This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/authorsrights

Author's personal copy

Progress in Oceanography 124 (2014) 12–27

Contents lists available at ScienceDirect

Progress in Oceanography journal homepage: www.elsevier.com/locate/pocean

Interannual control of plankton communities by deep winter mixing and prey/predator interactions in the NW Mediterranean: Results from a 30-year 3D modeling study P.A. Auger a,⇑, C. Ulses a, C. Estournel a, L. Stemmann b, S. Somot c, F. Diaz d,e a

Université de Toulouse 3, CNRS/INSU, Laboratoire d’Aérologie (LA), UMR 5560, 14 avenue Edouard Belin, Toulouse, France Sorbonne Universités, UPMC Université Paris 06, Laboratoire d’Océanographie de Villefranche (LOV), UMR 7093, Observatoire océanologique, Villefranche/mer, France Météo-France-CNRS, Centre National de Recherches Météorologiques (CNRM-GAME), 42 avenue Coriolis, Toulouse, France d Aix-Marseille Université, CNRS/INSU, IRD, Mediterranean Institute of Oceanography (MIO), UM 110, Marseille, France e Université de Toulon, CNRS/INSU, IRD, Mediterranean Institute of Oceanography (MIO), UM 110, La Garde, France b c

a r t i c l e

i n f o

Article history: Received 24 August 2012 Received in revised form 10 April 2014 Accepted 11 April 2014 Available online 19 April 2014

a b s t r a c t A realistic modeling approach is designed to address the role of winter mixing on the interannual variability of plankton dynamics in the north-western (NW) Mediterranean basin. For the first time, a high-resolution coupled hydrodynamic–biogeochemical model (Eco3m-S) covering a 30-year period (1976–2005) is validated on available in situ and satellite data for the NW Mediterranean. In this region, cold, dry winds in winter often lead to deep convection and strong upwelling of nutrients into the euphotic layer. High nutrient contents at the end of winter then support the development of a strong spring bloom of phytoplankton. Model results indicate that annual primary production is not affected by winter mixing due to seasonal balance (minimum in winter and maximum in spring). However, the total annual water column-integrated phytoplankton biomass appears to be favored by winter mixing because zooplankton grazing activity is low in winter and early spring. This reduced grazing is explained here by the rarefaction of prey due to both light limitation and the effect of mixing-induced dilution on prey/ predator interactions. A negative impact of winter mixing on winter zooplankton biomass is generally simulated except for mesozooplankton. This difference is assumed to stem from the lower parameterized mortality, top trophic position and detritivorous diet of mesozooplankton in the model. Moreover, model suggests that the variability of annual mesozooplankton biomass is principally modulated by the effects of winter mixing on winter biomass. Thus, interannual variability of winter nutrient contents in the euphotic layer, resulting from winter mixing, would control spring primary production and thus annual mesozooplankton biomass. Our results show a bottom-up control of mesozooplankton communities, as observed at a coastal location of the Ligurian Sea. ! 2014 Elsevier Ltd. All rights reserved.

1. Introduction Over recent decades, in situ observations of several biological indicators have suggested a major impact of climate variability on pelagic ecosystem dynamics. In the North Atlantic and northwestern (NW) Mediterranean Sea, temperature changes have been

Abbreviations: NW, north-western; 1D, one-dimensional; 3D, three-dimensional; CZCS, Coastal Zone Color Scanner; SeaWiFS, Sea-viewing Wide Field-of-view Sensor; MLD, mixed layer depth; HPLC, High Performance Liquid Chromatography; CV, coefficient of variation; FC, fractional change; PB, Point B. ⇑ Corresponding author. Present address: Institut de Recherche pour le Développement, Laboratoire de Physique des Océans, BP 70, 29280 Plouzané, France. Tel.: +33 298498662. E-mail address: [email protected] (P.A. Auger). http://dx.doi.org/10.1016/j.pocean.2014.04.004 0079-6611/! 2014 Elsevier Ltd. All rights reserved.

suspected to enhance regime shifts of plankton communities. For example, a regime shift from cold to warm biotopes, with a turning point in 1987, was described in the Atlantic Ocean and related to the North-Atlantic Oscillation and surface temperature anomalies in the Northern Hemisphere (Beaugrand and Reid, 2003). In the open NW Mediterranean Sea, decadal evolution of surface phytoplankton biomass has been suggested by satellite data and attributed to changes in winter mixing (Morel and André, 1991; Bosc et al., 2004). Moreover, several authors have suggested that a regime shift in zooplankton abundances occurred at a coastal site in the Ligurian Sea (Point B, PB, - bay of Villefranche-sur-Mer) in the late 1980s (Molinero et al., 2008; Berline et al., 2012). However, zooplankton abundances almost recovered their 1980s values after 2000, suggesting a decadal variability rather than a long-term

Author's personal copy

13

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

change (Garcia-Comas et al., 2011; Vandromme et al., 2011). Higher abundances of zooplankton in the 1980s and after 2000 were correlated with cold, dry winters, resulting in high winter mixing and nutrient enrichment of the Northern Mediterranean Current, which potentially strengthened the phytoplankton spring bloom as observed in the central Ligurian Sea (Marty and Chiavérini, 2010). The NW Mediterranean basin is usually characterized by cold, dry winds in winter, which trigger strong winter mixing in the MEDOC area (MEDOC-group, 1970), potentially extending into the Ligurian Sea (Marshall and Schott, 1999). The decadal variability of weather regimes influences local heat fluxes at the surface of the Mediterranean Sea (Rixen et al., 2005; Josey et al., 2011; Papadopoulos et al., 2012). Several authors have highlighted the major role played by latent heat fluxes at the ocean–atmosphere interface on the interannual variability of dense water formation in the open NW Mediterranean Sea (Mertens and Schott, 1998; Grignon et al., 2010; Herrmann et al., 2010; Béranger et al., 2010). Long-term trends of meteorological conditions and hydrology have been reported for the 1980s and 1990s with decreasing wind-speed and rainfall accompanied by deep water warming (Béthoux et al., 1998). In consequence, the interannual evolution of phytoplankton biomass in the open NW Mediterranean Sea could be driven by winter vertical mixing as suggested by in situ observations at the DYFAMED station in the central Ligurian Sea (Gernez et al., 2011; Marty and Chiavérini, 2010; Marty et al., 2002) and satellite data (Morel and André, 1991; Bosc et al., 2004). On the one hand, the winter vertical mixing is responsible for nutrient inputs that support primary production in the euphotic zone (spring bloom); on the other hand, vertical motions due to winter mixing limit the light availability for phytoplankton growth in winter (Sverdrup, 1953) and control the prey–predator interactions through vertical dispersion of plankton and organic detritus (Behrenfeld, 2010). Several modeling studies have investigated the dynamics of plankton communities in the NW Mediterranean (see review in Crise et al., 1999), including the use of realistic one-dimensional (1D) simulations validated on field data (Allen et al., 2002; Raick et al., 2005). However, three-dimensional processes such as lateral advection, upwelling and eddy transport can profoundly modify the simplified interpretations made from a 1D approach. Recently,

three-dimensional (3D) annual simulations were carried out in the NW Mediterranean (Herrmann et al., 2013) in the aim of describing the response of the plankton ecosystem and associated carbon cycle to the atmospheric and oceanic interannual variability. In the present study, we focus on the processes linking winter vertical mixing (i.e. turbulent mixing and winter deep convection) and interannual plankton dynamics in the NW Mediterranean Sea using a realistic 3D modeling approach. A high-resolution simulation (2.5 km) covering a long period (30 years) is validated on available observations in the NW Mediterranean Sea. The impact of low versus high winter mixing conditions in such a deep convection area is examined. The effect of winter mixing on winter and spring/summer plankton biomass is investigated to address the interannual relationships between winter mixing and plankton dynamics. 2. Material and methods 2.1. Coupled hydrodynamic–biogeochemical model A biogeochemical model (Eco3m-S, presented in Auger et al., 2011) was forced by the 3D primitive equations, sigma-coordinate, free surface SYMPHONIE hydrodynamic model described in detail by Marsaleix et al. (2009, 2011, 2012). This hydrodynamic model was used to reproduce coastal circulations induced by river discharges (Reffray et al., 2004) and wind (Hu et al., 2009) in the dense water formation of the Gulf of Lions shelf in winter (Estournel et al., 2005; Ulses et al., 2008; Herrmann et al., 2008a), the along-slope circulation at regional scale in the NW Mediterranean (Rubio et al., 2009; Bouffard et al., 2008) and offshore convection (Herrmann et al., 2008b). The biogeochemical model had been previously used to study the dynamics of plankton communities impacted by freshwater discharge and their role in carbon export in the Gulf of Lions (Auger et al., 2011). Specific calibrations were adapted to this particular ecosystem characterized by very high N/P ratios. This concerned zooplankton grazing rates, half-saturation constants for phosphate uptake, maximum internal ratios and quantum yields of phytoplankton. In the present paper, a new calibration of these parameters is derived to reproduce the plankton dynamics observed at the DYFAMED station (7"510 E/ 43"250 N) in the Ligurian Sea (Tables 1 and 2).

Table 1 Biogeochemical parameters of the Eco3 m-S model modified from the version of Auger et al. (2011). Phy1: picophytoplankton, Phy2: nanophytoplankton, Phy3: microphytoplankton, Zoo1: nanozooplankton, Zoo2: microzooplankton, Zoo3: mesozooplankton. Symbol

Description

Unit

Value

Phytoplankton

umax;Phyi

ðN=CÞmax Phyi ðP=CÞmax Phyi ðSi=CÞmax Phyi max ðChl=NÞPhyi kPO4 ;Phyi T REF Phy

T REF Zoo

Phy2

Phy3

1.6d!4

1.9d!4

2.6d!4

1, 2, c

Maximum quantum yield

mmolC J!1

Maximal internal N/C quota

molN molC

0.2

0.2

0.2

3, 4, 5

Maximal internal P/C quota

molP molC!1

0.019

0.019

0.019

4, 5, 6, 7

Maximal internal Si/C quota

molSi molC!1

-

-

0.19

5, 7

Maximal internal Chl/N quota

molChl molN!1

2.3

2.3

2.3

8, 9, c

Half saturation constant for PO4 Reference temperature

mmolP m!3 "C

0.005 15

0.015 15

0.05 15

7, 10, 11, c c

Zoo1

Zoo2

Zoo3

Maximum grazing rate Reference temperature

d "C

3.89 15

2.59 15

1.30 15

Reference temperature

"C

15

c

Reference temperature for remineralization

"C

15

c

!1

Zooplankton g Zooi

References

Phy1

!1

12, 13, c c

Bacteria T REF Bac Non-living matter T REF rem

Parameters of the biogeochemical model and references: (c) calibration; (1) Babin et al. (1996); (2) Claustre et al. (2005); (3) Heldal et al. (2003); (4) Riegman et al. (2000); (5) Geider et al. (1998); (6) Bertilsson et al. (2003); (7) Sarthou et al. (2005); (8) Geider et al. (1997); (9) Sondergaard and Theil-Nielsen (1997); (10) Tyrrell and Taylor (1996); (11) Timmermans et al. (2005); (12) Christaki et al. (2002); (13) Nejstgaard et al. (1997).

Author's personal copy

14

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

2.2. Simulation setup 2.2.1. Simulation domain The western Mediterranean basin includes regions with strongly contrasted trophic regimes (D’Ortenzio and Ribera d’Alcala, 2009) schematically organized as an east–west gradient. The water exchanges between the NW Mediterranean basin and the Tyrrhenian and Algerian basins are intense. Moreover, the biogeochemistry of the different water masses is poorly known. Specifying boundary conditions for a regional simulation, as described in Section 2.2.3, was thus a challenging task and the coupled hydrodynamic–biogeochemical model was implemented on a large domain with restricted open boundaries. The biogeochemical model was forced by a hydrodynamic simulation performed over the western Mediterranean basin (0"400 W–11"400 E and 36"250 N– 44"250 N, Fig. 1). Our analysis was then restricted to the domain 2"E–9"E and 40"N–44"250 N (Fig. 1, dotted box, hereafter called the study domain), and also delimited by the 1500 m isobaths to avoid the signal of coastal zones. Levy et al. (1998) have highlighted the strong impact of mesoscale instabilities around the convection area on the spring bloom in the MEDOC area. Considering the small values of the Rossby radius during convection, a high resolution, of 2.5 km, was adopted, approximately the same as was used by (Herrmann et al., 2008b) (3 km) and (Madec et al., 1991) (2 km) to capture the impacts of mesoscale structures on the characteristics of convection and subsequent restratification. 2.2.2. Hydrodynamic simulation A 30-year hydrodynamic simulation was produced over the period 1976–2005 (November 1975–November 2005) when in situ and satellite data were available for validation. Each year was simulated by the high resolution model SYMPHONIE (the first year, 1976, was simulated from November 1975 to November 1976.). Initialization and boundary conditions were provided by the NEMOMED8 model (Herrmann et al., 2010), a Mediterranean Table 2 Zooplankton grazing parameters of the Eco3m-S model modified from the version of Auger et al. (2011). Phy1: picophytoplankton, Phy2: nanophytoplankton, Phy3: microphytoplankton, Zoo1: nanozooplankton, Zoo2: microzooplankton, Zoo3: mesozooplankton, sPOM: small particulate organic matter.

uPrey;Zooi

Bacteria

Phy1

Phy2

Phy3

Zoo1

Zoo2

sPOM

Zoo1 Zoo2 Zoo3

0.35 0.08 0

0.65 0.06 0

0 0.25 0

0 0.20 0.5

0 0.25 0

0 0.12 0.45

0 0.04 0.05

configuration of the NEMO ocean model at 1/8" resolution. This simulation was forced by the atmospheric fluxes from the ARPERA dataset (Herrmann and Somot, 2008; Herrmann et al., 2010) obtained from a dynamic downscaling of the ECMWF (European Centre for Medium-Range Weather Forecasts) ERA40 re-analysis dataset. Herrmann et al. (2010) showed that NEMOMED8 (10 km resolution) forced by ARPERA (50 km resolution) was able to reproduce a realistic interannual variability of the Western Mediterranean Deep Water formation. However, it is worth noting that Herrmann et al. (2008b) demonstrated that a 10 km ocean resolution was not sufficient to simulate the meso-scale structures associated with such an event. Momentum, heat, solar and water fluxes from ARPERA were also used to force the high resolution SYMPHONIE simulation at a daily time scale. A relaxation term toward NEMOMED8 daily sea surface temperature was applied with an 8-day restoring time scale for the heat flux. The water flux was the sum of the evaporation-minus-precipitation flux without any relaxation or correction, plus the Rhône River daily runoffs (Compagnie Nationale du Rhône). 2.2.3. Biogeochemical simulation The simulation methodology was designed to focus on the interannual variability. Annual outputs of the hydrodynamic model were aggregated to produce a 30-year set of 3D daily current velocities, vertical diffusivities and temperatures that were used to force the biogeochemical model. For practical considerations linked to availability of computer resources, the 30-year biogeochemical simulation was split into six time-slices, so a possible decadal variability could not be highlighted. Each time-slice was initialized with the same 3D fields of biogeochemical variables obtained from a two-year ‘spin-up simulation’ performed using forcings for the year 1981 which was characterized by high winter mixing. The 2-year duration of the spin-up simulation appeared sufficient to achieve a stabilized 3D field of biogeochemical variables. The biogeochemical model was then run over five consecutive years. It was checked that the choice of the time-slices did not affect model results at the interannual scale by computing a second set of time slices shifted of 2 years: differences less than 2% were obtained. This very low sensitivity to initial conditions (except inorganic and organic stocks in deep and intermediate waters) can be attributed to the effect of winter mixing that produces huge changes in the nutrients and planktonic biomass distribution in the euphotic zone: the surface layer is replenished with nutrients while the planktonic biomasses fall at very small concentrations in winter as detailed in the following of this paper.

Fig. 1. Simulation domain (bold box) and study domain delimited by the 1500 m isobaths (dotted box) superimposed on a spatial distribution of satellite-derived chl a as reported by D’Ortenzio and Ribera d’Alcala (2009). See the location of the DYFAMED station (black cross) in the Ligurian Sea.

Author's personal copy

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

Simulations were initialized before the winter mixing (early November) for this reason. Furthermore, Herrmann et al. (2013) clearly established the impact of winter mixing on the intensity of the following spring bloom. Initial conditions for the spin-up simulation were constructed to meet two key requirements, mainly concerning the vertical and horizontal distribution of nutrients. First, the vertical structure of nutrient profiles in the convection area was important to obtain a realistic enrichment of surface waters when the mixed layer deepened. Second, it was essential to capture the horizontal gradients between periodically well-mixed regions of the NW Mediterranean Sea and the low-nutrient Northern Mediterranean Current, which participates in restratification between two convective events. The distribution of nutrients in the Ligurian and Gulf of Lions regions shows marked similarities with the doming distribution of density as observed by Sournia et al. (1990). Beside these large scale features, vertical velocities deduced from displacements of isopycnals in the Ligurian Sea allowed the short-term vertical variations of the nitracline and phosphacline to be successfully represented in a 1D coupled model (Raybaud et al., 2011). These tight links between density and nutrients led us to build an initial 3D field of nutrients for the spin-up simulation from the 3D field of density, as already done by Prieur and Legendre (1988). The relations between density and both nitrate and phosphate concentrations were deduced from observations made during the DYNAPROC2 cruise in September–October 2004 (Goutx et al., 2009). These relations were used to initialize nitrate and phosphate profiles from potential density simulated by the hydrodynamic model (½NO3 %ðzÞ ¼ 26:5625 ' rðzÞ ! 764:47 and ½PO4 %ðzÞ ¼ 1:6667 ' rðzÞ ! 48:1 with rðzÞ the potential density (sigma) in kg m!3, [NO3] and [PO4] in mmol m!3). They were also used to derive the boundary conditions. The initial profiles for all other biogeochemical variables were taken from a climatology of profiles for December observed at DYFAMED from1991 to 2005 (http:// www.obs-vlfr.fr/dyfBase/). A zero-gradient condition was applied at the open boundaries. At the bottom, nutrient fluxes at the sediment interface were obtained by coupling the biogeochemical model to a simplified version of a vertically-integrated benthic model described by Soetaert et al. (2000). 2.3. Interannual variability analysis The mixed layer depth (MLD) simulated by the model was computed from the daily averaged vertical diffusivity profiles using a threshold value of 4 10!4 m2 s!1 (Herrmann et al., 2008b). Stratification dominated from April to October over the study domain (not shown). The mixed layer then deepened generally below 100-m depth from January to March with maximum intensity in the Médoc area (MEDOC-group, 1970), in agreement with the MLD climatology derived from in situ profiles of temperature and salinity by D’Ortenzio et al. (2005). The spatial mean of MLD in winter can be considered, at first order, as a proxy of the winter vertical mixing intensity in the study region. MLD was therefore averaged over the study domain each year from early November to the end of March so as to construct an index characterizing the interannual variability of winter mixing on the NW Mediterranean basin for the period 1976–2005 (MLD index, Fig. 2a). Values of the MLD index depend on both the initial stratification of the water column in fall and the intensity of air– sea fluxes in winter (Grignon et al., 2010; Herrmann et al., 2010). The relationships between the interannual variability of the MLD index and the corresponding evolution of annual and seasonal means (winter/spring) of water column-integrated plankton biomasses, primary production and zooplankton grazing rates were studied via Pearson correlations. Temporal autocorrelation was taken into account by calculating the ‘‘effective’’ number of degrees of freedom for each correlation (see Pyper and Peterman, 1998)

15

Fig. 2. (a) Interannual evolution of the mean of mixed layer depth over the study domain (see Fig. 1) averaged each year (1976–2005) from early November to end of March (MLD index, m, black); idem for nitrate concentration in the 0–100 m surface layer (mmolN m!3, gray line). The intensity of the winter convection (strong/weak) as reported in the literature (*) is represented by gray areas (up/down). (b) Daily variation of the mean of mixed layer depth simulated at DYFAMED (no filtering and 30-day running mean) superimposed on monthly observations at DYFAMED (1991– 2005) and interannual MLD index. *Herrmann et al. (2008b), Mertens and Schott (1998), Herrmann and Somot (2008), Sannino et al. (2009), Marty and Chiavérini (2010), Béranger et al. (2010), Hong et al. (2007), Grignon et al. (2010), Canals et al. (2006) and Herrmann et al. (2010).

and adjusting significance levels accordingly (p < 0.05). No correlation could be found between the MLD index and advective fluxes through the lateral boundaries of the study domain (not shown). So, horizontal advection processes within the NW Mediterranean basin could not bias the interannual relationships presented here. However, note that the modeling protocol (time-slice method) used in this study may have affected the real interannual variability of the nutrient contents of water masses. 3. Model evaluation In this section, an evaluation of model outputs versus field observations is made over the study domain (see Fig. 1). The interannual evolution of the MLD index is compared with a review of in situ data of deep-water formation over the period 1976–2005 (Fig. 2a). Mixed layer depth simulated at DYFAMED is also compared with monthly in situ measurements (Fig. 2b, MLD calculated as the depth having a density increase of 0.015 kg m!3 below 10 m depth). The interannual variability of surface chlorophyll is also confronted with both CZCS (Coastal Zone Color Scanner) and SeaWiFS (Sea-viewing Wide Field-of-view Sensor) satellite data sets. Depth profiles of phytoplankton and zooplankton biomasses simulated by the model are compared with in situ data from DYFAMED (Ligurian Sea). 3.1. Interannual variability of the winter mixing The mixed layer depth simulated at DYFAMED for the period 1991–2005 is in fair agreement with observations (r ( 0.38,

Author's personal copy

16

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

p < 0.01; Fig. 2b). However, the interannual variability observed on monthly measurements at DYFAMED is different from that of the MLD index, which is based on daily outputs and constrained by the convection intensity in the MEDOC area. During the period 1976–2005, the MLD index shows marked interannual variability superimposed on decadal variability with strong winter mixing during the first half of the 1980s and relatively lower winter mixing in the 1990s (Fig. 2a). Except for the years 1976 and 2002, the interannual evolution of the MLD index agrees reasonably well with observational evidence and previous model results and in the NW Mediterranean (Fig. 2a). In particular, the winter of 1987 is characterized by strong mixing as observed by Krahmann (1997) and simulated by Herrmann et al. (2008a,b). The study by Sannino et al. (2009) shows that the 1990s was a period of weak convection. Moreover, in situ data of mixed layer depth in the Ligurian Sea over the period 1991–1995 (DYFAMED, Marty et al., 2002) reveal that winter mixing was stronger in 1995 compared to previous years. More recent observations (DYFAMED, 1995– 2007; Marty and Chiavérini, 2010; Béthoux et al., 2002) and model results (Hong et al., 2007; Grignon et al., 2010) indicate moderate winter mixing during the winters of 1997 and 1998 followed by strong winter mixing in 1999, 2000, 2003, 2004 and 2005. This is reflected by our MLD index except for 2000. The MLD index is also

consistent with the results of Béranger et al. (2010), which show stronger winter mixing in 1999 than in 2000, 2001 and 2002. Finally, the extreme convection that occurred during winter 2005 (Canals et al., 2006; Herrmann et al., 2010) and the southward extension of the deep water formation area simulated by the model and attested by Argo float data in the Catalan sub-basin (Smith et al., 2008) are consistent with the high MLD index. 3.2. Representation of plankton communities 3.2.1. Satellite-derived chlorophyll Model outputs of surface chlorophyll biomass were confronted to the daily time-series of satellite images from CZCS (1979–1983) and SeaWiFS (1998–2002). This data set has been analyzed by several authors (Morel and André, 1991; Bosc et al., 2004), and was recently reprocessed by Antoine et al. (2005) to identify long-term trends in the world ocean. Given a maximum uncertainty of satellite estimation when chlorophyll levels are high (see Antoine et al., 2005), observed concentrations were limited to the maximum value observed in the open NW Mediterranean (i.e. Chl < 5 mg m!3, see Antoine et al., 2008). Satellite data and surface model outputs were averaged over the study domain (see Fig. 1) and both are presented versus time on the same diagram (Fig. 3a and b). Given the

Fig. 3. (a and b) Time-evolution of the surface chlorophyll biomass (mgChl m!3) averaged over the study domain (see Fig. 1), as simulated by the model (red line) and estimated from CZCS and SeaWiFS satellite data (black dots, courtesy of D. Antoine) respectively over the periods (a) 1979–1983 and (b) 1998–2002; (c) Taylor diagram: radius is standard deviation, angle is correlation and distance from the origin is root mean square (pattern error); (d and e) maps of Pearson correlation between model and data. The Pearson correlations are significant over the whole domain (p < 0.05, not shown). Temporal autocorrelation is taken into account by adjusting the significance level at each grid point (see Pyper and Peterman, 1998). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Author's personal copy

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

uncertainty of satellite chlorophyll products (±35%, see Moore et al., 2009), the seasonal cycle of surface chlorophyll biomass is well reproduced by the biogeochemical model, with minima in summer (June–September) and winter (January–February), a fall bloom in October/November and an intense spring bloom beginning in late winter (Morel and André, 1991; Bosc et al., 2004). Model results appear generally in phase with satellite data. Although the spring bloom starts too early in 2000, both fall and spring blooms are correctly localized in time. Following spring peaks, the simulated decrease in phytoplankton biomass until summer matches satellite data well for 1979–1983 but it is stronger in 1998–2000. The summer biomass is thus reduced in the recent period. A comparison with DYFAMED in situ data (Fig. 4) indicates that this may be a bias on satellite-derived chlorophyll during summer, as already noted in oligotrophic waters (Bricaud et al., 2002; Morel et al., 2007). Otherwise, the intensity of fall and spring blooms are reasonably well reproduced by the model especially in the 1998–2002 period, for which more satellite data are available. Most of the differences previously noted between CZCS and SeaWiFS (Bosc et al., 2004) are represented by the model. For example, whereas modeled winter minimum biomass can approach summer values during 1979–1983 (especially in January 1980 and 1981), winter modeled biomass is found to be consistently higher than that of summer in 1998–2002 as observed in the satellite data. Thus, the fall bloom and winter minimum are clearly more marked in 1979–1983 than in 1998–2002, a period for which a more gradual increase in biomass from fall to spring is simulated. Moreover, the spring bloom of 1998–2002 is found to be approximately one month in advance compared to that of 1979–1983, in agreement with CZCS and SeaWiFS observations (Bosc et al., 2004). Maps of point-by-point linear correlation (Fig. 3c,d) are presented for both CZCS and SeaWiFS observations. The lowest correlations are found within convective regions (MEDOC area and Ligurian Sea), where sub-mesoscale structures cause rapid changes that are difficult to reproduce without a data assimilation scheme. However, correlations are always significant (p < 0.05, not shown) which ensures the spatio-temporal coherence between surface model results and satellite data. A Taylor diagram (r ! h polar plot, Taylor, 2001) provides a more quantitative comparison of the model simulation against satellite data (Fig. 3e). If the model was perfect, the result of the comparison would lie along the X-axis, right on top of the (data) reference point. Significant correlations

Fig. 4. Time-evolution of the surface chlorophyll biomass as simulated by the model (red line), estimated from satellite data over 1998–2002 (SeaWiFS data, black dots, courtesy of D. Antoine), and observed in situ over 1991–2005 (blue crosses) at DYFAMED. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

17

are found with both CZCS and SeaWiFS datasets (r ( 0.49 and 0.52 respectively). However, the normalized standard deviations of model results are 15–35% higher than the satellite data, which is probably due to the coarser resolution in satellite data ((20 km) compared to the model (2.5 km). 3.2.2. Depth profiles of plankton biomass in the Ligurian Sea In this section, model results are compared to a monthly data set of chlorophyll biomass from DYFAMED (1991–2005) derived from HPLC pigment analyses (Claustre, 1994). We used the pigment grouping methodology proposed by Claustre (1994) and Vidussi et al. (2001), and recently improved by Uitz et al. (2006), to determine the chlorophyll biomasses of the three main phytoplankton groups modeled. In order to give a synthetic view, we present the interannual variance of the three phytoplankton groups on seasonal and annual mean (Fig. 5), and the corresponding contributions of each phytoplankton group to the total 0–200 m integrated phytoplankton biomass (Fig. 6). The shapes of observed profiles are globally correctly reproduced by the model. During winter, phytoplankton biomass is low, and maximum at the surface. Nano-and microphytoplankton appear overestimated whereas picophytoplankton is underestimated. In spring/ summer, the surface maximum of both microphytoplankton and nanophytoplankton presents a weaker variability in the model than in observations. In fall, the depth of the deep maximum is globally underestimated by the model (bias ( 10–15 m). As in winter, picophytoplankton is underestimated when larger phytoplankton is overestimated. On the whole, the phytoplankton community structure is reproduced reasonably well by the model, with an overall domination of nanophytoplankton and microphytoplankton (Fig. 6). However, the model underestimates the picophytoplankton contribution in summer and fall, and globally overestimates the microphytoplankton contribution. In a second part, model results are compared to the very few zooplankton depth profiles and surface time series available for the study area (see review by Siokou-Frangou et al., 2010). In the absence of long-term interannual monitoring of nano- and microzooplankton biomass in the open NW Mediterranean, the modeled seasonal evolution is first compared to that observed at DYFAMED from May 1999 to February 2000 (Fig. 7) by Tanaka and Rassoulzadegan (2002). The seasonal evolution of both zooplankton groups is correctly represented by the model, especially for microzooplankton (ciliates). A surface minimum in January is followed by a regular increase leading to a surface maximum in May, which migrates deeper ((40 m) in August. The order of magnitude of surface biomass is correctly reproduced by the model despite an overestimation in several profiles, especially for nanozooplankton. Nano- and microzooplankton biomass then sharply decreases below 60 m depth, and appears to be underestimated by the model below 100 m. Zooplankton biomass has been observed to concentrate in the 0–100 m surface layer in the Ligurian Sea (Andersen et al., 2001; Nowaczyk et al., 2011). The positive vertical gradient of microzooplankton to nanozooplankton biomass ratio found throughout the water column (Tanaka and Rassoulzadegan, 2002) is consistent with an increasing contribution of large organisms at greater depths (Licandro and Icardi, 2009). Therefore, the main features of nano-/microzooplankton distribution with depth are represented by our model. The seasonal evolution of near surface depth-averaged mesozooplankton biomass in the model is also confronted with recent monthly measurements of copepod biomass at DYFAMED (Fig. 8, L. Stemmann, personal communication; see Vandromme et al., 2011 for technical information). First, observed copepods and simulated mesozooplankton biomass time-series for 2001–2003 are presented in Fig. 8a (0–200 m). Second, a random 5-year seasonal model climatology calculated over 1976–2005 is compared with

Author's personal copy

18

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

Fig. 5. Mean (black crosses) and standard deviation (black lines) of phytoplankton biomass depth profiles observed at DYFAMED (1991–2005, black) in winter (November– January), spring (February–April), summer (May–July), fall (August–October) and over the whole year (Annual). Superimposed are the mean (red lines) and standard deviation (red areas) of model profiles corresponding temporally to DYFAMED profiles. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 6. Contributions of each phytoplankton group to the total 0–200 m integrated phytoplankton biomass (%) simulated by the model and observed at DYFAMED (1991–2005) on seasonal and annual means.

DYFAMED data for 2005–2009 (0–100 m, Fig. 8b). The above comparisons show a consistent modeled seasonal pattern with fall/ winter minimum and spring/summer maximum (red lines in Fig. 8), as already observed in the Ligurian Sea (see review in Licandro and Icardi, 2009). However, there is no observational evidence of the biomass peak simulated by the model in late fall. Model results analysis show that repeated simulated intrusions of the Northern Mediterranean Current (Bouffard et al., 2008) containing lower zooplankton concentration at DYFAMED point lead to a zooplankton drop in fall ((2 months in advance). If we avoid such hydrodynamic bias by looking at model results 30 km farther offshore (8"050 E/43"110 N, black lines in Fig. 8), the observed fall drop is less pronounced and the late fall simulated peak is absent. The bias is then limited to the vicinity of the Northern Mediterranean Current. Nonetheless, the seasonal evolution of mesozoo-

plankton biomass as observed at the DYFAMED station is smoothed by the model, leading to overestimation of the winter minimum and underestimation of the summer maximum. Such a mismatch could arise from the underestimation of large zooplankton (>few mm) because of collection with a WP2 net of 200 lm mesh size (Vandromme et al., 2012) at a time when larger carnivorous organisms usually develop in the NW Mediterranean (Garcia-Comas et al., 2011). Overall, the mesozooplankton mortality is probably underestimated in fall/winter. This includes too simplistic a formulation of mesozooplankton mortality, which is the closure term of our model (see Section 5.3). The dynamics of both zooplankton and fish appear to be affected by predation feedback (Megrey et al., 2007; Travers et al., 2009) and the non-linear mechanisms linking fish and zooplankton biomasses are not parameterized in our model.

Author's personal copy

19

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

Fig. 7. Zooplankton biomass profiles (mgC m!3) (a) observed at DYFAMED by Tanaka and Rassoulzadegan (2002) and (b) simulated by the model from May 1999 to February 2000. Ciliates (microzooplankton) left panel, and heterotrophic nanoflagellates (HNF, nanozooplankton) right panel.

Table 3 Overall mean of the 30-year time-series of water column-integrated annual minimum, maximum and annual mean of plankton biomass (mmolC m!2), primary production (mgC m!2 d!1 for annual optimum; gC m!2 y!1 for annual mean), and composition of the plankton assemblages (size-class distribution).

Fig. 8. Comparison of the simulated mesozooplankton integrated biomass (mgC m!3, red line) with data available at DYFAMED (7"510 E/43"250 N) over the periods 2001–2003 (a; 0–200 m, black) and 2005–2009 (b; 0–100 m, blue): mean and standard deviation of the different replicates. Model climatology was built by randomly choosing 5 years from our 30-year simulation (1000 iterations) to compare with 2005–2009 data. Black lines represent the simulated mesozooplankton integrated biomass at a point 30 km offshore of the DYFAMED position (8"050 E/ 43"110 N). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

4. Results 4.1. Interannual variability of plankton community structure and production In this section, we aim to describe the interannual variability of integrated plankton biomasses and production as simulated by the model in the NW Mediterranean. This is done in terms of annual

Annual minimum Mean

Annual maximum Mean

Annual mean Mean

Picophytoplankton Nanophytoplankton Microphytoplankton Total phytoplankton Nanozooplankton Microzooplankton Mesozooplankton Total zooplankton Primary production

6.9 47.7 28.7 83.3 8.2 22.6 109.3 140.2 178

38.1 165.6 158.8 362.5 46.7 110.1 141.9 298.7 1231

18.2 96.1 73.8 188.1 31 69.8 127.5 228.3 282

Size-class

Mean (%)

Mean (%)

Mean (%)

Picophytoplankton Nanophytoplankton Microphytoplankton Nanozooplankton Microzooplankton Mesozooplankton

8.2 57 34.8 5.7 15.6 78.7

10.5 45.8 43.7 15.7 36.8 47.5

9.7 51.1 39.2 13.6 30.5 55.9

minimum, annual maximum and annual mean. Thirty-year timeseries are built from daily model outputs. Their overall mean over the 30-year simulation and the coefficient of variation (CV = standard deviation/arithmetic mean expressed in %) are computed and presented in Table 3 and Fig. 9 respectively. Conceptually, the annual minimum informs on the immediate impact of winter mixing, which produces a ‘‘blue hole’’ (chlorophyll-depleted area) on satellite images of the NW Mediterranean (Bosc et al., 2004). On the other hand, the annual maximum represents the signature and intensity of the phytoplankton spring bloom.

Author's personal copy

20

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

The interannual variability of integrated primary production and phytoplankton biomass appears weaker on annual mean (CV < 7.9%) than those at seasonal extrema (e.g. CV > 8.3% for annual minima) (Fig. 9). This result suggests a seasonal balance in the functioning of the phytoplankton community. The phytoplankton assemblage (Table 3) is generally dominated by nanophytoplankton (>45.8%) and microphytoplankton (>34.8%), the picophytoplankton contribution remaining low ( 33.4%), which suggests a strong impact of winter mixing on the smallest zooplankton groups. The zooplankton community structure is constantly dominated by mesozooplankton (55.9% on annual mean), then microzooplankton make up 33.8% and nanozooplankton 12.6% on annual mean of the total zooplankton biomass. Meso- and microzooplankton are respectively favored during winter mixing and spring bloom periods, when they form 78.7% and 36.8% of the total zooplankton biomass. The low influence of winter mixing on the annual minimum of mesozooplankton (CV ( 4.2%) explains the dominance of this group in the depth-integrated zooplankton community structure.

4.2. MLD index vs. plankton communities Here, we present the Pearson correlations between the interannual time-series of MLD index (see Section 3.1) and the corresponding time-series of annual and seasonal means (winter/ spring) of water column-integrated plankton biomasses, primary production and zooplankton grazing rates (Fig. 10). The seasonal cycles of these variables are plotted together and colored according to the value of MLD index for each year to illustrate the impact of winter mixing (Fig. 11). The fractional change between strong and weak winter mixing conditions (FC in %, see legend Fig. 11) is also shown in order to assess the impact of winter mixing on plankton biomass more quantitatively. Winter total phytoplankton biomass is negatively correlated with the MLD index (r ( !0.49, FC ( !6%) and low correlations are found within each phytoplankton group in winter. Nonetheless, fall/winter pico- and nanophytoplankton biomasses globally decrease when winter mixing increases (FC ( !9.8%). On the contrary, nano- and microphytoplankton are positively correlated in spring/summer (resp. r ( +0.45 and +0.67, FC ( +2.1% and +17.8%) suggesting a positive control of the spring bloom by winter mixing (Marty et al., 2002). The annual total phytoplankton and microphytoplankton biomasses are positively correlated with the MLD index (r > +0.55, resp. FC ( +4.2% and +12.6%), unlike the picoand nanophytoplankton groups (|r| < 0.29, |FC| < 1.4%). This result highlights the major role played by microphytoplankton at the interannual scale. A striking result is the absence of correlation between annual primary production and the MLD index (r ( +0.02, FC ( !0.5%), which is explained by a winter/spring compensation (resp. r ( !0.77 and 0.56, FC ( !12.5% and 2.5%). The annual nano- and microzooplankton biomasses, which control total zooplankton grazing, are both negatively correlated with the MLD index (r < !0.57, resp. FC ( !4.2%/!3.1%) particularly through winter and early spring dynamics (r < !0.72, resp. FC ( !20.8%/!18.8%). In contrast, the annual mesozooplankton biomass is positively correlated with the MLD index (r ( +0.60, FC ( +4.9%) owing to its fall/winter and spring/summer dynamics (resp. r ( +0.20 and 0.71, FC ( +1.6% and 6.9%). 5. Discussion 5.1. Impact of mixing-induced dilution on the phytoplankton biomass

Fig. 9. Coefficient of variation (%) of the 30-year time-series of annual minimum (blue triangle), maximum (green triangle) and mean (black circle) of water columnintegrated plankton biomass, primary production and composition of the plankton assemblage. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The intensity of the phytoplankton spring bloom in the open NW Mediterranean Sea is known to be favored by winter mixing through nutrient supply in the euphotic zone (Marty et al., 2002). On the other hand, the surface phytoplankton biomass in winter is limited by winter mixing. The vertical orbital motions and turbulence induced by convection disperse phytoplankton organisms down to the deep zone (Backhaus et al., 1999, 2003), thus reducing the phytoplankton biomass in the surface layer. During convective years, the spring bloom can be delayed by one month, as observed on satellite-derived chlorophyll (Morel and André, 1991; Bosc et al., 2004). Interestingly, the positive impact of winter mixing on the surface phytoplankton spring bloom (by nutrient supply in the euphotic layer) is mostly counterbalanced in the model by the negative impact of dilution in winter, as indicated by a null correlation between surface annual total phytoplankton biomass and the MLD index (not shown). In contrast, the annual water column-integrated phytoplankton biomass increases with winter mixing (Fig. 11). The winter water columnintegrated phytoplankton biomass is less affected by winter mixing than surface biomass (not shown) because of the redistribution of phytoplankton within the mixed layer. In the model, light limitation is the primary mechanism explaining phytoplankton decline

Author's personal copy

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

21

Fig. 10. Pearson correlation coefficients between the MLD index and annual (black), fall/winter (white) and spring/summer (gray) means of water column-integrated plankton biomass, primary production and total zooplankton grazing. The limit between fall/winter and spring/summer is defined in early March for nutrients and phytoplankton biomass/production, and in early April for zooplankton biomass/grazing. Annual correlations (black) can be explained by winter (white) and/or spring/ summer (gray) correlations similar to the MLD index (similar bars). The 95% significance of the Pearson correlation coefficients (p < 0.05, white diamonds) was calculated from the Pearson table and the number of samples (Sokal and Rohlf, 1995) taking into account autocorrelation (see Pyper and Peterman, 1998).

Fig. 11. Annual cycles of surface nutrient contents (0–100 m, mmolC m!3), water column-integrated phytoplankton and zooplankton biomasses (mmolC m!2), primary production and total zooplankton grazing rates (mgC m!2 d!1) colored according to the annual value of MLD index (blue for low MLD, red for high MLD). Average fractional changes (FC in %) between the 15 highest and the 15 lowest annual values of MLD index are indicated on winter, spring/summer and annual means. The separation between fall/winter and spring/summer seasons is fixed in early March for phytoplankton (beginning of the spring bloom) and early April for zooplankton (1-month lag with phytoplankton). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

in winter. Nevertheless, the winter minimum of surface phytoplankton biomass deduced from satellite data (‘‘blue hole’’) does not reflect the state of the water column-integrated biomass. Our analysis demonstrates that the interannual control of phytoplankton biomass by winter mixing cannot be inferred only from the surface dynamics analysis.

5.2. Interannual control of phytoplankton production, biomass and assemblage Model results suggest that the interannual variability of water column-integrated primary production is weak (282 ± 6 gC m!2 y!1, CV ( 2.2%) in the NW Mediterranean Sea despite a strong sea-

Author's personal copy

22

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

sonal signal (Fig. 11), with minimum values in winter (178 ± 38 mgC m!2 d!1, CV ( 21.2%) and maximum values in spring (1231 ± 102 mgC m!2 d!1, CV ( 8.3%). Such values fall within the range of monthly data collected in the Ligurian Sea (Marty and Chiavérini, 2002; 100–1800 mgC m!2 d!1 over 1993– 1999), and also satellite-based estimates (Morel and André, 1991; Bosc et al., 2004). Assessments of annual primary production budgets, based on measurements in situ and satellite data, have been previously proposed in the Western Mediterranean basin (see Table 4) but reliable information on the interannual variability of primary production remains scarce. In a first review of primary production data on the Mediterranean Sea, obtained mostly in the western basin, Sournia (1973) found a ‘‘stupefying’’ regularity of yearly estimates of between 80 and 90 gC m!2 y!1, in spite of the heterogeneity of sources, sampling strategy and experimental methods. SeaWiFS satellite-based estimates also show some regularity of annual primary production (297 ± 7 gC m!2 y!1 over 1998–2001, Bosc et al., 2004) which confirms a moderate interannual variability at least at the end of our study period. More recent measurements in situ in the Ligurian Sea by Marty and Chiavérini (2002) rather suggest a large interannual variability of primary production (86–232 gC m!2 y!1 over 1993–1999). Nevertheless, such estimates could have been biased by a monthly-limited sam-

Table 4 Observed and simulated primary production rates in the NW Mediterranean Sea during recent decades. Primary production (gC m!2 y!1) Model (this study) Sournia (1973) Marty and Chiavérini (2002) Bosc et al. (2004) Olita et al. (2011)

282 ± 6 (1976–2005) 80–90 (review of data before 1970) 86–232 (DYFAMED, 1993–1999) 297 ± 7 (SeaWiFS, 1998–2001) 150–180 (SeaWiFS, 1998–2006)

pling strategy causing a large loss of information in the interannual signal. In the model, no significant correlation was found between annual primary production and winter mixing intensity (MLD index). This is not due to an increase in annual primary production with winter MLD up to an ‘‘optimum’’ above which primary production decreases (not shown). Actually, a seasonal compensation between fall/winter and spring/summer likely explains the weak interannual variability of primary production in the model. The interannual variability of the phytoplankton community structure is controlled by winter mixing which relatively favors microphytoplankton compared to nanophytoplankton (resp. FC = 12.6%/!0.8%). Despite an under-(resp. over-) estimation of the contribution of pico-(resp. micro-) phytoplankton to total biomass in winter (Fig. 6), this analysis looks robust as the low biomass values in winter have little influence on the annual mean. This result is in line with the increase in the microphytoplankton contribution at the DYFAMED station observed during the extreme convection event of 2006 in the Ligurian Sea (see Fig. 6 and Marty and Chiavérini, 2010). High nutrient contents and intermittent access to light, induced by winter mixing have been observed to favor a microphytoplankton spring bloom (Kiorboe, 1993; Sarthou et al., 2005). Light limitation is uniformly increased by winter mixing, due to a deepening of the mixed layer (Fig. 12b). In contrast, zooplankton grazing limitation in winter/spring is reduced by winter mixing for every groups of phytoplankton (Fig 12c). Mesozooplankton grazing pressure on microzooplankton is also reduced by winter mixing (data not shown), so microzooplankton can exert a stronger control of predation on small-sized phytoplankton groups. The positive impact of winter mixing on the microphytoplankton spring bloom in the model thus results from higher growth rate, both due to lower mesozooplankton grazing and more efficient competition for nutrients driven by stronger grazing on the smaller-sized phytoplankton groups.

Fig. 12. (a) Annual cycles of zooplankton mortality rates (mmolC m!3)2 d!1 for mesozooplankton and mmolC m!3 d!1 for smaller groups); (b) light limitation and (c) grazing limitation terms for phytoplankton growth (no dimension) colored according to the annual value of MLD index (same as Fig. 11). Limitation terms saturate to 1 when growth is not limited and remain inferior to 1 when growth is limited. The light limitation term was calculated following equation (A17) of Table A4 in Auger et al. (2011) ! " PARðzÞ , and the grazing limitation term is the difference between primary production and zooplankton grazing rates normalized by the primary kd 2 1þsPhy *rPhy *PARðzÞþsPhy * k *ðrPhy *PARðzÞÞ i

i

i

r

i

production rate. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Author's personal copy

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

5.3. Influence of prey–predator interactions on plankton biomass and assemblage In the model, unlike primary production (FC ( !0.5%, see Fig. 11), the annual water column-integrated phytoplankton biomass is favored by winter mixing (FC ( +4.2%, see Fig. 11). The total zooplankton grazing rate is negatively correlated with the MLD index because of the negative effect of winter mixing on total zooplankton biomass and grazing by dilution of both prey and predators (Fig. 10). This would explain the discrepancy between primary production and total phytoplankton biomass, and suggests a significant influence of zooplankton top-down control on the interannual variability of phytoplankton biomass. Zooplankton grazing rates are parameterized in the biogeochemical model by a mechanistic Holling type II formulation (Gentleman et al., # $ 2003) Predator'Prey , with kg;Zooi the half-saturation constant for prey kg;Zoo þPrey i consumption (see Table 1). Considering a twofold reduction of prey and predator the zooplankton grazing rates become # biomasses, $ equal to 12 Predator'Prey and are thus reduced by a factor of more 2kg;Zoo þPrey i than two. Such impact of dilution on the zooplankton grazing rates is further enhanced at low prey level (phyto- and zooplankton), which makes its variation with winter mixing nonlinear. The prey–predator decoupling induced by winter mixing concerns all zooplankton groups. However, in winter, the water column-integrated nano- and microzooplankton biomasses are anticorrelated to winter mixing, while the water column-integrated mesozooplankton biomass presents a weak positive correlation with winter mixing (Figs. 10 and 11, FC ( +1.6%). The effect of dilution on zooplankton mortality (loss term of zooplankton biomass) is different between mesozooplankton and smaller zooplankton groups. Mortality is linear for nano- and microzooplankton but quadratic for mesozooplankton in the model. Quadratic mortality expresses the observed low specific mortality of large zooplankton compared to smaller size-classes (Banse and Mosher, 1980) which is attributed to the prey–predator interactions within the mesozooplankton community (including grazing on eggs) and the grazing pressure of higher predators that are not modeled (Fasham et al., 2006; Steele and Henderson, 1992). Thus, the mechanical dilution of zooplankton biomass over the mixed layer should reduce mesozooplankton mortality as compared to the other groups. However, this remains not significant (Fig. 12a) suggesting that lower mesozooplankton mortality cannot explain why mesozooplankton better resists to winter mixing compared to smallersize zooplankton groups (Fig. 11), nor the overestimation with respect to the data (Fig. 8). Top trophic position, including zooplankton and detritivorous diet, is therefore proposed. Model results show that the duration of the zooplankton minimum in winter increases with the zooplankton size-class (Fig. 11) in agreement with the few available observations (Ribera d’Alcala et al., 2004; Vandromme et al., 2011; Romagnan, 2013). In the model, nano- and microzooplankton respond faster than mesozooplankton to the phytoplankton biomass increase during spring, certainly because of their contrasted grazing rates and feeding diet. In consequence, the phytoplankton spring bloom supports first the buildup of nano- and microzooplankton populations, and then that of mesozooplankton through feeding on the growing stocks of microplankton and, to a lesser extent, on organic detritus. In the model results, the spring/summer water column-integrated mesozooplankton biomass is favored by winter mixing (FC ( +6.9%, see Fig. 11) which is not the case for nano- and microzooplankton (FC < +1.7%). In the model, mesozooplankton is fueled by large amounts of organic detritus produced during the phytoplankton spring bloom and by top-down control on microzooplankton as already suggested by coastal observations in the Ligurian Sea (Garcia-Comas et al., 2011; Vandromme et al., 2011). Prey–predator interactions and the consumption of organic

23

detritus by mesozooplankton are probably the key factors explaining the integrator role of mesozooplankton in the model (Drinkwater et al., 2010). The impact of favorable conditions in winter/spring on the mesozooplankton biomass is still observed in summer and fall, which tends to confirm the hypothesis that the spring/summer zooplankton biomass actually depends on winter biomass, as proposed by Vandromme et al. (2011) for a coastal station (PB) in the Northern Mediterranean Current. Long-term observations of zooplankton biomass in the open NW Mediterranean Sea are missing for the period 1976–2005. Data at PB during the periods 1974–2003 and 1995–2005 suggest that nutrient inputs by winter mixing are concomitant with a reduction of phytoplankton biomass through strong top-down control by zooplankton (Garcia-Comas et al., 2011 and Vandromme et al., 2011), as indicated by increasing annual mesozooplankton biomass and abundance. This trend on phytoplankton is opposite to that observed at DYFAMED, where winter mixing is known to favor phytoplankton (Marty and Chiavérini, 2010). In the model, the time-series at DYFAMED and PB both show a positive correlation between winter mixing and surface nutrient concentrations, column-integrated phytoplankton and mesozooplankton biomass (not shown). Nonetheless, the positive impact of winter mixing on phytoplankton biomass is less clear at PB (not shown). While nutrient replenishment is achieved by deep vertical mixing at DYFAMED, it probably relies on wind-induced turbulent mixing and coastal upwelling of intermediate water along the continental slope at PB, as observed near the inner edge of the Northern Mediterranean Current off Marseille (Conan et al., 1998). On the other hand, the effect of winter mixing on zooplankton grazing rates appears weaker at PB than at DYFAMED (not shown). The shallowness of the water column at PB (90 m) compared to DYFAMED (2600 m) is then proposed to limit prey–predator decoupling, resulting in a stronger top-down control on phytoplankton. Note that our model resolution is not sufficient to represent accurately the dynamics at PB. A better representation of the plankton ecosystem dynamics at PB, by increasing the model resolution, would be necessary to validate this hypothesis. From an analysis of a nine-year satellite record of phytoplankton biomass in the North Atlantic, Behrenfeld (2010) recently suggested that the prey/predator decoupling induced by winter mixing could be a key factor controlling the onset of the phytoplankton spring bloom as low zooplankton grazing implies a high potential for phytoplankton growth. In regions where the water column is well mixed, it was first suggested that the onset of the spring bloom at the end of winter was triggered by improved light, temperature and stratification conditions (‘‘Critical Depth Hypothesis’’, Sverdrup, 1953; Chiswell, 2011), and reduction of the turbulent mixing of air–sea fluxes (Taylor and Ferrari, 2011). However, Behrenfeld (2010) explained the occurrence of phytoplankton blooms in the absence of mixed layer shoaling in the North Atlantic by a balance between phytoplankton growth and grazing. The winter stock of phytoplankton actually forms the inoculum for spring production in the open ocean (Backhaus et al., 2003). In the model, the phytoplankton spring bloom starts in early January, with a peak in early April during convective years (Fig. 11). Among all the factors that can limit phytoplankton growth, the model indicates that low grazing in winter is the most significant for explaining the intensity of the phytoplankton spring bloom during convective years. In that sense, the impact of winter mixing on prey–predator interactions is likely to trigger the phytoplankton spring bloom in the model. 5.4. Potential impacts on services (fishing and carbon sequestration) Deep convection is the main factor explaining the interannual variability of organic carbon export below the euphotic layer in

Author's personal copy

24

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

Fig. 13. Impact of winter mixing on plankton biomass and carbon pathways as suggested by the model: (fall–winter mean)/(spring–summer mean) ? (annual mean). (+) Positive impact, (!) negative impact, (=) no impact.

the model (200 m, Auger, 2011; Herrmann et al., 2013). Deep export is mainly related to advection processes during winter/ spring, as confirmed by observations at DYFAMED (Heimbürger et al., 2013). The particulate organic carbon biomass, which is enhanced by winter mixing during this period (not shown), is then likely sequestered in deep waters. Although dissolved organic carbon fuels bacterial communities (Santinelli et al., 2010), organic detritus should be less remineralized in the deep ocean than in upper layers. Moreover, winter mixing increases the deep sequestration of atmospheric inorganic carbon in the NW Mediterranean in the form of larger, faster-sinking organic detritus produced by larger plankton communities (i.e. microphytoplankton and mesozooplankton, see Fig. 13). Although large particles form (2% of the total particulate organic carbon pool (see Herrmann et al., 2013), they potentially contribute (74% to the sedimentation on the seafloor owing to high sinking rates (not shown, see also Auger et al., 2011). Fast sinking also limits the residence time of particulate organic carbon in the water column, and so its remineralization. The consequence is that the alteration of the plankton ecosystem structure toward large-size communities during well mixed winters probably has a significant impact on organic carbon storage and the carbon cycle in the NW Mediterranean. A positive correlation has been found, with an 18-month lag, between sardine catches and wind mixing in the NW Mediterranean (Lloret et al., 2004). Small pelagic sardine biomass would be bottom-up driven by the intensity of winter mixing in the NW Mediterranean (Palomera et al., 2007), as supported by a fair comparison between the evolution of our MLD index and annual catches of sardines in the Catalan Sea during 1990–2004 (see Palomera et al., 2007). Not only sardine but also anchovy are identified as bottom-up flow control groups (controlling predators, Cury et al., 2000, Hunter and Price, 1992), meaning that top predators are also concerned by interannual changes in zooplankton. Both these commercially exploited fishes feed on zooplankton organisms, principally copepods, cladocerans, euphausids, crustacean larvae, fish eggs and phytoplankton (Blaxter, 1969; Demirhindi, 1961; Petrakis et al., n.d.; Rasoanarivo et al., 1991) although the typical prey remain copepods (Blaxter and Hunter, 1982; Conway et al., 1991; Plounevez and Champalbert, 2000). As a result, winter mixing would tend to further increase the growth rates of small pelagic larvae and fish from spring to fall

and then enhance recruitment (Palomera et al., 2007) through higher availability of large plankton cells in spring/summer.

6. Conclusions The main originality of the present work is to provide a realistic simulation of the interannual variability of plankton dynamics in the NW Mediterranean Sea over a long period (30 years), so as to complete the partial picture built up from sparse in situ data on nutrients, phytoplankton and zooplankton, and remote sensing synoptic satellite data. Our work first demonstrates that the control of interannual plankton dynamics by winter mixing cannot be inferred only by analyzing surface dynamics, because of strong mixing-induced dilution. Thus, a biogeochemical model validated on field observations in the upper layers (0–200 m) was used to address the variability of the plankton ecosystem integrated over the whole water column. Model results and their implications are synthesized in Fig. 13. Our results indicate a positive impact of deep convection in winter on annual phytoplankton biomass, especially microphytoplankton. Higher prey–predator decoupling leads to lower grazing by small zooplankton size-classes in winter/spring. In contrast, annual mesozooplankton biomass is enhanced by deep convection through lower prey–predator decoupling in winter, top trophic position and detritivorous diet. The interannual variability of mesozooplankton biomass appears to be mainly controlled by the levels of winter biomass. This is not evident in upper layers where budgets are strongly constrained by mixing-induced dilution (Herrmann et al., 2013). It is worth noting that the understanding and the quantification of these mechanisms require the whole water column to be taken into account and not only the upper layers, due to the winter mixinginduced biomass dilution. According to model results, the interannual variability of organic carbon sequestration, and potentially of fish biomass, in the NW Mediterranean could be determined by the climate interannual variability in winter through deep convection. The control of winter mixing on the size-class distribution of plankton communities seems to play an important role. Extrapolating the results of our model hindcast to the impact of global warming should depend on how surface heat loss and winter mixing are modified, and how

Author's personal copy

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

this will affect the organic carbon biomass, the size-class structure of plankton communities and the prey–predator decoupling. In a warmer ocean, deep convection is expected to decrease, overall stratification to increase and the nutricline position to be nearer the surface. As in the case of low convective years simulated in this paper, lower prey–predator decoupling would lead to stronger domination of small zooplankton size-classes with higher growth rates. Organic carbon storage could consequently be reduced, as could small pelagic fish and larval populations that feed on large organic particles. Even though validation on available data supports the use of this simulation to build a statistical analysis of winter mixing impact on interannual plankton dynamics, we wish to emphasize the weak points of our simulation and suggest possible improvements. First, it could certainly gain in realism by the use of consistent lateral boundary forcings from lower resolution biogeochemical simulations of the entire Mediterranean Sea. The overestimated contribution of microphytoplankton to total phytoplankton biomass should be addressed. Moreover, the closure term of our model mesozooplankton mortality does not account for a potential control from higher trophic levels such as gelatinous and small pelagic fish. Despite of our effort to better model zooplankton dynamics and compare it with available data sets, zooplankton may have been overestimated in the model. Large uncertainties in our understanding of zooplankton ecology such as feeding or natural and predator-induced mortality, particularly in deep layers and in winter, need to be address to better model this pivotal group between primary production and higher trophic levels. Acknowledgments We thank the three anonymous reviewers for their comments, which greatly enriched the manuscript. We are deeply indebted to Patrick Marsaleix for discussions that helped us to improve the model, and David Antoine for providing reliable CZCS and SeaWiFS satellite data. We thank Cyril Nguyen and computer engineers from the Laboratoire d’Aérologie for their technical support. We also thank Michel Déqué and Florence Sevault (Météo-France/ CNRM) for running ARPERA and NEMOMED8 respectively. Field data for the Ligurian Sea were provided by the DYFAMED marine observation department, LOV, Villefranche-sur-Mer, France. This work was mainly supported by the SESAME project (Contract No. GOCE-2006-036949) from the sixth Framework Program of the European Commission and is a contribution to the MerMEx project (Marine Ecosystem Response in the Mediterranean Experiment) from the MISTRALS international program. Numerical simulations were partly performed using HPC resources from CALMIP (Grant 2011-P09115). References Allen, J.I., Somerfield, P.J., Siddorn, J., 2002. Primary and bacterial production in the Mediterranean Sea: a modelling study. Journal of Marine Systems 33–34, 473– 495. Andersen, V., Gubanova, A., Nival, P., Ruellet, T., 2001. Zooplankton community during the transition from spring bloom to oligotrophy in the open NW Mediterranean and effects of wind events. 2. Vertical distributions and migrations. Journal of Plankton Research 23, 243–261. Antoine, D., Morel, A., Gordon, H.R., Banzon, V.F., Evans, R.H., 2005. Bridging ocean color observations of the 1980s and 2000s in search of long-term trends. Journal of Geophysical Research 110. Antoine, D., d’ Ortenzio, F., Hooker, S.B., Bécu, G., Gentili, B., Tailliez, D., Scott, A.J., 2008. Assessment of uncertainty in the ocean reflectance determined by three satellite ocean color sensors (MERIS, SeaWiFS and MODIS-A) at an offshore site in the Mediterranean Sea (BOUSSOLE project). Journal of Geophysical Research – Oceans 113. Auger, P.A., 2011. Modélisation des écosystèmes planctoniques pélagiques en Méditerranée nord-occidentale: Impact des eaux du Rhône à l’échelle du

25

plateau du golfe du Lion et variabilité interannuelle à décennale au large. PhD Thesis, Université de Toulouse III – Paul Sabatier. Auger, P.A., Diaz, F., Ulses, C., Estournel, C., Neveux, J., Joux, F., Pujo-Pay, M., Naudin, J.J., 2011. Functioning of the planktonic ecosystem on the Gulf of Lions shelf (NW Mediterranean) during spring and its impact on the carbon deposition: a field data and 3-D modelling combined approach. Biogeosciences 8, 3231–3261. Babin, M., Morel, A., Claustre, H., Bricaud, A., Kolber, Z., Falkowski, P.G., 1996. Nitrogen- and irradiance-dependent variations of the maximum quantum yield of carbon fixation in eutrophic, mesotrophic and oligotrophic marine systems. Deep Sea Research Part I: Oceanographic Research Papers 43, 1241–1272. Backhaus, J.O., Wehde, H., Hegseth, E.N., Kampf, J., 1999. ‘‘Phyto-convection’’: the role of oceanic convection in primary production. Marine Ecology Progress Series 189, 77–92. Backhaus, J.O., Hegseth, E.N., Wehde, H., Irigoien, X., Hatten, K., Logemann, K., 2003. Convection and primary production in winter. Marine Ecology Progress Series 251, 1–14. Banse, K., Mosher, S., 1980. Adult body mass and annual production/biomass relationships of field populations. Ecological Monographs 50, 355–379. Beaugrand, G., Reid, P.C., 2003. Long-term changes in phytoplankton, zooplankton and salmon related to climate. Global Change Biology 9, 801–817. Behrenfeld, M.J., 2010. Abandoning Sverdrup’s critical depth hypothesis on phytoplankton blooms. Ecology 91, 977–989. Béranger, K., Drillet, Y., Houssais, M.N., Testor, P., Bourdallé-Badie, R., Alhammoud, B., Bozec, A., Mortier, L., Bouruet-Aubertot, P., Crépon, M., 2010. Impact of the spatial distribution of the atmospheric forcing on water mass formation in the Mediterranean Sea. Journal of Geophysical Research – Oceans 115. Berline, L., Siokou-Frangou, I., Marasovic, I., Vidjak, O., Fernandez de Puelles, M.L., Mazzocchi, M.G., Assimakopoulou, G., Zervoudaki, S., Fonda-Umani, S., Conversi, A., Garcia-Comas, C., Ibanez, F., Gasparini, S., Stemmann, L., Gorsky, G., 2012. Intercomparison of six Mediterranean zooplankton time series. Progress in Oceanography 97–100, 76–91. Bertilsson, S., Berglund, O., Karl, D.M., Chisholm, S.W., 2003. Elemental composition of marine Prochlorococcus and Synechococcus: implications for the ecological stoichiometry of the sea. Limnology and Oceanography 48, 1721–1731. Béthoux, J.P., Gentili, B., Tailliez, D., 1998. Warming and freshwater budget change in the Mediterranean since the 1940s, their possible relation to the greenhouse effect. Geophysical Research Letters 25, 1023–1026. Béthoux, J.P., Durrieu de Madron, X., Nyffeler, F., Tailliez, D., 2002. Deep water in the western Mediterranean: peculiar 1999 and 2000 characteristics, shelf formation hypothesis, variability since 1970 and geochemical inferences. Journal of Marine Systems 33–34, 117–131. Blaxter, J.H.S., 1969. Experimental rearing of pilchard larvae, Sardina pilchardus. Journal of the Marine Biological Association of the United Kingdom 49, 557– 575. Blaxter, J.H.S., Hunter, J.R., 1982. The biology of the clupeoid fishes. Advances in Marine Biology 20, 1–223. Bosc, E., Bricaud, A., Antoine, D., 2004. Seasonal and interannual variability in algal biomass and primary production in the Mediterranean Sea, as derived from 4 years of SeaWiFS observations. Global Biogeochemical Cycles 18. Bouffard, J., Vignudelli, S., Herrmann, M., Lyard, F., Marsaleix, P., Ménard, Y., Cipollini, P., 2008. Comparison of ocean dynamics with a regional circulation model and improved altimetry in the Northwestern Mediterranean. Journal of Terrestrial, Atmospheric and Oceanic, Sciences Special Issue ‘‘Satellite Altimetry over Land and Coastal Zones: Applications and Challenges’’ 19, 117–133. Bricaud, A., Bosc, E., Antoine, D., 2002. Algal biomass and sea surface temperature in the Mediterranean Basin: intercomparison of data from various satellite sensors, and implications for primary production estimates. Remote Sensing of Environment 81, 163–178. Canals, M., Puig, P., Durrieu de Madron, X., Heussner, S., Palanques, A., Fabres, J., 2006. Flushing submarine canyons. Nature 444, 354–357. Chiswell, S.M., 2011. Annual cycles and spring blooms in phytoplankton: don’t abandon Sverdrup completely. Marine Ecology Progress Series 443, 39–50. Christaki, U., Courties, C., Karayanni, H., Giannakourou, A., Maravelias, C., Kormas, K.A., Lebaron, P., 2002. Dynamic characteristics of Prochlorococcus and Synechococcus consumption by bacterivorous nanoflagellates. Microbial Ecology 43, 341–352. Claustre, H., 1994. The trophic status of various oceanic provinces as revealed by phytoplankton pigment signatures. Limnology and Oceanography 39, 1206– 1210. Claustre, H., Babin, M., Merien, D., Ras, J., Prieur, L., Dallot, S., Prasil, O., Dousova, H., Moutin, T., 2005. Toward a taxon-specific parameterization of bio-optical models of primary production: a case study in the North Atlantic. Journal of Geophysical Research 110. Conan, P., Pujo-Pay, M., Raimbault, P., Leveau, M., 1998. Variabilité hydrologique et biologique du golfe du Lion. I. Transports en azote et productivité potentielle. Oceanologica Acta 21, 751–765. Conway, D.V.P., Tranter, P.R.G., Fernandez de Puelles, M.L., Coombs, S.H., 1991. Feeding of larval sprat (Sprattus sprattus L.) and sardine (Sardina pilchardus Walbaum). ICES CM Biological Oceanography Committee 76. Crise, A., Allen, J.I., Baretta, J., Crispi, G., Mosetti, R., Solidoro, C., 1999. The Mediterranean pelagic ecosystem response to physical forcing. Progress in Oceanography 44, 219–243. Cury, P., Bakun, A., Crawford, R.J.M., Jarre, A., Quinones, R.A., Shannon, L.J., Verheye, H.M., 2000. Small pelagics in upwelling systems: patterns of interaction and structural changes in ‘‘wasp-waist’’ ecosystems. ICES Journal of Marine Science 57, 603–618.

Author's personal copy

26

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27

Demirhindi, U., 1961. Nutrition of the sardine (Sardina pilchardus Walb.). Proceedings and Technical Papers of the General Fisheries Council for the Mediterranean 6, 253–259. D’Ortenzio, F., Ribera d’Alcala, M., 2009. On the trophic regimes of the Mediterranean Sea: a satellite analysis. Biogeosciences 6, 139–148. D’Ortenzio, F., Iudicone, D., de, Boyer Montegut, C., Testor, P., Antoine, D., Marullo, S., Santoleri, R., Madec, G., 2005. Seasonal variability of the mixed layer depth in the Mediterranean Sea as derived from in situ profiles. Geophysical Research Letters 32. Drinkwater, K.F., Beaugrand, G., Kaeriyama, M., Kim, S., Ottersen, G., Perry, R.I., Pörtner, H.-O., Polovina, J.J., Takasuka, A., 2010. On the processes linking climate to ecosystem changes. Journal of Marine Systems 79, 374–388. Estournel, C., Zervakis, V., Marsaleix, P., Papadopoulos, A., Auclair, F., Perivoliotis, L., Tragou, E., 2005. Dense water formation and cascading in the Gulf of Thermaikos (North Aegean), from observations and modelling. Continental Shelf Research 25, 2366–2386. Fasham, M.J.R., Flynn, K.J., Pondaven, P., Anderson, T.R., Boyd, P.W., 2006. Development of a robust marine ecosystem model to predict the role of iron in biogeochemical cycles: a comparison of results for iron-replete and ironlimited areas, and the SOIREE iron-enrichment experiment. Deep Sea Research Part I: Oceanographic Research Papers 53, 333–366. Garcia-Comas, C., Stemmann, L., Ibanez, F., Berline, L., Mazzocchi, M.G., Gasparini, S., Picheral, M., Gorsky, G., 2011. Zooplankton long-term changes in the NW Mediterranean Sea: decadal periodicity forced by winter hydrographic conditions related to large-scale atmospheric changes? Journal of Marine Systems 87, 216–226. Geider, R.J., MacIntyre, H.L., Kana, T.M., 1997. Dynamic model of phytoplankton growth and acclimation: responses of the balanced growth rate and the chlorophyll a:carbon ratio to light, nutrient-limitation and temperature. Marine Ecology Progress Series 143, 187–200. Geider, R.J., Macintyre, H.L., Graziano, L.M., McKay, R.M.L., 1998. Responses of the photosynthetic apparatus of Dunaliella tertiolecta (Chlorophyceae) to nitrogen and phosphorus limitation. European Journal of Phycology 33, 315–332. Gentleman, W., Leising, A., Frost, B., Strom, S., Murray, J., 2003. Functional responses for zooplankton feeding on multiple resources: a review of assumptions and biological dynamics. Deep Sea Research Part II: Topical Studies in Oceanography 50, 2847–2875. Gernez, P., Antoine, D., Huot, Y., 2011. Diel cycles of the particulate beam attenuation coefficient under varying trophic conditions in the northwestern Mediterranean Sea: observations and modeling. Limnology and Oceanography 56, 17–36. Goutx, M., Guigue, C., Aritio, D., Ghiglione, J.F., Pujo-Pay, M., Raybaud, V., Duflos, M., Prieur, L., 2009. Short term summer to autumn variability of dissolved lipid classes in the Ligurian sea (NW Mediterranean). Biogeosciences 6, 1229–1246. Grignon, L., Smeed, D.A., Bryden, H.L., Schroeder, K., 2010. Importance of the variability of hydrographic preconditioning for deep convection in the Gulf of Lion, NW Mediterranean. Ocean Science 6, 573–586. Heimbürger, L.-E., Lavigne, H., Migon, C., D’Ortenzio, F., Estournel, C., Coppola, L., Miquel, J.-C., 2013. Temporal variability of vertical export flux at the DYFAMED time-series station (Northwestern Mediterranean Sea). Progress in Oceanography 119, 59–67. Heldal, M., Scanlan, D.J., Norland, S., Thingstad, F., Mann, N.H., 2003. Elemental composition of single cells of various strains of marine Prochlorococcus and Synechococcus using X-ray microanalysis. Limnology and Oceanography 48, 1732–1743. Herrmann, M., Somot, S., 2008. Relevance of ERA40 dynamical downscaling for modeling deep convection in the Mediterranean Sea. Geophysical Research Letters 35. Herrmann, M., Estournel, C., Déqué, M., Marsaleix, P., Sevault, F., Somot, S., 2008a. Dense water formation in the Gulf of Lions shelf: impact of atmospheric interannual variability and climate change. Continental Shelf Research 28, 2092–2112. Herrmann, M., Somot, S., Sevault, F., Estournel, C., Déqué, M., 2008b. Modeling the deep convection in the northwestern Mediterranean Sea using an eddypermitting and an eddy-resolving model: case study of winter 1986–1987. Journal of Geophysical Research – Oceans 113. Herrmann, M., Sevault, F., Beuvier, J., Somot, S., 2010. What induced the exceptional 2005 convection event in the northwestern Mediterranean basin? Answers from a modeling study. Journal of Geophysical Research 115. Herrmann, M., Diaz, F., Estournel, C., Marsaleix, P., Ulses, C., 2013. Impact of atmospheric and oceanic interannual variability on the Northwestern Mediterranean Sea pelagic planktonic ecosystem and associated carbon cycle. Journal of Geophysical Research – Oceans 118, 5792–5813. Hong, X., Hodur, R.M., Martin, P.J., 2007. Numerical simulation of deep-water convection in the Gulf of Lion. Pure and Applied Geophysics 164, 2101–2116. Hu, Z.Y., Doglioli, A.M., Petrenko, A.A., Marsaleix, P., Dekeyser, I., 2009. Numerical simulations of eddies in the Gulf of Lion. Ocean Modelling 28, 203–208. Hunter, M.D., Price, P.W., 1992. Playing chutes and ladders: heterogeneity and the relative roles of bottom-up and top-down forces in natural communities. Ecology 73, 724–732. Josey, S.A., Somot, S., Tsimplis, M., 2011. Impacts of atmospheric modes of variability on Mediterranean Sea surface heat exchange. Journal of Geophysical Research 116. Kiorboe, T., 1993. Turbulence, phytoplankton cell size, and the structure of pelagic food webs. Advances in Marine Biology 29, 1–72.

Krahmann, G., 1997. Saisonale und zwischenjährliche Variabilität im westlichen Mittelmeer - Analyse historischer Daten (PhD Thesis). Cristian Albrecthts Universität zu Kiel Levy, M., Mémery, L., André, J.M., 1998. Simulation of primary production and export fluxes in the Northwestern Mediterranean Sea. Journal of Marine Research 56, 197–238. Licandro, P., Icardi, P., 2009. Basin scale distribution of zooplankton in the Ligurian Sea (north-western Mediterranean) in late autumn. Hydrobiologia 617, 17–40. Lloret, J., Palomera, I., Salat, J., Sole, I., 2004. Impact of freshwater input and wind on landings of anchovy (Engraulis encrasicolus) and sardine (Sardina pilchardus) in shelf waters surrounding the Ebre (Ebro) River delta (north-western Mediterranean). Fisheries Oceanography 13, 102–110. Madec, G., Chartier, M., Crepon, M., 1991. The effect of thermohaline forcing variability on deep water formation in the Western Mediterranean sea: a high resolution three dimensional numerical study. Dynamics of Atmospheres and Oceans 15, 301–332. Marsaleix, P., Auclair, F., Estournel, C., 2009. Low-order pressure gradient schemes in sigma coordinate models: the seamount test revisited. Ocean Modelling 30, 169–177. Marsaleix, P., Auclair, F., Estournel, C., Nguyen, C., Ulses, C., 2011. An accurate implementation of the compressibility terms in the equation of state in a low order pressure gradient scheme for sigma coordinate ocean models. Ocean Modelling 40, 1–13. Marsaleix, P., Auclair, F., Duhaut, T., Estournel, C., Nguyen, C., Ulses, C., 2012. Alternatives to the Robert–Asselin filter. Ocean Modelling 41, 53–66. Marshall, J., Schott, F., 1999. Open-ocean convection: observations, theory, and models. Reviews of Geophysics 37, 1–64. Marty, J.C., Chiavérini, J., 2002. Seasonal and interannual variations in phytoplankton production at DYFAMED time-series station, northwestern Mediterranean Sea. Deep Sea Research Part II: Topical Studies in Oceanography 49, 2017–2030. Marty, J.C., Chiavérini, J., 2010. Hydrological changes in the Ligurian Sea (NW Mediterranean, DYFAMED site) during 1995–2007 and biogeochemical consequences. Biogeosciences 7, 2117–2128. Marty, J.C., Chiavérini, J., Pizay, M.D., Avril, B., 2002. Seasonal and interannual dynamics of nutrients and phytoplankton pigments in the western Mediterranean Sea at the DYFAMED time-series station (1991–1999). Deep Sea Research Part II: Topical Studies in Oceanography 49, 1965–1985. MEDOC-group, 1970. Observations of formation of deep-water in the Mediterranean Sea. Nature 227, 1037–1040. Megrey, B.A., Rose, K.A., Klumb, R.A., Hay, D.E., Werner, F.E., Eslinger, D.L., Smith, S.L., 2007. A bioenergetics-based population dynamics model of Pacific herring (Clupea harengus pallasi) coupled to a lower trophic level nutrient– phytoplankton–zooplankton model: description, calibration, and sensitivity analysis. Ecological Modelling 202, 144–164. Mertens, C., Schott, F., 1998. Interannual variability of deep-water formation in the northwestern Mediterranean. Journal of Physical Oceanography 28, 1410–1424. Molinero, J.C., Ibanez, F., Souissi, S., Buecher, E., Dallot, S., Nival, P., 2008. Climate control on the long-term anomalous changes of zooplankton communities in the Northwestern Mediterranean. Global Change Biology 14, 11–26. Moore, T.S., Campbell, J.W., Dowell, M.D., 2009. A class-based approach to characterizing and mapping the uncertainty of the MODIS ocean chlorophyll product. Remote Sensing of Environment 113, 2424–2430. Morel, A., André, J.M., 1991. Pigment distribution and primary production in the western Mediterranean as derived and modeled from coastal zone color scanner observations. Journal of Geophysical Research 96, 12685–12698. Morel, A., Claustre, H., Antoine, D., Gentili, B., 2007. Natural variability of bio-optical properties in Case 1 waters: attenuation and reflectance within the visible and near-UV spectral domains, as observed in South Pacific and Mediterranean waters. Biogeosciences 4, 913–925. Nejstgaard, J.C., Gismervik, I., Solberg, P.T., 1997. Feeding and reproduction by Calanus finmarchicus, and microzooplankton grazing during mesocosm blooms of diatoms and the coccolithophore Emiliania huxleyi. Marine Ecology Progress Series 147, 197–217. Nowaczyk, F., Carlotti, F., Thibault-Botha, D., Pagano, M., 2011. Distribution of epipelagic metazooplankton across the Mediterranean Sea during the summer BOUM cruise. Biogeosciences 8, 2159–2177. Olita, A., Sorgente, R., Ribotti, A., Fazioli, L., Perilli, A., 2011. Pelagic primary production in the Algero-Provencal Basin by means of multisensor satellite data: focus on interannual variability and its drivers. Ocean Dynamics. 61, 1005–1016. Palomera, I., Olivar, M.P., Salat, J., Sabatés, A., Coll, M., Garcia, A., Morales-Nin, B., 2007. Small pelagic fish in the NW Mediterranean Sea: an ecological review. Progress in Oceanography 74, 377–396. Papadopoulos, V.P., Josey, S., Bartzokas, A., Somot, S., Ruiz, S., Drakopoulou, P., 2012. Large scale atmospheric circulation favoring deep and intermediate water formation in the Mediterranean Sea. Journal of Climate 25, 6079–6091. Petrakis, G., Stergiou, K.I., Crhistou, E., Politou, C.-Y., Karkani, M., Simboura, N., Kouyoufas, P., n.d. Small Scale Fishery in the South Euboikos Gulf. Unpublished Final Report, Contract No. XIV-1/MED-91/007. National Centre for Marine Research, Athens, Hellas. Plounevez, S., Champalbert, G., 2000. Diet, feeding behaviour and trophic activity of the anchovy (Engraulis encrasicolus L.) in the Gulf of Lions (Mediterranean Sea). Oceanologica Acta 23, 175–192.

Author's personal copy

P.A. Auger et al. / Progress in Oceanography 124 (2014) 12–27 Prieur, L., Legendre, L., 1988. Oceanographic criteria for new phytoplankton production. NATO ASI Series. Series C, Mathematical and Physical Sciences 239, 71–112. Pyper, B.J., Peterman, R.M., 1998. Comparison of methods to account for autocorrelation in correlation analyses of fish data. Canadian Journal of Fisheries and Aquatic Sciences 55, 2127–2140. Raick, C., Delhez, E.J.M., Soetaert, K., Grégoire, M., 2005. Study of the seasonal cycle of the biogeochemical processes in the Ligurian Sea using a 1D interdisciplinary model. Journal of Marine Systems 55, 177–203. Rasoanarivo, R., Champalbert, J.F., Becker, B., 1991. Relations entre les communautés phytoplanctoniques et l’alimentation des larves de Sardina pilchardus Walb. dans le golfe de Fos (Méditerranée occidentale): influence de la lumière sur l’activité alimentaire des larves. Journal of Experimental Marine Biology and Ecology 151, 83–92. Raybaud, V., Nival, P., Prieur, L., 2011. Short time-scale analysis of the NW Mediterranean ecosystem during summer–autumn transition: a 1D modelling approach. Journal of Marine Systems 84, 1–17. Reffray, G., Fraunié, P., Marsaleix, P., 2004. Secondary flows induced by wind forcing in the Rhône region of freshwater influence. Ocean Dynamics 54, 179–196. Ribera d’Alcala, M., Conversano, F., Corato, F., Licandro, P., Mangoni, O., Marino, D., Mazzocchi, M.G., Modigh, M., Montresor, M., Nardella, M., Saggiomo, V., Sarno, D., Zingone, A., 2004. Seasonal patterns in plankton communities in a pluriannual time series at a coastal Mediterranean site (Gulf of Naples). Scientia Marina 68, 65–83. Riegman, R., Stolte, W., Noordeloos, A.A.M., Slezak, D., 2000. Nutrient uptake and alkaline phosphatase (ec 3:1:3:1) activity of emiliania huxleyi (Prymnesiophyceae) during growth under n and p limitation in continuous cultures. Journal of Phycology 36, 87–96. Rixen, M., Beckers, J.M., Levitus, S., Antonov, J., Boyer, T., Maillard, C., Fichaut, M., Balopoulos, E., Iona, S., Dooley, H., Garcia, M.J., Manca, B., Giorgetti, A., Manzella, G., Mikhailov, N., Pinardi, N., Zavatarelli, M., 2005. The Western Mediterranean deep water: a proxy for climate change. Geophysical Research Letters 32, L12608. Rubio, A., Taillandier, V., Garreau, P., 2009. Reconstruction of the Mediterranean northern current variability and associated cross-shelf transport in the Gulf of Lions from satellite-tracked drifters and model outputs. Journal of Marine Systems 78, S63–S78. Romagnan, J.B., 2013. Les communautés planctoniques des bactéries au macroplancton : dynamique temporelle en Mer Ligure et distribution dans l’océan global lors de l’expédition Tara Oceans - Approche holistique par imagerie. PhD Thesis, Université de Nice - Sophia-Antipolis. Sannino, G., Herrmann, M., Carillo, A., Rupolo, V., Ruggiero, V., Artale, V., Heimbach, P., 2009. An eddy-permitting model of the Mediterranean Sea with a two-way grid refinement at the Strait of Gibraltar. Ocean Modelling 30, 56–72. Santinelli, C., Nannicini, L., Seritti, A., 2010. DOC dynamics in the meso and bathypelagic layers of the Mediterranean Sea. Deep Sea Research Part II: Topical Studies in Oceanography 57, 1446–1459. Sarthou, G., Timmermans, K.R., Blain, S., Tréguer, P., 2005. Growth physiology and fate of diatoms in the ocean: a review. Journal of Sea Research 53, 25–42. Siokou-Frangou, I., Christaki, U., Mazzocchi, M.G., Montresor, M., Ribera d’Alcala, M., Vaqué, D., Zingone, A., 2010. Plankton in the open Mediterranean Sea: a review. Biogeosciences 7, 1543–1586. Smith, R.O., Bryden, H.L., Stansfield, K., 2008. Observations of new western Mediterranean deep water formation using Argo floats 2004–2006. Ocean Science 4, 133–149.

27

Soetaert, K., Middelburg, J.J., Herman, P.M.J., Buis, K., 2000. On the coupling of benthic and pelagic biogeochemical models. Earth-Science Reviews 51, 173– 201. Sokal, R.R., Rohlf, F.J., 1995. Biometry: The Principles and Practice of Statistics in Biological Research, third ed. W. H. Freeman and Co., New York. Sondergaard, M., Theil-Nielsen, J., 1997. Bacterial growth efficiency in lakewater cultures. Aquatic Microbial Ecology 12, 115–122. Sournia, A., 1973. La production primaire planctonique en Méditerranée, Essai de mise à jour. Bulletin Etude en commun de la Méditerranée, N" Spécial 5. Sournia, A., Brylinski, J.M., Dallot, S., Lecorre, P., Leveau, M., Prieur, L., Froget, C., 1990. Hydrological fronts off the coasts of France – a review. Oceanologica Acta 13, 413–438. Steele, J.H., Henderson, E.W., 1992. The role of predation in plankton models. Journal of Plankton Research 14, 157–172. Sverdrup, H.U., 1953. On conditions for the vernal blooming of phytoplankton. Journal du Conseil International pour l’Exploration de la Mer 18, 287–295. Tanaka, T., Rassoulzadegan, F., 2002. Full-depth profile (0–2000 m) of bacteria, heterotrophic nanoflagellates and ciliates in the NW Mediterranean Sea: vertical partitioning of microbial trophic structures. Deep Sea Research Part II: Topical Studies in Oceanography 49, 2093–2107. Taylor, K.E., 2001. Summarizing multiple aspects of model performance in a single diagram. Journal of Geophysical Research 106, 7183–7192. Taylor, J.R., Ferrari, R., 2011. Shutdown of turbulent convection as a new criterion for the onset of spring phytoplankton blooms. Limnology and Oceanography 56, 2293–2307. Timmermans, K.R., Van der Wagt, B., Veldhuis, M.J.W., Maatman, A., De Baar, H.J.W., 2005. Physiological responses of three species of marine pico-phytoplankton to ammonium, phosphate, iron and light limitation. Journal of Sea Research 53, 109–120. Travers, M., Shin, Y.-J., Jennings, S., Machu, E., Huggett, J.A., Field, J.G., Cury, P.M., 2009. Two-way coupling versus one-way forcing of plankton and fish models to predict ecosystem changes in the Benguela. Ecological Modelling 220, 3089– 3099. Tyrrell, T., Taylor, A.H., 1996. A modelling study of Emiliania huxleyi in the NE Atlantic. Journal of Marine Systems 9, 83–112. Uitz, J., Claustre, H., Morel, A., Hooker, S.B., 2006. Vertical distribution of phytoplankton communities in open ocean: an assessment based on surface chlorophyll. Journal of Geophysical Research 111. Ulses, C., Estournel, C., Puig, P., Durrieu de Madron, X., Marsaleix, P., 2008. Dense shelf water cascading in the northwestern Mediterranean during the cold winter 2005: Quantification of the export through the Gulf of Lion and the Catalan margin. Geophysical Research Letters 35. Vandromme, P., Stemmann, L., Berline, L., Gasparini, S., Mousseau, L., Prejger, F., Passafiume, O., Gurarini, J.M., Gorsky, G., 2011. Inter-annual fluctuations of zooplankton communities in the Bay of Villefranche-sur-mer from 1995 to 2005 (Northern Ligurian Sea, France). Biogeosciences 8, 3143–3158. Vandromme, P., Stemmann, L., Garcìa-Comas, C., Berline, L., Sun, X., Gorsky, G., 2012. Assessing biases in computing size spectra of automatically classified zooplankton from imaging systems: a case study with the ZooScan integrated system. Methods in Oceanography 1–2, 3–21. Vidussi, F., Claustre, H., Manca, B.B., Luchetta, A., Marty, J.C., 2001. Phytoplankton pigment distribution in relation to upper thermocline circulation in the eastern Mediterranean Sea during winter. Journal of Geophysical Research 106, 19939– 19956.