Krill aggregations vary in size, krill density and uniformity depending on the species of krill. These aggregations may be structured to allow individuals to sense the hydrodynamic cues of neighboring krill or to avoid the flow fields of neighboring krill, which may increase drag forces on an individual krill. To determine the strength and location of the flow disturbance generated by krill, we used infrared particle image velocimetry measurements to analyze the flow field of free-swimming solitary specimens (Euphausia superba and Euphausia pacifica) and small, coordinated groups of three to six E. superba. Euphausia pacifica individuals possessed shorter body lengths, steeper body orientations relative to horizontal, slower swimming speeds and faster pleopod beat frequencies compared with E. superba. The downward-directed flow produced by E. pacifica has a smaller maximum velocity and smaller horizontal extent of the flow pattern compared with the flow produced by E. superba, which suggests that the flow disturbance is less persistent as a potential hydrodynamic cue for E. pacifica. Time record analysis reveals that the hydrodynamic disturbance is very weak beyond two body lengths for E. pacifica, whereas the hydrodynamic disturbance is observable above background level at four body lengths for E. superba. Because the nearest neighbor separation distance of E. superba within a school is less than two body lengths, hydrodynamic disturbances are a viable cue for intraspecies communication. The orientation of the position of the nearest neighbor is not coincident with the orientation of the flow disturbance, however, which indicates that E. superba are avoiding the region of strongest flow.

Organized social behavior (e.g. schooling), often exhibited in aquatic organisms, can improve fitness through reduced predation rates (Burgess and Shaw, 1979), increased foraging success (Baird et al., 1991; Foster et al., 2001) and reduced energy expenditures (Hamner and Parrish, 1997; Ritz, 2000). In contrast, there are costs to the individuals within schools including lowered food availability at certain positions within the school (Ritz and Metillo, 1998), increased risk of predation and increased risk of disease and parasitism (Hamner, 1984; Gómez-Guttiérrez et al., 2003). Krill are found in a variety of aggregations such as schools (Hamner and Hamner, 2000), swarms, superswarms (Tarling et al., 2009) and layers (Watkins and Murray, 1998). Schooling krill have been described as uniformly oriented with rapid collective behavior and synchronous execution (Hamner, 1984; Hamner and Hamner, 2000). The distinction between swarms and schools is not consistent within the literature, perhaps because the structure of krill aggregations varies from coordinated to unorganized in both field (Hamner et al., 1983; Hamner, 1984; Tarling et al., 2009) and laboratory observations (Kawaguchi et al., 2010). Because there are costs and benefits to schooling, the diversity in aggregation structure may reflect a changing ratio of costs and benefits that occur for different species of krill, krill size or life stages.

Despite having similar morphology and propulsion mechanisms, the schooling abilities of Euphausia pacifica and Euphausia superba are different. Euphausia superba (Antarctic krill) has been described as an obligate schooling species that routinely migrates hundreds of kilometers (Kils, 1983; Hamner, 1984). Aggregations of E. superba are segregated by size, gender, molt state and feeding state (Marr, 1962; Nicol, 1984; Johnson and Tarling, 2008; Tarling et al., 2009). Euphausia pacifica (Pacific krill) is a smaller species of krill that is only occasionally found in aggregations (generally described as swarms), which may be restricted to mature krill during reproduction (Endo, 1981; Nicol, 1984). Hanamura et al. observed that E. pacifica has the ability to form aggregations similar to schools (Hanamura et al., 1984), but visual and in situ acoustic observations have shown that E. pacifica groups predominantly occur in uncoordinated aggregations or layers (Hanamura et al., 1984; de Robertis et al., 2003). Zhou et al. observed that individuals of a larger species of euphausiids are more agile swimmers and able to maintain tighter, less random aggregations than smaller species of euphausiids (Zhou et al., 2005). This observation suggests that larger krill (E. superba) may be better able to control their position in the environment and more likely be found in schools.

Swimming capacity is dependent on size (Johnson and Tarling, 2008) and organism size is directly related to the spatial extent of the hydrodynamic disturbance, which potentially provides a sensory field for prey, predators and conspecifics (Abrahamsen et al., 2010). Krill flow fields have been identified as a source of hydrodynamic sensory cues between individuals within schools (Hamner, 1984; Wiese and Ebina, 1995; Wiese, 1996; Yen et al., 2003) and it is suspected that schooling krill will orient in positions within a school to receive hydrodynamic cues from their neighbors (Wiese, 1996). The dominant frequency within the flow field is equal to the species-specific pleopod stroke frequency, and this dominant frequency is larger for smaller species of krill (Wiese and Ebina, 1995). Additionally, in experiments on tethered specimens, only the largest euphausiid species, E. superba, was found to produce a hydrodynamic disturbance of a spatial extent longer than several body lengths (Ebina and Miki, 1996). Thus, there may be an organism size limitation for schooling that is caused by the limited spatial extent of the sensory field. A comparison of quantitative flow field data of these two species may provide insight to costs and benefits of krill orientation within aggregations during schooling.

A few studies have been published on the flow fields generated by tethered, individual specimens of E. pacifica (Yen et al., 2003) and E. superba (Kils, 1982; Ebina and Miki, 1996). However, flow fields collected from tethered krill specimens have limited value because tethering alters the behavior of the krill (Yen et al., 2003) and the flow fields produced by the specimen (Catton et al., 2007). To date, no studies have examined the flow fields around free-swimming coordinated groups of krill. In this study, we quantified the flow fields produced by free-swimming solitary E. pacifica, solitary E. superba and small coordinated groups of E. superba to investigate the differences in the flow field structure of these two species of krill and quantify the spatial extent of a potential hydrodynamic cue.

Our interest is in the sensory ecology of krill, focusing here on the creation, structure, transmission and persistence of the hydrodynamic signal produced by two species of krill that differ in size, kinematics and habitat. Specifically, we aim to address the following questions: (1) in what aspects do the flow fields generated by two species of euphausiids differ, (2) where would individual krill of both species align to use the propulsion-generated flow field as a hydrodynamic cue and (3) how does aggregative behavior alter the flow fields?

Animal collection and care

Euphausia pacifica Hansen 1911 individuals were collected in July 2007 with a plankton tow off the continental shelf approximately 25 miles off the coast of Newport, OR, USA. The collected individuals were kept in a 100 l tank inside a 10°C cold room at the Hatfield Marine Station in Newport, OR, USA, for a period of 2 weeks during the experiments [courtesy of W. Peterson, National Oceanic and Atmospheric Administration (NOAA) Fisheries Department at the Hatfield Marine Station, Newport, OR, USA]. Euphausia superba Dana 1850 individuals in this study were in culture at the Australian Antarctic Division in Kingston, Tasmania. The krill were originally collected using a rectangular midwater trawl net in the Southern Ocean in February 2005 and March 2006 and kept onboard the RSV Aurora Australis in 200 l tanks until arriving in Hobart, Tasmania (Kawaguchi et al., 2010). The krill were maintained in 1000 l tanks with recirculating, chilled, filtered seawater at a temperature of 0.5°C and salinity of 34.5‰. The flow field data reported herein were collected in December 2006 and January 2007.

Experimental setup

The flow fields generated by free-swimming euphausiids were measured using particle image velocimetry (PIV) with infrared illumination. PIV is a non-intrusive flow visualization technique that quantifies a velocity field by recording the location of tiny suspended particles via a digital camera and laser sheet illumination of the measurement plane (Raffel et al., 1998; Catton et al., 2007). The velocity vector is measured by quantifying the displacement of the particles over a known time period using a pair of digital images. The PIV system in the present study consisted of an 808 nm infrared laser (Oxford HSI-500, Oxford Lasers Inc., Shirley, MA, USA), a 1280×1024 pixel CMOS camera (VDS Vosskühler CMC-1300, VDS Vosskühler GmbH, Osnabrück, Germany) and an image acquisition laptop. In addition, a Pulnix TM-745i camera (JAI Inc., San Jose, CA, USA) was located in a perpendicular orientation to the primary digital CMOS camera in order to observe the position of the krill within the laser sheet. The laser sheet was 1 mm thick and illuminated an imaging region of 80×60 mm (vertical×horizontal). The laser was located above the tank for all data collection. For the majority of cases, the laser formed a light sheet in a plane defined by the vertical and swimming directions. One variation of this arrangement was employed to capture data for the dorsal view. In this case, the laser was rotated by 90 deg to form a light sheet defined by the vertical and transverse directions. Krill behaved naturally while swimming in and near the infrared laser sheet, which is consistent with our previous observations with other zooplankton. The time lapse between paired images ranged from 7 to 11 ms, and image pairs were collected at a rate of 25 Hz.

Collection of data for solitary krill was performed in a square glass tank (15×15×15 cm) that was maintained at the required temperature of 10°C [kinematic fluid viscosity (ν)=1.37 mm2 s–1] for E. pacifica and 0.5°C (ν=1.79 mm2 s–1) for E. superba. The seawater in this tank was quiescent, such that flow fields generated by krill could be quantified and compared between species.

Euphausia superba were housed in a circular white plastic tank that was conducive to group forming and schooling behavior (tank diameter=600 mm, tank height=400 mm). White surroundings have been shown to enhance the schooling behavior of E. superba, resulting in tighter, more cohesive schools (Kawaguchi et al., 2010). A 150×400 mm glass window was added to the tank sidewall to allow optical access. Data were collected during daytime with background illumination from the room. Prior to collection of flow field data, the temperature of the tank was maintained using a chilled water supply. The water supply was stopped at least 10 min prior to data collection; hence any fluid motion in the tank was generated by the krill. The experiments were conducted for a time period of less than 1 h such that the temperature in the tank rose by less than 1°C.

Krill handling

For measurements of solitary krill, specimens were introduced individually into the measurement region through a PVC pipe (funnel) aligned with the laser sheet at a distance of 4 cm from the field of view. Typically, krill exited the pipe and swam across the tank, which increases the probability of collecting PIV data with krill swimming parallel to the laser sheet in the camera imaging region. The grouped krill data were collected in a tank that contained 100 swimming krill that segregated into coordinated groups of three to six krill.

Kinematics analysis

To connect krill behavior to the resulting flow field, a kinematic analysis was performed using the same digital images pairs used for the PIV analysis. The length of the krill (l) was measured in the images as the distance between the furthest extent of the antennae and tail. The swimming speed (U) of the krill was calculated by measuring the magnitude of the displacement of the krill over an image pair and dividing this magnitude by the time lapse between the images. To reduce random error of this measurement, swimming speed data were averaged over five image pairs (collected at 25 Hz) for each replicate. The Reynolds number (Re) is a non-dimensional parameter used to categorize the flow regime around swimming organisms that combines the swimming speed of the organism, the length of the organism and the kinematic viscosity of the fluid (ν):
formula
(1)
The Reynolds number is a key indicator of flow stability in shear flows, and its value usually defines the transition between laminar and turbulent conditions. Previous observations indicate that euphausiids swim in a flow regime of intermediate Reynolds number (Yen et al., 2003) where the flow is neither viscous-dominated nor fully turbulent.
The pleopod beat frequency (fbeat) was estimated from the image sequences (which were collected at a frequency of 25 Hz). The krill body angle (θswim) during swimming was calculated relative to horizontal as:
formula
(2)
where ΔX and ΔY refer to the horizontal distance and vertical distance, respectively, between the base of the tail and the center of the eye.

Coordinated groups of E. superba

The nearest neighbor distance (NND) and the nearest neighbor elevation (NNE) were measured between krill in the images of coordinated groups. NNE is represented as an angle, where 0 deg is directly behind of the reference krill, 90 deg is directly below the reference krill and –90 deg is directly above the reference krill. NND and NNE were measured from the center of the eye of the reference krill to the center of the eye of the neighbor krill. Because these quantities are measured in the PIV images, the values represent a two-dimensional projection of relative position of the organisms.

Flow field calculations

The displacement of particles between PIV image pairs was found via cross-correlation analysis. The pattern of particles within a 32×32 pixel region in the first image was compared with the pattern of particles in the corresponding region in the second image. The location of the peak value of the cross-correlation function relative to the center of the region corresponds to the displacement of the particles located in the region. The details of the validation and filtering algorithms used in the data analysis programs are explained in Catton et al. (Catton et al., 2007). The planar PIV system is used to collect velocity vectors in a plane defined by the X (horizontal) and Y (vertical) directions. The symbol u represents the X-direction velocity component and v represents the Y-direction velocity component. Vorticity (ω) describes the local rotation of the fluid, and the component calculated in the measurement plane is defined as:
formula
(3)
The spatial gradients of velocity in this equation are calculated via a central difference calculation of the measured velocity field.

The quantities extracted from the flow fields include maximum velocity, maximum vorticity, wake angle, vertical extent of the flow disturbance and horizontal extent of the flow disturbance. Normalized maximum velocity is also reported, where the maximum velocity is divided by the swimming speed of the krill. The wake angle is the orientation of the core of the flow disturbance relative to horizontal. The vertical and horizontal extents of the flow disturbance are defined as the respective distance from the krill eye location to the point where the velocity of the flow disturbance equaled the background velocity. The background velocity is defined as the mean velocity magnitude of the flow field prior to entry of the krill into the field of view. For several of the data sets, the horizontal and vertical extents of the flow disturbance continued beyond the measurement region, hence the value was not estimated. In these cases, a different value of N is noted to show the number of replicates used in the analysis. The horizontal extent and vertical extent of the flow disturbance were normalized by the length of the krill and presented with the absolute horizontal and vertical extents.

Statistical analysis

A two-tailed t-test with unequal variance was used to identify the variables that were statistically significantly different within the following hypotheses: (1) there is no difference in mean values of the kinematics (i.e. l, U, Re, fbeat and θswim) or flow field variables (i.e. maximum velocity, maximum vorticity, wake angle and extent of the flow perturbation) between species (E. pacifica versus E. superba); and (2) there is no difference in mean values of the kinematics or flow field variables between behaviors of E. superba (solitary versus coordinated group behavior).

A multivariate ANOVA (MANOVA) analysis was performed on the two hypotheses and the null hypothesis was rejected for the species comparison (Hypothesis 1). Because the null hypothesis was rejected, the variables identified in the t-test are not likely to be artifacts of multiple comparisons (Timm, 2002). The significance level was set at P<0.05 for each hypothesis.

For the first time, simultaneous measurements of krill swimming kinematics and high-resolution flow fields are reported for free-swimming krill of two species (E. superba and E. pacifica). Data also are reported for one species of krill (E. superba) performing two different behaviors: solitary swimming and coordinated group swimming. This section presents comparisons of swimming krill kinematics and characteristics of the flow field.

Krill kinematics

The krill kinematic parameters measured in this study (Table 1) were the length of each krill, swimming speed, Reynolds number, body angle and pleopod beat frequency. All of these quantities, except normalized swimming speed, were significantly different between species (E. pacifica versus E. superba). The length of E. pacifica individuals was approximately half the length of E. superba individuals, and E. pacifica swimming speed was approximately one-third the swimming speed of E. superba. As a result, the Reynolds number of the swimming E. pacifica was roughly 4.5 times smaller than that of E. superba. Therefore, E. pacifica swim in an effectively more viscous flow regime compared with the larger E. superba despite the temperature-induced difference in the viscosity of the ambient fluid. In addition, the body angle compared with horizontal is greater (i.e. steeper orientation) for E. pacifica. The pleopod beat frequency is also greater in E. pacifica than E. superba. The kinematics data match the observation during the experiments that the different krill species have different swimming behaviors. During the experiments, E. pacifica individuals circled the tank with their body positioned at a steep vertical angle, whereas E. superba individuals swam in a straight line out of the funnel until they reached the wall of the tank, where they dropped to the bottom of the tank.

Unlike the species comparison above, there was no effect of coordinated behavior on the length, swimming speed, Reynolds number or body orientation associated with the krill (Table 1). The krill length was measured on specimens of the same species and ages; hence, the similar values were expected. The mean values of swimming speed and Reynolds number were lower for coordinated krill compared with solitary krill, but the inter-individual variation was large enough that the mean values of the parameters were not significantly different. The coordinated group behavior was associated with decreased pleopod beat frequency. These results indicate that E. superba in a coordinated group swim with a lower pleopod beat frequency to generate the same swimming speed and Reynolds number as solitary E. superba.

The spatial arrangement of krill in the coordinated groups was quantified in terms of NND and NNE (Fig. 1). The closest neighbor to the reference krill was located at angles ranging from –105 deg (above the reference krill) to 90 deg (below the reference krill), with a maximum distance from the reference krill of less than 10 cm (i.e. less than two body lengths).

Euphausia pacifica flow field

The side-view flow field of E. pacifica can be characterized as a downward-directed jet with distinct regions of high velocity that result from each stroke of the pleopods (Fig. 2). Fluid is entrained from the front and sides of the krill and expelled to the rear and below the krill (Figs 2, 3). The mean maximum velocity in the flow fields is similar to the swimming speed of the krill (Tables 1, 2). The high velocity regions merge into a single region with a larger vertical extent (Fig. 2). The mean horizontal and vertical extents from replicate side-view flow fields of E. pacifica are in the range of one to two body lengths (Table 2). The transverse width of the flow disturbance for E. pacifica was roughly equal to the width of the krill, hence the flow perturbation created by the krill decreases rapidly in the transverse direction (Fig. 3). In this analysis, the vortex identification method of Jeong and Hussain (Jeong and Hussain, 1995) was employed to distinguish a region of large vorticity from a vortex. Regions of positive and negative vorticity align along the edges of the downward-directed jet, although no distinct vortices are present in the flow field (Fig. 5A).

Euphausia superba flow field

Sequences of four instantaneous side-view velocity fields generated by a solitary free-swimming E. superba are shown in Fig. 4. The E. superba flow field has multiple, non-uniform, separated regions of high velocity produced by the beating motion of the pleopods. The flow in these regions is generally directed downward and rearward. Fluid is entrained from several directions (i.e. above, behind and from the side of the krill) into the region surrounding the pleopods and into the resulting fluid disturbance. The flow field of the krill penetrates a small distance in the vertical direction and is not visible beyond a distance of a half of a body length below the specimen. The restricted vertical extent of the krill flow field is roughly a half of a body length (Table 2). The flow field persists in the horizontal direction at a mean value of nearly four body lengths (Fig. 4, Table 2). Dorsal views of E. superba (not shown) indicate that the transverse width of the flow disturbance is again limited to approximately the width of the krill. The high velocity regions in the flow disturbance are associated with high values of vorticity (Fig. 5B). The mean maximum velocity is similar to the swimming speed of the krill (Tables 1, 2) and the mean maximum value of vorticity varies substantially among individuals (Table 2). Opposing positive and negative vorticity regions align horizontally in the krill flow field, and again no distinct vortices are present (Fig. 5B).

Flow field comparisons

Qualitatively, the flow fields of E. pacifica and E. superba are different in terms of the uniformity of the velocity vectors, the shape of the flow field and the direction of fluid entrainment. The flow field of E. superba is characterized by persistent high-velocity regions with less uniform velocity vectors that merge at a later time than the high velocity regions in the E. pacifica flow field (Figs 2, 4). Quantitatively, the significant differences in the flow fields are the maximum velocity, normalized vertical extent, and dimensional and normalized horizontal extent of the flow field (Table 2). The maximum velocity is significantly smaller in the E. pacifica flow field than the E. superba flow field (Table 2). The horizontal extent of the E. superba flow field is approximately four times greater than the horizontal extent of the E. pacifica flow field. Given the larger size of E. superba, the increase in absolute horizontal extent is expected. The mean horizontal extent normalized by the krill length is two times larger for E. superba than E. pacifica after accounting for the differences in krill size. The decreased body angle and increased Reynolds number explain why the flow disturbance pattern of E. superba is more directed in the horizontal direction and has more irregular and persistent flow features (Table 1). Similarly, the wake angle was smaller for E. superba, but the results were not significantly different because of the small sample size (Table 2). The dimensional vertical extents of the flow fields for both species were similar in length, whereas the normalized vertical extent was significantly smaller for E. superba. The transverse width of the flow disturbance for both species of krill was roughly equal to the width of the krill, hence the flow perturbation created by the krill decreased rapidly in the transverse direction (Fig. 3). The maximum value of vorticity in the flow disturbance of these two species of krill was significantly different (Table 2) and the shape of the vorticity fields differed between the two species (Fig. 5). The regions of large vorticity (positive and negative) were aligned with the horizontal direction in the flow disturbance of E. superba (Fig. 5B) whereas the regions of large vorticity were aligned vertically for E. pacifica (Fig. 5A).

A series of instantaneous velocity fields generated by E. superba swimming in a small coordinated group is shown in Fig. 6. The flow disturbance produced by E. superba in a coordinated group (Fig. 6) is visually similar in structure to the flow disturbance of solitary E. superba (Fig. 4), with several regions of high-velocity fluid located below and behind the pleopods. The maximum velocity, maximum vorticity and wake angle were not significantly different compared with the solitary E. superba flow fields (Table 2). Additionally, the horizontal and vertical extents of the flow fields (for the reported data) were not significantly different between solitary and coordinated E. superba (Table 2). However, in most of the coordinated group data sets, the spatial extent of the flow fields could not be accurately quantified because of the finite size of the observation region and the presence of neighboring krill. Generally, the horizontal and vertical extents of the flow perturbation of aggregating E. superba appeared larger than those of the solitary E. superba. The background velocity was greater for the coordinated flow fields because of the presence of over 100 krill in the tank. The vorticity field is more variable for the group E. superba compared with the solitary E. superba because of the presence of neighboring krill (Fig. 5). Nevertheless, the vorticity field appears to continue to distinguish the edge region of the propulsion jet for each krill. Because swimming speed and Reynolds number were not significantly different between the solitary individuals and group members, few differences in the flow field structure are expected. However, it appears that the flow disturbances of individual krill in the group interact to alter the flow structure. Thus, the flow field of a group of E. superba is not perfectly equivalent to a flow field of superimposed solitary flow fields.

Time series analysis

Time series of velocity were extracted from the flow fields in a moving frame of reference that represented the position of a krill swimming at a set distance and angle behind the krill generating the flow field. The time series were extracted at a set of points defined by distances of 1.5, two, three and four body lengths behind the swimming krill and at NNE of 0, 15, 30, 45 and 60 deg from horizontal (Fig. 7). This set of points was translated in space with the reference krill such that the points always remained at the original distance and angle relative to the reference krill position. The time records, therefore, correspond to the flow perturbation, which is potentially available to a neighbor as a hydrodynamic cue, at the particular point of extraction. The velocity time record is shown in Fig. 8. The time records of vorticity and deformation rate are not shown because they qualitatively follow the same trends observed in the velocity time record. To highlight the distances and angles that have a significant hydrodynamic perturbation, a background noise level is identified with a pink background in Fig. 8. Background noise levels for the velocity series were identified by measuring the mean velocity magnitude (V) of the flow field prior to the entry of the krill (V<0.1 cm s–1 for solitary E. pacifica, V<0.2 cm s–1 for solitary E. superba and V<0.6 cm s–1 for group E. superba). These velocity thresholds are above the known velocity sensitivity (70 μm s–1) of Euphausia flow-sensing structures (Wiese and Marschall, 1990). The significant hydrodynamic perturbation was interpreted as the velocity measurements that exceeded the background noise level.

For E. pacifica (Fig. 8), the most robust velocity perturbation was located at an NNE of 60 deg (which agrees with the wake angle reported in Table 2) and a distance of less than two body lengths from the individual. The flow perturbation also was evident at an NNE of 45 deg at distances less than two body lengths. Considering the replicate E. pacifica data, the mean NNE of the largest velocity perturbation was 66 deg (range=45–90 deg) and the perturbation was greater than the background fluid motion at a distance of less than 2.4 body lengths (N=5).

Strong velocity perturbations in the flow field of solitary E.superba were located at an NNE of 15 deg and a distance of at least four body lengths from the reference krill (Fig. 8). The results were consistent among the limited flow field replicates (N=3). In order for grouped E. superba to use a hydromechanical perturbation for cues, a preferential position is 15 deg and roughly less than four body lengths (note that we do not have data beyond this distance to evaluate the cue strength at greater distances). Power spectra calculated from the time series of velocity magnitude revealed a peak of energy at approximately 3 Hz (for positions less than four body lengths), which agrees with the pleopod beat frequency for E. superba (Table 1). In contrast, power spectra calculated based on the time records collected in the flow field of E. pacifica did not show a peak of energy at a specific frequency.

In the velocity field around the coordinated group, it is difficult to separate the background flow field produced by the general movement of the group from the cue produced by one krill within the group (Fig. 8). As a consequence, the time records extracted for the group E. superba reveal a greater flow perturbation compared with the solitary specimens (e.g. Fig. 8D). The peak perturbation appeared at an NNE of 0 deg and all NNEs shown had significant velocity perturbations (Fig. 8). From the replicate flow fields (N=5), the NNE of the peak perturbation was 30±15 deg and the perturbation was greater than the background fluid motion at a distance of 3.4±0.9 body lengths.

Swimming kinematics

Accurate assessments of krill hydrodynamic cues require that the behavior of the krill in the laboratory matches the behavior of krill in situ. In this study, we collected data on free-swimming krill performing pleopod swimming at mean swimming speeds of approximately 2.5 cm s–1 (E. pacifica) and 7 cm s–1 (E. superba), speeds that are similar to those of solitary E. pacifica and horizontal schooling E. superba measured in situ at 1.8 cm s–1 (de Robertis et al., 2003) and between 3 and 15 cm s–1 (Hamner, 1984; Zhou and Dorland, 2004), respectively. In this study, E. pacifica swam at lower swimming speeds at a steeper body angle than solitary E. superba. In the field, E. pacifica swim at oblique trajectories of less than 60 deg (de Robertis et al., 2003) whereas the body angle of E. superba was typically horizontal (Hamner, 1984). In conclusion, the behavior of krill in this study was similar to field observations and is suitable for further discussion.

The swimming behavior of E. pacifica and E. superba has been analyzed in laboratory studies, but none of these studies directly compared swimming behavior with the resulting flow fields for these two species. In previous laboratory studies, the mean swimming speeds of E. superba and E. pacifica for pleopod swimming were 6 cm s–1 (Kils, 1979a; Kils, 1979b) and less than 2 cm s–1 (Miyashita et al., 1996), respectively. Hence, the swimming speeds of E. superba and E. pacifica in this study were similar to findings in other laboratory studies. However, some studies found steeper body angles (Endo, 1993; Kils, 1982) than our study because these studies observed krill during hovering rather than during pleopod swimming. The swimming velocity threshold of hovering versus pleopod swimming appears to be arbitrary and the swimming speeds, body angle and behavior (i.e. hovering or pleopod swimming) of krill in situ varies throughout the day (de Robertis et al., 2003; Zhou and Dorland, 2004). Further, Miyashita et al. found that the swimming speed of E. pacifica is inversely related to swimming angle, such that hovering krill will have larger swimming angles than faster swimming krill (Miyashita et al., 1996). Previous studies have shown that pleopod beat frequency increases with increased swimming speed (Kils, 1982; Swadling et al., 2005) and decreases for larger body lengths (Kils, 1982; Johnson and Tarling, 2008). Comparisons between the species in this study are consistent with these previous observations. In summary, E. pacifica in this study had steeper body orientations, slower swimming speeds and faster pleopod beat frequencies than E. superba, as expected from previous krill swimming kinematics studies.

Flow fields

The Reynolds number around a free-swimming solitary E. superba (Re≈2300) is more than four times larger than the Reynolds number for a solitary E. pacifica (Re≈500). These findings are supported by previous studies that estimated the Reynolds number of E. superba as ranging from 500 to 3000 (Swadling et al., 2005) and that of E. pacifica as 175 (Yen et al., 2003). It should be noted that the mean Reynolds number of E. pacifica in our study was 265 when calculated using the definition in Yen et al. (Yen et al., 2003) (i.e. Reynolds number defined by maximum velocity in the wake and the width of the wake). The Reynolds number is generally interpreted as the non-dimensional ratio of inertial forces to viscous forces, and the value compared with unity indicates the relative importance of these effects. Marine organisms swim at a range of flow regimes from viscosity-dominated flow regimes (Re≤1) and intermediate flow regimes with both viscous and inertial forces acting on the organism, to inertia-dominated flow regimes where viscous drag is restricted to a thin boundary layer adjacent to the organism body (Re⪢1000). The Reynolds numbers in our study are representative of an intermediate flow regime.

Past studies on euphausiid flow fields are limited to studies of tethered specimens of Meganyctiphanes norvegica (Kils, 1982; Patria and Wiese, 2004), E. pacifica (Yen et al., 2003) and E. superba (Ebina and Miki, 1996). Tethered specimens generate flow fields with higher velocities and more rotation in the flow (Catton et al., 2007) because the tether imparts an unbalanced force on the fluid. Therefore, studies on tethered specimens do not provide an accurate representation of the induced flow fields. The free-swimming E. pacifica flow field more closely resembled the findings of tethered studies, with a distinct downward-directed jet (Kils, 1982; Yen et al., 2003; Patria and Wiese, 2004), whereas the non-uniformity and variation in direction produced by each stroke of the pleopods was more apparent in the E. superba flow fields. Unlike the study by Patria and Wiese (Patria and Wiese, 2004), the larger species (E. superba) did not produce vortex rings from the side view, which suggests that the observed vortex rings were an artifact of the aquarium or tethering in their study. Vortex rings were also not produced in the flow field of a model lobster (235 mm body length) performing pleopod swimming (Lim and DeMont, 2009), thus suggesting that vortex rings may not form, even at larger Reynolds numbers. Patria and Wiese identified the vortex as a potential benefit to propulsion because well positioned krill could take advantage of regions of flow disturbances that have an upward and forward moving component (Patria and Wiese, 2004). Because coherent rings and upward velocity components do not appear in the flow disturbances in our study, it is unlikely that free-swimming krill gain any propulsive advantage from such rings.

Hydrodynamic cues

Hydrodynamic cues are suspected to be important for individuals to maintain position within schools of krill (Hamner, 1984; Wiese and Ebina, 1995). To sense hydromechanical cues, krill have multiple antennules that are oriented to sense flow in the vertical and horizontal directions. The antennules are lined with hair-type sensilla 50 μm in length that sense high-frequency (>40 Hz) flow perturbations (Patria and Wiese, 2004). These small sensilla may act similarly to copepod setae in response to hydrodynamic cues. Velocity gradients have been shown to provide a hydrodynamic sensory cue to copepods via the deflection of setae (Fields and Yen, 1997; Kiørboe et al., 1999; Fields et al., 2002). In addition to the sensilla, the proprioreceptor at the base of the antennule has been hypothesized to be sensitive to fluid perturbations at 5–40 Hz, which roughly matches the frequency of flow perturbations generated by krill (Patria and Wiese, 2004). The proprioreceptor is activated by water motion along the length of the antennae (half of the body length of a krill) that is sufficient to produce movement of the hinge. For the proprioreceptor, regions of high fluid velocity act as the hydrodynamic cue because a uniform velocity generates deflection in the antennae. Wiese and Marschall (Wiese and Marschall, 1990) and Wiese (Wiese, 1996) report that the antennular flow sensors of E. superba are highly sensitive to fluid velocities of 70 μm s–1 at frequencies between 0.5 and 50 Hz. Patria and Wiese (Patria and Wiese, 2004) state that the receptor system of North Atlantic krill (Meganyctiphanes norvegica) is tuned to velocity at low frequency and to acceleration at higher frequency and that the threshold sensitivity for E. superba is 0.15 mm s–1. Krill mechanosensors have sensitivities similar to other arthropods [e.g. crickets sense 30 μm s–1 (Shimozawa et al., 2003); crayfish sense 20 μm s–1 (Mellon and Christison-Lagay, 2008); and copepods sense 10 nm displacements (Yen et al., 1992)]. Mechanoreceptor sensitivity often shows five orders of magnitude of sensitivity (Humphrey and Barth, 2008), hence the 70 μm s–1 detection threshold permits integration of flow information up to the order of magnitude of cm s–1, similar to the swimming speeds of krill. These studies suggest that fluid velocity rather than another fluid quantity (e.g. pressure) is the hydrodynamic cue used by krill and that the extent of the hydrodynamic cue can be estimated by the high-velocity region in the krill flow fields.

The sensory cue between krill has been hypothesized to be the high-velocity core of the downward-directed flow disturbance, where the largest velocities occur (Yen et al., 2003), or the edges of the flow disturbance, where the krill can avoid areas of high velocity (Wiese, 1996; Wiese and Ebina, 1995). In this study, the velocity cue of E. superba flow disturbances was persistent at a longer distance (four versus two body lengths) and at a shallower angle (15 versus 66 deg from horizontal) than E. pacifica flow disturbances. Because E. pacifica swim more vertically, they generate a downward-directed flow disturbance. Thus, an individual E. pacifica krill would need to be located under a neighboring krill to sense the hydrodynamic cue and would potentially experience a significant downward force from the flow field. In contrast, E. superba individuals within schools could be spaced farther apart and sample the more horizontally aligned neighbor-generated flow field without experiencing the negative consequences of a high downward-directed velocity. Individuals would experience less of the negative consequences of schooling, such as decreased oxygen supply and increased parasitism, as the individuals have less direct contact with each other. As one caveat, because these studies were performed under still tank conditions, we expect that the cues identified in this study are stronger than the cues present in the natural environment. Further, the data collected for grouped E. superba indicate that neighboring flow disturbances interact to produce flow perturbations above background levels at all angles and distances within the groups. Hence, a hydrodynamic cue that is distinct from the background water motion may be only available at short distances in the ocean.

Comparison with schooling behavior of E. superba

To assess the ability of krill to use hydrodynamic cues, it is important to compare the flow field cues with actual krill arrangements within a school. In situ measurements of E. pacifica swarms found that vertical migration was not synchronous (de Robertis et al., 2003) and only one study found a portion of an E. pacifica surface swarm with krill in a uniform orientation (Hanamura et al., 1984). It is assumed, therefore, that E. pacifica is not a obligate schooling species. Although data are not available on the internal arrangement of aggregations for this species, E. pacifica has a more vertically oriented flow field whereas E. superba has a horizontally oriented flow field that could contribute to the long layering noted in schools of the polar species versus the vertical migratory behavior of the temperate species. Euphausia superba have been reported to school in aquarium during the day (O'Brien, 1989; Strand and Hamner, 1990; Kawaguchi et al., 2010) and in the natural environment during the day and night (Hamner et al., 1983). The length of a school is reported to be up to 100 m (and larger for superswarms) (Tarling et al., 2009), and the schools tend to be narrow in one dimension such that an individual is only a few meters from clear water (Hamner et al., 1983; Hamner, 1984). Observations on the density of krill within a school range from 100 to 100,000 krill m–3 (Mauchline, 1980; Hamner, 1984).

Fig. 1 indicates that the NNDs are in the range of less than two body lengths for these data. Consistently, the NNDs from previous laboratory studies of schooling krill range from 0.4 to three body lengths (O'Brien, 1989; Kawaguchi et al., 2010). Because the hydrodynamic perturbation of E. superba is present at a distance of less than four body lengths, the potential cue is within the range of schooling individuals. Individuals positioned to the side of juvenile E. superba have a smaller NND (less than one body length) than individuals in front of or behind the krill (one to two body lengths) (O'Brien, 1989). O'Brien (O'Brien, 1989) reports that E. superba often swim at the same elevation as their neighbors, i.e. at an NNE of 0 deg. The NNE data in Fig. 1 indicate the location of the nearest neighbor is rarely in the –15 to 15 deg range. The location of the largest velocity disturbance was at an angle of 15 deg for the solitary E. superba data, and the perturbation cue for the coordinated group was also substantial at 0 deg (Fig. 8). Based on these data, E. superba appear to avoid the region of the strongest propulsion-generated flow because the krill are in positions associated with minimal flow disturbance. The limited spatial data of the coordinated groups in this study indicate that krill can be located directly above and below the krill (Fig. 1) where the hydrodynamic perturbation is minimal (Fig. 8). The internal arrangement of krill within a school will dictate the structure of the krill flow field and the potential for hydromechanical cues. However, the three-dimensional arrangement of adult krill within a school has not been reported and the data collected in this study contradict some findings from previous studies on krill arrangements. Thus, further investigation of the three-dimensional structure is needed to provide additional insight into the flow fields within schools.

Euphausia superba are known for forming schools that are large, often in layers of matched individuals, and that react quickly, synchronously and in a coordinated fashion. Our data show that hydrodynamic features present in the propulsive flow field can provide information about size, speed and direction. The small coordinated groups in this study consisted of approximately five krill, whereas natural krill schools consist of hundreds of individuals, usually with densities of tens of thousands of individuals per cubic meter. In this study, the flow fields of several krill interacted to make a stronger and more persistent hydrodynamic cue. The hydrodynamic disturbances of larger krill aggregations may have even more complicated interactions and more persistent features may be present. Current data on in situ krill aggregations are sparse in terms of data on inter-individual distances between krill and the orientation of krill to each other within schools. Additional studies are necessary to more completely address whether the hydrodynamic cues generated by krill are used during schooling. Simultaneous measurements of schooling krill behavior and the flow fields generated by the krill of different sizes, sex or stage in solitary swimming versus in schools of different density would allow us to test hypotheses on whether krill use this kind of information. Further, the flow field data in this study facilitate a rough estimate on the extent of chemical cues because chemical cues originating from an individual krill are transported via advection in the flow field. Modification of flow structure or behavior in the presence of chemical and fluid predator or prey cues under different optical conditions would allow us to evaluate the relative importance of hydrodynamic cues versus chemical or optical cues for intraspecies communication.

Financial support was provided by a National Science Foundation (NSF) grant CTS-0625898 awarded to J.Y., and a NSF-Integrative Graduate Education and Research Traineeship (IGERT) fellowship and international travel grant to K.B.C.

We wish to thank Dr William Peterson and his colleagues at the National Oceanic and Atmospheric Administration (NOAA) Fisheries Department at the Hatfield Marine Station in Newport, OR, USA, for the collection and maintenance of the E. pacifica used in these experiments. In addition, we thank Rob King of the Australian Antarctic Division in Kingston, Tasmania, for help in the construction of the tanks with optical access for the E. superba experiments.

Abrahamsen
M. B.
,
Browman
H. I.
,
Fields
D. M.
,
Skiftesvik
A. B.
(
2010
).
The three-dimensional prey field of the northern krill, Meganyctiphanes norvegica, and the escape responses of their copepod prey
.
Mar. Biol.
157
,
1251
-
1258
.
Baird
T. A.
,
Ryer
C. H.
,
Olla
B. L.
(
1991
).
Social enhancement of foraging on an ephemeral food source in juvenile walleye pollock, Theragra chalcogramma
.
Environ. Biol. Fish.
31
,
307
-
311
.
Burgess
J. W.
,
Shaw
E.
(
1979
).
Development and ecology of fish schooling
.
Oceanus
22
,
11
-
17
.
Catton
K. B.
,
Webster
D. R.
,
Brown
J.
,
Yen
J.
(
2007
).
Quantitative analysis of tethered and free-swimming copepodid flow fields
.
J. Exp. Biol.
210
,
299
-
310
.
de Robertis
A.
,
Schell
C.
,
Jaffe
J. S.
(
2003
).
Acoustic observations of the swimming behavior of the euphausiid Euphausia pacifica Hansen
.
ICES J. Mar. Sci.
60
,
885
-
898
.
Ebina
Y.
,
Miki
T.
(
1996
).
Range and biological significance of characteristic water currents produced by the shrimps Euphausia superba and Metapenaeus intermedius
.
Zoology
99
,
163
-
174
.
Endo
Y.
(
1981
).
Ecological studies on the euphausiids occurring on the Sanriku waters with special reference to their life history and aggregated distribution
.
PhD thesis
,
Tokuhu University
,
Sendhai, Japan
.
Endo
Y.
(
1993
).
Orientation of Antarctic krill in an aquarium
.
Nippon Suisan Gakkaishi
59
,
465
-
468
.
Fields
D. M.
,
Yen
J.
(
1997
).
The escape behavior of marine copepods in response to a quantifiable fluid mechanical disturbance
.
J. Plankton Res.
19
,
1289
-
1304
.
Fields
D. M.
,
Shaeffer
D. S.
,
Weissburg
M. J.
(
2002
).
Mechanical and neural responses from the mechanosensory hairs on the antennule of Gaussia princeps
.
Mar. Ecol. Prog. Ser.
227
,
173
-
186
.
Foster
E. G.
,
Ritz
D. A.
,
Osborn
J. E.
,
Swadling
K. M.
(
2001
).
Schooling affects the feeding success of Australian salmon (Arripis trutta) when preying on mysid swarms (Paramesopodopsis rufa)
.
J. Exp. Mar. Biol. Ecol.
261
,
93
-
106
.
Gómez-Gutiérrez
J.
,
Peterson
W. T.
,
de Robertis
A.
,
Brodeur
R. D.
(
2003
).
Mass mortality of krill caused by parasitoid ciliates
.
Science
301
,
339
.
Hamner
W. M.
(
1984
).
Aspects of schooling in Euphausia superba
.
J. Crust. Biol.
4
,
67
-
74
.
Hamner
W. M.
,
Hamner
P. P.
(
2000
).
Behavior of Antarctic krill (Euphausia superba): schooling foraging, and antipredatory behavior
.
Can. J. Fish. Aquat. Sci.
57
,
192
-
202
.
Hamner
W. M.
,
Parrish
J. K.
(
1997
).
Is the sum of the parts equal to the whole: the conflict between individuality and group membership
. In
Animal Groups in Three Dimensions: How Species Aggregate
(ed.
Parrish
J. K.
,
Hamner
W. M.
), pp.
165
-
173
.
Cambridge
:
Cambridge University Press
.
Hamner
W. M.
,
Hamner
P. P.
,
Strand
S. W.
,
Gilmer
R. W.
(
1983
).
Behavior of Antarctic krill, Euphausia superba: chemoreception, feeding, schooling, and molting
.
Science
220
,
433
-
435
.
Hanamura
Y.
,
Endo
Y.
,
Taniguchi
A.
(
1984
).
Underwater observation on the surface swarm of a euphausiid, Euphausia pacifica in Sendai Bay, Northeastern Japan
.
La Mer
22
,
63
-
68
.
Humphrey
J. A. C.
,
Barth
F. G.
(
2008
).
Medium flow sensing hairs: biomechanics and models
. In
Advances in Insect Physiology
, Vol.
34
,
Insect Mechanics and Control
(ed.
Casas
J.
,
Simpson
S. J.
), pp.
1
-
80
.
Burlington, MA
:
Academic Press
.
Jeong
J.
,
Hussain
F.
(
1995
).
On the identification of a vortex
.
J. Fluid Mech.
285
,
69
-
94
.
Johnson
M. L.
,
Tarling
G. A.
(
2008
).
Influence of individual state on swimming capacity and behaviour of Antarctic krill Euphausia superba
.
Mar. Ecol. Prog. Ser.
366
,
99
-
110
.
Kawaguchi
S.
,
King
R.
,
Meijers
R.
,
Osborn
J. E.
,
Swadling
K. M.
,
Ritz
D. A.
,
Nicol
S.
(
2010
).
An experimental aquarium for observing the schooling behaviour of Antarctic krill (Euphausia superba)
.
Deep Sea Res. Part 2
57
,
683
-
692
.
Kils
U.
(
1979a
).
Performance of Antarctic krill Euphausia superba, at different levels of oxygen saturation
.
Meeresforschung
27
,
35
-
47
.
Kils
U.
(
1979b
).
Swimming speed and escape capacity of Antarctic krill, Euphausia superba
.
Meeresforschung
27
,
264
-
266
.
Kils
U.
(
1982
).
The swimming behaviour, swimming performance and energy balance of Antarctic krill, Euphausia superba
.
BIOMASS Sci. Ser.
3
,
1
-
121
.
Kils
U.
(
1983
).
Swimming and feeding of Antarctic krill, Euphausia superba – some outstanding energetics and dynamics – some unique morphological details
.
Ber. Polarforschung
4
,
130
-
155
.
Kiørboe
T.
,
Saiz
E.
,
Visser
A.
(
1999
).
Hydrodynamic signal perception in the copepod Acartia tonsa
.
Mar. Ecol. Prog. Ser.
179
,
97
-
111
.
Lim
J. L.
,
DeMont
M. E.
(
2009
).
Kinematics, hydrodynamics and force production of pleopods suggest jet-assisted walking in the American lobster (Homarus americanus)
.
J. Exp. Biol.
212
,
2731
-
2745
.
Marr
J. S. W.
(
1962
).
The natural history and geography of the Antarctic krill (Euphausia superba Dana)
.
Discov. Rep.
32
,
33
-
464
.
Mauchline
J.
(
1980
).
Studies on patches of krill, Euphausia superba Dana
.
Biomass Handbook
6
,
1
-
35
.
Mellon
D.
,
Christison-Lagay
K.
(
2008
).
A mechanism for neuronal coincidence revealed in the crayfish antennule
.
Proc. Natl. Acad. Sci. USA
105
,
14626
-
14631
.
Miyashita
K.
,
Aoki
I.
,
Inagaki
T.
(
1996
).
Swimming behaviour and target strength of isada krill (Euphausia pacifica)
.
ICES J. Mar. Sci.
53
,
303
-
308
.
Nicol
S.
(
1984
).
Population structure of daytime surface swarms of the euphausiid Meganyctiphanes norvegica in the Bay of Fundy
.
Mar. Ecol. Prog. Ser.
18
,
241
-
251
.
O’Brien
D. P.
(
1989
).
Analysis of the internal arrangement of individuals within crustacean aggregations (Euphausiacea mysidacea)
.
J. Exp. Mar. Biol. Ecol.
128
,
1
-
30
.
Patria
M. P.
,
Wiese
K.
(
2004
).
Swimming in formation in krill (Euphausiacea), a hypothesis: dynamics of the flow field, properties of antennular sensor systems and a sensory-motor link
.
J. Plankton Res.
26
,
1315
-
1325
.
Raffel
M.
,
Willert
C.
,
Kompenhans
J.
(
1998
).
Particle Image Velocimetry – A Practical Guide
.
New York
:
Springer-Verlag
.
Ritz
D. A.
(
2000
).
Is social aggregation in aquatic crustaceans a strategy to conserve energy?
Can. J. Fish. Aquat. Sci.
57
,
59
-
67
.
Ritz
D. A.
,
Metillo
E. B.
(
1998
).
Costs and benefits of swarming behaviour in mysids: does orientation and position in the swarm matter?
J. Mar. Biol. Assoc. UK
78
,
1011
-
1014
.
Shimozawa
T.
,
Murakami
J.
,
Kumagai
T.
(
2003
).
Cricket wind receptors: thermal noise for the highest sensitivity known
. In
Sensors and Sensing in Biology and Engineering
(ed.
Barth
F. G.
,
Humphrey
J. A. C.
,
Secomb
T.
), pp.
145
-
157
.
Berlin
:
Springer-Verlag
.
Strand
S. W.
,
Hamner
W. M.
(
1990
).
Schooling behavior of Antarctic krill (Euphausia superba) in laboratory aquariums: reactions to chemical and visual stimuli
.
Mar. Biol.
106
,
355
-
359
.
Swadling
K. M.
,
Ritz
D. A.
,
Nicol
S.
,
Osborn
J. E.
,
Gurney
L. J.
(
2005
).
Respiration rate and cost of swimming for Antarctic krill, Euphausia superba, in large groups in the laboratory
.
Mar. Biol.
146
,
1169
-
1175
.
Tarling
G. A.
,
Klevjer
T.
,
Fielding
S.
,
Watkins
J.
,
Atkinson
A.
,
Murphy
E.
,
Korb
R.
,
Whitehouse
M.
,
Leaper
R.
(
2009
).
Variability and predictability of Antarctic krill swarm structure
.
Deep Sea Res. Part 1
56
,
1994
-
2012
.
Timm
N. H.
(
2002
).
Applied Multivariate Analysis
.
New York
:
Springer-Verlag
.
Watkins
J. L.
,
Murray
A. W. A.
(
1998
).
Layers of Antarctic krill, Euphausia superba: are they just long krill swarms?
Mar. Biol.
131
,
237
-
247
.
Wiese
K.
(
1996
).
Sensory capacities of euphausiids in the context of schooling
.
Mar. Freshw. Behav. Physiol.
28
,
183
-
194
.
Wiese
K.
,
Ebina
Y.
(
1995
).
The propulsion jet of Euphausia superba (Antarctic krill) as a potential communication cue among conspecifics
.
J. Mar. Biol. Assoc. UK
75
,
43
-
54
.
Wiese
K.
,
Marschall
H. P.
(
1990
).
Sensitivity to vibration and turbulence of water in context with schooling in Antarctic krill Euphausia superba
. In
Frontiers in Crustacean Neurobiology
(ed.
Wiese
K.
,
Krenz
W. D.
,
Tautz
J.
,
Reichert
H.
,
Mulloney
B.
), pp.
121
-
130
.
Basel
:
Birkhäuser Verlag
.
Yen
J.
,
Lenz
P. H.
,
Gassie
D. V.
,
Hartline
D. K.
(
1992
).
Mechanoreception in marine copepods: electrophysiological studies on the first antennae
.
J. Plankton Res.
14
,
495
-
512
.
Yen
J.
,
Brown
J.
,
Webster
D. R.
(
2003
).
Analysis of the flow field of the krill, Euphausia pacifica
.
Mar. Freshw. Behav. Physiol.
36
,
307
-
319
.
Zhou
M.
,
Dorland
R. D.
(
2004
).
Aggregation and vertical migration behavior of Euphausia superba
.
Deep Sea Res. Part II
51
,
2119
-
2137
.
Zhou
M.
,
Zhu
Y. W.
,
Tande
K. S.
(
2005
).
Circulation and behavior of euphasiids in two Norwegian sub-Arctic fjords
.
Mar. Ecol. Prog. Ser.
300
,
159
-
178
.