Physiology Paris ELSEVIER Journal of Physiology - Paris 96 (2002) 363-377

Home , Ampullae of Lorenzini

Available online at www.sciencedirect.com Journal of Physiology Paris ELSEVIER Journal of Physiology - Paris 96 (2002) 363-377 www.elsevier.com/locate/jphysparis

The electric sense of the paddlefish: a passive system for the detection and capture of zooplankton prey Lon A. Wilkensa,*, Michael H. Hofmannb, Winfried Wojteneka aCenter for Neurodynamics, Department of Biology, University of Missouri-St. Louis, 8001 Natural Bridge Road, St. Louis, Missouri 63121, USA bInstitute of Zoology, University of Bonn, Poppelsdorfer Schloss, 53115 Bonn, Germany

Abstract Behavioral and electrophysiological experiments have shown that the elongated paddlefish rostrum, with its extensive population of ampullae of Lorenzini, constitutes a passive electrosensory antenna of great sensitivity and spatial resolution. As demonstrated in juvenile paddlefish, the passive electrosense serves a novel function in feeding serving as the primary, if not exclusive sensory modality for the detection and capture of zooplanktonic prey. Ampullary receptors are sensitive to the weak electrical fields of plankton from distances up to 9 cm, and juvenile paddlefish capture plankton individually with great swimming dexterity in the absence of vision or other stimulus signals. Paddlefish also detect and avoid metal obstacles, the electrical signatures of which are a potential hindrance to their feeding and reproductive migrations. The ampullary receptors, their peripheral innervation and central targets in the dorsal octavolateral nucleus, are described. We also describe the ascending and descending neuronal circuitry of the electrosensory system in the brain based on tracer studies using dextran amines. © 2003 Elsevier Ltd. All rights reserved. Keywords: Paddlefish; Ampullae of Lorenzini; Rostrum; Electrosense; Zooplankton

1. Introduction passive electric sense in plankton feeding by the paddlefish, evidence that now provides a functional expla- The paddlefish, Polyodon spathula, is distinctive for its nation for the elongated rostrum and its role as an flat, elongated rostrum and large size. Its size can be 'antennal' organ. In addition, we present new informa- accounted for by a feeding strategy shared by a number tion concerning the neural architecture of the electro- of large aquatic 'grazing' animals including baleen sensory system, including peripheral innervation of the whales, the whale shark and basking sharks. All of the ampullary receptors and brain circuitry. aforementioned animals filter feed, straining large quantities of zooplankton from the water column, a rich resource near the bottom of the food chain. The raison 2. Methods d'etre for the rostrum has remained enigmatic, however, although it has long been known to be covered with A brief description is provided for each of the fol- "sensory pores" [18] of unknown function. Over 60 lowing experimental protocols beginning with the years later, and after considerable debate as to their methods used to establish the electrosensory basis of function, these pores were formally characterized as plankton feeding, the premise for our study of the pad- ampullae of Lorenzini [16], similar to the electrosensory dlefish. Detailed descriptions are available in the refer- ampullae of elasmobranchs and other primitive fish. enced literature. The university Institutional Animal Physiological recordings confirmed their electro- Care and Use Committee approved all experiments. sensitivity [29]. However, the association between the eye-catching rostrum, the electric sense, and plankton 2.1. Feeding behavior feeding was not made until more recently [43]. Here we review evidence that establishes a unique role for the These experiments utilize small paddlefish (12-17 cm long) that feed selectively by capturing individual Correpsonding author. plankton. Feeding was characterized by video analysis E-mail address: [email protected] (L.A. Wilkens). of plankton strikes by fish swimming in place in a

recirculating laminar stream of water driven by the overlying cartilage, followed by paralysis with curare propeller of a trolling motor [see also 43,44]. Fish were and washout of the anesthetic. Fish were artificially restricted to a viewing chamber with glass sides. Two irrigated by aerated water flowing over the gills. Stimu- IR-sensitive video cameras recorded their movements, lation of the electroreceptors was by a small silver wire one viewing the fish from the side, the other from electrode, serving as the cathode or anode, referenced to beneath via a 45°-angled mirror. Following baseline a large silver plate electrode at the side of the chamber. experiments in the light, all subsequent experiments Alternatively, a dipole electrode with wire tips separated have been performed in the dark using near infrared by 5 mm was used to deliver stimuli. Constant-current light (>880 nm) for illumination. Plankton, admitted to DC, or sinusoidal AC currents at low frequencies were the stream via a plastic tube, were added at a rate to used to modulate the firing of electroreceptor primary offset consumption and to maintain a density of 1-2 per afferents. We also recorded electrical responses to nat- liter. In some experiments we sought to obstruct poten- ural stimuli, i.e. to plankton glued to a thin glass fila- tial chemo- and mechanosensory signals by adding ment and swept over the receptive field by a pen motor concentrated plankton extract to the water, by encap- lever. Responses were recorded and analyzed using sulating live plankton in agarose, blocking the fish's Spike 2 software (CED, Cambridge, UK). nares, and by creating turbulent water flow. Feeding events were analyzed off line from video taped feeding 2.4. Neuroanatomy strikes. After each capture the tape was reversed to align the plankton with the tip of the rostrum. This reference Innervation of the ampullae was studied by clearing

frame was digitized and the coordinates of the plankton thin skin samples with 1% H2O2 over night and and rostrum tip were entered into a computer spread- immersing in a solution of 0.5% sudan black in ethylene sheet for analysis. Feeding strikes equivalent to those glycol containing 10% ethanol for 8-24 h. The tissue aimed at plankton could be elicited also by weak electric was rinsed in ethylene glycol (containing 10% ethanol) fields from dipole electrodes placed in the water of a several hours until the excess stain was washed out. The large holding tank (550 1) [46]. Strikes were recorded tissue was then placed on slides and coverslipped with from sets of four fish in response to weak oscillating ethylene glycol. currents at various intensities and frequencies, again Central projections of the electrosensory system were using infrared illumination and an IR-sensitive video studied by means of applying the neuronal tracer bioti- camera. nylated dextran amine (BDA, molecular weight 3000, Molecular Probes, Inc., Eugene, Oregon, USA) into the 2.2. Avoidance behavior lateral line nerves, dorsal octavolateral nucleus, cerebellum, and mesencephalic tectum. Before surgery, the Paddlefish sensitivity to a static electric field was tips of small insect pins were dipped into a paste made examined by introducing a 2.5-cm diameter aluminum of BDA dissolved in a small amount of distilled water. rod into the path of a fish swimming in circles near the The BDA paste dried on the tip of the pin. For surgery perimeter of a round tank (95-cm diameter) [14]. animals were anesthetized with MS 222 (0.01%, Sigma). Responses were compared to equivalent sized obstacles The skull was opened with a scalpel, reflecting back a of plastic and insulated aluminum, and to controls with piece of cartilage to expose the brain or nerves. The pins no obstacle. Swimming movements were recorded under were inserted into the tissue and the skull was closed infrared illumination using a computerized tracking and sealed with Vetbond Tissue Adhesive (3M, St. Paul, system (Chromotrack 4.02, San Diego, CA) that detec- MN, USA). After surgery fish were revived by perfusing ted a small reflector glued to the tip of the rostrum. The the gills with aerated water until they restarted respira- obstacles were lowered by a robotic manipulator while tory and swimming movements. Three days after sur- the fish was at the opposite side of the tank and an gery, the animal was deeply anesthetized with MS222 avoidance trial was recorded if the fish passed a criter- (0.05%) and transcardially perfused with phosphate ion point about 35 cm from the target. Approach dis- buffer (PB, pH 7.4, 0.1 M) followed by 4% para- tances and collisions were recorded for 20 fish, with 15 formaldehyde in PB (Fisher Scientific Company, New trials for each type of obstacle. Jersey, USA). After fixation the brain was removed from the skull and stored overnight in fixative in a 2.3. Electrophysiology refrigerator. The brains were transferred to PB for two or more hours and embedded in 7.5% gelatin (Fisher) in Electrically mediated responses were recorded from PB. The gelatin block was stored overnight in 4% par- electroreceptor primary afferents in the anterior lateral aformaldehyde in PB before it was cut on a vibratome line nerve (aLLN) at the level of the dorsal ganglion (OTS-4000, Electron Microscopy Sciences, USA). 100 adjacent to the medulla [43]. The brain was exposed μm sections were collected into PB. Sections were under anesthesia (0.01% MS222) by removal of the washed in PB and endogenous peroxidase destroyed by L.A. Wilkens et al. / Journal of Physiology - Paris 96 (2002) 363-377 365 incubation in 0.3% H2O2 for 10 min. Thereafter, the ducted in the dark, shows their position in vertical space sections were incubated with ABC-solution (Vectastain in front of the fish. As evident in this figure, the majority ABC Kit, Vector Laboratories, Inc, Burlingame, CA, of captures were within 20 mm of the rostrum (56%), USA) and stained with diaminobenzidine containing after which capture frequency falls off rapidly. Of this ammonium nickel sulfate and CoCl2. The sections were sample only 98 plankton (4.3%) were captured at dis- mounted on slides, dehydrated and coverslipped. Every tances greater than 40 mm, with a maximum at 83 mm. second section was counterstained with neutral red The distribution of captured plankton is centered fairly (Sigma, St. Louis, MO, USA). evenly on the rostrum with only a small increase in numbers below the rostrum (53%). Plankton captures are biased somewhat for the right side (59%), which we 3. Results speculate as being due to the fish swimming, on average, left of tank center. 3.1. Plankton feeding Capture distributions also reflect the shape of the rostrum, i.e., a somewhat flattened array that extends Juvenile paddlefish are particulate feeders, selectively laterally over greater distances that above or below the capturing individual zooplankton (mostly daphnids) rostrum. This is particularly evident for the most distant detected as they pass alongside the rostrum of the captures where the maximum lateral coordinate is 80.7 advancing fish. Paddlefish swim continuously through- mm compared to a vertical maximum of 45.0 mm, a out life as obligate ram ventilators [3]. The ontogeny of 79% increase in relative distance (63% when compen- non-selective suspension feeding by larger fish (>23 cm sated for by rostral dimensions, see below). This is also long) corresponds with the development of the gill represented in the mean lateral capture distance (13.3 rakers [35]. Our studies address only selective plankton mm, absolute value) in comparison to the mean vertical feeding. distance (9.0 mm). However, this 4.3 mm difference is A feeding strike is characterized as a "ram-gulp" more than offset by the mean half-width dimensions of capture mechanism where the mouth opens widely as the rostrum horizontally (9.8 mm) and vertically (1.4 the fish overtakes and engulfs the prey. Thus, there is no mm) (cf. rostral inset, Fig. 2A). The horizontal dimen- entrainment of prey in a high-velocity suction current, sion is overvalued, however, since it reflects the max- the creation of which is unnecessary for the slow-mov- imum rather than average width of the rostrum. Thus, ing plankton. For a plankter passing directly under- relative capture distances are approximately equidistant neath the rostrum the fish simply opens its mouth. For about the rostrum, i.e., in their proximity to the ampul- trajectories above, well below, or to the side of the ros- lary receptors. trum, a sequence of stereotyped motions ensue that The plankton capture distributions illustrated in reflect the hydrodynamic constraints imposed by the Fig. 2A are equivalent to those obtained in the light broad, flat rostrum. Lateral captures involve saccade- [42]. Therefore, we interpret these results as representing like yaws, i.e., slicing rostral motions that direct the the relative distances from the rostrum at which pad- mouth toward the plankton. For targets with vertical dlefish detect the electric fields of the plankton. To inclination, capture motions involve both roll and yaw, gauge the potential involvement of other sensory mod- again allowing the rostrum to slice through the water alities, additional control experiments were designed to with least resistance. Fig. 1 illustrates a feeding strike interfere with the chemical and/or hydrodynamic signals where the fish turns nearly upside down, using approxi- from the plankton, all of which were conducted in the mately 130-140° of roll to capture plankton passing dark [44]. Feeding was tested in the presence of high above the rostrum. These often acrobatic movements background concentrations of plankton extract in the are initiated by a combination of fin motions and/or an stream water, by plugging the nares of fish with agarose, arching flexion of the trunk, the onset of which begins and after coating plankton individually with agarose, primarily when the plankton has traversed 25-50% the procedures all designed to mask or eliminate chemo- length of the rostrum [36]. At the extremes, a strike may sensory signals. Turbulent water flow and the use of be initiated before the plankton reaches the tip of the agarose-encapsulated plankton were procedures con- rostrum or be delayed until after passing the mouth. trolling for hydrodynamic plankton signals. In each of For the latter, detection probably involves ampullae these experiments the capture profile was qualitatively located on the operculum, with fish reversing direction equivalent to the results shown in Fig. 2A. For quanti- or turning around to chase the plankton. tative comparison the capture coordinates were con- In analyzing feeding behavior, we have measured the verted into radial distances from the center of the position (prior to capture) of over 10,000 plankton, rostrum and plotted as normalized histograms of cap- both Daphnia and Artemia, relative to the long axis of ture frequency versus distance [44]. Slight differences the rostrum [37,43,44]. A representative sample of free- were noted for the background plankton extract and swimming Daphnia (Fig. 2A), from experiments con- nares block experiments. Otherwise there were no 366 L.A. Wilkens et al. Journal of Physiology - Paris 96 (2002) 363-377 significant differences between capture distributions in The detection and capture of plankton encapsulated the various control treatments and in unobstructed live in agarose is equivalent to the experiments first feeding. By another measure, feeding was enhanced by demonstrating passive electrosensory feeding in which the chemical extracts, as indicated by higher rates of sharks detected flatfish encased in an agar chamber [17]. plankton capture. A further indication of the high sensitivity of the

Fig. 1. Plankton feeding in juvenile paddlefish. A sequence of seven video frames showing the fish swimming in place in the artificial stream with the approach of a single Daphnia (white spot). Lateral and ventral views are shown in the left and right columns, respectively. Zero time (in ms) in the capture sequence is shown in the second frame, when fin movements signal the initial response of the fish to the plankton. All captures were analyzed based on the location of plankton in a vertical reference plane at the rostrum tip, shown here at t = -266 ms. Vertical lines in the reference frames illustrate the vertical and horizontal distances used to quantify capture distributions. Frames have been computer-enhanced to increase contrast and highlight the plankton. The plankton is obscured in the ventral view after t = 0 by the lateral movement of the fish. L.A. Wilkens et all Journal of Physiology - Paris 96 (2002) 363-377 367 paddlefish electrosense was demonstrated by feeding strikes in the dark, with preference under various con- choice experiments where equal numbers of encapsu- ditions for frequencies between 5 and 15 Hz. This fre- lated plankton and empty agarose particles were pre- quency preference overlaps closely with the low- sented in suspension. In experiments with Daphnia, a frequency signals recorded from Daphnia (maximum small number (21) of empty particles were captured power at 3 and 7 Hz), the natural prey of the paddlefish, initially, mostly within 10-20 mm of the rostrum, but and with the frequency-sensitivity of the paddlefish the paddlefish soon limited feeding strikes to agarose electrosensory afferents (see below). particles containing live plankton (368). Similar results A final set of experiments that supports the electro- were obtained with encapsulated Artemia, where only sensory feeding hypothesis has shown that prey capture three agarose particles were captured compared to 144 is directly affected by other, extraneous electrical signals with plankton (Fig. 2B). These results demonstrate not in the near-field environment. In these experiments only a very high sensitivity but also an ability to dis- paddlefish feeding (in the dark) was analyzed with criminate inanimate particles from 'live' particles. respect to electrical noise introduced as a uniform field Each of these experiments provides indirect evidence across the viewing chamber of the artificial stream [37]. for electrosensory-based planktivory. Two additional Noise intensities were presented between 0.05-μV cm-1 sets of experiments provide direct evidence for electro- r.m.s. (low-pass filtered at 50 Hz), a range encompassing sensory feeding. As with the shark, paddlefish were the threshold sensitivity of the ampullae. Low noise presented artificial electrical signals, here sinusoidal levels, 0.1 μV cm-1 or less, had little effect on plankton currents simulating the oscillating dipole fields of the capture. Higher intensity noise (>1.0 μV cm-1) limited plankton [43,46]. These stimuli readily elicited feeding prey capture to short distances from the rostrum, with feeding nearly abolished at 50 μV cm-1. Thus, whereas chemo- and mechanosensory background noise did not interfere with feeding, electrical noise did. Of particular interest, these experiments demonstrated that at inter- mediate levels (0.2-1.0 μV cm-1) noise actually enhanced plankton feeding, with an optimal result at 0.5 μV cm-1. Overall, the relative effect of electrical noise on feeding is non linear. This relationship, referred to as stochastic resonance (SR), has now been shown to extend to behavioral systems. As a non-linear dynamical phe- nomenon, SR defines an optimum noise level whereby the information content of a weak signal is enhanced. Low noise is without effect, whereas responses are overwhelmed by the noise at high intensities. Previously, biological SR effects have been demonstrated only in sensory systems, e.g., as increases in signal-to-noise ratios of the response outputs [5,7,21]. For paddlefish feeding, optimum noise increases the vertical range of captured plankton, a statistically significant increase in the variance of plankton distribution. This expansion of the capture distribution is interpreted as an increase in plankton 'detectability', as opposed to an increase in feeding rate (e.g., increased motivation) [36].

3.2. Avoidance behavior -60 -40 -20 0 20 40 60 Horizontal Distance (mm) In their paper describing the ampullae of Lorenzini, Fig. 2. Scatterplots of plankton distributions prior to capture in feed- Jorgensen et al. [16] noted briefly that paddlefish avoid ing experiments. All results were obtained in the dark under IR illumination. (A) Pooled results for free-swimming Daphnia. The an iron tube inserted in their tanks, but not comparably approximate cross-sectional size of the rostrum is indicated by the sized wooden rods, a behavioral sensitivity first investi- hatched ellipsoid at the intersection of the horizontal and vertical axes. gated in the catfish [31]. Having observed dramatic Modified from Ref. [44]. (B) Pooled results of experiments in which startle responses to metal rods by paddlefish, we decided equal numbers of agarose-encapsulated plankton (Artemia, circles) to quantify the avoidance response [14]. and agarose particles (diamonds) were presented. Encapsulated plankton were taken at a rate nearly 50 times greater than agarose Lowering an aluminum obstacle into the predicted particles. trajectory of a fish while it was on the opposite side of 368 L.A. Wilkens et all Journal of Physiology - Paris 96 (2002) 363-377 the tank caused the fish to turn sharply away at dis- after tag implant. In recent experiments by D. Hildreth tances up to 38 cm (Fig. 3). No fish approached within (unpublished), we have repeated the plankton feeding less than 10 cm resulting in a successful avoidance in all experiments and found similarly that the wire tag trials. In contrast, fish approached a plastic obstacle of implants have no effect on the electrosensory detection the same dimensions as if unaware of it, often colliding of plankton, based on the distribution of captures. with the obstacle. The mean minimum approach distance (MAD) of ~2 cm for plastic was significantly 3.3. Electrosensitivity different from that for the aluminum rod (MAD = 22.1 cm), as was that for the number of collisions. An elec- The ampullae of Lorenzini of paddlefish are very trically insulated aluminum rod resulted in MADs and much like those of the elasmobranchs in their physi- collisions equivalent to those for the plastic rod. How- ological as well as anatomical features, except for the ever, fish approached closer and 'collided' more fre- short canal length of the ampullae. As with the ampul- quently with the target area in control trials without an lary receptors of sharks, skates, and rays [24] and the obstacle. The differences although small were sig- more closely related sturgeon [40], paddlefish ampullae nificant, an effect due possibly to the detection by lateral are excited by a cathodal stimulus and inhibited by line receptors of water motions produced by the anodal stimulation. For example, a vigorous phasic- approaching fish that were reflected from the object. tonic burst of spikes is elicited in the primary afferent Compared to plankton, the electrical potential of the nerve at stimulus onset for a negative DC voltage step aluminum rod was large, varying from 2.73 mV at 1 cm [43]. The initial burst adapts rapidly, within 100 ms, and to 4.7 μV at 38 cm, the greatest MAD [14]. At 1 cm the for a sustained stimulus the response adapts completely aluminum potential is more than 100 times greater than over a period of several seconds. Thus, paddlefish corresponding values for Daphnia (~25 μV). The alu- receptors are not well suited for detecting steady state minum potential at 38 cm is within the range of thresh- potentials. Cathodal stimulus offset produces an inhibi- old sensitivity of the paddlefish ampullae, as estimated tory pause. Anodal stimuli elicit the corresponding off- from plankton potentials extrapolated for capture dis- on response, a near mirror image of the cathodal on-off tances up to 8-9 cm. response. In effect, the cathodal off response following In managing paddlefish populations and assessing the adaptation is equivalent to the onset of an anodal effect of migration on restocking efforts, recapture stimulus. studies have utilized small coded-wire tags inserted into Paddlefish electrosensory afferents are spontaneously the tip of the rostrum for future identification [30]. Low active in the absence of electrical stimulation. The firing recapture rates [10] prompted a second series of avoid- of action potentials is periodic, with fundamental fre- ance experiments to determine whether the wire quencies varying from 25-85 Hz, although most fall implants had any effect on the electrosensory function within the range of 40-60 Hz [27,32]. Time-interval his- of the ampullae [13]. None were found, as there was no tograms of spontaneous activity most often exhibit significant difference between fish tested prior to and sharp peaks with narrow half-height widths (Fig. 4A, left trace), e.g., a mean width of 5.6±3 ms for 22 primary afferents [32]. However, some cells have broader interval histograms (Fig. 4A, middle trace) or multiple peaks (Fig. 4A, arrow in right trace). These data are indicative of a noisy oscillator at or near spike threshold, with the noise responsible for the variation in histogram width, and the cycle skips for the resultant bimodal distributions. This interpretation is consistent with the model of Braun et al. [1] based on the ampullary receptors of the dogfish Scyliorhinus canicula. According to this model, endogenous activity is viewed as arising from a periodic oscillator in the dendritic terminals of primary afferents and subject to noisy

Fig. 3. Avoidance responses of paddlefish as illustrated by swimming synaptic input from the voltage-sensitive epithelial tracts recorded in the vicinity of an obstacle. Fish approaching the receptors. target area from either direction as they swim along the tank perimeter Further evidence for this model has been obtained in make abrupt turns in response to a metal obstacle (bold unbroken an analysis of spontaneous activity from paddlefish traces), and frequently collide when the obstacle is plastic (thin electroreceptors [26]. Power spectra of these spike trains unbroken) or plastic-coated metal (thin dashed). In the absence of an obstacle fish frequently swim through the target location (bold dashed show peaks at the fundamental oscillator frequency plus trace). Traces were selected to illustrate typical responses. Modified additional peaks correlated with potentials that can be from Ref. [14]. recorded in the lumen of the ampullae. These canal L.A. Wilkens et al. / Journal of Physiology - Paris 96 (2002) 363-377 369

oscillations arise from the receptor cells lining the base [26] suggests that endogenous firing in paddlefish pri- of the ampullae and are similar to those previously mary afferent neurons is the result of dual oscillators, reported in the skate [4,22]. Canal oscillations in the with an output representing a periodically-driven non- paddlefish range from 25-35 Hz and show strong linear oscillator with activity peaks at both fundamental coherence with peaks at those frequencies in the spike and driving frequencies. Further analysis [26] suggests train records. Thus, the work of Neiman and Russell that stronger coupling between the driving signal, i.e.,

Fig. 4. Spike train analysis of electroreceptor primary afferents. (A) Interspike interval histograms of spontaneous activity for three cells firing at mean spike rates of 62, 25, and 53 Hz (left to right), with the latter exhibiting a small peak at twice the interspike interval due to skips in firing at the fundamental frequency. (B1) Power spectrum for a cell firing spontaneously with maximum power at 75 Hz. (B2) Same cell as in B1 entrained by a weak 10 Hz periodic stimulus. (C) Frequency-response curves for two cells (circles, triangles). Frequency sensitivity is represented as the first har- monic of the instantaneous spike frequency (left ordinate) or as relative modulation of mean spike rate at stimulus frequency (right ordinate). (D) Frequency response curve for a cell based on the maximum firing rate less minimum firing rate per stimulus cycle over a range of stimulus frequencies. Note cells in C and D exhibit overlapping peak sensitivity at low frequencies in the range of 5-20 Hz. Data in A from Ref. [32]; unpublished results from X. Pei (B, C) and D. F. Russell (D). 370 L.A. Wilkens et al. / Journal of Physiology - Paris 96 (2002) 363-377

the noisy canal oscillations, and the afferent oscillator studies with catfish estimate threshold sensitivity at may explain the broader, noisier peaks in the interspike approximately 1μV cm-1 [33]. In paddlefish exact beha- interval histograms of some paddlefish electroreceptors. vioral thresholds are difficult to obtain since these fish The spontaneous activity of paddlefish electro- are ram ventilators and swim continuously [3]. As a receptors is readily entrained by weak periodic stimuli, result they do not establish a home base or territory as shown previously in response to the oscillating field from which to accurately position stimulating electro- potentials of Daphnia [43]. Entrainment by direct elec- des. However, sensitivity can be extrapolated from both trical stimulation with weak sinusoidal waveforms can feeding and avoidance behaviors. For example, Daphnia also be seen in power spectra of electrosensory activity. exhibit dipole surface potentials of 0.1-1.0 mV [43,45]. For example, a cell firing spontaneously at 75 Hz Assuming a cubic falloff with distance, field potentials

(Fig. 4B1) is entrained to a weak 10 Hz sine wave sti- detected at 10 cm, the approximate maximum distance mulus (Fig. 4B2), as indicated by the sharp 5 peak at for feeding detection, would be on the order of hun- stimulus frequency and elimination of the somewhat dreds of nV cm-1. In another measure of sensitivity it noisy endogenous peak. However, sensitivity is limited has been shown that electrical noise enhances feeding at primarily to low-frequency electrical signals [32,43], a amplitudes in the range of 0.2-1.0 μV cm-1 r.m.s. with characteristic property of ampullary receptors [2,23]. maximum effect at 0.5 μV cm-1 r.m.s. [37]. This indi- Peak sensitivity for paddlefish is in the range of 5-10 cates that paddlefish discriminate threshold events Hz, with a gradual decrease in sensitivity down to 0.5 varying in amplitude by less than 1.0 μV cm-1. Also, the Hz or less and an abrupt decrease for frequencies above cathodal responses reported previously [43] in response 20 Hz (Fig. 4C, D). Thus, the tuning characteristics of to voltage gradients of 5 μV cm-1 produced strong paddlefish electroreceptors closely match the frequency- suprathreshold responses that suggest lower threshold specific feeding strikes elicited by an artificial electric sensitivity. dipole current [43]. That paddlefish electroreceptors are sensitive to weak Several lines of evidence suggest that paddlefish elec- electrical signals such as would be useful in prey capture troreceptors are among the most sensitive of freshwater is best illustrated by the response of these electro- fish, with a threshold of 1 μV cm-1 or less. Behavioral receptors directly to plankton. In addition to the

Fig. 5. Dynamic sensitivity of an electroreceptor primary afferent to live Daphnia swept over its receptive field at 5.4 cm s-1, ~5 mm above the surface of the rostrum. Each panel shows a raster plot and peristimulus time histogram for 50 sweeps in the anterior-posterior direction (time in ms). Panel insets show plankton orientations for responses at 90° intervals of rotation. Inhibitory and excitatory responses at ±90° orientations corre- spond to a tail-negative and head-positive dipole field about the plankton. Ventral and dorsal plankton surfaces elicit mirror-image biphasic responses with excitation and inhibition again reflecting the dipole field of the advancing plankton. Modified from Ref. [45]. L.A. Wilkens et a/./Journal of Physiology - Paris 96 (2002) 363-377 371 synchronization of primary afferents with the oscillating sensory afferents innervate the dorsal part of the DON. field potentials of Daphnia held stationary near the The DON, however, receives a number of other inputs receptive field [43], recent experiments have shown that as well. After tracer application into the DON, retro- primary afferents respond strongly to plankton swept gradely labeled cells were found in the contralateral more naturally past their receptive fields, and that DON (Fig. 7B), in the nucleus preeminentialis (Fig. 7C), responses are specific for the orientation of the plankton and in the cerebellar auricles (Fig. 7D). (Fig. 5). Each of the four panels illustrate raster profiles Ascending fibers of the DON terminate in three of spikes and peristimulus time histograms for a plank- mesencephalic areas, the mesencephalic tectum (TM, ton rotated at 90° intervals in the sagittal plane and Fig. 7E), the torus semicircularis, and the lateral swept over the receptive field. With a different surface of mesencephalic nucleus. None of these areas appears to the plankton facing the rostrum, e.g., the head, tail, and have a direct projection back to the DON. Instead, dorsal or ventral surface, a different electrical signature tracer applications into the TM revealed a projection to is presented corresponding to the dipole electric field of the nucleus preeminentialis. Cells of this nucleus either the plankton [46]. In response, the spontaneous activity project directly to the DON or indirectly via the cere- is inhibited (upper panel) when the caudal end of the bellar auricles. The latter projection via the cerebellar plankton faces the rostrum whereas the head end of the auricles reaches the ventral part of the DON, which is plankton elicits an excitatory burst (lower panel), dipole termed cerebellar crest. Here, parallel fibers of the orientations equivalent respectively to anodal and granular cells in the auricles make contact with the cathodal stimulus polarities. Mirror image biphasic ventral dendrites of the large principle ascending cells in responses are elicited by dorsal and ventral plankton the DON. The cerebellar crest is much larger than the orientations. dorsal part of the DON where the primary electrosensory afferents terminate. 3.4. Distribution and innervation of the ampullae These descending feedback loops are very similar to other electrosensory fishes and may be important for The ampullae are distributed all over the rostrum and electrosensory information processing in general. The head, including the gill covers, but no electroreceptors ascending projections of the DON, however, are quite were found on the trunk. There are up to 75,000 different in the paddlefish compared to either the elas- ampullae [25], which is by far the highest number found mobranchs or the teleosts [15]. Whereas in elasmo- in any electrosensory animal. The ampullae are arran- branchs, the lateral mesencephalic nucleus and the TM ged in clusters of 6-20 (Fig. 6A,B). In a preliminary are innervated by DON fibers, only the torus semi- study, the innervation of these clusters has been exam- circularis receives input from the DON in teleosts. The ined by means of nerve stains with sudan black. The paddlefish, however, is the only species with DON pro- number of fibers innervating each cluster is lower than jections to all three mesencephalic targets (Fig. 7F). The the number of ampullae in the cluster. In Fig. 6C, a projection to the TM is particularly well developed and cluster of eight ampullae is shown. Fig. 6D is a close-up the topography of the DON-TM projections is currently of the nerve fibers innervating this cluster. Three fibers under investigation. In fishes and amphibians, the TM is leave the main bundle and split up into several smaller the most important structure for orienting movements fibers that course toward the ampullae. This means that and prey catching [6,41], which suggests that the projec- the information from several ampullae converges onto a tion from DON to TM in the paddlefish may be involved few nerve fibers. in the electrosensory guided prey catching behavior. The central projections of the electrosensory afferents were studied by means of biotinylated dextran amine injections into branches of the anterior lateral line nerve 4. Discussion (ALLn). The cell bodies of the primary afferents are located in large ganglia within the roots of the ALLn 4.1. Plankton feeding in juvenile paddlefish (Fig. 6E). The central processes of the electrosensory afferent enter the brain in the dorsal medulla and Paddlefish feed primarily on crustacean zooplankton innervate the dorsal octavolateral nucleus (DON) throughout their entire life cycle. As described here, (Fig. 6F, see also Fig. 7A). The DON is the sole target juveniles are particulate feeders selecting and capturing of the electrosensory efferents. individual planktonic prey, a feeding mechanism adop- ted in the early larval stages as the yolk sac is depleted. 3.5. CNS electrosensory circuits As they grow paddlefish switch to a non-selective suspension feeding mechanism that uses the same plank- Within the CNS, tracer was applied to the dorsal tonic resource. Yet, there is no metamorphic octavolateral nucleus (DON), the cerebellar auricles, and reorganization of the mouth and buccal apparatus from the mesencephalic tectum (Fig. 7A). Primary electro- larva to adult, except for the development of the gill 372 L.A. Wilkens et al../ Journal of Physiology - Paris 96 (2002) 363-3 77

Fig. 6. Electroreceptors and their innervation in the paddlefish. (A) Distribution of ampullae on the lower side of the rostrum. Scale bar 2 mm. (B- D) Photomicrographs of the innervation of the ampullae in the skin after clearing with peroxide and staining with sudan black. Scale bar in B 1 mm. (C) higher power magnification of one cluster consisting of eight ampullae. Scale bar 200 μm. (D) Close up of the nerve fibers innervating the cluster shown in (C). Three fibers leave the bundle and split up into several branches to innervate the eight ampullae of the cluster. Scale bar 40 μm. (E) Ganglion cells of the lateral line nerve were retrogradely filled after tracer application into the lateral line branch which innervates the ampullae of the rostrum. Scale bar 100 μm. (F) Central processes of the electrosensory afferents terminating in the dorsal octavolateral nucleus of the medulla. Scale bar 200 μm.

rakers. Planktivory in Polyodon is distinct from that of ram feeding mechanism in juvenile paddlefish, the its close relatives, e.g., the piscivorous Chinese paddle- capture sequence in Fig. 1 has been modified by aligning fish and the mostly bottom-feeding sturgeons. the plankton along a vertical axis (Fig. 8). This is a Both particulate and suspension feeding mechanisms more natural representation of feeding since the plank- are consistent with ram ventilation in the paddlefish, ton is effectively stationary and the fish is viewed as if i.e., a form of "ram feeding" whereby fish rely less on a swimming forward. Note that the swimming velocity of rapid lunge and/or oral suction to capture prey as the paddlefish is relatively constant. However, the shape opposed to simply gulping or straining plankton that of the paddle requires a sequence of complex motions they encounter while swimming. To better illustrate the when plankton are not in line with the swim path. As L.A. Wilkens et al. / Journal of Physiology - Paris 96 (2002) 363-377 373

illustrated (Figs. 1 and 8), paddlefish turns involve a unless in association with long extensions of the jaws combination of roll and yaw movements allowing the and olfactory organs. Various functions have been rostrum to slice through the water with least resistance attributed to the rostrum, most associated with its pre- in bringing the mouth toward the prey. sumed use as a spatula for extricating food from the substrate, as reflected in the common misnomers 4.2. Novelty of the passive electrosense in plankton feeding 'spoonbill' and 'shovelnose cat'. A digging role would be incompatible with the sensory nature of the rostral The paddlefish is unique in many respects, not the epithelium, its ampullary receptors and associated least being the presence of the elongated rostrum. Such innervation. The somewhat delicate structure of the a prominent rostral structure is found in no other fish rostrum is evident in the calluses that form at the tip of

Fig. 7. (A) Dorsal view of the brain of the paddlefish (scale bar 3 mm). (B-E) Cross sections through the brain at levels indicated in (A). (B-D) Unilateral tracer application into the dorsal octavolateral nucleus (DON) retrogradely filled cells in the contralateral DON (B, scale bar 200 μm), in the nucleus preeminentialis (C, scale bar 200 μm), and in the cerebellar auricles (D, scale bar lOOμm). (E) A strong projection of cells in the DON reaches the mesencephalic tectum (scale bar 200 μm). (F) Schematic drawing of the ascending (solid lines) and descending (dashed lines) projections of the electrosensory system in the paddlefish. Abbreviations: BO: bulbus olfactorius; Cer: cerebellum; DON: dorsal octavolateral nucleus; Imn: lateral mesencephalic nucleus; nLLd: dorsal root of the lateral line nerve; Npe: nucleus preeminentialis; Tel: telencephalon; TM: mesencephalic tectum; TS: torus semicircularis. 374 L.A. Wilkens et al. / Journal of Physiology - Paris 96 (2002) 363-377 the rostrum in fish that routinely bump into the sides of electroreception [20,28]. Other fish that incorporate the our holding tanks. It has been suggested also that the passive electrosense in their feeding strategies, e.g., paddle serves as a pitch stabilizer in swimming [19], an elasmobranchs and silurids, rely as well on additional hypothesis as yet untested. Our work with the paddlefish sensory modalities, including vision, olfactory homing, has shown that the rostrum does play a role in feeding, taste and tactile (barbels), and hydrodynamic cues. In but in a far different capacity than as a tool for excava- addition, these fish target mostly macroscopic, as tion. Rather, the rostrum serves as an antenna, in con- opposed to near-microscopic prey, as first described in junction with its population of electrosensory ampullae. the shark [17]. A second novel feature of the paddlefish is that they The paddlefish is one of only two surviving species in detect and capture zooplankton, their primary food, by the Polyodontidae, a family containing numerous fossil passive electroreception of planktonic electric fields. representatives [12] and closely related to sturgeon. Most planktivores are sight feeders [9], although weakly These mostly cartilaginous fish are considered primitive electric fish detect and capture plankton using active by most criteria, yet Polyodon is highly adapted as no other fresh water fish for plankton feeding through a combination of specialized features. As ram ventilating fish they are in constant motion through the water and its suspended planktonic food resources. The mouth and buccal structures open to expansive proportions for feeding, which results in ram engulfment of slow mov- ing plankton, whether by selective capture (particulate feeding) by juveniles, as we have studied, or by non selective filtration (suspension feeding) in larger fish. The key to the longevity and success of this ancient species is undoubtedly tied to the antennal function of the rostrum, which takes the place of vision as the primary means for prey detection [44]. The eyes, placed in the head at the base of the rostrum and directed ven- trolaterally, do not provide views either in front or above the fish, as evident in captive fish that fail to detect and frequently collide with the walls of their tanks. In contrast, paddlefish are adept at avoiding objects with an electrical signature [14], a behavioral extension of the rostral electrosense aside from its primary role in feeding. Thus, the rostrum and the passive electrosense have co-evolved as a specialized and highly sensitive apparatus designed to detect small prey, an important adaptation for planktivory in a non-visual, turbid habitat. Whereas the rostrum gives the appear- ance of a clumsy appendage in confinement, where fish bump into tank walls and get stuck in corners, it is exquisitely designed to function in its natural environment. Its elongated spatulate structure provides substrate for an extensive population of distributed receptors that in turn provide dynamic information centrally as they scan the aquatic environment for approaching plankton, information used for the spatial localization of their prey. Finally, the rostrum is ideally positioned in front of the fish. This is especially relevant for a ram ventilating Fig. 8. Lateral view of the capture sequence equivalent to feeding in fish in constant motion, providing ample reaction time the open environment. Panels from Fig. 1 have been adjusted to align for directing feeding strikes at plankton while they are the plankton in the vertical sequence. Thus, the fish is viewed as mov- in proximity with the mouth, thus avoiding the less effi- ing forward normally to engulf the plankton. Swimming velocity cient motions of reversing direction and/or interruption throughout the sequence is relatively constant, as indicated numeri- cally (mm s-1) between panels. Velocity was calculated with reference of forward ram swimming. This is in stark contrast with to the horizontal position of the eye and relative time interval between the active electrosensory scanning of plankton by weakly frames. electric fish. As described in Apteronotus albifrons L.A. Wilkens et al. j Journal of Physiology - Paris 96 (2002) 363-377 375

[20,28], scanning involves swimming backward past the In portions of their former range paddlefish are either prey, a direction that once again brings the mouth into extinct or endangered [11]. In Missouri, two dams cut proximity with the prey. Reverse scanning by these off access to spawning sites and prompted the develop- more sedentary species and forward scanning in the ment of aquaculture techniques and restocking pro- paddlefish are complimentary 'approaches' to electro- grams. To gauge the success of restocking and sensory feeding by very dissimilar fish. determine patterns of long-distance migration, hatchery The use of the passive electrosense in plankton feed- fish are implanted rostrally with coded wire tags at time ing by the paddlefish is a novel application among of release. Due to low recovery rates and concerns that aquatic organisms that possess the electric sense. So far, the electrosense was compromised, we tested both our conclusions are based solely on juvenile fish feeding plankton feeding and avoidance behavior to see if either by selective capture. We assume that larger fish feeding was influenced by the wire tag [13, D. Hildreth, unpub- non-selectively by straining plankton from the water lished]. No effect was found, a result removing tagging also rely on electrical cues. Supporting this prediction, from consideration in evaluating recapture success. larger fish begin to filter feed when they first encounter However, the fact that paddlefish are sensitive to and live plankton, as observed during daily feeding of our fish avoid relatively small metal objects raises concern about stocks. This feeding behavior involves accelerated swim- the presence of large metal structures in the habitat, i.e., ming and more frequent turns, in addition to the wide dams and weirs. For instance, there are 26 dams on the opening of the mouth and expansion of the buccal cavity. Mississippi River upstream from St. Louis, a major However, in the wild large fish are occasionally netted migratory route for paddlefish. Paddlefish reluctance to in good condition that have lost their rostrum, in most pass through dams where flow is regulated by partially cases as a result of mutilation by a boat propeller. That closed metal gates has been documented [39]. The fish survive without their rostrum seems to argue blockade of regional spawning sites or interruption of against the assertion that the rostral electrosense is key long-distance migrations, and therefore gene flow to their survival. This apparent contradiction is miti- among populations, is probable. In addition, seasonal gated by the fact that only large fish have been observed water flow to shallow backwaters along the rivers is without paddles. Fecundity surveys of young-of-year often regulated by metal structures. Natural flooding of paddlefish cohorts involving thousands of fish [8,38] these areas produces valuable plankton blooms una- have never reported small fish absent their rostrum. vailable to paddlefish if they avoid passages with metal This suggests that while large fish survive loss of the obstacles. Further analysis of paddlefish avoidance rostrum, small fish do not, with the implication that the behavior may be warranted for assessing management passive rostral electrosense is an adaptation benefiting strategies. primarily larval and juvenile fish. In support for this interpretation, we have observed healthy robust fish in 4.4. The paddlefish as a model for passive electrosensory the post yolk-sac larval stages measuring 20-50 mm in systems total length that suddenly appear gaunt if feeding is interrupted for only a few hours. These fish invariably In depth behavioral and electrophysiological research die. Larval paddlefish begin life as ram ventilators, on paddlefish would have been nearly impossible prior swimming with rapid, inefficient trunk undulations that to their availability from hatcheries. Knowledge of the appear to rapidly exhaust metabolic stores in the life history of these fish was incomplete until the early absence of feeding. As larvae, ampullae are profuse, 1960s [34] due in large part to the difficulty of sampling covering the head and opercula prior to development of juvenile paddlefish in the wild. At present paddlefish are the rostrum [47]. We conclude that efficient feeding in only available on an annual basis since they spawn once the early life stages, larval through juvenile, is critical in the spring, and grow rapidly. Nevertheless, as fresh- for survival and that the early development of the ros- water fish they can be cultured relatively easily and tral electrosense plays a key role. Successful feeding in trained to eat commercial fish pellets, although large adult fish without a rostrum may show reliance on the volume tanks and good water quality are essential. substantial number of ampullae that remain on the As a result, the paddlefish is a relatively recent addi- head, opercula, and lower jaw and/or the possibility tion to the menagerie of aquatic organisms under that schooling allows rostrally-challenged fish to find investigation for the specialized electric sense. Accord- adequate plankton resources. ingly, we know little about the neural mechanisms, sensory to motor, that underlie electrosensory behavior in 4.3. The paddlefish electrosense and fisheries management paddlefish and future research options are many. How- ever, the role of the electrosense in paddlefish biology Paddlefish once occupied rivers throughout the Mid- seems clear in serving a function essential for survival, west United States including the Mississippi, Missouri, namely feeding. The novel role of passive electrorecep- and Ohio Rivers and their larger tributaries in 22 states. tion in plankton feeding is of intrinsic interest for 376 L.A. Wükens et all Journal of Physiology - Paris 96 (2002) 363-377 comparative neuroethology, but the paddlefish offers [2] T.H. Bullock, An essay on the discovery of sensory receptors and additional experimental options. the assignment of their functions together with an introduction to Questions of a general nature that can be addressed electroreceptors, in: A. Fessard (Ed.), Electroreceptors and other Specialized Receptors in Lower Vertebrates, Springer, New York, with the paddlefish model include (i) What are the bio- 1974, pp. 1-12. physical properties and structural organization of the [3] W.W. Burggren, W.E. Bemis, Metabolism and ram gill ventila- peripheral receptors that underlie high sensitivity to tion in juvenile paddlefish, Polyodon spathula (Chondrostei: weak electrical fields? (ii) How are peripheral receptors Polyodontidae), Physiol. Zool. 65 (1992) 515-539. represented topographically in the brain, e.g., the [4] W.T. Clusin, M.V.L. Bennett, The oscillatory responses of skate electroreceptors to small voltage stimuli, J. Gen. Physiol. 73 ampullae of Lorenzini distributed over the rostrum and (1979) 685-702. head in Polyodon! (iii) How is sensory input trans- [5] JJ. Collins, Noise-enhanced tactile sensation. Nature 383 (1996) formed centrally to signal the path and trajectory of 770. objects in electrical space? (iv) What form do novel [6] U. Dicke, G. Roth, Tectal activation of premotor and motor evolutionary adaptations take in shaping the neural networks during feeding in salamanders, Eur. J. Morphol. 32 (1994) 106-116. architecture of the brain? An illustration of the utility of [7] J.K. Douglass, L. Wilkens, E. Pantazelou, F. Moss, Noise this preparation is the use of plankton feeding to inves- enhancement of information transfer in crayfish mechan- tigate the stochastic dynamics in sensory physiology and oreceptors by stochastic resonance, Nature 365 (1993) 337-340. behavior [37]. The electrosense is an ideal medium since [8] J.P. Fredericks, D.L. Scarnecchia, Use of surface visual counts electrical noise can be introduced in its purest form, for estimating relative abundance of age-0 paddlefish in Lake Sakakawea, N. Amer. J. Fisheries Management 17 (1997) 1014— easily regulated and quantified. Also, the paddlefish is a 1018. robust cartilaginous preparation making the brain [9] S.D. Gerking, Particulate feeding, Chapter 6, in: S.D. Gerking readily accessible for experimental study. This feature (Ed.), Feeding Ecology of Fish, Academic Press, San Diego, has been used to great advantage in our initial tracing of 1994, pp. 111-138. electrosensory pathways in the brain [15]. We now know [10] J.M. Grady, G.A. Conover, Mississippi Interstate Cooperative Research Association Mississippi River Basin Paddlefish the brain circuitry involved in electrosensory information Research Coded-Wire Tagging Project: 1997 Annual Report. processing and are able to investigate the function of the MICRA, 1998, Bettendorf, Iowa different electrosensory pathways by studying the effect [11] K. Graham, Contemporary status of the North American pad- of selective lesions on the two well defined and quantifi- dlefish, Polyodon spathula, Environ. Biol. Fishes 48 (1997) 279- able behavioral tasks, prey capture and object avoidance. 289. [12] L. Grande, W.E. Bemis, Osteology and phylogenetic relation- Paddlefish, members of an ancient group of acti- ships of fossil and recent paddlefishes (Polyodontidae) with com- nopterygian fish, the Chondrostei, have an exceptionally ments on the interrelationships of Acipenseriformes, J. Vert. well developed electrosensory system. Its investigation Paleontol. 11 (1991) 1-121. and comparison with other electrosensory fishes may [13] C. Gurgens, Electrosensory Avoidance Behavior of the Paddle- give important insights into the central mechanisms fish (Polyodon spathula) and the Effect of Coded-Wire Tags, Master's thesis, University of Missouri-St. Louis, 1998. underlying passive electroreception. Furthermore, a [14] C. Gurgens, D.F. Russell, L.A. Wilkens, Electrosensory avoid- greater understanding of the paddlefish electrosense will ance of metal obstacles by the paddlefish, J. Fish. Biol. 57 (2000) add to our knowledge concerning the evolution of a 277-290. specialized yet basic aquatic sensory modality. [15] M.H. Hofmann, W. Wojtenek, L.A. Wilkens, Central organization of the electrosensory system in the paddlefish (Polyodon spathula), J. Comp. Neurol. 446 (2002) 25-36. [16] J.M. Jorgensen, A. Flock, J.Z. Wersäll, The Lorenzinian ampul- Acknowledgements lae of Polyodon spathula, Z. Zellforsch. 130 (1972) 362-377. [17] A.J. Kalmijn, The electric sense of sharks and rays, J. Exp. Biol. Special thanks are due to the Blind Pony Fish Hatch- 55(1971)371-383. ery, Missouri Department of Conservation, for provid- [18] H.D. Kistler, The primitive pores of Polyodon spathula, J. Comp. Neurol. 16 (1906) 294-298. ing paddlefish throughout this project. We also thank [19] K.F. Lagler, J.E. Bardach, R.R. Miller, Ichthyology, John Wiley Tamara Wassoer for expert assistance in histological and Sons, New York, 1962. experiments and Frank Moss for comments on the [20] M.J. Lannoo, S.J. Lannoo, Why do electric fishes swim back- manuscript. This research was supported in part by wards? an hypothesis based on gymnotiform foraging behavior grants from the Department of Defense (URISP-ONR), interpreted through sensory constraints, Env. Biol. Fish 36 (1993) 157-165. the Department of Energy, and the National Science [21] J.E. Levin, J.P. Miller, Broadband neural encoding in the cricket Foundation-German Academic Exchange Service. cereal sensory system enhanced by stochastic resonance, Nature 380(1996) 165-168. [22] J. Lu, H.M. Fishman, Localization and function of the electrical References oscillation in electroreceptive ampullary epithelium from skates, Biophys. J. 69 (1995) 2458-2466. [1] H.A. Braun, H. Wissing, K. Schäfer, M.C. Hirsch, Oscillation [23] J.C. Montgomery, Frequency response characteristics of primary and noise determine signal transduction in shark multimodal and secondary neurons in the electrosensory system of the sensory cells, Nature 367 (1994) 270-273. thornback ray, Comp. Biochem. Physiol. 79A (1984) 189-195. L.A. Wükens et al. / Journal of Physiology - Paris 96 (2002) 363-377 377

[24] R.W. Murray, The response of the ampullae of Lorenzini of [37] D.F. Russell, L.A. Wilkens, F. Moss, Use of behavioural sto- elasmobranchs to electrical stimulation, J. Exp. Biol. 39 (1962) chastic resonance by paddlefish for feeding, Nature 402 (1999) 119-128. 291-294. [25] H.F. Nachtrieb, The primitive pores of Polyodon spalhula (Wal- [38] D. Scarnecchia, J. Fredericks, F. Ryckman, Unlocking the secrets baum), J. Exp. Biol. 9 (1910) 455-468. of Lake Sakakawea's young paddlefish, North Dakota Outdoors [26] A. Neiman, D.F. Russell, Stochastic biperiodic oscillations in the 57(1995) 10-14. electroreceptors of paddlefish, Phys. Rev. Lett. 86 (2001) 3443-3446. [39] P.O. Southall, W.A. Hubert, Habitat use by adult paddlefish in [27] A. Neiman, D.F. Russell, X. Pei, W. Wojtenek, J. Twitty, the upper Mississippi River, Trans. Amer. Fisheries Soc. 113 E. Simonotto, B.A. Wettring, E. Wagner, L.A. Wilkens, F. Moss, (1984) 125-131. Stochastic synchronization of electroreceptors in paddlefish, Int. [40] J.H. Teeter, R.B. Szamier, M.V.L. Bennett, Ampullary electro- J. Bifurcation Chaos 10 (2000) 2499-2527. receptors in the sturgeon Scaphirhynchus platoryunchus (Rafin- [28] M.E. Nelson, M.A. Maclver, Prey capture in the weakly electric esque), J. Comp. Physiol. 138 (1980) 213-223. fish Apteronotus albifrons: sensory acquisition strategies and [41] H. Vanegas, B. Williams, E. Essayag, Electrophysiological and electrosensory consequences, J. Exp. Biol. 202 (1999) 1195-1203. behavioral aspects of the teleostean optic tectum, in: H. Vanegas [29] J.G. New, D. Bodznick, Segregation of electrorecptive and (Ed.), Comparative Neurology of the Optic Tectum, Plenum, mechanoreceptive lateral line afferents in the hindbrain of chon- New York, 1984, pp. 121-162. drostean fishes, Brain Res. 336 (1985) 89-98. [42] L.A. Wilkens, M. Cox, D. Russell, Evidence for electrosensitivity [30] J.H. Oven, Mississippi Interstate Cooperative Research Associ- in planktivorous feeding in the paddlefish Polyodon spathula, ation Interjurisdictional Paddlefish Tagging Study Protocol, Amer. Zool. 34 (1994) 36A. MICRA, Iowa, 1995. [43] L.A. Wilkens, D.F. Russell, X. Pei, C. Gurgens, The paddlefish [31] G.H. Parker, A.P. Van Heusen, The responses of the catfish, rostrum functions as an electrosensory antenna in plankton Amirus nebulosus, to metallic and non-metallic rods, Amer. J. feeding, Proc. Roy. Soc. Lond. B264 (1997) 1723-1729. Physiol. 44 (1917) 405-120. [44] L.A. Wilkens, B.A. Wettring, E. Wagner, W. Wojtenek, [32] X. Pei, D.F. Russell, L.A. Wilkens, F. Moss, Dynamics of the D.F. Russell, Prey detection in selective plankton feeding by the electroreceptors in the paddlefish, Polyodon spathula, in: J. Bower paddlefish: is the electric sense sufficient, J. Exp. Biol. 204 (2001) (Ed.), Computational Neuroscience, Plenum, , 1998, pp. 245-249. 1381-1389. [33] R.C. Peters, F. van Wijland, Electro-orientation in the passive [45] W. Wojtenek, M.H. Hofmann, L.A. Wilkens, Primary afferent electric catfish, Ictalurus nebulosus LeS, J. Comp. Physiol. 92 electrosensory neurons represent paddlefish natural prey, Neuro- (1974)273-280. computing 38 (2001) 451-458. [34] C.A. Purkett Jr., Reproduction and early development of the [46] W. Wojtenek, X. Pei, L.A. Wilkens, Paddlefish strike at artificial paddlefish, Trans. Amer. Fisheries Soc. 90 (1961) 125-129. dipoles simulating the weak electric fields of planktonic prey, J. [35] R.A. Rosen, D.C. Hales, Feeding of paddlefish, Polyodon spa- Exp. Biol. 204(2001) 1391-1399. thula, Copeia 1981 (1981) 441^55. [47] B.L. Yeager, R. Wallus, Family Polyodontidae, paddlefish, in: [36] D.F. Russell, A. Tucker, B.A. Wettring, A. Neiman, L. Wilkens, R. Wallus (Ed.), Reproductive Biology and Early Life History of F. Moss, Noise effects on the electrosense-mediated feeding Fishes in the Ohio River Drainage, vol. 1, Acipenseridae through behavior of small paddlefish, Fluctuations and Noise Letters 1 Esocidae, Tennessee Valley Authority, Chattanooga, Tennessee, L71-L86 (2001). 1990, pp. 50-55.