In-Situ Feeding in the Northern Krill, Meganyctiphanes norvegica: A DNA Analysis of Gut Contents

The northern krill, Meganyctiphanes norvegica, is an important member of North Atlantic shelf ecosystems, serving both as prey for whales, fish and seabirds, and as a predator on phytoplankton and other zooplankton. However, understanding in-situ krill feeding is technically challenging; incubations may not be representative of krill feeding in-situ, while biomarkers and microscopic examination of gut contents suffer from limited prey type resolution and have a limited range of detectible prey. Analyzing DNA in gut contents may offer insight into M norvegica feeding in-situ unobtainable using other methodologies. The major technical difficulty in using DNA as a marker of gut contents is the overwhelming quantity of predator DNA; two approaches are taken here to exclude predator DNA from analysis, firstly the use of species specific primers, and secondly the use of a krill-specific peptide nucleic acid probe. Species specific primers were used to sequence and quantify known phytoplankton prey (Thalassiosira weissflogii and Rhodomonas sp.) from the guts of captive krill, showing that even low abundance prey items can be successfully detected using DNA in gut contents. A krill-specific peptide nucleic acid (PNA) probe was used to sequence all non-krill I 8S DNA present in the guts of krill collected in-situ in the Gulf of Maine. PNA is a synthetic, uncharged, DNA analog which binds strongly and specifically to complimentary DNA, and thus inhibits PCR amplification of the target sequence. Including a krill-specific PNA probe in a PCR using universal I 8S primers allowed for amplification of all eukaryotes present in the krill fore-gut. Gut contents amplicons were sequenced from a clone library and compared with known sequences to determine their identity. The most common prey item, found in M norvegica guts at every station, was an uncultured and poorly known protist, which previous studies suggest represents a novel kingdom of eukaryotes, and may suggest krill feeding on the sediment interface. M norvegica in the Gulf of Maine also consumed the copepod Calanus finmarchicus at 7 of the 8 stations samples, in agreement with results found using other methods. Additionally, Centropages sp., 4 other protists, another copepod, 5 phytoplankton and the salp Thalia democratica were found as gut contents. In addition to providing interesting information about M norvegica feeding in-situ in the Gulf of Maine, these results demonstrate the utility of the PNA-PCR clone library approach to gut contents analysis, elucidating prey, such as protists and T. democratica, which would have been difficult or impossible to detect with other methodologies.


Introduction:
Meganyctiphanes norvegica, the northern krill, is an important component of food webs in the North Atlantic, serving both as prey for whales, fish and seabirds, and as a predator on phytoplankton and smaller zooplankton (Kaartvedt et al 2002, Thomasson et al 2003, Cotte and Simard 2005, Lass et al 2001.
Understanding the feeding behavior of M norvegica is important to understanding the functioning of these ecosystems, and their potential to adapt to changing environmental conditions. However, measuring feeding of small organisms consuming microscopic prey in a marine environment is technically challenging. While many approaches have been used to measure krill feeding, all have advantages and drawbacks . Incubations may not be representative of in-situ feeding, biomarkers have low temporal and prey type resolution, and gut contents microscopy is limited to organisms with hard parts (McClatchie I 985, Torgersson 2001, Schmidt et al 2003, Dalpadado et al 2008, Kaartvedt et al 1998, Bamstedt and Karlson 1998. Recent studies with Antarctic krill, Euphausia superba, have used DNA as a marker of gut contents . In this work, the use of DNA as a marker of gut contents in M norvegica is investigated. Species specific primers were employed in sequencing and quantitative PCR reactions on krill fed a known phytoplankton diet.
For this study 18S ribosomal DNA is used. 18S rDNA is the DNA region encoding for the 18S rRNA which makes up a structural component of the small subunit of eukaryotic ribosomes . I 8S rDNA is around I ,900 nucleotides in length, although this varies somewhat between species . J 8S is also a multi-copy gene, that is, it occurs many times in the genome of a single cell unlike most genes which occur once per cell. This increases the probability of detecting an organism, which is particularly important when working with partially digested gut contents.
18S rDNA is often used as a "bar-coding" region of DNA, a region used to identify different species. Both previous studies using DNA to investigate krill feeding have used 18S . Due to its role in the structure of the ribosome, l 8S contains regions of sequence which are highly conserved, as well as quite variable regions. This is ideal for use as a "bar-coding" gene because it allows for the design of universal primers amplifying all eukaryotes in the highly conserved regions, while providing good information content and species resolution in the highly variable regions, as well as allowing for the design of species specific primers, as done in this study. The popularity of l 8S as a bar-coding marker also means that the greatest amount of sequence data is available for it. GenBank (National Center for Biotechnology Information, National Institute of Health), contains 273,609 sequences for" I 8S" (http://www.ncbi .nlm .nih.gov, January 3, 20 I 0). By contrast none of the other bar-coding type genes contained even 70% of that number, with "COI" having 187,353, "Cox" having 123,027, and "Cob" with 6,497 (http://www.ncbi .nlm.nih.gov, January 3, 2010). For gut contents studies an ideal marker will have a large database of sequences of the region from a variety of known organism, thus providing a way of identifying unknown sequences through homology with sequences of known origin.

Methods:
Krill for this study were collected on NOAA ship Delaware II during the Ecosystem Monitoring (ECOMON) cruise in January/February 2009. Live euphausiids were collected using a bongo net with 333 µm mesh and solid plastic I liter cod-ends in the Gulf of Maine (Table l ). Tows were oblique from the surface to l 00 meters, with a forward speed of 1.5 to 2 knots, and a wire speed of 50 m min-1 out, and 10 to 15 m min-1 in, and were conducted mainly at night. When the net reached the surface cod ends were immediately submerged in buckets of seawater, and their contents gently inverted to minimize sheer forces on the krill. Individual krill were transferred from the bucket into holding tanks on deck using large white plastic ladles.
Methods were designed to be as gentle as possible and minimize handling time of the krill, to improve krill survival (King et al 2003).
While at sea krill were maintained in two flowing seawater tanks on deck, with each tank containing approximately half the population, at maximum approximately 0.5 krill L-1 • These tanks were 120 liter insulated coolers, modified to include external adjustable height stand pipes, internal large surface area water outflows, and adjustable rate seawater inflow. While closed, these tanks were tightly shut using external ratchet straps, and water level was maintained completely full, with minimal air space to reduce sloshing due to ship movement. Tanks were secured on deck to maintain in-situ temperatures. Krill were fed Artemia salina larvae (2 to 7 days old) which were raised in a heated, aerated, 4 liter aquarium with a salinity around 50 psu.
Krill were fed approximately 500 ml of dense Artemia culture daily, additionally the seawater flowing through the tanks was coarsely filtered, so small prey items similar to those found in-situ were available to the krill. Krill were fed and checked daily; any dead individuals or molts were removed and recorded. Krill were maintained onboard up to one week.
In the Jab, the krill were maintained in a round tank, 70 cm high and 150 cm diameter, with a water volume of 1,240 liters. Krill were transferred from the ship's return to port in Woods Hole, MA to the lab in Narragansett, RI in 5 gallon buckets of seawater, which were then allowed to temperature acclimatize by floating in the laboratory tank, before being gently transferred to the tank. At maximum population size immediately upon return to shore, 123 krill were maintained in this tank, 0.1 krill L-1 • The tank had 5 µm cartridge filtered flowing filtered seawater from Narragansett Bay, initially 6-7° C warming to 12° C by April 22 when the experiment was concluded with a flow rate of approximately 3 L min-' . The central stand pipe outflow was modified with a large cone of plastic mesh, effectively increasing the inflow surface to approximately 50 linear cm. The tank was of dark blue slightly translucent polyethylene plastic and was covered by a teepee of thin black plastic in order to create a darkened environment for the krill which is optimal for maintaining euphausiids in captivity (King et al 2003). Krill were observed daily and any molts or dead individuals were removed and noted, as was excess dirt or fecal matter. Krill were fed two to ten day old larval Artemia salina daily at a final krill tank concentration of approximately 60 Artemia L-1 • Feeding experiments: Feeding experiments were carried out 10 days after the krill were transferred to the laboratory aquarium, ensuring that remaining krill were healthy and acclimated (King et al 2003). Three different feeding treatments were conducted: Rhodomonas sp., Thalassiosira weissflogii, and combined Rhodomonas sp. and T. weissflogii.
Rhodomonas sp. (CCMP 768) and T. weissflogii (CCMP I 048) were cultured in sterile f/2 media at 20° C and constant light (I 5 µmole photons m-2 s-1 ) to high concentrations in 2 L polycarbonate containers. Feeding experiments were conducted by filling 5 L wide mouth plastic jars approximately two thirds full of filtered seawater and a high concentration of prey (visibly colored the water), to promote high feeding rates . These jars were floated within the main tank to temperature acclimate. Krill were captured as gently as possible from the main tank using small aquarium nets and beakers without bringing them into the air. All feeding experiments were 5 hours, from approximately I 0 am to 3 pm in the dark. While this is a time when krill would normally be at depth and thus unlikely to be actively feeding on phytoplankton, work by Bamstedt and Karlson (1998) suggests that krill do not have endogenous rhythms in feeding activity. After 5 hours krill were removed from the feeding containers as rapidly as possible and with minimal stress to the animals and placed in 80% ethanol. Ethanol was changed once after 24 hours to maintain preservation.
Laboratory analysis: Prey DNA was sequenced from the same cultures used in the feeding experiments. 10 ml of culture was vacuum filtered onto 13 mm diameter 0.5 µm membrane filters (nucleopore). DNA was extracted using the DNeasy plant kit (Qiagen) as per manufacturer's instructions. DNA extracts were diluted 1 Ox to 1 OOx and amplified using universal eukaryote 18S primers Euk A and Euk B (Table 2) . Each 50 µl reaction contained 1 x GoTaq Green Master Mix (Promega), 0.5µM forward primer EukA, 0.5 µM reverse primer EukB, and 0.5 -1 ng µr' template DNA. Thermal cycling was run on a Mastercycler (Eppendorf) as follows: 95°C for 30s, 35 cycles of 94°C 30s, 60°C 1 min, 72°C 2 min, final extension 72°C 10 min, 4°C hold up to 12 hours.
PCR products were visualized on agarose gels (200 ml of 1 % (weight/volume) agarose in Ix TAE buffer, 125 volts for 90 minutes). PCR products were purified using the Qiaquick clean-up kit as per manufacturer's instructions (Qiagen) and DNA concentration was measured spectrophotometrically (Nanodrop). DNA samples were prepared for sequencing by combining 180 ng purified 18S PCR product, 10 picomoles primer, and ultra pure H 2 0 to a total volume of 24 µI. Each sample was sequenced with both forward (Euk A) and reverse (Euk B) primers to cover the entire 1,800 bp 18S amplicons. Sequencing was conducted by the Rhode Island Genomics and Sequencing Center using the Applied Biosystems ABI 3 I 30xl genetic analyzer and POP7 polymer, a 50 cm 16-capillary array and the KB Basecaller software (URI RIG SC).
Four krill were analyzed from each of the single prey feeding experiments, and six from the mixed prey experiment. Krill were patted dry, wet weighed using an electronic balance to the nearest 0.01 gram, and measured under the microscope for total length, using Mauchline's standard one length measure as the lateral or dorsal distance from the tip of the rostrum to the end of the uropods, excluding terminal setae (Mauchline I 980 in Everson 2000 A quantitative PCR (qPCR) experiment was run to investigate the sensitivity of the species specific primer gut contents sequencing approach using the Thalassiosira weissjlogii specific primers. Cycling was run on a Stratagene Mx3005P.
Each 25 µI reaction contained: Ix SYBR green qPCR master mix, 0.1 µM T.w. If forward primer, 0.1 µM T.w. Ir reverse primer, 0.03 µM Rox reference dye, and template DNA. Thermocycling was run as above, and fluorescence was detected at the end of each annealing step. A standard curve of l 8S T. weissjloggi amp I icons (as used in sequencing) was run at 10 7 , 10 5 , 10 3 , 10 1 , and 10° copies µr'; l8S copy number µr' was determined as follows: copies µI -t = (6.02 * I 0 23 ) * (C) * ( 1800*650) -1 , where C is the concentration of PCR product in ng µI -1 , 1800 is the bp length of the PCR amp I icon, and 650 is the average base pair weight. Gut extracts from every M norvegica individual were amplified, all samples and standards were run in duplicate.

Results:
The system developed for collecting and maintaining live euphausiids was Sequences derived from DNA extracts of krill guts using species specific primers were of good quality and were identical to the sequences derived from phytoplankton culture. All 4 of the krill who were fed only Thalassiosira weissjlogii resulted in sequences identical to that of pure culture. Additionally 3 of the 6 krill fed a mixed diet of T. weissjlogii and Rhodomonas sp. resulted in sequences identical to pure T. weissjlogii culture when amplified with the species specific primers. Using the Rhodomonas sp. specific primers gut DNA extracts of 3 of the 4 krill fed only Rhodomonas sp. resulted in sequences identical to that of pure culture and 3 of the 6 krill fed a mixed diet resulted in sequences identical to that derived from pure culture, including one individual which also yielded a sequence for T. weissflogii. The remaining 7 reactions had low signal level and did not produce any meaningful sequence data.
qPCR results indicated that prey DNA l 8S copy numbers in the krill gut extracts were very low (Fig. 2). The krill gut samples which resulted in successful sequences identical to that of pure phytoplankton culture contained approximately 0.1 to 2.5 copies µr 1 , about 1.5 to 30 copies per reaction, equivalent to 60 -1,200 copies guf 1. Samples which had not resulted in sequences showed little or no amplification.

Discussion:
The krill maintenance in captivity was fairly successful, with krill consuming at least some food, and swimming actively around the tank. The maximum.life span of the Meganyctiphanes norvegica maintained in this study, about 9 weeks, was equivelent to the longest time they have been maintained in captivity in published literature, (McDonald 1927) Other species of krill, notably Euphausia superba, have been maintained for significantly longer than this in captivity, however doing so tends to involve significant infrastructure and be fairly labor intensive (King et al 2003).
Earlier efforts for this study to maintain M norvegica in captivity suffered from high mortality rates, and were labor intensive. Factors that appear to be important to maintaining captive krill include stable, in-situ temperatures, relatively large volumes of water per krill, flowing water to maintain good water quality, and gentle capture of the krill. However, this system is probably less than ideal in some respects . Krill did suffer some mortality; microscopic necropsies did not provide a consistent answer on the cause of death. Additionally, krill exhibited behaviors in captivity different from those one would expect to find in-situ in a largely open ocean organism; krill were frequently observed near the bottom and sides of the tank, and particularly where the bottom and sides met, while they were relatively rarely seen in the central open water area of the tank. Krill feeding levels may also have been quite low, as discussed further in the qPCR.
This species specific primer technique was successful in sequencing two different prey items in the guts of Meganyctiphanes norvegica, including sequencing both items from the same krill individual. This suggests that species specific primers could be used in studies of in-situ krill feeding, focusing on specific prey items.
However, this technique does require significant a-priori assumptions as to what the krill are consuming, and so is probably most applicable to very directed questions, such as the role of krill grazing on HAB formation, krill predation on common prey items such as Calanusfinmarchicus, or predation of krill on eggs and larvae offish species.
qPCR results indicated that prey 18S copy number in Meganyctiphanes norvegica guts was very low, for which there are two main explanations. Either krill rapidly digest prey DNA leaving little to be detected by qPCR, or krill in captivity were eating at very low levels due to stress of being in a confined, laboratory environment. While it is certainly true that krill have particularly strong digestive enzymes, as seen by the rapid degradation of photosynthetic pigments in Euphausia superba guts, low feeding rates are also likely in this case . Previous studies of M norvegica feeding in captivity have shown low and variable feeding rates under laboratory conditions. For example at a copepod concentration of 1.6 copepods L-1 •  found a feeding rate of < I copepod krilr' hour"' in 4 L aquariums, whereas Torgerson (2001) found a feeding rate of 7 copepods krilr' hour" 1 in SOL aquariums. Phytoplankton may also not be a preferred prey of krill. M norvegica has been observed to consume significant amounts of photosynthetic prey, within an order of magnitude as much carbon as carnivorous feeding, (Kaartvedt et al 2002, McDonald 1927, Dalpadado et al. 2008. However, in incubation experiments M norvegica were not able to support metabolism feeding on phytoplankton alone, and mandible morphology suggests M norvegica is mainly carnivorous (McClatchie 1985, Bamstedt andKarlson 1998). Krill were accustomed to being provided A. salina as prey, and thus feeding on phytoplankton may not have been necessary or energy efficient for these krill.
Low prey I 8S copy numbers found in qPCR indicate the sensitivity of the technique. Samples containing as few as 2 copies of the target prey DNA resulted in successful sequences using species specific primers. This further suggests the utility of this technique in detecting relatively rare or rapidly digested prey items. Species specific primers may be applicable to detecting krill feeding on particular prey species in-situ, such as dominant copepods, or harmful algal bloom organisms.   stations. These sequences were classified by homology with known sequences. This technique offers unique insights into in-situ feeding, with minimal a-priori assumptions, and may be applicable to other small organisms.

Introduction:
Meganyctiphanes norvegica, the Northern krill, is a pelagic zooplankton important in food webs of the North Atlantic, as prey for whales, fish and seabirds, and as a predator on phytoplankton and smaller zooplankton (Kaartvedt et al 2002, Thomasson et al 2003, Cotte and Simard 2005, Lass et al 2001.
While copepods, and particularly Ca/anus finmarchicus, have been commonly found to be important in M norvegica feeding, a variety of other prey items have also been observed, including phytoplankton, tintinnids, terrestrial debris, and marine detritus (Bamstedt and Karlson 1998, Kaartvedt et al 2002, Dalpdado et al 2008, Lass et al 2001. Understanding the feeding behavior of these krill is important to understanding the functioning of this ecosystem, and yet is also challenging. Incubations may not be representative of in-situ feeding, and biomarkers, such as stable isotopes and fatty acids, have low temporal and prey type resolution , Torgersson 2001, Schmidt et al 2003,.
Current methods of gut contents analysis have a limited range of detectible prey, with gut contents microscopy limited to organisms with hard parts and gut pigments limited to photosynthetic prey, and suffer from challenges in identifying macerated prey items (Dalpadado et al 2008, Kaartvedt et al 1998, Bamstedt and Karlson 1998.
DNA as a marker of gut contents has the potential to provide information about in-situ krill feeding on a variety of prey. The major technological issue facing such studies is the overwhelming quantity of non-information containing predator DNA.
Previous studies of DNA in Antarctic krill gut contents have used either group specific diatom primers  or dissection of all krill tissue away from gut contents . An ideal method of gut contents DNA analysis would allow for the detection of all potential prey within the gut, while excluding or ignoring krill DNA. One such potential method is peptide nucleic acid mediated PCR clone library sequencing. This approach consists of sequencing clone I ibraries of "barcoding" genes from krill guts, incorporating a peptide nucleic acid probe to block amplification of krill DNA in the initial polymerase chain reaction (PCR). This approach has not previously been applied to gut contents studies, and this study will develop methods which may be applied to other organisms and ecosystems.
Peptide nucleic acid, PNA, is a synthetic DNA analog. Like DNA it consists of a relatively rigid helical backbone bearing a string of the nucleotide bases whose order determines binding kinetics . Unlike DNA, whose backbone is composed of sugar and phosphate groups bearing a negative charge, in PNA this backbone is composed of uncharged peptides .
This uncharged backbone has important effects on the binding kinetics of PNA-DNA interactions. Binding of complimentary PNA-DNA strands is stronger than binding of DNA-DNA strands . PNA-DNA has a higher binding specificity than DNA-DNA interactions . For DNA-DNA interactions the minimum llT 111 for a single base-pair mismatch is 4°, whereas for PNA-DNA interactions the minimum llT 111 for a single base pair mismatch is 8°, twice that observed for DNA-DNA interactions (Nielson and Egholm I 999). This means that a single base pair mismatch is much more destabilizing to PNA-DNA binding kinetics, and thus PNA-DNA interactions are more sequence specific than DNA-DNA interactions.
One application of the unique features of PNA is in PCR-clamping, the inclusion of a PNA probe in a PCR reaction (0rum 2000). PCR-clamping works in one of two ways, either through competition for the primer site, or through arresting polymerase elongation (0rum 2000). Both of these techniques depend on differences in binding kinetics between the PNA probe and the DNA primer. The PNA probe must have a higher Tm than the DNA primer, allowing the PNA to bind under conditions where the DNA primer will not (0rum 2000).
This study uses PNA to arrest polymerase elongation. In this approach a PNA probe is designed which binds the target DNA sequence between the two PCR primers, and blocks PCR amplification by stopping the progress of the polymerase and thus preventing the creation of full length complimentary strands (0rum 2000). PCR clamping employs a 4 step PCR. Initial high temperature denaturing is followed by a step at the PNA binding temperature, which is a temperature between the PNA Tm and the DNA primer Tm. Primer binding is then conducted at a temperature below the DNA primer Tm, and elongation is run at a temperature at least 10° below the Tm of the PNA to prevent PNA disassociation (0rum 2000, Nielson pers. com.) In previous studies, a PNA probe increased the sensitivity of a PCR-DHPLC  1 ). Krill were collected in oblique bongo tows to 5 meters from the bottom, or 200 meters in water depths greater than 205 meters, using was a 61 cm bongo frame fitted with 333 µm mesh nets without cod-ends, net ends folded and tied. The use of tied net ends as opposed to solid cod-ends should minimize the potential for net feeding, as the flow of water through the net keeps both krill and potential prey pushed against the back of the net and thus unable to filter feed. The net was towed with a forward velocity of between 1.5 and 2 knots, and wire speeds of 50 m min-1 out, 20 m min-1 in.
A bongo net is good for capturing euphausiids as it has no bridle in front of the net opening, thus minimizing the potential for net avoidance . Minimizing net avoidance is important in investigations of feeding as gear which allows for significant net avoidance may select for the weaker and less able to escape krill. Work with Euphausia superba has observed white-ish, poor condition krill individuals weakly swimming at the edges of krill schools, and these individuals may consume a very different diet from the healthy individuals making up the bulk of the krill population . In this work, all krill appeared to be still alive or recently dead when collected and no white-ish individuals were collected.
The net was immediately rinsed down, and contents washed from the net into a plankton sieve, from which krill were rapidly picked with forceps and placed directly into jars of 80% ethanol to preserve gut contents. The time from the net reaching the surface to the krill being placed in ethanol averaged less than 5 minutes. Most krill still appeared to be alive at the time they were placed in ethanol, actively attempting to escape the forceps and/or ethanol. Ethanol has been shown to more effectively preserve krill gut contents DNA than freezing, through immediately and permanently deactivating the digestive enzymes . 80% ethanol is used as opposed to 95% as 95% has been seen to result in krill to brittle for dissection . Ethanol was changed once after 12 to 24 hours to maintain effective preservation. Krill were preserved in 125 ml jars, with no more than one third of the jar composed of krill biomass; a maximum of approximately 15 to 20 krill per jar. The only selective criteria were an appearance of health, as indicated by clear and red coloration, and relatively large size. Different stages and sizes of krill may consume different diets, and this study focuses on adult krill, thus juvenile krill (as determined practically as small individuals, less than about 20 mm) were not collected.
Laboratory analysis of krill guts: Krill were dissected to remove their foreguts under a stereo microscope. Eight krill were analyzed from each station, with the exception of station 5, Bay of Fundy summer, where three replicate groups of eight krill were analyzed. Krill were selected at random from the adult krill collected, patted dry, and wet weighed using an electronic balance to the nearest 0.01 gram. Krill were measured under the microscope for total length, using Mauchline's standard one length measure as the lateral or dorsal distance from the tip of the rostrum to the end of the uropods, excluding terminal setae , and krill sex was determined morphologically .
Krill were dissected in disposable translucent weigh boats using flame sterilized sharp forceps. The front of the carapace was detached and the fore gut was removed with forceps, taking care not to break the outer membrane. Foregut fullness was classified as "empty", "< Y2 full", "> Y2 full " or "full", similar to the classification schemes used by Bamstedt and Karlson (1998), and Dalpadado et al (2008). Foreguts were placed in numbered individual microcentrifuge tubes and kept on ice for the duration of the dissections, up to 2 hours. Following dissections, krill gut DNA was immediately extracted.
DNA was extracted from whole krill fore-guts using the DNeasy Blood and Tissue kit as per manufacturer's instructions (Qiagen). This kit has been used successfully in previous work for DNA extraction from both zooplankton and phytoplankton, including dinoflagellates and diatoms. Krill guts were first mechanically disrupted using sterilized toothpicks, to ensure complete lysis of gut contents. Lysis incubation was conducted overnight, and DNA was eluted in 200µ1.
DNA extracts consistently had high DNA concentrations of 25 to 35 ng µi-1 (Nanodrop). 2µ1 aliquots from each of the eight krill at a station were combined for use in the analysis.
A PNA probe was designed from an alignment of Meganyctiphanes norvegica and 16 diverse potential prey items. The probe was designed to sit between well established universal eukaryote primers on the l 8S gene , to be complimentary to M norvegica, and to contain as many differences as possible with all potential prey items ( Table I). Design followed the recommendations of Applied Biosystems and associated sequence analyzer function for Tm, GC content and to minimize self complimentarity (ABI 2009). The primers developed by  were used in this study because they have been well established to amplify a wide range of Eukaryotic organisms, and create relatively short amp I icons, 250 bp, which helps to minimize the effects of digestion on sequence recovery.
The effectiveness of the PNA was tested using a quantitative PCR (qPCR) experiment. qPCR was run with samples of full length l 8S gene PCR products from Meganyctiphanes norvegica and Thalassiosira weissjlogii with and without PNA.
These PCR products were produced using universal eukaryote primers  in 50 µI reactions containing (final concentrations) Ix GoTaq Green Master Mix (Promega), .5µM forward primer EukA, .5µM reverse primer Euk B, and 0.5 -I ng µr' DNA template of DNA extracted using the DNeasy Blood and Tissue kit from M norvegica eyes, and DNeasy Plant kit from T. weissjlogii culture filters (Qiagen).
Thermal cycling was conducted on an Eppendorf Mastercycler as follows, initial denaturation at 95 ° C for 30s, followed by 35 cycles of 94°C for 30s, 60°C for I min, 72°C for 2 min, then a final extention of 72°C for I 0 min. I 8S copy number µr' of these I 8S PCR products was determined as follows: copies µI_, = (6.02 * I 0 23 ) * (C) * (I 800*650* 10 6 ) -1 , where C is the concentration of PCR product in ng µI -1 , 1800 is the bp length of the PCR amplicon, and 650 is the average base pair weight.
In a qPCR, a standard curve of M norvegica amp I icons was run at I 0 1 , I 0 4 , 10 6 , 10 8 , and I 0 10 gene copies µr', and these same M norvegica amplicons were run in the presence of PNA at 10 4 , 10 6 , and I 0 8 copies µr 1 . bp, were purified with ethanol precipitation, using a protocol modified from Zeugin and Hartley (1985). Samples were prepared for sequencing by combining 200 ng of PCR product, or 2 µl, whichever was greater, as determined spectrophotometrically (Nanodrop), 10 pmoles ofM13 forward primer, and H 2 0 to a final volume of24 µI.
Samples were sequenced at the Rhode Island Genomics and Sequencing Center (RIGSC) using the Applied Biosystems ABI 3 l 30xl genetic analyzer and POP7 polymer, a 50 cm. 16-capillary array and the KB Basecaller software.
Prey Field Sequencing: To identify the sequences obtained from gut contents, a variety of potential prey items for krill were sequenced to complement the data available on Genbank.  Table I).
These OTU A fir primers were also used in conjunction with universal EukA EukB primers to obtain a longer sequence for this organism from krill guts.
Samples were sequenced at the RIGSC as described above, with sequencing run starting at each primer, in order to cover the entire length of the l 8S gene, 1800 bp.
Sequence analysis: All sequences, gut contents and prey field, were visually checked for read quality using the Applied Biosystems Sequence Scanner (ABI). MegAlign and EditSeq (DNAstar) were used to align, assemble and crop sequences. Clone library sequences were cropped to remove plasmid DNA sequence, and the reverse compliment taken as necessary. Prey field zooplankton sequences were assembled with at least 50 overlapping identical base pairs. All replicates of the same prey species were identical. For sequences which were not sufficiently long to overlap, the region of interest lay entirely on the reverse strand, so this sequence alone was used in further analysis, and both are provided separately in GenBank (Table 4).
Sequences were aligned using ClustalW and default settings (gap penalty 15.0, gap length penalty 6.666, delay divergent seqs 30%, DNA transition weight .5). Gut contents clone library sequences were aligned over all stations, and classified into Operational Taxonomic Units (OTUs) based on a 3% divergence cut-off and assigned arbitrary letter names . OTU sequences were aligned with a variety of known sequences, both of species previously found to be taken as prey by krill, and the top named (not "uncultured eukaryote") BLAST hits for each OTU (Altschul et al 1990). Known prey item sequences were taken from Genbank and the prey field sequencing done as part of this study. OTUs were classified as the most closely related known species within 3%. In cases where there were no identified species within 3%, and it was unclear what the closest species was, OTUs were classified into the lowest taxonomic group where they could confidently be placed based on phylogenetic tree morphology.
Data on the mesozooplankton abundances in the environment was obtained from Ecosystem Monitoring Program of the NOAA Northeast Fisheries Science Center. This data is in the form of counts of each zooplankton species or group per volume filtered, as measured with flow meters mounted on net mouth openings . Counts come from the other side of the bongo net on the same tows where krill were collected. Mesozooplankton abundance data was obtained for every station where krill gut contents were analyzed for in-situ feeding .

Results:
The krill PNA probe reduced amplification of krill DNA to negligible levels, while having a minimal effect on non-target sequences (standard curve r 2 = 1.000) Mesozooplankton were common prey items for Meganyctiphanes norvegica in this region. Ca/anus finmarchicus was taken as prey by krill at every station except 7.
Centropages sp. was a gut contents item at station 3, and was the dominant sequence at this location, making up 87% of the prey sequences obtained. The copepod consumed at 2 was very similar to C. finmarchicus, more so than to any other available copepod sequence, but was a distinct cluster outside of 3% similarity. The salp Thalia democratica was taken as prey by krill in 2 of the 3 replicate cloning experiments at station 5, the Bay of Fundy during the summer (figure 5).
Phytoplankton prey were taken relatively infrequently. Of the five phytoplankton sequences found, only one could be identified to lower taxonomic groupings. This Prorocentrum sp. was found as a krill gut contents item at station 7, off the southern flank of Georges Bank in winter. Two of these remaining phytoplankton prey were found as gut content items at station 4. One of these same phytoplankton, and an additional two types were found as gut contents items at station 5 ( figure 6).
The most surprising finding of this study was the abundance of uncultured protists as krill gut contents. A specific uncultured protist was found as a gut contents item at every station investigated in this study. While the relative proportions of different clones within a station is not here considered as quantitative, it may be worth noting that this uncultured protist comprised 93 of the 255 prey sequences obtained in this study, 36%. In addition to this common uncultured protist, three other uncultured protists were found as gut contents, for a total of four distinct protists as krill gut contents items. Two of these additional protists were found as gut contents at station 7, and the third at station 5. Overall gut contents abundance data is presented in figure 9.
Further investigations into the sources of these protists sequenced protist DNA from sediment and surface waters of the Gulf of Maine. Sequences from Georges Basin sediment, surface water and nephloid water, as well as Wilkinson Basin sediment and surface water were identical to that of the abundant krill gut contents uncultured protist. The protist found at station 5 was 99% similar to the sequence obtained from Wilkinson Basin nephloid water. This sequencing was done using targeted primers, designed to amplify only the abundant krill gut contents protist (  3).

Discussion:
This study identified a variety of prey items taken in-situ by Meganyctiphanes norvegica in the Gulf of Maine. The two main prey items found were Ca/anus finmarchicus and an uncultured protist. Additional zooplankton prey included Centropages sp., another copepod and the salp Thalia democratica. Phytoplankton were encountered as prey relatively rarely, but three different phytoplankton were found including a Prorocentrum sp.(dinoflagellate) and a green alga. Heterotrophic alveolates (2) and other uncultured protists (3) were also encountered.
Ca/anus finmarchicus is a large, abundant and oil rich copepod in the Gulf of Maine region. This copepod has been previously found to be taken by Meganyctiphanes norvegica as prey in several studies , Bamstedt and Karlson 1998. C. finmarchicus is considered to be an important prey item for M norvegica, and in one study was observed to make up between 63% and I 00% of the copepod prey taken (Bamstedt and Karlson 1998 has been observed to consume salps in incubation experiments, and it has been suggested salps may be a preferred food . In those experiments, krill "attacked the salps several times within 5-10 minutes, and then grasped the salps with their thoracic endopodites and swam away .. . and ingested them efficiently" . Salps were only found at one of the stations studied, and only in two of three replicate groups of krill from that station, so it is unlikely T. democratica plays an important role in M norvegica nutrition, but is interesting none the less. T. democratica as a gut contents item also highlights some of Phylogenetic tree analysis suggests this sequence is from a protist of some description, however, it appears to be a poorly known organism. This protist is larger than 0.5 µm or particle associated as seen by its presence on 0.5 µm water filters, and of a related sequence in the >0.5µm size fraction, but not the 0.2 to 0.5 µm size fraction in Lopez-Garcia et al (200 I). Previous studies which have found closely related sequences have not been able to assign it to a known taxonomic grouping.
Takishita et al (2007) studying methane seep microeukaryotes found a sequence 98% identical to the protist found here over the overlapping region "that could not be assigned to major eukaryotic groups ... [and] possibly represent anoxic tolerant taxa".
That study also helps to put an upper size limit on this unknown protist, as they specifically excluded metazoans from DNA extractions (Takishita et al 2007).  found a sequence 89% similar to the krill gut contents protist, which was among the sequences they concluded were "not specifically affiliated with any molecularly described taxonomic group, and therefore indicate novel kingdom-level relatedness groups".  found a sequence 90% identical to the krill gut contents protist in the Guaymas Basin hydrothermal vent environment, and concluded that this and a few other sequences "are unrelated to those of any other eukaryotes, and they seem to represent early branches in the eukaryotic tree". A reanalysis of some of these sequences as well as additional sequences from river sediment (93% identical to krill gut contents protist) again concluded these sequences represent a "possibly novel high-level lineage" (Berney et al 2004). A sequence 90% identical to the gut contents protist was found in Weddell Sea deep water and "represents a new lineage emerging in the region of Archeozoa" (Lopez-Garcia et al.

2001)
This leads to the question of what is the role of this poorly known protist as a krill gut contents item. There are several possible explanations for this gut contents item, four of which will be discussed below. These are: I) this sequence is a parasitic or symbiotic organism resident in the guts of krill, 2) this sequence is a protist parasitic or endosymbiotic with a krill prey item, 3) this sequence is a protist filtered and consumed by krill in the water column, and 4) this sequence is a protist consumed by krill on the sediment or in the near bottom nephloid layer.
Krill endoparasites have been found in a few studies. A gregarine has been found in Meganyctiphanes norvegica guts (MacDonald 1927). There is evidence that M norvegica is an important intermediate host of the helminth Anisakis simplex (Everson 2000 p. 264). Endosymbiotic bacteria have also been shown to have a digestive function in M norvegica . The extremely divergent nature of this sequence, as compared to sequences of known organisms, has been suggested to imply it may come from a parasite, whose rRNA evolved rapidly (Lopez-Garcia et al 200 I). However, the gut contents protist sequence was not found in krill maintained in captivity. Krill captured in the same regions as those analyzed for in-situ feeding were maintained in captivity for I 0 days in filtered sea water fed known species of cultured phytoplankton or artemia larvae. PCR amplification of gut DNA extracts of these krill using primers amplifying specifically this uncultured protist gave no detectible amplification. This suggests that the uncultured protist is not a gut parasite or symbiont, as such parasites and symbionts would be expected to remian present in the guts of captive krill. Additionally, this sequence was found in mud and water filters, so for this sequence to come from a parasite it would have to be a parasite with ubiquitus free-living forms.

This gut contents protist could be a parasite on or in prey consumed by
Meganyctiphanes norvegica. However, this protist was found at every station sampled, and no other prey item was consistently detected, suggesting that the sequence does not belong to a copepod specific parasite. Canibalism is not detectible using the DNA methods in this study, thus the protist sequence could be a krill parasite. Ectoparasites noted in previous studies of M norvegica include a dinoflagellate, Staphylocystis racemosus, on the carapace, and a suctorian found on pleiopods (MacDonald I 927).
For this gut contents parasite to be consistent with observations here, it would also need to have a free living form found throughout the water column and in sediments.
Cannibalism has been observed in M norvegica, but is not thought to form a major component of the diet (Fisher and Goldie 1959, Lass et al 200 I). Thus this explanation also appears unlikely This uncultured protist may have been filtered by krill from the water column.
The sequence was obtained from filters of surface water in Georges Basin and Wilkinson Basin. This indicates that the sequence, and the organism from which it originates, were present in surface waters of the region where krill for this study were collected. It also provides a minimum estimate on the particle size associated with this sequence. The filters used in this study were 0.5 µm membrane filters, and the sequence was retained on these filters . Thus, either the organism containing this sequence is larger than 0.5 µm in some dimension, or it was particle-associated.
However, the species specific primer approach used to determine the presence or absence of this sequence in the surface waters is extremely sensitive and nonquantitative, thus the sequence may be in low abundance in these waters. A low abundance of this sequence in surface waters is also suggested by the absence of this sequence from clone library studies of surface waters, including such studies in this region . Additionally, this uncultured protist sequence was found in krill guts at every station, whereas phytoplankton was found in krill guts at only three of the eight stations. If krill were actively filtering sufficient volumes of water to consume this protist, one might expect that they would additionally be consuming phytoplankton.
Krill may have obtained this uncultured protist while feeding on or near the sediment interface. The sequence of this protist has been obtained from sediments of Georges Basin and Wilkinson Basin. To minimize the inhibitory effects of humics, these extractions contained very little starting material and were diluted ten to a thousand fold before PCR. This suggests that the krill gut content uncultured protist 48 may be relatively common at the sediment interface. All of the BLAST hits showing alignment with more than 70% of the krill gut uncultured protist sequence come from studies of sediment, with the exceptions of: a Euphausia superba gut contents item, a benthic bivalve gut contents item, and a study of Weddell Sea deep water (below 2,000 meters), all of which could be expected to contain some amount of sediment, consumed by the organisms, or re-suspended into the near bottom waters. Previous studies have observed E. superba feeding on the sediment interface in-situ, and Meganyctiphanes norvegica has been observed to re-suspend sediment and feed on this mud cloud in captivity (Clarke and Tyler 2008). Detritus has been observed as a gut contents item in-situ The data available for prey field mesozooplankton helps to explain some of the distributions in krill prey items observed. However, since only 4 of the 13 gut contents items found are mesozooplankton, and krill gut contents data obtained here is qualitative, but not quantitative, a thorough analysis of selectivity is not possible. The lower overall mesozooplankton abundance in winter may explain lower krill gut fullness in winter. Centropages sp. was consumed only at station 3, where these copepods (C. typicus and C. hamatus) made up a greater proportion of the available prey (27%) than at other stations, although in absolute abundance, nearly 2x greater concentrations occurred at station 4 where Centropages sp. was not found in krill gut contents. C. hamatus was only found at station 3 of all stations sampled. Because sequence data is not available for C. hamatus, these two congeners cannot be differentiated as krill gut contents items, however, they are morphologically similar, and thus seemingly unlikely to be differentiated by M norvegica. Prey field data also explains the apparently anomalous feature of krill not consuming Ca/anus finmarchicus at station 7. C. finmarchicus was in much lower abundance there that at other stations sampled, with only 1 copepod m-3 there, as compared to 77 to 370 copepods m-3 at other stations sampled where krill did consume C. finmarchicus. It is interesting to note that krill consumed T. democratica at station 120 where they were so not particularly abundant, but did not consume salps at other stations where they were more abundant, and more abundant relative to other available mesozooplankton.
The method employed in this study of PNA mediated PCR followed by cloning and sequencing has been successful in identifying a variety of prey items consumed by Meganyctiphanes norvegica in-situ. The advantages of this technique over other methods of gut contents analysis, including minimizing a-priori assumptions, and detecting morphologically indistinct prey are clear from these results. Gut contents items found in this study included soft bodied organisms not previously thought to be important to the diet of Meganyctiphanes norvegica, including the salp Thalia democratica, and uncultured protists. These items would have been difficult or impossible to detect using traditional methods of gut contents microscopy, as they contain no hard parts which would be identifiable after maceration by the krill. Additionally, as they were not expected to be prey items, they likely would not have been detected using more targeted gut contents studies, such as those employing group or species specific primers.
However, this method does currently have one major draw-back; it does not offer quantitative information on the relative amounts of different prey items consumed. The method could be made more quantitative in a variety of ways. qPCR could be applied to the sequences found to be present at many stations. Some of the newest sequencing technologies based on massively parallel independent sequencing reactions are considered to be quantitative, with abundances of sequences obtained representative of the abundance of the sequence in a starting sample. Incorporating a PNA probe into massively parallel sequencing techniques could potentially be used to obtain quantitative sequence information directly from krill guts for all prey items consumed. One caveat to increased quantification is the potential for bias from variations in l 8S copy number per cell, and differences in digestion rate for different prey types.
This method has provided interesting insights into the in-situ feeding of Meganyctiphanes norvegica in the Gulf of Maine, both confirming the importance of Ca/anus finmarchicus as a prey item, and suggesting the potential importance of protists to M norvegica feeding. The method could be applied to any small organism whose feeding is poorly known, and offers the potential for new insights into feeding behaviors, particularly feeding on small, soft bodied, or cryptic organisms.       The northern krill, Meganyctiphanes norvegica, is a pelagic zooplankton of great importance in North Atlantic ecosystems (Bamstedt and Karlson 1998. M norvegica plays a key role as prey for many diverse organisms in the region, including fish, seabirds and whales (Bamstedt andKarlson 1998, Cotte andSimard 2005). M norvegica may also be an important predator on smaller plankters in the region (Bamstedt andKarlson 1998, Kaartvedt et al 2002). In this study M norvegica feeding in-situ in the Gulf of Maine is investigated using DNA analysis of gut contents.
Meganyctiphanes norvegica is a euphaussid crustacean, and is fairly large for a zooplankton, 40 to 44 mm in length, with no sexual dimporphism , Thomasson et al 2003. M norvegica is thought to live between 2 and 3 years, reaching sexual maturity after year one, with spawning occurring in the summer, April through October, although this varies over their geographic range . Assessing the abundance and biomass of Meganyctiphanes norvegica is challenging due to net avoidance, and patchiness due to schooling and swarming behaviors, and diel vertical migration, however some limited estimates are available (Bamstedt and Karlson 1998). In Kosterfjorden, Sweden, krill biomass ranged from zero to 1,600 mg dry weight m-2 and was seasonally variable, with highest biomass observed in August/September (Bamstedt and Karlson 1998). In the Laurentian Channel krill density in patches ranged from approximately 25 grams m-3 to 500 grams m-3 , about 1,500 M norvegica m-3 and was correlated with tidal forcing (Cotte and Simard 2005). The density of krill swarms in the Bay of Fundy was estimated as ranging from 78 to 780 grams m-2 , with a total M norvegica biomass in the Bay of Fundy and Jordan Basin region of 15,000 tons .

Meganyctiphanes norvegica as prey
Meganyctiphanes norvegica is an important prey item for many fish, whales and seabirds of the Gulf of Maine region. Hake, including Silver and Red hake on Georges Bank, feed on M norvegica, and it is considered to be a preferred food for them . Hake have even been suggested to follow M norvegica prey in their daily vertical migrations . M norvegica is considered to be one of the most important food sources for Atlantic herring, on both sides of the basin, and areas of high krill concentration are thought to attract herring . Capelin and Atlantic mackerel also consume M norvegica (Cotte andSimard 2005, Everson 2000 p. 191 ). The physonect siphonophore, Nanomia cara, was shown to prey on M norvegica in the North Atlantic .
Whales feed extensively on krill. Seabirds also make use of Meganyctiphanes norvegica as prey. In the Western Atlantic the sooty shearwater, Puffinis griseus, takes as prey mainly krill (Everson 2000 p. 192). M norvegica is an important prey item for Leach's storm petrel, Oceanodroma leucorhoa and the razorbill, Alea torda, feeds on M norvegica and sculpins .
Humans can also be considered predators on Meganyctiphanes norvegica. A fishery for M norvegica was conducted in the Mediterranean in the 19 1 h century, with the catch used for fishing bait (Everson 2000, p. 228). In the North Atlantic an exploratory fishery in the Laurentian Channel in 1995 caught 6.3 tons of M norvegica, and the fishery in this region is estimated at a potential value of 3.75 million Canadian dollars (Nicol and Endo 1997). The proposed fishery in this area would catch krill for freezing and freeze-drying for home and public aquarium food, aquaculture feed, and as a flavourant for use in food for human consumption .

Meganyctiphanes norvegica as predators
Meganyctiphanes norvegica is omnivorous, consuming a variety of phytoplankton, zooplankton and other prey, but carnivory is thought to make up the majority of the diet, more so than for other species of Euphausiid (Torgerson 200 I).
M norvegica's mandibles with sharp pars incivia may be an adaptation to carnivorous feeding, (Bamstedt and Karlson 1998). The degree to which krill are carnivorous may be seasonally and spatially variable, depending on the available prey and the dietary needs of the krill. In spring and summer, when phytoplankton is more abundant, carnivorous feeding is relatively less important, with carnivory accounting for 23% of the diet in the coastal waters around Norway in the summer time (Kaartvedt et al 2002, Bamstedt andKarlson 1998). showed that M norvegica could not meet its metabolic needs feeding on phytoplankton alone, nor when feeding on small copepods such as Acartia spp. or Pseudocalanus spp., suggesting M. norvegica relied on high concentration 2.265 mg OW L-1 , about 190,000 C. finmarchicus-sized copepods m-3 , patches of large copepods, such as Calanus spp. or Centropages spp., or was unable to feed efficiently in his small incubation chambers . All of these studies point to the importance of copepods to the diet of M norvegica throughout its geographic range.

Calanus finmarchicus is probably the most important single prey species for
Meganyctiphanes norvegica. C. finmarchicus is a relatively large and oil rich copepod which is abundant and biomass dominant in many of the same areas of the North Atlantic where M norvegica is found, and has commonly been observed as a gut contents item , Bamstedt and Karlson 1998. C. finmarchicus, with a prosome length distribution centered around 1.00 mm, stage three copepodites (C3), made up 64% to 100% of the copepods consumed by M norvegica in the coastal waters of Norway (Bamstedt and Karlson 1998). M norvegica was calculated based on measured in-situ feeding rates to consume .3 to 6.4% in Norway and 1.3% to 2. 7% in the Gulf of Maine of the total C. finmarchicus biomass daily (Bamstedt andKarlson 1998, Thal 2004). Tintinnid lorica have also been seen as M norvegica gut contents (Dalpadado et al 2008).
Cannibalism is another form of carnivory which has been observed in M norvegica, though to a fairly limited extent. Compound eyes identified as belonging to euphausiids were found relatively infrequently, and more commonly in winter than of magnitude as much carbon as carnivory (Kaartvedt et al 2002). It was concluded that "M norvegica is very versatile in its ability and motivation to exploit algal food" (Kaartvedt et al 2002).
Diatoms have been observed as prey in the Barents Sea, Clyde and Kattegat Seas and Loch Fyne (Dalpadado et al 2008, Virtue et al 2000. Diatoms, especially Thallasiossira spp., and less commonly pennates, including Fragilariopsis spp., Pseudonitschia spp., and Navicula spp., were the most numerous photosynthetic prey in Meganyctiphanes norvegica guts in the Barents Sea (Dalpadado et al 2008). In Loch Fyne a variety of diatoms were found as prey items in gut contents (MacDonald 1927, Fisher and. and Heterocapsa spp., were consumed in winter Goldie 1959, MacDonald 1927). Thus, while not as common a krill prey as diatoms, dinoflagellates may also be seasonally important.
In addition to zooplankton and phytoplankton, Meganyctiphanes norvegica have been observed to consume a variety of other, less common, prey, including marine detritus, benthic prey, and terrigenous prey. Detritus can be consumed in the water column, either as small particulate detritus, or from aggregates or marine snow. In the Barents Sea microscopic examination of M norvegica gut contents indicated krill had consumed detritus (Dalpadado et al 2008). Fyne organic debris and inorganic detritus were found in 60% to I 00% of the krill samples each week, and it was concluded that "the large amounts usually found indicated that this material probably forms the bulk of the diet"   . Later work in the same region confirmed these observations with fem sporangia found in up to 25% of the krill guts sampled . Fem sporangia were most common as prey in the fall, and more commonly at night than during the day, suggesting they were consumed in the water column . "The fem sporangia found in small numbers in the stomachs of M norvegica, indicate that there may be an inkling of truth in the long held belief that the bracken-clad hills bordering the loch nourish Loch Fyne herring, one of the predators of M norvegica, and impart to them and the kippers made from them their renowned high quality" . In addition to fem pieces, dipteran egg membranes were found as M norvegica gut contents, most commonly in spring, when they were found in up to 35% of the guts sampled, coinciding with the breeding season of these insects . In the Kattegat, a much more open ocean like environment, pine pollen was found to be a common gut contents item in the summer (Lass et al 2001 ). Pine pollen was present in all sampled krill guts in summer, and was fairly abundant in the guts (Lass et al 2001 ).
This observation was further substantiated by the lipid analysis of the gut contents which showed sterol profiles typical of higher plants (Lass et al 2001 ).
While detritus can occasionally be abundant in the water column or as neuston, it is also present at the sediment surface. The above studies did not differentiate between detritus captured in the water column, and that potentially grazed from the sediment surface. Krill are considered to be mainly filter feeders, consuming prey in the water column, and they are well adapted to such a lifestyle with their filtering basket of thoracic limbs. However, some studies have observed more benthic feeding behaviors in Meganyctiphanes norvegica and other krill species. M norvegica maintained in clear tanks containing natural sediment were found to feed on this sediment . They were observed to lie flat on the sediment and beat their pleiopods, thus raising a cloud of re-suspended sediment and then filtered and consumed particles from this sediment cloud by creating a feeding current, . Euphausia superba have been observed to consume benthic materials in-situ. They have been seen to use their extended thoracic appendages to graze on benthic materials . ROY observations of E.
superba have seen them actively feeding on benthic material down to abyssal depths of 3,000 meters near Marguerite Bay, and the researchers "frequently observed a characteristic behavior whereby the krill would nosedive into the sediment and then rise up and feed actively on the re-suspended sediment. Typically krill would dive head first from a height of less than I meter above the seabed and at a fairly steep angle of 30° -50°" (Clarke and Tyler 2008).
It is an open question to what extent Meganyctiphanes norvegica is a selective feeder. Some authors have suggested that M norvegica consumes whatever prey is available, whereas other studies have suggested M norvegica feeds selectively on preferred prey. Selectivity may be based on detection, with larger organisms and more motile organisms easier to detect, or it may represent selection for higher food quality. Suggestions that M norvegica is not a selective feeder tend to be based on similarities between krill diet and water column plankton abundances, "M norvegica are opportunists, feeding on whatever they may find" (Dalpadado et al 2008). In incubation studies of M norvegica using a mixture of copepod species, 75% Centropages typicus, 15% Calanus finmarchicus, and I 0% Pseudocalanus sp., varying in size and swimming behaviors no selection was found .
Some studies have seen krill gut contents mirror changes in water column abundances over time or space leading to suggestions that krill will readily consume whatever is available; "the northern krill can switch between herbivory and carnivory quite opportunistically, depending on food availability" (Lass et al 200 I). One study which illustrated the non-selective feeding behavior of krill was aimed at assessing gut evacuation rate in Euphausia superba and offered krill phytoplankton-sized charcoal particles in an effort to create a continuous feeding environment while observing the evacuation of photosynthetic prey, and surprisingly "Charcoal particles are readily ingested by krill and gradually displace previously digested food" (Perissinotto and Pakhamov 1996). As charcoal particles are clearly a poor food for krill and contain little if any useable nutrients, this consumption of inert particles suggests that krill are not picky eaters, at least in the absence of preferred prey.
Other studies have suggested Meganyctiphanes norvegica is indeed a selective feeder, and consumes prey in different proportions to those found in the environment. Selectivity was seen for Temora longicornis and to a lesser extent for Ca/anus spp., and against cyclopoid copepods (Kaartvedt et al 2002). Several possible explanations were offered for this selection for Temora, swimming behavior, patchiness, and pigmentation (Kaartvedt et al 2002). Temora swims continuously, which would make it relatively easy for a krill to detect, is darkly pigmented, which makes it more visually conspicuous and is distributed particularly patchily, which while making the potential searching distance for a krill greater, might optimize foraging by allowing for rapid feeding once a patch was located (Kaartvedt et al (Kaartvedt et al 2002). Para/Pseudocalanus were the preferred and selected prey of krill studied in the Kattegat and Clyde Sea areas, with Temora as the second choice, and consumed when Para/Pseudocalanus was below the levels required for optimal foraging efficiency (Lass et al 2001 ). M norvegica is a visual predator, in addition to using mechanosensory reception (Torgerson 2001 ). Krill showed much higher predation rates under dim light than in total darkness, and the selective pressure on different copepod species was different in the light as opposed to dark regimes with Metridia lucens consumed significantly more than Calanus in the dark, because it tends to swim faster and more constantly (Torgerson 200 I). This selectivity was still evident in dim light conditions, with Metridia consumed more than Ca/anus, but the difference was much less, suggesting that visual predation is both more efficient than mechanosensory predation, and also allows for the detection and capture of a wider range of prey (Torgerson 2001).
Results are mixed with respect to when during the diel cycle Meganyctiphanes norvegica does most of its feeding. The traditional hypothesis has been that krill feed actively during the night while in the zooplankton rich surface waters, and little if at all in the food poor deep waters where they spend their days. Some studies confirm this hypothesis, finding that M norvegica consumes more during the night than the day (Bamstedt andKarlson 1998, Lass et al 2001). These studies found that the copepod mandibles found in krill guts belonged to copepods living in the surface waters, and therefore must have been consumed at night (Lass et al 2001 ). However, they also note with caution that gut fullness may not be a measure of recent feeding activity, as when feeding activity decreases or ceases digestion rate may also decrease, such that krill captured at deep day-time depths contained many copepod mandibles, but these mandibles were of surface dwelling copepods, and thus were the remains of the previous nights feeding (Lass et al. 200 I).
Other studies, such as those finding benthic detritus as a major food item, suggest daytime feeding may be significant, at least in relatively shallow areas where M norvegica's die! vertical migration takes it to the sediment interface, and have concluded that M norvegica do not show consistent die! rhythms in feeding activity .

Measuring krill feeding
Measuring feeding in very small pelagic organisms such as zooplankton is challenging, and four basic approaches and varying methodologies within them have been taken to gain an understanding of krill feeding in-situ, direct observation, incubations, biomarkers, and gut contents analysis. All of these approaches have their advantages and limitations. Direct observation is challenging in oceanic environments, Incubations may induce behaviors different from those found in-situ and are difficult to scale appropriately, biomarkers suffer from poor temporal and prey type resolution, and gut contents analysis are often limited in prey type and prey resolution and must contend with partially digested samples.
Direct observation of krill feeding has been done using ROVs, scuba divers and surface based observers. ROY observations discovered Antarctic krill, Euphausia superba, feeding on benthic phytodetritius at abyssal depths, as deep as 3,000 m (Clarke and Tyler 2008). Scuba divers have observed E. superba feeding on ice algae . Surface based observers noted a previously undiscovered feeding behavior in E. superba, in which the krill swim just below the surface and "hold one branch of each antennules out of the water. Floating particles are flicked out of the surface film for inspection and sometimes eaten" (Hamner et al 1983 Incubations have been used to look at several aspects of krill feeding. Incubations have several advantages. They are highly controlled, with prey available, light, temperature, feeding time, and pre-feeding treatment all controlled by the investigator. Incubations can also be very quantitative, individual prey can be counted before and after a certain time of krill feeding to determine exactly how many of each prey item the krill consumed during that time. McClatchie conducted a series of feeding experiment incubations in the early 1980s looking at feeding rate and selectivity of Meganyctiphanes norvegica on a variety of copepod and phytoplankton prey in 4 liter containers . When fed different concentrations of the diatom Thalassiosira weissflogii, M norvegica fed at low rates and could not meet its metabolic needs feeding on this phytoplankton alone . In incubations with a mixture of copepods, M norvegica fed at rates correlated with prey concentration, and showed no selectivity amongst different copepods, despite order of magnitude size differences . Torgerson investigated the extent to which krill are visual predators using illuminated and dark incubations in 501 slightly conical tanks (Torgerson 2001 ). He found that krill were indeed visual predators and fed at significantly higher rates under low light than under total darkness. Torgerson also found differences in the rates of different copepod prey under different light conditions, with Metridia relatively more susceptible than Ca/anus in the dark as compared to illuminated conditions, which may be due to differences in swimming behavior, and hence differences in the ability of krill to detect the copepod
This discrepancy highlights one of the potential problems with incubation experiments: captive krill may not exhibit the same feeding behaviors as krill in-situ.
Krill may be damaged in capture and handling, resulting in lower feeding rates than in-situ, and potentially shift to easier to capture prey. Light levels in incubations may not mirror those found in-situ, which would also affect the feeding behavior of the krill, potentially changing the detectability of different prey items. Additionally, krill are relatively large and mobile organisms, potentially able to cover large distances and seek out patchy prey. Thus the prey field presented to the krill in a relatively small incubation may not be representative of the prey field available to them in-situ.
Biomarkers, including stable isotopes and fatty acids, are molecules in prey which are incorporated into the body of the predator, and can be measured to assess in-situ feeding on a variety of different prey. Stable isotopes were used on Euphausia superba to asses feeding on different prey in the West Antarctic Peninsula region (Schmidt et al 2003, Schmidt et al 2006. Carbon isotopes are used as an indicator of the source of original autotrophic production leading to a predator, and Nitrogen isotopes are used as a marker of trophic level, with() 15 N increasing by ~3% with each increasing trophic level (Schmidt et al 2003). The use of stable isotopes allowed for distinguishing between otherwise indistinguishable prey, notable the same species of diatom growing as ice algae or freely floating plankton (Schmidt et al 2003 Prymnesiophyceae, but fatty acid ratios did not consistently support these interpretations . High levels of the fatty acid 22: 1 (n-11) were suggested to indicate krill feeding on C. finmarchicus  confuse interpretation is the ability of some higher trophic level organisms to synthesize fatty acids de-novo . Another potential drawback of the fatty acid analysis technique is the difficulty in determining the linkages of the markers. For example in the Clyde and Kattegat diatom fatty acid markers were found in krill, but it is unclear whether these diatoms were consumed directly by the krill, or whether there diatoms were consumed by copepods, and these copepods then consumed by the krill (Virtue et al 2000).
Biomarkers, both stable isotopes and fatty acids, integrate feeding over time. This can be seen as an advantage as it allows for a more average diet, and potentially more representative. However, this time integration may be a disadvantage in a spatially or temporally heterogeneous environment, such as areas with high seasonality or small scale spatial patchiness. The stable isotope baseline, the signature of the lowest trophic level, was found to be quite variable, with differences up to 10 0 10 0 within only a few weeks time, as the seasonal bloom progressed (Schmidt et al 2003). While these temporal changes in the stable isotope signature were reflected in copepod stable isotopes, the krill, which integrate biomarker signals over longer time periods, did not show similar changes, and this baseline variability may have confused interpretation of krill diet (Schmidt et al 2003). Fatty acid profiles of potential prey can also vary spatially and temporally (Virtue et al 2000). For relatively large and mobile zooplankton such as krill, this may make it difficult to resolve what krill are consuming, and may mask any seasonal or day/night changes in diet.
Another potential difficulty in the use of biomarkers is the low prey type resolution, and complex interpretation of results in species rich ecosystems. Stable isotopes differentiate mainly the original source of carbon fixation and the trophic level of the organism. This may leave many unanswered questions in diverse and complex ecosystems where there may be a variety of different potential prey of roughly the same trophic level and original carbon source, but different behaviors and morphologies. This same relatively poor resolution affects fatty acid analysis, with resolution only to the level of diatoms vs. dinoflagellates .
Interpretation of biomarkers may be particularly challenging in organisms such as krill which potentially feed at multiple trophic levels.
Gut contents analysis is one of the more direct approaches to understanding zooplankton feeding. Gut contents analysis represents a snap shot in time, what the krill consumed in the minutes or hours before it was caught. Several techniques have been applied to analyzing krill gut contents, including microscopic examination, gut pigments, lipids, antibodies, and most recently DNA.
The most commonly used of these techniques has been microscopic examination of gut contents, which involves dissecting the krill , and removing its gut contents, which are stained, and affixed to microscope slides and identifiable prey remains categorized and enumerated. After maceration and partial digestion by the krill, often the only identifiable gut contents are hard parts, such as copepod mandibles and diatom frustules. Microscopic examination of gut contents has been used to investigate feeding by Meganyctiphanes norvegica in the Marginal ice zone in the Barents Sea, in the Skagerrak (Norway), and in the Clyde and Kattegat Seas (Dalpadado et al 2008, Bamstedt and Karlson 1998, Lass et al 2001. Microscopy of gut contents offers some advantages over other techniques, most notably it can offer information about the size or life stage of prey taken, as opposed to simply the species or type, and may have resolution down to species level in low diversity environments (Bamstedt andKarlson 1998, Dalpadado et al 2008). Gut contents microscopy requires few a-priori assumptions about the type of prey taken, assuming mainly that the important prey items contain hard parts and are within the size range to be visible under compound light microscopy.
Potential drawbacks of gut contents microscopy include biases from krill consuming partial prey, difficulties detecting soft bodied or non-descript prey, biases from differences in prey digestibility, and difficulties identifying macerated prey. If prey enumeration is based on specific hard parts, such as copepod mandibles, and krill preferentially eat only the back end of copepods, or suck out the insides of copepods predation will be underestimated, whereas if krill preferentially decapitate their prey and consume only heads, mandible enumeration will over-estimate copepod predation (Bamstedt and Karlson 1998). Different prey will have different detectibilities in gut contents microscopy, with organisms featuring many distinctive hard parts most identifiable and identifiable after the longest digestion time, and organisms with few or no hard parts difficult to detect (Dalpadado et al 2008, Haberman et al 2002. This may lead to overestimation of the importance of hard part containing species, and missing soft bodied prey, such as salps, naked pteropods, or athecate dinoflagellates. Identifying macerated and partially digested prey is one of the challenges of this technique, and often results in a significant proportion of the gut contents being classified simply as unidentifiable, green fluff, digested green or similar (Dalpadado et al 2008, Lass et al 200 I, Bamstedt and Karlson I 998). In gut contents microscopy of M norvegica in Scandinavian waters Lass et al found that "in most cases the major part (80 -90% in the Clyde and 70% in the Kattegat in summer) of the stomach contents was unidentifiable", and SEM imagery suggested this unidentifiable stuff was mainly fragments of centric diatoms (Lass et al 2001 ).
Gut flourescence has been applied to detecting krill feeding on photosynthetic prey. In essence, the krill is dissected and the gut placed in an organic solvent, typically acetone, to extract pigments, and this extract is read on a flourometer. Kaartvedt et al (2002) used gut fluorescence to quantify the relative contributions of photosynthetic and heterotrophic prey for Meganyctyiphanes norvegica in Oslofjorden . The advantages are relative simplicity, and low cost. Potential disadvantages include, a limited range of detectible prey, limited prey type resolution, and high pigment destruction in the guts of krill. Pigment destruction in krill guts is especially high, among the highest recorded for zooplankton, and has been measured at between 58. l % and 98.1 %, so much of the prey signal may be completely lost due to digestion .
Two other methods which have been applied infrequently to analyzing krill gut contents are antibodies and gut content lipids. Gut content lipid measures were combined with microscopy to investigate feeding in Meganyctiphanes norvegica: stomachs were dissected, lipids were extracted, and total lipids, sterols and fatty acid methyl esters were measured by gas chromatography (Lass et al 200 I).
While this method does not have particularly high prey type resolution, it does offer insight into some prey items which may not be detectible by microscopic examination of gut contents, such as bacteria or detritus (Lass et al 2001). An enzyme linked immunosorbent assay (ELISA) was used to detect Euphausia superba feeding on Phaeocysitis antarctica (Haberman et al 2002). Antibodies were developed in rabbits specifically for P. antarctica, tested on lab reared krill, and then used to assess in-situ feeding of E. superba on P. antarctica, (Haberman et al 2002). Antibodies are very sensitive and able to detect even low abundance, and visually inconspicuous prey in krill guts. However, antibodies are time consuming and expensive to design, and results are difficult to replicate in different labs. Additionally, an antibody approach must necessarily include a-priori assumptions about what species, or groups the krill are consuming, and are likely to be limited to a small number of potential prey targets in any given study.
Dioxyribose nucleic acid (DNA) offers many advantages as a gut contents marker, and has been used in a few recent studies of feeding in Euphausia superba krill. As a universal information carrying molecule DNA can be used to detect any living, or recently living, prey in krill guts. Interspecies variability in nucleotide sequence can be used to identify the species source of the DNA in krill guts.
Passmore et al investigated Euphausia superba feeding on diatoms using DNA from krill guts . l 8S genes from all diatom DNA present in extracts from krill guts were amplified using a polymerase chain reaction (PCR) and diatom group specific primers . Primers can be designed to amplify species, or groups, with different levels of phylogenetic relatedness by taking advantage of the different regions of the DNA which show different levels of variation. Comparing the abundance and diversity of diatom species found by sequencing of gut contents DNA clone libraries, to that found by gut contents microscopy, showed similar species compositions, but in different proportions .
Martin et al investigated Euphausia superba feeding on a range of prey items around Anvers Island . Krill gut contents DNA was amplified using a PCR reacting with universal eukaryote 18S primers designed to amplify any eukaryotic organism and then run through a denaturing gradient gel electrophoresis (DGGE)  DGGE is a gel electrophoresis containing a gradient of formamide and urea used to separate DNA amplicons based on length and sequence . They identified 26 different prey phylotypes, and sequenced many of these to determine their identity. However, issues with this technique included the necessity to dissect away all krill tissue before DNA extraction, co-migration of some bands, and multiple bands derived from the same prey sequence . Frisher Detection and discovery of crustacean parasites in blue crabs (Callinectes sapidus) by using 18SrRNA gene-targeted denaturing high performance liquid chromatography, Applied and Environmental Microbiology, pp. 4346-4353, (2008)