pollack imaizumi bio essays 99

Upload: le-duc

Post on 05-Apr-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/31/2019 Pollack Imaizumi Bio Essays 99

    1/9

    N e ural a nalysis of soundfre que ncy in inse c tsGerald S. Pollack * and Kazuo Imaizumi

    Su mmary

    Insects, like other hearing animals, must extract information from the sounds they

    hear so that they may respond appropriately. One parameter of sound that carries

    information is its frequency content. Insects analyze sound frequency to identify

    mates, to judge the distance to potential competitors, and to detect p redators and

    prey. We review how frequency is analyzed in the insect nervous system, focusing

    on two taxa in which this problem has been studied most intensively, katydids and

    crickets. BioEssays21:295303, 1999. 1999 John Wiley & Sons, Inc.

    IntroductionAn important function of nervous systems is sensory process-

    ing, that is, the extraction of behaviorally relevant information

    from the environment. Insects provide favorable model sys-

    tems for studying sensory processing, for a number of

    reasons. First, they offer the usual advantage of invertebrate

    model systems: relatively few neurons, many of which are

    identifiable. An insect nervous system may contain a few

    hundred thousand neurons (and a single segmental ganglion,

    which is often the appropriate level of analysis, may have only

    a few thousand), whereas a mammalian brain contains

    billions of neurons. Moreover, many of the neurons of insects

    can be identified by their physiology and anatomy as unique

    cells that can be recognized and studied in specimen afterspecimen. Second, many behaviors of insects are tightly

    related to specific sensory inputs in a robust manner, allowing

    one to formulate clear hypotheses about the sorts of informa-

    tion processing that might be taking place.Third, insects are a

    highly diverse group in which sensory capabilities have

    evolved repeatedly; hearing, for example, has arisen indepen-

    dently at least 14 times.(1) This diversity provides a rich

    source of material for a comparative approach that can help

    demonstrate how neural circuits are shaped by natural

    selection.

    In this review we examine the neural processing of one

    important feature of auditory stimuli, sound frequency. Al-

    though analysis of sound frequency is important in manyinsect groups, this problem, and in particular its relationship

    to behavior, has been studied most thoroughly in Ensifera, a

    suborder of Orthoptera that includes katydids (Tettigoniidae)

    and crickets (Gryllidae). These insects will form the focus of

    this review.

    Ensiferan a uditory system s

    The ears of ensiferans are situated on the front pair of legs.

    (Insect ears occur in a variety of locations that seem strange

    from a vertebrate perspective, e.g., wings, mouths, abdo-

    mens(2).) Externally, the ear consists of a pair of tympanal

    membranes, which serve as the animals physical interfacewith its auditory world. Sound causes the thin membranes to

    vibrate, and these vibrations are passed on to a population of

    sense organs, called scolopoid sensilla, which are the basic

    units of hearing. Each scolopoid sensillum consists of a

    receptor neuron and several accessory cells, and the ear

    includes a few dozen sensilla arrayed along the length of the

    leg (Fig. 1A).

    A single receptor is most sensitive, or tuned, to a

    particular sound frequency (see Fig. 2B, for examples).

    Different receptors may be tuned to different sound frequen-

    cies, and there is a systematic relationship between the

    position of the receptor within the ear and the frequency to

    which it is most sensitive,(36) similar to the tonotopic organiza-

    tion of frequency sensitivity along the basilar membrane of

    the vertebrate cochlea.(7)

    What determines the frequency sensitivities of different

    receptors? In some insects, notably mole crickets (which are

    also ensiferans) and locusts, the underlying mechanism

    resides in the tympanal membrane itself.(810) The mechanical

    Department of Biology, McGill University, Montreal, Canada. Funding

    agency: NSERC; Funding agency: Whitehall Foundation.

    *Correspondence to: Gerald S. Pollack, Department of Biology, McGill

    University, 1205 Doctor Penfield Avenue, Montreal QC H3A B1,

    Canada. E-mail: [email protected]

    Re v ie w a r tic le s

    BioEssays 21:295303, 1999 John Wiley & Sons, Inc. BioEssays 21.4 295

  • 7/31/2019 Pollack Imaizumi Bio Essays 99

    2/9

    properties of the tympanal membrane and associated struc-

    tures are such that different sound frequencies cause differ-

    ent regions of the membrane to vibrate most strongly.

    Auditory sensilla attach at different sites on the membrane,

    and thus receptor neurons acquire different frequency sensi-

    tivity. In crickets and katydids, however, the tympanum

    responds in a spatially uniform way with respect to sound

    frequency.(1114) Moreover, the scolopoid sensilla do not

    attach directly to the tympanal membrane. Instead, sound

    energy is transmitted to them through internal membranes

    and air cavities,(15,16) the mechanical properties of which are

    unknown. There are two main views about how individual

    receptors become tuned, each with some supporting evi-

    dence. One is that the mechanics of the inner ear as a whole

    (i.e., the membranes, air cavities, etc.) respond in a spatially

    nonuniform way to sounds of different frequencies, so that, for

    example, one end of the structure would be most efficientlystimulated by high sound frequencies, and the other by lower

    frequencies.(17) This sort of mechanism is similar to what

    occurs in the inner ears of higher vertebrates, where the

    mechanics of the basilar membrane results in a spatial

    separation of sound frequencies along its length.(7) The

    alternative view is that the tuning of each receptor neuron is

    determined by factors intrinsic to the scolopoid sensillum in

    which it resides, rather than by the structures that deliver

    sound energy to the sensillum.(18) Resolution of this issue will

    require detailed measurements of the movements of struc-

    tures within the inner ear as they respond to sound; such

    measurements are not yet available. An extensive treatment

    of the biomechanics of insect hearing is beyond the scope ofthis brief review. However, several more thorough discus-

    sions of this topic are available.(8,19,20)

    Although many of the acoustical details of katydid and

    cricket ears are still somewhat murky, the structural similari-

    ties suggest that the ears are similar in overall operation. The

    similarities extend into the central nervous system, where the

    same identified auditory interneurons can be recognized in

    both taxa (Fig. 1BE). It is unclear whether these similarities

    of cricket and katydid auditory systems represent an instance

    of homology(21) or of duplicate, but independent, evolution of

    similar auditory systems from the same exaptations.(22)

    Auditory behaviors of ensifera

    Foremost among the sounds that ensiferans listen to are the

    songs of their own species. Their songs are of course

    conspicuous to humans as well, the words cricket and

    katydid being onomatopoeic descriptions of the songs of

    particular species. Songs are used to attract mates, to induce

    copulation, and to declare and defend territories.(20,23) A

    second type of behaviorally relevant sound is produced by

    echolocating bats, which hunt insects by emitting ultrasonic

    cries and detecting the echoes that are reflected from the

    bodies of their prey. Many insects, including ensiferans,

    exhibit escape responses when stimulated with bat-like

    sounds, which they recognize in part by their ultrasonic

    frequencies.(1)

    The remainder of this review presents a few

    examples that illustrate how sound frequency affects behav-

    ior, and how it is analyzed in the nervous system.

    Frequency analysis in ka tydids

    One probable role of frequency analysis is estimation of

    distance to the sound source. As sound propagates through

    the environment, its spectrum changes. The attenuation of

    sound with distance increases with sound frequency, both

    because high frequencies are better blocked by small objects

    between the source and target, such as twigs and leaves, and

    because frictional loss of energy is greater for higher frequen-

    cies.(24,25) In the katydid species Mygalopsis marki, males

    situate their territories according to their acoustical judgments

    of the distances to other singing males,(26,27) and it is likely

    that they make this assessment by analyzing song spectrum.

    The song has a broad spectrum (ca. 830 kHz) close to its

    Figure 1. Ensiferan auditory systems. A: Auditory receptors

    are arrayed along the surface of a tracheal branch in the

    prothoracic tibia.(16) The ear of a katydid is illustrated, but

    cricket ears are similar in overall structure. The receptor

    neurons project to the prothoracic ganglion, where theyinteract with several interneurons, some of which are shown in

    BE. B,C: Ascending neurons from a cricket(65) (B) and a

    katydid(31) (C) . Both neurons are named ascending neuron 1

    (AN1). D,E: Local interneurons of a cricket(65) (D ) and a

    katydid(30) (E). Both neurons are designated omega neuron 1

    (ON1).

    Re v ie w a r tic le s

    296 BioEssays 2 1.4

  • 7/31/2019 Pollack Imaizumi Bio Essays 99

    3/9

    source, but the higher-frequency components become in-

    creasingly faint with distance. At a distance of 20 m, frequen-

    cies above 12 kHz are barely distinguishable from ambient

    noise, even though the lower frequency components of the

    song are still prominent.(28)

    Neural analysis of song spectrum depends both on the

    anatomical organization of the receptor neuron terminals inthe central nervous system and on circuitry among interneu-

    rons. As in most sensory systems, the central target regions

    of receptor neurons are related to their positions at the

    periphery. Auditory receptors form a tonotopic map in the

    central nervous system, mirroring the systematic relationship

    in the ear between a receptor neurons location and its

    frequency sensitivity. The receptors terminate in a region of

    the prothoracic ganglion called the auditory neuropil, where

    their branches form a curved band that starts at the anterior

    edge of the neuropil and proceeds posteriorly along its ventral

    border, eventually wrapping dorsally around the posterior

    edge and continuing anteriorly. Receptors that respond best

    to low sound frequencies terminate at the anterior end of the

    band and receptors tuned to progressively higher frequencies

    terminate at progressively posterior (and eventually anterodor-

    sal) positions(5,6,2830) (Fig. 2A). This anatomical organization

    provides the framework within which receptors make synaptic

    connections with interneurons. Different interneurons sample

    different regions of the array of receptor terminals with their

    dendrites, and thus receive input from receptors tuned to

    different sound frequencies.(30)

    Auditory receptors differ not only in frequency tuning but

    also in absolute sensitivity. Minimum thresholds of receptors

    tuned to different frequencies can vary by as much as 40 dB

    (Fig. 2B). When these sensitivity differences are consideredtogether with the distance-dependent spectrum of the song, it

    becomes apparent that the tonotopic organization of receptor

    terminals can also be interpreted as an odotopic map: there is

    a systematic relationship between the distance from which a

    song arises and the region of auditory neuropil where recep-

    tor activity is most intense (Fig. 2C). Similarly, the frequency

    selectivity of interneurons can also be seen as selectivity for

    songs arising from different distances. For example, a particu-

    lar interneurons dendrites are restricted to the low-frequency

    portion of the tonotopic map, and the neuron is accordingly

    tuned to low frequencies. The intensity of low frequency

    components in the song is low, and receptors tuned to low

    frequencies are rather insensitive.Consequently, this interneu-

    ron only responds to songs arising within 1015 m. By

    contrast, another interneuron with dendrites throughout most

    of the tonotopic array responds well to songs 30 m away.(31)

    Perhaps the most interesting interneurons, from the point

    of view of distance selectivity, are a group (of undetermined

    size) of morphologically indistinguishable neurons similar to

    that shown in Figure 1C. Even though their dendrites invade

    most of the auditory neuropil, different neurons within the

    group are maximally sensitive to sounds from different dis-

    tances (Fig. 2D). This selectivity thus comes about not

    through selective sampling of the receptor array, but instead

    is sculpted by inhibitory inputs from other interneurons (as

    yet unidentified) that are tuned to distances above and

    below a particular cells optimum distance.(31)

    Another role of frequency analysis is to sharpen selectivity

    for conspecific signals. The males calling song in another

    species of katydid, Ancistrura nigrovittata, has a fairly narrow

    spectrum, with energy concentrated between 10 and 17 kHz.Behavioral measurements show that females are most sensi-

    tive to frequencies in this range.(32) Thus, although a female

    can hear sounds over a wider range (perhaps allowing her to

    hear bats or other predators), her nervous system is neverthe-

    less specialized to detect and analyze frequencies in the

    narrower band used for intraspecific communication.A particu-

    Figure 2 . Tonotopic and odotopic organization of katydid

    receptor neurons. A: Drawings of parasagittal sections of the

    auditory neuropil, showing the terminal branches of receptors

    tuned to different frequencies, shown at right. (30) B: Threshold

    tuning curves of different receptors in M. markii. Each curve

    plots, for a single receptor neuron, the minumum sound

    intensity (threshold) needed to excite the receptor as a

    function of sound frequency. Each receptor is tuned to that

    frequency at which threshold is lowest.(31) C: The effect of

    distance to the source on the activation of receptors. Each

    curve in the graph corresponds to a different receptor. The

    sketches below illustrate the extent to which different regions

    of the auditory neuropil (shown in sagittal view) are activated

    by songs arising from different distances; darker areas are

    more heavily activated.(31) D: Distance selectivity of three

    different interneurons.(31)

    Re v ie w a r tic le s

    BioEssays 21.4 297

  • 7/31/2019 Pollack Imaizumi Bio Essays 99

    4/9

    lar interneuron, designated AN1 (Fig. 1C), is sharply tuned to

    the conspecific frequency range and may play a role in tuning

    the behavioral responses of females. AN1 is morphologically

    similar to the distance-selective interneurons of M. marki, and

    one is tempted to speculate that it may be a homologue of

    these. In both species, AN1s dendrites overlap with the

    terminals of most receptor neurons and frequency selectivity

    results mainly from inhibition by neurons tuned to higher and

    lower frequencies.(31,33,34) It thus appears likely that similar (or

    identical?) neural circuitry is used for different tasks in the two

    species; distance estimation in M. markii, and selectivity for

    conspecific song spectrum in A. nigrovittata.

    Frequency analysis in cricket s

    Cricket songs are dominated by a single narrow frequency

    band. This spectral purity (which is responsible for the songs

    musical quality) may be an adaptation that facilitates localiza-

    tion of the singer by conspecific listeners. Crickets determine

    sound direction by comparing sound intensity at the two

    ears.(35,36) The insects are small with respect to the wave-length of the dominant frequency in their songs (510 cm)

    and because of this a song originating on one side can readily

    diffract around the body and reach the offside ear with little or

    no attenuation. Interaural intensity differences are thus too

    small for accurate sound localization. Crickets solve this

    problem with an acoustical trick (the operation of which is

    beyond the scope of this article) that works only over a very

    narrow frequency range,(37) and the energy of the song is

    concentrated within this range. (The localization problem is

    presumably less severe for katydids, the songs of which

    generally include short-wavelength components that gener-

    ate larger interaural intensity differences.)

    The dominant frequency in the songs of the cricket

    species Teleogryllus oceanicus is about 4.5 kHz(38) and, not

    surprisingly, the insects are most sensitive to this fre-

    quency.(35) They are also highly sensitive to frequencies

    within the range 2550 kHz, however, even though these are

    present only at very low intensity in intraspecific signals.

    These ultrasonic frequencies are used by echolocating bats,

    which hunt insect prey on the wing by emitting ultrasonic cries

    and analyzing the echoes reflected by the insects bodies.

    When stimulated with ultrasound, T. oceanicus, like a number

    of other insects, exhibit escape responses.(1,35) Behavioral

    experiments indicate that T. oceanicus divide the entire

    audible range of frequencies, which extends at least from 3 to100 kHz, into only two frequency classes, namely low

    (cricket-like) and high (bat-like) frequencies, with the border

    between the two classes at around 15 kHz.(39)

    The dual specialization of the auditory system to analyze

    both cricket-like and bat-like sound frequencies is apparent at

    the level of interneurons, most of which show enhanced

    sensitivity to one or both of these frequency ranges.(4043) At

    the level of receptor neurons, representation of the two

    ranges is heavily skewed in favor of cricket-like stimuli.

    Recordings from the whole auditory nerve consist of com-

    pound action potentials (CAPs), which represent the summed

    extracellular potentials produced by synchronously firing

    receptor neurons. CAPs evoked by cricket-like frequenciesare 45 times larger than those evoked by ultrasound,

    suggesting that cricket-like frequencies excite more

    receptors(44)(Fig. 3A). A difference in CAP amplitude could

    also arise, however, if extracellular potentials were larger for

    low-frequency receptors, as would be the case if they had

    larger-diameter axons.(45) Larger axons would also be ex-

    pected to conduct action potentials more rapidly.(46) However,

    conduction velocity does not differ between low- and high-

    frequency CAPs (Fig. 3B), ruling out this explanation for the

    difference in their amplitude. Another possible explanation is

    that receptors that respond to low frequencies might be more

    tightly synchronized than those responding to ultrasound,

    resulting in more efficient summation of their individualpotentials. If this were the case, low-frequency CAPs would

    be expected to be shorter in duration, but they are not (Fig.

    3C). The difference in CAP amplitude thus arises simply

    because more receptors respond to low frequencies than to

    ultrasound. This conclusion is supported by single-unit record-

    ings. In a sample of 86 individual receptor neurons (recorded

    in as many crickets), 55 were tuned within the range 45.5

    kHz, whereas only 14 were tuned to 20 kHz.(47) Thus,

    although more than 80% of the neurons were tuned to one or

    the other of the two behaviorally relevant ranges, four times

    as many were tuned to the lower range.

    The numerical imbalance in receptors tuned to cricket-like

    and bat-like sounds is at odds with the equally robust

    behavioral and interneuronal responses to these stimuli.

    Crickets perform positive phonotaxis (locomotion toward the

    sound source) when stimulated with conspecific songs, and

    negative phonotaxis (locomotion away from the sound source)

    when stimulated with ultrasound.(35) Both responses are

    strong and reliable (although the threshold for negative

    phonotaxis is somewhat higher), and there is little hint of a

    four- to fivefold difference in the numbers of sensory neurons

    involved. Two main questions arise from this situation: (1)

    why are these two ranges represented unequally by recep-

    tors, and (2) how can only a few ultrasound-tuned receptors

    drive behavior as strongly as a much larger number ofneurons tuned to low-frequencies?

    Why are so few receptors tuned to ultrasound?

    There are no obvious functionaladvantages to detecting bat

    sounds with only a few receptors while devoting many more

    Re v ie w a r tic le s

    298 BioEssays 2 1.4

  • 7/31/2019 Pollack Imaizumi Bio Essays 99

    5/9

    to encoding intraspecific signals. Both tasks present similar

    problems, namely detecting, recognizing, and localizing the

    stimulus, and one might expect that these could be better

    accomplished with more rather than fewer receptors, if only

    because a larger number of receptor neurons would allow

    greater dynamic range and increased signal-to-noise ratio.

    Analysis of intraspecific signals does appear to be more

    demanding than bat detection, however, perhaps accounting

    in part for the differing numbers of receptors involved. For

    example, behavioral responses to cricket song, in addition to

    showing frequency selectivity, are tuned to the species-

    specific temporal structures of the signals, whereas the

    escape response elicited by ultrasound is less selective for

    temporal pattern.(48)

    Responses to intraspecific signals mayalso be oriented more precisely than escape responses.

    Although no direct comparison of the directional accuracy of

    these two responses is yet available, experiments on other

    model systems suggest that escape responses may be less

    precise than target-oriented behaviors.(49) Direct evidence

    that the sensory requirements to detect and evade bats are

    minimal is provided by some moths, which accomplish this

    with only a single receptor neuron in each ear.(50) So, one

    reason for the small number of ultrasound receptor neurons

    in crickets may simply be that only a few receptors are

    needed to get the job done.

    Another reason may be the different evolutionary histories

    of listening to conspecifics and listening to bats. Ears andsound-producing structures were present in the very earliest

    crickets, which appeared some 250300 million years

    ago.(21,51,52) Hearing in crickets has thus been associated with

    intraspecific communication throughout the entire course of

    this long evolutionary history. Bats, however, appeared only

    about 50 million years ago.(53) Selective pressure to detect

    and respond to bat cries thus arose relatively recently and

    must have been incorporated into a system that was already

    highly specialized for the analysis of intraspecific signals.

    One can imagine that this was accomplished by coopting

    some of the preexisting circuitry for this new task, and

    perhaps only a few receptors could be spared for this.

    How do only a few ultrasound receptors drive

    behavior as strongly as many low-frequency

    receptors?

    That a small number of ultrasound-tuned receptors can

    influence behavior as strongly as a much larger number that

    are tuned to low frequencies implies that each ultrasound

    neuron must speak more loudly to central neural circuits.

    Recent studies on an identified interneuron, ON1, have

    confirmed that this indeed takes place.

    The ON1 neurons (Fig. 1D) exist as a bilateral pair in the

    prothoracic ganglion. Each neuron receives auditory input on

    one side (the side on which its cell body is located) andprovides inhibitory input to a number of contralateral targets,

    including its mirror-image partner. ON1 enhances bilateral

    auditory contrast, thereby facilitating the crickets determina-

    tion of the direction of the sound source.(5456) ON1s fre-

    quency sensitivity mirrors that of behavioral responses,(41)

    suggesting that it receives input from the entire array of

    receptors.

    Although ON1 responds strongly to both cricket-like and

    bat-like sounds, recordings of its membrane potential show

    that the underlying physiological events differ for the two

    frequency ranges.(57) Cricket-like frequencies evoke excita-

    tory postsynaptic potentials (EPSPs) with smooth waveforms

    that increase in amplitude in a graded manner with increasing

    stimulus intensity (Fig. 4A). Ultrasound, by contrast, evokes

    trains of large, discrete EPSPs that increase in frequency with

    increasing intensity (Fig. 4B). A simple hypothesis to account

    for this difference in waveform is that the EPSPs evoked by

    low frequencies are compound responses, formed by the

    Figure 3. Unequal representation of cricket-like and bat-like

    soundsin cricket ears. A: Compound action potentials (CAPs)

    recorded from the auditory nerve are larger for stimuli with a

    cricket-like frequency (4.5 kHz) than for those with a bat-likefrequency (30 kHz). Numbers to the left indicate stimulus

    intensity relative to threshold.(44) B: Conduction velocities of

    CAPs are similar for the two frequencies, indicating that the

    difference in their amplitude is not due to differences in axon

    diameter.(44) C: CAPs evoked by 4.5 kHz (solid line) and 30

    kHz (dashed line) are similar in duration, indicating that the

    difference in amplitude (note scaling factors) is not due to a

    difference in the degree of synchronization among the contrib-

    uting receptor neurons.(44)

    Re v ie w a r tic le s

    BioEssays 21.4 299

  • 7/31/2019 Pollack Imaizumi Bio Essays 99

    6/9

    summation of a number of smaller EPSPs that cannot be

    resolved individually, but that the discrete EPSPs observed in

    response to ultrasound are unitary responses, each resulting

    from a single action potential in a single receptor neuron. The

    unitary nature of ultrasound-elicited EPSPs is confirmed in by

    simultaneous recordings from the auditory nerve and ON1. In

    Figure 4C, the large EPSPs were used as time markers to

    divide a long recording into segments, each corresponding to

    a single EPSP. These segments were then aligned according

    to the timing of the EPSP and averaged. The clear actionpotential in the nerve recording seen in the lower trace of

    Figure 4C survives the averaging because it is time-locked

    to the EPSP; other events in the nerve that occur randomly

    with respect to the EPSP average to zero. The constant

    latency between action potentials in the nerve and EPSPs in

    ON1 indicates that receptors tuned to ultrasound synapse

    directly onto ON1, and the large size of the EPSPs indicates

    that these connections are powerful. Another indication of

    their potency comes from comparing the timing of action

    potentials in ON1 and another ultrasound-sensitive interneu-

    ron, AN2. If single EPSPs are sufficiently large to bring both

    interneurons to firing threshold, the timing of action potentials

    in the interneurons would be determined in part by the timingof the EPSPs. If the same small population of receptors is

    responsible for the responses of both interneurons to ultra-

    sound, it follows that action potentials in the two interneurons

    would be temporally correlated. This is in fact observed, but

    only when the neurons are driven by ultrasound (Fig. 4D).

    These experiments show that the paucity of ultrasound

    receptor neurons is compensated by their increased synaptic

    potency.

    The cellular mechanisms for increased synaptic efficacy

    are unknown, but several possibilities present themselves.

    Perhaps the most obvious of these is differential passive

    filtering of inputs from low- and high-frequency receptors. The

    transmission of synaptic potentials through a neurons den-dritic arbor is affected by factors such as the diameters and

    lengths of neuronal branches. If ultrasound-tuned receptors

    terminated closer to the site at which action potentials are

    initiated in ON1, or if they terminated on larger-diameter

    processes, the EPSPs they produce would suffer less attenu-

    ation by the passive electrical properties of the postsynaptic

    cell and would thus have a greater influence on ON1s

    behavior. Recent anatomical studies have begun to address

    this possibility.

    Unlike katydids, in which receptor neurons form a continu-

    ous map in the central nervous system, cricket auditory

    receptor neurons fall into two distinct anatomical classes.

    One class terminates in a cluster of branches near the

    midline, in the same region where the best known auditory

    interneurons arborize most extensively (Fig. 5A). These

    receptor neurons are thus well positioned to make monosyn-

    aptic connections with the interneurons.(58,59)(Fig. 4C). Prelimi-

    nary data indicate that all the ultrasound-tuned receptors, and

    about one-third of those tuned to low frequencies, fall into this

    class.(60) Both low- and high-frequency neurons of this ana-

    tomical type terminate in essentially the same region, making

    differential passive filtering of their inputs unlikely (Fig. 5A).

    The second anatomical type, comprising the remaining low-

    frequency receptor neurons, has a characteristic bifurcating

    projection that has little or no overlap with the dendrites of thebest known auditory interneurons (Fig. 5B,C). Interestingly,

    this is the most frequently encountered type of receptor,

    suggesting that a substantial amount of sensory input to

    auditory interneurons may come not directly from sensory

    neurons but rather from interposed interneurons. So, part of

    the explanation for the differing potency of ultrasound and

    Figure 4. Difference in excitatory postsynaptic potentials

    (EPSPs) evoked by cricket-like and bat-like sounds. A: EP-

    SPs evoked in omega neuron 1 (ON1) by cricket-like stimuli

    arerelatively smooth andare gradedin intensity.(57) B: Bat-like

    sounds elicit summating trains of large, discrete EPSPs.(57) C:

    Large, ultrasound-elicited EPSPs (top trace) are preceded,

    with constant latency, by actionpotentials in the auditory nerve

    (bottom trace). This figure was prepared by aligning succes-

    sive segments of the recording according to the timing of theEPSPs, and then averaging across these segments. A total of

    770 segments of recording were averaged, derived from

    responses to a series of 50-dB, 30-kHz stimuli (G. Pollack,

    unpublished observations). D: Temporal coupling of spikes in

    two ultrasound-sensitive interneurons. Extracellular record-

    ings were made both from omega neuron 1 (ON1) (top two

    traces) and from another identified interneuron, ascending

    neuron 2 (AN2) (bottom trace), in response to both 5-kHz and

    30-kHz stimuli.(57) The traces are aligned according to the

    timing of the spikes in AN2. The top traceshows 23 superim-

    posed sweeps in a response to a 5-kHz, 100-dB stimulus. The

    spikes of ON1 occur nearly randomly with respect to those of

    AN2. The middle trace shows 25 sweeps in a response to 30

    kHz, 90 dB. Here, ON1 spikes are clustered around those of

    AN2, revealingtemporal coupling of spikesin the twoneurons.

    Re v ie w a r tic le s

    300 BioEssays 2 1.4

  • 7/31/2019 Pollack Imaizumi Bio Essays 99

    7/9

    low-frequency receptors may lie in differences in the design ofthe afferent circuitry.(61)

    Another part of the explanation may reside in the cellular

    properties of the receptors. Sensory terminals are character-

    ized by numerous swellings, or varicosities, which are be-

    lieved to be sites of transmitter release. Preliminary data

    indicate that receptors that are tuned to ultrasound have more

    varicosities than those tuned to low-frequencies, (60) suggest-

    ing that each action potential may release a greater amount of

    neurotransmitter. In addition, firing rates (i.e., number of

    action potentials per unit time) are about 2030% higher for

    ultrasound receptors.(47) This too might enhance their influ-

    ence on interneurons.

    Concluding remarks

    The examples summarized illustrate a few ways in which

    insect auditory systems implement the specializations that

    are required to analyze behaviorally important signals. Al-

    though this review has focused on only one suborder of

    insects, similar work is being pursued in other groups as well,

    including grasshoppers,(30,62) mantises(63) and flies.(64) And

    whereas we have focused here only on frequency analysis,

    these same insect models are being exploited to study other

    problems of hearing, such as the analysis of sound direction

    and of the temporal structure of stimuli.(23) What ties these

    various insects and investigations together is the explicit

    focus on the role of the auditory system in behavior. Naturalselection operates at the level of phenotype which, in the

    case of the nervous system, is behavior. Insects, with their

    wide array of auditory systems, each with their own special

    features that can be assigned clear behavioral functions,

    provide ideal material for studying the cellular mechanisms by

    which nervous systems respond to selection pressure.

    Ackn o wled g men ts

    We thank Dr. Zen Faulkes, Dr. Adam Wilkins and two

    anonymous referees for their helpful comments. Research

    not previously published was supported by grants (to G.S.P.)

    from NSERC and from the Whitehall Foundation.

    References1. Hoy RR, Robert D. Tympanal hearing in insects. Annu Rev Entomol

    1996;41:433450.

    2. Fullard JE, Yack JE. The evolutionary biology of insect hearing. Trends

    Ecol Evol 1993;8:248252.

    3. Oldfield BP. Tonotopic org anization of auditory recep tors in Tettigoniid ae

    (Orthoptera: Ensifera). J Comp Physiol A 1982;147:461469.

    4. Oldfield BP, Kleindienst HU, Huber F. Physiology and tonotopic organiza-

    tion of auditory receptors in the cricket Gryllus b imaculatus DeGeer. J

    Comp Physiol A 1986;159:457464.

    5. Stumpner A. Tonotopic org anization of the hearing organ in a bushcric ket.

    Physiological characterization and complete staining of auditory receptor

    cells. Naturwissenschaften 1996;83:8184.

    6. Stol ting H, Stumpner A. Tonotopic org anization of auditory recep tor cells

    in the bushcricket Pholidoptera griseoaptera (De Geer 1771) (Tettigoni-idae, Decticini). Cell Tissue Res 1998;294:377386.

    7. Bekesy G von. Experiments in hearing. New York: McGraw-Hill; 1960.

    p 745.

    8. Michelsen A. The physiology of the locust ear. I. Frequency sensitivity of

    single cells in the isolated ear. Z Vergl Physiol 1971;71:4962.

    9. Michelsen A. The physiology of thelocustear. II. Frequency discrimination

    based upon resonances in the tympanum. Z Vergl Physiol 1971;71:63

    101.

    Figure 5. Central projections of cricket auditory receptors. A:

    All ultrasound receptors (right) and some low-frequency receptors

    (left) terminate in a dense cluster of branches close to the midline,

    in the same region of the auditory neuropil in which well-described

    auditory interneurons arborize. B: The majority of low-frequency

    receptors terminate more laterally and posteriorly, and have few, ifany, branches that overlap with the dendrites of the well-described

    auditory interneurons. C: Confocal reconstruction of a low-

    frequency receptor of the type shown in B (green) and of

    processes of omega neuron 1 (ON1) (red), shown in horizon-

    tal view. The receptor neuron was stained with Lucifer Yellow,

    and ON1 was stained with neurobiotin/Texas-Red avidin. Most of

    the receptor terminals are lateral and posterior of ON1 processes.

    The fewreceptorbranches that appear to overlap with ON1 are, in

    fact, situated dorsal of ON1 dendrites (G. Pollack, unpublished

    observations). a, anterior; p, posterior; m, medial; l, lateral.

    Re v ie w a r tic le s

    BioEssays 21.4 301

  • 7/31/2019 Pollack Imaizumi Bio Essays 99

    8/9

    10. Michelsen A. Insect ears as mechanical systems. Am Sci 1979;67:696

    706.

    11. Paton JA, Capranica RR, Dragsten PR, Webb WW. Physical basis for

    auditory frequency analysis in field crickets (Gryllidae). J Comp Physiol A

    1977;119:221240.

    12. Michelsen A, Larsen ON. Biophysics of the ensiferan ear. I. Tympanal

    vibrations in bushcrickets (Tettigoniidae) studied with laser vibrometry. J

    Comp Physiol A 1978;123:193203.

    13. Larsen ON, Michelsen A. Biophysics of the ensiferan ear. III. The c ricketear as a four-input system. J Comp Physiol A 1978;123:217227.

    14. Bangert M, Kalmring K, Sickmann T, Stephen R, Jatho M, Lakes-Harlan R.

    Stimulus transmission in the auditory receptor organs of the forleg of

    bushcrickets (Tettigoniidae). I. The role of the tympana. Hearing Res

    1998;115:2738.

    15. Ball EE, Oldfield BP, Michel-Rudolph K. Auditoryorgan structure, develop-

    ment, and function. In: Huber F, Moore TE, Loher W, editors. Cricket

    behavior and neurobiology. Ithaca: Comstock; 1989. p 391422.

    16. Lakes R, Schikorski T. Neuroanatomy of tettigoniid s. In: Bailey WJ, Rentz

    DCF, editors.The Tettigoniidae: biology, systematics and evolution. Berlin:

    Springer-Verlag; 1990. p 166190.

    17. Kalmring K, Jatho M. The effectsof blocking inputs of theacoustic trachea

    on the frequency tuning of primary auditory receptors in two species of

    tettigoniids. J Exp Zool 1994;270:360371.

    18. Oldfield BP. The tuning of auditory receptors in bushc rickets. Hearing Res

    1985;17:2735.

    19. Michelsen A, Larsen ON. Hearing and sound. In: Kerkut GA, Gilbert LJ,

    editors. Comprehensive insect physiology, biochemistry and pharmacol-

    ogy, Vol 6. New York: Pergamon Press; 1985. p 495556.

    20. Ewing AR. Arthropod bioacoustics. Neurobiology and behaviour. Ithaca,

    NY: Cornell University Press; 1989. 260 pp.

    21. SharovAG. Phylogenyof the Orthopteroidea. (Transl1971, IsraelProgram

    for Scientific Translation, Jerusalem.) Trans Paleontol Inst Acad Sci USSR

    1968;118:1216.

    22. Gwynne DR.Phylogeny of the Ensifera (Orthoptera):a hypothesis support-

    ing multipleoriginsof acoustical signalling, complex spermatophores and

    maternal care in crickets, katydids, and weta. J Orthop Res 1995;4:203

    218.

    23. Pollack GS. Neural processing of ac oustic signals. In: Hoy RR, Popper

    AN, Fay RR, editors. Comparative hearing: insects. Springer handbook of

    auditory research. Berlin: Springer-Verlag; 1998. p 139196.

    24. Romer H. 1993;Environmental and b iologic al constraints for the evolution

    of long-range signalling and hearing in acoustic insects. Philos Trans RSoc Lond B 340:179185.

    25. Bennet-Clark HC. Size and scale effects as c onstraints in insect sound

    communication. Philos Trans R Soc Lond B 1998;353:407419.

    26. Thiele D, Bailey WJ. The function of sound in male spacing behaviour in

    bush-crickets (Tettigoniidae, Orthoptera). Aust J Ecol 1980;5:275286.

    27. Romer H, BaileyWJ. Insect hearing in thefield. II. Male spacing behaviour

    and correlated acoustic cues in the bushcricket Mygalopsis marki. J

    Comp Physiol A 1986;159:627638.

    28. Oldfield BP. Central projections of primary auditory fibres in Tettigoniidae

    (Orthoptera; Ensifera). J Comp Physiol A 1983;151:389395.

    29. Romer H. Tonotopic organization of the auditory neuropile in the bush-

    cricket Tettigonia viridissima. Nature 1983;306:6062.

    30. Romer H, Marquart V, Hardt M. Organization of a sensory neurop ile in the

    auditory pathway of two g roups of orthoptera. J Comp Neurol 1988;275:

    201215.

    31. Romer H. Representation of auditory distanc e within a central neuropil of

    the bushcricket Mygalopsis marki. J Comp Physiol A 1987;161:3342.

    32. Dobler S, StumpnerA, Heller K-G. Acoustic communication in theduetting

    bushcricket Ancistrura nigrovittata(Orthoptera, Phaneropteridae). II. Sex

    specific b ehavioural tuning for the partners song. J Comp Physiol A

    1994;175:303310.

    33. StumpnerA. Anauditory interneurone tuned to themale song frequency in

    the duetting bushcricket Ancistrura nigrovittata(Orthopera, Phaneropteri-

    dae). J Exp Biol 1997;200:10891011.

    34. Stumpner A. Picrotoxin eliminates frequency selectivity of an auditory

    interneuron in a bushcricket. J Neurophysiol 1998;79:24082415.

    35. Moiseff A, Pollack GS, Hoy RR. Steering responses of flying crickets to

    sound and ultrasound: mate attraction and predator avoidance. Proc Natl

    Acad Sci USA 1978;75:40524056.

    36. Schmitz B, Scharstein H, Wendler G. Phonotaxis in Gryllus c ampestrisL.

    (Orthoptera, Gryllidae). II. Acoustic orientation of female c rickets after

    occlusion of single sound entrances. J Comp Physiol A 1982;152:257

    264.37. Michelsen A, Popov AV, Lewis B. Physics of d irectional hearing in the

    cricket Gryllus bimaculatus. J Comp Physiol A 1994;175:153164.

    38. Balakrishnan R, Pollack GS. Recognition of courtship song in the field

    cricket, Teleogryllus oc eanicus. Anim Behav 1996;51:353366.

    39. Wyttenbach RA, May ML, Hoy RR. Categorical perception of sound

    frequency by crickets. Science 1996;273:15421544.

    40. Moiseff A, Hoy RR. Sensitivity to ultrasound in an identified auditory

    interneuron in the cricket: a possible neural link to phonotactic behavior. J

    Comp Physiol A 1983;131:113120.

    41. Atkins G, Pollack GS. Age dependent occurrence of an ascending axon

    on the omega neuron of the cricket, Teleogryllus oc eanicus. J Comp

    Neurol 1986;243:527534.

    42. Atkins G, Pollack GS. Response properties of p rothoracic, intergangli-

    onic, sound-activated interneurons in the cricket Teleogryllus oceanicus. J

    Comp Physiol A 1987;161:681693.

    43. BrodfuehrerPD, Hoy RR. Ultrasound sensitive neurons in the cricket brain.J Comp Physiol A 1990;166:651662.

    44. Pollack GS, Faulkes Z. Representation of behaviorally relevant sound

    frequencies by auditory receptors in the cricket Teleogryllus oceanicus. J

    Exp Biol 1998;201:155163.

    45. Rushton WAH. A theory of the effects of fibre size in medullated nerve. J

    Physiol 1951;115:101122.

    46. Hodgkin AL. A note on cond uction velocity. J Physiol 1954;125:221224.

    47. Imaizumi K, Pollack GS. Physiological properties of auditory receptors in

    the Australian field cricket Teleogryllus oc eanicus. Soc Neurosci Abs

    1997;23:1571.

    48. Pollack GS, El-Feghaly E.Calling song recognition in the cricket Teleogryl-

    lus oceanicus: c omparison of the effects of stimulus intensity and sound

    spectrum on selectivity for temporal pattern. J Comp Physiol A 1993;171:

    759765.

    49. Roche King J, Comer CM. Visually elicited turning behavior in Rana

    pipien: comparative organization and neural control of escape and prey

    capture. J Comp Physiol A 1996;178:293305.

    50. Surlykke A. Hearing in notodontid moths: a tympanic organ with a single

    auditory neurone. J Exp Biol 1984;113:1128.

    51. Hoy RR. The evolution of hearing in insects as an adaptation to predation

    from bats. In: Webster DB, Fay RR, Popper AN, editors. The evolutionary

    biology of hearing. Berlin: Springer-Verlag; 1992. p 115129.

    52. Otte D. Evolution of cricket songs. J Orthop Res 1992;1:2446.

    53. Novacek MJ. Mammalian phylogeny: shaking the tree. Nature 1992;356:

    121125.

    54. Selverston AI, Kleindienst HU, Huber F. Synaptic connectivity between

    cricket auditory interneurons as studied b y selective p hotoinactivation. J

    Neurosci 1985; 5:12831292.

    55. Horseman G, Huber F. Sound localisation in crickets. I. Contralateral

    inhibition of an ascending auditoryinterneuron (AN1) in thec ricket Gryllus

    bimaculatus. J Comp Physiol A 1994a; 174:389398.

    56. Horseman G, Huber F. Sound localisation in cric kets. II. Modeling the role

    of a simple neural network in the prothoracic ganglion. J Comp Physiol A

    1994b;175:399413.

    57. Pollack GS. Synaptic inputs to the omega neuron of the cricket Teleogryl-

    lus oceanicus: differences in EPSP waveforms evoked by low and high

    sound frequencies. J Comp Physiol A 1994;174:8389.

    58. Hennig RM. Ascending auditory interneurons in the cricket Teleogryllus

    commodus(Walker): comparative physiology and direct connections with

    afferents. J Comp Physiol A 1988;163:135143.

    Re v ie w a r tic le s

    302 BioEssays 2 1.4

  • 7/31/2019 Pollack Imaizumi Bio Essays 99

    9/9

    59. Hirtz R, Wiese K. Ultrastructure of synaptic contacts b etween identified

    neurons of the auditory pathway in Gryllus bimaculatusDe Geer. J Comp

    Neurol 1997;386:347357.

    60. Imaizumi K, Pollack GS. Anatomy and physiology of auditory receptors in

    the Australian field cricket Teleogryllus oc eanicus. Soc Neurosci Abs

    1996;22:1082.

    61. Faulkes Z, Pollack GS. Sound frequency specific responses of cricket

    omega neuron. Soc Neurosci Abs 1997;23:1570.

    62. Halex H, Kaiser W, Kalmring K. Projectionareas and branching patterns of

    the tympanal receptor cells in migratory locusts, Locusta migratoria and

    Schistocerca gregaria. Cell Tissue Res 1988;253:517528.

    63. Yager DD. Serially homologous ears p erform frequency range fraction-

    ation in the paying mantis, Creobroter (Mantodea, Hymenopodidae). J

    Comp Physiol A 1996;178:463475.

    64. Oshinksy ML, Hoy RR. Response properti es of auditory afferents in the fly

    Ormia ochracea. In: Burrows M, Matheson T, Newland PL, Schuppe J,

    editors. Nervous systems and behaviour. Proceedings of the fourth

    international congress of neuroethology. Stuttgart: Georg Thieme Verlag;

    1995. p 359.

    65. Wohlers DW, Huber F. Intracellular recording and staining of c ricket

    auditory interneurons (Gryllus campestris L., Gryllus bimaculatus De-

    Geer). J C omp Physiol 1978;127:1128.

    Re v ie w a r tic le s

    BioEssays 21.4 303