Morphological Variability of Pseudo-nitzschia pungens Clade I (Bacillariophyceae) in the Northwestern Adriatic Sea

Pseudo-nitzschia pungens is a common component of the phytoplankton community of the northern Adriatic Sea. In this study, an in-depth morphological analysis of P. pungens was carried out in both cultured strains isolated in different periods and field samples, revealing a surprisingly wide variability in a number of details, with both the gross morphology and ultrastructural levels deviating from the nominal P. pungens. Colonies showed an overlap (from one-third to one-sixth) and a transapical axis (rarely reaching 3 µm), strongly differing from the original description of the species. Moreover, valves may be either symmetrical or slightly asymmetrical, with striae almost always biseriate but sometimes uniseriate or triseriate. Poroids’ morphology in cingular bands was characterized by a wide variability (square, circular, or rectangular poroids without or with up to two hymen sectors), with several combination of them, even within the same cingular band. Phylogenetic analyses based on ITS rDNA showed that the P. pungens of the northern Adriatic Sea belonged to clade I. Domoic acid was not detected.


Introduction
Planktonic diatom species of the genus Pseudo-nitzschia are recorded in coastal regions worldwide. Among the 53 known species [1], 26 have been shown to produce domoic acid (DA) [2,3], causing neurologic disorders and memory loss in vertebrates linked to the consumption of contaminated shellfish (Amnesic Shellfish Poisoning) [4,5]. In this genus, the number of genetic lineages is markedly higher than the number of taxa discernible by light microscope (LM). With the description of several new Pseudo-nitzschia species, detailing and comparing additional ultrastructural characters has become necessary [6][7][8]. Indeed, several species not easily distinguishable using LM have been described using electron microscopy (EM) coupled with molecular techniques [9][10][11][12][13][14][15][16][17].
Traditionally, species of the genus Pseudo-nitzschia have been subdivided in two LM-discernible groups based on cell width in valve view: All the species wider than 3 µm have been combined into the seriata group, while those less than 3 µm are in the delicatissima group [18]. Pseudo-nitzschia pungens, with its varieties, is a cosmopolitan species detected from temperate to tropical waters [8,[19][20][21], ascribed to the seriata group. The combination of a few morphological characters (including cell size, overlap of cells in colonies [22]) has usually been enough to distinguish P. pungens from other   and field samples, respectively. The number of fibulae in 10 µm was 5-18, and the number of striae in 10 µm was 9-16 ( Table 1). The number of fibulae was generally the same as the number of striae ( Figures 2D and 3A). At times, the number of fibulae was higher ( Figures 2B and 3C), but it was rarely lower.     Figures 2B and 4E). The number of poroids in 1 µm ranged between 1 and 4 ( Table 1). Cells with a very low density of poroids were not rare (22%, Figure 3D), and a decreasing density of poroids from the valve center toward the apices was often observed ( Figures 2B and 3B,E). Cells formed stepped colonies of several cells with an overlap that ranged from one-third to one-sixth of the AA length ( Figure 1A,B, Table 1). Cells were strongly silicified. The interstriae and fibulae were discernible in LM ( Figure 1A,C). Although colonies were generally symmetrical in valve view ( Figure 1C), asymmetrical colonies were not rare ( Figure 1D).
In valve view, valves were slightly asymmetrical with minor hemivalve to major hemivalve area ratios ranging from 0.56 to 0.99 (n = 49, 0.82 ± 0.12) and from 0.57 to 0.94 (n = 13, 0.80 ± 0.12) in cultured and field samples, respectively. The number of fibulae in 10 µm was 5-18, and the number of striae in 10 µm was 9-16 ( Table 1). The number of fibulae was generally the same as the number of striae ( Figures 2D and 3A). At times, the number of fibulae was higher ( Figures 2B and 3C), but it was rarely lower.
Striae were almost always (72% of the observations) biseriate ( Figure Figures 2B and 4E). The number of poroids in 1 µm ranged between 1 and 4 ( Table 1). Cells with a very low density of poroids were not rare (22%, Figure 3D), and a decreasing density of poroids from the valve center toward the apices was often observed ( Figures 2B and 3B,E).  In girdle view, cingular bands showed a wide morphological variability both in the shape of poroids and in the number and size of sectors within them ( Figure S10A,B, respectively), even within the same cingular band ( Figures 5 and 6). In general, poroids' dimensions showed a decreasing trend in the abvalvar direction (i.e., poroids in the third cingular band were narrower than those in the first two, and sometimes (25%), no poroids were detected), even if, often (40%), the dimensions and the shape of poroids in the first two cingular bands did not differ so much (but the second band had more striae with smaller poroids than the valvocopula, e.g., Figure 5A). Band striae (12-23 in 10 µm, mean 16.2 ± 2.7, n = 88) were perforated, with (a) oval to rectangular, (b) square, or (c) circular poroids ( Figure S10), showing one, two (partially to completely divided), or no hymen sectors ( Figure 6). Within each cingular band, either poroids characterized by only one shape ( Figure 5C or Figure 5B(a)) or poroids with different shapes ( Figure 5D-F) could occur. Nevertheless, cingular bands with only circular poroids were never observed.
All the patterns of hymenation were observed in poroids ( Figure 6), except for circular poroids that were observed only without hymen sectors ( Figure 6A) or with one sector. In girdle view, cingular bands showed a wide morphological variability both in the shape of poroids and in the number and size of sectors within them ( Figure S10A,B, respectively), even within the same cingular band (Figures 5 and 6). In general, poroids' dimensions showed a decreasing trend in the abvalvar direction (i.e., poroids in the third cingular band were narrower than those in the first two, and sometimes (25%), no poroids were detected), even if, often (40%), the dimensions and the shape of poroids in the first two cingular bands did not differ so much (but the second band had more striae with smaller poroids than the valvocopula, e.g., Figure 5A). Band striae (12-23 in 10 µm, mean 16.2 ± 2.7, n = 88) were perforated, with (a) oval to rectangular, (b) square, or (c) circular poroids ( Figure S10), showing one, two (partially to completely divided), or no hymen sectors ( Figure 6). Within each cingular band, either poroids characterized by only one shape ( Figure 5C or Figure 5B(a)) or poroids with different shapes ( Figure 5D-F) could occur. Nevertheless, cingular bands with only circular poroids were never observed.
All the patterns of hymenation were observed in poroids ( Figure 6), except for circular poroids that were observed only without hymen sectors ( Figure 6A) or with one sector.

Molecular Analyses
BLAST results of the 13 LSU sequences confirmed the identification of Pseudo-nitzschia pungens (showing from 99 to 100% of identity with Pseudo-nitzschia pungens LSU sequences from GenBank).
The final alignment was obtained from a total 94 ITS1-5.8s-ITS2 rDNA sequences of P. pungens

Molecular Analyses
BLAST results of the 13 LSU sequences confirmed the identification of Pseudo-nitzschia pungens (showing from 99 to 100% of identity with Pseudo-nitzschia pungens LSU sequences from GenBank).
The final alignment was obtained from a total 94 ITS1-5.8s-ITS2 rDNA sequences of P. pungens from different geographical locations, including four sequences from this study.
The complete alignment was rooted with P. multiseries (AY257844). The alignment comprised 632 characters, of which 116 were variable sites and 31 were parsimony-informative. Three clades were recovered (Figure 7). ML and BI analysis revealed the P. pungens strains of this study fell into clade I with a strongly supported bootstrap value (88), and there was a p-distance value of 0.0004 between them and the other sequences of clade I. Clade I was closely related to clade II (mean p-distance value: 0.0143), while the highest p-distance was observed with clade III (mean p-distance value: 0.0319). The highest p-distance among P. pungens clades were observed between clade II and clade III (mean p-distance value: 0.0356).

Toxin Content
None of the tested strains by LC-MS/MS produced DA in detectable amounts. The LOD varied between 0.09 and 0.02 fg cell −1 .

Discussion
In this study, a significant morphological variability in Pseudo-nitzschia pungens populations from the northern Adriatic Sea was highlighted by analyzing a consistent number of samples from both field and cultured material. ML and BI analysis revealed the P. pungens strains of this study fell into clade I with a strongly supported bootstrap value (88), and there was a p-distance value of 0.0004 between them and the other sequences of clade I. Clade I was closely related to clade II (mean p-distance value: 0.0143), while the highest p-distance was observed with clade III (mean p-distance value: 0.0319). The highest p-distance among P. pungens clades were observed between clade II and clade III (mean p-distance value: 0.0356).

Toxin Content
None of the tested strains by LC-MS/MS produced DA in detectable amounts. The LOD varied between 0.09 and 0.02 fg cell −1 .

Discussion
In this study, a significant morphological variability in Pseudo-nitzschia pungens populations from the northern Adriatic Sea was highlighted by analyzing a consistent number of samples from both field and cultured material.
The molecular characterization revealed that the Adriatic population belonged to the clade I, as the P. pungens nominal variety [24]. However, a number of morphological and ultrastructural details differed from the original description. In the earliest description of P. pungens, an overlap of one-third of the cells in colony (or more) and a transapical axis (TA) wider 3 µm has been reported [47], as well as for all the other species belonging to the seriata group. Nevertheless, the P. pungens from this study revealed an overlap (from one-third to one-sixth) and a TA (rarely reaching 3 µm), strongly differing from the original description of the species, but in agreement with what previously reported for several P. pungens strains, irrespective of clade/variety (Table 1). Indeed, the increasing studies focusing on this species have shown that a number of morphological features, previously indicated as key characters, would be not strictly respected. Some authors have already reported an overlap much lower (ranging from one-fourth to one-sixth) in Canada, the Bay of Fundy [53], the northeastern Adriatic Sea [36], the North Sea [54], the Gulf of Mexico [56], the Danish coastal waters [48], and the Atlantic coast of Portugal (P. pungens var. aveirensis) [27], and a TA often less than 3 µm (down to 1.9), especially in the Mediterranean coast of Greece [29], the northeastern Adriatic Sea [36], and the Atlantic coast of Portugal (P. pungens var. aveirensis) [27] (Table 1).
Although striae are generally reported as biseriate, an incomplete third row of poroids was very common. Such third row of poroids has been detected in other strains included in all other clades/varieties [24,27,47,48]. On the contrary, the presence of uniseriate striae was reported for the first time in this study.
All the main morphological data of the Adriatic P. pungens matched with those of the other clades/varieties, except for (i) the density of band striae in 10 µm (12-23) that slightly diverges from that reported in Pacific coast of USA (California) and in Atlantic coast of Portugal (for P. pungens var. cingulata and P. pungens var. aveirensis, 20-24 and 21-25, respectively) [26,27], (ii) the density of fibulae in 10 µm (5-18) that slightly diverges from that reported in Danish coastal waters (10-20) [48], and (iii) the density of poroids in 1 µm (1-4) that slightly diverges from that reported in the Pacific coast of USA (Washington State) (4-5) [51] (Table 1).
In this study, the cell asymmetry of P. pungens in valve view has been highlighted for the first time, although it could be noticed looking at the TEM micrographs of P. pungens var. aveirensis ([27], Figure 47).
Several ultrastructural features have been indicated as useful to discriminate among P. pungens varieties, such as the shape and pattern of poroids in the cingular bands ( Figure S11) [24,27]. While the nominal variety (P. pungens var. pungens) has three cingular bands, all with one row of oval to rectangular poroids [22], P. pungens var. cingulata is characterized by different cingular bands, i.e., the valvocopula has square poroids having two rows of 2-3 hymen sectors, while the second band has rectangular poroids characterized by 1-2 hymen sectors; no descriptions have been reported for the third band [26]. A further poroids' pattern was described by Churro et al. [27] who established the variety P. pungens var. aveirensis, having two types of cingular bands, valvocopula with square poroids split into two to three parts and the second band with one row of oval (sometimes split) poroids ( [27], Figure 51 and Figure 51 insert).
The morphology of the poroids in cingular bands of the Adriatic P. pungens was characterized by a wide variability among strains and within strains and among field samples, showing a number of different combinations of ultrastructural details previously used to discriminate the P. pungens varieties described so far. In fact, cingular bands could have square, circular, or rectangular poroids without or with 1-2 hymen sectors, with several combination of them, even within the same cingular band. As a consequence, this high variability in ultrastructural detail patterns makes such details uninformative for discriminating the N Adriatic P. pungens (belonging to clade I, as the P. pungens nominal variety [24]) from P. pungens var. aveirensis. On the contrary, N Adriatic P. pungens clearly differs from P. pungens var. cingulata.
Poroids' dimensions in the cingular bands of P. pungens var. cingulata and P. pungens var. aveirensis showed a decreasing trend in the abvalvar direction (i.e., poroids in the third cingular band were narrower than those in the first two cingular bands) [26,27], as often observed in this study for P. pungens from the Adriatic Sea. However, differently from the other varieties, in our samples, it was not rare to observe that, in the dimensions of poroids, the first two cingular bands did not differ, so much that valvocopula and the second cingular band were not always easily discernable. In these cases, the second cingular band could be distinguished because of the slightly higher number of band striae in 10 µm.
Until now, the P. pungens varieties have been approximatively ascribed to the three clades: P. pungens var. pungens~clade I, P. pungens var. cingulata~clade II, and P. pungens var. aveirensis~clade III [24,26,27]. Nevertheless, results of this study suggest that a clade does not necessary correspond to a morphological variety and vice versa. Indeed, Adriatic P. pungens clade I showed a wide morphological variability, covering at least two varieties (i.e., P. pungens var. pungens and P. pungens var aveirensis).
Field and experimental studies showed that diatom frustules can be significantly modified by environmental conditions such that genetically identical individuals could be identified as different species [58,59]. For example, salinity and temperature, among other conditions, have strong effects on frustule morphology, clearly demonstrating the flexibility in diatom morphogenesis [58,60,61]. Some of the morphological features, defined as key characteristics for diatom taxonomic identification, have been shown to be more variable than previously thought [62,63]. In this regard, morphological variability should be investigated under different environmental conditions and in the highest possible number of individuals in order to cover the entire morphological variability within the same population.
The wide morphological and morphometrical variability observed in the P. pungens clade I population from Adriatic Sea, often overlapping characteristics proper of different varieties, could be explained by taking in account the great number of observations that were performed in this study compared with the previous ones (Table 1). Moreover, the analyses were conducted under a wide spectrum of conditions (i.e., from field and cultured samples sampled in different periods and with different strains of different ages). Nevertheless, this wide variability was detected also between different specimens belonging to the same sample, suggesting that this variability was intrinsic and only partially ascribable to the different environmental conditions. Finally, although some strains of P. pungens clade I were recorded to be toxigenic [29], none of the cultured strains of P. pungens clade I from this study produced DA in detectable amounts, in accordance with what previously observed in NW Adriatic strains [30] and with the results from the official monitoring of shellfish production sites [64] in the NW Adriatic Sea, that only sporadically revealed the presence of DA in shellfish and at very low levels [7,40].

Study Area and Sampling
The study area is the coastal station SG01 (43 • 45.86 N, 13 • 13.00 E) of the Senigallia-Susak transect located in the southern part of the northern Adriatic subbasin, 1.2 nM from the Italian coastline (bottom depth: 12 m), and is included in the LTER (Long-Term Ecological Research) Italian sites, where phytoplankton and environmental parameters have been sampled since 1988.
Sampling was carried out with at about a monthly frequency, from January 2018 to December 2019. Water samples were collected at surface by Niskin bottles, in 250 mL dark glass bottles and preserved by adding 0.8% formaldehyde prefiltered and neutralized with hexamethylenetetramine [65], and stored at 4 • C until analysis was performed. Moreover, net (20 µm mesh) samples were collected for cell isolation (see below).

Pseudo-Nitzschia Strain Isolation
The isolation of single cells of Pseudo-nitzschia pungens was carried out in 24-well plates following the capillary pipette method [66]. Cultures were maintained at 21 • C with a 12:12 h of light:dark photoperiod and an irradiance of 100 µmol m −2 s −1 , in sterile filtered seawater enriched with f/2 nutrients [67]. Every month, the algal cultures were checked for their purity and quality and refreshed with fresh culture medium. A total of 14 Pseudo-nitzschia pungens strains were set up.

DNA Extraction from Algal Culture PCR Amplification and Sequencing
Of the total 14 strains set up, 13 were used for the molecular analyses. Algal cultures were harvested during their late exponential phase and centrifuged at 4000× g for 15 min in order to obtain the pellets. Pellets were extracted using CTAB (N-cetyl-N,N,N-trimethylammoniumbromide) buffer (2% CTAB, 1 M Tris pH 8.0, 0.5 M EDTA pH 8.0, 5M NaCl, 1%) modified from Doyle and Doyle [68].
Extracted DNA was amplified by Polymerase Chain Reaction (PCR) technique, carried out with a SimpliAmp TM Thermal Cycler.
PCR products were visualized with UV from an agarose gel (1%). The PCR conditions for LSU and ITS region were 94 • C for 4 min, followed by 35 cycles of 94 • C for 30 sec, annealing at 60 and 58 • C (for LSU and ITS regions, respectively) for 45 sec, and elongation at 72 • C for 1 min, followed by further elongation at 72 • C for 5 min.

Sequence Analyses
Taxonomic assignation was performed by blasting each LSU sequences against the GenBank database (NCBI on-line BLAST web interface version 2.9.0+ [71]) to determine the closest known sequences.
Sequences were adjusted for the presence of double peaks by eye with BioEdit [72]. Among the ITS sequences from this study, 4 were aligned with 90 sequences retrieved from GenBank. Pseudo-nitzschia multiseries was the outgroup sequences (Table S2). The selection of outgroup sequences was based on the findings by Lim et al. [21].
Alignments were made with ClustalW [73] using the default setting and were then edited manually. Regions that did not fit with the others were excluded from the phylogenetic analyses. Two independent analyses were used to conduct the ITS1-5.8s-IT2 phylogeny: Maximum Likelihood (ML) and Bayesian Inference (BI). The best nucleotide substitution model was tested with Partitionfinder 2 [74]. The generalized time-reversible evolution model (GTR+G) was used for the construction of the RAxML phylogenetic analysis, and Kimura's two-parameter model (K80 + I) was used for the Bayesian inference tree. ML analyses were carried out with RAxML [75] 1000 pseudo replicates through Cipress portal [76].
Bayesian analyses were carried out using MrBayes 3.2 [77] with 3,000,000 Markov chain and Monte Carlo generations, a sample frequency of 1500, and a diagnosing frequency of 1000. The 50% majority rule consensus tree was constructed discarding the first 25% of samples. Posterior probabilities were calculated to measure tree strength.
The distance estimation matrix between groups was calculated with the p-distance method using the default setting of MEGA 7 [78].

Light Microscopy Analyses
Pseudo-nitzschia cells were measured at 1000× magnification using an inverted microscope (ZEISS Axiovert 135) equipped with phase contrast. The Apical Axis (AA) and the overlapping region of the cells in a chain were measured in at least 100 cells from cultured strains and field samples.

Ultrastructural Characterization (TEM and SEM)
Samples for TEM and SEM analyses were harvested from cultured strains in exponential growth phase, collected during 2018-2019 (Table S3) and from field net samples collected in February and May 2019.
Samples were acid-cleaned following von Stosch's protocol [18]. A drop (2 µL) of the cleaned material was placed on a grid and on a stub and observed with a Philips TEM 400 microscope and a SEM (FE-SEM; Zeiss Supra 40, Carl Zeiss AG, Oberkochen, Germany), respectively.
Several cells were measured (see Table 1 for the number of cells used for each measurement) both from cultured and field samples for Transapical Axis (TA), fibulae, striae, and poroids' density in both valves and cingular bands with particular focus to valvocopula.
A measurement of valval symmetry was performed on SEM micrographs, calculating the cell surface with an image analysis software and using the formula as follows: Cells in valval view were divided into two hemivalves by the apical axis, crossing the half of the transapical axis. Then, the symmetry was expressed as the ratio of the two hemivalve areas (minor hemivalve:major hemivalve, Figure S12). Valves were asymmetric when the ratio 1.

Chemicals and Standards
The acetonitrile (MeACN) and formic acid (FA) were of LC-MS grade, and the methanol (MeOH) was of HPLC grade. Water was distilled and passed through a MilliQ water purification system (DIW) (Millipore Ltd., Bedford, MA, USA).
Certified reference material for DA, CRM-DA-g (103.3 µg mL −1 ), was purchased from the Institute of Biotoxin Metrology at the National Research Council of Canada (NRCC, Halifax, Nova Scotia, Canada). Calibration solutions of DA were prepared from serial dilutions of the reference material in DIW.

DA Extraction
Chemical analysis of Pseudo-nitzschia pungens needs a large quantity of cells, so each strain was grown in an increasing volume up to 2 L to achieve abundances, ranging from 17 × 10 4 to 61 × 10 4 cells mL −1 among the cultured strains.
The strains were grown in the same culture conditions reported above. Cells were harvested from the early stationary growth phase. Algal pellets of 4 P. pungens strains (Table S3) were extracted using a mixture of MeOH/H 2 O (50:50 v/v), following the official EU-RL RP-LC-UV method (EURLMB 2008), for the determination of DA in shellfish and finfish.
All culture volume (2 L) was centrifuged for 20 min at 2500× g (4 • C) in 40 centrifuge tubes (50 mL volume). Pellets were combined and extracted with 5 mL of MeOH/H 2 O (50:50 v/v), vortex-mixed for 1 min, and bath-sonicated for 10 min. After sonication, the aliquot was centrifuged for 10 min at 2500× g (4 • C), and the supernatant was transferred to a 100 mL evaporation flask. Pellet extraction was repeated three times, and the supernatants were combined and evaporated to dryness. The residue was reconstituted in 1 mL of MeOH/H 2 O (50:50 v/v) and filtered through a 0.2 µm syringe filter (Minisart, Sartorius, Germany) for LC-MS/MS analysis.

LC-MS/MS Analysis
LC-MS/MS analyses were performed using a hybrid triple-quadrupole/linear ion trap 3200 QTRAP mass spectrometer (AB Sciex, Darmstadt, Germany) equipped with a Turbo V source and an electrospray ionization (ESI) probe. The mass spectrometer was coupled to an Agilent model 1200 LC instrument (Palo Alto, CA, USA), which included a solvent reservoir, inline degasser, quaternary pump, refrigerated autosampler, and column oven.
The method was implemented following the conditions described by Mafra et al. [46], which were properly modified. LC separation was performed using a Gemini ®® NX-C18 column (2 mm × 100 mm, 3 µm particle size; Phenomenex, Torrance, CA, USA), set at 40 • C, with a flowrate of 0.4 mL min −1 . Mobile phase A was DIW and B MeACN, both containing 0.2% of FA. Gradient elution was adopted, as described below: From 10% to 20% B in 5 min, from 20% to 35% B in 1 min, then hold for 6 min, return to the original conditions at 13 min, and hold for 7 min before the next injection.
Infusion experiments were performed using CRM-DA-g to set the turbo IonSpray source parameters as follows: Nebulizer Gas (GS1) 50  Type II: one hymen sectors. Type III: two hymen sectors. Type IV: one partially divided hymen sector, Figure S11: Drawing representing the ultrastructure of poroids in the cingular bands of the three P. pungens varieties: (A) var. pungens [22]; (B) var. cingulata [26]; (C) var. aveirensis [27]. VC: valvocopula; II: second cingular band; III: third cingular band. No information about the third cingular band is available in var. cingulata and var. aveirensis, Figure S12: Pseudo-nitzschia pungens cell divided into two hemivalves. Green area: minor hemivalve; red area: major hemivalve, Table S1: Morphometric characteristics of Pseudo-nitzschia pungens strains from the coastal site SG1 of LTER Senigallia transect. n.r., not reported, Table S2: List of sequences retrieved from Genbank for the construction of the Bayesian consensus tree. Strains from this study are in bold. (c) strains of which PCR products were cloned, Table S3: List of Pseudo-nitzschia pungens strains from the coastal site SG1 of