INTRODUCTION
Primary production is nitrogen-limited in most marine ecosystems (Granéli et al., Reference Granéli, Wallstrom, Larsson, Graneli and Elmgren1990; Kivi et al., Reference Kivi, Kaitala, Kuosa, Kuparinen, Leskinen, Lignell, Marcussen and Tamminen1993; Tuomainen et al., Reference Tuomainen, Hietanen, Kuparinen, Martikainen and Servomaa2003). Nitrogen fixation, nitrification and denitrification, which are all microbially mediated, are the key processes for determining the availability of nitrogen (Tuomainen et al., Reference Tuomainen, Hietanen, Kuparinen, Martikainen and Servomaa2003; Sorokovikova et al., Reference Sorokovikova, Belykh, Gladkikh, Kotsar, Tikhonova, Timoshkin and Parfenova2013; Stief et al., Reference Stief, Fuchs-Ocklenburg, Kamp, Manohar, Houbraken, Boekhout, Beer and Stoeck2014). Cyanobacteria are the major biomass producers both in aquatic and terrestrial ecosystems and represent more than 50% of the biomass in many aquatic ecosystems (Häder et al., Reference Häder, Kumar, Smith and Worrest2007). Cyanobacteria appear to be responsible for most of planktonic N2 fixing in aquatic ecosystems and this ability gives a significant competitive advantage to these organisms during periods of nitrogen limitation (Tilman et al., Reference Tilman, Kilham and Kilham1982; Howarth et al., Reference Howarth, Cole, Marino and Lane1988; Leppänen et al., Reference Leppänen, Niemi and Rinne1988; Gallon, Reference Gallon1992; Zehr et al., Reference Zehr, Waterbury, Turner, Montoya, Omoregie, Steward, Hansen and Karl2001; Paerl & Otten, Reference Paerl and Otten2013). Diazotrophic (nitrogen-fixing) cyanobacteria are important contributors of new nitrogen to oligotrophic environments and greatly influence oceanic productivity (Berman-Frank et al., Reference Berman-Frank, Quigg, Finkel, Irwin and Haramaty2007). The factors controlling N2 fixation are still poorly known in the Mediterranean Sea (Ridame et al., Reference Ridame, Le Moal, Guieu, Ternon, Biegala, L'Helguen and Pujo-Pay2011). As the Mediterranean Sea has been described as a phosphate-depleted basin, phosphorus can be logically suspected to be the limiting nutrient for diazotrophic activity (Ridame et al., Reference Ridame, Le Moal, Guieu, Ternon, Biegala, L'Helguen and Pujo-Pay2011). The Mediterranean Sea is strongly impacted by episodic Saharan dust deposition (e.g. Guerzoni et al., Reference Guerzoni, Chester, Dulac, Herut, Loye-Pilot, Measures, Migon, Molinaroli, Moulin, Rossini, Saydam, Soudine and Ziveri1999; Guieu et al., Reference Guieu, Loye-Pilot, Benyaya and Dufour2010). It has been shown that new atmospheric nutrients associated with Saharan dust pulses significantly stimulate N2 fixation in the Mediterranean Sea and that N2 fixation is a key process in the carbon cycle in oligotrophic environments such as the Gulf of Gabes (Drira et al., Reference Drira, Hamza, Bel Hassen, Ayadi, Bouaïn and Aleya2008; Ridame et al., Reference Ridame, Guieu and L'Helguen2013; Elloumi et al., Reference Elloumi, Drira, Guermazi, Hamza and Ayadi2015). Despite being oligotrophic, the Gulf of Gabes accounts for 65% of Tunisian fish production (DGPA, 2005–2009) and is a well-known habitat for marine turtles such as Caretta caretta and Chelonia mydas (Maffucci et al., Reference Maffucci, Kooistra and Bentivegna2006; Lotze & Worm, Reference Lotze and Worm2009). We hypothesized that, although the Gulf of Gabes is an oligotrophic ecosystem, the diazotrophic phenomena as well as the Saharan dust inputs stimulate the primary production in this ecosystem and could explain the important national fish production in Tunisia. The purpose of this study was therefore to examine over space and time variations of the marine nitrogen-fixing cyanobacteria in the Gulf of Gabes in relation to the hydrographic water properties and the nutrient contents during four cruises carried out in the gulf between 2005 and 2007.
MATERIALS AND METHODS
Field sampling
Samplings were carried in the Gulf of Gabes, Tunisia (Eastern Mediterranean Sea, between 35°N and 33°N) aboard the RV ‘Hannibal’ during four oceanographic cruises: 9–14 July 2005, 27 May – 9 June 2006, 7–10 September 2006 and 16–19 March 2007 (Figure 1). The Gulf of Gabes has a 700 km coastline stretching from the coastal zone of Tunis to the industrial town of Gabes. Its climate is dry (average annual precipitation: 210 mm year−1) and sunny with strong easterly winds resulting in severe aeolian erosion. The Gulf of Gabes opens to the offshore area and has a wide continental shelf. The tide is semidiurnal, with a maximum range of about 2 m. The shallow waters run around Kerkennah, Djerba islands, and the lagoons of Boughrara and El Bibane (Hattour et al., Reference Hattour, Sammari and Ben Nassrallah2010). As the Gulf of Gabes is located between the eastern and western parts of the Mediterranean Sea, the dynamics of its water masses are strongly linked to the general circulation of the Mediterranean Sea.
Fig. 1. Geographic map focusing on the planktonic cyanobacteria sampling stations in the Gulf of Gabes during the four cruises between 2005 and 2007.
To examine spatial trends, 26–34 stations were sampled per cruise encompassing the continental shelf area between 20 and 200 m. In each station, temperature, salinity and density (sigma-t) were measured with a Conductivity-Temperature-Depth profiler (CTD: SBE 9, Sea-Bird Electronics, USA) equipped with a 12 Niskin bottle rosette sampler lowered from the surface to the near bottom. Water samples for physical and chemical analyses and phytoplankton assessment were collected at three depths (surface, mid-water column and near bottom) in coastal stations less than 50 m in depth and at 5 depths (surface, −10 m, −20 m, thermocline and near bottom) in offshore stations where depth exceeded 50 m. Samples for nutrient analyses (120 mL) were preserved immediately upon collection (−20°C, in the dark), and those for phytoplankton counts (1 L) were preserved with Lugol iodine solution (Bourrelly, Reference Bourrelly1985) and stored in the dark at low temperature (4°C) until analysis. Water samples (2 L) for chlorophyll a analyses were filtered by vacuum filtration onto Whatman GF/F fibreglass filter, and filters were then immediately stored at −20°C.
Samples analysis
The pH was measured immediately after sampling using a Met Röhm® type pH meter. Nutrients (NO2 −, NO3 −, NH4 +, PO4 3−, Si(OH)4, T-N and T-P) were analysed with a BRAN and LUEBBE type 3 autoanalyser and their concentrations were determined colorimetrically using a UV-visible (6400/6405) spectrophotometer (APHA, 1992). We calculated the Ni/Pi = DIN/DIP = NO2 − + NO3 − + NH4 + /PO4 3−. Turbidity was estimated by the measurement of the suspended matter rate (mg L−1) and euphotic layer by using a Secchi disk (m). The concentrations of suspended matter were determined by measuring the dry weight of the residue after filtration of 1 L of seawater onto Whatman GF/C membrane filters. Chlorophyll a analyses were performed by HPLC according to Pinckney et al. (Reference Pinckney, Richardson, Millie and Paerl2001), the analytical method was fully described in Bel Hassen et al. (Reference Bel Hassen, Drira, Hamza, Ayadi, Akrout and Issaoui2008).
Cyanobacteria enumeration
Cyanobacteria in sub-samples were counted with an inverted microscope after settling for 24–48 h using the Utermöhl's method (Reference Utermöhl1958). The identification of cyanobacteria taxa was achieved according to Bourrelly (Reference Bourrelly1985) and Baker (Reference Baker1991, Reference Baker1992). Biovolumes, which were calculated from microscopic measurements of length and width, and assuming simple geometrical shapes (Lohman, Reference Lohman1908; Hillebrand et al., Reference Hillebrand, Durselen, Kirschtel, Pollingher and Zohary1999), were converted into carbon biomass using the conversion factors proposed by Menden-Deuer & Lessard (Reference Menden-Deuer and Lessard2000): 1 µm3 = 0.216 × 10−6 µg C, for cyanobacteria.
The dominance index for the cyanobacteria was calculated according to the formula δ = (n 1 + n 2)/N, where δ is the dominance index that is equal to the percentage of contribution of the two most important species (n 1 + n 2) of the total standing stock and N the total individual abundance.
Statistical analyses
Pearson's rank-correlation was performed using XL-Stat software to determine, for each sampling period, the correlations between the different biological and physico-chemical parameters. To evaluate differences between inshore and offshore areas for each studied biological, physical and chemical parameters with a confidence level of 95%, a non-paired Student's t-test was carried out. Differences were considered significant at P < 0.05. Analysis of variance (ANOVA) was applied to identify significant differences between the sampled periods. The co-inertia analysis which was a direct extension of multiple regressions to the modelling of a multivariate response matrix (Legendre & Legendre, Reference Legendre and Legendre1998) was performed to examine the correlation between an array of response variables (in this study the four sampled periods) and of independent explanatory variables (planktonic cyanobacteria communities) conditional to a third matrix (here physical and chemical parameters), keeping the physico-chemical effect constant. The overall canonical relationship between data matrices was tested with a permutation test and the total variation was partitioned into variations due to main cyanobacteria taxa, to physical and chemical variables and to the co-variation between the four sampling periods, following Peres-Neto et al. (Reference Peres-Neto, Legendre, Dray and Borcard2006). Computing and graphical displays were performed using R 2.4 (R-Development Core Team, 2006), the packages ade4 1.4.2 (Chessel et al., Reference Chessel, Dufour and Thioulouse2004) and vegan 1.8–3 (Oksanen et al., Reference Oksanen, Kindt, Legendre and O'Hara2006). In order to prepare the analysis, data matrices were explored using principal component analysis to picture the co-variations between the four sampled periods and between the cyanobacteria data and the physical and chemical parameters.
RESULTS
Physical and chemical parameters
The mean values of the different physical features such as temperature, salinity, sigma-t, pH and suspended matter determined from inshore (less than 50 m, in depth) and offshore samples (more than 50 m, in depth), during the entire survey period, are given in Table 1. The highest temperature was observed in inshore samples during September 2006 (26.56 ± 1.36°C) and the lowest one was recorded in the offshore sample during March 2007 (16.16 ± 0.29°C). The ANOVA analysis showed that temperature (F = 5.576, df = 119) differed significantly (P < 0.001) between the sampled periods (Table 1). Salinity, which showed quite similar variations in both inshore and offshore samples during the survey period, varied from 37.38 ± 0.69 in July in the offshore area to 38.15 ± 0.38 detected in September in the inshore area. In the spring season (May–June 2006) the thermocline was at 20 m in depth, while it was located deeper than 25 m depth during the summer period (July 2005) and was even more pronounced and deeper than 30 m in September 2006. The mean salinity profiles were similar in July and September, both in inshore and offshore stations, but they differed markedly in May–June 2006 and March 2007. In May–June 2006, salinity at 20 m was higher in the inshore than in the offshore areas but the reverse case was found in deeper water samples. In March 2007, the water column was well-mixed and the salinity decreased with depth in the inshore area but increased in the open sea. The mean sigma-t values showed a slight difference between the inshore and offshore areas only in May–June 2006. The highest values were recorded in the inshore area during March 2007 and the lowest ones at the inshore area in September 2006. Sigma-t was negatively correlated with temperature at the beginning of the water stratification in May–June (r = −0.745, P < 0.0001, df = 28), July (r = −0.825, P < 0.0001, df = 31) and September (r = −0.903, P < 0.0001, df = 32). However, sigma-t and salinity of inshore waters were positively correlated (r = 0.992, P < 0.05, df = 24) under the well-mixed conditions in March 2007. In the offshore region, sigma-t was correlated with both temperature (r = 0.416, P < 0.05, df = 24) and salinity (r = 0.967, P < 0.05, df = 24).
Table 1. Mean values and standard deviation (SD) of physico-chemical and biological parameters in inshore and offshore areas of the Gulf of Gabes during the four cruises between 2005 and 2007. Student's t-test was carried out to evaluate differences between inshore and offshore areas for each studied parameters. Analysis of variance (ANOVA) was applied to identify significant differences between the sampled periods with df equal to 119. F values: between-groups mean square/within-groups mean square. *Significant difference between inshore and offshore areas: *(P < 0.05); **(P < 0.001); ***(P < 0.0001).
The mean pH values were generally more alkaline in the inshore samples than in the offshore samples, suggesting a more pronounced photosynthetic activity along the coast mainly during May–June 2006 (Student's t, P < 0.0001). The mean concentrations of the suspended matter ranged from 19.44 mg L−1 (May–June 2006) to 64.93 mg L−1 (September 2006) in offshore area and from 20.21 ± 1.61 mg L−1 (May–June 2006) to 33.62 ± 24.43 mg L−1 (July 2005) in the inshore stations. The Secchi disk values showed highest values in the offshore area during the beginning of stratification in May–June 2006 (21.85 ± 3.76 m) and the lowest ones during the mixing period in March 2007 in the offshore area (16 ± 2.94 m). The mean nutrient values were more important in inshore than in offshore sampled stations during all the survey period. The concentrations of NO3 − + NO2 − were significantly higher during May–June 2006 and July 2005 than during September 2006 and March 2007, whereas NH4 + concentrations exhibited a reverse trend. PO4 3− concentrations were low (0.06–0.1 µM) during the well-stratified conditions (July 2005 and September 2006), but reached a maximum of 0.44 ± 0.21 µM during May–June in inshore area. So, N/P ratios were higher (>10) during stratification but were lower than the Redfield ratio (16), suggesting a potential N limitation during May–June 2006 (<10).
Biological parameters
The mean chlorophyll-a concentrations determined during the thermal stratification from both inshore and offshore samples were <3 × 10−4 mg L−1 (Table 1). They were high during May–June 2006 with a sub-surface chlorophyll maximum found up to 15 m in the inshore area, and a deep chlorophyll maximum, DCM (5 × 10−4 mg L−1), found up to 40 m in the offshore area. The phytoplankton community consisted of Dinophyceae (120 taxa), Bacillariophyceae (40 taxa), Cyanobacteria (seven taxa) and other groups such as Dictyochophyceae, Euglenophyceae, Coccolithophorideae and Chlorophyceae were represented by one species. The phytoplankton abundance varied from 2.69 × 103 to 47.63 × 103 cells L−1, these being recorded respectively in July 2005 in offshore area and May–June 2006 in inshore area. The highest cyanobacteria abundance was recorded in May–June 2006 in the inshore area (1.79 × 103 ± 1.94 × 103 cells l−1), with Pseudoanabaena sp. contributing to 10% of the total phytoplankton and 68% of total cyanobacteria abundances (Table 1, Figures 2A & 3B). The lowest cyanobacteria abundance was observed in September 2006 (80 ± 60 cells L−1). The ANOVA analysis showed that cyanobacteria abundance differed significantly (F = 15.359, df = 119, P < 0.0001) between the sampled periods (Table 1). The study of the marine planktonic cyanobacteria species throughout the sampling period showed the presence of seven different species (Table 2). The cyanobacteria were distributed throughout the water column in both inshore and offshore areas (Figure 2A) but they exhibited no clear trend in their vertical distribution. They were more abundant during the water mixing period (March 2007) and preferentially during the semi-mixing period (May–June 2006) mainly in the inshore area at 34 m in depth (Figure 2A). At this latter period a high proliferation of planktonic cyanobacteria was observed at the offshore stations near the thermocline level coinciding with the euphotic layer (21.85 ± 3.76 m) allowing access of light radiation. During this period, we also noted the lowest ammonium concentration decreasing with depth but the planktonic cyanobacteria was weakly correlated with the NH4 + amount (r = 0.098, P < 0.05, df = 28).
Fig. 2. Average vertical distribution of cyanobacteria abundance (A), biomass (B) and ammonium concentration (C) in coastal and open sea areas in the Gulf of Gabes between 2005 and 2007.
Fig. 3. Relationships between the dominance index of the planktonic cyanobacteria species and distance to the coast during July 2005 (A), May–June 2006 (B), September 2006 (C) and March 2007 (D) (Ana: Anabaena sp., Pseudo: Pseudoanabaena sp., Tricho: Trichodesmium erythraeum, Micro: Microcystis sp., Spiru: Spirulina subsalsa).
Table 2. List, abundance (Abun), mean density (D) and length (L) of the marine cyanobacteria species observed between 2005 and 2007 (V = very abundant. 30–100%; A = abundant. 10–30%; C = common. 5–10%; R = rare. 1–5%; X = present occasionally. 0–1; – Not detected).
The mean values of cyanobacteria biomass varied from 0.56 ± 0.60 µg C L−1 in September 2006 in the offshore area to 46.59 ± 13.9 µg C L−1 in July 2005 in the offshore area (Table 1, Figure 2B). This important biomass in July 2005 was due to the large size of the dominant species that were present at this time, such as Trichodesmium erythraeum (134.3 ± 123.85 µm) and Pseudoanabaena sp. (82.41 ± 53.20 µm) with an abundance varying from 1 to 5% of total phytoplankton and representing 41 and 50% of total cyanobacteria abundance, respectively (Table 2). During the semi-mixed period (May–June 2006), the biomass of cyanobacteria was still important with large-sized opportunist species such as Pseudoanabaena sp. (128.75 ± 66.12 µm) (300 ± 129 µm) accounting for 68% of the total cyanobacteria abundance (Table 2). The lowest biomass recorded in March 2007 in the offshore area was due to small-sized opportunist species such as Anabaena sp. (35.11 ± 5.05 µm) and Spirulina subsalsa (54 ± 12.72 µm) (Table 2). The ANOVA analysis showed that cyanobacteria biomass differed significantly (F = 50.701, df = 119, P < 0.0001) between the sampled periods (Table 1). During all the survey period, the dominance index showed that the biodiversity of cyanobacteria increased gradually along the inshore-offshore gradient (Figure 3). This index increase depended on a coastal–open sea distance during the stratified period (July 2005 and September 2006), the mixing period (March 2007) and at the beginning of stratification (May–June 2006). In fact, during July 2005, Trichodesmium erythraeum dominated in the inshore zone, while Anabaena sp. proliferated in the offshore zone. Pseudoanabaena sp. was found in substantial amounts between the inshore and offshore areas (Figures 3A & 4). During May–June 2006, we noted the dominance of Pseudoanabaena sp. and Anabaena sp. in substantial amounts between the inshore and offshore areas (Figures 3B & 4). During September 2006, the inshore zone was dominated by Trichodesmium erythraeum while the offshore zone was characterized rather by Microcystis sp. Anabaena sp. was found in substantial amounts between the inshore and offshore areas (Figures 3C & 4). During March 2007, the inshore zone was dominated by Microcystis sp. while the offshore zone was characterized rather by Trichodesmium erythraeum. Pseudoanabaena sp. and Anabaena sp. proliferated at the intermediate zone (Figures 3D & 4).
Fig. 4. The main cyanobacteria species taken with inverted microscopy (40×) recorded in the Gulf of Gabes during the survey period (2005–2007).
The co-inertia plot illustrated a close relationship between the composition of planktonic cyanobacteria species and the physico-chemical features of the water column during the four sampling periods (Figure 5A). The overall model explained 77% of the total variation (permutation test, P = 0.24, 1000 replicates). This variation was due to cyanobacteria taxa (22%) and to physico-chemical variability (17%) (Figure 5B). The May–June 2006 and March 2007 sampling periods showed close links between sigma-t, NO3 −, PO4 3− and T-P concentrations and the phytoplankton composition which was illustrated by the position of cyanobacteria species such as Pseudoanabaena sp., Anabaena sp., Microcystis sp., Spirulina sp. and Chroococcus sp. around the Y axis (Figure 5A). In contrast, during July 2005 and September 2006, this component axis was surrounded by the numerically dominant Trichodesmium erythraeum and Spirulina subsalsa. To summarize, the co-inertia indicated the clear differences between the four studied periods for the distribution of planktonic cyanobacteria species. In fact, we recorded a clear dominance of Pseudoanabaena sp. as opportunistic species during the mixing (March 2007; 68% of the total cyanobacteria abundance) and the begging of stratification (May–June 2006; 88% of the total cyanobacteria abundance). This species characterizing the period of bloom showed a significantly negative correlation with the nitrate concentration (r = −0.09; P < 0.05; df = 28) probably indicating nitrate consumption. Trichodesmium erythraeum dominated the cyanobacteria community when the water column was well stratified (July 2005: 41% of the total cyanobacteria abundance and September 2006: 21% of the total cyanobacteria abundance).
Fig. 5. Co-inertia plot for the biotic and abiotic parameters changes for the four sampling periods (A) and partition of cyanobacteria species and physico-chemical features.
DISCUSSION
As described in previous studies (Drira et al., Reference Drira, Bel Hassen, Hamza, Rebai, Bouain, Ayadi and Aleya2009, Reference Drira, Bel Hassen, Ayadi and Aleya2014a, Reference Drira, Elloumi, Guermazi, Bel Hassen, Hamza and Ayadib), the hydrological features showed that during early spring (March 2007) the water column is well mixed and that thermal stratification of the water column starts in late spring (May–June 2006) until its complete establishment in summer (July 2005 and September 2006). The nitrogen availability in July 2005 was similar to that in May–June 2006 with a dominance of nitrate over ammonium concentrations. Moreover, in July 2005, the low phosphate concentration was similar to that in September 2006. If it is assumed that the July cruise constituted the continuum between May–June and September, it could be inferred that phosphate was the first limiting nutrient factor in the Gulf of Gabes, as reflected by the high N/P ratio during the strong stratification period in July and September. This strong stratification resulted in a shift in nitrogen sources from nitrates to ammonium as well as to phosphate depletion (<2.5 µM) and to a decrease in silicate concentrations (<3 µM) (Bel Hassen et al., Reference Bel Hassen, Drira, Hamza, Ayadi, Akrout, Messaoudi, Issaoui, Aleya and Bouain2009; Drira et al., Reference Drira, Bel Hassen, Hamza, Rebai, Bouain, Ayadi and Aleya2009). N/P ratio was lower in May–June 2006 (<16) than in September 2006 (>16). In May–June 2006, when the cyanobacteria abundance was highest, the N/P ratio was clearly below the Redfield standard molar ratio of 16/1. The same result was observed in the Baltic Sea where cyanobacteria were favoured by low N/P-ratios and P was probably in excess during the bloom formation (Westman et al., Reference Westman, Borgendahl, Bianchi and Chen2003; Vahtera et al., Reference Vahtera, Conley, Gustafsson, Kuosa, Pitkanen, Savchuk, Tamminen, Viitasalo, Voss, Wasmund and Wulff2007; Walve et al., Reference Walve, Gelting and Ingri2014). This situation was also observed in the Mediterranean Sea in two sites on the Sicilian coastline in 1993 during which low N/P ratios (<10:l) coincided with high development of planktonic cyanobacteria in the subsurface layer (Giacobbe et al., Reference Giacobbe, Oliva, La Ferla, Puglisi, Crisafi and Maimone1995). In addition, high values of N/P ratio widely exceeding the Redfield ratio (N/P > 20), suggesting an overall phosphate depletion during July, September and March cruises, are consistent with other reports of P limitation in the Mediterranean Sea (Jacques et al., Reference Jacques, Cahet, Fiala and Panouse1973; Minas et al., Reference Minas, Minas, Coste, Gostan, Nival and Bonin1988; Thingstad & Rassoulzadegan, Reference Thingstad and Rassoulzadegan1995; Thingstad et al., Reference Thingstad, Zweifel and Rassoulzadegan1998). During our survey, cyanobacteria were found throughout the water column and were more concentrated at the thermocline within the deeper zone as commonly observed (Ribera d'Alcalà et al., Reference Ribera d'Alcalà, Conversano, Corato, Licandro, Mangoni, Marino, Mazzocchi, Modigh, Montresor, Nardella, Saggiomo, Sarno, Zingone, Ros, Packard, Gili, Pretus and Blasco2004). Cyanobacteria proliferated during the semi-mixed conditions (May–June 2006 and July 2005) mainly in the coastal area. They seemed to be governed by the nitrogen availability in the offshore area of the Gulf of Gabes favoured by the presence of a thermocline between 20 and 25 m (Drira et al., Reference Drira, Bel Hassen, Hamza, Rebai, Bouain, Ayadi and Aleya2009). In fact, cyanobacteria developed throughout the whole water column with a high density recorded at the thermocline and associated with Trichodesmium erythraeum and Pseudoanabaena sp. which contributed to 41 and 50% of the total cyanobacterial abundance, respectively. In these layers, cyanobacteria abundance was correlated with both NO3 − (r = 0.343, P < 0.05, df = 119) and NH4 +(r = 0.344, P < 0.05, df = 119), as also demonstrated by Drira et al. (Reference Drira, Hamza, Bel Hassen, Ayadi, Bouaïn and Aleya2008) and a significant correlation was recorded between cyanobacteria and N/P (r = 0.566, P < 0.05, df = 119). This indicates that the cyanobacterial growth was probably induced by the nitrate replenishment of the upper stratified layers after release by the thermocline.
Our study showed that cyanobacteria in the Gulf of Gabes was made up of seven different species, with some of them being N2-fixing such as Anabaena sp., Chroococcus sp., Trichodesmium erythraeum, Spirulina sp. and Spirulina subsalsa, and others being non N2-fixing such as Pseudoanabaena sp. and Microcystis sp. (Berman-Frank et al., Reference Berman-Frank, Quigg, Finkel, Irwin and Haramaty2007; O'Neil et al., Reference O'Neil, Davis, Burford and Gobler2012; Paerl & Otten, Reference Paerl and Otten2013). All these cyanobacteria species were also recorded in the Gulf of Gabes in summer (July) 2005 (Drira et al., Reference Drira, Hamza, Bel Hassen, Ayadi, Bouaïn and Aleya2010). These species, except Microcystis sp., were also observed in the coast of Chebba (Gulf of Gabes, East of Tunisia) during a summer cruise (August 2011) (Mabrouk et al., Reference Mabrouk, Hamza and Bradai2014). In addition, Anabaena spherica and Microcystis aeruginosa developed in both inshore and offshore stations in July 2006 during the summer water stratification in the Gulf of Hammamet (Tunisia, eastern Mediterranean Sea) (Hannachi et al., Reference Hannachi, Drira, Bel Hassen, Hamza, Ayadi and Aleya2011). Furthermore, a preferential occurrence of cyanobacteria under the semi-mixed conditions was also reported in a coastal north-western Mediterranean site (Bustillos-Guzman et al., Reference Bustillos-Guzman, Claustre and Marty1995; Marty et al., Reference Marty, Chiaverini, Pizay and Avril2002). In fact, the recorded species were generally large colony-forming prokaryotes such as the filamentous Anabaena and Pseudoanabaena, which were able to overcome sedimentation losses under stratified conditions (Drira et al., Reference Drira, Bel Hassen, Hamza, Rebai, Bouain, Ayadi and Aleya2009). Anabaena sp. was also the dominant species (36% of total cyanobacteria abundance) at 20 nearshore stations of the Gulf of Gabes during a 10-year survey with the greatest abundance (5.3 × 104 cells L−1) being recorded in 1997 (Feki et al., Reference Feki, Hamza, Frossard, Abdennadher, Hannachi, Jacquot, Bel Hassen and Aleya2013). In our study the highest cyanobacteria abundance (1.79 × 103 ± 1.94 × 103 cells L−1) was recorded during May–June 2006 in the inshore area with Pseudoanabaena sp. contributing to 10% of the total phytoplankton and 68% of total cyanobacteria abundances. The development of cyanobacteria near the thermocline (−25 m) was associated with high nitrate concentrations (r = 0.312, P < 0.05, df = 31) suggesting that nutrients played an important role in governing their vertical distribution (Drira et al., Reference Drira, Bel Hassen, Hamza, Rebai, Bouain, Ayadi and Aleya2009).
The bloom of cyanobacteria in the Gulf of Gabes during May–June 2006 was associated with the lowest ammonium concentration decreasing with depth. For this reason, we suggest that the blooms of cyanobacteria consumed the full amount of available ammonium. In fact, we recorded a weak correlation between cyanobacteria with the NH4 + amount (r = −0. 202, P < 0.05, df = 28). During May–June 2006, we noted the dominance of Pseudoanabaena sp. and Anabaena sp. in substantial amounts between the inshore and offshore areas. In the Gulf of Gabes, the summer season is characterized by high temperature and salinity which promote the stratification of the water column and induce the appearance of phytoplankton blooms with prominent presence of such cyanobacterial taxa. The phenomenon of bloom occurred frequently in several regions of the Gulf of Gabes: near Kerkennah Islands in offshore areas (Hamza & Ben Maiz, Reference Hamza and Ben Maiz1990; Fremy & Feldmann, Reference Fremy and Feldmann1935) and in Bougrara Lagoon and inshore areas due to high overgrowth of Trichodesmium erytreum (Turki et al., Reference Turki, Harzallah and Sammari2006). This important proliferation of Trichodesmium erytreum, which is termed a harmful species, was responsible for algal bloom-induced large-scale mortality and stranding of large quantities of fish, eels, cuttlefish, etc. in the southern Tunisian coasts (Hamza & Ben Maiz, Reference Hamza and Ben Maiz1990; Turki & El Abed, Reference Turki and El Abed2001; Hansen et al., Reference Hansen, Erard-Le Denn, Daugbjerg and Rodriguez2004; Turki et al., Reference Turki, Harzallah and Sammari2006).
CONCLUSION
In conclusion, the nutrient availability and the water column stability seemed to be among the major factors determining cyanobacteria dynamics in the Gulf of Gabes. Indeed, cyanobacteria were prominent near the thermocline during the period of strong stratification. Cyanobacteria were found throughout the water column and were more concentrated at the thermocline within the deeper zone. The bloom of cyanobacteria recorded in May–June 2006 was promoted by low N/P-ratios. The factor limiting the diazotrophic activity during the study period (2005–2007) was shown to be variable in the Gulf of Gabes: phosphorus as DIP was shown to be the limiting nutrient of N2 fixation and strongly influenced the cyanobacterial bloom dynamics.
ACKNOWLEDGEMENTS
The authors wish to thank the crew of the RV ‘Hannibal’ for their assistance.
FINANCIAL SUPPORT
This work was supported by the Tunisian funded project POEMM (LR02INSTM04) conducted in the National Institute of Marine Sciences and Technologies (INSTM) and Plankton and Microbiology of Aquatic Ecosystems Research Unit of the University of Sfax.
