Estuarine, Coastal and Shelf Science 60 (2004) 37e48
                                                        www.elsevier.com/locate/ECSS




             The relationship of reef fish densities to the
            proximity of mangrove and seagrass nurseries
         M. Dorenbosch, M.C. van Riel, I. Nagelkerken), G. van der Velde
      Department of Animal Ecology & Ecophysiology, University of Nijmegen, Toernooiveld 1, 6525 ED, Nijmegen, The Netherlands

                        Received 20 May 2003; accepted 28 November 2003




Abstract

  Visual census surveys were used to study the distribution of coral reef fishes that are associated with seagrass beds and mangroves
                                                    x
in their juvenile phase, on various coral reef sites along the coast of the Caribbean island of Curacao (Netherlands Antilles). The
hypothesis tested was that various reef fish species occur in higher densities on coral reefs adjacent to nursery habitats than on reefs
located at some distance to these habitats. Of 17 coral reef fish species that are known to use bays with seagrass beds and mangroves
as nurseries (nursery species), 15 were observed in quadrats on the reef. Four nursery species, Haemulon sciurus, Lutjanus apodus,
Ocyurus chrysurus and Scarus coeruleus occurred in significantly higher densities on coral reefs adjacent to bays with seagrass beds
and mangroves. Lutjanus analis, Lutjanus mahogoni and Sphyraena barracuda also had their highest densities on reefs adjacent to
these bays, although differences between the distinguished reef categories were not always significant. It is suggested that these seven
species are highly dependent on the presence of bays with seagrass beds and mangroves as nurseries on an island scale. Eight other
species that are known to use seagrass beds and mangroves as nurseries did not have their highest densities on reefs adjacent to bays
with seagrass beds and mangroves. For six of these species, juveniles were also observed on the reef. It is suggested that these species
are able to use the reef as an alternative nursery and do not depend strictly on the presence of bays with seagrass beds and
mangroves as nurseries.
Ó 2004 Elsevier Ltd. All rights reserved.

Keywords: nursery grounds; mangrove swamps; seagrasses; coral reef fishes; migration; juveniles




                                                     `
                                      2000a; Cocheret de la Moriniere et al., 2002; Adams and
1. Introduction
                                      Ebersole, 2002; Nagelkerken and van der Velde, 2002).
                                                   x
  In various parts of the world, shallow coastal areas          On the island of Curacao (Netherlands Antilles),
containing mangroves and seagrass beds are considered           Nagelkerken et al. (2000b) showed that an inland
important nurseries for juvenile fish (Pollard, 1984;            marine bay with seagrass beds and mangroves served
Parrish, 1989; Baelde, 1990; Robertson and Blaber,             as a nursery habitat for at least 17 coral reef species
1992). Pelagic fish larvae settle into these habitats, and         (indicated below as nursery species). It has been shown
grow from juveniles to subadults or adults that leave           on various islands that a reduced density of several of
these habitats by means of post-settlement migrations           these nursery species on the coral reef is related to the
(Jones, 1991; Blaber, 2000). In the Caribbean, shallow           absence of seagrass beds and mangroves (Nagelkerken
waters with mangroves and seagrass beds are charac-            et al., 2002). This suggests that these nursery species
terised by the presence of high densities of juveniles of         depend on the presence of seagrass beds and mangroves
several coral reef species that are assumed to migrate to         as a nursery habitat. If this is the case, coral reefs ad-
the coral reef on reaching the (sub)adult stage (Austin,          jacent to mangrove and seagrass nursery areas might be
1971; Louis and Guyard, 1982; Nagelkerken et al.,             expected to harbour higher densities of adults of these
                                      nursery species than reefs located at greater distance to
                                      these nursery areas, assuming that adult migration along
 ) Corresponding author.
                                      the coast between reefs is limited.
  E-mail address: i.nagelkerken@sci.kun.nl (I. Nagelkerken).

0272-7714/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ecss.2003.11.018
38              M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48

            x
  The island of Curacao provides an opportunity to test       of 80e90 m. The southwestern coast features eight large
this hypothesis along the coast of a single island. The        inland bays (Fig. 1), which are dominated by man-
occurrence of both seagrass beds and mangroves is           groves, seagrass beds and a muddy/sandy seabed
restricted to several shallow inland marine bays situated       (Table 1). Rocky substratum, in the form of boulders
at the southwestern part of the island, allowing a clear       and erosional notches, is present to some degree only in
distinction to be made between reefs adjacent to bays         Spanish Water Bay. Notches are formed at and under
with seagrass beds and mangroves, reefs adjacent to          the water line through biochemical solution of the fossil
                                                            ´
bays without seagrass beds and mangroves, and reefs          reef terrace along the shoreline (de Buisonje and
located at some distance from bays. In a pilot study,         Zonneveld, 1960). Fringing mangroves grow in stands
Nagelkerken et al. (2000b) already observed reduced          along the sandy shoreline of the bays and consist of
densities of six nursery species on the reef at an in-        Rhizophora mangle (see Nagelkerken et al., 2000b and
creasing distance from a single bay with nursery ha-         Nagelkerken et al., 2001 for a detailed description of
bitats. However, their study focused on only a few          these habitats). Seagrass beds in Spanish Water Bay and
species and a small part of the reef, and did not consider      Fuik Bay consist of Thalassia testudinum whereas those
the possible relation with fish size.                 in Piscadera Bay consist of Syringodium filiforme. All
  While subadult or adult bay-to-reef migrations are         bays have a narrow entrance from the open sea. The
likely to supply coral reefs adjacent to bays with nursery      water of Zakito Bay is polluted with heavy metals from
species, reefs at some distance from these habitats can be      a desalination plant and has an elevated temperature
colonised either by fish dispersal on reefs along the coast      and salinity (Nagelkerken, unpubl. data). The average
                                                 x
or by small populations of juvenile fish larvae that settle      daily tidal range in Curacao is about 30 cm (de Haan
and survive on these reefs. Several studies (Tulevech and       and Zaneveld, 1959), and the bays are not subject to
Recksiek, 1994; Macpherson, 1998; Zeller, 1998) suggest        strong tidal currents.
that it is predominantly the larger individuals that
undertake migrations along the reef over larger dis-         2.2. Study design
tances. Whereas the population of nursery species on
coral reefs adjacent to bays with seagrass beds and           The distribution of the 17 nursery species (listed in
mangroves is represented by older juveniles, subadults        Table 2) was studied at 11 coral reef sites in a gradient
and adults (Nagelkerken et al., 2000b; Cocheret de la         along the southwestern coast at varying distances from
     `
Moriniere et al., 2002; Nagelkerken and van der Velde,        two types of bays. The 11 reef sites were subdivided into
2002), it might be expected that the population of nur-        four ‘reef categories’ (Fig. 1): (1) three coral reef sites
sery species on coral reefs at great distances to bays with      adjacent to bays featuring major seagrass beds and
seagrass beds and mangroves would consist predomi-          mangrove habitats, indicated below as sgemg bays
nantly of adults.                           (distance to the bay !1 km); (2) three coral reef sites
  The present study tested the hypothesis that juveniles       adjacent to bays dominated by bare sediment without
and adults of nursery species occur in higher densities on      marine vegetation (distance to the bay !1 km), but
coral reefs adjacent to nursery habitats than on reefs        situated at some distance to sgemg bays, indicated
located at some distance to these habitats. In accordance       below as mud/sand bays (distance to nearest sgemg bay
with this, reduced densities of adults and the absence of       between 3.2 and 25.6 km); (3) two coral reef sites
juveniles on coral reefs away from these bays, are ex-        situated between sgemg bays (distance to nearest
pected. The degree to which nursery species might utilise       sgemg bay between 3.1 and 3.5 km, and to nearest
the coral reef as an alternative juvenile habitat instead of     mud/sand bay between 8.0 and 15.5 km); and (4) three
seagrass and mangrove habitats was also investigated.         coral reef sites located at greater distance to sgemg bays
                                   (distance to nearest sgemg bay between 11.6 and
                                   38.5 km, and to nearest mud/sand bay between 4.7
                                   and 13.4 km). The reef at Holiday Beach was located
2. Materials and methods
                                   close to a bay (St. Anna Bay), but was nevertheless
                                   defined as a reef situated between sgemg bays (Fig. 1).
2.1. Study area
                                   Due to industrial activities in St. Anna Bay (involving
  The present study was carried out on the coral reef at       the presence of a large harbour, oil refinery and
the leeward southwestern coast of the Caribbean island        shipyards), all natural marine vegetation and muddy/
     x
of Curacao, Netherlands Antilles (Fig. 1). The coast on        sandy habitats have been destroyed, and the water is
this side of the island is characterised by the presence of      highly polluted (van den Hoek et al., 1972). Therefore,
a continuous fringing coral reef that consists of a small       the ecological function of this bay cannot be considered
surf zone and a reef flat that gradually slopes down to        typical for a mud/sand bay, and the reef close to this bay
a ‘drop-off’ at 7e12 m (Bak, 1975). At the drop-off, the        cannot be considered typical for a reef adjacent to an
reef slopes off steeply and ends in a sandy plain at depths      unpolluted mud/sand bay.
                                                                           39
                  M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48




                                               x
Fig. 1. Locations of the eight largest bays and 11 reef sites sampled on the island of Curacao (latitude 12# N, longitude 68# W). The density pattern
                                                            x
of nursery species and their non-nursery congeners along the gradient of reef sites is shown below the map of Curacao. Separate patterns are shown
(a) for pooled densities of the seven nursery species that had their highest densities at reef sites adjacent to bays with seagrass beds and mangroves
(see Table 1) and their congeners, and (b) for pooled densities of the eight nursery species that did not have higher densities at reef sites adjacent to
bays with seagrass beds and mangroves (see Table 1) and their congeners. Error bars indicate SEM. The table shows the mean coral cover (%) of
each depth zone.



  Besides the 17 nursery species, the densities of nine            (2000b) it is assumed that juveniles of these congeners
common non-nursery congeners of the nursery species               do not use seagrass and mangrove habitats as a nursery.
were also determined on the reef sites: Acanthurus                 Data on the reef fish community structure were
                                        collected by visual census in quadrats using SCUBA and
bahianus, Acanthurus coeruleus, Chaetodon striatus,
                                        a stationary point-count method (Polunin and Roberts,
Haemulon carbonarium, Haemulon chrysargyreum, Sca-
                                        1993) by two independent observers. Square quadrats of
rus taeniopterus, Scarus vetula, Sparisoma aurofrenatum
and Sparisoma viride. Based on Nagelkerken et al.                10 ! 10 m were surveyed at four depth zones: shallow
                                                                                                             40
Table 1
                                               x
Main shallow-water habitats of the eight largest bays along the southwestern coastline of Curacao, and the abundance of nursery species




                                                                                                             M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48
       Total   Bay    Bay area   Length of Value as Lutjanus Lutjanus Lutjanus Lutjanus Gerres Chaetodon Haemulon   Haemulon Haemulon Ocyurus Scarus Sparisoma   Sphyraena
       bay    area    consisting  inundated nursery analis  apodus griseus mahogoni cinereus capistratus flavolineatum sciurus parra  chrysurus iserti chrysopterum barracuda
       area   covered  of muddy/  mangroves
       (m2)   by     sandy    along
            seagrass  seabeds   shoreline
            beds    (%)     (m)
            (%)
Sta. Martha  569,238       100           Low    **    *     *          ***                             *                 ***
            e            e                         e          e      e       e     e           e    e
  Bay
San Juan   159,060       100     60     Low    *     *     *          **
            e                                     e          e      e       e     e      e     e    e       e
  Bay
St. Michiel  193,640       100           Very        *               ***                                              *
            e            e          e          e     e          e      e       e     e      e     e    e
  Bay                           low
Piscadera   726,168  2      98     3964    High   ***    ***    ***         **    *      ***      **           *     ***   ***      ***
                                                  e                             e
  Bay
Zakito    140,151       100     2267    Very                        *
            e                      e     e     e     e          e      e       e     e      e     e    e       e
  Bay                           low
St. Anna   4,190,000 e     100           Very   nd    nd    nd    nd     nd    nd     nd       nd     nd     nd     nd   nd       nd
                        e
  Bay                           low
                                   ea              ea                             ea           ea   ea
Spanish    2,846,511 15     82     8702    High        *     *          *     ***     ***      ***          ***                ***
  Water
  Bay
Fuik Bay   687,556  3      97     3200    High   *     *          ***    *     **     *       *           *     *    *       *
                                             e                                  e
The presence of 13 nursery species is based on Nagelkerken et al. (2001) and unpublished data (Nagelkerken) for which the bays were sampled using a beach seine net. Based on estimated total standing stocks of
juveniles on seagrass beds and muddy/sandy seabeds, presence of species is expressed as absent (e), low (*), high (**) or very high (***). Classes are distinguished per species by dividing the highest total standing
stock by three. Based on mean abundance and mean species richness of nursery species in the main nursery habitats of the bays, Nagelkerken (unpubl. data) classified the nursery function of the bays as high, low
or very low. No data are available for St. Anna Bay, but its nursery function is assumed to be very low (see text). nd, no data.
  a
   Presence in seagrass/mangrove habitats demonstrated by means of visual census (Nagelkerken et al., 2000b).
                                                                  41
                  M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48

                                        For each species, data were also analysed separately
Table 2
Size classes (cm) used to define juveniles for each nursery species, based  for juveniles, based upon their maturation size (Table 2).
upon half the length of the smallest maturation sizes obtained from
                                      Maturation sizes were obtained from FishBase World
FishBase World Wide Web (Froese and Pauly, 2002) and Munro
                                      Wide Web (Froese and Pauly, 2002) and Munro (1983).
(1983) (for Lutjanus analis, the maturation size of Ocyurus chrysurus
                                      If this database gave maturation size as a range, the
was used to distinguish the juveniles (see text))
                                      smallest observed maturation size was used. Juveniles
Species          Juveniles Species          Juveniles
                                      were defined as individuals smaller than half the ma-
             0e10                 0e10
Acanthurus chirurgus        Lutjanus griseus
                                      turation size (i.e., maturation size divided by two) to be
             0e5                 0e12.5
Chaetodon capistratus        Lutjanus mahogoni
                                      able to distinguish them from larger subadults. Matu-
             0e10                 0e12.5
Gerres cinereus           Ocyurus chrysurus
                                      ration size for Lutjanus analis was 37.5 cm, which is
             0e5                 0e15
Haemulon flavolineatum        Scarus coeruleus
             0e12.5                no data
Haemulon parra           Scarus guacamaia            much larger than that of the other Lutjanidae studied
             0e10                 0e10
Haemulon plumieri          Scarus iserti
                                      (i.e., 17.5e22.5 cm). This value was based on only one
             0e10                 0e12.5
Haemulon sciurus          Sparisoma chrysopterum
                                      study (quoted in FishBase World Wide Web), and may
             0e12.5                0e30
Lutjanus analis           Sphyraena barracuda
                                      therefore not be very reliable. The same maturation size
             0e12.5
Lutjanus apodus
                                      for L. analis as for Ocyurus chrysurus was therefore
                                      used. This was based on the fact that O. chrysurus and
                                      L. analis have almost the same maximum length, and
reef flat (2.5 m), reef flat (5 m), drop-off (10 m) and reef
slope (15 m). A single 10 m line was used as a reference          because for O. chrysurus a large number of studies have
for the size of a complete quadrat. At each site, ten            determined the maturation size (quoted in FishBase
quadrats (placed in a direction parallel to the coastline)         World Wide Web).
per depth zone were surveyed, to a total of 40 quadrats             Since fish densities are often correlated to the degree
per site. These 40 quadrats were surveyed during three           of coral cover (Luckhurst and Luckhurst, 1978; Hixon
visual census rounds: 16 quadrats at each site in              and Beets, 1993; Grigg, 1994) the total hard coral cover
December 1999, 16 quadrats in January 2000 and 8              (both living and dead corals) at each site for each depth
                                      zone was visually quantified. To estimate coral cover of
quadrats in February 2000. After placing the quadrat
line, the observer waited for 5 min to minimise fish             the quadrat, the 10 ! 10 m quadrat was divided into
disturbance. All nursery species within or passing             four quarters of 5 ! 5 m. For each quarter, coral cover
through the quadrat were then counted over a period             was estimated separately and was averaged for the
of 10 min. During fish counting the observer was at the           whole quadrat afterwards. The 10 m quadrat line was
edge of the quadrat for 8 min. After 8 min, the observer          marked with a red label in the middle to visually
moved through the quadrats to search for and/or               estimate the size of each quarter. Because the number of
estimate sizes of possible small juvenile fish hiding            quadrats for which the cover was estimated was not
                                      constant for each site (between 6 and 10 estimations per
behind or between coral boulders. Care was taken to
ensure that fishes that regularly moved in and out of the          depth zone per site), cover was averaged for quadrats
quadrat were not counted twice. Fishes were classified            and expressed as mean hard coral cover per depth zone
into size classes of 2.5 cm. Each reef site was visited by         per site.
the two observers simultaneously and each observer
collected a total number of 20 quadrats. The location on          2.3. Statistical analysis
the reef, within a reef site, where an observer would
place the quadrats was randomly allocated to each of              Principal Component Analysis (PCA) was used to
the observers during each census round, making sure             study the spatial distribution pattern of nursery species
not to recount the same area of reef. Species identifica-          along the gradient of reef sites. PCA was carried out on
                                      log10-transformed mean fish densities (with all size
tion and quantification were first thoroughly and
simultaneously practised by the two observers. Estima-           classes pooled) per reef site, using the Canoco 4.0
tion of size classes was trained by repeatedly estimating          ordination program (ter Braak and Smilauer, 1998).
the sizes of 40 pieces of electrical wires of known length         Default options were used for the analysis: scaling was
(range 2.5e50 cm, in classes of 2.5 cm) under water.            focused on inter-species correlations (to focus more on
Training was continued until differences in size-estima-           the relationships between species), species scores were
tion were minimal (maximum difference of one size class           divided by the standard deviation (to reduce the
of 2.5 cm for wire sizes !15 cm and two size classes for          influence of species with a large variance in density),
sizes O15 cm) between the two observers. Training in            and the data were centred by species (used for ordinary
fish species identification was continued until it was the          PCA, where each species is weighted by its variance).
same between the observers. The training procedure               To test the influence of coral cover on fish density,
started two weeks before the census and was repeated            separate linear regressions were run for each species at
before each census round (three census rounds over             each depth zone. Since Haemulon parra occurred only at
a period of three months).                         one reef site, no regression analysis could be performed
42              M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48

for this species. For each species, mean fish density (with      a fourth cluster, in which none of the species had their
all size classes pooled) at each site (N ¼ 11) was used as      highest densities.
the dependent variable and mean hard coral cover was          Generalized linear models were significant for 14
used as the regression. Regression analyses were           species (Table 3). Post-hoc comparisons showed signif-
performed using SPSS version 11.5.                  icantly higher counts of Ocyurus chrysurus, Lutjanus
  The influence of the presence of a bay nursery habitat      apodus, Haemulon sciurus and Scarus coeruleus in the
on the occurrence of nursery species on the reef was         category reefs adjacent to sgemg bays than in the other
tested using generalized linear models. Because the data       three categories (Fig. 3a, Table 3). Ocyurus chrysurus
consisted of counts, a model based on a Poisson dis-         had decreasing counts on reefs located at increasing
tribution was used. For each quadrat, visual census         distances from sgemg bays. Lutjanus mahogoni and
counts of all size classes were pooled. Because the 10        Lutjanus analis also had their highest densities in the
quadrats of a depth zone were laid out in a line parallel      category reefs adjacent to sgemg bays (Fig. 3a). For
to those in other depth zones, counts of quadrats dis-        these two species, fish counts in the category reefs
tributed over the four depth zones were pooled to one        adjacent to sgemg bays differed significantly from those
count. Therefore, data for each site consisted of 10         in the categories reefs between sgemg bays and reefs
counts (i.e., each a sum of counts over four depth          adjacent to mud/sand bays, but not from reefs at great
zones). These fish counts were used as the dependent         distance from sgemg bays. Sphyraena barracuda had its
variable in the model. The factor ‘reef category’ was        highest density in the category reefs adjacent to sgemg
used as a fixed factor. Because data were collected          bays, but a significant difference between counts was
during three time periods (visual census rounds), a three-      only found between reefs adjacent to sgemg bays and
level block was added to the model, each level being one       reefs at great distance from sgemg bays.
visual census round. The log link function and type 3          Of the other eight nursery species, two had their
analysis were used in the model. Post-hoc comparisons        highest density in the category reefs between sgemg
between reef categories were made by calculating dif-        bays (Chaetodon capistratus and Sparisoma chrysopte-
ferences of least squares means. Statistics were per-        rum) and two in the category reefs adjacent to mud/
formed using the SAS system for Windows V8.             sand bays (Haemulon flavolineatum and Scarus iserti)
                                   (Table 3). Three species had their highest densities in
                                   the category reefs at great distance from sgemg bays
                                   (Gerres cinereus, Lutjanus griseus, and Haemulon parra).
3. Results
                                   Densities of Acanthurus chirurgus were highest on reefs
                                   adjacent to sgemg bays and on reefs adjacent to mud/
3.1. Total fish density
                                   sand bays.
                                    Pooled densities of the seven nursery species occur-
  In the present study, 15 of the 17 known nursery
                                   ring in higher densities on reefs adjacent to sgemg bays
species were observed in the quadrats on the reef.
                                   were higher at all reef sites adjacent to sgemg bays than
Haemulon plumieri and Scarus guacamaia were not
                                   at other reef sites (Fig. 1a). This pattern was not found
observed.
                                   for the other eight nursery species observed on the reef
  Of the 56 linear regressions between fish density and
                                   (Fig. 1b). Non-nursery congeners of species with higher
coral cover, only three were significant: Haemulon
sciurus in the 15 m zone (P ! 0:01; R2 ¼ 0:63; Y ¼          densities on reefs adjacent to sgemg bays, had their
0:91 ÿ 1:20X), Scarus coeruleus in the 5 m zone (P !         highest densities on reef sites in the southwestern part of
0:01; R2 ¼ 0:65; Y ¼ 0:60C1:57X) and Lutjanus mahog-         the gradient along the coast of the island, at great
oni in the 5 m zone (P ! 0:05; R2 ¼ 0:37; Y ¼ ÿ2:63C         distance from bays with sgemg (Fig. 1a). Non-nursery
                                   congeners of species without higher densities on reefs
11:08X).
                                   adjacent to sgemg bays did not show higher densities in
  PCA allowed the reef sites to be divided into four
                                   any particular part of the gradient of reef sites examined
clusters (Fig. 2). One cluster was formed by the three
                                   (Fig. 1b).
reef sites adjacent to sgemg bays and was characterised
by nine nursery species. Compared with the other reef
sites, the mean densities of seven of these species were       3.2. Juvenile fish density
highest on reefs adjacent to sgemg bays (Table 3). A
second cluster was formed by the reefs between sgemg          For the seven nursery species which had their highest
bays and was characterised by high densities of           densities (for the entire size range) on reefs adjacent to
Chaetodon capistratus. A third cluster was formed by         sgemg bays, juveniles were also observed on the coral
two reefs adjacent to mud/sand bays and one reef at         reef (Fig. 3b). An exception was Lutjanus analis, for
great distance from sgemg bays, and harboured five          which only adults were observed on the reef. Juveniles of
species. Two reefs located at great distance from sgemg       Haemulon sciurus were only observed on reefs adjacent
bays and one reef adjacent to a mud/sand bay formed         to sgemg bays, and those of Sphyraena barracuda only
                                                                         43
                  M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48




Fig. 2. Principal Component Analysis (PCA) of mean densities of the 15 nursery species at various reef sites. The horizontal axis represents the first
PCA axis, the vertical axis the second PCA axis. The first two axes accounted for 67.9% of the total variance. Abbreviations: sgemg bays: bays with
seagrass beds and mangroves; mud/sand bays: bays dominated by muddy/sandy seabeds; Achi: Acanthurus chirurgus; Ccap: Chaetodon capistratus;
Gcin: Gerres cinereus; Hfla: Haemulon flavolineatum; Hpar: Haemulon parra; Hsci: Haemulon sciurus; Lana: Lutjanus analis; Lapo: Lutjanus apodus;
Lgri: Lutjanus griseus; Lmah: Lutjanus mahogoni; Ochr: Ocyurus chrysurus; Scoer: Scarus coeruleus; Sise: Scarus iserti; Schr: Sparisoma chrysopterum;
Sbar: Sphyraena barracuda. On the basis of sites and species which showed the highest similarity in composition and density distribution (using PCA),
four clusters of sites and species were identified and bordered by lines.




on reefs between sgemg bays. Despite the presence of             densities in seagrass/mangrove habitats and in reef
juveniles of six of these seven nursery species on the            habitats (Fig. 4b).
coral reef, densities of their juveniles were much higher
in seagrass beds and mangroves than on the reef
(Fig. 3b). An exception was Scarus coeruleus, for which            4. Discussion
juvenile densities on the coral reef and those in seagrass
beds in Spanish Water Bay were similar.                     The present study showed significantly higher densi-
  For the eight nursery species which did not show              ties of four nursery species on reefs adjacent to sgemg
highest densities (for the entire size range) on reefs ad-          bays than in all three other reef categories, whereas three
jacent to sgemg bays, juveniles were also found on the            other nursery species showed significantly higher densi-
coral reef, except Lutjanus griseus and Haemulon parra            ties at reefs adjacent to sgemg bays than in two of the
(Fig. 4a). The eight species can be divided into two             three other reef categories. This is probably caused by the
groups. Densities of juveniles of Chaetodon capistratus,           very high densities in the bays (summarised in Table 1)
Haemulon flavolineatum, Gerres cinereus, L. griseus, and            of juveniles, which migrate to the adjacent reef when
H. parra were considerably higher in seagrass beds or             reaching adulthood. This connectivity between nursery
mangroves in Spanish Water Bay than on the reef                habitats in a bay and the reef adjacent to a bay has been
(Fig. 4a) whereas juveniles of Sparisoma chrysopterum,            indicated before for Spanish Water Bay (Nagelkerken
Scarus iserti, and Acanthurus chirurgus showed similar            et al., 2000b; Nagelkerken and van der Velde, 2002;
44                 M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48

Table 3
Results of the generalized linear models with reef category as fixed factor and survey time as random block
             Model       Block       Mean density per      P-values of post-hoc comparisons
                                reef category
                                (# ind. 100 mÿ2)
             X2         X2        1   2   3   4   1e2    1e3    1e4    2e3    2e4    3e4
                  P         P
Species with highest density for reef category 1
              654.50 !0.001 0.95    ns        1.9  1.3  0.5             !0.001 0.009
Ocyurus chrysurus                       6.2            !0.001  !0.001             !0.001 !0.001
              245.36 !0.001 1.20    ns        0.7  1.5  1.7                       !0.001 ns
Lutjanus apodus                        4.0            !0.001  !0.001   !0.001 !0.001
              54.66 !0.001 9.39    0.009      0.1  0.4  0.1       0.006   !0.001 0.005    ns   0.001
Haemulon sciurus                        0.7            !0.001
              55.69 !0.001 13.25    0.001      0.0  0.2      0.001   0.026       0.020
Scarus coeruleus                        0.4         e
              23.13 !0.001 58.76            1.8  1.3  1.9  0.041        ns   0.026    ns    0.004
Lutjanus mahogoni                       2.3
                           !0.001                     !0.001
              11.87   0.009 5.94  ns        0.0  0.0  0.1  0.033   0.011   ns   ns      ns    ns
Lutjanus analis                        0.2
              10.47   0.015 9.13  0.010         0.1  0.1  ns    ns     0.006 ns      0.045   ns
Sphyraena barracuda                      0.2  0.2

Other species
             501.77  !0.001 3.20 ns   1.8         1.1  2.2             0.027
Chaetodon capistratus                         7.3         !0.001  !0.001        !0.001  !0.001  !0.001
             106.78  !0.001 3.36 ns   1.0         0.9  0.2       ns
Sparisoma chrysopterum                        1.7         !0.001        !0.001  !0.001  !0.001  !0.001
              53.40  !0.001 3.08 ns   6.3      4.7     5.8       0.001   ns         0.003
Haemulon flavolineatum                            7.6      !0.001             !0.001       !0.001
             210.51  !0.001 84.45 !0.001 9.3      6.0     5.0       ns               0.012
Scarus iserti                                9.9      !0.001        !0.001  !0.001       !0.001
              31.08  !0.001 0.90 ns   0.2      0.2  0.5      ns    0.006        0.002        ns
Gerres cinereus                                  0.6             !0.001       !0.001
              22.52  !0.001 5.25 ns   0.1         0.1           ns     ns              ns
Lutjanus griseus                                  0.2
                                   e
             np
Haemulon parra                                   0.1
                            e       e   e
              28.00  !0.001 91.24 !0.001 1.5      0.9     0.8  0.001   ns     !0.001 0.002    ns
Acanthurus chirurgus                            1.5                               !0.001
             np
Haemulon plumieri                   e       e   e   e
             np
Scarus guacamaia                    e       e   e   e
P-values of post-hoc comparisons (differences of least mean squares) between the four types of reef categories are shown. Fish counts were converted
into mean fish densities per reef category; highest mean density is printed in bold. Abbreviations and symbols: np: not enough counts to perform the
test; ns: non-significant (P > 0:05); e: not observed; 1: reefs in front of bays with seagrass beds and mangroves; 2: reefs between bays with seagrass
beds and mangroves; 3: reefs in front of bays dominated by bare sediment; 4: reefs at great distances from bays with seagrass beds and mangroves.


             `
Cocheret de la Moriniere et al., 2002). The present study           new individuals on the reef, resulting in high densities on
suggests that all sgemg bays along the southwestern coast           reefs adjacent to these bays.
            x
of the island of Curacao show this type of connectivity             An exception was Lutjanus mahogoni, for which den-
for certain coral reef fish species. A direct interlinkage           sity differences between reefs adjacent to sgemg bays
between these habitats by fish life-cycle migration is             and the other types of reef categories were not as large
difficult to show, but studies using otolith microchemistry           as those for the other six species. A possible explanation
(Gillanders, 2002; Gillanders and Kingsford, 1996) have            may be found in the ability of this species to spend its
confirmed the existence of these life-cycle migrations             juvenile phase on the reef. Based on observations of
between juvenile habitats and adult habitats in temperate           juveniles on the reef in the present study and by Wilson
marine fish species.                              (2001) and Nagelkerken et al. (2000a), ‘‘local recruit-
  Regarding these seven species with the highest den-            ment’’ on the reef may be an important source of new
sities on reefs adjacent to sgemg bays, Nagelkerken              individuals. The higher densities on reefs adjacent to
et al. (2002) found that densities of Haemulon sciurus,            sgemg bays might be a result of an additional input of
Lutjanus apodus and Ocyurus chrysurus were greatly              individuals from these habitats onto the reef. Compar-
reduced on coral reefs of islands lacking seagrass and            isons of densities of this species between islands with and
mangrove habitats relative to islands where these hab-            without seagrass beds and mangroves did not reveal any
itats were present, indicating that these species are             differences (Nagelkerken et al., 2002) and are consistent
highly dependent on these nursery habitats. For Lut-             with this hypothesis.
janus analis, Sphyraena barracuda and Scarus coeruleus,             If sgemg bays function as the main source of new
Nagelkerken et al. (2002) found a possible dependence             individuals on the reef, the presence of these six species
on mangrove and/or seagrass nurseries. The present              on reefs not adjacent to sgemg bays may partly result
study suggests that the presence of sgemg bays strongly            from fish dispersal along the coast. This may explain
influences the distribution pattern of these six species on          why the three types of reef located at great distance from
the coral reef along the coast of a single island. Since           sgemg bays showed much lower densities for six of
mud/sand bays that lack seagrass and mangrove                 these nursery species. Studies have shown that fishes
habitats have a limited nursery function (Nagelkerken             are able to migrate along reefs over distances ranging
et al., 2001; Table 1), sgemg bays are likely to function           from hundreds of metres to several kilometres (Tulevech
as the main, and for some species the only, source of             and Recksiek, 1994; Kanashiro, 1998; Mazeroll and
                                                                            45
                   M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48




Fig. 3. Mean densities of (a) the entire size range and (b) juveniles of the seven nursery species that had higher densities on reefs adjacent to bays with
seagrass beds and mangroves than at other locations (see Table 3). (b) Also shows densities of juveniles in mangroves and seagrass beds in Spanish
Water Bay (data recalculated from Nagelkerken and van der Velde, 2002), to allow comparison with densities on the reef. Note that the Y-axis of
(b) is on a log10-scale. Error bars indicate SEM. mg bay: mangrove habitat in Spanish Water Bay; sg bay: seagrass habitat in Spanish Water Bay;
Reef sgemg: reefs adjacent to bays with seagrass beds and mangroves; Reef between: reefs between bays with seagrass beds and mangroves; Reef
mud/sand: reefs adjacent to bays dominated by bare sediment; Reef distance: reefs at great distances to bays with seagrass beds and mangroves.



Montgomery, 1998; Zeller, 1998; Chapman and                   reefs, rather than in seagrass or mangrove habitats.
Kramer, 2000). Long-distance dispersal of Haemulon                Although it has been shown, for example, that predation
                                         pressure results in low survival of Haemulidae on reefs
sciurus, Lutjanus analis, Lutjanus apodus, Ocyurus chrys-
urus, and Sphyraena barracuda may have contributed to              (Beets, 1997), some individuals may survive and con-
the presence of small fish populations on reefs located at            tribute to small populations on reefs at some distance
some distance from their main nursery habitats.                 from seagrass and mangrove habitats (Shulman and
  The presence of adults of species that had their highest           Ogden, 1987). In the specific case of Scarus coeruleus,
densities on reefs adjacent to sgemg bays in the other              which showed its highest densities on reefs adjacent to
reef categories may also be explained by the survival of             sgemg bays, local recruitment can play a major role
juveniles that have settled and grown up directly on these            because juvenile densities on the reef were comparable to
46                 M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48




Fig. 4. Mean densities of juveniles of the eight nursery species that did not have higher densities on reefs adjacent to bays with seagrass beds and
mangroves than at other locations. Densities are shown on a log10-scale for the coral reef (this study) and for the mangroves and seagrass beds of
Spanish Water Bay (data recalculated from Nagelkerken and van der Velde, 2002). Species with higher juvenile densities in seagrass beds/mangroves
than on the reef (a) are distinguished from species with similar densities in seagrass beds/mangroves and on the reef (b). Error bars indicate SEM. For
abbreviations see the legend to Fig. 3.



those in seagrass beds. Other studies have also observed            a role, the influence of the presence/absence of nursery
juveniles of S. coeruleus on patch reefs (Overholtzer and           bays on the fish community structure of various reef fish
Motta, 1999). These observations suggest that this spe-            species is greater than these other factors. Firstly, and
cies can also use the coral reef as a nursery.                 most importantly, if other factors were primarily
  One problem with the interpretation of the present             responsible, then non-nursery congeners of the nursery
results is that all reefs in front of bays with seagrass bed          species would also show significantly elevated densities
and mangrove nurseries were located on the southeast-             at reefs in front of nursery bays. This was not the case.
ern part of the coast, whereas all reefs in front of mud/           Secondly, coral cover at 2, 5, and 10 m depth and overall
sand bays and reefs at great distances from bays with             coral cover did not differ significantly between the
mangroves and seagrass beds were located on the north-             southeastern and northwestern reefs (P > 0:213, t-test).
western part of the island. Factors other than absence/            Only at 15 m depth was the coral cover significantly
                                        higher at the latter reefs than at the former (p ¼ 0:047,
presence of bays with mangrove and seagrass beds may
therefore also influence the reef fish communities at these           t-test), but the data indicated that with the exception of
reef categories. It is argued that even if such factors play          one fish species no high positive correlation was present
                                                                   47
               M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48

between coral cover and fish densities. Thirdly, Ocyurus       reef sites. Ontogenetic migrations from sgemg bays to
chrysurus, Lutjanus apodus and Haemulon sciurus which        reefs located much farther away are therefore not likely.
showed the highest difference in density between the           Various studies have demonstrated a close correla-
reefs in front of the bays with nursery habitats and the       tion between habitat complexity and total fish density
other three reef categories, were three of the four         (Luckhurst and Luckhurst, 1978; Bell and Galzin, 1984;
nursery species for which Nagelkerken et al. (2002)         Grigg, 1994). In the present study, however, the relation
indicated that they showed a very high dependence of         between coral cover and fish density was only evident
mangrove/seagrass nurseries at various islands through-       for Scarus coeruleus, suggesting that this species favours
out the Caribbean. Environmental factors such as water        reefs with a high coral cover. For the two other species
temperature, salinity and turbidity do not vary in          which showed a significant relation between density and
a systematic way at the two parts of the island, partly       coral cover, the relation was only significant in one
due to the ocean currents which run straight along the        depth zone, and was negative for Haemulon sciurus,
entire southwestern coast of the island. The island does       whereas for Lutjanus mahogoni the degree of variation
not have any fishing reserves, and fishing takes place         explained by the regression line was very low. Further-
along the entire sheltered southwestern coast. It is         more, the non-nursery congeners of the nursery species
therefore concluded that the presence of nursery bays is       showed different distribution patterns among the reef
in this case the best possible explanation for the elevated     sites than the nursery species. It is therefore likely that in
densities of seven nursery species on reefs in front of       this study coral complexity did not influence the dis-
sgemg bays.                             tribution of the sampled nursery species along the coast.
  Among the eight nursery species that did not occur in        The results of the present study indicate that the
higher densities as mainly adults on reefs adjacent to        distribution of Haemulon sciurus, Lutjanus apodus,
sgemg bays, two groups were distinguished: one in-          Ocyurus chrysurus and Scarus coeruleus on the coral
cluding species with higher juvenile densities in seagrass      reef along the coast of a single island is significantly
beds/mangroves than on the coral reef, and one in-          related to the presence of sgemg bays. Lutjanus analis,
cluding species with similar juvenile densities in seagrass     Lutjanus mahogoni and Sphyraena barracuda showed
beds/mangroves and on the reef. The first group in-          a similar trend but densities at reefs adjacent to sgemg
cludes two species (Chaetodon capistratus and Haemulon        bays were only significantly higher than those at two of
flavolineatum) for which local recruitment is probably        the three reef categories. Six of these seven nursery
the main source of adults, because juveniles were found       species showed much higher juvenile densities in
on the entire reef while no higher total density was         seagrass/mangrove habitats than on the reef, but were
observed on reefs adjacent to sgemg bays. Nagelkerken        nevertheless also found as adults on reef locations at
et al. (2000a) also found juveniles of both species on the      some distance from these nursery habitats, suggesting
reef. Nagelkerken et al. (2001) showed a major nur-         dispersal along the reef. Acanthurus chirurgus, Scarus
sery function of mud/sand bays for Gerres cinereus (see       iserti and Sparisoma chrysopterum showed comparable
Table 1). And since mud/sand bays are present over a         juvenile densities in seagrass/mangrove habitats and reef
large part of the coast, the observations of juveniles of      habitats, and were also found as adults at various reef
this species at the various reef sites at great distance       sites, suggesting that they can complete their entire life
from sgemg bays might be explained by the presence of        cycle on the reef and are not highly dependent on
these bays. Juveniles of Lutjanus griseus and Haemulon        seagrass beds and mangroves.
parra were predominantly observed in sgemg bays
(Table 1) and not on the coral reef. The presence of these
                                   Acknowledgements
species on reefs at some distance to sgemg bays might
therefore be explained by dispersal along the coast.
                                    The management and staff of the Carmabi Founda-
  For the second group, local recruitment is thought to
                                        x
                                   tion Curacao is thanked for the use of their facilities and
be the main source of adults on reef sites other than reefs
                                   for their support. Dr. A. Debrot provided information
adjacent to sgemg bays. Nagelkerken et al. (2002)
                                   and literature. The manuscript benefited by the com-
described both Acanthurus chirurgus and Sparisoma
                                   ments of two referees. This study was financially sup-
chrysopterum as species that do not depend on man-
                                   ported by a grant from the Schure-Beijerinck-Popping
groves or seagrass beds as nurseries. However, the same
                                   Foundation, The Netherlands.
study indicated that Scarus iserti depends heavily on the
presence of seagrass beds and mangroves as nurseries.
The results of the present study suggest that around         References
   x
Curacao, the species is well capable of using the reef as
an alternative nursery and is therefore not restricted to      Adams, A.J., Ebersole, J.P., 2002. Use of back-reef and lagoon
seagrass beds and mangroves as nurseries. Small juve-         habitats by coral reef fishes. Marine Ecology Progress Series 228,
niles (0e5 cm) were also frequently observed on most          213e226.
48                  M. Dorenbosch et al. / Estuarine, Coastal and Shelf Science 60 (2004) 37e48

Austin, H.M., 1971. A survey of the ichtyofauna of the mangroves of      Munro, J.L. (Ed.), 1983. Caribbean coral reef fishery resources.
  western Puerto Rico during December, 1967eAugust, 1968.            ICLARM Studies and Reviews 7. International Center for Living
  Caribbean Journal of Science 11, 27e39.                    Aquatic Resources Management, Manila, pp. 1e276.
Baelde, P., 1990. Differences in the structures of fish assemblages in     Nagelkerken, I., van der Velde, G., Gorissen, M.W., Meijer, G.J.,
  Thalassia testudinum beds in Guadeloupe, French West Indies, and       van‘t Hof, T., den Hartog, C., 2000a. Importance of mangroves,
  their ecological significance. Marine Biology 105, 163e173.          seagrass beds and the shallow coral reef as a nursery for important
Beets, J., 1997. Effects of a predatory fish on the recruitment and         coral reef fishes, using a visual census technique. Estuarine, Coastal
  abundance of Caribbean coral reef fishes. Marine Ecology Progress       and Shelf Science 51, 31e44.
  Series 148, 11e21.                            Nagelkerken, I., Dorenbosch, M., Verberk, W.C.E.P., Cocheret
                                                 `
Bak, R.P.M., 1975. Ecological aspects of the distribution of reef corals in    de la Moriniere, E., van der Velde, G., 2000b. Importance of
  the Netherlands Antilles. Bijdragen tot de Dierkunde 45, 181e190.       shallow-water biotopes of a Caribbean bay for juvenile coral reef
Blaber, S.J.M., 2000. Tropical estuarine fishes. Ecology, exploitation       fishes: patterns in biotope association, community structure and
  and conservation. Fisheries and Aquatic Resources Series 7,          spatial distribution. Marine Ecology Progress Series 202, 175e193.
  Blackwell Science, Oxford, 372 pp.                    Nagelkerken, I., Kleijnen, S., Klop, T., van den Brand, R.A.C.J., Cocheret
                                                `
Bell, J.D., Galzin, R., 1984. Influence of live coral cover on coral-reef     de la Moriniere, E., van der Velde, G., 2001. Dependence of Caribbean
  fish communities. Marine Ecology Progress Series 15, 265e274.         reef fishes on mangroves and seagrass beds as nursery habitats:
Chapman, M.R., Kramer, D.L., 2000. Movements of fishes within and         a comparison of fish faunas between bays with and without mangroves/
  among fringing coral reefs in Barbados. Environmental Biology of       seagrass beds. Marine Ecology Progress Series 214, 225e235.
  Fishes 57, 11e24.                             Nagelkerken, I., Roberts, C.M., van der Velde, G., Dorenbosch, M.,
             `                                                 `
Cocheret de la Moriniere, E., Pollux, B.J.A., Nagelkerken, I., van der      van Riel, M.C., Cocheret de la Moriniere, E., Nienhuis, P.H., 2002.
  Velde, G., 2002. Post-settlement life cycle migration patterns and      How important are mangroves and seagrass beds for coral-reef
  habitat preference of coral reef fish that use seagrass and mangrove      fish? The nursery hypothesis tested on an island scale. Marine
  habitats as nurseries. Estuarine, Coastal and Shelf Science 55,        Ecology Progress Series 244, 299e305.
  309e321.                                 Nagelkerken, I., van der Velde, G., 2002. Do non-estuarine mangroves
       ´
de Buisonje, P.H., Zonneveld, J.I.S., 1960. De kustvormen van           harbour higher densities of juvenile fish than adjacent shallow-
     x                                                          x
  Curacao, Aruba en Bonaire. Natuur Wetenschappelijke Werkgroep         water and coral reef habitats in Curacao (Netherlands Antilles)?
  Nederlandse Antillen, 11: 1e24, Martinus Nijhoff, ‘s Gravenhage.        Marine Ecology Progress Series 245, 191e204.
de Haan, D., Zaneveld, J.S., 1959. Some notes on tides in Annabaai      Overholtzer, K.L., Motta, P.J., 1999. Comparative resource use by
          x
  harbour, Curacao, Netherlands Antilles. Bulletin of Marine Science      juvenile parrotfishes in the Florida Keys. Marine Ecology Progress
  of the Gulf and Caribbean 9, 224e236.                     Series 177, 177e187.
Froese, R., Pauly, D. (Eds.), 2002. FishBase. World Wide Web elect-      Parrish, J.D., 1989. Fish communities of interacting shallow-water
  ronic publication. http://www.fishbase.org, version February 2002.       habitats in tropical oceanic regions. Marine Ecology Progress
Gillanders, B.M., 2002. Connectivity between juvenile and adult fish        Series 58, 143e160.
  populations: do adults remain near their recruitment estuaries?      Pollard, D.A., 1984. A review of ecological studies on seagrass-fish
  Marine Ecology Progress Series 240, 215e223.                 communities, with particular reference to recent studies in
Gillanders, B.M., Kingsford, M.J., 1996. Elements in otoliths may         Australia. Aquatic Botany 18, 3e42.
  elucidate the contribution of estuarine recruitment to sustaining     Polunin, N.V.C., Roberts, C.M., 1993. Greater biomass and value of
  coastal reef populations of a temperate reef fish. Marine Ecology       target coral-reef fishes in two small Caribbean marine reserves.
  Progress Series 141, 13e20.                          Marine Ecology Progress Series 100, 167e176.
Grigg, R.W., 1994. Effects of sewage discharge, fishing pressure and      Robertson, A.I., Blaber, S.J.M., 1992. Plankton, epibenthos and fish
  habitat complexity on coral ecosystems and reef fishes in Hawaii.       communities. In: Robertson, A.I., Alongi, D.M. (Eds.), Tropical
  Marine Ecology Progress Series 103, 25e34.                  mangrove ecosystems. Coastal and Estuarine Studies 41, 173e224.
Hixon, M.A., Beets, J.P., 1993. Predation, prey refuges, and the structure  Shulman, M.J., Ogden, J.C., 1987. What controls tropical reef fish
  of coral-reef fish assemblages. Ecological Monographs 63, 77e101.       populations: recruitment or benthic mortality? An example in the
Kanashiro, K., 1998. Settlement and migration of early stage spangled       Caribbean reef fish Haemulon flavolineatum. Marine Ecology
  emperor, Lethrinus nebulosus (Pisces: Lethrinidae), in the coastal      Progress Series 39, 233e242.
  waters off Okinawa island, Japan. Nippon Suisan Gakkaishi 64,       ter Braak, C.J.F., Smilauer, P., 1998. Reference Manual and User’s
  618e625.                                   Guide to Canoco for Windows: Software for Canonical Commu-
                        `´
Louis, M., Guyard, A., 1982. Contribution a l’etudes des peuplements       nity Ordination (Version 4). Microcomputer Power, New York.
  ichtyologiques dans les mangroves de Guadeloupe (Antilles         Tulevech, S.M., Recksiek, C.W., 1994. Acoustic tracking of adult
     x
  Francaises). Bulletin of Ecology 13, 9e24.                  white grunt, Haemulon plumieri, in Puerto Rico and Florida.
Luckhurst, B.E., Luckhurst, K., 1978. Analysis of the influence of         Fisheries Research 19, 301e319.
  substrate variables on coral reef fish communities. Marine Biology     van den Hoek, C., Colijn, F., Cortel-Breeman, A.M., Wanders,
  49, 317e323.                                 J.B.W., 1972. Algal vegetation-types along the shores of inner bays
                                                      x
Jones, G.P., 1991. Postrecruitment processes in the ecology of coral reef     and lagoons of Curacao, and of the lagoon Lac (Bonaire),
  fish populations: a multifactorial perspective. In: Sale, P.F. (Ed.),     Netherlands Antilles. Verhandelingen van de Koninklijke Neder-
  The Ecology of Fishes on Coral Reefs. Academic Press, New York,        landse Akademie der Wetenschappen, Afdeling Natuurkunde,
  pp. 294e328.                                 tweede reeks 61, 1e72.
Macpherson, E., 1998. Ontogenetic shifts in habitat use and          Wilson, D.T., 2001. Patterns of replenishment of coral-reef fishes in the
  aggregation in juvenile sparid fishes. Journal of Experimental         nearshore waters of the San Blas Archipelago, Caribbean Panama.
  Marine Biology and Ecology 220, 127e150.                   Marine Biology 139, 735e753.
Mazeroll, A.I., Montgomery, W.L., 1998. Daily migrations of a coral      Zeller, D.C., 1998. Spawning aggregations: patterns of the coral trout
  reef fish in the Red Sea (Gulf of Aqaba, Israel): initiation and        Plectropomus leopardus (Serranidae) as determined by ultrasonic
  orientation. Copeia 1998, 893e905.                      telemetry. Marine Ecology Progress Series 162, 253e263.