Journal of Comparative Physiology A

, Volume 199, Issue 6, pp 421–440

Hydrodynamic perception in true seals (Phocidae) and eared seals (Otariidae)

Authors

    • Institute for BiosciencesChair of Sensory and Cognitive Ecology, Rostock University
  • Sven Wieskotten
    • Institute for BiosciencesChair of Sensory and Cognitive Ecology, Rostock University
  • Christopher Marshall
    • Department of Marine BiologyTexas A&M University
  • Guido Dehnhardt
    • Institute for BiosciencesChair of Sensory and Cognitive Ecology, Rostock University
Review

DOI: 10.1007/s00359-012-0778-2

Cite this article as:
Hanke, W., Wieskotten, S., Marshall, C. et al. J Comp Physiol A (2013) 199: 421. doi:10.1007/s00359-012-0778-2

Abstract

Pinnipeds, that is true seals (Phocidae), eared seals (Otariidae), and walruses (Odobenidae), possess highly developed vibrissal systems for mechanoreception. They can use their vibrissae to detect and discriminate objects by direct touch. At least in Phocidae and Otariidae, the vibrissae can also be used to detect and analyse water movements. Here, we review what is known about this ability, known as hydrodynamic perception, in pinnipeds. Hydrodynamic perception in pinnipeds developed convergently to the hydrodynamic perception with the lateral line system in fish and the sensory hairs in crustaceans. So far two species of pinnipeds, the harbour seal (Phoca vitulina) representing the Phocidae and the California sea lion (Zalophus californianus) representing the Otariidae, have been studied for their ability to detect local water movements (dipole stimuli) and to follow hydrodynamic trails, that is the water movements left behind by objects that have passed by at an earlier point in time. Both species are highly sensitive to dipole stimuli and can follow hydrodynamic trails accurately. In the individuals tested, California sea lions were clearly more sensitive to dipole stimuli than harbour seals, and harbour seals showed a superior trail following ability as compared to California sea lions. Harbour seals have also been shown to derive additional information from hydrodynamic trails, such as motion direction, size and shape of the object that caused the trail (California sea lions have not yet been tested). The peculiar undulated shape of the harbour seals’ vibrissae appears to play a crucial role in trail following, as it suppresses self-generated noise while the animal is swimming.

Keywords

Hydrodynamic perceptionPinnipedsSensory biologyMarine mammalsVibrissae

Introduction

Pinnipeds are aquatic or to some degree semiaquatic mammals from three different lineages, the true seals (Phocidae), the eared seals (Otariidae), and the walruses (Odobenidae). They belong to the order carnivora and feed on a rich diversity of aquatic animals, some species being more specialized feeders than others. For example, within the family of Phocidae (adult) harbour seals (Phoca vitulina) are food generalists with a preference for small to medium-sized fish, both pelagic and benthic (e.g. Bowen et al. 2002; Sharples et al. 2009), while bearded seals (Erignathus barbatus) feed on invertebrates and vertebrates with a strong preference for benthic prey (Hjelset et al. 1999; Dehn et al. 2007); crabeater seals (Lobodon carcinophagus) feed mainly, but not exclusively on krill (Green and Williams 1986; Hückstädt et al. 2012), and leopard seals (Hydrurga leptonyx) include warm-blooded prey such as penguins and young seals in their diet (Siniff et al. 1979; Hall-Aspland and Rogers 2007; Casaux et al. 2009). Eared seals have a broad prey spectrum with regard to prey size and species composition as well. Pelagic and benthic fish and squids from relatively shallow depths of usually up to 100 m, but also well above 200 m, constitute a high percentage of the diet of California sea lions (Zalophus californianus) (Heath and Perrin 2009). At least 72 prey types have been identified in the diet of Steller sea lions Eumetopias jubatus (McKenzie and Wynne 2008). Fur seals have been reported to have a tendency to feed in deep water beyond the continental shelf on small squid and fish (Gentry 2009). However, e.g. the South African fur seal (Artocephalus pusillus) feeds opportunistically mainly on teleosts, on rare occasions more than 1 m in length (Mecenero et al. 2006), as well as on seabirds (Mecenero et al. 2005). The only member of the family of walruses, the walrus (Odobenus rosmarus), feeds on all kinds of benthic invertebrates, predominantly bivalves (Fay 1982).

Hydrodynamic perception is the sensing of water movements (e.g. Dehnhardt et al. 2004). The respective water movements are called hydrodynamic stimuli (for review see Bleckmann 1994). Usually, and also in this review, the water movements commonly known as sound are excluded from this definition, that is waves of compression and decompression that propagate through the water according to the wave equation. Surface waves (waves at the water–air interface) are often included in the definition of hydrodynamic stimuli. They are an important means of prey localization for many surface-feeding animals including insects (Wiese 1974), spiders (Bleckmann et al. 1994), amphibians (Görner 1973) and fish (Bleckmann et al. 1981), but are not covered here, as there is no indication that they might be used by pinnipeds. Hydrodynamic stimuli and hydrodynamic perception as the terms are used here refer to mid-water fluid movements other than sound.

Many of the hydrodynamic stimuli that are most relevant to aquatic predators are generated by moving animals such as conspecifics or prey (Bleckmann 1993; Conley and Coombs 1998). Hydrodynamic stimuli are for physical reasons short-ranged by nature (Bleckmann 1994). If the stimulus generator, such as a prey item, moves away from the site of stimulus generation while the stimulus remains, forming a so-called hydrodynamic trail, hydrodynamic stimuli can also be used for medium to long range object detection (Hanke et al. 2000; Dehnhardt et al. 2001).

True seals (Phocidae) and eared seals (Otariidae) can sense hydrodynamic stimuli using their whiskers, or vibrissae (Dehnhardt et al. 1998a). Vibrissae are tactile hairs that are found in almost all mammals (Ling 1977; Ahl 1986). In true seals, they are found on the snout (mystacial vibrissae), above the nares (rhinal vibrissae), and above the eyes (supraorbital vibrissae) (Ling 1977). Eared seals lack the rhinal vibrissae (Ling 1977). All vibrissae of true seals and eared seals, in contrast to the vibrissae of terrestrial mammals, are flattened, i.e. their cross section is elliptical rather than circular. True seals, with few exceptions, feature an additional specialization: their vibrissae are undulated (or beaded), that means both the greater and the smaller diameter of the ellipse vary along the length of the vibrissa (with a wavelength in the order of millimeters, out of phase by approximately 180°) (Dehnhardt and Mauck 2008; Hanke et al. 2010). This specialization, as we will see, has important consequences for the perception of hydrodynamic stimuli while the seal actively swims.

Hydrodynamic perception in invertebrates and anamniota

Hydrodynamic perception is known in most major aquatic and semiaquatic animal phyla (Wiese 1976; Bleckmann 1994; Breithaupt et al. 1995; Dehnhardt et al. 1998a; Engelmann et al. 2002; Crespo 2011). The peripheral receptors of all known hydrodynamic sensory systems consist of small structures on the animal’s skin or cuticle that move with the fluid and cause a depolarization or hyperpolarization in sensory cells. Cephalopods possess epidermal lines to sense water flow, where the only moveable structures are the kinocilia of the sensory cells (Budelmann and Bleckmann 1988). The best studied invertebrate group are probably the crustaceans. In crustaceans, the peripheral receptors consist of small appendices of the cuticle, the sensory hairs or setae (Wiese 1976; Yen et al. 1992; Fields et al. 2002). They move with the fluid and are innervated at their bases (Wiese 1976). In this regard, they are very similar to aerodynamic sensors such as those found in insects or spiders (Humphrey and Barth 2008), and in a broader sense also similar to the totally independently developed lateral line system of fish and amphibians (Coombs et al. 1989). All fish and most aquatic amphibians perceive hydrodynamic stimuli with the lateral line system (for reviews see Coombs et al. 1989; Bleckmann et al. 2004; Bleckmann and Zelick 2009). The basic unit of the lateral line system is the neuromast, an aggregation of hair cells whose kinocilia and stereovilli protrude into a jelly-like cupula. The cupula moves with the flow, and the associated movement of the kinocilia and stereovilli either hyperpolarizes or depolarizes the hair cells. Each hair cell is directionally sensitive. In each neuromast, two populations of hair cells with opposing directionality are present. If the cupula is deflected in a direction where one population of hair cells is maximally depolarized, the other population of hair cells is maximally hyperpolarized The opposite is true for a deflection in the opposite direction; and in case of a deflection orthogonal to these two, the hair cells are neither hyper- nor depolarized. In between, excitability follows a cosine law, meaning that the excitability of one hair cell population is its maximal excitability multiplied by the cosine of the angle between the direction of the actual deflection and the maximally exciting direction of deflection (Bleckmann 2004). The lateral line system of fish consists of two subsystems, the superficial lateral line and the canal lateral line. The superficial lateral line consists of neuromasts on the surface of the skin. The water flow around these neuromasts can sometimes be modified by structures such as grooves or pits. Superficial neuromasts respond proportionally to the water velocity at the neuromast’s cupula; it must be noted, however, that the water velocity at the cupula is generally lower than the free water velocity, as the cupula is predominantly or fully located within the fish’s boundary layer. By contrast, the canal lateral line consists of neuromasts that are situated in fluid-filled canals in the fish’s skin. These canals are connected to the surrounding water via canal pores (which are sometimes covered by membranes) (Coombs et al. 1988). The lateral line canals constitute accessory structures that modulate the stimulus prior to perception in such a way that the neuromasts do not respond to the water velocity across the skin, but to pressure differences between two adjacent pores (Denton and Gray 1985, 1988).

Two types of hydrodynamic stimuli: dipole stimuli and hydrodynamic trails

Hydrodynamic sensory systems in cephalopods, crustaceans, fish and mammals have often been studied using hydrodynamic dipole stimuli (hereafter called dipole stimuli). Dipole stimuli correspond to the water movements caused by a small sphere that oscillates sinusoidally in the water along one axis (the dipole axis). They have two distinct advantages: they are easy to generate, and the water movements can be calculated using known analytical equations. Dipole stimuli are suitable to mimic many naturally occurring hydrodynamic stimuli to a satisfying degree. This is due to the fact that all other components of the water movements caused by moving or deforming objects decrease with the distance from the object faster than the dipole component does (Milne-Thompson 1968), as long as the object does not change its volume (i.e. its monopole component is zero). This condition is probably often met in good approximation in animals that move using body appendages, even in animals that possess gas-filled swimbladders, as indicated by the absence of recordable underwater sound in the vicinity of swimming fish (personal observations). However, the role of water vortices and their perception is not covered by studies that use dipole stimuli. Studies that used dipole stimuli include early (Harris and van Bergeijk 1962) and more recent (Coombs et al. 2001; Bleckmann 2008; Coombs and Patton 2009; Braun and Coombs 2010; Mogdans and Nauroth 2011; Meyer et al. 2012) research on fish, as well as studies on crustaceans (Tautz et al. 1981; Heinisch and Wiese 1987; Killian and Page 1992) and cephalopods (Bleckmann et al. 1991b), to name a few examples. Dipole stimuli were also used for the initial assessment of the sensitivity of harbour seals (P. vitulina) to water movements (Dehnhardt et al. 1998a).

Hydrodynamic trail is a collective term for water movements that are generated in the near field of a moving object, but, as the object moves away, are after some time found at a considerable distance from the object. They can contain complex patterns of water movements that are often best described as vortices, including vortex rings that are relatively persistent structures (Shariff and Leonard 1992). The hydrodynamic trails caused by small (10 cm) fish that swim on a straight path have been shown to potentially last for 5 min or more under calm water conditions (Hanke et al. 2000). The trails of larger (25–30 cm) fish performing fast-starts contain water velocities of at least 1.5 m/s and have by extrapolation been estimated to last 25 min and more even in the presence of background turbulence (Niesterok and Hanke 2012). A harbour seal swimming at moderate speed produced hydrodynamic trails that contained water velocities of at least 30 mm/s after 30 s (Schulte-Pelkum et al. 2007). Different fish species can under certain circumstances be distinguished by the hydrodynamic trails that they leave (Hanke and Bleckmann 2004). In summary, hydrodynamic trails have a high potential to inform animals that use hydrodynamic sensory systems about the presence and certain features of an object that has passed by at an earlier point in time, but may now be several meters or tens of meters away.

Hydrodynamic perception in harbour seals (P. vitulina)

Dipole stimuli

The ability of harbour seals to perceive water movements has been investigated using dipole stimuli (Dehnhardt et al. 1998a). A harbour seal was trained to position in a station (to station, for short) at a defined distance from a stimulus generator (Fig. 1a, b). The stimulus generator was a small sphere that could be made to oscillate in the water at a single frequency, generating water movements, constituting a hydrodynamic dipole stimulus. In a go/no-go response paradigm, the harbour seal was trained to leave the station (go response) when it perceived the hydrodymanic stimulus, and to stay in the station (no-go response) otherwise. This way, the perception thresholds of the harbour seal, defined as the stimulus strength where the stimulus elicited a go-response in 50 % of the trials, were assessed at seven different frequencies from 10 to 100 Hz. Displacement, velocity and acceleration of the water at the position of the vibrissae could be calculated from known equations. An example of the water velocity vector field generated by the moving sphere is shown in Fig. 1c. Water velocities are strongest close to the moving sphere, and fall off steeply in the innermost part with the third power of the distance from the sphere.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig1_HTML.gif
Fig. 1

Hydrodynamic dipole experiment with a harbour seal. a Photograph of the experimental setup. b Drawing of the experimental setup. A harbour seal was trained to position its neck in a stationing hoop and its chin at a jaw station. This way a reproducible position of the muzzle and the whiskers was achieved. At a defined distance in front of the seal, a sphere (diameter 10 cm) oscillated sinusoidally in the water on a vertical axis, driven by a shaker motor. c Example of the water velocity vector field around the sphere, calculated for the sphere being at its maximum speed passing its zero displacement position, oscillation amplitude 1 mm, oscillation frequency 50 Hz. d Behavioural threshold of the harbour seal detecting the dipole stimuli in a go/no-go experimental paradigm, as a function of dipole oscillation frequency (replotted from Dehnhardt et al. 1998a, b). Here, water acceleration was chosen as the parameter on the abscissa (for plots of the same data with water displacement or velocity on the abscissa, see Dehnhardt et al. 1998a, b). The nearly constant thresholds in the range between 10 and 50 Hz indicate that in this frequency range, the vibrissal system acted as an acceleration detector

The harbour seal was equipped with an eye mask to prevent visual detection of the movement of the sphere. Auditory noise was played from headphones to mask any sounds from the shaker motor that moved the sphere.

The harbour seal responded to very weak hydrodynamic stimuli. In terms of water velocity, its best sensitivity was found at a frequency of 50 Hz, where it amounted to about 250 μm/s. In terms of water acceleration, the perception threshold of the seal was roughly constant from 10 to 50 Hz (Fig. 1d). For a sinusoidal movement as applied here, 250 μm/s is equivalent to a displacement amplitude of 0.8 μm and an acceleration of 77 mm/s2 (Dehnhardt et al. 1998a). In other words, the harbour seal had approximately the same sensitivity to dipole stimuli as some fish achieve with their lateral line (Bleckmann and Münz 1990). Some fish are approximately ten (Münz 1985; Bleckmann et al. 1989; Bleckmann 2008) or even a hundred times more sensitive than that (Coombs and Janssen 1990), the latter probably using their canal lateral line system.

A harbour seal was also tested for its ability to discriminate between dipole stimuli of different displacement amplitude. Presented with a stimulus of 3 μm displacement amplitude at 40 Hz, the seal was able to detect a change in amplitude of 0.8 μm (Weber fraction 0.26) (Dehnhardt and Mauck 2008). Fish appear to be able to discriminate between dipole stimuli of different amplitudes as well (Dailey and Braun 2011). However, the design of the study by Dailey and Braun (2011) aimed at testing for generalization across different amplitudes and does not allow for direct comparison with the discrimination performance of harbour seals, and no attempt was made so far to distinguish between the contributions of the lateral line and the inner ear system to this ability, where the inner ear may often be dominant (Dailey and Braun 2009; Nauroth and Mogdans 2009; Mogdans and Nauroth 2011).

Artificial hydrodynamic trails

Harbour seals have been tested for their ability to detect, to follow, and to analyse hydrodynamic trails, as these have the potential to play an important role by enhancing the detection range of hydrodynamic sensory systems. In most experiments, artificial hydrodynamic trails were used.

First, extended hydrodynamic trails were generated by the use of miniature submarines (Dehnhardt et al. 2001; Wieskotten et al. 2010b). Figure 2 shows the submarine type used in the study of Wieskotten et al. (2010b), two examples of a harbour seal following this submarine, and the decay of water velocity in the submarine’s trail measured with particle image velocimetry (compare Westerweel 1997). The miniature submarine in the first study (Dehnhardt et al. 2001) ran autonomously on either linear or curved paths, the latter being induced by inclination switches in a way unpredictable to the experimenter (double-blind experiment). The submarine generated a hydrodynamic trail that was similar to the hydrodynamic trails of swimming fish in two fundamental respects, namely, the water velocity and the width of the trail, where values for a typical 30-cm-long prey fish were extrapolated from flow measurements using smaller (6–10 cm) goldfish (Hanke et al. 2000). The experiments showed that harbour seals followed linear and curved hydrodynamic trails equally well. Letting the linear trails age up to 20 s before the seal was allowed to follow them did not affect the animals’ performance. However, when the animal’s vibrissae were covered by a stocking mask, it never successfully followed a hydrodynamic trail. In a follow-up study (Wieskotten et al. 2010b), another fundamental characteristic of the natural hydrodynamic trails generated by fish was introduced. Fish often do not swim continuously, but in burst-and-glide mode, meaning that single tail flicks or short series of tail flicks are alternated with glide phases where the fish keeps its body rigid (Blake 1983). This feature was included in the experiments by the use of modified miniature submarines. The new miniature submarines still produced hydrodynamic trails of similar water flow velocity and width as a medium-sized prey fish, but now were remote controlled to swim on defined paths. The remote control also allowed to switch the submarine’s engine off, letting the submarine glide, thus producing a hydrodynamic trail more similar to that of a gliding fish. Hydrodynamic trails that were in part generated by the actively propelled submarine (“burst trail”), but about half-way changed into trails produced by the gliding submarine (“coast trails”) were applied in part of the experiments. The percentage of successive trail following by the seal was assessed, as well as the areas where the seal lost a trail. Figure 3 summarizes the results from flow measurements in the hydrodynamic trail of the submarine, and the performance of the harbour seal when following burst trails and coast trails (adapted from Wieskotten et al. 2010b). The study showed that harbour seals are able to follow the trails of gliding fish-like objects quite well, with just a slightly lower success rate than following the burst trails. There was however a tendency for the seals to lose track of a trail around the switching point from a burst to a glide trail. Importantly, it must be noted that the trails of the gliding submarines can be considered similar in flow velocity to the trails of gliding fish of similar size (which have not yet been measured). This can be concluded from the fact that the submarine, after the engine had been switched off, easily glided a distance of more than 7 m; it thus had only little flow resistance.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig2_HTML.gif
Fig. 2

Hydrodynamic trail following in harbour seals investigated with a miniature submarine. a The remote controlled miniature submarine. For details see Wieskotten et al. (2010b). b Example of the swim paths of the miniature submarine (red dots) and the harbour seal (white dots) that follows it. View from a camera on an 8-m-high mast above the experimental pool. The curved hydrodynamic trail of the submarine is followed by the animal with only slight deviations. c Example of the harbour seal (white dots) intersecting the swim path of the submarine (red dots). The seal detects and then follows the hydrodynamic trail. d Water velocities in the hydrodynamic trail of the submarine measured with particle image velocimetry as described in Wieskotten et al. (2010b) and Gläser et al. (2011), who both used the same submarines. Data from Gläser et al. (2011) replotted and fitted with an exponential decay function

https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig3_HTML.gif
Fig. 3

Performance of a harbour seal following the hydrodynamic trail of a miniature submarine, where in part of the trials the engine of the submarine was switched off approximately halfway, leaving the submarine gliding. a Typical examples of the hydrodynamic trail of the miniature submarine measured with particle image velocimetry. Left burst trail, the engine is running; right glide trail, the engine is switched off and the submarine is gliding. Top vector fields of water velocity, circles indicating the origin of the vector and lines indicating direction and amount of the water velocity. Bottom temporal development of the amount of water velocity in the trail. For this plot, each vector field (25 vector fields per second) was reduced to a single row by averaging velocity over the columns of the vector field. These rows where then assembled in temporal order, with the velocity colour coded. The trail of the propelled submarine (left) lasts longer and spreads wider than the trail of the gliding submarine (right) Redrawn from (Wieskotten et al. 2010b). b Trail following performance of a harbour seal as a function of the delay between the start of the miniature submarine and the start of the animal. Solid line performance in burst trials, broken line performance in glide trials. The harbour seal was able to follow both burst and glide trails quite well, but with delays of 15–25 s, performance was significantly lower in glide than in burst trials. Replotted from Wieskotten et al. (2010b)

The experiments with hydrodynamic trails generated by miniature submarines showed not only that harbour seals can perceive this kind of stimuli, but also that they readily apply a search strategy where extended trails have to be followed. To modify single parameters of the hydrodynamic trails and to learn more about the details of the detection process, an experimental approach was developed where hydrodynamic stimuli were generated inside an experimental box (1.8 m × 2 m × 1.3 m, L × W × H) that was positioned in the shallow water (depth 1.1 m) (Wieskotten et al. 2010a, 2011). The harbour seal was trained to wait outside the experimental box while the stimulus was being generated. The seal was equipped with an eye mask for the whole experimental trial and with headphones displaying noise while it was waiting, so that it had to solve the following tasks using its vibrissae. After stimulus generation, the seal was allowed to swim into the experimental box up to its foreflippers, where it encountered the hydrodynamic stimulus. An additional time span was introduced between the generation of the hydrodynamic stimulus and the start signal for the seal to swim into the box; this way, the stimulus could age similar to the hydrodynamic trail of a prey fish having passed by at an earlier point in time.

Using this approach, it was shown that a harbour seal was able to determine the motion direction (from left to right or from right to left) of an artificial fish fin at least 35 s after the fin had passed by, using the fin’s hydrodynamic trail (Wieskotten et al. 2010a). The artificial fish fin (size 6 cm × 7 cm, W × H) was moved through the experimental box on a straight path from left to right or from right to left at a speed of approximately 20 cm/s. After the fin had stopped, a delay of 5–50 s was introduced. Then the harbour seal was allowed to enter the experimental box and to attempt to find out the movement direction using the fin’s hydrodynamic trail only. Psychometric data showed that the probability of correct answers was dependent on the delay (the age of the hydrodynamic trail). At a delay of 35 s, the seal’s performance was 71 % correct choices and was significantly different from chance (p = 0.012) (Fig. 4a).
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig4_HTML.gif
Fig. 4

Performance of a harbour seal in direction, size and shape discrimination experiments with hydrodynamic trails. Hydrodynamic trails were generated inside an experimental box which the animal was allowed to enter after a certain delay (see text). a Direction discrimination. A fish-like paddle was moved either from the right to the left or from the left to the right, and the animal had to indicate the movement direction using only the hydrodynamic trail. The percentage of the correct choices by the animal is plotted as a function of the delay (replotted from Wieskotten et al. 2010a). b Size discrimination. Flat vertical paddles of varying width (=size) were moved through the water, and the animal had to tell by the hydrodynamic trail if it was a paddle of standard width, or a different width. The Weber fraction, i.e. the smallest detectable width difference divided by the standard width of the respective experimental series, is plotted as a function of the standard width. Numbers at the data points indicate absolute size differences. Adapted from Wieskotten et al. (2010a). c Performance of a harbour seal discriminating vertical paddles or cylinders of different shape moving at the same speed. Experimental procedure equivalent to b. The animal’s success rate for object pairs of different shape is presented. Significant performance is indicated by an asterisk. Adapted from Wieskotten et al. (2010a)

In two other sets of experiments (Fig. 4b, c), a harbour seal was trained to distinguish between differently sized or differently shaped objects using their hydrodynamic trails only. Again, a two-alternative forced choice procedure was used. The seal learned to respond by touching one of two paddles, but this time not to indicate the directions left or right. Instead, touching one paddle corresponded to recognizing one of the learned sizes or shapes, touching the other paddle corresponded to recognizing a different size or shape. The objects were vertical rods or paddles 30 cm high and up to 8 cm wide. They were moved on horizontal circular paths of 1.4 m diameter inside the experimental box, passing the front door where the seal would enter.

In the experiments on size discrimination, the objects were flat paddles, 2–8 cm wide. It was tested which differences in size (defined as the width of the object) the seal could resolve. When the objects were moved at a constant speed of 55 cm/s, the harbour seal was able to discriminate size differences down to 2.8 cm. When the object speed was varied randomly between 31, 43, 55, 69 and 85 cm/s between trials in order to exclude maximum or mean water velocity as a cue, the seal was still able to discriminate size differences down to 4.1 cm (Fig. 4b).

In the experiments on shape discrimination, the objects were flat paddles as before, or flat paddles with undulated edges, or rods with a circular or triangular cross section (called flat, undulated, triangular or cylindrical objects). Again, experiments started with an easier approach, where object speed and size (the size defined as the width of the object) were held constant. Under these conditions, the seal could readily distinguish flat from cylindrical, flat from triangular, flat from undulated, and undulated from cylindrical (Fig. 4c). However, within the experimental season, it did not learn to distinguish undulated from triangular, or cylindrical from triangular. In an advanced task, the size of the differently shaped objects was varied randomly. This way, the harbour seal could not use the spatial extent of the water disturbance as a cue for the object’s shape, as it might have done before; still, the seal could distinguish the flat from the triangular shape. A possible explanation for the seal’s performance is that it used the spatial arrangement of vortices in the object’s wake.

Currently, harbour seals are being tested in a similar experimental setup using quasi-natural, but highly reproducible and stable hydrodynamic stimuli, namely vortex rings. Vortex rings have been shown to occur in the hydrodynamic trails of fishes (Drucker and Lauder 2002; Nauen and Lauder 2002), and isolated vortex rings are prominent in the hydrodynamic trails of fishes that perform escape responses (Niesterok and Hanke 2012).

In summary, the experiments on the discrimination of size and shape of moving objects using their hydrodynamic trails show that harbour seals have way more abilities to analyse hydrodynamic trails than merely detecting a water disturbance and moving towards it. This strongly supports our notion that the vibrissal system of harbour seals serves to form an image of the hydrodynamic environment within the reach of the vibrissae by integrating the input from multiple vibrissae, and to identify contiguous structures or patterns in the flow field. The vibrissae, including the vibrissal hair shafts and the highly innervated follicle–sinus complexes at their bases, form a multi-sensor array that samples many points of the environment simultaneously. We have only started to assess the potential of the subsequent central nervous processing.

Natural hydrodynamic trails

Experiments with artificial hydrodynamic trails that mimic natural hydrodynamic trails allow us to quantify the seal’s performance in various respects. However, the harbour seals’ ability to follow hydrodynamic trails has also been put to a test using natural hydrodynamic trails. Those were the hydrodynamic trails produced by other harbour seals (Schulte-Pelkum et al. 2007) (Figs. 5, 6).
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig5_HTML.jpg
Fig. 5

Following of biogenic hydrodynamic trails. View from a camera on an 8-m-high mast above the experimental pool. A harbour seal (Malte) was trained to swim from a platform (lower left corner) to the opposing wall on varying paths. This seal’s swim path is drawn in blue. A second harbour seal (Nick) was trained to follow the first seal using hydrodynamic information only. This seal’s swim path is drawn in red. a The second seal follows on as straight path. b The second seal follows on an undulated path. From Schulte-Pelkum et al. (2007)

https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig6_HTML.gif
Fig. 6

a Trail following performance of a harbour seal following the hydrodynamic trail of another harbour seal, plotted as a function of the swimming direction of the trail generating animal, where 0° is the direction straight ahead from the start platform. Success rate was independent of the angle. b Frequency of trials where the trail follower showed a linear search pattern as compared to an undulated search pattern. Undulated search patterns occured more frequently at large and small angles, possibly due to the angle under which the trail follower encountered the trail

A harbour seal (the trail generator) was trained to swim on an approximately straight path from a starting platform at the edge of the experimental pool to an end point near the opposing wall. This end point was varied pseudorandomly [i.e., end points were chosen randomly, but the same point was not chosen more than three times in a row, in this regard following the principles of Gellermann (1933)]. Swim paths were 10.5–15 m in length. After reaching the end point, the trail generator left the water and stayed out of the water, so that it could not give additional cues. At the same time, another harbour seal (the trail follower) was waiting near the starting platform. It was equipped with an eye mask and headphones that displayed pink noise in order not to perceive any cues on how and where the trail was generated. (Pink noise is noise with a spectral power density inversely proportional to the frequency; this way, the lower frequencies that are expected to originate from swimming animals (Bleckmann et al. 1991a) are masked without exposing the experimental subject to unnecessary high-frequency noise.) After the trail generator had left the water, the trail follower, still wearing the eye mask, was asked to find the end point of the trail using its vibrissae. Delays between the start of the trail generator and the trail follower were approximately 15 s.

The trail follower proved quite successful in finding the end points of the hydrodynamic trails generated by its conspecific. It found the end point in 90 % (444 out of 495) of the trials. Its success rate did not depend on the position of the end point of the trail along the opposing wall. However, the position of the end point had an influence on the trail follower’s strategy. It tended to follow the hydrodynamic trail more or less exactly along the swim path of the trail generator when the end point lay straight ahead from the starting platform. This was called a linear search pattern (Fig. 5a). When the end point was situated more to the left or to the right, an increasing tendency arose that the trail follower, while proceeding along the trail, crossed the trail multiple times, as if to investigate its borders. This was called an undulatory search pattern (Fig. 5b). Both search patterns were equally effective (Fig. 6a), the linear search pattern saving a small amount of energy and time. It was hypothesized that the angle of encounter between the swim path of the trail generator and the trail follower may have been influential on the choice of search patterns, as the undulated pattern occurred more frequently when the hydrodynamic trail led to the left or to the right rather than straight ahead from the experimental platform (Fig. 6b). Interestingly, when performing linear search patterns, the trail follower did not keep a quite constant heading, but followed the trail generator even on small deviations from a straight path. As flow measurements in separate experiments showed that the hydrodynamic trail could grow more than 2 m wide, it can be concluded that the trail follower was able to analyse the inner structure of the hydrodynamic trail to some degree.

Hydrodynamic perception in the California sea lion (Zalophus californianus)

Dipole stimuli

Dehnhardt and Mauck (2008) tested an eared seal, the California sea lion (Z. californianus), with a hydrodynamic dipole stimulus using an experimental setup and procedure equivalent to the experiments with harbour seals (Dehnhardt et al. 1998a). The California sea lion was tested at frequencies of the hydrodynamic dipole stimulus of 20 and 30 Hz. A comparison of the performance of the harbour seal and the California sea lion is presented in Fig. 7. Harbour seals can detect water velocities of 0.37 and 0.35 mm/s at 20 and 30 Hz, respectively. The California sea lion was even more sensitive at both frequencies.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig7_HTML.gif
Fig. 7

Sensitivity of a California sea lion (Zalophus californianus) and, for comparison, sensitivity of a harbour seal (Phoca vitulina) to dipole stimuli. The experimental setup and procedure are equivalent to those presented for harbour seals in Fig. 1. The calculated displacement amplitude of the water movements at the vibrissae where the sensory threshold of the animal was reached is plotted against the frequency of the sphere oscillation. Zalophus was tested at two different frequencies, where it was more sensitive than Phoca. Data for Zalophus replotted from Dehnhardt and Mauck (2008), data for Phoca from Dehnhardt et al. (1998a)

Artificial hydrodynamic trails

Trail following experiments with a California sea lion (Gläser et al. 2011) were performed in the Zoo Duisburg, Germany, where the animal lived in a group of sea lions in a pool of 230 m2. The same remote controlled miniature submarine and similar eye masks as in the recent harbour seal study (Wieskotten et al. 2010b) were used. The sea lion was trained to follow the hydrodynamic trail generated by the submarine while blindfolded. 209 hydrodynamic trails of length 1–6 m with different curvature were generated and the sea lion was asked to follow them after delays (time between the start of the submarine and the animal) of up to 7 s. Hydrodynamic trails were quantified using particle image velocimetry (PIV) as in the study of Wieskotten et al. (2010b).

The sea lion was able to follow the hydrodynamic trails reliably regardless of their length or curvature. An example from the data set analysed in Gläser et al. (2011) is presented in Fig. 8a. However, the time delay between the start of the submarine and the animal had a significant effect on the trail following performance. In contrast to harbour seals, the sea lion was unable to follow this kind of hydrodynamic trails after a delay of 7 s (Gläser et al. 2011). Figure 8b–d (replotted from Gläser et al. 2011) depicts the percentage of successive trail following in the California sea lion in three types of submarine trails: linear trails, where the submarine was steered on a straight path (Fig. 8b); curved trails, where the submarine’s path had one bend to the left or to the right (Fig. 8c); and complex trails, which contained more than one bend (Fig. 8d). Specifically in the curved trails, it becomes apparent that the sea lion’s performance declined rapidly for trails older than approximately 5 s. Attempts with 7-s-old trails were performed (not recorded) and confirmed that the animal could not follow a trail at this delay during any trial.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig8_HTML.gif
Fig. 8

Hydrodynamic trail following in a California sea lion (Zalophus californianus) investigated with a miniature submarine. The same submarine as presented in Fig. 2a was used. a Swim paths of the miniature submarine and the experimental animal in a typical example. The swim path of the submarine is shown in yellow. Sections showing the sea lion were cut out from video images and superimposed to show the animal’s position at different points in time (time indicated in the figure; t = 0 is the time when the submarine was started from the platform in the lower left corner). Reassembled from the data set presented in Gläser et al. (2011). bd Trail following performance of the California sea lion in three types of trials. bLinear trails the submarine was steered on a linear path. cCurved trails the submarine was steered on a curved path (one single bent). dComplex trails the submarine was steered on a complex path including multiple curves; path generation took so long that the submarine’s engine was still running when the animal started its search; the animal, however, preferred following the hydrodynamic trail over taking an acoustically guided shortcut. b The animal’s success rate declined when the delay between the start of the submarine and the start of the animal exceeded 5 s. It must be noted that the animal was completely unable to follow any trails after 7 s and more, which was unequivocally observed by the experimenter and by the submarine operator, but was not recorded on video and is therefore not included in the figure

Flow measurements showed that after 7 s the hydrodynamic trail still contained water velocities higher than 50 mm/s (Gläser et al. 2011). By contrast, harbour seals can follow the hydrodynamic trails of the same type for at least 20 s, corresponding to water velocities close to 20 mm/s (compare Fig. 2d).

Hydrodynamics of vibrissal hair shafts

The cross section of true seal and eared seal vibrissae is elliptical rather than circular. With few exceptions (namely, the monk seals, Monachus spp., and the bearded seal, E. barbatus, compare “Introduction”), true seals have undulated vibrissae (Fig. 9a), that means both the greater and the smaller diameter of the ellipse oscillate sinusoidally (out of phase by approximately 180°) along the length of the vibrissa. Eared seals, by contrast, have smooth vibrissae (Fig. 9b). To test the hypothesis that the distinct shape of the vibrissal hair shaft plays a crucial role in the detection process, four sets of experiments were performed (Hanke et al. 2010; Miersch et al. 2011).
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig9_HTML.jpg
Fig. 9

Different shapes of vibrissae from true seals (Phocidae) and eared seals (Otariidae). a The undulated vibrissa of a harbour seal (Phoca vitulina, Phocidae). b The smooth vibrissa of a California sea lion (Zalophus californianus, Otariidae). Scale bar 1 mm. Reassembled from photographs presented in Hanke et al. (2010)

First, a harbour seal was equipped with a head-mounted camera. The camera recorded the position and possible movements of the mystacial vibrissae while the harbour seal performed a hydrodynamic trail following task as described above. Recordings showed that the harbour seal constantly held its mystacial vibrissae in a protracted position during trail following, and that any vibrations of the vibrissae induced by the forward motion of the animal were below the resolution of the camera (i.e., below 0.2 mm in amplitude).

Second, in a comparative approach, vibrissae from harbour seals (P. vitulina) and from California sea lions (Z. californianus) were investigated in a rotational flume (Miersch 2002; Hanke et al. 2010; Miersch et al. 2011). The flume consisted of a round tank 124 cm in diameter, filled with water to a level of 20 cm. The tank rotated around its axis, inducing a rotation of the water body of the same angular speed after a break-in period. This way a turbulence-free water flow of only minimal curvature with respect to the length scale of a vibrissal diameter was established. Vibrissae were mounted one at a time in a piezoceramic holder that was positioned above the water surface in a way that the vibrissa protruded into the water. Forces exerted on the piezoceramic holder were measured in relative scale via the charge induced in the ceramic. Measurements were taken with flow speeds at the position of the vibrissae from 0.323 to 0.550 m/s. To take different vibrissal diameters into account, Reynolds numbers (Re) were calculated based on vibrissal diameters. Considering only the cases where the Re values differed by 5 % or less, the forces acting on the base of the harbour seal vibrissae were on average 6.2 times lower than the forces acting on the vibrissae of the sea lion (9.5 times lower considering all flow speeds tested).

Third and forth, to test the hypothesis that the specialized shape of harbour seal vibrissae (see above) plays a crucial role in this difference in force, the vortex streets behind vibrissae exposed to uniform flow were investigated. Third, micro-scale stereoscopic flow measurements using the principle of particle image velocimetry (PIV) (Westerweel 1997) were performed (Hanke et al. 2010; Witte et al. 2012). Short sections of harbour seal vibrissae were investigated in a micro-stereo PIV system developed at the department of fluid mechanics at the University of Rostock. Reynolds numbers based on the vibrissal diameter were 300, corresponding to a swim speed of approximately 0.5 m/s. The flow structures behind the vibrissae showed a considerable three-dimensional deformation as compared to a smooth cylinder, which was induced by the specific three-dimensional structure of the vibrissa.

Fourth, to look at the flow in more detail, computational fluid dynamics (CFD) was applied (Hanke et al. 2010; Witte et al. 2012). These models revealed details of the flow structures behind the harbour seal vibrissa and made it possible to calculate the lift and drag forces exerted on the vibrissae by the flow, that is, the lateral and the flow-wise forces on the vibrissae of harbour seals and on a cylinder of equivalent diameter. Lateral forces are usually exerted on a cylinder in a medium velocity water flow by the shedding of vortices from the cylinder; these vortex-induced forces lead to vibrations of the cylinder (vortex-induced vibrations or VIV) that can pose severe problems in engineering. Lateral forces were reduced by 90 % and flow-wise forces were reduced by 40 % in the vibrissa as compared to a cylinder. Examination of the flow structures showed that lateral forces and thus the vortex-induced vibrations were reduced by at least three effects: vortices being shed by the vibrissa are weaker, they form at a greater distance from the vibrissa, and they are arranged more symmetrically behind the vibrissa than behind a cylinder.

In summary, the four experimental approaches together show that the special undulated shape of harbour seal vibrissae, as opposed to the smooth shape of eared seals, significantly reduces the vibrations of the vibrissa that would be induced by the flow as the animal swims forward. Such a reduction of self-induced vibrations appears suitable to increase the signal-to-noise ratio when the seal is searching for prey-generated hydrodynamic stimuli. This is consistent with the different performance of harbour seal and the California sea lions found in the behavioural experiments on hydrodynamic trail following where the animal was actively swimming forward. It is also consistent with the fact that such differences were not apparent or even reversed in the experiments with hydrodynamic dipoles where the animal was at rest.

Pinniped whisker diversity

Although most mammals possess vibrissae, the depth of our knowledge regarding their function is limited and mainly directly toward laboratory animal models (e.g., Halata 1975; Halata and Munger 1980; Rice et al. 1986, 1993, 1997; Ebara et al. 2002; Hartmann et al. 2003; Neimark et al. 2003; Mitchinson et al. 2004) Although much work has been conducted recently by sensory ecologists on a few species of marine mammals, the full diversity of vibrissal function remains to be explored. Pinnipeds are an interesting group to study vibrissal function since this taxonomic group possesses the largest vibrissae among mammals, and their vibrissae are among the most diverse in terms of morphology and physiology. Among pinnipeds, the number, geometric arrangement of whiskers on the body, size, morphology, and shape exhibit considerable variation (Ling 1977). A blood filled follicle–sinus complex (F-SC) is a diagnostic characteristic of mammalian vibrissae. The F-SCs in some pinniped species are large enough to span the epidermis, dermis, and hypodermis (Marshall et al. 2006). The organization of these blood sinuses within pinniped F-SCs is tripartite and the upper cavernous sinus can comprise as much as 60 % of the entire follicle (Ling 1977; Dehnhardt 2002; Marshall et al. 2006). A vibrissal hair shaft emerges from the base of the F-SC, out the apex to extend out into the environment beyond the muzzle and rostrum of the individual. This hair shaft is the bridge that relays vibrotactile information from the environment (e.g., hydrodynamic or other tactile cues) to the mechanoreceptors and innervation deep within the F-SC (Burgess and Perl 1973; Gottschaldt et al. 1973; Dykes 1975; Halata 1975). The vibrissal hair shaft functions as a biomechanical filter; variations in the morphology and biomechanics of the vibrissal hair shaft are of extreme importance in modulating environmental stimuli that result in vibrotactile reception. However, we are just beginning to appreciate the functional and ecological consequences that lie within this diversity (Ling 1977; King 1983; Watkins and Wartzok 1985; Dehnhardt et al. 2001; Ginter et al. 2010, 2012; Miersch et al. 2011). More comparative studies on the morphology of pinniped vibrissal hair shafts from a comparative perspective are needed. The morphology of harbour seal hair shafts have been shown to have important functional consequences regarding fluid flow over the hair shaft (Hanke et al. 2010), and it is likely that any divergence in hair shaft morphology will also change the functional hydrodynamics.

The typical structure of most phocid vibrissal hair shafts has synonymously been called corrugated (Watkins and Wartzok 1985), waved (Watkins and Wartzok 1985; Dehnhardt and Kaminski 1995; Dehnhardt 2002), undulated (Hanke et al. 2010; Miersch et al. 2011), or beaded (Ling 1977; King 1983; Hyvärinen et al. 2009; Ginter et al. 2010, 2012). It should be noted that the term beaded is not meant to imply that both the diameter of the vibrissa in dorsal view and in frontal view are thickest in certain locations and thinner in between. Rather the greater and the smaller diameter of the elliptical cross section change out of phase (Dehnhardt and Mauck 2008; Hanke et al. 2010).

Traditionally, the hair shafts of pinniped vibrissae have been characterized as either beaded [with the exception of bearded seals (E. barbatus) and monk seals (Monachus sp.)] or smooth (Ling 1977; King 1983). However, recent studies demonstrate that this classification is too simplistic. Among phocids, there are species-specific variations in the beaded morphology that are likely to have functional consequences. A small comparative study using traditional morphometrics of beaded phocid vibrissal hair shafts (peak-to-peak; crest and trough width) from harp (Pagophilus groenlandicus), hooded (Cystophora cristata) and gray seals (Halichaoerus grypus) demonstrated differences in the beaded morphology of these phocid vibrissal hair shafts (Ginter et al. 2010). Figure 10 gives the definitions of these morphometrics. Four regions of vibrissal hair shafts (proximal, middle, distal and tip) from each species were compared. Peak-to-peak distance was found to be greatest in the one gray seal studied, followed by harp and hooded seals, respectively. This same pattern held true for crest width. The trough width of the gray seal vibrissae was greater than both harp and hooded seals, and all measures were more consistent along the entire length of gray seal vibrissal hair shafts than either harp or hooded seals.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig10_HTML.gif
Fig. 10

Close-up segments of vibrissae from harp (top), hooded (middle), and gray (bottom) seals. Arrows depict crest width, peak-to-peak, and trough width distances. All three measurements were collected from every vibrissa. Photo credit: Carly Ginter

A second more comparative study (Ginter et al. 2012) used these same traditional morphometric measurements (peak-to-peak distance, crest width, and trough width) and geometric morphometrics to investigate vibrissal morphology. Geometric morphometrics is a powerful quantitative tool that is often preferable to traditional morphometrics, which often does not capture true shape differences. This study incorporated five phocids with beaded vibrissal hair shafts (gray seal, Halichoerus grypus; harbour seal, P.vitulina; harp seal, P. groenlandicus; ringed seal, Pusa hispida; and spotted seal, Phoca largha), a phocid with smooth vibrissal hair shafts (bearded seal, E. barbatus); and five otariid pinnipeds, which all have smooth (non-beaded) vibrissal hair shafts (California sea lion, Z. californianus; Guadalupe fur seal, Arctocephalus townsendi; Northern fur seal, Callorhinus ursinus; South American fur seal, Arctocephalus australis; and Steller sea lion, E. jubatus). Since the beaded morphology of most pinniped vibrissal hair shafts lacks distinguishable and homologous landmarks, an outline-based geometric morphometric analysis, elliptical Fourier analysis (EFA), was used. This method generated EFA harmonics (and their coefficients) that are added sequentially until the morphological outline of vibrissal hair shafts was matched. In this study, a good fit to the beaded morphology occurred within 15 harmonics independently of size, orientation, and rotation. Because four Fourier coefficients are generated to describe each harmonic (plus the two zeroth harmonics), a total of 62 shape variables were generated. Subsequent principal component analysis (PCA) reduced this dataset to generate 18 principal components (PCs) that summarized 99 % of the variance within the dataset. These PCs were then used in the subsequent quadratic discriminant function analyses (QDFA).

Traditional morphometric analysis results

Traditional morphometrics demonstrated that the number of crests per cm of hair shaft length significantly differed among some of the species and ranged from 1.9 (spotted seals) to 2.7 (gray seals; Ginter et al. 2012). The caudally directed curvature of the hair shafts did not change the mean values of peak-to-peak distances on either the convex or concave sides of the hair shafts. Among phocids with beaded hair shafts, spotted seals demonstrated the greatest mean crest and trough widths, whereas ringed seals possessed the lowest mean crest and trough widths. Interestingly, ringed seals also exhibited the most pronounced beaded appearance of any phocid in the study as indicated by the large crest width to trough width ratio of 1.44, while spotted seals had the second-to-lowest crest to trough width ratio (1.22; the lowest found in gray seals with 1.21). Harbour and harp seals were intermediate among the phocids in this aspect, and were nearly identical to each other. Since the beaded morphology has been demonstrated to influence fluid flow over the hair shafts in harbour seals (Hanke et al. 2010), the diversity in the beaded morphology among phocids suggest a diversity in hydrodynamic function and therefore differences in sensory perception that likely has ecological consequences. Quadratic discriminant functional analysis (QDFA) of traditional morphometrics of all pinnipeds separated species with smooth vibrissa from those with a beaded profile (Fig. 11a). That is, the canonical centroid plots of otariids separated from those of most phocids. Not unexpectedly the centroid plots of bearded seals overlapped with otariids, but gray seals were intermediate between otariids and phocids. Among the remaining phocids in this study (harbour, spotted, harp, and ringed), harbour seals and ringed seals did not overlap with each other in canonical space. Spotted seals exhibited the greatest variability and their centroid plots overlapped all other phocids in the study except gray seals. QDFA of the traditional morphometric dataset successfully classified approximately 80 % of all individuals to species (Table 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig11_HTML.gif
Fig. 11

Centroid plots for the quadratic discriminant function analyses (QDFA) on the three morphometric methodologies. Crosses mark the mean for each species; ellipses are 95 % confidence regions. a Results of QDFA on the traditional morphometric measures. b Results of QDFA on elliptic Fourier harmonic coefficients (geometric morphometrics). c Results of QDFA on the combined traditional and geometric morphometrics. From Ginter et al. (2012)

Table 1

Mean values and standard deviation for measured characteristics of seal vibrissae, presented by species of seal and location along the vibrissae (Ginter et al. 2010)

 

P. groenlandicus

C. cristata

H. grypus

Number of vibrissae

33–41 (38 ± 4)

24–39 (32 ± 11)

25

Overall length (mm)

39.3 ± 17.3

33.4 ± 16.8

37.8 ± 13.1

Total bumps

12 ± 5

10 ± 6

9 ± 3

Peak-to-peak distance

 Proximal

2.65 ± 0.42

2.24 ± 0.13

3.50 ± 0.40

 Middle

3.00 ± 0.42

2.50 ± 0.48

3.90 ± 0.37

 Distal

2.70 ± 0.55

2.26 ± 0.24

3.80 ± 0.57

 Tip

2.50 ± 0.50

2.00 ± 0.34

3.90 ± 0.49

Crest width (mm)

 Proximal

0.63 ± 0.07

0.58 ± 0.08

0.94 ± 0.17

 Middle

0.61 ± 0.04

0.57 ± 0.09

0.97 ± 0.19

 Distal

0.45 ± 0.08

0.45 ± 0.06

0.89 ± 0.18

 Tip

0.34 ± 0.06

0.31 ± 0.06

0.91 ± 0.15

Trough width (mm)

 Proximal

0.47 ± 0.06

0.46 ± 0.09

0.80 ± 0.16

 Middle

0.43 ± 0.05

0.42 ± 0.07

0.76 ± 0.15

 Distal

0.33 ± 0.07

0.36 ± 0.04

0.72 ± 0.14

 Tip

0.25 ± 0.06

0.25 ± 0.06

0.74 ± 0.14

For definition of parameters see Fig. 10

Elliptical Fourier analysis results

As in the traditional morphometric analysis, the QDFA on the geometric morphometric dataset obtained by EFA also separated otariids from phocids, including bearded seals, which share the characteristic smooth vibrissal hair shaft of otariids (Fig. 11b). In addition, among otariids the canonical plots of fur seals separated from sea lions. It is unclear as to what characteristic is involved in this separation. The focus of the study by Ginter et al. (2012) was to characterize phocid beaded morphology, and the smaller sample of otariids was included as a morphological baseline. Bearded seals grouped intermediate and closer to phocids (but still separate from phocids) compared to the traditional morphometric dataset. Among phocids with beaded vibrissae, there was much overlap in these groups but gray seals still appear to be divergent. The main effects appear to be smooth versus beaded (Canonical Axis 1) followed by hair shaft length or an interaction of several characters instead (Canonical Axis 2). QDFA of the geometric morphometric dataset was better at correctly assigning individuals to species than QDFA of the traditional morphometric data set (approximately 98 % as compared to 80 %).

Although EFA was better than traditional morphometric methods in assigning individuals to the correct species, it is clear that this analysis focused on very different aspects of vibrissal hair shaft morphology than traditional morphometrics. Although some overlap in the methodologies exist, both have merit in that they focus on different aspects of quantifying vibrissal hair shaft morphology. Therefore, a fused traditional-geometric morphometric analysis was developed in which both datasets were used to characterize morphology and shape. This involved scaling each morphometric dataset to their combined true ranks (see Ginter et al. 2012 for methodological details). The combined dataset generated 19 PCs that summarized 99 % of the variance in the dataset, which substantiated the view that some overlap existed between the two methodologies. As observed in both the traditional and geometric morphometric analyses, the fused method separated the otariids from phocids, and the otariids overlapped to a small degree with each other (Fig. 11c). Once again bearded seals were intermediate between otariids and phocids; bearded seals did not overlap with phocids but were located closer to phocids in canonical space. As in the prior two analyses, vibrissal length appears to be responsible for some separation among all pinnipeds, but among phocids with beaded vibrissae small-scale morphologies appear to create some separation of canonical plots due to several species-specific differences. In the combined analysis, the position of gray seals in canonical space closer to pinnipeds with smooth hair shafts was supported, and this is likely related to significant differences in the small-scale beaded morphology. That is, the relatively small peak-to-peak distance and small ratio of crest to trough width of gray seals likely account for their location. Harbour seals had some overlap with gray and ringed seals but were independent of harp and spotted seals. Ringed and harp seals overlapped greatly with each other followed by spotted seals. Once again spotted seals appear to have the greatest variation but did not overlap with either harbour or gray seals. The fused dataset proved superior to both other methods in correctly classifying individuals to species; the QDFA correctly classified 100 % of all individuals to species.

In summary, phocids with a beaded vibrissal hair shaft morphology showed specific differences within an overall conserved sinusoidal profile. Although phocids with a beaded hair shaft cluster together in canonical space and were separate from smooth vibrissae species, distinct differences are apparent in both small-scale and large-scale morphologies. Congeneric species in this study (harbour and spotted seals) did not overlap. In addition, gray seals appear intermediate among phocids between beaded and smooth hair shaft morphologies indicating a potential for divergent performance and perhaps foraging ecology compared to other phocids with beaded morphologies. The fact that bearded seals, which share the smooth hair shaft morphology with otariids, always separated from otariids strongly suggested divergent uses for smooth hair shafts. Indeed, the geometric arrangement of bearded seal vibrissae on the muzzle is directed more interiorly (Marshall et al. 2006) and is important for their benthic foraging mode that includes significant suction performance at the level observed in walruses (Fay 1982; Kastelein et al. 1994; Marshall et al. 2008). In contrast, otariid vibrissae are located more laterally on the muzzle and serve very different functions. It is suspected that there may be more interesting morphological and performance differences related to vibrissae among otariids, but additional data are needed. The diversity of phocid vibrissal hair shaft morphologies strongly suggests the potential for a diversity of hydrodynamic functions. Future functional studies of the hydrodynamics of sensory perception by pinnipeds using vibrissae should include comparative studies to fully explore the full diversity of vibrissal function.

Ecological considerations and predictions on hydrodynamic perception in other pinniped species

The diversity of pinniped whiskers may be interpreted to indicate a different role of the vibrissal system in different species. Of the two behaviourally investigated species, harbour seals often inhabits turbid waters (Weiffen et al. 2006) and may for this reason be more dependent on hydrodynamic trail following than California sea lions. California sea lions lives in Pacific waters where vision may play a more prominent role. It is however not known at this point what the consequences of subtle differences among whisker morphologies are. These differences have the potential to influence the hydrodynamics of the interaction between the vibrissa and the fluid as well as the mechanics of the vibrissa itself, that is, its stiffness and spanwise distribution of stiffness, and consequently its resonance frequency (or frequencies).

One trait, however, that clearly allows to classify pinniped whiskers into two categories is the undulated versus smooth structure. Figure 12 presents two representatives of the species currently under investigation at the Marine Science Center Rostock, the harbour seal (P. vitulina) with undulated and the South African fur seal (Arctocepahlus pusillus) with smooth vibrissae. So far, each category has been investigated hydrodynamically, biomechanically and behaviourally using one representative species, the harbour seal (P. vitulina) for the undulated vibrissal type and the California sea lion (Z. californianus).for the smooth vibrissal type. The undulated vibrissae of harbour seals improve hydrodynamic perception as the animal swims forward by reducing self-generated noise; it appears reasonable to assume that the same mechanism is effective to some degree in all pinnipeds that possess undulated vibrissae, that is all true seals (Phocidae) except monk seals (Monachus spp.) and bearded seals (E. barbatus). From the point of view of the feeding ecology of monk seals and bearded seals, it is plausible that they are less dependent on the perception of hydrodynamic stimuli than other phocid species. Monk seals are the only phocid species that inhabits exclusively tropical and subtropical waters, where visibility is on average much better than in temperate waters (compare e.g. Weiffen et al. 2006). Vision is well developed in phocids (Hanke et al. 2006a, b, 2009, 2011) and can therefore strongly supplement or largely substitute hydrodynamic perception in the monk seals’ habitat. In addition, monk seals have some preference for benthic prey with little need to pursue prey in the water column. Benthic organisms may cause water flow as well (Hanke et al. 2012; Schwalbe et al. 2012), but many of them are more sessile than pelagic organisms and do not produce extended hydrodynamic trails. However, this feeding preference is not a strict rule; monk seals are opportunistic feeders that include also mesopelagic prey (Goodman-Lowe 1998).
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-012-0778-2/MediaObjects/359_2012_778_Fig12_HTML.jpg
Fig. 12

Illustration of the different types of vibrissae and their arrangement in a true seal and an eared seal. a Harbour seal (Phoca vitulina), a true seal. The characteristic undulated structure of the vibrissae is visible. Rhinal vibrissal are discernible above the nose, supraorbital vibrissae above the eyes. b South African fur seal (Arctocephalus pusillus), an eared seal, stationing its muzzle at a training target. The vibrissae have a smooth outline. Photos: Marine Science Center Rostock

Bearded seals (E. barbatus) are clearly benthic foragers (Hjelset et al. 1999; Dehn et al. 2007; Marshall et al. 2008). They can grow the longest vibrissae of all phocids when in captivity; however, in the wild, their vibrissae are worn down extensively (C.D.M., personal observation). They appear to use them for the detection of food items by direct touch. The smooth structure of monk seal and bearded seal vibrissae may make them stiffer in comparison to beaded vibrissae, possibly providing an advantage when detecting and discriminating objects by direct touch. Stiffer vibrissae may specifically be advantageous when only the distal ends of the long vibrissae are in contact with the object (but compare the good discrimination abilities of harbour seals as long as also short vibrissae are used; Dehnhardt and Kaminski 1995; Dehnhardt et al. 1998b). From both vibrissal morphology and feeding ecology, one would predict that hydrodynamic perception may be less accomplished in bearded seals and monk seals than in harbour seals and other phocids.

Gray seals (H. grypus) are largely demersal and benthic foragers (Hammond et al. 1994a, b; McConnell et al. 1999). As in most phocids, their preferred prey species vary by availability. In the sites studied, gadoids, flatfish and sandeels constituted large percentages of the gray seals’ prey. It would appear plausible that the relatively weakly beaded vibrissae of gray seals reflects a reduced role of hydrodynamic trail following as compared to seal species that spend more of their feeding time in the water column. However, more data on the feeding strategy of gray seals would be needed to support this hypothesis.

Ringed seals, which have so far been found to have a distinctly beaded vibrissal structure with short peak-to-peak distance and a high crest to trough width ratio, feed on both benthic and pelagic prey (Simpkins et al. 2001a, b, c). Although the actual feeding events during dives were not recorded, these studies strongly indicate that ringed seals apply various search strategies including forward swimming on convoluted paths, revisiting of locations, sit-and-wait predation, and passive drifting with the water flow. Horizontally convoluted dives, likely indicative of feeding behaviour, occurred at all depths and were not limited to dives that reached the benthos (Simpkins et al. 2001b). The diet of ringed seals includes pelagic and benthic amphipodes (Welch et al. 1992), sandeel, and sticklebacks (Suuronen and Lehtonen 2012). The scarce data on prey item size (Suuronen and Lehtonen 2012) is consistent with the presumption that ringed seals may in general prey on smaller animals than gray seals do. This may imply a need for a higher sensitivity for hydrodynamic stimuli in ringed seals, and possibly a shift to higher best frequencies of the sensory system. Hydrodynamical and biomechanical laboratory studies on the vibrissae as well as behavioural laboratory and field studies with ringed and gray seals would be needed to investigate in how far different prey preferences and the associated hunting strategies are related to the observed differences in the vibrissal system.

Conclusions and future directions

Hydrodynamic perception is one of the major sensory modalities in pinnipeds. It is mediated by the vibrissal system, that is the whiskers, the highly innervated follicle–sinus complexes at their bases, the vibrissal part of the trigeminal nerve, and the respective central nervous structures. Little is known about the central nervous system so far. In one study (Ladygina et al. 1985) the cortex of a pinniped, the fur seal C. ursinus, was investigated electrophysiologically. It was found, not surprisingly, that the vibrissae are represented in a large area of the cortex, and that they are somatotopically mapped.

The vibrissae of pinnipeds are morphologically highly diverse. In some species they show an undulated structure that reduces vibrations caused by the water flow when the animal is swimming. The undulated structure is found only in true seals, the exceptions being species from tropical waters and benthic feeders.

Hydrodynamic perception is not only possible at close distance from the stimulus-generating object as the physical nature of dipole stimuli appears to indicate, but also at distances in the order of tens of meters if hydrodynamic trails are used.

Two species of pinnipeds have been investigated behaviourally for the perception of hydrodynamic stimuli, the harbour seal as an example of a true seal with the typical undulated vibrissal shape, and the California sea lion as an example of an eared seal, which always have smooth vibrissae. Sensitivity to dipole stimuli was in harbour seals about as good as the sensitivity of some fish using their lateral line system (other fish can be two orders of magnitude more sensitive). The California sea lion exhibited even slightly higher sensitivity. In these experiments, the animal was at rest. The relation was reversed when the animals were asked to follow hydrodynamic trails, requiring them to swim forward. The harbour seals performed considerably better than the one California sea lion tested, consistent with the theory that the undulated vibrissal shape reduces self-generated noise in the swimming animal.

In harbour seals, it was found that they can not only detect water movements and follow hydrodynamic trails, but also analyse artificial hydrodynamic trails with regard to the moving direction, the size and the shape of the trail generator.

Behavioural data with both dipole stimuli and hydrodynamic trails from more true seal and eared seal species would be needed to further assess the role of hydrodynamic perception in pinnipeds. In the light of the high diversity of morphology, physiology and feeding ecology of pinnipeds, it can be expected that their ability to perceive and analyse hydrodynamic stimuli is highly diverse as well.

The ability of harbour seals to discriminate different hydrodynamic trails is currently being studied using reproducible hydrodynamic stimuli that closely mimic fish trails, as well as stimuli from life fish.

Regarding the hydrodynamics and biomechanics of single pinniped vibrissae, three-dimensional morphological data are being assessed to enable computational modelling. The transduction of force and momenta in the flow should be examined experimentally in low-turbulence flumes such as the rotational flume using single vibrissae and vibrissa arrays.

In summary, we are just beginning to understand the role, the mechanisms, and the possibilities of hydrodynamic perception with the vibrissal system. This recently discovered mechanosensory modality faces similar challenges and offers similar potential for the animal’s orientation in its environment as the lateral line system for fish, or the auditory system for all vertebrates.

Acknowledgments

The authors’ original work was funded by grants of the German Research Foundation (DFG) to W.H. and G.D., and the Volkswagenstiftung to G.D. We thank the DFG and the Office of Naval Research Global for supporting a conference on Sensory Biology of Aquatic Mammals in October 2012 at the Marine Science Center Rostock, accompanying this special issue of the Journal of Comparative Physiology A.

Copyright information

© Springer-Verlag Berlin Heidelberg 2012