proportion of maximum r across all
predictor variables was selected as the characteristic
scale for each species/assemblage attribute. Following
Holland et al. (2004), we took the scale of maximum
correlation to be the scale of maximum response. In
addition, the presence of spatial dependence in the
response variables was tested using the global Moran’s
I statistic (Moran 1950).
To examine the association between multivariate
seascape structure and the spatial patterns of fish den-
sity and number of species, we first defined seascape
structural types at characteristic scales using hierarchi-
cal cluster analysis with complete linkage applied to a
Bray-Curtis sample dissimilarity matrix (Primer v5 soft-
ware, Plymouth Marine Laboratory). Bray-Curtis dis-
similarities were calculated on the untransformed pro-
portion of each habitat type (mangroves, seagrasses,
coral reefs and unvegetated sediments) in the sea-
scapes surrounding all mangrove sample sites. This
technique partitioned the multiple gradients of habitat
abundance across the study area into distinct struc-
tural types that were then represented on a dendro-
gram. Similarity percentages (SIMPER) analysis was
used to identify the seascape variables that contributed
most to the dissimilarity between groups (Clarke &
Warwick 1994).
Non-metric multidimensional scaling (nMDS) plots
were used to (1) delineate the seascape structural
types in ordination space and (2) display overlays of
fish density and number of species by sample site. Sites
in close proximity to one another in the ordination
space are the most similar in structure, whereas sites
that plot farther away from each other are less similar
(Clarke & Warwick 1994). The goodness-of-fit of the
ordination is ranked by the stress index, which mea-
sures how well a particular plot of similarities corre-
sponds to the observed distance matrix. Stress values
range from 0 to 1, where low values (<0.05) indicate
a good fit and high values (>0.2) indicate a poor
fit (Kruskal 1964). The nonparametric Kruskal-Wallis
1-way analysis of variance test, followed by Dunn’s
multiple pairwise comparison test, was used to exam-
ine differences in the fish response among seascape
structural types.
RESULTS
Within the mangroves of SW Puerto Rico 176 188
individual fish of 96 species from 32 families were
observed. The most frequently occurring fish species
were Lutjanus apodus (100% of samples), Stegastes
leucostictus (76%), Haemulon sciurus (77%), H. flavo-
lineatum (77%), Sphyraena barracuda (65%), Abu-
defduf saxatilis (59%) and Chaetodon capistratus
(58%). Lutjanus griseus and Ocyurus chrysurus were
observed in 46 and 12% of samples, respectively.
Spatial autocorrelation
The response data were not significantly spatially
autocorrelated (Moran’s I = –0.49 to 0.14, p > 0.1) and
therefore the geographical distribution of samples was
not considered to have biased correlation coefficients
calculated to determine the characteristic scales.
Determining a characteristic scale of response
For density and number of species of assemblages,
and for density of Haemulon flavolineatum and Ocyu-
rus chrysurus, correlation coefficients were highest
at the 100 m radius scale for 3 or more of the 6 seascape
variables, and thus the 100 m radius scale was con-
sidered the characteristic scale for subsequent analysis
(Table 2). In contrast, Lutjanus griseus density was
most strongly correlated at the 600 m scale for 4 out of
6 seascape variables, and thus the 600 m radius scale
was considered the characteristic scale for subsequent
analysis (Table 2). More ambiguous was Haemulon
sciurus, which showed significant correlations at
the 600 m (seagrass cover) and 200 m scales (unvege-
276
Species Life stage No. Mean (SD) Mean (SD) max. Location Source
of fish duration (d) distance (m)
Haemulidae
Haemulon flavolineatum Juvenile 25 90.7 (30.9) 105.3 (66.8) Puerto Rico Bouwmeester (2005)
H. flavolineatum Sub-adult 28 198.6 (129.6) 93.6 (68.1) Puerto Rico Bouwmeester (2005)
H. sciurus Sub-adult/adult 10 11.3 (9.3) 279.2 (227.9) St. John, USVI Beets et al. (2003)
Lutjanidae
Lutjanus griseus Juvenile/sub-adult 10 64.3 (0.48) 648 (1216) Florida www.adoptafish.net
Table 1. Approximate scale of movements for haemulids and lutjanids from a range of studies using mark/recapture and
acoustic tracking
Pittman et al.: Seascape types and spatial patterns of fish in mangroves
tated cover), but with strongest correlation at the
600 m scale (Table 2).
Defining seascape structural types
At both the 100 and 600 m spatial scales, cluster
analysis identified 7 groupings of samples that were
>35% dissimilar to each other, of which the largest 6
groups were labelled Seascape Types A, B, C, D, E & F.
The groups were easily delineated on nMDS ordination
plots, with low stress values of 0.05 (100 m seascapes)
and 0.09 (600 m seascapes), indicating ‘excellent’ and
‘good’ 2-dimensional representation of the data struc-
ture, respectively (Fig. 2). Mean (±SD) dissimilarity was
54 ± 10.2% for all pairs of 100 m seascape types and
50 ± 12.1% for 600 m seascape types.
100 m radius seascape types
To improve interpretation, the six 100 m seascape
types were further grouped into 2 highly dissimilar
groups based on seascape characteristics. This division
was delineated on the ordination with a broken line
(Fig. 2). Highest pairwise percentage dissimilarity for
100 m seascapes was between 100A & F (76%), 100B &
F (67%), 100C & F (63%) and 100A & E (65%).
Seascape Structural Types 100E & F were the least
structurally heterogeneous 100 m radius seascapes,
composed of onshore mangroves (<800 m from shore),
surrounded by relatively shallow water (<2 m), with
low seagrass cover (<40%) and no coral reefs in close
proximity. The most structurally heterogeneous 100 m
radius seascapes were 100A & B, composed of offshore
(>800 m from shore) mangrove islands, with high sea-
grass cover (>40%) in close proximity. In addition,
Seascape Type 100A had the largest proportion of coral
reefs (>15%), and Seascape Type 100B had the deepest
adjacent waters and the largest proportion of seagrass
cover (>70%). Seascape Types 100C & D were mostly
onshore mangroves, with very similar structural char-
acteristics, but 100C mangroves had higher adjacent
seagrass cover (mean of 64 versus 50.6%) (Table 3).
Mean pairwise similarity percentages determined that
distance to shore (39 ± 5.6%) and the proportion of
mangroves (25 ± 5.8%) and unvegetated cover (22 ±
8.4%) in the seascape contributed most to the differ-
ences between Seascape Types 100A, B & C and 100E
& F (Table 3).
600 m radius seascape types
The six 600 m radius seascape types were divided
into 3 types with maximum dissimilarity and delineated
with a broken line on the ordination (Fig. 2). Highest
pairwise dissimilarity was calculated for 600A & E
(62%), 600A & C (57%), 600A & B (53%), 600B & D
(54%) and 600B & C (47%). Seascape Structural Types
600A & F were the least heterogeneous 600 m radius
seascapes, composed of onshore mangroves (<800 m
from shore), with low to medium seagrass and man-
grove cover (<40%) and very low (<5%) or no coral
reefs (Table 3). Types 600D & C were considered the
most highly heterogeneous of the onshore seascapes,
with higher seagrass cover (>40%), low unvegetated
cover (<25%), low coral reef cover (<8%) and medium
habitat richness (4 to 8 classes). Types 600B & E were
the most highly heterogeneous seascapes, composed of
relatively small offshore mangrove islands, surrounded
by deeper water (>15 m mean water depth), with the
largest proportion of coral reefs (>10%), low-medium
seagrass cover (>30%), relatively high unvegetated
cover (>20%) and habitat richness (6 to 13). Mean pair-
wise similarity percentages determined that distance to
277
Response Seascape variable
% Habitat Mean water
Mangrove Seagrass Coral reef Unvegetated richness depth
Assemblage
Density 100 (–0.31) 100 (0.33) 100 (0.16) 100 (–0.31) 600 (0.31) 300 (0.31)
Number of species 100 (–0.23) 500 (0.33) 50 (–0.14) 100 (–0.22) 100 (–0.17) 200 (–0.16)
Haemulidae
Haemulon flavolineatum 100 (–0.27) 100 (0.33) n.s. 100 (–0.21) 500 (0.16) 300 (0.25)
H. sciurus n.s. 600 (0.25) n.s. 200 (–0.26) 50 (–0.16) n.s.
Lutjanidae
Ocyurus chrysurus 100 (–0.22) 100 (0.18) 100 (0.12) 500 (0.14) 50 (–0.11) 300 (0.19)
Lutjanus griseus 600 (0.43) 100 (–0.34) 500 (–0.45) 600 (–0.21) 600 (–0.38) 600 (–0.44)
L. apodus n.s. n.s. n.s. n.s. n.s. n.s.
Table 2. Characteristic scales (bold) for the fish response to seascape variables and the associated maximum correlation
coefficient (in parentheses). All correlation coefficients are statistically significant (p < 0.05). n.s.: not significant
Mar Ecol Prog Ser 348: 273–284, 2007
shore (42 ± 4.1%) and the proportion of seagrasses in
the seascape (27 ± 8.2%) contributed most to the differ-
ences between all 3 groups of seascape types.
Linking fish density and number of species to
seascape types
Assemblages
Fish assemblage density within mangroves was
significantly higher (Kruskal-Wallis, K = 22.6, p <
0.001) in seascapes with high seagrass cover (>40%) in
close proximity (<100 m). This was demonstrated by
significantly higher fish densities in mangroves of
Seascape Types 100A, B, C & D when compared to
Seascape Types 100E & F (Fig. 3). Overall, highest
mean assemblage density was recorded for the most
structurally heterogeneous seascape types (100B & A),
composed of offshore mangroves, with high seagrass
cover and with close proximity to coral reefs and
deeper water areas. Lowest mean density was
recorded for the least heterogeneous seascapes (100E
& F), composed of onshore mangroves, with little or no
seagrass cover or coral reefs in close proximity. Mean
number of species was also higher in the more struc-
turally heterogeneous seascape types, but
significant differences were only detected
between Seascape Types 100B and 100E & F
(Kruskal-Wallis, K = 17.7, p < 0.001). Highest
mean number of species was reported for the
100B seascapes, characterised as offshore
mangroves with deeper adjacent water and
the highest seagrass cover (Fig. 3).
Haemulidae
Haemulon flavolineatum were observed
using mangroves in all seascape structural
types, but with markedly higher mean densi-
ties found within mangroves of seascape
types with high adjacent seagrass cover
(Fig. 4). Significantly higher mean density
was recorded for the 100B seascape type
when compared with the lowest mean den-
sity recorded for 100E & F seascapes
(Kruskal-Wallis, K = 13.5 p < 0.05). In fact, H.
flavolineatum was only recorded in 1 sample
from Seascape Type 100F. Haemulon sciurus
were widely distributed amongst mangroves
across the region, with a relatively high
mean density recorded in both offshore
(600B & E) and onshore mangroves (600C &
D) with adjacent seagrasses (Fig. 5). Lowest
mean density was found in mangroves (600A
& F) that were closest to shore, with no coral
reefs within 600 m (Kruskal-Wallis, K = 11.2,
p < 0.05).
Lutjanidae
Ocyurus chrysurus were absent from
onshore mangroves in Seascape Type 100F,
and only 1 individual was sighted from the
17 samples in 100E seascapes (Fig. 4). High-
est mean density was reported from the
278
Stress: 0.05
B
A
C
D
E
F
Shallow water onshore
High mangrove
Low seagrass in group E
No seagrass in group F
No coral reef
Low habitat richness
Offshore mangroves in groups A & B
Low-medium mangrove
High seagrass
High coral reef in group A
Medium-high habitat richness
Shallow water onshore
Medium-high mangrove & seagrass
Low unvegetated
Low coral reef
Medium habitat richness
Stress: 0.09
A
E
B
F
D
C
Deeper water offshore
Low mangrove
Low-med seagrass
High unvegetated
Low-medium coral reef
High habitat richness
Shallow water onshore
Low-medium mangrove & seagrass
No coral reef & low habitat richness
a
b
Fig. 2. Structural dissimilarity amongst (a) 100 m seascape structural types
and (b) 600 m seascape structural types using non-metric multidimen-
sional scaling plots (nMDS) showing the relative dissimilarity in seascape
composition (percent cover of seagrasses, mangroves, coral reefs and
unvegetated sediments) for all samples. Seascape structural types are
encircled and labelled (A, B, C, D, E & F) according to cluster groups iden-
tified using a hierarchical cluster analysis. The groups of seascapes with
highest structural dissimilarity are separated with broken lines
Pittman et al.: Seascape types and spatial patterns of fish in mangroves
279
B
A
C
D
E
F
B
C
D
A
E
F
a
b
0
2
4
6
8
10
12
14
16
ACE
Seascape types (cluster groups)
0
20
40
60
80
100
120
ACE
BDF
BDF
Seascape types (cluster groups)
Mean fish density (per 100 m
2
)Mean no. species (per 100 m
2
)
Fig. 3. Distribution of (a) fish assemblage density and (b) number of fish species within mangroves overlaid on an nMDS ordina-
tion of seascape structural dissimilarity at the 100 m radius scale. Grey circles are proportional in size to the value of density or
number of fish species observed at each survey site. Seascape structural types are encircled and labelled (A, B, C, D, E & F)
according to cluster groups identified using a hierarchical cluster analysis. The accompanying charts show mean (±SE) of (a) fish
assemblage density and (b) number of species for each seascape type
Habitat variable Seascape type
ABCDEF
100 m radius spatial extent
Mangrove (%) 22.6 (1.3) 14.2 (0.7) 24.7 (1.1) 42.3 (0.9) 57.2 (2.1) 56 (10.6)
Seagrass (%) 52.3 (2.8) 85.5 (1.5) 64 (1.6) 50.6 (1.2) 29.8 (2.5) <1 (<0.1)
Coral reef (%) 27.4 (1.6) 2.1 (0.6) 1.4 (0.5) 0.5 (0.3) 0 0
Unvegetated (%) 0.3 (0.3) 0.1 (<0.1) 1.7 (1.1) 6 (1.7) 6.9 (2.7) 39.8 (9.4)
Number of habitat types 4.2 (0.1) 2.9 (0.1) 3.1 (0.1) 2.7 (0.1) 2.9 (0.2) 2.4 (0.2)
Mean depth (m) 1.6 (0.1) 4.5 (0.5) 4.1 (0.5) 2.4 (0.2) 1.4 (0.1) 1.4 (0.2)
Distance to shore (km) 1.2 (0.1) 0.9 (0.1) 0.4 (0.1) 0.3 (<0.1) 0.3 (<0.1) 0.2 (0.01)
600 m radius spatial extent
Mangrove (%) 21.5 (1.4) 3.7 (0.4) 8.3 (0.6) 32.7 (1.9) 2.9 (0.4) 8.3 (1.5)
Seagrass (%) 15.9 (1.5) 26.2 (0.6) 66.8 (2.2) 53.4 (1.2) 42.8 (1.2) 32.5 (0.8)
Coral reef (%) 0 14.1 (0.8) 2.1 (0.5) 0.26 (0.1) 15.2 (1.2) 1.2 (0.7)
Unvegetated (%) 24.9 (3.4) 49.1 (2.7) 6.6 (0.6) 5.91 (0.8) 32.5 (2.2) 25.9 (5.2)
Number of habitat types 4.10 (0.2) 8.1 (0.3) 5.9 (0.2) 5.8 (0.2) 8.3 (0.4) 5.4 (0.6)
Mean depth (m) 4.7 (0.6) 27.2 (1.7) 13.1 (0.6) 5.0 (0.3) 21.6 (1.0) 14.5 (2.1)
Distance to shore (km) 0.15 (<0.1) 0.9 (0.03) 0.3 (<0.1) 0.47 (<0.1) 0.7 (0.1) 0.06 (<0.1)
Table 3. Characterisation of 100 and 600 m seascape structural types using mean (±SE) value for environmental variables
from samples grouped by cluster grouping
Mar Ecol Prog Ser 348: 273–284, 2007
mangrove islands (100A & B) farthest offshore, with
high seagrass cover and close proximity to high coral
reef cover. In fact, Seascape Types 100A & B accounted
for 71% of the O. chrysurus sighted within mangroves
of the study area. Mean density followed a similar
pattern to Haemulon flavolineatum, but no significant
differences (Kruskal-Wallis, K = 6.9, p > 0.05) were
detected for density of O. chrysurus between seascape
structural types. Lutjanus griseus were observed using
mangroves in all seascape types, with highest mean
density in the most heterogeneous of the shallow
onshore mangroves of 600C (Fig. 5). In contrast to the
other species studied here, significantly lower mean
density of L. griseus was recorded for offshore man-
groves 600B & E (Kruskal-Wallis, K = 29.2, p < 0.001).
L. apodus was not significantly correlated with any
seascape variables.
DISCUSSION
In the present study, we developed a multi-scale
seascape approach to explain spatial patterns in fish
density and number of species. Our study revealed
considerable spatial complexity in the densities of fish
throughout the mangroves of SW Puerto Rico, which
could not be explained adequately by any single envi-
ronmental variable alone. We found that an under-
standing of the distribution patterns of fish with spa-
tially complex life cycles and multi-habitat home-
range movements necessitates consideration of both
the geographical location of the mangrove and the
structure of the surrounding seascape, particularly the
amount of seagrasses. Furthermore, it appears that
quality and suitability of a particular seascape type is
species specific and likely determined by the biologi-
280
a
B
A
C
F
D
E
Haemulon flavolineatum
0
10
20
30
40
50
60
70
ACE
ACE
Seascape types (cluster groups)
Seascape types (cluster groups)
Mean fish density (per 100 m
2
)Mean fish density (per 100 m
2
)
A
B
C
F
E
D
Ocyurus chrysurus
b
0
1
2
3
4
BDF
BDF
Fig. 4. Distribution of (a) Haemulon flavolineatum density and (b) Ocyurus chrysurus density within mangroves overlaid on an
nMDS ordination of seascape structural dissimilarity at the 100 m radius scale. Grey circles are proportional in size to the value
of density observed at each survey site. Seascape structural types are encircled and labelled (A, B, C, D, E & F) according to
cluster groups identified using a hierarchical cluster analysis. The accompanying charts show mean (±SE) of (a) H. flavolineatum
density and (b) O. chrysurus density for each seascape type. Fish illustrations provided courtesy of Bohlke & Chaplin (1993)
Pittman et al.: Seascape types and spatial patterns of fish in mangroves
cal characteristics of a species (e.g. mobility, life-
history strategy and resource requirements).
Determining a characteristic scale of response
Our multi-scale approach allowed us to identify the
scale(s) at which seascape structure had the most influ-
ence on fish. Fish that were likely to have exhibited rel-
atively fine-scale movements responded more strongly
to seascape structure immediately surrounding man-
groves. In contrast, juveniles of fish capable of broader
scale movements were found to respond more strongly
to seascape structure at broader spatial scales. Overall,
we found that the strongest bivariate correlations more
frequently occurred at the 100 m radius spatial scale for
fish assemblage density and number of species, as well
as for density of Haemulon flavolineatum and Ocyurus
chrysurus. In contrast, Lutjanus griseus and Haemulon
sciurus appeared to respond to seascape structure at
markedly broader spatial scales, suggesting that selec-
tion of a single scale for multi-species studies may be
inappropriate and that characteristic scales may vary
markedly between species even within a family.
Our results show good agreement between the char-
acteristic scale selected for each species and the
281
A
E
B
F
D
C
Haemulon sciurus
a
0
5
10
15
20
25
ACE
ACE
Mean fish density (per 100 m
2
)Mean fish density (per 100 m
2
)
Lutjanus griseus
A
E
B
F
D
C
b
0
2
4
6
8
10
12
14
BDF
BDF
Seascape types (cluster groups)
Seascape types (cluster groups)
Fig. 5. Distribution of (a) Haemulon sciurus density and (b) Lutjanus griseus density within mangroves overlaid on an nMDS ordi-
nation of seascape structural dissimilarity at the 600 m radius scale. Grey circles are proportional in size to the value of density
observed at each survey site. Seascape structural types are encircled and labelled (A, B, C, D, E & F) according to cluster groups
identified using a hierarchical cluster analysis. The accompanying charts show mean and standard error of (a) H. sciurus density
and (b) L. griseus density for each seascape type. Fish illustrations provided courtesy of Bohlke & Chaplin (1993)
Mar Ecol Prog Ser 348: 273–284, 2007
known spatial extent of fish movements estimated by
fish tracking and tagging studies. For example, mark/
recapture experiments on juvenile and sub-adult
Haemulon flavolineatum in SW Puerto Rico revealed
a mean maximum distance moved of 105.3 m for 25
small juveniles (7.6 to 12.1 cm FL) and 93.6 m for 28
larger juveniles (10.9 to 14.9 cm FL) (Bouwmeester
2005). In addition, underwater observations of H.
flavolineatum at St. Croix have shown that schools of
juvenile grunts will move distances of 100 to 300 m to
feed at night in seagrass beds (Ogden & Ehrlich 1977).
Juvenile Ocyurus chrysurus is also thought to exhibit
relatively high site fidelity. Watson et al. (2002) carried
out daytime observations of newly settled O. chrysurus
(<7 cm FL) in seagrass beds of the British Virgin
Islands, with mean home ranges of <35 m
2
. Further-
more, Lindholm et al. (2005) reported high site fidelity
for 9 sub-adult O. chrysurus (21.5 to 26.5 cm FL) in the
Florida Keys, with the majority of acoustic detections
occurring at a single reef.
For Haemulon sciurus juveniles, our analysis found a
stronger response at broader spatial scales than for
H. flavolineatum, indicating that the characteristic scale
may vary between species within a family. Interestingly,
acoustic tracking of H. sciurus in St. John, US Virgin
Islands, showed that fish were moving greater distances
(mean 279 m, max. 767 m) than has been shown for sim-
ilar-sized H. flavolineatum (Beets et al. 2003). Support for
the broader scale response of Lutjanus griseus also
comes from acoustic tracking in Florida, with a mean
maximum distance travelled of 648 m, with the smallest
juveniles showing relatively high site fidelity and larger
juveniles making extensive movements (S. Whitcraft un-
publ. data). Unfortunately, insufficient data is available
on juvenile Lutjanus apodus to determine relevant spa-
tial scales. In our study, L. apodus did not respond signif-
icantly to individual seascape variables at any scale, and
additional analysis revealed that densities of juveniles
were not significantly different between seascape struc-
tural types. This indicates that either L. apodus does not
respond to the structure of the surrounding seascape, or
may respond to unmeasured variability such as the pres-
ence/absence of mangroves or perhaps to seascape
structure outside the scale range included in this study.
The importance of a relatively fine scale structure
(<100 m) for fish assemblages may relate to the pre-
ponderance of relatively small fish (mean 9.4 cm FL)
found using mangroves of SW Puerto Rico. In our
study, the species with the finest scale response were
also species with the largest proportion of small juve-
niles (Haemulon flavolineatum <10 cm = 90.8%; Ocy-
urus chrysurus <15 cm = 88.9%) of all focal species.
Kramer & Chapman (1999) found a strong positive lin-
ear allometric scaling relationship (r
2
= 0.73) between
body size and home range size for 29 tropical marine
fish, implying that, in general, smaller bodied fish sam-
ple their environment at finer spatial scales than do
larger bodied fish. Alternatively, scale differences may
be related to species-specific functional connectivity
between inshore juvenile habitat and offshore adult
habitat. Cocheret de la Morinière et al. (2002) pro-
posed that Lutjanus griseus (as well as L. apodus and
H. sciurus) are capable of long-distance, post-settle-
ment, life-cycle migrations (PLCM) that connect
inshore nursery habitats to adult habitats on offshore
coral reefs. In contrast, in the same study, the authors
differentiated H. flavolineatum and O. chrysurus as
exhibiting a stepwise PLCM with multiple spatial
shifts, with each sub-movement occurring at finer spa-
tial scales than those in long distance PLCM species.
Linking fish density and number of species to
seascape types
Across SW Puerto Rico, mangroves with extensive
seagrass beds (>40% of the seascape) in close proxim-
ity (<100 m) supported higher mean fish density and
number of species than mangroves with little or no sea-
grasses in close proximity. Elsewhere, several
seascape ecology studies that have incorporated the
distribution of surrounding habitat types have also
found that the presence, amount and proximity of sea-
grasses has emerged as an important explanatory vari-
able for fish spatial distributions in both mangroves
(Pittman et al. 2004) and coral reefs (Kendall et al.
2003, Grober-Dunsmore 2007).
However, we also found that not all mangrove-sea-
grass combinations were used equally, with offshore
mangrove-seagrass-dominated seascapes supporting
higher mean assemblage density and higher Haemu-
lon flavolineatum and Ocyurus chrysurus density than
onshore mangroves with adjacent seagrasses. O. chry-
surus was the most restricted in distribution, with only
offshore mangroves in close proximity (100 m) to high
coral reef (>15%) and seagrass cover (>40%) support-
ing high densities. Our results, therefore, indicate that
the preference of O. chrysurus for mangroves and
adjacent seagrasses is more specific than that indi-
cated by Nagelkerken et al. (2002), and the assignment
of equal ecological function to all mangrove-seagrass-
dominated seascapes across a region may be ecologi-
cally unrealistic.
A different spatial pattern of mangrove use was
found for Lutjanus griseus, with higher densities in
onshore mangroves. At the spatial extents examined in
this study (≤ 600 m radial extents), the presence of coral
reefs in close proximity to mangroves was not an
important determinant of fish densities for L. griseus.
Instead, higher densities of L. griseus were found in
282
Pittman et al.: Seascape types and spatial patterns of fish in mangroves
seascapes with extensive mangroves and extensive
seagrasses, but not necessarily in close proximity to
one another. A third response type was distinguished
for Haemulon sciurus densities, which did not differ
significantly between onshore or offshore mangroves
and seagrasses, but exhibited lowest densities within
onshore mangroves >600 m from coral reefs.
The functional significance of close proximity of man-
groves and seagrasses requires further investigation,
although it is likely that useable resources, such as sea-
grasses in close proximity, will provide a higher quality
habitat for many species than mangroves alone. This
may result from a sensitive dependence on the require-
ment for seagrass as settlement substratum (Pollux et
al. 2007), combined with the well-documented impor-
tance of seagrasses as feeding grounds for haemulids
and lutjanids (Ogden & Ehrlich 1977, Verweij et al.
2006). Furthermore, close proximity of highly struc-
tured habitat types may also provide an effective corri-
dor that enhances survival during spatial transitions be-
tween habitat types (Dorenbosch et al. 2004).
Evidence now exists demonstrating that many early
stage fish are capable of active habitat selection
through well-developed sensory and locomotory func-
tions (Leis & McCormick 2002), and that behavioural
preferences for settlement substratum may be medi-
ated by the amount of substratum (McDermott &
Shima 2006). Yet virtually nothing is known about the
process of hierarchical selection of habitat from region
to seascape to individual patch. For example, larval or
spawning adult fish may select patches suitable for
settlement and spawning that are also surrounded by
resources capable of supporting the sequence of
developmental stages and post-settlement movements
required to complete the life cycle. Further studies are
required to elucidate on the complex functional inter-
action between seascape structure and the various
ecological processes (settlement, predation, competi-
tion, growth rate, etc.) that determine fish species dis-
tribution patterns in nearshore marine environments.
Furthermore, future seascape studies should investi-
gate the influence of spatial and thematic resolution
when using benthic habitat maps and may also use-
fully incorporate additional environmental variables at
multiple scales, including substratum rugosity and the
physical and chemical properties of the surrounding
water.
Implications for resource management
The tendency to homogenise or subsume ecologically
important habitat variability in order to increase opera-
tional effectiveness has serious implications for the way
that we understand and manage our environment.
There is now an urgent need to understand the influ-
ence of seascape structure and to begin to incorporate
seascape spatial patterning into resource management
strategies. For instance, when designating manage-
ment units such as critical habitat for multi-habitat
users, it is important to look beyond individual habitat
types toward identification of critical seascapes at
scales appropriate to the movements of target species.
However, the apparent diversity of the species-specific
response to seascape structure, combined with a lack of
information on which structural types form optimal
habitat, highlight the need for a wider application of the
multi-scale seascape approach, with the specific objec-
tive of informing management. With regard to environ-
mental change, if seascape structure as a whole is
important, then its functional integrity may change
through the loss or gain of a single habitat type. Loss of
seagrasses adjacent to mangroves may mean that man-
groves can no longer be used by some species due to
reduction of supplementary or complementary food re-
sources, or of critical settlement substratum, or by dis-
rupting the functional connectivity between mangroves
and coral reefs. The results presented here are consis-
tent with the recommendation for conservation efforts
to protect seascapes with high connectivity between
mangroves, seagrasses and coral reefs (Mumby et al.
2004). In addition, we add the caveat that a seascape
type providing optimal function for one species may be
sub-optimal for another species.
Acknowledgements. This research was supported by NOAA’s
Coral Reef Conservation Program. We thank the staff of the
Biogeography Team at NOAA’s Center for Coastal Monitor-
ing and Assessment for field data collection; the Department
of Marine Science at the University of Puerto Rico
(Mayaguez) for field support; Captain Nazario for boat trans-
portation whilst in Puerto Rico; and E. Finnen for GIS techni-
cal support. We also extend a massive thank you to K. M.
Pittman and 4 anonymous reviewers for extensive comments
that helped in the refinement of this manuscript.
LITERATURE CITED
Adams AJ, Dahlgren CP, Kellison GT, Kendall MS, Layman
CA, Ley JA, Nagelkerken I, Serafy JE (2006) Nursery
function of tropical back-reef systems. Mar Ecol Prog Ser
318:287–301
Addicott JF, Aho JM, Antolin MF, Padilla DK, Richardson JS,
Soluk DA (1987) Ecological neighborhoods: scaling envi-
ronmental patterns. Oikos 49:340–346
Beets J, Muehlstein L, Haught K, Schmitges H (2003) Habitat
connectivity in coastal environments: patterns and move-
ments of Caribbean coral reef fishes with emphasis on
bluestriped grunt, Haemulon sciurus. Gulf Caribb Res
14(2):29–42
Blaber SJM (2000) Tropical estuarine fishes: ecology,
exploitation, and conservation. Blackwell Science, London
Bohlke JE, Chaplin CCG (1993) Fishes of the Bahamas and
adjacent tropical waters. University of Texas Press,
Austin, TX
283
Mar Ecol Prog Ser 348: 273–284, 2007
Bouwmeester BLK (2005) Ontogenetic migration and growth
of French grunt (Teleostei: Haemulon flavolineatum) as
determined by coded wire tags. Masters thesis, University
of Puerto Rico, Mayagüez
Brock VF (1954) A preliminary report on a method of estimat-
ing reef fish populations. J Wildl Manage 18:299–308
Christensen JD, Jeffrey CFG, Caldow C, Monaco ME,
Kendall MS, Appeldoorn RS (2003) Cross-shelf habitat uti-
lization patterns of reef fishes in southwestern Puerto Rico.
Gulf Caribb Res 14(2):9–27
Clarke KR, Warwick RM (1994) Change in marine communi-
ties: an approach to statistical analyses and interpretation.
Plymouth Marine Laboratory, Plymouth
Cocheret de la Morinière E, Pollux BJA, Nagelkerken I, van
der Velde G (2002) Post-settlement life cycle migration
patterns and habitat preference of coral reef fish that use
seagrass and mangrove habitats as nurseries. Estuar Coast
Shelf Sci 55:309–321
Dorenbosch M, van Riel MC, Nagelkerken I, van der Velde G
(2004) The relationship of reef fish densities to the proxim-
ity of mangrove and seagrass nurseries. Estuar Coast Shelf
Sci 60:37–48
Dunning JB, Danielson BJ, Pulliam HR (1992) Ecological pro-
cesses that affect populations in complex landscapes.
Oikos 65:169–175
Faunce CH, Serafy JE (2006) Mangroves as fish habitat: 50
years of field studies. Mar Ecol Prog Ser 318:1–18
García-Cagide AR, Claro R, Koshelev BV (1994) Reproduc-
ción. In: Claro R (ed) Ecología de los peces marinos de
Cuba. Inst. Oceanol. Acad. Cienc. Cuba. and Cen. Invest.
Quintana Roo (CIQRO) México, Chetumal, p 187–262
Grober-Dunsmore R, Frazer TK, Lindberg WJ, Beets J (2007)
Reef fish and habitat relationships in Caribbean
seascapes: the importance of reef context. Coral Reefs
26(1):210–216
Holland JD, Bert DG, Fahrig L (2004) Determining the spatial
scale of species’ response to habitat. BioScience 54:227–233
Kendall MS, Kruer CR, Buja KR, Christensen JD, Finkbeiner
M, Warner R, Monaco ME (2002) Methods used to map the
benthic habitats of Puerto Rico and the US Virgin Islands.
NOAA Tech Mem 152. Silver Spring, MD
Kendall MS, Christensen JD, Hillis-Starr Z (2003) Multi-scale
data used to analyze the spatial distribution of French
grunts, Haemulon flavolineatum, relative to hard and soft
bottom in a benthic landscape. Environ Biol Fish 66:19–26
Kramer DL, Chapman MR (1999) Implications of fish home
range size and relocation for marine reserve function.
Environ Biol Fish 55:65–79
Kruskal JB (1964) Nonmetric multidimensional scaling: a
numerical method. Psychometrika 29:115–129
Leis JM, McCormick MI (2002) Recent developments in the
biology, behaviour and ecology of the pelagic, larval stage
of coral-reef fishes. In: Sale PF (ed) Coral reef fishes.
Dynamics and diversity in a complex ecosystem. Acade-
mic Press, San Diego, CA
Lindholm J, Kaufman L, Miller S, Wagschal A, Newville M
(2005) Movement of yellowtail snapper (Ocyurus chry-
surus Block, 1790) and black grouper (Mycteroperca
bonaci Poey, 1860) in the northern Florida Keys National
Marine Sanctuary as determined by acoustic telemetry.
Marine Sanctuaries Conservation Series MSD-05–4,
US Department of Commerce, National Oceanic and
Atmospheric Administration, Marine Sanctuaries Divi-
sion, Silver Spring, MD
Manson FJ, Loneragan NR, Harch BD, Skilleter GA, Williams
L (2005) A broad-scale analysis of links between coastal
fisheries production and mangrove extent: a case-study
for northeastern Australia. Fish Res 74:69–85
McDermott CJ, Shima JS (2006) Ontogenetic shifts in micro-
habitat preference of the temperate reef fish Forsterygion
lapillum: implications for population limitation. Mar Ecol
Prog Ser 320:259–266
Moran PA (1950) Notes on continuous stochastic phenomena.
Biometrika 37:17–23
Mumby PJ, Edwards AJ, Arias-Gonzalez JE, Lindeman KC and
8 others (2004) Mangroves enhance the biomass of coral reef
fish communities in the Caribbean. Nature 427:533–536
Nagelkerken I, Dorenbosch M, Verberk WCEP, Cocheret de
la Morinière E, van der Velde G (2000) Day–night shifts of
fishes between shallow-water biotopes of a Caribbean
bay, with emphasis on the nocturnal feeding of Haemuli-
dae and Lutjanidae. Mar Ecol Prog Ser 194:55–64
Nagelkerken I, Roberts CM, van der Velde G, Dorenbosch M,
van Riel MC, de la Morinière EC, Nienhuis PH (2002) How
important are mangroves and seagrass beds for coral-reef
fish? The nursery hypothesis tested on an island scale.
Mar Ecol Prog Ser 244:299–305
Ogden JC, Ehrlich PR (1977) The behavior of heterotypic rest-
ing schools of juvenile grunts (Pomadasyidae). Mar Biol
42:273–280
Parrish JD (1989) Fish communities of interacting shallow-
water habitats in tropical oceanic regions. Mar Ecol Prog
Ser 58:143–160
Pittman SJ, McAlpine CA (2003) Movement of marine fish
and decapod crustaceans: process, theory and application.
Adv Mar Biol 44:205–294
Pittman SJ, McAlpine CA, Pittman KM (2004) Linking fish
and prawns to their environment: a hierarchical landscape
approach. Mar Ecol Prog Ser 283:233–254
Pollux BJA, Verberk WCEP, Dorenbosch M, Cocheret de la
Morinière E, Nagelkerken I (2007) Habitat selection dur-
ing settlement of three Caribbean coral reef fishes: indica-
tions for directed settlement to seagrass beds and man-
groves. Limnol Oceanogr 52(2):903–907
Robbins BD, Bell SS (1994) Seagrass landscapes: a terrestrial
approach to the marine environment. Trends Ecol Evol 9:
301–304
Rooker JR, Dennis GD (1991) Diel, lunar and seasonal
changes in a mangrove fish assemblage off southwestern
Puerto Rico. Bull Mar Sci 49:684–698
Sheaves M (2005) Nature and consequences of biological
connectivity in mangrove systems. Mar Ecol Prog Ser 302:
293–305
Turner MG (1989) Landscape ecology: the effect of pattern on
process. Annu Rev Ecol Syst 20:171–197
Verweij MC, Nagelkerken I, Wartenbergh SLJ, Pen IR,
van der Velde G (2006) Caribbean mangroves and sea-
grass beds as daytime feeding habitats for juvenile French
grunts, Haemulon flavolineatum. Mar Biol 149(6):
1291–1299
Watson M, Munro JL, Gell FR (2002) Settlement, movement
and early juvenile mortality of the yellowtail snapper
Ocyurus chrysurus. Mar Ecol Prog Ser 237:247–256
Wiens JA (1989) Spatial scaling in ecology. Funct Ecol 3:
385–397
Wiens JA, Stenseth NC, Van Horne B, Ims RA (1993) Eco-
logical mechanisms and landscape ecology. Oikos 66:
369–380
284
Editorial responsibility: Otto Kinne (Editor-in-Chief),
Oldendorf/Luhe, Germany
Submitted: December 5, 2006; Accepted: May 14, 2007
Proofs received from author(s): October 2, 2007
