|Home | About | Journals | Submit | Contact Us | Français|
Three-dimensional data on the mesoscale distribution of hydrography and mesozooplankton were collected at the Polar Front, northwestern Barents Sea, in spring 2008 (29 April–15 May) using a combination of multinet and towed instrument platform equipped with Laser Optical Plankton Counter, fluorometer and CTD. Trophic levels (TLs) within the zooplankton community (whole community and size-separated) were analysed for three consecutive periods using biovolume spectrum theory, which proved to be a powerful tool in the physically and biologically variable frontal system. Trophic structure was highly variable in time and across the Polar Front, but was mostly related to the phytoplankton bloom (as determined by fluorescence). High TLs of 5.5 within the zooplankton community were observed outside bloom situations (mostly in Atlantic Water) and were likely due to increased omnivory of Calanus spp., which dominated the large zooplankton size group that had a lower TL (2.2) during the bloom than outside blooms (max. TL 5.6). A strong input of herbivorous barnacle nauplii (Cirripedia) into the upper layer (35 000 ind. m−3 in net samples) substantially decreased mean TL in the marginal ice zone. Differences in TL estimates based on biovolume spectrum theory and other methods (stable isotopes, lipid markers, dietary analyses) are discussed.
The development of biomass spectrum theories has given us a strong theoretical background to understand important processes within mesozooplankton communities based on semi-automatic sampling (Platt and Denman, 1978; Heath, 1995; Zhou and Huntley, 1997). Using time-sequences of measured zooplankton biomass spectra, it is possible to estimate in situ growth and mortality at the mesoscale (Silvert and Platt, 1978; Zhou and Huntley, 1997; Edvardsen et al., 2002; Zhou et al., 2004). The slope of a biomass spectrum provides the information on the trophic structure of a mesozooplankton community (Zhou, 2006). These mathematical theories and interpretations of biomass spectrum features allow us not only to observe the relationship between sizes and taxonomy, but also to understand the population and trophic dynamics.
Energy fluxes within aquatic systems determine the shape of the biomass spectrum (Platt and Denman, 1978; Silvert and Platt, 1978; Sprules and Munawar, 1986; Heath, 1995; Zhou and Huntley, 1997; Zhou, 2006). Platt and Denman (Platt and Denman, 1977, 1978) described the energy flux through a given size interval as being governed mainly by individual growth within the size interval and by metabolism. Heath (Heath, 1995) characterized the flow of energy as a balance between individual and population growth on one side and mortality on the other side. Zhou and Huntley (Zhou and Huntley, 1997) unified both approaches and developed a mathematical theory of population dynamics that describes the energy flow through the biomass spectrum based on the distribution function of abundance and the law of the conservation of mass. The biomass spectrum theory developed by Zhou and Huntley (Zhou and Huntley, 1997) has thus a firm theoretical base and does not rely on any empirical assumptions, such as, for example, a constant predator–prey size ratio. Furthermore, the theory includes all sinks and sources that contribute to the flow of biomass through a given size interval. Recent research has pointed out the coherence between biomass spectrum theory and an individual-based model (IBM) (Law et al., 2009), thereby further confirming biomass spectrum theory. The stochastic IBM was built starting from processes at the level of individuals and yielded a deterministic limit that was in close agreement to the partial differential equations of biomass spectrum theory (Law et al., 2009).
Productive systems are characterized by a high intercept of the biomass spectrum. Enhanced primary production leads to an accumulation of biomass at small sizes and hence a high intercept (Zhou, 2006). In a time-varying system, accumulated biomass at small sizes can propagate along the biomass spectrum towards larger sizes. A high intercept of the biomass spectrum thus also denotes potentially higher secondary production. During a typical spring-bloom situation, for example, large amounts of phytoplankton biomass are transferred first to reproducing herbivores and then to developing mesozooplankton cohorts. These evolving mesozooplankton cohorts that feed and grow can be identified in the spectrum as waves propagating along the biomass spectrum (Silvert and Platt, 1978; Zhou and Huntley, 1997).
The slope of the biomass spectrum and the assimilation efficiency of the community provide information on the biomass recycling internally in the pelagic system, i.e. how many times biomass has been transferred from one pelagic organism to the next (Zhou, 2006). The slope of the biomass spectrum thus gives an instant picture of the pelagic system, but simultaneously contains information that was integrated in the system over time. A flat slope of the biomass spectrum indicates that the biomass has been recycled internally several times (Zhou, 2006). Trophic levels (TLs) computed by biomass spectrum theory can reach relatively high values in the plankton community because all fluxes are taken into account. This reflects the understanding today of the trophic coupling between phytoplankton, the microbial loop and mesozooplankton (Kirchmann, 2000; Fenchel, 2008). In a pelagic system where phytoplankton biomass is channelled via microzooplankton to copepods and back to detritus, microzooplankton and copepods, for example, the copepods have TL 6.
The number of trophic links between the microbial loop and the classic pelagic food web varies over short time scales. Trophic coupling is weakest during bloom periods when the most important mesozooplankton grazers feed nearly exclusively herbivorously (Levinsen et al., 2000) and indeed have a low TL (Tamelander et al., 2008). Outside the bloom, when the availability of phytoplankton is low and microzooplankton is relatively more abundant, mesozooplankton grazers have a more omnivorous diet (Levinsen et al., 2000; Basedow and Tande, 2006) leading to an increase in TL (Tamelander et al., 2008).
The Barents Sea is occupied by three main water masses: Arctic Water on the banks and in the northern Barents Sea, Atlantic Water in the deeper parts of the central Barents Sea and Coastal Water in the South along the Norwegian Coast (Loeng, 1991). The Polar Front, which separates Arctic from Atlantic Water, is thus associated not only with a change in salinity and water temperature but also with a gradient in bottom depth. At the front, mesoscale currents, meanders and eddies produce upwelling and vertical mixing of nutrients or enhance stratification. This leads to enhanced short-term primary production at the mesoscale, and strong gradients of plankton communities. During spring, the retreating ice-edge is found in close vicinity to the Polar Front. The abiotic environment thus differs greatly over small spatial scales, making it challenging to sample the frontal system adequately. The high inter-annual and local variability in the development of the phytoplankton bloom in the marginal ice zone further increases sampling complexity (Engelsen et al., 2002; Reigstad et al., 2002).
Our aim was to analyse the impact of mesoscale physical processes on food web dynamics within the plankton community at the Polar Front during spring 2008. We sampled the frontal area using a combination of semi-automatic high-resolution mapping and detailed analyses at stations. Applying biovolume spectrum theories to the data, we investigated differences between water masses in the trophic structure of the zooplankton community in the northwestern Barents Sea.
The Polar Front in the Barents Sea meanders along topography and separates Atlantic Water from colder, less saline Arctic Water (Loeng, 1991). In an area of the Polar Front southeast of Hopen Island (Fig. 1), data on the spatial and temporal distributions of hydrography and mesozooplankton were collected during the IPY-NESSAR project in spring 2008 (29 April–2 May and 9–14 May). For data collection, a towed instrument platform (Scanfish; GMI, Denmark) was equipped with a Laser Optical Plankton Counter (LOPC; Brooke Ocean Technology Ltd, Canada), a CTD (SBE 911plus, Seabird Electronics Inc., USA) and a fluorometer (F; Seapoint Chlorophyll Fluorometer, Seapoint Sensors Inc., USA). The Scanfish undulated between surface and 75 m (or between surface and 15 m above bottom depth if bottom depth was below 90 m) while it was towed along transects (Fig. 1) at a speed of 6–7 knots. Each tow of the Scanfish lasted for 2–3 days during which LOPC-CTD-F data were logged continuously (2 Hz). Before and after the tows, CTD profiles and discrete water and zooplankton net samples (Multinet, 180 µm mesh width, 0.25 m2 mouth opening) were collected at 12 stations in the study area to compare with and interpret the LOPC-CTD-F data (Table I). The depth layers sampled by multinet (vertical tows) were chosen based on hydrography and fluorescence profiles at each station to facilitate data interpretation.
Zooplankton samples were preserved in a solution of 20% fixation agent (50% formalin buffered with hexamine, 50% anti-bactericide propandiol) and 80% seawater. In the laboratory, zooplankton was counted and identified to species and stage or to the lowest taxonomical level feasible under a stereo-microscope. Calanus finmarchicus and Calanus glacialis as well as the younger stages of Calanus hyperboreus were distinguished by size (Daase and Eiane, 2007). Abundance was calculated based on filtered volume obtained from the flow meters attached to the multinet.
The LOPC was designed to overcome some of the problems associated with the OPC, it counts and measures particles that pass through a laser beam inside the instrument as the LOPC moves through the water (Herman et al., 2004). The size of particles is returned as digital size, which can be converted into equivalent spherical diameter (ESD), a quantity that yields the diameter that a particle had if it were an opaque sphere. ESD is thus a property describing the size as well as the transparency of a particle. We calculated the ESD as described in Gaardsted et al. (Gaardsted et al., in review). LOPC data were analysed using the python programming language (version 2.6.2) and basically following the instructions in the LOPC manual (Anonymous, 2006). The LOPC detects living and dead particles in the size range of ca. 0.1–35 mm ESD (Herman et al., 2004). Major non-living particles in the water column are marine snow and phytoplankton aggregates. As discussed at length in Edvardsen et al. (Edvardsen et al., 2002), marine snow and phytoplankton aggregates >0.1 mm are very likely eroded when towing an OPC at 6 knots because of micro-scale turbulence and associated shear stress at the opening of the OPC and there is no reason to believe that shear stress should be less at the opening of the LOPC. Therefore, though we cannot completely rule out that some phytoplankton aggregates might have been counted, we are assured that the majority of particles counted by the LOPC in this study were zooplankton particles. For each individual zooplankton particle, body volume was computed from its ESD. For biovolume spectra analyses, data were then grouped into 50 size groups of equal (body volume) increments to increase statistics and to clarify data presentation.
The ESD of all zooplankton genera that were abundant in net samples was estimated either based on literature concerning the OPC (Edvardsen et al., 2002; Basedow et al., 2008) or, for genera for which no literature data exist, estimated based on the ESD of similar sized genera. For data interpretation and presentation (Fig. 3), the data collected by LOPC were then divided into four size groups: S, M, L and XL (Table III). To our knowledge, no data on the calibration of the LOPC for northern marine plankton in spring/summer exist; therefore, we chose relatively coarse size ranges, each of which incorporates different species and stages. Particles <0.25 mm ESD were not included into our analyses because very small particles that result from the erosion of phytoplankton aggregates and marine snow would fall into this group and would corrupt our analyses. Few animals >5.5 mm were observed in Arctic and Polar Front Water and none in Atlantic Water. The XL size group thus consisted mostly of particles between 2 and 5.5 mm ESD.
Data on hydrography and zooplankton distribution are presented from selected transects (Figs 2 and and3).3). These figures were prepared by gridding and interpolating all data points collected along the transects. For interpolation, we used the natural neighbour interpolation that is built in pythons matplotlib library. The salinity plot overlaid onto the map (Fig. 1) was prepared much the same way, the only difference being that data points between 20 and 30 m from all transects were interpolated. All other figures were also prepared using the matplotlib library (version 0.98.5.2) (Hunter, 2007) in python. To simplify data presentation and aid comparison with LOPC data, abundance data obtained from Multinet are presented from the upper layer (0–ca. 60 m) and lower layer (ca. 60 m–bottom) in Table II and Fig. 4. The division at ca. 60 m was chosen because net samples were obtained from above and below 60 m at most stations (Table I) and because 0–60 m is close to the 0–75 m depth layer that was sampled by the LOPC.
Biovolume spectra are analogue to biomass spectra and are used if only the size and no weight of individuals is available (Edvardsen et al., 2002; Quinones et al., 2003; Zhou et al., 2004). The normalized biovolume spectrum b (Platt and Denman, 1978; Zhou and Huntley, 1997; Edvardsen et al., 2002) is defined as
with w being the body volume of an individual zooplankton in cubic millimetre. We computed the biovolume spectra for each of the three sampling times (30 April–2 May, 10–12 May, 14–15 May) and each of the three water masses (Arctic Water, Polar Front Water, Atlantic Water) separately in order to analyse differences in time and between water masses. To compare the spectra with each other, the slope of the observed biovolume spectra was calculated by fitting (least-squares) a linear function to each spectrum. For each biovolume spectrum, one slope was fitted to the whole size range and four to each of the size ranges of the four zooplankton groups S, M, L and XL (Table III). Because the LOPC measured few particles >5.5 mm ESD, the slope for the XL size group was fitted to the biovolume spectrum between 2.0 and 5.5 mm ESD only. On the basis of the slope of the biovolume spectrum and the mean assimilation efficiency of zooplankton (ηn), we estimated the number of TLs within the zooplankton community (Zhou, 2006):
In contrast to general biomass spectrum theory, which does not depend on any empirical assumptions, the computation of TLs is based on the assumption that the biovolume spectrum can be linearized on a logarithmic scale (Zhou, 2006). Furthermore, the assimilation efficiency of the zooplankton community has to be known to calculate TLs (Zhou, 2006). We applied a mean assimilation efficiency of 70%, a value commonly used for copepods. Data on assimilation efficiency in zooplankton are limited, and the existing data show a great variability depending on food source and species (Mauchline, 1998). For Pseudocalanus spp., mean assimilation efficiency may be lower than 70%, values reported range from 10% to 71% (Harris and Paffenhöfer, 1976; Corkett and McLaren, 1978; Koski et al., 1998), whereas for carnivores assimilation efficiency may be as high as 98% (Mauchline, 1998). TL estimates do not depend strongly on assimilation efficiency (Zhou, 2006, their Fig. 2), assuming a slope of the biovolume spectrum of −1.5, for example, and changing the assimilation efficiency from 70% to 50%, the estimated TL would increase from 1.6 to 2.0 [Equation (2)]. We therefore chose to hold assimilation efficiency constant with size. Thus, though the estimated TLs may not represent exact TLs, the variations of TL represent the differences in trophic structure between plankton communities. To analyse food web dynamics within the zooplankton community, we first estimated the number of TLs within the whole zooplankton community (size groups S to XL). We then estimated the number of TLs within those size groups for which a significant slope could be fitted to the biovolume spectra.
We worked in the frontal area and most of our study area was occupied by Polar Front Water with a salinity between 34.8 and 35.0 (Loeng, 1991). Polar Front Water was mainly found between the 150 and 200 m isobaths. Arctic Water with a salinity between 34.3 and 34.8 occupied the shelf in the western part and Atlantic Water with a salinity >35.0 was found in the very East of the study area (Figs 1 and and2,2, top panel). Temperatures were around −1°C in Arctic Water and around 3°C in Atlantic Water and they remained relatively stable throughout the 2 weeks (Fig. 2, middle panels). The westernmost stations in Arctic Water were close to the ice edge where the stratification of the water column had started, while the water column in Atlantic Water was still well mixed (Fig. 2) down to the bottom (CTD data, not shown).
On 1 May, fluorescence was highest in Arctic Water close to the ice edge. On 10 May, highest values were still found in Arctic Water but farther west and on 14 May highest fluorescence values had increased by one order of magnitude and were now observed in Polar Front Water in the centre of the study area (Fig. 2, lower panels). In Atlantic Water, very low fluorescence was observed throughout most of the study area (Fig. 2), only at the southeastern-most corner were fluorescence values slightly elevated at the end of the study (values around 1, data not shown).
In general, abundances of all mesozooplankton groups ranged from low to very low in Atlantic Water, whereas those in Polar Front Water and in Arctic Water were relatively high (Table II, Fig. 3). The zooplankton community in Arctic Water on the shelf was completely dominated by meroplanktonic larvae (Fig. 4) and especially by barnacle larvae, which reached extremely high abundances of up to 35 000 ind. m−3 (Table II). Also unidentified eggs (0.2–0.4 mm) were abundant in Arctic Water (Table II, Fig. 4). Barnacle larvae and unidentified eggs were also abundant in Polar Front Water on the shelf, but were mostly absent in deeper areas and in Atlantic Water (Table II). The distribution of small copepods (Oithona spp., Microcalanus) did not follow any obvious pattern in relation to water masses or depth (Table II); however, they were more important below 60 m, where eggs and larvae were comparatively rare (Fig. 4). Abundances of small zooplankton measured by LOPC (S size group) revealed maximum abundances of 400 000 ind. m−3, i.e. one order of magnitude higher than measured by multinet (Fig. 3, first row). The mesoscale pattern of the distribution of small zooplankton was closely linked to areas with high fluorescence. At a small scale, however, small zooplankton was more patchily distributed than fluorescence (Figs 2 and and33).
Of the medium sized-copepods, Pseudocalanus spp. was most abundant; highest abundances (100–330 ind. m−3) were found in Arctic Water, but also in Polar Front Water Pseudocalanus spp. was important (Table II). Again, during the first sampling, abundances of the medium size group (M) measured by LOPC were one order of magnitude higher than abundances measured by multinet (Table II; Fig. 3, second row). Later on, during the second and third sampling in mid-May, abundances of medium-sized zooplankton had increased to around 10 000–25 000 ind. m−3 in Arctic Water and Polar Front Water. From these areas, unfortunately, no net samples from mid-May are available to directly compare with the LOPC.
The larger copepods C. glacialis and C. finmarchicus occurred as nauplii, older copepodites (CIV–V and some few CIII) and adults in the study area (Table II). Both Calanus species were observed in Arctic Water on the shelf (Table II) where they were reproducing (own observation), and where nauplii were most abundant at the end of our study (Table II). Calanus finmarchicus was also found at depths (>150 m) in Atlantic Water and in Polar Front Water (Table II, Fig. 4) but was in diapause there (own observation). The LOPC measured abundances of mostly between 100 and 250 ind. m−3 during 1 May (Fig. 3, third row left), i.e. ca. three to five times higher than abundances obtained from multinet (Table II). Large zooplankton occurred mostly in small-scale patches in Arctic Water and in Polar Front Water (Fig. 3). High abundances of about 4000 ind. m−3 were observed in a larger patch in Polar Front Water during 14 May (Fig. 3, third row right). The patch measured ca. 1.5 km along the transect and stretched from 10 to 40 m depth.
The krill species Thysanoessa inermis and Thysanoessa longicaudata were caught by multinet in Polar Front Water and in Arctic Water; however, samples yielded low abundances and only krill nauplii were more abundant (Table II). Neither nauplii nor older krill were caught during the first sampling at the end of April. The LOPC measured small-scale patches of zooplankton in the size range of krill (XL size group) in Polar Front Water and in Arctic Water during all three sampling periods (Fig. 3, fourth row). During 14 May, in Polar Front Water, a patch was observed that sharply ended at the border between Polar Front Water and Arctic Water (Fig. 3, fourth row right). Abundances of XL zooplankton were mostly around 500 ind. m−3 but maximum values of 1400 were reached locally (Fig. 3, fourth row).
The biovolume spectra from Arctic Water had a higher intercept than the spectra from Polar Front Water and these again had a higher intercept than those from Atlantic Water (Fig. 5, upper panel), reflecting the observed high abundances in Arctic Water and very low abundances in Atlantic Water (Fig. 3). In all three water masses, the intercepts of the biovolume spectra increased with time (Fig. 5), indicating an increase in production in the frontal system over time.
In Arctic Water, the biovolume spectrum became flatter in the course of the two sampling weeks (Fig. 5, upper panel left). The slope that was fitted to the biovolume spectra of the whole zooplankton community thus decreased (Fig. 5, lower panel left; Table IV), and TLs in the community increased from 3.6 at the end of April to 5.5 in the mid of May (Table V). A flattening of the spectra and an increase in TLs were also observed in Polar Front Water (Fig. 5, centre), though the increase in TLs was not as pronounced as in Arctic Water (Table V). In Atlantic Water, the opposite pattern was observed: in mid-May the slope of the biovolume spectrum was steeper than at the end of April (Fig. 5, bottom right; Table IV). The zooplankton community in Atlantic Water was thus characterized by more TLs in the beginning of the study than at the end (Table V).
In all water masses, the TL of medium-sized zooplankton was around 1 at the end of April (Table V); these very low TLs may indicate that medium-sized zooplankton in the upper layer was feeding purely herbivorously during that time. In Arctic Water, TL within the medium-sized zooplankton group increased to 4.7 on 10 May, while the increase in Polar Front Water was more moderate to 1.3 on 10 May and 2.2 in the mid of May (Table V). In Atlantic Water, medium-sized zooplankton had a TL of ~1 during the whole sampling period, though it was a little higher during the second sampling (Table V).
Within the large zooplankton group, TL increased in Arctic Water during the course of sampling, reaching a TL of 4.4 in the mid of May (Table V). In contrast, in Polar Front Water, TL of large zooplankton was high (5.5) at the end of April and gradually decreased towards mid-May (Table V). A similar pattern was observed in Atlantic Water, where the large size group had a high TL of 5.6 during the second sampling and a lower TL of 2.9 during the third sampling (Table V).
The surface manifestation of the Polar Front and its associated physical and biological features southeast of Hopen Island in the Barents Sea were well resolved by our sampling strategy. The data showed a small-scale variability of nearly all measured parameters along transects. This stresses the importance of adequate high-resolution sampling in order to capture a true picture of the frontal system.
High fluorescence values proceeded from Arctic Water, which had recently become ice-free, towards Polar Front Water following stratification of the water column during the 2 weeks of our study. During spring, the phytoplankton bloom in the Barents Sea is dominated by chlorophyll-rich species (Rey et al., 1987; von Quillfeldt, 2000; Hodal and Kristiansen, 2008) and fluorescence is then a good indicator of phytoplankton (Signorini and McClain, 2009). In polar ecosystems, the bloom often progresses from the ice edge and early-stabilized Arctic Water towards well-mixed open water (Rey and Loeng, 1985; Waniek et al., 2005). We can therefore be reasonably sure that we encountered a phytoplankton bloom that proceeded from Arctic Water towards Polar Front Water in the course of our study. In the Barents Sea, blooms in Atlantic Water might appear with increased solar radiation in unstratified water columns (Eilertsen, 1993) as late as June (Wassmann et al., 1999). In our case, Atlantic Water remained in a winter situation throughout the sampling period in large parts, but not in the southeastern-most corner of the study area. Accordingly, the zooplankton community in Arctic Water corresponded to a spring situation with high abundances of larval stages, whereas the Atlantic community corresponded mostly to a winter situation with generally lower mesozooplankton abundances and Calanus sp. being located in diapause at depth.
The abundance of small (0.25–0.6 mm ESD) and medium (0.6–1.0 mm ESD) particles measured by LOPC was about one magnitude higher than that in the same size range observed in multinet samples (180 µm mesh width). This is in line with recent analyses estimating that <10% of zooplankton <0.8 mm length is retained by a 200 µm net (Gallienne and Robins, 2001), and it is also consistent with earlier studies showing that a significant proportion of copepods is lost if sampled with a mesh width larger than 75% of their body width (Nichols and Thompson, 1991). Locally, the LOPC also measured higher abundances of large zooplankton (1–2 mm ESD) than those observed from multinet samples. This contradicts findings showing that abundances of older stages of Calanus sp. are comparable, measured either by multinet or by LOPC (Gaardsted et al., in review). On the basis of the patchy distribution of large zooplankton (Fig. 3), we suppose that we missed the patches of high abundances of Calanus spp. with our net sampling in Arctic Water. We cannot rule out, however, that the large size group also contained some juvenile krill and this may indeed have been the case in Polar Front Water on 14 May (Fig. 3, third row right). In general, the multinet is not geared towards catching larger, more motile species (Sameoto et al., 2000), it was therefore not unexpected that the LOPC, with a higher towing speed, indicated higher abundances of krill (extra large size group, 2–14 mm ESD) compared with abundances of Thysanoessa spp. obtained from multinet samples. To conclude, in all probability, the LOPC was more reliable in capturing true abundances of all size groups in the upper 75 m than the multinet.
The abundances of Balanus sp. nauplii observed in Arctic shelf waters and in Polar Front Water were up to 35 000 ind. m−3 in net samples. These high abundances are comparable to abundances of barnacle nauplii observed in the Western Atlantic off Rhode Island during peak spawning (Lang and Ackenhusen-Johns, 1981), and to abundances observed in May on the shelf north of Svalbard (Blachowiak-Samolyk et al., 2007), and thus indicate that peak spawning of Balanus sp. was taking place during our survey period. The input of biomass from the benthic community into the upper layer was clearly visible in the biovolume spectra. In Arctic Water and in Polar Front Water, the pronounced increases of the biovolume spectra observed in the size range of small zooplankton from the end of April to 10–12 May and from 10–12 May to 14 May exceeded any potential increases associated with body and population increases internally within the community. This indicates that external biomass had to be supplied into the upper layer and highlights the importance of benthic–pelagic coupling during spring on the Arctic shelf. In Atlantic Water, the increase in the biovolume spectrum in the size range of small zooplankton was most pronounced from 10–12 May to 14–15 May, indicating that barnacle nauplii reached the upper layer later here. It is not clear from our data if barnacle nauplii were released in Atlantic Water or if they were advected from the Arctic shelf into Atlantic Water.
The intercepts of the biovolume spectra in Arctic and Polar Front Water were high, compared with the few biovolume spectra that exist in the literature. Intercepts were comparable to the spectra obtained from a fjord in northern Norway in May/June (Edvardsen et al., 2002) but higher than observed in a highly productive region offshelf the Norwegian coast in May/June (Zhou et al., 2009). The biovolume spectra from Arctic and Polar Front Water thus indicate high productivity on the Arctic Shelf and in the frontal region during our study. Phytoplankton blooms in the Arctic are restricted to the short ice-free season (Sakshaug, 2004) and productivity in the marginal ice-zone is then temporally high on shelves (Sakshaug, 2004) and in the stabilized marginal ice-zone in general (Carmack and Wassmann, 2006). The biovolume spectra collected here thus agree well with expectations and demonstrate a straightforward way to use the towed LOPC in combination with a CTD for quickly assessing spatial and temporal distributions of physical and biological variables prevailing in the study area during sampling.
Intercepts of the biovolume spectra from the end of April in Atlantic Water resembled the low intercepts of biovolume spectra observed in December in the Barents Sea south of Bjørnøya (Bear Island) (Zhou et al., 2009) and thus further confirm that the zooplankton community in Atlantic Water was still in a winter situation. By mid-May, intercepts in Atlantic Water had slightly increased, indicating that a pre-bloom situation may have developed in parts of the Atlantic Water. This corresponds well to the observed slightly higher fluorescence values in the southeastern-most part of the study area on 15 May.
The number of TLs within the zooplankton community increased in Arctic Water from 3.6 to 5.5, following the apparent development of the phytoplankton bloom. Top predators within the zooplankton community in Arctic Water were chaetognaths, amphipods (mostly Parathemisto libellula) and ctenophores. A TL of 3.6 agrees well with TLs determined by stable isotope analyses for these carnivores in spring (Søreide et al., 2006), and indicates the dominance of a classic food chain where phytoplankton is ingested by Pseudocalanus spp. and Calanus spp., and these in turn are ingested by carnivorous zooplankton. For older stages of C. glacialis and C. finmarchicus, which clearly dominated the large zooplankton size group in Arctic Water, TL increased from 2.2 to 4.4 at the end of the sampling. This indicates that Calanus spp. fed nearly exclusively as a herbivore during the bloom, while phytoplankton biomass likely was channelled through the microbial loop, e.g. via flagellates and ciliates, late in the bloom. This is consistent with an observed increase in the number of trophic links between microzooplankton and Calanus as the phytoplankton bloom progresses (Levinsen et al., 2000) and corresponds also well with an observed increase in TL of C. glacialis with the progression of the bloom (Tamelander et al., 2008). The higher number of TLs within the zooplankton community at the end of sampling in Arctic Water may then reflect that carnivorous zooplankton ingested copepods that had derived their energy from omnivorous diets.
In Polar Front Water, TLs computed by biovolume spectrum theory gave contradictory results: At the end of April, the maximum number of TLs within the zooplankton community was 3.9, while a TL of 5.5 was computed for the large zooplankton group (Calanus or possibly juvenile krill) alone. This can be understood as follows: computing TLs according to biovolume spectrum theory is based on the assumption that the biovolume spectrum can be linearized on a logarithmic scale (Zhou, 2006). We only computed TLs for significant slopes so as not to violate this assumption; however, looking at the biovolume spectra from Polar Front Water (Fig. 5), cycling deviations from the fitted line are obvious. Hence, a significant fit of the regression line was obtained, in spite of a non-linear biovolume spectrum. For Polar Front Water, and to a lesser degree also for Atlantic Water, the TLs computed for the whole zooplankton community thus have to be viewed as mean trophic values within the community, rather than as the maximum number of TLs within the zooplankton community. No deviations from a linear logarithmic biovolume spectrum were observed in the medium and large zooplankton size groups, and thus for these groups, a maximum TL could be computed.
Before the bloom reached Polar Front Water, older stages of Calanus spp. had a high TL reflecting an omnivorous diet. Later on, the TL decreased to 3.1, likely reflecting a more herbivorous diet in Calanus, as was observed in Arctic Water in the beginning of the bloom. Possibly also an increased abundance of juvenile, mostly herbivorous Thysanoessa spp. (Søreide et al., 2006) contributed to the decrease in TL in the large zooplankton group. Within the whole zooplankton community, mean TL increased towards the mid of May. This might be attributed to the large patch of XL zooplankton, which was observed in the study area then. We cannot say with certainty if this patch consisted of euphausids or not, but regard it as likely based on (i) net samples showing krill larvae and (ii) the observed distribution in the Polar Front, which is typical for Thysanoessa spp. (Dalpadado and Skjoldal, 1996). Older stages of T. inermis and T. longicaudata are omnivorous to a variable degree (Falk-Petersen et al., 2000; Søreide et al., 2006; Dalpadado et al., 2008) and thus have the potential to increase mean TL within the zooplankton community.
In all water masses, surprisingly low TLs between 1.0 and 1.2 were computed at the end of April for the medium-sized zooplankton group, which mostly consisted of Pseudocalanus spp. These low TLs indicate that TLs were underestimated in the medium zooplankton group. Assimilation efficiency was assumed to be 70%, but might have been lower in Pseudocalanus spp., as was discussed in the Method section. Assuming an assimilation efficiency of 50% would increase TL by 0.2. Independent of assimilation efficiency, the very low TLs computed for the medium-sized zooplankton group indicate that Pseudocalanus spp. fed mostly herbivorously.
The prevailing winter situation in Atlantic Water becomes clear when looking at TLs. The number of TLs in the zooplankton community was high at the beginning of the study, reflecting a community in which biomass is being recycled several times. The few large zooplankton found in the upper layer (C. finmarchicus) had a high TL of 5.6, similar to the TL in the late bloom/post bloom situation in Arctic Water. Later on, mean TL in the whole zooplankton community decreased, likely due to an increase in herbivorous Balanus sp. nauplii (Turner et al., 2001) in the upper layer. Also the TL in the large zooplankton group decreased with time, and this probably shows the ascent of the first C. finmarchicus from overwintering depths, which then started feeding on phytoplankton.
Trophic structure within the pelagic community was thus highly variable across the Polar Front. The variability was closely related to the stage of the phytoplankton bloom with generally higher TLs outside bloom situations, likely due to (i) increased abundances of herbivorous meroplanktonic larvae during the bloom and (ii) increased omnivory of Calanus spp. during pre- and post-bloom. Estimating TLs based on biovolume spectrum theory is a new approach (Zhou, 2006) that has not been extensively tested using field data. As has been shown in this study, it yields reasonable values for TLs within the zooplankton community as long as the underlying assumptions are met. Applying biovolume spectrum theory to data collected by semi-automatic sampling proved thus to be a powerful method to analyse the effect of mesoscale hydrography on trophic dynamics within the pelagic community, not at least in a dynamic frontal area. In the future, it would be desirable to combine the estimation of TLs using biovolume spectrum theory with methods that determine TLs based on direct sampling of a variety of planktonic organisms, e.g. stable isotope analyses. This would allow direct comparison of TL estimates from both methods for different groups of zooplankon, namely herbivores, omnivores and carnivores, and thus to obtain a greater certainty that estimated TLs reliably describe the trophic interactions within the zooplankton community.
It has to be noted that detritus is incorporated into the estimation of TLs when using biovolume spectrum theory. This leads to higher TLs compared with approaches where detritus is assigned TL one (Pauly et al., 1998), or where the background level of the food web (TL one) is determined based on particulate organic matter that includes detritus and phytoplankton alike, as is common in stable isotope analyses of marine pelagic food webs (Hobson et al., 1995; Søreide et al., 2006). It is complex to define the first TL in food webs (Post, 2002), especially for systems with many omnivorous species (Williams and Martinez, 2004) and a high variability on spatial scales (Tamelander et al., 2009), such as the pelagic system. It is debatable whether it is feasible to include detritus into the food web or not. On the one hand, detritus is important for the flow of energy within food webs (Moore et al., 2004), on the other hand food chains may become impractically long leading to very high TLs (Williams and Martinez, 2004). In the euphotic zone, detritus forms a vital source of energy for many organisms and is thus recycled frequently (Wexels Riser et al., 2007). When addressing the trophic linkages within the zooplankton community, we therefore think that incorporating detritus into the food web gives a more complete view of the trophic situation than levelling detritus to TL one.
This work was supported by the Research Council of Norway as part of the NESSAR project [# 176057/S30]. Funding to pay the Open Access publication charges for this article was provided by the Centre for Gender Research at the University of Tromsø.
We thank Malin Daase for quick and thorough processing of the zooplankton samples, John Terje Eilertsen and Richard Buvang for excellent technical assistance during the cruise, Ken Drinkwater for ample sampling time and a friendly atmosphere during the cruise and Tobias Tamelander for inspiring discussions.