Marine phytoplankton generates approximately half of the global primary production and affects the fate of sea life, with respect to fishery yields and sustainable marine ecosystems (Boyce et al., 2010). Domestic and industrial discharges cause nutrient enrichment and consequently, alterations of phytoplankton composition in terms of biodiversity and abundance in many coastal areas (Smith et al., 1999). Annual nitrogen and phosphorus input to ecosystems has more than doubled globally during the last fifty years (Falkowski et al., 2000; Smil, 2000). Since the availability of nutrients, necessary for growth and reproduction of phytoplankton, is often lower than the biological demand, additional nutrient supply is necessary for phytoplankton growth (Corbett, 2010). Therefore, in addition to common sources (e.g., terrestrial inflows, airborne particulates) anthropogenic discharges (e.g., industrial and domestic effluents, agricultural runoffs) and sediment resuspensions affect nutrient enrichment, leading to an increase in phytoplankton abundance. Under favorable oceanographic conditions (e.g., temperature and salinity), sufficient nutrient levels may cause excessive phytoplankton reproduction and harmful algal blooms may occur (Heisler et al., 2008). These enrichments mostly happen in coasts near urbanized and industrialized settlements (Ferreira et al., 2011; Heisler et al., 2008). However, biogeochemical structure and environmental pressures may vary for a coastal marine ecosystem and exact mechanisms of algal bloom occurrences remain unknown, as in the Izmit Bay.

The Izmit Bay, one of the most polluted marine ecosystems of Turkey, contains a wide range of contaminants including organic and inorganic compounds such as heavy metals, radionuclides, POPs, and nutrients (Balkis et al., 2007; Ergül et al., 2013a, b; Karademir et al., 2013; Tolun et al., 2006). The bay is located to the northeast of the Marmara Sea (Turkey) and a permanent pycnocline and a two-layered stratification are observed throughout year. Upper layer salinity is generally less than 24.0, and this reaches 36.0 below the halocline layer (Ergül, 2016). There are around 2 million inhabitants living in Kocaeli Province surrounding the Izmit Bay. Population and industrial activities have been increasing rapidly since the 1980s, and the bay receives a considerable amount of domestic and industrial discharge from its drainage basin. Since the 1960s, more than 400 large industrial plants have been built around the bay, including the largest shipyards, metallurgy, paper mill, fertilizer, and petrochemical facilities of Turkey. In addition to 70 shipyards, there are 40 ports located along the coast of the bay. In 2015, the bay received approximately 400 and 80 tonnes of total nitrogen (TN) and phosphorus (TP) discharges from wastewater treatment plants, respectively. Moreover, almost all effluents were discharged without treatment before 2000, and 41.0% of domestic and industrial discharge has remained untreated until 2010 (ECR, 2013; ISU, 2016). At the current time, there are no available data for untreated TN and TP discharges from other sources (i.e., untreated sewage discharges, agricultural runoffs, etc.) in the Izmit Bay. Eutrophication is also a serious problem (Morkoç et al., 2001; Okay et al., 2001) and phytoplankton blooms were recently reported in the bay (Ergül et al., 2014, 2010).

Phytoplankton blooms, causing brownish-red surface water with bad odor and mucilage formation, drew public attention to domestic and industrial discharges as potential triggering elements. Moreover, meteorological factors likely have an effect on algal bloom occurrence. Therefore, investigation into mechanisms of algal blooms and their harmful effects on the ecosystem will be crucial for evaluating effective environmental factors and solving future problems in an enclosed marine ecosystem, close to a populated and industrialized city, such as the Izmit Bay.

This study evaluated (1) serial phytoplankton blooms with mucilage formation, (2) accompanied physicochemical and meteorological conditions in 2015, (3) evaluation of possible reasons for harmful algal blooms in the Izmit Bay, and (4) a comparison between the present results and remarkable diatom and dinoflagellate blooms around nearby Mediterranean seas.

On March 19, 2015, the Izmit Bay which is located at the eastern end of the Marmara Sea, was covered by phytoplankton and a light red appearance was observed over almost all the bay’s surface (Fig. 2a). This appearance was disappeared after approximately two days. On April 12, phytoplankton abundance, which was dominated by Prorocentrum micans Ehrenberg, 1834, increased, surface water partly reached a brownish-red appearance, and mucilage formation was observed (Fig. 2b).

Six different phytoplankton species (Achnantes brevipes C. Agardh, 1824, Cerataulina pelagica (Cleve) Hendey, 1937, Dictyocha speculum Ehrenberg, 1834, Melosira sp., Navicula sp., and Prorocentrum micans), which belong to three different groups (four Bacillariophyceae, one Dictyochophyceae, and one Dinophyceae) were observed intermittently in the samples on April 12, April 30 and May 3, 2015 with average surface water temperatures of (14±1.0)°C, and salinity of 21±3.0. Regarding abundance, dinoflagellates were dominant with rates of over 99.0% and P. micans, which is known as a potential high biomass-forming dinoflagellate (Jeong et al., 2005) was the most abundant species in those three blooms (average 1.10×10⁶, 5.40×10⁶ and 17.9×10⁶ ind./L, respectively; Table 1, Fig. 3a). Mucilage formations occurred in these blooms and a milky light-brown flocculated layer was observed on the sea surface. After the blooms, red appearances and mucilage layers disappeared within a few days from surface waters. In the microscopy slides, mucus aggregates were also observed amongst P. micans clusters (Fig. 2c). Therefore, formation of mucilage should be excreted by P. micans after their excessive proliferation (Figs 2c, d). On May 3, the mucilage layer was very intense, and dead amphipods, which were coated with amorphous mucus pellets, were observed in the samples in Karamürsel ((72±12) ind./L, Fig. 2d) with the highest surface water turbidity ((287±46.0) NTU) and Chl a concentrations ((180±15.0) μg/L, Fig. 4b). Although several phytoplankton blooms have been reported from the Izmit Bay since 1986 (Taş et al., 2016), to our knowledge, this incident was the most intense. Moreover, the presence of dead amphipods after a phytoplankton bloom was reported for the first time in the present study. Mortis causa of the amphipods was not clear, and there are no (open) scientific literature regarding marine mucilage-dependent amphipod mortality. However, Bruno et al. (1989) reported that farmed and wild fish may also be killed by the smothering of gills due to phytoplankton mucus production. Therefore, given its appearance under stereomicroscope, it is reasonable that mucilage aggregates probably blocked amphipod gills and thus respiration was prevented, leading to their death (Fig. 2d). Also, oxygen concentration decrease may be considered as another reason of amphipod mortality. However we have no simultaneous dissolved oxygen data to prove this claim. Amphipod and copepods are known predators of dinoflagellates including bloom causative species (i.e., P. micans), therefore high amphipod abundance in the samples are most likely connected to their feeding behavior (Yi et al., 2017). It should be noted that phytoplankton abundance as well as nutrient concentrations were approximately three fold higher near the shore because of conglomeration after drifting via waves and those values were excluded from further assessment in the present study.

In the bloom conditions on May 3, 2015, nutrient levels remarkably increased, and raised inorganic nitrogen, and excessive o-PO₄^3– concentrations were measured (Figs 3c, d). On that day, most intense dinoflagellate bloom occurred concurrently with the highest concentrations of DIN and o-PO₄^3– ever recorded during the study. Therefore, the existence of elevated levels of nutrients should have triggered excessive dinoflagellate (i.e., P. micans) proliferation and caused excretion (i.e., mucilage formation) from P. micans cells, under the favorable meteorological conditions. It should be noted that, very calm weather and light winds were recorded for two days immediately before the bloom on May 3. Therefore, increased nutrients, which are most likely derived from undesirable or illegal discharges, could not be dispersed by winds, currents, precipitations, etc. from the surface water and finally were consumed by phytoplankton. In fact, the Izmit Bay’s renewal capacity is known to be insufficient because of a relatively long residence time of the water masses (Morkoç et al., 2001) and in this incident, light winds, and accordingly weak currents, allowed nutrients accumulation in surface waters. Thus, appropriate conditions occurred for excessive P. micans proliferation.

Following phytoplankton increases, which occurred on May 14, July 11 and November 6, 2015 and unlike other blooms, biodiversity reached 10, 14, and 10 species, respectively, during these incidents in the Izmit Bay. During the study, 17 taxa which have not bloom formation were also recorded in different classes (Bacillariophyceae, Dinophyceae, Euglenophyceae, and Oligotrichea) in addition to six bloom taxa (Table 1, Fig. 3a).

On May 14, both DIN and o-PO₄^3– concentration decreased compared to May 3 and in this event biodiversity reached ten species (Figs 3a, c, d). The dominant species at this time was Noctiluca scintillans (Macartney) Kofoid & Swezy, 1921 (Figs 3a, d) with surface water temperatures of 15.0°C and salinity of 25.9 in the Hereke station, which was the highest salinity level recorded during the study (Fig. 4). Corresponding to individual number, Bacillariophyceae (diatom) was dominant in the second order at a rate of 16.6%, following Dinophyceae (dinoflagellate) (75.5%). Euglenophyceae and Oligotrichea species were also observed in this red tide with rates of 7.70% and <1.00%, respectively. On that day, although the lowest phytoplankton abundance was found (54.0×10³ ind./L), the highest phytoplankton biomass ((268±26.0) mg/L) was recorded, because of the presence of N. scintillans (34.0×10³ ind./L), which is known for its large cell volume (Figs 3a, b). During the red tide, the N. scintillans bloom was also reported from the Marmara Sea on May 17 (NASA, 2015). This organism is not known as toxin producer, but it is, however, able to accumulate toxic levels of ammonia and can thus cause massive mortality to marine organisms because of ammonia excretion into the ambient waters (Faust and Gulledge, 2002; Okaichi and Nishio, 1976). In fact, the highest NH₃ concentration in second order was measured on May 14; however, its concentration was 12 fold higher than the sum of NO₂^–-N and NO₃^–-N and was the highest rate during the study (Fig. 3c). Therefore, it is reasonable that NO₃^–-N and NO₂^–-N were indirectly used by N. scintillans, and elevated levels of NH₃ were connected with their metabolic activities during the bloom. In addition, phytoplankton abundances and ammonia concentrations remarkably decreased with increased depth in the same station, considering the relationship between decrease in N. scintillas abundance and accordingly ammonia excretion (Fig. 3f). In previous studies, N. scintillans increase were reported following a P. micans bloom in late spring in the Izmit Bay as well as from the Marmara Sea (Aktan et al., 2005; Artüz and Baykut, 1987; Ergül et al., 2014) and those results coincide with the present study.

After late spring, N. scintillans was observed in other samplings; however, the following red tides were dominated by P. micans in 2015. On July 11, phytoplankton abundance increased again (average 480×10³ ind./L) and a light red appearance was observed on the sea surface without mucilage formation, while surface water temperatures of 25.0°C and salinity of 21.4 were recorded at Marina station in the Izmit Bay. With respect to individual count, Dinophyceae was the dominant group, with rates of 86.8%. Bacillariophyceae and Oligotrichea species were also observed with rates of 12.9% and <1.00%, respectively. In this red tide, the highest phytoplankton diversity, with 14 species (eight Dinophyceae, four Bacillariophyceae and two Oligotrichea) were observed simultaneously with the second highest DIN:P ratio as 0.5 (Fig. 3a). Thus, different phytoplankton species found an opportunity to proliferate. Although the highest DIN:P ratio was determined on May 14, as was discussed above, this rate was related to the elevated level of ammonia, which possibly derived from N. scintillans excretion. Regarding the increased diatom abundance, the lowest silica concentration ((0.22±0.01) mg/L) was determined on July 11, because of the silica usage in the cell wall of diatoms. On that day, one of the most common diatom species, Skletonema costatum reached 60.0×10³ ind./L which was the highest abundance determined among diatom species throughout the study.

After July 2015, a significant phytoplankton increase was not observed until late autumn in the Izmit Bay. On November 6, a new bloom which was dominated by P. micans was observed again, without mucilage formation. Although the incident remained limited and almost no red appearance was observed in the central and western basins, phytoplankton abundance reached 1.10×10⁶ ind./L and its diversity reached ten species (eight Dinophyceae and two Bacillariophyceae) in the Marina station located in the eastern basin of the bay. Interestingly, this bloom occurred a few days after a relatively long term strong wind blew. During that time, northeastern gusts, with an average speed of 35 km/h (ranged from 33 to 41 km/h) blew around the Izmit Bay for 5 d between 28 October and 2 November (Fig. 4a-right panel). Over the following days, the wind speed gradually decreased, and on November 6, the last phytoplankton bloom was observed in 2015, when average wind speed was 12 km/h, surface water temperatures were 15°C, and salinity was 22.0 (Fig. 4a). It is well known that the Izmit Bay has been exposed to treated and/or untreated wastewater discharges since the 1960s. Hence, an elevated level of organic and inorganic contaminants including nutrients accumulated in the surface sediment of the bay, is present (Aktan et al., 2005; Ergül et al., 2013a; Karademir et al., 2013; Morkoç et al., 2001). Beside the unknown amount of land based effluents from several sources (e.g., agricultural runoffs, untreated anthropogenic wastewater, underground water), ~400 t total nitrogen and ~100 t total phosphorus were routinely discharged from wastewater treatment plants located around the Izmit Bay in 2015 (Fig. 5). The northern and southern sides of the bay are surrounded by hills, while the eastern and western edges are open from either side. Due to this geographical structure, a corridor occurs in the east-west direction. Therefore, following the ~35 km/h northeastern gusts for 5 d, wind induced sediment resuspensions occurred in the east, the shallowest basin of the Izmit Bay, and ambient water was contaminated with nutrients that were used by phytoplankton for their reproduction. Unlike other cases, NO₃^–-N concentration ((0.040±0.003) mg/L) was slightly higher than the NH₃ level ((0.030±0.005) mg/L) in this bloom. This high rate of nitrate may be related to oxidation of resuspended ammonia after nitrification processes in the surface sediment (Fig. 3c). In terms of numbers of individuals, dinoflagellates were dominant with rates of over 99.0%, as in the first three blooms. Therefore, on November 6, 2015, silica content was remained relatively high ((0.76±0.02) mg/L) due to low usage by diatoms (Fig. 3d).

Since DIN:Si ratios ranged between 0.08 and 0.56 (i.e., <1), and DIN:P ratios ranged between 0.03 to 3.11 (i.e., below 16:1) in all samplings, nitrogen was the limiting factor during bloom conditions. It is known that N broadly limits phytoplankton growth in coastal marine ecosystems (Nixon, 1986) and N limitations have been reported from the Izmit Bay during bloom conditions (Aktan and Dede, 2008; Aktan et al., 2005), and other parts of the Marmara Sea (Turkoglu, 2013, 2016; Balkis et al., 2010) as in the present study.

According to previous studies (Morkoç et al., 2001) and Kocaeli Metropolitan Municipality records (ISU, 2016), discharges of total suspended matter (TSM), biological oxygen demand (BOD), total nitrogen (TN) and total phosphorus (TP) into the Izmit Bay gradually decreased 40, 40, 14 and 20 fold, respectively, since 1984, owing to established wastewater treatment plants (Fig. 5). Currently, it is estimated that more than 80.0% of discharge from the wastewater treatment plants has domestic origin, whereas the remaining discharge is from industrial sources. Nevertheless, the present results show that, despite precautions in place to decrease nutrient inputs via these plants, algal blooms still occur, with harmful effects including mucilage presence and invertebrate deaths.

The trophic index (TRIX), which was developed by Vollenweider et al. (1998) to determine the trophic state and quality of coastal marine ecosystems, ranged from 6.1 to 10 (6.9, 10, 6.3, 6.1, and 6.8 for April 30, May 3, May 14, July 11, and November 6, 2015, respectively). Therefore, based on TRIX values, the eutrophication status of the Izmit Bay was classified as degraded and there was a very high trophic level during the bloom conditions. It should be noted that the eutrophic state of the surface waters of the Izmit Bay fluctuated during the study and in some periods reached good quality and low trophic levels. On the other hand, because of the low current regime and the relatively long residence time of the water masses in the Izmit Bay (Morkoç et al., 2001), nutrient accumulation over certain periods of the year caused deterioration of the trophic state. Also, consecutive high TRIX values may be an important sign for future environmental problems for the bay. In fact, significant correlations between TRIX and total phytoplankton abundance (TPpA) (r=0.98, p<0.01, n=5) indicate that contents of the water encouraged the bloom conditions and following that the water quality of the bay was highly affected by the blooms. Positive significant correlations between o-PO₄^3– and TPpA (r=0.96, p<0.01, n=6) and between NO₃^–-N and TPpA (r=0.86, p<0.05, n=6) revealed that variations of total phytoplankton density were linked to the nutrient concentrations. Also, a positive significant correlation between total phytoplankton biomass and the DIN:P ratio (r=0.98, p<0.01, n=6) indicates nitrogen limitation. Consequently, despite the improvement work over the last few decades by the municipality of the city around the Izmit Bay, based on adopted legislation in the frame of European Union directives, harmful algal blooms still occur with the abovementioned harmful effects.

In fact, the marine eutrophication issue and phytoplankton blooms, encountered in coastal marine ecosystems of Mediterranean seas, has been addressed in a number of studies since the 1960s (Karydis and Kitsiou, 2012). Through regulation and prevention, based on the Barcelona convention and the directives regarding the European Union policy on eutrophication the Water Framework Directive 2000/60/EC (EC, 2000), Mediterranean countries have modeled their legislation to reduce nutrient inputs to the marine ecosystems (Karydis and Kitsiou, 2012; Saliba, 1995). Nevertheless, despite the precautions, numerous intense phytoplankton blooms have been observed in Mediterranean coastal waters. Hence, we have listed recent intense dinoflagellate and diatom blooms (i.e., >10⁶ ind./L) with temperature, salinity and nutrient data in the adjacent seas (i.e., the Marmara and Black Seas) and near Mediterranean coastal waters (i.e., the Aegean, Ionian and Adriatic Seas), in order to evaluate recent cases since 2000 and to compare them with the present results from the Izmit Bay (Table 2).

Temporal distribution of intense diatom and dinoflagellate blooms spread predominantly over three seasons: spring, summer and winter, in the Mediterranean coastal surface waters since 2000. In terms of frequency, the bloom conditions occurred mostly in spring (five diatom and five dinoflagellate blooms) followed by summer (four diatom and two dinoflagellate blooms) and winter (two diatom and three dinoflagellate blooms), whereas intense autumn blooms were recorded less frequently (two diatom and one dinoflagellate blooms). Typically, dinoflagellate blooms occurred in spring and early summer, while diatom blooms occured in winter and early spring. However, in some cases, dinoflagellate blooms were reported in winter (Scrippsiella. trochoidea (Stein) Loeblich III, 1976, and Prorocentrum redfeldii Bursa, 1959 from Trabzon Shore and Thermaikos Bay, respectively) whereas diatom blooms were reported in summer seasons (Pseudo-nitzschia Pungens (Grunow ex Cleve) G. R. Hasle, 1993, Thalassionema nitzschioides Schiller in Hustedt, 1930, Leptocylindrus minimus Gran, 1915 and Chaetoceros vixvisibilis Schiller in Hustedt, 1930 from the Çanakkale Stratit, Constanta Bay, Kastela Bay and Rovinj coasts, respectively; Table 2).

Phytoplankton blooms may occur in a wide range of temperatures and salinity (8.00–27.0°C and 11.0–36.0, respectively) and species-specific characters seem to be determined in bloom formations. Nutrient values also had diversified levels (<0.01–0.56, 0.01–0.55, 0.01–42.0 and 0.12–3.42 for NO₂^–-N + NO₃^–-N, NH₃, PO₄^3– and SiO₂ or SiO₄, respectively) and in almost all blooms, the ratios of nitrogen to phosphorus were remarkably different from the classical Redfield ratio (i.e., 16:1; Table 2). Consequently, these data suggest that a combination of nutrients is more important than its levels for diatom and dinoflagellate bloom formations, and termination of proliferation is determined by limiting nutrients. In addition, the blooms are predominantly influenced by temperature and salinity while sufficient light intensity and favorable amounts of nutrients exist in the water.

The spatial distribution of these blooms revealed that dense incidents mainly arose in the coastal waters of the northern parts of the Mediterranean. Although bloom cases have been reported, the abundance of diatoms and dinoflagellates did not exceed 10⁶ ind./L in the Levantine Sea. Extreme proliferations of both groups were reported from the Black, Marmara, Aegean and Ionian Seas. However, despite the remarkable dinoflagellate abundance, incidents that exceeded 10⁶ ind./L were reported merely for diatoms from the Adriatic Sea since 2000 (Table 2). Because the northern parts of the Mediterranean are richer in nutrients than the southern parts (Ignatiades, 2005), denser phytoplankton blooms are expected in the northern regions. In fact, the densest dinoflagellate bloom (70×10⁶ ind./L) was reported from the Golden Horn Estuary, adjacent to the Istanbul Strait as well as the Black and Marmara Seas, in summer 2001 (Taş and Okuş, 2011). Besides being a semi-enclosed basin, the Golden Horn Estuary was known for its polluted water with foul odor because of untreated industrial and domestic discharges. However, the estuary was rehabilitated in the mid-2000s and bloom conditions have not been reported since 2010. Another dense bloom (40×10⁶ ind./L) due to proliferation of diatoms was recorded in Pesaro coast in the north Adriatic in autumn 2000. Beside local land-based sources, the Po River was considered as the responsible nutrient carrier for this bloom (Penna et al., 2004). Like abovementioned blooms, most cases occurred under the land based anthropogenic inputs around the northern Mediterranean. However, legislations, based on EU directives, were set up to prevent further deterioration, and the decreasing bloom densities can be considered possible improvements in many Mediterranean coasts.

A comparison with blooms from nearby coastal surface waters of the Mediterranean seas showed that dinoflagellate blooms in the Izmit Bay were the densest consecutive blooms ever reported in recent years (Table 2). Like other incidents, domestic and industrial inputs were effective in those blooms. Additionally, unlike other incidents, wind-induced sediment resuspension caused a dinoflagellate bloom in the Izmit Bay, and this case indicates that accumulated contaminants in the surface sediment can be a potential nutrient source. This represents long term effects of untreated discharges on an elongated coastal marine ecosystem.

Beside low-speed water mass movement and routinely discharged nutrients via wastewater treatment plants from a populated city, various reasons might have contributed to the occurrence of the algal bloom conditions in the Izmit Bay, such as the inadequacy of advanced treatment technology, untreated and/or illegal domestic and industrial discharges, agricultural runoffs, and wind-induced resuspensions. Therefore, the results of the present study suggest that there is a need for increased action to prevent harmful algal blooms, in accordance with the environmental deterioration in the Izmit Bay. This can be achived by improvements to wastewater treatment technology, including nutrient removal capability, prevention of untreated discharges, illegal bilge bailing and controlling agricultural fertilizer usage. Finally, the removal of the upper layer of the eastern basin’s surface sediment should be considered to prevent undesirable effects of wind induced resuspensions.