Introduction

During recent decades, many coral reefs have shifted from coral to algal dominance (Done 1992; McManus and Polsenberg 2004; Hughes et al. 2007; Cheal et al. 2010). The algal-dominated state often consists of macroalgae, but small turf algae have also increased (Barott et al. 2009; Vermeij et al. 2010). Algal turf communities (henceforth called ‘turfs’) are heterogeneous assemblages of Chlorophyta, Phaeophyta, Rhodophyta, and filamentous Cyanobacteria (Steneck and Dethier 1994; Diaz-Pulido and McCook 2002; Fricke et al. 2011). Turfs are important primary producers on reefs that are intensively grazed (Adey and Goertemiller 1987; Carpenter and Williams 2007), compete for space against corals (Barott et al. 2009; Vermeij et al. 2010), provide shelter for various meiofauna (Zeller 1988; Logan et al. 2008), and may play a key role in the recycling of nutrients (Charpy-Roubaud et al. 2001; Charpy et al. 2007; Fricke et al. 2011; Charpy et al. 2012). Turfs currently represent the single most dominant benthic component on many coral reefs in the Caribbean (Kramer 2003; Vermeij et al. 2010), Central Pacific (Barott et al. 2009, 2012), Red Sea (Haas et al. 2010), and Indonesia (Wangpraseurt et al. 2012).

Many of the cyanobacteria in turfs are capable of nitrogen (N2) fixation (Charpy et al. 2012). Nitrogen is one of the key elements often limiting primary productivity on coral reefs (Littler et al. 1991; Delgado and Lapointe 1994; Larned 1998; Den Haan et al. 2013), and hence, N2 fixation provides access to an important additional N source. N2 fixation is very sensitive to oxygen, and diazotrophic cyanobacteria evolved various strategies to accommodate the incompatible processes of oxygenic photosynthesis and N2 fixation (e.g., Bergman et al. 1997; Zehr 2011). Some cyanobacteria developed specialized cells known as heterocysts to separate N2 fixation from oxygen production, and these heterocystous species often show higher N2 fixation activity during daytime than at night (Mullineaux et al. 1981; Staal et al. 2002). In contrast, many non-heterocystous diazotrophic cyanobacteria fix N2 during the night, fuelled by respiration of the carbohydrates built up by photosynthesis during the preceding daytime (Schneegurt et al. 1994; Dron et al. 2012; Brauer et al. 2013). The N2 fixed by cyanobacteria may be released as dissolved inorganic and organic N through leakage from the cells, cell death, or lysis (Berman and Bronk 2003; Benavides et al. 2013) and thereby becomes available for other organisms in the community (Mulholland et al. 2006; Agawin et al. 2007).

Eutrophication of coastal waters is a major problem in many densely populated (sub)tropical regions (Nyström et al. 2000) and is often associated with a phase shift from coral to algal dominance (Done 1992; Hughes 1994; Cheal et al. 2010). To what extent eutrophication may affect the relative importance of N2 fixation on coral reefs is still an open question. Generally speaking, N enrichment may alleviate ecosystems from N limitation, which would diminish the need for N2 fixation, and hence decrease the abundance of N2-fixing cyanobacteria (Smith 1983; Conley et al. 2009). However, if N continues to be limiting (e.g., if externally supplied N is utilized as quickly as it is supplied), then N2 fixation will not be suppressed and N2-fixing organisms may still prevail. Indeed, lake studies have shown that N2 fixation rates can be high in eutrophic waters, especially during episodes with a relative shortage of N (Howarth et al. 1988; Scott et al. 2009; Paerl 2009).

N2 fixation on coral reefs has been studied for almost 40 yrs (Webb et al. 1975; Wiebe et al. 1975). Most studies have focused on N2 fixed by solitary or mat-forming benthic cyanobacteria (e.g., O’Neil and Capone 1989; Charpy-Roubaud et al. 2001; Charpy et al. 2007, 2010), whereas only a few studies have investigated N2 fixation in turfs (Wilkinson and Sammarco 1983; Wilkinson et al. 1984; Larkum et al. 1988; Williams and Carpenter 1997, 1998; Davey et al. 2008). Little is known about the major environmental determinants of N2 fixation by turfs. For instance, it is not known to what extent depth-dependent variation in cyanobacterial composition and light conditions affect the N2 fixation activity of the turfs. Furthermore, it is not known whether turfs fix N2 mainly during the day or night, or whether N2 fixation by turfs is affected by the degradation of coral reefs associated with coastal eutrophication. The main goals of this study were therefore (1) to quantify N2 fixation by turfs in relation to water depth and the day–night cycle and (2) to compare N2 fixation by turfs on a degraded reef near a densely inhabited area versus a less degraded coral reef at a more remote site. We compared our results with published data from solitary cyanobacteria, mat-forming cyanobacteria, and turfs on reefs elsewhere in the world.

Materials and methods

Research sites and composition of turfs

This study was conducted on the leeward side of the island of Curaçao, Southern Caribbean, during June–August of 2011. We investigated two sites: ‘Buoy 0’ (12°7′29.07″N, 68°58′22.92″W) and ‘Playa Kalki’ (12°22′30.9″N, 69°9′31.49″W; Fig. 1). Buoy 0 is a degraded reef only a few kilometres downstream of the capital Willemstad with its industrial harbour and near the outlet of the eutrophied Piscadera Bay. Over the past 40 yr, coral cover of the shallow reefs at Buoy 0 has decreased dramatically (Bak et al. 2005). Playa Kalki is located ~35 km from the major population centre of Willemstad. It is a less degraded reef with higher hard coral cover, and lower algal and cyanobacterial cover, and lower dissolved nutrient concentrations than at Buoy 0 (see Results).

Fig. 1
figure 1

Map indicating research sites Buoy 0 and Playa Kalki on Curaçao, Southern Caribbean (12°10′N, 68°58′W). Dark grey zones are urbanized areas, including the capital Willemstad and its industrial harbour (hatched area)

Full size image

Turfs on the reefs at Buoy 0 consist of a mixed community of algae belonging to the Chlorophyta, Rhodophyta, Phaeophyta, and Cyanobacteria (Fricke et al. 2011). This study found that at 5-m depth, the turf community at Buoy 0 was dominated by Cyanobacteria, notably the heterocystous species Dichothrix utahensis (Fricke et al. 2011). At 15-m depth, the relative abundances of Cyanobacteria, Chlorophyta, Rhodophyta, and Phaeophyta were more evenly distributed (Fricke et al. 2011). At 25-m depth, Cyanobacteria represented ~40 % of the turf community, dominated by the non-heterocystous cyanobacteria Lyngbya spp. and Oscillatoria spp., whereas the heterocystous species D. utahensis was rare (Fricke et al. 2011). At Playa Kalki, we observed the same N2-fixing cyanobacteria as at Buoy 0 (i.e., with Dichothrix spp. near the surface, and Lyngbya spp. and Oscillatoria spp. in deeper parts of the reef).

Reef characteristics

Surveys comparing the dominant benthic communities at both Buoy 0 and Playa Kalki were conducted along 100-m transects, placed alongside the 5, 10, 20, and 30-m isobath. Benthic cover of hard corals, macroalgae, turfs, and cyanobacterial mats was determined from photographs of 60 quadrats (1.5 m2) randomly placed along both sides of the transects. The photographs were analysed using the program Coral Point Count with Excel Extensions (CPCe; Kohler and Gill 2006).

Light profiles were measured with a Hydrolab DS5 Sonde (OTT Messtechnik GmbH & Co., Kempten, Germany; n = 5 at both Buoy 0 and Playa Kalki). Furthermore, water samples for nutrient analysis were taken along each horizontal transect at 10 cm above the reef using a 50-ml Terumo syringe (Terumo Europe, Leuven, Belgium; n = 9 at both Buoy 0 and Playa Kalki). Water samples were immediately filtered at the diving site using 0.22-µm Acrodisc filters and stored in 6-ml polyethylene vials (PerkinElmer, MA, USA) at −20 °C until further analysis. Concentrations of NO3 − (Grasshoff et al. 1983), NH4 + (Helder and De Vries 1979), and PO4 3− (Murphy and Riley 1962) were analysed at the Royal Netherlands Institute for Sea Research (NIOZ), Texel, the Netherlands, using continuous flow analysis via a Quatro auto-analyzer (Seal Analytical, UK).

Sampling and incubation of turfs

The N2 fixation measurements required controlled incubation of turf samples. For this purpose, turfs were not collected directly off the reef, because scraping turfs off the rocks would damage the algal tissue and the rock underneath. Instead, turfs were grown on the exterior of 1.5-L square plastic bottles (FIJI Water Company, CA, USA), which were placed inside 1-m3 chicken-wired cages (mesh ∅2.5 cm). Growing turfs on plastic bottles ensured that the turfs were of approximately similar size and age and resulted in minimal damage to the live tissue of the turfs. Encaging prevented grazing by large herbivores (i.e., >2.5 cm in diameter) to ensure sufficient turf growth on the bottles. In May 2011, the encaged bottles were positioned at depths of 5, 10, 20, and 30 m using SCUBA diving. The bottles were placed 0.5 m above the reef to avoid overgrowth by benthic cyanobacterial mats or macroalgae and remained on the reefs of Buoy 0 and Playa Kalki for 63–79 d and 49–77 d, respectively. This growth period provided ample time for the establishment of turf communities with benthic cyanobacteria.

Plastic bottles were retrieved from each depth (5, 10, 20, and 30 m) between 23 June and 3 August 2011. Small strips (1.5 × 2.0 cm; six strips per depth) containing turfs were cut out of the plastic bottles using scissors. In the laboratory, each strip was placed inside a 10-ml headspace crimp vial (VWR, the Netherlands) containing 7 ml of 0.22 µm filtered ambient seawater (Whatman cellulose acetate membrane filters, ∅25 mm). Vials were then sealed with a gastight liner and an aluminium cap (Agilent Technologies, Santa Clara, CA, USA) using a vial crimper and stored in a flow-through aquarium to keep the turfs at the same temperature as on the reef (27–29 °C). The average light intensity inside the aquarium, measured with a Hydrolab DS5 Sonde (OTT Messtechnik GmbH & Co., Kempten, Germany), was ~100 µmol photons m−2 s−1 during the daytime. This light level is comparable to light intensities measured on the reefs at 20–30-m depth during sunny days. However, it is below the saturation light intensity of 200 µmol photons m−2 s−1 of the in situ photosynthetic activity of the turfs, measured according to Ralph and Gademann (2005) with a Diving-PAM/B Underwater Fluorometer (Walz Mess- und Regeltechnik, Effeltrich, Germany; J. Den Haan, unpublished results). The vials were acclimated inside the flow-through aquarium until the acetylene reduction assay (ARA, see below) commenced at 1900 hrs on the day of collection.

We incubated six strips per depth and per research site, each in a separate vial. In addition, we used one clean strip (without turfs) per depth and per research site as a negative control. Furthermore, the mat-forming cyanobacterium Oscillatoria spp. (n = 6) was collected off sandy sediments at Buoy 0 and solitary tufts of Symploca spp. (n = 6) were collected from rocks at Playa Kalki, both at 5-m depth. Oscillatoria and Symploca spp. are known to be active N2 fixers (Charpy et al. 2010), and their N2 fixation rates were compared to the turfs.

Acetylene reduction assay

The ARA was used to measure the nitrogenase activity of the N2-fixing organisms residing in the turfs (Stewart et al. 1967; Charpy et al. 2007). Briefly, 2.5 ml of acetylene gas was injected into the vials using a 2.5 MR-GT gastight syringe (SGE Analytical Science, UK) giving an acetylene concentration in the headspace of 33 % (Stal 1988; Zuckermann et al. 1997). Directly after injecting the acetylene, which was always at 1900 hrs (sunset), a 1-ml gas sample was taken from the vial using the same syringe to correct for background ethylene. Subsequent samples were taken after 12 h (at 0700 hrs; sunrise) and after 24 h (at 1900 hrs the next day) to quantify N2 fixation during night-time and the full 24-h period, respectively. Daytime N2 fixation was calculated by subtracting the night-time N2 fixation from the total N2 fixation after 24 h. Once a gas sample was taken from a vial, the sample was immediately transferred into a 6-ml Vacuette (Greiner Bio-One, The Netherlands), which was completely filled with a saturated salt (NaCl) solution, to properly conserve the gas samples until further analysis.

The ethylene content of the gas samples was analysed with a gas chromatograph (GC-14B, Shimadzu, Kyoto, Japan), using acetylene as an internal standard (Stal 1988). The gas chromatograph was equipped with a flame ionization detector and a 3-m Porapak R column (2 mm inner diameter; Chrompack). The temperatures of the injector and detector were set to 70 and 90 °C, respectively. From the headspace of each Vacuette, 0.1 ml was injected into the gas chromatograph using a gastight disposable syringe. The retention time for ethylene (C2H4) and acetylene (C2H2) were 1.9 and 2.7 min, respectively. Ethylene production was converted to N2 fixation using a conversion factor of four (Stal 1988; Mulholland et al. 2004).

Chlorophyll a analysis

After 24 h of incubation, the plastic strips were removed from the vials to determine the amount of chlorophyll a as a measure of turf biomass. Chlorophyll a was extracted by grinding the samples in 10 ml of 100 % methanol using a mortar and pestle in an ice bath and under dim light. Extraction was continued for 24 h in a refrigerator (4 °C) inside test tubes covered with aluminium foil. Subsequently, the extracts were centrifuged for 15 min at 3,500 rpm to separate the plastic particles from the extraction solvent. The chlorophyll a concentration of the extracts was measured on a T60 UV/Vis spectrophotometer (PG Instruments Ltd, Wibtoft, UK) according to Porra et al. (1989).

Statistical analysis

We used a two-way analysis of variance to test whether the measured nutrient concentrations, turf cover and turf biomass varied significantly with depth (5, 10, 20, 30 m) and research site (Buoy 0, Playa Kalki). A three-way analysis of variance was applied to test whether N2 fixation rates varied significantly with depth, research site, and time of day (day vs. night). We used type III sum of squares to account for unbalanced data, because some strips were damaged or lost during the N2 fixation analysis. The dependent variables were log-transformed if this improved the homogeneity of variance, as tested by Levene’s test. Post hoc comparisons of the means were based on Tukey’s HSD test using a significance level (α) of 0.05.

Results

Reef characteristics

The coral reef ecosystem at Buoy 0 had a significantly lower cover of corals and the macroalga Dictyota spp. than Playa Kalki and a significantly higher cover of the macroalga Lobophora variegata and cyanobacterial mats (Table 1). The water at Buoy 0 was slightly but significantly more turbid than at Playa Kalki, as revealed by comparison of the slopes of the log-transformed light profiles (ANCOVA, F 1,59 = 20.176, p < 0.001; Fig. 2a). NO3 − concentrations were significantly higher at Buoy 0 than at Playa Kalki at 5- and 10-m depth, but not at 20- and 30-m depth (Figs. 2b, d; Table 2). PO4 3− concentrations were significantly higher at Buoy 0 than at Playa Kalki at 5-m depth only (Fig. 2d; Table 2). NH4 + concentrations were similar at both sites (Fig. 2c; Table 2). The ambient N/P ratio of 21.2 ± 1.7 (mean ± SE) at Buoy 0 was significantly higher than the N/P ratio of 15.3 ± 1.0 at Playa Kalki (Table 2).

Fig. 2
figure 2

Light and nutrient conditions. a Light profiles at Buoy 0 (open diamonds with dashed trend line) and Playa Kalki (circles with solid trend line). b Nitrate, c ammonium, and d phosphate concentrations in water samples from different depths at Buoy 0 (dark grey bars) and Playa Kalki (light grey bars). Error bars represent SE of the means. Bars that do not share the same letter are significantly different, as tested by two-way analysis of variance followed by post hoc comparison of the means (Table 2; ESM Tables 1 and 2)

Full size image
Table 1 Comparison of percentage cover by hard corals, macroalgae (dominated by Lobophora variegata and Dictyota spp.), turfs and cyanobacterial mats at research sites Buoy 0 and Playa Kalki (mean ± SE)
Full size table
Table 2 Two-way analysis of variance of nutrient concentrations, with depth and location as independent variables
Full size table

Algal turfs

Turf cover on the reef decreased significantly with depth at both Buoy 0 and Playa Kalki (Fig. 3a; Table 3). We did not find systematic differences in turf cover between the two research sites. Turf cover at 5-m and 30-m depth was significantly higher at Buoy 0, whereas turf cover at 10 m was significantly higher at Playa Kalki. A similar pattern was found for turfs collected from the plastic bottles incubated on the reef, which showed a significant decrease in turf biomass (expressed as mg Chl a m−2) with depth that was comparable at both research sites (Fig. 3b; Table 3).

Fig. 3
figure 3

a Percentage cover by algal turfs on the reefs (based on n = 60 quadrates per depth and research site). b Biomass of algal turfs growing on incubated plastic bottles (n = 6 bottles per depth and research site). Data are from the research sites Buoy 0 (dark grey bars) and Playa Kalki (light grey bars). Error bars represent SE of the means. Bars that do not share the same letter are significantly different, as tested by two-way analysis of variance followed by post hoc comparison of the means (Table 3; ESM Tables 3 and 4)

Full size image
Table 3 Two-way analysis of variance of the turf cover on the reef and turf biomass on the plastic bottles, with depth and location as independent variables
Full size table

N2 fixation by turfs

N2 fixation by turfs did not differ significantly between Buoy 0 and Playa Kalki (Fig. 4; Table 4). However, we found significant main effects of depth and time (day vs. night) on N2 fixation by turfs (Table 4). More specifically, post hoc comparison of the means showed that daytime N2 fixation was significantly higher than night-time N2 fixation at each of the four depths (Fig. 4). Furthermore, during daytime, the post hoc comparison showed a tendency for higher N2 fixation activity at 5- and 10-m depth than at 30-m depth at a significance level of p < 0.10, but did not reveal significant differences at p < 0.05 between depths (Fig. 4; for details of the post hoc comparison, see Electronic Supplementary Material, ESM, Table 5). Averaged across all four depths and both research sites, the mean N2 fixation by turfs was 17.4 ± 2.8 (mean ± SE) mg N2 m−2 12 h−1 during daytime and 2.9 ± 0.4 mg N2 m−2 12 h−1 at night. Combined, this resulted in a total daily N2 fixation of 20.3 ± 2.9 mg N2 m−2 d−1.

Fig. 4
figure 4

N2 fixation by turfs at Buoy 0 (dark grey bars) and Playa Kalki (light grey bars) as function of depth. a Daytime N2 fixation (0700–1900 hrs); b night-time N2 fixation (1900–0700 hrs). Error bars represent SE of the means. Bars that do not share the same letter are significantly different, as tested by three-way analysis of variance followed by post hoc comparison of the means (Table 4; ESM Table 5). Because the rate of N2 fixation was not significantly different between Buoy 0 and Playa Kalki, both research sites are aggregated in the post hoc comparison as indicated by the braces

Full size image
Table 4 Three-way analysis of variance of the N2 fixation rate by turfs (mg N2 m−2 turf 12 h−1), with depth, location (Buoy 0 vs. Playa Kalki) and time of day (day vs. night) as independent variables
Full size table

For comparison, we also measured N2 fixation by two benthic non-heterocystous cyanobacteria collected at 5 m. Mat-forming Oscillatoria spp. collected off sandy sediments at Buoy 0 fixed substantially more N2 than the turfs, with a N2 fixation of 153.0 ± 60.3 mg N2 m−2 12 h−1 during daytime and 13.7 ± 2.8 mg N2 m−2 12 h−1 at night. Solitary-growing Symploca spp. collected from rocks at Playa Kalki also showed high N2 fixation activity, of 66.9 ± 45.7 mg N2 m−2 12 h−1 during daytime and 4.9 ± 2.7 mg N2 m−2 12 h−1 at night.

We calculated the total N2 fixation of turfs on a reef-wide scale as the product of their daily N2 fixation and the cover of turfs on the coral reef. This revealed that total N2 fixation at the reef scale was higher in the top 10 m than at greater depths, but did not differ between the degraded reef at Buoy 0 and the less degraded reef at Playa Kalki (Fig. 5).

Fig. 5
figure 5

Total daily N2 fixation on 100 m2 of reef at Buoy 0 (dark grey bars) and Playa Kalki (light grey bars). Error bars represent SE of the means

Full size image

Discussion

N2 fixation in algal turf communities

Our results show substantial N2-fixation activity in algal turf communities on the reefs of Curaçao, extending to a depth of at least 30 m. At the reef scale, the highest N2 fixation activity by turfs was in the upper 10 m, which was mainly due to the higher cover of turfs in these shallow parts of the reef. The strong proliferation of turfs in shallow waters was confirmed by the high turf biomass that developed on the plastic bottles incubated within the upper 10 m of the water column. The incubated bottles were protected from grazing by herbivores larger than 2.5 cm, and therefore, the prevalence of turfs in shallow waters was probably not due to depth-dependent variation in grazing rates of mid- and large-sized herbivores. Several studies have pointed out that turfs have high light requirements (e.g., Carpenter 1985; Klumpp et al. 1987), and therefore at greater depth, they have difficulty competing with other algae (Steneck and Dethier 1994). This would offer a plausible explanation for the observed depth distribution of turfs and is also consistent with the higher abundances of turfs in the shallower parts of other reefs (Van den Hoek et al. 1975; Adey and Goertemiller 1987; Steneck and Dethier 1994).

We observed a much higher N2 fixation activity during the day than at night. This contrasts with several studies of microbial mats in the temperate zone, which typically show higher night-time N2 fixation activity (Villbrandt et al. 1990; Severin and Stal 2008). However, cyanobacterial mats dominated by heterocystous species showed higher levels of N2 fixation during daytime on coral reefs in St. Croix and New Caledonia (Williams and Carpenter 1997; Charpy-Roubaud et al. 2001; Charpy et al. 2007). Heterocystous cyanobacteria were also the main diazotrophs at our sites, especially in turfs growing in the upper 10 m of the reef, where Dichothrix utahensis dominated the turf communities (Fricke et al. 2011). Although most heterocystous cyanobacteria can fix N2 in the dark, they usually exhibit higher N2 fixation activity during daytime (Mullineaux et al. 1981; Staal et al. 2002). Hence, the dominance of heterocystous cyanobacteria in the turfs likely explains the observed higher daytime N2 fixation.

The data were obtained by growing turf algae on artificial substrates incubated at different depths. Main advantages of this approach are (1) that the plastic bottles offered controlled and replicated conditions, as the turfs were grown on an identical substrate and were of similar size and age, (2) damage to algal tissues during sampling was minimal, and (3) the plastic strips could be easily handled during subsequent laboratory analyses. However, the use of an artificial substrate might have selected turf communities different from those found on the surrounding reef. We observed a similar composition of algae and cyanobacteria on the bottles as described by Fricke et al. (2011) for colonized turf algae on ceramic tiles in the same study area and after similar incubation times. The bottles were incubated for more than 7 weeks prior to sampling, which ensured the establishment of turf communities comprising all major taxa that were also observed in natural turfs, including Chlorophyta, Rhodophyta, Phaeophyceae, and Cyanobacteria (J. den Haan, pers. obs.). Hence, although future studies may consider a more detailed taxonomic analysis, the turf communities that developed on the bottles seemed representative of the natural turf assemblages on the reefs.

Non-heterocystous diazotrophic cyanobacteria, including Lyngbya spp. and Oscillatoria spp., prevailed in the turfs at 20–30-m depth (Fricke et al. 2011). Most non-heterocystous cyanobacteria avoid the problem of oxygen inhibition of nitrogenase by fixing nitrogen during the night (Gallon 1992; Bergman et al. 1997). Interestingly, however, in our study, the non-heterocystous cyanobacteria Oscillatoria spp. and Symploca spp. showed highest rates of N2 fixation during the day. In a laboratory study, Fredriksson et al. (1998) demonstrated that the nitrogenase activity of Symploca PCC 8002 was indeed highest during daytime, consistent with our field observations.

In addition to cyanobacteria, many other Bacteria and Archaea are also capable of N2 fixation and diazotrophic microbial communities are known to be highly diverse (Zehr et al. 1995; Olson et al. 1999; Severin and Stal 2010). Hence, it is possible that not only cyanobacteria but also other bacterial groups contributed to the N2 fixation activity that we observed in algal turfs at Curaçao. However, the observation that the N2 fixation activity was much higher during daytime than at night suggests that phototrophic organisms played a major role. Given the high abundance of diazotrophic cyanobacteria, it is therefore likely that most N2 fixation activity in the turfs can be attributed to cyanobacteria.

Comparison with coral reefs around the world

The rates of N2 fixation in the benthic cyanobacterial mats of Oscillatoria spp. and the solitary tufts of Symploca spp. on the reefs of Curaçao were higher than those of the turfs and are also high compared to those of turfs, cyanobacteria, and bacteria on coral reefs in other parts of the world (Table 5). Similarly high N2 fixation rates by tropical benthic cyanobacteria have been reported from the Marshall Islands (Wiebe et al. 1975), the Red Sea (Shashar et al. 1994), La Réunion (Casareto et al. 2008), and the Ryukyu Islands (Casareto et al. 2008). These findings are consistent with a recent review of benthic N2 fixation on coral reefs, which revealed that microbial (cyanobacterial) mats tend to have higher rates of N2 fixation than other benthic reef components (Cardini et al. 2014).

Algal turf communities comprise a mixture of diazotrophic and non-diazotrophic organisms, and it is therefore not surprising that their rates of N2 fixation were lower than those of the cyanobacteria Oscillatoria spp. and Symploca spp. (Table 5). The N2 fixation rates of the turfs in our study are in a similar range as those reported for other algal turf communities (Table 5). We note that the high N2 fixation rates of turfs reported by Carpenter et al. (1991) and Williams and Carpenter (1998) were associated with high flow rates, which tend to enhance N2 fixation activity, whereas our study was performed under stagnant conditions. Although turfs tend to have lower N2 fixation rates than cyanobacterial mats, algal turfs are likely to play an important role in the nitrogen budget of coral reefs. In particular, cyanobacterial mats and tufts are of a relatively ephemeral nature (Nagle and Paul 1998; Kuffner and Paul 2001; Albert et al. 2005; Paul et al. 2005), whereas algal turfs are present throughout the year. Moreover, at least in our study area, algal turfs cover a much larger area of the reef than cyanobacterial mats (Table 1).

Table 5 Nitrogen fixation rates by turfs and benthic (cyano-) bacteria from other coral reefs around the world
Full size table

Relationship with coastal eutrophication

Several decades ago turfs were already recognized as important primary producers and potential N2 fixers on coral reefs of the Caribbean (e.g., Adey and Goertemiller 1987). Since then, many Caribbean reefs have shown a decrease in coral cover (Gardner et al. 2003; Burke et al. 2011) and are now often dominated by turfs (Kramer 2003; Vermeij et al. 2010). Our results show that the coastal waters of Buoy 0 are more eutrophic than those of Playa Kalki. This conclusion is based on the fact that Buoy 0 shows significantly higher concentrations of nitrate and phosphate in the upper layers of the water column, a significantly higher N/P ratio, and its water is significantly more turbid in comparison with the more remote site of Playa Kalki. Moreover, Buoy 0 has a lower cover of hard corals and higher cover of the macroalga L. variegata and cyanobacterial mats than Playa Kalki. Over the past 40 yr, the live coral cover of shallow reefs in the vicinity of Buoy 0 has decreased by more than 50 % (Bak et al. 2005).

One might hypothesize that eutrophication would result in lower rates of N2 fixation. High N availability (especially in the form of ammonium) usually decreases N2 fixation, because the assimilation of combined N is energetically cheaper than acquiring N via the energy-costly N2 fixation process (Zevenboom et al. 1981; Holl and Montoya 2005; Brauer et al. 2013). N enrichment indeed decreased N2 fixation of benthic cyanobacteria in a field study on One Tree Island, Great Barrier Reef (Koop et al. 2001). Yet, in our study, we did not find differences in the rate of N2 fixation between the turfs of Buoy 0 and Playa Kalki, despite the different trophic status of the two reef systems. Similarly, no relationship was found between N2 fixation by benthic cyanobacteria and N enrichment from sewage discharge in the Kaneohe Bay, Hawaii (Hanson and Gundersen 1976). An explanation for the high N2 fixation activity in these degraded reefs is that the utilization of N might be close to its supply rate, thus maintaining low concentrations of dissolved inorganic N in the overlying water column. The dense cover by turfs, macroalgae, and cyanobacterial mats may enable effective removal of externally supplied N. In addition, phytoplankton above the reefs may also rapidly assimilate N (Van Duyl et al. 2002; Furnas et al. 2005), and denitrification may contribute to N losses from the system (Gruber and Sarmiento 1997; Deutsch et al. 2007). Even at Buoy 0, near the city of Willemstad and its industrial harbour, the concentrations of ammonium and nitrate remained below 1 μmol L−1, which is sufficiently low for continued N2 fixation by marine diazotrophic cyanobacteria (Mulholland et al. 2001; Agawin et al. 2007). These results mirror observations in eutrophic lakes, which also show high N2 fixation rates by cyanobacteria during periods of low N availability, despite the high nutrient loads to these systems (Howarth et al. 1988; Scott et al. 2009).

Previously, we showed that the photosynthetic activity of the macroalga L. variegata is co-limited by N and P at both Buoy 0 and Playa Kalki (Den Haan et al. 2013), which is in agreement with the low concentrations of dissolved inorganic N and P at a N/P ratio close to the Redfield ratio of 16:1. Co-limitation by N and P has also been reported for primary producers in other marine ecosystems (Elser et al. 2007), including macroalgae on the Great Barrier Reef (Schaffelke and Klumpp 1998; Schaffelke 1999). The high N2 fixation activity in combination with low ambient N and P concentrations and co-limitation by N and P indicates that, even at Buoy 0, the N input from N2 fixation and land-derived sources does not yet meet ecosystem N demands. Hence, enhanced inputs of N and P will most likely lead to further expansion of cyanobacterial and algal communities. Dual reduction in both N and P loads from terrestrial sources may offer a valuable management strategy (Paerl 2009; Lewis et al. 2011) that may help to curtail further degradation of these valuable reef ecosystems.