KILS, U 1982 Swimming behavior, Swimming Performance and Energy Balance of Antarctic Krill Euphausia superba. BIOMASS Scientific Series 3, BIOMASS Research Series, 122 p - complete translation of phD thesis of May 16 1979 of the Mathematisch Naturwissenschaftlichen Fakultaet der CHRISTIAN ALBRECHTS UNIVERSITAET zu KIEL, INSTITUT FUER MEERESKUNDE, also published as BERICHTE AUS DEM INSTITUT FUER MEERESKUNDE No.163, 1979

other languages available: German

this is the complete translation of the doctor thesis with Gotthilf Hempel ALFRED WEGENER INSTITUTE for POLAR RESEARCH into English by Carl Boyd, DALHOUSIE UNIVERSITY, Halifax, CANADA - initiated, financed and published by Sayed ElSayed, A & M UNIVERSITY Texas USA - this english version can be ordered at the SCAR SCOR office for $10 - right now on this server are the first rescans and some links to images, we are working on a CD Rom version with updates, high resolution images, videos and links - if you have anything to contribute or found errors please email

Swimming behaviour, Swimming Performance and Energy Balance of Antarctic Krill Euphausia superba


Scientific Series No. 3


Uwe Kils


Foreword by Sayed Z. El-Sayed

In early 1980 Professor Carl Boyd of the Department of Oceanography, Dalhousie University, brought to my attention the translation into English he made of a Ph.D. thesis entitled "Schwimmverhalten, Schwimmleistung and Energiebilanz des antarktischen Krills, Euphausia superba." (The Swimming Behavior, Swimming Performance and Energy Balance of Antarctic Krill, Euphausia superba). I requested that copy of this translation be sent to me, and upon reading it I immediately recognized its significance. I then got in touch with its author, Dr. U. Kils, asking him if he would like to share his interesting work with the BIOMASS community by allowing us to publish it in the BIOMASS Research Series. In March, 1982, Dr. Kils prepared for publication the manuscript presented here.

As Dr. Kils indicated in the "Preface," the research reported on here was carried out mainly during the second German Antarctic Expedition on board the fisheries research vessel WALTHER HERWIG during the austral summer of 1977/78. The state of knowledge and literature is, therefore, that of 1979. With the current interest in the Living Resources of the Southern Ocean, in particular those of the krill, Euphausia superba, Dr. Kils work which has earned him the "Maier Leibnitz Award" for 1979 for the best publication of a junior scientist in West Germany will have a great impact on current and future research on the physiology and behavioral biology of that organism, and will certainly elucidate the vital role it plays in the Antarctic marine ecosystem.

I am indeed most grateful to Dr. Kils for sharing his valuable contribution with the BIOMASS community and for allowing us to publish his work as a BIOMASS publication. I also wish to acknowledge with thanks the financial contributions of SCAR and SCOR which have made this publication possible.

Sayed Z. El-Sayed, Convenor

Group of Specialists on Southern Ocean Ecosystems

and their Living Resources

Editor, BIOMASS Research Series

College Station, Texas August 1982


This SCAR BIOMASS report series no. 30 is the direct translation of my Ph.D. thesis: "Schwimmverhalten, Schwimmleistung und Energiebilanz des antarktischen Krills, Euphausia superba. Ergebnisse der 2. deutschen Antarktis-Expedition des FFS Walther Herwig im Suedsommer 1977/78. Universitat Kiel, 1979. The state of knowledge and literature is that of the year 1979." Some of the results have been published separately already (see foot-note *), some will be published in elaboration with results from a recent expedition soon.

* KILS, U., 1979: Preliminary data on volume, density and cross section area of Antarctic Krill, Euphausia superba. Meeresforsch. 27, 207-209.

KILS, U., 1979: Swimming performance and escape capacity of Antarctic krill, Euphausia superba. Meeresforsch. 27, 264-266.

KILS, U., 1979: Aspects of physio1ogical ecology of Euphausia superba. ICES, C.M./L:54, 1979.

KILS, U., 1980: Size dissociation in Krill swarms. Proc. 15th Europ. Mar. Biol. Symp., Kieler Meeresforsch., Sonderh. 5, 272-273.

ALBERTI, G., KILS, U.,: The fi1tering basket of Euphausia superba. ICES, C.M./L:54, 1980.


The behavior and physiology of Antarctic krill Euphausia superba is the object of this investigation, which consists of four sections:

The combination of these four complexes allowed for the calculation of the energy budget of krill - employing two different, independent methods. It appears that the extraordinary body size of the Antarctic krill is the key factor which makes it possible for this animal to utilize the enormous, but extremely variable primary production of the Antarctic. On the other hand, the very large body size causes tremendous physiological and energetic liabilities: adult krill is much too heavy for its exclusively pelagic way of living; it does not possess a device to increase its buoyancy such as a swim-bladder or high oil content. Therefore, the krill has to defray a rather considerable amount of its metabolism to the struggle against sinking. This portion increases with body size, which has the following implication: metabolism in krill does not increase in proportion to a body surface, as is common in most other aquatic animals, but in proportion to body weight. This extraordinary performance can only be accomplished in very well oxygenated waters - a fact, which yields a correlation between krill distribution and hydrological regime.

The minimum food requirements and the swimming speed necessary for obtaining the minimum amount of plankton filtered are estimated. Any lack of sufficient oxygen or sufficient food or any weakening of the animal results in an inevitable sinking to the bottom, entailing a high natural mortality (M). This mechanism should provide the abundant bottom fauna with an abundant food supply. It could be demonstrated that from the energetic point of view, krill is capable of migrating several hundred kilometers per month. The results of experiments pertaining to sensory mechanisms (light and gravity) and many observations on krill behavior are added to increase the knowledge of the biology of Euphausia superba.


The Antarctic krill, Euphausia superba, is one of the most successful species of our earth; its biomass is probably larger than that of man. However, our knowledge about it is very incomplete. This is especially true for the physiology and behavioral biology of the krill, to which this investigation will be devoted. (Literature survey: See Discussion).

It was not only the collection of missing facts about a relatively unknown animal which attracted me to this work; Antarctic research offers the opportunity to investigate "Extremes", and the knowledge of "Extremes" often lays bare the principle of the "Normal" - for example, the biology of temperate zones which then allows the recognition of what actually makes the "Extreme" so extraordinary. For this s reason, the krill was not investigated in isolation, but rather some measurements were extended to other crustaceans

Through methods as diverse and independent of each other as possible, I have tried to represent some aspects of the way of life of the krill, as well as to verify them numerically, as a mosaic stone for an ecomodel of the Antarctic. Swimming mechanism, swimming behavior and swimming performance were analyzed with the help of aquarium- and underwater photography and filming. They served, together with hydrodynamic investigations, respiration measurements and diverse biometric data, to estimate the energy balance.

The investigations took place mainly during the Antarctic expedition on board the fisheries research ship WALTER HERWIG in the austral summer of 1977/78. For material provision and station data see KILS 1979.

A part of the experimental results has already been published: KILS 1979 Performance of Antarctic krill, Euphausia superba, at different levels of oxygen saturation, Meeresforsch 27 35-47. A knowledge of this publication is helpfull for the discussion of the energetic bilancing project; a copy is attached in the Appendix with permission of the editor.

l. Analysis of swimming

Materials and methods

To get a general impression of the swimming behavior of the krill, it was kept in 63 liter Plexiglas tanks on board. These tanks were closed with plexiglass covers and, airfree, filled with seawater (T = 1 degree Celsius). The laboratory was located at the height of the water line in the middle of the ship, so that the contents of the tanks scarcely moved. A parallel sheet of Plexiglas was hung behind the front pane at a distance of 4 cm (with a 5 cm measuring scanner). Only krill, which swam through this stratum (4 cm x 50 cm x 30 cm) were used for evaluation. A film camera was directed on this measuring zone, with a macro-objective of large focal distance (object distance = 48 x object size, so that the length error, affected by the 4 cm of play in the measuring zone, was smaller than 6 %). The picture progression was 18 ms; 25,000 pictures were taken. The lighting equipment could be set from 1400 - 88,000 lux. The film material was overexposed by 1.5 aperture settings, so that the fine bristle ends of the pleopods also would be within the contrast range. When full lighting was suddenly turned on out of the dark, the animals reacted with energetic flight movements. This behavior could be used to produce and measure maximum swim speeds. When the light became slowly stronger out of the dark, this reaction did not occur; even 88,000 lux did not seem to irritate the animals (noon sun = 100,000 lux at sea surface). In order to follow the flow of water produced, tiny floating bodies (density = 1.028) were mixed into the water.

In addition, one aquarium was observed over several days by a camera system, which took a photo of the whole tank every nine minutes. Later the swimming angles of the animals were evaluated with these photos. In the daytime the laboratory was lighted by neon tubes (650 lux), at night it was thoroughly dark. The exposure was done by two very weak flashes (together 'ca. 130 BCPS).

The position of the very fine bristles on the pleopods during the stroke was investigated under a dark field illumination system: living krill were gently squeezed between two glass plates underwater to fix the animal in space, while they continued with pleopod movements. During the movement the position of the bristles was recorded by a very short flashlight (t = 10 ps) on 60 x 60 mm high resolution film material with precision optics - the setup was especially ray-traced, corrected and constructed for this Antarctic mission.

To get an impression of swimming behavior in the natural environment, a conventional underwater television camera was used in depths from 0 - 30 m.

For investigations of micro dynamics and distributions a new underwater camera system was developed and applied in situ. High resolution optics and two high speed XENON flashes pointed forward from a streamlined aluminum body (diameter = 20 cm). The large aperture optics are calibrated so that only a narrow 4 cm area between 108 and 112 cm in front of the camera is scanned with high contrast gradients (fig.1; depth of focus).

Fig. 1. Autonomous underwater imaging system
Each picture is exposed by two XENON flash-lamps (dt = 20 ms) firing sequentially, so that a moving krill is imaged twice, allowing for an evaluation of speed and direction. Only objects with a threshold contrast are evaluated (i.e. at 108 - 112 cm in front of the lens). These scans give information about swimming direction, speed, distribution in space, animal length and school-concentration. This system was deployed in two strategies (fig.l):

l. as an autonomous free-fall imaging system from the drifting or slow moving research vessel:

built-in adjustable lead ballast and stabilizing wings at the trailing end allow the system to descent vertically at speed of 3 m per second. During the freefall every 0.33 seconds an image is scanned, frozen with a double flash, so that every meter a 4 cm thick zone is investigated. At a depth of 45 m the freefall is terminated by a thin nylon cord hitting its stop, the inertia turning off the optics, and recovering the system

2. as a towed imaging system aside/behind the traveling ship:

In a depth of 17 m the camera is sheer-dragged at 6 kn (= 3.1 meter per second) speed with a hydrofoil sheering it out of the ships turbulence and noise wake. Every 20 seconds (corresponding to about 62 m) an image is scanned.

The capacity of camera, flashes and electronics is 3600 images.

The flow field of the water produced in swimming was investigated with Meganyctiphanes norvegica. For this purpose the animal was glued (with cyano-acrylate) by the carapace to a thin cord and the path of water was photographically followed by means of drifting bodies (density = 1.028). All other experiments of this chapter were done with Euphausia superba on board the FFS "Walther Herwig" in January 1978 (catch time, catch area, catch method and choice, see KILS 1979).

For photometric evaluation the films were projected on the focusing screen grid of a simgle-photograph plotting apparatus.


The krill uses two methods to produce the propulsion for active forward movement: swimming with pleopods, which probably must be seen as the normal method of forward movement and which can be continued over very long periods of time, and a swimming behavior characterized by jerky driving of the tail (in the following called "tail swimming"), which is interpreted as a flight reaction and probably tires the animal quickly.

1.1 Pleopod swimming

The krill has five powerful pairs of legs on the first to the fifth abdominal segments (fig.2a and 2b), the pleopods.

Fig.2a Lateral view of an adult krill (length 58 mm)

Fig.2b Position and morphology of the pleopods

The pleopods can be compared with a paddle: the two-jointed protopodites as stalks carry large surfaces, formed by exo- and endopodites and their bristly edges. Fully spread out, the length of this "paddle" is almost 1/4 of the krill' s length. The pleopods are joined to the caudal-ventral end of the abdominal segments and are propelled by powerful muscle cards running diagonally after the cranial-dorsal (fig.3).

Fig.3 Pleopods during beating

The protopodites are rhythmically stroked forwards and backwards, the two legs belonging to each segment synchronized. During the backward stroke the exo- and endopodite are spread by the muscles of the protopodite. The bristly edges spread out because of the water pressure and a broad, deeply ventral forward-reaching surface is formed (fig.4 and 5).

Fig. 4-1. Front view of a pair of pleopods (schematic)  during the power stroke phase at maximum propulsion - the arrows animate forward or backward. The two endopodites are synchronized with the purple connection apparatus, attached in the centerline with a "vrelco" hookfield which allows for temporary opening  (in preserved animals usually lost)

Fig. 4-2. Front view of a pair of pleopods (schematic)  during the beginning of the backstroke folding the setae close to the endo- and exopodite and laying those on top of each other

Fig. 4-3. Front view of a pair of pleopods (schematic)  during the backstroke of the paddle (which is directed from the back to the front part of the animal - the endo- and exopodites are folded close to the body to reduce drag

Fig. 4-4. Front view of a pair of pleopods (schematic)  shortly before the begin of a new power stroke during unfolding all structures again

Fig. 5. Pleopods during beating

During the forward stroke the exo- and endopodite are laid one over the other (overlapping). The bristles fold backwards because of the now opposing water pressure and lie close to the exo- and endopodite. These are also brought into a position parallel to the body (not shown in fig.4, see fig.6: intense beat at 200 ms, and fig.3: Second pleopod), so that, as seen from the front, only the protopodites are left as resistance surfaces. This change in surface is contributed to by the fact that the exo- and endopodite as well as its bristles are movable backwards, but forwards they cannot be folded over any further than the plane of the paddle surface. The difference of the forces between the forward and the backward stroke produces the propulsion.

The locomotive apparatus, which seems so clumsy and ineffective in preserved animals, proves, through investigations with living Krill, to be a highly developed, efficient propulsion organ (fig.5). With the help of this, as will be shown, considerable swimming performance is achieved.

The pleopods strike metachronally from backwards to forwards (fig.7), and a single amplitude lasts exactly as long as the stroke wave takes to traverse the distance from the 5th to the 1st pleopod (= Phase); i. e., when the first pleopod has ended its stroke, the fifth is just beginning anew, so that at any given point in time one of the pairs of legs is in the position of maximum thrust. The six exopodites of the thoracopods (which are irrelevant for the swimming process) also traverse a metachronized series of strokes from backward to forward with the same phase. The stroke of the sixth exopodite follows immediately the stroke of the first pleopod, so that a wave of agitation traverses the animal from back to front (during two phases).

Fig. 6. Pleopod beat timeseries. lateral view during soft beat, medium beat (middle) and intense beat (below) - upper line protopodite, lower line exo- plus endopodite

Fig. 7. Amplitudes of the pleopods (below) and the exopodites of the thoracopods (above) in resolution over time. Numbers count from the front end of the animal

In the forward position, the protopodites of the pleopods are almost parallel to the body. The angle which the protopodites pass through in a stroke is shown in fig.8 as a black surface. It is to be noticed that this angle increases from front to back: from 62 degree for the first pleopod to 132 degree for the fifth pleopod. The surface passed through by the endo-exopodites of the fourth pleopod is shown in gray (shaded); these are drawn correspondingly in the most forward and most backward positions. Also the path lines of the leg tips on the backward stroke (outer curve) and the forward stroke (inner curve) are shown. The areas passed over by the exo- and endopodites are each given in am~ (the figures in parentheses correspond to the difference between forward and backward stroke). The figure at the top gives the sum of the surfaces ( A u 8 u ...), the figure under it gives the total surface without overlapping ( A u 8 u... - A n B - B n C - ...).

Fig. 8. Path of pleopod tips and the covered area - lateral view

Fig. 9. Path of pleopod tips and the covered area - ventral view

Fig.9 shows the paths -of the pleopod tips in a ventral view - the inner line for the forward stroke, the outer line for the backward stroke. Thus i t becomes cl ear how far the pleopods reach laterally into the water and also how closely they can be pulled together during the forward movement. Shown in gray is the surface passed aver by the right fourth pleopod. The points are 18-ms time markers (only on the outer curve).

The flow field produced by the pleopods is shown in fig.10. The length of the arrows is a measure of the flowing speed. The water is flung out in a relatively wide area. The thick lines indicate the flow boundary of the "propulsion stream". It is remarkable that the water is not propelled directly backwards, but rather on the average at ca. 40' to the back line, diagonally downwards. This will be discussed more extensively later.

Fig. 10. (left) and Fig. 11. (right) Flow field of a soft pleopod beat (left) of an intense beat. . The length of the arrows is a measure for the flowing speed. The thick lines indicat ethe flow boundary of the "propulsion stream"

Fig.12. Flow field at a changed telson angle

When the pleopod strokes become stronger, the fluid dynamic field changes as plotted in fig.11): the stream is no longer so widely fanned out and the average direction decreases to 30 degree relative to the dorsal line. In the center of the wake the flow increases to over 6 cm per second.

The influence of the telson is to be seen in fig.12: The forward flow boundary remains relatively unchanged, while the backward one is displaced forward creating a rather concentrated stream with an average direction of over 40 degrees to the back line.

In the forward area of the krill a flow runs towards the head of the krill, into which the animal only needs to hold its filtering basket to be able to filter, even without forward movement.

1.2 Tail swimming

The second means of fast movement is tail swimming: The spread-out telson is flipped towards the area of the filtering-basket, accelerating the animal backwards. In this streamlined shape the krill escapes with ca. 100 cm per second backwards. After a distance traveled of about 1.5 - 2 body lengths the animal stretches its tail out again while the telson is folded together. It then repeats the same process with several more strokes (fig.13 and 14). An average speed up to 60 cm per second is reached, which can be maintained over at least 50 cm, probably more. The stroke frequency of seven strokes per second is astonishingly high and cannot be seen with the naked eye.

That, even at this high speed, the krill is still able to keep control over its path is indicated by fig.15. In the path of the startling krill another animal is positioned, (drawn darkly). Shortly before the collision the darting krill slows down and changes the direction a little; then continues to dart away at a higher speed again without having touched the other animal. The stalk eyes are held high above the body during tail swimming, so that they can look in the swimming direction in spite of the backward movement.

During the stretching of the tail against the movement direction, the speed necessarily is lowered somewhat; so this kind of movement is more irregular than pleopod swimming, as the time analysis of speed in fig.16 shows.

Fig.13 Tail swimming "lobstering" (resolution over time); contour and eye drawn at dt = 18 milliseconds, some stages shaded: a = acceleration phase, b = speed phase. Below: the moment of the strike in two stages - a click into the lower image toggles to the next time stages

Fig. 14. and 15. Tail swimming "lobstering" (resolution over time)

Fig. 16. speed over time during high speed pleopod swimming and tail escape swimming

1.3 Swimming speeds

The attainable swimming speeds depends, as with all animals, on body size. Fig.17 shows this relationship. The points mark pleopod swimming (the larger points show maximum performance), the squares tail swimming. Krill can reach about 8 times its body length per second with pleopod swimming and 11 times its body length per second with tail swimming. In spite of the first impression that its body does not appear very streamlined, krill compares in its swimming performances even with fast-swimming fish of comparable size (ALEYEV, 1977).

Pleopod swimming is clearly the normal method of forward movement, tail-swimming the exception. In the aquarium the latter was triggered by sudden intensive lighting.

When freshly caught krill were touched with the hand in the aquarium, they reacted with tail swimming. Whether they use their potential for flight against fish and fishing net is not certain. MARR (1962) observed fish snapping up krill near the surface from below as easy prey, without any flight reaction being noticeable. The krill reacted with tail startles to the underwater TV camera and the pulled camera, both approached from the side, as is indicated by the typical position in fig.18. The krill did not react with tail swimming to the free fall camera, which approached from above.

Fig. 17. Maximum swimming velocity of pleopod swimming (points) and tail swimming (squares) - x and s average speed and variance of undisturbed krill in aquaria

Fig.18  Double-flash photography of the drag camera, showing krill at the typical posture of tail swimming (digitally enhanced)

1.4 Methods to increase speed

The investigated krill showed two dynamics used while migrating faster: an increase in stroke frequency and a change of the stroke. Fig.19 plots the increase stroke frequency of the pleopods (block points, left ordinate) over cruising speed.

Fig. 19. Stroke frequency and transport traveled per stroke over increasing cruising speed - length of animals 40 - 50 mm


In the range from 15 to 40 cm per second the relationship is linear, with an extrapolation running through the zero point. For speeds under 15 cm per s, the frequencies are above this simple linear relation; the ordinate is cut at a minimal f of 2.5 beats per second. None of the investigated animals (in size ranging between 40 and 50 mm) showed a lower frequency than 2.4 beats per second. Even dying animals maintained this frequency. The stroke rhythm seems to be automatic, because dissected abdomens stroked autonomously in the typical way for several hours (metachronized and as described above).

However, the krill can stop stroking completely; the pleopods of adult Euphausia superba stroke, therefore, either not at all or with a frequency above 2.4 beats per second. This minimal frequency depends on the size of the animals, as plotted in fig.20. The circles stand for Euphausia superba, the points for Meganyctiphanes norvegica.

Fig.20 Minimal stroke frequency of pleopods with increasing body sizes

                 fmin = 1.69 + 31.2 e - 0.0789 L                       r = 0.983

                (units: beats per second, mm)

In the speed range from 0 to 15 cm per s, in which the linear zero-point-intercepting relationship is not kept up to, other factors must influence the swimming dynamics. If the speed is compared with the distance accomplished per stroke (fig.19, circles, right ordinate), this distance increases slowly starting from zero, i.e. for the achievement of a higher forward velocity, a single pleopod-stroke transports the animal over a greater distance - the stroke must therefore have become more efficient. At 15 cm per second this relation has a bend, it becomes parallel to the abscissa; i.e., from here on there is no more increase in the stroke efficiency; over 15 cm per s the above described linear & zero-intercepting relationship of velocity to frequency is shown.

The reasons for the change in stroke efficiency are demonstrated in fig.21 with the example of the second pleopod of a 48 mm long krill. From the soft- over the medium- to the intense stroke the frequency changes very little (from 2.5 over 2.8 to 3 beats per s); however, the attained swimming speed changes from 0 over 4 to 7.5 cm per second . The following changes were observed: the covered angle of the protopodite increases from 34 degrees over 61 degrees to 84 degrees. The surface swept by the exo- and endopodites increases from 29 mm2 over 53 mm2 to 63 mm2 (partial surfaces in parentheses). On the forward stroke the exo- and endopodites are withdrawn closer to the body, from 6.7 mm over 5.9 mm to 4.9 mm. The displacement of the pleopod tips increases from 11.2 cm per s to over 20 cm per s

Fig.21 Changes in stroke execution of the 2. pleopod at different stroke intensities

Indicated in fig. 21 are also the centers of the areas swept by the exo- and endopodites, and the theoretical force-vector tangential to the center of the abdominal joint. Since the increase in angle size of the protopodite only occurs backwards, the angle of the force to the dorsal line becomes smaller, which is confirmed also in the flow-field (fig.10, 11); in parentheses the length of the drawn-in theoretical force vectors as relative figures weighed on the areas (the areas shown cross shaded calculated with double weight and double power, as the covered water is 3 dimensional).

Krill regulates its speed in the range of 0 - 15 cm per s predominantly by changes in stroke execution with an optimum reached at 15 cm per second. In the range above 15 up to 40 cm per second the stroke frequency alone controls the speed. At 40 cm per s the pleopods propel the animal with over 10 strokes per second, an astonishing performance for such lightly built constructions in the rather viscous medium water. Higher speeds can only be achieved by tail swimming, which the krill uses between 40 and 60 cm per second. In tail swimming (backwards), the antennae, antenna scales and the filtering basket are trailed behind the body, and are thus better protected from the fluid-dynamic pressure at such high velocities.

goto chapter 2 verhalten oekologie behaviour ecology southern ocean suedpolarmeer sueden sued south wale wahlnahrung global warming carbon cycle kohlenstoff kreislauf suedhalbkugel southern hemisphere animation bewegung propulsion paddel paddeln einfache animation teachware interaktives lernen remote education remote teaching distance learning cyberschool cyber school web schule webschool daumenkino flipmovie flip movie daumen kino iceshrimp ice shrimp bio energetics eco energetics image bank behavior ice life exobiology exo biology europa jupiter moon callisto callistro jo io moon pluto extratrrestrial life ausserirdisches leben mond planeten planets space weltraum mission weltraumfahrt leben unter dem eis ocean ozean entstehung des lebens exobiology icelife astro biology astro biologie astrobiology carbon sequestration kohlenstoff bindung kohlenstoff festnahme kohlenstoff falle kohlensotff senke carbon sink co2 sink co2sink