Journal of Sea Research 57 (2007) 19 – 35
www.elsevier.com/locate/seares
Causes of variability in diatom and Phaeocystis blooms in Belgian
coastal waters between 1989 and 2003: A model study
Nathalie Gypens a,⁎, Geneviève Lacroix b , Christiane Lancelot a
b
a
Université Libre de Bruxelles, Ecologie des Systèmes Aquatiques, CP-221, Bd du Triomphe, B-1050 Brussels, Belgium
Management Unit of the North Sea Mathematical Model, Royal Belgian Institute for Natural Sciences, Gulledelle 100, B-1200 Brussels, Belgium
Received 3 April 2006; accepted 18 July 2006
Available online 29 July 2006
Abstract
Massive blooms of Phaeocystis colonies usually occur in the Belgian coastal zone (BCZ) between spring and summer diatom
blooms but their relative magnitude varies between years. In order to understand this interannual variability, we used the
biogeochemical MIRO model to explore the link between diatom and Phaeocystis blooms and changing nutrient loads and
meteorological conditions over the last decade. For this application, MIRO was implemented in a simplified 3-box representation
of the domain between the Baie de Seine and the BCZ. MIRO was run over the 1989–2003 period using actual photosynthetic
active radiation (PAR), seawater temperature and riverine nutrient loads as forcing. The water mass residence time was calculated
for each box based on a monthly water budget estimated from 1993–2003 water flow simulations of the three-dimensional
hydrodynamical model COHSNS-3D. Overall MIRO simulations compare fairly well with nutrient and phytoplankton data
collected in the central BCZ but indicate the importance of the hydrodynamical resolution frame for correctly describing the
extremely high nutrient concentrations and biomass observed in the BCZ. Analysis of model results suggests that while interannual
variability in diatom biomass depends on both meteorological conditions (light and temperature) and nutrient loads, Phaeocystis
blooms are mainly controlled by nutrients. Further sensitivity tests with varying N and P loads suggest that only N reduction will
result in significantly decreased Phaeocystis blooms without negative affects on diatoms, while P reduction will negatively affect
diatoms. Moreover, Atlantic nutrient loads play such a great role in BCZ enrichment that reduction of Scheldt nutrient loads only is
not sufficient to significantly decrease phytoplankton blooms in the BCZ. It is concluded that future nutrient reduction policies
aimed to decrease Phaeocystis blooms in the BCZ without impacting diatoms should target the decrease of N loads in both the
Seine and the Scheldt rivers.
© 2006 Elsevier B.V. All rights reserved.
Keywords: North Sea; Eutrophication; Phaeocystis; Diatoms; Ecological modelling; Interannual variability
1. Introduction
Due to their impact on the marine ecosystem,
Phaeocystis colony blooms are generally reported as
undesirable (e.g. Smayda, 1990; Granéli et al., 1999).
⁎ Corresponding author.
E-mail address: [email protected] (N. Gypens).
1385-1101/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.seares.2006.07.004
Phaeocystis colonies have formed recurrent spring
blooms in the nutrient-enriched continental coastal
waters of the Southern Bight of the North Sea at least
since the early 1970s (Lancelot et al., 1987). In this area
Phaeocystis colonies bloom every spring between midApril and May–June (Rousseau, 2000; Cadée and
Hegeman, 2002; Rousseau et al., 2002) after an earlyspring diatom bloom but their magnitude varies from
20
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
year to year. This variability has been attributed to either
climatic changes (Owens et al., 1989) or changing
nutrient inputs due to human activity (Billen et al., 1991,
1999; Cadée and Hegeman, 2002). Interaction between
these driving forces has recently been suggested for the
Belgian coastal waters, based on a statistical analysis of
contemporary nutrient and phytoplankton time series
(Breton et al., 2006).
Biogeochemical models which are based on chemical
and biological principles and describe ecosystem carbon
and nutrient cycles as a function of environmental
forcing are ideal tools to investigate the link between
phytoplankton blooms and changing environmental
conditions. Such a model — the MIRO model (Lancelot
et al., 2005) — has been developed to describe diatom
and Phaeocystis blooms in the eastern Channel and
Southern Bight of the North Sea. In the present paper we
use the MIRO model to describe and understand the last
decade's variability in diatom and Phaeocystis colony
blooms in the Belgian coastal zone (BCZ) in response to
changing meteorological conditions and river loads. The
BCZ, located in the Southern Bight of the North Sea, is
appropriate to explore the dual role of natural and human
factors in Phaeocystis bloom magnitude. The BCZ is a
highly dynamic system with water masses resulting from
mixing between inflowing southwest Atlantic waters
through the Strait of Dover and Scheldt freshwater and
nutrient inputs (Fig. 1). Atmospheric deposition and
local benthic remineralisation also contribute to BCZ
enrichment. An annual budget of inorganic nitrogen
(DIN), phosphate (PO4) and dissolved silicate (DSi) has
been established on the scale of the BCZ, using riverine,
atmospheric and transboundary loads estimated for the
year 1992 (Rousseau et al., 2004). This budget evidences
the major contribution of transboundary loads to the
nutrient enrichment of the BCZ under average hydro-
Fig. 1. Map of the area studied with position of the monitored station
330 (⁎) and the MIRO multi-box frame.
dynamical conditions. Supporting this, De Jonge et al.
(1996) conclude that eutrophication of coastal waters in
the whole of the southern North Sea is not only related to
the nutrient loads of the local rivers but also to the
nutrient levels in the sea governed by the nutrient loads
from the Strait of Dover. This nutrient enrichment
fluctuates on an annual and a decadal scale, depending
on changes in human activity in the drainage basin
(Billen et al., 2005; Soetaert et al., 2006) and
meteorological conditions (rainfall and wind direction
and force). The latter also influences English Channel
water inflows to the BCZ, modifying the water mass
budget in the coastal zone (Reid et al., 2003; Breton et
al., 2006). Human pressure and rainfall conditions
determine river discharges and nutrient loads. The
geographical extent of the Scheldt plume in the BCZ is
mainly driven by wind force and direction (Yang, 1998;
Lacroix et al., 2004; Breton et al., 2006).
Between 1989 and 2003, monitoring of nutrients and
phytoplankton was performed in the central BCZ
(Station 330 51°26.05 N; 02°48.50 E, Fig. 1). The
usual diatom-Phaeocystis succession was clearly
observed every year but the magnitude of their bloom
showed marked interannual fluctuations (Breton et al.,
2006). Over this period, riverine PO4 loads to the North
Sea were reduced by 50% compared to the late 1980s
(OSPAR, 2003) but no direct responding trend in diatom
and Phaeocystis blooms could be identified from the
phytoplankton time-series at station 330 (Breton et al.,
2006). This lack of visible trends was recently related
with the variable influence of Scheldt nutrient loads at
station 330, modulated by meteorological conditions
and nutrient discharges (Breton et al., 2006).
In the present paper we describe simulations over the
1989–2003 period of the seasonal cycles of nutrients (N,
P, Si) and phytoplankton (diatom and Phaeocystis) in
the BCZ as obtained with the multi-box MIRO model
constrained with Global Solar Radiation (GSR), seawater temperature and actual riverine nutrient loads.
The MIRO simulations are compared with monthlyaveraged nutrient and phytoplankton observations
recorded at station 330 between 1989 and 2003 with
focus on diatom and Phaeocystis blooms which
represent more than 90% of phytoplankton in the
investigated area (V. Rousseau, pers. comm., 2006).
Model results are further analysed to understand the link
between the variability of diatom and Phaeocystis
blooms and the changing riverine nutrient loads and
meteorological conditions. Finally, as a first step towards
Phaeocystis mitigation, N or P reduction scenarios are
run to better understand how nutrient loads control the
magnitude of diatom and Phaeocystis blooms.
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
2. Material and methods
2.1. Model description
The ecological MIRO model describes C, N, P and Si
cycling through aggregated components of the planktonic and benthic realms of the Phaeocystis-dominated
ecosystem. It includes 38 state variables assembled in
four modules describing the dynamics of phytoplankton
(diatoms, nanoflagellates and Phaeocystis colonies),
zooplankton (copepods and microzooplankton), dissolved and particulate organic matter (each with two
classes of biodegradability) degradation and nutrients
(NO3, NH4, PO4 and Si(OH)4) regeneration by bacteria
in the water column and the sediment. Equations and
parameters are formulated based on current knowledge
on the kinetics and the factors controlling the main autoand hetero-trophic processes involved in the functioning
of the coastal marine ecosystem. These are fully
documented in Lancelot et al. (2005). The mathematical
formulation of processes is available at www.int-res.
com/journals/suppl/appendix_lancelot.pdf. MIRO was
first calibrated for climatological conditions (1989–
1999) of river loads, GSR and temperature; the
prediction capability of the model was demonstrated
by its ability to reproduce the SW-NE nutrient
enrichment gradient observed from the Western Channel
to the Belgian coastal water as well as the mean seasonal
nutrient and ecological features recorded in the central
BCZ during the last decade (Lancelot et al., 2005).
2.2. Model runs
The MIRO model was implemented in a multi-box
frame delineated on the basis of the hydrological regime
and river inputs. In order to take into account the
cumulated nutrient enrichment of Atlantic waters by the
Seine and Scheldt rivers, three successive boxes,
assumed to be homogeneous (the ‘oceanic’ Western
Channel WCH, the French Coastal Zone FCZ and the
BCZ; Fig. 1), were chosen in the domain between the
Baie de Seine and the northern limit of the BCZ. Each
box has its own morphological characteristics (see Table
1 in Lancelot et al., 2005) and is treated as an open
system, receiving waters from the SW adjacent box and
exporting water to the NE. The model was then run
successively in the WCH, the FCZ influenced by the
Seine and Atlantic waters from the WCH, and, finally, in
the BCZ influenced by the Scheldt loads and Atlantic
waters from the FCZ. Monthly residence time of water
mass in the FCZ and BCZ was calculated based on
monthly budgets of water masses (Atlantic water and
21
Table 1
Yearly-averaged stoichiometry (mol:mol) of Scheldt nutrient loads for
selected years of the 1989–2003 period
1990 (dry)
1994 (wet)
1996 (dry)
1999 (wet)
DIN :PO4
Si :PO4
DIN :Si
42
72
57
76
11
22
16
29
0.24
0.28
0.26
0.38
freshwater discharge of Seine and Scheldt, respectively)
calculated from 1993–2003 simulations of the threedimensional hydrodynamical model COHSNS (Lacroix
et al., 2004). Results obtained for the BCZ, assuming
that the Scheldt waters spread over the entire box
domain, showed large monthly variations in water
residence time of between 13 and 216 days over the
1993–2003 period, with lowest and highest values in
winter and summer, respectively. For the years 1989 to
1992 (no hydrodynamical simulations available), a
monthly climatological water residence time computed
from the 1993–2003 values was imposed.
Model simulations were performed over the 1989–
2003 period after a 3-y spin-up run. The boundary
conditions were provided by the results of the calculations performed for the conditions existing in the WCH
(Fig. 1), considered to be a quasi-oceanic closed system.
The seasonal variation of the state variables was
calculated by solving the different equations expressing
mass conservation according to the Euler procedure. A
time step of 15 min was adopted for the computation of
the numerical integration.
Forcing functions included the actual daily GSR,
seawater temperature and Seine and Scheldt nutrient
loads for the 1989–2003 period. However, daily
climatological forcing (computed for the 1996–2003
period) was used for seawater temperature between
1989 and 1994 due to insufficient data. The surface
incident Photosynthetic Active Radiation (PAR) was
calculated from the daily GSR data collected over
1989–2003 at the Oostende station of the Royal Institute
of Meteorology in Belgium making use of the empirical
relation: GSR = 3.43 10− 6 PAR2 + 0.0805 PAR (Rousseau et al., 2000) in which GSR is in Jcm− 2 30 min− 1
and PAR in μmole quanta m− 2 s− 1. Seawater temperature was derived from weekly sea surface temperature
obtained from the Bundesamt fuer Seeschiffahrt und
Hydrographie (Lacroix et al., 2004). River Seine
freshwater and nutrient inputs were obtained from
bimonthly measurements at the downstream monitoring
station ‘Caudebec’ described in Billen et al. (2001).
Scheldt nutrient loads were calculated making use of
nutrient concentrations at the station Doel (51°21 09 N;
22
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
04°13 50 E) and runoff at the upstream station Schelle.
The latter was multiplied by an empirically determined
correction factor of 1.15 in order to include lateral
freshwater inputs between the Schelle and Doel stations
(Rousseau et al., 2004). A monthly average was used for
the daily river forcing in the model applications and a
correction factor was applied in order to take into
account the modulating effect of wind on the river
plume extension in the BCZ box. This factor was
estimated from 3D-COHSNS simulations (Lacroix et
al., 2004) as the ratio between the influence of Scheldt
loads on the BCZ and the entire North Sea, respectively,
and ranged between 0.27 and 1 for the 1993–2003
period. For the 1989–1992 period, a full spreading of
the river in the BCZ is assumed.
standard deviation of the annual mean derived from
monthly (annual) data; n is the number of monthly
(annual) values; r is the correlation between Mt and Dt.
Validation assessment is taken from Radach and Moll
(2006) rating 0 < C < 1 as ‘very good’; 1 < C < 2 as
‘good’; 2 < C < 3 as ‘reasonable’ and C > 3 as poor. The
cost function calculation was, however, limited to the
1992–2000 data set which has the necessary time
resolution.
Anomalies, computed as the difference between
actual and mean forcing or MIRO phytoplankton values,
were used to assess the extent to which the different
forcings determined the phytoplankton interannual
variability simulated by the model. The mean MIRO
values corresponded to those obtained when a climatological forcing was used for the 1989–2003 period.
2.3. Model validation
3. Results and discussion
Station 330 located in the central BCZ (51°26.05 N;
002°48.50 E; Fig. 1) was chosen for the assessment of
MIRO results. For the 1989–2000 period, nutrient and
phytoplankton (Chl-a and diatom and Phaeocystis
carbon biomass) data were obtained from the high
time resolution monitoring at the station 330 (Rousseau, 2000, i.e. a weekly sampling frequency except
during winter (two per month)). This time series was
completed with available 2001–2003 data of nutrients
and Chl-a downloaded from the Belgian Marine Data
Centre (http://www.mumm.ac.be/datacentre/) and 2001
Phaeocystis and diatom carbon biomass (Antajan et al.,
2004). Nutrient and phytoplankton analysis methods
are detailed in Rousseau et al. (2002).
The goodness of fit between 1989–2003 MIRO
simulations was estimated based on the cost function C
recommended by OSPAR (Villars et al., 1998) and
discussed in Radach and Moll (2006):
C¼
RjMt −Dt j=n ⁎
ð0:5 þ 0:5ð1−rÞÞ
SD
In which:
Mt and Dt are, respectively, the model and data
monthly (annual) value at station 330; SD is the
3.1. Interannual variations in incident light, temperature and riverine nutrient loads over the 1989–2003
period
The year-to-year variation of daily-averaged PAR
over the photoperiod (Fig. 2) shows little variation with
maximum summer values ranging between 1100 and
1340 μmol quanta m − 2 s − 1 in 1991 and 2003,
respectively. Annual mean PAR over the period is
about 500 μmol quanta m− 2 s− 1. Annual anomaly
calculation for the 1989–2003 period indicates up to
10% variability between years and shows 2003 and
1998 to be the most sunny and cloudy year, respectively.
The year-to-year variation of seawater temperature
shows larger variation for winter temperatures than for
summer maxima (Fig. 3). Winter temperature minima
fluctuate between 1.2 and 6.6 °C, while summer
maxima vary between 18.6 and 21.9 °C. Annual
anomaly calculation shows up to 13% variability
between years, with 1996 and 1999 being the coldest
and warmest year, respectively.
The 1989–2003 variation of Scheldt nutrient loads in
Fig. 4 results from fluctuations in both the river
Fig. 2. 1989–2003 evolution of daily-averaged Photosynthetic Active Radiation (PAR) over the photoperiod.
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
23
Fig. 3. 1989–2003 evolution of daily surface water temperature.
discharge (Fig. 5a) and the downstream river nutrient
concentrations (not shown). Over the investigated
period, the monthly Scheldt discharge varied between
31 and 500 m3 s− 1 (Fig. 5a). Annual anomalies (Fig. 5b)
were strongly positive in 1994, 1995, 1999, 2000 and
2001 and negative in 1990, 1991, 1996, 1997 and 2002.
This calculation allows identifying particularly wet
(2001) or dry (1996 and 2002) years (Fig. 5b) where
annual anomalies show up to 71% and 35%, respec-
tively, of variability compared to the 1989–2003 mean
discharge.
Monthly total N, NO3 (Fig. 4a) and Si (Fig. 4c)
Scheldt loads show strong interannual variations that
significantly correlate with monthly Scheldt discharge
(NO3: r2 = 0.95,p < 0.01; Si: r2 = 0.96, p < 0.01; total N:
r2 = 0.85, p < 0.01). The decrease observed in total P
and PO4 (Fig. 4b) loads between 1989 and 2003 clearly
reflects reduced nutrient discharge due to waste water
Fig. 4. 1989–2003 evolution of monthly Scheldt nutrient loads: total N, NH4 and NO3 (a), total P and PO4 (b) and SiO (c).
24
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
Fig. 5. 1989–2003 evolution of (a) monthly Scheldt discharges and (b) annual discharge anomalies.
treatment and substitution of PO4 in washing powders
(Billen et al., 2001, 2005). Similarly NH4 loads
decrease after 1991–1992 due to waste water treatment
(Fig. 4a). As expected from their point source origin,
no significant correlation is to be found between PO4
and NH4 loads and the river discharge. The decrease in
PO4 discharge has a significant impact on the N:P and
Si:P balance of nutrients delivered to the sea (Table 1).
Considering a molar N:Si:P stoichiometry of 16:16:1
for marine phytoplankton (Redfield et al., 1963;
Brzezinski, 1985), an increasing over-enrichment of
NO3 relative to PO4 loads is observed over the
investigated period (Table 1). The N:P and Si:P ratios
of Scheldt waters discharging into the BCZ have more
than doubled between 1980 and 2002 (Soetaert et al.,
2006). This NO3 excess is exacerbated at high river
flow (Table 1) because NO3 loads strongly depend on
water discharge. Interestingly, the Si:P ratio in the river
Scheldt approached potential PO4 limitation in the
second half of the investigated period, especially for
wet years (Table 1). Over the simulated period the Si:N
ratio varied between 0.2 and 0.4 for the river Scheldt
(Table 1).
3.2. MIRO simulations of nutrients and phytoplankton
over the 1989–2003 period
Fig. 6 compares 1989–2003 MIRO simulations of
nutrient concentrations (Fig. 6a–c), total phytoplankton
(mg Chl-a m−3, Fig. 6d) and diatom (Fig. 6e) and
Phaeocystis (Fig. 6f) carbon biomass with available
observations collected at station 330 (Fig. 1). As a
general trend, MIRO results describe reasonably well
the order of magnitude and the timing of the observed
seasonal and interannual variations in nutrients and
phytoplankton.
Supporting visual comparison, the cost functions
calculated to evaluate the seasonal and interannual
resolving capability of MIRO to reproduce observed
nutrient and phytoplankton data are reported in Tables 2
and 3. Based on the ranking of Radach and Moll (2006),
the cost functions calculated for DIN, PO4 and Si(OH)4,
Chl-a, diatoms and Phaeocystis are ‘good’ to ‘very
good’, except for PO4 simulations in 1992 and 1996,
which are classified as ‘reasonable’ (Table 2). The
generally ‘very good’ or ‘good’ rating obtained suggests
that model results are valuable for capturing seasonal
signals as well as interannual trends in nutrient
enrichment and algal blooms. These are discussed in
details below.
3.2.1. Nutrients
MIRO maxima of winter DIN (NO3 + NH4) range
between 25.5 in 1992 and 36 mmol m− 3 in 1996 (Fig.
6a). Comparison with DIN Scheldt loads suggests that
the large variability of river DIN loads (Fig. 4a) is not
directly propagated in MIRO simulations. For instance,
the highest modelled winter value of 1996 corresponds
with dry winter (Fig. 6a) when river discharge is low.
Yet, the highest observed concentrations are not well
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
Fig. 6. MIRO simulations (solid line) and observations (◊) of DIN (a), Si(OH)4 (b), PO4 (c), Chl-a (d), diatoms (e) and Phaeocystis (f).
25
26
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
Table 2
Cost function (C) for seasonal resolving of MIRO simulations of
nutrients and phytoplankton at station 330
Year
1992
1993
1994
1995
1996
1997
1998
1999
2000
Cost function C
DIN
PO4
Si(OH)4
Chl-a
Diatom
Phaeocystis
1.9
0.65
0.84
1.36
0.74
0.94
0.75
0.98
0.73
2.5
1.3
1.1
0.72
2.0
0.8
0.76
0.66
0.88
0.8
0.59
0.85
1.4
1.6
0.99
1.26
1.71
0.83
1.5
0.84
0.88
1.44
0.39
1.09
0.7
1.06
1.21
0.67
1.08
1.3
1.95
0.42
0.66
0.67
1.0
0.67
0.67
0.66
0.31
0.83
0.52
0.64
0.46
Calculation years are defined by data availability.
captured by model simulations, especially in 1997 and
1998 when observations report up to 40–55 mmol m− 3
(Fig. 6a). Conversely, for the 1994 and 1995 wet years,
observed and MIRO winter DIN concentration are not
particularly elevated (Fig. 6a). In fact, due to a shorter
residence time of Scheldt waters in the BCZ, the model
simulates relatively lower winter DIN concentrations
during wet years despite higher Scheldt nutrient loads
(Figs. 4a, 6a).
Similarly, the model also underestimates the observed
winter concentrations of Si(OH)4 for years of low
Scheldt discharge (Figs. 6b, 5c). Highest model values
fluctuate between 8.5 and 14.1 mmol m− 3 in 1998 and
1996, respectively, (Fig. 6b), slightly underestimating
the observed maxima of 17 mmol m− 3 (Fig. 6b).
As a general trend, the model quite well reproduces
the measured maximum winter PO4 concentration.
MIRO maximum winter concentrations of PO4 fluctuate
between 1.7 and 1.2 mmol m− 3 in 1989 and 2002,
respectively, (Fig. 6c), suggesting a small decreasing
trend of PO4 enrichment in the BCZ as observed in the
data set. Trend curves calculated for the evolution of
PO4 concentration show similar decrease trend for data
(slope = − 0.024) and model results (slope = − 0.03) between 1989 and 2003.
Overall, the simplified description of hydrodynamics
and river forcing in MIRO seem to preclude the right
simulation of observed maximum winter concentrations.
It is therefore suggested that the monthly scale of the
river input forcing and the related water mass residence
time estimation might be too long for considering the
short-term hydrological variations resulting from the
wind-forced and tidally dependent freshwater influence
captured in the data set.
The timing and magnitude of the spring nutrient
decrease (Fig. 6a–c), concomitant with the phytoplank-
ton bloom development (Fig. 6d) observed between
March and May are well described by MIRO simulations. Observed minimal concentrations of PO4 (Fig. 6c)
and DIN (Fig. 6a) in spring are fairly well simulated.
However, summer MIRO DIN concentrations are
significantly higher than observations, with most
measured and MIRO concentrations being respectively
less than 10 and close to 20 mmol m− 3 until September
(Fig. 6a). In agreement with data, near-zero DIN
concentration is simulated in spring for years before
1994 after which DIN exhaustion has never been
reached again (minimal spring DIN between 3 and
8 mmol m− 3; Fig. 6a). Conversely, observed and
simulated minimal PO4 in spring are higher than
0.2 mmol m− 3 until 1992 and exhausted after that
year (Fig. 6c). The model fails correctly to reproduce the
observed spring decrease of Si(OH)4, except for a few
years at the beginning of the decade (e.g. 1990, 1992
and 1993) (Fig. 6b). The simulated Si(OH)4 minimum in
summer (2 and 8 mmol m− 3) generally overestimates
the data range (1.5–4 mmol m− 3; Fig. 6b). Overall, the
simulated summer nutrient consumption slightly underestimates the nutrient depletions estimated from data
(Fig. 6a–c).
3.2.2. Phytoplankton
In agreement with observations, the MIRO seasonal
cycle of Chl-a shows two phytoplankton blooming
events, one reaching elevated biomass in spring and one
of less importance in summer (Fig. 6d). The maximum
Chl-a simulated during the spring period fluctuates
between 13 and 21 mg Chl-a m− 3 (Fig. 6d) underestimating observed extremes reached in 1990, 1993,
1994 and 2000 (Fig. 6d). Summer phytoplankton blooms,
reaching concentrations of 5 and 12 mg Chl-a m− 3, are
generally in good agreement with data except in 1993,
1998 and 2000 when model results underestimate Chl-a
records (Fig. 6d).
Supporting observations, the model correctly reproduces the seasonal phytoplankton dynamics characterised, between March and July, by a diatomPhaeocystis-diatom succession (Fig. 6e, f). The seasonal
and interannual evolution of diatom biomass (Fig. 6e) is
Table 3
Cost function (C) for interannual resolving of R-MIRO simulations of
winter nutrients and phytoplankton spring maxima at station 330 for
the 1992–1998 period
Average winter concentration
NO3
PO4
Si(OH)4
1.17
0.74
1.28
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
reproduced reasonably well. The maximum diatom
biomass modelled in spring and summer ranges between
45 and 155 mgC m− 3 and 83 and 328 mgC m− 3,
respectively, with both lowest values in 1998 (Fig. 6e).
As observed for Chl-a, the model fails to reproduce the
extreme diatom biomass observed as for instance in
spring 1994 and summer 1998 and 2000 (Fig. 6e).
Each year Phaeocystis colonies are simulated
between the spring and summer diatoms and form the
bulk of the spring phytoplankton biomass (Fig. 6e, f).
While the timing of the Phaeocystis bloom is well
captured by the model, the observed range of amplitude
variability is less well described (Fig. 6f). However, the
ability of the MIRO model to properly describe the
interannual variability is difficult to assess from the
limited available data set compared to the Phaeocystis
dynamics. Growth and decay of Phaeocystis blooms in
the area are very transient (Breton et al., 2006) and it
might be that the monthly time scale of nutrient loads is
inadequate to reflect extreme events properly.
3.2.3. Primary production
Over the simulated period, the MIRO-derived
annual net primary production computed based on
the time-integration of phytoplankton (diatoms,
Phaeocystis and nanoflagellates) daily growth rates
varies between 150 and 232 gC m− 2 y− 1 without a
clear interannual trend (Fig. 7). These values agree
with the field-based annual estimates of 199 and
221 gC m− 2 y− 1 suggested by Joint and Pomroy
(1993) and Van Beusekom and Diel-Christiansen
(1994), respectively, for the Southern Bight of the
North Sea and for the simulated period. Moreover
MIRO primary productions are in agreement with
other model estimations (Lenhart et al., 1997; Moll,
1998; Moll and Radach, 2003; Skogen and Moll,
2000; Skogen et al., 1995, 2004). As observed in the
western Dutch Wadden Sea (De Jonge et al., 1996; De
27
Jonge, 1997; Cadée and Hegeman, 1993), our MIRObased primary production remains high despite a
significant decrease in the phosphorus loads (Fig. 4b).
Together, diatoms and Phaeocystis represent more
than 95% of the annual primary production and range
between 44 and 100 gC m− 2 y− 1 and between 92 and
158 gC m− 2 y− 1 respectively (Fig. 7). A particularly
high diatom production as simulated in 1994 and 1999 is
accompanied by low annual Phaeocystis production and
vice versa (e.g. 1997, 1998, 2000 and 2001). Altogether,
the contribution of Phaeocystis to the annual primary
production varies between 50 and 76% for the 1989–
2003 period, being the highest in the second half of the
simulated period (Fig. 7).
3.3. Relative contribution of PAR, seawater temperature
and nutrient loads to the year-to-year variability of
diatoms and Phaeocystis blooms
Nutrients, light availability and temperature constitute bottom-up controls of phytoplankton blooms and
show significant interannual variability during the
considered period (Figs. 2 and 3). The contribution of
each factor to the simulated 1989–2003 variability of
maximal spring diatoms, Phaeocystis and summer
diatom biomass is investigated based on the comparison
between actual MIRO 1989–2003 results and those
obtained by running MIRO with a constant forcing of
either PAR, temperature or nutrient loads averaged over
the simulated period.
Figs. 8 and 9 compare the spring/summer diatom and
Phaeocystis maximum biomass anomalies (computed as
the difference between model results obtained with
actual and mean forcing value) with light and
temperature anomalies corresponding to the blooming
period of each of the three phytoplankton groups. The
contribution of each forcing to the biomass variability of
spring/summer diatoms and Phaeocystis (not shown) is
Fig. 7. 1989–2003 evolution of MIRO annual diatom (grey block) and Phaeocystis (black block) production and the relative contribution of annual
Phaeocystis production to annual primary production (black dots).
28
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
Fig. 8. PAR anomalies (black diamonds, right axis) and contribution (bars, left axis) of incident PAR to the predicted variability of (a) maximum
spring diatom, (b) Phaeocystis and (c) summer diatom biomass.
given by the ratio multiplied by 100 of the calculated
anomaly normalised to the corresponding maximum
biomass obtained under average forcing. PAR anomalies computed for the spring diatom bloom period vary
between − 124 in 1998 and +112 μmole quanta m− 2 s− 1
in 2003 (Fig. 8a), i.e. some 22% of the spring mean
PAR. Except in 1989, the maximum spring diatom
biomass is positively and significantly related to PAR
anomaly (Fig. 8a; r2 = 0.63, p = 0.012). The minimal and
maximal biomass deviations are simulated in 2001 and
in 1990, respectively (Fig. 8a), suggesting that PAR can
explain between −38% (24 gC m− 3) and + 50% (69 gC
m− 3) variability of the simulated spring diatom maximum. Conversely, no relationship exists between the
variability of maximum summer diatoms and Phaeocystis biomass and their respective PAR anomalies (Fig.
8b, c). In spite of considerable anomalies, PAR
variability explains less than 20% of the simulated
variability of Phaeocystis maximal biomass and the
significant PAR anomalies recorded in 1989, 1990 and
1998 have little impact on the bloom magnitude
deviations (Fig. 8b). This allows concluding that
incident light variability during Phaeocystis growth
has little impact on the magnitude of their bloom. The
contribution of PAR variability to the simulated biomass
variability of summer diatom biomass is more complex
(Fig. 8c). In general, a positive PAR anomaly has a
positive effect on the simulated maximum biomass of
summer (Fig. 8c), but the reverse is not true. The lack of
a positive relationship between the simulated variability
of summer diatom (Fig. 8c) and Phaeocystis (Fig. 8b)
biomass and the corresponding PAR anomalies could be
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
Fig. 9. Temperature anomalies (black diamonds, right axis) and
contribution (bars, left axis) of seawater temperature to the predicted
variability of maximum (a) spring diatom, (b) Phaeocystis and (c)
summer diatom biomass.
explained by the added control of temperature and/or
nutrients.
Due to the use of a daily climatological forcing
before 1995, the impact of temperature variability on the
diatom and Phaeocystis bloom variability is analysed
only for the 1995–2003 period (Fig. 9). Temperature
anomalies during the spring diatom bloom (Fig. 9a) vary
between − 3.2 and + 1.1 °C, i.e. 41% of the mean.
Because spring diatoms adapt so easily to low
temperatures (Lancelot et al., 2005), a positive spring
diatom anomaly is associated with negative temperature
anomalies while the reverse holds for positive tempera-
29
ture anomalies (r 2 = − 0.75, p = 0.019; Fig. 9a). In
contrast, a positive temperature anomaly during the
summer diatom bloom contributes positively to the
maximum biomass variability (r2 = 0.91, p = 0.001; Fig.
9c), in spite of results obtained in 2000. While
temperature greatly influences diatoms, this forcing
has little effect on the variability of the maximal
Phaeocystis biomass and no clear relationship holds
between the simulated anomaly of Phaeocystis biomass
and that of temperature (Fig. 9b). Altogether for the
considered period, temperature anomaly can explain a
maximum of 10% of the simulated maximum of
Phaeocystis biomass. This corresponds to the coldest
year 1996 when the Phaeocystis biomass deviation
reaches − 96 gC m − 3 for a negative temperature
anomaly of 1.2 °C (Fig. 9b).
Beside meteorological conditions (light and temperature), nutrient loads are expected to explain much of
phytoplankton dynamics and variability in eutrophied
waters. Fig. 10 compares the 1989–2003 nutrient-driven
deviations of maximal spring diatoms, Phaeocystis and
summer diatom biomass with the corresponding
anomalies of winter nutrient loads to the BCZ (inflowing Atlantic and Scheldt river waters). In Fig. 10,
phytoplankton biomass deviations correspond to the
difference between actual MIRO 1989–2003 results and
those obtained by running MIRO with nutrient loads
averaged over the simulated period and applying a
corresponding monthly averaged water residence time.
Clearly, deviations of N and Si winter loads are
positively correlated (r2 = 0.96, p = 0.0001; Fig. 10a)
while P anomalies show anti-correlation with respect to
N and Si up to 1992, after which they evolve similarly to
N and Si (Fig. 10a). This shift in 1992 is to be related to
the large decrease in P discharges from point sources in
the early 1990s (Billen et al., 2001, 2005). The highest
positive and negative nutrient anomalies are computed
for the wet 1995 (+ 2.2 mmol P d− 1, + 65 mmol N d− 1,
+ 18 mmol Si d− 1; Fig. 10a) and dry 1996 (− 1.6 mmol P
d− 1, − 41 mmol N d− 1 , − 11 mmol Si d− 1; Fig. 10a)
years.
Nutrient deviations are suggested to explain up to
25% (50 mgC m− 3 y− 1), 20% (210 mgC m− 3 y− 1) and
137% (190 mgC m− 3 y− 1) of the variability of the
maximum biomass reached by spring diatoms (Fig.
10b), Phaeocystis (Fig. 10c) and summer diatoms (Fig.
10d), respectively. Overall, phytoplankton biomass
anomalies are positively related to nutrient load
deviations but the most likely nutrient responsible
differs from year to year. For instance, Phaeocystis
variations are positively linked with N anomalies except
in 1989 when a positive Phaeocystis biomass deviation
30
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
to P anomalies (1989 to 1992) or negatively to Si or N
anomalies in 1993 (Fig. 10a, b). Summer diatoms
anomalies are either related to P (1989 and 1993) or Si
(1990 and 1992) anomalies (Fig. 10a, d).
Beside bottom-up controls, zooplankton predation
could add to the interannual variability of the only
diatom blooms, Phaeocystis colonies not being grazed
by copepods in the simulated domain (Lancelot et al.,
2005). Over the simulated period, grazing on diatoms is
significantly correlated with diatom production
(r2 = 0.96, p = 0.001). Grazing thus modulates diatom
production but cannot explain the diatom-Phaeocystis
interannual variability.
Summarising, the analysis of our simulations indicates incident light and temperature as important
triggers of the diatom spring bloom while variations of
these forcings during the simulated period had little
impact on Phaeocystis growth. This is caused by the
easy adaptation of Phaeocystis to a wide spectrum of
light conditions (Schoemann et al., 2005). Conversely,
the variations of nutrient loads are suggested to impact
on the magnitude of both diatoms and Phaeocystis
blooms but in a complex pattern. This should be related
to the changing N:P:Si balance of nutrient loads during
the simulated period and to the nutrient assimilation
characteristics of Phaeocystis and spring and summer
diatoms.
3.4. Nutrient limitation of diatom and Phaeocystis
blooms over the simulated period
Fig. 10. (a) Winter nutrient (Ntot ( , left axis), Ptot (◊, right axis) and
Si (♦, left axis)) load anomalies, (b) contribution of nutrient load to the
predicted variability of maximum spring diatom, (c) Phaeocystis and
(d) summer diatom biomass.
is associated with a positive P anomaly (Fig. 10a, c). For
diatoms, the link between biomass deviations and
nutrient anomalies is unclear and a great nutrient
anomaly does not necessarily result in significant
biomass deviation (Fig. 10a, b, d). Globally since
1994, the deviations of spring and summer diatom
biomass are positively linked with all nutrient anomalies
(except in 2000) (Fig. 10a, b, d). Before 1994, the
variability of spring diatoms is either positively related
Which nutrient limits the growth of spring diatoms,
Phaeocystis, or summer diatoms can be found from
MIRO daily simulations of nutrients and nutrient uptake
characteristics specific to each phytoplankton group.
The nutrient (N, P and Si) limiting factors were therefore
computed based on MIRO nutrient concentration at the
bloom maximum of each of the three phytoplankton
groups, making use of the phytoplankton-specific halfsaturation constant for each nutrient assimilation
(Lancelot et al., 2005, www.int-res.com/journals/suppl/
appendix_lancelot.pdf). The limitation factor is then
expressed as 1 minus the calculated nutrient limitation
and varies between 0 (no limitation) to 1 (full
limitation). Fig. 11 shows the 1989–2003 variation of
the N, P and Si limitation factors calculated for the
spring diatoms (Fig. 11a), Phaeocystis (Fig. 11b) and
summer diatoms (Fig. 11c) blooms. As expected, spring
diatoms are never limited by N availability (Fig. 11a).
Before 1994, co-limitation of Si and PO4 holds, except
in 1990 when Si appears to be the most limiting nutrient
(Fig. 11a). After 1994, the spring diatom growth is
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
31
N during the dry years 1997 and 1998 and by P during
the wet years 1994, 1995 and 1999 (Fig. 11b). Finally,
the nutrient limitations computed for summer diatom
blooms seem to be less severe and similar for all
nutrients, but PO4 is more likely than Si to be the
nutrient potentially limiting such summer blooms after
1989 (Fig. 11c).
3.5. Nutrient loads in the BCZ and consequences for
nutrient reduction policy
Fig. 11. Evolution of MIRO values of the nutrient limitation function
for (a) spring diatoms, (b) Phaeocystis, and (c) summer diatoms over
the 1989–2003 period. N: grey solid line with grey diamonds, SiO:
black solid line with black diamonds and PO4: black dashed line with
empty diamonds.
limited by PO4 availability with a maximum value of
0.75 in 1997 (Fig. 11a). The increased importance of
PO4 over Si limitation of spring diatoms after 1993 is to
be related to the observed decreased trend in PO4 loads
(Fig. 4b) and the simulated winter enrichment of the
BCZ (Fig. 6c). A shift from N to P limitation is also
modelled for the Phaeocystis bloom in 1994. However,
this shift is followed by a transition period of N or P
limitation in 1997–2000 and again a clear P limitation
after 2000 (Fig. 11b). Moreover, limitation varies
between dry and wet years. This is particularly evident
for Phaeocystis growth, which is potentially limited by
The above nutrient limitation calculations indicate
the high sensitivity of the coastal phytoplankton
community structure to changing nutrient loads. Over
the simulated period, nutrient loads varied considerably
due to changing meteorological conditions and human
activity. While the former determine the water balance
in the BCZ, i.e. the mixing between SW inflowing
Atlantic and Scheldt waters, the latter relies on nutrient
discharges in the Seine and Scheldt watersheds with the
signature of the Seine nutrient inputs being propagated
in the BCZ via the entering Atlantic waters. The
importance of these two sources of nutrients for the
BCZ and for the simulated period is estimated based on
a nutrient budget of MIRO daily nutrient fluxes in the
BCZ. Fig. 12 compares the relative importance of N
(total N and NO3), P (Total P and PO4), and Si Atlantic
loads with those of the Scheldt in annual BCZ nutrient
enrichment between 1989 and 2003. Over the simulated period, the Scheldt contributes 10 to 30% of total
N (25 to 68 mmol N m− 3 y− 1) and P (0.9 to 3.8 mmol
P m− 3 y− 1) loads in the BCZ (Fig. 12a, b). NO3
constitutes 30 to 40% (48 to 80 mmol N m− 3 y− 1) of
Atlantic N loads and up to 90% (44 mmol N m− 3 y− 1)
of total N Scheldt loads (Fig. 12a). The direct
contribution of NO3 by the Scheldt represents 10 to
20% of total N loads in the BCZ and 20 to 47% the
total NO3 inputs to the BCZ. Conversely, PO4 Atlantic
loads constitute 60 to 75% (5.1 to 7.7 mmol P m− 3
y− 1) of total P Atlantic loads and 30 to 70% (0.55 to
0.84 mmol P m− 3 y− 1) of Scheldt loads (Fig. 12b).
These loads, therefore, represent less than 10 and 16%
of total P and PO4 loads, respectively, in the BCZ
during the period considered. Si Scheldt loads represent
20 to 40% (8.8 to 18 mmol Si m− 3 y− 1) of Si annual
loads in the BCZ (Fig. 12c).
Several MIRO simulations with changing nutrient
loads by the river Seine and/or Scheldt have only been
performed to explore the response of diatoms and
Phaeocystis to possible nutrient reduction scenarios
under realistic meteorological conditions met during the
1989–2003 period. Chosen scenarios therefore include
32
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
Fig. 12. Annual Atlantic (left block) and Scheldt (right block) loads to
the BCZ of (a) Ntot (black) and NO3 (grey), (b) Ptot (black) and PO4
(grey) and (c) Si(OH)4.
50% reduction of NO3 and/or PO4 Seine and/or Scheldt
loads. The resulting effect on phytoplankton blooms is
estimated by comparison with the nominal run under
1989–2003 actual conditions. Tables 4–6 show the
effect of nutrient reduction scenarios on the annual
diatom and Phaeocystis biomass for selected dry and
wet years corresponding to the first (high P loads) and
second (decreased P loads) half of the simulated period.
First, changing nutrient loads has no effect on the
diatom-Phaeocystis-diatom succession (not shown) but
acts on the magnitude of their blooms. The extent of the
phytoplankton bloom decrease depends on the target
river(s), nutrient(s) and phytoplankton group as N or P
reduction act differently on the diatom and Phaeocystis
biomass (Tables 4–6). According to model scenarios
applied to both the Seine and the Scheldt (Table 4), P
reduction results in a decrease of diatoms and this
decrease is particularly enhanced during wet years (30 to
40%, i.e. − 5.6 gC m− 3 in 1999 and − 8.4 gC m− 3 in
1994; Table 4). Conversely, Phaeocystis biomass is
generally little affected by PO4 reduction (< 12%, i.e.
∼ 3 gC m−3, for selected years, Table 4). N reduction
deeply affects Phaeocystis blooms by decreasing their
biomass by more than 38% (Table 4). The greatest
reduction of Phaeocystis biomass is, however, obtained
during the particularly dry years 1990 and 1996 (Table
4). Overall diatoms are less affected by N than P
reductions and an increasing biomass is even simulated
in 1996, i.e. when Phaeocystis had decreased considerably (Table 4). This suggests a strong competition
between Phaeocystis and diatoms for low PO4. The
combined NO3 and PO4 reduction decreases both
diatoms and Phaeocystis biomass. These drops are
similar (35–40%) during wet years while Phaeocystis
biomass seems to be more affected than diatoms during
dry years (Table 4). In order to better appraise the effect
of nutrient reduction on the phytoplankton blooms we
compared the obtained MIRO decreases of diatoms and
Phaeocystis biomass with observational errors. This
comparison shows that, except for the impact of the PO4
reduction on Phaeocystis biomass, the simulated
phytoplankton biomass decrease is significant (not
shown).
Model results obtained by changing nutrient loads
the Scheldt (Table 5) or the Seine (Table 6) show that the
effect of reduction in nutrient loads of only one river
(Seine or Scheldt) is lower than that obtained by
reducing nutrient loads in both rivers (Table 4). This is
Table 4
Effect of Scheldt and Seine nutrient reduction on annual diatom and Phaeocystis biomass in BCZ, expressed in %
PO4 reduction
1990 (dry)
1994 (wet)
1996 (dry)
1999 (wet)
NO3 reduction
PO4 + NO3 reduction
Diatoms
Phaeocystis
Diatoms
Phaeocystis
Diatoms
Phaeocystis
− 20
− 40
− 24
− 31
− 11
−6
− 11
−2
−9
− 22
14
− 22
− 43
− 38
− 43
− 34
− 20
− 37
− 11
− 32
− 41
− 36
− 31
− 34
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
33
Table 5
Effect of Scheldt nutrient reduction on annual diatom and Phaeocystis biomass in BCZ, expressed in %
PO4 reduction
1990 (dry)
1994 (wet)
1996 (dry)
1999 (wet)
NO3 reduction
PO4 + NO3 reduction
Diatoms
Phaeocystis
Diatoms
Phaeocystis
Diatoms
Phaeocystis
−5
−6
−3
−5
0
−3
−6
−1
−3
−3
0
−2
− 20
−3
−3
0
−7
−9
−4
−7
− 21
−4
−5
−5
particularly evident for Scheldt nutrient loads, whose
reduction affects phytoplankton blooms by less than
20% (Table 5). A 50% reduction of Scheldt NO3 loads
leads to a maximum decrease of annual Phaeocystis
biomass of 20% without affecting diatoms significantly
(Table 5). The 50% Scheldt PO4 reduction has little
impact on diatoms (≤6%, Table 5). Conversely, PO4
reduction in Seine loads greatly affects diatom biomass
(> 20%, Table 6). Comparison with results obtained by
reducing PO4 loads in the Seine and Scheldt (Table 4)
suggests that diatoms rely on river Seine-enriched
Atlantic waters inflows for their P requirements.
Similarly, NO3 reduction only in the Seine loads
(Table 6) leads to a maximum Phaeocystis decrease of
21 to 24% (i.e. 50% of the decrease computed with the
decrease in Seine and Scheldt NO3 loads).
Nutrient reduction scenarios suggest that only N
reduction would lead to a significant drop in Phaeocystis blooms without affecting diatoms. The latter are,
however, considerably decreased after P reduction. As
also suggested by Skogen et al. (2004), a decrease of P
loads with respect to present-day discharges will not
have a positive effect on the ecosystem. Moreover, due
to the importance of Atlantic nutrient loads (70 to 90%
of total N and P loads and 60 to 80% of Si loads) to the
BCZ, the reduction of Scheldt nutrient loads only is not
sufficient to significantly decrease nutrient concentrations and phytoplankton blooms in the BCZ. Rather, a
similar nutrient load reduction only in the Seine will
have a larger impact on diatom and Phaeocystis blooms
than can be obtained by reducing nutrients only in the
Scheldt.
4. Conclusions
The multi-box MIRO model has proved useful to
investigate the link between changing nutrient loads and
meteorological (light and temperature) conditions over
the past 13 years and diatom/Phaeocystis blooms in the
BCZ under direct and indirect influence of Scheldt and
Seine nutrient loads.
While diatom growth is highly sensitive to changes
in light, water temperature and nutrient loads, the
variability of Phaeocystis blooms is mainly caused by
nutrient loads. N loads (and mainly NO3 loads) control
the Phaeocystis growth (Lancelot et al., 1998; Rousseau, 2000; Breton et al., 2006) while P or Si loads limit
diatom blooming during the period considered. Coastal
eutrophication problems in the Southern Bight of the
North Sea are generally perceived as a shift from diatom
towards Phaeocystis dominance due to an over-enrichment of N and P with respect to Si, the latter being
considered a limiting factor for diatom growth. Our
model results now suggest that, as a result of decreased
P discharge during the period investigated (Billen et al.,
2001, 2005; Rousseau et al., 2004; Soetaert et al.,
2006), P is at present the limiting nutrient not only for
diatom but also for Phaeocystis growth in the BCZ. It is
shown that these two phytoplankton groups now
compete for P and, based on model scenarios, further
PO4 load reductions without decreasing N loads would
favour Phaeocystis. Only N reduction will result in a
significant drop in Phaeocystis blooms without negatively affecting diatoms. Moreover, due to the importance of Atlantic nutrient loads for BCZ enrichment,
Table 6
Effect of Seine nutrient reduction on annual diatom and Phaeocystis biomass in BCZ, expressed in %
PO4 reduction
1990 (dry)
1994 (wet)
1996 (dry)
1999 (wet)
NO3 reduction
PO4 + NO3 reduction
Diatoms
Phaeocystis
Diatoms
Phaeocystis
Diatoms
Phaeocystis
− 15
− 36
− 20
− 27
−6
−5
−8
2
−5
− 15
10
− 16
− 24
− 24
− 21
− 22
− 15
− 32
− 13
− 28
− 22
− 22
−2
− 23
34
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
reduction of Scheldt nutrient loads only is not sufficient
to significantly decrease phytoplankton blooms in the
BCZ.
These results confirm the importance of nutrient
limitation for the phytoplankton dynamics and the need
to consider all nutrient species when exploring the
impact of nutrient reductions on diatom and Phaeocystis
blooms. Moreover, future management of nutrient
discharge reduction aiming to at decrease Phaeocystis
blooms in the BCZ without impacting diatoms should
target a decrease in NO3 loads from both the Seine and
the Scheldt. However, the actual impact of a nutrient
reduction will vary due to meteorological conditions
(dry vs wet years). These first scenarios allow assessing
the effect of decreasing river nutrient loads on diatom
and Phaeocystis blooms in the coastal zone. However,
as such these scenarios do not consider the feasibility of
reaching the tested nutrient reductions. In future, the
coupled RIVERSTRAHLER-MIRO model (Lancelot et
al., in press) constitutes a powerful mathematical tool
for testing the effect of realistic scenarios of nutrient
reductions (point and diffuse sources) in the Seine and
Scheldt watershed on Phaeocystis blooms in the BCZ.
Acknowledgements
The present work is a contribution to the AMORE
(Advanced MOdeling and Research on Eutrophication)
project funded by the Belgian Program “Scientific
Support Plan for a Sustainable Development PolicySustainable Management of the North Sea” (contract #
EV-11-19) and to the LITEAU project N° CV04000018
≪Modélisation intégrée des transferts de nutriments
depuis les bassins versants de la Seine, Somme et Escaut
jusqu'en Manche-Mer du Nord≫ funded by the French
Ministry of Ecology and Sustainable Development
(MEDD). We are indebted to Dr V. Rousseau for
providing the 330 data base and fruitful comments on
model simulations. N. Gypens received financial support
from the ‘Fond pour la Formation à la Recherche dans
l'Industrie et dans l'Agriculture’ (FRIA, Belgium). We
are grateful to F. Colijn, M. Schartau and two
anonymous reviewers for their constructive comments
on the first version of the manuscript.
References
Antajan, E., Chrétiennot-Dinet, M.-J., Leblanc, C., Daro, M.-H.,
Lancelot, C., 2004. 19′-hexanoyloxyfucoxanthin may not be the
appropriate pigment to trace occurrence and fate of Phaeocystis:
the case of P. globosa in Belgian coastal waters. J. Sea Res. 52,
165–177.
Billen, G., Lancelot, C., Meybeck, M., 1991. N, P and Si retention
along the Aquatic Continuum from Land to Ocean. In: Mantoura,
R.F.C., Martin, J.-M., Wollast, R. (Eds.), Ocean Margin Processes
in Global Change, Dahlem Worshop Reports. Wiley, New York,
pp. 19–44.
Billen, G., Garnier, J., Deligne, C., Billen, C., 1999. Estimates of earlyindustrial inputs of nutrients to river systems: implication for
coastal eutrophication. Sci. Total Environm. 243–244, 43–52.
Billen, G., Garnier, J., Ficht, A., Cun, C., 2001. Modeling the response
of water quality in the Seine river estuary to human activity in its
watershed over the last 50 years. Estuaries 24 (6B), 977–993.
Billen, G., Garnier, J., Rousseau, V., 2005. Nutrient fluxes and water
quality in the drainage network of the Scheldt basin over the last 50
years. Hydrobiologia 540, 46–67.
Breton, E., Rousseau, V., Parent, J.Y., Ozer, J., Lancelot, C., 2006.
Hydroclimatic modulation of diatom/Phaeocystis blooms in the
nutrient-enriched Belgian coastal waters (North Sea). Limnol.
Oceanogr. 51, 1401–1409.
Brzezinski, M.A., 1985. The Si:C:N ratio of marine diatoms:
interspecific variability and the effect of some environmental
variables. J. Phycol. 21, 347–357.
Cadée, G.C., Hegeman, J., 1993. Persisting high levels of primary
production at declining phosphate concentrations in the Dutch
coastal area (Marsdiep). Neth. J. Sea Res. 31, 147–152.
Cadée, G.C., Hegeman, J., 2002. Phytoplankton in the Marsdiep at the
end of the 20th century; 30 years monitoring biomass, primary
production, and Phaeocystis blooms. J. Sea Res. 48, 97–110.
De Jonge, V.N., 1997. High remaining productivity in the Dutch
Western Wadden sea despite decreasing nutrient inputs from
riverine sources. Mar. Pollut. Bull. 34, 427–436.
De Jonge, V.N., Bakker, J.F., Van Stralen, M., 1996. Recent changes in
the contributions of river Rhine and North Sea to the eutrophication of the western Dutch Wadden sea. Neth. J. Aquat. Ecol. 30,
27–39.
Granéli, E., Carlsson, P., Turner, J.T., Tester, P.A., Béchemin, C.,
Dawson, R., Funari, E., 1999. Effects of N:P:Si ratios and
zooplankton grazing on phytoplankton communities in the northern Adriatic Sea: I. Nutrients, phytoplankton biomass, and
polysaccharide production. Aquat. Microb. Ecol. 18, 37–54.
Joint, I., Pomroy, A., 1993. Phytoplankton biomass and production in
the southern North Sea. Mar. Ecol. Prog. Ser. 99, 169–182.
Lacroix, G., Ruddick, K., Ozer, J., Lancelot, C., 2004. Modelling the
impact of the Scheldt and Rhine/Meuse plumes on the salinity
distribution in Belgian waters (southern North Sea). J. Sea Res. 52,
149–163.
Lancelot, C., Billen, G., Sournia, A., Weisse, T., Colijn, F., Veldhuis,
M., Davies, A., Wassmann, P., 1987. Phaeocystis blooms and
nutrient enrichment in the continental coastal zone of the North
Sea. Ambio 16, 38–46.
Lancelot, C., Keller, M., Rousseau, V., Smith Jr., W.O., Mathot, S.,
1998. Autoecology of the marine haptophyte Phaeocystis sp. In:
Anderson, D.A., Cembella, A.M., Hallegraef, G. (Eds.), NATO
Advanced Workshop on the Physiological Ecology of Harmful
Algal Blooms. NATO-ASI Series, Series G: Ecological Science,
vol. 41, pp. 209–224.
Lancelot, C., Spitz, Y., Gypens, N., Ruddick, K., Becquevort, S.,
Rousseau, V., Lacroix, G., Billen, G., 2005. Modelling diatom and
Phaeocystis blooms and nutrient cycles in the Southern Bight of
the North Sea: the MIRO model. Mar. Ecol. Prog. Ser. 289, 63–78.
Lancelot, C., Gypens, N., Billen, G., Garnier, J., Roubeix, V., in press.
Testing an integrated river-ocean mathematical tool for linking
marine eutrophication to land use: the Phaeocystis-dominated
N. Gypens et al. / Journal of Sea Research 57 (2007) 19–35
Belgian coastal zone (Southern North Sea) over the past 50 years.
AMEMR Proceedings. J. Mar. Syst., doi:10.1016/j.marsys.2006.
03.010.
Lenhart, H.J., Radach, G., Ruardij, P., 1997. The effects of river input
on the ecosystem dynamics in the continental coastal zone of the
North Sea using ERSEM. J. Sea Res. 38, 249–274.
Moll, A., 1998. Regional distribution of primary production in the
North Sea simulated by a three-dimensional model. J. Mar. Syst.
16, 151–170.
Moll, A., Radach, G., 2003. Review of three-dimensional ecological
modelling related to the North Sea shelf system: Part 1. Models and
their results. Prog. Oceanogr. 57, 175–217.
OSPAR, 2003. Rapport intégré OSPAR de 2003 sur l'état d'eutrophisation de la zone maritime OSPAR, basé sur la première
application de la Procédure exhaustive. Séries d'Eutrophisation.
OSPAR Commission, vol. 190. 56 pp.
Owens, N.J.P., Cook, D., Colebrook, M., Hunt, H., Reid, P.C., 1989.
Long term trends in the occurrence of Phaeocystis sp. in the NorthEast Atlantic. J. Mar. Biol. Ass. UK 69, 813–821.
Radach, G., Moll, A., 2006. Review of the three-dimensional
ecological modelling related to the North Sea shelf system: Part
2. Model validation and data needs. Oceanogr. Mar. Biol. Ann.
Rev. 44, 1–60.
Redfield, A.C., Ketchum, B.A., Richards, F.A., 1963. The influence of
organisms on the composition of sea-water. In: Hill, M.N. (Ed.),
The Sea. Wiley, New York, pp. 26–77.
Reid, P.C., Edwards, M., Beaugrand, G., Skogen, M., Stevens, D.,
2003. Periodic changes in the zooplankton of the North Sea during
the twentieth century linked to oceanic inflow. Fish. Oceanogr. 12,
260–269.
Rousseau, V., 2000. Dynamics of Phaeocystis and diatom blooms in
the eutrophicated coastal waters of the Southern Bight of the North
Sea. Ph.D. thesis, Université Libre de Bruxelles, Belgium,
unpublished.
Rousseau, V., Becquevort, S., Parent, J.Y., Gasparini, S., Daro, M.H.,
Tackx, M., Lancelot, C., 2000. Trophic efficiency of the planktonic
food web in a coastal ecosystem dominated by Phaeocystis
colonies. J. Sea Res. 43, 357–372.
Rousseau, V., Leynaert, A., Daoud, N., Lancelot, C., 2002. Diatom
succession, silicification and silicic acid availability in Belgian
coastal waters (Southern North Sea). Mar. Ecol. Prog. Ser. 236,
61–73.
35
Rousseau, V., Breton, E., De Wachter, B., Beji, A., Deconinck, M.,
Huijgh, J., Bolsens, T., Leroy, D., Jans, S., Lancelot, C., 2004.
Identification of Belgian maritime zones affected by eutrophication
(IZEUT). Towards the establishment of ecological criteria for the
implementation of the OSPAR Common Procedure to combat
eutrophication. Belgian Science Policy, Brussels, Final Report.
77 pp.
Schoemann, V., Becquevort, S., Stefels, J., Rousseau, V., Lancelot, C.,
2005. Phaeocystis blooms in the global ocean and their controlling
mechanisms: a review. J. Sea Res. 53, 43–66.
Skogen, M.D., Moll, A., 2000. Interannual variability of the North Sea
primary production: comparison from two model studies. Cont.
Shelf Res. 20, 129–151.
Skogen, M.D., Svendsen, E., Berntsen, J., Asknes, D., Ulvestad, K.B.,
1995. Modelling the primary production in the North Sea using a
coupled three-dimensional physical-chemical-biological ocean
model. Estuar. Coast. Shelf Sci. 41, 545–565.
Skogen, M.D., Søiland, H., Svendsen, E., 2004. Effects of changing
nutrient loads to the North Sea. J. Mar. Syst. 46, 23–38.
Smayda, T.J., 1990. Novel and nuisance phytoplankton blooms in the
sea: evidence for a global epidemic. In: Granéli, E., Sundström, B.,
Edler, L., Anderson, D.M. (Eds.), Toxic Marine Phytoplankton.
Elsevier, Amsterdam, pp. 20–40.
Soetaert, K., Middelburg, J.J., Heip, C., Meire, P., Van Damme, S.,
Maris, T., 2006. Long-term change in dissolved inorganic nutrients
in the heterotrophic Scheldt estuary (Belgium, The Netherlands).
Limnol. Oceanogr. 51, 409–423.
Van Beusekom, J., Diel-Christiansen, S., 1994. A Synthesis of Phytoand Zooplankton Dynamics in the North Sea environment.
Godalming: WWF_World Wide Fund for Nature, Godalming,
UK. 148 pp.
Villars, M., De Vries, I., Bokhorst, M., Ferreira, J., Gellers-Barkman,
S., Kelly-Gerreyn, B., Lancelot, C., Menesguen, A., Moll, A.,
Pätsch, J., Radach, G., Skogen, M., Soiland, H., Svendsen, E.,
Vested, H.J., 1998. Report of the ASMO modelling workshop on
eutrophication Issues, 5–8 November 1996. The Hague, The
Netherlands. OSPAR Commission Report, RIKZ, The Hague.
90 pp.
Yang, L., 1998. Modelling of hydrodynamic processes in the Belgian
coastal zone. Ph.D. Thesis, Katholieke Universiteit Leuven,
Belgium, unpublished.