INTRODUCTION
Seasonal changes and global warming considerably affects the biological structure of seas (Goffart et al., Reference Goffart, Hecq and Legendre2002). Mucilage formation in seas is the aggregation of organic substances that are produced by various marine organisms under special seasonal and trophic conditions (Innamorati et al., Reference Innamorati, Nuccio, Massi, Mori and Melley2001; Mecozzi et al., Reference Mecozzi, Acquistucci, Di Noto, Pietrantonio, Amirici and Cardarilli2001). Having serious effects on such human activities as fisheries, tourism and aquaculture, mucilage formation was reported to occur in the Adriatic Sea since the 17th Century (Totti et al., Reference Totti, Cangini, Ferrari, Kraus, Pompei, Pugnetti, Romagnoli, Vanucci and Socal2005). The correlation between mucilage formation and phytoplankton was demonstrated in the studies carried out in the late 19th Century and early 20th Century (Mingazzini & Thake, Reference Mingazzini and Thake1995). Rinaldi et al. (Reference Rinaldi, Vollenweider, Montanari, Ferrari and Ghetti1995) stated that diatoms, known to produce extracellular polysaccharide substance, are effective on mucilage formation, and bacteria were reported to participate in this formation (Herndl et al., Reference Herndl, Arrietta and Stoderegger1999; Azam & Long, Reference Azam and Long2001). Subsequent studies showed that dinoflagellates also produce extracellular mucilages (MacKenzie et al., Reference MacKenzie, Sims, Beuzenberg and Gillespie2002) and it was also stated that the dinoflagellate Gonyaulax fragilis (Schütt) Kofoid, 1911 takes part in mucilage formation in the Adriatic Sea (Pompei et al., Reference Pompei, Mazziotti, Guerrini, Cangini, Pigozzi, Benzi, Palamidesi, Boni and Pistocchi2003; Pistocchi et al., Reference Pistocchi, Cangini, Totti, Urbani, Guerrini, Romagnoli, Sist, Palamidesi, Boni and Pompei2005). In addition, mucilage aggregates are the result of a hyperproduction as a pathological response of the algae to a microbial infection and this hypothesis was supported by some laboratory experiments (Innamorati et al., Reference Innamorati, Nuccio, Massi, Mori and Melley2001).
In Turkish territorial waters, mucilage formation began to be observed firstly in İzmit Bay in the Marmara Sea in October 2007 and mainly fisheries and tourism have been damaged seriously (Tüfekçi et al., in press). The Marmara Sea is a small basin with an approximate size of 70 km in width and 250 km in length, a surface area of 11,500 km2 and a maximum depth of 1.390 m. Situated between the European and Asian continents, the Marmara Sea forms the Turkish Straits System along with the Bosphorus and the Dardanelles (Beşiktepe et al., Reference Beşiktepe, Sur, Özsoy, Abdul Latif, Oğuz and Ünlüata1995). It is connected to the Black Sea through the Bosphorus in the north-east and to the Aegean Sea through the Dardanelles in the south-west and, as a result, possesses two different water masses. The surface layer of the Marmara Sea is composed of brackish waters (22–26 ppt) originating from the Black Sea and the bottom layer of Mediterranean originated saline waters (38.5–38.6 ppt) (Ünlüata et al., Reference Ünlüata, Oğuz, Latif, Özsoy and Pratt1990; Tuğrul & Polat, Reference Tuğrul and Polat1995).
In this study, it was aimed to find and demonstrate the species and environmental factors causing mucilage event in the Marmara Sea, and temporal changes in the abundance of total bacteria, pico-, nano- and microphyoplankton were identified.
MATERIALS AND METHODS
Sampling was conducted monthly in a single station determined in the coastal waters of Büyükada Island between January and June 2008 (Figure 1). Samples necessary for water analyses were collected from 7 different depths (0.5, 5, 10, 15, 20, 25 and 30 m) using a water sampler with 3 l capacity, and the temperature of the samples were measured with the thermometer on the water sampler. The salinity was determined by the Mohr–Knudsen method (Ivanoff, Reference Ivanoff1972) and the dissolved oxygen by the Winkler method (Winkler, Reference Winkler1888). The samples necessary for nutrient analyses were put into polyethylene containers of 100 ml, and frozen in a deep freezer at −20°C and measured with an autoanalyser (APHA, 1999).

Fig. 1. Sampling station in the coastal zone of Büyükada Island.
1 l of water, collected for chlorophyll-a analyses used for determining phytoplankton biomass, was filtered through Whatman GF/F filter papers, kept in 90% aceton solution overnight, centrifuged, and then measured with a spectrophotometer at different wave lengths (Parsons et al., Reference Parsons, Maita and Lalli1984). Trophic index (TRIX) values were calculated in order to determine the eutrophication level of the sampling area and the quality of waters (Vollenweider et al., Reference Vollenweider, Giovanardi, Montanari and Rinaldi1998). The index is given by:

Chl-a = chlorophyll-a (µg l−1), D%O = oxygen as an absolute deviation (%) from saturation, N = dissolved inorganic nitrogen N-NO3 + NO2 (µg-at l−1), P = total phosphorus P-PO4 (µg-at l−1). Calculated TRIX values were not compared with other regions since NH4, as dissolved inorganic nitrogen, in the original formula was not used.
In order to determine the qualitative and quantitative situation of phytoplanktonic organisms responsible for mucilage formation, an additional 1 l of water was collected from the aforementioned depths with the water sampler; it was preserved by adding 2.5 ml of Lugol's iodine solution and left for sedimentation in the laboratory for a week. The overlying excess water was siphoned until 15 ml subsamples and then the subsamples were fixed by neutral formaldehyde with a final concentration of 2–4% (Throndsen, Reference Throndsen and Sournia1978). Cell counting was carried out under an inverted phase-contrast microscope (Olympus CK2) in a Sedgwick–Rafter cell.
DAPI (4′,6-diamidino-2-phenylindole) stain was used for determining total bacteria in water samples. 100 µl DAPI stain was added to 900 µl seawater taken from the sample, kept at 28°C for one hour, 3 ml of sterile bidistilled water was added to terminate the reaction and ensure homogenization, stirred and then filtered through polycarbonate filter (Millipore) with 0.2 µm pore diameter. The filter surface was observed under a microscope with epifluorescent attachment (Nikon 80i) and UV-2A filter cube, and photographs were taken from 10 different microscopical fields using 1000X magnification. The signals obtained from the photographs were counted and the mean values were calculated. Then these values were multiplied with their coefficients and the number of bacteria per millilitre was determined (Rodriguez et al., Reference Rodriguez, Phipps, Ishiguro and Ridgway1992).
Single celled organisms in the water samples that show autofluorescent properties were analysed in three groups according to their dimensions. Water samples were filtered through a 20 µm pore size membrane filter to separate microphytoplankton, then through a 2 µm filter for capturing nanophytoplankton and finally through a 0.2 µm filter to detect picophytoplankton. The autofluorescent plankton that were divided by sequential filtration according to their size were analysed with a relevant filter cube and photographs were taken from 10 different microscopical fields using 1000X magnification. The signals obtained from the photographs were counted and the mean values were calculated. Then these values were multiplied with relevant coefficients and the number of autofluorescent microorganisms per millilitre was determined (Totti et al., Reference Totti, Cangini, Ferrari, Kraus, Pompei, Pugnetti, Romagnoli, Vanucci and Socal2005).
The Spearman's rank-correlation coefficient was used to detect any correlation among biotic and abiotic variables (Siegel, Reference Siegel1956). The density of mucilage was assessed under seven categories according to the observations with an underwater camera (1 no mucilage, 2 rare, 3 less dense, 4 dense, 5 filamentous, 6 small masses and 7 big masses).
RESULTS
Environmental variables
During the sampling period, temperature values ranged between 7.0 and 21.5°C. The highest value was observed in June (21.5°C, 0.5 m) and the lowest in January (7.0°C, 0.5 m, 5 m) (Figure 2A). In January, February and March temperature values showed an increasing pattern from the surface to 30 m depth. In April, an irregular temperature differentiation by depth was observed. On the other hand, in May and June generally a decreasing pattern was observed from the surface to 30 m depth.

Fig. 2. Monthly variations of environmental variables: (A) temperature; (B) salinity; (C) dissolved oxygen; (D) nitrogen; (E) phosphorus; (F) silica; (G) N/P ratio; (H) N/Si ratio; (I) Si/P ratio along the water column at the sampling station.
Salinity ranged between 20.9 and 37.4 ppt and showed an increase from the surface to deeper layers in all months (Figure 2B). The highest salinity (37.4 ppt) was recorded at 30 m depth in March and the lowest (20.9 ppt) at 0.5 m depth in June. After 15–20 m depth, salinity showed a sudden increase due to the saline Mediterranean waters.
Dissolved oxygen values ranged from 2.75 mg l−1 to 12.75 mg l−1 (Figure 2C). The highest value was recorded in January in the surface layer that is in contact with the atmosphere and the lowest at 25 m depth in April. A decreasing pattern was observed in oxygen values from the surface to the bottom along the water column.
Of nutrients NO2 + NO3-N values ranged between 0.02 µg-at l−1 and 7.67 µg-at l−1, PO4-P between 0.11 µg-at l−1 and 0.96 µg-at l−1 and SiO4-Si between 0.37 µg-at l−1 and 8.93 µg-at l−1 (Figure 2D–F). The highest nutrient values were recorded generally at the bottom layer where there is aggregation. The N:P ratio ranged between 0.1 and 11.3 (Figure 2G). The minimum value (0.1) was recorded at the depths of 5 and 10 m in January and the maximum (11.3) at 25 m depth in May. The increase of nitrogen in proportion to phosphorus in May caused the N:P ratio to increase more remarkably compared to other months. The N:Si ratio ranged between 0.01 (January, 5 m and 10 m) and 1.10 (January, 20 m) (Figure 2H) while the Si:P ratio ranged between 2.92 (June, 0.5 m) and 52.33 (May, 15 m) (Figure 2I).
During the sampling period, chlorophyll-a amounts ranged between 0.10 µg l−1 and 6.35 µg l−1. Two remarkable peaks were observed, one (6.35 µg l−1) at 15 m depth in January and the other (4.89 µg l−1) at 10 m depth in April (Figure 3A).

Fig. 3. Monthly variations of (A) chlorophyll-a and (B) trophic index along the water column at the sampling station.
Trophic index (TRIX) values ranged between 1.06 (January, 10 m) and 3.30 (January, 20 m) and all values obtained were between 0 and 4 which indicates a state of high quality but low trophic level (Figure 3B).
Pico-, nano- and microphytoplankton and total bacteria
Picophytoplankton abundances during the study periods ranged from 64 ± 3.8 cells ml−1 to 2254 ± 29.5 cells ml−1 at 30 m depth in May and 10 m depth in January, respectively. Especially, at 5–20 m depth in the winter period (January and February), having intense aggregate appearance, picophytoplankton abundances were higher than in any other periods (Figure 4). Nanophytoplankton was at its lowest level (44 ± 1.9 cells ml−1) at 30 m depth in May as in the case of picophytoplankton and its highest density (1562 ± 22 cells ml−1) at 15 m depth in January (Figure 4). During the sampling, the abundance of microphytoplankton ranged between 33 ± 2.4 cells ml−1 (June, 30 m) and 712 ± 18.2 cells ml−1 (January, 15 m) (Figure 4). The highest value of total bacteria (6655 ± 44.4 cells ml−1) was recorded at 25 m depth in April and the lowest (1077 ± 26.1 cells ml−1) at 0.5 m depth in June. During all sampling periods, total bacteria values generally reached the highest density at the middle layer (10–20 m).

Fig. 4. Monthly distribution (cells ml−1) of pico-, nano-, microphytoplankton and total bacteria abundance at the sampling station.
Phytoplankton composition, succession and abundance
During the mucilage formation on the shores of Büyükada Island, a total of 62 species from 5 different microalgae groups were identified: 24 diatoms (38.7%), 34 dinoflagellates (54.9%), two dictyochophyceans (3.2%), one euglenophycean (1.6%) and one prasinophycean (1.6%). During the winter period, with dense mucilage formation, diatoms were dominant over dinoflagellates in terms of the number of species and individuals.
In January, mucilages were observed most densely at 15–20 m depth. In this month, the dominant group in terms of individual number was diatom (80.99–97.79%; Figure 5A), and this group reached the highest individual number at 15 m depth (4.75 × 105 cells l−1, 96.54%). Especially Skeletonema costatum (Greville) Cleve, 1878 played an important role in this increase (1.60 × 105 cells l−1). This species was followed by Cylindrotheca closterium (Ehrenberg) Reimann & Lewin, 1964 (8.86 × 104 cells l−1), Thalassiosira rotula Meunier, 1910 (7.62 × 104 cells l−1) and Pseudo-nitzschia sp. (5.22 × 104 cells l−1). In addition, Thalassiosira anguste-lineata (Schmidt) Fryxell & Hasle, 1977 (2.44 × 104 cells l−1), Dactyliosolen fragilissimus (Bergon) Hassle, 1991 (1.80 × 104 cells l−1) and Leptocylindrus danicus Cleve, 1889 (1.62 × 104 cells l−1) were among the important species observed during mucilage formation at 15 m depth. Dinoflagellates were found more abundantly at the depths of 10 m (1.92 × 104 cells l−1) and 15 m (1.48 × 104 cells l−1). Within this group, the dominance of G. fragilis, a species that produces mucilage, compared to other species is important (maximum abundance at 10 m depth: 1.82 × 104 cells l−1). The highest individual number (4.92 × 105 cells l−1) at 15 m depth was determined according to total phytoplankton values.

Fig. 5. Percentage composition of phytoplankton categories according to abundance in sampling months.
In February, mucilage formation was observed very densely at 15–25 m depth and large masses were formed at these depths. The dominance of diatoms continued (60.37–100%; Figure 5B), and diatoms reached the highest individual number at the surface (1.34 × 105 cells l−1, 86.69%). Cylindrotheca closterium played a significant role in this increase with its individual number of 1.14 × 105 cells l−1. It was followed by Pseudo-nitzschia sp. (1.06 × 104 cells l−1) and T. rotula (6.70 × 103 cells l−1). Besides, C. closterium showed considerable increases at the depths of 5 m (4.02 × 104 cells l−1) and 20 m (4.13 × 104 cells l−1). Dinoflagellates reached the highest individual number (1.84 × 104 cells l−1) at the surface (0.5 m) and were found more densely at the depths of 5 m (1.75 × 104 cells l−1) and 10 m (1.79 × 104 cells l−1) compared to other layers. Within this group G. fragilis (1.58 × 104 cells l−1, 10 m) and Protoperidinium paulseni (Pavillard) Balech, 1974 (1.23 × 104 cells l−1, 0.5 m) showed higher increases in comparison to other species. According to total phytoplankton values, the highest individual number was recorded at the surface layer (1.55 × 105 cells l−1).
In March, while there was no finding about mucilage until 15 m depth, signs were seen very densely at 20–25 m depth, large masses of this depth transformed into filamentous structures at 30 m depth. While dinoflagellates were dominant in terms of species number, diatoms were dominant at some depths (26.97–92.96%; Figure 5C) and dinoflagellates at others (6.74–71.50%; Figure 5C) in terms of individual number. Dinoflagellates reached the highest individual number at 10 m depth (6.60 × 104 cells l−1, 67.84%). Besides, they were dominant at 5 m depth compared to other depths (4.50 × 104 cells l−1, 71.50%). Gymnodinium sp. (2.43 × 104 cells l−1, 10 m), Scrippsiella trochoidea (Stein) Loeblich III, 1976 (2.42 × 104 cells l−1, 5 m; 1.76 × 104 cells l−1, 10 m) and G. fragilis (1.13 × 104 cells l−1, 10 m) played a significant role in this increase in dinoflagellates. On the other hand, diatoms reached the highest individual number at 20 m depth (5.76 × 104 cells l−1, 79.34%) and this depth was followed by 25 m (3.80 × 104 cells l−1) and 10 m (3.02 × 104 cells l−1). Within this group, the dominance of C. closterium at 20 m depth was important (4.26 × 104 cells l−1). Moreover, this species reached a significant density also at 25 m depth (3.26 × 104 cells l−1). According to total phytoplankton values, the higher individual numbers were recorded at 10 m depth (9.74 × 104 cells l−1) due to the effect of dinoflagellates and at 20 m depth (7.26 × 104 cells l−1) due to the effect of diatoms.
In April, the effect of mucilage began to decline and filamentous structures were observed at 20–25 m depth. In this month, dinoflagellates were the dominant group in terms of species number and diatoms in terms of individual number (94.17–98.32%; Figure 5D). Diatoms reached the higher individual numbers at the depths of 20 m (1.35 × 105 cells l−1, 97.42%), 5 m (1.16 × 105 cells l−1, 96.52%) and 10 m (1.09 × 105 cells l−1, 95.63%). Cylindrotheca closterium played a significant role in the increase in diatoms. This species reached the highest individual number at 20 m depth (7.20 × 104 cells l−1). The individual number of dinoflagellates decreased considerably in this period despite the warming of waters (1.40–5.83%; Figure 5D). Like diatoms, the higher individual numbers of dinoflagellates were recorded at the depths of 5 m (4.00 × 103 cells l−1) and 10 m (4.70 × 103 cells l−1). The dominant species of this group were Prorocentrum micans Ehrenberg, 1834 and S. trochoidea but they did not reach a significant individual number. According to total phytoplankton values, the higher individual numbers were recorded at the depths of 20 m (1.39 × 105 cells l−1), 5 m (1.20 × 105 cells l−1) and 10 m (1.14 × 105 cells l−1).
In May, filamentous structures were observed at 10–20 m depth. In May, like in April, the dominant groups were dinoflagellates in terms of species number and diatoms in terms of individual number (71.79–95.17%; Figure 5E). Diatoms reached the higher individual numbers at the depths of 15 m (4.25 × 105 cells l−1, 95.17%), 0.5 m (3.99 × 105 cells l−1, 89.95%) and 5 m (3.01 × 105 cells l−1, 82.34%). Pseudo-nitzschia sp. was the most abundant species of diatoms and it reached the highest individual number at 15 m depth (4.23 × 105 cells l−1). The individual numbers of other species belonging to this group were considerably low. The species number of dinoflagellates increased compared to other months but a remarkable increase was not seen in terms of individual numbers (4.83–28.21%; Figure 5E). This group reached the higher individual numbers at the depths of 5 m (6.47 × 104 cells l−1) and 0.5 m (4.46 × 104 cells l−1) and P. micans was responsible for the increase. According to total phytoplankton values, the higher individual numbers were recorded at the depths of 15 m (4.46 × 105 cells l−1) and 0.5 m (4.44 × 105 cells l−1).
In June, the mucilage event lost its effect considerably, and it was observed very rarely at 10–30 m depth. Dinoflagellates were seen dominant over diatoms in terms of the number of species and individuals (19.51–80%; Figure 5F) while only at 0.5 m depth diatoms showed a considerable increase compared to dinoflagellates (1.72 × 104 cells l−1, 80.49%). While C. closterium was responsible for the increase in diatoms at this depth (1.48 × 104 cells l−1), dinoflagellates reached the highest individual number at 15 m depth (1.32 × 104 cells l−1) and P. micans (4.10 × 103 cells l−1) and S. trochoidea (5.00 × 103 cells l−1) were the species responsible for this increase. According to total phytoplankton values, the higher individual numbers were recorded at the depths of 10 m (2.35 × 104 cells l−1), 0.5 m (2.13 × 104 cells l−1) and 15 m (2.03 × 104 cells l−1).
Distribution of individual numbers of species known to cause mucilage formation and dominant in this study and diatom:dinoflagellate ratio
The changes by month and depth observed in this study in the individual numbers of the species known to cause mucilage formation (C. closterium, Pseudo-nitzschia sp., S. costatum, T. rotula and G. fragilis) are shown in graphics below (Figure 6).

Fig. 6. Monthly distribution of abundance (cells l−1) of species causing mucilage according to depth.
As can be seen in Figure 6, of diatoms C. closterium is the most dominant species and it is followed by S. costatum. Peaks are remarkable especially at 15 m depth in January, at 0.5 m depth in February and at 20–25 m depth in March, when mucilage was observed. In April, increases were seen in both upper and lower water column. Of dinoflagellates, G. fragilis was observed in all months until April while it was not seen in May and June with the decrease in mucilage. Of diatoms, Pseudonitzschia sp. played the most important role in the increase in May. In June, a considerable decrease was observed in the individual numbers of the species.
The diatom:dinoflagellate ratio ranged between 0.3 (June, 30 m) and 109.4 (February, 20 m). No dinoflagellate species was recorded at 30 m depth in February. In January and February, when mucilage formation was observed densely, especially at 20 m depth this ratio was found to be very high (44.2–109.4). In June, when mucilage formation disappeared completely, this ratio was measured to be very low (0.3–4.1) since the individual numbers of species were low and dinoflagellates showed increase compared to diatoms.
Statistical analysis
Spearman's rank-correlation analysis determined some positive and negative relations between some biotic and abiotic variables (Table 1). The density of mucilage increased meaningfully in correlation with nitrogen (P < 0.01), phosphorus, silica (P < 0.05) and also with the N:P ratio (P < 0.01). It was determined that the increase in the percentage of individual numbers of 5 species (C. closterium, Pseudo-nitzschia sp., S. costatum, T. rotula and G. fragilis) known to cause mucilage formation and found dominantly in this study (P < 0.05) and the increase in the percentage of individual number of diatoms (P < 0.01) were in positive correlation with mucilage formation. Especially at 15 m depth in January, when mucilage formation was observed most densely, a meaningful correlation was observed between S. costatum and T. rotula, which increased excessively, and mucilage formation (P < 0.05).
Table 1. Spearman's rank-correlation matrix (rs) to correlate some biotic (DM, density of mucilage; DA, diatom abundance; DFA, dinoflagellate abundance; PI5, percentage of individual numbers of five species causing to mucilage; PDI, percentage of individual numbers of diatoms; PPHY, picophytoplankton; NPHY, nanophytoplankton; MPHY, microphytoplankton; TB, total bacteria; TPHY, total phytoplankton) and abiotic (T, temperature; S, salinity; DO, dissolved oxygen; D, depth; N, nitrogen; P, phosphorus; Si, silica) variables in the study area (**P < 0.01, *P < 0.05; ns, not significant; N = 42).

Of diatoms, the percentage of C. closterium was found to be in correlation with only phosphorus (P < 0.01) while the individual numbers of the species belonging to dinoflagellates were negatively affected by nitrogen, phosphorus (P < 0.01), silica (P < 0.05) and also the N:P ratio (P < 0.01). In addition, the abundance of diatoms (P < 0.05) and dinoflagellates (P < 0.01) was negatively correlated with salinity.
Pico-, nano-, microphytoplankton and total bacteria were negatively correlated with temperature (P < 0.01) and positively correlated with chlorophyll-a (P < 0.01). Picophytoplankton was negatively correlated with nitrogen (P < 0.05) and pico- and nanophytoplankton with the N:P ratio (P < 0.01). In addition, the density of mucilage positively correlated with pico-, nano-, microphytoplankton (P < 0.01) and total bacteria (P < 0.05).
Trophic index (TRIX) values were positively correlated with depth, salinity, all nutrients, the N:P ratio and the density of mucilage (P < 0.01) and negatively correlated with the abudance of dinoflagellate (P < 0.01).
DISCUSSION
The chemical oceanography of the Marmara Sea is remarkably affected by the Black Sea and the Aegean Sea, and the basin includes two different water masses. One of these masses consists of the low salinity waters originating from the Black Sea and observed as a relatively thin layer (10–15 m) on the surface, and the other of waters with higher salinity originating from the Mediterranean Sea, flowing at the bottom and separated by a sharp intermediate layer with a thickness of about 10–20 m (pycnocline) (Ünlüata et al., Reference Ünlüata, Oğuz, Latif, Özsoy and Pratt1990; Tuğrul & Polat, Reference Tuğrul and Polat1995). In this study, after 15–20 m depth an increase was observed in the salinity values and the effect of the Mediterranean waters on this increase was remarkable. In addition, the increase in temperature at 30 m depth during cold periods indicates the effect of the Mediterranean waters. On the other hand, the increase in surface waters during warm periods is due the effect of light and the contact of this layer with the atmosphere.
A decrease was observed in oxygen values generally from the surface to the bottom along the water column. The main reasons for this decrease, remarkable especially at 25–30 m depth, are bacterial decomposition and less occurrence of phytoplankton activity (photosynthesis) due to insufficient light at these depths.
The highest nutrient values were recorded generally at the bottom layer where there was aggregation. The reason for the determination of low amounts in the upper water column compared to the bottom layer is the consumption of nutrients due to the increase in phytoplankton at these depths. Especially, the increase in nitrogen amount at 20–30 m depth is remarkable since this element comes out due to the bacterial decomposition of the organic substances aggregated at the bottom. The N:P ratios recorded during the study are below 16:1 and N is the limiting element in the sampling area. The increase in the N:P ratio is remarkable at lower layers in comparison to the surface water, which shows that phosphorus increases in proportion to nitrogen in the surface water while nitrogen increases in proportion to phosphorus in lower layers. Mucilage formation in the North Adriatic Sea and Tyrrhenian Sea is observed more densely especially in the deficiency of P and less densely in the deficiency of N (Innamorati et al., Reference Innamorati, Nuccio, Massi, Mori and Melley2001; Pompei et al., Reference Pompei, Mazziotti, Guerrini, Cangini, Pigozzi, Benzi, Palamidesi, Boni and Pistocchi2003). According to Innamorati et al. (Reference Innamorati, Nuccio, Massi, Mori and Melley2001), in nature the deficiency of P can be considered as a necessary condition for the appearance of mucilages, but not the only factor.
During the sampling period, the N:Si ratio was below 1 except for two months of the period and two depths, which showed that silica is not a limiting element for the sampling area. These values allowed sufficient development of diatoms. The main reason for the high N:Si ratio especially in the lower layer waters is that in the upper water column silica is consumed by diatoms in building cell wall. According to the Spearman's rank-correlation analysis, the density of mucilage increased meaningfully with the N:P ratio besides nitrogen, phosphorus and silica in seawater. This shows that mucilage is closely correlated with the increase in the amount of nutrition element in seawater.
Diatoms are found densely in waters rich in nutrients and with a good mixture (Haris, Reference Haris1986) and are one of the groups playing an essential role in mucilage formation (Rinaldi et al., Reference Rinaldi, Vollenweider, Montanari, Ferrari and Ghetti1995). High nutrient concentrations accelerate cell division and the mixture decelerates the sinking of cells to the bottom (Arin et al., Reference Arin, Moran and Estrada2002). Since the levels of nutrients are high in coastal areas, diatoms might reach considerably high levels in comparison to dinoflagellates in terms of density in these areas. While silica, which plays a role especially in building the cell wall of diatoms, was found in high amounts in the study area, it did not limit growth. That nutrient values were at sufficient levels for phytoplankton growth provided adequate conditions for the increase of diatoms which needs high nutrient levels. The recorded diatom:dinoflagellate ratios also indicated the dominance of diatoms. In January and February, when mucilage formation was observed densely, at especially 15–20 m depth this ratio was found to be very high (44.2–109.4). In June, when mucilage formation was not observed, this ratio was measured to be rather low (0.5) since the individual numbers of species was low and dinoflagellates increased in comparison to diatoms. Besides, in January, when increase in chlorophyll-a was observed, at 15 m depth (6.35 µg l−1) and in April at 10 m depth (4.89 µg l−1) an increase was observed in the abundance of phytoplankton, and diatoms played a significant role in this increase.
In this study conducted on Büyükada Island mucilage formation was observed most densely in January, February and March, and the density decreased gradually in April and May, and in June almost no mucilage was observed. A meaningful correlation was found between C. closterium, T. rotula and S. costatum, main species among those known to cause mucilage formation, and a mucilage event according to the Spearman's rank-correlation analysis. These species were also found dominantly in the studies carried out in İzmit Bay (Tüfekçi et al., in press). However, although the dinoflagellate G. fragilis had not been recorded until 2007, it appeared with the beginning of mucilage formation in the Marmara Sea, increased in some periods and disappeared in May and June, when mucilage begins to disappear. These observations indicate that this species is also effective on mucilage formation as was reported in the previous studies conducted in the Adriatic Sea. Pompei et al. (Reference Pompei, Mazziotti, Guerrini, Cangini, Pigozzi, Benzi, Palamidesi, Boni and Pistocchi2003) detected the capacity of this organism to produce large amounts of extracellular carbohydrates in culture and stated that a few thousand G. fragilis cells release the same amount of carbohydrates as that produced by tens of millions of C. closterium cells. In August 2005, the maximum individual number of this species on Greek coasts was determined to be 7.92 × 103 cells l−1 and the species was stated to cause mucilage formation (Nikolaidis et al., Reference Nikolaidis, Aligizaki, Koukaras and Moschandreou2006). In this study, the maximum individual number of G. fragilis at 10 m depth in January was determined to be 1.82 × 104 cells l−1.
While the mucilage event was in positive correlation with all nutrients, it was in negative correlation with the individual number of dinoflagellates. Dinoflagellates were in negative correlation with nutrients, which indicates that dinoflagellates are not so effective on the mucilage event. The real reason responsible for the mucilage event is the species belonging to diatoms that grow with the increase in nutrients. In addition, it is also known that diatoms grow better in mucilage mass in comparison to dinoflagellates (Pompei et al., Reference Pompei, Mazziotti, Guerrini, Cangini, Pigozzi, Benzi, Palamidesi, Boni and Pistocchi2003; Tinti et al., Reference Tinti, Boni, Pistocchi, Riccardi and Guerrini2007). In the mucilage event that occurred in the Adriatic Sea in 1988 and 1999, Revelante & Gilmartin (Reference Revelante and Gilmartin1991) found an enrichment of species, such as Nitzschia longissima and C. closterium, at higher levels than in the surrounding seawater. The aggregates represent a microenvironment very suitable for the development of a rich community of microorganisms which is separated from the surrounding water (Simon et al., Reference Simon, Grossart, Schweitzer and Ploug2002), but phytoplankton communities associated with mucilage aggregates have been shown to vary, with different dominating species, depending on sampling area and period (Revelante & Gilmartin, Reference Revelante and Gilmartin1991; Cabrini et al., Reference Cabrini, Fonda-Umani and Honsell1992; Totti et al., Reference Totti, Cangini, Ferrari, Kraus, Pompei, Pugnetti, Romagnoli, Vanucci and Socal2005).
A positive increase was observed in the number of pico-, nano-, microphytoplankton (P < 0.01) and total bacteria (P < 0.05) by the density of mucilage. Negro et al. (Reference Negro, Crevatin, Larato, Ferrari, Totti, Pompei, Giani, Berto and Fonda-Umani2005) determined that exopolysaccharides in mucilage aggregates could meet the carbohydrate need of bacteria and bacteria that are in relation with mucilage aggregates are metabolically more active according to enzyme analyses. As a result, it might be stated that the presence of mucilage could lead to an increase in the bacteria in water. In April, when the number of total bacteria is the highest at 25 m depth (6655 cells ml−1), the dissolved oxygen value is the lowest (2.75 mg l−1) due to the bacterial activity.
It is our opinion that the mucilage formation in the Marmara Sea is mainly due to the excretory activity of some diatoms together with bacteria, the dinoflagellate G. fragilis, the presence of sharp pycnocline and thermocline caused by the two-layered water system of the Marmara Sea; besides which, the weather conditions and the status of currents during that time are effective on this formation.