Untreated sewage outfalls do not promote Trichodesmium blooms in the coasts of the Canary Islands

During the summer of 2017, recurrent extensive blooms of the diazotrophic cyanobacterium Trichodesmium invaded the beaches and coastal waters of the Canary Islands, causing great social alarm. Some local media and public sectors ascribed, without any strong scientific evidence, the origin and reactivation of these blooms to untreated sewage outfalls distributed along the coasts. In order to test whether sewage outfalls could have any influence on the metabolic activity of Trichodesmium, we performed 13C and 15N2 uptake experiments with colonies experiencing three different bloom development stages, incubated both with clear seawater and sewage water from an outfall south of Gran Canaria island. Our results showed that sewage outfalls did not promote any increase in dinitrogen (N2) fixation in Trichodesmium, supporting the hypothesis that decaying blooms were generated offshore and transported shoreward by local currents and winds, accumulating mostly leeward of the islands. The combination of unusually warm seawater temperatures, enhanced and sustained stratification of the upper water column and recurrent dust deposition events would have favored the development of the Trichodesmium blooms, which lasted for at least four months.


Mar Benavides 1 & Javier Arístegui 2*
During the summer of 2017, recurrent extensive blooms of the diazotrophic cyanobacterium Trichodesmium invaded the beaches and coastal waters of the Canary Islands, causing great social alarm. Some local media and public sectors ascribed, without any strong scientific evidence, the origin and reactivation of these blooms to untreated sewage outfalls distributed along the coasts. In order to test whether sewage outfalls could have any influence on the metabolic activity of Trichodesmium, we performed 13 C and 15 N 2 uptake experiments with colonies experiencing three different bloom development stages, incubated both with clear seawater and sewage water from an outfall south of Gran Canaria island. Our results showed that sewage outfalls did not promote any increase in dinitrogen (N 2 ) fixation in Trichodesmium, supporting the hypothesis that decaying blooms were generated offshore and transported shoreward by local currents and winds, accumulating mostly leeward of the islands. The combination of unusually warm seawater temperatures, enhanced and sustained stratification of the upper water column and recurrent dust deposition events would have favored the development of the Trichodesmium blooms, which lasted for at least four months.
Trichodesmium is a colonial filamentous cyanobacterium capable of atmospheric dinitrogen (N 2 ) fixation. It abounds in subtropical and tropical oceanic waters, often forming blooms large enough to be detected by satellite images 1 , rendering a source of fixed nitrogen that fuels primary production significantly 2 . Its growth may be limited however by the availability of iron 3 and phosphorus [4][5][6] . In the open waters of the subtropical North Atlantic Ocean, the seasonal meridional shift of the North African dipole controls the deposition of iron-rich Saharan dust on surface waters 7 . Dust events are known to enhance the proliferation and diazotrophic activity of Trichodesmium leading to the formation of massive blooms [8][9][10][11] , with colonies adopting three-dimensional shapes known as "puffs" or "tufts". Their organic matter coating as well as the microbial epibiont community provide Trichodesmium colonies the ability to degrade atmospheric dust particles to obtain the iron they contain 12,13 , which is likely not possible for free single Trichodesmium filaments 14 .
Although it is not clear what mechanism, or combination of mechanisms, triggers the development of Trichodesmium blooms, warm (> 20 °C) and stratified waters as well as the availability of phosphorus and iron, are essential requirements to sustain extensive blooms 15 . Under such conditions, Trichodesmium develops swiftly, fixing high amounts of carbon and nitrogen 16 . This intense activity may exhaust limiting nutrients, forcing Trichodesmium filaments to release exopolymeric substances while aggregating and forming dense colonies 17 . This strategy optimizes nutrient uptake and repartition of resources between filaments of a same colony 17,18 . When all limiting resources are depleted, Trichodesmium cells induce apoptosis, also referred to as programmed cell death (PCD) 19 , a process that can make Trichodesmium blooms almost disappear in a matter of hours to days 20 .
Traditionally considered an autotroph 16 , Trichodesmium is now known to use dissolved organic matter (DOM) molecules which may provide it with alternative nutrient resources when their inorganic forms are unavailable or their use becomes too energetically demanding. Trichodesmium is able to use dissolved organic phosphorus (DOP) forms including phosphoesters and phosphonates when inorganic phosphate availability is low 5,6 . The RuBisCo enzyme of Trichodesmium has a low affinity for CO 2 21 , which is thought to explain its enhanced carbon fixation when exposed to high CO 2 levels and/or lower pH 22,23  www.nature.com/scientificreports/ from the Southwest Pacific Ocean have been observed to obtain as much carbon via amino acids as via CO 2 24 . Cultures of Trichodesmium have been reported to grow faster on combined nitrogen forms such as urea than on N 2 25 . More recent laboratory experiments have shown that when grown under limiting concentrations of iron and/or phosphorus, Trichodesmium downregulates N 2 fixation and obtains nitrogen from trimethylamine or ammonium 26 . Collectively, these studies suggest that Trichodesmium adopts a mixotrophic nutrition mode when its basic metabolic activities are reduced due to environmental stress or a limitation of resources. Mixotrophy thus confers Trichodesmium metabolic plasticity and adaptation to dynamic environmental nutrient scenarios.
Throughout the summer of 2017, recurrent blooms of Trichodesmium were observed in coastal and offshore waters around the Canary Islands 27 . These blooms appeared first in the western side of the archipelago, coinciding with dense Saharan dust deposition events, enhanced water column stratification and seawater temperatures > 23 °C, later spreading to the eastern side of the archipelago, when those conditions extended eastwards 27 . The massive accumulations of Trichodesmium along the coasts of the Canary Islands (see Fig. 1) caused great social alarm, and were attributed by some media and public sectors to uncontrolled sewage outfalls 28 , for which the Spanish Government has been sued by the European Commission 29 . The attribution was partly based on a recent coastal study in the eastern Mediterranean Sea, that reported an increase in the abundance and N 2 fixation activity of Trichodesmium after a sporadic outburst of a local wastewater treatment plant 30 . In this study the authors suggested that, due to the low N 2 fixation activity measured, Trichodesmium could have shifted to a mixotrophic nutrition using wastewater-derived DOM compounds. Although the magnitude of the bloom in the Canary Islands was orders of magnitude larger in terms of density compared to that of the Mediterranean Sea 26 , we wanted to verify whether wastewater outfalls could have promoted or helped maintain Trichodesmium blooms in the nearshore waters of the Canary Islands. With this aim, we collected Trichodesmium colonies at different stages of development of the bloom (Fig. 2), and performed incubations with clear seawater and wastewater from an urban sewage outfall to the south of Gran Canaria Island and examined its impact on N 2 and carbon fixation rates.

Results and discussion
We incubated Trichodesmium from three different bloom development stages (sparse, slick and PCD-like, Fig. 2) with clear and sewage waters. Our experiments showed that N 2 fixation rates in clear water incubations were not significantly different from those of sewage incubations (Fig. 3, t-test p = 0.98), which indicates that sewage did not promote N 2 fixation in Trichodesmium. The concentrations of dissolved organic carbon (DOC) and nitrogen (DON) in sewage waters were double than in clear waters (Table 1), which is likely responsible for the increase in bacterial abundance in sewage waters with respect to clear waters ( Table 2). In support of these www.nature.com/scientificreports/ results, the abundance of the most active high nucleic acid bacteria (HNA) tripled when incubated with sewage for 24 h (Fig. 4e) coinciding with a net consumption of DOC (Fig. 4f), a trend which was not observed in clear water incubations (Fig. 4b,c). Trichodesmium could have responded positively to DOP inputs as observed elsewhere 31,32 , but in our study DOP concentrations of clear and sewage waters were not significantly different (Table 1). We cannot exclude however, the possibility that there were toxic components in sewage waters that  www.nature.com/scientificreports/ may have locally promoted the death of Trichodesmium close to the coast, as wastewaters are known to impair the metabolic activity of phytoplankton 33 . N 2 fixation rates were ~ 200, 70 and 2 nmol N L −1 day −1 in sparse, slick and PCD samples, respectively, in both clear or sewage water incubations (Fig. 3a,b). We confidently attribute these rates to Trichodesmium, since our incubations included concentrated Trichodesmium biomass and clear or sewage water prefiltered by 20 µm (see "Methods"). Diazotrophs smaller than 20 µm such as UCYN-A have been reported in the surroundings of the Canary Islands 34 , but their N 2 fixation activity is one to two orders of magnitude lower than that measured here. Our N 2 fixation rates were three, two and one order of magnitude higher than those observed during a sewage outburst in the eastern Mediterranean Sea 30 . The N 2 fixation rates of sparse colonies were at the high end of volumetric rates compiled in the global N 2 fixation database 35 , and in the range of those observed in hotspots of diazotrophy such as the western tropical South Pacific 36 . This indicates that Trichodesmium were healthy and active at the time of sampling. Between June and August 2017, the Canary Islands experienced optimal conditions for the development of Trichodesmium: unusually warm seawater temperatures (> 23 °C), enhanced stratification of the water column and several dust deposition peaks superimposed on a lower but steady dust supply throughout the summer months 27 . There was no evidence that Trichodesmium consumed DOC and DON despite their higher concentrations in sewage waters than in clear waters (Table 1). This agrees with previous studies that point towards a mixotrophic nutrition in Trichodesmium only when inorganic nutrients are not available 24,26 .
The N 2 fixation rates of the three bloom development stages were significantly different (one-way ANOVA p = 1.5 × 10 -5 and p = 0.003 for clear and sewage waters, respectively). This sequential decrease in N 2 fixation between bloom development stages agrees well with those observed in culture and field experiments 19,20,37,38 , and is clearly depicted by an increase in the release of PO 4 3− , NH 4 + and DOC as cells die and the bloom decays (Table 1).
Carbon fixation rates were not significantly different between clear and sewage water incubations either (t-test, p = 0.07). Contrary to N 2 fixation rates, carbon fixation was 1.4 and ~ 2 times higher in slick samples than in sparse stages in clear and sewage water incubations, respectively (Fig. 3c,d). However, this enhancement was only statistically significant in clear water incubations (one-way ANOVA p = 0.0002). Carbon fixation is performed by the bulk phytoplanktonic community and not only by Trichodesmium, however, the abundance of picophytoplankton in slick samples was not higher than that observed in the other bloom stages ( Table 2).
The recurrent PCD-like status of the Trichodesmium accumulations observed along the Canary Islands' coasts in the summer of 2017 (Fig. 1) further reinforce our hypothesis that sewage outfalls were not causing bloom reactivation and raises the question of whether they were, on the contrary, toxic to Trichodesmium. We however note that the chemical composition of sewage waters may be highly variable depending on their source (e.g. urban vs. rural domestic sewage, hotel resorts, golf courses, desalinization plants, etc.), potentially leading to differences in their impact on bacteria, non N 2 fixing phytoplankton and diazotrophs.
In conclusion, our results support the hypothesis that sewage outfalls south of the islands did not trigger the formation of new blooms, nor sustain or enhance blooms originated previously offshore. Although this study www.nature.com/scientificreports/ examines only the potential effects of sewage waters on the growth of Trichodesmium, other results (which will form the basis of a complementary paper) provide evidence that the 2017 bloom originated in open ocean waters of a large part of the Canary Current region, and the colonies were transported towards the coasts by the regional circulation and local winds. The progression from sparse colonies in open ocean waters, followed by surface slicks nearer the islands to finally decaying blooms accumulated in the lee of the islands strongly supports this hypothesis 27 . Trends in ocean temperature in the past two decades 39 indicate that the surface waters of the Canary Current are warming unabatedly, presumably favoring the more common appearance of Trichodesmium blooms in the near future. This has unknown consequences for the Canary region, as blooms have also occurred in the years 2018-2020, although with less intensity than in 2017. In order to predict the development of these blooms it is therefore necessary to determine which factors (or combination of factors) trigger and maintain the blooms from their origin to their collapse. The results of this study shed some light on this issue by clearly showing that coastal sewage outfalls do not enhance or maintain these blooms near the coast.

Carbon and N 2 fixation measurements.
In order to test the effects of sewage waters on the N 2 and carbon fixation activity of Trichodesmium, we sampled colonies from near surface waters south of Gran Canaria Island in September 2017 (Table 1). To evaluate the behavior of colonies experiencing different developmental stages, we sampled waters with Trichodesmium under three different bloom development phases (Fig. 2): (i) sparse colonies -'sparse'-, (ii) colonies accumulated in the surface as slicks-'slick'-, and (iii) collapsing accumulated colonies -'PCD-like'-, collected at three different locations south of Gran Canaria Island (Table 1). Clear seawater devoid of Trichodesmium colonies was sampled further offshore and sewage water was collected from a coastal outlet off Puerto Rico village (27° 47′ 17″ N 15° 42′ 40″ W, Table 1). In situ temperature was measured using a Hydrolab LH4 probe.
Sparse, slick and PCD-like colonies were concentrated using a 20 µm mesh sieve. Subsequently, 5 mL of the concentrate was distributed in triplicate acid-washed 2.3 L polycarbonate bottles (Nalgene) containing either clear seawater or sewage water from separate coastal areas (Table 1), previously filtered through 20 µm mesh to remove predators. Each bottle was spiked with 2.5 mL 15 N 2 (98 atom %, Euriso-top) and 13 C-labeled bicarbonate www.nature.com/scientificreports/ (NaH 13 CO 3 ; ≥ 98 atom %, Sigma Aldrich, 10 atom % final enrichment) as previously described 40 . The bottles were incubated in shaded incubators with surface seawater for 24 h. At the end of incubations, the content of the bottles was filtered through pre-combusted (5 h, 450 °C) GF/F filters (Whatman) and stored at -20 °C until analysis. The concentration of particulate nitrogen and carbon as well as the isotopic ratio of samples ( 15 N/ 14 N and 13 C/ 12 C) were obtained by means of a Thermo Flash 1112 elemental analyzer interfaced by a Conflo III with a Thermo Delta V Advantage isotope ratio mass spectrometer. To ensure an accurate calculation of N 2 fixation rates, background dissolved 15 N atom % enrichments were determined in all incubations by membrane inlet mass spectrometry as previously described 41 .
Nutrient and dissolved organic matter concentrations. Seawater samples for the analysis of inorganic nutrients and dissolved organic nitrogen and phosphorus (DON and DOP, respectively) were collected from clear waters, sewage-affected waters and within the Trichodesmium blooms (sparse, slick and PCD-like, see above). Samples for the determination of dissolved organic carbon (DOC) concentrations were collected from within Trichodesmium blooms as above, but also at the start and end of incubations for carbon and N 2 fixation measurements. All samples were filtered through pre-combusted GF/F filters (Whatman) before storage at -20 °C. Samples for the analysis of nitrate and nitrite (NO 3 − + NO 2 − ), phosphate (PO 4 3− ) and ammonium (NH 4 + ) were collected in 15 mL polyethylene tubes and stored at − 20 °C. Nutrient concentrations were determined using a Technicon II segmented-flow autoanalyzer. Samples for DOC analyses were collected in HCl-washed 20 mL polycarbonate tubes (Nalgene), stored at -20 °C, and DOC concentrations determined with a TOC-V Shimadzu as detailed in Santana-Falcón et al. 42 . DON and DOP concentrations were collected in 50 mL polyethylene tubes and analyzed by the wet oxidation of total dissolved nitrogen and phosphorus (TDN and TDP, respectively), with subsequent subtraction of NO 3 − + NO 2 − and PO 4 3− concentrations, respectively using standard wet oxidation methods 43 . Autotrophic and heterotrophic picoplankton. Abundances of autotrophic (Prochlorococcus and Synechococcus type cyanobacteria and pigmented picoeukaryotes) and heterotrophic prokaryote assemblages were determined by flow cytometry. Samples (1.6 mL) were preserved with paraformaldehyde (2% final concentration), left 15 min at 4 °C in the dark to fix, deep frozen in liquid nitrogen and stored at -80 °C until analyzed. Fixed samples were thawed, stained in the dark for a few minutes with a DMS-diluted SYTO-13 (Molecular Probes Inc.) stock (10:1) at 2.5 µM final concentration, and run through a BD FACSCalibur cytometer with a laser emitting at 488 nm. High and Low Nucleic Acid content prokaryotes (HNA, LNA) were identified in bivariate scatter plots of side scatter (SSC-H) versus green fluorescence (FL1-H). Autotrophic picoplankton were discriminated in plots of orange fluorescence (FL2) versus red fluorescence (FL3) and picocyanobacteria (Prochlorococcus and Synechococcus) were subtracted from HNA prokaryote counts. Samples were run at low or medium speed until 10.000 events were captured. A suspension of yellow-green 1 µm latex beads (10 5 -10 6 beads ml −1 ) was added as an internal standard (Polysciences, Inc.).