INTRODUCTION
At finfish aquaculture sites, organic waste (e.g. faecal matter and unconsumed feed), if not adequately dispersed, can alter seafloor biogeochemistry and benthic community structure (Ye et al., Reference Ye, Ritz, Fenton and Lewis1991; Pereira et al., Reference Pereira, Black, McLusky and Nickell2004; Shahidul Islam & Tanaka, Reference Shahidul Islam and Tanaka2004; Carvalho et al., Reference Carvalho, Barata, Pereira, Gaspar, da Fonseca and Pousão-Ferreira2006; Yokoyama et al., Reference Yokoyama, Abo and Ishihi2006; Kutti et al., Reference Kutti, Ervik and Hansen2007; Borja et al., Reference Borja, Rodrìguez, Black, Bodoy, Emblow, Fernandes, Forte, Karakassis, Muxika, Nickell, Papageorgiou, Pranovi, Sevastou, Tomassetti and Angel2009, Husa et al., Reference Husa, Kutti, Ervik, Sjøtun, Hansen and Aure2013). Sedimentary infauna beneath aquaculture cages typically show a decreased biodiversity and a high abundance of opportunistic species compared with adjacent sites; consequently, ecosystem functioning in such sediments is affected (Ye et al., Reference Ye, Ritz, Fenton and Lewis1991; Crawford et al., Reference Crawford, Mitchell and Macleod2001; Pereira et al., Reference Pereira, Black, McLusky and Nickell2004; Carvalho et al., Reference Carvalho, Barata, Pereira, Gaspar, da Fonseca and Pousão-Ferreira2006; Kutti et al., Reference Kutti, Ervik and Høisæter2008; Borja et al., Reference Borja, Rodrìguez, Black, Bodoy, Emblow, Fernandes, Forte, Karakassis, Muxika, Nickell, Papageorgiou, Pranovi, Sevastou, Tomassetti and Angel2009). In contrast, little is known on the impacts of aquaculture wastes on benthic community function at sites dominated by hard substrates.
Along the south coast of Newfoundland (NL), Canada, salmonid aquaculture takes place in bays and fjords with steep slopes and a patchy substrate, often dominated by coarse particles or bedrock (Hamoutene et al., Reference Hamoutene, Mabrouk, Sheppard, MacSween, Coughlan and Grant2013). Using seafloor imaging, white microbial mats (likely Beggiatoa spp.) and opportunistic polychaete complexes (OPC) were found to be directly associated with aquaculture production in this region (Hamoutene et al., Reference Hamoutene, Mabrouk, Sheppard, MacSween, Coughlan and Grant2013; Hamoutene, Reference Hamoutene2014). OPC are found at aquaculture sites worldwide and are composed of species tolerant of organic matter enrichment and associated reduced conditions (e.g. increased sulphide and methane concentrations), such as the capitellids Heteromastus filiformis Claparède, 1864 and Capitella capitata Fabricius, 1780 (Nickell et al., Reference Nickell, Black, Hughes, Overnell, Brand, Nickell, Breuer and Martyn Harvey2003; Kutti et al., Reference Kutti, Ervik and Høisæter2008). In NL, OPC beneath salmon and steelhead trout cages consist of a single new dorvilleid species, Ophryotrocha cyclops Salvo et al., Reference Salvo, Wiklund, Dufour, Hamoutene, Pohle and Worsaae2014, with conspecifics also found on whalebones in Greenland (Salvo et al., Reference Salvo, Wiklund, Dufour, Hamoutene, Pohle and Worsaae2014). Other Ophryotrocha species colonize extreme (and often sporadic) habitats such as wood and whale-falls (Wiklund et al., Reference Wiklund, Glover and Dahlgren2009a, Reference Wiklund, Glover, Johannessen and Dahlgrenb, Reference Wiklund, Altamira, Glover, Smith, Baco and Dahlgren2012), aquaculture sites (Paxton, Reference Paxton2009; Paxton & Davey, Reference Paxton and Davey2010), cold methane seeps (Sahling et al., Reference Sahling, Rickert, Lee, Linke and Suess2002; Levin et al., Reference Levin, Ziebis, Mendoza, Growney-Cannon and Walther2006, Reference Levin, Mendoza, Konotchick and Lee2009, Reference Levin, Ziebis, Mendoza, Bertics, Washington, Gonzalez, Thurber, Ebbe and Lee2013; Thurber et al., Reference Thurber, Kröger, Neira, Wiklund and Levin2010) or sites with high forest litter accumulation (McLeod et al., Reference McLeod, Wing and Skilton2010).
Dorvilleids are highly tolerant of organic enrichment and even sulphidic conditions: in cold seeps, 80% of the total abundance of dorvilleids was restricted to sulphide patches (Levin et al., Reference Levin, Ziebis, Mendoza, Growney-Cannon and Walther2006). In such habitats, they are taxonomically diverse and very abundant (>8000 ind. m−2) (Thornhill et al., Reference Thornhill, Struck, Ebbe, Lee, Mendoza, Levin and Halanych2012; Levin et al., Reference Levin, Ziebis, Mendoza, Bertics, Washington, Gonzalez, Thurber, Ebbe and Lee2013) compared with other taxa, and likely play a role in the sulphur cycle. Dorvilleids are often associated with Beggiatoa spp. mats (Levin et al., Reference Levin, Mendoza, Konotchick and Lee2009, Reference Levin, Ziebis, Mendoza, Bertics, Washington, Gonzalez, Thurber, Ebbe and Lee2013), and some species within the family selectively feed on sulphide-oxidizing bacteria or methanotrophic archaea (Levin et al., Reference Levin, James, Martin, Rathburn, Harris and Michener2000, Reference Levin, Mendoza, Konotchick and Lee2009, Reference Levin, Ziebis, Mendoza, Bertics, Washington, Gonzalez, Thurber, Ebbe and Lee2013; Levin & Michener, Reference Levin and Michener2002; Decker & Olu, Reference Decker and Olu2010; Thornhill et al., Reference Thornhill, Struck, Ebbe, Lee, Mendoza, Levin and Halanych2012; Thurber et al., Reference Thurber, Levin, Orphan and Marlow2012). At NL aquaculture sites, the accumulation of organic matter has likely triggered the development of OPC: using video-imaging, Ophryotrocha cyclops presence was mainly documented close to fish cages, where they were typically in high abundance and often co-occurring with Beggiatoa spp. mats (Hamoutene et al., Reference Hamoutene, Mabrouk, Sheppard, MacSween, Coughlan and Grant2013; Hamoutene, Reference Hamoutene2014). The food source(s), habitat requirements, and physiology of this dorvilleid species are not well known. A better understanding of the biology of O. cyclops could indicate whether this species plays a role in the remediation of benthic habitats at aquaculture sites following organic waste accumulation.
Here, we examine trophic relationships between Ophryotrocha cyclops at NL salmonid farm sites and fish pellets, flocculent matter (a complex mixture of sedimented material; Chou et al., Reference Chou, Haya, Paon, Burridge and Moffatt2002; Yokoyama et al., Reference Yokoyama, Abo and Ishihi2006) and other potential food sources such as macroalgae and suspended particulate organic matter (SPOM) using stable isotope analysis (SIA) (δ13C, δ15N and δ34S) and trace element analyses (TEA). The ratios of stable isotopes, especially those of nitrogen (δ15N) and carbon (δ13C), are often used in ecological studies to describe linkages between organisms and their food sources (e.g. Peterson & Fry, Reference Peterson and Fry1987; Cranford et al., Reference Cranford, Dowd, Grant, Hargrave and McGladdery2003; Carlier et al., Reference Carlier, Ritt, Rodrigues, Sarrazin, Olu, Grall and Clavier2010). The δ13C composition of primary producers is a function of available CO2 and the degree of isotope fractionation during carbon fixation (which varies between types of primary producers), and animals show δ13C signatures that are similar to those of their diet (DeNiro & Epstein, Reference DeNiro and Epstein1978; Peterson & Fry, Reference Peterson and Fry1987). The δ15N values of primary producers reflect the available nitrogen source (Peterson & Fry, Reference Peterson and Fry1987), and animal δ15N values increase by an average of 2.3‰ in successive trophic levels (McCutchan et al., Reference McCutchan, Lewis, Kendall and McGrath2003). As fish pellets and fish faecal matter often bear characteristic δ13C and δ15N signatures depending on their composition (Chou et al., Reference Chou, Haya, Paon, Burridge and Moffatt2002; Yokoyama et al., Reference Yokoyama, Abo and Ishihi2006) and the relative importance of marine and terrestrial-derived products within them, the latter can be used as biomarkers of aquaculture-derived organic matter transfer along food webs. SIA of nitrogen and carbon have been used to examine the dispersion of aquaculture waste, its contribution to sedimented or suspended organic matter (Ye et al., Reference Ye, Ritz, Fenton and Lewis1991; McGhie et al., Reference McGhie, Crawford, Mitchell and O'Brien2000; Franco-Nava et al., Reference Franco-Nava, Blancheton, Deviller and Le-Gall2004; Sarà et al., Reference Sarà, Scilipoti, Mazzola and Modica2004; Vizzini et al., Reference Vizzini, Savona, Caruso, Savona and Mazzola2005; Yokoyama et al., Reference Yokoyama, Abo and Ishihi2006) and its incorporation into organisms (Grey et al., Reference Grey, Waldron and Hutchinson2004). δ34S can further clarify trophic relationships: sulphur isotopes have been used to discriminate between primary producers (Connolly et al., Reference Connolly, Guest, Melville and Oakes2004), benthic and pelagic organisms (Peterson, Reference Peterson1999), or terrestrial and marine sources (Moreno et al., Reference Moreno, Jover, Munilla, Velando and Sanpera2010). Little to no S fractionation occurs along the food web (Peterson, Reference Peterson1999; McCutchan et al., Reference McCutchan, Lewis, Kendall and McGrath2003). At aquaculture sites, δ34S is likely to be informative given the presence of mats of sulphur-oxidizing bacteria, a potential food source for O. cyclops. During bacterial sulphate reduction, isotopic fractionation results in sulphide depleted in 34S (Canfield, Reference Canfield2001). The sulphur-oxidizing bacteria that assimilate this sulphide show 34S depletion, as do organisms consuming these bacteria (Carlier et al., Reference Carlier, Ritt, Rodrigues, Sarrazin, Olu, Grall and Clavier2010).
Zinc (from fish pellets) and copper (released during fish net cleaning) are two aquaculture-related trace elements that may accumulate in sediments (Chou et al., Reference Chou, Haya, Paon, Burridge and Moffatt2002; Brooks & Mahnken, Reference Brooks and Mahnken2003; Mendiguchía et al., Reference Mendiguchía, Moreno, Mánuel-Vez and García-Vargas2006; Dean et al., Reference Dean, Shimmield and Black2007; Sutherland et al., Reference Sutherland, Petersen, Levings and Martin2007) and have been used as tracers of aquaculture waste (Chou et al., Reference Chou, Haya, Paon, Burridge and Moffatt2002). Here, we compare the trace element composition of Ophryotrocha cyclops with that of their potential food sources (including fish pellets) to explore relationships between these dorvilleids and aquaculture waste.
MATERIALS AND METHODS
Sample collection
Newfoundland salmonid aquaculture occurs mainly along the south coast of the island in a complex of bays and fjords called Fortune Bay. Sampling was performed within 10 m from cages at aquaculture sites located in three bays: site S1 (depth: 54 m) in October 2012, site S2 (depth: 72 m) in November 2012, and site S3 (depth could not be precisely determined due to poor weather conditions but is estimated to be 35–40 m based on previous records) in August 2013. Sites S1 and S2 were in the second year of production while at site S3, salmon had been harvested the previous month. The coordinates of sampling sites are not disclosed herein at the aquaculture industry's request. There is a linear distance of 17.5 km between sites S1 and S3, of 44 km between sites S1 and S2, and of 25.6 km between sites S2 and S3.
We sampled Ophryotrocha cyclops during mandatory benthic monitoring surveys performed at aquaculture sites (as described in DFO, 2013). At site S1, O. cyclops were first located on the seafloor by video monitoring, at which point sampling was attempted. As grab sampling is inefficient at Newfoundland aquaculture sites (Hamoutene et al., Reference Hamoutene, Mabrouk, Sheppard, MacSween, Coughlan and Grant2013; Hamoutene, Reference Hamoutene2014), we used a different strategy for worm collection. A net with a 0.2 μm mesh size, held open with a loop of iron, was affixed to one of the edges at the bottom of the cage frame used for video monitoring. We dragged the bottom while observing worm sampling using the video camera. At sites S2 and S3, O. cyclops presence was not confirmed prior to sampling (only flocculent matter was observed at site S2, and weather conditions precluded video sampling at site S3); however, we nonetheless attempted to collect worms from these two sites. Ophryotrocha cyclops identity was confirmed using genetic analyses as described in Salvo et al. (Reference Salvo, Wiklund, Dufour, Hamoutene, Pohle and Worsaae2014).
At each site, seawater was collected using a Niskin bottle at both 1 m from the surface and close to the seafloor, for the analysis of SPOM.
Sample processing
Upon collection, dorvilleids were immediately isolated using a transfer pipette. Some were fixed in 95% EtOH, 4% formaldehyde or 2.5% glutaraldehyde in a 0.1 M sodium cacodylate buffer for identification, size determination or electron microscopy. Other specimens were kept at 4°C in separate 50 mL tubes with filtered seawater (0.7 μm) from the sampling site for a minimum of 24 h to allow gut content evacuation. Each replicate, consisting of 1 or 2 individuals, was frozen at −20°C in combusted vials. During the fasting period, filtered seawater was renewed and any mucus and faeces in the seawater were collected on pre-weighted and combusted (4 h, 450°C) GFF 47 mm filters (0.7 μm porosity), then dried for 48 h at 60°C. Dorvilleids were not rinsed with distilled water prior to freezing as it caused massive tissue rupturing. Separate frozen samples were used for either C&N SIA, for S SIA or for TEA.
All distinguishable types of organic matter found in the net were assumed to be potential food sources for the dorvilleids: pieces of macroalgae were cleaned of epibionts in filtered seawater, rinsed with distilled water and frozen at −20°C and flocculent matter was collected directly from the net and frozen at −20°C in 50 mL tubes. At site S2, partly degraded fish pellets were isolated from flocculent matter and frozen (−20°C) separately. Due to the sampling strategy, we were not able to isolate the bacterial mats that are known to coexist with OPC at NL aquaculture sites (Hamoutene et al., Reference Hamoutene, Mabrouk, Sheppard, MacSween, Coughlan and Grant2013).
Fish pellets were graciously provided to us by the aquaculture companies managing sites S1 and S2, where production was occurring at the time of sampling. At site S2, two types of pellets were collected: with and without medication. At site S3, salmon had been removed from the site about a month prior to OPC sampling, and we did not obtain pellets.
Stable isotope analysis
All samples (except filters) were freeze-dried, manually ground, homogenized and weighed in tin capsules for SIA of carbon and nitrogen (run simultaneously), or sulphur. Filters (containing either SPOM or faeces/mucus) were scraped to collect contents for SIA of nitrogen or sulphur; for carbon analysis, dried filters (with SPOM) were fumigated under HCl vapours to remove carbonates, then dried and scraped. Prior to sulphur SIA, up to 0.200 μg of V2O5 was added to each sample. We used a Finnigan MAT252 interfaced with a Carlo Erba NA1500 Series II elemental analyser and an OI Analytical Aurora 1030 TOC Analyser at the CREAIT TERRA Facility Stable Isotope Lab (Memorial University, Canada) for all stable isotope determinations.
Stable isotope ratios are expressed as per convention: R=13C/12C or 15N/14N or 34S/32S, with the reference being Vienna Pee Dee Belemnite for δ13C, atmospheric air for δ15N and Vienna-Canyon Diablo Triolite for δ34S (see Coplen et al., Reference Coplen, Böhlke, De Bievre, Ding, Holden, Hopple, Krouse, Lamberty, Peiser and Revesz2002). The maximal accuracy was < 0.3‰ for δ13C, <0.4‰ for δ15N and 1.1‰ for δ34S.
Trace element analysis
A subset of freeze-dried samples was selected for TEA. The dorvilleids considered were exclusively from site S2 and each sample consisted of 1–2 (pooled) individuals (after 48 h of fasting); fish pellets analysed were also from site S2.
Samples were weighed in Teflon screw cap tubes, then bathed in 8N nitric acid for 1 or 2 days at 60°C until dissolution was complete. Then, 1 mL of 30% hydrogen peroxide was added to the samples, which were heated at 60°C for 12 h. Samples were then dissolved in ultrapure water and a 10% dilution of each sample in 0.2N nitric acid was made the following day. TEA were performed at the CREAIT trace element lab (Memorial University, Canada) using a Perkin Elmer Elan DCR II Inductively Coupled Plasma Mass Spectrometer. Each sample except dorvilleids was run twice, with mussels (NIST 2976 and NIST 2977) used as standards. Replicates from dorvilleid samples consisted of one or two individuals per sample and are not analytical replicates.
To localize any accumulations of Zn or Cu in dorvilleid tissues, three individuals from site S2 that had been fixed in glutaraldehyde and post-fixed in 1% osmium tetroxide were processed for elemental analysis using an X-ray detector attached to an environmental scanning electron microscope (ESEM). Dorvilleids were sectioned into anterior, median and posterior fragments, dehydrated and separately embedded in Epon resin. 1 μm thick sections were mounted on aluminium stubs and carbon coated prior to elemental analysis using a standard (solid state) backscattered electron detector (Bruker XFlash SSD 5030) in an FEI Quanta 650F ESEM. Additional 60 nm thick sections were mounted on Cu grids, post-stained with uranyl acetate and lead citrate, and observed using a Philips 300 transmission electron microscope.
Statistical analyses
Comparisons of the isotopic signatures of worms were made using Mann–Whitney and Kruskal–Wallis analyses in Statistica 7.1. We could not statistically compare stable isotope ratios of different food sources or replicates due to low sample size (N < 5). Given site-specific differences in isotopic signatures, we chose to analyse sites separately rather than pooling sites and sources within a single analysis. Our data did not meet the necessary conditions for determining either a single, or three site-specific Bayesian mixing models (Caut et al., Reference Caut, Angulo and Courchamp2008, Reference Caut, Angulo and Courchamp2009; Moore & Semmens, Reference Moore and Semmens2008): available sources differed among sites and dates, and some important components (e.g. bacteria, fish faeces) could not be isolated. Moreover, mixing models need to consider isotopic fractionation, which varies between species or tissues considered (McCutchan et al., Reference McCutchan, Lewis, Kendall and McGrath2003; Caut et al., Reference Caut, Angulo and Courchamp2009). Although Thurber et al. (Reference Thurber, Kröger, Neira, Wiklund and Levin2010) assumed a null fractionation in some dorvilleid species from hydrothermal vents, fractionation has yet to be determined for the species investigated here. For these reasons, our interpretations are based on graphical representations of stable isotope ratios.
RESULTS
Using video monitoring, Ophryotrocha cyclops colonies were observed either directly underneath, or within 10 m from finfish cages. In the first bay (site S1), high abundances of dorvilleids were visible at the surface of the rocky substrate (Figure 1) whereas in the second bay (site S2), worms were not visible using video sampling but were present within flocculent matter. No direct seafloor observation was possible at site S3; however, dorvilleids were collected from this site in association with flocculent matter.
Fig. 1. Image of the substrate beneath aquaculture site B (June 2012) in Fortune Bay, NL, extracted from benthic monitoring surveys, showing abundant masses of Ophryotrocha cyclops. The inner frame measures 25 × 25 cm.
Stable isotope analysis
CARBON
Dorvilleids from sites S1 and S2 differed from those at site S3 in δ13C values (Table 1), the latter being about 2.5‰ lighter; Mann–Whitney U-tests revealed a significant difference between specimens from sites S2 and S3 (P < 0.001; S1 was not included in the analysis because N = 2). Compared with dorvilleid tissues, the faeces and mucus of dorvilleids were depleted in δ13C (Table 1).
Table 1. Carbon, nitrogen and sulphur stable isotope ratios of samples from the study sites. Values in δ15N+2.3 represent those expected in a particular source's consumer, assuming a fractionation of 2.3. Values are averages±SD, with number of replicates in parentheses when greater than 1.
FP, fish pellets; NA, not available; SPOM, suspended particulate organic matter.
Fish pellets from different sites had a similar carbon isotopic composition (average throughout sites and samples, with or without medication: −19.52 ± 0.35‰). A degraded fish pellet (site S2, −20.05‰) and flocculent matter had slightly lighter carbon isotope values (averages: site S1, −21.31 ± 0.10‰ and site S3, −21.22 ± 1.28‰) than fresh pellets (Table 1).
SPOM differed less in δ13C between sites than between surface and bottom water samples: the carbon signature at the bottom was lighter (−27.96 ± 1.20‰) than at the surface (−25.47 ± 1.31 ‰) at sites S1 and S2. At site S3, SPOM at the bottom had heavier δ13C values closer to surface SPOM samples from sites S1 and S2. The various macroalgae sampled had δ13C values characteristic of estuaries, ranging from −20.48 to −16.88‰, except for red algae, which had lighter δ13C values (approximately −34‰).
NITROGEN
The nitrogen isotopic composition of dorvilleids from sites S1 and S2 was similar (5.92 ± 0.63‰) and differed from that of worms at site S3 (3.77 ± 0.26‰); a Mann–Whitney U-test revealed a significant difference between samples from sites S2 and S3 (P < 0.05; S1 was not included in the analysis because N = 2, Table 1). The δ15N composition of dorvilleid faeces was close to that of dorvilleid tissues.
The δ15N composition of fish pellets differed between sites, with values in site S1 being heavier than at site S2. The δ15N of the degraded fish pellet was slightly heavier than that of fresh pellets (4.88‰) and flocculent matter was lighter at site S3 than at site S1.
The δ15N values of SPOM ranged from 3.84 to 7.20‰ according to site and depth in the water column. Macroalgae had light δ15N values ranging from 2.94 to 5.23‰, as expected for marine primary producers.
SULPHUR
Sulphur stable isotope data were not obtained for all material types due to the restricted number of samples available and the limited quantity of sulphur within samples.
The sulphur stable isotopic composition of dorvilleid worms varied greatly between individuals (coefficient of variation = 51%) and among sites: no site-specific significant differences were found using Kruskal–Wallis comparisons (P > 0.05). Dorvilleid faeces are more 34S enriched than are individuals (Table 1). The δ34S signature of algae ranges from 16.86 to 22‰.
Fish pellets are distinctive in δ34S (5.21 ± 1.85‰) and are the isotopically lightest of samples examined. The degraded fish pellet from site S2 was more 34S enriched (18.03‰) than fresh pellets, as was flocculent matter from S1 site (14.52‰). SPOM (site S2 only) had a δ34S signature typical of marine phytoplankton, near 20‰.
Relationships between worms and potential food sources at each site
At site S1, the two dorvilleid δ13C signatures were close to those of fresh fish pellets and green algae (Figure 2). Dorvilleids also had the heaviest δ15N values (with the exception of surface SPOM and one of the dorvilleid faeces samples), roughly one trophic level above many of the potential sources, considering a fractionation of 2.3 (Table 1). In plots of carbon vs. sulphur isotopic signatures, all green algae were grouped together, with sulphur values corresponding to a pelagic marine source whereas: (1) the dorvilleids were placed between fresh fish pellets (light in δ34S) and marine algae, and (2) dorvilleids were closer to, but slightly lower in δ34S than flocculent matter and their own faeces (Figure 2).
Fig. 2. Stable isotope composition of samples from sites S1 (A, D), S2 (B, E) and S3 (C, F), with data presented as averages±SD. (A–C) Nitrogen and carbon stable isotope ratios; (D–F) Sulphur and carbon isotope ratios.
At site S2, dorvilleids showed the same pattern as in site S1 in δ13C, their signature being close to fresh and degraded fish pellets, their own faeces and algae, and dissimilar to red algae and SPOM (Figure 2). The δ15N signature of dorvilleids was close to SPOM from the surface. Dorvilleids showed the lightest sulphur isotope values, which did not correspond to typical marine signatures and were closest to fish pellets.
No fish pellets were collected from site S3 because no salmon were present at the time of sampling, but dorvilleids showed isotopic carbon values similar to those of flocculent matter (Figure 2). As in sites S1 and S2, dorvilleids from site S3 are highly variable in δ34S.
Trace metal element analysis
Dorvilleid tissues had zinc concentrations ranging from 23 to 96 ppm (Table 2), whereas Zn concentrations were higher in flocculent matter (433 ± 51 ppm). Copper concentrations were relatively low (<10 ppm) in most samples, with flocculent matter again showing the highest values (135 ppm). Fe is most concentrated in flocculent matter (7 × 103 ppm), while in other samples Fe concentrations ranged from 2 to 5 × 102 ppm.
Table 2. Average and SD of Cu, Zn, Fe, Mn and S concentrations (ppm) using trace element analysis. N, number of analytical replicates, except for Ophryotrocha cyclops, where each replicate consists of either 1 or 2 whole individuals.
The lowest concentrations of sulphur were found in fish pellets (7 × 104 ppm). Dorvilleids showed a great variability in sulphur concentration (2.19 ± 8.15 × 104 ppm) and range, as did flocculent matter (1 × 104 to 2 × 104 ppm).
The X-ray analysis of dorvilleid tissues using SEM revealed no localized accumulations of Zn or Cu. However, we noted abundant lipid droplets in gut epithelial cells using transmission electron microscopy (Figure 3).
Fig. 3. Transmission electron micrograph of gut epithelial cells of Ophryotrocha cyclops. Lipid droplets appear as dark structures within epithelial cells. Scale bar: 5 μm.
DISCUSSION
We studied several potential Ophryotrocha cyclops food sources (SPOM, macroalgae, fish pellets both fresh and degraded, and flocculent matter) collected from aquaculture sites. Despite site-specific variations in isotopic composition, we were able to identify likely and unlikely contributors to the diet of this annelid at aquaculture sites, as detailed below.
Macroalgae and SPOM probably contribute very little, if at all, to the diet of Ophryotrocha cyclops. The macroalgae retrieved using the net were rare, mostly fragmented, and occasionally degraded according to their shape and colour but with δ13C signatures similar to those reported in a study of salmonid farms in Tasmania (Ulva: – 19.82‰, Rhodophyceae: −30‰; Ye et al., Reference Ye, Ritz, Fenton and Lewis1991). While most macroalgae are close in their δ13C and δ15N composition to O. cyclops, both δ34S differences (≈10‰) and rarity indicate that macroalgae are unlikely to constitute a perennial food source for the worms, considering a null fractionation for sulphur along the food web (McCutchan et al., Reference McCutchan, Lewis, Kendall and McGrath2003). It is also unlikely that O. cyclops feeds extensively on suspended or freshly deposited POM, as the δ13C signature of SPOM at sites S1 and S2 was lighter (by at least 6‰) than that of O. cyclops, while the SPOM δ34S was heavier (≈10‰) and typically marine. Moreover, the shape of the jaw in dorvilleids species is not suggestive of a deposit or suspension-feeding behaviour (Salvo et al., Reference Salvo, Wiklund, Dufour, Hamoutene, Pohle and Worsaae2014).
The δ13C and δ15N signatures of fresh fish pellets were respectively about 1 and 2‰ lower than those of Ophryotrocha cyclops from sites S1 and S2, indicating that fish pellets are a likely food source for those worms once average fractionation is considered. Also, the δ34S of fresh fish pellets was notably lighter (>10‰) than other marine sources, suggesting a terrestrial sulphur component (i.e. plant matter, FAO, 2014) and providing a means of tracing fresh fish pellet-derived organic matter at our study sites. Sulphur isotopic signatures indicate that fresh fish pellets likely form a major part of the diet of O. cyclops. The changes in isotopic signature that take place as fish pellets degrade on the seafloor are noteworthy: their δ13C becomes slightly lighter, while δ15N and δ34S become heavier (the latter remarkably so, see Figure 2, S2). These changes may be due to differential decay of various organic matter components (Lehmann et al., Reference Lehmann, Bernasconi, Barbieri and McKenzie2002), and/or to colonization by microbes; the widely ranging O. cyclops δ34S signatures may reflect their consumption of fish pellets (and associated microbes) at various stages of decay. Moreover, the fish pellets used by aquaculture companies at the time of sampling (or, for site S3, prior to harvesting), were oil-rich (the pellets most commonly used by aquaculture companies in NL, Skretting Optiline, have a lipid content >30%; http://www.skretting.ca). The large, electron-dense droplets observed in gut epithelial cells of O. cyclops suggest that they may be assimilating large quantities of fish pellet-derived lipids. A high proportion of lipids could modify the fractionation between food and consumer (Post et al., Reference Post, Layman, Arrington, Takimoto, Quattrochi and Montaña2007).
Ophryotrocha cyclops may also consume a fraction of the materials present in flocculent matter. Although the exact composition of flocculent matter at our study site is unknown, its isotopic and trace element composition indicate relationships with fish pellets. Flocculent matter is composed of roughly 60% organic matter (unpublished data) and shows high concentrations of Zn and Cu (Table 2), likely due to Zn enrichment in fish pellets, selective Cu excretion in fishes, and Cu use in antifouling paints, as observed at other aquaculture sites (Chou et al., Reference Chou, Haya, Paon, Burridge and Moffatt2002; Brooks & Mahnken, Reference Brooks and Mahnken2003; Mendiguchía et al., Reference Mendiguchía, Moreno, Mánuel-Vez and García-Vargas2006; Dean et al., Reference Dean, Shimmield and Black2007; Sutherland et al., Reference Sutherland, Petersen, Levings and Martin2007). Ophryotrocha cyclops, however, does not accumulate Zn and Cu to a large extent.
Flocculent matter contains microbes, fish faeces, dorvilleid mucus and other sedimented organic matter, which collectively contribute to the viscosity and thickness of this material and explains the intermediate position of flocculent matter in Figure 2. The isotopic signature of flocculent matter is variable at different scales (within S3 and between sites), distinct from SPOM, and likely influenced by resident microbes: a mixture of heterotrophs and chemoautotrophs including sulphur oxidizers such as Beggiatoa spp. Chemoautotrophic bacteria show different degrees of specificity for 12C and 13C during carbon fixation, depending on which form of the Rubisco enzyme they contain (Robinson & Cavanaugh, Reference Robinson and Cavanaugh1995). The relative importance of heterotrophic to chemoautotrophic processes in flocculent matter should vary according to redox conditions and influence the carbon isotope signature. The δ13C of microbes is also influenced by available inorganic carbon: for instance, isotopically light methane at some seeps leads to light δ13C in microbes and in organisms that consume them, including dorvilleids (see Levin & Michener, Reference Levin and Michener2002; Levin et al., Reference Levin, Ziebis, Mendoza, Bertics, Washington, Gonzalez, Thurber, Ebbe and Lee2013).
The δ13C signature of Ophryotrocha cyclops was about 2‰ heavier than that of flocculent matter collected from sites S1 and S3, suggesting that the worms may have consumed part of this organic matter (considering fractionation). Also, the slightly more positive δ15N values of O. cyclops compared with flocculent matter are concordant with assimilation of this resource and a trophic level increase. However, the difference in δ34S (4.5 and 1‰ for S1 and S3, respectively) between flocculent matter and O. cyclops may indicate selective feeding (possibly on bacteria with lighter δ34S; Canfield, Reference Canfield2001) within the flocculent matter pool. Interestingly, the δ34S of flocculent matter from site S1 is similar to that of the sample of dorvilleid mucus and faeces from the same site, supporting the idea that O. cyclops feeds on flocculent matter (i.e. material that has gone through the gut shares the signature of flocculent matter). However, the difference in δ34S between dorvilleids and their faeces may indicate that they assimilate a particular fraction of flocculent matter (possibly the microbial component) and reject the rest (including most of the Cu and Zn-enriched fraction). Other observations support the likely consumption of microbes by O. cyclops: specimens collected from whalebones in Greenland were observed to consume white microbial filaments (K. Worsaae, pers. comm.) and dorvilleids from cold seeps contained filamentous microbes in their gut (Levin & Michener, Reference Levin and Michener2002).
The relatively high abundance of sulphur in Ophryotrocha cyclops provides further evidence that this dorvilleid likely consumes filamentous bacteria, and accumulates sulphur. The storage of elemental sulphur in Beggiatoa spp. is a defining feature of the genus (Teske & Nelson, Reference Teske, Nelson, Dworkin, Falkow, Rosenberg, Schleifer and Stackebrandt2006). While different species of Beggiatoa differ metabolically, elemental sulphur likely acts as a reservoir of electron donors within cells (Teske & Nelson, Reference Teske, Nelson, Dworkin, Falkow, Rosenberg, Schleifer and Stackebrandt2006); in some strains, elemental sulphur accumulation was linked to the availability of sulphide or thiosulphate (Nelson & Castenholz, Reference Nelson and Castenholz1981). The high variability in sulphur concentrations among O. cyclops samples may reflect small-scale differences in elemental sulphur content between sulphur-metabolizing bacteria at the sampling site.
Factors leading to observed spatial and temporal differences in Ophryotrocha cyclops isotopic signatures are not yet known. The δ13C of individuals from sites S1 and S2 was different from that of individuals from site S3, possibly reflecting differences in: (1) fish pellet composition and abundance (those sites were managed by different companies); (2) state of fish pellet degradation (there were no freshly deposited pellets at site S3); (3) sulphate reduction rates, which vary spatially and temporally around cages according to fine-scale conditions (Holmer & Kristensen, Reference Holmer and Kristensen1994, Reference Holmer and Kristensen1996); and (4) microbial species composition. The lower δ15N values of both O. cyclops and flocculent matter at site S3 also suggest that chemoautotrophic bacteria may comprise a relatively larger fraction of this organic matter pool at this site, and that dorvilleids are consuming these bacteria. Dorvilleids may also be opportunistic and able to prey on different food sources, as suggested by the presence of O. cyclops conspecifics at both aquaculture sites in NL and whalebones in Greenland (Salvo et al., Reference Salvo, Wiklund, Dufour, Hamoutene, Pohle and Worsaae2014). In cold seep dorvilleids, wide isotopic ranges were observed among conspecifics within a site, suggesting some degree of dietary flexibility; however, community analyses revealed trophic partitioning such that dorvilleid species specialized on different types of microbes (Levin et al., Reference Levin, Ziebis, Mendoza, Bertics, Washington, Gonzalez, Thurber, Ebbe and Lee2013). Compared with these cold seep dorvilleids (and considering differences in the inorganic carbon pool at cold seeps and at aquaculture sites), O. cyclops from this study most resemble, in their isotopic signature, cold seep species such as Ophryotrocha maciolekae and O. platykephale which are thought to consume mainly sulphur-oxidizing filamentous bacteria (Levin et al., Reference Levin, Ziebis, Mendoza, Bertics, Washington, Gonzalez, Thurber, Ebbe and Lee2013).
Our investigations at Newfoundland aquaculture sites lead us to conclude that: (1) there are trophic linkages between Ophryotrocha cyclops and fish pellets (reflected in Zn signatures, isotopic composition and lipid accumulation in gut epithelial cells); (2) O. cyclops select specific components from the flocculent matter and likely consume Beggiatoa-like bacteria (isotopic composition and sulphur content); and (3) both fish pellets and bacteria contribute to the complex nature of flocculent matter. Fatty acid or lipid composition/stable isotope analyses may be useful in determining the relative importance of microbes and fish pellets to the diet of O. cyclops.
At aquaculture sites, Ophryotrocha cyclops presence often coincides with that of Beggiatoa spp., based on visual observations of benthic images (Bungay, Reference Bungay2013; Hamoutene et al., Reference Hamoutene, Mabrouk, Sheppard, MacSween, Coughlan and Grant2013). Considering that O. cyclops most likely consume aquaculture-derived organic matter (including fish pellets undergoing degradation) and microbes such as Beggiatoa spp., we can put forward some hypotheses regarding the functional role of O. cyclops at these sites. First, feeding on organic waste can help accelerate the remineralization of this excess organic matter and aid in the recovery of benthic habitats at aquaculture sites. Through their feeding activities, O. cyclops may facilitate the transfer of energy and organic matter to higher trophic levels in these environments. Second, selective feeding on microbes can stimulate microbial productivity and accelerate nutrient cycling at the seafloor; in particular, O. cyclops may play important roles in the sulphur cycle. Third, the ability of O. cyclops to live and move within the flocculent matter layer may further enhance remineralization by increasing oxygenation rates. Further research is needed to clarify the relative importance of O. cyclops to nutrient cycling during periods of aquaculture production and fallowing, considering fluxes in population size for this species.
ACKNOWLEDGEMENTS
T. Bungay, K. Burt and D. Drover (Fisheries and Oceans Canada) assisted with initial specimen sampling; additional sampling was generously facilitated by B. Sweeney and J. Arsenault at Sweeney International Corporation. A. Pye assisted with stable isotope analysis, P. King and L. Hewa helped with trace element analysis, and M. Schaffer is thanked for SEM work. This project could not have been conducted without the aquaculture companies' authorization for site access and their provision of fish pellets.
FINANCIAL SUPPORT
Financial support was provided by an NSERC Engage grant (S.C.D., grant number EGP 446559-13, with industry partner Sweeney International Corporation) and by PARR project funding, Fisheries and Ocean Canada (D.H).
