Harmful phytoplankton diversity and dynamics in an upwelling region (Sagres, SW Portugal)

revealed by ribosomal RNA microarray combined with microscopy.

Danchenko, Sergei1,2; Fragoso, Bruno1,2,4, Guillebault, Delphine3; Icely, John 1,4; Berzano, Marco5;

Newton, Alice6,7.

1CIMA – Centre for Marine and Environmental Research, University of Algarve, Campus de Gambelas, Faro 8005-139, Portugal 2Facultad de Ciencias del Mar y Ambientales, University of Cadiz, Campus de Puerto Real, Polígono San Pedro s/n, Puerto Real, 11510, Cadiz, Spain 3 Microbia Environnement, Avenue Pierre Fabre, 66650 Banyuls sur Mer, France 4Sagremarisco Lda., Apartado 21, Vila do Bispo 8650–999, Portugal 5Polytechnic University of Marche, Piazza Roma, 22, 60121 Ancona AN, Italy 6FCT – University of Algarve, Campus de Gambelas, Faro 8005-139, Portugal 7NILU-IMPEC, Box 100, 2027 Kjeller, Norway

Corresponding author: Sergei Danchenko, CIMA, FCT, University of Algarve, ed. 7, Piso 1, Cacifo 32, Campus de Gambelas, Faro 8005–139, Portugal. Tel.: +351 935076102 (E-mail address: danchenko- [email protected]; [email protected]). E-mail addresses: [email protected], [email protected] (S. Danchenko)

Abstract

The study region in Sagres, SW Portugal, is subject to natural eutrophication of coastal waters by wind- driven upwelling, which stimulates high primary productivity facilitating the recent economic expansion of bivalve aquaculture in the region. However, this economic activity is threatened by harmful algal blooms

(HAB) caused by Pseudo-nitzschia spp., Dinophysis spp. and other HAB , all of which can produce toxins, that can induce Amnesic Shellfish Poisoning (ASP), Diarrhetic Shellfish Poisoning (DSP) and

Paralytic Shellfish Poisoning (PSP). This study couples traditional microscopy with 18S/28S rRNA microarray to improve the detection of HAB species and investigates the relation between HAB and the specific oceanographic conditions in the region. Good agreement was obtained between microscopy and microarray data for diatoms of genus Pseudo-nitzschia and dinoflagellates Dinophysis spp., Gymnodinium catenatum and raphidophyte Heterosigma akashiwo, with less effective results for Prorocentrum. Microarray provided detection of flagellates Prymnesium spp., Pseudochattonella spp., Chloromorum toxicum and the important

HAB dinoflagellates of the genera Alexandrium and Azadinium, with the latter being the first recorded from the study region. Seasonality and upwelling induced by northerly winds were found to be the driving forces of

HAB development, with Pseudo-nitzschia spp. causing the risk of ASP during spring and summer upwelling 1 season, and dinoflagellates causing the risk of DSP and PSP during upwelling relaxation, mainly in summer and autumn. The findings were in agreement with the results from toxicity monitoring of shellfish by the

Portuguese Institute for Sea and Atmosphere (IPMA, 2014-2016) and confirm the suitability of the RNA microarray method for HAB detection and aquaculture management applications.

Keywords: microarray; HAB; shellfish aquaculture; upwelling; Pseudo-nitzschia; Dinophysis; Azadinium.

Introduction

Harmful algal blooms (HAB) threaten public health through the contamination of seafood with biotoxins causing economic damage to aquaculture, as well as threatening tourism through negative impacts on the environment (Shumway et al., 2018; Davidson et al., 2016; Trainer et al., 2010). HAB regularly occur in marine and brackish waters worldwide, especially in the regions affected by anthropogenic or natural nutrient enrichment, such as upwelling zones along the western continental margins (Kudela et al., 2005, 2010). The

Sagres area is located at the south-west extreme of the Iberian peninsula (Fig. 1) and is subject to seasonal upwelling as part of the West Iberian upwelling system (Fiúza et al., 1982), mostly active during summer (June

– September). Despite the threat by HAB events, shellfish aquaculture is an economic activity recently introduced into the area, with its development directly dependent on phytoplankton primary production.

Previous studies in the area have shown the regular presence of toxic species and demonstrated the dependence of phytoplankton blooms on wind-induced upwelling. The presence of HAB species, including diatoms Pseudo-nitzschia Peragallo, has been reported generally from March to August (Goela et al., 2014;

Loureiro et al., 2011, 2005). This genus includes species responsible for the production of domoic acid causing amnesic shellfish poisoning (ASP) in humans. Monitoring programs carried out by the Portuguese Institute of

Sea and Atmosphere (IPMA) regularly detect significant concentrations of toxins in the tissue of shellfish from the study area (IPMA, 2013) and corresponding HAB phytoplankton, including ASP-producing Pseudo- nitzschia species, DSP-producing Dinophysis Ehrenberg, and PSP-producing Gymnodinium catenatum

H.W.Graham, 1943. Therefore, in the interest of supporting the aquaculture industry and public health, it is important to study the composition and the conditions for development of HAB in the study region.

2

Monitoring of phytoplankton communities in the marine environment is required by the European Union

(EU) legislation, such as the Marine Strategy Framework Directive (MSFD, 2008/56/EC and 2017/845/EC) and

Water Framework Directive (WFD, 2000/60/EC), as well as the national law of the member states (Garmendia et al., 2013). The recent legislation is focused on the indicators for marine ecosystem functioning, such as biodiversity descriptors, and at the same time requires the methods of data acquisition to be both time- and cost- effective (Borja et al., 2016). Traditional identification by light microscopy (LM) is dependent on expert training and experience; therefore, the reproducibility rate of identification between experts can be low, e.g. as low as 38% for certain species of dinoflagellates (Dinophysis, Culverhouse et al., 2003). Molecular tools can successfully complement the traditional microscopy methods, and even provide more precise detection of target species, such as harmful groups (Bourlat et al., 2013). Microarrays are state of the art technology in molecular biology for the processing of biological samples capable of the detection of target RNA/DNA sequences, although they are not commonly used in environmental monitoring (Medlin and Orozco, 2017). The universal microarrays capable of rapidly detecting the presence of specific harmful algal species are especially useful for monitoring. In contrast to DNA- based analysis, the RNA -based method has the advantage of detecting physiologically active cells, and is affected only to a lesser degree by dead organisms or their biochemical traces in the environment (Medlin, 2013). Thus, RNA-based genomic methods correspond well with the legal requirements of the recent legislation (MSFD) for the assessment of the functional state of ecosystems.

In the current study, a multispecies RNA microarray (IEMArray) was used to identify and quantify simultaneously several HAB species on a single array (Medlin et al., 2013; McCoy et al., 2015) which could be a novel method for the routine detection of toxic species in environmental samples. The IEMArray is the result of further development of the MIDTAL microarray project, and is commercially available from Microbia

Environnement (patent number EP20130773308). This tool offers higher taxonomic resolution, and is potentially more time- and cost-effective and less subjective compared to previous research on harmful phytoplankton in the study region, using LM (Loureiro et al., 2005, 2011) and chemotaxonomy (CHEMTAX,

Goela et al., 2014).

The aim of this study was to improve detection of toxic phytoplankton species in an upwelling area designated for bivalve aquaculture by applying the ribosomal RNA microarray method in combination with 3 conventional inverted microscopy. The objectives were: 1) to study the phytoplankton community at the aquaculture-production site; 2) to compare traditional microscopic with molecular methods in terms of identification and detection of harmful species; 3) to relate the phytoplankton community dynamics to oceanographic parameters, such as sea surface temperature and upwelling index.

The research questions were as follows. 1) Which species of toxic phytoplankton can be detected by the microarray in the study area? 2) How does species identification obtained by microscope counts compare to the molecular method? 3) What are the oceanographic conditions in which harmful species tend to develop in the study area?

2. Methods

2.1. Study area and sampling

The study area (Fig. 1) is located near the Sagres fishing harbour (Porto da Baleeira) in the vicinity of

Cape Saint Vincent, at the south-west of the Iberian Peninsula. The oceanography of the region is driven by seasonal upwelling along the west coast, induced by persistent north winds in spring and summer with maxima during July to September (Fiúza et al., 1982, Haynes et al., 1993). Upwelling along the west coast may influence the areas further east around Cape Saint Vincent (Cravo et al., 2010). Because of the specific pattern of water mass circulation along the south-western coast of Iberia and the Gulf of Cadiz (Relvas and Barton,

2005; Sánchez et al., 2006), the study area is subject to rapid changes in oceanographic conditions caused by the combination of wind stress induced water transport, warm coastal counter-current from the Gulf of Cadiz

(Garel et al., 2016) and other factors, especially during winter months, November to February (Relvas et al.,

2007).

Sampling was carried out at the station located approximately 5 km east from Cape Saint Vincent, and 1 km south of Sagres port (37o00’39’’ N, 8o53’58’’ W), near offshore installations for bivalve aquaculture. Sea water samples were collected from the surface layer at a 0.5 m depth. Samples for microscopy were collected in

330 ml plastic bottles, fixed with acidic Lugol iodine solution and kept in the dark and cool until analysis.

Between 0.5 and 1.0 L of water was filtered through 0.45 µm Millipore nitro cellulose or polycarbonate filters, using a low pressure pump; each filter was placed into a 2.5 ml cryovials, fixed with 1-2 ml of TRI-reagent 4

(Ambion - Sigma) and stored in liquid nitrogen, or frozen at -20 oC if liquid nitrogen was not available at the time of sampling, and transferred later for storage in liquid nitrogen until analysis.

Twenty six samples were collected between the 7th July 2014 and the 14th June 2016 for the microscopic

(26 samples) and ribosomal RNA microarray analysis (25 samples); nutrient samples were collected from 23rd

March to 14th June 2016.

2.2. Sea surface temperature (SST), wind, upwelling index and nutrients

To quantitatively describe the upwelling conditions in the study area, upwelling indices were derived from wind stress, based on the Ekman transport. The Ekman transport (Qx,y) of surface water under wind forcing was calculated following Bakun (1973) and Cropper et al., (2014), as:

Qx= τy/ρwf = (v w ρa Cd)1000/f ρw and

Qy= - τx/ρwf = - (u w ρa Cd)1000/f ρw ,

where τx,y is the wind stress, u and v are the wind speeds in longitudinal and meridional directions

-3 respectively, w is the module of wind speed, ρa is the air density (1.22 kg m ), Cd is the dimensionless drag coefficient (1.14×10-3, Large and Pond, 1982), f is the Coriolis parameter (8.78×10-5 s-1 for the study area) and

-3 ρw is the density of sea water (1025 kg.m ).

The orientation of the coastline in the study area (Fig.1) is generally parallel to the meridian (west coast) and the equator (south coast). The study site is located near the point where the coastline abruptly changes direction from sub meridional to sub lateral. Therefore, the latitudinal (Qx) and longitudinal (Qy) components of

Ekman transport were considered to represent the upwelling indices for the western and southern coasts respectively. Negative values of Qx and Qy indicated upwelling favourable conditions along the western and southern coasts in the study area, respectively.

Data used for wind speed and direction were acquired from the Blended Daily Averaged 0.25-degree Sea

Surface Winds product (at 10 m level) by the National Oceanic and Atmospheric Administration (NOAA) and

National Climatic Data Center (Zhang et al., 2006). The sea surface temperature (SST) time series used were extracted from the NOAA Optimum Interpolation (OI) daily SST (AVHRR only) at 0.25-degree resolution

5 product (Reynolds et al., 2007). The data was accessed via the NOAA Environmental Research Division Data

Access Program (ERDDAP, http://upwell.pfeg.noaa.gov/erddap/index.html).

For nutrient analysis, water was sampled and stored in 0.5 L plastic bottles at -18 oC. The methods for the determination of nutrients were based on Molecular Absorption Spectrophotometry UV/VIS, and the techniques used were from the manual Standard Methods for the Examination of Water and Wastewater 20st

Ed. (APHA, SMEWW, 2005): nitrate SMEWW 4500-NO3 - F (sulfanilamide, automated cadmium reduction column), nitrite SMEWW 4500-NO2 - B (sulfanilamide), ammonia SMEWW 4500-NH3 - G (indophenol blue), phosphate: SMEWW 4500-P - F (ascorbic acid), silicate SMEWW 4500-SiO2 - E (molybdate-reactive silica). In situ Chl a sampling and analysis was carried out as described in Goela et al. (2014).

2.3. Microscope counts

The microscopic study of the phytoplankton was based on the Utermohl (1931) method modified by

(Lund et al., 1958). Each sample was fixed with acid Lugol solution, sedimented over 24 hours in a 50 ml chamber. Phytoplankton taxonomic identification and quantification was carried out using Zeiss Axiovert 15 inverted microscope. In order to achieve a representative sample, 20 to 70 fields of view were examined at 400x magnification, with not less than 300 cells in total of the most abundant species identified and counted. The minimum size of the cells detected was approximately 3 µm. The whole chamber area (530.29 mm2) was examined at 100x magnification to detect the rarer specimens at sizes equal to or greater than 20 µm; at high densities the area of the chamber under examination was reduced by half. A single count per sample was performed, as the above counting rules were considered to provide sufficient replication (Lund et al., 1958).

Taxonomic identifications were made according to Dodge (1982), Tomas (1997), Hasle and Syvertsen (1997),

Steidinger and Jangen (1997) and Throndsen (1997); synonymy of taxa was checked using AlgaeBase (Guiry and Guiry, 2018). The taxonomic authority of the dinoflagellates in this study is referenced according to

Gómez (2005), unless stated otherwise. Pseudo-nitzschia spp. cells were identified into two groups: Pseudo- nitzschia group delicatissima, width <3 μm, linear shape, although often distorted under Lugol fixation, and

Pseudo-nitzschia group seriata, width >3 μm, lanceolate shape of visually more robust cells (following the approach of e.g. Fehling et al., 2006 and Paterson et al., 2017). 6

2.4. Microarray laboratory procedure

RNA extraction, labelling and microarray hybridisation were performed generally following the procedures described in Lewis and Medlin (2012), Medlin et al. (2013) and Taylor et al. (2013).

Total RNA was extracted from samples using Tri-Reagent following the manufacturer’s instructions

(BCP extraction) and precipitated with isopropanol. Dunaliella tertiolecta (500,000 cells) was added to the sample at the bead-beating step as a RNA extraction and hybridisation control. After the final centrifugation step, the pellet was suspended in RNase free water and stored at -80°C. Total RNA was labelled using a Cy5

Platinum Bright 647 Infrared Nucleic Acid kit (Leica Biosystems, Germany), fragmented to a size range of 100 base pairs and hybridised for the prepared microarray slides at 65°C for 30 minutes (IEMArray, Microbia

Environnement, France). Hybridisations were done in two technical duplicates for most of the samples (two microarray spotted grids located on one slide), except samples 16-Oct-2015, 23-Mar-2016, 17-May-2016 and

27-May-2016. The 4X hybridisation buffer was prepared to a 1x final concentration (1M NaCl/10 mM Tris, pH

8/0.005% Triton X-100/0.5 mg.ml-1 BSA/ 0.1 µg.µl-1 HS-DNA, 5 ng Positive Control) then mixed with 1000 ng of labelled total rRNA. A TBP-fragment of Sacchromyces cerevisiae Meyen ex E.C. Hansen was added as the positive control. The hybridisation solution was placed at 95°C for 5 minutes to denature the RNA secondary structures then immediately placed on ice. Kreablock (Leica Biosystems; Germany) was added to the hybridisation mixture for a final volume of 37 µL. A cover slip (Lifter Slip; Implen, Munich, Germany), was placed on each array of the slide. A volume of 18.5 µl hybridisation solution was pipetted under the cover slip, and capillary action ensured an even dispersal of hybridisation solution throughout each array. Slides were placed in a humid chamber and the hybridisation was conducted at 65°C for 30 minutes. After hybridisation, coverslips were withdrawn and slides were sequentially washed in saline sodium citrate containing washing buffers (2x SSC/10 mM EDTA/0.05% SDS; 1x SSC/10 mM EDTA) 1 and 2 for 10 minutes at room temperature, than in washing buffer 3 for 10 minutes at 50°C. Slides were dried by centrifugation for 1 min at minimum speed.

2.5. Image and data analysis 7

The chips were scanned using a GenePix 4000B scanner (Molecular Devices LLC, USA). Analysis of the fluorescent signal intensities was performed by superimposing a grid of circles, stored as a GenePix Array List

(GAL) file format, onto the scanned image to identify all the names and positions of the probes on the arrays in order to calculate their respective fluorescent signals and the associated background intensity. Image analysis was performed using Spotxel Image Analysis software (Sicasys, GmbH, Germany) with the integrated

Microbia Environnement program module (plug-in) version 1.5.2.17 for all samples except for samples from

14-Jul-2014 until 19-Aug-2014. For these three samples, image analysis was performed using the GenePix®

(Axon) with contrast increase, resulting fluorescent signals data stored in GPR format was analysed using the

GPR-Analyzer software version 1.28 (Dittami and Edvardsen, 2012). Where applicable, the microarray data related to these three samples are shown on a different Y-scale in the figures from those processed with the

Microbia analysis. Signal intensity of each probe was normalized by dividing its total signal intensity by the intensity of the positive control probe POSITIVE_25_dT (TBP) or DunGS02_25_dT probe (following the approach of Dittami and Edvardsen, 2012), responding to a control standard of 500,000 cells of Dunaliella tertiolecta, added to each sample before RNA extraction. The normalised signal to noise ratios were used in further analysis. Data visualization shown in all figures is based on the positive control (POSITIVE_25_dT) normalised signal. Statistical analysis was carried out using MS Excel 2013 and Statistica 6.0.

3. Results

3.1. Oceanographic conditions: in-situ and satellite data

In July and August 2014, the seasonal upwelling was strongly pronounced, but it was also interrupted by short periods of relaxation (sample 14-Jul-2014, Fig. 2, 3) followed by further conditions favourable to upwelling (samples 22-Jul-2014 and 19-Aug-2014). In late November, the region was under the influence of relatively strong westerly winds, where the Qy component was elevated, possibly causing water column mixing in combination with upwelling (sample 18-Nov-2014).

In the second half of January 2015, north-westerly winds produced high upwelling indices and a decrease in SST (Fig. 2) (sample 27-Jan-2015). During the sampling period in July to October 2015, the study area was almost constantly under the influence of northerly and westerly winds that resulted in high upwelling indices, 8 interrupted by frequent short periods of relaxation during July and August. In September, high magnitude of upwelling components caused SST to decrease with a short periodicity of 5 – 8 days in response to wind forcing. In the beginning of October, there was a period of upwelling relaxation that was interrupted by a marked decrease in SST during the second week of October.

The SST increased during the sampling period from late March to July 2016 with several intermittent phases of upwelling events. The water temperature dropped continuously from late March to early April

(samples 24-Mar-2016, 2-Apr-2016 and 9-Apr-2016), and then from the middle of April the temperature showed some increase in the absence of significant upwelling (Fig. 3). At the end of April, upwelling conditions returned, followed quickly by a period of relaxation, resulting in a rapid increase in SST (samples

26-Apr-2016 and 5-May-2016). Later, the upwelling components increased after the middle of May (17-May-

2016), followed by decrease at the end of the month (sample 27-May-2016). June 2016 was characterized by the typical summer upwelling conditions (samples 2-Jun-2016 and 14-Jun-2016).

Dynamics for nutrient concentrations measured in 2016 were consistent with upwelling, showing higher concentrations on the dates preceded by the upwelling conditions (Table 1). The ammonia, nitrate and nitrite forms of nitrogen were the main nutrients that varied with upwelling conditions, concentrations of phosphorous and silica were typically low and close or below the quantification limits. Following the phytoplankton blooms, and in the absence of upwelling supply, nutrients were depleted by the phytoplankton. The nutrient concentrations were elevated in March to early April, then decreased to a minimum between 26 April and 5

May, but increased again in May and June after the return to upwelling conditions. In general, oceanographic conditions observed during late spring 2016 were characterised by relatively more intensive upwelling during spring than had been observed previously (described in Goela et al., 2014, 2016).

Figure 3 A-C shows images of SST based on remote sensing data, in conditions of active upwelling, when colder and nutrient rich water masses spread along the coast. In conditions of upwelling relaxation (Fig. 3D),

SST increased and water mass become more stratified. Details of oceanographic parameters on sampling dates and respective phytoplankton conditions are shown in Fig. 4 and Table 2. Negative Ekman transport corresponded to SST decrease and coincident Chl a increase, indicating activation of phytoplankton growth under upwelling conditions. 9

3.2. Phytoplankton community composition by microscopy

Observations and counts by inverted microscopy showed that on the basis of abundance nanoflagellates, diatoms and dinoflagellates dominated the phytoplankton assemblage (Fig. 5). Nanoflagellates were the most numerous group attaining abundances between 250 and 3500 cells ml-1, diatoms ranged from almost absent during winter (e.g. 27-Jan-2015) up to 650 –1730 cells ml-1 during spring and summer blooms. Dinoflagellates typically had abundances of around 10 – 100 cells ml-1, but on some occasions (22-Jul-2014) reached 518 cells ml-1. In cases of high abundance, the community was dominated by small unarmoured

Gymnodinioid species. Among flagellates, cryptophytes were an important contributor to the community, and when identified separately from nanoflagellates, typically ranged around 30 – 130 cells ml-1 but did attain a maximum of 674 cells ml-1.

Phytoplankton community composition was highly variable, even on a time scale of several days.

Development of high abundance communities was observed usually during favourable conditions for upwelling

(Table 2). The community typical of the beginning of the bloom (e.g. on 03-Jul-2015, 13-Jul-2015) was dominated by nanoflagellates (with significant contribution of cryptophytes) and small chain-forming diatoms, such as Chaetoceros, and in some cases Pseudo-nitzschia cf. delicatissima; small unarmoured dinoflagellates were also frequently present in such a community. At their peak, well-developed diatom blooms (e.g. on 19-

Aug-2014, 30-Jul-2015, 28-Aug-2015, 24-Sep-2015, 14-Jun-2016) consisted of Pseudo-nitzschia cf. delicatissima and cf. seriata, Guinardia spp., Leptocylindrus spp., Chaetoceros spp. and other genera. Diatoms were the main group contributing to Chl a (Fig. 6).

Nanoflagellate abundances were usually also high, with the presence of cryptophytes and coccolithophorids. Dinoflagellates were represented by the small size (<25 µm) Gymnodinioidea group, that may include harmful genera Karenia and Karlodinium, and also typically by the large species of genera

Protoperidinium, Diplopsalis and Tripos (Gómez, 2013), with occasional Dinophysis spp. and Prorocentrum spp. When diatom blooms declined (e.g. 10-Aug-2015 and 9-Apr-2016), they were sometimes succeeded (14-

Jul-2014, 26-Apr-2016 and 5-May-2016) with communities dominated by nanoflagellates and a high diversity

10 of dinoflagellates, often including HAB species (Dinophysis spp., Alexandrium spp., and Gymnodinium catenatum).

3.3. HAB species detected by LM and microarray

Twenty five samples between 14th July 2014 and 14th June 2016 were used for phytoplankton identification using light microscopy and microarray targeting ribosomal 18S/28S RNAs.

Pseudo-nitzschia

This HAB diatom genus was found microscopically to be typically present in all seasons throughout the sampling period, with abundances that varied between almost zero and up to 250 – 300 cells ml-1 (Fig. 7A).

Blooms nearing or reaching the IPMA regulatory threshold of 80 cell ml-1 of Pseudo-nitzschia seriata group

(shellfishery alert limit) were observed during the warmer half of the year in the samples from 07-Jul-2014, 19-

Aug-2014, 30-Jul-2015, 28-Aug-2015, 10-Sep-2015, 19-Sep-2015, 24-Sep-2015, 09-Apr-2016, 02-Jun-2016 and 14-Jun-2016. The proportions between P. seriata and P. delicatissima varied considerably in the samples.

The highest concentrations were observed on 19-Aug-2014 (246 cells ml-1 dominated by P. seriata), 30-Jul-

2015 (118 cells ml-1 of P. delicatissima and 27 cells ml-1 of P. seriata), 19-Sep-2015 (236 cell ml-1 dominated by P. seriata) and 5 days later on 24-Sep-2015 (combined bloom of circa 140 cell ml-1 of P. seriata and circa

100 cells ml-1 of P. delicatissima). During spring 2016, blooms that consisted of both Pseudo-nitzschia types were observed during late March - April with maximum on 09-Apr-2016 (106 cells ml-1), and after break in early May (6.7 cell ml-1 on 5-May-2016) increased during May to 02-Jun-2016 (85 cell ml-1) and 14-Jun-2016

(125 cells ml-1). Genus level microarray probes showed a significant signal for all the bloom dates (Fig. 7A) and a weaker signal on the dates when Pseudo-nitzschia spp. abundance was lower. The normalised signal for multispecies probes (Fig.7B-D) was in accordance with genus probes, and demonstrated the high diversity of this genus composition during blooms. Probes detecting P. delicatissima group (Fig. 7B) were active during both late summer 2015 and spring 2016 bloom pulses, while diversity seemed to be higher during 2015, as more different multispecies probes attained high signal. Probes for P. seriata also demonstrated high diversity, especially during peak bloom periods in September 2015 and April 2016, when seriata type also dominated the cell counts; P. delicatissima seemed to be favoured during less pronounced blooms in July 2015 (sample 30- 11

Jul-2015) and from 17 to 27 May 2016. General agreement between LM and microarray was good, as all significant blooms of more than 80 cells ml-1 were detected by the microarray probes. Microarray probes even responded to relatively small cell densities on the order of 5 - 10 cell ml-1 observed e.g. on 18-Nov-2014, 13-

Jul-2015, 9-Oct-2015 and 24-Mar-2016, although produced less pronounced signal on 17-May-2016 as compared to cell counts (circa 43 cell ml-1).

Dinophysis and Phalacroma

The Dinophysis genus was found to be present throughout the study period in LM samples (except samples on 27-Jan-2015, 28-Aug-2015 and 17-May-2016), with densities varying from 0.02 to 0.92 cell ml-1.

The highest cell concentrations were counted on 14-Jul-2014 (D. caudata and D. acuminata, 0.52 cell ml-1) and on 13-Jul-2015 (mainly D. caudata, 0.30 cell ml-1 ), from 10-Sep-2015 to 16-Oct-2015 ( D. ovum, 0.88 cell ml-

1), during spring 2016 on 26-Apr-2016 and 05-May-2016 (D. ovum and D. acuminata, 0.64 cell ml-1 ). Counts data was generally in good agreement with microarray detection (Fig. 8), with genus probes DphyFS02_25_dT and DphyexacutaFS01_25_dT showing S/N ratios that correlated with total Dinophysis counts. D. acuminata +

D.dens + D.sacculus probes DacumiS01_25_dT and DacumiD02_25_dT had significant signal for most samples from 2015, including September and October, when significant abundancies of Dinophysis were microscopically identified as D. ovum.

The Phalacroma genus was represented by Phalacroma rotundatum that was microscopically identified in the samples in July 2014, July to September 2015, 26-Apr-2016, 05-May-2016 and 2-Jun-2016 (Fig. 8C); the abundancies were always low with maximum of 160 cell L-1. Species probe ProtuS01_25_dT signal was usually low, but was confirmed hierarchically by the DphyGS03_25_dT probe on 22-Jul-2014, 19-Aug-2014,

3-Jul-2015, 21-Jul-2015, all samples in September 2015, 26-Apr-2016, 5-May-2016 and 27-May-2016. The cell counts mostly coincided with microarray detection, although a low signal suggested that cell abundance was near the detection limit, or cell activity was decreased.

Prorocentrum

Prorocentrum was found to be present in almost all samples during the study period, with high abundances for Prorocentrum micans, P. scutellum and P. minimum recorded during July - October 2015 (Fig.

9B, C); P. triestinum was recorded in sample on 10-Aug-2015 (10 cells ml-1). The microarray genus probes 12 generally demonstrated some positive signal (Fig. 9A), including the dates with high (2015) and low microscope counts (2016). Benthic species probes showed positive signal in July and September 2015, 9-Apr-

16 and 27-May-2016 (Fig. 9A), that was not in agreement with LM counts that were dominated by planktonic clade species. In many cases signal of species probes was not confirmed by the higher level hierarchy probes due to a low signal to noise ratio. The genus probes ProroFBS01_25dT (detecting the benthic clade, based on

Prorocentrum lima) and ProroFPS01_25dT (detecting the planktonic clade for Prorocentrum) signals were below the cut-off value of 2 in most of the samples, causing hierarchy test of species probes to fail.

Prorocentrum minimum probe had slightly elevated values in many samples (Fig. 9B), with maximum just below 0.02 on 28-Aug-2015 that corresponded with the maximum 29.4 cells ml-1 count. Even though, the signal of P. minimum probe remained very low, close to the negative controls, and cannot be considered reliable.

Signal of P. micans probe PmicaD02_25dT (Fig. 9C) was highlighted in samples from 2014 and August –

October 2015, and in general was in line with LM counts. Quantification of P. micans was based on the

PmicaD02_25dT probe, and corresponded with elevated LM counts during summer 2015 from 10-Aug-2015 to

16-Oct-2015, but in the samples with high cell counts (1.02 – 1.52 cells ml-1) quantified density values were considerably smaller (Fig. 9D). In the rest of the samples from 2016, both LM counts and probe quantification showed low densities or absence of P. micans, and were in agreement.

Alexandrium

Alexandrium genus was detected in cell counts in samples from 13-Jul-2015 to October 2015 (up to 480 cells L-1 on 10-Aug-2015), and then in lower densities (20 – 40 cells L-1) from 26-Apr-2016 to 14-Jun-2016; in

2014 samples counts of A. were unavailable. On the microarray, Alexandrium genus level probe

AlexGD01_25_dT produced significant signal in all samples (Fig. 10). The hierarchical approach confirmed the signal of species-specific probes for A. minutum, A. ostenfeldii and A. tamarense in many samples (Fig. 10A), including all samples from 2014, and 27-Jan-2015, 03-Jul-2015, 21-Jul-2015, 10-Aug-2016, 10 to 24-Sep-2015, but only A. minutum and A. ostenfeldii on 26-Apr-2016. Quantification of cell numbers (Fig. 10B) based on

AlexGD01_25_dT probe produced results above the limit of quantification (LOQ = 280 cells L-1) during the period from 10-Aug-2015 to 10-Sep-2015. Counts corresponded well with it on 10-Sep-2015 and 24-Sep-2015, when Alexandrium density up to 440 cells L-1 was within the LOQ range. Alexandrium was not detected in the 13 counts on 28-Aug-2015 and 19-Sep-2015, despite significant microarray signal on that dates. Low densities of

Alexandrium in the counts during April – June 2016 corresponded to weak microarray signal that was below the

LOQ.

Azadinium

Azadinium Elbrachter and Tillmann is a genus of armoured dinoflagellates with the size range 12 – 45 µm

(Tillmann et al., 2007), small size and lack of easily discernible diagnostic features making it difficult to identify using LM. Results of the LM and microarray detection of Azadinium are shown in Fig. 11. Cells resembling Azadinium 12-20 µ in size were identified in the samples 10-Aug-2015 (1.7 cells ml-1) and 28-Aug-

2015 (8 cells ml-1). The larger species Azadinium caudatum syn. Amphidoma caudata Haldall (see Nézan et al.,

2012) was identified on 14-Jul-2014 and 22-Jul-2014 (up to 0.64 cell ml-1). In several samples, small dinoflagellates < 20 µm in size with theca, displaced cingulum, and convex sides of epitheca, sometimes with pore, were frequently observed and enumerated (reported as dinoflagellates cf. Heterocapsa/Azadinium, form

F-3), and may belong to Azadinium or other morphologically similar genera. On the microarray there were four probes detecting Azadinium spp., (one of them, AzaGS02_25_dT can also react with Karenia mikimotoi).

AzaGS01_25_dT is higher hierarchy level probe, which should have S/N ratio of more than 2 to confirm positive signals for the probes AzaGD01_25_dT and AzaGD03_25_dT. Microarray probes have passed the hierarchy test for Azadinium on 22-Jul-2014, 19-Aug-2014, 18-Nov-2014, 03-Jul-2015, 21-Jul-2015, 19-Sep-

2015, 24-Sep-2015 and partly on 14-Jul-2014, 10-Sep-2015 and 26-Apr-2016. The probes showed some signal supporting cf. Azadinium spp. cells identifications on 28-Aug-2015, but not on 10-Aug-2015. Counts of dinoflagellates cf. Heterocapsa/Azadinium (F-3) and thecate dinoflagellates <20 um, that could contain

Azadinium, corresponded with microarray on 22-Jul-2014, 19-Aug-2014, 3-July-2015, 21-Jul-2015, 19-Sep-

2015 and 24-Sep-2015. In the samples from 2016, these dinoflagellates forms were observed in LM, but microarray probes signal was hierarchically confirmed only on 16-Apr-2016, and therefore did not correspond well with counts during that period.

Gymnodinioid dinoflagellates

Gymnodinioid dinoflagellates >20 µm in size (Fig. 12E) were frequently observed, reaching maximum of

20.22 cells ml-1 on 14-Jul-2014, significant concentrations were also observed on 24-Sep-2015 (13.55 cells ml- 14

1), 26-Apr-2016 (11.68 cells ml-1), 5-May-2016 (16.73 cells ml-1) and 14-Jun-2016 (8.96 cells ml-1).

Gymnodinioid dinoflagellates less than 20 µm in size were detected in all samples, with maximum on 22-Jul-

2014 (136.87 cells ml-1), had lower concentration during 2015 (not more than 20 cells ml-1) that increased in

2016 and peaked on 24-Mar-2016 (51.19 cells ml-1) and 9-Apr-2016 (66.76 cells ml-1).

Gymnodinium catenatum was detected microscopically on four sampling dates (14-Jul-2014 and 22-Jul-

2014, 10- Sep-2015 and 27-May-2016), reaching maximum abundance of 2.48 cells ml-1 on 14-Jul-2014, and

0.04 to 0.26 on other dates (Fig. 12A). The LM observations were in partial agreement with microarray detection. Species probe GcateS01_25_dT and higher hierarchical level probe LSGcat0270A24_dT, both showed a significant S/N ratio from 14-Jul-2014 to 19-Aug-2014, corresponding with the highest G. catenatum cell count. Probes signal was weaker in other samples. GcateS01_25_dT was highlighted from 10-Sep-2015

(coincident with cell count 0.26 cells ml-1) until 24-Sep-2015, then showed weak signal on 27-May-2016, when

G. catenatum was present in counts (0.04 cells ml-1). The LSGcat0270A24_dT probe showed significant signal and passed the hierarchy test on 14-Jul-2014, 22-Jul-2014, 18-Nov-2014, 27-Jan-2015, 24-Sep-2015, 24-Mar-

2016, 9-Apr-2016 and 26-Apr-2016; but this probe is suspected to have specificity issues, therefore, these results should be taken with caution. Although no cells of Gymnodinium catenatum per se were detected, unidentified Gymnodinioid dinoflagellates reached high abundance in the same period (Fig. 12E, 44 cells ml-1 on 10-Sep-2015 and 13 cells ml-1 on 24-Sep-2015). A small peak for the probes signal on 26-Apr-2016 also corresponded with an elevated count of unidentified Gymnodinioid dinoflagellates >20 µm (12 cells ml-1). The hierarchy of the species probe GcateS01_25_dT, that depends on the LSGcat0270A24_dT probe was only confirmed in the samples 14-Jul-2014, 22-Jul-2014, 24-Sep-2015 and 26-Apr-2016.

Karenia probes showed significant signal in July – August 2014, 27-Jan-2015, July and September 2015, in April 2016 and on 27-May-2016 (Fig. 12B, C). K. brevis probes were highlighted and passed hierarchy test on 14-Jul-2014, 19-Aug-2014, 21-Jul-2015 and 24-Sep-2016. K. mikimotoi species probe KmikiD01_25_dT had significant signal when its hierarchy was supported on 27-Jan-2015, 3-Jul-2015, 21-Jul-2015, 19-Sep-2015,

24-Sep-2015, 26-Apr-2016 and 27-May-2016. These results in part corresponded with high cell densities of

Gymnodinioid cells >20 µm detected in July-August 2014, 24-Sep-2015 and 26-Apr-2016. High microarray signal detected during July 2015 and on 27-May-2016 was not reflected in counts. 15

Karlodinium was not identified in counts because of the difficulty with identification using LM.

Gymnodinioid cells ranging 10 – 15 µm in size, which were enumerated, could possibly include Karlodinium.

(Gymnodinioid <20 µm in Fig. 12E). K. veneficum was detected by the microarray probes (Fig. 12C) mainly during 2014 and August – September 2015. A hierarchy of Karlodinium veneficum species probes was confirmed in all samples from 2014, 3-Jul-2015, 21-Jul-2015, 19-Sep-2015 and 24-Sep-2015, several probes were active on 18-Nov-2014. Counts of small Gymnodinioid dinoflagellates were generally higher during 2016 than 2015, and therefore did not show correspondence with Karlodinium probes microarray signal.

Flagellates

Total nanoflagellate assemblage counts under LM comprising of the cryptophytes, prasinophytes, prymnesiophytes and other unidentified nanoflagellates demonstrated abundances between 133 to 3485 cells ml-1 (22-Jul-2014) and was 680 cells ml-1 on average. Local peak values were also recorded in samples 10-Sep-

2015 (1079 cells ml-1), 10-Oct-2015 (1229 cells ml-1). In the 2016 sampling period, nanoflagellates were abundant from 24-Mar-2016 to 09-Apr-2016 (550 – 800 cells ml-1), then declined to minimum of 177 cells ml-1 on 5-May-2016, and increased again reaching 632 cells ml-1 on 2-Jun-2016.

Prymnesium was assumed to occur in the nanoflagellate community and its microarray results were compared with counts (Fig. 13A). Microarray data identified the frequent presence of Prymnesium spp. at the study site throughout different seasons. Clade level probes Clade01old_25_dT and Clade01new_25_dT passed hierarchy in all samples and showed elevated signal in samples from July - August 2014, 10-Sep-2015, 19-Sep-

2015 and 9-Apr-2016. Group probes PrymS01_25_dT and PrymS03_25_dT passed hierarchy test across all samples and showed high signal in 2014, 27-Jan-2015, 21-Jul-2015, 28-August to 9-Oct-2016. The species probe for P. polylepis CpolyS01_25_dT was confirmed in 14-Jul-2014, 18-Nov-2014, 27-Jan-2015, 3-Jul-2015,

21-Jul-2015, from 28-Aug-2015 to 16-Oct-2015, 9-Apr-2016, 26-Apr-2016, 27-May-2016 and 14-Jun-2016.

The probe for P. parvum PparvD01_25_dT had very low values but was confirmed by a hierarchy test in several samples. Prymparv01_25_dT species probe did not pass hierarchy test in all samples. Prymnesium spp. group and species probes had confirmed signal typically in the samples with higher nanoflagellate cell counts, especially in summer 2014, July and early September 2015 and April 2016 (Fig. 13A). In some cases, e.g. 27-

Jan-2015, 19-Sep-2015 and 26-Apr-2016, high probes signal coincided with low nanoflagellates cells counts. 16

Pseudochattonella was not targeted by LM counts. Genus level probes PschGS01_25_dT,

PschGS04_25_dT and PschGS05_25_dT showed significant signal (Fig. 13B) and confirmed the species probes PfarD01_25_dT (for Pseudochattonella farcimen Eichrem) and PverD01_25_dT (for Pseudochattonella verruculosa Tanabe-Hosoi et al.) to have valid signal on the following dates: 18-Nov-2014, 27-Jan-2015, 3-Jul-

2015, from 10-Sep-2015 to 9-Oct-2015, 9-Apr-2016, 17-May-2016 (except P. farcimen) and 27-May-2016.

These probes showed signals in the samples from summer 2014, although hierarchy was not confirmed.

Chloromorum. Microarray probes detecting Chloromorum toxicum Tomas et al. were highlighted and confirmed the hierarchy test depending on the CtoxiS09_25_dT probe, indicating presence of this species on

22-Jul-2014, 19-Aug-2014, 3-Jul-2015, 21-Jul-2015 and from 10-Sep-2015 to 24-Sep-2015 (Fig. 13C). This species was not targeted by LM counts.

Heterosigma akashiwo Y. Hara was microscopically detected in the samples 22-Jul-2014 (16.67 cells ml-

1), 30-Jul-2015 (5.34 cell ml-1), from 2-Apr-2016 to 26-Apr-2016 (6.68 – 13.35 cells ml-1) and 27-May-2016

(4.45 cell ml-1). Multiple probes produced significant signal in samples from July-August 2014, 3-Jul-2015, 19-

Sep-2015 through 16-Oct-2016, April 2016 and 27-May-2016 (Fig. 13D). Species probes SSHaka0200A25_dT and SSHaka0200A25_dT signal were confirmed by genus probes on 14-Jul-2014, 22-Jul-2014, 24-Sep-2015

(only SSHaka0200A25_dT), 9-Apr-2016 and 26-Apr-2016. LM identification demonstrated good agreement with genetic probes, in sample 22-Jul-2014 and for four samples from April - May 2016. Despite the confirmed detection of H. akashiwo by species probes on 14-Jul-2014 and 24-Sep-2015, it did not appear in cell counts; on the contrary, LM count of 5.34 cell ml-1 on 30-Jul-2015 did not correspond with the probes signal. Therefore, in five out of six cases (83%) of microscopic detection, it was confirmed by microarray.

4. Discussion

4.1. HAB species detected in the study area

Previous studies in the area have demonstrated the dependence of phytoplankton bloom development and composition on the wind-induced upwelling, and noted the importance of the seasonal factor. Spring and summer phytoplankton blooms are provoked by the nutrient enrichment induced by upwelling conditions and occur from March to August (Loureiro et al., 2011, 2005). These blooms are dominated by diatoms of the 17 genera Chaetoceros, Guinardia , as well as Pseudo-nitzschia which is responsible for domoic acid production, the cause of amnesic shellfish poison in humans (Shumway et al., 2018). Previous studies (Goela et al., 2015;

Loureiro et al., 2011, 2005) also reported the frequent presence of toxic dinoflagellates species, including the genera Dinophysis, Gymnodinium, Prorocentrum and Lingulodinium, and their relation with seasonal upwelling events. Cysts of the dinoflagellates Gymnodinium catenatum and Lingulodinium polyedrum have been reported from the coastal sediments along the S and SW coasts of Portugal (Amorim et al., 2002; Ribeiro et al., 2016), suggesting a regular occurrence of these toxic species in the Sagres area.

Results of this study confirm that Pseudo-nitzschia spp. and HAB dinoflagellates of the genera

Dinophysis, Gymnodinium, Prorocentrum and Alexandrium are frequently occurring in the study area, with

Pseudo-nitzschia diatoms being the most common and abundant HAB organism. Microarray provided information of possible presence in the study region of invasive Pseudochattonella verruculosa, Chloromorum toxicum, and has confirmed previous observations of Heterosigma akashiwo in Sagres (Loureiro et al., 2011).

Azadinium spp. was for the first time (to the knowledge of the authors) detected in this study along the south-western coast of Portugal, using molecular probes in combination with LM. Previous biotoxin monitoring studies in Portugal (Vale et al., 2008b, 2008a) have detected (since 2006) azaspiracid toxins (AZA) at levels below the regulatory limit in various commercial mollusc species, including mussels, mostly from the north of

Portugal (Aveiro lagoon). However, the Sagres area has not been included in these studies, which only considered these toxins in mollusc tissue. Azadinium has been relatively recently described (Tillmann et al.,

2009) and identified as an azaspiracid (AZA) toxin producer (Krock et al., 2009a). This genus has since been observed from the North Sea (Krock et al., 2009b), British Isles (Luo et al., 2017; Salas et al., 2011; Tillmann et al., 2014, 2012), Norway (Tillmann et al., 2018), China (Gu et al., 2013; Krock et al., 2014), France,

Mediterranean (Luo et al., 2016), Korea (Potvin et al., 2012), Argentina (Tillmann et al., 2016), South-East

Pacific (Tillmann et al., 2017b) and the U.S.A. (Kim et al., 2017). Amphidoma languida Tillmann, Salas &

Elbrächter taxonomically related to Azadinium (Nézan et al., 2012), was recently reported from the Spanish south-western coast in the gulf of Cadiz, approximately 150 km east from Sagres (Tillmann et al., 2017a); A. languida had a specific azaspiracid toxins profile, and the accumulation of AZA in the mussel tissue was confirmed (together with other DSP toxins), resulting in DSP syndrome detected by mouse bioassay. Taking 18 into account the above reports and the results of this study, it is suggested that possible occurrence of

Azadinium spp. and Amphidoma languida represent a potential threat of azaspiracid toxicity in shellfish for

Portuguese coastal waters. Further research is required to identify the occurrence of these HAB genera in shellfish producing areas, their regional geographical distribution, toxin accumulation, and whether their occurrence and range is affected by climatological and oceanographic factors.

4.2. Microarray detection of HAB species compared to LM

The microarray complemented with observations by LM have in this study provided a better detection of dinoflagellates from the genera Alexandrium, Azadinium, Karenia, Karlodinium and flagellates Prymnesium,

Pseudochattonella, Chloromorum and Heterosigma. In the case of diatoms Pseudo-nitzschia, presence of more species could be detected by multispecies probes in the microarray than the two morphological species complexes, P. seriata and P. delicatissima, distinguished by microscopy, although the discrimination between them in natural mixture is complicated by the lack of species-specific probes for Pseudo-nitzschia, and therefore, relies on probes that detect clades or a group of several species simultaneously (multispecies probes).

Observation of the multiple probes reactions and the use of hierarchical probe validation approach reduced the probability of false positive or negative results. In the samples with high LM counts of Pseudo-nitzschia many multispecies probes were typically active (Fig. 7), detecting both seriata and delicatissima groups, therefore, it is likely that this genus was represented by multispecies complexes. P. seriata group, based on microarray results, was more diverse during strong bloom period, such as September 2015 and April 2016, while P. delicatissima seemed to be better represented during less pronounced or shorted blooms, e.g. 30-Jul-2015 and colder season, such as on 18-Nov-2014 and 27-Jan-2015; this agrees with seasonal differentiation of Pseudo- nitzschia species reported by (Ruggiero et al., 2015). Significantly high probes signal corresponding to low cell count (3.7 cell ml-1) on 9-Oct-2015, at the decay of very active bloom, may be explained by the presence of genetic material of destroyed cells in the environment, or by the elevated physiological activity of the cells that were adapting to changing conditions. Significant LM counts of mainly P. delicatissima on 17-May-2016 were reflected only in reaction of relatively low number of probes; this may hypothetically reflect the early bloom start, when less species in assemblage are present, and grow with different speed. Regardless of that, the 19 presence of potentially toxic genus, Pseudo-nitzschia as a complex of species was successfully identified by the microarray, with results well comparable to LM counts.

Dinophysis genus with its more common species D. acuminata, D. caudata, D. ovum were detected by both light microscopy and microarray. As these species did not attain high abundances during the study, with typical concentrations of tens to hundreds cells L-1, it is an advantage of the microarray method that its sample volume of 0.5 to 1.0 L is larger compared to the 50 ml sample for microscopy, ensuring a higher probability for detection of low abundance species. In several samples from summer 2015, microarray probes detecting D. acuminata + D.dens + D.sacculus corresponded with D. ovum counts (Fig. 8). These results may be explained, besides from the LM identification error, by a possible probes cross-specificity; indeed Kavanagh et al. (2010) were unable to discriminate between D. acuminata, D.dens, D.sacculus and D. norvegica, due to similarities in the ribosomal large subunit (LSU) region; this may also be the situation for D.ovum.

Prorocentrum genus probes had mixed performance on the microarray, as the genus level probes used for hierarchical validation of species probes tended to have S/N ratios below the cut-off limit, even in cases when clade and species probes had significant signal. The species probes, except the probe for P. micans, also often showed low signals. This indicated that genus probes were less sensitive and gave weaker signal compared to the rest, and may perform better under conditions of higher Prorocentrum cell density. The limit of detection of these probes was assessed by the developers to be around 500 cells L-1, with LOQ of 4000 cells per sample, while microscopic counts showed much lower densities (Fig. 9). Benthic clade probes were occasionally highlighted, although counts were dominated by planktonic members of the genus. The probe for P. minimum corresponded with its presence in counts, but its low normalised signal and lack of genus probe confirmation, suggest that these results should be taken with caution. Quantification of P. micans using PmicaD02_25_dT probe was possible and reflected well the elevated cell densities during summer 2015, but considerably underestimated cell numbers in comparison with LM counts (Fig. 9).

Dinoflagellates such as Alexandrium, Azadinium, Karenia and Karlodinium, which could not be easily identified microscopically due to the lack of visible under Lugol’s fixation diagnostic features, were supposedly better detected by the microarray probes (Figs. 8 – 12). Alexandrium was detected in many samples, however this identification could not always be confirmed by microscopy, because of the low number of cells and 20 limited volume (50 ml) used for screening, reducing the probability for detection. At the same time, in several samples there were significant numbers of thecate dinoflagellates, ranging from approximately 20 to 60 µm in size which could not be identified decisively by LM, that could contain Alexandrium spp. Based on the high reported specificity of the Alexandrium probes (Taylor et al., 2013), it is likely that this genus was frequently present at the study site (Fig. 10), even on the dates without observations by LM.

Considering the difficulties with microscopic identification of Azadinium, it is likely that microarray provided better indication of its presence. Azadinium probes have not been fully tested against all the species of this genus, as these species has only been described recently (e.g. Tillmann et al., 2018) and cultures may not be easily available. Therefore, it is possible that the available probes do not detect the presence of all species of this genus; also the possibility of false positive reactions should be considered too and the microarray will require further testing when cultures or well characterized contaminated samples will be available. LM identification of Azadinium for the monitoring purpose does not seem be feasible, requiring high expertise and not providing exact diagnostics. It highlights the importance of development of molecular based tools for this

HAB genus.

Gymnodinium catenatum was detected during July 2014 by both LM and MA, then by LM on 10-Seo-

2015 and 27-May-2016, with MA and LM signal partly corresponding. But LSGcat0270A24_dT probe is suspected to have specificity issues, therefore, these results should be taken with caution. The observed (Fig.

12) coincidence of signal of G. catenatum probes with counts of Gymnodinioid dinoflagellates may point out at lack of probe specificity, but as well at the possibility that G. catenatum was indeed present in assemblage together with other taxonomically related dinoflagellates, and that it was not detected in LM counts because of sample volume effect or other reason. This species is frequently reported in the study region and was detected in monitoring zone L7c during study period on 22-Jul-2014 and other dates (IPMA, 2014-2016).

Gymnodinium-like cells were frequently observed microscopically, but small size, lack of diagnostic features and existence of several genera similar to harmful Karlodinium spp. and Karenia spp. prevents conclusive identification by LM. The microarray confirmed the presence of this potentially harmful genera in several samples (Fig. 12), but only indirect comparison with LM could be done, i.e. comparing probes signal to total counts of Gymnodinioid dinoflagellates, that may include target species. 21

Flagellates such as Prymnesium spp., Pseudochattonella spp., Chloromorum toxicum were not targeted with LM in this study, therefore, a direct comparison with microarray could not be made. Prymnesium MA signal in general corresponded with elevated counts of total nanoflagellate community. Pseudochattonella spp. and Chloromorum toxicum presence was indicated by the MA signal (Fig. 13), and seemed to appear during the general phytoplankton bloom dates in July – August 2014, September 2015, April – June 2016, but also on the dates with no significant blooms on 27-Jan-2015 and July 2015.

Heterosigma akashiwo was previously reported from the study site (Loureiro et al., 2011), and detected by both LM and MA in this study, with good (83%) correspondence between the two methods.

Possible sources contributing to differences between LM and microarray were: 1) errors of LM counting and taxa identification; 2) intrinsic variability of microarray method, originating from inefficiencies in RNA extraction (mitigated by addition of Dunaliella tertiolecta control before extraction), degree of labelling, degree of hybridisation and errors introduced during image processing, and 3) cross-reactivity between probes and unspecific binding, that can be increased in cases where the RNA is degraded. For instance, cross-reactivity between several probes of the earlier MIDTAL microarray have been found (Dittami et al., 2013; Kegel et al.,

2013a) when applied to the laboratory cultures. RNA extracted from a natural sample may contain many other sequences originating from different species present in the phytoplankton community. This may lead to a higher degree of variability in microarray results than would be observed with laboratory algal cultures, and may increase false positive results. The issue of the false positive microarray signal is to some degree minimised in the microarray data processing by means of a hierarchical approach, where several probes reinforce each other starting with higher taxonomic levels (order, class, genus) and to the multispecies and species-specific probes

(Medlin et al., 2013). Furthermore, data analysis software could improve the output results by applying a signal threshold individual for each probe, whereby the signal would be considered valid, and thereby decreasing false positive signals. The implementation of this data treatment in combination with hierarchy approach would facilitate the interpretation of the results. Incrementing the microarray probes database through systematic analysis of contaminated environmental samples or cultures will also contribute to reinforce the robustness and quantitative aspect of this molecular tool by improving probe specificity, LOD and LOQ.

22

4.3. Microarray performance

The observations on the performance of the microarray in this study are improved compared to those described in Kegel et al. (2013a, 2013b) for the earlier generations of the MIDTAL microarray, which represented the earlier version of the IEMArray microarray used in this study, and has been tested on environmental samples from the Atlantic coasts of France (Kegel et al., 2013a, 2013b), Ireland (McCoy et al.,

2014, 2015), Norway (Dittami et al., 2013b) and the Galician coast of Spain (Dittami et al., 2013a). The laboratory workflow procedures used in the present study have been improved and standardised compared to the MIDTAL project; some variations observed in microarray performance described in this section may have been induced by somewhat different procedures that were followed in the earlier studies.

In cases when microarray and toxin analysis (using ELISA or HPLC) has been carried out simultaneously, the causative species are usually detected by the microarray (Dittami et al., 2013b). Difference between LM counts and microarray detection has been reported that in some cases. Typically, more toxic species are detected by the microarray rather than by microscopy. This is especially frequent for the genera

Alexandrium, Dinophysis and benthic clades of Prorocentrum. The possible reasons for such discrepancies, discussed by the same authors are: 1) different sampling volume for LM and microarray, 2) imperfect taxonomic identification by LM, 3) lack of specificity or false-positive reactions of the early versions of microarray probes, 4) shortcomings of the extraction protocol when some robust cells (e.g. Prorocentrum spp.) are not broken and the RNA is not available for hybridisation, and 5) LM detects all cells whether they are alive, dormant or dead, whilst microarray detects only physiologically active cells.

The sampling volume defines the minimum limit of the detectable abundance of a species. The volume used for inverted LM is typically 50 ml, as used in this study. Therefore, the limit of detection of rare species was 20 cell L-1. The volume of sample used for microarray analysis was one order of magnitude higher (0.5 L for most of the samples), and thus have increased probability of detection of rare species by microarray compared to LM. The species affected by the sampling volume could belong to genera Dinophysis,

Prorocentrum, Alexandrium that often occurred in densities of tens to hundreds of cells L-1 in the sampling area

(Loureiro et al., 2005, 2011). Rarer small sized species (<20 µm), counted at 400x magnification, using the minimum 20 fields of view, are at the limit of detection of approximately 3300 cells L-1. For many samples, the 23 actual number of observed fields is 35 and up to 70, improving the limit of detection to approximately 1000 –

2000 cell L-1. Thus, the problem of the limit of detection may have affected the relatively rare and small sized target organisms, such as Heterosigma spp., Azadinium spp., Prorocentrum minimum, Karlodinium veneficum and Alexandrium spp. In such cases, the microarray results would demonstrate higher reslution potential than

LM, considering the higher probability of detecting rare organisms in a greater sample volume, and the high specificity of most of the species-level probes on the microarray as reported by Medlin et al. (2013).

One of the reasons for discrepancy between dinoflagellate class-level probes for

(DinoB_25_dT, DinoE12_25_dT) was defined by Kegel et al. (2013a) as the presence of unidentified dinoflagellates sized <10 - 15 µm, that are not always microscopically enumerated in many studies. In this study, the small dinoflagellates have been estimated, although the majority of them were not taxonomically identified to species or genus level and thus have been reported as a morphological forms, such as naked

Gymnodinioid cells <20 µm, thecate dinoflagellates <20 µm or as a category of unidentified dinoflagellates

<20 µm.

Non-specificity and cross reactions with untargeted regions of the RNA are the other important possible sources of errors for the microarray. It has been reported (Dittami et al., 2013b) that the presence of

Alexandrium pseudogonyaulax was not detected because of the variation in the LSU rDNA sequence of the strain occurring in the study area in Oslo Fjord (Norway). In the current study, possibly because of the high reactivity of the Alexandrium genus level probes, they had positive signals in many of the samples, even when

LM counts did not detect significant abundance of Alexandrium spp. cells. The same effect was reported by

Dittami et al. (2013b) in Norway and Kegel et al., (2013b) from Arcachon bay, France. There is an indication

(Microbia Environnement, unpublished) that the cut-off level for these probes should be set higher than S/N=2.

Probes have different strengths and therefore, weakly reacting ones can only detect high concentrations of target cells, such as DnorvS01_25_dT Dinophysis norvegica probe that is reported to be weaker compared to other Dinophysis probes and is only able to detect high concentrations of 1200-1400 cells L-1 (Dittami et al.,

2013; Edvardsen et al., 2013). The issue of varying degree of reactivity of different probes can be mitigated at the stage of data analysis, by means of applying different levels of S/N ratio as a cut-off values for each individual probe, depending on its specific properties. 24

The lack of exact species-probe specificity for the majority of the Pseudo-nitzschia probes (except P. multistriata) has been described (Kegel et al., 2013a) and tested with field samples and cultures of known species (e.g. in Naples bay, Italy by Barra et al., 2013). This shortcoming, however, is partly resolved by using the hierarchical approach, when genus and clade level probes assist the interpretation of the multi-species probes data. Toxicity is found in many species of Pseudo-nitzschia, and is also subject to environmental conditions, so from the point of view of seafood security monitoring, the exact identification of species might be excessive for the purpose, especially considering that this diatom genus often appears in multi-species mixtures that follow seasonal dynamics (Ruggiero et al., 2015). The presence of many multi-species Pseudo- nitzschia probes on the array that may co-react (Barra et al., 2013), can reduce the possibility of false positive or false negative outcomes, especially if the task is merely to confirm the presence or absence of Pseudo- nitzschia bloom in the marine environment.

High ratio of target to non-target RNA in the samples that occurs if the sample is dominated by the bloom abundance of other species, e.g. non-toxic diatoms, may have also reduced the capability of detection by the original MIDTAL microarray (Dittami et al., 2013). However, this effect has not been noticed in the most recent implementation of IEMArray, likely due to optimisation and standardisation of the laboratory procedures, applied on the later stage of microarray development.

Previously there were reported problems of RNA extraction, when robust cells of Prorocentrum micans

(Kegel et al., 2013a) and Dinophysis (Edvardsen et al., 2013) and Pseudo-nitzschia (Dittami et al., 2013) have not been broken up and consequently the RNA was not available for hybridisation. This issue is resolved in the current microarray protocol, by including of a bead-beating step in TRI-reagent, and optionally sonicating to break apart the cells and to provide complete nucleic acid extraction. The rRNA content in the living cells of the same species may be variable under various growth conditions; for instance, variation in the cellular RNA contents in Dinophysis could be up to six fold (Edvardsen et al., 2013) and may represent an important source of variation of microarray results.

Cells physiological state may be the important difference between LM and microarray. RNA microarray primarily detects physiologically active cells, with significant amount of ribosomal RNA. In contrast, LM does not readily discriminate between physiologically active, dormant or dead cells. LM counts would include cells 25 from all physiological states, whilst microarray would only include active, live cells. Therefore, a low number of highly active cells versus a high number of non-active cells in microarray would result in the same signal.

Generally, in this study, the IEMArray provided a consistent performance. In comparison with previous studies, it represents an improvement on the earlier developmental versions of the MIDTAL project microarrays. The consistency and specificity of the microarray as well as extraction and hybridisation protocols have been improved and standardized. The method has already demonstrated the capability to detect target toxic phytoplankton in geographically diverse locations on the Mediterranean and Atlantic coasts of Europe

(Norway, Ireland, France, Spain, Italy and in this study, Portugal), suggesting its applicability on a wider geographical scale; indeed it has been designed using sequences from globally distributed strains of HAB species. Further development in species quantification ability using microarray will increase its potential in

HAB monitoring applications.

4.4. Upwelling conditions and HAB species

Most of the sampling dates occurred during the warmer (spring to early autumn) period of the year, when phytoplankton was not supposed to be limited by the solar radiation or temperature, but rather by the nutrient availability (Reynolds, 2006), making the upwelling an important factor in controlling its temporal change over short periods. The oceanographic conditions during the study period changed several times between wind- driven upwelling, relaxation and mixing. The periods of the most pronounced upwelling included summer

2014, August – September 2015, April and late May to June 2016. Samples such as on 19-Aug-2014, 19-Sep-

2015, 24-Sep-2015, 9-Apr-2016, and 27-May-2016 were highlighted as bloom events, when both microarray probes and LM simultaneously detected development of a complex of harmful species, from Pseudo-nitzschia to small dinoflagellates and raphidophytes. Relaxation of upwelling shifted the assemblage structure to develop community with high diversity of dinoflagellates (14-Jul-2014, 26-Apr-2016 and 5-May-2016). Two conditions for the development of HAB species in the study area are identified: 1) spring and summer upwelling associated with blooms of chain-forming diatoms, that are often dominated by the ASP-producing Pseudo- nitzschia; 2) conditions of upwelling relaxation breaks, especially in summer and autumn, including over relatively short periods (5-10 days) following upwelling end, enable the development of HAB dinoflagellates, 26 favouring the more stratified water column and advection from the offshore frontal zones (Moita et al., 2003;

Oliveira et al., 2009).

As diatom bloom conditions suitable for Pseudo-nitzschia during the spring and summer upwelling contrast with the conditions suitable for the dinoflagellates Dinophysis, Prorocentrum, Alexandrium,

Lingulodinium, Gymnodinium catenatum during upwelling relaxation, there is an interchange between the main

HAB groups that can produce an extensive period of toxicity throughout the year. Azadinium, Karenia and

Karlodinium and toxic flagellates seem to be favoured by the same conditions suitable for diatom development, enabling these species to occur together. Extensive periods of toxicity impose considerable difficulties for the management of the shellfish production in the Sagres region.

4.5. Relation with shellfish harvest bans

The results of regular monitoring of phytoplankton carried out by the Portuguese Institute of Sea and

Atmosphere (IPMA, 2013) using inverted microscopy frequently detect the presence Pseudo-nitzschia species,

Dinophysis spp. and Prorocentrum spp., Gymnodinium catenatum and Alexandrium spp., in the Sagres area, typically during the warm part of the year (May – October), in agreement with the results of this study.

Harvested mollusc tissue is also analysed for ASP, DSP,YTX and PSP biotoxins presence using HPLC (Vale et al., 2008b). Above a certain threshold of toxins in the tissue, or concentration of HAB species cells in the water, a ban on harvesting shellfish will be imposed which will negatively impact bivalve aquaculture. Therefore, knowledge regarding phytoplankton diversity and dynamics in the study area is important for monitoring the development of HAB and to provide information for aquaculture management.

Over the duration of the study, Table 3 shows a comparison between HAB species detected, by both microscopy and microarray, and the periods when shellfish harvesting bans have been imposed by the national monitoring authority (IPMA, 2014-2016). A full or partial ban on mussel collection occurred in July – August

2014, July 2015 a September – October 2015, and from the end of April until the end of the study period in

June 2016. Phytoplankton sampling stations of IPMA, as of the time of the study, are located at a significant distance from the Sagres study site (Arrifana, 30 km North for L7a and Porto do Mos, 20 km East for L7c, Fig.

1), and therefore may have not exactly reflected its oceanographic and phytoplankton conditions. Comparative 27 analysis (Table 3) shows a clear correspondence between the closures of the shellfishery in monitoring zones

L7a and L7c and the presence of HAB species at Sagres study site. The correspondence with bans in zone L7a

(western coast) is particularly strong for the ASP-producing Pseudo-nitzschia, which is the usual HAB source, causing shellfish bans during spring and summer diatom blooms, related to upwelling. This co-correspondence and oceanographic studies (Relvas et al., 2007; Relvas and Barton, 2005; Sánchez et al., 2006) highlight the hydrographic connectivity between the west coast of SW Portugal and the Sagres region, that is affected by the upwelling along the western coast.

Conclusions

The conclusions on the results of this study linked to the original research questions are as follows.

1. The most common HAB species in Sagres are diatom Pseudo-nitzschia, dinoflagellates of the genera

Dinophysis, Prorocentrum, Alexandrium, and Gymnodinium catenatum. Microarray results have pointed out the possible presence of Azadinium, Karenia, Karlodinium, and flagellates Prymnesium and raphidophytes

Heterosigma akashiwo and Chloromorum toxicum, although these were not always supported by LM results.

2. The 18S/28S rRNA microarray method in combination with observations by microscopy has provided more specific identification of the genera Alexandrium, Azadinium, Karenia, Karlodinium, Gymnodinium catenatum, Prymnesium and Heterosigma akashiwo. The presence of Pseudo-nitzschia, as a complex of species has been successfully identified by the microarray in all samples, and found to be comparable to LM counts.

The results demonstrate that, without LM, microarray alone would successfully detect Pseudo-nitzschia spp. and the main HAB dinoflagellate genera. The cells quantification capability, an important feature of the RNA microarray, was tested and provided promising results with Alexandrium but underestimated cells in the case of

Prorocentrum. This has a potential to improve with the development of microarray probes quantitative calibration.

3. The oceanographic conditions in which HABs tend to develop in the study area occur in two situations: first, usually during spring and summer, blooms of diatoms are associated with upwelling, these blooms often contain by Pseudo-nitzschia and bring ASP risk; second, during upwelling relaxation, high diversity of HAB

28 dinoflagellates responsible for DSP and PSP can develop, even over relatively short breaks of 5 – 10 days that often interrupt the upwelling during the warm period of the year (May – October).

The overall conclusion is that the applied microarray (IEMArray) demonstrated to be a useful tool for harmful phytoplankton monitoring that can provide more precise detection of HAB species that are difficult to identify by microscopy.

Acknowledgements

Authors acknowledge the financial support by the EC Erasmus Mundus MACOMA PhD grant to S.

Danchenko; A. Newton and M. Berzano were supported by the EU FP7 DEVOTES project (no308392). J. Icely and B. Fragoso were supported by the projects EU FP7 AquaUsers (nº 607325), Horizon 2020 AquaSpace (nº

633476), Horizon 2020 Ceres (nº 678193), and Horizon 2020 GAIN (nº 773330); A. Newton also thanks Future

Earth and Coasts. This work was partially supported by the Portuguese Science Foundation (FCT) through the grants UID/MAR/00350/2013 and UID/MAR/00350/2019 attributed to CIMA of the University of Algarve. We also thank Dr. Carmem Lara-Manes for help with manuscript preparation and Dr. Ruth Paterson (Scottish

Association for Marine Science) for help with Azadinium identification.

29

References Amorim, A., Moita, M., Oliveira, P., 2002. Dinoflagellate blooms related to coastal upwelling plumes off Portugal. Harmful Algae 89–91. APHA. Standard methods for the examination of water and wastewater, 21sted. Washington, DC, New York: American Public Health Association; 2005. Bakun, A., 1973. Coastal upwelling indices, West coast of North America. Technical report NMFS SSRF-671, NOAA. Barra, L., Ruggiero, M.V., Sarno, D., Montresor, M., Kooistra, W.H.C.F., 2013. Strengths and weaknesses of microarray approaches to detect Pseudo-nitzschia species in the field. Environ. Sci. Pollut. Res. 20, 6705– 6718. https://doi.org/10.1007/s11356-012-1330-1 Borja, A., Elliott, M., Snelgrove, P.V.R., Austen, M.C., Berg, T., Cochrane, S., Carstensen, J., Danovaro, R., Greenstreet, S., Heiskanen, A.-S., Lynam, C.P., Mea, M., Newton, A., Patrício, J., Uusitalo, L., Uyarra, M.C., Wilson, C., 2016. Bridging the Gap between Policy and Science in Assessing the Health Status of Marine Ecosystems. Front. Mar. Sci. 3. https://doi.org/10.3389/fmars.2016.00175 Bourlat, S.J., Borja, A., Gilbert, J., Taylor, M.I., Davies, N., Weisberg, S.B., Griffith, J.F., Lettieri, T., Field, D., Benzie, J., Glöckner, F.O., Rodríguez-Ezpeleta, N., Faith, D.P., Bean, T.P., Obst, M., 2013. Genomics in marine monitoring: New opportunities for assessing marine health status. Mar. Pollut. Bull. 74, 19–31. https://doi.org/10.1016/j.marpolbul.2013.05.042 Cravo, A., Relvas, P., Cardeira, S., Rita, F., Madureira, M., Sánchez, R., 2010. An upwelling filament off southwest Iberia: Effect on the chlorophyll a and nutrient export. Cont. Shelf Res. 30, 1601–1613. https://doi.org/https://doi.org/10.1016/j.csr.2010.06.007 Cropper, T.E., Hanna, E., Bigg, G.R., 2014. Spatial and temporal seasonal trends in coastal upwelling off Northwest Africa, 1981-2012. Deep. Res. Part I Oceanogr. Res. Pap. 86, 94–111. https://doi.org/10.1016/j.dsr.2014.01.007 Culverhouse, P.F., Williams, R., Reguera, B., Herry, V., Gonzalez Gil, S., 2003. Do experts make mistakes? A comparison of human and machine labelling of dinoflagellates. Mar. Ecol. Prog. Ser. 247, 17–25. https://doi.org/10.3354/meps247017 Davidson, K., Anderson, D.M., Mateus, M., Reguera, B., Silke, J., Sourisseau, M., Maguire, J., 2016. Forecasting the risk of harmful algal blooms. Harmful Algae 53, 1–7. https://doi.org/10.1016/j.hal.2015.11.005 Directive 2008/56/EC of the European Parliament and of the Council of 17 June 2008. Establishing a framework for community action in the field of marine environmental policy (Marine Strategy Framework Directive). Official Journal of the European Union, 164/19. Dodge, J.D. (1982) Marine Dinoflagellates of the British Isles. Her Majesty's Stationery Office, London. 303 pp. Dittami, S., Edvardsen, B., 2012. GPR-Analyzer: A simple tool for quantitative analysis of hierarchical multispecies microarrays, Environmental science and pollution research international. https://doi.org/10.1007/s11356-012-1051-5

30

Dittami, S.M., Hostyeva, V., Egge, E.S., Kegel, J.U., Eikrem, W., Edvardsen, B., 2013. Seasonal dynamics of harmful algae in outer Oslofjorden monitored by microarray, qPCR, and microscopy. Environ. Sci. Pollut. Res. 20, 6719–6732. https://doi.org/10.1007/s11356-012-1392-0 Dittami, S.M., Pazos, Y., Laspra, M., Medlin, L.K., 2013. Microarray testing for the presence of toxic algae monitoring programme in Galicia (NW Spain). Environ. Sci. Pollut. Res. 20, 6778–6793. https://doi.org/10.1007/s11356-012-1295-0 Edvardsen, B., Dittami, S.M., Groben, R., Brubak, S., Escalera, L., Rodríguez, F., Reguera, B., Chen, J., Medlin, L.K., 2013. Molecular probes and microarrays for the detection of toxic algae in the genera Dinophysis and Phalacroma (Dinophyta). Environ. Sci. Pollut. Res. 20, 6733–6750. https://doi.org/10.1007/s11356-012-1403-1 Fehling, J., Davidson, K., Bolch, C., Tett, P., 2006. Seasonality of Pseudo-nitzschia spp.(Bacillariophyceae) in western Scottish waters. Mar. Ecol. Prog. Ser. 323, 91–105. Fiúza, a. F.D., de Macedo, M.E., Guerreiro, M.R., 1982. Climatological space and time variation of the Portuguese coastal upwelling. Oceanol. Acta 5, 31–40. Garel, E., Laiz, I., Drago, T., Relvas, P., 2016. Characterisation of coastal counter-currents on the inner shelf of the Gulf of Cadiz. J. Mar. Syst. 155, 19–34. https://doi.org/10.1016/j.jmarsys.2015.11.001 Garmendia, M., Borja, Á., Franco, J., Revilla, M., 2013. Phytoplankton composition indicators for the assessment of eutrophication in marine waters: Present state and challenges within the European directives. Mar. Pollut. Bull. 66, 7–16. https://doi.org/https://doi.org/10.1016/j.marpolbul.2012.10.005 Goela, P., Danchenko, S., Icely, J., Lubián, L., Cristina, S., Newton, A., 2014. Using CHEMTAX to evaluate seasonal and interannual dynamics of the phytoplankton community off the South-west Coast of Portugal. Estuar. Coast. Shelf Sci. 151, 112–123. https://doi.org/10.1016/j.ecss.2014.10.001 Goela, P.C.P.C., Icely, J., Cristina, S.S., Danchenko, S., DelValls, T.A., Newton, A., Angel DelValls, T., Newton, A., 2015. Using bio-optical parameters as a tool for detecting changes in the phytoplankton community (SW Portugal). Estuar. Coast. Shelf Sci. 167. https://doi.org/http://dx.doi.org/10.1016/j.ecss.2015.07.037 Gómez, F., 2005. A list of free-living dinoflagellate species in the world’s oceans. Acta Bot. Croat 64, 129– 212. Gu, H., Luo, Z., Krock, B., Witt, M., Tillmann, U., 2013. Morphology, phylogeny and azaspiracid profile of Azadinium poporum (Dinophyceae) from the China Sea. Harmful Algae 21–22, 64–75. https://doi.org/10.1016/J.HAL.2012.11.009 Guiry, M.D. & Guiry, G.M. 2018. AlgaeBase. World-wide electronic publication, National University of Ireland, Galway. http://www.algaebase.org; searched on 01 December 2018. Hasle, G.R., Syvertsen, E.E., 1997. Chapter 2 - Marine Diatoms, in: Tomas, C.R.B.T.-I.M.P. (Ed.), . Academic Press, San Diego, pp. 5–385. https://doi.org/https://doi.org/10.1016/B978-012693018-4/50004-5 Haynes, R., Barton, E.D., Pilling, I., 1993. Development, persistence, and variability of upwelling filaments off the Atlantic coast of the Iberian Peninsula. J. Geophys. Res. 98, 22681. https://doi.org/10.1029/93JC02016 Lewis, J., Medlin, Linda K. R.R. (Ed.), 2012. MIDTAL (Microarrays for the Detection of Toxic Algae): A 31

Protocol for a Successful Microarray Hybridisation and Analysis. Gantner Verlag. IPMA - Instituto Português do Mar e de Atmosfera, I.P. Plano de açao. Sistema nacional de monotorização de moluscos bivalves. Novembro 2013, Lisboa. IPMA - Instituto Português do Mar e de Atmosfera, I.P. Resultados das Análises de Fitoplâncton na água. 2014 – 2016. http://www.ipma.pt/pt/bivalves, accessed on 06 December 2018. Kavanagh, S., Brennan, C., O’Connor, L., Moran, S., Salas, R., Lyons, J., Silke, J., Maher, M., 2010. Real-time PCR Detection of Dinophysis Species in Irish Coastal Waters. Mar. Biotechnol. 12, 534–542. https://doi.org/10.1007/s10126-009-9238-6 Kegel, J.U., Del Amo, Y., Costes, L., Medlin, L.K., 2013a. Testing a Microarray to Detect and Monitor Toxic Microalgae in Arcachon Bay in France. Microarrays 2, 1–23. Kegel, J.U., Del Amo, Y., Medlin, L.K., 2013b. Introduction to project MIDTAL: its methods and samples from Arcachon Bay, France. Environ. Sci. Pollut. Res. 20, 6690–6704. https://doi.org/10.1007/s11356- 012-1299-9 Kim, J.-H., Tillmann, U., Adams, N.G., Krock, B., Stutts, W.L., Deeds, J.R., Han, M.-S., Trainer, V.L., 2017. Identification of Azadinium species and a new azaspiracid from Azadinium poporum in Puget Sound, Washington State, USA. Harmful Algae 68, 152–167. https://doi.org/10.1016/J.HAL.2017.08.004 Krock, B., Tillmann, U., Cembella, A., 2009a. Isolation and toxin composition of the azaspiracid-producing dinoflagellate from the Danish west coast. Molluscan Shellfish Saf. 74. Krock, B., Tillmann, U., John, U., Cembella, A.D., 2009b. Characterization of azaspiracids in plankton size- fractions and isolation of an azaspiracid-producing dinoflagellate from the North Sea. Harmful Algae 8, 254–263. https://doi.org/10.1016/J.HAL.2008.06.003 Krock, B., Tillmann, U., Witt, M., Gu, H., 2014. Azaspiracid variability of Azadinium poporum (Dinophyceae) from the China Sea. Harmful Algae. https://doi.org/10.1016/j.hal.2014.04.012 Kudela, R., G. Pitcher, T. Probyn, F. Figueiras, T. Moita, and V. Trainer. 2005. Harmful algal blooms in coastal upwelling systems. Oceanography 18(2):184–197. Kudela, R.M., Seeyave, S., Cochlan, W.P., 2010. The role of nutrients in regulation and promotion of harmful algal blooms in upwelling systems. Prog. Oceanogr. 85, 122–135. https://doi.org/10.1016/j.pocean.2010.02.008 Large, W.G., Pond, S., 1982. Sensible and Latent Heat Flux Measurements over the Ocean. J. Phys. Oceanogr. 12, 464–482. https://doi.org/10.1175/1520-0485(1982)012<0464:SALHFM>2.0.CO;2 Loureiro, S., Newton, A., Icely, J.D., 2005. Microplankton composition , production and upwelling dynamics in Sagres ( SW Portugal ) during the summer of 2001. Sci. Mar. 69, 323–341. https://doi.org/10.3989/scimar.2005.69n3323 Loureiro, S., Reñé, A., Garcés, E., Camp, J., Vaqué, D., 2011. Harmful algal blooms (HABs), dissolved organic matter (DOM), and planktonic microbial community dynamics at a near-shore and a harbour station influenced by upwelling (SW Iberian Peninsula). J. Sea Res. 65, 401–413. https://doi.org/10.1016/j.seares.2011.03.004 Lund, J.W.G., Kipling, C., Le Cren, E.D., 1958. The inverted microscope method of estimating algal numbers 32

and the statistical basis of estimations by counting. Hydrobiologia 11, 143–170. https://doi.org/10.1007/BF00007865 Luo, Z., Krock, B., Mertens, K.N., Nézan, E., Chomérat, N., Bilien, G., Tillmann, U., Gu, H., 2017. Adding new pieces to the Azadinium (Dinophyceae) diversity and biogeography puzzle: Non-toxigenic Azadinium zhuanum sp. nov. from China, toxigenic A. poporum from the Mediterranean, and a non-toxigenic A. dalianense from the French Atlantic. Harmful Algae 66, 65–78. https://doi.org/https://doi.org/10.1016/j.hal.2017.05.001 Luo, Z., Krock, B., Mertens, K.N., Price, A.M., Turner, R.E., Rabalais, N.N., Gu, H., 2016. Morphology, molecular phylogeny and azaspiracid profile of Azadinium poporum (Dinophyceae) from the Gulf of Mexico. Harmful Algae 55, 56–65. https://doi.org/10.1016/j.hal.2016.02.006 McCoy, G.R., McNamee, S., Campbell, K., Elliott, C.T., Fleming, G.T.A., Raine, R., 2014. Monitoring a toxic bloom of Alexandrium minutum using novel microarray and multiplex surface plasmon resonance biosensor technology. Harmful Algae 32, 40–48. https://doi.org/https://doi.org/10.1016/j.hal.2013.12.003 McCoy, G.R., Touzet, N., Fleming, G.T.A., Raine, R., 2015. Evolution of the MIDTAL microarray: the adaption and testing of oligonucleotide 18S and 28S rDNA probes and evaluation of subsequent microarray generations with Prymnesium spp. cultures and field samples. Environ. Sci. Pollut. Res. 22, 9704–9716. https://doi.org/10.1007/s11356-015-4152-0 Medlin, L., 2013. Molecular tools for monitoring harmful algal blooms. Environ. Sci. Pollut. Res. 20, 6683– 6685. https://doi.org/10.1007/s11356-012-1195-3 Medlin, L.K., Montresor, M., Graneli, E., Reugera, B., Raine, R., Edvardsen, B., Lewis, J., Elliott, C., Pazos, Y., Maranda, L., 2013. MIDTAL (Microarrays for the Detection of toxic Algae). Phytotaxa 127, 201–210. https://doi.org/10.11646/phytotaxa.127.1.19 Medlin, L.K., Orozco, J., 2017. Molecular Techniques for the Detection of Organisms in Aquatic Environments, with Emphasis on Harmful Algal Bloom Species. Sensors (Basel). 17. https://doi.org/10.3390/s17051184 Moita, M.T., Oliveira, P.B., Mendes, J.C., Palma, A.S., 2003. Distribution of chlorophyll a and Gymnodinium catenatum associated with coastal upwelling plumes off central Portugal. Acta Oecologica 24, S125–S132. https://doi.org/10.1016/S1146-609X(03)00011-0 Nézan, E., Tillmann, U., Bilien, G. l, Boulben, S., Chèze, K., Zentz, F., Salas, R., Chomérat, N., 2012. Taxonomic revision of the dinoflagellate Amphidoma caudata: transfer to the genus Azadinium (Dinophyceae) and proposal of two varieties, based on morphological and molecular phylogenetic analyses. J. Phycol. 48, 925–939. https://doi.org/10.1111/j.1529-8817.2012.01159.x Oliveira, P.B., Nolasco, R., Dubert, J., Moita, T., Peliz, Á., 2009. Surface temperature, chlorophyll and advection patterns during a summer upwelling event off central Portugal. Cont. Shelf Res. 29, 759–774. https://doi.org/10.1016/j.csr.2008.08.004 Paterson, R.F., McNeill, S., Mitchell, E., Adams, T., Swan, S.C., Clarke, D., Miller, P.I., Bresnan, E., Davidson, K., 2017. Environmental control of harmful dinoflagellates and diatoms in a fjordic system. Harmful Algae 69, 1–17. https://doi.org/https://doi.org/10.1016/j.hal.2017.09.002 Potvin, É., Jeong, H.J., Kang, N.S., Tillmann, U., Krock, B., 2012. First report of the photosynthetic 33

dinoflagellate genus Azadinium in the Pacific Ocean: Morphology and molecular characterization of Azadinium cf. poporum. J. Eukaryot. Microbiol. https://doi.org/10.1111/j.1550-7408.2011.00600.x Relvas, P., Barton, E.D., 2005. A separated jet and coastal counterflow during upwelling relaxation off Cape São Vicente (Iberian Peninsula). Cont. Shelf Res. 25, 29–49. https://doi.org/10.1016/j.csr.2004.09.006 Relvas, P., Barton, E.D., Dubert, J., Oliveira, P.B., Peliz, Á., da Silva, J.C.B., Santos, A.M.P., 2007. Physical oceanography of the western Iberia ecosystem: Latest views and challenges. Prog. Oceanogr. 74, 149–173. https://doi.org/10.1016/j.pocean.2007.04.021 Reynolds, C.S., 2006. The ecology of phytoplankton, The Ecology of Phytoplankton. https://doi.org/10.1017/CBO9780511542145 Reynolds, R.W., Smith, T.M., Liu, C., Chelton, D.B., Casey, K.S., Schlax, M.G., 2007. Daily High-Resolution- Blended Analyses for Sea Surface Temperature. J. Clim. 20, 5473–5496. https://doi.org/10.1175/2007JCLI1824.1 Ribeiro, S., Amorim, A., Abrantes, F., Ellegaard, M., 2016. Environmental change in the Western Iberia Upwelling Ecosystem since the preindustrial period revealed by dinoflagellate cyst records. Holocene 26, 874–889. https://doi.org/10.1177/0959683615622548 Ruggiero, M.V., Sarno, D., Barra, L., Kooistra, W.H.C.F., Montresor, M., Zingone, A., 2015. Diversity and temporal pattern of Pseudo-nitzschia species (Bacillariophyceae) through the molecular lens. Harmful Algae 42, 15–24. https://doi.org/https://doi.org/10.1016/j.hal.2014.12.001 Salas, R., Tillmann, U., John, U., Kilcoyne, J., Burson, A., Cantwell, C., Hess, P., Jauffrais, T., Silke, J., 2011. The role of Azadinium spinosum (Dinophyceae) in the production of azaspiracid shellfish poisoning in mussels. Harmful Algae. https://doi.org/10.1016/j.hal.2011.06.010 Sánchez, R.F., Mason, E., Relvas, P., da Silva, A.J., Peliz, Á., 2006. On the inner-shelf circulation in the northern Gulf of Cádiz, southern Portuguese shelf. Deep. Res. Part II Top. Stud. Oceanogr. 53, 1198– 1218. https://doi.org/10.1016/j.dsr2.2006.04.002 Shumway, S., M Burkholder, J., Morton, S., 2018. Harmful Algal Blooms: A Compendium Desk Reference. Steidinger, K.A., Jangen, K., 1997. Chapter 3 - Dinoflagellates, in: Tomas, C.R.B.T.-I.M.P. (Ed.), . Academic Press, San Diego, pp. 387–584. https://doi.org/https://doi.org/10.1016/B978-012693018-4/50005-7 Taylor, J.D., Berzano, M., Percy, L., Lewis, J., 2013. Evaluation of the MIDTAL microarray chip for monitoring toxic microalgae in the Orkney Islands, U.K. Environ. Sci. Pollut. Res. 20, 6765. https://doi.org/10.1007/s11356-012-1393-z Throndsen, J., 1997. Chapter 5 - The Planktonic Marine Flagellates, in: Tomas, C.R.B.T.-I.M.P. (Ed.), . Academic Press, San Diego, pp. 591–729. https://doi.org/https://doi.org/10.1016/B978-012693018- 4/50007-0 Tillmann, U., Borel, C.M., Barrera, F., Lara, R., Krock, B., Almandoz, G.O., Witt, M., Trefault, N., 2016. Azadinium poporum from the Argentine Continental Shelf, Southwestern Atlantic, produces azaspiracid-2 and azaspiracid-2 phosphate. Harmful Algae 51, 40–55. https://doi.org/10.1016/J.HAL.2015.11.001 Tillmann, U., Edvardsen, B., Krock, B., Smith, K.F., Paterson, R.F., Voß, D., 2018. Diversity, distribution, and azaspiracids of (Dinophyceae) along the Norwegian coast. Harmful Algae 80, 15–34. 34

https://doi.org/10.1016/J.HAL.2018.08.011 Tillmann, U., Elbrächter, M., Krock, B., John, U., Cembella, A., 2009. Azadinium spinosum gen. et sp. nov. (Dinophyceae) identified as a primary producer of azaspiracid toxins. Eur. J. Phycol. 44, 63–79. https://doi.org/10.1080/09670260802578534 Tillmann, U., Gottschling, M., Nézan, E., Krock, B., Bilien, G., 2014. Morphological and Molecular Characterization of Three New Azadinium Species (Amphidomataceae, Dinophyceae) from the Irminger Sea. Protist 165, 417–444. https://doi.org/https://doi.org/10.1016/j.protis.2014.04.004 Tillmann, U., Jaén, D., Fernández, L., Gottschling, M., Witt, M., Blanco, J., Krock, B., 2017a. Amphidoma languida (Amphidomatacea, Dinophyceae) with a novel azaspiracid toxin profile identified as the cause of molluscan contamination at the Atlantic coast of southern Spain. Harmful Algae 62, 113–126. https://doi.org/10.1016/J.HAL.2016.12.001 Tillmann, U., Sánchez-Ramires, S., Krock, B., Bernales, A., 2017b. A bloom of Azadinium polongum in coastal waters off Peru, Revista de Biologia Marina Y Oceanografia. https://doi.org/10.4067/S0718- 19572017000300015 Tillmann, U., Soehner, S., Nézan, E., Krock, B., 2012. First record of the genus Azadinium (Dinophyceae) from the Shetland Islands, including the description of Azadinium polongum sp. nov. Harmful Algae 20, 142–155. https://doi.org/https://doi.org/10.1016/j.hal.2012.10.001 Tomas, C.R.B.T.-I.M.P. (Ed.), 1997. Front Matter. Academic Press, San Diego, p. iii. https://doi.org/https://doi.org/10.1016/B978-012693018-4/50012-4 Trainer, V.L., Pitcher, G.C., Reguera, B., Smayda, T.J., 2010. The distribution and impacts of harmful algal bloom species in eastern boundary upwelling systems. Prog. Oceanogr. 85, 33–52. https://doi.org/10.1016/j.pocean.2010.02.003 Utermohl, V., (1931) (Mit besondere Beriicksichtigung des Ultraplanktons). Verhandlungen. Neue Wege in der quantitativen Erfassung des Planktons, 5. Internationale Vereinigung fur Theoretische und Angewandte Limnologie, pp. 567-595. Vale, P., Bire, R., Hess, P., 2008a. Confirmation by LC–MS/MS of azaspiracids in shellfish from the Portuguese north-western coast. Toxicon 51, 1449–1456. https://doi.org/10.1016/J.TOXICON.2008.03.022 Vale, P., Botelho, M.J., Rodrigues, S.M., Gomes, S.S., Sampayo, M.A. de M., 2008b. Two decades of marine biotoxin monitoring in bivalves from Portugal (1986-2006): A review of exposure assessment. Harmful Algae 7, 11–25. https://doi.org/10.1016/j.hal.2007.05.002 Zhang, H., Reynolds, R., Bates, J., 2006. Blended and gridded high resolution global sea surface wind speed and climatology from multiple satellites: 1987–present. Am. Meteorol. Soc. -- 2006 Annu. Meet. Pap. #P2.23, Atlanta, GA.

35

Fig.1. Study site at the offshore mussel farm at Sagres, SW Iberia (Portugal); IPMA shellfish monitoring zones L7a and L7c, as of the time of this study (2014 – 2016); distances from IPMA toxic phytoplankton monitoring stations (Arrifana, L7a and Porto de Mos, L7c) are denoted by arrows.

36

Fig. 2. SST and upwelling components (Qx and Qy) A) during 2015 and B) during 2016; eastward Ekman transport Qx is converted to positive for illustration. In this figure positive values of Qx and negative values of

Qy are indicative of upwelling conditions, and are usually followed by a decrease in SST.

37

Fig. 3. SST maps on (A) 22-Jul-2014 and (B) 10-Set-2015 shows the spread of cold water along the SW coast of Portugal under pronounced upwelling conditions. SST analysis on (C) 26-Apr-2016 shows upwelling relaxation, whilst on (D) 27-May-2016 shows SST increase in the end of upwelling period followed by relaxation conditions. Data courtesy: NASA JPL Multi-scale Ultra-high resolution daily SST analysis, Global 0.01o, 2002 – present.

0 3 -1 Fig. 4. Change of SST ( C) and upwelling components (eastward Qx and northward Qy Ekman transport, m s km-1) during 5 days preceding the sampling dates, and in-situ Chl a (mg L-1). Negative Ekman transport corresponds to SST decrease and coincident Chl a increase, indicating upwelling conditions.

38

Fig. 5. Cell counts of phytoplankton: composition of the main taxonomic groups (cells ml-1) and their co- occurrence with toxic diatom genus Pseudo-nitzschia during the study period 2014 - 2016. Phytoplankton was dominated mostly by nanoflagellates and diatoms.

Fig. 6. Diatoms abundance (cells ml-1) and in-situ Chl a concentration (mg L-1, ±SD). Dependence of Chl a concentration on diatom abundance, R2=0.53 (n=25), demonstrates that diatoms were the major contributor to Chl a.

39

40

Fig. 7. Microarray and LM results of Pseudo-nitzschia, microarray probes signal normalised to the positive control (POSITIVE_25_dT). (A) Genus level probes; (B – D) multispecies probes detecting: (B) P. delicatissima and P. pseudodelicatissima, (C) P. caciantha, calliantha, australis, galaxiae and (D) P. seriata, multistriata, mannii, pungens. The hatched line in (A) shows the regulatory level at which the shellfishery is closed (80 cells ml-1 of P. seriata). The first series of samples from 2014 (i) is shown at a different scale from 2015 - 2016 (ii) due to the different image processing (see Methods).

41

Fig. 8. (A) Dinophysis genus level probes, (B) D. acuminata and D. acuta species probes, and (C) Phalacroma rotundatum species probes signal, normalised to positive control (POSITIVE_25_dT) compared to microscopy counts (cells ml-1). The first series of samples from 2014 (i) is shown at a different scale from 2015 - 2016 (ii) due to the different image processing (see Methods).

42

Fig. 9. Microarray and LM results of Prorocentrum detection: (A) genus and clade probes signal (normalized by POSITIVE_25_dT), (B) P. minimum and (C) P. micans probes signal compared to LM counts (cells ml-1), (D) quantification of P. micans cells based on microarray probe compared to LM cell counts (cells ml-1). The first series of samples from 2014 (i) is shown at a different scale from 2015 - 2016 (ii) due to the different image processing (see Methods).

43

Fig. 10. (A) Microarray signals of Alexandrium probes, (B) Quantification of Alexandrium spp. based on AlexGD01 probe signal (cells L-1) compared to cell counts. Cell numbers inferred from probes signal intensity are corrected to 1 L sample volume.The first series of samples from 2014 (i) is shown at a different scale from 2015 - 2016 (ii) due to the different image processing (see Methods).

Fig. 11. Microarray signals of Azadinium genus probes compared to LM counts, including counts of resembling A. dinoflagellates cf. Azadinium spp., and morphologically similar thecate dinoflagellates <20 µm. The first

44 series of samples from 2014 (i) is shown at a different scale from 2015 - 2016 (ii) due to the different image processing (see Methods).

45

Fig. 12. Microarray probes signal: (A) Gymnodinium catenatum with cell counts, (B) Karenia brevis, (C) K. mikimotoi, (D) Karlodinium spp. and (E) cell counts of unidentified Gymnodinioid dinoflagellates (cells ml-1). 46

The first series of samples from 2014 (i) is shown at a different scale from 2015 - 2016 (ii) due to the different image processing (see Methods).

47

Fig.13. Microarray detection of flagellates: (A) Prymnesium spp. and P. parvum probes compared to LM counts of total nanoflagellates (cells ml-1), (B) Pseudochattonella spp. probes, (C) Chloromorum toxicum probes, (D) Heterosigma akashiwo probes compared to LM counts (cells ml-1). The first series of samples from 2014 (i) is shown at a different scale from 2015 - 2016 (ii) due to the different image processing (see Methods).

48

Table 1. Nutrient concentrations from 24th March to 14th June 2016, µmol L-1. LOD – limit of detection, LOQ – limit of quantification.

+ - - 3- 4- Sample date NH4 NO3 NO2 PO4 SiO4 24-Mar-2016 0.53 5.82 0.19 < LOD 1.39 02-Apr-2016 1.25 4.15 0.18 < LOD 1.38 09-Apr-2016 0.91 4.81 0.20 < LOD 1.20 26-Apr-2016 1.25 1.98 (0.12) < LOQ < LOD 0.88 05-May-2016 1.11 0.53 < LOD < LOD 0.72 17-May-2016 0.70 1.53 (0.07) < LOQ < LOD 0.56 27-May-2016 1.53 0.57 (0.05) < LOQ < LOD 0.83 02-Jun-2016 1.12 5.28 0.21 (0.20) < LOQ 1.58 14-Jun-2016 (0.21) < LOQ 2.88 (0.12) < LOQ < LOD 1.52

49

Table 2. Oceanographic conditions and phytoplankton community general composition obtained by LM during 2014 – 2016 at Sagres study site.

Sample date Chl a, Oceanographic Phytoplankton community mg m-3 conditions

14-Jul-2014 1.5 Non-Upwelling Nanoflagellates; highest diversity of HAB dinoflagellates

22-Jul-2014 1.8 Upwelling Nanoflagellates bloom, Gymnodinioid dinoflagellates, moderate diatom development.

19-Aug-2014 5.7 Upwelling Diatom bloom, nanoflagellates, dinoflagellates present

18-Nov-2014 3.3 Mild Upwelling, mixing Nanoflagellates, small-cell diatoms and small athecate dinoflagellates

27-Jan-2015 0.3 Upwelling, Mixing Low abundance nanoflagellates and small-cell diatoms

03-Jul-2015 0.3 Mild Upwelling Nanoflagellates and small-cell diatoms

13-Jul-2015 2.6 Upwelling Medium abundance of diatoms, nanoflagellates, dinoflagellates

21-Jul-2015 0.4 Mild Upwelling Nanoflagellates incl. Cryptophytes and small chain-forming diatoms development

30-Jul-2015 4.0 Upwelling Diatom bloom

10-Aug-2015 0.7 Non- Upwelling Nanoflagellates and dinoflagellates dominated community; diatoms collapsed

28-Aug-2015 1.3 Mild Upwelling Nanoflagellates and diatoms dominate, significant abundance of dinoflagellates

10-Sep-2015 4.8 Upwelling Bloom of nanoflagellates, cryptophytes, diatoms, significant abundance of dinoflagellates

19-Sep-2015 4.2 UW Relaxation start Nanoflagellates and diatoms bloom, complemented by dinoflagellates

24-Sep-2015 5.01 Upwelling Nanoflagellates (incl. Cryptophytes); Moderate diatom bloom, complemented by dinoflagellates

09-Oct-2015 1.22 Upwelling Nanoflagellates, significant abundance of diatoms

16-Oct-2015 1.69 UW Relaxation Nanoflagellates bloom, some dinoflagellates

24-Mar-2016 1.86 Upwelling Centric diatoms moderate bloom; Nanoflagellates; Gymnodinioid dinoflagellates

02-Apr-2016 1.76 Moderate Upwelling Moderate diatom and nanoflagellates development bloom

09-Apr-2016 4.50 Upwelling Diatom bloom incl. Pseudo-nitzschia, Nanoflagellates;

26-Apr-2016 1.77 Upwelling relaxation Diatom bloom largely decreased, dinoflagellates increased diversity

50

05-May-2016 0.57 Relaxation or mild Nanoflagellates, high diversity of HAB dinoflagellates Upwelling

17-May-2016 3.24 Upwelling in the narrow Moderate diatom and nanoflagellates bloom near-shore zone

27-May-2016 1.29 Upwelling Nanoflagellates, diatoms bloom

02-Jun-2016 4.31 Mild Upwelling Diatoms incl. Pseudo-nitzschia and nanoflagellates dominate

14-Jun-2016 4.87 Upwelling Diatom incl. Pseudo-nitzschia and nanoflagellates dominate

51

Table 3. HAB species detected by microscopy (LM) and microarray (MA) methods in relation to shellfish harvesting ban on bivalves imposed by the Portuguese regulating authority (IPMA): B – mussel Mytilus edulis banned, PB – only wedge clam Donax trunculus banned. Harvesting restrictions may be imposed based on direct toxins analysis in the mollusc tissue samples, as well as on results of phytoplankton monitoring. Zones borders and sampling stations are shown in Fig.1: L7a – Aljezur – Cape Saint Vincent, L7c – Cape Saint Vincent – Lagos. Microarray (MA) denoted as detected if hierarchy tests of genus, group or at least several species probes passed (except Prorocentrum). Pseudo-Nitzschia spp. including P. seriata and P. delicatissima concentrations in LM at or above 10 cells ml-1 are shown as significant. LM counts of Azadinium spp. may include other similar dinoflagellates (see text for details).

Zone L7a Zone L7c Sample date

West coast South coast spp. Nitzschia

-

catenatum

Phalacroma

rotundatum

Gymnodinium

Azadinium spp.Azadinium

Dinophysis spp. Dinophysis

Alexandrium spp.Alexandrium

Prorocentrum spp.

Pseudo

Heterosigma akashiwo 14-Jul-2014 B - LM, MA LM, MA LM, MA LM, MA MA MA LM,MA MA

22-Jul-2014 B - LM LM, MA MA LM, MA LM, MA LM, MA

19-Aug-2014 B LM, MA LM, MA MA MA LM, MA MA

18-Nov-2014 B until 13-Nov PB LM, MA MA LM, MA MA MA

27-Jan-2015 MA MA MA

03-Jul-2015 B LM, MA LM, MA LM, MA LM, MA

13-Jul-2015 B B LM, MA LM LM LM

21-Jul-2015 B B LM, MA LM LM, MA LM, MA LM, MA

30-Jul-2015 B PB until 27-Jul LM, MA LM LM LM

10-Aug-2015 B until 9-Aug LM LM, MA LM, MA LM, MA LM

28-Aug-2015 LM, MA LM, MA LM

10-Sep-2015 B LM, MA LM, MA MA LM, MA LM, MA LM LM

19-Sep-2015 B LM, MA LM, MA MA LM, MA MA MA

24-Sep-2015 B 25-Sep B until 23-Sep LM, MA LM, MA MA LM, MA LM, MA LM, MA MA MA

09-Oct-2015 B B MA LM, MA LM, MA LM

16-Oct-2015 B B LM, MA LM, MA MA

24-Mar-2016 LM, MA LM

02-Apr-2016 LM, MA LM LM LM LM

09-Apr-2016 LM, MA LM MA MA LM LM, MA

26-Apr-2016 B LM, MA LM, MA LM, MA LM, MA LM, MA MA MA LM, MA

05-May-2016 B B LM, MA LM, MA LM, MA LM

17-May-2016 B B LM, MA LM LM

27-May-2016 B B LM, MA LM, MA MA LM, MA LM, MA LM LM LM, MA

02-Jun-2016 B B LM, MA LM LM

14-Jun-2016 B PB LM, MA LM LM, MA LM, MA

52

Zone L7a Zone L7c Sample date

West coast South coast spp. Nitzschia

-

catenatum

Phalacroma

rotundatum

Gymnodinium

Azadinium spp.Azadinium

Dinophysis spp. Dinophysis

Alexandrium spp.Alexandrium

Prorocentrum spp.

Pseudo

Heterosigma akashiwo 14-Jul-2014 Closed N/A LM, MA LM, MA LM, MA LM, MA MA MA LM,MA MA

22-Jul-2014 Closed N/A LM LM, MA MA LM, MA LM, MA LM, MA

19-Aug-2014 Open Closed LM, MA LM, MA MA MA LM, MA MA

Closed until 18-Nov-2014 Partly Closed 13-Nov LM, MA MA LM, MA MA MA

27-Jan-2015 Open Open MA MA MA

03-Jul-2015 Open Closed LM, MA LM, MA LM, MA LM, MA

13-Jul-2015 Closed Closed LM, MA LM LM LM

21-Jul-2015 Closed Closed LM, MA LM LM, MA LM, MA LM, MA

Partly Closed 30-Jul-2015 Closed until 27-Jul LM, MA LM LM LM

Closed until 10-Aug-2015 Open 9-Aug LM LM, MA LM, MA LM, MA LM

28-Aug-2015 Open Open LM, MA LM, MA LM

10-Sep-2015 Open Closed LM, MA LM, MA MA LM, MA LM, MA LM LM

19-Sep-2015 Open Closed LM, MA LM, MA MA LM, MA MA MA

Closed on 25- Closed until 24-Sep-2015 Sep 23-Sep LM, MA LM, MA MA LM, MA LM, MA LM, MA MA MA

09-Oct-2015 Closed Closed MA LM, MA LM, MA LM

16-Oct-2015 Closed Closed LM, MA LM, MA MA

24-Mar-2016 Open Open LM, MA LM

02-Apr-2016 Open Open LM, MA LM LM LM LM

09-Apr-2016 Open Open LM, MA LM MA MA LM LM, MA

26-Apr-2016 Closed Open LM, MA LM, MA LM, MA LM, MA LM, MA MA MA LM, MA

05-May-2016 Closed Closed LM, MA LM, MA LM, MA LM

17-May-2016 Closed Closed LM, MA LM LM

27-May-2016 Closed Closed LM, MA LM, MA MA LM, MA LM, MA LM LM LM, MA

02-Jun-2016 Closed Closed LM, MA LM LM

14-Jun-2016 Closed Partly Closed LM, MA LM LM, MA LM, MA

53