Santa Giusta Lagoon (Italy, western Mediterranean Sea) is located along the west central coast of Sardinia Island (Fig. 1). It is a research station in part of the site “14 Sardinian marine ecosystems” of the LTER-Italy network (www.lteritalia.it). Santa Giusta Lagoon has an area of 8 km² and a mean depth of 1 m. The two primary freshwater inputs are located on the lagoon’s east side and sea exchanges are on the western side. Santa Giusta underwent substantial human modification during the last century, resulting in profound ecosystem alterations (Sechi et al. 2001, Lugliè et al. 2002). Sechi et al. (2001) signalled its hypertrophy and reported several fish kill events associated with harmful algal blooms. Moreover, Satta et al. (2014) reported the presence of harmful dinoflagellate cysts in the sediments.

The 27 archived samples analysed in this study (Table 1) belong to the LTER phytoplankton samples collection maintained at the Dipartimento di Architettura, Design e Urbanistica of the University of Sassari. They were collected from the water surface layer (–30 cm), from 3 to 5 stations (Fig. 1) during Chattonella blooms in the summers of 1994, 1998, 1999, 2010 and 2013, the first four coinciding with extensive fish mortalities. All samples were immediately fixed with Lugol’s iodine solution and analysed within ten days from sampling to assess Chattonella cell densities, following the Utermöhl method (Utermöhl 1958) and using an inverted Axiovert Zeiss 25 microscope. Further, within a few hours from the collection, live samples were always observed under the microscope for species identification.

Five clonal cultures of C. cf. subsalsa (Adriatic genotype) were established from samples collected from Santa Giusta Lagoon in July 2013 (UNISS7, UNISS8, UNISS9, UNISS10, UNISS11). Vegetative cells were isolated with glass micropipettes and transferred into IWAKI tissue culture multiplates. Plates were filled with L1 medium (Guillard and Hargraves 1993) prepared with filtered seawater adjusted to a salinity of 35, and maintained at 20±1°C with a 12:12 light:dark cycle. Illumination was provided by a photon irradiance of 100 µmol m^–2 s^–1.

Reference cultures of C. subsalsa (CCMP217; Global genotype) from the Scandinavian Culture Collection of Algae & Protozoa (SCCAP) and C. antiqua (C. marina var. antiqua, NIES 1) (fixed with Lugol’s iodine solution) from the Provasoli-Guillard National Centre for Marine Algae and Micobiota (NCMA, formerly CCMP) were also acquired.

Three artificial phytoplankton samples (Lugol-fixed, hereinafter named artificial samples) were used to create controlled conditions of presence of Global and Adriatic C. subsalsa genotypes and other algae, as could happen in natural conditions. The first sample contained CCMP217 and UNISS8 strains (sample A; i.e. both Global and Adriatic genotypes), the others only one of the two genotypes, respectively sample B the Global genotype (CCMP217) and sample C the Adriatic genotype (UNISS8; Table 2).

Further, we mixed in different proportions the DNA extracted from two different cultures of C. subsalsa genotypes (CCMP217 and UNISS8, respectively) by performing six tests (hereinafter named DNA tests, Table 3) to verify whether different DNA concentrations might affect the ITS-5.8S rDNA amplification region.

DNA was extracted with the DNeasy Plant Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s instructions from 50 mL of archived samples, artificial samples and from 15 mL of fixed cultures (Fig. 2).

PCR analyses were performed on archived samples from at least two of the sampled stations for each sampling date (Table 1). An archived sample with C. subsalsa abundance below the detection limit (<10 cells L^–1) and from a season not favourable to C. subsalsa growth (Station 3 of 5/4/2012) was used as a negative control. A first PCR was performed with 1 µL of extracted DNA (Fig. 2) in a total of 40.5 µL reaction mixture containing the following: 0.25 mM of each dNTP; 0.1µM of each primer; 2.5 mM MgCl₂; 1x HotMaster Taq Buffer (PRIME, Hamburg, Germany); and 2.5 U Taq DNA polymerase (PRIME). ITSA and ITSB primers (Adachi et al. 1994) were used to amplify the internal transcribed spacer (ITS) regions and 5.8S rDNA. PCR conditions were as follows: an initial denaturation step at 94°C for 5 min, followed by 35 cycles at 94°C for 20 s, 57°C for 10 s, and 70°C for 30 s; and a final elongation step at 70°C for 5 min. Six PCR replicates for each sample were performed, three with undiluted DNA and three with 1:10 diluted DNA (Fig. 2). Subsequently, 1 µL of each PCR product was used in two distinct nested PCR amplifications (Fig. 2), with the same mixture condition as above. These two different nested PCRs were carried out with specific primers: oBTG-005-F (CTGGGGAAGGATCATTACC) and oBTG-027-R (GCCGATTGCTTCCAGAGA) for C. marina, and oBTG-005-F and oBTG-028-R(CGCCACTCGTTGCCCAGT) for C. subsalsa (Connell 2002). Nested PCR conditions were as follows: an initial denaturation step at 95°C for 5 min; then 35 cycles at 95°C for 30 s, 55°C for 10 s (using species-specific primers for C. marina) or 45°C for 10 s (using species-specific primers for C. subsalsa), and 72°C for 30 s; and a final elongation step at 72°C for 10 min. The DNA extracted from the cultures C. subsalsa CCMP217 and C. antiqua (C. marina var. antiqua) NIES 1 were used as a positive control in each PCR on the analysed archived samples.

Genetic analyses were also conducted to confirm species and genotype identification of the five C. cf. subsalsa UNISS7, UNISS8, UNISS9, UNISS10, UNISS11 cultures. For the ITS-5.8S rDNA sequences, the first PCR was performed using ITSA and ITSB primers (Adachi et al. 1994) and the nested PCR with primers oBTG-005-F and oBTG-028-R (Connell 2002), with the same PCR protocol as that described above. PCR primers D1R and D2C (Scholin et al. 1994) were used to amplify the LSU rDNA. PCR was carried out in 50-μL reactions containing 1 μL of DNA extract, 0.8 µM of each primer, 200 µM of dNTPs (Qiagen mix), PCR Buffer 1X (Qiagen) containing 1.5 mM of MgCl₂, and 1.25 U of Taq DNA polymerase. Thermocycling included one initial step at 95°C for 5 min followed by 40 cycles at 95°C for 20 s, 55°C and at 72°C for 1 min, followed by a final extension at 72°C for 10 min.

The whole protocol applied on archived samples was also tested on the three artificial samples (Fig. 2, Table 2). The same PCR-based assay was also applied on the six DNA tests (Table 3). The DNA concentration was evaluated with the SmartSpec^TM Plus Spectrophotometer (Bio-rad) following the manufacturer’s instruction.

All PCR amplifications were performed in a DNA Engine^® Thermal Cycler.

All PCR products were resolved on a 1.8% (80v) agarose gel. All nested PCR products were purified and sequenced by an external service (Macrogen Inc., Europe) using both primers, and a 3730XL DNA sequencer.

Sequences obtained in this study were compared with sequences in the NCBI Nucleotide Collection (BLAST Algorithm; http://www.ncbi.nlm.-nih.gov) to determine the closest known sequences. Sequences were also aligned with those obtained from GenBank (Table 4 and 5) using the MAFFT v.6 program (Katoh et al. 2002) under FFT-NS-i (slow; iterative refinement method). Alignments were manually checked with BioEdit v. 7.0.5 (Hall 1999). Phylogenetic relationships, based on the LSU rDNA data (Table 5), were inferred using maximum likelihood (ML) method and the GTRGAMMA evolution model on Randomized Axelerated Maximum Likelihood v. 7.0.4 (RA×ML) (Stamatakis 2006). All model parameters were estimated by RA×ML, using Pseudochattonella verruculosa strains from public databases as an outgroup. The tree with the best topology (the one with the greatest likelihood of 1000 alternative trees) was selected by repeated runs on distinct starting trees. Bootstrap ML analysis was done with 1000 pseudo-replicates and the consensus tree was computed with the RA×ML software.

Through the sequencing and BLAST analysis of the nested PCR products, C. subsalsa was detected in all archived samples, which resulted positive also for microscope analysis (Table 1). The only exception was a sample taken before the beginning of the bloom in 2010 (17/7/2010), for which PCR results were positive and microscope analysis negative, and the sample was assumed as negative (Station 3 of 5/4/2012).

All PCR products unequivocally belonged to C. subsalsa, with control DNA of 380 bp length for C. subsalsa CCMP217 (Fig 3a, GenBank accession number KR709218) and 181 bp for C. antiqua (C. marina var. antiqua) NIES 1 (Fig 3b).

ITS-5.8S rDNA sequences comparison among sequences from the 24 archived samples and those of C. subsalsa deposited in GenBank (Table 4) yielded a BLAST analysis with a 99%-100% similarity with the Global genotype sequences, 96%-97% sequence identity with the C. cf. subsalsa strains CRIM E and CRIM F (i.e. Adriatic genotype) and 98%-99% sequence identity with the C. cf. subsalsa strain CHPI36.

The nested PCR on cellular cultures produced six ITS-5.8S rDNA sequences (Table 4), four of which were of 573 bp (UNISS7, UNISS8, UNISS9, UNISS10 strains) and one of 576 pb (UNISS11 strain). These five sequences were longer than the nested PCR product of CCMP217 strain (380 pb) (Fig. 4).

The MAFT alignment among the oBTG-028-R primer and the ITS-5.8S rDNA sequences of C. subsalsa strains (UNISS7, UNISS9, CCMP217, CRIM E and C. Tomas Sardinia) indicated seven differences for UNISS7, UNISS9, CRIM E strains (i.e. Adriatic genotype) with respect to the CCMP217 and C. Tomas Sardinia strains (i.e. Global genotype). The BLAST analysis of 5.8S-ITS rDNA sequences showed a 99%-100% similarity of UNISS7, UNISS8, UNISS9, UNISS10 and UNISS11 strains with C. cf. subsalsa CRIM E and CRIM F strains and 94%-96% sequence identity with all C. subsalsa Global sequences (Table 4).

The six LSU rDNA partial sequences were long 652 pb for UNISS7, 650 pb for UNISS8, UNISS10 and UNISS11, 651 pb for UNISS9 and 643 pb for CCMP217. The ML phylogenetic tree obtained for LSU rDNA sequences showed that C. subsalsa strains divided into two distinct groups (Fig. 5). The first group showed a branch with the only C. cf. subsalsa CHPI36 and another with C. subsalsa CCMP217 and C. subsalsa AF409126. The second group included all C. cf. subsalsa Adriatic genotype (Klöpper et al. 2013) and UNISS7, UNISS8, UNISS9, UNISS10, UNISS11 strains, with a bootstrap value of 100%. All strains of C. marina var. marina, C. marina var. ovata, and C. marina var. antiqua grouped together with a 100% bootstrap support.

The nested PCR of the samples A and B produced sequences with a length of 380 pb belonging to C. subsalsa CCMP217. The nested PCR of sample C produced a sequence with a length of 573 pb belonging to C. cf. subsalsa UNISS8.

The nested PCR of DNA tests produced sequences with a length of 380 pb belonging to the CCMP217 strain even when C. cf. subsalsa UNISS8 had a concentration one hundred times higher.