Content uploaded by Gabriel Reygondeau
Author content
All content in this area was uploaded by Gabriel Reygondeau on May 21, 2015
Content may be subject to copyright.
PERSEUS Deliverable Nr. 1.6
-1-
Definition of the Mediterranean eco-regions and maps
of potential pressures in these eco-regions
Deliverable Nr. 1.6
PERSEUS Deliverable Nr. 1.6
-2-
Project Full title
Policy-oriented marine Environmental Research in
the Southern EUropean Seas
Project Acronym
PERSEUS
Grant Agreement No.
287600
Coordinator
Dr. E. Papathanassiou
Project start date and duration
1st January 2012, 48 months
Project website
www.perseus-net.eu
Deliverable Nr.
1.6
Deliverable Date
31-10-2014
Work Package No
1
Work Package Title
Pressures and Impacts at Basin and Sub-basin Scale
Responsible
Xavier Durrieu de Madron - CNRS
Project Full title
Policy-oriented marine Environmental Research in
the Southern EUropean Seas
Project Acronym
PERSEUS
Grant Agreement No.
287600
Coordinator
Dr. E. Papathanassiou
Project start date and duration
1st January 2012, 48 months
Project website
www.perseus-net.eu
Deliverable Nr.
1.6
Deliverable Date
31-10-2014
Work Package No
1
Work Package Title
Pressures and Impacts at Basin and Sub-basin Scale
Responsible
Xavier Durrieu de Madron - CNRS
PERSEUS Deliverable Nr. 1.6
-3-
Authors &
Institutes Acronyms
D1.6 Scientific team
UPMC-LOV: G. Reygondeau, S.D. Ayata, S.
Gasparini, C. Guieu, J-O. Irisson, P. Koubbi
UMR-BOREA: (UPMC) P. Koubbi
UQ: C. Albouy
LEMAR: T. Hattab
Status:
Final (F)
!
Draft (D)
Revised draft (RV)
Dissemination level:
Public (PU)
!
Restricted to other program
participants (PP)
Restricted to a group specified
by the consortium (RE)
Confidential, only for members
of the consortium (CO)
PERSEUS Deliverable Nr. 1.6
-4-
Subtask 1.6: Definition of the Mediterranean ecoregions and maps
of potential pressures in these ecoregions
Gabriel Reygondeau1,2*, Jean-Olivier Irisson1,2, Sakina-Dorothée Ayata1,2, Stéphane Gasparini1,2,
Fabio Benedetti1,2, Camille Albouy3, Tarek Hattab4, Cécile Guieu1,2, and Philippe Koubbi1,2,5.
1 Sorbonne Universités, UPMC Univ Paris 06, UMR 7093, LOV, Observatoire Océanologique, F-
06230, Villefranche-Sur-Mer, France.
2 CNRS, UMR 7093, LOV, Observatoire Océanologique, F-06230, Villefranche-Sur-Mer, France.
3 Université du Québec à Rimouski, Département de biologie, 300, Allée des Ursulines,
Rimouski, Qc, G5L 3A1.
4 Laboratoire des Sciences de l'Environnement Marin, UMR 6539 LEMAR
(CNRS/UBO/IRD/IFREMER), Institut de Recherche pour le Developpement, Institut
Universitaire Européen de la Mer, Technopôle Brest-Iroise, Rue Dumont d'Urville, 29280
Plouzané - FRANCE
5 Sorbonne Universités, BOREA UMR 7208 MNHN/CNRS/IRD/UPMC, F-75005 Paris, France.
* Now at the Centre of Macroecology, Evolution and Climate in University of Copenhagen.
Department of Biology, Universitetsparken 15, DK-2100 Copenhagen and DTU Aqua,
Jægersborg Alle 1, 2920 Charlottenlund, Danemark.
PERSEUS Deliverable Nr. 1.6
-5-
Table of Contents
Executive summary / Abstract .......................................................................................... 6
1. General Introduction .................................................................................................. 7
2. Biogeochemical regions of the Mediterranean Sea ................................................ 9
2.1. Scope of the study ........................................................................................................... 9
2.2. Materials and methods ......................................................................................................... 9
2.2.1 Environmental data ........................................................................................................... 9
2.2.2 Quantification of the vertical boundaries ........................................................................... 9
2.2.3 Characterization of the environmental condition of each vertical layer ........................... 12
2.2.4 Identification of the spatial distribution of the biogeochemical regions ........................... 12
2.3 Results .................................................................................................................................. 13
2.3.1 Spatial and vertical environmental gradients of the Mediterranean Sea ......................... 13
2.3.2 Biogeochemical regions of the Mediterranean Sea ........................................................ 17
2.4 Conclusions on the biogeochemical regions ................................................................... 19
3. Ecoregionalisation of the Mediterranean Sea: characterisation of the marine
ecosystems and their potential threats ......................................................................... 20
3.1 Scope of the study ............................................................................................................... 20
3.2 Materials and Methods ........................................................................................................ 20
3.2.1 Biological data ................................................................................................................. 20
3.2.2 Anthropogenic pressures ................................................................................................ 22
3.2.3 Environmental niche modelling ....................................................................................... 23
3.2.4 Identification of the marine ecoregions ........................................................................... 26
3.3 Results and Discussion ...................................................................................................... 27
3.3.1 Mediterranean marine communities ................................................................................ 27
3.3.2 Distribution of the Mediterranean marine ecosystems .................................................... 28
3.3.3 Anthropogenic pressures on each marine ecosystem .................................................... 28
3.4 Conclusions on marine ecoregions ................................................................................... 30
4. General conclusions on the ecoregionalisation of the of the Mediterranean Sea
31
5. References ................................................................................................................ 40
5.1. Cited references ................................................................................................................. 40
5.2. Presentation to International conferences ...................................................................... 42
PERSEUS Deliverable Nr. 1.6
-6-
Executive summary / Abstract
In recent decades, it has been found useful to partition the ocean using the concept of
ecoregionalisation where within each region it is assumed that environmental conditions and
species associations are distinguishable and unique. Indeed, all partitions of the ocean that has
been proposed aimed to delineate the main oceanographical, ecological patterns and
discontinuities in order to provide a geographical framework for ecological studies and
management purposes. The aim of the present work is to integrate and process existing
environmental data and biological observations (from phytoplankton to top predators) in order to
define and characterize the Mediterranean Sea’ ecosystems. The first step was to gather a
comprehensive database informed on environmental conditions (22 parameters), biological
observations (more than 1500 species from plankton to whales) and human pressure (Halpern et
al., 2008) from online database, cruises and published articles. Based on a novel multi-clustering
methodology and on environmental niche modelling, a two levels partition of the Mediterranean
Sea the: biogeochemical regions (biotopes) and the ecoregions (associated biocenoses) are
proposed. This work allows us to characterize the main environmental divisions of the basin as
well as the biodiversity and mean organisms size gradient at each trophic level. Finally, an
ecological characterization of each ecoregion is proposed along with a perturbation index based on
13 human pressures.
Keyword: biogeography, ecoregionalisation, macroecology, human pressure, biodiversity,
biogeochemical regions, ecoregions.
Scope
The aim of the subtask 1.1.6 was to synthetize all the available data in the Mediterranean
Sea (and also from the other subtasks of the task 1.1 of PERSEUS) in order to give a
geographical framework for the work packages : WP3, WP4, and WP6.
PERSEUS Deliverable Nr. 1.6
-7-
1. General Introduction
Anthropogenic pressures strongly influence the physical and biological systems of the
oceanic realm. Recent modifications of the natural range and dynamic of environmental
factors that regulates the global marine ecosystems have introduced drastic modifications
over the biogeochemical division of the ocean (Reygondeau et al. 2013) and hence on
marine habitats with consequences on distributions and dynamics of species (Cheung et
al., 2012). In recent years, the combined effects of increasing observations and growing
demands of effective ecosystem based management have made possible to characterize
the present state of marine ecosystems which is a prerequisite to predict their future
changes.
In this context, a geographical framework based on environmental and biological
discontinuities is required to characterize marine ecosystem at a basin-scale. Over the last
decades, several types of oceanic partitions have been proposed such as Large Marine
Ecosystems (LME, Sherman, 2005), Marine Ecoregions of the World (MEOWS) (Spalding
et al., 2007) or Biogeochemical provinces (BGCP, Longhurst, 2007). Due to the growing
availability of marine observations with a global or basin-scale coverage combined with the
application of multidimensional or exploratory analyses in ecology, several novel
approaches, based on statistical modelling? have been developed to partition in a more
accurate way the global ocean.
Such approaches named ecoregionalisation can be defined by as “the process and
output of identifying and mapping broad spatial patterns based on physical and/or
biological attributes through classification methods used for planning and
management purposes” (Vierros et al. 2008). Ecoregionalisation is provides marine
ecological geography rests on significant changes in environmental forcing. Ecoregions are
thus considered as unique macro-assemblages of flora, fauna and the supporting
geophysical environment contained within distinct but dynamic spatial boundaries (Vierros
et al., 2008). Such ecological partition clearly helps understanding biogeochemical or
ecological processes at local, regional and macro-scale providing a framework for
ecosystem management.
Historically, the Mediterranean Sea (MS) has been divided into 8 geographical zones
(UNEP-MAP-RAC/SPA, 2010): Adriatic Sea, Aegean Sea, Alboran Sea, Levantine Sea,
Ionian Sea, Tyrrhenian Sea, Algerian-provencal basin and Tunisian-Syrian gulf. Each
division delineates specific parts of the basin topography and coastline morphology and are
now mainly used for European marine strategy consortium or for economical and political.
Nonetheless, this original partition does not take into account the biogeochemistry,
oceanography and ecology features of the Mediterranean Sea. More recently, several
studies have attempted to partition the basin either using abiotic parameters (Gabrié et al.,
2012) or biotic parameters (D’Ortenzio & D’Alacala, 2009). These studies have confirmed a
non-stationary picture of the dynamic of the basin, and significant biogeochemical
differences between regions that have consequences on the local dynamic of the primary
production for example.
In this subtask, a two-level ecoregionalisation of the Mediterranean Sea is proposed.
Based on both open access and data gathered from the subtask 1 of the PERSEUS project
for environmental and biological datasets, an ecological spatial reference is investigated.
The ecoregionalisation will present two level partitions: (1) a first 3D partition based on the
main environmental characteristics named biogeochemical regions following the concept
PERSEUS Deliverable Nr. 1.6
-8-
from Longhurst (2007; MEDI province) but here revisited using novel statistical approach
(Reygondeau et al., in revision) and (2) a subdivision based on the identification of main
biological assemblages named ecoregions. Finally, the impact of anthropogenic
pressures will be including into the spatial analysis to identify the most endangered
systems of the Mediterranean Sea.
PERSEUS Deliverable Nr. 1.6
-9-
2. Biogeochemical regions of the Mediterranean Sea
2.1. Scope of the study
Previous partitioning studies of the Mediterranean Sea have mainly based their
methodology on a few descriptors and without consideration of the vertical dimension and
thus they do not provide a full picture of the basin biogeochemical complexity.
In this section, we attempt to provide an objective division of the pelagic and seafloor
compartments of the Mediterranean Sea into what we call the biogeochemical regions.
This biogeochemical framework needs to be interpreted as the biogeochemical provinces
described by Longhurst (2007) but forced by mesoscale environmental features. Therefore,
our approach was to (1) identify the vertical limits of each pelagic layer; (2) quantify the
environmental characteristics of each layer; (3) identify the all-significant types of
environmental conditions in the l; and (4) identify the strength of the regions' boundaries.
!
2.2. Materials and methods
All the steps of the methodology are summarized in figure 1.
2.2.1 Environmental data
To capture the main environmental conditions and oceanographical features of the
Mediterranean basin, several annual and seasonal climatologies have been downloaded
from open access databases. Datasets are mainly composed by remote sensing
observations and merged oceanographic campaign data. Each parameter has been
selected to depict and/or characterize specific oceanographic features such as gyral
system, frontal structure, continental shelf, river runoff, water masses, coastal upwelling
and Low nutrient low chlorophyll (LNLC) areas. All relevant information of the dataset are
summarized in table 1.
Temperature, salinity, nitrites, nitrates, orthophosphate, silicate, pH, Chlorophyll-a
concentration, and dissolved oxygen concentration were gathered from the
MEDAR/MEDATLAS datasets (Medar Group, 2002). These datasets were computed into
seasonal and annual climatology. These parameters are gathered because of the full
spatial coverage of the basin at a 0.2° resolution ranging from 9.3°W to 36.5°E of longitude
and from 30°N to 46°N of latitude, and also the information on the vertical gradients of each
parameters (26 depths with non linear width). Available dataset for particulate organic
carbon fluxes, chlorophyll-a concentrations, euphotic depth (i.e., depth where only 1% of
the surface photosynthetically active radiation is available), mixed layer depth, thermocline
depth and intensity, wind speed, were gathered from publications. They were first
interpolated using a spline interpolation on a similar grid than the MEDAR/MEDATLAS
data. Seasonal and annual climatologies were computed from these data.
2.2.2 Quantification of the vertical boundaries
Three vertical boundaries have been defined to separate the epipelagic, mesopelagic,
and bathypelagic vertical layers. The shallower boundary represents the depth where the
epipelagic and the mesopelagic domains are separated. This boundary is approximated at
the maximal depth where primary production is available. Based on previous work on
vertical limitation of photosynthesis activity (Sverdrup, 1953; Behrenfeld,
PERSEUS Deliverable Nr. 1.6
-10-
Table 1. Information on the Environmental variables taken into account in this study.
PERSEUS Deliverable Nr. 1.6
-11-
Figure 1. Sketch diagram of the methodology of the part 1 on Biogeochemical regions. All step are
detailed in the 2.2 Materials and Methods
PERSEUS Deliverable Nr. 1.6
-12-
2010) the upper mesopelagic boundary is set at the shallowest depth between the euphotic
depth and the mixed layer depth for each geographical cell of the global ocean. The annual
climatology and latitudinal variation of the upper mesopelagic boundary were mapped for
the MS following this concept for the global ocean (Figure 2a).
The bottom boundary separates the mesopelagic and the bathypelagic layers. In
absence of any consensual definition, this boundary has been based on the shape of the
flux of particular organic carbon (FPOC). Indeed, it has been shown that it reflects the main
biogeochemical vertical gradients in the water column (see Reygondeau et al., in revision).
The bottom boundary has been defined at the depth where FPOC tends to an asymptote
(decrease of POC is not significant between two consecutive depth). First, each profile is
interpolated every 5 meters between 0 and 5000m using a cubic spline interpolation.
Second, to identify the depth of the boundary, two tests are performed on each FPOC
profiles. The derivative function of FPOC against depth is computed and the depth of the
boundary is set at the depth where the decrease of FPOC between 5 consecutive points is
not significant. This resulting depth is compared to the one identified at the depth
corresponding to 5% of the surface FPOC. The average depth resulting from the two
methods is calculated and mapped on Figure 2b.
2.2.3 Characterization of the environmental condition of each vertical layer
In the global ocean, environmental gradients over the vertical dimension are usually
more pronounced comparatively to the horizontal dimension. Consequently, the partitioning
of the variance in a three-dimensional analysis is biased for any ordinary statistics
methodology, as the vertical gradient will be detected over the horizontal gradient. The
biogeochemical division of the mesopelagic layer with consideration of both horizontal and
vertical environmental gradients was performed on the basis of previous methodologies
(Oliver et al., 2004; Beaugrand et al. 2011; Reygondeau et al. 2013). This allowed
obtaining distinct environmental matrices for each of the four vertical layers (epipelagic,
mesopelagic, bathypelagic, and seafloor).
To identify the main environmental parameters driving the spatial variance of each
layer, a principal component analysis (PCA) was performed on the environmental matrix of
each layer. The first and second components of each PCA are mapped on Figure 3 as well
as the result of the PCA. The environmental factors that contribute the most to the first two
principal components (i.e. that explain most of the total variance of the environmental
matrix of each layer) were identified (table 2).
2.2.4 Identification of the spatial distribution of the biogeochemical regions
To propose an objective environmental spatial division, we applied the methodology
proposed by Oliver et al. (2004). This numerical procedure uses four types of clustering
methodologies, here applied on the normalized environmental matrices of each vertical
layer: K-means (Hartigan and Wong, 1979), C-means (Quackenbush, 2001), agglomerative
with ward linkage (Ward, 1963) and with complete linkage (Legendre & Legendre, 1998)
(step 2.2; Fig. 1). These four types of clustering algorithm were selected for their ability to
synoptically group similar environmental data and for their differences to handle low
dissimilarity clusters (i.e. sensible cluster). Each clustering algorithm was run to retrieve
from 2 to 50 clusters using a Euclidian distance. The C-means and K-means were repeated
999 times per iterations and the division the most retrieved was selected. The next step of
the methodology consisted in the identification of the optimal number of cluster to consider
(following the analysis called the figure of merit (FOM) in Yeung et al., 2001) in the next
PERSEUS Deliverable Nr. 1.6
-13-
steps of the procedure. The strength of the boundaries between the identified regions
obtained from different clustering was estimated in order to provide an objective definition
of the biogeochemical regions.
2.3 Results
The objectives of the study were to depict a synoptical view of the environmental
condition for the whole basin by taking into account and summarizing the main
environmental features of the Mediterranean Sea. The use of exploratory statistic allowed
to objectively detangling the links between variables for each layer, to identify and describe
the main 3D biotopes (i.e. multivariate environmental intervals) and quantify the strength of
the vertical and horizontal boundaries between BGCRs. In this study, a set of
methodologies is selected for these purposes because they decompose the vertical and
horizontal variance of the whole Mediterranean Sea environmental gradients. Interestingly,
even if result are logically link to database used, any biases on the absolute values of a
given parameters does not affect the methodology if the spatial gradient are close to reality.
Consequently, each result needs to be taken as a general biogeochemical trend and not as
a fine scale resolution study that investigate on a given process.
2.3.1 Spatial and vertical environmental gradients of the Mediterranean
Sea
Figure 2. Depth of the vertical boundaries of the water column: (a) between the epipelagic layer and
the mesopelagic layer, (b) between the mesopelagic layer and the bathypelagic layer.
Using the shallowest depth between the mixed layer depth and the euphotic depth,
we found that the epipelagic/mesopelagic boundary varies over a 65 m range between
10 m and 75 m (Fig. 2a). The deepest upper mesopelagic boundaries are found in open
ocean areas and particularly in the oriental basin and Tyrrhenian Sea. These areas are
typically oligotrophic with low primary productivity and low detritus concentration implying a
high light penetration. In contrast, more productive areas located in the occidental basin
and continental shelves exhibit shallower depth of epipelagic/mesopelagic boundary (down
to 40 m). In those regions, the first 200 m are more turbulent waters resulting in shallower
epipelagic layer than in tropical areas (Fig. 2a).
PERSEUS Deliverable Nr. 1.6
-14-
Figure 3.Results of the PCA on the environmental conditions of each considered layer and map of
the PC1 and PC2
PERSEUS Deliverable Nr. 1.6
-15-
The lower boundary separating the mesopelagic and the bathypelagic layers are
presented Fig. 2b.The annual climatology (Fig. 2b) shows that this limit has strong variation
between 140 m and 1500 m. The deepest boundaries are localized in open ocean areas
with low biological activity. These area are also the deepest of the basin and the more
stable in term of mesoscale activity. Thus, regions near rift show a lowest mesopelagic
bottom boundary than oceanic basin.
The map of the first principal component (PC1) reveals a clear longitudinal gradient
from the occidental to the oriental basin retrieved for epi- and mesopelagic layer and to a
lesser extend for bathypelagic layer (Fig. 3). This gradient is mainly supported by the
important contribution of the temperature to each layer environmental variance. Therefore,
this environmental can be directly attributed to the principal atmospherical (i.e. climate) and
ocean circulation (main current) forcing. Regional features such as influence of the Atlantic
Ocean (impacting strongly on T and S) on the occidental basin or influence of river run-off
or high primary productive regions are also retrieved on PC1 of pelagic layers. These
regions are carried by the high contribution of several parameters to the total variance that
varies between pelagic layer (table. 2): epipelagic: temperature, chlorophyll-a
concentration, oxygen concentration, mesopelagic: temperature, euphotic depth, nutrients,
salinity, bathypelagic: bathymetry, nutrients.
Logically, the regions with the higher primary production like in the Ligurian sea is
retrieved because of the importance of chlorophyll-a concentration in the epipelagic layer,
high nutrients concentration and stratification variation in deeper layers. The second
component of the principal component analysis (PC2) mainly detects areas with strong
wind stress and seasonal variability in water column stability that influence the deeper
layers nutrient concentration. PCA performs on the seafloor layer reveal a strong opposition
in the environmental conditions between continental shelf (warmer with higher nutrient
concentration) and open sea areas.
Table 2.Contribution of each parameter in the first two principal components PC1 and PC2 for each
vertical layer (in %).
PERSEUS Deliverable Nr. 1.6
-16-
Figure 4.Biogeochemical regions of the Mediterranean Sea defined for each vertical layer.
PERSEUS Deliverable Nr. 1.6
-17-
2.3.2 Biogeochemical regions of the Mediterranean Sea
The previous analysis highlighted clear environmental gradients and regional features
for the whole basin. This resulted in the identification of 63 biogeochemical regions: 12 for
the epipelagic, 12 for the mesopelagic, 13 for the bathypelagic and 26 for the seafloor (Fig.
4). The strength of the spatial boundaries of each layer is represented on supplementary
Figure 5 and the environmental intervals of each biogeochemical regions are shown on
supplementary Figures 6 to 9.
The longitudinal pattern identified by the previous analysis (Fig. 2) is retrieved on Fig.
4, but here significant changes in environmental conditions are quantified allowing to
cluster the sea into 3D biotopes. Strongest frontiers of all pelagic layers are distributed
nearby the Gibraltar strait, the Sicilian strait, and between the Ionian Sea and the Levantine
Sea. Also, marginal seas such as the Adriatic Sea, Aegean Sea and Ligurian Sea are
systematically delineated from the occidental and oriental basin in each layer. We can than
assume that the environmental pattern observed on Fig. 2 geographically vary by regional
/stepwise shift over the basin (strength of shifts being quantified on Fig. S5). These
significant geographical changes highlight well-known local/regional specificities driven by
the continental morphology, the atmospheric forcing and the hydrological/oceanographical
dynamics. For instance the strength of the environmental shift being the contour of each
resulting division show a high similarity with the main circulation (Mermex group, 2011)
from the bottom to the surface. These spatial modifications of environmental conditions can
thus be linked to the surface climate forcing combined with time of water retention’s.
Therefore, each biogeochemical region is characterized by homogenous environmental
conditions that are significantly different from surrounding regions. Each biogeochemical
region represents a characteristic biotope (pelagic / benthic habitat) for marine species.
2.3.3 Caveat and validation of the spatial clustering
The present study is an attempt to provide a robust and objective environmental
partition of the Mediterranean Sea by the implementation of a novel and objective statistical
methodology. To decrease the potential bias of merging several types of dataset, only
parameters coming from sources of comparable quality have been retained. Main datasets
originated from the MEDAR/MEDATLAS or from published datasets because these data (1)
presented a similar spatial and vertical grid, (2) have been computed from same sampling
methodology and (3) have been interpolated using similar methodologies.
However, the spatial heterogeneity of the in situ sampling used as well as the
variety of measuring techniques lead to some uncertainty on the resulting outputs. These
different types of biases need to be kept in mind for the interpretation of the result,
especially in under sampled areas that are mainly located in the southern regions of the
basin. To decrease the potential effects of these biases, the methodology selected was
performed after a normalization of the raw values. Therefore, the spatial partition of each
methodology is based on changes in the distributions of each parameter rather than the
raw amplitude value.
PERSEUS Deliverable Nr. 1.6
-18-
Figure 5. Upper panel: BOUM cruise track superimposed to the map of the epipelagic
biogeochemical regions of the MS . Lower panel: Variation of the habitat suitability index (HSI:
Habitat Suitability index which is the probability to retrieve a given regions according to the local
environmental conditions) associated with in situ measurement of Sea Surface Temperature (SST),
Salinity, Chlorophyll-a concentration [Chla] and dissolved oxygen. Dash line represent change in
epipelagic biogeochemical region.
PERSEUS Deliverable Nr. 1.6
-19-
To test the effectiveness of the boundaries detected (Figure 4), the environmental
partition obtained has been confronted to an independent set of data. Owing to the difficulty
to retrieve deep environmental sampling originating from the same cruise and covering
both occidental and oriental basins, only the epipelagic layer has been tested using data
(SST, SSS, Chlorophyll-a and oxygen concentration) collected during the BOUM cruise
(Moutin et al., 2012) in the epipelagic layer (Figure 5). Probabilities of each biogeochemical
region (Habitat Suitability Index, HIS) crossed by the cruise track were confronted to actual
surface parameters measured during the cruise. Each boundary between biogeochemical
regions (dash line, figure 5) coincides indeed with significant change (defined as a variation
of more than 5% of the mean of the 5 previous sampling site) of at least one of the
parameters. However, significant changes of the in situ data are also detected within the
same biogeochemical regions. Significant variations within biogeochemical regions can be
either attributed to: (1) sub-mesoscale biogeochemical structures that are not detected by
our methodology owing to the resolution of the used parameters; (2) temporal variations of
the environmental parameters during the one-month summer cruise that are not considered
in the environmental envelopes defined in our study that consider annual means.
2.4 Conclusions on the biogeochemical regions
Based on a very large dataset and using a novel statistical methodology, the
present study propose objective maps of the all distinct multivariate environmental
conditions types that can be encountered in the Mediterranean basin with associated
strength of the boundaries and stability index. These biogeochemical regions have been
environmentally characterized for all parameters considered (see supplementary figure 6 to
9).
As each biogeochemical region is delineated by complex oceanographical features
that influence environmental conditions in an anisotropic way, the map proposed provide a
geographical framework of environmentally homogenous regions. Nonetheless, as effects
of temporal fluctuations of each parameter on the distribution of the biogeochemical regions
is not considered, the dynamic of the systems is not yet fully captured. Each
biogeochemical region can thus be considered as potential habitats where adapted species
can maintain their populations. Results also suggest that previous 2D and univariate
partitions of the basin cannot be extrapolated to the whole water column.
Results also aim to be used as a geographical framework of Mediterranean
environmental conditions to (1) optimize the design of new sampling cruises, (2) spatially
merge heterogeneous dataset and (3) improve the comparison of modelled and observed
data.
PERSEUS Deliverable Nr. 1.6
-20-
3. Ecoregionalisation of the Mediterranean Sea:
characterisation of the marine ecosystems and their
potential threats
3.1 Scope of the study
The ocean is composed of a jigsaw of ecological units separated by hydrological
frontiers. Each unit is characterized by a given biotope (environmental characteristics) and
a biocenoses (adapted species in non-exclusive competition that form a trophic web) and
are defined as « ecosystem ». In 1995, on the basis of 15 biogeochemical variables,
Longhurst proposed a two level partition of the global ocean, commonly accepted by the
scientific community. The first level of partition represents the main ocean climatic types
named biomes and the second level of division subdivides each biome according to
regional features and oceanic basin shape. A more detailed description of such
biochemical regions, for the Mediterranean Sea, was presented in the previous section.
The goal of this part of the study is now to delineate ecoregions of the MS.(regions that are
homogenous in terms of environmental characteristics and biological communities). This
partition will then provide a framework to evaluate the impact of anthropogenic pressures
on each identified marine ecosystems.
Growing anthropogenic pressures (over-fishing, pollution, climate change, etc.) are
assumed to have modified the natural equilibrium of the MS marine ecosystems. In this
context, international consortia have advocated for a holistic approach that integrates all
components of ecosystems for the implementation of an effective conservation planning
(MEA, 2003; Boyen et al., 2012; IPCC, 2013). However, the Mediterranean Sea is
characterized by an important influence of mesoscale activity and localised human
pressures on regional biogeochemistry, hydrology and ecosystem trophodynamics (Coll et
al., 2012). Consequently, large-scale partitions such as biogeochemical provinces
(Longhurst, 2007) cannot be applied in this ecosystem-level-management perspective. In
recent years, the increasing availability of observations (biotic and abiotic) and advances in
numerical techniques has made it to define ecoregions appropriate to informing ecosystem-
based management.
The objectives of the present study are to (1) map the distribution of marine taxa,
from phytoplankton to top predators, over the Mediterranean Sea through environmental
niche modelling; (2) delineate marine ecoregions based on the distributions hence
modelled; (3) quantify anthropogenic pressures for each ecoregion.
3.2 Materials and Methods
3.2.1 Biological data
Biological data is extracted from online databases for the Mediterranean basin. The
harvested databases are:
• Intergovernmental Oceanographic Commission (IOC) of UNESCO. The Ocean
Biogeographic Information System (OBIS, http://www.iobis.org/)
• Global Biodiversity Information Facility (GBIF, 2012 ;http://www.gbif.org/)
PERSEUS Deliverable Nr. 1.6
-21-
• Ocean Biogeographic Information System for megavertabrate (OBIS-SeaMAP,
http://seamap.env.duke.edu/)
• Fishbase (http://www.fishbase.org/)
• The Coastal and Oceanic Plankton Ecology Production and Observation database
(http://www.st.nmfs.noaa.gov/plankton/-).
• LEFE-CYBER (http://www.obs-vlfr.fr/proof/index_vt.htm)
• SESAME database (http://isramar.ocean.org.il/sesamemeta/)
• CLIOTOP database (http://www.imber.info/index.php/Science/Regional-
Programmes/CLIOTOP)
• Digitalized fishes and mammals atlas used in Mouillot et al (2011).
After initial duplicate deletion, the final dataset comprises only taxa (1) that are
informed down to species level, (2) for which at least 30 observations were available and
(3) which maps of occurrence were reviewed by experts. This resulted in localised 17 606
777 occurrences covering 1280 species from phytoplankton to top predators with high
heterogeneity of sampling effort (Figure 6). Due to the heterogeneity of the sampling
methodologies used in this dataset, all abundance data were converted into “presence
only”.
Figure 6.Spatial distribution of number of observations for low and high trophic level species. Areas
with white colour mean that no data have been retrieved.
To better explore ecological patterns in the distribution of species, additional
information were gathered for each species: taxonomy, depth range (minimum and
maximum depth where the species can occur), size (minimum, mean and maximum), main
habitat of species (i.e. epipelagic, mesopelagic, bathypelagic, demersal, benthic) and mean
trophic level (according to Pauly et al, 1998). Online databases and publications focussing
on each species are used to inform these metadata: fishbase (www.fishbase.org),
encyclopedia of life (http://eol.org/), OBIS-SEAMAP (http://seamap.env.duke.edu/), World
Register of Marine Species (http://www.marinespecies.org/).
PERSEUS Deliverable Nr. 1.6
-22-
3.2.2 Anthropogenic pressures
Several comprehensive and validated maps of anthropogenic pressures were
gathered to quantify the effect of human activities on marine ecosystems. These maps
were originally computed by Halpern et al. (2009) and reinterpreted by Coll et al. (2012) by
integrating new observations. The original data is spline-interpolated from the 10 x 10 km
grid to our spatial grid: from 9.3°W to 36.5°E of longitude and from 30°N to 46°N of latitude
with a resolution of 0.2° x 0.2° (i.e., spatial grid of the environmental parameters in
MEDAR/MEDATLAS). To summarize the information, pressures listed in Table 3 were
grouped into three categories: ‘Climate change’, ‘Fishery activities’ and ‘Pollution’. In each
category, pressures were summed and normalize by the maximal value obtained;new
pressure layer (0 <values<;1) are displayed Figure 7).
Pressure type in this paper
Pressure defined by Halpern et al 2009 and Coll et
at 2012
Climate change
UV Radiation
Sea Surface Temperature Increase
Ocean Acidification
Fishery activities
Fishing Pelagic Low Bycatch
Fishing Pelagic High Bycatch
Fishing Demersal Non Destructive Low Bycatch
Fishing Demersal Non Destructive High Bycatch
Fishing Demersal Destructive
Artisanal Fishing
Pollution and euthrophication index
Urban Runoff
Risk Of Hypoxia
Organic Pollution Pesticides
Nutrient Input Fertilizers
Invasive Species
Commercial Shipping
Coastal Population Density
Benthic Structures Oil Rigs
Table 3. Groups of anthropogenic pressures
PERSEUS Deliverable Nr. 1.6
-23-
Figure 7. Spatial distribution of the groups of anthropogenic pressures: Pollution, Climate change
and Fishery activities
3.2.3 Environmental niche modelling
Although we use a very large database, observations were too scarce and too
heterogeneously distributed to simply interpolate the distribution of the 1280 selected
species over the whole basin. Alternatively, we use environmental niche modelling to map
their potential distribution. The principle of niche modelling based on presences-only is to
define the conditions in which a species is most often found and then to map the probability
of occurrence of this species according to these conditions. In a data poor context (which is
the general case in biological studies) this is a commonly accepted way to obtain
continuous distribution map. The construction of our multi-model of ecological niche
required the following steps (Figure 8).
step 1. Collection of the environmental conditions related to observed occurrences.
Environmental conditions were based on the climatologies of the same 19 variables
used to define bioregions (Section 2): Temperature, Salinity, Chla, O2, pH, NO3, NO2, PO4,
SiO2, depth of euphotic zone, intensity and depth of thermocline (and seasonal variability
thereof), mixed layer depth and its seasonal variability, wind speed, bathymetry and class
of bathymetry. When relevant, these variables were divided in 4 vertical layers (epi, meso,
bathypelagic and seafloor) and in seasons or months.
Each of these variables was interpolated to the position of each occurrence, in
terms of latitude-longitude but also depth layer. When the time of sampling was available,
the corresponding monthly/seasonal climatology was used; otherwise, an annual
climatology was used.
PERSEUS Deliverable Nr. 1.6
-24-
Figure 8. Sketch diagram of the methodology of the Part 2 on Ecoregionalisation. All step are
detailed in 3.2.3
PERSEUS Deliverable Nr. 1.6
-25-
As several models are sensitive to the number of variables, only 4 variables were
finally selected. The best combination has been obtained using an Escoufier vector
methodology (Legendre and Legendre, 1998) and a principal component analysis. In
addition, according to expert knowledge, only parameters with an ecological meaning for
the considered species and/or layer were retained.
Step 2. Environmental niche models
In this study, a multi model approach was adopted to best approximate the
environmental tolerance of each species. By accounting the spatial variability of each
model due to the difference in the statistical approach used, we can indeed assume that
the environmental range of the species is better delineate. However, owing to the quality of
the data, only Environmental Niche Models (ENM) dealing with presence only data were
selected : Domain, Bioclim (from Biomod package Thuillier et al., 2008), Maxent (Phillips et
al., 2004), NPPEN (Beaugrand et al., 2011) and ENFA
Step 3. Numerical validation of the distribution
The spatial distribution of each species and ENM has been numerically tested using
the Hirzel et al. (2006) methodology. This test was selected, as it is one of the most
accurate for ENM on presence only. This test consists in the confrontation of relative
presence (sum of the presence divided by the total of observations of the species in the
MS) with associated probability of occurrence at different spatial resolution. The test was
permutated by changing the spatial resolution of both parameters and thus by re-computing
both probability of occurrence and relative occurrence. Then, a coefficient of regression
(here a spearman coefficient of regression) was calculated using all data from all
permutations. Values of the coefficient of regression were therefore considered as an index
of quality of the model. All specificity of the methodology can be retrieved in Hirzel et al.
(2006).
Step 4. Model averaging
For each species, probabilities of presence computed from the five ENM have been
averaged using weight values computed in step 3. Only ENM with a Hirzel’s index value >
0.5 have been considered for the averaging procedure. In addition, each model was
weighted by the value of the index. The annual mean and standard deviation of the spatial
distribution of the probability of presence of 6 emblematic species resulting from this
computation are presented Figure 9
PERSEUS Deliverable Nr. 1.6
-26-
Figure 9.Spatial distribution of several emblematic species of the Mediterranean Sea extracted from
the average probability of presence of the 5 ENM computed.
3.2.4 Identification of the marine ecoregions
To characterize the Mediterranean ecoregions that represent specific environmental
and ecological composition, the clustering methodology has been used as for the
identification of the biogeochemical regions (Figure 1). Firstly, the clustering methodology
was performed to identify the primary producers (i.e., Trophic level (TL) between 1 and 2),
primary consumers (TL between 2 and 3), secondary consumers (TL between 3 and 4) and
top predators (TL between 4 and 5) assemblages (Figure 10). Secondly, based on the
spatial distributions of each trophic assemblage (4 layers) and biogeochemical regions (4
layers), the clustering methodology (see step 2.3 of §2.2 Material and Methods) was used
to find a trade off between all agglomerative methodologies. Therefore, each ecoregion
represents a specific species association and environmental interval (Figure 11). In
addition, to visualise the main anthropogenic pressures per ecoregion, the spatial
coordinates of each ecoregion are used to average merged anthropogenic pressures.
Mean anthropogenic pressure are mapped on figure 12 as well as the sum of the
pressures.
PERSEUS Deliverable Nr. 1.6
-27-
3.3 Results and Discussion
3.3.1 Mediterranean marine communities
Figure 10.Spatial distribution of primary producers (i.e., Trophic level between 1 and 2), primary
consumers (TL between 2 and 3), secondary consumers (TL between 3 and 4) and top predators
(TL between 4 and 5).
Figure 10 maps the different trophic assemblages (here named communities). Each
cluster represents a specific marine species association from primary producer (PP) to top
predator (TP). Open seas areas appear more divided by the lower trophic communities (PP
and Primary Consumers) while coastal areas are more partitioned by high trophic levels
(Secondary Consumer and TP). This result is attributed to the opposite pattern of
biodiversity between trophic level driven by an important physiological evolution (increase
of thermal regulation capacity) and functional trait changes (planktonic versus nektonic)
throughout the trophic levels. Therefore, differences in communities’ distribution can be
attributed to change of forcing factors throughout the trophic web. For example, PP
community’s distribution can be directly attributed to nutrient types and concentration, index
of stratification and euphotic depth that influence both their abundance and species
occurrence, while TP community’s distribution will be more influenced by temperature and
food availability.
PERSEUS Deliverable Nr. 1.6
-28-
3.3.2 Distribution of the Mediterranean marine ecosystems
Figure 11.Spatial distribution of the Mediterranean marine ecosystems.
Based on the spatial distribution of each trophic community and environmental
variable, the distributions of the marine ecosystems of the Mediterranean Sea have been
captured (Figure 11). Each ecoregion detected here represents a characteristic species
association from primary producers to top predators (i.e., biocenoses) forced by similar
environmental conditions (i.e., biotopes).
The map of the 25 ecoregions delineates well the characteristics of open seas and
more coastal features, such as the Northern current or the Levantine basin, as well as
known ecosystems, such as the Gulf of Lions, and the southern area off Tunisia including
Gulf of Gabes. More surprisingly, several ecoregions are distributed as boundaries of
bigger and well-known systems. This result confirmed the assumption of Van der Spoel et
al. (1996) and Odum (1974) that boundaries between two main ecosystems need to be
considered as an ecosystem as well. Indeed, ecotones (i.e., boundary between two
ecosystems) are characterized by a trade off between species associations originated from
adjacent ecosystems.
The global distribution of the ecoregions reveals a coherent ecosystemic spatial
framework at the exception of the Aegean Sea. Indeed, this particular sea shows a patchy
distribution of the clusters. This could reflect high heterogeneity of marine habitats likely
influenced by inputs from the Black Sea. Indeed, owing to the scarce resolution of the data
used, sub-mesoscale and local features cannot be captured and thus can bias the
methodology in some specific areas.
3.3.3 Anthropogenic pressures on each marine ecosystem
The mean anthropogenic impacts have been computed for each of the 25
ecoregions identified in the Mediterranean Sea (Figure 12). Globally, climate change
pressures are high everywhere with a marked gradient from west to east. Pollution and
fisheries pressures are more localized in coastal areas and in the western basin. The
cumulated pressure map shows that coastal areas appear as more pressured than open
sea areas both in northern and southern parts. Hot spots of potential perturbed ecosystem
are very localized and encompass: the Algero-Tunisian coast, the Adriatic Sea, the Aegean
Sea, the Gulf of Gabes, the Catalan coasts, the Gulf of Lions and the Egyptian coasts.
PERSEUS Deliverable Nr. 1.6
-29-
Figure 12. Mean anthropogenic pressure and cumulative impact per ecoregion.
PERSEUS Deliverable Nr. 1.6
-30-
Interestingly, some large open sea regions are characterized by low cumulated
pressures, mainly in the Central part of the MS: in the Southern and Northern Ionian Sea
and in the Tyrrhenian sea
The present analysis reveals that all the ecoregions of the Mediterranean basin
undergo at least one type of pressure (see Supplementary Figure 10). Indeed, relation
between fisheries and pollution pressures per ecoregions tend to be highly positively
correlated suggesting that regions with the highest fishing activities are also the most
polluted. Climate change and others pressures are negatively correlated, which suggests
that stronger climate change likely impact regions with relatively low pollution and fishery
activities. It has to be noted nonetheless, that these last relations are hardly significant.
Consequently, spatial distributions of hot spots of perturbation can directly be attributed to
regions with high human density that mechanically increase the number of anthropogenic
activities in near coast areas; while open sea areas are mainly pressured by long term
change of environmental conditions.
3.4 Conclusions on marine ecoregions
The present study have succeed to identify all the marine species (with enough
observations to be modelled) that can be encountered in the Mediterranean basin from
primary producers to top predators, in both pelagic and neritic domains and to characterize
their potential spatial distributions. This work allows to better capture the trophic complexity
of marine ecosystems providing a more realistic picture of the Mediterranean spatial trophic
organisation. Indeed, known macro-environmental structures already defined in previous
studies are here refined on the basis of trophic assemblages. 25 ecoregions subdivide the
Mediterranean macro-environmental partition defined previously by delineating all
significant environmental and trophic forcing affecting the biogeochemical processes and
biodiversity patterns.
PERSEUS Deliverable Nr. 1.6
-31-
4. General conclusions on the ecoregionalisation of
the of the Mediterranean Sea
For the subtask 1.6.6 of the PERSEUS FP7-Project, we have: (i) partitioned the
Mediterranean Sea into biogeochemical regions (both in 2- and 3-dimensions), (ii)
partitioned the Mediterranean Sea into 25 ecoregions, and (iii) quantified the anthropogenic
pressures for each of the ecoregion. The ecoregionalisation approach performed in this
study provides applicable tools in a context of ecosystemic fisheries management and
biodiversity management, especially for the on-going implementation of the EU’s Marine
Strategy Framework Directive (MSFD). We have hence provided an ecological
geographical framework characterizing the main species assemblages and environmental
features, with a quantification of the various environmental and anthropogenic forcing. The
regions undergoing the highest risks of perturbation have been identified and mapped, and
now require adapted conservation strategies to reach a stable equilibrium for the
sustainability of marine ecosystems.
PERSEUS Deliverable Nr. 1.6
-32-
Supplementary materials
Supplementary Figure 1. Annual climatology of Temperature, salinity and [Chla] for each identified layer
Supplementary Figure 2. Annual climatology of NO3, PO4 and SiO2 concentration for each identified layer
PERSEUS Deliverable Nr. 1.6
-33-
Supplementary Figure 3. Annual climatology of dissolved Oxygen concentration,
pH and NO2 concentration for each identified layer
Supplementary Figure 4. Annual climatology of Thermocline Depth and Intensity,
Mixed layer and Euphotic depth and wind speed
PERSEUS Deliverable Nr. 1.6
-34-
Supplementary Figure 5. FOM and distribution of the effectiveness of the boundaries for each layer
PERSEUS Deliverable Nr. 1.6
-35-
Supplementary Figure 6. Environmental chacterisation of each epipelagic biogeochemical regions
PERSEUS Deliverable Nr. 1.6
-36-
Supplementary Figure 7. . Environmental chacterisation of each mesopelagic biogeochemical regions
PERSEUS Deliverable Nr. 1.6
-37-
Supplementary Figure 8. . Environmental chacterisation of each bathypelagic biogeochemical regions
PERSEUS Deliverable Nr. 1.6
-38-
Supplementary Figure 9. . Environmental chacterisation of each seafloor biogeochemical regions
PERSEUS Deliverable Nr. 1.6
-39-
Supplementary Figure 10. Mean merged Anthropogenic impact per ecoregions. The orientation of the
histogram shows if the value is higher or lower than the mean of the whole of the parameter in the whole basin.
PERSEUS Deliverable Nr. 1.6
-40-
5. References
5.1. Cited references
Beaugrand, G., Lenoir, S., Ibañez, F., Manté, C., 2011. A new model to assess the probability of
occurrence of a species, based on presence-only data. Marine Ecology Progress Series,
424, 175-190.
Behrenfeld, M.J., 2010. Abandoning Sverdrup's Critical Depth Hypothesis on phytoplankton
blooms. Ecology, 91, 977-989.
Boyen, C., Heip, C., Cury, P., Baisnée, P.-F., Brownlee, C., Tesmar-Raible, K., Allen, I.,
Arvanitidis, C., Austen, M., Bolhuis, H., 2012. EuroMarine Research Strategy Report:
Deliverable 3.2. Seventh Framework Programme Project EuroMarine.
Coll, M., Piroddi, C., Albouy, C., Ben Rais Lasram, F., Cheung, W.W.L., Christensen, V., Karpouzi,
V.S., Guilhaumon, F., Mouillot, D., Paleczny, M., 2012. The Mediterranean Sea under
siege: spatial overlap between marine biodiversity, cumulative threats and marine reserves.
Global Ecology and Biogeography, 21, 465-480.
Cheung, W.W.L., Sarmiento, J.L., Dunne, J., Frölicher, T.L., Lam, V.W.Y., Palomares, M.L.D.,
Watson, R., Pauly, D., 2012. Shrinking of fishes exacerbates impacts of global ocean
changes on marine ecosystems. Nature Climate Change.
Durrieu de Madron, X., Guieu, C., Sempéré, R., Conan, P., Cossa, D., D’Ortenzio, F., Estournel,
C., Gazeau, F., Rabouille, C., Stemmann, L., 2011. Marine ecosystems’ responses to
climatic and anthropogenic forcings in the Mediterranean. Progress in Oceanography, 91,
97-166.
D'Ortenzio, F., Ribera D'Alcalã, M., 2009. On the trophic regimes of the Mediterranean Sea: a
satellite analysis. Biogeosciences, 6, 139-148.
Gabrié C., Lagabrielle E., Bissery C., Crochelet E., Meola B., Webster C., Claudet J., Chassanite
A., Marinesque S., Robert P.,Goutx M., Quod C. 2012. Statut des Aires Marines Protégées
en mer Méditerranée. MedPAN & CAR/ASP.Ed: MedPAN Collection. 260 pp.
GBIF. 2012. Recommended practices for citation of the data published through the GBIF Network.
Version 1.0 (Authored by Vishwas Chavan), Copenhagen: Global Biodiversity Information
Facility. Pp.12
Hartigan, J.A., Wong, M.A., 1979. A k-means clustering algorithm. JR Stat. Soc., Ser. C, 28, 100-
108.
Halpern, B.S., Walbridge, S., Selkoe, K.A., Kappel, C.V., Micheli, F., D'Agrosa, C., Bruno, J.F.,
Casey, K.S., Ebert, C., Fox, H.E., 2008. A global map of human impact on marine
ecosystems. Science, 319, 948.
Hirzel, A.H., Le Lay, G., Helfer, V., Randin, C., Guisan, A., 2006. Evaluating the ability of habitat
suitability models to predict species presences. ecological modelling, 199, 142-152.
PERSEUS Deliverable Nr. 1.6
-41-
Intergouvernemental Panel on Climate Change (I.P.C.C)., 2013. Impacts, Adaptation, and
Vulnerability. Cambridge, United Kingdom and New York, NY, USA: Cambridge University
Press.
Legendre, P., Legendre, L., 1998. Numerical Ecology. Amsterdam: Elsevier Science BV.
Longhurst, A., 2007. Ecological geography of the Sea. London: Academic Press.
MEDAR Group, 2002 - MEDATLAS/2002 database. Mediterranean and Black Sea database of
temperature salinity and bio-chemical parameters. Climatological Atlas. IFREMER Edition
Millenium Ecosystem Assessment (M.E. A.), 2005. Ecosystems and human well-being: Island
Press.
Mouillot, D., Albouy, C., Guilhaumon, F., Ben Rais Lasram, F., Coll, M., Devictor, V., Meynard,
C.N., Pauly, D., Tomasini, J.A., Troussellier, M., 2011. Protected and threatened
components of fish biodiversity in the Mediterranean Sea. Current Biology, 21, 1044-1050.
Moutin, T., Van Wambeke, F., Prieur, L., 2012. Introduction to the Biogeochemistry from the
Oligotrophic to the Ultraoligotrophic Mediterranean (BOUM) experiment. Biogeosciences, 9,
3817-3825.
Oliver, M.J., Glenn, S., Kohut, J.T., Irwin, A.J., Schofield, O.M., Moline, M.A., Bissett, W.P., 2004.
Bioinformatic approaches for objective detection of water masses on continental shelves.
Journal of Geophysical Research, 109, C07S04.
Odum, E.P., 1971. Fundamentals of ecology. Philadelphia: Saunders.
Pauly, D., Christensen, V., Dalsgaard, J., Froese, R., Torres, F., 1998. Fishing down marine food
webs. Science, 279, 860-863.
Phillips, S.J., Dudík, M., Schapire, R.E., 2004. A maximum entropy approach to species
distribution modeling. Proceedings of the twenty-first international conference on Machine
learning (p. 83): ACM.
Quackenbush, J., 2001. Computational analysis of microarray data. Nature Reviews Genetics, 2,
418-427.
Reygondeau, G., Beaugrand G., Guidi L., Stemann L., Koubbi P., Henson S. A., Maury, O.in
revision. First approach for a biogeography of the mesopelagic layer. Plos One.
Reygondeau, G., Longhurst, A., Martinez, E., Beaugrand, G., Antoine, D., Maury, O., 2013.
Dynamic biogeochemical provinces in the global ocean. Global Biogeochemical Cycles, 27,
1046-1058.
Sherman, K., 2005. The Large Marine Ecosystem Approach for Assessment and Management of
Ocean Coastal Waters. Sustaining Large Marine Ecosystems: The Human Dimension, 3-
16.
Spalding, M.D., Fox, H.E., Allen, G.R., Davidson, N., FerdaÑA, Z.A., Finlayson, M.A.X., Halpern,
B.S., Jorge, M.A., Lombana, A.L., Lourie, S.A., 2007. Marine Ecoregions of the World: A
Bioregionalization of Coastal and Shelf Areas. BioScience, 57, 573-583.
PERSEUS Deliverable Nr. 1.6
-42-
Sverdrup, H.U., 1953. On Conditions for the Vernal Blooming of Phytoplankton. ICES Journal of
Marine Science, 18, 287-295.
Thuiller, W., Lafourcade, B., Engler, R., Araújo, M.B., 2009. BIOMOD–a platform for ensemble
forecasting of species distributions. Ecography, 32, 369-373.
UNEP-MAP-RAC/SPA, 2010b. Fisheries conservation and vulnerable ecosystems in the
Mediterranean open seas, including the deep seas. By S. De Juan and J. Lleonart, Ed.
RAC/SPA, Tunis, 103p.
van der Spoel, S., 1994. A biosystematic basis for pelagic biodiversity. Bijdragen tot de Dierkunde,
64, 3-31.
Vierros, M., Bianchi, G., & Skjoldal, H. R. (2008). The Ecosystem Approach of the Convention on
Biological Diversity. The Ecosystem Approach to Fisheries, Berlin Heidelberg, Federal
Environmental Agency.
Yeung, K.Y., Haynor, D.R., Ruzzo, W.L., 2001. Validating clustering for gene expression data.
Bioinformatics, 17, 309-318.
Ward, J.H., 1963. Hierarchical grouping to optimize an objective function. Journal of the American
statistical association, 58, 236-244.
5.2. Presentation to International conferences
April 2013 European Geosciences Union 5egu° General Assembly, Vienna, Austria
G. Reygondeau, J-O. Irisson, C. Guieu, S. Gasparini, S. D. Ayata & P. Koubbi. Toward a
dynamic biogeochemical division of the Mediterranean Sea in a context of global climate
change. Oral communication.
October 2013 CIESM, Marseille, France
G. Reygondeau, J-O. Irisson, C. Guieu, S. Gasparini, S. D. Ayata & P. Koubbi.
Ecoregionalisation of the Mediterranean sea: Environmental, ecological features and
potential threats. Oral communication.
June 2014 IMBER, Bergen, Norway
G. Reygondeau, J-O. Irisson, C. Guieu, S. Gasparini, S. D. Ayata & P. Koubbi.
Ecoregionalisation of the Mediterranean sea: Environmental, ecological features and
potential threats. Oral communication.
September 2014 ICES, Malaga, Spain
G. Reygondeau, C. Albouy, T. Hattab, F. Benedetti, J O Irisson, S D Ayata, S Gasparini,
B Mckenzie C Guieu and P Koubbi.. Mediterranean biodiversity (from phytoplankton to top
predators) and present threats. Oral communication.
PERSEUS Deliverable Nr. 1.6
-43-
October 2014 Global challenge and sustainable solutions, Copenhaguen, Denmark
G. Reygondeau, C. Albouy, T. Hattab, F. Benedetti, J O Irisson, S D Ayata, S Gasparini,
B Mckenzie C Guieu and P Koubbi.. Mediterranean biodiversity (from phytoplankton to top
predators) and present threats. Oral communication.