Microplankton composition, production and upwelling dynamics in Sagres (SW Portugal) during the summer of 2001*

Microplankton community, production, and respiration were studied alongside physical and chemical conditions at Sagres (SW Portugal) during the upwelling season, from May to September 2001. The sampling station was 5 km east of the upwelling centre off Cabo S. Vicente, and 2 km west of an offshore installation for bivalve aquaculture. Three major periods were distinguished according to sea surface temperature (SST): period 1 (P1; May and June), characterised by high temperature values (17.0±1.8oC); period 2 (P2; July), characterised by lower temperatures (14.6±0.3oC), identified as an upwelling-blooming stage; and period 3 (P3; August), characterised by a high temperature pattern (16.25±1.14oC). Chaetoceros spp., Thalassiosira spp., Lauderia spp., Detonula spp. and Pseudo-nitzschia spp. were the major taxa contributing to the dissimilarities between P2 (July) and the other periods. In July (P2), the average gross production (GP; 52.5±12.3 μM O2 d-1) and net community production (NCP; 46.9±15.3 μM O2 d-1) peaked with the maximal concentrations of diatom-chl a. Dark community respiration (DCR) remained low and more constant throughout (4.6±3.6 μM O2 d-1). The plankton assemblage was dominated by diatoms throughout the survey. Physical events were the primary factors determining the microplankton structure and distribution at this location.


INTRODUCTION
Coastal fertilisation by cold nutrient-rich upwelled waters stimulates productivity and phytoplankton blooms (Barber and Smith, 1981).These blooms are dominated initially by nonmotile diatoms (Officer and Ryther, 1980) that are preferentially selected under the turbulent conditions produced by strong winds, which are responsible for the upwelling.As the turbulence is reduced, optimal conditions develop for the more motile dinoflagellates, establishing the plankton succession pattern (Margalef, 1978;Smayda, 2000).The ocean biota is sustained by the balance between the autotrophic (i.e.production) and heterotrophic (i.e.respiration) processes (e.g.Williams, 1984Williams, , 1998)).In coastal systems where inputs from terrestrial sources are limited, such as the studied location, phytoplankton primary production represents the main source of organic matter.Size fractionation studies (Williams, 1981) have associated the dominant respiratory activity in coastal waters with small non-photosynthetic organisms, such as heterotrophic bacteria and microflagellates.
Northerly winds along the west coast of the Iberian Peninsula produce conditions for seasonal upwelling from early spring to late summer (e.g.Wooster et al., 1976;Fiúza et al., 1982), whilst occasional upwelling occurs along the southern coast of Portugal (Algarve) with favourable westerly winds.After a prolonged period of northerly winds, fertile water can circulate around the Cabo S. Vincente, the southwestern tip of the peninsula, and flow eastwards along the southern coastal shelf (Fiúza, 1983;Sousa and Bricaud, 1992;Relvas and Barton, 2002).In contrast, a warm counter current, originating in the Gulf of Cadiz (Fig. 1) flows westwards to the Algarve coast and, during periods of prolonged southeasterly winds, can circulate around Cabo S. Vincente and flow northwards (Relvas and Barton, 2002).In relation to the overall patterns of ocean circulation in the eastern Atlantic, the northern part of the west coast of the Iberian Peninsula is influenced by the subpolar branch of the Eastern North Atlantic Central Water (ENACWsp), whereas the southern upwelled waters have characteristics of the ENACW subtropical branch (Fiuza, 1984;Ríos et al., 1992).
The variations in phytoplankton abundance and composition between the northern and southern part of the west coast are primarily a consequence of the distinct topography of the continental shelves and river runoff (Peliz and Fiúza, 1999).In winter, the freshwater runoff induces salinity stratification on the wider and shallower shelf of the northwest coast, favouring the development of phytoplankton blooms.The peak for seasonal phytoplankton abundance occurs in spring and summer.The summer upwelling community is composed of chain-forming diatoms such as Pseudonitzschia spp.and Chaetoceros spp.(Moita, 2001).
These upwelling systems have supported an important fishery resource for the west coast of the Iberian peninsular.In the case of the Algarve, 12.1% of the total licensed fleet is located at Sagres (Martins and Carneiro, 1997;Pita et al., 2002).Furthermore, in recent years, a significant contribution to the local economy has come from the production of 300 tons of oysters at Sagres (Cachola, 1995;European Commission, 1999;pers. comm. Tessier).This aquaculture is dependent on the enrichment of the coastal waters by upwelling as there are no permanent rivers or streams in the area and the anthropogenic contribution is minimal because of the low resident population and limited agriculture.
Despite the importance of the Sagres region for Portuguese fisheries and bivalve culture, studies of production and associated phytoplankton community are scarce.Villa et al. (1997) reported a peak in May and September for phytoplankton based on estimates of chlorophyll a (chl a), and maxima for zooplankton between July and September based on plankton tows.Moita et al., (1998) have observed episodic blooms of toxic dinoflagellates east of Cabo S. Vicente along the Algarve coast.Sampayo et al. (1997) have detected biotoxins, leading to the temporary closure of oyster sales from Sagres.
This study was undertaken during the upwelling season, from May to September 2001, at Sagres, in order to understand the influence of the circulation and upwelling events on the local microplanktonic population and primary production.The monitoring includes several of the elements required by the European Water Framework Directive (WFD, 2000) to assess the ecological status of coastal waters including physico-chemical parameters (temperature, salinity, oxygen and transparency data) and biological parameters (composition, abundance and biomass of the phytoplanktonic community).

Study area
The Algarve coast along southern Portugal extends between 7°20'W and 9°W, along 37°N indented by two major canyons: S. Vicente and Portimão.The west coast off Algarve has an even narrow shelf, about 10 km wide.The sampling station (Fig. 1) was 5 km east of the upwelling centre off Cabo S. Vicente, at the entrance to the Porto Baleeira at Sagres (37°00'63" N and 8°55'62"W), and 3 km west of an offshore "long-line" system for oyster culture (37°00'40"N and 8°53'75"W).Following the requirements of the Water Framework Directive (WFD, 2000), this area is classified as a mesotidal, moderately exposed, coastal water of the Atlantic type (Bettencourt et al., 2004).The location was recently selected as an intercalibration site for the Common Implementation Strategy of the WFD.

Sampling
The Sagres station was sampled weekly, between the end of May and the beginning of September, with an interruption of 19 days in June.Surface water was collected early in the morning, independently of the tidal phase, and filtered through a 200µm mesh size net, to select for the microplankton community and remove the larger grazing organisms and particles.Aliquots for nutrients determination were frozen at -20 ºC for later analysis of ammonium, nitrite, nitrate, phosphate, and silicate, according to the methods described in Grasshoff et al. (1983).Chl a concentration was determined by further filtering 1 l of water sample, through a Whatman GF/F filter, for measurement with a Jasco FP-777 based on the fluorometric methods described by JGOFS (1994).
Water transparency was determined by Secchidisc depth and used for the estimation of the percentage irradiation depth profile.In general, the euphotic zone (defined as the depth at which the light intensity is 1% of the intensity of the surface) was greater than the overall depth of the sample site, which averaged 20 ± 3m depending on tidal fluctuations.Water for the determination of the dissolved oxygen concentration was collected with a Niskin bottle from depths at which the light intensity was 100, 50, 25 and 10% of that at the surface.Oxygen concentrations were determined with triplicates of each sample by the Winkler method (Strickland and Parsons, 1972;Bryan et al., 1976) using a Brand microburette for the titrations and expressing the final concentrations as µM O 2 (± SE).
Sea surface temperature (SST) was recorded with a Tinytalk PT 100 logger attached to a "longline" for oyster culture.Total daily solar irradiance (KJ m -2 ) was recorded by the Portuguese Instituto de Meteorologia (IM) at the Sagres station (8°57'W, 37°00'N, 25 m).Irradiance was converted to photosynthetically available radiation (PAR) using the criteria that PAR roughly represents 45% of total solar radiation (Kirk, 1994).PAR values for the surface layer were estimated based on the equation: where I 0 is the incident radiation, I z the radiation at z depth, and k the Secchi extinction coefficient (Kirk, 1994).
Apart from the 24 July, temperature and salinity profiles were recorded with a Seacat SBE 19 CTD between July and the end of the survey in September.The density (σ t ) was calculated from temperature and salinity data according to the algorithms of Fofonoff and Millard (1983).

Upwelling indices
The Ekman transport of surface water was estimated according to Bakun's (1973) method, and used as a coastal upwelling index: (2) q f C V V f where τ x,y is the wind stress vector, ρ a is the air density (1.22 Kg m -3 ), C D is an empirical dimensionless drag coefficient (1.14x10 -3 , see Large and Pond, 1982), V x,y is the wind speed vector on the sea surface, with magnitude |V|, ƒ is the Coriolis parameter (8.78x10 -5 s -1 for Sagres), and ρ w is the density of seawater (~1025 Kg m -3 ).
Wind direction and magnitude were obtained from the IM station at Sagres.The wind stress vector was divided into its two components (τ x the eastward component, and τ y the northward component), giving an estimation of q x and q y (m 3 s -1 km -1 ) for Ekman transport.Positive values for q x indicate upwelling-favourable offshore Ekman transport along the south coast, whereas negative values of q x represent inshore Ekman transport on the south coast.Conversely, positive values of q y indicate downwelling on the west coast, whilst negative values of q y indicate upwelling-favourable offshore Ekman transport along the west coast.

Production and respiration rates
Production and respiration rates were estimated by the oxygen light-dark bottle technique (Strickland and Parsons, 1972).The filtered samples were siphoned carefully into 300 ml Winkler bottles with silicon tubing to reduce turbulence.Triplicates were fixed immediately for measurement of initial dissolved oxygen concentrations.Triplicates of light and dark bottles were suspended along a 'long-line' and incubated for 24 h, after which they were fixed.
Gross production (GP), net community production (NCP) and dark community respiration (DCR) were determined from the difference between the means of the light, dark, and initial time replicates; rates are expressed as µM O 2 d -1 (±SE) .Rates were converted to carbon units using 1.4 as the photosynthetic quotient (Laws, 1991).

Microplankton identification and carbon content
Microplankton samples were preserved with acidified Lugol's iodine solution.Each sample was placed in a 100 ml sedimentation chamber and settled for observation with a Zeiss Axiovert 25 inverted microscope.Qualitative and quantitative analyses of the samples were based on the methods of Utermöhl (1958).Smaller cells were identified (Tomas, 1997) and counted at 400x magnification up to a total of 100 optical fields, whereas the less abundant and larger organisms were observed over 326 S. LOUREIRO et al. the entire chamber at 100x magnification.Organisms were generally identified down to genus and whenever possible to species level; whenever this classification was not possible cells were included in wider groups (see Table 3).Cell volumes were determined by approximation to the nearest geometric shape (Hillebrand et al., 1999), and converted to biomass carbon units on the basis of formulae devised by Verity and Langdon (1984) and Verity (1992).

Analysis of microplankton assemblage
A statistical study of the microplankton community was completed with PRIMER © software (Plymouth Routines In Multivariate Ecological Research) for a multivariate analysis of the microplankton community.An assessment of natural groupings within the community was completed by multi-dimensional scaling (MDS) ordination using the Bray-Curtis similarity matrices of square-root abundance and biomass data.Significance tests for differences between the a priori established groups were carried out using one-way analysis of similarities (ANOSIM in Clarke and Warwick, 2001).The contribution of taxa to dissimilarities between the different periods (see results for period's definition) were evaluated using the routine for similarity percentages (SIMPER in Clarke, 1993).The non-parametric statistical tests were done with the STATIS-TICA © 6 program.

Stages of the upwelling season
Three periods were distinguished on the basis of the changes in SST during the survey (Table 1): period 1 (P1), from 24 May to 10 July, corresponded to a high temperature stage prior to a persistent upwelling event; period 2 (P2), from 11 July to 31 July, was marked by lower temperatures corresponding to a major upwelling event; finally, period 3 (P3), from 1 August to 3 September, corresponded to a further stage of higher temperatures.

Wind and hydrographic conditions
Figure 2 summarises both the speed distribution and the direction of the wind, and Figure 3b is a stick vector diagram of the time-series for coastal wind speeds.Both figures show the prevailing northerly wind regime, from May to September, with average velocities of 6-10 m s -1 .At the beginning of May (P1), favourable conditions for upwelling on the south coast (inferred by q x >0, Fig. 3a) induced a period of low SST (14ºC), followed by conditions favourable for upwelling on the west coast (inferred by q y < 0).The increasing SST by the end of May was related to a brief reversal in wind direction, leading to the replacement of cold water by warmer waters from the intrusion of the counter current from Cadiz towards the study site at Sagres.In June (P1), the generally high wind velocities and the persistent upwelling on the west coast (q y decreased to -700 m 3 s -1 km -1 ) were linked to a decline in local SST (min.15ºC), suggesting the influence of western, upwelled cold waters on the Sagres site.An increase in SST was recorded in the last week of June (max.18°C), probably reflecting the relaxation of the upwelling conditions on the south coast (q x < 0) followed by the intrusion of the warmer coastal counterflow (07 and 25 June in Fig. 4).During the first few days of July (P2) an upwelling plume extended eastward from Cabo S. Vicente (03 June in Fig. 4) but had still not arrived at Sagres sampling station.In July (P2), SST reached the minimal values, associated with q x and q y favourable to offshore transport.On 11 July cold waters were mainly located south of the Cabo S. Vicente region, with a slight eastward advection extending up to Sagres (Fig. 4).During the rest of the month cold waters extended along the shelf off the Algarve.In August (P3), the SST at Sagres (Fig. 3d) reflected the variability of the wind regime with a cycle of upwelling / relaxation events with a duration SAGRES: MICROPLANKTON COMPOSITION AND PRODUCTION 327 of 14 days.Along the south continental shelf, episodes of relaxation were associated with the influence of the warm counter current.The selected SST satellite images (Fig. 4) are representative of P1 (24 May-3 July), P2 (11 July-28 August) and P3 (4 August-1 September).
Local SST was negatively correlated (Spearman; p<0.05) with average weekly values for q x taken from the previous 7 days, and positively correlated with average weekly values for q y (Table 2) over the period of the survey.Overall, the upwelling events adjacent to Sagres seemed to be influenced by the interplay between water circulation driven by the winds along the west and south coasts.

Time-series of depth profiles
Figure 5 shows a series of depth profiles for O 2 and temperature measured during the survey.
24 May -10 July (P1).At the beginning of the study the water column was homogeneously oxygenated (248±0.3µM O 2 , n = 12).On 31 May, the SST maximum was complemented by a minimal oxygen concentration at the surface (233±0.2µM O 2 ).In June, the two available oxygen profiles presented similar distribution patterns, and by 3 July a subsurface (9.5 m) minimal oxygen value (222±0.2µM O 2 ) indicated the possibility of intrusion at Sagres of oxygen-deficient, upwelled waters.
11 July -31 July (P2).This period of matureupwelling was characterised by colder temperatures at all depths.On 11 July, the thermocline (0.2°C m -1 ) was at a depth of 11 m.By 18 July, the surface water was warmer, and a steeper (0.4°C m - 1 ), shallower (6 m) thermocline had developed: salinity and density (σ t ) profiles (Fig. 6) demonstrated a stratification on this date.However, oxygen profiles were generally homogeneous.On 31 July, high pelagic oxygen concentrations (277±2 µM O 2 , n = 12) probably reflected a recent active blooming phase.The pycnocline was associated with a less saline surface layer.
1 August -3 September (P3).The oxygen decreased, as is typical of post-blooming periods.On 14 August there was an increase in surface stratification (pycnocline of 5-10 m) associated with a warmer, less saline layer.This probably reflects the intrusion of the warm coastal countercurrent coming from the Gulf of Cádiz.On 20 August there was a steep shallow thermocline (4 m), below which was a layer of both low oxygen (227±1 µM O 2 , n = 6) and low temperature (13.8ºC).The subsurface oxygen SAGRES: MICROPLANKTON COMPOSITION AND PRODUCTION 329 ) and biomass (BMdiat, BMdino, BMnano, BMcilia, Bmtotal, µgC l -1 ), oxygen concentration ([O 2 ], µM), eastward (q x ) and northward (q y , m 3 s -1 km -1 ) Ekman transport component.Bold figures are significant at p < 0.05; n represents the number of samples.

Physical, biological and chemical parameters
Table 1 summarises the ranges of surface physical, chemical and biological parameters.PAR was high throughout the survey, with a maxima during P1 (Fig. 7b).The density attained a maximum (26.7 kg m -3 ) in July (P2), confirming the upwelling of denser water masses.The highest transparency val-ues for the water column (Fig. 7a) were recorded at the beginning of the study and on 20 August (11 m).The depth of the euphotic layer, calculated from Secchi disk data, was 19-40 m in May-June (P1).This was reduced to 19-24 m during the July (P2) upwelling/blooming event and then increased in August (P3) to 19-30 m.
In May-June (P1), chla surface values averaged 1.8±0.5 µg l -1 , followed by a significant increase during the July (P2; ANOVA p <0.0001, post hoc LSD Fisher test) upwelling episode (Fig. 7a), with the maximum of 6.2 µg l -1 .The August (P3) decline (min.1.7 µg l -1 ) was followed by a steady rise until the end of the survey, implying the development of a new bloom.
The changes in relative composition of the systematic groups in each period are evaluated in Figure 11a, b.There was a clear dominance of diatom abundance throughout the survey, reaching a peak (95%) in July (P2).May-June (P1) showed a biomass with a balanced composition of diatoms (37%), dinoflagellates (38%) and ciliates (22%).The biomass contribution of dinoflagellates in May-June (P1) was mainly due to Protoperidinium spp.and Ceratium spp.Tintinnina and oligotrichida were confirmed as the principle contributors to the biomass of ciliates.

Statistical assemblage analysis
The MDS plots evidenced three distinct groupings corresponding to the May-June (P1), July (P2), and August (P3) stages, both for abundance (Fig. 12 SAGRES: MICROPLANKTON COMPOSITION AND PRODUCTION 333 a, b) and for biomass (Fig. 12 c, d).The global R (a statistical measure of the degree of separation of groups) resulting from the one-way ANOSIM tests for abundance data (Table 4) implied the rejection of the null hypothesis (no assemblage differences between P1, P2 and P3) at the 0.002 significance level.However, the pairwise R values (resulting from the comparison of the specific pairs of groups) showed a weak separation (R = 0.37) between the community structures in May-June (P1) and August (P3).The May-June (P1) and July (P2) groups were significantly different (R > 0.5); finally, July (P2) and August (P3) showed a well-separated community composition for abundance (R > 0.75).Biomass followed a similar statistical pattern to the community composition.
The result of SIMPER analysis is represented on Table 5.The highest average dissimilarities were found between July (P2) and August (P3) for abundance (δ = 54.45), and between May and June (P1) and July (P2) for biomass (δ = 66.01).May-June (P1) and August (P3) were the most similar periods for both abundance and biomass data, confirming the values obtained in the ANOSIM test and the MDS ordination.Chaetoceros spp., Thalassiosira spp., Lauderia spp., Detonula spp., and Pseudonitzschia spp.were the main taxa contributing to the dissimilarities between July (P2) and the other periods.Figure 13 shows the temporal distribution of the main taxa contributing to the abundance and biomass dissimilarities between the different periods.

Potentially HAB organisms
Identification was mainly done down to genus level, so differentiation of harmful species within each taxa was not detected.Nevertheless, Table 6 presents the temporal distribution of algal taxa associated with harmful algal bloom (HAB) events identified during the survey at Sagres.Pseudo-nitzschia spp., a taxon that includes toxic species associated with amnesic shellfish poisoning (ASP; Bates et al., 1998), had the highest values for abundance in July (P2; 178±58 x10 3 cell.l -1 ).Water discolorations, commonly called red tides, are produced by SAGRES: MICROPLANKTON COMPOSITION AND PRODUCTION 335 TABLE 5. -Taxa contribution (%) to the average (a) abundance and (b) biomass Bray-Curtis dissimilarity (δ), between the three defined sampling periods (P1, P2 and P3).Data were square root transformed.Taxa were selected until ~50% of the cumulative dissimilarity was attained (for taxa codes see Table 3).-Temporal distribution of (a) abundance and (b) biomass of the main taxa contributing to Bray-Curtis dissimilarities between the defined sampling periods: P1, P2 and P3 (see Table 3 for taxa codes).Circles are proportional to abundance (max.637 10 3 cell.l -1 ) and biomass (max.390 µgC l -1 ) values; to avoid overlapping of circles, they represent 50% of their original size; as such, the absence of a bubble does not necessarily mean no occurrence, but that the relative abundance is low.TABLE 6. -Abundance (10 3 cell.l -1 ) of potentially HAB organisms (Hallegraeff, 1995;Pitcher and Calder, 2000;Smayda, 2000) from May to September (2001) at Sagres station.See Table 3  Ceratium spp., Gonyaulax spp., and Scrippsiella spp., amongst other organisms (Pitcher and Calder, 2000;Smayda, 2000).These blooms, although nontoxic, are undesirable because they may cause fish and invertebrate killings due to oxygen depletion, following the decay of the blooms.Ceratium spp.occurred at low values (< 4 x10 3 cell.l -1 ) and was basically characteristic of May-June (P1) and August (P3).Gonyaulax spp.only occurred once in May-June (P1), and Scrippsiella spp. was prominent in August (P3; 0.3-16.8x10 3 cell.l -1 ).Organisms with the potential to cause paralytic shellfish poisoning (PSP), such as Alexandrium spp.and Gymnodinium spp., were also recorded.Alexandrium spp.occurred only in low numbers in August, whilst Gymnodinium spp.occurred throughout the survey, with the greatest abundance in May-June (P1) and August (P3; 47 x 10 3 cell l -1 ).Dinophysis spp.and Prorocentrum spp., related to diarrhetic shellfish poisoning (DSP), were absent in July (P2), but occurred in May-June (P1) and August (P3).

Physical events and microplankton assemblage
Although the lack of sufficient vertical data limits an understanding of the whole dynamics in threedimensions of the study site, the results show that during the upwelling season the Sagres region is influenced by the wind-driven circulation along the south and west coast, which forces cold, upwelled water into the surface layer.Upwelled water masses have characteristics of the ENACW subtropical branch (temperature > 13°C, σ t < 27.1 kg m -3 ).These findings are consistent with the patterns already described for the Algarve coast (Fiúza, 1983(Fiúza, , 1984;;Sousa and Bricaud, 1992).Winds are mostly moderate (6-8 m s -1 ), and relatively intense velocities (8-10 m s -1 ) are only registered in July (P2), revealing a decrease in wind stress conditions in comparison with previous years (Relvas and Barton, 2002).The influence of the warm counterflow on the south coast during episodes of relaxation (Relvas and Barton, 2002) has been noticed on several occasions.
Chl a peaks earlier (July) than has been reported (September) for the same area by Villa et al. (1997), probably owing to the interannual variability of physical factors (Peliz and Fiúza, 1999).The seasonal values for chl a and chemical parameters are in general agreement with the ranges described for the Cabo S. Vicente region (Moita, 2001).However, lower values of phosphate and silicate may imply the occurrence of a spring-bloom before the beginning of the survey.The maximal values for chl a (6.2 µg l -1 ) attained in July are similar to those reported for the upwelling regions of NW Spain -La Coruña (6.7 µg l -1 Casas et al., 1999) and Chile (6.2 µg l -1 , Daneri et al., 2000), but lower than those of other upwelling systems such as Orgeon (1-57 µg l -1 , Dickson and Wheeler, 1995), Benguela, NW Africa, and off Peru (5-50 µg l -1 , Andrews and Hutchings, 1980;Estrada, 1974;Blasco, 1971 respectively).The lack of correlation between the Secchi-depth and chl a (Fig. 7), particularly during the bloom stage in July (P2) when Secchi values did not decrease as expected, may be due to several factors.The Secchi-disk depth is a measure of the concentration of light attenuating particles in the water column, whether from phytoplankton or non-phytoplankton sources.Factors contributing to the variation in Secchi-depth include the sun angle, sea surface reflectance and tidal height (Edmonson, 1980;Preisendorfer, 1986, Borkman andSmayda, 1998).
In the current study, observations have been made independently of tidal phase.It may also be associated with variability in changes in chl a content per cell, carbon:chl a ratio, or chl a to accessory pigment ratio (Falkwoski and LaRoche, 1991).
Several upwelling pulses were registered from late spring to late summer (Fig. 3).The first pulse in June (P1) fertilises the surface water with nutrients, but its evolution was not followed by this survey.A more persistent-active upwelling event develops in July (P2), fertilising the surface with concentrations of nitrate up to 19 µM.This value is higher than that reported for NW Spain (La Coruña, 9.8 µM, Casas et al., 1999;Ria de Vigo, 12 µM, Moncoiffé et al., 2000).
Diatom biomass and density is dominant throughout the survey, and its temporal evolution is positively correlated (Spearman, p<0.05) with chl a and negatively correlated with SST, implying an association with cold waters supplied by upwelling.The maximal diatom abundance (1366 x10 3 cell.l -1 ) is typical for other upwelling regions (10 6 cell.l -1 , refs. in Moita, 2001): NW Iberian-Galicia (Estrada, 1984), NW Africa (Blasco et al., 1980), Peru (Blasco, 1971) and Benguela (Giraudeau et al., 1993).The persistent diatom-chl a peak (≈ 21 days in July, P2) is related to prolonged conditions favourable to upwelling.This group is adapted to turbulent conditions (Margalef, 1978).The fact that ammonium peaks are not coincident with oxygen minima, together with the predominance of low ammonium levels (< 0.5 µM), may imply pelagic nutrient regeneration as a secondary process during the survey period.Positively or neutrally buoyant diatoms could also partially explain the persistent bloom (refs. in Tremblay et al., 2002).
The bloom collapse seems to be associated with a decrease in conditions favourable to upwelling, together with episodes of stratification in the water column, probably caused by the influence of the warm countercurrent.Nevertheless, the transition to a well-established stratified surface layer, which is a condition for the development of the classical diatom-dinoflagellate succession (Margalef, 1978), does not occur because of the fortnightly cycles of upwelling and relaxation, typical of temperate upwelling conditions (Walsh et al., 1977).
Dinoflagellate abundance is positively correlated (Spearman, p<0.05) with temperature, suggesting an association with the warm waters of the countercurrent.Lingulodinium polyedrum has been described for this location by Amorim et al. (2004).Its absence from the samples in this study may be due to the sampling hour (early morning), when diel vertical migration limits its presence in surface waters, or to the inclusion of this species in higher classification groups.This species seems to be associated with coastal retention conditions in the Sagres area that may develop at times of relaxation when the cold waters are replaced by the warm waters of the countercurrent.Water retention has been reported in several upwelling areas (Graham and Largier, 1997;Demarcq and Faure, 2000;Marín et al., 2003).Coccolithophorids have been observed in the Cabo S. Vicente region (Abrantes and Moita, 1999;Cachão and Moita, 2000), but they have not been quantified because the calcareous plates may be damaged by preservation with acidic Lugol's solution.

Potentially HAB
The Pseudo-nitzschia spp.reached high abundances (171 x10 3 cell.l -1 ) during this study.Nevertheless, this taxon includes toxic and nontoxic organisms.In order to evaluate the potential harmful effects of this species, a joint study of occurrence of organisms and detection of total biotoxin and biotoxin per cell must be undertaken.In Portugal, IPIMAR is the National Reference Laboratory for biotoxins.Potentially harmful dinoflagellate taxa (Alexandrium spp., Ceratium spp., Dinophysis spp., Gonyaulax spp., Gymnodinium spp., Prorocentrum spp., and Scripsiella spp.) were also recorded.Since 1994, Gymonidinium catenatum blooms have been registered east of Cabo S. Vicente, and their presence seemed to be dependent on upwelling nutrient enrichment (Moita et al., 1998).Regarding Dinophysis spp., concentrations of < 500 cell.l -1 were already reported as agents of human intoxication in Portugal, leading to the closure of bivalve harvest (Vale, 1999).During this survey, higher concentrations were attained (max.2600 cell.l -1 ).These values fall within ranges previously described for the Portuguese coast (Moita and Silva, 2000;Palma et al., 1998).
In a region such as Sagres where bivalve culture occurs, precautionary closure of the zone should be carried out for abundances of 200-1000 cell.l -1 for Dinophysis spp., Gymnodinium catenatum, and Alexandrium minutum and > 100000 cell.l -1 for Pseudo-nitzschia spp.(European Commission, 2002).The closure should be maintained until the respective biotoxin analysis is found to be negative.

Production and respiration rates
Production maxima are attained in July, concurrent with the diatom-chl a peak.The seasonal average of volumetric GP (25.4±19.8µM O 2 d -1 ) is higher than for the systems of Chile (11.5 µM O 2 d -1 ; Daneri et al., 2000), Arabian Sea (6.8 µM O 2 d -1 ; Robinson and Williams, 1999), NW Africa and Benguela (15.2 µM O 2 d -1 , 14.4 µM O 2 d -1 respectively, Robinson et al., 2002), but lower than for the Ría de Vigo-NW Spain area (37.3±30.7 µM O 2 d -1 , Moncoifée et al., 2000).DCR, on the other hand, is generally lower than reported for the above systems, representing only 17% of the GP, which reflects the predominance of the autotrophic component throughout the survey.The high significant correlations between total microplankton, chl SAGRES: MICROPLANKTON COMPOSITION AND PRODUCTION 337 a, diatoms, production and oxygen data (Table 2) also suggest a dominant and active community of diatom-producers.
Following the approach of Blight et al. (1995), GP was plotted against respiration to study the phasing of these parameters (Fig. 14).It is generally observed that the autotrophic peaks are not coupled with heterotrophic maxima, denoting a temporal lag between the two processes.This feature has been reported for other coastal areas (e.g.Blight et al., 1995;Robinson et al., 1999) and is probably associated with natural physical loss mechanisms (dispersion, sedimentation) in upwelling areas.However, from date 4-5 (May-June, P1), and date 12-13 (August, P3), the increase in GP is related to an increase in respiration rates.Although the lack of bacterioplankton data limits the interpretation of these findings, high temperatures were recorded on day 5 and 13, which usually favours picoheterotrophic activity (Wiebe et al., 1993).Additionally, on both occasions there was a peak for ciliate abundance, the best biological predictor of DCR according to Spearman's correlation.A more efficient transfer from the auto-to the heterotrophic communities can be associated with a low molecular weight (LMW) pool of organic matter, originating from algal exudation, readily assimilated by heterotrophs (Blight et al., 1995).
The autotrophic maximum (18 July) is coincident with the diatom bloom in July (P2).The heterotrophic maximum (31 July) is associated with a ciliate peak, together with a diatom maximum, a silicate minimum and a low PAR value, which sug-gests a co-limitation of light and nutrient on the diatom-photosynthetic rate (Kudela and Dugdale, 2000).The net heterotrophic period (NCP < 0) on 14 August occured during an episode in which nutrients were not limiting (nitrate: 8.5 µM; phosphate: 0.3 µM; silicate: 1.9 µM) but the value for PAR is low.This can be interpreted as a light limitation of the production rate (e.g.Ryther, 1956;Kirk, 1994).Cloud coverage can affect rates of production by a factor of up to 4.5 (Riegman and Colijn, 1991).Also, the decline in diatoms by this date is accompanied by an increase in the remaining functional groups (ciliate, dino-and nanoflagellate), contributing to a higher heterotrophic component.This transition period of the microplankton composition is probably associated with the intrusion of the warm coastal counterflow and the consequent stratification described above.As suggested for other systems (Moncoifée et al., 2000;Robinson et al., 2002), the observed heterotrophy could have been sustained by the accumulation of organic substrates from a recent bloom.The persistently high oxygen saturation (107%) measured at this time corroborates this hypothesis (Robinson et al., 2002).A contribution to the dissolved organic matter pool from the excretion of hanging mussels has also been reported (Álvarez-Salgado et al., 1996).

CONCLUSIONS
The Sagres area is subjected to the upwelling of cold waters in spring to late summer, originating in the wind-driven circulation patterns off the south and west coast.The temporal variation of these physical events regulates the influx of nutrients to the surface waters and subsequent microalgal growth, sustaining the phytoplankton biomass and production of the system.The long-lived diatom-chl a peak throughout July is probably associated with the persistence of the upwelling event.The collapse of this diatom bloom appears to be related to the decrease in upwelling conditions and the stratification of the water column, probably induced by the intrusion of the warm inshore water mass.These features imply a physical control of the biological development.Chaetoceros spp., Thalassiosira spp., Lauderia spp., Detonula spp., and Pseudo-nitzschia spp.can be considered as an upwelling proxy for this site.The progression of an upwelling / relaxation cycle determined the attained succession stage,  therefore regulating the composition of the microplankton assemblage and the subsequent nature of transfer to higher trophic levels, sediments and export.Low respiration rates (17% of GP) and uncoupling with production peaks appear to stem mainly from the interplay of the predominant autotrophic component and physical loss factors.
Altogether, physical events seem to be the main factor influencing microplankton structure and production in this area.
More work needs to be done to understand the whole dynamic of this ecological productive system, including water-column studies of new production, bacterial rates, regeneration processes and grazing pressure.Also, the benthic and atmospheric domain awaits further study to improve the understanding of the ecosystem behaviour.Nevertheless, the present study brings a valuable insight into the productive waters of the Sagres area.

FIG. 2
FIG. 2. -Chart of wind direction and speed distribution (%) from May to September 2001 at Sagres.
.2 µM O 2 ) was at 3 m adjacent to the thermic surface layer.The water column temperature rose towards the end of August (max.18°C), but declined rapidly in September (min.16°C), reflecting the relaxation / upwelling cycles referred to previously.

TABLE 2 .
-Spearman rank-order correlation between biological, chemical and physical parameters determined during the sampling season: net community production (NCP, µM O

TABLE 3 .
-List of identified microplankton taxa, its codes, and frequency of occurrence during the survey.Most frequent taxa (≥50%) in bold type.

TABLE 4 .
-One-way ANOSIM test for microplankton assemblage differences (in square root transformed abundance and biomass data) between the three a priori groups (P1, P2 and P3).