Metabolically active, non-nitrogen fixing, Trichodesmium in UK coastal waters during winter

Trichodesmium, a colonial cyanobacterium typically associated with tropical waters, was observed between January and April 2014 in the western English Channel. Sequencing of the heterocyst differentiation (hetR) and 16S rRNA genes placed this community within the Clade IV Trichodesmium, an understudied clade previously found only in low numbers in warmer waters. Nitrogen fixation was not detected although measurable rates of nitrate uptake and carbon fixation were observed. Trichodesmium RuBisCO transcript abundance relative to gene abundance suggests the potential for viable and potentially active Trichodesmium carbon fixation. Observations of Trichodesmium when coupled with a numerical advection model indicate that Trichodesmium communities can remain viable for >3.5 months at temperatures lower than previously expected. The results suggest that Clade IV Trichodesmium occupies a different niche to other Trichodesmium species, and is a cold- or low-light-adapted variant.

marine N fixation budget of 100-200 Tg N per year (Karl et al., 2002), making this an important group of organisms for the productivity and functioning of the warm, oligotrophic environment within which their distribution is largely limited. Although the occurrence of metabolically active Trichodesmium appears to be mostly restricted to water temperatures of .208C, Carpenter (Carpenter, 1983) introduces a number of accounts of Trichodesmium appearing in colder conditions. McCarthy and Carpenter (McCarthy and Carpenter, 1979) reported active populations of Trichodesmium at 18.38C in the North Atlantic, and Farran (1932) reported collecting Trichodesmium in 75% of plankton samples from the south of Ireland between October and March in temperatures of 9-128C. Indications are that Trichodesmium occurs episodically in northern latitudes of the Atlantic following passive transport along ocean currents (Laroche and Breitbarth, 2005 and references therein) and that cells are capable of surviving for several weeks at temperatures at least as low as 178C (Breitbarth et al., 2007) and possibly even in Arctic environments (Díez et al., 2012).
This paper describes observations made of the persistent occurrence of Trichodesmium at stations L4 and E1 of the Western Channel Observatory (WCO, http://www .westernchannelobservatory.org.uk/) over 12 weeks between January and April 2014. The winter of 2013 into 2014 for the UK, western English Channel and the greater north east Atlantic was characterized by persistent periods of low pressure which resulted in consistently strong winds originating in the south and west and prolonged periods of heavy rainfall (http://www.metoffice.gov .uk/climate/uk/summaries/2014). Seawater temperatures at L4 decreased from 12.28C in early December to 8.98C by 5 March and then increased up to 10.28C by 9 April, which is typical for this region (Smyth et al., 2010). Nitrate concentrations were generally typical of winter conditions: 6.4 mmol L 21 on 2 December 2013, 7.9 mmol L 21 on 18 February 2014 and 8.6 mmol L 21 on 9 April (see http://www.westernchannelobservatory.org. uk/l4_nutrients.php for the full dataset). Mean and median concentrations between 2 December 2013 and 9 April 2014 were 9.2 and 8.5 mmol L 21 , respectively.
Colonies of suspected Trichodesmium were first recorded in zooplankton samples collected at station L4 on 14 January 2014. These samples are routinely collected as part of the WCO monitoring programme with standardized protocols to provide a semi-quantitative assessment of plankton abundance or biomass. Full details are provided elsewhere (UNESCO, 1968, p. 153-157;Eloire et al., 2010). The sampling protocol involves vertical hauls of a WP2 net (57-cm diameter, 200-mm mesh size) from 50 m (bottom depth 55 m) to the surface at a winch speed of 20 cm s 21 . Samples were kept in cool, dark conditions, returned to the laboratory and processed under ambient conditions. A number of colonies were either collected directly onto 25-mm GF/F filters (Fig. 1a) or gently removed from the concentrated net samples using forceps and examined in 3-mL aliquots of filtered seawater using a Leica DMI4000B light microscope at magnifications between Â4 and Â20 ( Fig. 1b and c). Trichodesmium was observed on 11 of 13 sampling occasions between 14 January and 9 April 2014. Colony abundance was enumerated from the WP2 net hauls on three occasions, with counts of 663, 194 and 17 colonies m 23 estimated on 10 February, 18 February and 5 March, respectively, using light microscopy. As a flow-meter is not fitted to WP2 nets, the volume of water filtered is estimated from the product of net opening area (0.25 m 2 ) Â depth sampled (50 m) Â sampling efficiency coefficient (0.95).
In order to identify the population, DNA was extracted from colonies from three separate occasions (14 January, 29 January and 10 February) which had been collected onto 25-mm, 0.8-mm polycarbonate filters which were frozen at 2808C. DNA was extracted from the filaments using the method of Hoffman and Winston (Hoffman and Winston, 1987) and amplified using three different PCR primer pairs previously used to identify Trichodesmium sp. (Hynes et al., 2012) (molecular methods can be found in Supplementary Information). Sequences from each of the three different genes studied had . 99% sequence similarity, suggesting that only one species was present. Hynes et al. (Hynes et al., 2012) identified four clades of Trichodesmium from cultured representatives and field samples. The filaments found at Station L4 were most closely related to Trichodesmium, with the 16S rRNA and hetR ( Fig. 1d) gene sequences both indicating the colonies grouped with Clade IV. Clade IV hetR sequences were derived from Trichodesmium colonies collected from the Caribbean Sea (T. contortum AF013031), close to the equator in the Atlantic Ocean (T. tenue AF013033, Janson et al., 1999), and also station ALOHA (KF960514) in the Pacific. The Trichodesmium colonies found at L4 had 97% hetR sequence similarity to Clade IV Trichodesmium previously isolated from the Caribbean Sea (Janson et al., 1999). Similarly, the L4 16S rRNA gene sequence had 99% sequence similarity to the Clade IV Trichodesmium. Other studies that have included descriptions of Clade IV Trichodesmium are limited. However, one study of the diversity of Trichodesmium at station ALOHA at 10 m depth over a 2-year period demonstrated only 2% of hetR gene sequences grouped with Clade IV (Gradoville et al., 2014). It is possible that Clade IV may be more suited to cooler waters than those associated with tropical regions. Alternatively, this group may be adapted to lower light conditions. The Clade IV T. tenue identified by Janson et al. (Janson et al., 1999) close to the equator in the Atlantic Ocean was isolated from a depth of 75 m, and it was noted that this type was not found at depths shallower than 25 m. Together, this suggests that Clade IV Trichodesmium may occupy a different niche than other Trichodesmium species.
Following the observation of Trichodesmium-like cyanobacteria in low-temperature environments (Carpenter, 1983;LaRoche and Breitbarth, 2005;Díez et al., 2012), the question arises as to whether these populations are growing, dormant or dead. Two approaches were taken to investigate their activity. One method involved using stable isotopes to follow the incorporation of N 2 , NO 3 2 and CO 2 into colonies. As it is possible that this approach may be influenced by the complex consortia of phytoplankton and epibiotic bacteria often associated with Trichodesmium sp. colonies (Sheridan et al., 2002;Hmelo et al., 2012), molecular techniques were also used to investigate RuBisCO activity to provide a more direct insight into photosynthetic activity. 15 N and 13 C isotopes were used during a single investigation which was performed on colonies collected on 30 January. Colonies were picked using forceps and a  (Hynes et al., 2012).Oscillatoria sancta PCC7515 was used as an outgroup to root the tree. Numbers at nodes reflect percent bootstrap consensus .50%. binocular microscope, rinsed in filtered sea water and then distributed into 3Â125 mL serum bottles where they were re-suspended in 0.2-mm filtered seawater collected immediately before plankton net hauls. A single bottle containing 60 colonies was filled and crimp sealed with a Teflon backed butyl septa. A second bottle containing 60 colonies was amended with 620 mmol L 21 of 13 C-labelled bicarbonate to determine carbon fixation. This was capped as previously and then further amended with the addition of 1 mL of 98 atom% 15 N 2 gas (Cambridge Isotope Laboratories, Inc., LOT I1-11785A) to determine nitrogen fixation (e.g. Rees et al., 2009). A final bottle containing 30 colonies was amended with additions of 620 mmol L 21 of 13 C bicarbonate and 0.4 mmol L 21 of 15 N-labelled nitrate to determine carbon fixation and nitrate uptake (e.g. Bury et al., 2012). All three bottles were transferred to a laboratory incubator for 24 h and maintained at 108C for 12 h in the light ( 100 mE m 22 s 21 ) and 12 h in the dark. At the end of the incubation, they were filtered onto 25-mm GF/F filters and dried for 12 h at 508C. 13 C and 15 N atom% was then determined using stable isotope mass spectrometry (SERCON 20:20 and GSL); precision of analysis was 0.26 and 0.03% coefficient of variation for 15 N and 13 C, respectively.
Relative to the control treatment, which received no additions, it is not surprising that there was no enrichment of 15 N in cells contained within the bottle amended with 15 N-N 2 . This supports the accepted wisdom that Trichodesmium does not fix nitrogen under lowtemperature and high-nutrient conditions (Carpenter, 1983;Capone et al., 1997;Holl and Montoya, 2005;Breitbarth et al., 2007). With the major caveat that there was likely to be a microbial community associated with incubated Trichodesmium (Sheridan et al., 2002;Hmelo et al., 2012), indications are that the populations of the Clade IV Trichodesmium were capable of the fixation of carbon and uptake of NO 3 2 (Fig. 2). Enrichments of 0.0088 atom% 15 N and 0.0604 and 0.0742 atom% 13 C were observed which equate to rates in the order of 0.05 nmol NO 3 2 colony 21 day 21 and 0.42 nmol C colony 21 day 21 , and a ratio of carbon fixation:nitrate uptake of 8:1. With other nitrogen sources available including NH 4 þ (mean concentration of 0.52 mmol L 21 for December 2013 to February 2014), it is not unreasonable to assume that C:N uptake rates were in balance relative to Redfield stoichiometry. Whether attributable solely to Trichodesmium or to an associated microbial community, these rates do not present a significant contribution to the productivity of the western English Channel. The observed C fixation was an order of magnitude lower than the lowest rates of primary production reported by Carpenter et al. (Carpenter et al., 2004) for the tropical North Atlantic, and based on the maximum observed abundance of 663 colonies m 23 contributes ,0.03% of wintertime primary production of 12.7 (+7.1) mgC m 23 day 21 reported by Barnes et al. (Barnes et al., 2015).
To confirm that the Trichodesmium were able to fix carbon, quantitative PCR (qPCR) was used to measure the abundance of transcripts of the large subunit of RuBisCO (rbcL) relative to gene numbers. Colonies collected on two occasions (29 January and 10 February) were subjected to a combined DNA and RNA extraction procedure, as described in Supplementary Information. Primers specific to the rbcL gene of T. erythraeum were designed, PCR amplified from cDNA and DNA and the resultant PCR product was cloned and sequenced. Eight clones were sequenced from each time-point: all contained the same gene sequence which had 92 and 99% identity to the T. erythraeum rbcL gene and protein sequence, respectively (accession number KU221396). Quantification of rbcL mRNA and DNA indicated the rbcL mRNA:DNA ratio to be 25.8 (+10.04 SD; n ¼ 2). Measurements of both rbcL gene and transcript abundance from other studies are rare. For diatom species in the Bering Sea, diatom-specific rbcL transcripts relative to genes numbered .1000 (Endo et al., 2015). A recent study of the impact of CO 2 and temperature on Trichodesmium gene expression reported expression of rbcL relative to 16S rRNA abundance, in the range of 10 0 -10 1 copies of rbcL:16S rRNA (Hewson et al., 2009). Although the expression levels of rbcL measured in this current study were low, the presence of rbcL mRNA does indicate that the Trichodesmium sp. found at L4 were capable of fixing carbon.
Trichodesmium spp. are generally associated with oligotrophic sub-tropical and tropical waters with a temperature range of 20-308C (Breitbarth et al., 2007). To assess the transport of this population to the WCO, we have considered satellite observations of surface temperatures and currents (Fig. 3) to provide some constraints as to how long these trichomes may have experienced waters cooler than 208C. The seasonal variations in sea surface temperature in the northern hemisphere peak in September, with the isotherms at their northernmost extent. The mean currents in the region, including the eastward-flowing North Atlantic Current and the poleward-flowing European Slope Current in the Bay of Biscay (Xu et al., 2015) provide some transport to the WCO site. To include the action of eddies, an advective model was used to transport individual particles from the realm bounded by the 20 and 218C isotherms in September 2013. The numerical model uses horizontal surface current velocities produced at daily resolution by the SURCOUF component of GlobCurrent (www .globcurrent.org). The current field includes both the geostrophic component, calculated from a number of altimeters, and the surface Ekman current calculated from the vector wind field (Rio et al., 2014). The inclusion of the eddy field increases the dispersion of particles compared with simply using a set of mean currents, but even so requires more than the 3.5 months until the start of 2014 to explain the presence of viable Trichodesmium at the WCO (Fig. 3).
We also examined the SST distribution for September during other years to note whether 2013 was in any way different. In fact, although the 208C isotherm at 208W was at its furthest north in 2013, for regions east of 158W, the temperatures were cooler than normal, with that isotherm displaced southwards, which indicates that a contribution via the European Slope Current (Xu et al., 2015) can also be discounted.
We find that RuBisCO transcript abundance relative to gene abundance suggests the potential for viable and potentially active Clade IV Trichodesmium in the western English Channel. Indications are that this group can remain viable for many months under environmental conditions previously considered to preclude its activity. Although the rates of carbon fixation and nitrate uptake observed do not indicate a major role for this community in the biogeochemical cycling of these waters, their occurrence is ecologically unprecedented and informs on the transport and distribution of this globally significant cyanobacterium.

AC K N OW L E D G E M E N T S
Elaine Fileman and Angela Rodriguez picked out colonies for experiments and Malcolm Woodward and Carolyn Harris provided nitrate concentrations. AMSR data are produced by Remote Sensing Systems and sponsored by the NASA Earth Science MEaSUREs DISCOVER Project and the NASA AMSR-E Science Team. Data are available at www.remss.com. We are grateful to Marie-Helene Rio for early access to the daily surface current fields generated for the GlobCurrent project. We thank Pia Moisander and two anonymous referees whose comments have helped in the publication of this work.

F U N D I N G
This work was supported by the UK Natural Environment Research Council (NERC) through National Capability funding of the Western Channel Observatory (WCO). The authors acknowledge the sustained contribution made by the crew of the RV Plymouth Quest. Funding to pay the Open Access publication charges for this article was provided by the UK Natural Environment Research Council.