Temporal and spatial variability of the potentially toxic Pseudo-nitzschia spp. in a eutrophic estuary (Sea of Marmara)

Abstract

Spatial and temporal variability and bloom formation of the potentially toxic diatom Pseudo-nitzschia spp. was investigated weekly to monthly from October 2009 to October 2010 in a eutrophic estuary, the Golden Horn. Pseudo-nitzschia spp. were detected in 195 of 512 samples (38%) collected throughout the year. Two species, P. calliantha and P. pungens, were identified based on the SEM examination. Blooms of Pseudo-nitzschia occurred in the lower and middle estuary in January and May. The bloom in January mainly comprised P. calliantha. In the bloom in early May, P. calliantha made up 72% of the Pseudo-nitzschia cells and P. pungens 28%. However, the contribution of P. pungens increased to 83% in late May. The Pseudo-nitzschia blooms occurred at low temperature (9–15°C) and moderate salinity (17–18), and for P. calliantha a significant negative correlation was found with temperature and a significant positive correlation with salinity. The percentage of Pseudo-nitzschia cells decreased gradually from lower to upper estuary (59–14%), correlating with a decrease in Secchi depth (5.5–0.5 m). Principal components analyses (PCA) were used to explore the spatial and temporal variability of environmental factors in relation to Pseudo-nitzschia abundances, and showed that NH₄, pH, Secchi depth and DO values were the most important factors reflecting spatial differences, while temperature, salinity, Chl-a and Si:N were more important factors showing temporal differences. High abundances of P. pungens correlated mainly with pH, Secchi depth and DO values, whereas P. calliantha also correlated with NO₃ + NO₂. Low light availability due to high concentrations of suspended material and very variable environmental conditions (e.g. pH, DO and NH₄) may have limited growth of Pseudo-nitzschia in the upper estuary. Regular monitoring of Pseudo-nitzschia is important for improving the understanding of the influence of environmental parameters on bloom dynamics in the study area.

INTRODUCTION

Estuaries are known as highly productive ecosystems, being often nutrient-rich and having multiple sources of organic carbon to sustain populations of heterotrophic organisms. Phytoplankton production may, however, be very low in river-dominated estuaries with high turbidity caused by river inputs of suspended particulate matter (SPM) and/or resuspension of bottom sediments. High SPM concentrations in estuaries result in rapid attenuation of light in the water column and phytoplankton photosynthesis is thus confined to a shallow photic zone. As a consequence, phytoplankton dynamics (including productivity and spatial and temporal changes in biomass) will be largely controlled by light availability (Cloern, 1987). Peterson & Festa (1984) concluded that estuarine productivity becomes strongly depressed as SPM concentration increases from 10 to 100 mg l⁻¹.

Under certain circumstances, some phytoplankton species can form high-biomass and/or toxic blooms, thereby causing harm to aquatic ecosystems (Kudela et al., 2015). Marine diatoms of the genus Pseudo-nitzschia H. Peragallo have been confirmed as producers of domoic acid (DA), the toxin responsible for amnesic shellfish poisoning (ASP) (Trainer et al., 2012). Domoic acid may enter the food chain via zooplankton or filter-feeding shellfish and subsequently accumulate in, for example, marine invertebrates, birds and mammals, and harmful effects have been documented in marine birds, mammals and humans (Trainer et al., 2012; Lelong et al., 2012). Presently, at least 16 Pseudo-nitzschia species and two Nitzschia species have been found to produce DA (Smida et al., 2014; Teng et al., 2014).

Pseudo-nitzschia species have been commonly observed in studies exploring phytoplankton distribution in Turkish coastal waters (Koray, 1995; Eker & Kıdeyş, 2000; Polat et al., 2000; Türkoğlu & Koray, 2002; Balkis, 2003; Deniz & Tas, 2009; Tas et al., 2009; Tas & Okus, 2011, 2014). Harmful algal blooms, including Pseudo-nitzschia, have been reported in Turkish coastal waters (Türkoğlu & Koray, 2002; Koray, 2004). Blooms of P. delicatissima (April) and P. pungens (late July to mid-August) were observed in Turkish coastal waters of the southern Black Sea (Türkoğlu & Koray, 2002). No negative impacts have been reported during the blooms, but no shellfish or fish farms presently exist in the area. The Pseudo-nitzschia species mentioned above have seldom been identified to species level, and not using state of the art methods like electron microscopy or molecular methods, except for P. calliantha Lundholm, Moestrup & Hasle which has been recorded from the Black Sea (Bargu et al., 2002). Pseudo-nitzschia calliantha is a toxin-producing species (Lundholm et al., 2003) which has been found widely from the Baltic Sea to the Adriatic Sea and Black Sea, indicating a cosmopolitan distribution (Lundholm et al., 2003). In a previous study carried out in the Golden Horn Estuary (GHE), P. cf. delicatissima and P. cf. pungens were commonly observed (Tas et al., 2009), but these identifications were based on light microscopy only.

Pseudo-nitzschia blooms have been associated with environmental conditions such as upwelling (Trainer et al., 2000, 2002), coastal runoff or discharges (Smith et al., 1990), low Si:N (Sommer, 1994) or submarine groundwater discharge (Liefer et al., 2009), all indicating that Pseudo-nitzschia thrive at nutrient repletion, but that the nutrient ratios are also important (Trainer et al., 2012). Anthropogenic eutrophication has in the last decades affected the ecosystem of GHE considerably. The effects of eutrophication on the GHE ecosystem including the phytoplankton should be considered to evaluate the ecosystem change.

The unplanned urbanization, increase in settlements and industrial facilities around the GHE since the 1950s have caused a high degree of pollution, particularly from wastewaters of pharmaceutical, detergent, dye and leather industries and domestic discharges. By the early 1990s, estuarine life was limited to the lower estuary, and the upper estuary had almost no eukaryotic phytoplankton due to anoxia and heavy sedimentation. The GHE became a severely polluted environment, where the water column was characterized by frequent anoxic episodes (Kıratlı & Balkıs, 2001). In 1997, the Rehabilitation Project in the GHE was initiated in order to improve water quality by reducing nutrient loading. For this purpose, surface discharges were gradually taken under control, and 4.25 × 10⁶ m³ of anoxic sediment was removed from the completely filled upper estuary, resulting in an increase of at least 5 m in depth in this region, and a bridge floating on pontoons was partially opened in order to provide possibilities for surface water circulation. These changes resulted in rapid renewal of water and oxygenation of the anoxic sediment of the highly eutrophic upper estuary (Yüksek et al., 2006).

The present study aimed to investigate the spatial and temporal variability as well as bloom formation of potentially toxic Pseudo-nitzschia spp. in relation to environmental parameters and to understand better which factors are important for the presence of these organisms in the GHE.

MATERIALS AND METHODS

Location

The study area (GHE) is located south-west of the Strait of Istanbul extending in a north-west–south-east direction and is ~7.5 km long and 700 m wide, with a surface area of 2.6 km² (Figure 1). The study area is divided in three parts based on the hydrographic structure: lower (LE), middle (ME) and upper estuary (UE). The LE is the deepest section (40 m) and it is strongly influenced by interaction with the Strait of Istanbul (Bosphorus). The depth rapidly decreases to 14 m in the ME, where a bridge operating on buoys limits the upper layer circulation between the LE and the UE. The UE has a depth of 4 m due to a high degree of sedimentation (Figure 1). Although the two streams (Alibey and Kağıthane) carry fresh water to the estuary, the amount of the flow decreased remarkably by the end of 1990s following the construction of a series of dams. Therefore, the main source of the fresh water flowing into the GHE is rainfall (Sur et al., 2002). The lower part of the GHE is characterized by a two-layered stratification similar to the neighbouring Strait of Istanbul, whose upper layer has a salinity of ~18 originating from the Black Sea and lower layer, with a salinity of ~38, originating from the Mediterranean Sea (Özsoy et al., 1988).

Fig. 1. The study area and sampling stations (TH1 to TH5). Insert shows the position of the GHE.

Seawater sampling and analysis

Seawater samples were collected monthly (October 2009 to March 2010) or weekly (April 2010 to September 2010) at five sampling stations (Figure 1) from 0.5, 2.5 and 5 m depth using 5 l Niskin bottles. The parameters salinity, temperature, dissolved oxygen (DO) and pH were measured using a multi-parameter probe (YSI Incorporated Professional Pro Plus), and light transparency was measured using a Secchi disc. Chl-a analyses were carried out by an acetone extraction method according to Parsons et al. (1984). Inorganic nutrient (NO₃ + NO₂, NH₄, PO₄, SiO₂) analyses were measured using a Bran + Luebbe AA3 auto-analyser according to standard methods (APHA, 1999). All data were afterwards used as the average of values from 0.5, 2.5 and 5 metres.

Water samples (250 ml) for phytoplankton counts were taken from the Niskin bottle samples, and preserved with acidic Lugol's solution (2%). Aliquots of 20 ml were left to settle overnight in Utermöhl sedimentation chambers (Utermöhl, 1958). The cell counts were performed at 200 × magnification using a Leica DM IL LED inverted microscope equipped with phase contrast optics. At least 300 phytoplankton cells (>20 µm) or two or more transects in the settling chamber were counted. Measurements of cell dimensions, and observations of cell shape and chain formation were mainly done using a Leica DM 2500 LM light microscope at needed magnifications. The percentage of Pseudo-nitzschia cells of total phytoplankton abundance was calculated. All data were used as the average of 0.5, 2.5 and 5 m.

A total of 64 net samples for qualitative analyses of phytoplankton were collected using a Nansen plankton net (0.57 m diameter, 55 µm mesh size) towed vertically from 10 m to the surface of the LE and ME and preserved with borax-buffered formalin (4%). All net samples were collected simultaneously with the water samples.

Species identification

For routine observations of Pseudo-nitzschia species, a Leica DM 2500 LM light microscope with brightfield optics was used. For ultrastructural identification of Pseudo-nitzschia species, organic material was removed using an acid-wash treatment prior to SEM observations as described by Bargu et al. (2002). A total of 10 net samples corresponding to the bloom periods in January and May were used for species identification. Samples were concentrated onto 0.2 µm pore size isopore polycarbonate membrane filters (Millipore). Salt was removed from samples by rinsing with deionized water under low vacuum (150 mm Hg). To remove organic material, saturated KMnO4 was added until the filters were covered and the samples were allowed to digest for 15 min. Twelve M HCl (3 ml) was then added to the samples and held for a total of 30 min to complete the oxidation process. Samples were then vacuumed gently and rinsed with deionized water. This process was repeated twice. The filters were air-dried in a desiccator for 24 h and then mounted onto SEM stubs with double-sided tape and sputter coated with gold palladium. All micrographs were taken with a field-emission SEM (FEI-QUANTA FEG 450) at an accelerating voltage of 10 kV. Species identification of phytoplankton was based on Cupp (1943), Hendey (1964), Drebes (1974), Delgado & Fortuna (1991) and Tomas (1997). Pseudo-nitzschia species were morphologically identified based on SEM examination of the frustules (Lundholm et al., 2012). At least 20 valves were studied in detail in each sample. In SEM, the morphometric parameters measured were: width and length of valves, densities of interstriae, fibulae and poroids, and structure of the poroids.

Data analysis

Principal components analyses (PCA) were used to explore the spatial and temporal variability of environmental factors to relate Pseudo-nitzschia abundances to environmental variability. Prior to all PCAs, environmental data were transformed to fourth root to reduce the heterogeneity in the data and to normalize the data matrix using Primer v6 program. The relationships among abundance of Pseudo-nitzschia and environmental parameters were analysed by Spearman rank correlation following transformations to natural logarithms by using PASW v18 program.

RESULTS

Morphology of Pseudo-nitzschia cells

Ultrastructural examination by SEM revealed that the bloom-forming Pseudo-nitzschia species obtained from field material in January was mainly P. calliantha, while in May both P. calliantha and P. pungens were observed. The cells of P. calliantha were linear in valve view and overlapping in colonies. Apical axis of valves ranged from 45 to 50 µm, transapical axis of valves was between 1.5 and 2.0 µm. Overlap of cells in chains were short (about one-ninth of cell length) (Figures 2 & 3). The fibulae were regularly spaced, 18–20 in 10 µm, and interrupted by a central nodule in the middle of the cell. The valves had 36–40 interstriae in 10 µm. The striae comprised one row of poroids, 5–6 poroids in 1 µm. Each poroid was divided in 4–10 sectors (Figure 3).

Fig. 2. Light micrographs of P. calliantha (A, B) and P. pungens (C, D) overlapping cells in colonies (Scale bars: 10 µm).

Fig. 3. Scanning electron micrographs of P. calliantha (A–D) and P. pungens (E–G) from the GHE. A large part of the valve (A), the tip of the valve (B), parts of the valve showing fibulae, interstriae and poroid structures (C) and a central nodule of P. calliantha (D). Whole valve (E), parts of the valve showing the ultrastructure (F) and overlapping valves of P. pungens (G). Scale bars: B, C, D and F: 1 µ m; A, G: 5 µm; E: 10 µm.

The cells of P. pungens were linear-lanceolate, and symmetrical about the apical axis. Overlap of cells in chains was considerable, close to one fourth of cell length (Figure 2). The apical axis ranged from 90 to 100 µm, and the transapical axis was 3.5–5.5 µm. The fibulae were regularly spaced with a density of 13–14 in 10 µm. The density of interstriae was 13–14 in 10 µm. The striae contained two rows of poroids, 4–5 poroids in 1 µm (Figure 3).

Cell abundance of Pseudo-nitzschia species

The major groups of phytoplankton in the study area were diatoms, dinoflagellates and other flagellates (silicoflagellates, chrysophytes, raphidophytes, cryptophytes, euglenophytes, prasinophytes and chlorophytes), constituting 54, 35.5 and 10.5%, respectively, of the total number of phytoplankton species. The most abundant species were the diatoms Pseudo-nitzschia spp., Skeletonema marinoi, Ceratoneis closterium (=Cylindrotheca closterium) and Thalassiosira sp., the dinoflagellates Heterocapsa triquetra, Scrippsiella trochoidea and Prorocentrum cordatum (=Prorocentrum minimum); the raphidophyte Heterosigma akashiwo, the euglenophytes Euglena viridis and Eutreptiella marina, the cryptophyte Plagioselmis prolonga and the prasinophyte Pyramimonas grossii.

Pseudo-nitzschia cells were detected in 195 of 512 water samples (38%) analysed. It was evident that Pseudo-nitzschia spp. were absent or occurred in low abundances in the UE, as seen in the abundance distribution over the entire study period (Figure 4). The frequency of Pseudo-nitzschia spp. in all water samples decreased considerably from LE (found in 59% of the examined samples) towards UE (14%), correlating with a decreasing Secchi depth (Figure 4).

Fig. 4. The frequency of Pseudo-nitzschia occurrence and mean Secchi depths throughout the study area.

Two Pseudo-nitzschia species, P. calliantha and P. pungens, were detected in the GHE during the study period. The two species were common in the LE and ME particularly during winter and spring. The abundance of P. calliantha started to increase in December and reached bloom densities (average 1.2 × 10⁶ cells l⁻¹) in January in the LE, and also high densities in the ME (~9 × 10⁵ cells l⁻¹), whereas P. pungens was not observed in January (Figure 5). Pseudo-nitzschia calliantha made up 99% of the total phytoplankton in January in both the LE and ME (Figure 6). Only very low concentrations (20 × 10³ cells l⁻¹) of Pseudo-nitzschia were observed in the UE in January. Pseudo-nitzschia spp. were almost absent in February and March, however in April, P. calliantha was detected again in the LE in sub-bloom densities (average 5.4 × 10⁵ cells l⁻¹) (Figures 5 & 6). Concentrations were slightly lower in the ME and very low in the UE (Figure 5).

Fig. 5. The abundance of Pseudo-nitzschia calliantha and P. pungens throughout the study area.

Fig. 6. The contribution of Pseudo-nitzschia abundance to the total phytoplankton throughout the study area.

The second Pseudo-nitzschia bloom occurred in the ME during the first two weeks of May where cell densities reached an average of 1.1 × 10⁶ cells l⁻¹. During the first week of this bloom, P. calliantha made up 72 and P. pungens 28% of the cell counts. An evident increase in the contribution of P. pungens took place during the third and fourth week of May, with P. pungens contributing with 40 and 83%, respectively, of the total Pseudo-nitzschia cell counts, with densities reaching 5.8 × 10⁵ cells l⁻¹ in late May (Figure 5). More or less the same appeared in May in LE although at lower cell densities and with a lower contribution of P. pungens. Pseudo-nitzschia cells were frequently observed until early June. The cell abundances decreased clearly from ME to UE and Pseudo-nitzschia spp. were observed in much lower densities in the UE than in the other parts of the estuary (Figure 5). During the winter bloom, cell densities were highest in the LE, while during the spring bloom they were highest in the ME (Figure 5).

The average annual contribution of diatoms to the total phytoplankton abundance was 81% in the LE and it decreased gradually towards the ME (65%) and was 37.7% in the UE (not illustrated). Similarly, the average annual contribution of Pseudo-nitzschia cells to the total phytoplankton abundance was on average higher in the LE (18.1%) than in the ME (11.6%) and it decreased markedly in the UE (0.4%) (Figure 6).

Abundance of Pseudo-nitzschia in relation to physical variables

Fluctuations in temperature, salinity and Secchi depth measured throughout the sampling period and the relationships between Pseudo-nitzschia densities in the three parts of the estuary are shown in Figure 7. Temperature showed a clear seasonal pattern with minor differences between the three parts of the estuary. Overall the temperature values ranged from 5.9°C (February) to 28.4°C (August) during the study period. The UE had a slightly relatively higher temperature (~2.2°C) than the LE. The mean annual temperatures and standard deviations were 17.11°C ± 5.73 for LE and 19.32°C ± 5.90 for UE.

Fig. 7. Relationships between Pseudo-nitzschia abundances and physico-chemical variables during the study period.

Pseudo-nitzschia species were observed at a wide range of water temperatures from 9.3°C (January) to 23.5°C (August) in this study area, but the peaks in cell density appeared at a more narrow temperature between 9.3°C (January) and 15.5°C (early June). The mean abundance of Pseudo-nitzschia cells was 140 × 10³ cells l⁻¹ in December at 11°C, with 70% of the total density composed of P. calliantha and 30% P. pungens. The first bloom of Pseudo-nitzschia, comprising only P. calliantha, occurred at 9.3°C in January. In early April, the sub-bloom density of P. calliantha occurred at 9.8°C. The second bloom of Pseudo-nitzschia spp. in the first half of May happened at 14–15°C. Even though P. calliantha was found over a broad range of temperatures, it was more abundant at a temperature range between 9 and 15°C. There was a significant negative correlation (P < 0.001) between Pseudo-nitzschia spp. abundance and temperature (Table 1). This also agrees with Pseudo-nitzschia species being found only in low abundances during the warmer months, July–October, where the temperatures were between 19.5 and 23°C (Figure 7).

Table 1. Spearman correlation coefficients (rho) calculated between Pseudo-nitzschia spp. and environmental factors.

Statistically significant correlations are indicated by symbols: *P < 0.05; ‡P < 0.01; †P < 0.001 (N = 96).

The surface salinity varied between 2.5 (December, UE) and 21.6 (May, LE) throughout the study period. Exceptionally, salinity decreased to less than 1.0 in the UE following a heavy rainfall in early June. Pseudo-nitzschia spp. were found at salinities from 11.6 (TH5, January) to 21.6 (TH1, May), but the higher densities were found at salinities from 15.7 to 18.7. Salinity generally differed between the LE and the UE. The UE had a variable salinity (15.15 ± 3.05) which was always lower than the LE, which had relatively stable salinity (17.83 ± 1.08). Bloom densities of P. calliantha were detected at salinities of 18 and 19. There was a highly significant positive correlation (P < 0.001) between Pseudo-nitzschia abundance and salinity (Table 1, Figure 7), with P. calliantha in total making up the majority of the Pseudo-nitzschia abundance (average 80.2%) during the whole year.

Precipitation was the major freshwater input for the GHE. A high amount of suspended particulate material (SPM) carried by streams causes very low water transparency especially in the UE. Moreover, sludge dredging activities in the UE in combination with insufficient water circulations result in increased turbidity even in periods without rainfall. Water transparency based on Secchi depth decreased considerably during the rainy periods. It ranged from 0.1m (February, UE) to 10m (August, LE) and it was always much lower in the UE than the LE (Figure 7). The mean Secchi depths were 5.51 ± 2.34, 1.59 ± 0.22, 0.54 ± 0.22 in the LE, ME, UE, respectively. Spearman correlation coefficients showed a significant positive relationship (P < 0.01) between Pseudo-nitzschia spp. abundance and Secchi depth (Figure 4).

Abundance of Pseudo-nitzschia in relation to chemical variables

The mean annual pH values were 8.2 ± 0.26 in the LE and 7.5 ± 0.34 in the UE and pH generally decreased gradually from the LE to the UE. The minimum pH values in the surface were 6.3 in the UE in winter, while the maximum pH values were measured to 8.5 in the LE in spring. The fluctuations in pH showed a spatial and temporal variation (Figure 8). The pH values were more variable in the UE than ME and LE. In April and May, pH increased up to 8.5 in the LE and 8.1 in the UE. Even though the Pseudo-nitzschia spp. cells were observed at a wide range of pH values, the high cell densities were found at pH values over 8.0 and lower densities at the pH values below 8.0 particularly in the UE. There was a highly significant positive correlation (P < 0.001) between Pseudo-nitzschia spp. abundance and pH (Table 1).

Fig. 8. PCA ordinations of environmental variables expressed as stations (A) and months (B) and superimposed abundances (cells l⁻¹) of Pseudo-nitzschia calliantha (C) and P. pungens (D).

The mean annual dissolved oxygen DO concentrations were 8.1 ± 1.75 mg l⁻¹ for LE and 3.0 ± 2.57l⁻¹ for UE, and DO were generally very low (<3 mg l⁻¹) in the UE except during bloom periods, while they were consistently high in the LE due to the strong hydrodynamic structure and interaction with the Strait of Istanbul. Although the UE has low DO values, the increase in DO values in May is based on phytoplankton abundance, and algal production is one of the main factors affecting DO level in the UE (Figure 8). Spearman correlation coefficients showed a highly significant positive relationship (P < 0.001) between Pseudo-nitzschia abundance and DO values (Table 1).

Chl-a concentrations increased considerably from the LE to the UE and ranged between 0.7 and 10.6 µg l⁻¹ in the LE, 0.5 and 35.4 µg l⁻¹ in the ME and 0.6 and 121.4 µg l⁻¹ in the UE. Chl-a values showed two peaks in spring and summer when flagellate blooms occurred. Chl-a measured as 2.8 µg l⁻¹ in the LE in January when Pseudo-nitzschia dominated the phytoplankton with abundances reaching 1.2 × 10⁶ cells l⁻¹. In May, chl-a measured as 7.3 µg l⁻¹ in the ME when the blooms of Pseudo-nitzschia and cryptophyte Plagioselmis prolonga occurred. No clear correlation was observed between blooms of Pseudo-nitzschia spp. and the major chl-a peaks observed in spring and summer (Figure 7). At these periods flagellates such as euglenophyceans and cryptophyceans dominated.

Inorganic nutrient concentrations increased in winter due to the high amounts of terrestrial inputs from the streams and due to precipitation, while they decreased in spring because of phytoplankton activity. All nutrient values increased remarkably from lower to upper estuary (Figure 7). Mean concentrations of nutrients were 3.3 ± 5.2 µM at LE and 6.4 ± 12.4 µM at UE for NO₃ + NO₂ (NO_×); 0.39 ± 0.36 µM at LE and 3.5 ± 3.4 µM at UE for PO₄; 6.2 ± 7.0 µM at LE and 37.1 ± 47.2 µM at UE for SiO₂. Silicate concentrations decreased to 1.8 µM during the bloom in May, while the mean annual value was 17.7 µM. There were significantly negative correlations between Pseudo-nitzschia abundance and PO₄ (P < 0.001) and SiO₂ concentration (P < 0.01). No relationship was found between Pseudo-nitzschia abundance and NO_x concentrations (Table 1, Figure 7). The mean annual N:P ratio based on DIN (NO₃ + NO₂-N and NH₄-N) and DIP (PO₄-P) varied between 26.2 ± 20.5 for the LE and 35.3 ± 35.6 for the UE, increasing towards the UE, and it was always higher than the Redfield ratio.

Data analysis

PCA analyses showed a significant spatial and temporal variation in environmental variables during the study period. The first two PCs explained 61.7% of the total variation (PC1: 41.2%, PC2: 20.5%). Projection of stations on the PCA ordination (Figure 8A) showed an upper to lower estuary separation of stations in the ordination plane along a diagonal transect, indicating that the highest variation between stations was in TH4 and TH5 (upper estuary) and that the lowest variation was in TH1 (lower estuary). Projection of sampling periods (months) on the PCA ordination (Figure 8B) provided a seasonal separation in the ordination plane along a diagonal transect, indicating the lowest variation between April and May, while the highest variation was observed in January and February depending on the effect of precipitation. Considering coefficients in the linear combinations of variables, it is seen that NH₄, pH, Secchi depth and DO (in the direction of grey arrow) have the greatest importance on the first PC (Figure 8A), while temperature, salinity, Chl-a and Si:N (in the direction of grey arrow) have most importance on the second PC (Figure 8B). The results of PCAs showed that the main factors causing spatial variation between stations were NH₄, Secchi depth, pH and DO values, indicating a very variable environmental condition in the upper estuary. According to PCAs, the factors causing temporal variation among months were temperature, salinity, Chl-a and Si:N. PCA analyses showed also a significant spatial and temporal variation in Pseudo-nitzschia abundances. The first two PCs explained 57.2% of the total variation (PC1: 37.6%, PC2: 19.6%). Projection of P. calliantha and P. pungens on the PCA ordination (Figure 8C & D) showed that the main factors causing spatial and temporal variation in both species were Secchi depth, DO, pH and NO_x. Notably P. pungens was clearly correlated with Secchi depth, DO and pH. These factors thus are important for the distribution of two Pseudo-nitzschia species. As can be seen in PCA plots, P. calliantha was more abundant at TH1 and TH2 in January and May, while P. pungens was more abundant at TH1 and TH2 particularly in May.

DISCUSSION

Domoic acid (DA) producing Pseudo-nitzschia species were commonly detected in the GHE, particularly two species, P. calliantha and P. pungens, which were found forming two blooms. Although Pseudo-nitzschia species were observed at a wide range of water temperatures from 9.3°C (January) to 23.5°C (August), peaks in abundance appeared between 9.3°C (January) and 15.5°C (early June) and abundances were very low, rarely exceeding 10³ cells l⁻¹, in the summer months. A significant negative correlation between Pseudo-nitzschia spp. abundance and temperature indicates that low temperatures in nutrient-rich coastal areas and estuaries like the GHE may stimulate growth and bloom formation in different Pseudo-nitzschia species. Prior studies have shown a wide temperature range for presence of Pseudo-nitzschia spp. (Dortch et al., 1997; Liefer et al., 2009), consistent with the results of the present study. Pseudo-nitzschia spp. abundance also decreased considerably at high temperature (>20°C), which apparently limits the growth of Pseudo-nitzschia spp. in this area. Pseudo-nitzschia calliantha was found at temperatures between 9.3 and 21.7°C with maximum abundance at 9.3°C in January and at 14.7°C in May, whereas P. pungens was found at temperatures between 11.6°C and 22.8°C with maximum abundance at 16°C in late May. This is in agreement with several other studies with regard to P. pungens (Stonik et al., 2001) and P. calliantha (Terenko & Terenko, 2012), but does not agree with reports of P. pungens showing a peak in abundance at a water temperature of 21.0°C (Terenko & Terenko, 2012). But P. pungens comprises several varieties, and the temperature preference of these varieties have been found to differ (Kim et al., 2015), which may explain this disagreement. The contribution of P. calliantha to total phytoplankton was higher in January than in late May decreasing from 99% to 17%. During the same period, the contribution of P. pungens to total phytoplankton increased to 83%. This might reveal that these species shows different temperature responses as also indicated by Fryxell et al. (1990).

Pseudo-nitzschia spp. were found at a salinity range of 11.6 to 21.6, but the higher densities were detected at 15.7 to 18.7. Pseudo-nitzschia spp. were almost not observed during periods of low salinity (<10). A highly significant positive correlation between Pseudo-nitzschia abundance and salinity indicates that low salinities below 15 may not be suitable for growth and bloom formation of Pseudo-nitzschia spp. The same is seen in the Baltic Sea, where Pseudo-nitzschia decreases in abundance with a decrease in salinity (Hällfors, 2004). In general, Pseudo-nitzschia spp. demonstrate euryhaline characteristics, growing over a wide range of salinities (1 to >30) but they occur more frequently at higher rather than lower salinities (Thessen et al., 2005). This general statement is supported by the present study, which found that Pseudo-nitzschia spp. was found to thrive in brackish waters, but preferred the higher salinities. Our results are in agreement with laboratory studies of P. calliantha and P. pungens. Salinity preference experiments on P. calliantha found growth at salinities from 5–>25 with a growth optimum >25 (Lundholm et al., 1997), and P. pungens showed positive growth at salinities from 15 to >26, with optima at 20 or 26 (Cho et al., 2001; Doan-Nhu et al., 2008). On the contrary, other species like P. americana or P. circumpora showed no growth at salinities lower than 15 or 20, respectively (Miller & Kamykowski, 1986; Lim et al., 2012).

Studies on harmful algal blooms carried out in the GHE showed that water discolouration with different types of colour depending on the causative species, decreasing light intensity and rapid changes in DO concentrations had major effects on the ecosystem (Tas and Okus, 2011; Tas, 2015; Tas & Yilmaz, 2015). Before the rehabilitation of the GHE, it was reported that high amounts of suspended solid matter, toxic gases and anoxic conditions limited phytoplankton growth and Pseudo-nitzschia species were observed at very low densities (~10³ cells l⁻¹) in the LE and ME (Tas & Okus, 2003). During the rehabilitation of the GHE, the abundance of Pseudo-nitzschia cells was very low (~10³ cells l⁻¹) and found in the LE and ME only. After the rehabilitation, Pseudo-nitzschia density increased considerably based on the improving water quality and cell densities exceeded 10⁶ cells l⁻¹ in the LE in January 2002. However, Pseudo-nitzschia cells were not observed in the UE which had a high turbidity (Tas et al., 2009). The previous studies carried out in the GHE (Aslan-Yilmaz et al., 2004; Yüksek et al., 2006; Tas et al., 2009) revealed that the most important problem for this area is the high concentration of terrestrial materials entering the ecosystem with rainwater and the streams, thereby affecting physico-chemical parameters and the phytoplankton community structure particularly in the upper estuary. In the present study, there was a significant positive relationship (P < 0.01) between Pseudo-nitzschia abundance and Secchi depth, and blooms of Pseudo-nitzschia spp. were not observed in the UE. Low light availability due to high amount of SPMs at the surface of the UE most likely inhibits the photosynthesis and limits the spatial distribution and growth of Pseudo-nitzschia spp. Thus, light limitation should be considered one of the most important factors determining growth of Pseudo-nitzschia in the UE.

It is well-known that high abundances of Pseudo-nitzschia are found in coastal areas enriched by nutrients (Parsons et al., 2002; Lundholm et al., 2004; Trainer et al., 2012). High Pseudo-nitzschia abundances have also been associated with submarine ground discharge of nutrients (Liefer et al., 2009) and their blooms can be stimulated by nutrients from several sources, including upwelling or mixing events and riverine inputs (Trainer et al., 2000, 2012), suggesting a response to eutrophication (Parsons et al., 2002). These findings agree with our observations of high nutrient concentrations supporting growth and blooms of Pseudo-nitzschia in the GHE. The increasing nutrient concentrations from the LE to the UE are the effects of two streams carrying high nutrient levels into GHE at the UE. Low silicate concentrations during summer due to lack of precipitation may have limited the growth of Pseudo-nitzschia cells.

The effect of pH on the growth of marine phytoplankton has been studied (Taraldsvik & Myklestad, 2000; Hansen, 2002; Lundholm et al., 2004; Havskum & Hansen, 2006) and it has been suggested that high pH (8.7 to 9.1) inhibits growth of most Pseudo-nitzschia species (Lundholm et al., 2004). Such high pH levels were never observed in the present study, where pH levels from 6.3 to 8.5 were found.

Multivariate analyses (PCA) performed on environmental factors reflected the deviating environmental conditions found in the upper estuary, where the abundance of Pseudo-nitzschia spp. was very low. The present study showed that the spatial variability was particularly structured along Secchi depth, pH, DO and NH₄, while the temporal variability was structured along temperature, salinity, Chl-a and Si:N. Our results agree with Tas & Yilmaz (2015), which suggested that a projection of stations and environmental variables provided an upper to lower estuary separation and that the variability was particularly structured along Secchi depth and temperature. Secchi depth, DO and pH were the main factors causing spatial and temporal variation in Pseudo-nitzschia and these factors have a significant positive relationship with Pseudo-nitzschia abundances as stated in Spearman correlation coefficients. Light limitation and the very variable environmental conditions caused by several factors (e.g. Secchi depth, DO, pH and NH₄) particularly in the upper estuary may be the most important limiting factors for growth of Pseudo-nitzschia in the study area. The presence of the potentially toxic Pseudo-nitzschia blooms in GHE pose a risk for toxin accumulation in local fauna with associated problems for the ecosystem as well as for human consumers of marine products from GHE. As both P. calliantha and P. pungens have been found to produce toxins in other parts of the worlds (Bates et al., 1989; Smith et al., 1990; Bargu et al., 2002; Hallegraeff, 2002; Besiktepe et al., 2008; Moestrup et al., 2009; Lundholm, 2011), they may, due to the high cell concentrations, pose a potential risk for toxic events at LE and ME. Future studies are needed to explore the toxic potential of the local strains at different environmental conditions. The cell concentrations found at the GHE suggest that accumulation of domoic acid in, for example, shellfish is a realistic scenario to consider, but as toxin production is affected by environmental and biological parameters in various ways and probably depends on the genetic potential of the local strains, a complete understanding of the potential risk requires monitoring of Pseudo-nitzschia, domoic acid, the ecosystem and the chemical-physical parameters.

In conclusion, available environmental conditions (i.e. high nutrient, low water circulations) in the GHE promote the potential toxic Pseudo-nitzschia blooms in the region. Therefore, rehabilitation efforts are very important to provide a healthy ecosystem in the region by reducing nutrient inputs, terrestrial runoffs and increasing upper layer circulation.