Marine Geology 208 (2004) 265 – 280
                                                    www.elsevier.com/locate/margeo




  Half a century of dynamic coastal change affecting mangrove
   shorelines of French Guiana. A case study based on remote
        sensing data analyses and field surveys
                    F. Fromard a,*, C. Vega a,1, C. Proisy b
    a
                          ´               ´
      Laboratoire Dynamique de la Biodiversite (LADYBIO), CNRS, Universite Paul Sabatier, 29 rue Jeanne Marvig, BP 4349,
                            31055 Toulouse cedex 4, France
            b
             UMR AMAP, IRD, Route de Montabo, BP 165, 97323 Cayenne cedex, Guyane francaise, France
                                                    ß
                      Received 17 September 2002; accepted 14 April 2004



Abstract

  The mobile mud banks, several kilometres wide and about 30 km long, which form the sedimentary environment of the
coast of the Guianas are a consequence of the huge particulate discharge of the Amazon. These mud banks shift towards the
northwest, influenced by the combined action of accretion and erosion, a process also affected by periodic variability. Because
of this movement, the coastline is unstable and continuously changing. Such changes determine the structure and composition
of mangrove forests, the only type of vegetation adapted to this dynamic environment. The objectives of this study were to
identify coastal changes that took place over the last 50 years, and to relate them to natural processes of turnover and
replenishment of mangrove forests. These objectives have been achieved through a combination of remote sensing techniques
(aerial photographs and SPOT satellite images) and field surveys in the area of the Sinnamary Estuary, French Guiana. Ground
data were collected in representative mangrove forest stands, chosen as a function of their growth stages and their structural
features, from pioneer and young stages to adult, mixed and declining formations. The coastline changes and the mangrove
dynamics over the 1951 – 1999 period are analyzed through production of synthetic digital maps, showing an alternation of net
accretion (1951 – 1966) and erosion periods (1966 – 1991), followed by the present accretion phase. Based on this structural,
functional and historic information, a global scenario of mangrove forest dynamics is proposed, including a model of forest
development, forest gap processes and sedimentological dynamics. The results of this research are discussed within the context
of regional (coastal Amazonian area) and global climate.
D 2004 Elsevier B.V. All rights reserved.

Keywords: mangrove forest; coastal dynamics; Amazonian dispersal system; mangrove development model; French Guiana; remote sensing




                                      1. Introduction
  * Corresponding author. Tel.: +33-5-62-26-99-72; fax: +33-5-
62-26-99-99.                                 The Amazon River discharges into the Atlantic
  E-mail addresses: Francois.Fromard@cict.fr (F. Fromard),
                                      Ocean considerable quantities of sediments originat-
proisy@cayenne.ird.fr (C. Proisy).
  1
   Present address: Institut des Science de l’Environnement,      ing from the Andean mountain range and collected by
     ´    ´   `    ´
Universite du Quebec a Montreal C.P. 8888, succ. Centre-Ville,       the river’s huge catchment area. These materials,
   ´   ´
Montreal, Quebec, Canada H3 C3P8.                     partly carried northwestwards by the marine currents,

0025-3227/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.margeo.2004.04.018
266                F. Fromard et al. / Marine Geology 208 (2004) 265–280

flow along the coastlines of North Brazil, of French      of the accretion processes. More recently, Marchand
Guiana, of Surinam, and even further up to the mouth      et al. (2002) showed that the sediments of the
of the Orinoco River, Venezuela. This vast coastal       Guianese coastline had geochemical characteristics
region under Amazonian influence is thus character-      testifying both to their Amazonian origin and the
ized by specific dynamic conditions; that is, the sedi-    alterations brought about by the mangrove vegeta-
ments form mobile mud banks, which migrate along        tion, which grows there. Allison and Lee (2004),
the coast and generate alternate phases of intense       analyzing the mechanisms of sediment exchange
accretion and spectacular erosion.               between Guianese mud banks and shore-fringing
  The hydrodynamic and sedimentological setting       mangroves, propose a new mechanism for the mi-
of this ‘‘Amazonian Dispersal System’’ is complex       gration of mud bank along the Guianese coastline,
and has been analyzed by various authors. Thus,        suggesting that the mud banks would be disconnect-
Froidefond et al. (1988) defined the typology of the      ed from the mangrove fringe.
mud banks, and evaluated the velocity of their           On these unstable substrates, characterized by
migration along the Guianese coast to be 900 m         variable salinities and subject to the tidal cycle under
per year on average. These authors have estimated       a hot and humid climate, only mangrove vegetation
that the surface area gained from the sea by accretion     can develop. This ecosystem is characterized by very
balanced the erosional losses. Prost (1989, 1997)       low plant biodiversity, with species (mangrove trees)
showed that the dynamics of the Holocene sand         displaying adaptations to these strong ecological
banks of the Guianese coastline are associated with      constraints: specific root systems enabling them to
those of the mud banks, and described the relations      stabilize themselves (stilt roots of Rhizophora) or to
between coastal vegetation and sedimentological        ensure a continuity in the gas exchanges throughout
processes. Allison et al. (1995a,b, 2000), from ra-      the tidal cycles (pneumatophores of Avicennia) (Tom-
diochemical activity profiles of sediments collected      linson, 1986; Ball, 1988). The vivipary of a large
on the Northeast coast of the Amapa (North Brazil),      number of mangrove species, their dispersal over
showed a seasonal and decadal periodicity of sedi-       great distances by flotation, and their rapid growth
ment supplies related to the variations in intensity of    also allow them optimal development in these habitats
the trade winds. The close correlation between the       (Duke et al., 1998). Morphological (salt secretion
characteristics of the trade winds and the coastal       glands) and physiological (osmoregulation process)
sedimentological processes have also been demon-        adaptations enable them to tolerate variable and
strated by Eisma et al. (1991) and by Augustinus        sometimes high salinity levels (Ball, 1988).
(2004). In the framework of the large multidisciplin-       The mangrove forest thus is able to thrive along
ary study ‘‘A Multidisciplinary Amazon Shelf SED-       the French Guianese coast, covering an area of
iment Study’’ (AMASSEDS), various authors have         approximately 70 000 ha. Mangroves colonize recent
investigated the oceanic processes at work around       and present marine deposits between the Holocene
the mouth of the Amazon river (Nittrouer et al.,        sand bar and the ocean, on the outer limit of the
1996), analyzed the deposition and accumulation of       coastal plain (Froidefond et al., 1988; Prost, 1989).
sediment in the same area (Kuehl et al., 1996), and      They can also spread upstream several kilometres
studied fluid-mud distribution and processes on the      inland, following the riverbanks, to the limit of tidal
Amazon continental shelf (Kineke and Sternberg,        influence.
1995; Kineke et al., 1996). In another context, Parra       The Guianese mangrove ecosystem is mainly com-
and Pujos (1998) confirmed, via the analysis of the      posed of four species of mangrove trees (Fromard
constituent clay minerals, that the origin of the       et al., 1998):
marine muds was Amazonian and that the climatic
variations affecting the Amazon Basin could influ-       – Laguncularia racemosa (L) Gaert. (Combretaceae),
ence coastal dynamics. For example, the drought         a fast-growing pioneer heliophilous species, which
episodes, which occurred in Western Amazon during        occupies the sea front and estuarine stages.
the late Holocene, have led to a reduction in the       – Avicennia germinans (L.) Stearn (Avicenniaceae),
solid fluxes reaching the coast, and a slowing down       the predominant species of the Guianese man-
                  F. Fromard et al. / Marine Geology 208 (2004) 265–280               267

 grove swamps. It forms pioneer stands on its own or      (Fromard et al., 1998), primary productivity (Betoulle
 in association with L. racemosa. It also constitutes     et al., 2001), and certain ecological features (Fabre et
 the often monospecific and homogeneous in age         al., 1999). The use of radar remote sensing data
 adult stands, which characterize most of Guiana’s       enabled us to initiate a spatialization of these results
 coastal mangrove (Lescure and Tostain, 1989).         (Mougin et al., 1999; Proisy et al., 2000, 2002).
– Rhizophora racemosa Meyer and Rhizophora
 mangle L. (Rhizophoraceae), two species very
 closely related morphologically, present in the        2. Methods
 inland mangrove and along the riverbanks where
 sea water is sufficiently diluted by constant         2.1. Study area
 freshwater input. They are also found, at the limit
 of tidal influence, in mixed mangrove-swamp           The study area corresponds to the estuary of the
 forest communities.                      Sinnamary River and the adjacent coastal zone
                                (52j50V– 53jW, 5j23V– 5j28V (Fig. 1). This region
                                                N)
  Thus, the coastal dynamics, and the biological and     of Guiana has been well documented by aerial- and
ecological features peculiar to each species, lead to      satellite-photographic data since 1951.We have been
zonation of the vegetation, establishing the typologi-     able to monitor this scene of intensive accretion,
cal and functional diversity of these mangrove         directly in the field, since 1992.
swamps. Distance from the sea or the estuary bank
is the major discriminatory parameter of this organi-      2.2. Field analysis
sation; this parameter integrates both the variations in
salinity and the duration of the substrate’s submersion.      Six characteristic stages were distinguished in this
  The typological analysis of the mangrove, paired      area where the structural diversity of the mangrove is
with the study of coastal changes, must enable its       particularly high, from pioneer, young and adult
dynamics and history to be reconstructed, and possi-      stages, then mature and mixed mangrove formations
bly its evolution to be predicted in a regional Ama-      (transition with brackish swamp forest) and to the
zonian context. Eventually, these local and regional      declining or dead mangrove stands (‘‘mangrove cem-
dynamics will need to be placed in the context of        etery’’). Mangrove stages in French Guiana, and in the
global climatic changes and expected sea-level rise.      Sinnamary area, have been previously analyzed in
This constitutes the aim of our research launched in      detail from ecological and structural points of view
the framework of a joint French – US CNRS/NSF          (Fromard et al., 1998; Fabre et al., 1999).
research project and carried on within the French          In each stage, one or several stands (a total of 15)
‘‘National Coastal Environment Program’’.            have been selected for both their representative nature
  Three successive stages are required to reach this     and their accessibility. Each stand was analyzed by
objective: (i) describe the current mangrove forest in     delimiting of one or several adjacent plots (a total of
terms of structure and functioning, (ii) analyze its      51), organized in transects perpendicular to the coast-
recent history using remote sensing tools, (iii) propose    line or to the riverbanks. The surface area of the plots
scenarios for this evolution taking into account this      varied as a function of tree density: from 3 Â 3 m for
structural, functional and historic information.        sea front pioneer stages to 20 Â 20 m or 30 Â 30 m for
  In this article, we analyze the evolution of the      adult and decaying formations. All individuals were
Guianese coastal line over the past 50 years, using       identified, their density, diameter and height were
available aerial and satellite remote sensing data, and     measured, the basal areas of the plots were calculated
examine the typology and dynamics of mangroves         (sum of the surface areas of the sections of tree trunks,
associated with the Sinnamary River estuary over that      expressed in m2 ha-1, each surface area being calcu-
same period.                          lated using the diameter measurements).
  Previous work based on intensive field studies         From a balanced sampling between the diameter
have enabled us to establish an overall typology of       classes of individuals, specific allometric relationships
the Guianese mangrove and to quantify its phytomass       between tree diameter and biomass had been defined
268                   F. Fromard et al. / Marine Geology 208 (2004) 265–280




Fig. 1. Estuary of the Sinnamary River, French Guiana, based on the 1999 coastline situation. Location of the sampling stands in the mangrove
domain (stands 1 – 15).


beforehand and validated for each of Guiana man-               x: tree diameter, expressed in cm, measured at 1.30
grove species: A. germinans, L. racemosa, Rhizo-               m (dbh) for adults, at half-height for individuals
phora spp. (Fromard et al., 1998). Various forms of             under 2 m tall, and above the uppermost
regression were tested in this previous work, and were            intersection of the prop roots for Rhizophora,
compared to existing literature data (Cintron and
Schaeffer-Novelli, 1984; Saenger and Snedaker,                The coefficients ao and a1 are characteristic of each
1993; Woodroffe, 1985; Mackey, 1993; Golley et              species and compartment (total biomass, leaf biomass,
al., 1975; Komiyama et al., 1988, etc.). More recently,          branches and trunk biomass), their values are given in
Saenger (2003) published a review of these relation-           Table 1.
ships, taking into account our Guianese data.                 The application of these relations to the Sinnamary
  Based on these studies, we chose the following            site enabled the biomass of each studied stage to be
logarithmic equation as giving the best description of          established.
the relationships between biomass and diameter, i.e.:
                                     2.3. Remote sensing data
   y: aoxa1 where
   y: above-ground biomass, in tons of dry organic                      ´
                                       Series of Institut Geographique National (IGN)
   matter per hectare (t ha-1),                    aerial photographs for the years 1951, 1955, 1966,
                    F. Fromard et al. / Marine Geology 208 (2004) 265–280               269

Table 1                                3.1. Structuring of the mangrove forest
Regression coefficients computed for biomass estimations in
mangrove stands, for each species and plant compartment,
according to the equation: y = aoxa1 where y is above-ground      Six mangrove stages were recognized in the study
biomass and x is tree diameter (from Fromard et al., 1998)      area, their structural characteristics and standing bio-
Mangrove species     Plant component    ao     a1    mass are given in Table 2.
Avicennia germinans   Leaves (g)       40.27   1.50
 (1 cm < dbh < 4 cm)  Branches (g)      66.07   1.66   3.1.1. Pioneer mangrove
             Trunk (g)       149.97   2.00     Stands 1– 3 describe this stage established on
             Total (g)       200.44   2.09   the stabilized mud banks along the sea front. L.
Avicennia germinans   Leaves (g)       40.55   1.77   racemosa and A. germinans make up different
 (dbh>4 cm)       Branches (g)      33.11   2.33
                                   substages according to their respective densities.
             Trunk (kg)        0.07   2.59
             Total (kg)        0.14   2.44   We have shown elsewhere (Fromard et al., 1998)
Laguncularia racemosa  Leaves (g)       18.07   1.89   that, when these two species are associated, L.
             Branches (g)      24.49   2.22   racemosa is progressively overtaken in height by
             Trunk (g)        72.95   2.56   A. germinans, which grows faster, and disappears
             Total (g)       102.30   2.53
                                   while the stands continue to evolve. The still-
Rhizophora ssp.     Leaves (g)       27.35   2.00
             Branches (g)      54.83   2.50   standing dead individuals in these stands bear
             Trunk (g)        58.75   2.62   witness to this process (stand 3). These pioneer
             Total (g)       128.23   2.61   stages form very dense (up to 30 000 ha -1) and
                                   homogeneous stands, ranging from 2.50 to 5 m
                                   high, the mean diameter of the individuals being
1976 and 1987 (1:25 000 and 1:30 000) are available         under 3 cm. Their biomass varies from 11.5 t ha-1 on
for the study area. These photos have been digitized at       sites made up of the youngest individuals (low basal
600 dpi, assembled in mosaic, rectified and georefer-        area, limited tree height), to 57 t ha-1 at more advanced
enced with ground control points and exported into          stages.
MapInfo software for analysis.
  They are supplemented by multispectral satellite         3.1.2. Young mangrove
digital images SPOT 3 (1991, 1993, 1997) and SPOT            In these stands (plots 4 –8) the density is lower
4 (1999), geometrically corrected. Their 20-m spatial        (under 10 000 ha-1) with the disappearance of the
resolution corresponds to a 1: 100 000 scale.            weakest L. racemosa and A. germinans. The remain-
  All the data, converted to a common scale, were          ing individuals have increased in diameter (around 4.5
exported into a geographical information system           cm) and in height (5– 6 m), the stands are monospe-
(MapInfo software) in which our field sampling plots         cific and very homogeneous. The biomass calculated
were located.                            for this stage is not very different from that of the
                                   pioneer mangrove; that is, the decrease in the number
                                   of individuals is compensated by the increase in their
3. Results                              diameter. The stand number 8 is characterized by
                                   densities and basal areas of live individuals which
  The results obtained are presented in two forms:         are especially low, and by a large number of dead
                                   trees (approximately 800 per ha). This stand consti-
– A description of the main mangrove stages and           tutes a structural transition between young and adult
 their dynamics within the Sinnamary site,             stages.
– A series of maps synthesizing the ecological
 (evolution of the mangrove stages) and sedimen-          3.1.3. Adult mangrove
 tological (dynamics of the coastline) information.          This stage (stands 9 – 11), covering vast areas, is
                                   the most characteristic of the Guianese mangrove. A.
  Global diagrams of the evolution of the mangroves         germinans constitutes the predominant stratum in
will be presented in the discussion part.              these stands, whose diversity increases with the
270                   F. Fromard et al. / Marine Geology 208 (2004) 265–280

Table 2
Structural characteristics of the study mangrove stages, Sinnamary area
Mangrove       Stand    Plot number     Number of      Tree   Basal    Tree    Stand   Above ground
stage        number    and surface     tree species    density  area     diameter  height  biomass
                                     (haÀ 1)  (m2 haÀ 1)  (cm)    (m)    (t haÀ 1)
Pioneer        1      5 (3 Â 3 m)     2          31,100  12.50     2.40    2.80   35.10
           2      5 (5 Â 5 m)     2          29,000  21.08     2.70    5.00   56.60
           3      3 (5 Â 5 m)     1          8400   4.08     2.33    2.50   11.42
Young         4      5 (5 Â 5 m)     2          9200   21.40     4.30    5.00   61.40
           5      5 (5 Â 5 m)     1          8400   18.04     4.50    5.00   50.20
           6      3 (5 Â 5 m)     1          8000   18.03     4.80    5.50   73.10
           7      3 (5 Â 5 m)     1          5151   8.60     4.30    5.00   32.39
           8      2 (10 Â 10 m)    1          2400   4.00     4.36    6.10   14.58
Adult         9      3 (20 Â 20 m)    4           917  24.60    23.60    20.00   180.00
           10      3 (20 Â 20 m)    6           663  22.50    44.90    22.00   214.40
           11      2 (20 Â 20 m)    2           450  26.86    24.20    18.22   228.84
Mature        12       (20 Â 20 m)    3           162  51.40    67.10    24.80   431.90
Mixed        13      5 (10 Â 10 m)    6          3047   17.81    21.70    19.00   122.20
Cemetery stand    14      3 (30 Â 30 m)    3           267  18.50    28.50    15.00   110.00
           15      3 (20 Â 20 m)    3           825  13.80    31.10    17.00   77.60




formation of a lower stratum of Rhizophora spp.                The fern Acrostichum aureum makes up a lower
(stand 10). The tree density is under 1000 trees per             and dense stratum. This stage with a high basal
ha, their height (18 –22 m) and their biomass (180 –             area (51.40 m2 ha-1), and a very strong aerial
228 t ha-1) is relatively homogeneous. At this stage,             phytomass (432 t ha-1), is not very common in the
natural disturbances such as pathogen attacks, wood              Sinnamary area, nor is it in the mangroves of
boring by insects or very local wind storms, may               Guiana as a whole.
create canopy openings, through the decaying and              – An enrichment with swamp forest species leads to
death of one or more contiguous trees. Erosion by               the formation of mixed stages (stand 13). The
minor tidal channels may also open forest gaps.                upper stratum of these formations is composed of
These small forest gaps are different from large-scale            A. germinans, frequently showing signs of decay
disturbances caused by hurricanes or typhoons that              (dry and broken treetops, reiterations). The con-
are known to destroy vast mangrove stands in some               stant freshwater input (proximity of the riverbank
tropical areas, but do not occur on the Guianese               or of marshy areas) determines the development of
coast. Small gaps have been locally observed in                these stages, which are very frequent in the study
adult and mature mangrove stages in the Sinnamary               area.
area, directly in the field or through aerial photo-            – A trend towards a dead-standing mangrove forest
graphs. An analysis of their structural characteristics            or ‘‘cemetery stand’’ (stand 14). This stage is
and dynamics is presently in progress in French                characteristic of the Guianese coast, in particular
Guiana.                                    near the river mouths. The fast and massive
  From this adult stage, three types of development             sediment input which prevents respiratory
of the mangrove can be observed:                       exchanges at root level (pneumatophores) seems
                                       to be the origin of the decaying processes, leading
– Development into a mature mangrove (stand 12),               to the death of all the trees (Fromard et al., 1998).
 which is clearly differentiated from the other stages            A new colonization phase may appear in these
 by its structural features: A. germinans is the only            formations, with the arrival of propagules brought
 mangrove tree in the stand, represented by very               by the tides. Young individuals of A. germinans
 wide-diameter trees, covered with lianas and                and L. racemosa then grow among the standing
 epiphytes (Phyllodendron spp. and Bromeliaceae).              dead trunks (stand 15).
                 F. Fromard et al. / Marine Geology 208 (2004) 265–280               271

  Based on the structural analysis of the stands, a      pioneer stage could still be distinguished. The internal
typology of the mangrove can thus be set up. The        swamp surface area contracted.
evolution of one stage into another takes place          For 1966, only one type of mangrove can be
through a reduction of the density of these stands,      detected on the aerial photographs, corresponding to
increase in the diameter of the remaining individuals,     an adult stage. The surface area of mangrove had
and possibly recruiting new cohorts to diversify the      increased again (158.8 km2) and the internal swamp
system. A dynamic model of forestry development        areas had been colonized. At that date, the mangroves
can be constructed from these data (cf. Discussion). In    of the Sinnamary sector reached their maximum
Guiana particularly, the coastal sedimentological pro-     (7.4 km of maximum width). The convex aspect
cesses can perturb this evolutionary cycle at any time.    of the coastal line and the homogeneity of the man-
                                grove indicate a stabilization of the sedimentological
3.2. Evolution of the coastline from 1951 to 1999       processes.
                                  For 1976, it is noticeable that the general aspect of
  Fig. 2, obtained by analysis of aerial photographs     the area is very different, indicating that a major
and of SPOT images, shows the evolution of the         sedimentological change occurred between 1966 and
coastline in the Sinnamary region from 1951 to         1976. Erosion affected the entire study area. The
1999. Table 3 indicates the corresponding surface       coastline was marked by many indentations. It had
areas of mangrove. We have made an overall differ-       clearly retreated towards the right bank of the river. A
entiation on these maps of two mangrove classes        vast area of swamp, open onto the sea, was expanding
defined as: (1) ‘‘same since the previous map’’,        again and the residual patches of mangrove forest
mainly corresponding to adult stands (dark purple)       were becoming isolated (southeast of the image). The
and (2) ‘‘new since the previous map’’ corresponding      mangrove surface area was only 76.3 km2.
to pioneer or young stages (light pink) on the sea         In 1987, the erosion process had continued and a
front. Furthermore, textural and colour features of      large mangrove patch was visible near the river
‘‘old’’ and ‘‘new’’ mangrove stands showed enough       mouth. However, although mangroves continued to
differences to be easily distinguished on every image.     regress overall, new stands appeared at the southeast
Mud banks have been drawn (white) only when their       of the area, in contact with old stands, which had
outlines were clearly visible on images, i.e., during     resisted erosion. This local development of mangroves
accretion periods and at low tide (1951, 1955, 1997).     illustrated the formation of new substrates favourable
The arrows indicate the general direction of the        to colonization. Over the study area as a whole, the
sedimentological process, accretion or erosion. Out-      mangrove surface area contracted until it reached
side the mangrove domain, the mainland forest frag-      36.7 km2.
ments appear in dark green and the areas of savannah        In 1991, a heavy cloud cover impaired the imaging
that are temporarily or constantly flooded by fresh-      of the whole Sinnamary mouth. Despite these con-
water are shown in light green.                straints accretion and erosion processes could still be
  In 1951, mangroves were extensive around the        demonstrated: at the level of the estuary, the erosion
Sinnamary estuary, and the coastline was situated well     cycle continued, whereas the development of the
beyond the Holocene sand bar. A pioneer mangrove        coastline was now perceptible in the southeast of the
stand on the sea front, characterized by specific       zone, with new mangrove stands gained in the swamp
texture and colour on the aerial photograph, indicated     area. These changes confirm the arrival of a new mud
the existence of a recent accretion phase. The bare      bank in the area. As a whole, the mangrove surface
mud bank was not very visible on the photo (white       area had regressed once more, to reach its lowest level
area). The surface area of the mangrove is estimated,     over the period studied, with 33.4 km2 and a maxi-
for that date and for the area studied, at 99.2 km2 and    mum width of 2.3 km.
its maximum extension between the sand bar and the         In 1993, the Northwest part of the study area was
sea front is 6.5 km.                      not visible (the dashed line on the map corresponds to
  In 1955, mangroves had expanded to reach 128.9       the boundary of the SPOT image), but an amplifica-
km2. The accretion process continued and a new         tion of the accretion process is detectable throughout
272  F. Fromard et al. / Marine Geology 208 (2004) 265–280
                      F. Fromard et al. / Marine Geology 208 (2004) 265–280                         273

Table 3
Coastal dynamics and mangrove areas evolution from 1951 to 1999, in the Sinnamary area




the entire image. The mud bank had extended to the               – A phase of mangrove erosion and global retreat of
Sinnamary mouth. The swamp was entirely colonized                the coastal line during the 1976 –1991 period, with
by new mangroves. The mangrove surface area in-                 different processes between the estuarine part and
creased to reach 39.8 km2.                            the coastal area, i.e., continuous erosion of the
  In 1997, accretion had increased still more. Man-               estuarine part over the entire period considered (15
groves were clearly expanding to the southeast of the              years), erosion and then beginning of recoloniza-
area, the bare mud bank being particularly visible in              tion for the coastal area.
this image taken at low tide. An area of 52.9 km2 was             – An accretion phase, particularly in the coastal area
occupied by mangroves.                              until 1999.
  Finally, in the most recent image we have been
able to analyze (1999), the process had continued to               Overall, there has been greater stability in the
follow the same course, rapidly increasing the man-              estuarine region compared with the coastal sector.
grove surface area (59.9 km2).                         The structure of the mangrove stands is the result of
  The image-by-image analysis thus enables the                this differential dynamic between the estuarine and
evolution of the coastal line and that of the associated            coastal sectors.
mangroves to be monitored and demonstrates their
dynamics. By comparison, the stability of the Holo-
cene sand bar, colonized by specific vegetation (sandy             4. Discussion
forest), is noticeable. This sand bar marks a clear
limit between the mangroves growing on the unsta-               4.1. Evaluation of the age of mangrove formations
ble marine mud and the mainland vegetation forma-
tions (freshwater flooded savannahs, various forest                The successive maps representing the time evolu-
fragments).                                  tion of the coastline (Fig. 2) also characterize the
  There has been a succession of three phases in the             evolution with time of the surface area and of man-
study area (Table 3):                             grove locations, which in the field results in the
                                        coexistence of vegetation stages of different ages
– An accretion phase, already started in 1951 and               and structures. In order to visualize this situation in
 which continued until 1966, with a 60-km2                  a single image, the following analysis was carried out:
 increase in the mangrove surface area over this               the most recent map, i.e., 1999, was overlain by the
 period,                                   1997 map. The parts common to the two images were

Fig. 2. Maps of coastal changes and mangrove stands dynamics, in the Sinnamary area, interpreted from time series of aerial photographs
(1951 – 1987) and SPOT satellite images (1991 – 1999). Two mangrove classes have been distinguished on maps: «Same since the previous
map» mainly corresponding to adult stands (dark blue) and «New since the previous map» corresponding to pioneer or young stages (light
pink). Mud banks have been drawn only when their outlines were clearly visible on images, i.e., at low tide and during accretion periods. The
arrows indicate the general direction of the sedimentological process, accretion or erosion. Dashed line on the 1991 map indicates presence of
local cloud cover preventing mapping of the mangrove islet on the Sinnamary mouth. On the 1993 map, dashed line indicates the boundary of
the SPOT image used. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
274                    F. Fromard et al. / Marine Geology 208 (2004) 265–280

mapped and dated 1997. The 1993 map was super-               regular. Uncertainty is greater for the 1951 – 1987
imposed on this first mosaic, and the areas common to            period when photographic data were less abundant.
both images were mapped and dated 1993. The same                It is difficult to directly attribute an age to man-
step by step procedure was followed back to the oldest           grove trees using classic dendrochronological meth-
image, i.e., 1951; that is, for each date, only the parts          ods, because these species usually lack datable growth
common to all the previous steps were kept. The final            rings (Gill, 1971; Tomlinson, 1986). When these
map obtained is a mosaic on which the mangrove               growth rings exist (for A. germinans among others),
stands are juxtaposed, differentiated by the date on            our observations show that their irregular growth and
which the substrate was deposited (Fig. 3). This image           structure do not allow them to be counted precisely
constitutes an original map of the successive sedimen-           and their numbers to be correlated with the ages of the
tological states. It can also be converted into a map of          individuals. In certain cases and for certain species of
the ‘‘theoretical ages’’ of the mangrove, considering            mangrove trees (Rhizophora apiculata, Avicennia
the mangrove development is subsequent to the setup             alba, Sonneratia caseolaris), other techniques of
of the substrate. These theoretical ages correspond             morphological analysis can enable only young indi-
rather well with the real ages of the stands for the most          viduals to be aged (Duarte et al., 1999). More recently,
recent dates (from 1991 to 1999), the time step               Menezes et al. (2003) have demonstrated, with den-
between two successive images being small and                drochronology and radiocarbon analyses, that R. man-




Fig. 3. The 1999 map of mangrove area showing ages of different subareas based on imagery back to 1951, or map of «theoretical ages of the
mangrove stands», considering the mangrove development is subsequent to the substrate setup. This theoretical age corresponds relatively well
to the real age of the stands for the most recent dates (from 1991 to 1999), the time step between two successive images being small and regular.
Uncertainty is greater for the 1951 – 1987 period when photographic data were less abundant.
                      F. Fromard et al. / Marine Geology 208 (2004) 265–280                       275

gle in North Brazil formed annual rings, correlated              clear in the southeast of the study area, where
with tree age in certain environmental conditions. In             transitions between 91 –93 and 93– 97 age groups
French Guiana, where A. germinans is the main                 reflect this process, as well as on the left bank of
mangrove species, only an indirect and overall esti-             the Sinnamary River for the older stages (51 –55
mate of the stand ages can therefore be obtained.               and 55– 66). In this configuration, the age of the
  The resulting map demonstrates a majority of old              mangroves decreases overall from inland towards
mangroves (1951 –1966) on the Sinnamary estuary,               the sea front.
whereas the coastal area is young, with formations             – A colonization by ‘‘patch expansion’’. This
mainly dating from 1991. The latter area also presents            colonization mode appears particularly from 1987
a more heterogeneous arrangement, with a composite              onwards (Fig. 2), during the colonization of the
of different aged stages, reflecting a more varied              internal swamp by the mangrove. The pioneer
sedimentological history. On the other hand, the               stages developed from residual mud patches that
structure of the estuarine area is simpler, owing to             escaped erosion.
the greater stability of this sector. The origin of this          – An arc-shaped colonization linked both to the
dynamic differentiation may be related to the Sinna-             existence of residual mud patches and with an
mary River’s sand inputs, which have locally formed              east – west direction of Amazonian sediment
an offshore bar protecting the extreme tip of the right            movement. This movement leads to the formation
riverbank. This narrow bar, colonized by mangrove, is             of sedimentary bars, which are rapidly colonized
not distinguishable on the remote sensing documents.             by the pioneer mangrove.
The interaction of local sand dynamics and of the
regional process of mud bank movements has also                The last two types of structuring, whose occurrence
been described at other localities on the Guianese             is higher in the coastal area, also explain the great
coast (Anthony et al., 2001).                        heterogeneity of this sector apparent on the theoretical
                                      age map. Thus, a precise analysis of the organization
4.2. Mangrove colonization process                     of the mangrove stages in the coastal landscape
                                      enables the recent history and dynamics to be known.
  Besides a description of the stages, the analysis of
the map of theoretical ages of the mangrove stands             4.3. Global evolution scenario
enables different colonization processes to be distin-
guished, which are outlined in Fig. 4:                     The integrated analysis of the structure of man-
                                      groves and of the evolution of the coastline has shown
– A colonization in bands parallel to the coastline,            us that these two processes are closely linked. The
 which establishes a regular zonation of the                internal organization of the stages and their respective
 mangrove. This organization is characteristic of             layout are consequences of the biological character-
 coastal mangrove formations; it is particularly              istics of the species, but also of the sedimentological




Fig. 4. Colonization patterns of mangrove stands in the Sinnamary area. (a) Colonization by regular zonation, in bands parallel to the coastline.
(b) Colonization by expansion of patches, from residual mud patches not taken away by previous erosion period. (c) Colonization by «arc» of
vegetation, in an east – west direction linked to the Amazonian sediment movement.
276                   F. Fromard et al. / Marine Geology 208 (2004) 265–280

history. The conception of a combined development            population to its biomass. The subsequent part of this
model then becomes possible. Fig. 5 illustrates the           evolution may be decay of the mangrove into a
suggested model, made up of three complementary             ‘‘cemetery’’ stage, or a transformation into swamp
parts.                                  forest. In the case of decay, a new arrival of prop-
  Part A: On the basis of the analysis of Guianese           agules can induce a new cycle of mangrove develop-
mangroves, we have proposed several models, based            ment between the standing dead trunks. This forestry
on the relationships between structural parameters            development model is represented in Fig. 5(A). The
(Fromard et al., 1998). The reduction in stand densi-          transformation of the mangrove structure is also
ties by death of the weakest individuals, the increase          accompanied by a progressive evolution in substrates
in diameter and height of the remaining trees, corre-          (Marchand et al., 2002) and an increase in crab
lated to a progressive increase in the standing bio-           population densities (Amouroux, pers. commun.).
mass, are the mechanisms of regulation in these               Part B: Extending this forest development model,
models. We have shown elsewhere (Fromard et al.,             Duke (2001) introduced the forest gap process, which
1998) that one of these models adjusts perfectly to the         is common in many mangrove forests, brought about
‘‘Self-thinning Rule’’, described by various authors           by small-scale climatic or biological disturbances
(Westoby, 1984; Midgley, 2001; Enquist and Niklas,            (Sherman et al., 2000; Pinzon et al., 2003). We have
2001) and recently applied to the dynamics of man-            seen that such natural small gaps could locally affect
grove stands in Brazil by Berger and Hildenbrandt            adult or mature Guianese mangrove stages. Following
(2000) and Berger et al. (2002). This model links,            the decline and death of individual mangrove trees, the
according to a simple logarithmic law, the density of a         opening up of the canopy leads to rapid germination of




Fig. 5. A new combined model of Guianese mangrove dynamics. (A): Forest development model, mainly based on growth and self-thinning
processes. (B): Forest gaps dynamics, brought about by local decaying and death of individual mangrove trees (adapted from Duke, 2001). (C):
Sedimentological dynamics, the major driving force in French Guiana as in the entire coastal area under Amazonian influence.
                 F. Fromard et al. / Marine Geology 208 (2004) 265–280              277

the soil’s seed bank and the heliophilous Avicennia      5. Conclusions and perspectives
propagules can then develop. The structural evolution
continues according to the phases of ‘‘gap filling’’,      A similar analysis over the entire Guianese coast,
then of ‘‘gap closure’’. These processes are well       and on the scale of the coastal region under Amazo-
described by Duke for various mangrove types, in-       nian influence, now needs to be carried out to improve
cluding Australian mangroves where species succes-      our understanding of coupled sedimentary and bio-
sion is more complex than in French Guiana. The        logical processes and to develop models at a regional
mangrove gap process that we are analyzing in differ-     and global level. Such an integrated approach requires
ent areas in French Guiana is outlined in Fig. 5(B).     the joint development of two types of actions, which
  Part C: If gap formations are to be properly        we are planning to start up as part of interregional
integrated into a mangrove evolution model, it is the     cooperation programmes:
sedimentological dynamics, which remains the driv-
ing force behind its evolution in Guiana and in the      – The setting up of a permanent and homogeneous
entire coastal area under Amazonian influence. We        network of observation and analysis plots on a
have seen that the arrival of a mud bank constitutes       regional scale, enabling regular monitoring of the
the first phase of the cycle. Colonization by man-        mangrove and coastal dynamics to take place.
groves takes place as soon as the stabilized bank is      – The increasing use of remote sensing methods and
subjected to tidal emersion –immersion cycles. Man-       tools, in particular of new airborne and satellite
grove tree propagules are brought in large numbers by      captors, in the visible and radar wave fields.
flotation, and their adaptations enable them to grow
rapidly. Our observations have confirmed that a ho-       The implications of this research concern different
mogeneous, dense formation of young mangrove trees      levels and domains:
could have entirely covered a mud bank only a few
months after it was set up. The stand transformation      – At the local level and in the socioeconomic domain:
then takes place as in the forestry development model      the cyclic silting up of the harbours, estuaries and
described above (part A), locally modified by gap        beaches of Guiana are problematic for the country’s
processes (part B). An erosion phase may occur at any      life. Modelling of these processes will not enable
stage during the development of this model. The         local consequences to be avoided, but should help
occurrence of a sedimentological change increases        foresee them, and thus also allow littoral infra-
as the mangroves develop, with an erosion period         structures to be adapted. On the other hand, proven
following an accretion period. This is what we out-       functional relationships between mangrove pro-
lined in Fig. 5(C). When the erosion process reaches a      ductivity and the richness of coastal fish stocks are
given place on the coast, marine alluvial deposits are      particularly important for Guiana, shrimp fishing
remobilized through the action of the swell and the       representing a major part of its income. Predictive
marine currents. Trees are uprooted, then carried away      data on movement of mud banks and mangroves,
as the coastline moves back, and the mangrove          which are feeding areas of many crustacean and
surfaces regress. With the formation and stabilization      fish species, are also expected from the profes-
of a new mud bank, a new cycle of biological           sional bodies concerned, and are of primary
development of the mangrove can start up again with       importance for the regional economy.
the arrival of a new cohort of propagules.           – At a regional level and in the more fundamental
  On the Sinnamary site, we can evaluate the erosion      sphere of the main ecological and climatic
cycle as a period ranging from 15 to 25 years, the        balances, the overall dynamics of the Guianese
uncertainty resulting from the irregularity in the        coastal region are therefore related to the dynamics
remote sensing data available (no data between 1966       of the Amazon River and its catchment area. From
and 1976, for example). The duration of accretion        the approximately 1.2 Â 109 t year-1 of sediment
cycles is at least 10 (1993 until now) to 15 years        discharged into the Atlantic Ocean, about 15 –20%
(1951 – 1966). Studies being carried out on other        are advected along the Amazonian coast (Meade
Guianese sites (Kaw River) give comparable figures.       et al., 1985) and constitute mobile mud banks
278                F. Fromard et al. / Marine Geology 208 (2004) 265–280

 which are locally and temporarily stabilized by          Werth and Avissar, 2002). We don’t know if,
 mangrove vegetation. Different types of variability        and how, these phenomena could interfere in the
 may have an influence on these processes and, as a         long term with Amazonian coastal dynamics.
 result, on Guianese mangrove dynamics:               However they should also be taken into account
 – The northeast trade winds, which blow towards          in any global model of coastal dynamics.
  the Guianas coast, generate and strengthen wave       – At a global level the influence of recurring climatic
  action according to a seasonal cycle, with direct                 ˜       ˜
                                 phenomena El Nino and La Nina events and the
  consequences on sediment influx and mud banks         expected processes of sea-level change should also
  migration. A long-term trade wind cyclicity ap-        interact with the Amazonian dispersal system and
  pears also to be demonstrated along the Amazo-        therefore on the coastal dynamics of this region.
  nian coastline (Eisma et al., 1991; Allison et al.,      The implications of such a study reach well beyond
  2000; Augustinus, 2004). In any case, trade winds       the Guianese context.
  patterns, associated with tides and wave action,
  appear to be the main factors directly controlling
  the present movement of the mud banks and the
                                Acknowledgements
  cyclical alternation of accretion and erosion
  phases along the Guianese coast.
                                 This work is part of a research program supported
 – Annual rainfall in the Amazon basin, above 2000
                                by both NSF (USA) and CNRS (France), in
  mm around the equator with a maximum above
                                collaboration with R. Aller (State University of New
  3000 mm on the Atlantic coast, shows important
                                York, Stony Brook, USA), M. Allison (Tulane
  interannual variability and anomalies, them-
                                University, New Orleans, USA), F. Baltzer (Universite´
  selves directly correlated to variations in sea-
                                Paris Sud, Orsay, France) and E. Lallier-Verges  `
  surface temperatures (Ronchail et al., 2002). It
                                      ´
                                (ISTO, Orleans, France). We wish to thank the
  has also been established that this rainfall
                                ENGREF center in Kourou, the HYDRECO labora-
  variability mainly explains variability in Ama-
                                tory in Petit Saut and the IRD center in Cayenne for
  zonian discharge (Molinier et al., 1996) and,
                                offering facilities during fieldwork. This work was
  thus, could interfere with the sedimentary
                                completed within the Guiana workshop of the PNEC
  processes throughout the Amazon basin. Second,
                                (National Coastal Environment Program).
  and in the very longer term, these phenomena
                                 The mosaic of aerial photographs have been
  could have an influence on the global coastal
                                realized in collaboration with L. Cadamuro (LET
  dynamics as it has been demonstrated for the late
                                Toulouse) and Mike Lee (Texas A&M University,
  Holocene period (Parra and Pujos, 1998).
                                Galveston).We thank Dr. Peter Saenger, Dr. Norman
  Nevertheless, the present mobility of the mud
                                Duke, and an anonymous reviewer for providing
  banks, seems to be mainly controlled by marine
                                very constructive comments on a previous version
  currents strength, trade winds and wave action
                                                 ´ ´
                                of the manuscript. Uta Berger, Frederic Baltzer and
  (Allison and Lee, 2004), and not directly linked
                                Robert Aller also provided useful remarks on this
  to fluctuations in Amazon discharge.
                                manuscript.
 – The evolution of land cover in the Amazon
   catchment area essentially consists of changes
   from primary forest to pasture and subsequently
   to secondary forest (Fearnside, 1996). This could     References
   result in an accentuation of erosional processes,
   and therefore in a downstream increase in the       Allison, M.A., Lee, M.T., 2004. Sediment exchange between Am-
   sediment load of the Amazon River. On the other        azon Mudbanks and shore-fringing mangroves in French Gui-
                                  ana. Mar. Geol. 208, 169 – 190, this issue.
   hand, simulations of the climatic changes result-
                                Allison, M.A., Nittrouer, C.A., Faria Jr., L.E.C. 1995a. Rates and
   ing from Amazon deforestation predicted sub-         mechanisms of shoreface progradation and retreat downdrift of
   stantial decreases in rainfall, evapotranspiration      the Amazon River mouth. Mar. Geol. 125, 373 – 392.
   and runoff (Dickinson and Kennedy, 1992;          Allison, M.A., Nittrouer, C.A., Kineke, G.C., 1995b. Seasonal sedi-
                     F. Fromard et al. / Marine Geology 208 (2004) 265–280                      279

  ment storage on mudflats adjacent to the Amazon River. Mar.       grove ecosystems: new data from French Guiana. Oecologia
  Geol. 125, 303 – 328.                          115, 39 – 53.
Allison, M.A., Lee, M.T., Ogston, A.S., Aller, R.C., 2000. Origin of  Gill, M., 1971. Endogenous control of growth-ring development in
  Amazon mudbanks along the northern coast of South America.       Avicennia. For. Sci. 17, 462 – 465.
  Mar. Geol. 163, 241 – 256.                      Golley, F.B., McGinnis, J.T., Child, G.I., Duever, M.J., 1975. Mi-
Anthony, E., Dolique, F., Guiral, D., Gardel, A., Lefebvre, J.P.,     neral Cycling in a Tropical Moist Forest ecosystem. Univers.
  2001. The influence of migrating mud banks on the sediment       Georgia Press, Athens, GA.
  dynamics of a headland-bound beach in Cayenne, French Gui-      Kineke, G.C., Sternberg, R.W., 1995. Distribution of fluid muds on
          `         ´
  ana. 8e Congre s Franc ais de Se dimentologie—Livre des
              ß                       the Amazon continental shelf. Mar. Geol. 125, 193 – 233.
   ´  ´
  resumes. ASF, Paris 36, 15 – 16.                   Kineke, G.C., Sternberg, R.W., Trowbridge, J.H., Geyer, W.R.,
Augustinus, P.G.E.F., 2004. The influence of the trade winds on      1996. Fluid-mud processes on the Amazon Continental shelf.
  the coastal development of the Guianas: a synthesis. Mar. Geol.     Cont. Shelf Res. 16, 667 – 696.
  208, 145 – 151, this issue.                     Komiyama, A., Ogino, K., Aksornkeo, S., Sabhasri, S., 1988. Root
Ball, M.C., 1988. Ecophysiology of mangroves. Trees 2, 129 – 142.     biomass of a mangrove forest in south Thailand. Estimation by
Berger, U., Hildenbrandt, H., 2000. A new approach to spatially      the trench method and the zonal structure of root biomass. J.
  explicit modelling of forest dynamics: spacing, ageing and       Trop. Ecol. 3, 97 – 108.
  neighbourhood competition of mangrove trees. Ecol. Model.      Kuehl, S.A., Nittrouer, C.A., Allison, M.A., Faria, L.E.C., Dukat,
  132, 287 – 302.                             D.A., Jaeger, J.M., Pacioni, T.D., Figueiredo, A.G., Underkof-
Berger, U., Hildenbrandt, H., Grimm, V., 2002. Towards a standard     fler, E.C., 1996. Sediment deposition, accumulation, and sea-
  for the individual-based modeling of plant populations: self-      bed dynamics in an energetic fine-grained coastal environment.
  thinning and the field-of-neighborhood approach. Nat. Res.       Cont. Shelf Res. 16, 787 – 815.
  Model. 15, 39 – 51.                         Lescure, J.P., Tostain, O., 1989. Les mangroves guyanaises. Bois
                            ´
Betoulle, J.L., Fromard, F., Puig, H., 2001. Caracterisation des      For. Trop. 220, 35 – 42.
         `
  chutes de litiere et des apports au sol en nutriments dans une    Mackey, A.P., 1993. Biomass of the mangrove Avicennia marina
  mangrove de Guyane francaise. Can. J. Bot. 79, 238 – 249.
                ß                      (Forsk.) Vierh. near Brisbane, south-eastern Queensland. Aust.
Cintron, G., Schaeffer-Novelli, S.Y., 1984. Methods for studying      J. Mar. Freshw. Res. 44, 721 – 725.
  mangrove structure. In: Snedaker, S.C., Snedaker, J.G. (Eds.),                  `
                                    Marchand, C., Lallier-Verges, E., Baltzer, F., 2002. The composi-
  The Mangrove Ecosystem: Research Method. UNESCO, Paris,         tion of sedimentary organic matter in relation to the dynamic
  pp. 91 – 113.                              features of a mangrove-fringed coast in French Guiana. Estuar.
Dickinson, R.E., Kennedy, P., 1992. Impacts on regional climate of     Coast. Shelf Sci. 56, 119 – 130.
  Amazon deforestation. Geophys. Res. Lett. 19, 1947 – 1950.      Meade, R.H., Dunne, T., Richey, J.E., Santos, U.M., Salati, E.,
Duarte, C.M., Thampanya, U., Terrados, J., Geertz-Hansen, O.,       1985. Storage and remobilization of suspended sediment in
  Fortes, M.D., 1999. The determination of the age and growth       the lower Amazon River of Brazil. Science 228, 488 – 490.
  of SE Asian mangrove seedlings from internodal counts. Man-     Menezes, A., Berger, U., Worbes, M., 2003. Annual growth rings
  grove Salt Marshes 3, 251 – 257.                    and long-term growth patterns of mangrove trees from Braganca  ß
Duke, N.C., 2001. Gap creation and regenerative processes driving     peninsula, North Brazil. Wetlands Ecol. Manage. 11, 233 – 242.
  diversity and structure of mangrove ecosystems. Wetl. Ecol.     Midgley, J.J., 2001. Do mixed-species mixed-size indigenous for-
  Manag. 9, 257 – 269.                          ests also follow the self-thinning line? Trends Ecol. Evol. 16,
Duke, N.C., Ball, M.C., Ellison, J.C., 1998. Factors influencing      661 – 662.
  biodiversity and distributional gradients in mangroves. Glob.    Molinier, M., Guyot, J.L., De Oliveira, E., Guimaraes, V., 1996. Les
  Ecol. Biogeogr. Lett. 7, 27 – 47.                     ´
                                      regimes hydrologiques de l’Amazone et de ses affluents. Hydro-
Eisma, D., Augustinus, P.G.E.F., Alexander, C., 1991. Recent and                ´               ´
                                      logie tropicale: Geoscience et outil pour le developpement.
  subrecent changes in the dispersal of Amazon mud. Neth. J. Sea     IAHS, Paris, pp. 209 – 222.
  Res. 28, 181 – 192.                         Mougin, E., Proisy, C., Marty, G., Fromard, F., Puig, H., Betoulle,
Enquist, B.J., Niklas, K., 2001. Invariant scaling relations across    J.L., Rudant, J.P., 1999. Multifrequency and multipolarization
  tree-dominated communities. Nature 410, 655 – 660.           radar backscattering from mangrove forests. IEEE Trans. Geo-
Fabre, A., Fromard, F., Trichon, V., 1999. Fractionation of phos-     sci. Remote Sens. 37, 94 – 102.
  phate in sediments of four representative mangrove stages      Nittrouer, C.A., David, J., DeMaster, D.J., 1996. The Amazon shelf
  (French Guiana). Hydrobiologia 392, 13 – 19.              setting: tropical, energetic, and influenced by a large river. Cont.
Fearnside, P.M., 1996. Amazonian deforestation and global warm-      Shelf Res. 16, 553 – 841.
  ing: carbon stocks in vegetation replacing’s Brazil Amazon for-   Parra, M., Pujos, M., 1998. Origin of late Holocene fine-grained
  est. For. Ecol. Manag. 80, 21 – 34.                   sediments on the French Guiana shelf. Cont. Shelf Res. 8,
Froidefond, J.M., Pujos, M., Andre, X., 1988. Migration of mud       1613 – 1629.
  banks and changing coastline in French Guiana. Mar. Geol. 84,    Pinzon, Z.S., Ewel, K.C., Putz, E., 2003. Gap formation and forest
  19 – 30.                                regeneration in a Micronesian mangrove forest. J. Trop. Ecol.
Fromard, F., Puig, H., Mougin, E., Marty, G., Betoulle, J.L., Cada-    19, 143 – 153.
  muro, L., 1998. Structure and above-ground biomass of man-      Proisy, C., Mougin, E., Fromard, F., Karam, M.A., 2000. Interpre-
280                    F. Fromard et al. / Marine Geology 208 (2004) 265–280

  tation of polarimetric signatures of mangrove forest. Remote     Saenger, P., 2003. Mangrove Ecology, Silviculture and Conserva-
  Sens. Environ. 71, 56 – 66.                       tion. Kluwer Academic Publishing, Dordrecht Hardbound.
Proisy, C., Mougin, E., Fromard, F., Karam, M.A., Trichon, V.,      Saenger, P., Snedaker, S.C., 1993. Pantropical trends in mangrove
  2002. On the influence of canopy structure on the polarimetric      above ground biomass and annual litterfall. Oecologia 96,
  radar response from mangrove forest. Int. J. Remote Sens. 23       293 – 299.
  (20), 4197 – 4210.                          Sherman, R.E., Fahey, T.J., Battles, J.J., 2000. Small-scale distur-
Prost, M.T., 1989. Coastal dynamics and chenier sands in French       bances and regeneration dynamics in a neotropical mangrove
  Guiana. Mar. Geol. 90, 259 – 267.                    forest. J. Ecol. 88, 165 – 178.
Prost, M.T., 1997. La mangrove de front de mer en Guyane: ses      Tomlinson, P.B., 1986. The Botany of Mangroves. Cambridge
                       `
  transformations sous l’influence du systeme de dispersion ama-      Univ. Press, Cambridge.
              ´´ ´
  zonien et son suivi par teledetection. In: Kjerfve, B., de Lacerda,  Werth, D., Avissar, R., 2002. The local and global effects of Ama-
  D., Diop, H.S. (Eds.), Mangrove Ecosystems Studies in Latin       zon deforestation. J. Geophys. Res. 107, D20.
  America and Africa. UNESCO, ISME, Paris, pp. 111 – 126.        Westoby, M., 1984. The self-thinning rule. Adv. Ecol. Res. 14,
Ronchail, J., Cochonneau, G., Molinier, M., Guyot, J.L., De Mi-       167 – 225.
  randa Chaves, A.G., Guimaras, V., De Oliveira, E., 2002. Inter-    Woodroffe, C.D., 1985. Studies of a mangrove basin, Tuff Crater,
  annual rainfall variability in the Amazon Basin and sea-surface     New-Zealand: I. Mangrove biomass and production of detritus.
  temperatures in the equatorial pacific and the tropical Atlantic     Estuar. Coast. Shelf Sci. 20, 265 – 280.
  oceans. Int. J. Climatol. 22, 1663 – 1686.