3.2.1. Seasonal Variability
 SeaWiFS images (Figure 4) show that the first leg of P1 took place at a time when zonal average surface chlorophyll concentrations were low (0.3 mgChl m−3), with concentrations even lower during P3 (less than 0.1 mgChl m−3). In contrast, during the first leg of P2, surface chlorophyll was higher (1 mgChl m−3 and more), but the satellite images show that north of 42°N the bloom was interrupted for about 2 weeks during the second leg of P2. The sites 3 and 4 were sampled during this period of relatively low average activity [Lévy et al., 2005a].
Figure 4. Annual evolution of the zonal average of Sea-viewing Wide Field-of-view Sensor (SeaWiFS) chlorophyll against latitude in the POMME area. The numbers indicate the location of time and latitude of the process study stations of the second legs (white, P1; black, P2; yellow, P3). North of 42°N, the bloom has been “stopped” at the end of April for several weeks, and the end of P2 occurred during that period. Figure by Y. Lehahn and M. Lévy (LODYC, Paris).
Download figure to PowerPoint
 Observed nitrate and phosphate distributions suggest that, at least in the southwest, nutrient uptake had already begun before cruise P1 (beginning of February), whereas, in agreement with the satellite data, the bloom was not over yet north of the 41°N front at the beginning of May (end of P2) [Fernández et al., 2005a]. During cruise P3, nutrients were depleted at the ocean surface: the system was strongly oligotrophic. This seasonal evolution is also observed in the rates of biological production and in variations in the phytoplankton taxonomy.
 An increase of primary production (PP) is observed between P1 and P2 (almost by a factor of 3 from 456 mgC m−2 d−1 during the first leg of P1 (February) compared to 1236 mgC m−2 d−1 during the first leg of P2 (April) [Fernández et al., 2005b]). It is related to the receding mixed layer and an increase in near surface phytoplankton biomass, but is not as strong as one could expect from the large change in mixed layer depth between the two cruises. During P3 (September), the PP values were close to the winter estimates (420 mgC m−2 d−1). In general, the contribution of nanophytoplankton to PP is largest, although during the strongly oligotrophic period sampled during P3, it is exceed that of picophytoplankton (about 50% [Claustre et al., 2005]). Even during P2, the contribution of picophytoplankton does not dominate PP (around 30% for cells larger than 10 μm). This is consistent with results from plankton taxonomy which indicate that the diatoms that provide a large contribution to the biomass during P2 (Pseudo nitzchia spp) are often relatively small; e.g., part of the nanophytoplankton production could originate from small diatoms [Leblanc et al., 2005]. This unexpected phenomenon may result from silicate being a limiting element which is suggested by the incubation experiments performed to estimate the half saturation constant of silicate uptake during P2 but also during P1 [Leblanc et al., 2005]. Incubation experiments also indicate that iron is limiting, although episodic deposition of dust of Saharan origin might relieve this limitation [Blain et al., 2004]. The iron limiting effect is larger in winter, and could together with other colimitations decrease the uptake rates of carbon by 70%. This result contrasted with the expectation that the POMME region in the NE Atlantic Ocean is not an HLNC region.
 At the end of P2, and mostly during P3, macronutrients also limit the production: the molar nitrogen/phosphate ratio decreases down to 5 at the surface (T. Moutin, personal communication, 2004), the diatom concentrations and biogenic silicate production are extremely low [Leblanc et al., 2005], and the few large cells which can be observed within some mesoscale structures are dinoflagellates (HPLC data [Claustre et al., 2005]). The distribution of species results in a low F ratio (NO3 uptake/NH4 uptake). Moreover, as nitrification rates are strong (between 5 and up to more than 50% of nitrate uptake), part of the nitrate uptake is in fact regenerated production, even in winter [Fernández et al., 2005a], and the corrected F ratio estimates are on the order of 0.5 during P1 and P2 (and 0.2 during P3 [Fernández et al., 2005b]).
 Cruise to cruise differences in bacterial production (BP) tend to mimic those of primary production, but the BP/PP ratio increases with time (from 10% during P1 and P2 up to 28% during P3 (F. Van Wambeke, personal communication, 2004)). This increase goes along with an increase in the Bacterial Growth Efficiency (BGE), from 8% during P1 and 14% during P2 to 24% during P3. The major part of this BP is consumed by heterotrophic nanoflagellates [Karayanni et al., 2005]. Grazing pressure exerted by these small heterotrophs, and by ciliates is equivalent to about 30% of the daily primary production, but it can reach more than 80% per day on some occasions (F. Van Wambeke, personal communication, 2004). Given a refractory background DOC concentration of around 41 mmolC m−3, the POMME area is characterized by an accumulation of semilabile DOC of up to 15% of the background value during P3.This semilabile pool can potentially be exported into the permanent thermocline during spring subduction, after deep winter mixing [Sohrin and Sempéré, 2005].
 The zooplankton community shows a very pronounced seasonal variability, in terms of species, functional impact, grazing pressure and remineralization fluxes. The Bioness results show that the average macroplankton biomass in the 0–700 m depth range is 4 (3) times higher during P2 (P3) than during P1 (V. Andersen, personal communication, 2003). Moreover, in terms of numbers, omnivorous species are mostly predominant during P2, whereas carnivorous species dominate during P3. Significant stocks of herbivorous species are found in the north during P1 and in the south during P2. Large copepods, as well as euphausiids, are mostly present in April, during P2, whereas the copepod concentrations are very low in September and those of euphausiids are low in February. The WP2 biomass results present a high variability and somewhat conflicting results (V. Andersen, personal communication, 2003). One set of hauls indicated somewhat higher biomass during the fall cruise (2.6 g.m−2), while the other set suggested a spring maximum of mesozooplankton biomass. This contradiction can be resolved with an integrated view of the area possible by the OPC transects, which showed that biomass varied from 4 g m−2 in spring to 2.5 g m−2 in fall. An earlier study in the same sampling zone in 2000 yielded similar spring biomass estimates. In term of carbon, the WP2 data of P2 give an average of 3.4 mgC m−3 (between 0.7 and 7.7 mgC m−3) over the 0–200 m depth range. This is comparable to the results obtained by Dam et al.  during NABE during the same period (average of 4.4 mgC m−3 with variations from 2.1 to 6.1 mgC m−3): although the sampling during NABE was restricted to the mixed layer (shoaling from 120 m at the beginning (end of April) to 30 m at the end (end of May)), the range of biomass measurements is comparable between the two surveys. Another major observation during POMME was the large number of gelatinous species (pteropods and salps) during P2, and also to a smaller extent during P3. The grazing pressure on phytoplankton averaged over the top 200 m was higher during P2 (around 30% of the PP), and much lower during P1 (10% of the PP) and P3 (25% of a much smaller PP). During P2, the remineralization rates can be quite high and vary between 8 and 26 mmoleCO2 m−2 d−1, or around 25% of the PP (P. Mayzaud, personal communication, 2004): these estimates have still to be made for the other cruises.
 The data return from the sediment traps is good (90%): only one trap at 400 m did not collect during 2 months. The moorings display a clear seasonal cycle, with a maximum particle flux in March–April 2001, and 1 month later in 2002 [Guieu et al., 2005]: the interannual variability is strong, with higher particulate matter fluxes during the bloom in 2001 than in 2002 (at 400 m, largest average fluxes over 10 days reaching 500 mg m−2 d−1 during March 2001 compared to 350 mg m−2 d−1 in April–June 2002). Data have been corrected using Thorium measurements: the efficiency of the sediment traps is rather low and highly variable varying from 18 to 55% (with the smallest value for the northeast mooring associated with high horizontal velocities, resulting certainly from the presence of the quasi permanent A1 structure nearby). A first estimate of particulate carbon export at 400 m during 2001 gives 5.2 mgC m−2 d−1 and 4.9 mgC m−2 d−1 at 1000 m: these fluxes are at the lower end of the range of other data obtained in the North Atlantic Ocean, and are also fairly close to the estimates found at the Eumeli oligotrophic site [Morel, 1996; Bory et al., 2001]. Data from the drifting traps launched during the process study stations are consistent with the data of the moored sediment traps [Goutx et al., 2005]. These drifting trap data, as well as the mooring data [Guieu et al., 2005], show a shift from an export dominated by carbonate during P1 to an increased share of organic matter during P2 (more than half of the exported carbon at 200 m, the carbonate representing still the major part at 400 m) and of opal (although this flux is underestimated in the trap data). During P3, the fluxes are much lower (by a factor 4), with a very low contribution of biogenic silicate [Mosseri et al., 2005] and an equipartition of carbonate and organic carbon. These results are coherent with the time evolution of phytoplankton, presenting large concentrations of diatoms during P2, a larger contribution of coccolithophores during P1, and an oligotrophic situation during P3 (H. Claustre, personal communication, 2004; B. Quéguiner, personal communication, 2004).
 Net balances of oxygen and carbon variations show that the community biological activity in the POMME area results in a system which is strongly autotrophic during P1 and P2, and which changes into heterotrophic conditions in summer (P3) [Maixandeau et al., 2005]. Together with the role played by ocean dynamics and sea surface temperature (SST) variability, this biological production is responsible for an annual atmospheric CO2 uptake of the ocean [Gonzalez Davila et al., 2005]. The CARIOCA drifter data and the data from the surveys indicate a strong CO2 sink during P1 and P2, and a slight degassing during P3 (L. Merlivat, personal communication, 2003). Although the average flux is close to the flux obtained by Takahashi [Takahashi et al., 2002], the seasonal variation in the POMME region is much stronger during 2001, whereas in Takahashi et al.'s  data compilation study the ocean is continuously taking up atmospheric CO2, all year round [Gonzalez Davila et al., 2005].
3.2.2. Submeso- and Mesoscale Variability
 Because of the temporal variability during the first leg surveys, an interpretation of the data collected solely in terms of spatial variability is difficult to do. Nevertheless, the mesoscale structures (eddies) and in particular the anticylonic eddies were rather stable: some of these structures were sampled during each cruise. This can be used for a first analysis of the seasonal evolution of the ecosystems within cyclonic and anticyclonic eddies, and of the differences between these structures. Preliminary numerical simulations show nevertheless that water mass exchanges between the eddies and the surrounding environment can be substantial over the several months separating successive surveys (M. Lévy, personal communication, 2004). This indicates that eddies cannot be considered in isolation of their environment, in particular between P2 and P3.
 A preliminary analysis shows that the seasonal evolution within cyclonic (C4) and anticyclonic eddies (A1, A2) is quite similar. Nevertheless, when A2 is compared to C4 (Figure 2), it seems that the bloom (P1) as well as the transition toward oligotrophy (P2) occur sooner in the anticyclonic structure, characterized by a stronger and shallower stratification in A2 during winter. During summer (P3) the biological activity is stronger at the station located within the cyclone C4 than to its south, possibly because of the uplift of the isopycnal surfaces.
 Interestingly, the major signal in biological activity and tracer concentrations is obtained either within or close to the A1 anticyclone. On the basis of the float data, this eddy is extremely stable. It is characterized by very high levels of biomass and primary production, along with high diatom and biogenic silicate concentrations during P2, and large particle export during P1 and P2 (more than three times the export obtained at the other 2 day stations, although this statement should be considered with caution, as the position of these stations is not exactly localized at the very centre of the core, and as it might vary in time relatively to the structure). This anticyclone is clearly seen during the first leg of P3, as it is characterized by high nutrient levels [Fernández et al., 2005a] (around 6 μmol kg−1 at 50 m compared to less than 1.5 μmol kg−1 everywhere else) and by surface pigment concentrations much higher than elsewhere in the POMME area, associated with large dinoflagellates [Claustre et al., 2005]. This eddy was moreover always close to the NE sediment trap mooring, which shows also the largest particulate fluxes in the four moorings, with the largest proportion of CaCO3 export production and the smallest opal export, although diatom production is very large [Guieu et al., 2005]. Unfortunately, no long time series station could be occupied during the second leg of P3 for logistical reasons (emergency evacuation).
 The stations during the first legs barely resolved the mesoscale features. The interpretation of the data obtained during these surveys has to be moderated somewhat by the observations obtained from high-resolution surveys (Seasoar/OPCT, Tow-Yo), or high-frequency observations (Carioca buoys). These, yet unpublished observations as well as the modeling studies [Lévy et al., 2005b] suggest that the main variability in the biological variables is found at scales smaller than the mesoscale. At least in the late winter, these might be associated with fronts and eddy-eddy interactions, where ageostrophic processes, vertical velocities and spatial variability of mixed layer depth are expected to be important. These scales are also associated with strong horizontal currents. Keeping in mind the resolution issue during the first legs, this horizontal transport can be seen in the patterns of nutrient [Fernández et al., 2005a] or bacterial distributions [Thyssen et al., 2005], for instance within the northward current east of C4. Examples of submesoscale variability are shown in Figures 5– 8. Figure 5 is a SeaWiFS image obtained on 31 March with a 1 km horizontal resolution in the POMME area (H. Loisel, personal communication, 2003): although some medium-scale features can be seen, most of the sea color gradients are associated with filaments. Figure 6 shows a Tow-Yo transect between the centers of the cyclonic patterns C4 and C5B, crossing strong meridional fronts to the east of the anticyclones A1 and A4 (Figure 3) (L. Prieur, personal communication, 2003): a deep tongue of high oxygen concentration can be seen close to these fronts. This tends to show that ventilation and mode water formation could occur at small scales, e.g., through subduction associated with anticyclones. Figure 7 results from a Seasoar/OPCT survey during P3 between A4 and C4, along 42°N. Maximum concentrations of zooplankton are obtained within small areas, possibly close to fronts (not shown): they can be associated with either maximum or minimum values of fluorescence. These observations indicate that whereas an increase of biomass might be strongly correlated with specific dynamical features at the submesoscale, the coupling between primary and secondary producers differs significantly from place to place, probably in relation to the intensity of ageostrophic vertical motions, the growth rate of phytoplankton and the grazing rate of zooplankton. Finally, Figure 8 shows the evolution of some parameters measured by a Carioca drifter during the bloom (L. Merlivat, personal communication, 2003). The fast variability of pCO2 corrected for the solubility (temperature) effect is strong (several tens of μatm in a few days), and can be partly explained by biological activity (biological pump), seen with the abrupt increase of fluorescence, but also by the impact of different water masses encountered by the drifter, possibly on both sides of the jet associated with a front.
Figure 5. SeaWiFS image obtained on 31 March 2001, with a 1 km resolution. The cloudy areas are in white. Figure by H. Loisel (Université Lille, Lille, France).
Download figure to PowerPoint
Figure 6. Tow-Yo section between the two cyclones C4 and C5B (white line on the figure on the right) during P2. (top) Density. (middle) Fluorescence. (bottom) Dissolved oxygen. Figure by L. Prieur (LOV, Villefranche-sur-mer, France).
Download figure to PowerPoint
Figure 7. Seasoar/OPCT section during P3 across A1 to A4 (northern segment of the ship track). (top) Fluorescence from Seasoar (relative unit). (middle) Zooplankton biovolume from OPCT data. (bottom) Vertically integrated zooplankton biovolume. Figure by J.-P. Labat (LOV, Villefranche-sur-mer, France).
Download figure to PowerPoint
Figure 8. Carioca drifter observations. (top left) Drifter trajectory between day 49 (18 February) and 212 (1 July). The small numbers indicate the day associated with the position of the buoy. The letters C, A, and F indicate the approximate locations of cyclonic, anticyclonic, and frontal patterns. (top right) Time evolution of fluorescence (green) (relative units) and normalized dissolved inorganic carbon (DIC) (red) (μmol/kg). The large decrease in DIC observed at the end of March to the beginning of April is associated with a large increase of fluorescence, e.g., with the upset of a bloom. (bottom right) Time evolution of normalized DIC (red) (μmol/kg) and sea surface salinity (SSS) (black) (PSU). The DIC oscillations observed between day 140 and day 180 are strongly anticorrelated with SSS, which tends to show that this variability is mostly driven by the characteristics of the different water masses that the buoy crosses during its southward journey (see Figure 8 (top left)) within a meridional front. (bottom left) Air (black) and sea surface (red) CO2 partial pressure. When sea surface pCO2 is smaller (larger) than atmospheric pCO2, the ocean behaves like a CO2 sink (source), which is the case in winter and spring (in summer). Figure by L. Merlivat (LODYC, Paris).
Download figure to PowerPoint