Fatty acid trophic markers and trophic links among seston , crustacean zooplankton and the siphonophore Nanomia cara in Georges Basin and Oceanographer Canyon ( NW Atlantic )

1 institut de Ciència i tecnologia ambientals, universitat autònoma de barcelona, Campus Cn uab s/n, Cerdanyola del Vallés (barcelona) 08193, Spain. e-mail: Sergio.rossi@uab.cat 2 harbor branch oceanographic institution, Fort Pierce, Florida, 34946, uSa. 3 Department of Fisheries and aquatic Sciences, university of Florida, gainesville, Florida, 32653, uSa. 4 institut de Ciències del Mar (CSiC), Passeig Marítim de la barceloneta 37-49, 08003 barcelona, Spain.

the roles of gelatinous zooplankton in food webs mainly have been determined from the stomach contents of specimens caught in nets.unfortunately, such sampling can bias estimates of prey consumption because gelatinous zooplankton may feed on prey concentrated in cod ends, digest stomach contents during tows, or regurgitate prey (Youngbluth and båmstedt, 2001). in addition, gelatinous plankters are typically abraded or fragmented in nets to an extent that complicates identification and enumeration.in some cases, in situ observations provide reliable data on gut contents of gelatinous predators, but such observations are limited (robison, 2004).
analyses of fatty acid trophic markers complement analyses of stomach contents.For example, fatty acid composition integrates feeding behaviour over longer time scales and is not biased by digestion times (Dalsgaard et al., 2003).however, fatty acids are seldom unique to an organism, and changes in environmental conditions that affect metabolic rates can alter the production, storage or conversion of fatty acids (Dalsgaard et al., 2003).therefore, in the absence of data on metabolism, the most reliable evidence of trophic links arises from the transfer of multiple fatty acids in reasonable quantities, and fatty acid compositions should be viewed primarily as qualitative indicators of trophic links rather than quantitative indicators of the strength of such links (Dalsgaard et al., 2003).
this study examined fatty acids in the physonect siphonophore, Nanomia cara, two crustacean zooplankters (stage V copepodites of the calanoid copepod Calanus finmarchicus and adult euphausiids, Meganyctiphanes norvegica), and seston in an effort to identify trophic links in georges basin and ocea-nographer Canyon, nW atlantic.the ecological role of N. cara in this coastal region is poorly known, but this species should exert significant effects on its prey and competitors when it reaches densities of 10-100 colonies m -3 (rogers et al., 1978;Mills, 1995;Sommer et al., 2002).

Collection of samples
Seston, copepods, euphausiids and siphonophores were collected at depths of between 56 m and 831 m in georges basin and oceanographer Canyon from 11 to 25 September 2003 (Fig. 1; table 1). at each sampling station, depth, temperature, salinity and dissolved oxygen were measured with Seabird Sbe 25 Sealoggers.
Seston was extracted from 1-2 l of water collected in niskin bottles by filtering it through Whatman gF/F filters that had been precombusted for 5 h at 450ºC.individual filters with seston were placed into cryotubes and frozen in liquid nitrogen.Four sets of duplicate samples from georges basin (n = 8) and three sets of duplicate samples from oceanographer Canyon (n = 6) were analysed for particulate organic carbon and nitrogen.Fatty acid compositions were determined for three sets of duplicate samples and one unreplicated sample from georges Fatty acid compositions were determined for pooled samples of Calanus finmarchicus stage V copepodites from georges basin (n = 3) and pooled samples of adult Meganyctiphanes norvegica from oceanographer Canyon (n = 2).Pooled samples consisted of either 20 C. finmarchicus or 1-2 M. norvegica that had been placed in separate cryotubes and frozen in liquid nitrogen.Crustaceans for pooled samples were captured with forceps from collections taken with a suction sampler attached to a Johnson-Sea-Link submersible (Youngbluth, 1984). in total, six colonies of the physonect siphonophore Nanomia cara were collected in georges basin (n = 5) and oceanographer Canyon (n = 1).each colony was captured in a different 6.5-l acrylic sampler that had been washed with 1n hydrochloric acid prior to use.Colonies without prey visible in their gastrozooids were kept at 6-10ºC on a bed of ice under a dissecting microscope while nectosomes were separated from siphosomes (Fig. 2).each nectosome or siphosome was placed in a separate cryotube and frozen in liquid nitrogen.

Analysis of particulate organic carbon and nitrogen in seston
Filters containing seston were removed from liquid nitrogen and dried at 60ºC for 24 h.inorganic material was destroyed by keeping the dried filters in air saturated with hydrochloric acid for 48 h (rossi and gili, 2005).Filters were dried further at 60ºC for another 24 h.Particulate organic carbon and nitrogen were measured with a Perkin-elmer 2400 autoanalyser (Doval et al., 1999).

Extraction and quantification of fatty acids
Fatty acids were extracted from samples of seston, copepods, euphausiids and siphonophores that had been lyophilised for 12 h at -100ºC and 100 mbar.after lyophilisation, glass fibre filters with seston were sonicated in 2:1 dichloromethane-methanol three times for 10 min each time.all other samples were ground gently in a 5 ml glass homogeniser, and sonicated three times for 20 min each time in 1 ml of 2:1 dichloromethane-methanol. after each sonication, the solvent was separated from particles by centrifugation.the extracts for each sample were combined, evaporated under vacuum to 0.5 ml, and hydrolysed overnight with 2 ml of 6% potassium hydroxide and methanol.neutral fractions were recovered with three 2 ml extractions using n-hexane, and then acidic fractions were recovered using nhexane that had been acidified to ph 2 with aqueous 6n hydrochloric acid.the acidic fractions were reduced to 0.5 ml and esterified overnight with 3 ml of 10% boron trifluoride-methanol. the resulting complexes were destabilised with 2 ml of water, and fatty acids were recovered as their methyl esters by extracting three times with 2 ml of n-hexane (rossi et al., 2006).
Quantitative gas chromatography was performed with an agilent 5890 Series ii instrument equipped with a flame ionisation detector and a splitless injector.the Db-5 column was 30 m long with an internal diameter of 0.25 mm and a 0.25 µm coating of phenyl-methylpolysiloxane. helium was used as a carrier gas at 33 cm s -1 .the oven temperature was programmed to increase from 60 to 300ºC at 6ºC min -1 .injector and detector temperatures were 270 and 310ºC, respectively.Methyl esters of fatty acids were identified by comparing their retention times to those of standard fatty acids (Supelco®).Fatty acids were quantified by integrating areas under peaks in the gas chromatograph traces, with calibrations derived from an external standard containing different methyl esters.

Evaluation of trophic links
Semi-strong, hybrid multidimensional scaling was used to ordinate relative fatty acid concentra-tions expressed as percentages of the total pool of fatty acids (belbin, 1989).Separate ordinations were conducted using data from georges basin and oceanographer Canyon to ensure that differences in fatty acid compositions between locations did not obscure patterns within a location.Means were calculated for duplicate seston samples, which yielded 4 values for seston from georges basin and 3 values for seston from oceanographer Canyon.ordinations were based on bray-Curtis dissimilarities, with linear regression applied to dissimilarities below 0.9 and ordinal regression applied to values above 0.9.ordinations in three dimensions yielded stress values below 0.1, which were considered acceptable representations of the data.
if ordinations indicated that fatty acid compositions of samples within a trophic level were more similar to each other than to different trophic levels, then comparisons of relative concentrations across trophic levels were used to elucidate trophic markers.Fatty acids were classed as potential trophic markers if they: 1) represented approximately 2% or more of the relevant fatty acid pools (Dalsgaard et al., 2003) and 2) occurred in similar percentages in different trophic levels or at a higher percentage in the higher trophic level, which indicated that they were transferred conservatively or accumulated through trophic links.

Environmental conditions
all samples were collected below the thermocline, which was at approximately 50 m at both locations (table 2).at these depths, temperatures, salinities and dissolved oxygen concentrations were similar and stable during the sampling period, so spatiotemporal variations in fatty acid metabolism were unlikely.
twenty-one fatty acids were isolated from Nanomia cara, with siphosomes yielding more fatty acids than nectosomes (table 4).Fatty acids in individual nectosomes and siphosomes comprised 34-82% SFas, 7-45% PuFaS and less than 20% MuFas.Sums of relative fatty acid concentrations indicated contributions from bacillariophyceae, Dinophyceae and Prymnesiophyceae. in contrast, ratios of selected fatty acid concentrations did not support this interpretation.as expected for a predator, the mean ratios of 18:1 (n-7) to 18:1 (n-9) for N. cara were lower than those recorded for their potential prey, i.e.Calanus finmarchicus in georges basin and Meganyctiphanes norvegica in oceanographer Canyon. in contrast, the mean ratios of 20:5 (n-3) to 22:6 (n-3) and PuFas to SFas did not support the conclusion that N. cara fed at a higher trophic level than its potential prey.

Characterisation of trophic links
three-dimensional ordinations confirmed that samples of seston, Calanus finmarchicus stage V copepodites and Meganyctiphanes norvegica had fatty acid compositions that were more similar to each other than to samples from different trophic levels (Fig. 3a and b). as shown by the separation of relevant points in the ordinations, samples of Nanomia cara nectosomes and siphosomes had the most variable fatty acid compositions (Fig. 3a and b).however, multiple samples of N. cara from georges basin did not display a consistent pattern related to the date or time of sampling (Fig. 3a).overall, the data indicated that comparisons of relative fatty acid analyses of all samples yielded 22 fatty acids (table 4).individual fatty acids represented less than 1% to over 30% of total fatty acid pools.in an effort to elucidate trophic markers, fatty acids were classified as: 1) compounds with uncertain value as indicators if they represented less than 2% of every fatty acid pool; 2) trophic markers if they occurred in similar percentages in different trophic levels or at higher percentages in higher trophic levels, which indicated conservative transfer or accumulation; 3) indicators of other food sources or de novo synthesis if they were found primarily in a higher trophic level; 4) compounds that were not transferred conservatively if they appeared primarily in a lower trophic level.

DiSCuSSion
the transfer of fatty acids to higher trophic levels is a complex process.Detection of trophic markers is enhanced if higher trophic levels feed extensively on the foods investigated and samples are taken during a period of anabolism rather than catabolism (Falk-Petersen et al., 1987;St. John and lund, 1996;Kirsch et al., 1998;Fukuda and naganuma, 2001;Falk-Petersen et al., 2002;Dalsgaard and St. John, 2004).
Without detailed information about metabolism, trophic links should not be derived from fine-scale comparisons of quantitative differences in relative concentrations of fatty acids.instead, identifying trophic links must rely primarily on the apparent transfer of multiple fatty acids in reasonable quantities among samples collected during a time without significant variation in environmental conditions.We interpreted our results within this context by looking for environmental variability that could mask trophic links, comparing our data to previous reports to establish their reliability, and applying a consistent process to identify trophic markers.

Environmental variability
environmental variability can alter physiological responses of organisms and mask trophic links.however, fatty acid compositions have been reported to be stable unless environmental conditions changed noticeably.For example, fatty acid concentrations of mixed phytoplankton were stable when the trophic links for Gadus morhua larvae were analysed over 9-10 d in the absence of a phytoplankton bloom (Klungsøyr et al., 1989). in addition, changes in fatty acid compositions of fishes, copepods, phytoplankton and seston following phytoplankton blooms remained stable and detectable for 2-3 months (Pedersen et al., 1999;reuss and Poulsen, 2002;Parrish et al., 2005).
During our 15-d sampling period, environmental conditions remained stable below the thermoclines in georges basin and oceanographer Canyon; therefore, fatty acid compositions were not expected to change to an extent that would mask trophic markers.ordinations confirmed that the relative fatty acid concentrations of seston, Calanus finmarchicus stage V copepodites, adult Meganyctiphanes norvegica and Nanomia cara did not vary in a consistent pattern across the sampling interval.

Comparisons with previous reports
Seston in georges basin and oceanographer Canyon was sparse and refractory, which indicated that this particulate matter was a poor source of nutrition.in fact, concentrations of particulate organic carbon (93-103 µg C l -1 ) were low and C:n ratios (25) were high compared with values reported for a mixed water column on georges bank between January and June 1999 (150-300 µg C l -1 ; ratios from 3 to 8; townsend and thomas, 2002).one explanation for these findings arises from visual observations made from the Johnson-Sea-Link submersible.these observations suggested that marine snow aggregates formed a large percentage of the seston that we sampled.During non-bloom conditions, such particles can become enriched in organic carbon and depleted in nitrogen due to bacterial activity (Silver and alldredge, 1981;alldredge and Youngbluth, 1985).overall, low concentrations of organic carbon and high C:n ratios were consistent with previous reports of low nutrient levels and decreased primary production in the photic layer between June and october accompanied by rapid recycling of elements in shallow water (roman et al., 1995; townsend and thomas, 2002; bisagni, 2003).
Compared with seven sets of data reported previously, our seston samples contained different fatty acids and different relative concentrations of some fatty acids (appendix 1).our coefficients of variation ranged from 0.06 to 2.65, which indicated that the reliability of our measurements was similar to that of reports in the literature, with coefficients of variation ranging from 0.02 to 3.00 (Klungsøyr et al., 1989;Mayzaud et al., 1989;reuss and Poulsen, 2002;Parrish et al., 2005).out of 27 fatty acids with relative concentrations of 1% or higher in any report, samples from georges basin matched 15 fatty acids reported elsewhere, and samples from oceanographer Canyon matched 12 fatty acids reported elsewhere.our samples were the only ones that contained 12:0 in relative concentrations of 1% or more, and 22:0 was found in relative concentrations that were at least 10 times higher in our samples.Sums and ratios of relative concentrations considered indicative of contributions from various classes of phytoplankton yielded inconsistent evidence of such contributions to the seston that we sampled (Fahl and Kattner, 1993;reuss and Poulsen, 2002;Dalsgaard et al., 2003). in fact, our samples most closely matched those taken from oligotrophic antarctic waters, with relatively high concentrations of SFas and low concentrations of PuFas (Fahl and Kattner, 1993). in addition, fatty acids become saturated as particulate organic matter is oxidised in the water column, especially during periods with low nutrient availability, high levels of detritus, and limited phytoplankton growth (goutx and Saliot, 1980;Mayzaud et al., 1989;Fahl and Kattner, 1993;baldi et al., 1997;Parrish et al., 2005). in contrast to those of seston, relative fatty acid compositions of Calanus finmarchicus stage V copepodites and Meganyctiphanes norvegica were similar to each other and to reports in the literature (appendices 2 and 3). the coefficients of variation among our replicates (0.02-1.73) were within the ranges reported elsewhere (0.01-2.25), which demonstrated the reliability of our measurements.our samples of Calanus finmarchicus and M. norvegica shared 16 fatty acids that occurred in relative concentrations of 1% or more. in addition, our samples contained approximately 60% of the fatty acids found in relative concentrations of 1% or more in previous studies (14 out of 22 fatty acids for C. finmarchicus and 13 out of 21 fatty acids for M. norvegica).our samples of C. finmarchicus contained 20:3 (n-6), which had not been reported previously in concentrations of 1% or more, and 3-5 times the relative concentrations of 15:0, 20:0, 18:1 (n-7) and 18:3 (n-3).our samples of M. norvegica contained 20 times the relative concentration of 20:3 (n-6). in addition, our samples of C. finmarchicus had approximately twice the relative concentration of SFas reported in all but two studies.in contrast, relative concentrations of MuFas were consistent with five of seven previous reports and lower than the other two, and our concentrations of PuFas were consistent with four previous reports, lower than two and higher than one.the relative concentration of 22:1 (n-11) in C. finmarchicus agreed with reports that this fatty acid is synthesised by herbivorous copepods, but 20:1 (n-9), another fatty acid reported to be synthesised by herbivorous copepods, was absent from our samples.the relative concentrations of SFas, MuFas and PuFas in our samples of M. norvegica were approximately equal, which was consistent with most other reports.
our data represent the first report of relative fatty acid concentrations for Nanomia cara. in general, the relative concentrations of fatty acids in nectosomes and siphosomes of N. cara were similar to those reported for 14 species of gelatinous zooplankton from arctic and antarctic regimes (appendix 4).Coefficients of variation (0.19-2.45) overlapped ranges reported elsewhere (0.03-2.00). in addition, our samples contained 18 out of 30 fatty acids with relative concentrations of 1% or more in any report.Nanomia cara contained 22:0, 23:0, 26:0, 22:2 (n-6) and 20:3 (n-6), which had not been reported to occur in relative concentrations of 1% or more in any other species.Nanomia cara also had more SFas than all species other than the ctenophore Pleurobrachia pi-leus, and fewer MuFas than all other species, which may be related to SFas being transferred through the trophic web from seston. in summary, the relative concentrations of fatty acids in our samples provided a reliable, qualitative basis for interpreting trophic links.our samples confirmed that, in September, seston found below the thermocline in georges basin and oceanographer Canyon was a poor source of nutrition.in general, our samples of Calanus finmarchicus, Meganyctiphanes norvegica and Nanomia cara were qualitatively similar to previous reports.a detailed interpretation of quantitative differences in relative concentrations of fatty acids was obviated by a lack of information about metabolism in all studies.

Implied trophic links
evidence for trophic links between seston and two common crustaceans in georges basin and oceanographer Canyon was inconsistent.only four fatty acids, i.e. 14:0, 16:0, 16:1 (n-7) and 18:1 (n-7), were classified as trophic markers linking Calanus finmarchicus stage V copepodites or Meganyctiphanes norvegica to seston.two of these trophic markers were SFas, and the relatively high concentrations of SFas in both our seston and C. finmarchicus samples also suggested a trophic link.however, three other SFas, i.e. 12:0, 18:0 and 22:0 were not transferred from seston to crustacean grazers.Furthermore, both C. finmarchicus and M. norvegica contained six fatty acids that did not appear to be derived from the seston sampled in this study.in addition, sums and ratios of relative fatty acid concentrations did not indicate strong contributions from bacillariophyceae, Dinophyceae or Prymnesiophyceae (Pedersen et al., 1999;reuss and Poulsen, 2002;Dalsgaard et al., 2003). in combination with C:n ratios indicating that the seston was refractory, these results did not strongly suggest a trophic link from seston to crustaceans in georges basin or oceanographer Canyon.
inconsistent evidence of trophic links between seston and common crustaceans may be related to physiological requirements or feeding behaviour.in September, the metabolic demands and feeding rates of Calanus finmarchicus stage V copepodites found below the thermocline may have been decreasing due to the onset of diapause (Miller et al., 1991;Durbin et al., 1997;Saumweber and Durbin, 2006).Meganyctiphanes norvegica appeared to feed at a higher trophic level than C. finmarchicus, as shown by mean ratios of PuFas to SFas, 18:1 (n-7) to 18:1 (n-9), and 20:5 (n-3) to 22:6 (n-3). in addition, adult M. norvegica contained relatively high concentrations of 22:1 (n-11), which has been reported as a trophic marker indicating predation on C. finmarchicus (Dalsgaard et al., 2003).thus, the indications of trophic links to both seston and C. finmarchicus confirmed previous reports that M. norvegica is omnivorous and preys on copepods (båmstedt and Karlson, 1998;lass et al., 2001).
nine or ten common fatty acids represented potential trophic markers linking Nanomia cara to Calanus finmarchicus stage V copepodites in georges basin and Meganyctiphanes norvegica in oceanographer Canyon, respectively.these trophic markers represented approximately 70-85% of the fatty acid pools for N. cara.however, ratios of fatty acids provided inconsistent evidence that N. cara fed at a higher trophic level than C. finmarchicus or M. norvegica, which raises questions about the reliability of such ratios.in addition, as reported for seston and crustaceans, sums and ratios of fatty acid compositions did not indicate that N. cara derived fatty acids from bacillariophyceae, Dinophyceae or Prymnesiophyceae (Pedersen et al., 1999;reuss and Poulsen, 2002;Dalsgaard et al., 2003).
the transfer of significant amounts of fatty acids from potential prey to Nanomia cara was not surprising.Siphonophores are known to assimilate over 90% of the carbon and nitrogen found in their prey (Purcell, 1983). in addition, siphonophores have been reported to prey on small, crustacean zooplankters, with the capacity to ingest 3-69% of available copepod biomass in areas where colonies occurred in high densities (rogers et al., 1978;robison et al., 1998;Pagès et al., 2001;Youngbluth et al., personal observation).in fact, gastrozooids of N. cara captured in georges basin and oceanographer Canyon contained Calanus finmarchicus stage V copepodites and Meganyctiphanes norvegica (Youngbluth et al., personal observation).Furthermore, oil sacs have been observed in overwintering C. finmarchicus stage V copepodites taken from the gulf of Maine in autumn and oil droplets have been observed in N. cara, particularly in gastrozooids and palpons (rogers et al., 1978;Fig. 2).other gelatinous zooplankters also contained highly refractive, lipid droplets in the lumens of their digestive systems, but the droplets we observed were not confined to the digestive system (larson and harbison, 1989; Fig. 2). in fact, N. cara could have been accumulating fatty acids at the time of our study because we observed evidence of reproduction.other invertebrates have been reported to accumulate fatty acids before reproducing, especially PuFas for vitellogenesis and development (Mayzaud et al., 1999;albessard et al., 2001;hudson et al., 2004).
in conclusion, our study suggested conservative transfer or accumulation of fatty acids from Calanus finmarchicus stage V copepodites and Meganyctiphanes norvegica to Nanomia cara in georges basin and oceanographer Canyon.the data indicated that these common, secondary consumers contributed significantly to the diet of this siphonophore.in contrast, seston appeared to contribute little to its diet at these two locations.aCKnoWleDgeMentS We greatly appreciate the assistance of the crew of the r/V Seward Johnson, the crew of the Johnson-Sea-Link II submersible, and all the participants in the research cruise.Pilar teixidor provided tools and guidance for gC-MS analyses.a grant to MJY from the national Science Foundation (nSF-0002493), the european Project eurogel, and uSDa CriS Project Fla-FaS-03978 supported this work.this is contribution no.1696 to the harbor branch oceanographic institution.this paper is dedicated to the memory of our friend and colleague, Francesc Pagès, whose curiosity and enthusiasm did and always will inspire us.reFerenCeS ackman, r.g., C.a. eaton, J.C. Sipos, S.n.hooper and J.D. Castell.

Table 1 .
-Details for collection of samples.aM, daytime; PM, nighttime; PoC:Pon, ratio of particulate organic carbon to particulate organic nitrogen.
a Position of initial double bond not reported.Appendix 3. -Fatty acid concentrations as percentages of the total fatty acid pool in Meganyctiphanes norvegica.1,northatlantic from fin whale stomachs in october and november(n = 3; ackman et al., 1970); 2, balsfjorden, norway in november and December (Falk-Petersen et al., 2002)entrations as percentages of the total fatty acid pool for gelatinous zooplankton.1,AtollawyvilleiStygiomedusagigantea from antarctic waters in January and February(nelson et al.,   2000).13,Mertensiaovatum and 14, B. cucumis from Kongsfjorden, norway in august and September(Falk-Petersen et al., 2002).nectosomes, Nanomia cara nectosomes (n = 6) and Siphosomes, N. cara siphosomes (n = 6) from georges basin and oceanographer Canyon in September.Fatty acids with concentrations ≥ 1.0% in one or more references.