Estuarine, Coastal and Shelf Science 73 (2007) 617e629
                                                               www.elsevier.com/locate/ecss




         Long term changes in Zostera meadows in the Berre lagoon
                (Provence, Mediterranean Sea)
             Guillaume Bernard a,b,*, Charles F. Boudouresque a, Philippe Picon b
       a
         UMR DIMAR, Centre d’Oceanologie de Marseille, Universite de la Mediterranee, Parc Scientifique et Technologique de Luminy,
                      ´                 ´    ´    ´
                               13288 Marseille cedex 09, France
          b                                                    ´
           Groupement d’Interet Public pour la Rehabilitation de l’Etang de Berre, Cours Mirabeau, 13130 Berre l’Etang, France
                   ´ ˆ         ´
                          Received 7 November 2006; accepted 6 March 2007
                               Available online 25 April 2007




Abstract

  The Berre lagoon (Provence, France), one of the largest Mediterranean brackish lagoons (155 km2), was occupied, at the turn of the 20th
century, by extensive Zostera meadows (Zostera marina and probably Zostera noltii; perhaps over 6000 ha). Subsequently, the lagoon was dis-
turbed by urban and industrial pollution and, from 1966, by the diversion of the Durance River. This resulted in a 10e49-fold and 8e31-fold
increase of the freshwater and silt inputs, respectively. By means of digital analysis of aerial photographs for the years 1944, 1992, 1998 and
2004, coupled with ground truth for the last three dates, we mapped the Zostera meadows. The replacement of Z. marina by Z. noltii, the latter
species being already dominant in the 1970s, was completed in 1990. In parallel to this substitution, the Zostera beds underwent a dramatic
decline. Their depth limit, which was (6e9) m in the early 20th century, withdrew to 3.5, 3, 1 and less than 1 m by 1944, the 1970s, 1992
and 1998, respectively. Since 1998, Zostera must be considered as functionally extinct. The total surface area of Zostera meadows was of
the order of 1.5 ha in 2004. In an attempt to alleviate disturbance, the input of freshwater and silt from the Durance River was significantly
reduced from the early 1980s and 1990s respectively. Similarly, from the 1970s to the 1990s, urban and domestic pollution was drastically re-
duced. Despite these steps, Zostera meadows continued to shrink to near extinction. The lagoon has shifted from a system dominated by seagrass
beds to a system with bare silt bottoms, which now occupy most of the lagoon. The reasons could be, in addition to continuing nutrient inputs,
the resuspension of silt, no longer trapped under the seagrass canopy, during wind episodes, which are frequent in the area, and/or the release of
nutrients from the bare silt habitat, which would constitute an indication of a possible hysteresis of the system. However, since 2000, the es-
tablishment of the mussel Mytilus galloprovincialis, a drop in turbidity and a slight, inconspicuous progression of Z. noltii could be the harbinger
of a reverse shift of the system.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Zostera noltii mapping; Zostera marina; brackish lagoon; disturbance; phase shift; France; Provence; Berre lagoon




1. Introduction                                 are usually more or less well identified: eutrophication and
                                        organic pollution through increasing agriculture and urbanisation
  Coastal lagoons and estuaries have, since the early or middle        in river catchments, port facilities, aquaculture, turbidity and
20th century, become among the most disturbed coastal ecosys-          over-sedimentation (Giesen et al., 1990; Valiela et al., 1997;
tems throughout the world. The initial causes for the disturbance        Bowen and Valiela, 2001; Cardoso et al., 2004). In addition, sea-
                                        grasses, which are common dwellers of these habitats, are de-
                                        clining throughout the world (Short and Wyllie-Echeverria,
                                        1996). However, the subsequent dynamics of the ecosystems
                           ´
* Corresponding author. UMR DIMAR, Centre d’Oceanologie de Marseille,
     ´    ´    ´
Universite de la Mediterranee, Parc Scientifique et Technologique de Luminy,
                                        in response to further disturbances and/or to the improvement
13288 Marseille cedex 09, France.                        of water quality remains poorly understood in most habitats.
  E-mail address: guillaume.bernard@univmed.fr (G. Bernard).          The Berre brackish lagoon (Provence, Southern France,

0272-7714/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ecss.2007.03.003
618                G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629

Mediterranean Sea) offers the opportunity of a time series of       study sites were chosen according to the occurrence of present
over one century, at least for some data. Furthermore, in addition    day Zostera stands. The aerial photographs were selected
to the above mentioned disturbances, the diversion of a river       within the archives of the National Geographic Institute
towards the lagoon resulted in a huge freshwater input.          (IGN) and the National Centre for Scientific Research
  In the late 19th and early 20th centuries, the Berre Lagoon,     (CNRS Centre Camille Julian) on the basis of the following
one of the largest Mediterranean coastal lagoons (155 km2),        criteria: (1) photographs must have been taken with no sun
was occupied by very extensive Zostera meadows (Magnolio-         glint, no surface waves and optimum water transparency.
phyta, Plantae), from a few centimetres below the mean water       Water transparency was considered sufficient when bottom de-
level to 6 m depth, and sometimes even deeper (down to 9 m),       tails where perceptible down to 4 m depth, within or outside
close to the maximum depth of the lagoon (Marion, 1887;          the limits of the mapped area (see below); (2) photographs
Gourret, 1907). Depth contours make it possible to estimate        must have been taken at the season of maximum seagrass
that their surface area was then over 6000 ha, a rather conser-      leaf development, i.e. late springeearly summer; and (3)
vative figure. The Zostera species dwelling in the lagoon was       a site (east of the Pointe de Berre) commonly subject to accu-
Zostera marina Linnaeus, according to Gourret (1907). How-        mulation of drift macroalgae (such as Ulva spp.) and/or sea-
ever, this author, after a good description of Z. marina, pointed     grass leaves has been discarded since they may be confused
out that some specimens exhibited narrow and 3-nerved           with in situ seagrass in the image analysis. Four sites with
leaves, which suggests that a second species also occurred at       a suite of aerial photographs fully matching these conditions
that time, Zostera noltii Hornemann (¼Nanozostera noltii         were chosen on the western and eastern shores of the Berre la-
(Hornemann) Tomlinson and Posluzny).                   goon (Fig. 1). A specific campaign was designed for aerial
  In 1925, the 6 m deep channel which linked the Berre lagoon      photograph acquisition in June 2004 (GIPREB-AERIAL) ac-
to the sea was deepened to 9 m (Mars, 1966). Subsequently, ur-      cording to a standardized protocol (altitude, lens, time, angle,
ban development and industrialisation (especially petrochemi-       resolution, contrast; see McKenzie et al., 2001) optimising the
cals) of the lagoon region resulted in a steady increase in        quality of photographs for seagrass identification. The surface
                       `
pollution (Mars, 1949; Schachter, 1954; Febvre, 1968). From        area of the study sites ranges from 48.9 to 283.5 ha (Table 1).
1966, the diversion of the Durance River towards the Saint Cha-      All sites are shallow (less than 4 m depth).
mas hydroelectric power plant then into the Berre lagoon re-
sulted in: (1) a heavy input of freshwater (up to seven times       2.2. Seagrass mapping from aerial photographs
the volume of the lagoon per year); (2) the decline of surface wa-
ter salinity from 24e36 to 1e22 (Riouall, unpublished data;          The aerial photograph scale was 1:22,500 (1944), 1:20,000
Kim, 1985); (3) stratification with low salinity water down to       (1992 and 1998) and 1:8000 (2004). The photographs were re-
5 m and more salty water at depth (under calm conditions);        spectively scanned at 1143, 1016, and 406 dpi in order to get
and (4) eutrophication and unstable ecological conditions         a pixel size of 0.5 m. Colour photos (1998 and 2004) were
                   ´
(Minas, 1974; Stora et al., 1995; Nerini et al., 2000, 2001). In     converted to B/W, in order to use the same mapping method.
the years following the putting into operation of the diversion      ArcGISÒ georeferencing tool was used for rectification of the
of the Durance River to the lagoon, the decline of Zostera ma-      photographs, according to a single reference (IGN BDORTHO
rina and Zostera noltii meadows was reported (Riouall, 1971,       1998, reported accuracy of 1 m). The rectification error,
       ´
1972; Huve et al., 1973). Subsequently, this decline became        expressed as the RMS (root mean square) distance between
more pronounced, and Z. marina disappeared from the lagoon        original and modelled position of control points, varied from
(Pergent-Martini et al., 1995; Bernard et al., 2005).           1.11 to 5.48 m. Identification of seagrass beds was performed
  To date, the only attempt to map the Zostera meadows of the                            ´
                                     through published sources and maps (Huve and Huve, 1954;´
Berre lagoon is a rough sketch (scale 1:100,000) published by       Mars, 1966; Pergent-Martini et al., 1995) and ground truth
Mars (1966). In addition, Pergent-Martini et al. (1995) men-       (observation by snorkelling: 1992, M. Brugeaille unpublished
tioned the presence or absence of Zostera along the shoreline.      data, 1998, 2004). Depth, size and GPS position of each sea-
  In the present study, by means of digital analysis of aerial     grass patch observed were recorded. At the study sites, the po-
photographs for the years 1944, 1992, 1998 and 2004, coupled       sition of the depth lines (down to the 4 m one) was similar
with ground truth for the last three dates, we have mapped the      between 1955 (SHOM maps) and 2004 (GPS position).
Zostera meadows in 4 sectors of the Berre lagoon in an            The photographs were manually analysed and digitalised,
attempt to assess on the basis of factual and quantitative        through visual interpretation of different grey-tones corre-
data the patterns of change over time and to connect them         sponding to seagrasses, with GIS (ArcGisÒ). Contrast stretching
with the changes in the lagoon environment.                was applied when necessary and seagrass was defined as pixel
                                     with grey tones up to a certain threshold value, to obtain the
2. Materials and methods                         probable surface area of seagrass beds. The accuracy of the
                                     mapping process was determined by creating a maximum and
2.1. Study sites and photograph acquisition                minimum estimate of seagrass cover in addition to the normal
                                     mapping procedure as described by Frederiksen et al. (2004).
 The whole shoreline of the Berre lagoon was explored in         In the maximum estimate, we decreased the grey tone threshold
1998 and 2004 (from a small boat and by snorkelling). The         to include even sparse seagrass patches at the risk of including
                   G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629                   619




        Fig. 1. Location of the study sites. Arrow: outfall of freshwater from the Durance River via the Saint Chamas power plant.



other features such as dark sand or macroalgal stands; the min-        nutrients due to either the diversion of the Durance River or to
imum estimate included only the darkest pixel values, represent-       rivers flowing into the lagoon, come from the literature (Minas,
ing the most distinct seagrass areas and therefore might           1974; Arfi, 1989; Kim and Travers, 1997a,b), unpublished data
underestimate seagrass surface area. The seagrass areas of the        (Roma~a et al., Gosse et al.) and from the French Ministry of
                                           n
original mapping results ranged from À19.3 to þ4.7% relative         Environment databases (Banque HYDRO, 2006; Reseau     ´
to the mean of the corresponding min-max interval.

                                       Table 1
2.3. Hydrological data                            Data on the study sites. Surface area only concerns bottoms between the mean
                                       sea level and the 4 m depth line
  The data on inflow of fresh water and silt due to the diversion       Study site        Surface area (ha)     Times series
of the Durance River towards the Berre lagoon, via the Saint         Pointe de l’Arc      57.5           1992e1998e2004
Chamas hydroelectric plant (1966e2004), has been provided           Pointe de Berre     283.5           1944e1992e1998e2004
         ´
by EDF (Electricite de France). Data on the inflow of fresh water       Martigues         48.9           1944e1992e2004
                                       Figuerolles        55.6           1944e1998e2004
due to rivers flowing into the lagoon, and on the inflow of
620                     G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629

Table 2
Long-term changes in Zostera surface area at Pointe de l’Arc, Pointe de Berre, Figuerolles and Martigues. Above, the probable surface area in hectares. [ ] indicates
the minimum and maximum estimates (see text). Below, the remaining surface area, as a percentage of the baseline value (i.e. 100%, in 1944 or 1992). md ¼ miss-
ing data
            Pointe de l’Arc            Pointe de Berre             Martigues                Figuerolles
1944          md                  84.17 [76.01e87.75]           13.59 [12.28e14.17]           22.43 [20.26e23.38]
                               100%                  100%                   100%
1992          6.32 [5.21e6.47]           3.47 [2.86e3.56]            0.24 [0.19e0.24]             md
            100%                 4.1%                  1.7%
1998          0.10 [0.09e0.10]           0.51 [0.47e0.53]            md                    0.00
            1.6%                 0.6%
2004          0.22 [0.18e0.22]           0.81 [0.66e0.82]            0.02 [0.01e0.02]             0.02 [0.02e0.02]
            3.5%                 0.9%                  0.2%                   0.1%



National de Bassin, 2006). Salinity (1994e2004) was measured             was rejected by a ShapiroeWilk test), with post hoc compar-
with a CTD probe YSIÒ, every 50 cm down to 4 m.                   isons using the Dunn method (Zar, 1999).


2.4. Statistics                                   3. Results

  Size-frequency distribution of patches was tested for nor-            3.1. Changes in Zostera distribution
mality using ShapiroeWilk test. Statistical analyses (correla-
tions and KolmogoroveSmirnov) were conducted using                   At all study sites, at least in 1992, 1998 and 2004, years for
STATISTICAÓ. Hydrological data were compared between                 which ground observations were performed, only one species
years using a non-parametric KruskaleWallis test (normality             was present: Zostera noltii. Zostera meadows underwent




                     Fig. 2. Zostera distribution at Pointe de l’Arc study site in 1992, 1998 and 2004.
                  G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629              621

dramatic losses between 1944 and 1992 or 1998, with a reduc-        3.2. Hydrological data
tion of covered areas of between 98 and 100%. This decline
concerned the four sites studied (Table 2). It was less pro-          The flow of fresh water from the tributary rivers into the
nounced along the eastern shore (Pointe de l’Arc and Pointe         Berre lagoon exhibits a high degree of inter-annual variability
de Berre; Figs. 2 and 3) than along the western shore of the        (Fig. 6a). The mean annual input into the Berre lagoon is
Berre lagoon (Martigues and Figuerolles; Figs. 4 and 5).          0.2 Gm3 yrÀ1 due to inflow from the rivers Arc, Cadiere and
                                                                   `
  In 2004 the Zostera meadows located at Pointe de l’Arc and       Touloubre (69%, 19% and 12% respectively).
Pointe de Berre exhibited a slight recovery. The increase is          The inflow of fresh water from the Durance River diversion
significant at the Pointe de Berre site (KolmogoroveSmirnov         began in 1966 (2 Gm3 yrÀ1) (Fig. 6b). It fluctuated between
test, p < 0.05). At the Figuerolles site, where Zostera was ab-       3 and 4 Gm3 yrÀ1, from the late 1960s to early 1990s, with
sent in 1998, isolated patches were observed in 2004. The          a peak in 1977 (6.6 Gm3) and a minimum in 1989 (the hydro-
2004 survey concerned almost all the sites where Zostera noltii       electric plant was out of order for several months). Finally,
is still present, with the exception of a site discarded due to       from the early 1990s to 2004, the mean flow was reduced to
drift macrophyte accumulation (see Section 2) and small           2 Gm3 yrÀ1, in an attempt to reduce its impact on the lagoon
patches located between Pointe de l’Arc and Pointe de Berre,        habitats. For the 1984e2004 period, the Durance fresh water
and between Martigues and Figuerolles, not exceeding a few         input was 10e49-fold that of the tributary rivers, depending
tens of square meters. Considering that the surface area of         on the year. Salinity is homogeneous within the 4-m thick sur-
Z. noltii in the study areas represents 1.07 ha (Table 2), the to-     face layer and has fluctuated, since 1994, from 6 to 27 (Fig. 7).
tal surface area of Zostera meadows of the Berre lagoon can          The silt input from the tributary rivers strongly fluctuated
therefore be considered as less than 1.5 ha, a very conservative      from one year to the next, depending upon rainfall and was
figure.                                   on average 25,000 t yrÀ1 from 1966 to 1998 (data from Imbert
  The structure of the Zostera meadows also showed major         et al., 1999).
changes since 1944 (Table 3). The decline of the covered            Input of silt from the Durance River (Fig. 8) culminated in
area resulted from splitting of the largest stands together         1977 (1.6 Mt). A first decline occurred in late 1970s, after the
with the reduction of the mean size of the patches.             setting up of a settling basin, and a second after 1994, when it




                 Fig. 3. Zostera distribution at Pointe de Berre study site in 1944, 1992 and 2004.
622               G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629




                 Fig. 4. Zostera distribution at Martigues study site in 1944, 1992 and 2004.



was decided to interrupt the diversion of the Durance River        appearance of Zostera since then, possibly misidentifying
during high turbidity events (more than 2 kg mÀ3), which         Z. noltii as ‘‘stunted Z. marina’’. Nevertheless, in the early
had previously resulted in high inputs of silt into the lagoon.      1960s, both species were still in more or less equal abundance
The inflow of silt from the Durance River was on average          (Ledoyer, 1966). After 1966, parallel to the overall dramatic
24, 31, 13, 8 and 2-fold that of the tributary rivers, for the      decline of the Zostera beds, Z. noltii was clearly more abun-
1966e1973, 1974e1980, 1981e1992, 1993e1998 and                                     ´
                                     dant than Z. marina (Riouall, 1971; Huve et al., 1973), the dis-
1999e2004 periods, respectively.                     appearance of the latter being completed in 1990 (Pergent-
                                     Martini et al., 1995; Bernard et al., 2005). Consequently, we
4. Discussion                               cannot infer which of the species was (or were) present in
                                     1944.
  There is no doubt that in the late 19th and early 20th cen-       The ancient depth limit of Zostera meadows was said to be
tury, Zostera marina was widespread and dominant in the          6 m, sometimes even deeper (down to 9 m) (Marion, 1887;
Berre Lagoon, though a second species, Zostera noltii, proba-       Gourret, 1907; Chevallier, 1916). In 1944, our interpretation
bly also occurred (Gourret, 1907). A conservative figure of        of aerial photographs revealed an already shallow limit
6000 ha can be proposed, based upon the 6 m depth contour         (3.5 m depth). In the absence of ground truth, the presence
and the ancient literature. The fate of Z. marina from that        of deeper Zostera beneath the penetration depth of aerial pho-
time to the early 1960s, before the diversion of the Durance       tographs cannot be ruled out. However, the indented shape of
River towards the lagoon, is poorly known. Authors mentioned       the limit, together with visible patches beyond the limit, sug-
only generically the presence of Zostera, without specifying       gests that this is not the result of an artefact due to water trans-
the species name (Chevallier, 1916; Mars, 1966). Of interest       parency. This is consistent with the general trend of an upward
is the abrupt breakdown of Zostera meadows after an excep-        withdrawal of this limit since at least 1938 (Mars, 1949): 3 m
tionally warm summer (1911) and an exceptionally cold win-        in the early 1970s (Riouall, unpublished data), 1 m in 1992,
                  ´
ter (1956) (Chevallier, 1916; Huve, 1960). The replacement of       less than 1 m in 1998 and 2004 (this work).
Z. marina by Z. noltii may have been in progress as early as         Between 1992 and 2004, the decline of the Zostera
1938, as Mars (1949) emphasized the more and more stunted         meadows continued. This withdrawal can be considered as
                  G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629              623




                 Fig. 5. Zostera distribution at Figuerolles study site in 1944, 1998 and 2004.




representative of the whole lagoon, since the present study        salinity, climate warming, competition with other macro-
concerns almost all the sites where Zostera noltii is still pres-     phytes, pollution and turbidity.
ent. It therefore appears that the reduction of fresh water and        As far as salinity is concerned, Zostera noltii is a euryhaline
silt input since early 1990s, the latter being conspicuous,        species which thrives from near freshwater to salinity over 30,
had little effect on the Zostera beds, which became function-       including rapid changes of salinity (Hartog den, 1970; Hem-
ally extinct from 1998.                          minga and Duarte, 2000; Charpentier et al., 2005). In addition,
  The year-to-year variations in the extent of Zostera marina      surface water salinity increased since the early 1990s, while Z.
and Zostera noltii beds may be considerable, due both to nat-       noltii continued its decline.
ural factors and human impact (e.g. Rasmussen, 1977; Ris-           Climate warming can hinder Zostera marina, a species with
mondo et al., 2003; Frederiksen et al., 2004). The best          cold water affinities (Hartog den, 1970). However, this is not
known variation in the extent of the seagrass beds took place       the case for Zostera noltii, whose temperature range is rela-
in the 1930s. It was known as the ‘‘wasting disease’’ and rav-       tively wide (Hartog den, 1970). Climate warming cannot
aged the seagrass beds on both sides of the North Atlantic. The      therefore account for its dramatic decline.
Mediterranean seems to have been unaffected. Its cause re-          The habitat of Zostera has not been occupied by other mac-
mains controversial (Labyrinthula zosterae, a stramenopile         rophytes, such as Potamogeton pectinatus and Ruppia sp.
parasite, or a climatic episode). Since then, more localized los-     Localized and shallow stands of these species occurred up to
ses, often unexplained, have been recorded, e.g. Chesapeake        1995 (Pergent-Martini et al., 1995; Soltan and Francour,
Bay (USA), Helford River (Cornwall, UK), Terenez Bay (Brit-        2000), but they disappeared between 1996 and 1998 (personal
tany, France) (Rasmussen, 1977; Hartog den, 1994, 1996; Har-        observation). So competition with newly established species
tog den et al., 1996).                           occupying the habitat cannot be accepted as an explanation.
  In contrast to these cases of decline followed (or not) by        The concentrations of heavy metals in the Berre sediments (e.g.
recovery, that of the Berre Zostera beds runs over at least        Cd, Hg, Cu, Pb, Zn) are similar to those observed in other
six decades and possibly almost one century. Several hypoth-        Mediterranean coastal lagoons and even lower than values ob-
eses could account for this dramatic decline: the drop in         served in Thau (France) and Venice (Italy) lagoons (A. Accornero,
624                   G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629

Table 3                                    of the Durance River diversion silt inputs, which were up to
Number and size (mean, median, 1st, 3rd quartile and skewness) over time of  31-fold those from the rivers, did not leave the lagoon towards
Zostera patches at the four study sites. md ¼ missing data)
                                        the sea but on the contrary accumulated, mainly in the deepest
Sites        Year  Pointe   Pointe   Figuerolles  Martigues   areas of its northern part: the mean rate of sedimentation was
               de l’Arc  de Berre
                                        0.6e0.8 and 1.8e3.6 cm yrÀ1 before and after the setting up
Number of patches  1944  md     188    16      34      of the diversion, respectively (Imbert et al., 1999). During
           1992  180     449    md      154
           1998   18     114    0       md
                                        the last period (2000e2004), suspended solids in the water
           2004   21     173    18      92      column reverted to the pre-diversion mean values (Table 4).
                                        The explanation could lie in the establishment of dense cover
Mean size (m2)    1944  md     4477    14,018    3999
           1992  351     77    md      79      of the mussel Mytilus galloprovincialis on the shallow (down
           1998   54     49    0       md      to 4 m depth) sediment (GB, personal observations). Mussel
           2004  104     47    10      2       beds have been suggested as a means to help restoration of la-
Median (m2)     1944  md     216    242      912      goon habitats (Katwijk van, 2003).
           1992   32     13    md      16        Light reduction could also be related to eutrophication of
           1998   17      9    0       md      the lagoon (Table 4), which started in the 1920s with the
           2004   18     10    4       1       urbanisation of the lagoon shore and the setting up of petro-
1st quartile (m2)  1944  md      78    131      252      leum refining and chemical plants. Eutrophication results in
           1992   13      6    md      10      micro- and macro-phytoplankton (Ulva sp.) blooms together
           1998   7      5    0       md
                                        with high levels of colonization of Zostera leaves by epiphytes
           2004   6      4    2       1
                                        (Silberstein et al., 1986; De Casabianca et al., 2003): epiphyte
3rd quartile (m2)  1944  md     735    940      1559     biomass can be higher than leaf biomass (GB, personal obser-
           1992  116     35    md      38
           1998   40     26    0       md
                                        vations). The suffocation by an enteromorph-like Ulva of
           2004   39     22    17      2       a mixed meadow of Zostera marina and Zostera noltii (Hay-
                                        ling Island, Hampshire, UK) and its disappearance has been
Skewness       1944  md      9    4       3
           1992   7      15    md      3       observed (Hartog den, 1994). Surprisingly, the nitrogen con-
           1998   3      19    0       md      centration did not clearly decline over time (Table 4), despite
           2004   3      10    3       5       the reduction of the Durance River inflow (the nitrate input to
                                        the Berre lagoon due to the Durance River diversion is propor-
                                        tional to the water inflow; r2 ¼ 0.99) and the improvement of
                                        the tributary river water quality from the 1970s to the 2000s
     `
Universita degli Studi di Napoli, pers. comm.), where extensive                     ˆ   ´    ´
                                        (Agence de l’Eau Rhone Mediterranee Corse, 2006) with the
stands of Zostera occur.                            setting up of sewage treatment plants; the percentage of urban
  The present day extent of the Zostera noltii meadows, re-         sewage undergoing treatment was 10%, 18%, 40% and 95% in
stricted to very shallow waters, suggests light as the limiting        1970, 1980, 1990 and 2000, respectively, while the population
factor (Valiela et al., 1997; Vermaat et al., 2000; Peralta          of the catchment area increased less than two fold (INSEE,
et al., 2002; Brun et al., 2003; Charpentier et al., 2005).          2006). In addition, between 2000 and 2004, all the sewage
This is consistent with the shrinking of the euphotic zone, cal-        previously flowing directly into the lagoon has been diverted
culated from Secchi disk data by means of the Poole and At-          towards sewage treatment plants. Overall, the nitrogen input
kins relation (1929): mean depth 12.2 m, 10.9 m, 5.1 m and           to the Berre Lagoon was 4665 t, 2514 t, 2021 t and 1338 t in
3.5 m in 1912, 1965, 1966e1969 and 1978e1980, respec-             1977, 1983e84, 2000 and 2004, respectively (Kim and Tra-
tively (Chevallier, 1916; Minas, unpublished data; Kim,            vers, 1997a; Romana et al., unpublished data; Banque HY-
unpublished data) and the amount of suspended solids in the                   ´
                                        DRO, 2006; Reseau National de Bassin, 2006). The release
surface water (Table 4). Zostera marina is also very sensitive         of nutrients trapped within the lagoon sediments, as observed
to light reduction, via the turbidity; shoots die after 3 weeks of       in Orbetello lagoon, Italy (Lardicci et al., 2001) and in the
light limitation (Giesen et al., 1990; Cabello-Pasini et al.,         Greifswalder Bodden, a Baltic estuary (Munkes, 2005), could
2002). Since silt input overwhelmingly decreased, light reduc-         account for the concentration of nutrients which remains
tion could be due to either the silt input being still too high or       higher than that recorded before the diversion of the Durance
to sediment resuspension. Once the major part of seagrass           River. During wind episodes, the vertical mixing of the water
beds, prone to trapping sediment and to hindering its resuspen-        column provides a nutrient input in the photic zone inducing
sion (see Gacia and Duarte, 2001; Charpentier et al., 2005)          an extremely intense bloom of phytoplankton: up to
have disappeared, the wind easily resuspends them in shallow          50 mg LÀ1 chl a (Minas et al., 1976). Be that as it may, present
waters. It is worth noting that Provence is a very windy region,        day concentrations of nutrients are similar to those recorded in
with a wind called the Mistral blowing southwards on average          comparable Mediterranean lagoons, e.g. Venice lagoon, Italy
142 days per year, up to 6 days running, 25e100 km hÀ1, in           (Sfriso and Marcomini, 1997) and Thau lagoon, France (Lau-
                      `
addition to frequent easterly winds (Febvre, 1968; Nerini   ´        gier et al., 1999). It is worth noting that both Venice and Thau
et al., 2000, 2001). So the decline of Zostera could be            lagoons, despite high levels of organic and nutrient load, still
a self-maintained process. It must be emphasized that most           harbour extensive Zostera meadows (Laugier et al., 1999;
                        G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629                     625

                                        0.3
                                           (a)
                                       0.25


                                        0.2




                                    Gm3
                                       0.15


                                        0.1


                                       0.05


                                        0
                                           1984  1986  1988  1990  1992  1994  1996  1998  2000  2002  2004

      7
         (b)
      6

      5

      4
   Gm3




      3

      2

      1

      0
        1966  1968  1970    1972  1974  1976  1978  1980  1982   1984  1986  1988  1990  1992  1994  1996  1998  2000  2002

                   3  À1
Fig. 6. Flow of freshwater (in Gm a ) from: (a) the rivers flowing into the Berre Lagoon, from 1984 to 2004; and (b) from the Durance River via the diversion of
the Saint Chamas hydroelectric plant, from 1966 to 2004.


Rismondo et al., 2003). According to Bowen and Valiela                     The reduction of shoot density in response to decreased
(2001) and Hauxwell et al. (2003), substantial Z. marina loss                light availability is also a well-known response of seagrasses
occurs at loads of w30 kg N haÀ1 yrÀ1, and total disappear-                 to reduce self-shading and therefore to enhance light harvest-
ance at loads !60 kg N haÀ1 yrÀ1; in the Berre Lagoon, the ni-               ing efficiency (Hemminga and Duarte, 2000). Furthermore,
trogen load declined from 301 to 86 kg N haÀ1 yrÀ1, between                 a high ratio of above-ground/below-ground biomass would
1977 and 2004, but still lies above the threshold of Z. marina               be favoured at low-light conditions (Hemminga, 1998). Low
disappearance, which accounts for the lack of recolonization.                shoot-density and high above-ground/below-ground ratio
Several newly established patches of Z. marina, which were                 observed for the Zostera noltii beds in the Berre Lagoon
observed in 2001 in the southern part of the lagoon (Bernard                (GB, personal observations), compared to other Mediterranean
et al., 2005), eventually disappeared. Unfortunately, no data                                    ´
                                              lagoons (Laugier et al., 1999; Menendez et al., 2002; Brun
on nitrogen sensitivity are available for Z. noltii, but its persis-            et al., 2003), support the hypothesis of light limitation.
tence during the period of highest nitrogen load suggests a far                 The present day surviving Zostera noltii stands in the Berre
higher threshold.                                      lagoon mostly consist of small patches, with a skewed patch
  Whatever the reason for light reduction (turbidity and/or                size distribution (Table 3) which is consistent with the distri-
eutrophication), Zostera noltii may prove to be more sensitive               bution pattern usually reported for other species or populations
than other seagrasses. The length of time a seagrass species                (Duarte and Sand-Jensen, 1990; Olesen and Sand-Jensen,
can survive below its minimum light requirement is related                 1994; Vidondo et al., 1997; Ramage and Schiel, 1999). Skew-
to its ability to store carbohydrates, especially in the rhizomes              ness toward low values is indicative of fast patch formation
(Alcoverro et al., 1999; Cabello-Pasini et al., 2002). The stor-              (mostly through seedlings) and high mortality rates observed
age capacity and the clonal integration (sensu Hartnett and                 in seagrass populations depending largely on sexual reproduc-
Bazzaz, 1983) is largely seagrass size-dependent (Hemminga                 tion (Duarte and Sand-Jensen, 1990). Such a high patch mor-
and Duarte, 2000). Small species like Z. noltii have presum-                tality rate is consistent with the poor environmental conditions
ably a lower capacity than those with thick and long-lived rhi-               in the Berre Lagoon.
zomes, conferring a very limited tolerance to light deprivation                 Patch mortality is size-dependent. As patch growth pro-
         `
episodes (Marba and Duarte, 1998; Peralta et al., 2002).                  ceeds, mortality rate decreases and heterogeneity (i.e. within
626                            G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629

             30


             25


             20


             15


             10


             5


             0
                8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 12
                  1994            1995            1996               1997             1998                1999

             30

             25

             20

             15

             10

             5

             0
                1  2   3  4  5  6  7   8  9  10  12  1  2  3  4  5  6   7  8   9  10  11  12  1  2  3   4  5   6  7   8  9   10  11  12
                             2002                        2003                             2004

             Fig. 7. Mean salinity of the surface layer down to 4 m depth from 1994 through 1999 and 2002 to 2004 in the Berre Lagoon.


patch variability) increases (Duarte and Sand-Jensen, 1990).                           other than the mutually sheltering structure phenomenon can
Several studies support the notion of a minimum patch size                            operate.
above which the probability of patch mortality decreases                              The present day near extinction of Zostera in the Berre
(Duarte and Sand-Jensen, 1990; Olesen and Sand-Jensen,                              lagoon probably results from several causes, operating over
1994) due to enhanced anchoring, mutual physical protection                           decades in synergy or successively, namely, pollution (includ-
and physiological integration (‘‘mutually sheltering struc-                           ing nutrients), low salinity and turbidity. There is no doubt that
ture’’) (Thayer et al., 1984). For Zostera novazelandica Setch-                         the decline of the Zostera beds began before the diversion of
ell, this minimum patch size is 0.4 m2 (Ramage and Schiel,                            the Durance River towards the lagoon. However, the inrush
1999). Our results do not provide an adequate basis for sug-                           of huge amounts of water and sediment was obviously the rea-
gesting a minimum patch size for Zostera noltii, though                             son for the dramatic withdrawal of their lower limit and their
many patches disappeared from one map to the next, as factors                          eventual near extirpation. Overall, up to 2000, the lagoon


         1800

         1600

         1400

         1200
   t x 1000




         1000

         800

         600

         400

         200

          0
            1966  1968    1970  1972   1974   1976   1978  1980  1982    1984    1986  1988    1990    1992    1994    1996    1998    2000    2002  2004

                   3
Fig. 8. Flow of silt (in 10 metric tons) from the Durance River via the diversion of Saint Chamas hydroelectric plant, into the Berre lagoon, from 1966 to 2004.
                                                                                                                     G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629              627




Mean (year round) content in NO3, PO4, suspended solids and chlorophyll a of the surface water (less than 1 m depth) of the Berre lagoon, calculated from published and unpublished data. md ¼ missing data.




                                                                                                        (unpubl. data)
                                                                                                                                        shifted from a system dominated by benthic primary producers




                                                                                                        P. Raimbault




                                                                                                                       9.9 (10.4)cd




                                                                                                                               12.8 (8.7)b
                                                                                                                                        (seagrasses) to a system with bare silt bottoms, no longer trap-




                                                                                                                       0.3 (0.2)b
                                                                                                                       7.7 (2.7)a
                                                                                                        2000e04
                                                                                                                                        ped under the seagrass canopy and therefore prone to resus-
( ) ¼ SD, where available. Statistical analysis by KruskaleWallis with post hoc comparisons using a Dunn method. Periods with different letters below the SD were significantly different ( p < 0.01)                                      pension, dominated by plankton primary producers. A




                                                                                                                   10
                                                                                                                   12
                                                                                                                                        similar shift has been described in a shallow lake in Denmark
                                                                                                                                        (McGowan et al., 2005) and in a Baltic Sea estuary (Munkes,
                                                                                                        (unpubl. data)
                                                                                                        P. Raimbault
                                                                                                                                        2005). The threshold level of the forcing variables allowing




                                                                                                                       16.3 (12.3)d

                                                                                                                               43.2 (46.1)c
                                                                                                                       6.0 (8.8)b
                                                                                                                       0.2 (0.3)d
                                                                                                        1994e99



                                                                                                                                        a natural shift back from the apparently ‘‘stable’’ bare silt hab-
                                                                                                                   10                    itats to the previous ‘‘stable’’ Zostera state remains unknown
                                                                                                                   12                    (see Knowlton, 2004; Schroder et al., 2005). Could the slight
                                                                                                                                                       ¨
                                                                                                                                        and inconspicuous progression of Zostera noltii since 2000,
                                                                                                                                        parallel to mussel development and turbidity reduction, be
                                                                                                        (unpubl. data)




                                                                                                                       34.2 (39.0)c           considered as the harbinger of a new shift towards a previous
                                                                                                        and R. Arfi
                                                                                                        M. Minas
                                                                                                        1984e85




                                                                                                                                        state? Or is it just a casual episode in the context of a phase
                                                                                                                                        which could be long-lasting, due to a possible hysteresis of
                                                                                                                       md
                                                                                                                       md


                                                                                                                               md
                                                                                                                   12
                                                                                                                    2




                                                                                                                                        the system in relation with silt resuspension (beyond the pres-
                                                                                                                                        ent-day interlude) or release of nutrient trapped within the sed-
                                                                                                        (unpu-bl. data)




                                                                                                                                        iments, or both?
                                                                                                                       10.7 (8.9)d
                                                                                                                       0.6 (0.5)c
                                                                                                        1978e80




                                                                                                                                        5. Conclusion
                                                                                                        RNO*




                                                                                                                       md

                                                                                                                               md
                                                                                                                   12
                                                                                                                    2




                                                                                                                                          The decline of the extensive Zostera meadows which occu-
                                                                                                                                        pied a large part (possibly over 6000 ha) of the Berre Lagoon
                                                                                                        Kim (unpubl. data),




                                                                                                                                        in the early 20th century possibly began more than 60 years
                                                                                                        Kim and Travers




                                                                                                                                        ago. It has been attributed to pollution and, from 1966,
                                                                                                                                        when the Saint Chamas power plant went into service, to the
                                                                                                                               12.6 (7.2)b
                                                                                                                       8.3 (4.1)a
                                                                                                        (1997a,b)
                                                                                                        1977e78




                                                                                                                                        diversion of the Durance River, which resulted in a heavy
                                                                                                                                        input of freshwater, nitrogen and silt into the Berre Lagoon.
                                                                                                                       17.6
                                                                                                                        0.5
                                                                                                                   12
                                                                                                                    5




                                                                                                                                          Subsequently, a significant reduction of silt (from the late
                                                                                                                                        1970s) and freshwater (from the early 1990s) inputs occurred,
                                                                                                                                        in an attempt to reduce their impact on the lagoon habitats.
                                                                                                        Minas (unpubl.




                                                                                                                                        Concomitantly, urban and industrial sewage outputs were dras-
                                                                                                        data; 1974)




                                                                                                                       9.5 (16.4)b

                                                                                                                               11.1 (6.9)b
                                                                                                                       6.6 (8.2)bc
                                                                                                                       0.3 (0.2)b
                                                                                                        1966e69




                                                                                                                                        tically reduced, though the nitrogen concentration of the body
                                                                                                                                        water did not conspicuously change. As far as the Zostera
                                                                                                                   11
                                                                                                                    5




                                                                                                                                        meadows are concerned, their decline has continued steadily,
                                                                                                                                        to near extirpation from 1998 onward (less than 1.5 ha over-
                                                                                                                                        all), despite a very slight recovery in 2004.
                                                                                                        Blanc et al. (1967),




                                                                                                                                          The present day localization of Zostera noltii, restricted to
                                                                                                        Minas (unpubl.




                                                                                                                                        very shallow waters, suggests that light could be the limiting
                                                                                                        data; 1974)




                                                                                                                       0.6 (0.2)a
                                                                                                                       8.5 (3.4)a

                                                                                                                               5.2 (2.1)a




                                                                                                                                        factor, either due to silt resuspension or eutrophication.
                                                                                                                                          Our results suggest that freshwater, silt and nutrient inputs
                                                                                                        1965




                                                                                                                        1.8
                                                                                                                   12
                                                                                                                    5




                                                                                                                                        were the forcing variables responsible for the phase shift from
                                                                                                                                        seagrass meadows to bare silt habitats. However they do not
                                                                                                                                        provide a basis for forecasting whether we are on the brink
                                                                                                        Schachter (1961)




                                                                                                                                        of a reverse shift or in the context of a long-lasting alternative
                                                                                                        Nisbet and




                                                                                                                                        ‘‘stable’’ state.
                                                                                                                       1.8 (3.5)a
                                                                                                                       0.9 (0.9)a
                                                                                                        1955e56




                                                                                                                                        Acknowledgment
                                                                                                                       md

                                                                                                                               md
                                                                                                                   16
                                                                                                                   10




                                                                                                                                          The authors are indebted to Patrick Bonhomme and Jean-
                                                                                                                   Number of sampled sites




                                                                                                                                         ´
                                                                                                                                        Remy Bravo (GIS Posidonie) for field assistance, to Michael
                                                                                                                   Chlorophyll a (mg LÀ1)




                                                                                                                                        Paul for improving the English text, to EDF and the ‘‘Mission
                                                                                                                   Number of sampled




                                                                                                                   Suspended solids




                                                                                                                                                ˆ    ´
                                                                                                                                        pour la reconquete de l’etang de Berre’’ for data on freshwater
                                                                                                                   NO3 (mmol LÀ1)
                                                                                                                   PO4 (mmol LÀ1)
                                                                                                                    months/year




                                                                                                                                        input to the Berre Lagoon, to Pierre Boissery (Agence de l’Eau
                                                                                                        Data source




                                                                                                                    (mg LÀ1)




                                                                                                                                          ˆ   ´    ´
                                                                                                                                        Rhone Mediterranee Corse) for urban sewage data, to Robert
Table 4



                                                                                                        Period




                                                                                                                                        Arfi, M. Brugeaille, Philippe Gosse, Patrick Raimbault and
                                                                                                                                        Alexandre Roma~a for unpublished data and to Michele
                                                                                                                                                  n                    `
628                     G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629

Perret-Boudouresque and Francoise Cubizolles for biblio-
                 ¸                         Hartnett, D.C., Bazzaz, F.A., 1983. Physiological integration among intraclo-
graphical assistance. The authors also acknowledge the help               nal ramets in Solidago canadensis. Ecology 64, 779e788.
                                           Hartog den, C., 1970. The Seagrasses of the World. North Holland Publ. Co.,
of 3 anonymous referees and the editor for their constructive              Amsterdam, 275 pp.
suggestions.                                     Hartog den, C., 1994. Suffocation of a littoral Zostera bed by Enteromorpha
  This study is a part of a more extensive monitoring program             radiata. Aquatic Botany 47, 21e28.
of the Berre and Va€ lagoons operated by GIS Posidonie
           ıne                              Hartog den, C., 1996. Sudden declines of seagrass beds: ‘‘wasting disease’’ and
(Parc Scientifique et Technologique de Luminy, Marseille                 other disasters. In: Kuo, J., Phillips, R.C., Walker, D.I., Kirkman, H. (Eds.),
                                            Seagrass Biology. Proceedings of an International Workshop, Rottnest
France) and funded by GIPREB (Berre l’Etang, France).                  Island. Univ. of Western Australia Publ., Australia, pp. 307e314.
                                           Hartog den, C., Vergeer, L.H.T., Rismondo, A.F., 1996. Occurrence of Laby-
References                                       rinthula zosterae in Zostera marina from Venice Lagoon. Botanica Marina
                                            39, 23e26.
           ˆ   ´    ´
Agence de l’Eau Rhone Mediterranee Corse, 2006. http://rdb.eaurmc.fr.                  ´
                                           Hauxwell, J., Cebrian, J., Valiela, I., 2003. Eelgrass Zostera marina loss in tem-
Alcoverro, T., Zimmerman, R.C., Kohrs, D.G., Alberte, R.S., 1999. Resource       perate estuaries: relationship to land-derived nitrogen loads and effect of
  allocation and sucrose mobilization in light-limited eelgrass Zostera ma-      light limitation imposed by algae. Marine Ecology Progress Series 247,
  rina. Marine Ecology Progress Series 187, 121e131.                  59e73.
Arfi, R., 1989. Annual cycles and budget of nutrients in Berre Lagoon (Med-      Hemminga, M.A., 1998. The root/rhizome system of seagrasses: an asset and
  iterranean, France). Internationale Revue der Gesamten Hydrobiologie 74,       a burden. Journal of Sea Research 39, 183e196.
  29e49.                                      Hemminga, M.A., Duarte, C.M., 2000. Seagrass Ecology. Cambridge Univer-
Banque HYDRO, 2006. http://www.hydro.eaufrance.fr.                   sity Press, Cambridge, 298 pp.
Bernard, G., Bonhomme, P., Boudouresque, C.F., 2005. Recovery of the sea-        ´                   ´
                                           Huve, H., 1960. Sur l’envahissement recent d’une portion de l’Etang de Berre
  grass Zostera marina in a disturbed Mediterranean lagoon (Etang de Berre,                      ´    ´
                                            (Etang de Va€ par une Ceramiacee du genre Spermothamnion. Rapport
                                                   ıne)
          ˆ
  Bouches-du-Rhone, Southern France). Hydrobiologia 539, 157e161.                `         ´
                                            et Proces-verbaux des Reunions de la Commission Internationale pour
Blanc, F., Coste, B., Minas, H.J., Szekielda, K.H., 1967. Distribution des prin-                            ´   ´
                                            l’Exploration Scientifique de la Mer Mediterranee 15, 141e145.
  cipaux facteurs hydrobiologiques dans un milieu de forte production or-        ´     ´                     ˆ
                                           Huve, P., Huve, H., 1954. Zonation superficielle des cotes rocheuses de l’Etang
        ´
  ganique: l’etang de Berre. Marine Biology 1, 43e55.                 de Berre et comparaison avec celles du Golfe de Marseille (de Carry a     `
Bowen, J.L., Valiela, I., 2001. The ecological effects of urbanization of coastal    Sausset). Vie et Milieu 5, 330e344.
  watersheds: historical increases in nitrogen loads and eutrophication of       ´
                                           Huve, H., Kiener, A., Riouall, R., 1973. Modifications de la flore et des pop-
  Waquoit Bay estuaries. Canadian Journal of Fisheries and Aquatic                            ´
                                            ulations ichtyologiques des etangs de Berre et de Va€ (Bouches-du-
                                                                            ıne
  Sciences 58, 1489e1500.                                  ˆ                           ´´
                                            Rhone) en fonction des conditions hydrologiques creees par le deverse- ´
       ´     ´       ´
Brun, F.G., Perez-Llorens, J.L., Hernandez, I., Vergara, J.J., 2003. Patch distri-                         ´
                                            ment de la Durance. Bulletin du Museum d’Histoire Naturelle de Marseille
  bution and within-patch dynamics of the seagrass Zostera noltii Hornem. in      33, 123e134. þ Tables IeIII.
                ´
  Los Toru~os salt-marsh, Cadiz Bay, natural park, Spain. Botanica Marina
       n                                   Imbert, G., Kerambrun, P., Degiovanni, C., 1999. Hydrodynamics and sedi-
  46, 513e524.                                     mentation linked to anthropogenic inputs in a Mediterranean littoral basin.
Cabello-Pasini, A., Lara-Turrent, C., Zimmerman, R.C., 2002. Effect of storms                     ´
                                            Comptes rendus de l’Academie des Sciences 329, 205e209.
  on photosynthesis, carbohydrate content and survival of eelgrass popula-     INSEE, 2006. http://www.insee.fr.
  tions from a coastal lagoon and the adjacent open ocean. Aquatic Botany      Katwijk van, M.M., 2003. Reintroduction of Eelgrass (Zostera marina L.) in the
  74, 149e164.                                     Dutch Wadden Sea: A Research Overview and Management Vision. Pro-
Cardoso, P.G., Pardal, M.A., Lillebø, A.I., Ferreira, S.M., Raffaelli, D.,       ceedings of the 10th International Scientific Wadden Sea Symposium, Gro-
  Marques, J.C., 2004. Dynamic changes in seagrass assemblages under eu-        ningen, 2000. Ministry of Agriculture, Nature Management and Fisheries/
  trophication and implications for recovery. Journal of Experimental Marine      University of Groningen, Dept. of Marine Biology Publ., NL, pp. 173e196.
  Biology and Ecology 302, 233e248.                                       ´        ´       ´
                                           Kim, K.T., 1985. La salinite et la densite des eaux des etangs de Berre et de
Charpentier, A., Grillas, P., Lescuyer, F., Coulet, E., Auby, I., 2005. Spatio-          ´     ´
                                            Va€ (Mediterranee nord-occidentale). Relations avec les affluents et le
                                               ıne
  temporal dynamics of a Zostera noltii dominated community over a period       milieu marin voisin. Journal of Natural Sciences 5, 221e246.
  of fluctuating salinity in a shallow lagoon, Southern France. Estuarine,      Kim, K.T., Travers, M., 1997a. Les nutriments de l’Etang de Berre et des mi-
  Coastal and Shelf Sciences 64, 307e315.                                                 ˆ         ´
                                            lieux aquatiques contigus (eaux douces, saumatres et marines; Mediterra-
             ´                    ´
Chevallier, A., 1916. L’etang de Berre. Annales de l’Institut Oceanographique       ´
                                            nee NW). 2 Les nitrates. Marine Nature 5, 35e48.
  7, 1e91.                                     Kim, K.T., Travers, M., 1997b. Les nutriments de l’Etang de Berre et des mi-
De Casabianca, M.L., Rabotin, M., Rigault, R., 2003. Preliminary results on                                ˆ         ´
                                            lieux aquatiques contigus (eaux douces, saumatres et marines; Mediterra-
  eelgrass regression and red seaweed dominance under increasing eutrophi-        ´
                                            nee NW). 4 Les nitrites. Marine Nature 5, 65e78.
  cation (Thau Lagoon, France). Acta Adriatica 44, 33e40.              Knowlton, N., 2004. Multiple ‘‘stable’’ states and the conservation of marine
Duarte, C.M., Sand-Jensen, K., 1990. Seagrass colonisation: patch formation       ecosystems. Progress in Oceanography 60, 387e396.
  and patch growth in Cymodocea nodosa. Marine Ecology Progress Series       Lardicci, C., Como, S., Corti, S., Rossi, F., 2001. Recovery of the macrozo-
  65, 193e200.                                     benthic community after severe dystrophic crises in a Mediterranean
`                                ´
Febvre, J., 1968. Etude bionomique des substrats meubles de l’etang de Berre.      coastal lagoon (Orbetello, Italy). Marine Pollution Bulletin 42, 202e214.
  Recueil des Travaux de la Station Marine d’Endoume 44, 298e355.          Laugier, T., Rigollet, V., De Casabianca, M.L., 1999. Seasonal dynamics in
Frederiksen, M., Krause-Jensen, D., Holmer, M., Laursen, J.S., 2004. Long-       mixed eelgrass beds, Zostera marina L. and Z. noltii Hornem., in a Mediter-
  term changes in area distribution of eelgrass (Zostera marina) in Danish       ranean coastal lagoon (Thau lagoon, France). Aquatic Botany 63, 51e69.
  coastal waters. Aquatic Botany 78, 167e181.                                                       ´
                                           Ledoyer, M., 1966. Ecologie de la faune vagile des biotopes mediterraneens   ´
Gacia, E., Duarte, C.M., 2001. Sediment retention by Mediterranean Posido-                                 ´
                                            accessibles en scaphandre autonome. II. Donnees analytiques sur les herb-
  nia oceanica meadow: the balance between deposition and resuspension.               ´
                                            iers de Phanerogames. Recueil des Travaux de la Station Marine d’En-
  Estuarine, Coastal and Shelf Sciences 52, 505e514.                  doume 41, 135e164.
Giesen, W.B.J.T., Katwijk van, M.M., Hartog den, C., 1990. Eelgrass condition       `
                                           Marba, N., Duarte, C.M., 1998. Rhizome elongation and seagrass clonal
  and turbidity in the Dutch Wadden Sea. Aquatic Botany 37, 71e85.           growth. Marine Ecology Progress Series 174, 269e280.
                         ´
Gourret, P., 1907. Topographie zoologique des etangs de Caronte, de Labillon,                     ´       ˆ
                                           Marion, A.F., 1887. Etude des etangs saumatres de Berre (Bouches-du-Rhone).  ˆ
  de Berre et de Bolmon. Flore, faune, migrations, etc. Annales du Museum  ´                                 ´
                                            Faune ichtyologique. Comptes Rendus de l’Academie des Sciences 104,
  d’Histoire Naturelle de Marseille 11, 1e166.                     1306e1308.
                      G. Bernard et al. / Estuarine, Coastal and Shelf Science 73 (2007) 617e629                       629

                ` ´            ´    ´
Mars, P., 1949. Contribution a l’etude biologique des etangs mediterraneens. ´   Rasmussen, E., 1977. The wasting disease of eelgrass (Zostera marina) and its
              ´       ´
  Quelques aspects de l’evolution de l’etang de Berre. Bulletin de la Societe´´    effects on environmental factors and fauna. In: McRoy, C.P., Helfferich, C.
     ´
  Linneenne de Provence 17, 8e16.                           (Eds.), Seagrass Ecosystems. Dekker, pp. 1e51.
                      ´         ´    ´
Mars, P., 1966. Recherches sur quelques etangs du littoral mediterraneen fran-    ´
                                          Reseau National de Bassin, 2006. http://sandre.eaufrance.fr.
  cais et sur leurs faunes malacologiques. Vie et Milieu 20, 1e359.
  ¸                                                                ´
                                          Riouall, R., 1971. Inventaire floristique des etangs de Berre et de Va€ (an-ıne
McGowan, S., Leavitt, P.R., Hall, R.L., Anderson, N.J., Jeppesen, E.,           ´                     ´´
                                            nees 1970e1971). Bulletin de la Societe des Sciences Naturelles et
  Odgaard, B.V., 2005. Controls of algal abundance and community compo-            ´
                                            d’Archeologie de Toulon et du Var 23, 153e160.
  sition during ecosystem state change. Ecology 86, 2200e2211.                               ´         ´
                                          Riouall, R., 1972. A propos des Enteromorphes des etangs de Berre et de
McKenzie, L.J., Finkbeiner, M.A., Kirkman, H., 2001. Methods for mapping                     ´
                                            Va€ Bulletin du Museum d’Histoire Naturelle de Marseille 32, 153e
                                              ıne.
  seagrass distribution. In: Short, F.T., Coles, R.G. (Eds.), Global Seagrass     160.
  Research Methods. Elsevier Science B.V., Amsterdam, pp. 101e121.        Rismondo, A., Curiel, D., Scarton, F., Mion, D., Caniglia, G., 2003. A New
   ´
Menendez, M., Hernandez, O., Comin, F.A., 2002. Spatial distribution and                                ¨
                                            Seagrass Map for the Venice Lagoon. In: Ozhan, E. (Ed.), Proceedings
  ecophysiological characteristics of macrophytes in a Mediterranean coastal     of the Sixth International Conference on the Mediterranean Coastal Envi-
  lagoon. Estuarine, Coastal and Shelf Science 55, 403e413.              ronment (2). Ravenna, Italy. Medcoast 03. Middle East Technical Univer-
                        ´       ´´
Minas, M., 1974. Distribution, circulation et evolution des elements nutritifs,     sity Publ., Ankara, Turkey, pp. 843e852.
                   ´       ´
  en particulier du phosphore mineral, dans l’etang de Berre. Influence des                                      ´
                                          Schachter, D., 1954. A propos d’une pollution accidentelle de l’etang de Berre
  eaux duranciennes. Internationale Revue der Gesamten Hydrobiologie               ´     ´
                                            provoquee par un deversement massif d’hydrocarbures. Rapport et Proces-   `
  59, 509e542.                                            ´
                                            verbaux des Reunions de la Commission Internationale pour l’Exploration
                                     `
Minas, M., Bonin, M.C., Coste, B., David, P., Minas, H.J., 1976. Caractere par-                   ´     ´
                                            Scientifique de la Mer Mediterranee 12, 251e254.
          ´                  ´
  ticulier du mecanisme de l’eutrophisation dans l’etang de Berre. Annales    Schroder, A., Persson, L., De Roos, A.M., 2005. Direct experimental evidence
                                             ¨
           ´
  de l’Institut Oceanographique 52, 153e161.                     for alternative stable states: a review. Oikos 110, 3e19.
Munkes, B., 2005. Eutrophication, phase shift, the delay and the potential return  Sfriso, A., Marcomini, A., 1997. Macrophytes production in a Mediterranean
  in the Greifswalder Bodden, Baltic Sea. Aquatic Sciences 67, 372e381.        coastal lagoon. Part I. Coupling with physico-chemical parameters and nu-
 ´               ´
Nerini, D., Durbec, J.P., Mante, C., 2000. Analysis of oxygen rate time series     trients concentration in water. Marine Environmental Research 44, 351e
  in a strongly polluted lagoon using a regression tree method. Ecological      375.
  Modelling 133, 95e105.                             Short, F.T., Wyllie-Echeverria, S., 1996. Natural and human-induced distur-
 ´         ´                       ´
Nerini, D., Mante, C., Durbec, J.P., Garcia, F., 2001. Une methode statis-       bance of seagrasses. Environmental Conservation 23, 17e27.
        ´         ´        ´          ´
  tique de determination de sequences caracteristiques dans une serie tem-    Silberstein, K., Chiffings, A.W., McComb, A.J., 1986. The loss of seagrasses
                          `
  porelle de plusieurs variables. Application a la physico-chimie des eaux      in cockburn sound, Western Australia. III. The effect of epiphytes on pro-
     ´                       ´
  de l’etang de Berre. Comptes-Rendus de l’Academie des Sciences 332,         ductivity of Posidonia australis Hook f. Aquatic Botany 24, 355e371.
  457e464.                                    Soltan, D., Francour, P., 2000. Monitoring system of benthic macrophytes
                         ` ´    ´       ´
Nisbet, M., Schachter, D., 1961. Contribution a l’etude ecologique des etangs      communities in Berre and Va€ ponds: development of a new strategy.
                                                             ıne
    ´     ´                          ´
  mediterraneens. Constituants chimiques des eaux de quelques etangs littor-                   ´
                                            Journal de Recherche Oceanographique 25, 53e57.
                  ´
  aux. Bulletin de l’Institut Oceanographique de Monaco 1207, 4e46.        Stora, G., Arnoux, A., Galas, M., 1995. Time and spatial dynamic of Mediter-
Olesen, B., Sand-Jensen, K., 1994. Patch dynamic of eelgrass Zostera marina.      ranean lagoon macrobenthos during an exceptionally prolonged interrup-
  Marine Ecology Progress Series 106, 147e156.                    tion of freshwater inputs. Hydrobiologia 300e301, 123e132.
        ´     ´       ´
Peralta, G., Perez-Llorens, J.L., Hernandez, I., Vergara, J.J., 2002. Effects of  Thayer, G.W., Kenworthy, W.J., Fonseca, M.S., 1984. The ecology of eelgrass
  light availability on growth, architecture and nutrient content of the sea-     meadows of the Atlantic Coast: a Community Profile. U.S. Fish Wildlife
  grass Zostera noltii Hornem. Journal of Experimental Marine Biology         Service FWS/OBS-84/02, 1e147.
  and Ecology 269, 9e26.                             Valiela, I., Collins, G., Kremer, J., Lajtha, M., Geist, M., Seely, B., Brawley, J.,
Pergent-Martini, C., Semroud, R., Rico-Raimondino, V., Pergent, G., 1995.        Sham, C.H., 1997. Nitrogen loading from coastal watersheds to receiving
            ´                  ´
  Localisation et Evolution des Peuplements de Phanerogames Aquatiques        estuaries: review of methods and calculation of loading to Waquoit, Bay.
      ´                 ˆ
  de l’Etang de Berre (Bouches du Rhone e France). Proceedings of the         Ecology Applications 7, 358e380.
  39th National Meeting of the Association Francaise de Limnologie. Uni-
                           ¸               Vermaat, J.E., Verhagen, F.C.A., Lindenburg, D., 2000. Contrasting responses
      ´                   ˆ   ´    ´
  versite de Corse & Agence de l’Eau Rhone-Mediterranee-Corse Publ.,         in two populations of Zostera noltii Hornem. to experimental photoperiod
  France, pp. 169e179.                                manipulation at two salinities. Aquatic Botany 67, 179e189.
Poole, H.H., Atkins, W.R.G., 1929. Photo electric measurements of submarine     Vidondo, B., Duarte, C.M., Middelboe, A.L., Stefansen, K., Lutzen, T.,  ¨
  illumination throughout the year. Journal of Marine Biology Assessment       Nielsen, S.L., 1997. Dynamics of a landscape mosaic: size and age distri-
  16, 297e324.                                    butions, growth and demography of seagrass Cymodocea nodosa patches.
Ramage, D.L., Schiel, D.R., 1999. Patch dynamic and response to disturbance       Marine Ecology Progress Series 158, 131e138.
  of the seagrass Zostera novazelandica on intertidal platforms in southern    Zar, J.H., 1999. Biostatistical Analysis, fourth ed. Prenctice-Hall Inc., London,
  New Zealand. Marine Ecology Progress Series 189, 275e288.              663 pp.